d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
5882166b78
freebsd-dev/sys/dev/iwn/if_iwn.c

8989 lines
247 KiB
C
Raw Normal View History

Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
/*-
* Sort the copyright lines by date * Ok ok, I've touched this enough to claim part of it.
2013-12-02 05:45:11 +00:00
* Copyright (c) 2007-2009 Damien Bergamini <damien.bergamini@free.fr>
* Copyright (c) 2008 Benjamin Close <benjsc@FreeBSD.org>
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
* Copyright (c) 2008 Sam Leffler, Errno Consulting
* Sort the copyright lines by date * Ok ok, I've touched this enough to claim part of it.
2013-12-02 05:45:11 +00:00
* Copyright (c) 2011 Intel Corporation
* Copyright (c) 2013 Cedric GROSS <c.gross@kreiz-it.fr>
* Copyright (c) 2013 Adrian Chadd <adrian@FreeBSD.org>
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
/*
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
* Driver for Intel WiFi Link 4965 and 1000/5000/6000 Series 802.11 network
* adapters.
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Include opt_wlan.h so the net80211 RX debug API compiles correctly.
2012-11-30 03:14:11 +00:00
#include "opt_wlan.h"
Now that conf/options knows about if_iwn.h, add it to if_iwn.c. This allows for IWN_DEBUG (and maybe more stuff later) to be a build time configure option.
2013-08-01 21:50:50 +00:00
#include "opt_iwn.h"
Include opt_wlan.h so the net80211 RX debug API compiles correctly.
2012-11-30 03:14:11 +00:00
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
#include <sys/param.h>
#include <sys/sockio.h>
#include <sys/sysctl.h>
#include <sys/mbuf.h>
#include <sys/kernel.h>
#include <sys/socket.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/bus.h>
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
#include <sys/conf.h>
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
#include <sys/rman.h>
#include <sys/endian.h>
#include <sys/firmware.h>
#include <sys/limits.h>
#include <sys/module.h>
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
#include <sys/priv.h>
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
#include <sys/queue.h>
#include <sys/taskqueue.h>
#include <machine/bus.h>
#include <machine/resource.h>
#include <machine/clock.h>
#include <dev/pci/pcireg.h>
#include <dev/pci/pcivar.h>
#include <net/if.h>
The r48589 promised to remove implicit inclusion of if_var.h soon. Prepare to this event, adding if_var.h to files that do need it. Also, include all includes that now are included due to implicit pollution via if_var.h Sponsored by: Netflix Sponsored by: Nginx, Inc.
2013-10-26 17:58:36 +00:00
#include <net/if_var.h>
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
#include <net/if_dl.h>
#include <net/if_media.h>
#include <netinet/in.h>
#include <netinet/if_ether.h>
#include <net80211/ieee80211_var.h>
#include <net80211/ieee80211_radiotap.h>
#include <net80211/ieee80211_regdomain.h>
net80211 rate control framework (net80211 ratectl). This framework allows drivers to abstract the rate control algorithm and just feed the framework with the usable parameters. The rate control framework will now deal with passing the parameters to the selected algorithm. Right now we have AMRR (the default) and RSSADAPT but there's no way to select one with ifconfig, yet. The objective is to have more rate control algorithms in the net80211 stack so all drivers[0] can use it. Ideally, we'll have the well-known sample rate control algorithm in the net80211 at some point so all drivers can use it (not just ath). [0] all drivers that do rate control in software, that is. Reviewed by: bschmidt, thompsa, weyongo MFC after: 1 months
2010-04-07 15:29:13 +00:00
#include <net80211/ieee80211_ratectl.h>
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
#include <dev/iwn/if_iwnreg.h>
#include <dev/iwn/if_iwnvar.h>
Break out the iwn(4) device IDs into if_iwn_devid.h, as well as add IDs for new devices. * Add new device IDs * Extend the ID probe code to include the newer range of bits used by later model devices Tested: * Intel 5100, STA mode TODO: * Test on Intel 4965, just to be sure Submitted by: Cedric GROSS <cg@gross.info>
2013-08-02 21:23:28 +00:00
#include <dev/iwn/if_iwn_devid.h>
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
#include <dev/iwn/if_iwn_chip_cfg.h>
Break out the debug code into a new include file in preparation for some more iwn work.
2013-10-24 01:03:42 +00:00
#include <dev/iwn/if_iwn_debug.h>
Track and expose the latest statistics from the firmware. Tested: * Intel Centrino 6205
2014-02-24 02:37:04 +00:00
#include <dev/iwn/if_iwn_ioctl.h>
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
struct iwn_ident {
uint16_t vendor;
uint16_t device;
const char *name;
};
Now that all bits are in for 1030/6230 adapters enable those. While here pull the adapter names from the Linux driver and sort the list by ID.
2011-04-20 17:49:05 +00:00
static const struct iwn_ident iwn_ident_table[] = {
Break out the iwn(4) device IDs into if_iwn_devid.h, as well as add IDs for new devices. * Add new device IDs * Extend the ID probe code to include the newer range of bits used by later model devices Tested: * Intel 5100, STA mode TODO: * Test on Intel 4965, just to be sure Submitted by: Cedric GROSS <cg@gross.info>
2013-08-02 21:23:28 +00:00
{ 0x8086, IWN_DID_6x05_1, "Intel Centrino Advanced-N 6205" },
{ 0x8086, IWN_DID_1000_1, "Intel Centrino Wireless-N 1000" },
{ 0x8086, IWN_DID_1000_2, "Intel Centrino Wireless-N 1000" },
{ 0x8086, IWN_DID_6x05_2, "Intel Centrino Advanced-N 6205" },
{ 0x8086, IWN_DID_6050_1, "Intel Centrino Advanced-N + WiMAX 6250" },
{ 0x8086, IWN_DID_6050_2, "Intel Centrino Advanced-N + WiMAX 6250" },
{ 0x8086, IWN_DID_x030_1, "Intel Centrino Wireless-N 1030" },
{ 0x8086, IWN_DID_x030_2, "Intel Centrino Wireless-N 1030" },
{ 0x8086, IWN_DID_x030_3, "Intel Centrino Advanced-N 6230" },
{ 0x8086, IWN_DID_x030_4, "Intel Centrino Advanced-N 6230" },
{ 0x8086, IWN_DID_6150_1, "Intel Centrino Wireless-N + WiMAX 6150" },
{ 0x8086, IWN_DID_6150_2, "Intel Centrino Wireless-N + WiMAX 6150" },
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
{ 0x8086, IWN_DID_2x00_1, "Intel(R) Centrino(R) Wireless-N 2200 BGN" },
{ 0x8086, IWN_DID_2x00_2, "Intel(R) Centrino(R) Wireless-N 2200 BGN" },
/* XXX 2200D is IWN_SDID_2x00_4; there's no way to express this here! */
Break out the iwn(4) device IDs into if_iwn_devid.h, as well as add IDs for new devices. * Add new device IDs * Extend the ID probe code to include the newer range of bits used by later model devices Tested: * Intel 5100, STA mode TODO: * Test on Intel 4965, just to be sure Submitted by: Cedric GROSS <cg@gross.info>
2013-08-02 21:23:28 +00:00
{ 0x8086, IWN_DID_2x30_1, "Intel Centrino Wireless-N 2230" },
{ 0x8086, IWN_DID_2x30_2, "Intel Centrino Wireless-N 2230" },
{ 0x8086, IWN_DID_130_1, "Intel Centrino Wireless-N 130" },
{ 0x8086, IWN_DID_130_2, "Intel Centrino Wireless-N 130" },
{ 0x8086, IWN_DID_100_1, "Intel Centrino Wireless-N 100" },
{ 0x8086, IWN_DID_100_2, "Intel Centrino Wireless-N 100" },
Add support for the Intel Centrino Wireless-N 105 chipset. MFC after: 2 weeks Relnotes: yes
2014-05-27 19:39:57 +00:00
{ 0x8086, IWN_DID_105_1, "Intel Centrino Wireless-N 105" },
{ 0x8086, IWN_DID_105_2, "Intel Centrino Wireless-N 105" },
Add support for the Intel Centrino Wireless-N 135 chipset. MFC after: 2 weeks
2014-01-08 13:59:33 +00:00
{ 0x8086, IWN_DID_135_1, "Intel Centrino Wireless-N 135" },
{ 0x8086, IWN_DID_135_2, "Intel Centrino Wireless-N 135" },
Break out the iwn(4) device IDs into if_iwn_devid.h, as well as add IDs for new devices. * Add new device IDs * Extend the ID probe code to include the newer range of bits used by later model devices Tested: * Intel 5100, STA mode TODO: * Test on Intel 4965, just to be sure Submitted by: Cedric GROSS <cg@gross.info>
2013-08-02 21:23:28 +00:00
{ 0x8086, IWN_DID_4965_1, "Intel Wireless WiFi Link 4965" },
{ 0x8086, IWN_DID_6x00_1, "Intel Centrino Ultimate-N 6300" },
{ 0x8086, IWN_DID_6x00_2, "Intel Centrino Advanced-N 6200" },
{ 0x8086, IWN_DID_4965_2, "Intel Wireless WiFi Link 4965" },
{ 0x8086, IWN_DID_4965_3, "Intel Wireless WiFi Link 4965" },
{ 0x8086, IWN_DID_5x00_1, "Intel WiFi Link 5100" },
{ 0x8086, IWN_DID_4965_4, "Intel Wireless WiFi Link 4965" },
{ 0x8086, IWN_DID_5x00_3, "Intel Ultimate N WiFi Link 5300" },
{ 0x8086, IWN_DID_5x00_4, "Intel Ultimate N WiFi Link 5300" },
{ 0x8086, IWN_DID_5x00_2, "Intel WiFi Link 5100" },
{ 0x8086, IWN_DID_6x00_3, "Intel Centrino Ultimate-N 6300" },
{ 0x8086, IWN_DID_6x00_4, "Intel Centrino Advanced-N 6200" },
{ 0x8086, IWN_DID_5x50_1, "Intel WiMAX/WiFi Link 5350" },
{ 0x8086, IWN_DID_5x50_2, "Intel WiMAX/WiFi Link 5350" },
{ 0x8086, IWN_DID_5x50_3, "Intel WiMAX/WiFi Link 5150" },
{ 0x8086, IWN_DID_5x50_4, "Intel WiMAX/WiFi Link 5150" },
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
{ 0x8086, IWN_DID_6035_1, "Intel Centrino Advanced 6235" },
{ 0x8086, IWN_DID_6035_2, "Intel Centrino Advanced 6235" },
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
{ 0, 0, NULL }
};
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
static int iwn_probe(device_t);
static int iwn_attach(device_t);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
static int iwn4965_attach(struct iwn_softc *, uint16_t);
static int iwn5000_attach(struct iwn_softc *, uint16_t);
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
static int iwn_config_specific(struct iwn_softc *, uint16_t);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_radiotap_attach(struct iwn_softc *);
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
static void iwn_sysctlattach(struct iwn_softc *);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
static struct ieee80211vap *iwn_vap_create(struct ieee80211com *,
Fix some net80211 enum nits: - ic_vap_create() uses an ieee80211_opmode argument - ieee80211_rate2media() takes an ieee80211_phymode argument - ieee80211_plcp2rate() takes an ieee80211_phytype argument - cast to enum ieee80211_protmode and ieee80211_roamingmode to silence compiler warnings Submitted by: arundel@
2011-12-17 10:23:17 +00:00
const char [IFNAMSIZ], int, enum ieee80211_opmode, int,
const uint8_t [IEEE80211_ADDR_LEN],
const uint8_t [IEEE80211_ADDR_LEN]);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
static void iwn_vap_delete(struct ieee80211vap *);
Add some missing protoypes. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:46:42 +00:00
static int iwn_detach(device_t);
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
static int iwn_shutdown(device_t);
static int iwn_suspend(device_t);
static int iwn_resume(device_t);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn_nic_lock(struct iwn_softc *);
static int iwn_eeprom_lock(struct iwn_softc *);
static int iwn_init_otprom(struct iwn_softc *);
static int iwn_read_prom_data(struct iwn_softc *, uint32_t, void *, int);
Add some missing protoypes. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:46:42 +00:00
static void iwn_dma_map_addr(void *, bus_dma_segment_t *, int, int);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
static int iwn_dma_contig_alloc(struct iwn_softc *, struct iwn_dma_info *,
Remove the flags argument of iwn_dma_contig_alloc(), it is always set as BUS_DMA_NOWAIT. While here also set BUS_DMA_COHERENT.
2011-04-16 08:57:17 +00:00
void **, bus_size_t, bus_size_t);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
static void iwn_dma_contig_free(struct iwn_dma_info *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn_alloc_sched(struct iwn_softc *);
static void iwn_free_sched(struct iwn_softc *);
static int iwn_alloc_kw(struct iwn_softc *);
static void iwn_free_kw(struct iwn_softc *);
static int iwn_alloc_ict(struct iwn_softc *);
static void iwn_free_ict(struct iwn_softc *);
static int iwn_alloc_fwmem(struct iwn_softc *);
static void iwn_free_fwmem(struct iwn_softc *);
static int iwn_alloc_rx_ring(struct iwn_softc *, struct iwn_rx_ring *);
static void iwn_reset_rx_ring(struct iwn_softc *, struct iwn_rx_ring *);
static void iwn_free_rx_ring(struct iwn_softc *, struct iwn_rx_ring *);
static int iwn_alloc_tx_ring(struct iwn_softc *, struct iwn_tx_ring *,
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
int);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_reset_tx_ring(struct iwn_softc *, struct iwn_tx_ring *);
static void iwn_free_tx_ring(struct iwn_softc *, struct iwn_tx_ring *);
static void iwn5000_ict_reset(struct iwn_softc *);
static int iwn_read_eeprom(struct iwn_softc *,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint8_t macaddr[IEEE80211_ADDR_LEN]);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn4965_read_eeprom(struct iwn_softc *);
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
#ifdef IWN_DEBUG
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn4965_print_power_group(struct iwn_softc *, int);
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
#endif
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn5000_read_eeprom(struct iwn_softc *);
Add some missing protoypes. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:46:42 +00:00
static uint32_t iwn_eeprom_channel_flags(struct iwn_eeprom_chan *);
static void iwn_read_eeprom_band(struct iwn_softc *, int);
static void iwn_read_eeprom_ht40(struct iwn_softc *, int);
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
static void iwn_read_eeprom_channels(struct iwn_softc *, int, uint32_t);
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
static struct iwn_eeprom_chan *iwn_find_eeprom_channel(struct iwn_softc *,
struct ieee80211_channel *);
static int iwn_setregdomain(struct ieee80211com *,
struct ieee80211_regdomain *, int,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
struct ieee80211_channel[]);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_read_eeprom_enhinfo(struct iwn_softc *);
static struct ieee80211_node *iwn_node_alloc(struct ieee80211vap *,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
const uint8_t mac[IEEE80211_ADDR_LEN]);
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
static void iwn_newassoc(struct ieee80211_node *, int);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn_media_change(struct ifnet *);
static int iwn_newstate(struct ieee80211vap *, enum ieee80211_state, int);
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
static void iwn_calib_timeout(void *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_rx_phy(struct iwn_softc *, struct iwn_rx_desc *,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_rx_data *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_rx_done(struct iwn_softc *, struct iwn_rx_desc *,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_rx_data *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_rx_compressed_ba(struct iwn_softc *, struct iwn_rx_desc *,
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
struct iwn_rx_data *);
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
static void iwn5000_rx_calib_results(struct iwn_softc *,
struct iwn_rx_desc *, struct iwn_rx_data *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_rx_statistics(struct iwn_softc *, struct iwn_rx_desc *,
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_rx_data *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn4965_tx_done(struct iwn_softc *, struct iwn_rx_desc *,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_rx_data *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn5000_tx_done(struct iwn_softc *, struct iwn_rx_desc *,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_rx_data *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_tx_done(struct iwn_softc *, struct iwn_rx_desc *, int,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint8_t);
Inform the rate control code if a single frame AMPDU transmission succeeds but has some retries. Without this, single frame transmission in AMPDU will always look like it succeeded fine, and thus AMRR will think it's totally fine to just keep upping the rate upwards. Now, this is still not quite right! For multi-frame aggregates the completion happens in two parts - the TX done and the BA received. The driver is currently double accounting those a little - there's no way to say to the rate control code "I completed X frames, Y worked fine, there were Z retries." And it's a bit odd with iwn, as the firmware retransmits frames for us so we don't get to see how many retransmits happened; only that it took longer than normal. I may have to extend the rate control API to properly track that. So this may keep the rate lower than it should be, but that's better than keeping it higher than it should be. Tested: * 5100, STA mode
2014-08-28 07:44:59 +00:00
static void iwn_ampdu_tx_done(struct iwn_softc *, int, int, int, int, void *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_cmd_done(struct iwn_softc *, struct iwn_rx_desc *);
static void iwn_notif_intr(struct iwn_softc *);
static void iwn_wakeup_intr(struct iwn_softc *);
static void iwn_rftoggle_intr(struct iwn_softc *);
static void iwn_fatal_intr(struct iwn_softc *);
static void iwn_intr(void *);
static void iwn4965_update_sched(struct iwn_softc *, int, int, uint8_t,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint16_t);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn5000_update_sched(struct iwn_softc *, int, int, uint8_t,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint16_t);
iwn5000_reset_sched() is not used; #ifdef it. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:47:46 +00:00
#ifdef notyet
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn5000_reset_sched(struct iwn_softc *, int, int);
iwn5000_reset_sched() is not used; #ifdef it. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:47:46 +00:00
#endif
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn_tx_data(struct iwn_softc *, struct mbuf *,
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
struct ieee80211_node *);
static int iwn_tx_data_raw(struct iwn_softc *, struct mbuf *,
struct ieee80211_node *,
const struct ieee80211_bpf_params *params);
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
static void iwn_xmit_task(void *arg0, int pending);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
static int iwn_raw_xmit(struct ieee80211_node *, struct mbuf *,
const struct ieee80211_bpf_params *);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
static int iwn_transmit(struct ieee80211com *, struct mbuf *);
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
static void iwn_watchdog(void *);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
static int iwn_ioctl(struct ieee80211com *, u_long , void *);
static void iwn_parent(struct ieee80211com *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn_cmd(struct iwn_softc *, int, const void *, int, int);
static int iwn4965_add_node(struct iwn_softc *, struct iwn_node_info *,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
int);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn5000_add_node(struct iwn_softc *, struct iwn_node_info *,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
int);
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
static int iwn_set_link_quality(struct iwn_softc *,
struct ieee80211_node *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn_add_broadcast_node(struct iwn_softc *, int);
Rename some stuff in favour of the OpenBSD names: - prefer EDCA over WME - qid for a TXQ ID - reg for register values
2011-04-16 13:40:32 +00:00
static int iwn_updateedca(struct ieee80211com *);
Change three methods in struct ieee80211com, namely ic_updateslot, ic_update_mcast and ic_update_promisc, to pass pointer to the ieee80211com, not to the ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 19:53:29 +00:00
static void iwn_update_mcast(struct ieee80211com *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_set_led(struct iwn_softc *, uint8_t, uint8_t, uint8_t);
static int iwn_set_critical_temp(struct iwn_softc *);
static int iwn_set_timing(struct iwn_softc *, struct ieee80211_node *);
static void iwn4965_power_calibration(struct iwn_softc *, int);
static int iwn4965_set_txpower(struct iwn_softc *,
Revert iwn channel-argument handling of set_txpower() to r201822 Pass the channel argument as a real argument and not through RXON. The RXON version seems to have problem with both the older 228.57.2.23 firmware and the latest 228.61.2.24 version resulting in device initialization errors like: wpa_supplicant[2928]: Failed to initiate AP scan. kernel: firmware error log: kernel: error type = "SYSASSERT" (0x00000005) kernel: program counter = 0x0000147C kernel: source line = 0x0000058B kernel: error data = 0x0000058B00000000 kernel: branch link = 0x0000145A00001492 kernel: interrupt link = 0x000006DE00000000 kernel: time = 7310 ... kernel: iwn0: iwn_config: could not set TX power kernel: iwn0: iwn_init_locked: could not configure device, error 35 By passing the current channel to hal->set_txpower() the firmware error is fixed, at least for the 4965 chipset of my Thinkpad and the ones tested by Bernhard. Submitted by: Bernhard Schmidt <bschmidt@techwires.net> Approved by: rpaulo
2010-01-09 04:00:22 +00:00
struct ieee80211_channel *, int);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn5000_set_txpower(struct iwn_softc *,
Revert iwn channel-argument handling of set_txpower() to r201822 Pass the channel argument as a real argument and not through RXON. The RXON version seems to have problem with both the older 228.57.2.23 firmware and the latest 228.61.2.24 version resulting in device initialization errors like: wpa_supplicant[2928]: Failed to initiate AP scan. kernel: firmware error log: kernel: error type = "SYSASSERT" (0x00000005) kernel: program counter = 0x0000147C kernel: source line = 0x0000058B kernel: error data = 0x0000058B00000000 kernel: branch link = 0x0000145A00001492 kernel: interrupt link = 0x000006DE00000000 kernel: time = 7310 ... kernel: iwn0: iwn_config: could not set TX power kernel: iwn0: iwn_init_locked: could not configure device, error 35 By passing the current channel to hal->set_txpower() the firmware error is fixed, at least for the 4965 chipset of my Thinkpad and the ones tested by Bernhard. Submitted by: Bernhard Schmidt <bschmidt@techwires.net> Approved by: rpaulo
2010-01-09 04:00:22 +00:00
struct ieee80211_channel *, int);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn4965_get_rssi(struct iwn_softc *, struct iwn_rx_stat *);
static int iwn5000_get_rssi(struct iwn_softc *, struct iwn_rx_stat *);
static int iwn_get_noise(const struct iwn_rx_general_stats *);
static int iwn4965_get_temperature(struct iwn_softc *);
static int iwn5000_get_temperature(struct iwn_softc *);
static int iwn_init_sensitivity(struct iwn_softc *);
static void iwn_collect_noise(struct iwn_softc *,
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
const struct iwn_rx_general_stats *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn4965_init_gains(struct iwn_softc *);
static int iwn5000_init_gains(struct iwn_softc *);
static int iwn4965_set_gains(struct iwn_softc *);
static int iwn5000_set_gains(struct iwn_softc *);
static void iwn_tune_sensitivity(struct iwn_softc *,
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
const struct iwn_rx_stats *);
Begin fleshing out some code to handle tracking PLCP error rates in preparation for the scan based retune logic. The linux iwlwifi driver does a rescan (onto a non-active channel) to force an RF retune when the PLCP error rates exceed a certain threshold. * Add code to track HT PLCP rate errors; * Separate out the PLCP error count fetch and update so the delta can be used when checking for PLCP error rates; * Implement the PLCP error logic from iwlwifi; * For now, just print out whenever the error rate exceeds the threshold. The actual scan based retune will take a bit more effort; the scan command code right now assumes that a scan state is passed in. This does need to change to be more flexible (both for this and in preparation for scanning multiple channels at once.) Tested: * 5100 (STA mode) * 2200 (STA mode) * 2230 (STA mode)
2013-12-07 08:03:10 +00:00
static void iwn_save_stats_counters(struct iwn_softc *,
const struct iwn_stats *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn_send_sensitivity(struct iwn_softc *);
Begin fleshing out some code to handle tracking PLCP error rates in preparation for the scan based retune logic. The linux iwlwifi driver does a rescan (onto a non-active channel) to force an RF retune when the PLCP error rates exceed a certain threshold. * Add code to track HT PLCP rate errors; * Separate out the PLCP error count fetch and update so the delta can be used when checking for PLCP error rates; * Implement the PLCP error logic from iwlwifi; * For now, just print out whenever the error rate exceeds the threshold. The actual scan based retune will take a bit more effort; the scan command code right now assumes that a scan state is passed in. This does need to change to be more flexible (both for this and in preparation for scanning multiple channels at once.) Tested: * 5100 (STA mode) * 2200 (STA mode) * 2230 (STA mode)
2013-12-07 08:03:10 +00:00
static void iwn_check_rx_recovery(struct iwn_softc *, struct iwn_stats *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn_set_pslevel(struct iwn_softc *, int, int, int);
Split out bluetooth coexistence setup.
2011-04-15 17:10:52 +00:00
static int iwn_send_btcoex(struct iwn_softc *);
Add basic support for advanced bluetooth coexistence required for 6005 gen2b (1030/6030) adapters.
2011-04-20 16:59:27 +00:00
static int iwn_send_advanced_btcoex(struct iwn_softc *);
The DC calibration result obtained during initialization can't be passed over to the runtime firmware on 6050 devices. Instead let the runtime firmware do the calibration itself. This fixes support for the 6050 series devices. Obtained from: OpenBSD Submitted by: kevlo Tested by: lx, Tz-Huan Huang(earlier version)
2011-11-21 22:19:12 +00:00
static int iwn5000_runtime_calib(struct iwn_softc *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn_config(struct iwn_softc *);
Refactor out the scan id and scan vap as part of the scan work. Make the scan state optional - we'll obviously need a vap, but we now won't require the scan state. the only thing the scan state is needed for is to check for the list of SSIDs to scan - which we can now just plain ignore by passing in NULL as the scan state pointer. Tested: * Intel 5100 (STA)
2013-12-07 08:32:15 +00:00
static int iwn_scan(struct iwn_softc *, struct ieee80211vap *,
struct ieee80211_scan_state *, struct ieee80211_channel *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn_auth(struct iwn_softc *, struct ieee80211vap *vap);
static int iwn_run(struct iwn_softc *, struct ieee80211vap *vap);
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
static int iwn_ampdu_rx_start(struct ieee80211_node *,
struct ieee80211_rx_ampdu *, int, int, int);
static void iwn_ampdu_rx_stop(struct ieee80211_node *,
struct ieee80211_rx_ampdu *);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
static int iwn_addba_request(struct ieee80211_node *,
struct ieee80211_tx_ampdu *, int, int, int);
static int iwn_addba_response(struct ieee80211_node *,
struct ieee80211_tx_ampdu *, int, int, int);
Add some missing protoypes. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:46:42 +00:00
static int iwn_ampdu_tx_start(struct ieee80211com *,
struct ieee80211_node *, uint8_t);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
static void iwn_ampdu_tx_stop(struct ieee80211_node *,
struct ieee80211_tx_ampdu *);
Add some missing protoypes. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:46:42 +00:00
static void iwn4965_ampdu_tx_start(struct iwn_softc *,
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
struct ieee80211_node *, int, uint8_t, uint16_t);
static void iwn4965_ampdu_tx_stop(struct iwn_softc *, int,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
uint8_t, uint16_t);
Add some missing protoypes. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:46:42 +00:00
static void iwn5000_ampdu_tx_start(struct iwn_softc *,
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
struct ieee80211_node *, int, uint8_t, uint16_t);
static void iwn5000_ampdu_tx_stop(struct iwn_softc *, int,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
uint8_t, uint16_t);
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
static int iwn5000_query_calibration(struct iwn_softc *);
static int iwn5000_send_calibration(struct iwn_softc *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn5000_send_wimax_coex(struct iwn_softc *);
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
static int iwn5000_crystal_calib(struct iwn_softc *);
fix the order of the prototypes from the previosu commit
2011-04-15 20:40:49 +00:00
static int iwn5000_temp_offset_calib(struct iwn_softc *);
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
static int iwn5000_temp_offset_calibv2(struct iwn_softc *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn4965_post_alive(struct iwn_softc *);
static int iwn5000_post_alive(struct iwn_softc *);
static int iwn4965_load_bootcode(struct iwn_softc *, const uint8_t *,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
int);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn4965_load_firmware(struct iwn_softc *);
static int iwn5000_load_firmware_section(struct iwn_softc *, uint32_t,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
const uint8_t *, int);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn5000_load_firmware(struct iwn_softc *);
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
static int iwn_read_firmware_leg(struct iwn_softc *,
struct iwn_fw_info *);
static int iwn_read_firmware_tlv(struct iwn_softc *,
struct iwn_fw_info *, uint16_t);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int iwn_read_firmware(struct iwn_softc *);
static int iwn_clock_wait(struct iwn_softc *);
static int iwn_apm_init(struct iwn_softc *);
static void iwn_apm_stop_master(struct iwn_softc *);
static void iwn_apm_stop(struct iwn_softc *);
static int iwn4965_nic_config(struct iwn_softc *);
static int iwn5000_nic_config(struct iwn_softc *);
static int iwn_hw_prepare(struct iwn_softc *);
static int iwn_hw_init(struct iwn_softc *);
static void iwn_hw_stop(struct iwn_softc *);
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
static void iwn_radio_on(void *, int);
static void iwn_radio_off(void *, int);
Make iwn(4) able to get itself back into working condition after "fatal firmware error" happens. Previously it was neccessary to reset it manually, using "/etc/rc.d/netif restart". Approved by: adrian@ MFC after: 1 month Sponsored by: The FreeBSD Foundation
2014-05-22 15:29:25 +00:00
static void iwn_panicked(void *, int);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_init_locked(struct iwn_softc *);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
static void iwn_init(struct iwn_softc *);
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void iwn_stop_locked(struct iwn_softc *);
static void iwn_stop(struct iwn_softc *);
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
static void iwn_scan_start(struct ieee80211com *);
static void iwn_scan_end(struct ieee80211com *);
static void iwn_set_channel(struct ieee80211com *);
static void iwn_scan_curchan(struct ieee80211_scan_state *, unsigned long);
static void iwn_scan_mindwell(struct ieee80211_scan_state *);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
static void iwn_hw_reset(void *, int);
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
#ifdef IWN_DEBUG
static char *iwn_get_csr_string(int);
static void iwn_debug_register(struct iwn_softc *);
#endif
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
static device_method_t iwn_methods[] = {
/* Device interface */
DEVMETHOD(device_probe, iwn_probe),
DEVMETHOD(device_attach, iwn_attach),
DEVMETHOD(device_detach, iwn_detach),
DEVMETHOD(device_shutdown, iwn_shutdown),
DEVMETHOD(device_suspend, iwn_suspend),
DEVMETHOD(device_resume, iwn_resume),
- There's no need to keep track of resource IDs. - Simplify MSI allocation and release. For a single one, we don't need to fiddle with the MSI count and pci_release_msi(9) is smart enough to just do nothing in case of INTx. - Don't allocate MSI as RF_SHAREABLE. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. MFC after: 1 week
2013-12-29 19:32:27 +00:00
DEVMETHOD_END
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
};
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
static driver_t iwn_driver = {
"iwn",
iwn_methods,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
sizeof(struct iwn_softc)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
};
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
static devclass_t iwn_devclass;
- There's no need to keep track of resource IDs. - Simplify MSI allocation and release. For a single one, we don't need to fiddle with the MSI count and pci_release_msi(9) is smart enough to just do nothing in case of INTx. - Don't allocate MSI as RF_SHAREABLE. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. MFC after: 1 week
2013-12-29 19:32:27 +00:00
DRIVER_MODULE(iwn, pci, iwn_driver, iwn_devclass, NULL, NULL);
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
Add module version to iwi/ipw/wpi and iwn. The version is used to check if a module is already preset, not setting it results in: can't re-use a leaf (ipw)! module_register: module pci/ipw already exists! Module pci/ipw failed to register: 17 while trying to load the module due to an entry in loader.conf. With this commit we get the expected: module ipw already present! Reported by: Dru Lavigne, bz Tested by: bz MFC after: 1 week
2011-05-31 19:08:25 +00:00
MODULE_VERSION(iwn, 1);
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
MODULE_DEPEND(iwn, firmware, 1, 1, 1);
MODULE_DEPEND(iwn, pci, 1, 1, 1);
MODULE_DEPEND(iwn, wlan, 1, 1, 1);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
static d_ioctl_t iwn_cdev_ioctl;
static d_open_t iwn_cdev_open;
static d_close_t iwn_cdev_close;
static struct cdevsw iwn_cdevsw = {
.d_version = D_VERSION,
.d_flags = 0,
.d_open = iwn_cdev_open,
.d_close = iwn_cdev_close,
.d_ioctl = iwn_cdev_ioctl,
.d_name = "iwn",
};
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
static int
iwn_probe(device_t dev)
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
const struct iwn_ident *ident;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
for (ident = iwn_ident_table; ident->name != NULL; ident++) {
if (pci_get_vendor(dev) == ident->vendor &&
pci_get_device(dev) == ident->device) {
device_set_desc(dev, ident->name);
- Probe with BUS_PROBE_DEFAULT instead of 0. - Remove clearing PCIM_CMD_INTxDIS; pci(4) will do that as appropriate since r189367. MFC after: 1 week
2013-12-30 16:46:50 +00:00
return (BUS_PROBE_DEFAULT);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
}
return ENXIO;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
static int
iwn_is_3stream_device(struct iwn_softc *sc)
{
/* XXX for now only 5300, until the 5350 can be tested */
if (sc->hw_type == IWN_HW_REV_TYPE_5300)
return (1);
return (0);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
static int
iwn_attach(device_t dev)
{
struct iwn_softc *sc = (struct iwn_softc *)device_get_softc(dev);
struct ieee80211com *ic;
- There's no need to keep track of resource IDs. - Simplify MSI allocation and release. For a single one, we don't need to fiddle with the MSI count and pci_release_msi(9) is smart enough to just do nothing in case of INTx. - Don't allocate MSI as RF_SHAREABLE. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. MFC after: 1 week
2013-12-29 19:32:27 +00:00
int i, error, rid;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
sc->sc_dev = dev;
Support setting up the iwn debug flags during probe/attach by checking "hint.iwn.X.debug". Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-24 13:31:31 +00:00
#ifdef IWN_DEBUG
error = resource_int_value(device_get_name(sc->sc_dev),
device_get_unit(sc->sc_dev), "debug", &(sc->sc_debug));
if (error != 0)
sc->sc_debug = 0;
#else
sc->sc_debug = 0;
#endif
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: begin\n",__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Get the offset of the PCI Express Capability Structure in PCI
* Configuration Space.
*/
Do a sweep of the tree replacing calls to pci_find_extcap() with calls to pci_find_cap() instead.
2011-03-23 13:10:15 +00:00
error = pci_find_cap(dev, PCIY_EXPRESS, &sc->sc_cap_off);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (error != 0) {
device_printf(dev, "PCIe capability structure not found!\n");
return error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Clear device-specific "PCI retry timeout" register (41h). */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
pci_write_config(dev, 0x41, 0, 1);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Enable bus-mastering. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
pci_enable_busmaster(dev);
- There's no need to keep track of resource IDs. - Simplify MSI allocation and release. For a single one, we don't need to fiddle with the MSI count and pci_release_msi(9) is smart enough to just do nothing in case of INTx. - Don't allocate MSI as RF_SHAREABLE. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. MFC after: 1 week
2013-12-29 19:32:27 +00:00
rid = PCIR_BAR(0);
sc->mem = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &rid,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
RF_ACTIVE);
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if (sc->mem == NULL) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(dev, "can't map mem space\n");
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
error = ENOMEM;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
sc->sc_st = rman_get_bustag(sc->mem);
sc->sc_sh = rman_get_bushandle(sc->mem);
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
- There's no need to keep track of resource IDs. - Simplify MSI allocation and release. For a single one, we don't need to fiddle with the MSI count and pci_release_msi(9) is smart enough to just do nothing in case of INTx. - Don't allocate MSI as RF_SHAREABLE. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. MFC after: 1 week
2013-12-29 19:32:27 +00:00
i = 1;
rid = 0;
if (pci_alloc_msi(dev, &i) == 0)
rid = 1;
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Install interrupt handler. */
- There's no need to keep track of resource IDs. - Simplify MSI allocation and release. For a single one, we don't need to fiddle with the MSI count and pci_release_msi(9) is smart enough to just do nothing in case of INTx. - Don't allocate MSI as RF_SHAREABLE. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. MFC after: 1 week
2013-12-29 19:32:27 +00:00
sc->irq = bus_alloc_resource_any(dev, SYS_RES_IRQ, &rid, RF_ACTIVE |
(rid != 0 ? 0 : RF_SHAREABLE));
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (sc->irq == NULL) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(dev, "can't map interrupt\n");
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
error = ENOMEM;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
goto fail;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
IWN_LOCK_INIT(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
/* Read hardware revision and attach. */
Break out the iwn(4) device IDs into if_iwn_devid.h, as well as add IDs for new devices. * Add new device IDs * Extend the ID probe code to include the newer range of bits used by later model devices Tested: * Intel 5100, STA mode TODO: * Test on Intel 4965, just to be sure Submitted by: Cedric GROSS <cg@gross.info>
2013-08-02 21:23:28 +00:00
sc->hw_type = (IWN_READ(sc, IWN_HW_REV) >> IWN_HW_REV_TYPE_SHIFT)
& IWN_HW_REV_TYPE_MASK;
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->subdevice_id = pci_get_subdevice(dev);
Remove debugging printf that crept in in r266770. Submitted by: adrian, nwhitehorn MFC after: 2 weeks
2014-05-27 23:05:12 +00:00
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
/*
* 4965 versus 5000 and later have different methods.
* Let's set those up first.
*/
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
if (sc->hw_type == IWN_HW_REV_TYPE_4965)
error = iwn4965_attach(sc, pci_get_device(dev));
else
error = iwn5000_attach(sc, pci_get_device(dev));
if (error != 0) {
device_printf(dev, "could not attach device, error %d\n",
error);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
goto fail;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
/*
* Next, let's setup the various parameters of each NIC.
*/
error = iwn_config_specific(sc, pci_get_device(dev));
if (error != 0) {
device_printf(dev, "could not attach device, error %d\n",
error);
goto fail;
}
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_hw_prepare(sc)) != 0) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
device_printf(dev, "hardware not ready, error %d\n", error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
goto fail;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Allocate DMA memory for firmware transfers. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_alloc_fwmem(sc)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(dev,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
"could not allocate memory for firmware, error %d\n",
error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
goto fail;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Allocate "Keep Warm" page. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_alloc_kw(sc)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(dev,
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
"could not allocate keep warm page, error %d\n", error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
goto fail;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Allocate ICT table for 5000 Series. */
if (sc->hw_type != IWN_HW_REV_TYPE_4965 &&
(error = iwn_alloc_ict(sc)) != 0) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(dev, "could not allocate ICT table, error %d\n",
error);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
goto fail;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Allocate TX scheduler "rings". */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_alloc_sched(sc)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(dev,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"could not allocate TX scheduler rings, error %d\n", error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
goto fail;
}
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Allocate TX rings (16 on 4965AGN, 20 on >=5000). */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
for (i = 0; i < sc->ntxqs; i++) {
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_alloc_tx_ring(sc, &sc->txq[i], i)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(dev,
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
"could not allocate TX ring %d, error %d\n", i,
error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
goto fail;
}
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Allocate RX ring. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_alloc_rx_ring(sc, &sc->rxq)) != 0) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(dev, "could not allocate RX ring, error %d\n",
error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
goto fail;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Clear pending interrupts. */
IWN_WRITE(sc, IWN_INT, 0xffffffff);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ic = &sc->sc_ic;
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
ic->ic_softc = sc;
Make net80211 drivers supply their device name to the net80211 layer, so that the latter doesn't need to go through struct ifnet to get their name. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 13:51:13 +00:00
ic->ic_name = device_get_nameunit(dev);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ic->ic_phytype = IEEE80211_T_OFDM; /* not only, but not used */
ic->ic_opmode = IEEE80211_M_STA; /* default to BSS mode */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Set device capabilities. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ic->ic_caps =
Minor cleanup of vap create work: o add IEEE80211_C_STA capability to indicate sta mode is supported (was previously assumed) and mark drivers as capable o add ieee80211_opcap array to map an opmode to the equivalent capability bit o move IEEE80211_C_OPMODE definition to where capabilities are defined so it's clear it should be kept in sync (on future additions) o check device capabilities in clone create before trying to create a vap; this makes driver checks unneeded o make error codes return on failed clone request unique o temporarily add console printfs on clone request failures to aid in debugging; these will move under DIAGNOSTIC or similar before release
2008-05-12 00:15:30 +00:00
IEEE80211_C_STA /* station mode supported */
| IEEE80211_C_MONITOR /* monitor mode supported */
Disable bgscan for now - it interferes with 11n activity and general stability. I'll re-enable it once the scan overhaul is done - the NIC itself can do bgscan, but not how we're doing it. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-04 05:37:07 +00:00
#if 0
The firmware of 4965 series adapters seems to die while trying to send probe requests at 1Mbps while being associated on a 5GHz channel. Sending those at 6Mbps does work, so use that instead during a background scan. This workaround allows us to re-enable background scan support for the 4965 adapters. Also, just enabling one antenna on 5GHz results in better reception of beacons: test 00:26:5a:c6:14:1a 40 54M -71:-95 200 E WME HTCAP ATH vs test 00:26:5a:c6:14:1a 40 54M -92:-95 200 E WME HTCAP ATH Due to roam:rssi thresholds set to 7 by default it might have been impossible to associate to that network. While here use IEEE80211_IS_CHAN_5GHZ() to determine the band.
2011-06-04 11:43:09 +00:00
| IEEE80211_C_BGSCAN /* background scanning */
Disable bgscan for now - it interferes with 11n activity and general stability. I'll re-enable it once the scan overhaul is done - the NIC itself can do bgscan, but not how we're doing it. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-04 05:37:07 +00:00
#endif
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
| IEEE80211_C_TXPMGT /* tx power management */
| IEEE80211_C_SHSLOT /* short slot time supported */
| IEEE80211_C_WPA
| IEEE80211_C_SHPREAMBLE /* short preamble supported */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
#if 0
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
| IEEE80211_C_IBSS /* ibss/adhoc mode */
#endif
| IEEE80211_C_WME /* WME */
Enable the station-side power management flag. It's not enabled by default in net80211 so this is a no-op unless if you enable it (ifconfig wlan0 powersave). Tested: * iwn0: <Intel WiFi Link 5100> mem 0xf4300000-0xf4301fff irq 17 at device 0.0 on pci3 TODO: * .. test on all the other NICs * See if I have to disable it during scan and such * Make it configurable live, rather than only after it's done its initial receive calibration.
2013-07-04 19:45:34 +00:00
| IEEE80211_C_PMGT /* Station-side power mgmt */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
;
Disable background scan support for 4965 adapters. On legacy channels every once in a while the firmware throws a SYSASSERT on line 208. On HT channels though this does always happen and I'm not aware of any workaround currently.
2011-05-08 11:01:53 +00:00
Read chainmask information before announcing it.
2011-05-08 11:05:03 +00:00
/* Read MAC address, channels, etc from EEPROM. */
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
if ((error = iwn_read_eeprom(sc, ic->ic_macaddr)) != 0) {
Read chainmask information before announcing it.
2011-05-08 11:05:03 +00:00
device_printf(dev, "could not read EEPROM, error %d\n",
error);
goto fail;
}
/* Count the number of available chains. */
sc->ntxchains =
((sc->txchainmask >> 2) & 1) +
((sc->txchainmask >> 1) & 1) +
((sc->txchainmask >> 0) & 1);
sc->nrxchains =
((sc->rxchainmask >> 2) & 1) +
((sc->rxchainmask >> 1) & 1) +
((sc->rxchainmask >> 0) & 1);
if (bootverbose) {
device_printf(dev, "MIMO %dT%dR, %.4s, address %6D\n",
sc->ntxchains, sc->nrxchains, sc->eeprom_domain,
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ic->ic_macaddr, ":");
Read chainmask information before announcing it.
2011-05-08 11:05:03 +00:00
}
Enable 11n (sans HT40) support.
2011-05-08 12:23:01 +00:00
if (sc->sc_flags & IWN_FLAG_HAS_11N) {
ic->ic_rxstream = sc->nrxchains;
ic->ic_txstream = sc->ntxchains;
Cap the number of streams supported to two for now. I haven't yet reviewed the Intel driver(s) in more depth to see if there are 1x1 NICs that report they support 2 transmit/receive chains.. if so then we'll have to update this. Tested: * Intel 4965, which is a 2x2 device with 3 RX and 2 TX chains. PR: kern/181132
2013-08-08 05:52:41 +00:00
/*
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
* Some of the 3 antenna devices (ie, the 4965) only supports
* 2x2 operation. So correct the number of streams if
* it's not a 3-stream device.
Cap the number of streams supported to two for now. I haven't yet reviewed the Intel driver(s) in more depth to see if there are 1x1 NICs that report they support 2 transmit/receive chains.. if so then we'll have to update this. Tested: * Intel 4965, which is a 2x2 device with 3 RX and 2 TX chains. PR: kern/181132
2013-08-08 05:52:41 +00:00
*/
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
if (! iwn_is_3stream_device(sc)) {
if (ic->ic_rxstream > 2)
ic->ic_rxstream = 2;
if (ic->ic_txstream > 2)
ic->ic_txstream = 2;
}
Cap the number of streams supported to two for now. I haven't yet reviewed the Intel driver(s) in more depth to see if there are 1x1 NICs that report they support 2 transmit/receive chains.. if so then we'll have to update this. Tested: * Intel 4965, which is a 2x2 device with 3 RX and 2 TX chains. PR: kern/181132
2013-08-08 05:52:41 +00:00
Enable 11n (sans HT40) support.
2011-05-08 12:23:01 +00:00
ic->ic_htcaps =
IEEE80211_HTCAP_SMPS_OFF /* SMPS mode disabled */
| IEEE80211_HTCAP_SHORTGI20 /* short GI in 20MHz */
| IEEE80211_HTCAP_CHWIDTH40 /* 40MHz channel width*/
| IEEE80211_HTCAP_SHORTGI40 /* short GI in 40MHz */
Enable HT40 (40MHz channel width) support.
2011-06-04 15:22:01 +00:00
#ifdef notyet
Enable 11n (sans HT40) support.
2011-05-08 12:23:01 +00:00
| IEEE80211_HTCAP_GREENFIELD
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
#if IWN_RBUF_SIZE == 8192
Enable 11n (sans HT40) support.
2011-05-08 12:23:01 +00:00
| IEEE80211_HTCAP_MAXAMSDU_7935 /* max A-MSDU length */
#else
| IEEE80211_HTCAP_MAXAMSDU_3839 /* max A-MSDU length */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
#endif
disable HT capabilities until we sort out firmware issues
2008-04-29 21:43:16 +00:00
#endif
Enable 11n (sans HT40) support.
2011-05-08 12:23:01 +00:00
/* s/w capabilities */
| IEEE80211_HTC_HT /* HT operation */
| IEEE80211_HTC_AMPDU /* tx A-MPDU */
#ifdef notyet
| IEEE80211_HTC_AMSDU /* tx A-MSDU */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
#endif
Enable 11n (sans HT40) support.
2011-05-08 12:23:01 +00:00
;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ieee80211_ifattach(ic);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ic->ic_vap_create = iwn_vap_create;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ic->ic_ioctl = iwn_ioctl;
ic->ic_parent = iwn_parent;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ic->ic_vap_delete = iwn_vap_delete;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ic->ic_transmit = iwn_transmit;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ic->ic_raw_xmit = iwn_raw_xmit;
ic->ic_node_alloc = iwn_node_alloc;
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
sc->sc_ampdu_rx_start = ic->ic_ampdu_rx_start;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
ic->ic_ampdu_rx_start = iwn_ampdu_rx_start;
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
sc->sc_ampdu_rx_stop = ic->ic_ampdu_rx_stop;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
ic->ic_ampdu_rx_stop = iwn_ampdu_rx_stop;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
sc->sc_addba_request = ic->ic_addba_request;
ic->ic_addba_request = iwn_addba_request;
sc->sc_addba_response = ic->ic_addba_response;
ic->ic_addba_response = iwn_addba_response;
sc->sc_addba_stop = ic->ic_addba_stop;
ic->ic_addba_stop = iwn_ampdu_tx_stop;
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
ic->ic_newassoc = iwn_newassoc;
Rename some stuff in favour of the OpenBSD names: - prefer EDCA over WME - qid for a TXQ ID - reg for register values
2011-04-16 13:40:32 +00:00
ic->ic_wme.wme_update = iwn_updateedca;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
ic->ic_update_mcast = iwn_update_mcast;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
ic->ic_scan_start = iwn_scan_start;
ic->ic_scan_end = iwn_scan_end;
ic->ic_set_channel = iwn_set_channel;
ic->ic_scan_curchan = iwn_scan_curchan;
ic->ic_scan_mindwell = iwn_scan_mindwell;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
ic->ic_setregdomain = iwn_setregdomain;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_radiotap_attach(sc);
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
callout_init_mtx(&sc->calib_to, &sc->sc_mtx, 0);
callout_init_mtx(&sc->watchdog_to, &sc->sc_mtx, 0);
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
TASK_INIT(&sc->sc_reinit_task, 0, iwn_hw_reset, sc);
TASK_INIT(&sc->sc_radioon_task, 0, iwn_radio_on, sc);
TASK_INIT(&sc->sc_radiooff_task, 0, iwn_radio_off, sc);
Make iwn(4) able to get itself back into working condition after "fatal firmware error" happens. Previously it was neccessary to reset it manually, using "/etc/rc.d/netif restart". Approved by: adrian@ MFC after: 1 month Sponsored by: The FreeBSD Foundation
2014-05-22 15:29:25 +00:00
TASK_INIT(&sc->sc_panic_task, 0, iwn_panicked, sc);
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
TASK_INIT(&sc->sc_xmit_task, 0, iwn_xmit_task, sc);
mbufq_init(&sc->sc_xmit_queue, 1024);
Make iwn(4) able to get itself back into working condition after "fatal firmware error" happens. Previously it was neccessary to reset it manually, using "/etc/rc.d/netif restart". Approved by: adrian@ MFC after: 1 month Sponsored by: The FreeBSD Foundation
2014-05-22 15:29:25 +00:00
sc->sc_tq = taskqueue_create("iwn_taskq", M_WAITOK,
taskqueue_thread_enqueue, &sc->sc_tq);
error = taskqueue_start_threads(&sc->sc_tq, 1, 0, "iwn_taskq");
if (error != 0) {
device_printf(dev, "can't start threads, error %d\n", error);
goto fail;
}
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_sysctlattach(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Hook our interrupt after all initialization is complete.
*/
error = bus_setup_intr(dev, sc->irq, INTR_TYPE_NET | INTR_MPSAFE,
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
NULL, iwn_intr, sc, &sc->sc_ih);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (error != 0) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(dev, "can't establish interrupt, error %d\n",
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
error);
goto fail;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
#if 0
device_printf(sc->sc_dev, "%s: rx_stats=%d, rx_stats_bt=%d\n",
__func__,
sizeof(struct iwn_stats),
sizeof(struct iwn_stats_bt));
#endif
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
if (bootverbose)
ieee80211_announce(ic);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
/* Add debug ioctl right at the end */
sc->sc_cdev = make_dev(&iwn_cdevsw, device_get_unit(dev),
UID_ROOT, GID_WHEEL, 0600, "%s", device_get_nameunit(dev));
if (sc->sc_cdev == NULL) {
device_printf(dev, "failed to create debug character device\n");
} else {
sc->sc_cdev->si_drv1 = sc;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return 0;
fail:
iwn_cleanup() is just a wrapper around iwn_detach(), call it directly instead.
2011-04-14 17:31:34 +00:00
iwn_detach(dev);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end in error\n",__func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
/*
* Define specific configuration based on device id and subdevice id
* pid : PCI device id
*/
static int
iwn_config_specific(struct iwn_softc *sc, uint16_t pid)
{
switch (pid) {
/* 4965 series */
case IWN_DID_4965_1:
case IWN_DID_4965_2:
case IWN_DID_4965_3:
case IWN_DID_4965_4:
sc->base_params = &iwn4965_base_params;
sc->limits = &iwn4965_sensitivity_limits;
sc->fwname = "iwn4965fw";
/* Override chains masks, ROM is known to be broken. */
sc->txchainmask = IWN_ANT_AB;
sc->rxchainmask = IWN_ANT_ABC;
/* Enable normal btcoex */
sc->sc_flags |= IWN_FLAG_BTCOEX;
break;
/* 1000 Series */
case IWN_DID_1000_1:
case IWN_DID_1000_2:
switch(sc->subdevice_id) {
case IWN_SDID_1000_1:
case IWN_SDID_1000_2:
case IWN_SDID_1000_3:
case IWN_SDID_1000_4:
case IWN_SDID_1000_5:
case IWN_SDID_1000_6:
case IWN_SDID_1000_7:
case IWN_SDID_1000_8:
case IWN_SDID_1000_9:
case IWN_SDID_1000_10:
case IWN_SDID_1000_11:
case IWN_SDID_1000_12:
sc->limits = &iwn1000_sensitivity_limits;
sc->base_params = &iwn1000_base_params;
sc->fwname = "iwn1000fw";
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
/* 6x00 Series */
case IWN_DID_6x00_2:
case IWN_DID_6x00_4:
case IWN_DID_6x00_1:
case IWN_DID_6x00_3:
sc->fwname = "iwn6000fw";
sc->limits = &iwn6000_sensitivity_limits;
switch(sc->subdevice_id) {
case IWN_SDID_6x00_1:
case IWN_SDID_6x00_2:
case IWN_SDID_6x00_8:
//iwl6000_3agn_cfg
sc->base_params = &iwn_6000_base_params;
break;
case IWN_SDID_6x00_3:
case IWN_SDID_6x00_6:
case IWN_SDID_6x00_9:
////iwl6000i_2agn
case IWN_SDID_6x00_4:
case IWN_SDID_6x00_7:
case IWN_SDID_6x00_10:
//iwl6000i_2abg_cfg
case IWN_SDID_6x00_5:
//iwl6000i_2bg_cfg
sc->base_params = &iwn_6000i_base_params;
sc->sc_flags |= IWN_FLAG_INTERNAL_PA;
sc->txchainmask = IWN_ANT_BC;
sc->rxchainmask = IWN_ANT_BC;
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
/* 6x05 Series */
case IWN_DID_6x05_1:
case IWN_DID_6x05_2:
switch(sc->subdevice_id) {
case IWN_SDID_6x05_1:
case IWN_SDID_6x05_4:
case IWN_SDID_6x05_6:
//iwl6005_2agn_cfg
case IWN_SDID_6x05_2:
case IWN_SDID_6x05_5:
case IWN_SDID_6x05_7:
//iwl6005_2abg_cfg
case IWN_SDID_6x05_3:
//iwl6005_2bg_cfg
case IWN_SDID_6x05_8:
case IWN_SDID_6x05_9:
//iwl6005_2agn_sff_cfg
case IWN_SDID_6x05_10:
//iwl6005_2agn_d_cfg
case IWN_SDID_6x05_11:
//iwl6005_2agn_mow1_cfg
case IWN_SDID_6x05_12:
//iwl6005_2agn_mow2_cfg
sc->fwname = "iwn6000g2afw";
sc->limits = &iwn6000_sensitivity_limits;
sc->base_params = &iwn_6000g2_base_params;
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
/* 6x35 Series */
case IWN_DID_6035_1:
case IWN_DID_6035_2:
switch(sc->subdevice_id) {
case IWN_SDID_6035_1:
case IWN_SDID_6035_2:
case IWN_SDID_6035_3:
case IWN_SDID_6035_4:
sc->fwname = "iwn6000g2bfw";
Add some initial support for the Intel 6235. Tested: * Intel 5100 * Intel 6235 Obtained from: mav, others
2013-12-09 03:40:02 +00:00
sc->limits = &iwn6235_sensitivity_limits;
sc->base_params = &iwn_6235_base_params;
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
/* 6x50 WiFi/WiMax Series */
case IWN_DID_6050_1:
case IWN_DID_6050_2:
switch(sc->subdevice_id) {
case IWN_SDID_6050_1:
case IWN_SDID_6050_3:
case IWN_SDID_6050_5:
//iwl6050_2agn_cfg
case IWN_SDID_6050_2:
case IWN_SDID_6050_4:
case IWN_SDID_6050_6:
//iwl6050_2abg_cfg
sc->fwname = "iwn6050fw";
sc->txchainmask = IWN_ANT_AB;
sc->rxchainmask = IWN_ANT_AB;
sc->limits = &iwn6000_sensitivity_limits;
sc->base_params = &iwn_6050_base_params;
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
/* 6150 WiFi/WiMax Series */
case IWN_DID_6150_1:
case IWN_DID_6150_2:
switch(sc->subdevice_id) {
case IWN_SDID_6150_1:
case IWN_SDID_6150_3:
case IWN_SDID_6150_5:
// iwl6150_bgn_cfg
case IWN_SDID_6150_2:
case IWN_SDID_6150_4:
case IWN_SDID_6150_6:
//iwl6150_bg_cfg
sc->fwname = "iwn6050fw";
sc->limits = &iwn6000_sensitivity_limits;
sc->base_params = &iwn_6150_base_params;
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
/* 6030 Series and 1030 Series */
case IWN_DID_x030_1:
case IWN_DID_x030_2:
case IWN_DID_x030_3:
case IWN_DID_x030_4:
switch(sc->subdevice_id) {
case IWN_SDID_x030_1:
case IWN_SDID_x030_3:
case IWN_SDID_x030_5:
// iwl1030_bgn_cfg
case IWN_SDID_x030_2:
case IWN_SDID_x030_4:
case IWN_SDID_x030_6:
//iwl1030_bg_cfg
case IWN_SDID_x030_7:
case IWN_SDID_x030_10:
case IWN_SDID_x030_14:
//iwl6030_2agn_cfg
case IWN_SDID_x030_8:
case IWN_SDID_x030_11:
case IWN_SDID_x030_15:
// iwl6030_2bgn_cfg
case IWN_SDID_x030_9:
case IWN_SDID_x030_12:
case IWN_SDID_x030_16:
// iwl6030_2abg_cfg
case IWN_SDID_x030_13:
//iwl6030_2bg_cfg
sc->fwname = "iwn6000g2bfw";
sc->limits = &iwn6000_sensitivity_limits;
sc->base_params = &iwn_6000g2b_base_params;
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
/* 130 Series WiFi */
/* XXX: This series will need adjustment for rate.
* see rx_with_siso_diversity in linux kernel
*/
case IWN_DID_130_1:
case IWN_DID_130_2:
switch(sc->subdevice_id) {
case IWN_SDID_130_1:
case IWN_SDID_130_3:
case IWN_SDID_130_5:
//iwl130_bgn_cfg
case IWN_SDID_130_2:
case IWN_SDID_130_4:
case IWN_SDID_130_6:
//iwl130_bg_cfg
sc->fwname = "iwn6000g2bfw";
sc->limits = &iwn6000_sensitivity_limits;
sc->base_params = &iwn_6000g2b_base_params;
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
/* 100 Series WiFi */
case IWN_DID_100_1:
case IWN_DID_100_2:
switch(sc->subdevice_id) {
case IWN_SDID_100_1:
case IWN_SDID_100_2:
case IWN_SDID_100_3:
case IWN_SDID_100_4:
case IWN_SDID_100_5:
case IWN_SDID_100_6:
sc->limits = &iwn1000_sensitivity_limits;
sc->base_params = &iwn1000_base_params;
sc->fwname = "iwn100fw";
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
Add support for the Intel Centrino Wireless-N 105 chipset. MFC after: 2 weeks Relnotes: yes
2014-05-27 19:39:57 +00:00
/* 105 Series */
/* XXX: This series will need adjustment for rate.
* see rx_with_siso_diversity in linux kernel
*/
case IWN_DID_105_1:
case IWN_DID_105_2:
switch(sc->subdevice_id) {
case IWN_SDID_105_1:
case IWN_SDID_105_2:
case IWN_SDID_105_3:
//iwl105_bgn_cfg
case IWN_SDID_105_4:
//iwl105_bgn_d_cfg
sc->limits = &iwn2030_sensitivity_limits;
sc->base_params = &iwn2000_base_params;
sc->fwname = "iwn105fw";
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
Add support for the Intel Centrino Wireless-N 135 chipset. MFC after: 2 weeks
2014-01-08 13:59:33 +00:00
/* 135 Series */
/* XXX: This series will need adjustment for rate.
* see rx_with_siso_diversity in linux kernel
*/
case IWN_DID_135_1:
case IWN_DID_135_2:
switch(sc->subdevice_id) {
case IWN_SDID_135_1:
case IWN_SDID_135_2:
case IWN_SDID_135_3:
sc->limits = &iwn2030_sensitivity_limits;
sc->base_params = &iwn2030_base_params;
sc->fwname = "iwn135fw";
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
/* 2x00 Series */
case IWN_DID_2x00_1:
case IWN_DID_2x00_2:
switch(sc->subdevice_id) {
case IWN_SDID_2x00_1:
case IWN_SDID_2x00_2:
case IWN_SDID_2x00_3:
//iwl2000_2bgn_cfg
case IWN_SDID_2x00_4:
//iwl2000_2bgn_d_cfg
sc->limits = &iwn2030_sensitivity_limits;
sc->base_params = &iwn2000_base_params;
sc->fwname = "iwn2000fw";
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice) \n",
pid, sc->subdevice_id, sc->hw_type);
return ENOTSUP;
}
break;
/* 2x30 Series */
case IWN_DID_2x30_1:
case IWN_DID_2x30_2:
switch(sc->subdevice_id) {
case IWN_SDID_2x30_1:
case IWN_SDID_2x30_3:
case IWN_SDID_2x30_5:
//iwl100_bgn_cfg
case IWN_SDID_2x30_2:
case IWN_SDID_2x30_4:
case IWN_SDID_2x30_6:
//iwl100_bg_cfg
sc->limits = &iwn2030_sensitivity_limits;
sc->base_params = &iwn2030_base_params;
sc->fwname = "iwn2030fw";
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
/* 5x00 Series */
case IWN_DID_5x00_1:
case IWN_DID_5x00_2:
case IWN_DID_5x00_3:
case IWN_DID_5x00_4:
sc->limits = &iwn5000_sensitivity_limits;
sc->base_params = &iwn5000_base_params;
sc->fwname = "iwn5000fw";
switch(sc->subdevice_id) {
case IWN_SDID_5x00_1:
case IWN_SDID_5x00_2:
case IWN_SDID_5x00_3:
case IWN_SDID_5x00_4:
case IWN_SDID_5x00_9:
case IWN_SDID_5x00_10:
case IWN_SDID_5x00_11:
case IWN_SDID_5x00_12:
case IWN_SDID_5x00_17:
case IWN_SDID_5x00_18:
case IWN_SDID_5x00_19:
case IWN_SDID_5x00_20:
//iwl5100_agn_cfg
sc->txchainmask = IWN_ANT_B;
sc->rxchainmask = IWN_ANT_AB;
break;
case IWN_SDID_5x00_5:
case IWN_SDID_5x00_6:
case IWN_SDID_5x00_13:
case IWN_SDID_5x00_14:
case IWN_SDID_5x00_21:
case IWN_SDID_5x00_22:
//iwl5100_bgn_cfg
sc->txchainmask = IWN_ANT_B;
sc->rxchainmask = IWN_ANT_AB;
break;
case IWN_SDID_5x00_7:
case IWN_SDID_5x00_8:
case IWN_SDID_5x00_15:
case IWN_SDID_5x00_16:
case IWN_SDID_5x00_23:
case IWN_SDID_5x00_24:
//iwl5100_abg_cfg
sc->txchainmask = IWN_ANT_B;
sc->rxchainmask = IWN_ANT_AB;
break;
case IWN_SDID_5x00_25:
case IWN_SDID_5x00_26:
case IWN_SDID_5x00_27:
case IWN_SDID_5x00_28:
case IWN_SDID_5x00_29:
case IWN_SDID_5x00_30:
case IWN_SDID_5x00_31:
case IWN_SDID_5x00_32:
case IWN_SDID_5x00_33:
case IWN_SDID_5x00_34:
case IWN_SDID_5x00_35:
case IWN_SDID_5x00_36:
//iwl5300_agn_cfg
sc->txchainmask = IWN_ANT_ABC;
sc->rxchainmask = IWN_ANT_ABC;
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
/* 5x50 Series */
case IWN_DID_5x50_1:
case IWN_DID_5x50_2:
case IWN_DID_5x50_3:
case IWN_DID_5x50_4:
sc->limits = &iwn5000_sensitivity_limits;
sc->base_params = &iwn5000_base_params;
sc->fwname = "iwn5000fw";
switch(sc->subdevice_id) {
case IWN_SDID_5x50_1:
case IWN_SDID_5x50_2:
case IWN_SDID_5x50_3:
//iwl5350_agn_cfg
sc->limits = &iwn5000_sensitivity_limits;
sc->base_params = &iwn5000_base_params;
sc->fwname = "iwn5000fw";
break;
case IWN_SDID_5x50_4:
case IWN_SDID_5x50_5:
case IWN_SDID_5x50_8:
case IWN_SDID_5x50_9:
case IWN_SDID_5x50_10:
case IWN_SDID_5x50_11:
//iwl5150_agn_cfg
case IWN_SDID_5x50_6:
case IWN_SDID_5x50_7:
case IWN_SDID_5x50_12:
case IWN_SDID_5x50_13:
//iwl5150_abg_cfg
sc->limits = &iwn5000_sensitivity_limits;
sc->fwname = "iwn5150fw";
sc->base_params = &iwn_5x50_base_params;
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id :"
"0x%04x rev %d not supported (subdevice)\n", pid,
sc->subdevice_id,sc->hw_type);
return ENOTSUP;
}
break;
default:
device_printf(sc->sc_dev, "adapter type id : 0x%04x sub id : 0x%04x"
"rev 0x%08x not supported (device)\n", pid, sc->subdevice_id,
sc->hw_type);
return ENOTSUP;
}
return 0;
}
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
static int
iwn4965_attach(struct iwn_softc *sc, uint16_t pid)
{
struct iwn_ops *ops = &sc->ops;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
ops->load_firmware = iwn4965_load_firmware;
ops->read_eeprom = iwn4965_read_eeprom;
ops->post_alive = iwn4965_post_alive;
ops->nic_config = iwn4965_nic_config;
ops->update_sched = iwn4965_update_sched;
ops->get_temperature = iwn4965_get_temperature;
ops->get_rssi = iwn4965_get_rssi;
ops->set_txpower = iwn4965_set_txpower;
ops->init_gains = iwn4965_init_gains;
ops->set_gains = iwn4965_set_gains;
ops->add_node = iwn4965_add_node;
ops->tx_done = iwn4965_tx_done;
ops->ampdu_tx_start = iwn4965_ampdu_tx_start;
ops->ampdu_tx_stop = iwn4965_ampdu_tx_stop;
sc->ntxqs = IWN4965_NTXQUEUES;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
sc->firstaggqueue = IWN4965_FIRSTAGGQUEUE;
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
sc->ndmachnls = IWN4965_NDMACHNLS;
sc->broadcast_id = IWN4965_ID_BROADCAST;
sc->rxonsz = IWN4965_RXONSZ;
sc->schedsz = IWN4965_SCHEDSZ;
sc->fw_text_maxsz = IWN4965_FW_TEXT_MAXSZ;
sc->fw_data_maxsz = IWN4965_FW_DATA_MAXSZ;
sc->fwsz = IWN4965_FWSZ;
sc->sched_txfact_addr = IWN4965_SCHED_TXFACT;
sc->limits = &iwn4965_sensitivity_limits;
sc->fwname = "iwn4965fw";
/* Override chains masks, ROM is known to be broken. */
sc->txchainmask = IWN_ANT_AB;
sc->rxchainmask = IWN_ANT_ABC;
Begin fleshing out a knob to enable/disable bluetooth coexistence. Some firmware versions seem to get very unhappy if they're sent btcoex commands when they don't actually have bluetooth hardware in them. So, disable sending them those commands. Tested: * 5100 (which has bluetooth, no problems) * 4965 (which doesn't have bluetooth, but didn't seem to crash) * 6200 (no bluetooth, seems to get unhappy being sent bluetooth commands.)
2013-10-25 19:46:52 +00:00
/* Enable normal btcoex */
sc->sc_flags |= IWN_FLAG_BTCOEX;
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "%s: end\n",__func__);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
return 0;
}
static int
iwn5000_attach(struct iwn_softc *sc, uint16_t pid)
{
struct iwn_ops *ops = &sc->ops;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
ops->load_firmware = iwn5000_load_firmware;
ops->read_eeprom = iwn5000_read_eeprom;
ops->post_alive = iwn5000_post_alive;
ops->nic_config = iwn5000_nic_config;
ops->update_sched = iwn5000_update_sched;
ops->get_temperature = iwn5000_get_temperature;
ops->get_rssi = iwn5000_get_rssi;
ops->set_txpower = iwn5000_set_txpower;
ops->init_gains = iwn5000_init_gains;
ops->set_gains = iwn5000_set_gains;
ops->add_node = iwn5000_add_node;
ops->tx_done = iwn5000_tx_done;
ops->ampdu_tx_start = iwn5000_ampdu_tx_start;
ops->ampdu_tx_stop = iwn5000_ampdu_tx_stop;
sc->ntxqs = IWN5000_NTXQUEUES;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
sc->firstaggqueue = IWN5000_FIRSTAGGQUEUE;
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
sc->ndmachnls = IWN5000_NDMACHNLS;
sc->broadcast_id = IWN5000_ID_BROADCAST;
sc->rxonsz = IWN5000_RXONSZ;
sc->schedsz = IWN5000_SCHEDSZ;
sc->fw_text_maxsz = IWN5000_FW_TEXT_MAXSZ;
sc->fw_data_maxsz = IWN5000_FW_DATA_MAXSZ;
sc->fwsz = IWN5000_FWSZ;
sc->sched_txfact_addr = IWN5000_SCHED_TXFACT;
Pull some features out of the firmware: - If a ENH_SENS TLV section exit the firmware is capable of doing enhanced sensitivity calibration. - Newer devices/firmwares have more calibration commands therefore hardcoding the noise gain/reset commands no longer works. It is supposed to use the next index after the newest calibration type support. Read the command index of the TLV section if available.
2011-04-19 19:47:41 +00:00
sc->reset_noise_gain = IWN5000_PHY_CALIB_RESET_NOISE_GAIN;
sc->noise_gain = IWN5000_PHY_CALIB_NOISE_GAIN;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
return 0;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
/*
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
* Attach the interface to 802.11 radiotap.
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_radiotap_attach(struct iwn_softc *sc)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ieee80211_radiotap_attach(&sc->sc_ic,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
&sc->sc_txtap.wt_ihdr, sizeof(sc->sc_txtap),
IWN_TX_RADIOTAP_PRESENT,
&sc->sc_rxtap.wr_ihdr, sizeof(sc->sc_rxtap),
IWN_RX_RADIOTAP_PRESENT);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
static void
iwn_sysctlattach(struct iwn_softc *sc)
{
Support setting up the iwn debug flags during probe/attach by checking "hint.iwn.X.debug". Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-24 13:31:31 +00:00
#ifdef IWN_DEBUG
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
struct sysctl_ctx_list *ctx = device_get_sysctl_ctx(sc->sc_dev);
struct sysctl_oid *tree = device_get_sysctl_tree(sc->sc_dev);
SYSCTL_ADD_INT(ctx, SYSCTL_CHILDREN(tree), OID_AUTO,
Support setting up the iwn debug flags during probe/attach by checking "hint.iwn.X.debug". Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-24 13:31:31 +00:00
"debug", CTLFLAG_RW, &sc->sc_debug, sc->sc_debug,
"control debugging printfs");
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
#endif
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
static struct ieee80211vap *
Fix some net80211 enum nits: - ic_vap_create() uses an ieee80211_opmode argument - ieee80211_rate2media() takes an ieee80211_phymode argument - ieee80211_plcp2rate() takes an ieee80211_phytype argument - cast to enum ieee80211_protmode and ieee80211_roamingmode to silence compiler warnings Submitted by: arundel@
2011-12-17 10:23:17 +00:00
iwn_vap_create(struct ieee80211com *ic, const char name[IFNAMSIZ], int unit,
enum ieee80211_opmode opmode, int flags,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
const uint8_t bssid[IEEE80211_ADDR_LEN],
const uint8_t mac[IEEE80211_ADDR_LEN])
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
struct iwn_softc *sc = ic->ic_softc;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_vap *ivp;
struct ieee80211vap *vap;
if (!TAILQ_EMPTY(&ic->ic_vaps)) /* only one at a time */
return NULL;
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ivp = malloc(sizeof(struct iwn_vap), M_80211_VAP, M_WAITOK | M_ZERO);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
vap = &ivp->iv_vap;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ieee80211_vap_setup(ic, vap, name, unit, opmode, flags, bssid);
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
ivp->ctx = IWN_RXON_BSS_CTX;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
vap->iv_bmissthreshold = 10; /* override default */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Override with driver methods. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ivp->iv_newstate = vap->iv_newstate;
vap->iv_newstate = iwn_newstate;
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->ivap[IWN_RXON_BSS_CTX] = vap;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
net80211 rate control framework (net80211 ratectl). This framework allows drivers to abstract the rate control algorithm and just feed the framework with the usable parameters. The rate control framework will now deal with passing the parameters to the selected algorithm. Right now we have AMRR (the default) and RSSADAPT but there's no way to select one with ifconfig, yet. The objective is to have more rate control algorithms in the net80211 stack so all drivers[0] can use it. Ideally, we'll have the well-known sample rate control algorithm in the net80211 at some point so all drivers can use it (not just ath). [0] all drivers that do rate control in software, that is. Reviewed by: bschmidt, thompsa, weyongo MFC after: 1 months
2010-04-07 15:29:13 +00:00
ieee80211_ratectl_init(vap);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Complete setup. */
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ieee80211_vap_attach(vap, iwn_media_change, ieee80211_media_status,
mac);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ic->ic_opmode = opmode;
return vap;
}
static void
iwn_vap_delete(struct ieee80211vap *vap)
{
struct iwn_vap *ivp = IWN_VAP(vap);
net80211 rate control framework (net80211 ratectl). This framework allows drivers to abstract the rate control algorithm and just feed the framework with the usable parameters. The rate control framework will now deal with passing the parameters to the selected algorithm. Right now we have AMRR (the default) and RSSADAPT but there's no way to select one with ifconfig, yet. The objective is to have more rate control algorithms in the net80211 stack so all drivers[0] can use it. Ideally, we'll have the well-known sample rate control algorithm in the net80211 at some point so all drivers can use it (not just ath). [0] all drivers that do rate control in software, that is. Reviewed by: bschmidt, thompsa, weyongo MFC after: 1 months
2010-04-07 15:29:13 +00:00
ieee80211_ratectl_deinit(vap);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ieee80211_vap_detach(vap);
free(ivp, M_80211_VAP);
}
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
static void
iwn_xmit_queue_drain(struct iwn_softc *sc)
{
struct mbuf *m;
struct ieee80211_node *ni;
IWN_LOCK_ASSERT(sc);
while ((m = mbufq_dequeue(&sc->sc_xmit_queue)) != NULL) {
ni = (struct ieee80211_node *)m->m_pkthdr.rcvif;
ieee80211_free_node(ni);
m_freem(m);
}
}
static int
iwn_xmit_queue_enqueue(struct iwn_softc *sc, struct mbuf *m)
{
IWN_LOCK_ASSERT(sc);
return (mbufq_enqueue(&sc->sc_xmit_queue, m));
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
iwn_cleanup() is just a wrapper around iwn_detach(), call it directly instead.
2011-04-14 17:31:34 +00:00
iwn_detach(device_t dev)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
struct iwn_softc *sc = device_get_softc(dev);
Rename some stuff in favour of the OpenBSD names: - prefer EDCA over WME - qid for a TXQ ID - reg for register values
2011-04-16 13:40:32 +00:00
int qid;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
if (sc->sc_ic.ic_softc != NULL) {
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
/* Free the mbuf queue and node references */
IWN_LOCK(sc);
iwn_xmit_queue_drain(sc);
IWN_UNLOCK(sc);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ieee80211_draintask(&sc->sc_ic, &sc->sc_reinit_task);
ieee80211_draintask(&sc->sc_ic, &sc->sc_radioon_task);
ieee80211_draintask(&sc->sc_ic, &sc->sc_radiooff_task);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_stop(sc);
Make iwn(4) able to get itself back into working condition after "fatal firmware error" happens. Previously it was neccessary to reset it manually, using "/etc/rc.d/netif restart". Approved by: adrian@ MFC after: 1 month Sponsored by: The FreeBSD Foundation
2014-05-22 15:29:25 +00:00
taskqueue_drain_all(sc->sc_tq);
taskqueue_free(sc->sc_tq);
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
callout_drain(&sc->watchdog_to);
callout_drain(&sc->calib_to);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ieee80211_ifdetach(&sc->sc_ic);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Uninstall interrupt handler. */
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
if (sc->irq != NULL) {
bus_teardown_intr(dev, sc->irq, sc->sc_ih);
- There's no need to keep track of resource IDs. - Simplify MSI allocation and release. For a single one, we don't need to fiddle with the MSI count and pci_release_msi(9) is smart enough to just do nothing in case of INTx. - Don't allocate MSI as RF_SHAREABLE. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. MFC after: 1 week
2013-12-29 19:32:27 +00:00
bus_release_resource(dev, SYS_RES_IRQ, rman_get_rid(sc->irq),
sc->irq);
pci_release_msi(dev);
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Free DMA resources. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_free_rx_ring(sc, &sc->rxq);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
for (qid = 0; qid < sc->ntxqs; qid++)
iwn_free_tx_ring(sc, &sc->txq[qid]);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_free_sched(sc);
iwn_free_kw(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (sc->ict != NULL)
iwn_free_ict(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_free_fwmem(sc);
if (sc->mem != NULL)
- There's no need to keep track of resource IDs. - Simplify MSI allocation and release. For a single one, we don't need to fiddle with the MSI count and pci_release_msi(9) is smart enough to just do nothing in case of INTx. - Don't allocate MSI as RF_SHAREABLE. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. MFC after: 1 week
2013-12-29 19:32:27 +00:00
bus_release_resource(dev, SYS_RES_MEMORY,
rman_get_rid(sc->mem), sc->mem);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
if (sc->sc_cdev) {
destroy_dev(sc->sc_cdev);
sc->sc_cdev = NULL;
}
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_LOCK_DESTROY(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return 0;
}
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
static int
iwn_shutdown(device_t dev)
{
struct iwn_softc *sc = device_get_softc(dev);
iwn_stop(sc);
return 0;
}
static int
iwn_suspend(device_t dev)
{
struct iwn_softc *sc = device_get_softc(dev);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ieee80211_suspend_all(&sc->sc_ic);
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
return 0;
}
static int
iwn_resume(device_t dev)
{
struct iwn_softc *sc = device_get_softc(dev);
/* Clear device-specific "PCI retry timeout" register (41h). */
pci_write_config(dev, 0x41, 0, 1);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ieee80211_resume_all(&sc->sc_ic);
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
return 0;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_nic_lock(struct iwn_softc *sc)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
int ntries;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Request exclusive access to NIC. */
IWN_SETBITS(sc, IWN_GP_CNTRL, IWN_GP_CNTRL_MAC_ACCESS_REQ);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Spin until we actually get the lock. */
for (ntries = 0; ntries < 1000; ntries++) {
if ((IWN_READ(sc, IWN_GP_CNTRL) &
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
(IWN_GP_CNTRL_MAC_ACCESS_ENA | IWN_GP_CNTRL_SLEEP)) ==
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_GP_CNTRL_MAC_ACCESS_ENA)
return 0;
DELAY(10);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return ETIMEDOUT;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
static __inline void
iwn_nic_unlock(struct iwn_softc *sc)
{
IWN_CLRBITS(sc, IWN_GP_CNTRL, IWN_GP_CNTRL_MAC_ACCESS_REQ);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
static __inline uint32_t
iwn_prph_read(struct iwn_softc *sc, uint32_t addr)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_PRPH_RADDR, IWN_PRPH_DWORD | addr);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_BARRIER_READ_WRITE(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return IWN_READ(sc, IWN_PRPH_RDATA);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
static __inline void
iwn_prph_write(struct iwn_softc *sc, uint32_t addr, uint32_t data)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_PRPH_WADDR, IWN_PRPH_DWORD | addr);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_BARRIER_WRITE(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_PRPH_WDATA, data);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
static __inline void
iwn_prph_setbits(struct iwn_softc *sc, uint32_t addr, uint32_t mask)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_prph_write(sc, addr, iwn_prph_read(sc, addr) | mask);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
static __inline void
iwn_prph_clrbits(struct iwn_softc *sc, uint32_t addr, uint32_t mask)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_prph_write(sc, addr, iwn_prph_read(sc, addr) & ~mask);
}
static __inline void
iwn_prph_write_region_4(struct iwn_softc *sc, uint32_t addr,
const uint32_t *data, int count)
{
for (; count > 0; count--, data++, addr += 4)
iwn_prph_write(sc, addr, *data);
}
static __inline uint32_t
iwn_mem_read(struct iwn_softc *sc, uint32_t addr)
{
IWN_WRITE(sc, IWN_MEM_RADDR, addr);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_BARRIER_READ_WRITE(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return IWN_READ(sc, IWN_MEM_RDATA);
}
static __inline void
iwn_mem_write(struct iwn_softc *sc, uint32_t addr, uint32_t data)
{
IWN_WRITE(sc, IWN_MEM_WADDR, addr);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_BARRIER_WRITE(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_MEM_WDATA, data);
}
static __inline void
iwn_mem_write_2(struct iwn_softc *sc, uint32_t addr, uint16_t data)
{
uint32_t tmp;
tmp = iwn_mem_read(sc, addr & ~3);
if (addr & 3)
tmp = (tmp & 0x0000ffff) | data << 16;
else
tmp = (tmp & 0xffff0000) | data;
iwn_mem_write(sc, addr & ~3, tmp);
}
static __inline void
iwn_mem_read_region_4(struct iwn_softc *sc, uint32_t addr, uint32_t *data,
int count)
{
for (; count > 0; count--, addr += 4)
*data++ = iwn_mem_read(sc, addr);
}
static __inline void
iwn_mem_set_region_4(struct iwn_softc *sc, uint32_t addr, uint32_t val,
int count)
{
for (; count > 0; count--, addr += 4)
iwn_mem_write(sc, addr, val);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_eeprom_lock(struct iwn_softc *sc)
{
int i, ntries;
for (i = 0; i < 100; i++) {
/* Request exclusive access to EEPROM. */
IWN_SETBITS(sc, IWN_HW_IF_CONFIG,
IWN_HW_IF_CONFIG_EEPROM_LOCKED);
/* Spin until we actually get the lock. */
for (ntries = 0; ntries < 100; ntries++) {
if (IWN_READ(sc, IWN_HW_IF_CONFIG) &
IWN_HW_IF_CONFIG_EEPROM_LOCKED)
return 0;
DELAY(10);
}
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end timeout\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return ETIMEDOUT;
}
static __inline void
iwn_eeprom_unlock(struct iwn_softc *sc)
{
IWN_CLRBITS(sc, IWN_HW_IF_CONFIG, IWN_HW_IF_CONFIG_EEPROM_LOCKED);
}
/*
* Initialize access by host to One Time Programmable ROM.
* NB: This kind of ROM can be found on 1000 or 6000 Series only.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_init_otprom(struct iwn_softc *sc)
{
Fix for the Intel WiFi Link 1000. The EEPROM image is in the OTPROM block before the last block, not in the last block itself. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 3 weeks
2010-02-15 19:18:46 +00:00
uint16_t prev, base, next;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
int count, error;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Wait for clock stabilization before accessing prph. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_clock_wait(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
iwn_prph_setbits(sc, IWN_APMG_PS, IWN_APMG_PS_RESET_REQ);
DELAY(5);
iwn_prph_clrbits(sc, IWN_APMG_PS, IWN_APMG_PS_RESET_REQ);
iwn_nic_unlock(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Set auto clock gate disable bit for HW with OTP shadow RAM. */
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->shadow_ram_support) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_SETBITS(sc, IWN_DBG_LINK_PWR_MGMT,
IWN_RESET_LINK_PWR_MGMT_DIS);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_CLRBITS(sc, IWN_EEPROM_GP, IWN_EEPROM_GP_IF_OWNER);
/* Clear ECC status. */
IWN_SETBITS(sc, IWN_OTP_GP,
IWN_OTP_GP_ECC_CORR_STTS | IWN_OTP_GP_ECC_UNCORR_STTS);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/*
Fix for the Intel WiFi Link 1000. The EEPROM image is in the OTPROM block before the last block, not in the last block itself. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 3 weeks
2010-02-15 19:18:46 +00:00
* Find the block before last block (contains the EEPROM image)
* for HW without OTP shadow RAM.
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
*/
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (! sc->base_params->shadow_ram_support) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Switch to absolute addressing mode. */
IWN_CLRBITS(sc, IWN_OTP_GP, IWN_OTP_GP_RELATIVE_ACCESS);
Fix for the Intel WiFi Link 1000. The EEPROM image is in the OTPROM block before the last block, not in the last block itself. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 3 weeks
2010-02-15 19:18:46 +00:00
base = prev = 0;
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
for (count = 0; count < sc->base_params->max_ll_items;
count++) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
error = iwn_read_prom_data(sc, base, &next, 2);
if (error != 0)
return error;
if (next == 0) /* End of linked-list. */
break;
Fix for the Intel WiFi Link 1000. The EEPROM image is in the OTPROM block before the last block, not in the last block itself. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 3 weeks
2010-02-15 19:18:46 +00:00
prev = base;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
base = le16toh(next);
}
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (count == 0 || count == sc->base_params->max_ll_items)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
return EIO;
/* Skip "next" word. */
Fix for the Intel WiFi Link 1000. The EEPROM image is in the OTPROM block before the last block, not in the last block itself. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 3 weeks
2010-02-15 19:18:46 +00:00
sc->prom_base = prev + 1;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return 0;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_read_prom_data(struct iwn_softc *sc, uint32_t addr, void *data, int count)
{
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
uint8_t *out = data;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint32_t val, tmp;
int ntries;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
addr += sc->prom_base;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
for (; count > 0; count -= 2, addr++) {
IWN_WRITE(sc, IWN_EEPROM, addr << 2);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
for (ntries = 0; ntries < 10; ntries++) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
val = IWN_READ(sc, IWN_EEPROM);
if (val & IWN_EEPROM_READ_VALID)
break;
DELAY(5);
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (ntries == 10) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
device_printf(sc->sc_dev,
"timeout reading ROM at 0x%x\n", addr);
return ETIMEDOUT;
}
if (sc->sc_flags & IWN_FLAG_HAS_OTPROM) {
/* OTPROM, check for ECC errors. */
tmp = IWN_READ(sc, IWN_OTP_GP);
if (tmp & IWN_OTP_GP_ECC_UNCORR_STTS) {
device_printf(sc->sc_dev,
"OTPROM ECC error at 0x%x\n", addr);
return EIO;
}
if (tmp & IWN_OTP_GP_ECC_CORR_STTS) {
/* Correctable ECC error, clear bit. */
IWN_SETBITS(sc, IWN_OTP_GP,
IWN_OTP_GP_ECC_CORR_STTS);
}
}
*out++ = val >> 16;
if (count > 1)
*out++ = val >> 24;
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return 0;
}
static void
iwn_dma_map_addr(void *arg, bus_dma_segment_t *segs, int nsegs, int error)
{
if (error != 0)
return;
KASSERT(nsegs == 1, ("too many DMA segments, %d should be 1", nsegs));
*(bus_addr_t *)arg = segs[0].ds_addr;
}
static int
iwn_dma_contig_alloc(struct iwn_softc *sc, struct iwn_dma_info *dma,
Remove the flags argument of iwn_dma_contig_alloc(), it is always set as BUS_DMA_NOWAIT. While here also set BUS_DMA_COHERENT.
2011-04-16 08:57:17 +00:00
void **kvap, bus_size_t size, bus_size_t alignment)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
{
int error;
dma->tag = NULL;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
dma->size = size;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
error = bus_dma_tag_create(bus_get_dma_tag(sc->sc_dev), alignment,
0, BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL, size,
Remove the flags argument of iwn_dma_contig_alloc(), it is always set as BUS_DMA_NOWAIT. While here also set BUS_DMA_COHERENT.
2011-04-16 08:57:17 +00:00
1, size, BUS_DMA_NOWAIT, NULL, NULL, &dma->tag);
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
if (error != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
goto fail;
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
error = bus_dmamem_alloc(dma->tag, (void **)&dma->vaddr,
Remove the flags argument of iwn_dma_contig_alloc(), it is always set as BUS_DMA_NOWAIT. While here also set BUS_DMA_COHERENT.
2011-04-16 08:57:17 +00:00
BUS_DMA_NOWAIT | BUS_DMA_ZERO | BUS_DMA_COHERENT, &dma->map);
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
if (error != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
goto fail;
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
Remove the flags argument of iwn_dma_contig_alloc(), it is always set as BUS_DMA_NOWAIT. While here also set BUS_DMA_COHERENT.
2011-04-16 08:57:17 +00:00
error = bus_dmamap_load(dma->tag, dma->map, dma->vaddr, size,
iwn_dma_map_addr, &dma->paddr, BUS_DMA_NOWAIT);
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
if (error != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
goto fail;
Add missing bus_dmamap_sync calls as well as remove two duplicate ones. Obtained from: OpenBSD
2011-04-16 10:52:11 +00:00
bus_dmamap_sync(dma->tag, dma->map, BUS_DMASYNC_PREWRITE);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (kvap != NULL)
*kvap = dma->vaddr;
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return 0;
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
fail: iwn_dma_contig_free(dma);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_dma_contig_free(struct iwn_dma_info *dma)
{
- Don't check the dma map address for a static DMA mapping against NULL to determine if the mapping is valid. - Don't explicitly destroy the dmap map for static DMA mappings. Reviewed by: adrian
2014-06-13 19:40:02 +00:00
if (dma->vaddr != NULL) {
bus_dmamap_sync(dma->tag, dma->map,
BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(dma->tag, dma->map);
bus_dmamem_free(dma->tag, dma->vaddr, dma->map);
dma->vaddr = NULL;
Make sure to destroy all DMA tags and maps.
2011-04-16 10:35:02 +00:00
}
if (dma->tag != NULL) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
bus_dma_tag_destroy(dma->tag);
Make sure to destroy all DMA tags and maps.
2011-04-16 10:35:02 +00:00
dma->tag = NULL;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_alloc_sched(struct iwn_softc *sc)
{
/* TX scheduler rings must be aligned on a 1KB boundary. */
Remove the flags argument of iwn_dma_contig_alloc(), it is always set as BUS_DMA_NOWAIT. While here also set BUS_DMA_COHERENT.
2011-04-16 08:57:17 +00:00
return iwn_dma_contig_alloc(sc, &sc->sched_dma, (void **)&sc->sched,
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
sc->schedsz, 1024);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_free_sched(struct iwn_softc *sc)
{
iwn_dma_contig_free(&sc->sched_dma);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_alloc_kw(struct iwn_softc *sc)
{
/* "Keep Warm" page must be aligned on a 4KB boundary. */
Remove the flags argument of iwn_dma_contig_alloc(), it is always set as BUS_DMA_NOWAIT. While here also set BUS_DMA_COHERENT.
2011-04-16 08:57:17 +00:00
return iwn_dma_contig_alloc(sc, &sc->kw_dma, NULL, 4096, 4096);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_free_kw(struct iwn_softc *sc)
{
iwn_dma_contig_free(&sc->kw_dma);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_alloc_ict(struct iwn_softc *sc)
{
/* ICT table must be aligned on a 4KB boundary. */
Remove the flags argument of iwn_dma_contig_alloc(), it is always set as BUS_DMA_NOWAIT. While here also set BUS_DMA_COHERENT.
2011-04-16 08:57:17 +00:00
return iwn_dma_contig_alloc(sc, &sc->ict_dma, (void **)&sc->ict,
IWN_ICT_SIZE, 4096);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_free_ict(struct iwn_softc *sc)
{
iwn_dma_contig_free(&sc->ict_dma);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_alloc_fwmem(struct iwn_softc *sc)
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Must be aligned on a 16-byte boundary. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
return iwn_dma_contig_alloc(sc, &sc->fw_dma, NULL, sc->fwsz, 16);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_free_fwmem(struct iwn_softc *sc)
{
iwn_dma_contig_free(&sc->fw_dma);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_alloc_rx_ring(struct iwn_softc *sc, struct iwn_rx_ring *ring)
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
bus_size_t size;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
int i, error;
ring->cur = 0;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Allocate RX descriptors (256-byte aligned). */
size = IWN_RX_RING_COUNT * sizeof (uint32_t);
Remove the flags argument of iwn_dma_contig_alloc(), it is always set as BUS_DMA_NOWAIT. While here also set BUS_DMA_COHERENT.
2011-04-16 08:57:17 +00:00
error = iwn_dma_contig_alloc(sc, &ring->desc_dma, (void **)&ring->desc,
size, 256);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
"%s: could not allocate RX ring DMA memory, error %d\n",
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
__func__, error);
goto fail;
}
Unify TX/RX ring allocation, finish the descriptior DMA stuff before starting with data.
2011-04-16 10:38:27 +00:00
/* Allocate RX status area (16-byte aligned). */
error = iwn_dma_contig_alloc(sc, &ring->stat_dma, (void **)&ring->stat,
sizeof (struct iwn_rx_status), 16);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
"%s: could not allocate RX status DMA memory, error %d\n",
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
__func__, error);
goto fail;
}
Unify TX/RX ring allocation, finish the descriptior DMA stuff before starting with data.
2011-04-16 10:38:27 +00:00
/* Create RX buffer DMA tag. */
error = bus_dma_tag_create(bus_get_dma_tag(sc->sc_dev), 1, 0,
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL,
IWN_RBUF_SIZE, 1, IWN_RBUF_SIZE, BUS_DMA_NOWAIT, NULL, NULL,
&ring->data_dmat);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
"%s: could not create RX buf DMA tag, error %d\n",
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
__func__, error);
goto fail;
}
/*
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
* Allocate and map RX buffers.
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
*/
for (i = 0; i < IWN_RX_RING_COUNT; i++) {
struct iwn_rx_data *data = &ring->data[i];
bus_addr_t paddr;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
error = bus_dmamap_create(ring->data_dmat, 0, &data->map);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
"%s: could not create RX buf DMA map, error %d\n",
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
__func__, error);
goto fail;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Mechanically substitute flags from historic mbuf allocator with malloc(9) flags in sys/dev.
2012-12-04 09:32:43 +00:00
data->m = m_getjcl(M_NOWAIT, MT_DATA, M_PKTHDR,
OpenBSD uses IWN_RBUF_SIZE not MJUMPAGESIZE for the RX path, also replace caddr_t with void * to be in sync.
2011-04-16 08:59:21 +00:00
IWN_RBUF_SIZE);
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
if (data->m == NULL) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(sc->sc_dev,
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
"%s: could not allocate RX mbuf\n", __func__);
Prevent double-free, also use the same error codes as OpenBSD.
2011-04-16 11:11:22 +00:00
error = ENOBUFS;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
goto fail;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
error = bus_dmamap_load(ring->data_dmat, data->map,
OpenBSD uses IWN_RBUF_SIZE not MJUMPAGESIZE for the RX path, also replace caddr_t with void * to be in sync.
2011-04-16 08:59:21 +00:00
mtod(data->m, void *), IWN_RBUF_SIZE, iwn_dma_map_addr,
&paddr, BUS_DMA_NOWAIT);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0 && error != EFBIG) {
device_printf(sc->sc_dev,
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
"%s: can't map mbuf, error %d\n", __func__,
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
goto fail;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Set physical address of RX buffer (256-byte aligned). */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ring->desc[i] = htole32(paddr >> 8);
}
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
bus_dmamap_sync(ring->desc_dma.tag, ring->desc_dma.map,
BUS_DMASYNC_PREWRITE);
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return 0;
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
fail: iwn_free_rx_ring(sc, ring);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end in error\n",__func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_reset_rx_ring(struct iwn_softc *sc, struct iwn_rx_ring *ring)
{
int ntries;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (iwn_nic_lock(sc) == 0) {
IWN_WRITE(sc, IWN_FH_RX_CONFIG, 0);
for (ntries = 0; ntries < 1000; ntries++) {
if (IWN_READ(sc, IWN_FH_RX_STATUS) &
IWN_FH_RX_STATUS_IDLE)
break;
DELAY(10);
}
iwn_nic_unlock(sc);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ring->cur = 0;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
sc->last_rx_valid = 0;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_free_rx_ring(struct iwn_softc *sc, struct iwn_rx_ring *ring)
{
int i;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s \n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_dma_contig_free(&ring->desc_dma);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_dma_contig_free(&ring->stat_dma);
for (i = 0; i < IWN_RX_RING_COUNT; i++) {
struct iwn_rx_data *data = &ring->data[i];
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (data->m != NULL) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(ring->data_dmat, data->map,
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
BUS_DMASYNC_POSTREAD);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_unload(ring->data_dmat, data->map);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
m_freem(data->m);
Prevent double-free, also use the same error codes as OpenBSD.
2011-04-16 11:11:22 +00:00
data->m = NULL;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (data->map != NULL)
bus_dmamap_destroy(ring->data_dmat, data->map);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Make sure to destroy all DMA tags and maps.
2011-04-16 10:35:02 +00:00
if (ring->data_dmat != NULL) {
bus_dma_tag_destroy(ring->data_dmat);
ring->data_dmat = NULL;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_alloc_tx_ring(struct iwn_softc *sc, struct iwn_tx_ring *ring, int qid)
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
bus_addr_t paddr;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
bus_size_t size;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
int i, error;
ring->qid = qid;
ring->queued = 0;
ring->cur = 0;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Allocate TX descriptors (256-byte aligned). */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
size = IWN_TX_RING_COUNT * sizeof (struct iwn_tx_desc);
Remove the flags argument of iwn_dma_contig_alloc(), it is always set as BUS_DMA_NOWAIT. While here also set BUS_DMA_COHERENT.
2011-04-16 08:57:17 +00:00
error = iwn_dma_contig_alloc(sc, &ring->desc_dma, (void **)&ring->desc,
size, 256);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
"%s: could not allocate TX ring DMA memory, error %d\n",
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
__func__, error);
goto fail;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
size = IWN_TX_RING_COUNT * sizeof (struct iwn_tx_cmd);
Remove the flags argument of iwn_dma_contig_alloc(), it is always set as BUS_DMA_NOWAIT. While here also set BUS_DMA_COHERENT.
2011-04-16 08:57:17 +00:00
error = iwn_dma_contig_alloc(sc, &ring->cmd_dma, (void **)&ring->cmd,
size, 4);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
"%s: could not allocate TX cmd DMA memory, error %d\n",
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
__func__, error);
goto fail;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
error = bus_dma_tag_create(bus_get_dma_tag(sc->sc_dev), 1, 0,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL, MCLBYTES,
IWN_MAX_SCATTER - 1, MCLBYTES, BUS_DMA_NOWAIT, NULL, NULL,
&ring->data_dmat);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
"%s: could not create TX buf DMA tag, error %d\n",
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
__func__, error);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
goto fail;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
paddr = ring->cmd_dma.paddr;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (i = 0; i < IWN_TX_RING_COUNT; i++) {
struct iwn_tx_data *data = &ring->data[i];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
data->cmd_paddr = paddr;
data->scratch_paddr = paddr + 12;
paddr += sizeof (struct iwn_tx_cmd);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
error = bus_dmamap_create(ring->data_dmat, 0, &data->map);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Replace RX/TX ring allocation error messages with something more sane and remove those where the caller already prints one.
2011-04-16 11:14:42 +00:00
"%s: could not create TX buf DMA map, error %d\n",
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
__func__, error);
goto fail;
}
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return 0;
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
fail: iwn_free_tx_ring(sc, ring);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end in error\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_reset_tx_ring(struct iwn_softc *sc, struct iwn_tx_ring *ring)
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
int i;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->doing %s \n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (i = 0; i < IWN_TX_RING_COUNT; i++) {
struct iwn_tx_data *data = &ring->data[i];
if (data->m != NULL) {
Add missing bus_dmamap_sync calls as well as remove two duplicate ones. Obtained from: OpenBSD
2011-04-16 10:52:11 +00:00
bus_dmamap_sync(ring->data_dmat, data->map,
BUS_DMASYNC_POSTWRITE);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_unload(ring->data_dmat, data->map);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
m_freem(data->m);
data->m = NULL;
}
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
if (data->ni != NULL) {
ieee80211_free_node(data->ni);
data->ni = NULL;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Clear TX descriptors. */
memset(ring->desc, 0, ring->desc_dma.size);
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
bus_dmamap_sync(ring->desc_dma.tag, ring->desc_dma.map,
BUS_DMASYNC_PREWRITE);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
sc->qfullmsk &= ~(1 << ring->qid);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ring->queued = 0;
ring->cur = 0;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_free_tx_ring(struct iwn_softc *sc, struct iwn_tx_ring *ring)
{
int i;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s \n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_dma_contig_free(&ring->desc_dma);
iwn_dma_contig_free(&ring->cmd_dma);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
for (i = 0; i < IWN_TX_RING_COUNT; i++) {
struct iwn_tx_data *data = &ring->data[i];
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (data->m != NULL) {
bus_dmamap_sync(ring->data_dmat, data->map,
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(ring->data_dmat, data->map);
m_freem(data->m);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (data->map != NULL)
bus_dmamap_destroy(ring->data_dmat, data->map);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Make sure to destroy all DMA tags and maps.
2011-04-16 10:35:02 +00:00
if (ring->data_dmat != NULL) {
bus_dma_tag_destroy(ring->data_dmat);
ring->data_dmat = NULL;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn5000_ict_reset(struct iwn_softc *sc)
{
/* Disable interrupts. */
IWN_WRITE(sc, IWN_INT_MASK, 0);
/* Reset ICT table. */
memset(sc->ict, 0, IWN_ICT_SIZE);
sc->ict_cur = 0;
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Set physical address of ICT table (4KB aligned). */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
DPRINTF(sc, IWN_DEBUG_RESET, "%s: enabling ICT\n", __func__);
IWN_WRITE(sc, IWN_DRAM_INT_TBL, IWN_DRAM_INT_TBL_ENABLE |
IWN_DRAM_INT_TBL_WRAP_CHECK | sc->ict_dma.paddr >> 12);
/* Enable periodic RX interrupt. */
sc->int_mask |= IWN_INT_RX_PERIODIC;
/* Switch to ICT interrupt mode in driver. */
sc->sc_flags |= IWN_FLAG_USE_ICT;
/* Re-enable interrupts. */
IWN_WRITE(sc, IWN_INT, 0xffffffff);
IWN_WRITE(sc, IWN_INT_MASK, sc->int_mask);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_read_eeprom(struct iwn_softc *sc, uint8_t macaddr[IEEE80211_ADDR_LEN])
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint16_t val;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
int error;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Check whether adapter has an EEPROM or an OTPROM. */
if (sc->hw_type >= IWN_HW_REV_TYPE_1000 &&
(IWN_READ(sc, IWN_OTP_GP) & IWN_OTP_GP_DEV_SEL_OTP))
sc->sc_flags |= IWN_FLAG_HAS_OTPROM;
DPRINTF(sc, IWN_DEBUG_RESET, "%s found\n",
(sc->sc_flags & IWN_FLAG_HAS_OTPROM) ? "OTPROM" : "EEPROM");
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Adapter has to be powered on for EEPROM access to work. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_apm_init(sc)) != 0) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
device_printf(sc->sc_dev,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"%s: could not power ON adapter, error %d\n", __func__,
error);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
return error;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if ((IWN_READ(sc, IWN_EEPROM_GP) & 0x7) == 0) {
device_printf(sc->sc_dev, "%s: bad ROM signature\n", __func__);
return EIO;
}
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_eeprom_lock(sc)) != 0) {
device_printf(sc->sc_dev, "%s: could not lock ROM, error %d\n",
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
__func__, error);
return error;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (sc->sc_flags & IWN_FLAG_HAS_OTPROM) {
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_init_otprom(sc)) != 0) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
device_printf(sc->sc_dev,
"%s: could not initialize OTPROM, error %d\n",
__func__, error);
return error;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Add some new features: - 6000 series devices need enhanced sensitivity calibration. - 6000 series devices need a different setting for the shadow reg. - set the IWN_FLAG_HAS_11N bit if the EEPROM says the device has 11n support. Obtained from: OpenBSD
2011-04-16 14:56:13 +00:00
iwn_read_prom_data(sc, IWN_EEPROM_SKU_CAP, &val, 2);
DPRINTF(sc, IWN_DEBUG_RESET, "SKU capabilities=0x%04x\n", le16toh(val));
/* Check if HT support is bonded out. */
if (val & htole16(IWN_EEPROM_SKU_CAP_11N))
sc->sc_flags |= IWN_FLAG_HAS_11N;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_read_prom_data(sc, IWN_EEPROM_RFCFG, &val, 2);
sc->rfcfg = le16toh(val);
DPRINTF(sc, IWN_DEBUG_RESET, "radio config=0x%04x\n", sc->rfcfg);
Read RX/TX chainmasks directly of the EEPROM. Some chips are known to have the wrong/broken information stored, keep the hardcoded values for those. Obtained from: OpenBSD
2011-04-16 14:49:29 +00:00
/* Read Tx/Rx chains from ROM unless it's known to be broken. */
if (sc->txchainmask == 0)
sc->txchainmask = IWN_RFCFG_TXANTMSK(sc->rfcfg);
if (sc->rxchainmask == 0)
sc->rxchainmask = IWN_RFCFG_RXANTMSK(sc->rfcfg);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Read MAC address. */
iwn_read_prom_data(sc, IWN_EEPROM_MAC, macaddr, 6);
/* Read adapter-specific information from EEPROM. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
ops->read_eeprom(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_apm_stop(sc); /* Power OFF adapter. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_eeprom_unlock(sc);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return 0;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_read_eeprom(struct iwn_softc *sc)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
uint32_t addr;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint16_t val;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
int i;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Read regulatory domain (4 ASCII characters). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_read_prom_data(sc, IWN4965_EEPROM_DOMAIN, sc->eeprom_domain, 4);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Read the list of authorized channels (20MHz ones only). */
Use IWN_NBANDS rather than a hard-coded limit. Tested: * Intel 5100, STA
2013-11-11 08:54:45 +00:00
for (i = 0; i < IWN_NBANDS - 1; i++) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
addr = iwn4965_regulatory_bands[i];
iwn_read_eeprom_channels(sc, i, addr);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Read maximum allowed TX power for 2GHz and 5GHz bands. */
iwn_read_prom_data(sc, IWN4965_EEPROM_MAXPOW, &val, 2);
sc->maxpwr2GHz = val & 0xff;
sc->maxpwr5GHz = val >> 8;
/* Check that EEPROM values are within valid range. */
if (sc->maxpwr5GHz < 20 || sc->maxpwr5GHz > 50)
sc->maxpwr5GHz = 38;
if (sc->maxpwr2GHz < 20 || sc->maxpwr2GHz > 50)
sc->maxpwr2GHz = 38;
DPRINTF(sc, IWN_DEBUG_RESET, "maxpwr 2GHz=%d 5GHz=%d\n",
sc->maxpwr2GHz, sc->maxpwr5GHz);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Read samples for each TX power group. */
iwn_read_prom_data(sc, IWN4965_EEPROM_BANDS, sc->bands,
sizeof sc->bands);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Read voltage at which samples were taken. */
iwn_read_prom_data(sc, IWN4965_EEPROM_VOLTAGE, &val, 2);
sc->eeprom_voltage = (int16_t)le16toh(val);
DPRINTF(sc, IWN_DEBUG_RESET, "voltage=%d (in 0.3V)\n",
sc->eeprom_voltage);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
#ifdef IWN_DEBUG
/* Print samples. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (sc->sc_debug & IWN_DEBUG_ANY) {
Fix off-by-one. Sorry!
2013-11-11 08:55:38 +00:00
for (i = 0; i < IWN_NBANDS - 1; i++)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_print_power_group(sc, i);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
#endif
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
#ifdef IWN_DEBUG
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_print_power_group(struct iwn_softc *sc, int i)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn4965_eeprom_band *band = &sc->bands[i];
struct iwn4965_eeprom_chan_samples *chans = band->chans;
int j, c;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
printf("===band %d===\n", i);
printf("chan lo=%d, chan hi=%d\n", band->lo, band->hi);
printf("chan1 num=%d\n", chans[0].num);
for (c = 0; c < 2; c++) {
for (j = 0; j < IWN_NSAMPLES; j++) {
printf("chain %d, sample %d: temp=%d gain=%d "
"power=%d pa_det=%d\n", c, j,
chans[0].samples[c][j].temp,
chans[0].samples[c][j].gain,
chans[0].samples[c][j].power,
chans[0].samples[c][j].pa_det);
}
}
printf("chan2 num=%d\n", chans[1].num);
for (c = 0; c < 2; c++) {
for (j = 0; j < IWN_NSAMPLES; j++) {
printf("chain %d, sample %d: temp=%d gain=%d "
"power=%d pa_det=%d\n", c, j,
chans[1].samples[c][j].temp,
chans[1].samples[c][j].gain,
chans[1].samples[c][j].power,
chans[1].samples[c][j].pa_det);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
#endif
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_read_eeprom(struct iwn_softc *sc)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
* Rename bluetooth coexistence flags, no binary change. * Enable DC calibration and crystal calibration on Centrino Advanced-N 6250 parts. * Workaround for a HW bug (does not affect 4965AGN) that may sporadically affect latency under some rare circumstances. From a similar commit to iwlwifi. * Update sensitivity settings for 5000 series to workaround a performance bug in the DSP (1000 is not affected so we keep the old values for 5000). * Update sensitivity settings for 6000 series. * Set differential gains on 6250 too (but use a 1.0 factor, not 1.5). * Init OFDM sensitivity with min value (which depends on the chip) instead of hardcoding it to 90. * Read calibration version from ROM and set IWN_GP_DRIVER_CALIB_VER6 bit on 6x50 if version >= 6. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 2 weeks
2010-04-10 06:58:24 +00:00
struct iwn5000_eeprom_calib_hdr hdr;
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
int32_t volt;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
uint32_t base, addr;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint16_t val;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
int i;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Read regulatory domain (4 ASCII characters). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_read_prom_data(sc, IWN5000_EEPROM_REG, &val, 2);
base = le16toh(val);
iwn_read_prom_data(sc, base + IWN5000_EEPROM_DOMAIN,
sc->eeprom_domain, 4);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Read the list of authorized channels (20MHz ones only). */
Use IWN_NBANDS rather than a hard-coded limit. Tested: * Intel 5100, STA
2013-11-11 08:54:45 +00:00
for (i = 0; i < IWN_NBANDS - 1; i++) {
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
addr = base + sc->base_params->regulatory_bands[i];
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_read_eeprom_channels(sc, i, addr);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Read enhanced TX power information for 6000 Series. */
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->enhanced_TX_power)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_read_eeprom_enhinfo(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_read_prom_data(sc, IWN5000_EEPROM_CAL, &val, 2);
base = le16toh(val);
* Rename bluetooth coexistence flags, no binary change. * Enable DC calibration and crystal calibration on Centrino Advanced-N 6250 parts. * Workaround for a HW bug (does not affect 4965AGN) that may sporadically affect latency under some rare circumstances. From a similar commit to iwlwifi. * Update sensitivity settings for 5000 series to workaround a performance bug in the DSP (1000 is not affected so we keep the old values for 5000). * Update sensitivity settings for 6000 series. * Set differential gains on 6250 too (but use a 1.0 factor, not 1.5). * Init OFDM sensitivity with min value (which depends on the chip) instead of hardcoding it to 90. * Read calibration version from ROM and set IWN_GP_DRIVER_CALIB_VER6 bit on 6x50 if version >= 6. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 2 weeks
2010-04-10 06:58:24 +00:00
iwn_read_prom_data(sc, base, &hdr, sizeof hdr);
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"%s: calib version=%u pa type=%u voltage=%u\n", __func__,
hdr.version, hdr.pa_type, le16toh(hdr.volt));
Fix some small whitespace nits. MFC after: 3 days
2010-07-15 08:05:20 +00:00
sc->calib_ver = hdr.version;
* Rename bluetooth coexistence flags, no binary change. * Enable DC calibration and crystal calibration on Centrino Advanced-N 6250 parts. * Workaround for a HW bug (does not affect 4965AGN) that may sporadically affect latency under some rare circumstances. From a similar commit to iwlwifi. * Update sensitivity settings for 5000 series to workaround a performance bug in the DSP (1000 is not affected so we keep the old values for 5000). * Update sensitivity settings for 6000 series. * Set differential gains on 6250 too (but use a 1.0 factor, not 1.5). * Init OFDM sensitivity with min value (which depends on the chip) instead of hardcoding it to 90. * Read calibration version from ROM and set IWN_GP_DRIVER_CALIB_VER6 bit on 6x50 if version >= 6. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 2 weeks
2010-04-10 06:58:24 +00:00
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->calib_need & IWN_FLG_NEED_PHY_CALIB_TEMP_OFFSETv2) {
sc->eeprom_voltage = le16toh(hdr.volt);
iwn_read_prom_data(sc, base + IWN5000_EEPROM_TEMP, &val, 2);
sc->eeprom_temp_high=le16toh(val);
iwn_read_prom_data(sc, base + IWN5000_EEPROM_VOLT, &val, 2);
sc->eeprom_temp = le16toh(val);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (sc->hw_type == IWN_HW_REV_TYPE_5150) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Compute temperature offset. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_read_prom_data(sc, base + IWN5000_EEPROM_TEMP, &val, 2);
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
sc->eeprom_temp = le16toh(val);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_read_prom_data(sc, base + IWN5000_EEPROM_VOLT, &val, 2);
volt = le16toh(val);
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
sc->temp_off = sc->eeprom_temp - (volt / -5);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE, "temp=%d volt=%d offset=%dK\n",
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
sc->eeprom_temp, volt, sc->temp_off);
} else {
/* Read crystal calibration. */
iwn_read_prom_data(sc, base + IWN5000_EEPROM_CRYSTAL,
&sc->eeprom_crystal, sizeof (uint32_t));
DPRINTF(sc, IWN_DEBUG_CALIBRATE, "crystal calibration 0x%08x\n",
le32toh(sc->eeprom_crystal));
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/*
* Translate EEPROM flags to net80211.
*/
static uint32_t
iwn_eeprom_channel_flags(struct iwn_eeprom_chan *channel)
{
uint32_t nflags;
nflags = 0;
if ((channel->flags & IWN_EEPROM_CHAN_ACTIVE) == 0)
nflags |= IEEE80211_CHAN_PASSIVE;
if ((channel->flags & IWN_EEPROM_CHAN_IBSS) == 0)
nflags |= IEEE80211_CHAN_NOADHOC;
if (channel->flags & IWN_EEPROM_CHAN_RADAR) {
nflags |= IEEE80211_CHAN_DFS;
/* XXX apparently IBSS may still be marked */
nflags |= IEEE80211_CHAN_NOADHOC;
}
return nflags;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
static void
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_read_eeprom_band(struct iwn_softc *sc, int n)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
struct iwn_eeprom_chan *channels = sc->eeprom_channels[n];
const struct iwn_chan_band *band = &iwn_bands[n];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct ieee80211_channel *c;
Obtain the channel number directly from the laster RXON command, also chan is an uint8_t.
2011-04-16 07:17:03 +00:00
uint8_t chan;
int i, nflags;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
for (i = 0; i < band->nchan; i++) {
if (!(channels[i].flags & IWN_EEPROM_CHAN_VALID)) {
DPRINTF(sc, IWN_DEBUG_RESET,
"skip chan %d flags 0x%x maxpwr %d\n",
band->chan[i], channels[i].flags,
channels[i].maxpwr);
continue;
}
chan = band->chan[i];
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
nflags = iwn_eeprom_channel_flags(&channels[i]);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
c = &ic->ic_channels[ic->ic_nchans++];
c->ic_ieee = chan;
c->ic_maxregpower = channels[i].maxpwr;
c->ic_maxpower = 2*c->ic_maxregpower;
iwn4965_set_txpower() uses maxpwr from EEPROM to calculate the power to set, it make sense to actually initialize that array. This fixes some issues with 4965 adapters where the TX power is crucial. This got lost somewhere while merging with the OpenBSD code. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 2 weeks
2010-04-10 07:00:26 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (n == 0) { /* 2GHz band */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
c->ic_freq = ieee80211_ieee2mhz(chan, IEEE80211_CHAN_G);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* G =>'s B is supported */
c->ic_flags = IEEE80211_CHAN_B | nflags;
c = &ic->ic_channels[ic->ic_nchans++];
c[0] = c[-1];
c->ic_flags = IEEE80211_CHAN_G | nflags;
} else { /* 5GHz band */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
c->ic_freq = ieee80211_ieee2mhz(chan, IEEE80211_CHAN_A);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
c->ic_flags = IEEE80211_CHAN_A | nflags;
}
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
/* Save maximum allowed TX power for this channel. */
sc->maxpwr[chan] = channels[i].maxpwr;
DPRINTF(sc, IWN_DEBUG_RESET,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"add chan %d flags 0x%x maxpwr %d\n", chan,
channels[i].flags, channels[i].maxpwr);
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
if (sc->sc_flags & IWN_FLAG_HAS_11N) {
/* add HT20, HT40 added separately */
c = &ic->ic_channels[ic->ic_nchans++];
c[0] = c[-1];
c->ic_flags |= IEEE80211_CHAN_HT20;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
static void
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_read_eeprom_ht40(struct iwn_softc *sc, int n)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
struct iwn_eeprom_chan *channels = sc->eeprom_channels[n];
const struct iwn_chan_band *band = &iwn_bands[n];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct ieee80211_channel *c, *cent, *extc;
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
uint8_t chan;
int i, nflags;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s start\n", __func__);
if (!(sc->sc_flags & IWN_FLAG_HAS_11N)) {
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end no 11n\n", __func__);
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
return;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
for (i = 0; i < band->nchan; i++) {
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
if (!(channels[i].flags & IWN_EEPROM_CHAN_VALID)) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DPRINTF(sc, IWN_DEBUG_RESET,
"skip chan %d flags 0x%x maxpwr %d\n",
band->chan[i], channels[i].flags,
channels[i].maxpwr);
continue;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
chan = band->chan[i];
nflags = iwn_eeprom_channel_flags(&channels[i]);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Each entry defines an HT40 channel pair; find the
* center channel, then the extension channel above.
*/
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
cent = ieee80211_find_channel_byieee(ic, chan,
(n == 5 ? IEEE80211_CHAN_G : IEEE80211_CHAN_A));
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (cent == NULL) { /* XXX shouldn't happen */
device_printf(sc->sc_dev,
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
"%s: no entry for channel %d\n", __func__, chan);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
continue;
}
extc = ieee80211_find_channel(ic, cent->ic_freq+20,
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
(n == 5 ? IEEE80211_CHAN_G : IEEE80211_CHAN_A));
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (extc == NULL) {
DPRINTF(sc, IWN_DEBUG_RESET,
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
"%s: skip chan %d, extension channel not found\n",
__func__, chan);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
continue;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DPRINTF(sc, IWN_DEBUG_RESET,
"add ht40 chan %d flags 0x%x maxpwr %d\n",
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
chan, channels[i].flags, channels[i].maxpwr);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
c = &ic->ic_channels[ic->ic_nchans++];
c[0] = cent[0];
c->ic_extieee = extc->ic_ieee;
c->ic_flags &= ~IEEE80211_CHAN_HT;
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
c->ic_flags |= IEEE80211_CHAN_HT40U | nflags;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
c = &ic->ic_channels[ic->ic_nchans++];
c[0] = extc[0];
c->ic_extieee = cent->ic_ieee;
c->ic_flags &= ~IEEE80211_CHAN_HT;
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
c->ic_flags |= IEEE80211_CHAN_HT40D | nflags;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
static void
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_read_eeprom_channels(struct iwn_softc *sc, int n, uint32_t addr)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_read_prom_data(sc, addr, &sc->eeprom_channels[n],
iwn_bands[n].nchan * sizeof (struct iwn_eeprom_chan));
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (n < 5)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_read_eeprom_band(sc, n);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
else
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_read_eeprom_ht40(sc, n);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
ieee80211_sort_channels(ic->ic_channels, ic->ic_nchans);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
static struct iwn_eeprom_chan *
iwn_find_eeprom_channel(struct iwn_softc *sc, struct ieee80211_channel *c)
{
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
int band, chan, i, j;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
Hook HT channel setup.
2011-05-08 10:31:22 +00:00
if (IEEE80211_IS_CHAN_HT40(c)) {
band = IEEE80211_IS_CHAN_5GHZ(c) ? 6 : 5;
if (IEEE80211_IS_CHAN_HT40D(c))
chan = c->ic_extieee;
else
chan = c->ic_ieee;
for (i = 0; i < iwn_bands[band].nchan; i++) {
if (iwn_bands[band].chan[i] == chan)
return &sc->eeprom_channels[band][i];
}
} else {
for (j = 0; j < 5; j++) {
for (i = 0; i < iwn_bands[j].nchan; i++) {
if (iwn_bands[j].chan[i] == c->ic_ieee)
return &sc->eeprom_channels[j][i];
}
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
}
}
return NULL;
}
/*
* Enforce flags read from EEPROM.
*/
static int
iwn_setregdomain(struct ieee80211com *ic, struct ieee80211_regdomain *rd,
int nchan, struct ieee80211_channel chans[])
{
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
struct iwn_softc *sc = ic->ic_softc;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
int i;
for (i = 0; i < nchan; i++) {
struct ieee80211_channel *c = &chans[i];
struct iwn_eeprom_chan *channel;
channel = iwn_find_eeprom_channel(sc, c);
if (channel == NULL) {
Use ic_printf() instead of if_printf().
2015-05-25 15:12:23 +00:00
ic_printf(ic, "%s: invalid channel %u freq %u/0x%x\n",
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
__func__, c->ic_ieee, c->ic_freq, c->ic_flags);
return EINVAL;
}
c->ic_flags |= iwn_eeprom_channel_flags(channel);
}
return 0;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_read_eeprom_enhinfo(struct iwn_softc *sc)
{
struct iwn_eeprom_enhinfo enhinfo[35];
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Use the enhanced TX power information availabe on newer EEPROMs.
2011-05-08 10:35:16 +00:00
struct ieee80211_channel *c;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
uint16_t val, base;
int8_t maxpwr;
Use the enhanced TX power information availabe on newer EEPROMs.
2011-05-08 10:35:16 +00:00
uint8_t flags;
int i, j;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_read_prom_data(sc, IWN5000_EEPROM_REG, &val, 2);
base = le16toh(val);
iwn_read_prom_data(sc, base + IWN6000_EEPROM_ENHINFO,
enhinfo, sizeof enhinfo);
for (i = 0; i < nitems(enhinfo); i++) {
Use the enhanced TX power information availabe on newer EEPROMs.
2011-05-08 10:35:16 +00:00
flags = enhinfo[i].flags;
if (!(flags & IWN_ENHINFO_VALID))
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
continue; /* Skip invalid entries. */
maxpwr = 0;
if (sc->txchainmask & IWN_ANT_A)
maxpwr = MAX(maxpwr, enhinfo[i].chain[0]);
if (sc->txchainmask & IWN_ANT_B)
maxpwr = MAX(maxpwr, enhinfo[i].chain[1]);
if (sc->txchainmask & IWN_ANT_C)
maxpwr = MAX(maxpwr, enhinfo[i].chain[2]);
if (sc->ntxchains == 2)
maxpwr = MAX(maxpwr, enhinfo[i].mimo2);
else if (sc->ntxchains == 3)
maxpwr = MAX(maxpwr, enhinfo[i].mimo3);
Use the enhanced TX power information availabe on newer EEPROMs.
2011-05-08 10:35:16 +00:00
for (j = 0; j < ic->ic_nchans; j++) {
c = &ic->ic_channels[j];
if ((flags & IWN_ENHINFO_5GHZ)) {
if (!IEEE80211_IS_CHAN_A(c))
continue;
} else if ((flags & IWN_ENHINFO_OFDM)) {
if (!IEEE80211_IS_CHAN_G(c))
continue;
} else if (!IEEE80211_IS_CHAN_B(c))
continue;
if ((flags & IWN_ENHINFO_HT40)) {
if (!IEEE80211_IS_CHAN_HT40(c))
continue;
} else {
if (IEEE80211_IS_CHAN_HT40(c))
continue;
}
if (enhinfo[i].chan != 0 &&
enhinfo[i].chan != c->ic_ieee)
continue;
DPRINTF(sc, IWN_DEBUG_RESET,
"channel %d(%x), maxpwr %d\n", c->ic_ieee,
c->ic_flags, maxpwr / 2);
c->ic_maxregpower = maxpwr / 2;
c->ic_maxpower = maxpwr;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static struct ieee80211_node *
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_node_alloc(struct ieee80211vap *vap, const uint8_t mac[IEEE80211_ADDR_LEN])
{
return malloc(sizeof (struct iwn_node), M_80211_NODE,M_NOWAIT | M_ZERO);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Prepare for transmitting frames at MCS rates: - instead of calling iwn_plcp_signal() for every frame, map the expected value directly within wn->ridx - concat plcp, rflags and xrflags, there is no clean byte boundary within the flags, for example the antenna setting uses bit 6, 7 and 8 - there is still need for a custom rate to plcp mapping, as those expected by the hardware are not conform to the std
2011-05-08 11:49:50 +00:00
static __inline int
rate2plcp(int rate)
{
switch (rate & 0xff) {
case 12: return 0xd;
case 18: return 0xf;
case 24: return 0x5;
case 36: return 0x7;
case 48: return 0x9;
case 72: return 0xb;
case 96: return 0x1;
case 108: return 0x3;
case 2: return 10;
case 4: return 20;
case 11: return 55;
case 22: return 110;
}
return 0;
}
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
static int
iwn_get_1stream_tx_antmask(struct iwn_softc *sc)
{
return IWN_LSB(sc->txchainmask);
}
static int
iwn_get_2stream_tx_antmask(struct iwn_softc *sc)
{
int tx;
/*
* The '2 stream' setup is a bit .. odd.
*
* For NICs that support only 1 antenna, default to IWN_ANT_AB or
* the firmware panics (eg Intel 5100.)
*
* For NICs that support two antennas, we use ANT_AB.
*
* For NICs that support three antennas, we use the two that
* wasn't the default one.
*
* XXX TODO: if bluetooth (full concurrent) is enabled, restrict
* this to only one antenna.
*/
/* Default - transmit on the other antennas */
tx = (sc->txchainmask & ~IWN_LSB(sc->txchainmask));
/* Now, if it's zero, set it to IWN_ANT_AB, so to not panic firmware */
if (tx == 0)
tx = IWN_ANT_AB;
/*
* If the NIC is a two-stream TX NIC, configure the TX mask to
* the default chainmask
*/
else if (sc->ntxchains == 2)
tx = sc->txchainmask;
return (tx);
}
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
/*
* Calculate the required PLCP value from the given rate,
* to the given node.
*
* This will take the node configuration (eg 11n, rate table
* setup, etc) into consideration.
*/
static uint32_t
iwn_rate_to_plcp(struct iwn_softc *sc, struct ieee80211_node *ni,
uint8_t rate)
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
{
Prepare for transmitting frames at MCS rates: - instead of calling iwn_plcp_signal() for every frame, map the expected value directly within wn->ridx - concat plcp, rflags and xrflags, there is no clean byte boundary within the flags, for example the antenna setting uses bit 6, 7 and 8 - there is still need for a custom rate to plcp mapping, as those expected by the hardware are not conform to the std
2011-05-08 11:49:50 +00:00
struct ieee80211com *ic = ni->ni_ic;
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
uint32_t plcp = 0;
int ridx;
Prepare for transmitting frames at MCS rates: - instead of calling iwn_plcp_signal() for every frame, map the expected value directly within wn->ridx - concat plcp, rflags and xrflags, there is no clean byte boundary within the flags, for example the antenna setting uses bit 6, 7 and 8 - there is still need for a custom rate to plcp mapping, as those expected by the hardware are not conform to the std
2011-05-08 11:49:50 +00:00
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
/*
* If it's an MCS rate, let's set the plcp correctly
* and set the relevant flags based on the node config.
*/
Fix the PLCP lookup code in iwn(4) to base the 11n decision on whether the rate is 11n, rather than whether the channel is 11n. This correctly allows the PLCP lookup code to return the legacy rates even on an 11n channel. PR: kern/183430
2013-10-29 04:03:00 +00:00
if (rate & IEEE80211_RATE_MCS) {
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
/*
* Set the initial PLCP value to be between 0->31 for
* MCS 0 -> MCS 31, then set the "I'm an MCS rate!"
* flag.
*/
net80211 & wireless drivers: remove duplicate defines (noop) * IEEE80211_DIR_DSTODS(wh) -> IEEE80211_IS_DSTODS(wh). * N(a) -> nitems(a). * Remove LE_READ_2(p)/LE_READ_4(p) definitions (and include ieee80211_input.h instead). * <drvname>_TXOP_TO_US(txop) -> IEEE80211_TXOP_TO_US(txop). * Put IEEE80211_RV(v) into ieee80211_proto.h and remove local RV(v) definitions. Submitted by: Andriy Voskoboinyk <s3erios@gmail.com> Differential Revision: https://reviews.freebsd.org/D3705
2015-09-22 02:44:59 +00:00
plcp = IEEE80211_RV(rate) | IWN_RFLAG_MCS;
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
/*
* XXX the following should only occur if both
* the local configuration _and_ the remote node
* advertise these capabilities. Thus this code
* may need fixing!
*/
/*
* Set the channel width and guard interval.
*/
if (IEEE80211_IS_CHAN_HT40(ni->ni_chan)) {
plcp |= IWN_RFLAG_HT40;
if (ni->ni_htcap & IEEE80211_HTCAP_SHORTGI40)
Add support for transmitting frames at MCS rates.
2011-05-08 11:54:38 +00:00
plcp |= IWN_RFLAG_SGI;
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
} else if (ni->ni_htcap & IEEE80211_HTCAP_SHORTGI20) {
plcp |= IWN_RFLAG_SGI;
Add support for transmitting frames at MCS rates.
2011-05-08 11:54:38 +00:00
}
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
/*
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
* Ensure the selected rate matches the link quality
* table entries being used.
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
*/
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
if (rate > 0x8f)
plcp |= IWN_RFLAG_ANT(sc->txchainmask);
else if (rate > 0x87)
plcp |= IWN_RFLAG_ANT(iwn_get_2stream_tx_antmask(sc));
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
else
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
plcp |= IWN_RFLAG_ANT(iwn_get_1stream_tx_antmask(sc));
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
} else {
/*
* Set the initial PLCP - fine for both
* OFDM and CCK rates.
*/
plcp = rate2plcp(rate);
/* Set CCK flag if it's CCK */
/* XXX It would be nice to have a method
* to map the ridx -> phy table entry
* so we could just query that, rather than
* this hack to check against IWN_RIDX_OFDM6.
*/
ridx = ieee80211_legacy_rate_lookup(ic->ic_rt,
rate & IEEE80211_RATE_VAL);
if (ridx < IWN_RIDX_OFDM6 &&
IEEE80211_IS_CHAN_2GHZ(ni->ni_chan))
plcp |= IWN_RFLAG_CCK;
/* Set antenna configuration */
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
/* XXX TODO: is this the right antenna to use for legacy? */
plcp |= IWN_RFLAG_ANT(iwn_get_1stream_tx_antmask(sc));
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
}
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
DPRINTF(sc, IWN_DEBUG_TXRATE, "%s: rate=0x%02x, plcp=0x%08x\n",
__func__,
rate,
plcp);
return (htole32(plcp));
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
}
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
static void
iwn_newassoc(struct ieee80211_node *ni, int isnew)
{
/* Doesn't do anything at the moment */
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_media_change(struct ifnet *ifp)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
int error;
error = ieee80211_media_change(ifp);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* NB: only the fixed rate can change and that doesn't need a reset */
return (error == ENETRESET ? 0 : error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_newstate(struct ieee80211vap *vap, enum ieee80211_state nstate, int arg)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_vap *ivp = IWN_VAP(vap);
struct ieee80211com *ic = vap->iv_ic;
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
struct iwn_softc *sc = ic->ic_softc;
Pass errors that might happen during state transitions up to net80211.
2011-04-16 07:21:00 +00:00
int error = 0;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DPRINTF(sc, IWN_DEBUG_STATE, "%s: %s -> %s\n", __func__,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
ieee80211_state_name[vap->iv_state], ieee80211_state_name[nstate]);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IEEE80211_UNLOCK(ic);
IWN_LOCK(sc);
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
callout_stop(&sc->calib_to);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon = &sc->rx_on[IWN_RXON_BSS_CTX];
Handle RUN->ASSOC->RUN transition correctly, as in not trigger a firmware error. Convert if statements to a switch statement while I'm here. Tested by: Benjamin Kaduk <kaduk at mit.edu> MFC after: 2 weeks
2010-07-15 11:52:20 +00:00
switch (nstate) {
case IEEE80211_S_ASSOC:
if (vap->iv_state != IEEE80211_S_RUN)
break;
/* FALLTHROUGH */
case IEEE80211_S_AUTH:
if (vap->iv_state == IEEE80211_S_AUTH)
break;
/*
* !AUTH -> AUTH transition requires state reset to handle
* reassociations correctly.
*/
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->associd = 0;
sc->rxon->filter &= ~htole32(IWN_FILTER_BSS);
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
sc->calib.state = IWN_CALIB_STATE_INIT;
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
/* Wait until we hear a beacon before we transmit */
sc->sc_beacon_wait = 1;
Pass errors that might happen during state transitions up to net80211.
2011-04-16 07:21:00 +00:00
if ((error = iwn_auth(sc, vap)) != 0) {
device_printf(sc->sc_dev,
"%s: could not move to auth state\n", __func__);
}
Handle RUN->ASSOC->RUN transition correctly, as in not trigger a firmware error. Convert if statements to a switch statement while I'm here. Tested by: Benjamin Kaduk <kaduk at mit.edu> MFC after: 2 weeks
2010-07-15 11:52:20 +00:00
break;
case IEEE80211_S_RUN:
/*
* RUN -> RUN transition; Just restart the timers.
*/
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
if (vap->iv_state == IEEE80211_S_RUN) {
sc->calib_cnt = 0;
Handle RUN->ASSOC->RUN transition correctly, as in not trigger a firmware error. Convert if statements to a switch statement while I'm here. Tested by: Benjamin Kaduk <kaduk at mit.edu> MFC after: 2 weeks
2010-07-15 11:52:20 +00:00
break;
}
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
/* Wait until we hear a beacon before we transmit */
sc->sc_beacon_wait = 1;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* !RUN -> RUN requires setting the association id
* which is done with a firmware cmd. We also defer
* starting the timers until that work is done.
*/
Pass errors that might happen during state transitions up to net80211.
2011-04-16 07:21:00 +00:00
if ((error = iwn_run(sc, vap)) != 0) {
device_printf(sc->sc_dev,
"%s: could not move to run state\n", __func__);
}
Handle RUN->ASSOC->RUN transition correctly, as in not trigger a firmware error. Convert if statements to a switch statement while I'm here. Tested by: Benjamin Kaduk <kaduk at mit.edu> MFC after: 2 weeks
2010-07-15 11:52:20 +00:00
break;
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
case IEEE80211_S_INIT:
sc->calib.state = IWN_CALIB_STATE_INIT;
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
/*
* Purge the xmit queue so we don't have old frames
* during a new association attempt.
*/
sc->sc_beacon_wait = 0;
iwn_xmit_queue_drain(sc);
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
break;
Handle RUN->ASSOC->RUN transition correctly, as in not trigger a firmware error. Convert if statements to a switch statement while I'm here. Tested by: Benjamin Kaduk <kaduk at mit.edu> MFC after: 2 weeks
2010-07-15 11:52:20 +00:00
default:
break;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_UNLOCK(sc);
IEEE80211_LOCK(ic);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
if (error != 0){
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end in error\n", __func__);
Pass errors that might happen during state transitions up to net80211.
2011-04-16 07:21:00 +00:00
return error;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
}
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return ivp->iv_newstate(vap, nstate, arg);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
static void
iwn_calib_timeout(void *arg)
{
struct iwn_softc *sc = arg;
IWN_LOCK_ASSERT(sc);
/* Force automatic TX power calibration every 60 secs. */
if (++sc->calib_cnt >= 120) {
uint32_t flags = 0;
DPRINTF(sc, IWN_DEBUG_CALIBRATE, "%s\n",
"sending request for statistics");
(void)iwn_cmd(sc, IWN_CMD_GET_STATISTICS, &flags,
sizeof flags, 1);
sc->calib_cnt = 0;
}
callout_reset(&sc->calib_to, msecs_to_ticks(500), iwn_calib_timeout,
sc);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Process an RX_PHY firmware notification. This is usually immediately
* followed by an MPDU_RX_DONE notification.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_rx_phy(struct iwn_softc *sc, struct iwn_rx_desc *desc,
struct iwn_rx_data *data)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_rx_stat *stat = (struct iwn_rx_stat *)(desc + 1);
DPRINTF(sc, IWN_DEBUG_CALIBRATE, "%s: received PHY stats\n", __func__);
o add more notification strings in iwn_intr_str() o sync with OpenBSD code Submitted by: Bernhard Schmidt <bschmidt at techwires.net> MFC after: 3 days
2010-01-25 19:36:48 +00:00
bus_dmamap_sync(sc->rxq.data_dmat, data->map, BUS_DMASYNC_POSTREAD);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Save RX statistics, they will be used on MPDU_RX_DONE. */
memcpy(&sc->last_rx_stat, stat, sizeof (*stat));
sc->last_rx_valid = 1;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Process an RX_DONE (4965AGN only) or MPDU_RX_DONE firmware notification.
* Each MPDU_RX_DONE notification must be preceded by an RX_PHY one.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_rx_done(struct iwn_softc *sc, struct iwn_rx_desc *desc,
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_rx_data *data)
{
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_rx_ring *ring = &sc->rxq;
struct ieee80211_frame *wh;
struct ieee80211_node *ni;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct mbuf *m, *m1;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_rx_stat *stat;
caddr_t head;
bus_addr_t paddr;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint32_t flags;
int error, len, rssi, nf;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (desc->type == IWN_MPDU_RX_DONE) {
/* Check for prior RX_PHY notification. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (!sc->last_rx_valid) {
DPRINTF(sc, IWN_DEBUG_ANY,
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
"%s: missing RX_PHY\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return;
}
stat = &sc->last_rx_stat;
} else
stat = (struct iwn_rx_stat *)(desc + 1);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(ring->data_dmat, data->map, BUS_DMASYNC_POSTREAD);
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (stat->cfg_phy_len > IWN_STAT_MAXLEN) {
device_printf(sc->sc_dev,
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
"%s: invalid RX statistic header, len %d\n", __func__,
stat->cfg_phy_len);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (desc->type == IWN_MPDU_RX_DONE) {
struct iwn_rx_mpdu *mpdu = (struct iwn_rx_mpdu *)(desc + 1);
head = (caddr_t)(mpdu + 1);
len = le16toh(mpdu->len);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
} else {
head = (caddr_t)(stat + 1) + stat->cfg_phy_len;
len = le16toh(stat->len);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
flags = le32toh(*(uint32_t *)(head + len));
/* Discard frames with a bad FCS early. */
if ((flags & IWN_RX_NOERROR) != IWN_RX_NOERROR) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
DPRINTF(sc, IWN_DEBUG_RECV, "%s: RX flags error %x\n",
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
__func__, flags);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
counter_u64_add(ic->ic_ierrors, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Discard frames that are too short. */
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
if (len < sizeof (struct ieee80211_frame_ack)) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_RECV, "%s: frame too short: %d\n",
__func__, len);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
counter_u64_add(ic->ic_ierrors, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return;
}
Mechanically substitute flags from historic mbuf allocator with malloc(9) flags in sys/dev.
2012-12-04 09:32:43 +00:00
m1 = m_getjcl(M_NOWAIT, MT_DATA, M_PKTHDR, IWN_RBUF_SIZE);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (m1 == NULL) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_ANY, "%s: no mbuf to restock ring\n",
__func__);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
counter_u64_add(ic->ic_ierrors, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_unload(ring->data_dmat, data->map);
OpenBSD uses IWN_RBUF_SIZE not MJUMPAGESIZE for the RX path, also replace caddr_t with void * to be in sync.
2011-04-16 08:59:21 +00:00
error = bus_dmamap_load(ring->data_dmat, data->map, mtod(m1, void *),
IWN_RBUF_SIZE, iwn_dma_map_addr, &paddr, BUS_DMA_NOWAIT);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0 && error != EFBIG) {
device_printf(sc->sc_dev,
"%s: bus_dmamap_load failed, error %d\n", __func__, error);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
m_freem(m1);
In case a new mbuf can't be loaded, reuse the old one.
2011-04-16 09:34:40 +00:00
/* Try to reload the old mbuf. */
error = bus_dmamap_load(ring->data_dmat, data->map,
mtod(data->m, void *), IWN_RBUF_SIZE, iwn_dma_map_addr,
&paddr, BUS_DMA_NOWAIT);
if (error != 0 && error != EFBIG) {
panic("%s: could not load old RX mbuf", __func__);
}
/* Physical address may have changed. */
ring->desc[ring->cur] = htole32(paddr >> 8);
bus_dmamap_sync(ring->data_dmat, ring->desc_dma.map,
BUS_DMASYNC_PREWRITE);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
counter_u64_add(ic->ic_ierrors, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return;
}
m = data->m;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
data->m = m1;
/* Update RX descriptor. */
ring->desc[ring->cur] = htole32(paddr >> 8);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(ring->desc_dma.tag, ring->desc_dma.map,
BUS_DMASYNC_PREWRITE);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Finalize mbuf. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
m->m_data = head;
m->m_pkthdr.len = m->m_len = len;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Grab a reference to the source node. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
wh = mtod(m, struct ieee80211_frame *);
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
if (len >= sizeof(struct ieee80211_frame_min))
ni = ieee80211_find_rxnode(ic, (struct ieee80211_frame_min *)wh);
else
ni = NULL;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
nf = (ni != NULL && ni->ni_vap->iv_state == IEEE80211_S_RUN &&
(ic->ic_flags & IEEE80211_F_SCAN) == 0) ? sc->noise : -95;
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
rssi = ops->get_rssi(sc, stat);
RSSI related syncs with the OpenBSD code: - read RSSI only for the active chains - cast RSSI/NF to int8_t before passing it up to radiotap - remove the htole64() for the timestamp Obtained from: OpenBSD
2011-04-16 07:32:57 +00:00
Overhaul monitor mode handling: o replace DLT_IEEE802_11 support in net80211 with DLT_IEEE802_11_RADIO and remove explicit bpf support from wireless drivers; drivers now use ieee80211_radiotap_attach to setup shared data structures that hold the radiotap header for each packet tx/rx o remove rx timestamp from the rx path; it was used only by the tdma support for debugging and was mostly useless due to it being 32-bits and mostly unavailable o track DLT_IEEE80211_RADIO bpf attachments and maintain per-vap and per-com state when there are active taps o track the number of monitor mode vaps o use bpf tap and monitor mode vap state to decide when to collect radiotap state and dispatch frames; drivers no longer explicitly directly check bpf state or use bpf calls to tap frames o handle radiotap state updates on channel change in net80211; drivers should not do this (unless they bypass net80211 which is almost always a mistake) o update various drivers to be more consistent/correct in handling radiotap o update ral to include TSF in radiotap'd frames o add promisc mode callback to wi Reviewed by: cbzimmer, rpaulo, thompsa
2009-05-20 20:00:40 +00:00
if (ieee80211_radiotap_active(ic)) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_rx_radiotap_header *tap = &sc->sc_rxtap;
Overhaul monitor mode handling: o replace DLT_IEEE802_11 support in net80211 with DLT_IEEE802_11_RADIO and remove explicit bpf support from wireless drivers; drivers now use ieee80211_radiotap_attach to setup shared data structures that hold the radiotap header for each packet tx/rx o remove rx timestamp from the rx path; it was used only by the tdma support for debugging and was mostly useless due to it being 32-bits and mostly unavailable o track DLT_IEEE80211_RADIO bpf attachments and maintain per-vap and per-com state when there are active taps o track the number of monitor mode vaps o use bpf tap and monitor mode vap state to decide when to collect radiotap state and dispatch frames; drivers no longer explicitly directly check bpf state or use bpf calls to tap frames o handle radiotap state updates on channel change in net80211; drivers should not do this (unless they bypass net80211 which is almost always a mistake) o update various drivers to be more consistent/correct in handling radiotap o update ral to include TSF in radiotap'd frames o add promisc mode callback to wi Reviewed by: cbzimmer, rpaulo, thompsa
2009-05-20 20:00:40 +00:00
tap->wr_flags = 0;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (stat->flags & htole16(IWN_STAT_FLAG_SHPREAMBLE))
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
tap->wr_flags |= IEEE80211_RADIOTAP_F_SHORTPRE;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
tap->wr_dbm_antsignal = (int8_t)rssi;
tap->wr_dbm_antnoise = (int8_t)nf;
tap->wr_tsft = stat->tstamp;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
switch (stat->rate) {
/* CCK rates. */
case 10: tap->wr_rate = 2; break;
case 20: tap->wr_rate = 4; break;
case 55: tap->wr_rate = 11; break;
case 110: tap->wr_rate = 22; break;
/* OFDM rates. */
case 0xd: tap->wr_rate = 12; break;
case 0xf: tap->wr_rate = 18; break;
case 0x5: tap->wr_rate = 24; break;
case 0x7: tap->wr_rate = 36; break;
case 0x9: tap->wr_rate = 48; break;
case 0xb: tap->wr_rate = 72; break;
case 0x1: tap->wr_rate = 96; break;
case 0x3: tap->wr_rate = 108; break;
/* Unknown rate: should not happen. */
default: tap->wr_rate = 0;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
/*
* If it's a beacon and we're waiting, then do the
* wakeup. This should unblock raw_xmit/start.
*/
if (sc->sc_beacon_wait) {
uint8_t type, subtype;
/* NB: Re-assign wh */
wh = mtod(m, struct ieee80211_frame *);
type = wh->i_fc[0] & IEEE80211_FC0_TYPE_MASK;
subtype = wh->i_fc[0] & IEEE80211_FC0_SUBTYPE_MASK;
/*
* This assumes at this point we've received our own
* beacon.
*/
DPRINTF(sc, IWN_DEBUG_TRACE,
"%s: beacon_wait, type=%d, subtype=%d\n",
__func__, type, subtype);
if (type == IEEE80211_FC0_TYPE_MGT &&
subtype == IEEE80211_FC0_SUBTYPE_BEACON) {
DPRINTF(sc, IWN_DEBUG_TRACE | IWN_DEBUG_XMIT,
"%s: waking things up\n", __func__);
/* queue taskqueue to transmit! */
taskqueue_enqueue(sc->sc_tq, &sc->sc_xmit_task);
}
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
IWN_UNLOCK(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Send the frame to the 802.11 layer. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (ni != NULL) {
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
if (ni->ni_flags & IEEE80211_NODE_HT)
m->m_flags |= M_AMPDU;
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
(void)ieee80211_input(ni, m, rssi - nf, nf);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Node is no longer needed. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ieee80211_free_node(ni);
} else
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
(void)ieee80211_input_all(ic, m, rssi - nf, nf);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
IWN_LOCK(sc);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Process an incoming Compressed BlockAck. */
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_rx_compressed_ba(struct iwn_softc *sc, struct iwn_rx_desc *desc,
struct iwn_rx_data *data)
{
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
struct iwn_ops *ops = &sc->ops;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
struct iwn_node *wn;
struct ieee80211_node *ni;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
struct iwn_compressed_ba *ba = (struct iwn_compressed_ba *)(desc + 1);
struct iwn_tx_ring *txq;
Fix a TX aggregation issue, if after the last compressed BA notification the TX queue is empty, there won't be a TX done notification, effectly resulting in an mbuf leak. The correct way to handle this is to free up mbufs on both BA and TX done notifications up to the last sent seqno. Tested by: osa@ MFC after: 3 days
2012-06-27 15:55:34 +00:00
struct iwn_tx_data *txdata;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
struct ieee80211_tx_ampdu *tap;
Fix a TX aggregation issue, if after the last compressed BA notification the TX queue is empty, there won't be a TX done notification, effectly resulting in an mbuf leak. The correct way to handle this is to free up mbufs on both BA and TX done notifications up to the last sent seqno. Tested by: osa@ MFC after: 3 days
2012-06-27 15:55:34 +00:00
struct mbuf *m;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
uint64_t bitmap;
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
uint16_t ssn;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
uint8_t tid;
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
int ackfailcnt = 0, i, lastidx, qid, *res, shift;
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
int tx_ok = 0, tx_err = 0;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE | IWN_DEBUG_XMIT, "->%s begin\n", __func__);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
Add missing bus_dmamap_sync calls as well as remove two duplicate ones. Obtained from: OpenBSD
2011-04-16 10:52:11 +00:00
bus_dmamap_sync(sc->rxq.data_dmat, data->map, BUS_DMASYNC_POSTREAD);
Fix a TX aggregation issue, if after the last compressed BA notification the TX queue is empty, there won't be a TX done notification, effectly resulting in an mbuf leak. The correct way to handle this is to free up mbufs on both BA and TX done notifications up to the last sent seqno. Tested by: osa@ MFC after: 3 days
2012-06-27 15:55:34 +00:00
qid = le16toh(ba->qid);
txq = &sc->txq[ba->qid];
tap = sc->qid2tap[ba->qid];
Migrate the net80211 TX aggregation state to be from per-AC to per-TID. TODO: * Test mwl(4) more thoroughly! Reviewed by: bschmidt (for iwn)
2012-04-15 20:29:39 +00:00
tid = tap->txa_tid;
Fix a TX aggregation issue, if after the last compressed BA notification the TX queue is empty, there won't be a TX done notification, effectly resulting in an mbuf leak. The correct way to handle this is to free up mbufs on both BA and TX done notifications up to the last sent seqno. Tested by: osa@ MFC after: 3 days
2012-06-27 15:55:34 +00:00
wn = (void *)tap->txa_ni;
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
res = NULL;
ssn = 0;
if (!IEEE80211_AMPDU_RUNNING(tap)) {
res = tap->txa_private;
ssn = tap->txa_start & 0xfff;
}
Fix a TX aggregation issue, if after the last compressed BA notification the TX queue is empty, there won't be a TX done notification, effectly resulting in an mbuf leak. The correct way to handle this is to free up mbufs on both BA and TX done notifications up to the last sent seqno. Tested by: osa@ MFC after: 3 days
2012-06-27 15:55:34 +00:00
for (lastidx = le16toh(ba->ssn) & 0xff; txq->read != lastidx;) {
txdata = &txq->data[txq->read];
/* Unmap and free mbuf. */
bus_dmamap_sync(txq->data_dmat, txdata->map,
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(txq->data_dmat, txdata->map);
m = txdata->m, txdata->m = NULL;
ni = txdata->ni, txdata->ni = NULL;
KASSERT(ni != NULL, ("no node"));
KASSERT(m != NULL, ("no mbuf"));
(more) correctly account TX completion status for A-MPDU session frames. The rules turn out to be: * for non-aggregation session TX queues - it's either sent or not sent. * for aggregation session TX queues - if nframes=1, then the status reflects the completed transmission. * however, for nframes > 1, then this is just a status reflecting what the initial transmission did. The compressed BA (immediate or delayed) may not have yet been received, so the actual frame status is in the compressed BA updates. Whilst here, I fiddled with debugging and formatting a bit. There's also RTS attempts (what the atheros chips call "short retries") which weren't being logged and they aren't yet being used in the rate control statistics updates. For now, at least log them. TODO: * This still isn't 100% correct! So I have to tinker with this some more. (The failures aren't always failures..) * Extend the rate control API in net80211 so it can take both short and long retry counts. Tested: * Intel 5100, STA mode
2014-09-08 03:16:28 +00:00
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: freeing m=%p\n", __func__, m);
Migrate iwn(4) to use the new ieee80211_tx_complete() API. Tested: * Intel 5100, STA mode
2013-08-29 13:56:44 +00:00
ieee80211_tx_complete(ni, m, 1);
Fix a TX aggregation issue, if after the last compressed BA notification the TX queue is empty, there won't be a TX done notification, effectly resulting in an mbuf leak. The correct way to handle this is to free up mbufs on both BA and TX done notifications up to the last sent seqno. Tested by: osa@ MFC after: 3 days
2012-06-27 15:55:34 +00:00
txq->queued--;
txq->read = (txq->read + 1) % IWN_TX_RING_COUNT;
}
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
if (txq->queued == 0 && res != NULL) {
iwn_nic_lock(sc);
ops->ampdu_tx_stop(sc, qid, tid, ssn);
iwn_nic_unlock(sc);
sc->qid2tap[qid] = NULL;
free(res, M_DEVBUF);
return;
}
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
if (wn->agg[tid].bitmap == 0)
return;
shift = wn->agg[tid].startidx - ((le16toh(ba->seq) >> 4) & 0xff);
if (shift < 0)
shift += 0x100;
if (wn->agg[tid].nframes > (64 - shift))
return;
Inform the rate control code if a single frame AMPDU transmission succeeds but has some retries. Without this, single frame transmission in AMPDU will always look like it succeeded fine, and thus AMRR will think it's totally fine to just keep upping the rate upwards. Now, this is still not quite right! For multi-frame aggregates the completion happens in two parts - the TX done and the BA received. The driver is currently double accounting those a little - there's no way to say to the rate control code "I completed X frames, Y worked fine, there were Z retries." And it's a bit odd with iwn, as the firmware retransmits frames for us so we don't get to see how many retransmits happened; only that it took longer than normal. I may have to extend the rate control API to properly track that. So this may keep the rate lower than it should be, but that's better than keeping it higher than it should be. Tested: * 5100, STA mode
2014-08-28 07:44:59 +00:00
/*
(more) correctly account TX completion status for A-MPDU session frames. The rules turn out to be: * for non-aggregation session TX queues - it's either sent or not sent. * for aggregation session TX queues - if nframes=1, then the status reflects the completed transmission. * however, for nframes > 1, then this is just a status reflecting what the initial transmission did. The compressed BA (immediate or delayed) may not have yet been received, so the actual frame status is in the compressed BA updates. Whilst here, I fiddled with debugging and formatting a bit. There's also RTS attempts (what the atheros chips call "short retries") which weren't being logged and they aren't yet being used in the rate control statistics updates. For now, at least log them. TODO: * This still isn't 100% correct! So I have to tinker with this some more. (The failures aren't always failures..) * Extend the rate control API in net80211 so it can take both short and long retry counts. Tested: * Intel 5100, STA mode
2014-09-08 03:16:28 +00:00
* Walk the bitmap and calculate how many successful and failed
* attempts are made.
*
* Yes, the rate control code doesn't know these are A-MPDU
* subframes and that it's okay to fail some of these.
Inform the rate control code if a single frame AMPDU transmission succeeds but has some retries. Without this, single frame transmission in AMPDU will always look like it succeeded fine, and thus AMRR will think it's totally fine to just keep upping the rate upwards. Now, this is still not quite right! For multi-frame aggregates the completion happens in two parts - the TX done and the BA received. The driver is currently double accounting those a little - there's no way to say to the rate control code "I completed X frames, Y worked fine, there were Z retries." And it's a bit odd with iwn, as the firmware retransmits frames for us so we don't get to see how many retransmits happened; only that it took longer than normal. I may have to extend the rate control API to properly track that. So this may keep the rate lower than it should be, but that's better than keeping it higher than it should be. Tested: * 5100, STA mode
2014-08-28 07:44:59 +00:00
*/
Fix a TX aggregation issue, if after the last compressed BA notification the TX queue is empty, there won't be a TX done notification, effectly resulting in an mbuf leak. The correct way to handle this is to free up mbufs on both BA and TX done notifications up to the last sent seqno. Tested by: osa@ MFC after: 3 days
2012-06-27 15:55:34 +00:00
ni = tap->txa_ni;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
bitmap = (le64toh(ba->bitmap) >> shift) & wn->agg[tid].bitmap;
for (i = 0; bitmap; i++) {
if ((bitmap & 1) == 0) {
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
tx_err ++;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
ieee80211_ratectl_tx_complete(ni->ni_vap, ni,
IEEE80211_RATECTL_TX_FAILURE, &ackfailcnt, NULL);
} else {
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
tx_ok ++;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
ieee80211_ratectl_tx_complete(ni->ni_vap, ni,
IEEE80211_RATECTL_TX_SUCCESS, &ackfailcnt, NULL);
}
bitmap >>= 1;
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
(more) correctly account TX completion status for A-MPDU session frames. The rules turn out to be: * for non-aggregation session TX queues - it's either sent or not sent. * for aggregation session TX queues - if nframes=1, then the status reflects the completed transmission. * however, for nframes > 1, then this is just a status reflecting what the initial transmission did. The compressed BA (immediate or delayed) may not have yet been received, so the actual frame status is in the compressed BA updates. Whilst here, I fiddled with debugging and formatting a bit. There's also RTS attempts (what the atheros chips call "short retries") which weren't being logged and they aren't yet being used in the rate control statistics updates. For now, at least log them. TODO: * This still isn't 100% correct! So I have to tinker with this some more. (The failures aren't always failures..) * Extend the rate control API in net80211 so it can take both short and long retry counts. Tested: * Intel 5100, STA mode
2014-09-08 03:16:28 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE | IWN_DEBUG_XMIT,
"->%s: end; %d ok; %d err\n",__func__, tx_ok, tx_err);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
/*
* Process a CALIBRATION_RESULT notification sent by the initialization
* firmware on response to a CMD_CALIB_CONFIG command (5000 only).
*/
static void
iwn5000_rx_calib_results(struct iwn_softc *sc, struct iwn_rx_desc *desc,
struct iwn_rx_data *data)
{
struct iwn_phy_calib *calib = (struct iwn_phy_calib *)(desc + 1);
int len, idx = -1;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
/* Runtime firmware should not send such a notification. */
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
if (sc->sc_flags & IWN_FLAG_CALIB_DONE){
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s received after clib done\n",
__func__);
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
return;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
}
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
len = (le32toh(desc->len) & 0x3fff) - 4;
bus_dmamap_sync(sc->rxq.data_dmat, data->map, BUS_DMASYNC_POSTREAD);
switch (calib->code) {
case IWN5000_PHY_CALIB_DC:
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->calib_need & IWN_FLG_NEED_PHY_CALIB_DC)
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
idx = 0;
break;
case IWN5000_PHY_CALIB_LO:
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->calib_need & IWN_FLG_NEED_PHY_CALIB_LO)
idx = 1;
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
break;
case IWN5000_PHY_CALIB_TX_IQ:
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->calib_need & IWN_FLG_NEED_PHY_CALIB_TX_IQ)
idx = 2;
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
break;
case IWN5000_PHY_CALIB_TX_IQ_PERIODIC:
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->calib_need & IWN_FLG_NEED_PHY_CALIB_TX_IQ_PERIODIC)
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
idx = 3;
break;
case IWN5000_PHY_CALIB_BASE_BAND:
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->calib_need & IWN_FLG_NEED_PHY_CALIB_BASE_BAND)
idx = 4;
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
break;
}
if (idx == -1) /* Ignore other results. */
return;
/* Save calibration result. */
if (sc->calibcmd[idx].buf != NULL)
free(sc->calibcmd[idx].buf, M_DEVBUF);
sc->calibcmd[idx].buf = malloc(len, M_DEVBUF, M_NOWAIT);
if (sc->calibcmd[idx].buf == NULL) {
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"not enough memory for calibration result %d\n",
calib->code);
return;
}
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
"saving calibration result idx=%d, code=%d len=%d\n", idx, calib->code, len);
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
sc->calibcmd[idx].len = len;
memcpy(sc->calibcmd[idx].buf, calib, len);
}
Track and expose the latest statistics from the firmware. Tested: * Intel Centrino 6205
2014-02-24 02:37:04 +00:00
static void
iwn_stats_update(struct iwn_softc *sc, struct iwn_calib_state *calib,
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
struct iwn_stats *stats, int len)
Track and expose the latest statistics from the firmware. Tested: * Intel Centrino 6205
2014-02-24 02:37:04 +00:00
{
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
struct iwn_stats_bt *stats_bt;
struct iwn_stats *lstats;
/*
* First - check whether the length is the bluetooth or normal.
*
* If it's normal - just copy it and bump out.
* Otherwise we have to convert things.
*/
if (len == sizeof(struct iwn_stats) + 4) {
memcpy(&sc->last_stat, stats, sizeof(struct iwn_stats));
sc->last_stat_valid = 1;
return;
}
/*
* If it's not the bluetooth size - log, then just copy.
*/
if (len != sizeof(struct iwn_stats_bt) + 4) {
DPRINTF(sc, IWN_DEBUG_STATS,
"%s: size of rx statistics (%d) not an expected size!\n",
__func__,
len);
memcpy(&sc->last_stat, stats, sizeof(struct iwn_stats));
sc->last_stat_valid = 1;
return;
}
Track and expose the latest statistics from the firmware. Tested: * Intel Centrino 6205
2014-02-24 02:37:04 +00:00
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
/*
* Ok. Time to copy.
*/
stats_bt = (struct iwn_stats_bt *) stats;
lstats = &sc->last_stat;
/* flags */
lstats->flags = stats_bt->flags;
/* rx_bt */
memcpy(&lstats->rx.ofdm, &stats_bt->rx_bt.ofdm,
sizeof(struct iwn_rx_phy_stats));
memcpy(&lstats->rx.cck, &stats_bt->rx_bt.cck,
sizeof(struct iwn_rx_phy_stats));
memcpy(&lstats->rx.general, &stats_bt->rx_bt.general_bt.common,
sizeof(struct iwn_rx_general_stats));
memcpy(&lstats->rx.ht, &stats_bt->rx_bt.ht,
sizeof(struct iwn_rx_ht_phy_stats));
/* tx */
memcpy(&lstats->tx, &stats_bt->tx,
sizeof(struct iwn_tx_stats));
/* general */
memcpy(&lstats->general, &stats_bt->general,
sizeof(struct iwn_general_stats));
/* XXX TODO: Squirrel away the extra bluetooth stats somewhere */
Track and expose the latest statistics from the firmware. Tested: * Intel Centrino 6205
2014-02-24 02:37:04 +00:00
sc->last_stat_valid = 1;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Process an RX_STATISTICS or BEACON_STATISTICS firmware notification.
* The latter is sent by the firmware after each received beacon.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_rx_statistics(struct iwn_softc *sc, struct iwn_rx_desc *desc,
struct iwn_rx_data *data)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
struct iwn_calib_state *calib = &sc->calib;
struct iwn_stats *stats = (struct iwn_stats *)(desc + 1);
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
struct iwn_stats *lstats;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
int temp;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Ignore statistics received during a scan. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (vap->iv_state != IEEE80211_S_RUN ||
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
(ic->ic_flags & IEEE80211_F_SCAN)){
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s received during calib\n",
__func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
o add more notification strings in iwn_intr_str() o sync with OpenBSD code Submitted by: Bernhard Schmidt <bschmidt at techwires.net> MFC after: 3 days
2010-01-25 19:36:48 +00:00
bus_dmamap_sync(sc->rxq.data_dmat, data->map, BUS_DMASYNC_POSTREAD);
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE | IWN_DEBUG_STATS,
"%s: received statistics, cmd %d, len %d\n",
__func__, desc->type, le16toh(desc->len));
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
sc->calib_cnt = 0; /* Reset TX power calibration timeout. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
/*
* Collect/track general statistics for reporting.
*
* This takes care of ensuring that the bluetooth sized message
* will be correctly converted to the legacy sized message.
*/
iwn_stats_update(sc, calib, stats, le16toh(desc->len));
/*
* And now, let's take a reference of it to use!
*/
lstats = &sc->last_stat;
Track and expose the latest statistics from the firmware. Tested: * Intel Centrino 6205
2014-02-24 02:37:04 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Test if temperature has changed. */
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
if (lstats->general.temp != sc->rawtemp) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Convert "raw" temperature to degC. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
sc->rawtemp = stats->general.temp;
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
temp = ops->get_temperature(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE, "%s: temperature %d\n",
__func__, temp);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Update TX power if need be (4965AGN only). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (sc->hw_type == IWN_HW_REV_TYPE_4965)
iwn4965_power_calibration(sc, temp);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
if (desc->type != IWN_BEACON_STATISTICS)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return; /* Reply to a statistics request. */
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
sc->noise = iwn_get_noise(&lstats->rx.general);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE, "%s: noise %d\n", __func__, sc->noise);
/* Test that RSSI and noise are present in stats report. */
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
if (le32toh(lstats->rx.general.flags) != 1) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DPRINTF(sc, IWN_DEBUG_ANY, "%s\n",
"received statistics without RSSI");
return;
}
if (calib->state == IWN_CALIB_STATE_ASSOC)
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
iwn_collect_noise(sc, &lstats->rx.general);
Begin fleshing out some code to handle tracking PLCP error rates in preparation for the scan based retune logic. The linux iwlwifi driver does a rescan (onto a non-active channel) to force an RF retune when the PLCP error rates exceed a certain threshold. * Add code to track HT PLCP rate errors; * Separate out the PLCP error count fetch and update so the delta can be used when checking for PLCP error rates; * Implement the PLCP error logic from iwlwifi; * For now, just print out whenever the error rate exceeds the threshold. The actual scan based retune will take a bit more effort; the scan command code right now assumes that a scan state is passed in. This does need to change to be more flexible (both for this and in preparation for scanning multiple channels at once.) Tested: * 5100 (STA mode) * 2200 (STA mode) * 2230 (STA mode)
2013-12-07 08:03:10 +00:00
else if (calib->state == IWN_CALIB_STATE_RUN) {
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
iwn_tune_sensitivity(sc, &lstats->rx);
Begin fleshing out some code to handle tracking PLCP error rates in preparation for the scan based retune logic. The linux iwlwifi driver does a rescan (onto a non-active channel) to force an RF retune when the PLCP error rates exceed a certain threshold. * Add code to track HT PLCP rate errors; * Separate out the PLCP error count fetch and update so the delta can be used when checking for PLCP error rates; * Implement the PLCP error logic from iwlwifi; * For now, just print out whenever the error rate exceeds the threshold. The actual scan based retune will take a bit more effort; the scan command code right now assumes that a scan state is passed in. This does need to change to be more flexible (both for this and in preparation for scanning multiple channels at once.) Tested: * 5100 (STA mode) * 2200 (STA mode) * 2230 (STA mode)
2013-12-07 08:03:10 +00:00
/*
* XXX TODO: Only run the RX recovery if we're associated!
*/
Handle the newer-style bluetooth message format from the (at least) Centrino 2230 firmware. This fixes the general statistics block to be actually valid. I've verified this by contrasting the output of iwnstats before and after the change. The general block is now correct. Tested: * Intel 5100 (old format stats message) * Intel 2230 (new format stats message)
2014-03-04 08:01:56 +00:00
iwn_check_rx_recovery(sc, lstats);
iwn_save_stats_counters(sc, lstats);
Begin fleshing out some code to handle tracking PLCP error rates in preparation for the scan based retune logic. The linux iwlwifi driver does a rescan (onto a non-active channel) to force an RF retune when the PLCP error rates exceed a certain threshold. * Add code to track HT PLCP rate errors; * Separate out the PLCP error count fetch and update so the delta can be used when checking for PLCP error rates; * Implement the PLCP error logic from iwlwifi; * For now, just print out whenever the error rate exceeds the threshold. The actual scan based retune will take a bit more effort; the scan command code right now assumes that a scan state is passed in. This does need to change to be more flexible (both for this and in preparation for scanning multiple channels at once.) Tested: * 5100 (STA mode) * 2200 (STA mode) * 2230 (STA mode)
2013-12-07 08:03:10 +00:00
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Begin fleshing out some code to handle tracking PLCP error rates in preparation for the scan based retune logic. The linux iwlwifi driver does a rescan (onto a non-active channel) to force an RF retune when the PLCP error rates exceed a certain threshold. * Add code to track HT PLCP rate errors; * Separate out the PLCP error count fetch and update so the delta can be used when checking for PLCP error rates; * Implement the PLCP error logic from iwlwifi; * For now, just print out whenever the error rate exceeds the threshold. The actual scan based retune will take a bit more effort; the scan command code right now assumes that a scan state is passed in. This does need to change to be more flexible (both for this and in preparation for scanning multiple channels at once.) Tested: * 5100 (STA mode) * 2200 (STA mode) * 2230 (STA mode)
2013-12-07 08:03:10 +00:00
/*
* Save the relevant statistic counters for the next calibration
* pass.
*/
static void
iwn_save_stats_counters(struct iwn_softc *sc, const struct iwn_stats *rs)
{
struct iwn_calib_state *calib = &sc->calib;
/* Save counters values for next call. */
calib->bad_plcp_cck = le32toh(rs->rx.cck.bad_plcp);
calib->fa_cck = le32toh(rs->rx.cck.fa);
calib->bad_plcp_ht = le32toh(rs->rx.ht.bad_plcp);
calib->bad_plcp_ofdm = le32toh(rs->rx.ofdm.bad_plcp);
calib->fa_ofdm = le32toh(rs->rx.ofdm.fa);
/* Last time we received these tick values */
sc->last_calib_ticks = ticks;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Process a TX_DONE firmware notification. Unfortunately, the 4965AGN
* and 5000 adapters have different incompatible TX status formats.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_tx_done(struct iwn_softc *sc, struct iwn_rx_desc *desc,
struct iwn_rx_data *data)
{
struct iwn4965_tx_stat *stat = (struct iwn4965_tx_stat *)(desc + 1);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
struct iwn_tx_ring *ring;
int qid;
qid = desc->qid & 0xf;
ring = &sc->txq[qid];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: "
(more) correctly account TX completion status for A-MPDU session frames. The rules turn out to be: * for non-aggregation session TX queues - it's either sent or not sent. * for aggregation session TX queues - if nframes=1, then the status reflects the completed transmission. * however, for nframes > 1, then this is just a status reflecting what the initial transmission did. The compressed BA (immediate or delayed) may not have yet been received, so the actual frame status is in the compressed BA updates. Whilst here, I fiddled with debugging and formatting a bit. There's also RTS attempts (what the atheros chips call "short retries") which weren't being logged and they aren't yet being used in the rate control statistics updates. For now, at least log them. TODO: * This still isn't 100% correct! So I have to tinker with this some more. (The failures aren't always failures..) * Extend the rate control API in net80211 so it can take both short and long retry counts. Tested: * Intel 5100, STA mode
2014-09-08 03:16:28 +00:00
"qid %d idx %d RTS retries %d ACK retries %d nkill %d rate %x duration %d status %x\n",
__func__, desc->qid, desc->idx,
stat->rtsfailcnt,
stat->ackfailcnt,
stat->btkillcnt,
stat->rate, le16toh(stat->duration),
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
le32toh(stat->status));
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
Use correct bus_dma_tag_t for TX frames. Reported by: Andreas Nilsson <andrnils at gmail.com> Approved by: rpaulo (mentor) MFC after: 3 days
2010-04-21 17:38:16 +00:00
bus_dmamap_sync(ring->data_dmat, data->map, BUS_DMASYNC_POSTREAD);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
if (qid >= sc->firstaggqueue) {
iwn_ampdu_tx_done(sc, qid, desc->idx, stat->nframes,
Inform the rate control code if a single frame AMPDU transmission succeeds but has some retries. Without this, single frame transmission in AMPDU will always look like it succeeded fine, and thus AMRR will think it's totally fine to just keep upping the rate upwards. Now, this is still not quite right! For multi-frame aggregates the completion happens in two parts - the TX done and the BA received. The driver is currently double accounting those a little - there's no way to say to the rate control code "I completed X frames, Y worked fine, there were Z retries." And it's a bit odd with iwn, as the firmware retransmits frames for us so we don't get to see how many retransmits happened; only that it took longer than normal. I may have to extend the rate control API to properly track that. So this may keep the rate lower than it should be, but that's better than keeping it higher than it should be. Tested: * 5100, STA mode
2014-08-28 07:44:59 +00:00
stat->ackfailcnt, &stat->status);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
} else {
iwn_tx_done(sc, desc, stat->ackfailcnt,
le32toh(stat->status) & 0xff);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_tx_done(struct iwn_softc *sc, struct iwn_rx_desc *desc,
struct iwn_rx_data *data)
{
struct iwn5000_tx_stat *stat = (struct iwn5000_tx_stat *)(desc + 1);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
struct iwn_tx_ring *ring;
int qid;
qid = desc->qid & 0xf;
ring = &sc->txq[qid];
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: "
(more) correctly account TX completion status for A-MPDU session frames. The rules turn out to be: * for non-aggregation session TX queues - it's either sent or not sent. * for aggregation session TX queues - if nframes=1, then the status reflects the completed transmission. * however, for nframes > 1, then this is just a status reflecting what the initial transmission did. The compressed BA (immediate or delayed) may not have yet been received, so the actual frame status is in the compressed BA updates. Whilst here, I fiddled with debugging and formatting a bit. There's also RTS attempts (what the atheros chips call "short retries") which weren't being logged and they aren't yet being used in the rate control statistics updates. For now, at least log them. TODO: * This still isn't 100% correct! So I have to tinker with this some more. (The failures aren't always failures..) * Extend the rate control API in net80211 so it can take both short and long retry counts. Tested: * Intel 5100, STA mode
2014-09-08 03:16:28 +00:00
"qid %d idx %d RTS retries %d ACK retries %d nkill %d rate %x duration %d status %x\n",
__func__, desc->qid, desc->idx,
stat->rtsfailcnt,
stat->ackfailcnt,
stat->btkillcnt,
stat->rate, le16toh(stat->duration),
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
le32toh(stat->status));
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
#ifdef notyet
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Reset TX scheduler slot. */
iwn5000_reset_sched(sc, desc->qid & 0xf, desc->idx);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
#endif
o add more notification strings in iwn_intr_str() o sync with OpenBSD code Submitted by: Bernhard Schmidt <bschmidt at techwires.net> MFC after: 3 days
2010-01-25 19:36:48 +00:00
Use correct bus_dma_tag_t for TX frames. Reported by: Andreas Nilsson <andrnils at gmail.com> Approved by: rpaulo (mentor) MFC after: 3 days
2010-04-21 17:38:16 +00:00
bus_dmamap_sync(ring->data_dmat, data->map, BUS_DMASYNC_POSTREAD);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
if (qid >= sc->firstaggqueue) {
iwn_ampdu_tx_done(sc, qid, desc->idx, stat->nframes,
Inform the rate control code if a single frame AMPDU transmission succeeds but has some retries. Without this, single frame transmission in AMPDU will always look like it succeeded fine, and thus AMRR will think it's totally fine to just keep upping the rate upwards. Now, this is still not quite right! For multi-frame aggregates the completion happens in two parts - the TX done and the BA received. The driver is currently double accounting those a little - there's no way to say to the rate control code "I completed X frames, Y worked fine, there were Z retries." And it's a bit odd with iwn, as the firmware retransmits frames for us so we don't get to see how many retransmits happened; only that it took longer than normal. I may have to extend the rate control API to properly track that. So this may keep the rate lower than it should be, but that's better than keeping it higher than it should be. Tested: * 5100, STA mode
2014-08-28 07:44:59 +00:00
stat->ackfailcnt, &stat->status);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
} else {
iwn_tx_done(sc, desc, stat->ackfailcnt,
le16toh(stat->status) & 0xff);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Adapter-independent backend for TX_DONE firmware notifications.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_tx_done(struct iwn_softc *sc, struct iwn_rx_desc *desc, int ackfailcnt,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint8_t status)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
struct iwn_tx_ring *ring = &sc->txq[desc->qid & 0xf];
struct iwn_tx_data *data = &ring->data[desc->idx];
struct mbuf *m;
struct ieee80211_node *ni;
net80211 rate control framework (net80211 ratectl). This framework allows drivers to abstract the rate control algorithm and just feed the framework with the usable parameters. The rate control framework will now deal with passing the parameters to the selected algorithm. Right now we have AMRR (the default) and RSSADAPT but there's no way to select one with ifconfig, yet. The objective is to have more rate control algorithms in the net80211 stack so all drivers[0] can use it. Ideally, we'll have the well-known sample rate control algorithm in the net80211 at some point so all drivers can use it (not just ath). [0] all drivers that do rate control in software, that is. Reviewed by: bschmidt, thompsa, weyongo MFC after: 1 months
2010-04-07 15:29:13 +00:00
struct ieee80211vap *vap;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
KASSERT(data->ni != NULL, ("no node"));
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Unmap and free mbuf. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(ring->data_dmat, data->map, BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(ring->data_dmat, data->map);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
m = data->m, data->m = NULL;
ni = data->ni, data->ni = NULL;
net80211 rate control framework (net80211 ratectl). This framework allows drivers to abstract the rate control algorithm and just feed the framework with the usable parameters. The rate control framework will now deal with passing the parameters to the selected algorithm. Right now we have AMRR (the default) and RSSADAPT but there's no way to select one with ifconfig, yet. The objective is to have more rate control algorithms in the net80211 stack so all drivers[0] can use it. Ideally, we'll have the well-known sample rate control algorithm in the net80211 at some point so all drivers can use it (not just ath). [0] all drivers that do rate control in software, that is. Reviewed by: bschmidt, thompsa, weyongo MFC after: 1 months
2010-04-07 15:29:13 +00:00
vap = ni->ni_vap;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/*
* Update rate control statistics for the node.
*/
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
if (status & IWN_TX_FAIL)
net80211 rate control framework (net80211 ratectl). This framework allows drivers to abstract the rate control algorithm and just feed the framework with the usable parameters. The rate control framework will now deal with passing the parameters to the selected algorithm. Right now we have AMRR (the default) and RSSADAPT but there's no way to select one with ifconfig, yet. The objective is to have more rate control algorithms in the net80211 stack so all drivers[0] can use it. Ideally, we'll have the well-known sample rate control algorithm in the net80211 at some point so all drivers can use it (not just ath). [0] all drivers that do rate control in software, that is. Reviewed by: bschmidt, thompsa, weyongo MFC after: 1 months
2010-04-07 15:29:13 +00:00
ieee80211_ratectl_tx_complete(vap, ni,
IEEE80211_RATECTL_TX_FAILURE, &ackfailcnt, NULL);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
else
net80211 rate control framework (net80211 ratectl). This framework allows drivers to abstract the rate control algorithm and just feed the framework with the usable parameters. The rate control framework will now deal with passing the parameters to the selected algorithm. Right now we have AMRR (the default) and RSSADAPT but there's no way to select one with ifconfig, yet. The objective is to have more rate control algorithms in the net80211 stack so all drivers[0] can use it. Ideally, we'll have the well-known sample rate control algorithm in the net80211 at some point so all drivers can use it (not just ath). [0] all drivers that do rate control in software, that is. Reviewed by: bschmidt, thompsa, weyongo MFC after: 1 months
2010-04-07 15:29:13 +00:00
ieee80211_ratectl_tx_complete(vap, ni,
IEEE80211_RATECTL_TX_SUCCESS, &ackfailcnt, NULL);
Migrate iwn(4) to use the new ieee80211_tx_complete() API. Tested: * Intel 5100, STA mode
2013-08-29 13:56:44 +00:00
/*
* Channels marked for "radar" require traffic to be received
* to unlock before we can transmit. Until traffic is seen
* any attempt to transmit is returned immediately with status
* set to IWN_TX_FAIL_TX_LOCKED. Unfortunately this can easily
* happen on first authenticate after scanning. To workaround
* this we ignore a failure of this sort in AUTH state so the
* 802.11 layer will fall back to using a timeout to wait for
* the AUTH reply. This allows the firmware time to see
* traffic so a subsequent retry of AUTH succeeds. It's
* unclear why the firmware does not maintain state for
* channels recently visited as this would allow immediate
* use of the channel after a scan (where we see traffic).
*/
if (status == IWN_TX_FAIL_TX_LOCKED &&
ni->ni_vap->iv_state == IEEE80211_S_AUTH)
ieee80211_tx_complete(ni, m, 0);
else
ieee80211_tx_complete(ni, m,
(status & IWN_TX_FAIL) != 0);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
sc->sc_tx_timer = 0;
Remove the software queue, which is a remnant of ifnet ifqueue. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-09-02 15:23:51 +00:00
if (--ring->queued < IWN_TX_RING_LOMARK)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
sc->qfullmsk &= ~(1 << ring->qid);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Process a "command done" firmware notification. This is where we wakeup
* processes waiting for a synchronous command completion.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_cmd_done(struct iwn_softc *sc, struct iwn_rx_desc *desc)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
struct iwn_tx_ring *ring;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_tx_data *data;
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
int cmd_queue_num;
if (sc->sc_flags & IWN_FLAG_PAN_SUPPORT)
cmd_queue_num = IWN_PAN_CMD_QUEUE;
else
cmd_queue_num = IWN_CMD_QUEUE_NUM;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if ((desc->qid & IWN_RX_DESC_QID_MSK) != cmd_queue_num)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return; /* Not a command ack. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
ring = &sc->txq[cmd_queue_num];
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
data = &ring->data[desc->idx];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* If the command was mapped in an mbuf, free it. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (data->m != NULL) {
Add missing bus_dmamap_sync calls as well as remove two duplicate ones. Obtained from: OpenBSD
2011-04-16 10:52:11 +00:00
bus_dmamap_sync(ring->data_dmat, data->map,
BUS_DMASYNC_POSTWRITE);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_unload(ring->data_dmat, data->map);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
m_freem(data->m);
data->m = NULL;
}
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
wakeup(&ring->desc[desc->idx]);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
static void
iwn_ampdu_tx_done(struct iwn_softc *sc, int qid, int idx, int nframes,
Inform the rate control code if a single frame AMPDU transmission succeeds but has some retries. Without this, single frame transmission in AMPDU will always look like it succeeded fine, and thus AMRR will think it's totally fine to just keep upping the rate upwards. Now, this is still not quite right! For multi-frame aggregates the completion happens in two parts - the TX done and the BA received. The driver is currently double accounting those a little - there's no way to say to the rate control code "I completed X frames, Y worked fine, there were Z retries." And it's a bit odd with iwn, as the firmware retransmits frames for us so we don't get to see how many retransmits happened; only that it took longer than normal. I may have to extend the rate control API to properly track that. So this may keep the rate lower than it should be, but that's better than keeping it higher than it should be. Tested: * 5100, STA mode
2014-08-28 07:44:59 +00:00
int ackfailcnt, void *stat)
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
{
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
struct iwn_ops *ops = &sc->ops;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
struct iwn_tx_ring *ring = &sc->txq[qid];
struct iwn_tx_data *data;
struct mbuf *m;
struct iwn_node *wn;
struct ieee80211_node *ni;
struct ieee80211_tx_ampdu *tap;
uint64_t bitmap;
uint32_t *status = stat;
uint16_t *aggstatus = stat;
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
uint16_t ssn;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
uint8_t tid;
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
int bit, i, lastidx, *res, seqno, shift, start;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
(more) correctly account TX completion status for A-MPDU session frames. The rules turn out to be: * for non-aggregation session TX queues - it's either sent or not sent. * for aggregation session TX queues - if nframes=1, then the status reflects the completed transmission. * however, for nframes > 1, then this is just a status reflecting what the initial transmission did. The compressed BA (immediate or delayed) may not have yet been received, so the actual frame status is in the compressed BA updates. Whilst here, I fiddled with debugging and formatting a bit. There's also RTS attempts (what the atheros chips call "short retries") which weren't being logged and they aren't yet being used in the rate control statistics updates. For now, at least log them. TODO: * This still isn't 100% correct! So I have to tinker with this some more. (The failures aren't always failures..) * Extend the rate control API in net80211 so it can take both short and long retry counts. Tested: * Intel 5100, STA mode
2014-09-08 03:16:28 +00:00
/* XXX TODO: status is le16 field! Grr */
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Inform the rate control code if a single frame AMPDU transmission succeeds but has some retries. Without this, single frame transmission in AMPDU will always look like it succeeded fine, and thus AMRR will think it's totally fine to just keep upping the rate upwards. Now, this is still not quite right! For multi-frame aggregates the completion happens in two parts - the TX done and the BA received. The driver is currently double accounting those a little - there's no way to say to the rate control code "I completed X frames, Y worked fine, there were Z retries." And it's a bit odd with iwn, as the firmware retransmits frames for us so we don't get to see how many retransmits happened; only that it took longer than normal. I may have to extend the rate control API to properly track that. So this may keep the rate lower than it should be, but that's better than keeping it higher than it should be. Tested: * 5100, STA mode
2014-08-28 07:44:59 +00:00
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: nframes=%d, status=0x%08x\n",
__func__,
nframes,
*status);
tap = sc->qid2tap[qid];
tid = tap->txa_tid;
wn = (void *)tap->txa_ni;
ni = tap->txa_ni;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
(more) correctly account TX completion status for A-MPDU session frames. The rules turn out to be: * for non-aggregation session TX queues - it's either sent or not sent. * for aggregation session TX queues - if nframes=1, then the status reflects the completed transmission. * however, for nframes > 1, then this is just a status reflecting what the initial transmission did. The compressed BA (immediate or delayed) may not have yet been received, so the actual frame status is in the compressed BA updates. Whilst here, I fiddled with debugging and formatting a bit. There's also RTS attempts (what the atheros chips call "short retries") which weren't being logged and they aren't yet being used in the rate control statistics updates. For now, at least log them. TODO: * This still isn't 100% correct! So I have to tinker with this some more. (The failures aren't always failures..) * Extend the rate control API in net80211 so it can take both short and long retry counts. Tested: * Intel 5100, STA mode
2014-09-08 03:16:28 +00:00
/*
* XXX TODO: ACK and RTS failures would be nice here!
*/
/*
* A-MPDU single frame status - if we failed to transmit it
* in A-MPDU, then it may be a permanent failure.
*
* XXX TODO: check what the Linux iwlwifi driver does here;
* there's some permanent and temporary failures that may be
* handled differently.
*/
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
if (nframes == 1) {
If A-MPDU transmission fails entirely, then no BA is received from the NIC and pushed up to the driver. Unfortunately this means there's no rate control notification done. Thus, if the rate control code makes a decision that hits a crappy rate that can't succeed, the rate code would never lower the rate and packet loss would continue. So, fake some rate control notification in this case.
2013-11-11 09:08:22 +00:00
if ((*status & 0xff) != 1 && (*status & 0xff) != 2) {
#ifdef NOT_YET
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
printf("ieee80211_send_bar()\n");
#endif
If A-MPDU transmission fails entirely, then no BA is received from the NIC and pushed up to the driver. Unfortunately this means there's no rate control notification done. Thus, if the rate control code makes a decision that hits a crappy rate that can't succeed, the rate code would never lower the rate and packet loss would continue. So, fake some rate control notification in this case.
2013-11-11 09:08:22 +00:00
/*
* If we completely fail a transmit, make sure a
* notification is pushed up to the rate control
* layer.
*/
(more) correctly account TX completion status for A-MPDU session frames. The rules turn out to be: * for non-aggregation session TX queues - it's either sent or not sent. * for aggregation session TX queues - if nframes=1, then the status reflects the completed transmission. * however, for nframes > 1, then this is just a status reflecting what the initial transmission did. The compressed BA (immediate or delayed) may not have yet been received, so the actual frame status is in the compressed BA updates. Whilst here, I fiddled with debugging and formatting a bit. There's also RTS attempts (what the atheros chips call "short retries") which weren't being logged and they aren't yet being used in the rate control statistics updates. For now, at least log them. TODO: * This still isn't 100% correct! So I have to tinker with this some more. (The failures aren't always failures..) * Extend the rate control API in net80211 so it can take both short and long retry counts. Tested: * Intel 5100, STA mode
2014-09-08 03:16:28 +00:00
ieee80211_ratectl_tx_complete(ni->ni_vap,
ni,
IEEE80211_RATECTL_TX_FAILURE,
&ackfailcnt,
NULL);
} else {
/*
* If nframes=1, then we won't be getting a BA for
* this frame. Ensure that we correctly update the
* rate control code with how many retries were
* needed to send it.
*/
ieee80211_ratectl_tx_complete(ni->ni_vap,
ni,
IEEE80211_RATECTL_TX_SUCCESS,
&ackfailcnt,
NULL);
If A-MPDU transmission fails entirely, then no BA is received from the NIC and pushed up to the driver. Unfortunately this means there's no rate control notification done. Thus, if the rate control code makes a decision that hits a crappy rate that can't succeed, the rate code would never lower the rate and packet loss would continue. So, fake some rate control notification in this case.
2013-11-11 09:08:22 +00:00
}
}
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
bitmap = 0;
start = idx;
for (i = 0; i < nframes; i++) {
if (le16toh(aggstatus[i * 2]) & 0xc)
continue;
idx = le16toh(aggstatus[2*i + 1]) & 0xff;
bit = idx - start;
shift = 0;
if (bit >= 64) {
shift = 0x100 - idx + start;
bit = 0;
start = idx;
} else if (bit <= -64)
bit = 0x100 - start + idx;
else if (bit < 0) {
shift = start - idx;
start = idx;
bit = 0;
}
bitmap = bitmap << shift;
bitmap |= 1ULL << bit;
}
tap = sc->qid2tap[qid];
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
tid = tap->txa_tid;
wn = (void *)tap->txa_ni;
wn->agg[tid].bitmap = bitmap;
wn->agg[tid].startidx = start;
wn->agg[tid].nframes = nframes;
res = NULL;
ssn = 0;
if (!IEEE80211_AMPDU_RUNNING(tap)) {
res = tap->txa_private;
ssn = tap->txa_start & 0xfff;
On state changes from RUN to anything else the AGGR sessions are cleared/dropped leading to qid2tap[n] being NULL as there no longer is a tap. Now, if there have been lots of frames queued the firmware processes and returns those after the tap is gone. Tested by: osa MFC after: 1 week
2012-01-27 17:39:02 +00:00
}
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
(more) correctly account TX completion status for A-MPDU session frames. The rules turn out to be: * for non-aggregation session TX queues - it's either sent or not sent. * for aggregation session TX queues - if nframes=1, then the status reflects the completed transmission. * however, for nframes > 1, then this is just a status reflecting what the initial transmission did. The compressed BA (immediate or delayed) may not have yet been received, so the actual frame status is in the compressed BA updates. Whilst here, I fiddled with debugging and formatting a bit. There's also RTS attempts (what the atheros chips call "short retries") which weren't being logged and they aren't yet being used in the rate control statistics updates. For now, at least log them. TODO: * This still isn't 100% correct! So I have to tinker with this some more. (The failures aren't always failures..) * Extend the rate control API in net80211 so it can take both short and long retry counts. Tested: * Intel 5100, STA mode
2014-09-08 03:16:28 +00:00
/* This is going nframes DWORDS into the descriptor? */
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
seqno = le32toh(*(status + nframes)) & 0xfff;
for (lastidx = (seqno & 0xff); ring->read != lastidx;) {
data = &ring->data[ring->read];
/* Unmap and free mbuf. */
bus_dmamap_sync(ring->data_dmat, data->map,
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(ring->data_dmat, data->map);
m = data->m, data->m = NULL;
ni = data->ni, data->ni = NULL;
Fix a TX aggregation issue, if after the last compressed BA notification the TX queue is empty, there won't be a TX done notification, effectly resulting in an mbuf leak. The correct way to handle this is to free up mbufs on both BA and TX done notifications up to the last sent seqno. Tested by: osa@ MFC after: 3 days
2012-06-27 15:55:34 +00:00
KASSERT(ni != NULL, ("no node"));
KASSERT(m != NULL, ("no mbuf"));
(more) correctly account TX completion status for A-MPDU session frames. The rules turn out to be: * for non-aggregation session TX queues - it's either sent or not sent. * for aggregation session TX queues - if nframes=1, then the status reflects the completed transmission. * however, for nframes > 1, then this is just a status reflecting what the initial transmission did. The compressed BA (immediate or delayed) may not have yet been received, so the actual frame status is in the compressed BA updates. Whilst here, I fiddled with debugging and formatting a bit. There's also RTS attempts (what the atheros chips call "short retries") which weren't being logged and they aren't yet being used in the rate control statistics updates. For now, at least log them. TODO: * This still isn't 100% correct! So I have to tinker with this some more. (The failures aren't always failures..) * Extend the rate control API in net80211 so it can take both short and long retry counts. Tested: * Intel 5100, STA mode
2014-09-08 03:16:28 +00:00
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: freeing m=%p\n", __func__, m);
Migrate iwn(4) to use the new ieee80211_tx_complete() API. Tested: * Intel 5100, STA mode
2013-08-29 13:56:44 +00:00
ieee80211_tx_complete(ni, m, 1);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
ring->queued--;
ring->read = (ring->read + 1) % IWN_TX_RING_COUNT;
}
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
if (ring->queued == 0 && res != NULL) {
iwn_nic_lock(sc);
ops->ampdu_tx_stop(sc, qid, tid, ssn);
iwn_nic_unlock(sc);
sc->qid2tap[qid] = NULL;
free(res, M_DEVBUF);
return;
}
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
sc->sc_tx_timer = 0;
Remove the software queue, which is a remnant of ifnet ifqueue. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-09-02 15:23:51 +00:00
if (ring->queued < IWN_TX_RING_LOMARK)
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
sc->qfullmsk &= ~(1 << ring->qid);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Process an INT_FH_RX or INT_SW_RX interrupt.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_notif_intr(struct iwn_softc *sc)
{
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
uint16_t hw;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
bus_dmamap_sync(sc->rxq.stat_dma.tag, sc->rxq.stat_dma.map,
BUS_DMASYNC_POSTREAD);
hw = le16toh(sc->rxq.stat->closed_count) & 0xfff;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
while (sc->rxq.cur != hw) {
struct iwn_rx_data *data = &sc->rxq.data[sc->rxq.cur];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_rx_desc *desc;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(sc->rxq.data_dmat, data->map,
BUS_DMASYNC_POSTREAD);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
desc = mtod(data->m, struct iwn_rx_desc *);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_RECV,
Log the rx ring offset as part of the debug message.
2013-12-02 03:49:33 +00:00
"%s: cur=%d; qid %x idx %d flags %x type %d(%s) len %d\n",
__func__, sc->rxq.cur, desc->qid & 0xf, desc->idx, desc->flags,
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
desc->type, iwn_intr_str(desc->type),
le16toh(desc->len));
Replace the hard-coded RX queue value check with IWN_UNSOLICITED_RX_NOTIF.
2013-11-11 08:56:40 +00:00
if (!(desc->qid & IWN_UNSOLICITED_RX_NOTIF)) /* Reply to a command. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_cmd_done(sc, desc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
switch (desc->type) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
case IWN_RX_PHY:
iwn_rx_phy(sc, desc, data);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
break;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
case IWN_RX_DONE: /* 4965AGN only. */
case IWN_MPDU_RX_DONE:
/* An 802.11 frame has been received. */
iwn_rx_done(sc, desc, data);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
break;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
case IWN_RX_COMPRESSED_BA:
/* A Compressed BlockAck has been received. */
iwn_rx_compressed_ba(sc, desc, data);
break;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
case IWN_TX_DONE:
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* An 802.11 frame has been transmitted. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
ops->tx_done(sc, desc, data);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
break;
case IWN_RX_STATISTICS:
case IWN_BEACON_STATISTICS:
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_rx_statistics(sc, desc, data);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
break;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
case IWN_BEACON_MISSED:
{
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_beacon_missed *miss =
(struct iwn_beacon_missed *)(desc + 1);
o add more notification strings in iwn_intr_str() o sync with OpenBSD code Submitted by: Bernhard Schmidt <bschmidt at techwires.net> MFC after: 3 days
2010-01-25 19:36:48 +00:00
int misses;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(sc->rxq.data_dmat, data->map,
BUS_DMASYNC_POSTREAD);
o add more notification strings in iwn_intr_str() o sync with OpenBSD code Submitted by: Bernhard Schmidt <bschmidt at techwires.net> MFC after: 3 days
2010-01-25 19:36:48 +00:00
misses = le32toh(miss->consecutive);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_STATE,
"%s: beacons missed %d/%d\n", __func__,
misses, le32toh(miss->total));
/*
* If more than 5 consecutive beacons are missed,
* reinitialize the sensitivity state machine.
*/
Only handle beacon misses while in RUN state and not scanning.
2011-04-15 16:55:45 +00:00
if (vap->iv_state == IEEE80211_S_RUN &&
Fix bmiss notifications, events should be sent when NOT scanning. MFC after: 3 days
2011-10-13 18:17:03 +00:00
(ic->ic_flags & IEEE80211_F_SCAN) == 0) {
Only handle beacon misses while in RUN state and not scanning.
2011-04-15 16:55:45 +00:00
if (misses > 5)
(void)iwn_init_sensitivity(sc);
if (misses >= vap->iv_bmissthreshold) {
IWN_UNLOCK(sc);
ieee80211_beacon_miss(ic);
IWN_LOCK(sc);
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
break;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
case IWN_UC_READY:
{
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_ucode_info *uc =
(struct iwn_ucode_info *)(desc + 1);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* The microcontroller is ready. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(sc->rxq.data_dmat, data->map,
BUS_DMASYNC_POSTREAD);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_RESET,
"microcode alive notification version=%d.%d "
"subtype=%x alive=%x\n", uc->major, uc->minor,
uc->subtype, le32toh(uc->valid));
if (le32toh(uc->valid) != 1) {
device_printf(sc->sc_dev,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
"microcontroller initialization failed");
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
break;
}
if (uc->subtype == IWN_UCODE_INIT) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Save microcontroller report. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
memcpy(&sc->ucode_info, uc, sizeof (*uc));
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Save the address of the error log in SRAM. */
sc->errptr = le32toh(uc->errptr);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
break;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
case IWN_STATE_CHANGED:
{
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
/*
* State change allows hardware switch change to be
* noted. However, we handle this in iwn_intr as we
* get both the enable/disble intr.
*/
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(sc->rxq.data_dmat, data->map,
BUS_DMASYNC_POSTREAD);
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
#ifdef IWN_DEBUG
uint32_t *status = (uint32_t *)(desc + 1);
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
DPRINTF(sc, IWN_DEBUG_INTR | IWN_DEBUG_STATE,
"state changed to %x\n",
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
le32toh(*status));
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
#endif
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
break;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
case IWN_START_SCAN:
{
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(sc->rxq.data_dmat, data->map,
BUS_DMASYNC_POSTREAD);
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
#ifdef IWN_DEBUG
struct iwn_start_scan *scan =
(struct iwn_start_scan *)(desc + 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_ANY,
"%s: scanning channel %d status %x\n",
__func__, scan->chan, le32toh(scan->status));
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
#endif
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
break;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
case IWN_STOP_SCAN:
{
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(sc->rxq.data_dmat, data->map,
BUS_DMASYNC_POSTREAD);
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
#ifdef IWN_DEBUG
struct iwn_stop_scan *scan =
(struct iwn_stop_scan *)(desc + 1);
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
DPRINTF(sc, IWN_DEBUG_STATE | IWN_DEBUG_SCAN,
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
"scan finished nchan=%d status=%d chan=%d\n",
scan->nchan, scan->status, scan->chan);
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
#endif
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
sc->sc_is_scanning = 0;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_UNLOCK(sc);
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
ieee80211_scan_next(vap);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_LOCK(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
break;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
case IWN5000_CALIBRATION_RESULT:
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
iwn5000_rx_calib_results(sc, desc, data);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
break;
case IWN5000_CALIBRATION_DONE:
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
sc->sc_flags |= IWN_FLAG_CALIB_DONE;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
wakeup(sc);
break;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
sc->rxq.cur = (sc->rxq.cur + 1) % IWN_RX_RING_COUNT;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Tell the firmware what we have processed. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
hw = (hw == 0) ? IWN_RX_RING_COUNT - 1 : hw - 1;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_FH_RX_WPTR, hw & ~7);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Process an INT_WAKEUP interrupt raised when the microcontroller wakes up
* from power-down sleep mode.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_wakeup_intr(struct iwn_softc *sc)
{
int qid;
DPRINTF(sc, IWN_DEBUG_RESET, "%s: ucode wakeup from power-down sleep\n",
__func__);
/* Wakeup RX and TX rings. */
IWN_WRITE(sc, IWN_FH_RX_WPTR, sc->rxq.cur & ~7);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
for (qid = 0; qid < sc->ntxqs; qid++) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_tx_ring *ring = &sc->txq[qid];
IWN_WRITE(sc, IWN_HBUS_TARG_WRPTR, qid << 8 | ring->cur);
}
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
iwn_rftoggle_intr(struct iwn_softc *sc)
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint32_t tmp = IWN_READ(sc, IWN_GP_CNTRL);
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
IWN_LOCK_ASSERT(sc);
device_printf(sc->sc_dev, "RF switch: radio %s\n",
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
(tmp & IWN_GP_CNTRL_RFKILL) ? "enabled" : "disabled");
if (tmp & IWN_GP_CNTRL_RFKILL)
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
ieee80211_runtask(ic, &sc->sc_radioon_task);
else
ieee80211_runtask(ic, &sc->sc_radiooff_task);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Dump the error log of the firmware when a firmware panic occurs. Although
* we can't debug the firmware because it is neither open source nor free, it
* can help us to identify certain classes of problems.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_fatal_intr(struct iwn_softc *sc)
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_fw_dump dump;
int i;
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
IWN_LOCK_ASSERT(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Force a complete recalibration on next init. */
sc->sc_flags &= ~IWN_FLAG_CALIB_DONE;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Check that the error log address is valid. */
if (sc->errptr < IWN_FW_DATA_BASE ||
sc->errptr + sizeof (dump) >
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
IWN_FW_DATA_BASE + sc->fw_data_maxsz) {
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
printf("%s: bad firmware error log address 0x%08x\n", __func__,
sc->errptr);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return;
}
if (iwn_nic_lock(sc) != 0) {
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
printf("%s: could not read firmware error log\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return;
}
/* Read firmware error log from SRAM. */
iwn_mem_read_region_4(sc, sc->errptr, (uint32_t *)&dump,
sizeof (dump) / sizeof (uint32_t));
iwn_nic_unlock(sc);
if (dump.valid == 0) {
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
printf("%s: firmware error log is empty\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return;
}
printf("firmware error log:\n");
printf(" error type = \"%s\" (0x%08X)\n",
(dump.id < nitems(iwn_fw_errmsg)) ?
iwn_fw_errmsg[dump.id] : "UNKNOWN",
dump.id);
printf(" program counter = 0x%08X\n", dump.pc);
printf(" source line = 0x%08X\n", dump.src_line);
printf(" error data = 0x%08X%08X\n",
dump.error_data[0], dump.error_data[1]);
printf(" branch link = 0x%08X%08X\n",
dump.branch_link[0], dump.branch_link[1]);
printf(" interrupt link = 0x%08X%08X\n",
dump.interrupt_link[0], dump.interrupt_link[1]);
printf(" time = %u\n", dump.time[0]);
/* Dump driver status (TX and RX rings) while we're here. */
printf("driver status:\n");
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
for (i = 0; i < sc->ntxqs; i++) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_tx_ring *ring = &sc->txq[i];
printf(" tx ring %2d: qid=%-2d cur=%-3d queued=%-3d\n",
i, ring->qid, ring->cur, ring->queued);
}
printf(" rx ring: cur=%d\n", sc->rxq.cur);
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_intr(void *arg)
{
struct iwn_softc *sc = arg;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
uint32_t r1, r2, tmp;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
IWN_LOCK(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Disable interrupts. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_WRITE(sc, IWN_INT_MASK, 0);
/* Read interrupts from ICT (fast) or from registers (slow). */
if (sc->sc_flags & IWN_FLAG_USE_ICT) {
tmp = 0;
while (sc->ict[sc->ict_cur] != 0) {
tmp |= sc->ict[sc->ict_cur];
sc->ict[sc->ict_cur] = 0; /* Acknowledge. */
sc->ict_cur = (sc->ict_cur + 1) % IWN_ICT_COUNT;
}
tmp = le32toh(tmp);
* Rename bluetooth coexistence flags, no binary change. * Enable DC calibration and crystal calibration on Centrino Advanced-N 6250 parts. * Workaround for a HW bug (does not affect 4965AGN) that may sporadically affect latency under some rare circumstances. From a similar commit to iwlwifi. * Update sensitivity settings for 5000 series to workaround a performance bug in the DSP (1000 is not affected so we keep the old values for 5000). * Update sensitivity settings for 6000 series. * Set differential gains on 6250 too (but use a 1.0 factor, not 1.5). * Init OFDM sensitivity with min value (which depends on the chip) instead of hardcoding it to 90. * Read calibration version from ROM and set IWN_GP_DRIVER_CALIB_VER6 bit on 6x50 if version >= 6. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 2 weeks
2010-04-10 06:58:24 +00:00
if (tmp == 0xffffffff) /* Shouldn't happen. */
tmp = 0;
else if (tmp & 0xc0000) /* Workaround a HW bug. */
tmp |= 0x8000;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
r1 = (tmp & 0xff00) << 16 | (tmp & 0xff);
r2 = 0; /* Unused. */
} else {
r1 = IWN_READ(sc, IWN_INT);
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
if (r1 == 0xffffffff || (r1 & 0xfffffff0) == 0xa5a5a5a0) {
IWN_UNLOCK(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
return; /* Hardware gone! */
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
r2 = IWN_READ(sc, IWN_FH_INT);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_INTR, "interrupt reg1=0x%08x reg2=0x%08x\n"
, r1, r2);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (r1 == 0 && r2 == 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
goto done; /* Interrupt not for us. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Acknowledge interrupts. */
IWN_WRITE(sc, IWN_INT, r1);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (!(sc->sc_flags & IWN_FLAG_USE_ICT))
IWN_WRITE(sc, IWN_FH_INT, r2);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (r1 & IWN_INT_RF_TOGGLED) {
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
iwn_rftoggle_intr(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
goto done;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
if (r1 & IWN_INT_CT_REACHED) {
device_printf(sc->sc_dev, "%s: critical temperature reached!\n",
__func__);
}
if (r1 & (IWN_INT_SW_ERR | IWN_INT_HW_ERR)) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(sc->sc_dev, "%s: fatal firmware error\n",
__func__);
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
#ifdef IWN_DEBUG
iwn_debug_register(sc);
#endif
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Dump firmware error log and stop. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_fatal_intr(sc);
Make iwn(4) able to get itself back into working condition after "fatal firmware error" happens. Previously it was neccessary to reset it manually, using "/etc/rc.d/netif restart". Approved by: adrian@ MFC after: 1 month Sponsored by: The FreeBSD Foundation
2014-05-22 15:29:25 +00:00
taskqueue_enqueue(sc->sc_tq, &sc->sc_panic_task);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
goto done;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if ((r1 & (IWN_INT_FH_RX | IWN_INT_SW_RX | IWN_INT_RX_PERIODIC)) ||
(r2 & IWN_FH_INT_RX)) {
if (sc->sc_flags & IWN_FLAG_USE_ICT) {
if (r1 & (IWN_INT_FH_RX | IWN_INT_SW_RX))
IWN_WRITE(sc, IWN_FH_INT, IWN_FH_INT_RX);
IWN_WRITE_1(sc, IWN_INT_PERIODIC,
IWN_INT_PERIODIC_DIS);
iwn_notif_intr(sc);
if (r1 & (IWN_INT_FH_RX | IWN_INT_SW_RX)) {
IWN_WRITE_1(sc, IWN_INT_PERIODIC,
IWN_INT_PERIODIC_ENA);
}
} else
iwn_notif_intr(sc);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if ((r1 & IWN_INT_FH_TX) || (r2 & IWN_FH_INT_TX)) {
if (sc->sc_flags & IWN_FLAG_USE_ICT)
IWN_WRITE(sc, IWN_FH_INT, IWN_FH_INT_TX);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
wakeup(sc); /* FH DMA transfer completed. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (r1 & IWN_INT_ALIVE)
wakeup(sc); /* Firmware is alive. */
if (r1 & IWN_INT_WAKEUP)
iwn_wakeup_intr(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
done:
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Re-enable interrupts. */
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
if (sc->sc_flags & IWN_FLAG_RUNNING)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_WRITE(sc, IWN_INT_MASK, sc->int_mask);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
IWN_UNLOCK(sc);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Update TX scheduler ring when transmitting an 802.11 frame (4965AGN and
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
* 5000 adapters use a slightly different format).
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_update_sched(struct iwn_softc *sc, int qid, int idx, uint8_t id,
uint16_t len)
{
uint16_t *w = &sc->sched[qid * IWN4965_SCHED_COUNT + idx];
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*w = htole16(len + 8);
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
bus_dmamap_sync(sc->sched_dma.tag, sc->sched_dma.map,
BUS_DMASYNC_PREWRITE);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (idx < IWN_SCHED_WINSZ) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*(w + IWN_TX_RING_COUNT) = *w;
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
bus_dmamap_sync(sc->sched_dma.tag, sc->sched_dma.map,
BUS_DMASYNC_PREWRITE);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_update_sched(struct iwn_softc *sc, int qid, int idx, uint8_t id,
uint16_t len)
{
uint16_t *w = &sc->sched[qid * IWN5000_SCHED_COUNT + idx];
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*w = htole16(id << 12 | (len + 8));
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
bus_dmamap_sync(sc->sched_dma.tag, sc->sched_dma.map,
BUS_DMASYNC_PREWRITE);
if (idx < IWN_SCHED_WINSZ) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*(w + IWN_TX_RING_COUNT) = *w;
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
bus_dmamap_sync(sc->sched_dma.tag, sc->sched_dma.map,
BUS_DMASYNC_PREWRITE);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
iwn5000_reset_sched() is not used; #ifdef it. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:47:46 +00:00
#ifdef notyet
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_reset_sched(struct iwn_softc *sc, int qid, int idx)
{
uint16_t *w = &sc->sched[qid * IWN5000_SCHED_COUNT + idx];
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*w = (*w & htole16(0xf000)) | htole16(1);
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
bus_dmamap_sync(sc->sched_dma.tag, sc->sched_dma.map,
BUS_DMASYNC_PREWRITE);
if (idx < IWN_SCHED_WINSZ) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*(w + IWN_TX_RING_COUNT) = *w;
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
bus_dmamap_sync(sc->sched_dma.tag, sc->sched_dma.map,
Small whitespace cleanup. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-10 06:55:29 +00:00
BUS_DMASYNC_PREWRITE);
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
iwn5000_reset_sched() is not used; #ifdef it. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:47:46 +00:00
#endif
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Temporarily disable multi-rate retry (link quality) and eliminate rate index lookups. * My recent(ish) change to iwn(4) and the net80211 rate control API to support 11n rates broke the link quality table use. So, until I or someone else decides to fix it, let's just disable it for now. * Teach iwn_tx_data_raw() to use the iwn_rate_to_plcp() function. * Eliminate two uses of the net80211 rate index lookup functions - they are only for legacy rates and they're not needed here. This fixes some invalid looking rate control TX issues that showed up on my 4965 but it doesn't fix the two TX hangs I've noticed. Those look like DMA related issues. Tested: * 4965, STA mode * 5100, STA mode
2013-10-25 19:44:53 +00:00
/*
* Check whether OFDM 11g protection will be enabled for the given rate.
*
* The original driver code only enabled protection for OFDM rates.
* It didn't check to see whether it was operating in 11a or 11bg mode.
*/
static int
iwn_check_rate_needs_protection(struct iwn_softc *sc,
struct ieee80211vap *vap, uint8_t rate)
{
struct ieee80211com *ic = vap->iv_ic;
/*
* Not in 2GHz mode? Then there's no need to enable OFDM
* 11bg protection.
*/
if (! IEEE80211_IS_CHAN_2GHZ(ic->ic_curchan)) {
return (0);
}
/*
* 11bg protection not enabled? Then don't use it.
*/
if ((ic->ic_flags & IEEE80211_F_USEPROT) == 0)
return (0);
/*
Implement htprotmode handling. This is separate to 11g protection - the default is to RTS protect 11n frames, including A-MPDU frames. Tested: * Intel 5100, STA mode
2014-09-08 07:16:00 +00:00
* If it's an 11n rate - no protection.
* We'll do it via a specific 11n check.
Temporarily disable multi-rate retry (link quality) and eliminate rate index lookups. * My recent(ish) change to iwn(4) and the net80211 rate control API to support 11n rates broke the link quality table use. So, until I or someone else decides to fix it, let's just disable it for now. * Teach iwn_tx_data_raw() to use the iwn_rate_to_plcp() function. * Eliminate two uses of the net80211 rate index lookup functions - they are only for legacy rates and they're not needed here. This fixes some invalid looking rate control TX issues that showed up on my 4965 but it doesn't fix the two TX hangs I've noticed. Those look like DMA related issues. Tested: * 4965, STA mode * 5100, STA mode
2013-10-25 19:44:53 +00:00
*/
if (rate & IEEE80211_RATE_MCS) {
Implement htprotmode handling. This is separate to 11g protection - the default is to RTS protect 11n frames, including A-MPDU frames. Tested: * Intel 5100, STA mode
2014-09-08 07:16:00 +00:00
return (0);
Temporarily disable multi-rate retry (link quality) and eliminate rate index lookups. * My recent(ish) change to iwn(4) and the net80211 rate control API to support 11n rates broke the link quality table use. So, until I or someone else decides to fix it, let's just disable it for now. * Teach iwn_tx_data_raw() to use the iwn_rate_to_plcp() function. * Eliminate two uses of the net80211 rate index lookup functions - they are only for legacy rates and they're not needed here. This fixes some invalid looking rate control TX issues that showed up on my 4965 but it doesn't fix the two TX hangs I've noticed. Those look like DMA related issues. Tested: * 4965, STA mode * 5100, STA mode
2013-10-25 19:44:53 +00:00
}
/*
* Do a rate table lookup. If the PHY is CCK,
* don't do protection.
*/
if (ieee80211_rate2phytype(ic->ic_rt, rate) == IEEE80211_T_CCK)
return (0);
/*
* Yup, enable protection.
*/
return (1);
}
Fix up the link quality lookup and re-enable multi-rate retry. This is a terrible solution that at least behaves mostly correctly. It walks the currently active rate table looking for rates to match. It assumes that the code matches the setup path in the link quality setup code (much like the previous, much simpler but even more hackish math did.) It's O(n), but n<15, so we're okay for the time being. Tested: * Intel 5100, STA - 11a, 11n, 11bg modes.
2013-11-12 05:49:01 +00:00
/*
* return a value between 0 and IWN_MAX_TX_RETRIES-1 as an index into
* the link quality table that reflects this particular entry.
*/
static int
iwn_tx_rate_to_linkq_offset(struct iwn_softc *sc, struct ieee80211_node *ni,
uint8_t rate)
{
struct ieee80211_rateset *rs;
int is_11n;
int nr;
int i;
uint8_t cmp_rate;
/*
* Figure out if we're using 11n or not here.
*/
if (IEEE80211_IS_CHAN_HT(ni->ni_chan) && ni->ni_htrates.rs_nrates > 0)
is_11n = 1;
else
is_11n = 0;
/*
* Use the correct rate table.
*/
if (is_11n) {
rs = (struct ieee80211_rateset *) &ni->ni_htrates;
nr = ni->ni_htrates.rs_nrates;
} else {
rs = &ni->ni_rates;
nr = rs->rs_nrates;
}
/*
* Find the relevant link quality entry in the table.
*/
for (i = 0; i < nr && i < IWN_MAX_TX_RETRIES - 1 ; i++) {
/*
* The link quality table index starts at 0 == highest
* rate, so we walk the rate table backwards.
*/
cmp_rate = rs->rs_rates[(nr - 1) - i];
if (rate & IEEE80211_RATE_MCS)
cmp_rate |= IEEE80211_RATE_MCS;
Disable this debugging - it's far too verbose when doing TX rate debugging.
2013-11-29 22:36:00 +00:00
#if 0
Fix up the link quality lookup and re-enable multi-rate retry. This is a terrible solution that at least behaves mostly correctly. It walks the currently active rate table looking for rates to match. It assumes that the code matches the setup path in the link quality setup code (much like the previous, much simpler but even more hackish math did.) It's O(n), but n<15, so we're okay for the time being. Tested: * Intel 5100, STA - 11a, 11n, 11bg modes.
2013-11-12 05:49:01 +00:00
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: idx %d: nr=%d, rate=0x%02x, rateentry=0x%02x\n",
__func__,
i,
nr,
rate,
cmp_rate);
Disable this debugging - it's far too verbose when doing TX rate debugging.
2013-11-29 22:36:00 +00:00
#endif
Fix up the link quality lookup and re-enable multi-rate retry. This is a terrible solution that at least behaves mostly correctly. It walks the currently active rate table looking for rates to match. It assumes that the code matches the setup path in the link quality setup code (much like the previous, much simpler but even more hackish math did.) It's O(n), but n<15, so we're okay for the time being. Tested: * Intel 5100, STA - 11a, 11n, 11bg modes.
2013-11-12 05:49:01 +00:00
if (cmp_rate == rate)
return (i);
}
/* Failed? Start at the end */
return (IWN_MAX_TX_RETRIES - 1);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
iwn_tx_data(struct iwn_softc *sc, struct mbuf *m, struct ieee80211_node *ni)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
struct iwn_ops *ops = &sc->ops;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
const struct ieee80211_txparam *tp;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct ieee80211vap *vap = ni->ni_vap;
struct ieee80211com *ic = ni->ni_ic;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_node *wn = (void *)ni;
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
struct iwn_tx_ring *ring;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_tx_desc *desc;
struct iwn_tx_data *data;
struct iwn_tx_cmd *cmd;
struct iwn_cmd_data *tx;
struct ieee80211_frame *wh;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct ieee80211_key *k = NULL;
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
struct mbuf *m1;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
uint32_t flags;
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
uint16_t qos;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
u_int hdrlen;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
bus_dma_segment_t *seg, segs[IWN_MAX_SCATTER];
Temporarily disable multi-rate retry (link quality) and eliminate rate index lookups. * My recent(ish) change to iwn(4) and the net80211 rate control API to support 11n rates broke the link quality table use. So, until I or someone else decides to fix it, let's just disable it for now. * Teach iwn_tx_data_raw() to use the iwn_rate_to_plcp() function. * Eliminate two uses of the net80211 rate index lookup functions - they are only for legacy rates and they're not needed here. This fixes some invalid looking rate control TX issues that showed up on my 4965 but it doesn't fix the two TX hangs I've noticed. Those look like DMA related issues. Tested: * 4965, STA mode * 5100, STA mode
2013-10-25 19:44:53 +00:00
uint8_t tid, type;
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
int ac, i, totlen, error, pad, nsegs = 0, rate;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
IWN_LOCK_ASSERT(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
wh = mtod(m, struct ieee80211_frame *);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
hdrlen = ieee80211_anyhdrsize(wh);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
type = wh->i_fc[0] & IEEE80211_FC0_TYPE_MASK;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
/* Select EDCA Access Category and TX ring for this frame. */
if (IEEE80211_QOS_HAS_SEQ(wh)) {
qos = ((const struct ieee80211_qosframe *)wh)->i_qos[0];
tid = qos & IEEE80211_QOS_TID;
} else {
qos = 0;
tid = 0;
}
ac = M_WME_GETAC(m);
Use the M_AMPDU_MPDU flag to determine when to manually set the seqno and use a BA queue.
2012-04-15 18:25:17 +00:00
if (m->m_flags & M_AMPDU_MPDU) {
Add some code to double-check whether we're correctly populating the TX ring according to what the firmware requires. The firmware requires A-MPDU sub-frames to be at a very specific ring offset - that is, the ring slot offset should be (seqno % 256.) This holds for every NIC I've tested thus far except the 4965, which starts erroring out here shortly before the firmware panics. Which is good, it's doing what it's supposed to (read: capture that we've screwed up somewhere.) The specifics about getting this stuff right: * the initial seqno allocation should match up with the ringid. * .. yes, this means we can start at a ring offset that isn't zero. * .. because we program the start seqno in the firmware message to setup the AC. * The initial seqno allocation may be non-zero _and_ frames may be being transmitted during a-mpdu negotiation. I faced similar issues on ath(4) and had to software queue frames to that node+TID during A-MPDU negotiation. * seqno allocation should be in lockstep with ring increments. * If we fail to transmit some segment, no, we shouldn't reuse that ring slot. We should just transmit a BAR (which we aren't yet doing, sigh) and move onto the next seqno. * In theory there shouldn't be any holes in the seqno space when we are transmitting frames. Tested: * 4965 (throws problems, so yes we have to fix this); * 5100 (seems ok); * 6200 (seems ok); * 2200 (seems ok); * 2230 (seems ok).
2013-12-02 03:40:51 +00:00
uint16_t seqno;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
struct ieee80211_tx_ampdu *tap = &ni->ni_tx_ampdu[ac];
Discard frames after a DELBA which where queued during an active BA session. MFC after: 1 week
2012-05-20 10:00:36 +00:00
if (!IEEE80211_AMPDU_RUNNING(tap)) {
m_freem(m);
return EINVAL;
}
Add some code to double-check whether we're correctly populating the TX ring according to what the firmware requires. The firmware requires A-MPDU sub-frames to be at a very specific ring offset - that is, the ring slot offset should be (seqno % 256.) This holds for every NIC I've tested thus far except the 4965, which starts erroring out here shortly before the firmware panics. Which is good, it's doing what it's supposed to (read: capture that we've screwed up somewhere.) The specifics about getting this stuff right: * the initial seqno allocation should match up with the ringid. * .. yes, this means we can start at a ring offset that isn't zero. * .. because we program the start seqno in the firmware message to setup the AC. * The initial seqno allocation may be non-zero _and_ frames may be being transmitted during a-mpdu negotiation. I faced similar issues on ath(4) and had to software queue frames to that node+TID during A-MPDU negotiation. * seqno allocation should be in lockstep with ring increments. * If we fail to transmit some segment, no, we shouldn't reuse that ring slot. We should just transmit a BAR (which we aren't yet doing, sigh) and move onto the next seqno. * In theory there shouldn't be any holes in the seqno space when we are transmitting frames. Tested: * 4965 (throws problems, so yes we have to fix this); * 5100 (seems ok); * 6200 (seems ok); * 2200 (seems ok); * 2230 (seems ok).
2013-12-02 03:40:51 +00:00
/*
* Queue this frame to the hardware ring that we've
* negotiated AMPDU TX on.
*
* Note that the sequence number must match the TX slot
* being used!
*/
Use the M_AMPDU_MPDU flag to determine when to manually set the seqno and use a BA queue.
2012-04-15 18:25:17 +00:00
ac = *(int *)tap->txa_private;
Add some code to double-check whether we're correctly populating the TX ring according to what the firmware requires. The firmware requires A-MPDU sub-frames to be at a very specific ring offset - that is, the ring slot offset should be (seqno % 256.) This holds for every NIC I've tested thus far except the 4965, which starts erroring out here shortly before the firmware panics. Which is good, it's doing what it's supposed to (read: capture that we've screwed up somewhere.) The specifics about getting this stuff right: * the initial seqno allocation should match up with the ringid. * .. yes, this means we can start at a ring offset that isn't zero. * .. because we program the start seqno in the firmware message to setup the AC. * The initial seqno allocation may be non-zero _and_ frames may be being transmitted during a-mpdu negotiation. I faced similar issues on ath(4) and had to software queue frames to that node+TID during A-MPDU negotiation. * seqno allocation should be in lockstep with ring increments. * If we fail to transmit some segment, no, we shouldn't reuse that ring slot. We should just transmit a BAR (which we aren't yet doing, sigh) and move onto the next seqno. * In theory there shouldn't be any holes in the seqno space when we are transmitting frames. Tested: * 4965 (throws problems, so yes we have to fix this); * 5100 (seems ok); * 6200 (seems ok); * 2200 (seems ok); * 2230 (seems ok).
2013-12-02 03:40:51 +00:00
seqno = ni->ni_txseqs[tid];
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
*(uint16_t *)wh->i_seq =
Add some code to double-check whether we're correctly populating the TX ring according to what the firmware requires. The firmware requires A-MPDU sub-frames to be at a very specific ring offset - that is, the ring slot offset should be (seqno % 256.) This holds for every NIC I've tested thus far except the 4965, which starts erroring out here shortly before the firmware panics. Which is good, it's doing what it's supposed to (read: capture that we've screwed up somewhere.) The specifics about getting this stuff right: * the initial seqno allocation should match up with the ringid. * .. yes, this means we can start at a ring offset that isn't zero. * .. because we program the start seqno in the firmware message to setup the AC. * The initial seqno allocation may be non-zero _and_ frames may be being transmitted during a-mpdu negotiation. I faced similar issues on ath(4) and had to software queue frames to that node+TID during A-MPDU negotiation. * seqno allocation should be in lockstep with ring increments. * If we fail to transmit some segment, no, we shouldn't reuse that ring slot. We should just transmit a BAR (which we aren't yet doing, sigh) and move onto the next seqno. * In theory there shouldn't be any holes in the seqno space when we are transmitting frames. Tested: * 4965 (throws problems, so yes we have to fix this); * 5100 (seems ok); * 6200 (seems ok); * 2200 (seems ok); * 2230 (seems ok).
2013-12-02 03:40:51 +00:00
htole16(seqno << IEEE80211_SEQ_SEQ_SHIFT);
ring = &sc->txq[ac];
if ((seqno % 256) != ring->cur) {
device_printf(sc->sc_dev,
"%s: m=%p: seqno (%d) (%d) != ring index (%d) !\n",
__func__,
m,
seqno,
seqno % 256,
ring->cur);
}
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
ni->ni_txseqs[tid]++;
}
Use the M_AMPDU_MPDU flag to determine when to manually set the seqno and use a BA queue.
2012-04-15 18:25:17 +00:00
ring = &sc->txq[ac];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
desc = &ring->desc[ring->cur];
data = &ring->data[ring->cur];
/* Choose a TX rate index. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
tp = &vap->iv_txparms[ieee80211_chan2mode(ni->ni_chan)];
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (type == IEEE80211_FC0_TYPE_MGT)
rate = tp->mgmtrate;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
else if (IEEE80211_IS_MULTICAST(wh->i_addr1))
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
rate = tp->mcastrate;
else if (tp->ucastrate != IEEE80211_FIXED_RATE_NONE)
rate = tp->ucastrate;
Send EAPOL frames at the management rate, not the data rate. Without this, a far away station with low signal strength would associate using the management rate (by default the lowest rate) and then the EAPOL frames would go out at the current AMRR best guess. This would result in association failing authentication. Tested: * Intel 5100, STA * Intel 2230, STA
2013-11-11 08:53:20 +00:00
else if (m->m_flags & M_EAPOL)
rate = tp->mgmtrate;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
else {
net80211 rate control framework (net80211 ratectl). This framework allows drivers to abstract the rate control algorithm and just feed the framework with the usable parameters. The rate control framework will now deal with passing the parameters to the selected algorithm. Right now we have AMRR (the default) and RSSADAPT but there's no way to select one with ifconfig, yet. The objective is to have more rate control algorithms in the net80211 stack so all drivers[0] can use it. Ideally, we'll have the well-known sample rate control algorithm in the net80211 at some point so all drivers can use it (not just ath). [0] all drivers that do rate control in software, that is. Reviewed by: bschmidt, thompsa, weyongo MFC after: 1 months
2010-04-07 15:29:13 +00:00
/* XXX pass pktlen */
(void) ieee80211_ratectl_rate(ni, NULL, 0);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
rate = ni->ni_txrate;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Encrypt the frame if need be. */
Rename definition of IEEE80211_FC1_WEP to IEEE80211_FC1_PROTECTED. The origin of WEP comes from IEEE Std 802.11-1997 where it defines whether the frame body of MAC frame has been encrypted using WEP algorithm or not. IEEE Std. 802.11-2007 changes WEP to Protected Frame, indicates whether the frame is protected by a cryptographic encapsulation algorithm. Reviewed by: adrian, rpaulo
2014-01-08 08:06:56 +00:00
if (wh->i_fc[1] & IEEE80211_FC1_PROTECTED) {
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Retrieve key for TX. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
k = ieee80211_crypto_encap(ni, m);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (k == NULL) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
m_freem(m);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return ENOBUFS;
}
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* 802.11 header may have moved. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
wh = mtod(m, struct ieee80211_frame *);
}
totlen = m->m_pkthdr.len;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Overhaul monitor mode handling: o replace DLT_IEEE802_11 support in net80211 with DLT_IEEE802_11_RADIO and remove explicit bpf support from wireless drivers; drivers now use ieee80211_radiotap_attach to setup shared data structures that hold the radiotap header for each packet tx/rx o remove rx timestamp from the rx path; it was used only by the tdma support for debugging and was mostly useless due to it being 32-bits and mostly unavailable o track DLT_IEEE80211_RADIO bpf attachments and maintain per-vap and per-com state when there are active taps o track the number of monitor mode vaps o use bpf tap and monitor mode vap state to decide when to collect radiotap state and dispatch frames; drivers no longer explicitly directly check bpf state or use bpf calls to tap frames o handle radiotap state updates on channel change in net80211; drivers should not do this (unless they bypass net80211 which is almost always a mistake) o update various drivers to be more consistent/correct in handling radiotap o update ral to include TSF in radiotap'd frames o add promisc mode callback to wi Reviewed by: cbzimmer, rpaulo, thompsa
2009-05-20 20:00:40 +00:00
if (ieee80211_radiotap_active_vap(vap)) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_tx_radiotap_header *tap = &sc->sc_txtap;
tap->wt_flags = 0;
Prepare for transmitting frames at MCS rates: - instead of calling iwn_plcp_signal() for every frame, map the expected value directly within wn->ridx - concat plcp, rflags and xrflags, there is no clean byte boundary within the flags, for example the antenna setting uses bit 6, 7 and 8 - there is still need for a custom rate to plcp mapping, as those expected by the hardware are not conform to the std
2011-05-08 11:49:50 +00:00
tap->wt_rate = rate;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (k != NULL)
tap->wt_flags |= IEEE80211_RADIOTAP_F_WEP;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
ieee80211_radiotap_tx(vap, m);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Prepare TX firmware command. */
cmd = &ring->cmd[ring->cur];
cmd->code = IWN_CMD_TX_DATA;
cmd->flags = 0;
cmd->qid = ring->qid;
cmd->idx = ring->cur;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
tx = (struct iwn_cmd_data *)cmd->data;
/* NB: No need to clear tx, all fields are reinitialized here. */
tx->scratch = 0; /* clear "scratch" area */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
flags = 0;
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
if (!IEEE80211_IS_MULTICAST(wh->i_addr1)) {
/* Unicast frame, check if an ACK is expected. */
if (!qos || (qos & IEEE80211_QOS_ACKPOLICY) !=
IEEE80211_QOS_ACKPOLICY_NOACK)
flags |= IWN_TX_NEED_ACK;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if ((wh->i_fc[0] &
(IEEE80211_FC0_TYPE_MASK | IEEE80211_FC0_SUBTYPE_MASK)) ==
(IEEE80211_FC0_TYPE_CTL | IEEE80211_FC0_SUBTYPE_BAR))
flags |= IWN_TX_IMM_BA; /* Cannot happen yet. */
if (wh->i_fc[1] & IEEE80211_FC1_MORE_FRAG)
flags |= IWN_TX_MORE_FRAG; /* Cannot happen yet. */
/* Check if frame must be protected using RTS/CTS or CTS-to-self. */
if (!IEEE80211_IS_MULTICAST(wh->i_addr1)) {
/* NB: Group frames are sent using CCK in 802.11b/g. */
if (totlen + IEEE80211_CRC_LEN > vap->iv_rtsthreshold) {
flags |= IWN_TX_NEED_RTS;
Temporarily disable multi-rate retry (link quality) and eliminate rate index lookups. * My recent(ish) change to iwn(4) and the net80211 rate control API to support 11n rates broke the link quality table use. So, until I or someone else decides to fix it, let's just disable it for now. * Teach iwn_tx_data_raw() to use the iwn_rate_to_plcp() function. * Eliminate two uses of the net80211 rate index lookup functions - they are only for legacy rates and they're not needed here. This fixes some invalid looking rate control TX issues that showed up on my 4965 but it doesn't fix the two TX hangs I've noticed. Those look like DMA related issues. Tested: * 4965, STA mode * 5100, STA mode
2013-10-25 19:44:53 +00:00
} else if (iwn_check_rate_needs_protection(sc, vap, rate)) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (ic->ic_protmode == IEEE80211_PROT_CTSONLY)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
flags |= IWN_TX_NEED_CTS;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
else if (ic->ic_protmode == IEEE80211_PROT_RTSCTS)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
flags |= IWN_TX_NEED_RTS;
Implement htprotmode handling. This is separate to 11g protection - the default is to RTS protect 11n frames, including A-MPDU frames. Tested: * Intel 5100, STA mode
2014-09-08 07:16:00 +00:00
} else if ((rate & IEEE80211_RATE_MCS) &&
(ic->ic_htprotmode == IEEE80211_PROT_RTSCTS)) {
flags |= IWN_TX_NEED_RTS;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Temporarily disable multi-rate retry (link quality) and eliminate rate index lookups. * My recent(ish) change to iwn(4) and the net80211 rate control API to support 11n rates broke the link quality table use. So, until I or someone else decides to fix it, let's just disable it for now. * Teach iwn_tx_data_raw() to use the iwn_rate_to_plcp() function. * Eliminate two uses of the net80211 rate index lookup functions - they are only for legacy rates and they're not needed here. This fixes some invalid looking rate control TX issues that showed up on my 4965 but it doesn't fix the two TX hangs I've noticed. Those look like DMA related issues. Tested: * 4965, STA mode * 5100, STA mode
2013-10-25 19:44:53 +00:00
/* XXX HT protection? */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (flags & (IWN_TX_NEED_RTS | IWN_TX_NEED_CTS)) {
if (sc->hw_type != IWN_HW_REV_TYPE_4965) {
/* 5000 autoselects RTS/CTS or CTS-to-self. */
flags &= ~(IWN_TX_NEED_RTS | IWN_TX_NEED_CTS);
flags |= IWN_TX_NEED_PROTECTION;
} else
flags |= IWN_TX_FULL_TXOP;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (IEEE80211_IS_MULTICAST(wh->i_addr1) ||
type != IEEE80211_FC0_TYPE_DATA)
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
tx->id = sc->broadcast_id;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
else
tx->id = wn->id;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (type == IEEE80211_FC0_TYPE_MGT) {
uint8_t subtype = wh->i_fc[0] & IEEE80211_FC0_SUBTYPE_MASK;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Tell HW to set timestamp in probe responses. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (subtype == IEEE80211_FC0_SUBTYPE_PROBE_RESP)
flags |= IWN_TX_INSERT_TSTAMP;
if (subtype == IEEE80211_FC0_SUBTYPE_ASSOC_REQ ||
subtype == IEEE80211_FC0_SUBTYPE_REASSOC_REQ)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
tx->timeout = htole16(3);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
else
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
tx->timeout = htole16(2);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
} else
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
tx->timeout = htole16(0);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (hdrlen & 3) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* First segment length must be a multiple of 4. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
flags |= IWN_TX_NEED_PADDING;
pad = 4 - (hdrlen & 3);
} else
pad = 0;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
tx->len = htole16(totlen);
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
tx->tid = tid;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
tx->rts_ntries = 60;
tx->data_ntries = 15;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
tx->lifetime = htole32(IWN_LIFETIME_INFINITE);
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
tx->rate = iwn_rate_to_plcp(sc, ni, rate);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
if (tx->id == sc->broadcast_id) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Group or management frame. */
tx->linkq = 0;
} else {
Fix up the link quality lookup and re-enable multi-rate retry. This is a terrible solution that at least behaves mostly correctly. It walks the currently active rate table looking for rates to match. It assumes that the code matches the setup path in the link quality setup code (much like the previous, much simpler but even more hackish math did.) It's O(n), but n<15, so we're okay for the time being. Tested: * Intel 5100, STA - 11a, 11n, 11bg modes.
2013-11-12 05:49:01 +00:00
tx->linkq = iwn_tx_rate_to_linkq_offset(sc, ni, rate);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
flags |= IWN_TX_LINKQ; /* enable MRR */
}
Fix up the link quality lookup and re-enable multi-rate retry. This is a terrible solution that at least behaves mostly correctly. It walks the currently active rate table looking for rates to match. It assumes that the code matches the setup path in the link quality setup code (much like the previous, much simpler but even more hackish math did.) It's O(n), but n<15, so we're okay for the time being. Tested: * Intel 5100, STA - 11a, 11n, 11bg modes.
2013-11-12 05:49:01 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Set physical address of "scratch area". */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
tx->loaddr = htole32(IWN_LOADDR(data->scratch_paddr));
tx->hiaddr = IWN_HIADDR(data->scratch_paddr);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Copy 802.11 header in TX command. */
memcpy((uint8_t *)(tx + 1), wh, hdrlen);
/* Trim 802.11 header. */
m_adj(m, hdrlen);
tx->security = 0;
tx->flags = htole32(flags);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
error = bus_dmamap_load_mbuf_sg(ring->data_dmat, data->map, m, segs,
&nsegs, BUS_DMA_NOWAIT);
if (error != 0) {
if (error != EFBIG) {
device_printf(sc->sc_dev,
"%s: can't map mbuf (error %d)\n", __func__, error);
m_freem(m);
return error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
/* Too many DMA segments, linearize mbuf. */
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
m1 = m_collapse(m, M_NOWAIT, IWN_MAX_SCATTER - 1);
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
if (m1 == NULL) {
device_printf(sc->sc_dev,
"%s: could not defrag mbuf\n", __func__);
m_freem(m);
return ENOBUFS;
}
m = m1;
error = bus_dmamap_load_mbuf_sg(ring->data_dmat, data->map, m,
segs, &nsegs, BUS_DMA_NOWAIT);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
"%s: can't map mbuf (error %d)\n", __func__, error);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
m_freem(m);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
data->m = m;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
data->ni = ni;
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_XMIT,
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
"%s: qid %d idx %d len %d nsegs %d flags 0x%08x rate 0x%04x plcp 0x%08x\n",
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
__func__,
ring->qid,
ring->cur,
m->m_pkthdr.len,
nsegs,
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
flags,
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
rate,
tx->rate);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Fill TX descriptor. */
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
desc->nsegs = 1;
if (m->m_len != 0)
desc->nsegs += nsegs;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* First DMA segment is used by the TX command. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
desc->segs[0].addr = htole32(IWN_LOADDR(data->cmd_paddr));
desc->segs[0].len = htole16(IWN_HIADDR(data->cmd_paddr) |
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
(4 + sizeof (*tx) + hdrlen + pad) << 4);
/* Other DMA segments are for data payload. */
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
seg = &segs[0];
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (i = 1; i <= nsegs; i++) {
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
desc->segs[i].addr = htole32(IWN_LOADDR(seg->ds_addr));
desc->segs[i].len = htole16(IWN_HIADDR(seg->ds_addr) |
seg->ds_len << 4);
seg++;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(ring->data_dmat, data->map, BUS_DMASYNC_PREWRITE);
bus_dmamap_sync(ring->data_dmat, ring->cmd_dma.map,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
BUS_DMASYNC_PREWRITE);
bus_dmamap_sync(ring->desc_dma.tag, ring->desc_dma.map,
BUS_DMASYNC_PREWRITE);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Update TX scheduler. */
Only update the scheduler's byte count table for aggregation queues. The other queues, especially the command queue, uses the FIFO mode which doesn't require the byte count table because queued entries are processed in order. Pointed out by: Lucius Windschuh <lwindschuh at googlemail dot com>
2011-05-15 08:09:36 +00:00
if (ring->qid >= sc->firstaggqueue)
ops->update_sched(sc, ring->qid, ring->cur, tx->id, totlen);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Kick TX ring. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ring->cur = (ring->cur + 1) % IWN_TX_RING_COUNT;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_HBUS_TARG_WRPTR, ring->qid << 8 | ring->cur);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Mark TX ring as full if we reach a certain threshold. */
if (++ring->queued > IWN_TX_RING_HIMARK)
sc->qfullmsk |= 1 << ring->qid;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return 0;
}
static int
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_tx_data_raw(struct iwn_softc *sc, struct mbuf *m,
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
struct ieee80211_node *ni, const struct ieee80211_bpf_params *params)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
struct iwn_ops *ops = &sc->ops;
Overhaul monitor mode handling: o replace DLT_IEEE802_11 support in net80211 with DLT_IEEE802_11_RADIO and remove explicit bpf support from wireless drivers; drivers now use ieee80211_radiotap_attach to setup shared data structures that hold the radiotap header for each packet tx/rx o remove rx timestamp from the rx path; it was used only by the tdma support for debugging and was mostly useless due to it being 32-bits and mostly unavailable o track DLT_IEEE80211_RADIO bpf attachments and maintain per-vap and per-com state when there are active taps o track the number of monitor mode vaps o use bpf tap and monitor mode vap state to decide when to collect radiotap state and dispatch frames; drivers no longer explicitly directly check bpf state or use bpf calls to tap frames o handle radiotap state updates on channel change in net80211; drivers should not do this (unless they bypass net80211 which is almost always a mistake) o update various drivers to be more consistent/correct in handling radiotap o update ral to include TSF in radiotap'd frames o add promisc mode callback to wi Reviewed by: cbzimmer, rpaulo, thompsa
2009-05-20 20:00:40 +00:00
struct ieee80211vap *vap = ni->ni_vap;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_tx_cmd *cmd;
struct iwn_cmd_data *tx;
struct ieee80211_frame *wh;
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
struct iwn_tx_ring *ring;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_tx_desc *desc;
struct iwn_tx_data *data;
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
struct mbuf *m1;
bus_dma_segment_t *seg, segs[IWN_MAX_SCATTER];
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
uint32_t flags;
u_int hdrlen;
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
int ac, totlen, error, pad, nsegs = 0, i, rate;
Temporarily disable multi-rate retry (link quality) and eliminate rate index lookups. * My recent(ish) change to iwn(4) and the net80211 rate control API to support 11n rates broke the link quality table use. So, until I or someone else decides to fix it, let's just disable it for now. * Teach iwn_tx_data_raw() to use the iwn_rate_to_plcp() function. * Eliminate two uses of the net80211 rate index lookup functions - they are only for legacy rates and they're not needed here. This fixes some invalid looking rate control TX issues that showed up on my 4965 but it doesn't fix the two TX hangs I've noticed. Those look like DMA related issues. Tested: * 4965, STA mode * 5100, STA mode
2013-10-25 19:44:53 +00:00
uint8_t type;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
IWN_LOCK_ASSERT(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
wh = mtod(m, struct ieee80211_frame *);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
hdrlen = ieee80211_anyhdrsize(wh);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
type = wh->i_fc[0] & IEEE80211_FC0_TYPE_MASK;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
ac = params->ibp_pri & 3;
ring = &sc->txq[ac];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
desc = &ring->desc[ring->cur];
data = &ring->data[ring->cur];
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Temporarily disable multi-rate retry (link quality) and eliminate rate index lookups. * My recent(ish) change to iwn(4) and the net80211 rate control API to support 11n rates broke the link quality table use. So, until I or someone else decides to fix it, let's just disable it for now. * Teach iwn_tx_data_raw() to use the iwn_rate_to_plcp() function. * Eliminate two uses of the net80211 rate index lookup functions - they are only for legacy rates and they're not needed here. This fixes some invalid looking rate control TX issues that showed up on my 4965 but it doesn't fix the two TX hangs I've noticed. Those look like DMA related issues. Tested: * 4965, STA mode * 5100, STA mode
2013-10-25 19:44:53 +00:00
/* Choose a TX rate. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
rate = params->ibp_rate0;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
totlen = m->m_pkthdr.len;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Prepare TX firmware command. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
cmd = &ring->cmd[ring->cur];
cmd->code = IWN_CMD_TX_DATA;
cmd->flags = 0;
cmd->qid = ring->qid;
cmd->idx = ring->cur;
tx = (struct iwn_cmd_data *)cmd->data;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* NB: No need to clear tx, all fields are reinitialized here. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
tx->scratch = 0; /* clear "scratch" area */
flags = 0;
if ((params->ibp_flags & IEEE80211_BPF_NOACK) == 0)
flags |= IWN_TX_NEED_ACK;
if (params->ibp_flags & IEEE80211_BPF_RTS) {
if (sc->hw_type != IWN_HW_REV_TYPE_4965) {
/* 5000 autoselects RTS/CTS or CTS-to-self. */
flags &= ~IWN_TX_NEED_RTS;
flags |= IWN_TX_NEED_PROTECTION;
} else
flags |= IWN_TX_NEED_RTS | IWN_TX_FULL_TXOP;
}
if (params->ibp_flags & IEEE80211_BPF_CTS) {
if (sc->hw_type != IWN_HW_REV_TYPE_4965) {
/* 5000 autoselects RTS/CTS or CTS-to-self. */
flags &= ~IWN_TX_NEED_CTS;
flags |= IWN_TX_NEED_PROTECTION;
} else
flags |= IWN_TX_NEED_CTS | IWN_TX_FULL_TXOP;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (type == IEEE80211_FC0_TYPE_MGT) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint8_t subtype = wh->i_fc[0] & IEEE80211_FC0_SUBTYPE_MASK;
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Tell HW to set timestamp in probe responses. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (subtype == IEEE80211_FC0_SUBTYPE_PROBE_RESP)
flags |= IWN_TX_INSERT_TSTAMP;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (subtype == IEEE80211_FC0_SUBTYPE_ASSOC_REQ ||
subtype == IEEE80211_FC0_SUBTYPE_REASSOC_REQ)
tx->timeout = htole16(3);
else
tx->timeout = htole16(2);
} else
tx->timeout = htole16(0);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (hdrlen & 3) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* First segment length must be a multiple of 4. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
flags |= IWN_TX_NEED_PADDING;
pad = 4 - (hdrlen & 3);
} else
pad = 0;
if (ieee80211_radiotap_active_vap(vap)) {
struct iwn_tx_radiotap_header *tap = &sc->sc_txtap;
tap->wt_flags = 0;
tap->wt_rate = rate;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
ieee80211_radiotap_tx(vap, m);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
tx->len = htole16(totlen);
tx->tid = 0;
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
tx->id = sc->broadcast_id;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
tx->rts_ntries = params->ibp_try1;
tx->data_ntries = params->ibp_try0;
tx->lifetime = htole32(IWN_LIFETIME_INFINITE);
Temporarily disable multi-rate retry (link quality) and eliminate rate index lookups. * My recent(ish) change to iwn(4) and the net80211 rate control API to support 11n rates broke the link quality table use. So, until I or someone else decides to fix it, let's just disable it for now. * Teach iwn_tx_data_raw() to use the iwn_rate_to_plcp() function. * Eliminate two uses of the net80211 rate index lookup functions - they are only for legacy rates and they're not needed here. This fixes some invalid looking rate control TX issues that showed up on my 4965 but it doesn't fix the two TX hangs I've noticed. Those look like DMA related issues. Tested: * 4965, STA mode * 5100, STA mode
2013-10-25 19:44:53 +00:00
tx->rate = iwn_rate_to_plcp(sc, ni, rate);
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Group or management frame. */
tx->linkq = 0;
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Set physical address of "scratch area". */
scratch_paddr has the same address pre-assigned, use that instead.
2011-04-16 09:39:08 +00:00
tx->loaddr = htole32(IWN_LOADDR(data->scratch_paddr));
tx->hiaddr = IWN_HIADDR(data->scratch_paddr);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Copy 802.11 header in TX command. */
memcpy((uint8_t *)(tx + 1), wh, hdrlen);
/* Trim 802.11 header. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
m_adj(m, hdrlen);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
tx->security = 0;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
tx->flags = htole32(flags);
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
error = bus_dmamap_load_mbuf_sg(ring->data_dmat, data->map, m, segs,
&nsegs, BUS_DMA_NOWAIT);
if (error != 0) {
if (error != EFBIG) {
device_printf(sc->sc_dev,
"%s: can't map mbuf (error %d)\n", __func__, error);
m_freem(m);
return error;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
/* Too many DMA segments, linearize mbuf. */
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
m1 = m_collapse(m, M_NOWAIT, IWN_MAX_SCATTER - 1);
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
if (m1 == NULL) {
device_printf(sc->sc_dev,
"%s: could not defrag mbuf\n", __func__);
m_freem(m);
return ENOBUFS;
}
m = m1;
error = bus_dmamap_load_mbuf_sg(ring->data_dmat, data->map, m,
segs, &nsegs, BUS_DMA_NOWAIT);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
"%s: can't map mbuf (error %d)\n", __func__, error);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
m_freem(m);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
}
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
data->m = m;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
data->ni = ni;
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: qid %d idx %d len %d nsegs %d\n",
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
__func__, ring->qid, ring->cur, m->m_pkthdr.len, nsegs);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Fill TX descriptor. */
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
desc->nsegs = 1;
if (m->m_len != 0)
desc->nsegs += nsegs;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* First DMA segment is used by the TX command. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
desc->segs[0].addr = htole32(IWN_LOADDR(data->cmd_paddr));
desc->segs[0].len = htole16(IWN_HIADDR(data->cmd_paddr) |
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
(4 + sizeof (*tx) + hdrlen + pad) << 4);
/* Other DMA segments are for data payload. */
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
seg = &segs[0];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
for (i = 1; i <= nsegs; i++) {
Rewrite DMA segment handling to be more inline with the OpenBSD code. Also change the m_len == 0 hack to have less code churn.
2011-04-16 10:32:46 +00:00
desc->segs[i].addr = htole32(IWN_LOADDR(seg->ds_addr));
desc->segs[i].len = htole16(IWN_HIADDR(seg->ds_addr) |
seg->ds_len << 4);
seg++;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(ring->data_dmat, data->map, BUS_DMASYNC_PREWRITE);
bus_dmamap_sync(ring->data_dmat, ring->cmd_dma.map,
BUS_DMASYNC_PREWRITE);
bus_dmamap_sync(ring->desc_dma.tag, ring->desc_dma.map,
BUS_DMASYNC_PREWRITE);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Update TX scheduler. */
Only update the scheduler's byte count table for aggregation queues. The other queues, especially the command queue, uses the FIFO mode which doesn't require the byte count table because queued entries are processed in order. Pointed out by: Lucius Windschuh <lwindschuh at googlemail dot com>
2011-05-15 08:09:36 +00:00
if (ring->qid >= sc->firstaggqueue)
ops->update_sched(sc, ring->qid, ring->cur, tx->id, totlen);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Kick TX ring. */
ring->cur = (ring->cur + 1) % IWN_TX_RING_COUNT;
IWN_WRITE(sc, IWN_HBUS_TARG_WRPTR, ring->qid << 8 | ring->cur);
/* Mark TX ring as full if we reach a certain threshold. */
if (++ring->queued > IWN_TX_RING_HIMARK)
sc->qfullmsk |= 1 << ring->qid;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return 0;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
static void
iwn_xmit_task(void *arg0, int pending)
{
struct iwn_softc *sc = arg0;
struct ieee80211_node *ni;
struct mbuf *m;
int error;
struct ieee80211_bpf_params p;
int have_p;
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: called\n", __func__);
IWN_LOCK(sc);
/*
* Dequeue frames, attempt to transmit,
* then disable beaconwait when we're done.
*/
while ((m = mbufq_dequeue(&sc->sc_xmit_queue)) != NULL) {
have_p = 0;
ni = (struct ieee80211_node *)m->m_pkthdr.rcvif;
/* Get xmit params if appropriate */
if (ieee80211_get_xmit_params(m, &p) == 0)
have_p = 1;
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: m=%p, have_p=%d\n",
__func__, m, have_p);
/* If we have xmit params, use them */
if (have_p)
error = iwn_tx_data_raw(sc, m, ni, &p);
else
error = iwn_tx_data(sc, m, ni);
if (error != 0) {
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
if_inc_counter(ni->ni_vap->iv_ifp,
IFCOUNTER_OERRORS, 1);
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
ieee80211_free_node(ni);
}
}
sc->sc_beacon_wait = 0;
IWN_UNLOCK(sc);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
static int
iwn_raw_xmit(struct ieee80211_node *ni, struct mbuf *m,
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
const struct ieee80211_bpf_params *params)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
struct ieee80211com *ic = ni->ni_ic;
Migrate ifp->if_softc -> ic->ic_softc. Tested: * Lenovo T400 (Intel 5300) * make universe
2015-08-17 23:35:31 +00:00
struct iwn_softc *sc = ic->ic_softc;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
int error = 0;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
DPRINTF(sc, IWN_DEBUG_XMIT | IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
if ((sc->sc_flags & IWN_FLAG_RUNNING) == 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ieee80211_free_node(ni);
m_freem(m);
return ENETDOWN;
}
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
/* XXX? net80211 doesn't set this on xmit'ed raw frames? */
m->m_pkthdr.rcvif = (void *) ni;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
IWN_LOCK(sc);
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
/* queue frame if we have to */
if (sc->sc_beacon_wait) {
if (iwn_xmit_queue_enqueue(sc, m) != 0) {
m_freem(m);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
if_inc_counter(ni->ni_vap->iv_ifp,
IFCOUNTER_OERRORS, 1);
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
ieee80211_free_node(ni);
IWN_UNLOCK(sc);
return (ENOBUFS);
}
/* Queued, so just return OK */
IWN_UNLOCK(sc);
return (0);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (params == NULL) {
/*
* Legacy path; interpret frame contents to decide
* precisely how to send the frame.
*/
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
error = iwn_tx_data(sc, m, ni);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
} else {
/*
* Caller supplied explicit parameters to use in
* sending the frame.
*/
Fix WME/QoS handling: - move the TX queue selection into iwn_tx_data/iwn_tx_data_raw - extract traffic identifier and use it - do not expect ACKs for frames marked as such
2011-04-16 13:25:19 +00:00
error = iwn_tx_data_raw(sc, m, ni, params);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (error != 0) {
/* NB: m is reclaimed on tx failure */
ieee80211_free_node(ni);
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
} else
sc->sc_tx_timer = 5;
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_UNLOCK(sc);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE | IWN_DEBUG_XMIT, "->%s: end\n",__func__);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
static int
iwn_transmit(struct ieee80211com *ic, struct mbuf *m)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
{
Remove the software queue, which is a remnant of ifnet ifqueue. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-09-02 15:23:51 +00:00
struct iwn_softc *sc = ic->ic_softc;
struct ieee80211_node *ni;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
int error;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_LOCK(sc);
Remove the software queue, which is a remnant of ifnet ifqueue. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-09-02 15:23:51 +00:00
if ((sc->sc_flags & IWN_FLAG_RUNNING) == 0 || sc->sc_beacon_wait) {
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
IWN_UNLOCK(sc);
return (ENXIO);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Remove the software queue, which is a remnant of ifnet ifqueue. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-09-02 15:23:51 +00:00
if (sc->qfullmsk) {
IWN_UNLOCK(sc);
return (ENOBUFS);
First cut at attempting to buffer frames until we see a beacon. The iwn(4) firmware forgets most of its channel state after an RXON command. This means that any beacons its seen on passive 5GHz channels are forgotten upon an association/authorisation request. This unfortuantely means that 5GHz association almost always fails - the assoc and/or auth frames are dropped with a status of "passive channel, haven't seen a beacon yet." (0x90.) So: * add an xmit queue, global, to buffer frames * modify the xmit path to use the mbuf tag from net80211 to specify raw frame details * buffer xmit frames from both raw and non-raw paths * if a beacon is seen in the RX path, schedule a taskqueue to send said frames and un-buffer things. * flush frames during state change back to INIT, or NIC down/up/detach. This isn't the final shape I'd like this to be in but it certainly is better than 5GHz "not working at all". Tested: * Intel 5100, STA mode (before spilling coffee) * Intel 5300, STA mode (after spilling coffee) Story: * This has been bugging me at work for months, which I just worked around by throwing an ath(4) into my Lenovo T400 cardbus slot. * Our ops director discovered indeed FreeBSD runs well on the Lenovo T420p, except for that pesky 5GHz thing. So now developers also can have a T420p running FreeBSD to do work with. Their #1 feedback to me - "boy it'd be nice if 5GHz wifi worked." * .. then, I was at NANOG but stuck with 5GHz only wifi and no ath(4) NIC to put in a laptop - and I snapped. Thus, the reason this is actually work related. MFC after: 2 weeks Sponsored by: Norse Corp, Inc.
2015-06-19 01:44:17 +00:00
}
Remove the software queue, which is a remnant of ifnet ifqueue. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-09-02 15:23:51 +00:00
ni = (struct ieee80211_node *)m->m_pkthdr.rcvif;
error = iwn_tx_data(sc, m, ni);
if (error) {
if_inc_counter(ni->ni_vap->iv_ifp, IFCOUNTER_OERRORS, 1);
ieee80211_free_node(ni);
} else
sc->sc_tx_timer = 5;
IWN_UNLOCK(sc);
return (error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
static void
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
iwn_watchdog(void *arg)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
{
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
struct iwn_softc *sc = arg;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
IWN_LOCK_ASSERT(sc);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
KASSERT(sc->sc_flags & IWN_FLAG_RUNNING, ("not running"));
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
remove debug left-overs
2011-04-15 20:19:18 +00:00
if (sc->sc_tx_timer > 0) {
if (--sc->sc_tx_timer == 0) {
Use ic_printf() instead of if_printf().
2015-05-25 15:12:23 +00:00
ic_printf(ic, "device timeout\n");
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
ieee80211_runtask(ic, &sc->sc_reinit_task);
return;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
callout_reset(&sc->watchdog_to, hz, iwn_watchdog, sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
iwn_cdev_open(struct cdev *dev, int flags, int type, struct thread *td)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
return (0);
}
static int
iwn_cdev_close(struct cdev *dev, int flags, int type, struct thread *td)
{
return (0);
}
static int
iwn_cdev_ioctl(struct cdev *dev, unsigned long cmd, caddr_t data, int fflag,
struct thread *td)
{
int rc;
struct iwn_softc *sc = dev->si_drv1;
struct iwn_ioctl_data *d;
rc = priv_check(td, PRIV_DRIVER);
if (rc != 0)
return (0);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
switch (cmd) {
Track and expose the latest statistics from the firmware. Tested: * Intel Centrino 6205
2014-02-24 02:37:04 +00:00
case SIOCGIWNSTATS:
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
d = (struct iwn_ioctl_data *) data;
Track and expose the latest statistics from the firmware. Tested: * Intel Centrino 6205
2014-02-24 02:37:04 +00:00
IWN_LOCK(sc);
/* XXX validate permissions/memory/etc? */
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
rc = copyout(&sc->last_stat, d->dst_addr, sizeof(struct iwn_stats));
Track and expose the latest statistics from the firmware. Tested: * Intel Centrino 6205
2014-02-24 02:37:04 +00:00
IWN_UNLOCK(sc);
break;
case SIOCZIWNSTATS:
IWN_LOCK(sc);
memset(&sc->last_stat, 0, sizeof(struct iwn_stats));
IWN_UNLOCK(sc);
break;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
default:
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
rc = EINVAL;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
break;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
return (rc);
}
static int
iwn_ioctl(struct ieee80211com *ic, u_long cmd, void *data)
{
return (ENOTTY);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
}
static void
iwn_parent(struct ieee80211com *ic)
{
struct iwn_softc *sc = ic->ic_softc;
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
int startall = 0, stop = 0;
IWN_LOCK(sc);
if (ic->ic_nrunning > 0) {
if (!(sc->sc_flags & IWN_FLAG_RUNNING)) {
iwn_init_locked(sc);
if (IWN_READ(sc, IWN_GP_CNTRL) & IWN_GP_CNTRL_RFKILL)
startall = 1;
else
stop = 1;
}
} else if (sc->sc_flags & IWN_FLAG_RUNNING)
iwn_stop_locked(sc);
IWN_UNLOCK(sc);
if (startall)
ieee80211_start_all(ic);
else if (vap != NULL && stop)
ieee80211_stop(vap);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
/*
* Send a command to the firmware.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_cmd(struct iwn_softc *sc, int code, const void *buf, int size, int async)
{
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
struct iwn_tx_ring *ring;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_tx_desc *desc;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_tx_data *data;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_tx_cmd *cmd;
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
struct mbuf *m;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
bus_addr_t paddr;
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
int totlen, error;
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
int cmd_queue_num;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
if (async == 0)
IWN_LOCK_ASSERT(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->sc_flags & IWN_FLAG_PAN_SUPPORT)
cmd_queue_num = IWN_PAN_CMD_QUEUE;
else
cmd_queue_num = IWN_CMD_QUEUE_NUM;
ring = &sc->txq[cmd_queue_num];
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
desc = &ring->desc[ring->cur];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
data = &ring->data[ring->cur];
totlen = 4 + size;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
if (size > sizeof cmd->data) {
/* Command is too large to fit in a descriptor. */
if (totlen > MCLBYTES)
return EINVAL;
Mechanically substitute flags from historic mbuf allocator with malloc(9) flags in sys/dev.
2012-12-04 09:32:43 +00:00
m = m_getjcl(M_NOWAIT, MT_DATA, M_PKTHDR, MJUMPAGESIZE);
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
if (m == NULL)
return ENOMEM;
cmd = mtod(m, struct iwn_tx_cmd *);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
error = bus_dmamap_load(ring->data_dmat, data->map, cmd,
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
totlen, iwn_dma_map_addr, &paddr, BUS_DMA_NOWAIT);
if (error != 0) {
m_freem(m);
return error;
}
data->m = m;
} else {
cmd = &ring->cmd[ring->cur];
paddr = data->cmd_paddr;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
cmd->code = code;
cmd->flags = 0;
cmd->qid = ring->qid;
cmd->idx = ring->cur;
memcpy(cmd->data, buf, size);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
desc->nsegs = 1;
desc->segs[0].addr = htole32(IWN_LOADDR(paddr));
desc->segs[0].len = htole16(IWN_HIADDR(paddr) | totlen << 4);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_CMD, "%s: %s (0x%x) flags %d qid %d idx %d\n",
__func__, iwn_intr_str(cmd->code), cmd->code,
cmd->flags, cmd->qid, cmd->idx);
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
if (size > sizeof cmd->data) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(ring->data_dmat, data->map,
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
BUS_DMASYNC_PREWRITE);
} else {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
bus_dmamap_sync(ring->data_dmat, ring->cmd_dma.map,
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
BUS_DMASYNC_PREWRITE);
}
bus_dmamap_sync(ring->desc_dma.tag, ring->desc_dma.map,
BUS_DMASYNC_PREWRITE);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Kick command ring. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
ring->cur = (ring->cur + 1) % IWN_TX_RING_COUNT;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_HBUS_TARG_WRPTR, ring->qid << 8 | ring->cur);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
return async ? 0 : msleep(desc, &sc->sc_mtx, PCATCH, "iwncmd", hz);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_add_node(struct iwn_softc *sc, struct iwn_node_info *node, int async)
{
struct iwn4965_node_info hnode;
caddr_t src, dst;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* We use the node structure for 5000 Series internally (it is
* a superset of the one for 4965AGN). We thus copy the common
* fields before sending the command.
*/
src = (caddr_t)node;
dst = (caddr_t)&hnode;
memcpy(dst, src, 48);
/* Skip TSC, RX MIC and TX MIC fields from ``src''. */
memcpy(dst + 48, src + 72, 20);
return iwn_cmd(sc, IWN_CMD_ADD_NODE, &hnode, sizeof hnode, async);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_add_node(struct iwn_softc *sc, struct iwn_node_info *node, int async)
{
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Direct mapping. */
return iwn_cmd(sc, IWN_CMD_ADD_NODE, node, sizeof (*node), async);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
iwn_set_link_quality(struct iwn_softc *sc, struct ieee80211_node *ni)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
struct iwn_node *wn = (void *)ni;
Use the negotiated HT rate set when generating the link quality table.
2013-11-12 05:00:18 +00:00
struct ieee80211_rateset *rs;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_cmd_link_quality linkq;
Prepare for transmitting frames at MCS rates: - instead of calling iwn_plcp_signal() for every frame, map the expected value directly within wn->ridx - concat plcp, rflags and xrflags, there is no clean byte boundary within the flags, for example the antenna setting uses bit 6, 7 and 8 - there is still need for a custom rate to plcp mapping, as those expected by the hardware are not conform to the std
2011-05-08 11:49:50 +00:00
int i, rate, txrate;
Don't base the rate table selection based on the channel mode; it needs to check whether there are rate entries in there or not. PR: kern/183428
2013-10-31 02:21:48 +00:00
int is_11n;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
memset(&linkq, 0, sizeof linkq);
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
linkq.id = wn->id;
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
linkq.antmsk_1stream = iwn_get_1stream_tx_antmask(sc);
linkq.antmsk_2stream = iwn_get_2stream_tx_antmask(sc);
Correctly initialise the 2-chain antenna mask in the link quality table. The previous code simply hard-coded IWN_ANT_AB which is only correct for some of the NICs. Now, if the NIC is a 1-stream TX, you need to set IWN_ANT_AB and _not_ just a single antenna. The Intel 5100 firmware panics the moment the link quality table is updated. So! * no secondary antenna? Set it to IWN_ANT_AB; * two-stream device? Transmit on the full transmit antenna configuration. Tested: * Intel 5100, STA * Intel 2200 (eadler) Obtained from: Linux iwlwifi
2013-11-13 07:09:00 +00:00
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
linkq.ampdu_max = 32; /* XXX negotiated? */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
linkq.ampdu_threshold = 3;
linkq.ampdu_limit = htole16(4000); /* 4ms */
Correctly initialise the 2-chain antenna mask in the link quality table. The previous code simply hard-coded IWN_ANT_AB which is only correct for some of the NICs. Now, if the NIC is a 1-stream TX, you need to set IWN_ANT_AB and _not_ just a single antenna. The Intel 5100 firmware panics the moment the link quality table is updated. So! * no secondary antenna? Set it to IWN_ANT_AB; * two-stream device? Transmit on the full transmit antenna configuration. Tested: * Intel 5100, STA * Intel 2200 (eadler) Obtained from: Linux iwlwifi
2013-11-13 07:09:00 +00:00
DPRINTF(sc, IWN_DEBUG_XMIT,
"%s: 1stream antenna=0x%02x, 2stream antenna=0x%02x, ntxstreams=%d\n",
__func__,
linkq.antmsk_1stream,
linkq.antmsk_2stream,
sc->ntxchains);
Don't base the rate table selection based on the channel mode; it needs to check whether there are rate entries in there or not. PR: kern/183428
2013-10-31 02:21:48 +00:00
/*
* Are we using 11n rates? Ensure the channel is
* 11n _and_ we have some 11n rates, or don't
* try.
*/
Use the negotiated HT rate set when generating the link quality table.
2013-11-12 05:00:18 +00:00
if (IEEE80211_IS_CHAN_HT(ni->ni_chan) && ni->ni_htrates.rs_nrates > 0) {
rs = (struct ieee80211_rateset *) &ni->ni_htrates;
Don't base the rate table selection based on the channel mode; it needs to check whether there are rate entries in there or not. PR: kern/183428
2013-10-31 02:21:48 +00:00
is_11n = 1;
Use the negotiated HT rate set when generating the link quality table.
2013-11-12 05:00:18 +00:00
} else {
rs = &ni->ni_rates;
Don't base the rate table selection based on the channel mode; it needs to check whether there are rate entries in there or not. PR: kern/183428
2013-10-31 02:21:48 +00:00
is_11n = 0;
Use the negotiated HT rate set when generating the link quality table.
2013-11-12 05:00:18 +00:00
}
Don't base the rate table selection based on the channel mode; it needs to check whether there are rate entries in there or not. PR: kern/183428
2013-10-31 02:21:48 +00:00
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
/* Start at highest available bit-rate. */
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
/*
* XXX this is all very dirty!
*/
Don't base the rate table selection based on the channel mode; it needs to check whether there are rate entries in there or not. PR: kern/183428
2013-10-31 02:21:48 +00:00
if (is_11n)
Add support for transmitting frames at MCS rates.
2011-05-08 11:54:38 +00:00
txrate = ni->ni_htrates.rs_nrates - 1;
else
txrate = rs->rs_nrates - 1;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (i = 0; i < IWN_MAX_TX_RETRIES; i++) {
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
uint32_t plcp;
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
/*
* XXX TODO: ensure the last two slots are the two lowest
* rate entries, just for now.
*/
if (i == 14 || i == 15)
txrate = 0;
Don't base the rate table selection based on the channel mode; it needs to check whether there are rate entries in there or not. PR: kern/183428
2013-10-31 02:21:48 +00:00
if (is_11n)
Use the negotiated HT rate set when generating the link quality table.
2013-11-12 05:00:18 +00:00
rate = IEEE80211_RATE_MCS | rs->rs_rates[txrate];
Add support for transmitting frames at MCS rates.
2011-05-08 11:54:38 +00:00
else
net80211 & wireless drivers: remove duplicate defines (noop) * IEEE80211_DIR_DSTODS(wh) -> IEEE80211_IS_DSTODS(wh). * N(a) -> nitems(a). * Remove LE_READ_2(p)/LE_READ_4(p) definitions (and include ieee80211_input.h instead). * <drvname>_TXOP_TO_US(txop) -> IEEE80211_TXOP_TO_US(txop). * Put IEEE80211_RV(v) into ieee80211_proto.h and remove local RV(v) definitions. Submitted by: Andriy Voskoboinyk <s3erios@gmail.com> Differential Revision: https://reviews.freebsd.org/D3705
2015-09-22 02:44:59 +00:00
rate = IEEE80211_RV(rs->rs_rates[txrate]);
Prepare for transmitting frames at MCS rates: - instead of calling iwn_plcp_signal() for every frame, map the expected value directly within wn->ridx - concat plcp, rflags and xrflags, there is no clean byte boundary within the flags, for example the antenna setting uses bit 6, 7 and 8 - there is still need for a custom rate to plcp mapping, as those expected by the hardware are not conform to the std
2011-05-08 11:49:50 +00:00
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
/* Do rate -> PLCP config mapping */
plcp = iwn_rate_to_plcp(sc, ni, rate);
linkq.retry[i] = plcp;
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_XMIT,
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
"%s: i=%d, txrate=%d, rate=0x%02x, plcp=0x%08x\n",
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
__func__,
i,
txrate,
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
rate,
le32toh(plcp));
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
Comment what 'mimo' does in the link quality table.
2013-11-12 04:57:31 +00:00
/*
* The mimo field is an index into the table which
* indicates the first index where it and subsequent entries
* will not be using MIMO.
*
* Since we're filling linkq from 0..15 and we're filling
* from the higest MCS rates to the lowest rates, if we
* _are_ doing a dual-stream rate, set mimo to idx+1 (ie,
* the next entry.) That way if the next entry is a non-MIMO
* entry, we're already pointing at it.
*/
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
if ((le32toh(plcp) & IWN_RFLAG_MCS) &&
net80211 & wireless drivers: remove duplicate defines (noop) * IEEE80211_DIR_DSTODS(wh) -> IEEE80211_IS_DSTODS(wh). * N(a) -> nitems(a). * Remove LE_READ_2(p)/LE_READ_4(p) definitions (and include ieee80211_input.h instead). * <drvname>_TXOP_TO_US(txop) -> IEEE80211_TXOP_TO_US(txop). * Put IEEE80211_RV(v) into ieee80211_proto.h and remove local RV(v) definitions. Submitted by: Andriy Voskoboinyk <s3erios@gmail.com> Differential Revision: https://reviews.freebsd.org/D3705
2015-09-22 02:44:59 +00:00
IEEE80211_RV(le32toh(plcp)) > 7)
Add support for transmitting frames at MCS rates.
2011-05-08 11:54:38 +00:00
linkq.mimo = i + 1;
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
/* Next retry at immediate lower bit-rate. */
if (txrate > 0)
txrate--;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
/*
* If we reached the end of the list and indeed we hit
* all MIMO rates (eg 5300 doing MCS23-15) then yes,
* set mimo to 15. Setting it to 16 panics the firmware.
*/
if (linkq.mimo > 15)
linkq.mimo = 15;
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: mimo = %d\n", __func__, linkq.mimo);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
return iwn_cmd(sc, IWN_CMD_LINK_QUALITY, &linkq, sizeof linkq, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
/*
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
* Broadcast node is used to send group-addressed and management frames.
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_add_broadcast_node(struct iwn_softc *sc, int async)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_node_info node;
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
struct iwn_cmd_link_quality linkq;
uint8_t txant;
int i, error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon = &sc->rx_on[IWN_RXON_BSS_CTX];
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
memset(&node, 0, sizeof node);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
IEEE80211_ADDR_COPY(node.macaddr, ieee80211broadcastaddr);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
node.id = sc->broadcast_id;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DPRINTF(sc, IWN_DEBUG_RESET, "%s: adding broadcast node\n", __func__);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
if ((error = ops->add_node(sc, &node, async)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
/* Use the first valid TX antenna. */
txant = IWN_LSB(sc->txchainmask);
memset(&linkq, 0, sizeof linkq);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
linkq.id = sc->broadcast_id;
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
linkq.antmsk_1stream = iwn_get_1stream_tx_antmask(sc);
linkq.antmsk_2stream = iwn_get_2stream_tx_antmask(sc);
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
linkq.ampdu_max = 64;
linkq.ampdu_threshold = 3;
linkq.ampdu_limit = htole16(4000); /* 4ms */
/* Use lowest mandatory bit-rate. */
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
/* XXX rate table lookup? */
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
if (IEEE80211_IS_CHAN_5GHZ(ic->ic_curchan))
Prepare for transmitting frames at MCS rates: - instead of calling iwn_plcp_signal() for every frame, map the expected value directly within wn->ridx - concat plcp, rflags and xrflags, there is no clean byte boundary within the flags, for example the antenna setting uses bit 6, 7 and 8 - there is still need for a custom rate to plcp mapping, as those expected by the hardware are not conform to the std
2011-05-08 11:49:50 +00:00
linkq.retry[0] = htole32(0xd);
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
else
Prepare for transmitting frames at MCS rates: - instead of calling iwn_plcp_signal() for every frame, map the expected value directly within wn->ridx - concat plcp, rflags and xrflags, there is no clean byte boundary within the flags, for example the antenna setting uses bit 6, 7 and 8 - there is still need for a custom rate to plcp mapping, as those expected by the hardware are not conform to the std
2011-05-08 11:49:50 +00:00
linkq.retry[0] = htole32(10 | IWN_RFLAG_CCK);
linkq.retry[0] |= htole32(IWN_RFLAG_ANT(txant));
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
/* Use same bit-rate for all TX retries. */
for (i = 1; i < IWN_MAX_TX_RETRIES; i++) {
Prepare for transmitting frames at MCS rates: - instead of calling iwn_plcp_signal() for every frame, map the expected value directly within wn->ridx - concat plcp, rflags and xrflags, there is no clean byte boundary within the flags, for example the antenna setting uses bit 6, 7 and 8 - there is still need for a custom rate to plcp mapping, as those expected by the hardware are not conform to the std
2011-05-08 11:49:50 +00:00
linkq.retry[i] = linkq.retry[0];
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
return iwn_cmd(sc, IWN_CMD_LINK_QUALITY, &linkq, sizeof linkq, async);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Rename some stuff in favour of the OpenBSD names: - prefer EDCA over WME - qid for a TXQ ID - reg for register values
2011-04-16 13:40:32 +00:00
iwn_updateedca(struct ieee80211com *ic)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
#define IWN_EXP2(x) ((1 << (x)) - 1) /* CWmin = 2^ECWmin - 1 */
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
struct iwn_softc *sc = ic->ic_softc;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_edca_params cmd;
Rename some stuff in favour of the OpenBSD names: - prefer EDCA over WME - qid for a TXQ ID - reg for register values
2011-04-16 13:40:32 +00:00
int aci;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
memset(&cmd, 0, sizeof cmd);
cmd.flags = htole32(IWN_EDCA_UPDATE);
Rename some stuff in favour of the OpenBSD names: - prefer EDCA over WME - qid for a TXQ ID - reg for register values
2011-04-16 13:40:32 +00:00
for (aci = 0; aci < WME_NUM_AC; aci++) {
const struct wmeParams *ac =
&ic->ic_wme.wme_chanParams.cap_wmeParams[aci];
cmd.ac[aci].aifsn = ac->wmep_aifsn;
cmd.ac[aci].cwmin = htole16(IWN_EXP2(ac->wmep_logcwmin));
cmd.ac[aci].cwmax = htole16(IWN_EXP2(ac->wmep_logcwmax));
cmd.ac[aci].txoplimit =
htole16(IEEE80211_TXOP_TO_US(ac->wmep_txopLimit));
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IEEE80211_UNLOCK(ic);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
IWN_LOCK(sc);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
(void)iwn_cmd(sc, IWN_CMD_EDCA_PARAMS, &cmd, sizeof cmd, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
IWN_UNLOCK(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IEEE80211_LOCK(ic);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return 0;
#undef IWN_EXP2
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
static void
Change three methods in struct ieee80211com, namely ic_updateslot, ic_update_mcast and ic_update_promisc, to pass pointer to the ieee80211com, not to the ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 19:53:29 +00:00
iwn_update_mcast(struct ieee80211com *ic)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
{
/* Ignore */
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_set_led(struct iwn_softc *sc, uint8_t which, uint8_t off, uint8_t on)
{
struct iwn_cmd_led led;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
#if 0
/* XXX don't set LEDs during scan? */
if (sc->sc_is_scanning)
return;
#endif
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Clear microcode LED ownership. */
IWN_CLRBITS(sc, IWN_LED, IWN_LED_BSM_CTRL);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
led.which = which;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
led.unit = htole32(10000); /* on/off in unit of 100ms */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
led.off = off;
led.on = on;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
(void)iwn_cmd(sc, IWN_CMD_SET_LED, &led, sizeof led, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
/*
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
* Set the critical temperature at which the firmware will stop the radio
* and notify us.
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_set_critical_temp(struct iwn_softc *sc)
{
struct iwn_critical_temp crit;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
int32_t temp;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_UCODE_GP1_CLR, IWN_UCODE_GP1_CTEMP_STOP_RF);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (sc->hw_type == IWN_HW_REV_TYPE_5150)
temp = (IWN_CTOK(110) - sc->temp_off) * -5;
else if (sc->hw_type == IWN_HW_REV_TYPE_4965)
temp = IWN_CTOK(110);
else
temp = 110;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
memset(&crit, 0, sizeof crit);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
crit.tempR = htole32(temp);
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
DPRINTF(sc, IWN_DEBUG_RESET, "setting critical temp to %d\n", temp);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return iwn_cmd(sc, IWN_CMD_SET_CRITICAL_TEMP, &crit, sizeof crit, 0);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_set_timing(struct iwn_softc *sc, struct ieee80211_node *ni)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_cmd_timing cmd;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
uint64_t val, mod;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
memset(&cmd, 0, sizeof cmd);
memcpy(&cmd.tstamp, ni->ni_tstamp.data, sizeof (uint64_t));
cmd.bintval = htole16(ni->ni_intval);
cmd.lintval = htole16(10);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Compute remaining time until next beacon. */
Reuse net80211 code: - IWN_TXOP_TO_US is equal to IEEE80211_TXOP_TO_US - use IEEE80211_DUR_TU - ieee80211_add_rates/ieee80211_add_xrates are public, use em - copied ieee80211_add_ssid it is not public
2011-04-14 17:26:13 +00:00
val = (uint64_t)ni->ni_intval * IEEE80211_DUR_TU;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
mod = le64toh(cmd.tstamp) % val;
cmd.binitval = htole32((uint32_t)(val - mod));
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DPRINTF(sc, IWN_DEBUG_RESET, "timing bintval=%u tstamp=%ju, init=%u\n",
ni->ni_intval, le64toh(cmd.tstamp), (uint32_t)(val - mod));
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return iwn_cmd(sc, IWN_CMD_TIMING, &cmd, sizeof cmd, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_power_calibration(struct iwn_softc *sc, int temp)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Revert iwn channel-argument handling of set_txpower() to r201822 Pass the channel argument as a real argument and not through RXON. The RXON version seems to have problem with both the older 228.57.2.23 firmware and the latest 228.61.2.24 version resulting in device initialization errors like: wpa_supplicant[2928]: Failed to initiate AP scan. kernel: firmware error log: kernel: error type = "SYSASSERT" (0x00000005) kernel: program counter = 0x0000147C kernel: source line = 0x0000058B kernel: error data = 0x0000058B00000000 kernel: branch link = 0x0000145A00001492 kernel: interrupt link = 0x000006DE00000000 kernel: time = 7310 ... kernel: iwn0: iwn_config: could not set TX power kernel: iwn0: iwn_init_locked: could not configure device, error 35 By passing the current channel to hal->set_txpower() the firmware error is fixed, at least for the 4965 chipset of my Thinkpad and the ones tested by Bernhard. Submitted by: Bernhard Schmidt <bschmidt@techwires.net> Approved by: rpaulo
2010-01-09 04:00:22 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Adjust TX power if need be (delta >= 3 degC). */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE, "%s: temperature %d->%d\n",
__func__, sc->temp, temp);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (abs(temp - sc->temp) >= 3) {
/* Record temperature of last calibration. */
sc->temp = temp;
Revert iwn channel-argument handling of set_txpower() to r201822 Pass the channel argument as a real argument and not through RXON. The RXON version seems to have problem with both the older 228.57.2.23 firmware and the latest 228.61.2.24 version resulting in device initialization errors like: wpa_supplicant[2928]: Failed to initiate AP scan. kernel: firmware error log: kernel: error type = "SYSASSERT" (0x00000005) kernel: program counter = 0x0000147C kernel: source line = 0x0000058B kernel: error data = 0x0000058B00000000 kernel: branch link = 0x0000145A00001492 kernel: interrupt link = 0x000006DE00000000 kernel: time = 7310 ... kernel: iwn0: iwn_config: could not set TX power kernel: iwn0: iwn_init_locked: could not configure device, error 35 By passing the current channel to hal->set_txpower() the firmware error is fixed, at least for the 4965 chipset of my Thinkpad and the ones tested by Bernhard. Submitted by: Bernhard Schmidt <bschmidt@techwires.net> Approved by: rpaulo
2010-01-09 04:00:22 +00:00
(void)iwn4965_set_txpower(sc, ic->ic_bsschan, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
}
/*
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
* Set TX power for current channel (each rate has its own power settings).
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
* This function takes into account the regulatory information from EEPROM,
* the current temperature and the current voltage.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Revert iwn channel-argument handling of set_txpower() to r201822 Pass the channel argument as a real argument and not through RXON. The RXON version seems to have problem with both the older 228.57.2.23 firmware and the latest 228.61.2.24 version resulting in device initialization errors like: wpa_supplicant[2928]: Failed to initiate AP scan. kernel: firmware error log: kernel: error type = "SYSASSERT" (0x00000005) kernel: program counter = 0x0000147C kernel: source line = 0x0000058B kernel: error data = 0x0000058B00000000 kernel: branch link = 0x0000145A00001492 kernel: interrupt link = 0x000006DE00000000 kernel: time = 7310 ... kernel: iwn0: iwn_config: could not set TX power kernel: iwn0: iwn_init_locked: could not configure device, error 35 By passing the current channel to hal->set_txpower() the firmware error is fixed, at least for the 4965 chipset of my Thinkpad and the ones tested by Bernhard. Submitted by: Bernhard Schmidt <bschmidt@techwires.net> Approved by: rpaulo
2010-01-09 04:00:22 +00:00
iwn4965_set_txpower(struct iwn_softc *sc, struct ieee80211_channel *ch,
int async)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Fixed-point arithmetic division using a n-bit fractional part. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
#define fdivround(a, b, n) \
((((1 << n) * (a)) / (b) + (1 << n) / 2) / (1 << n))
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Linear interpolation. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
#define interpolate(x, x1, y1, x2, y2, n) \
((y1) + fdivround(((int)(x) - (x1)) * ((y2) - (y1)), (x2) - (x1), n))
static const int tdiv[IWN_NATTEN_GROUPS] = { 9, 8, 8, 8, 6 };
struct iwn_ucode_info *uc = &sc->ucode_info;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn4965_cmd_txpower cmd;
struct iwn4965_eeprom_chan_samples *chans;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
const uint8_t *rf_gain, *dsp_gain;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
int32_t vdiff, tdiff;
int i, c, grp, maxpwr;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint8_t chan;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon = &sc->rx_on[IWN_RXON_BSS_CTX];
Obtain the channel number directly from the laster RXON command, also chan is an uint8_t.
2011-04-16 07:17:03 +00:00
/* Retrieve current channel from last RXON. */
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
chan = sc->rxon->chan;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
DPRINTF(sc, IWN_DEBUG_RESET, "setting TX power for channel %d\n",
chan);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
memset(&cmd, 0, sizeof cmd);
cmd.band = IEEE80211_IS_CHAN_5GHZ(ch) ? 0 : 1;
cmd.chan = chan;
if (IEEE80211_IS_CHAN_5GHZ(ch)) {
maxpwr = sc->maxpwr5GHz;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
rf_gain = iwn4965_rf_gain_5ghz;
dsp_gain = iwn4965_dsp_gain_5ghz;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
} else {
maxpwr = sc->maxpwr2GHz;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
rf_gain = iwn4965_rf_gain_2ghz;
dsp_gain = iwn4965_dsp_gain_2ghz;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Compute voltage compensation. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
vdiff = ((int32_t)le32toh(uc->volt) - sc->eeprom_voltage) / 7;
if (vdiff > 0)
vdiff *= 2;
if (abs(vdiff) > 2)
vdiff = 0;
DPRINTF(sc, IWN_DEBUG_CALIBRATE | IWN_DEBUG_TXPOW,
"%s: voltage compensation=%d (UCODE=%d, EEPROM=%d)\n",
__func__, vdiff, le32toh(uc->volt), sc->eeprom_voltage);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Get channel attenuation group. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (chan <= 20) /* 1-20 */
grp = 4;
else if (chan <= 43) /* 34-43 */
grp = 0;
else if (chan <= 70) /* 44-70 */
grp = 1;
else if (chan <= 124) /* 71-124 */
grp = 2;
else /* 125-200 */
grp = 3;
DPRINTF(sc, IWN_DEBUG_CALIBRATE | IWN_DEBUG_TXPOW,
"%s: chan %d, attenuation group=%d\n", __func__, chan, grp);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Get channel sub-band. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (i = 0; i < IWN_NBANDS; i++)
if (sc->bands[i].lo != 0 &&
sc->bands[i].lo <= chan && chan <= sc->bands[i].hi)
break;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (i == IWN_NBANDS) /* Can't happen in real-life. */
return EINVAL;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
chans = sc->bands[i].chans;
DPRINTF(sc, IWN_DEBUG_CALIBRATE | IWN_DEBUG_TXPOW,
"%s: chan %d sub-band=%d\n", __func__, chan, i);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
for (c = 0; c < 2; c++) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
uint8_t power, gain, temp;
int maxchpwr, pwr, ridx, idx;
power = interpolate(chan,
chans[0].num, chans[0].samples[c][1].power,
chans[1].num, chans[1].samples[c][1].power, 1);
gain = interpolate(chan,
chans[0].num, chans[0].samples[c][1].gain,
chans[1].num, chans[1].samples[c][1].gain, 1);
temp = interpolate(chan,
chans[0].num, chans[0].samples[c][1].temp,
chans[1].num, chans[1].samples[c][1].temp, 1);
DPRINTF(sc, IWN_DEBUG_CALIBRATE | IWN_DEBUG_TXPOW,
"%s: Tx chain %d: power=%d gain=%d temp=%d\n",
__func__, c, power, gain, temp);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Compute temperature compensation. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
tdiff = ((sc->temp - temp) * 2) / tdiv[grp];
DPRINTF(sc, IWN_DEBUG_CALIBRATE | IWN_DEBUG_TXPOW,
"%s: temperature compensation=%d (current=%d, EEPROM=%d)\n",
__func__, tdiff, sc->temp, temp);
for (ridx = 0; ridx <= IWN_RIDX_MAX; ridx++) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Convert dBm to half-dBm. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
maxchpwr = sc->maxpwr[chan] * 2;
if ((ridx / 8) & 1)
maxchpwr -= 6; /* MIMO 2T: -3dB */
pwr = maxpwr;
/* Adjust TX power based on rate. */
if ((ridx % 8) == 5)
pwr -= 15; /* OFDM48: -7.5dB */
else if ((ridx % 8) == 6)
pwr -= 17; /* OFDM54: -8.5dB */
else if ((ridx % 8) == 7)
pwr -= 20; /* OFDM60: -10dB */
else
pwr -= 10; /* Others: -5dB */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Do not exceed channel max TX power. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (pwr > maxchpwr)
pwr = maxchpwr;
idx = gain - (pwr - power) - tdiff - vdiff;
if ((ridx / 8) & 1) /* MIMO */
idx += (int32_t)le32toh(uc->atten[grp][c]);
if (cmd.band == 0)
idx += 9; /* 5GHz */
if (ridx == IWN_RIDX_MAX)
idx += 5; /* CCK */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Make sure idx stays in a valid range. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (idx < 0)
idx = 0;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
else if (idx > IWN4965_MAX_PWR_INDEX)
idx = IWN4965_MAX_PWR_INDEX;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE | IWN_DEBUG_TXPOW,
"%s: Tx chain %d, rate idx %d: power=%d\n",
__func__, c, ridx, idx);
cmd.power[ridx].rf_gain[c] = rf_gain[idx];
cmd.power[ridx].dsp_gain[c] = dsp_gain[idx];
}
}
DPRINTF(sc, IWN_DEBUG_CALIBRATE | IWN_DEBUG_TXPOW,
"%s: set tx power for chan %d\n", __func__, chan);
return iwn_cmd(sc, IWN_CMD_TXPOWER, &cmd, sizeof cmd, async);
#undef interpolate
#undef fdivround
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Revert iwn channel-argument handling of set_txpower() to r201822 Pass the channel argument as a real argument and not through RXON. The RXON version seems to have problem with both the older 228.57.2.23 firmware and the latest 228.61.2.24 version resulting in device initialization errors like: wpa_supplicant[2928]: Failed to initiate AP scan. kernel: firmware error log: kernel: error type = "SYSASSERT" (0x00000005) kernel: program counter = 0x0000147C kernel: source line = 0x0000058B kernel: error data = 0x0000058B00000000 kernel: branch link = 0x0000145A00001492 kernel: interrupt link = 0x000006DE00000000 kernel: time = 7310 ... kernel: iwn0: iwn_config: could not set TX power kernel: iwn0: iwn_init_locked: could not configure device, error 35 By passing the current channel to hal->set_txpower() the firmware error is fixed, at least for the 4965 chipset of my Thinkpad and the ones tested by Bernhard. Submitted by: Bernhard Schmidt <bschmidt@techwires.net> Approved by: rpaulo
2010-01-09 04:00:22 +00:00
iwn5000_set_txpower(struct iwn_softc *sc, struct ieee80211_channel *ch,
int async)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
{
struct iwn5000_cmd_txpower cmd;
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
int cmdid;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* TX power calibration is handled automatically by the firmware
* for 5000 Series.
*/
memset(&cmd, 0, sizeof cmd);
cmd.global_limit = 2 * IWN5000_TXPOWER_MAX_DBM; /* 16 dBm */
cmd.flags = IWN5000_TXPOWER_NO_CLOSED;
cmd.srv_limit = IWN5000_TXPOWER_AUTO;
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE | IWN_DEBUG_XMIT,
"%s: setting TX power; rev=%d\n",
__func__,
IWN_UCODE_API(sc->ucode_rev));
if (IWN_UCODE_API(sc->ucode_rev) == 1)
cmdid = IWN_CMD_TXPOWER_DBM_V1;
else
cmdid = IWN_CMD_TXPOWER_DBM;
return iwn_cmd(sc, cmdid, &cmd, sizeof cmd, async);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
/*
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
* Retrieve the maximum RSSI (in dBm) among receivers.
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_get_rssi(struct iwn_softc *sc, struct iwn_rx_stat *stat)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn4965_rx_phystat *phy = (void *)stat->phybuf;
uint8_t mask, agc;
int rssi;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
mask = (le16toh(phy->antenna) >> 4) & IWN_ANT_ABC;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
agc = (le16toh(phy->agc) >> 7) & 0x7f;
rssi = 0;
RSSI related syncs with the OpenBSD code: - read RSSI only for the active chains - cast RSSI/NF to int8_t before passing it up to radiotap - remove the htole64() for the timestamp Obtained from: OpenBSD
2011-04-16 07:32:57 +00:00
if (mask & IWN_ANT_A)
rssi = MAX(rssi, phy->rssi[0]);
if (mask & IWN_ANT_B)
rssi = MAX(rssi, phy->rssi[2]);
if (mask & IWN_ANT_C)
rssi = MAX(rssi, phy->rssi[4]);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
DPRINTF(sc, IWN_DEBUG_RECV,
"%s: agc %d mask 0x%x rssi %d %d %d result %d\n", __func__, agc,
mask, phy->rssi[0], phy->rssi[2], phy->rssi[4],
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
rssi - agc - IWN_RSSI_TO_DBM);
return rssi - agc - IWN_RSSI_TO_DBM;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_get_rssi(struct iwn_softc *sc, struct iwn_rx_stat *stat)
{
struct iwn5000_rx_phystat *phy = (void *)stat->phybuf;
uint8_t agc;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
int rssi;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
agc = (le32toh(phy->agc) >> 9) & 0x7f;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
rssi = MAX(le16toh(phy->rssi[0]) & 0xff,
le16toh(phy->rssi[1]) & 0xff);
rssi = MAX(le16toh(phy->rssi[2]) & 0xff, rssi);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
DPRINTF(sc, IWN_DEBUG_RECV,
"%s: agc %d rssi %d %d %d result %d\n", __func__, agc,
Fix array overflow. Reviewed by: rpaulo
2010-01-08 21:02:12 +00:00
phy->rssi[0], phy->rssi[1], phy->rssi[2],
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
rssi - agc - IWN_RSSI_TO_DBM);
return rssi - agc - IWN_RSSI_TO_DBM;
}
/*
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
* Retrieve the average noise (in dBm) among receivers.
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_get_noise(const struct iwn_rx_general_stats *stats)
{
int i, total, nbant, noise;
total = nbant = 0;
for (i = 0; i < 3; i++) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if ((noise = le32toh(stats->noise[i]) & 0xff) == 0)
continue;
total += noise;
nbant++;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* There should be at least one antenna but check anyway. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return (nbant == 0) ? -127 : (total / nbant) - 107;
}
/*
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
* Compute temperature (in degC) from last received statistics.
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_get_temperature(struct iwn_softc *sc)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
struct iwn_ucode_info *uc = &sc->ucode_info;
int32_t r1, r2, r3, r4, temp;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
r1 = le32toh(uc->temp[0].chan20MHz);
r2 = le32toh(uc->temp[1].chan20MHz);
r3 = le32toh(uc->temp[2].chan20MHz);
r4 = le32toh(sc->rawtemp);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
if (r1 == r3) /* Prevents division by 0 (should not happen). */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return 0;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Sign-extend 23-bit R4 value to 32-bit. */
Don't timeout when stopping DMA channels. Obtained from: OpenBSD
2011-04-15 16:50:37 +00:00
r4 = ((r4 & 0xffffff) ^ 0x800000) - 0x800000;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Compute temperature in Kelvin. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
temp = (259 * (r4 - r2)) / (r3 - r1);
temp = (temp * 97) / 100 + 8;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
DPRINTF(sc, IWN_DEBUG_ANY, "temperature %dK/%dC\n", temp,
IWN_KTOC(temp));
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return IWN_KTOC(temp);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_get_temperature(struct iwn_softc *sc)
{
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
int32_t temp;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Temperature is not used by the driver for 5000 Series because
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
* TX power calibration is handled by firmware.
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*/
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
temp = le32toh(sc->rawtemp);
if (sc->hw_type == IWN_HW_REV_TYPE_5150) {
temp = (temp / -5) + sc->temp_off;
temp = IWN_KTOC(temp);
}
return temp;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
/*
* Initialize sensitivity calibration state machine.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_init_sensitivity(struct iwn_softc *sc)
{
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_calib_state *calib = &sc->calib;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint32_t flags;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
int error;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Reset calibration state machine. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
memset(calib, 0, sizeof (*calib));
calib->state = IWN_CALIB_STATE_INIT;
calib->cck_state = IWN_CCK_STATE_HIFA;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Set initial correlation values. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
calib->ofdm_x1 = sc->limits->min_ofdm_x1;
calib->ofdm_mrc_x1 = sc->limits->min_ofdm_mrc_x1;
* Rename bluetooth coexistence flags, no binary change. * Enable DC calibration and crystal calibration on Centrino Advanced-N 6250 parts. * Workaround for a HW bug (does not affect 4965AGN) that may sporadically affect latency under some rare circumstances. From a similar commit to iwlwifi. * Update sensitivity settings for 5000 series to workaround a performance bug in the DSP (1000 is not affected so we keep the old values for 5000). * Update sensitivity settings for 6000 series. * Set differential gains on 6250 too (but use a 1.0 factor, not 1.5). * Init OFDM sensitivity with min value (which depends on the chip) instead of hardcoding it to 90. * Read calibration version from ROM and set IWN_GP_DRIVER_CALIB_VER6 bit on 6x50 if version >= 6. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 2 weeks
2010-04-10 06:58:24 +00:00
calib->ofdm_x4 = sc->limits->min_ofdm_x4;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
calib->ofdm_mrc_x4 = sc->limits->min_ofdm_mrc_x4;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
calib->cck_x4 = 125;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
calib->cck_mrc_x4 = sc->limits->min_cck_mrc_x4;
calib->energy_cck = sc->limits->energy_cck;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Write initial sensitivity. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_send_sensitivity(sc)) != 0)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Write initial gains. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
if ((error = ops->init_gains(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
/* Request statistics at each beacon interval. */
flags = 0;
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE, "%s: sending request for statistics\n",
__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return iwn_cmd(sc, IWN_CMD_GET_STATISTICS, &flags, sizeof flags, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
/*
* Collect noise and RSSI statistics for the first 20 beacons received
* after association and use them to determine connected antennas and
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
* to set differential gains.
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_collect_noise(struct iwn_softc *sc,
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
const struct iwn_rx_general_stats *stats)
{
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_calib_state *calib = &sc->calib;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint32_t val;
int i;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Accumulate RSSI and noise for all 3 antennas. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (i = 0; i < 3; i++) {
calib->rssi[i] += le32toh(stats->rssi[i]) & 0xff;
calib->noise[i] += le32toh(stats->noise[i]) & 0xff;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* NB: We update differential gains only once after 20 beacons. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (++calib->nbeacons < 20)
return;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Determine highest average RSSI. */
val = MAX(calib->rssi[0], calib->rssi[1]);
val = MAX(calib->rssi[2], val);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Determine which antennas are connected. */
Detect active chains differently to work around a firmware bug which would mark non-existing chains as active. Obtained from: OpenBSD MFC after: 1 week
2010-07-15 09:34:00 +00:00
sc->chainmask = sc->rxchainmask;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (i = 0; i < 3; i++)
Detect active chains differently to work around a firmware bug which would mark non-existing chains as active. Obtained from: OpenBSD MFC after: 1 week
2010-07-15 09:34:00 +00:00
if (val - calib->rssi[i] > 15 * 20)
sc->chainmask &= ~(1 << i);
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE | IWN_DEBUG_XMIT,
Detect active chains differently to work around a firmware bug which would mark non-existing chains as active. Obtained from: OpenBSD MFC after: 1 week
2010-07-15 09:34:00 +00:00
"%s: RX chains mask: theoretical=0x%x, actual=0x%x\n",
__func__, sc->rxchainmask, sc->chainmask);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* If none of the TX antennas are connected, keep at least one. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if ((sc->chainmask & sc->txchainmask) == 0)
sc->chainmask |= IWN_LSB(sc->txchainmask);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
(void)ops->set_gains(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
calib->state = IWN_CALIB_STATE_RUN;
#ifdef notyet
/* XXX Disable RX chains with no antennas connected. */
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->rxchain = htole16(IWN_RXCHAIN_SEL(sc->chainmask));
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
if (sc->sc_is_scanning)
device_printf(sc->sc_dev,
"%s: is_scanning set, before RXON\n",
__func__);
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
(void)iwn_cmd(sc, IWN_CMD_RXON, sc->rxon, sc->rxonsz, 1);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
#endif
/* Enable power-saving mode if requested by user. */
Enable the station-side power management flag. It's not enabled by default in net80211 so this is a no-op unless if you enable it (ifconfig wlan0 powersave). Tested: * iwn0: <Intel WiFi Link 5100> mem 0xf4300000-0xf4301fff irq 17 at device 0.0 on pci3 TODO: * .. test on all the other NICs * See if I have to disable it during scan and such * Make it configurable live, rather than only after it's done its initial receive calibration.
2013-07-04 19:45:34 +00:00
if (ic->ic_flags & IEEE80211_F_PMGTON)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
(void)iwn_set_pslevel(sc, 0, 3, 1);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_init_gains(struct iwn_softc *sc)
{
struct iwn_phy_calib_gain cmd;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
memset(&cmd, 0, sizeof cmd);
cmd.code = IWN4965_PHY_CALIB_DIFF_GAIN;
/* Differential gains initially set to 0 for all 3 antennas. */
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"%s: setting initial differential gains\n", __func__);
return iwn_cmd(sc, IWN_CMD_PHY_CALIB, &cmd, sizeof cmd, 1);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_init_gains(struct iwn_softc *sc)
{
struct iwn_phy_calib cmd;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
memset(&cmd, 0, sizeof cmd);
Pull some features out of the firmware: - If a ENH_SENS TLV section exit the firmware is capable of doing enhanced sensitivity calibration. - Newer devices/firmwares have more calibration commands therefore hardcoding the noise gain/reset commands no longer works. It is supposed to use the next index after the newest calibration type support. Read the command index of the TLV section if available.
2011-04-19 19:47:41 +00:00
cmd.code = sc->reset_noise_gain;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
cmd.ngroups = 1;
cmd.isvalid = 1;
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"%s: setting initial differential gains\n", __func__);
return iwn_cmd(sc, IWN_CMD_PHY_CALIB, &cmd, sizeof cmd, 1);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_set_gains(struct iwn_softc *sc)
{
struct iwn_calib_state *calib = &sc->calib;
struct iwn_phy_calib_gain cmd;
int i, delta, noise;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Get minimal noise among connected antennas. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
noise = INT_MAX; /* NB: There's at least one antenna. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (i = 0; i < 3; i++)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (sc->chainmask & (1 << i))
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
noise = MIN(calib->noise[i], noise);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
memset(&cmd, 0, sizeof cmd);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
cmd.code = IWN4965_PHY_CALIB_DIFF_GAIN;
/* Set differential gains for connected antennas. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (i = 0; i < 3; i++) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (sc->chainmask & (1 << i)) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Compute attenuation (in unit of 1.5dB). */
delta = (noise - (int32_t)calib->noise[i]) / 30;
/* NB: delta <= 0 */
/* Limit to [-4.5dB,0]. */
cmd.gain[i] = MIN(abs(delta), 3);
if (delta < 0)
cmd.gain[i] |= 1 << 2; /* sign bit */
}
}
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"setting differential gains Ant A/B/C: %x/%x/%x (%x)\n",
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
cmd.gain[0], cmd.gain[1], cmd.gain[2], sc->chainmask);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return iwn_cmd(sc, IWN_CMD_PHY_CALIB, &cmd, sizeof cmd, 1);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_set_gains(struct iwn_softc *sc)
{
struct iwn_calib_state *calib = &sc->calib;
struct iwn_phy_calib_gain cmd;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
int i, ant, div, delta;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
* Rename bluetooth coexistence flags, no binary change. * Enable DC calibration and crystal calibration on Centrino Advanced-N 6250 parts. * Workaround for a HW bug (does not affect 4965AGN) that may sporadically affect latency under some rare circumstances. From a similar commit to iwlwifi. * Update sensitivity settings for 5000 series to workaround a performance bug in the DSP (1000 is not affected so we keep the old values for 5000). * Update sensitivity settings for 6000 series. * Set differential gains on 6250 too (but use a 1.0 factor, not 1.5). * Init OFDM sensitivity with min value (which depends on the chip) instead of hardcoding it to 90. * Read calibration version from ROM and set IWN_GP_DRIVER_CALIB_VER6 bit on 6x50 if version >= 6. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 2 weeks
2010-04-10 06:58:24 +00:00
/* We collected 20 beacons and !=6050 need a 1.5 factor. */
div = (sc->hw_type == IWN_HW_REV_TYPE_6050) ? 20 : 30;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
memset(&cmd, 0, sizeof cmd);
Pull some features out of the firmware: - If a ENH_SENS TLV section exit the firmware is capable of doing enhanced sensitivity calibration. - Newer devices/firmwares have more calibration commands therefore hardcoding the noise gain/reset commands no longer works. It is supposed to use the next index after the newest calibration type support. Read the command index of the TLV section if available.
2011-04-19 19:47:41 +00:00
cmd.code = sc->noise_gain;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
cmd.ngroups = 1;
cmd.isvalid = 1;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Get first available RX antenna as referential. */
ant = IWN_LSB(sc->rxchainmask);
/* Set differential gains for other antennas. */
for (i = ant + 1; i < 3; i++) {
if (sc->chainmask & (1 << i)) {
/* The delta is relative to antenna "ant". */
delta = ((int32_t)calib->noise[ant] -
* Rename bluetooth coexistence flags, no binary change. * Enable DC calibration and crystal calibration on Centrino Advanced-N 6250 parts. * Workaround for a HW bug (does not affect 4965AGN) that may sporadically affect latency under some rare circumstances. From a similar commit to iwlwifi. * Update sensitivity settings for 5000 series to workaround a performance bug in the DSP (1000 is not affected so we keep the old values for 5000). * Update sensitivity settings for 6000 series. * Set differential gains on 6250 too (but use a 1.0 factor, not 1.5). * Init OFDM sensitivity with min value (which depends on the chip) instead of hardcoding it to 90. * Read calibration version from ROM and set IWN_GP_DRIVER_CALIB_VER6 bit on 6x50 if version >= 6. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 2 weeks
2010-04-10 06:58:24 +00:00
(int32_t)calib->noise[i]) / div;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Limit to [-4.5dB,+4.5dB]. */
cmd.gain[i - 1] = MIN(abs(delta), 3);
if (delta < 0)
cmd.gain[i - 1] |= 1 << 2; /* sign bit */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
}
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE | IWN_DEBUG_XMIT,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
"setting differential gains Ant B/C: %x/%x (%x)\n",
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
cmd.gain[0], cmd.gain[1], sc->chainmask);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return iwn_cmd(sc, IWN_CMD_PHY_CALIB, &cmd, sizeof cmd, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
/*
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
* Tune RF RX sensitivity based on the number of false alarms detected
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
* during the last beacon period.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_tune_sensitivity(struct iwn_softc *sc, const struct iwn_rx_stats *stats)
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
#define inc(val, inc, max) \
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if ((val) < (max)) { \
if ((val) < (max) - (inc)) \
(val) += (inc); \
else \
(val) = (max); \
needs_update = 1; \
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
#define dec(val, dec, min) \
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if ((val) > (min)) { \
if ((val) > (min) + (dec)) \
(val) -= (dec); \
else \
(val) = (min); \
needs_update = 1; \
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
const struct iwn_sensitivity_limits *limits = sc->limits;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct iwn_calib_state *calib = &sc->calib;
uint32_t val, rxena, fa;
uint32_t energy[3], energy_min;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint8_t noise[3], noise_ref;
Updates to the iwn driver: * iwnfw has now been split into individual modules so autoloading of firmware module(s) does work again. * Changes have been made to RUN -> AUTH transition, this should fix the issue reported by Glen and others. * Brandon reported issues in iwn_cmd() with large commands, those have been fixed to. * DEAUTH is now handled correctly. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-24 09:55:11 +00:00
int i, needs_update = 0;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Check that we've been enabled long enough. */
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
if ((rxena = le32toh(stats->general.load)) == 0){
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end not so long\n", __func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Compute number of false alarms since last call for OFDM. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
fa = le32toh(stats->ofdm.bad_plcp) - calib->bad_plcp_ofdm;
fa += le32toh(stats->ofdm.fa) - calib->fa_ofdm;
Reuse net80211 code: - IWN_TXOP_TO_US is equal to IEEE80211_TXOP_TO_US - use IEEE80211_DUR_TU - ieee80211_add_rates/ieee80211_add_xrates are public, use em - copied ieee80211_add_ssid it is not public
2011-04-14 17:26:13 +00:00
fa *= 200 * IEEE80211_DUR_TU; /* 200TU */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (fa > 50 * rxena) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* High false alarm count, decrease sensitivity. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"%s: OFDM high false alarm count: %u\n", __func__, fa);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
inc(calib->ofdm_x1, 1, limits->max_ofdm_x1);
inc(calib->ofdm_mrc_x1, 1, limits->max_ofdm_mrc_x1);
inc(calib->ofdm_x4, 1, limits->max_ofdm_x4);
inc(calib->ofdm_mrc_x4, 1, limits->max_ofdm_mrc_x4);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
} else if (fa < 5 * rxena) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Low false alarm count, increase sensitivity. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"%s: OFDM low false alarm count: %u\n", __func__, fa);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
dec(calib->ofdm_x1, 1, limits->min_ofdm_x1);
dec(calib->ofdm_mrc_x1, 1, limits->min_ofdm_mrc_x1);
dec(calib->ofdm_x4, 1, limits->min_ofdm_x4);
dec(calib->ofdm_mrc_x4, 1, limits->min_ofdm_mrc_x4);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Compute maximum noise among 3 receivers. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (i = 0; i < 3; i++)
noise[i] = (le32toh(stats->general.noise[i]) >> 8) & 0xff;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
val = MAX(noise[0], noise[1]);
val = MAX(noise[2], val);
/* Insert it into our samples table. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
calib->noise_samples[calib->cur_noise_sample] = val;
calib->cur_noise_sample = (calib->cur_noise_sample + 1) % 20;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Compute maximum noise among last 20 samples. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
noise_ref = calib->noise_samples[0];
for (i = 1; i < 20; i++)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
noise_ref = MAX(noise_ref, calib->noise_samples[i]);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Compute maximum energy among 3 receivers. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (i = 0; i < 3; i++)
energy[i] = le32toh(stats->general.energy[i]);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
val = MIN(energy[0], energy[1]);
val = MIN(energy[2], val);
/* Insert it into our samples table. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
calib->energy_samples[calib->cur_energy_sample] = val;
calib->cur_energy_sample = (calib->cur_energy_sample + 1) % 10;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Compute minimum energy among last 10 samples. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
energy_min = calib->energy_samples[0];
for (i = 1; i < 10; i++)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
energy_min = MAX(energy_min, calib->energy_samples[i]);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
energy_min += 6;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Compute number of false alarms since last call for CCK. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
fa = le32toh(stats->cck.bad_plcp) - calib->bad_plcp_cck;
fa += le32toh(stats->cck.fa) - calib->fa_cck;
Reuse net80211 code: - IWN_TXOP_TO_US is equal to IEEE80211_TXOP_TO_US - use IEEE80211_DUR_TU - ieee80211_add_rates/ieee80211_add_xrates are public, use em - copied ieee80211_add_ssid it is not public
2011-04-14 17:26:13 +00:00
fa *= 200 * IEEE80211_DUR_TU; /* 200TU */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (fa > 50 * rxena) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* High false alarm count, decrease sensitivity. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"%s: CCK high false alarm count: %u\n", __func__, fa);
calib->cck_state = IWN_CCK_STATE_HIFA;
calib->low_fa = 0;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (calib->cck_x4 > 160) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
calib->noise_ref = noise_ref;
if (calib->energy_cck > 2)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
dec(calib->energy_cck, 2, energy_min);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (calib->cck_x4 < 160) {
calib->cck_x4 = 161;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
needs_update = 1;
} else
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
inc(calib->cck_x4, 3, limits->max_cck_x4);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
inc(calib->cck_mrc_x4, 3, limits->max_cck_mrc_x4);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
} else if (fa < 5 * rxena) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Low false alarm count, increase sensitivity. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"%s: CCK low false alarm count: %u\n", __func__, fa);
calib->cck_state = IWN_CCK_STATE_LOFA;
calib->low_fa++;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (calib->cck_state != IWN_CCK_STATE_INIT &&
(((int32_t)calib->noise_ref - (int32_t)noise_ref) > 2 ||
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
calib->low_fa > 100)) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
inc(calib->energy_cck, 2, limits->min_energy_cck);
dec(calib->cck_x4, 3, limits->min_cck_x4);
dec(calib->cck_mrc_x4, 3, limits->min_cck_mrc_x4);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
} else {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Not worth to increase or decrease sensitivity. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"%s: CCK normal false alarm count: %u\n", __func__, fa);
calib->low_fa = 0;
calib->noise_ref = noise_ref;
if (calib->cck_state == IWN_CCK_STATE_HIFA) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Previous interval had many false alarms. */
dec(calib->energy_cck, 8, energy_min);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
calib->cck_state = IWN_CCK_STATE_INIT;
}
if (needs_update)
(void)iwn_send_sensitivity(sc);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
#undef dec
#undef inc
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_send_sensitivity(struct iwn_softc *sc)
{
struct iwn_calib_state *calib = &sc->calib;
Add some new features: - 6000 series devices need enhanced sensitivity calibration. - 6000 series devices need a different setting for the shadow reg. - set the IWN_FLAG_HAS_11N bit if the EEPROM says the device has 11n support. Obtained from: OpenBSD
2011-04-16 14:56:13 +00:00
struct iwn_enhanced_sensitivity_cmd cmd;
int len;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
memset(&cmd, 0, sizeof cmd);
Add some new features: - 6000 series devices need enhanced sensitivity calibration. - 6000 series devices need a different setting for the shadow reg. - set the IWN_FLAG_HAS_11N bit if the EEPROM says the device has 11n support. Obtained from: OpenBSD
2011-04-16 14:56:13 +00:00
len = sizeof (struct iwn_sensitivity_cmd);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
cmd.which = IWN_SENSITIVITY_WORKTBL;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* OFDM modulation. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
cmd.corr_ofdm_x1 = htole16(calib->ofdm_x1);
cmd.corr_ofdm_mrc_x1 = htole16(calib->ofdm_mrc_x1);
cmd.corr_ofdm_x4 = htole16(calib->ofdm_x4);
cmd.corr_ofdm_mrc_x4 = htole16(calib->ofdm_mrc_x4);
cmd.energy_ofdm = htole16(sc->limits->energy_ofdm);
cmd.energy_ofdm_th = htole16(62);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* CCK modulation. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
cmd.corr_cck_x4 = htole16(calib->cck_x4);
cmd.corr_cck_mrc_x4 = htole16(calib->cck_mrc_x4);
cmd.energy_cck = htole16(calib->energy_cck);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Barker modulation: use default values. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
cmd.corr_barker = htole16(190);
Add some initial support for the Intel 6235. Tested: * Intel 5100 * Intel 6235 Obtained from: mav, others
2013-12-09 03:40:02 +00:00
cmd.corr_barker_mrc = htole16(sc->limits->barker_mrc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
o add more notification strings in iwn_intr_str() o sync with OpenBSD code Submitted by: Bernhard Schmidt <bschmidt at techwires.net> MFC after: 3 days
2010-01-25 19:36:48 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
"%s: set sensitivity %d/%d/%d/%d/%d/%d/%d\n", __func__,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
calib->ofdm_x1, calib->ofdm_mrc_x1, calib->ofdm_x4,
calib->ofdm_mrc_x4, calib->cck_x4,
calib->cck_mrc_x4, calib->energy_cck);
Add some new features: - 6000 series devices need enhanced sensitivity calibration. - 6000 series devices need a different setting for the shadow reg. - set the IWN_FLAG_HAS_11N bit if the EEPROM says the device has 11n support. Obtained from: OpenBSD
2011-04-16 14:56:13 +00:00
if (!(sc->sc_flags & IWN_FLAG_ENH_SENS))
goto send;
/* Enhanced sensitivity settings. */
len = sizeof (struct iwn_enhanced_sensitivity_cmd);
cmd.ofdm_det_slope_mrc = htole16(668);
cmd.ofdm_det_icept_mrc = htole16(4);
cmd.ofdm_det_slope = htole16(486);
cmd.ofdm_det_icept = htole16(37);
cmd.cck_det_slope_mrc = htole16(853);
cmd.cck_det_icept_mrc = htole16(4);
cmd.cck_det_slope = htole16(476);
cmd.cck_det_icept = htole16(99);
send:
return iwn_cmd(sc, IWN_CMD_SET_SENSITIVITY, &cmd, len, 1);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Begin fleshing out some code to handle tracking PLCP error rates in preparation for the scan based retune logic. The linux iwlwifi driver does a rescan (onto a non-active channel) to force an RF retune when the PLCP error rates exceed a certain threshold. * Add code to track HT PLCP rate errors; * Separate out the PLCP error count fetch and update so the delta can be used when checking for PLCP error rates; * Implement the PLCP error logic from iwlwifi; * For now, just print out whenever the error rate exceeds the threshold. The actual scan based retune will take a bit more effort; the scan command code right now assumes that a scan state is passed in. This does need to change to be more flexible (both for this and in preparation for scanning multiple channels at once.) Tested: * 5100 (STA mode) * 2200 (STA mode) * 2230 (STA mode)
2013-12-07 08:03:10 +00:00
/*
* Look at the increase of PLCP errors over time; if it exceeds
* a programmed threshold then trigger an RF retune.
*/
static void
iwn_check_rx_recovery(struct iwn_softc *sc, struct iwn_stats *rs)
{
int32_t delta_ofdm, delta_ht, delta_cck;
struct iwn_calib_state *calib = &sc->calib;
int delta_ticks, cur_ticks;
int delta_msec;
int thresh;
/*
* Calculate the difference between the current and
* previous statistics.
*/
delta_cck = le32toh(rs->rx.cck.bad_plcp) - calib->bad_plcp_cck;
delta_ofdm = le32toh(rs->rx.ofdm.bad_plcp) - calib->bad_plcp_ofdm;
delta_ht = le32toh(rs->rx.ht.bad_plcp) - calib->bad_plcp_ht;
/*
* Calculate the delta in time between successive statistics
* messages. Yes, it can roll over; so we make sure that
* this doesn't happen.
*
* XXX go figure out what to do about rollover
* XXX go figure out what to do if ticks rolls over to -ve instead!
* XXX go stab signed integer overflow undefined-ness in the face.
*/
cur_ticks = ticks;
delta_ticks = cur_ticks - sc->last_calib_ticks;
/*
* If any are negative, then the firmware likely reset; so just
* bail. We'll pick this up next time.
*/
if (delta_cck < 0 || delta_ofdm < 0 || delta_ht < 0 || delta_ticks < 0)
return;
/*
* delta_ticks is in ticks; we need to convert it up to milliseconds
* so we can do some useful math with it.
*/
delta_msec = ticks_to_msecs(delta_ticks);
/*
* Calculate what our threshold is given the current delta_msec.
*/
thresh = sc->base_params->plcp_err_threshold * delta_msec;
DPRINTF(sc, IWN_DEBUG_STATE,
"%s: time delta: %d; cck=%d, ofdm=%d, ht=%d, total=%d, thresh=%d\n",
__func__,
delta_msec,
delta_cck,
delta_ofdm,
delta_ht,
(delta_msec + delta_cck + delta_ofdm + delta_ht),
thresh);
/*
* If we need a retune, then schedule a single channel scan
* to a channel that isn't the currently active one!
*
* The math from linux iwlwifi:
*
* if ((delta * 100 / msecs) > threshold)
*/
if (thresh > 0 && (delta_cck + delta_ofdm + delta_ht) * 100 > thresh) {
Move the retune notification print to a debug print. Yes, I still have to do the retune. But I'm giving in to many people pestering me (very gently!) about this. Tested: * Intel Centrino 6205
2014-01-05 00:46:31 +00:00
DPRINTF(sc, IWN_DEBUG_ANY,
Begin fleshing out some code to handle tracking PLCP error rates in preparation for the scan based retune logic. The linux iwlwifi driver does a rescan (onto a non-active channel) to force an RF retune when the PLCP error rates exceed a certain threshold. * Add code to track HT PLCP rate errors; * Separate out the PLCP error count fetch and update so the delta can be used when checking for PLCP error rates; * Implement the PLCP error logic from iwlwifi; * For now, just print out whenever the error rate exceeds the threshold. The actual scan based retune will take a bit more effort; the scan command code right now assumes that a scan state is passed in. This does need to change to be more flexible (both for this and in preparation for scanning multiple channels at once.) Tested: * 5100 (STA mode) * 2200 (STA mode) * 2230 (STA mode)
2013-12-07 08:03:10 +00:00
"%s: PLCP error threshold raw (%d) comparison (%d) "
"over limit (%d); retune!\n",
__func__,
(delta_cck + delta_ofdm + delta_ht),
(delta_cck + delta_ofdm + delta_ht) * 100,
thresh);
}
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Set STA mode power saving level (between 0 and 5).
* Level 0 is CAM (Continuously Aware Mode), 5 is for maximum power saving.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_set_pslevel(struct iwn_softc *sc, int dtim, int level, int async)
{
struct iwn_pmgt_cmd cmd;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
const struct iwn_pmgt *pmgt;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint32_t max, skip_dtim;
Rename some stuff in favour of the OpenBSD names: - prefer EDCA over WME - qid for a TXQ ID - reg for register values
2011-04-16 13:40:32 +00:00
uint32_t reg;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
int i;
Implement basic 802.11n awareness in the PHY and AMRR rate control code. * Add 802.11n 2ghz and 5ghz tables, including legacy rates and up to MCS23 rates (3x3.) * Populate the rate code -> rate index lookup table with MCS _and_ normal rates, but _not_ the basic rate flag. Since the basic rate flag is the same as the MCS flag, we can only use one. * Introduce some accessor inlines that do PLCP and rate table lookup/access and enforce that it doesn't set the basic rate bit. They're not designed for MCS rates, so it will panic. * Start converting drivers that use the rate table stuff to use the accessor inlines and strip the basic flag. * Teach AMRR about basic 11n - it's still as crap for MCS as it is being used by iwn, so it's not a step _backwardS_. * Convert iwn over to accept 11n MCS rates rather than 'translate' legacy to MCS rates. It doesn't use a lookup table any longer; instead it's a function which takes the current node (for HT parameters) and the rate code, and returns the hardware PLCP code to use. Tested: * ath - it's a no-op, and it works that way * iwn - both 11n and non-11n
2013-07-04 21:16:49 +00:00
DPRINTF(sc, IWN_DEBUG_PWRSAVE,
"%s: dtim=%d, level=%d, async=%d\n",
__func__,
dtim,
level,
async);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Select which PS parameters to use. */
if (dtim <= 2)
pmgt = &iwn_pmgt[0][level];
else if (dtim <= 10)
pmgt = &iwn_pmgt[1][level];
else
pmgt = &iwn_pmgt[2][level];
memset(&cmd, 0, sizeof cmd);
if (level != 0) /* not CAM */
cmd.flags |= htole16(IWN_PS_ALLOW_SLEEP);
if (level == 5)
cmd.flags |= htole16(IWN_PS_FAST_PD);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Retrieve PCIe Active State Power Management (ASPM). */
Rename some stuff in favour of the OpenBSD names: - prefer EDCA over WME - qid for a TXQ ID - reg for register values
2011-04-16 13:40:32 +00:00
reg = pci_read_config(sc->sc_dev, sc->sc_cap_off + 0x10, 1);
if (!(reg & 0x1)) /* L0s Entry disabled. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
cmd.flags |= htole16(IWN_PS_PCI_PMGT);
cmd.rxtimeout = htole32(pmgt->rxtimeout * 1024);
cmd.txtimeout = htole32(pmgt->txtimeout * 1024);
if (dtim == 0) {
dtim = 1;
skip_dtim = 0;
} else
skip_dtim = pmgt->skip_dtim;
if (skip_dtim != 0) {
cmd.flags |= htole16(IWN_PS_SLEEP_OVER_DTIM);
max = pmgt->intval[4];
if (max == (uint32_t)-1)
max = dtim * (skip_dtim + 1);
else if (max > dtim)
max = (max / dtim) * dtim;
} else
max = dtim;
for (i = 0; i < 5; i++)
cmd.intval[i] = htole32(MIN(max, pmgt->intval[i]));
DPRINTF(sc, IWN_DEBUG_RESET, "setting power saving level to %d\n",
level);
return iwn_cmd(sc, IWN_CMD_SET_POWER_MODE, &cmd, sizeof cmd, async);
}
Split out bluetooth coexistence setup.
2011-04-15 17:10:52 +00:00
static int
iwn_send_btcoex(struct iwn_softc *sc)
{
struct iwn_bluetooth cmd;
memset(&cmd, 0, sizeof cmd);
cmd.flags = IWN_BT_COEX_CHAN_ANN | IWN_BT_COEX_BT_PRIO;
cmd.lead_time = IWN_BT_LEAD_TIME_DEF;
cmd.max_kill = IWN_BT_MAX_KILL_DEF;
DPRINTF(sc, IWN_DEBUG_RESET, "%s: configuring bluetooth coexistence\n",
__func__);
return iwn_cmd(sc, IWN_CMD_BT_COEX, &cmd, sizeof(cmd), 0);
}
Add basic support for advanced bluetooth coexistence required for 6005 gen2b (1030/6030) adapters.
2011-04-20 16:59:27 +00:00
static int
iwn_send_advanced_btcoex(struct iwn_softc *sc)
{
static const uint32_t btcoex_3wire[12] = {
0xaaaaaaaa, 0xaaaaaaaa, 0xaeaaaaaa, 0xaaaaaaaa,
0xcc00ff28, 0x0000aaaa, 0xcc00aaaa, 0x0000aaaa,
0xc0004000, 0x00004000, 0xf0005000, 0xf0005000,
};
struct iwn6000_btcoex_config btconfig;
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
struct iwn2000_btcoex_config btconfig2k;
Add basic support for advanced bluetooth coexistence required for 6005 gen2b (1030/6030) adapters.
2011-04-20 16:59:27 +00:00
struct iwn_btcoex_priotable btprio;
struct iwn_btcoex_prot btprot;
int error, i;
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
uint8_t flags;
Add basic support for advanced bluetooth coexistence required for 6005 gen2b (1030/6030) adapters.
2011-04-20 16:59:27 +00:00
memset(&btconfig, 0, sizeof btconfig);
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
memset(&btconfig2k, 0, sizeof btconfig2k);
flags = IWN_BT_FLAG_COEX6000_MODE_3W <<
IWN_BT_FLAG_COEX6000_MODE_SHIFT; // Done as is in linux kernel 3.2
if (sc->base_params->bt_sco_disable)
flags &= ~IWN_BT_FLAG_SYNC_2_BT_DISABLE;
else
flags |= IWN_BT_FLAG_SYNC_2_BT_DISABLE;
flags |= IWN_BT_FLAG_COEX6000_CHAN_INHIBITION;
/* Default flags result is 145 as old value */
/*
* Flags value has to be review. Values must change if we
* which to disable it
*/
if (sc->base_params->bt_session_2) {
btconfig2k.flags = flags;
btconfig2k.max_kill = 5;
btconfig2k.bt3_t7_timer = 1;
btconfig2k.kill_ack = htole32(0xffff0000);
btconfig2k.kill_cts = htole32(0xffff0000);
btconfig2k.sample_time = 2;
btconfig2k.bt3_t2_timer = 0xc;
for (i = 0; i < 12; i++)
btconfig2k.lookup_table[i] = htole32(btcoex_3wire[i]);
btconfig2k.valid = htole16(0xff);
btconfig2k.prio_boost = htole32(0xf0);
DPRINTF(sc, IWN_DEBUG_RESET,
"%s: configuring advanced bluetooth coexistence"
" session 2, flags : 0x%x\n",
__func__,
flags);
error = iwn_cmd(sc, IWN_CMD_BT_COEX, &btconfig2k,
sizeof(btconfig2k), 1);
} else {
btconfig.flags = flags;
btconfig.max_kill = 5;
btconfig.bt3_t7_timer = 1;
btconfig.kill_ack = htole32(0xffff0000);
btconfig.kill_cts = htole32(0xffff0000);
btconfig.sample_time = 2;
btconfig.bt3_t2_timer = 0xc;
for (i = 0; i < 12; i++)
btconfig.lookup_table[i] = htole32(btcoex_3wire[i]);
btconfig.valid = htole16(0xff);
btconfig.prio_boost = 0xf0;
DPRINTF(sc, IWN_DEBUG_RESET,
"%s: configuring advanced bluetooth coexistence,"
" flags : 0x%x\n",
__func__,
flags);
error = iwn_cmd(sc, IWN_CMD_BT_COEX, &btconfig,
sizeof(btconfig), 1);
}
Add basic support for advanced bluetooth coexistence required for 6005 gen2b (1030/6030) adapters.
2011-04-20 16:59:27 +00:00
if (error != 0)
return error;
memset(&btprio, 0, sizeof btprio);
btprio.calib_init1 = 0x6;
btprio.calib_init2 = 0x7;
btprio.calib_periodic_low1 = 0x2;
btprio.calib_periodic_low2 = 0x3;
btprio.calib_periodic_high1 = 0x4;
btprio.calib_periodic_high2 = 0x5;
btprio.dtim = 0x6;
btprio.scan52 = 0x8;
btprio.scan24 = 0xa;
error = iwn_cmd(sc, IWN_CMD_BT_COEX_PRIOTABLE, &btprio, sizeof(btprio),
1);
if (error != 0)
return error;
/* Force BT state machine change. */
Use the correct structure size when flipping the BT coex state machine. This showed up when doing some basic testing on the Intel 6230. Tested: * Intel 6230, STA mode
2013-08-11 03:39:28 +00:00
memset(&btprot, 0, sizeof btprot);
Add basic support for advanced bluetooth coexistence required for 6005 gen2b (1030/6030) adapters.
2011-04-20 16:59:27 +00:00
btprot.open = 1;
btprot.type = 1;
error = iwn_cmd(sc, IWN_CMD_BT_COEX_PROT, &btprot, sizeof(btprot), 1);
if (error != 0)
return error;
btprot.open = 0;
return iwn_cmd(sc, IWN_CMD_BT_COEX_PROT, &btprot, sizeof(btprot), 1);
}
The DC calibration result obtained during initialization can't be passed over to the runtime firmware on 6050 devices. Instead let the runtime firmware do the calibration itself. This fixes support for the 6050 series devices. Obtained from: OpenBSD Submitted by: kevlo Tested by: lx, Tz-Huan Huang(earlier version)
2011-11-21 22:19:12 +00:00
static int
iwn5000_runtime_calib(struct iwn_softc *sc)
{
struct iwn5000_calib_config cmd;
memset(&cmd, 0, sizeof cmd);
cmd.ucode.once.enable = 0xffffffff;
cmd.ucode.once.start = IWN5000_CALIB_DC;
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"%s: configuring runtime calibration\n", __func__);
return iwn_cmd(sc, IWN5000_CMD_CALIB_CONFIG, &cmd, sizeof(cmd), 0);
}
Attempt to make 5GHz HT/40 work on the 6xxx series NICs. The 6205 (Taylor Peak) in the Lenovo X230 works fine in 5GHz 11a and 11n HT20, but not 11n HT40. The NIC goes RX deaf the moment HT40 is configured. It's so RX deaf that it doesn't even hear beacons and the firmware sends "BEACON MISS" events. That's pretty deaf. I tried configuring up the HT40 flags in monitor mode and it worked - so I assumed that doing the transition from 20 -> 40MHz channel configuration when going auth->assoc (ie, after the NIC has been partially configured) is a problem. So for now, let's just always set them if they're available. Tested: * Intel 5300, STA mode, 5GHz HT/40 AP; 2GHz HT/20 AP * Intel 6205, STA mode, 5GHz HT/40, HT20, 11a AP; 2GHz HT/20 AP This was pointed out to me by coworkers trying to use FreeBSD-HEAD in the office on their Thinkpad T420p laptops. TODO: * I don't like how the HT40 flags are configured - the whole interop/ protection config should be re-checked. Notably, I think curhtprotmode is 0 in a lot of cases, which means "no interoperability" and i think that's busted. Sponsored by: Norse Corp, Inc.
2015-07-07 03:51:29 +00:00
static uint32_t
iwn_get_rxon_ht_flags(struct iwn_softc *sc, struct ieee80211_channel *c)
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Attempt to make 5GHz HT/40 work on the 6xxx series NICs. The 6205 (Taylor Peak) in the Lenovo X230 works fine in 5GHz 11a and 11n HT20, but not 11n HT40. The NIC goes RX deaf the moment HT40 is configured. It's so RX deaf that it doesn't even hear beacons and the firmware sends "BEACON MISS" events. That's pretty deaf. I tried configuring up the HT40 flags in monitor mode and it worked - so I assumed that doing the transition from 20 -> 40MHz channel configuration when going auth->assoc (ie, after the NIC has been partially configured) is a problem. So for now, let's just always set them if they're available. Tested: * Intel 5300, STA mode, 5GHz HT/40 AP; 2GHz HT/20 AP * Intel 6205, STA mode, 5GHz HT/40, HT20, 11a AP; 2GHz HT/20 AP This was pointed out to me by coworkers trying to use FreeBSD-HEAD in the office on their Thinkpad T420p laptops. TODO: * I don't like how the HT40 flags are configured - the whole interop/ protection config should be re-checked. Notably, I think curhtprotmode is 0 in a lot of cases, which means "no interoperability" and i think that's busted. Sponsored by: Norse Corp, Inc.
2015-07-07 03:51:29 +00:00
uint32_t htflags = 0;
if (! IEEE80211_IS_CHAN_HT(c))
return (0);
htflags |= IWN_RXON_HT_PROTMODE(ic->ic_curhtprotmode);
if (IEEE80211_IS_CHAN_HT40(c)) {
switch (ic->ic_curhtprotmode) {
case IEEE80211_HTINFO_OPMODE_HT20PR:
htflags |= IWN_RXON_HT_MODEPURE40;
break;
default:
htflags |= IWN_RXON_HT_MODEMIXED;
break;
}
}
if (IEEE80211_IS_CHAN_HT40D(c))
htflags |= IWN_RXON_HT_HT40MINUS;
return (htflags);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_config(struct iwn_softc *sc)
{
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
const uint8_t *macaddr;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
uint32_t txmask;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
uint16_t rxchain;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
int error;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if ((sc->base_params->calib_need & IWN_FLG_NEED_PHY_CALIB_TEMP_OFFSET)
&& (sc->base_params->calib_need & IWN_FLG_NEED_PHY_CALIB_TEMP_OFFSETv2)) {
device_printf(sc->sc_dev,"%s: temp_offset and temp_offsetv2 are"
" exclusive each together. Review NIC config file. Conf"
" : 0x%08x Flags : 0x%08x \n", __func__,
sc->base_params->calib_need,
(IWN_FLG_NEED_PHY_CALIB_TEMP_OFFSET |
IWN_FLG_NEED_PHY_CALIB_TEMP_OFFSETv2));
return (EINVAL);
}
/* Compute temperature calib if needed. Will be send by send calib */
if (sc->base_params->calib_need & IWN_FLG_NEED_PHY_CALIB_TEMP_OFFSET) {
The 6005 series devices need additional temperature offset calibration as well as the IWN_GP_DRIVER_CALIB_VER6 bit set. Obtained from: OpenBSD
2011-04-15 20:35:15 +00:00
error = iwn5000_temp_offset_calib(sc);
if (error != 0) {
device_printf(sc->sc_dev,
"%s: could not set temperature offset\n", __func__);
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
return (error);
}
} else if (sc->base_params->calib_need & IWN_FLG_NEED_PHY_CALIB_TEMP_OFFSETv2) {
error = iwn5000_temp_offset_calibv2(sc);
if (error != 0) {
device_printf(sc->sc_dev,
"%s: could not compute temperature offset v2\n",
__func__);
return (error);
The 6005 series devices need additional temperature offset calibration as well as the IWN_GP_DRIVER_CALIB_VER6 bit set. Obtained from: OpenBSD
2011-04-15 20:35:15 +00:00
}
}
The DC calibration result obtained during initialization can't be passed over to the runtime firmware on 6050 devices. Instead let the runtime firmware do the calibration itself. This fixes support for the 6050 series devices. Obtained from: OpenBSD Submitted by: kevlo Tested by: lx, Tz-Huan Huang(earlier version)
2011-11-21 22:19:12 +00:00
if (sc->hw_type == IWN_HW_REV_TYPE_6050) {
/* Configure runtime DC calibration. */
error = iwn5000_runtime_calib(sc);
if (error != 0) {
device_printf(sc->sc_dev,
"%s: could not configure runtime calibration\n",
__func__);
return error;
}
}
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Configure valid TX chains for >=5000 Series. */
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
if (sc->hw_type != IWN_HW_REV_TYPE_4965 &&
IWN_UCODE_API(sc->ucode_rev) > 1) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
txmask = htole32(sc->txchainmask);
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
DPRINTF(sc, IWN_DEBUG_RESET | IWN_DEBUG_XMIT,
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
"%s: configuring valid TX chains 0x%x\n", __func__, txmask);
error = iwn_cmd(sc, IWN5000_CMD_TX_ANT_CONFIG, &txmask,
sizeof txmask, 0);
if (error != 0) {
device_printf(sc->sc_dev,
"%s: could not configure valid TX chains, "
"error %d\n", __func__, error);
return error;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
/* Configure bluetooth coexistence. */
Begin fleshing out a knob to enable/disable bluetooth coexistence. Some firmware versions seem to get very unhappy if they're sent btcoex commands when they don't actually have bluetooth hardware in them. So, disable sending them those commands. Tested: * 5100 (which has bluetooth, no problems) * 4965 (which doesn't have bluetooth, but didn't seem to crash) * 6200 (no bluetooth, seems to get unhappy being sent bluetooth commands.)
2013-10-25 19:46:52 +00:00
error = 0;
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
/* Configure bluetooth coexistence if needed. */
if (sc->base_params->bt_mode == IWN_BT_ADVANCED)
Add basic support for advanced bluetooth coexistence required for 6005 gen2b (1030/6030) adapters.
2011-04-20 16:59:27 +00:00
error = iwn_send_advanced_btcoex(sc);
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->bt_mode == IWN_BT_SIMPLE)
Add basic support for advanced bluetooth coexistence required for 6005 gen2b (1030/6030) adapters.
2011-04-20 16:59:27 +00:00
error = iwn_send_btcoex(sc);
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
"%s: could not configure bluetooth coexistence, error %d\n",
__func__, error);
return error;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Set mode, channel, RX filter and enable RX. */
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon = &sc->rx_on[IWN_RXON_BSS_CTX];
memset(sc->rxon, 0, sizeof (struct iwn_rxon));
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
macaddr = vap ? vap->iv_myaddr : ic->ic_macaddr;
IEEE80211_ADDR_COPY(sc->rxon->myaddr, macaddr);
IEEE80211_ADDR_COPY(sc->rxon->wlap, macaddr);
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->chan = ieee80211_chan2ieee(ic, ic->ic_curchan);
sc->rxon->flags = htole32(IWN_RXON_TSF | IWN_RXON_CTS_TO_SELF);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (IEEE80211_IS_CHAN_2GHZ(ic->ic_curchan))
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->flags |= htole32(IWN_RXON_AUTO | IWN_RXON_24GHZ);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
switch (ic->ic_opmode) {
case IEEE80211_M_STA:
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->mode = IWN_MODE_STA;
sc->rxon->filter = htole32(IWN_FILTER_MULTICAST);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
break;
case IEEE80211_M_MONITOR:
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->mode = IWN_MODE_MONITOR;
sc->rxon->filter = htole32(IWN_FILTER_MULTICAST |
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_FILTER_CTL | IWN_FILTER_PROMISC);
break;
default:
/* Should not get there. */
break;
}
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->cck_mask = 0x0f; /* not yet negotiated */
sc->rxon->ofdm_mask = 0xff; /* not yet negotiated */
sc->rxon->ht_single_mask = 0xff;
sc->rxon->ht_dual_mask = 0xff;
sc->rxon->ht_triple_mask = 0xff;
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
/*
* In active association mode, ensure that
* all the receive chains are enabled.
*
* Since we're not yet doing SMPS, don't allow the
* number of idle RX chains to be less than the active
* number.
*/
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
rxchain =
IWN_RXCHAIN_VALID(sc->rxchainmask) |
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
IWN_RXCHAIN_MIMO_COUNT(sc->nrxchains) |
IWN_RXCHAIN_IDLE_COUNT(sc->nrxchains);
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->rxchain = htole16(rxchain);
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
DPRINTF(sc, IWN_DEBUG_RESET | IWN_DEBUG_XMIT,
"%s: rxchainmask=0x%x, nrxchains=%d\n",
__func__,
sc->rxchainmask,
sc->nrxchains);
Attempt to make 5GHz HT/40 work on the 6xxx series NICs. The 6205 (Taylor Peak) in the Lenovo X230 works fine in 5GHz 11a and 11n HT20, but not 11n HT40. The NIC goes RX deaf the moment HT40 is configured. It's so RX deaf that it doesn't even hear beacons and the firmware sends "BEACON MISS" events. That's pretty deaf. I tried configuring up the HT40 flags in monitor mode and it worked - so I assumed that doing the transition from 20 -> 40MHz channel configuration when going auth->assoc (ie, after the NIC has been partially configured) is a problem. So for now, let's just always set them if they're available. Tested: * Intel 5300, STA mode, 5GHz HT/40 AP; 2GHz HT/20 AP * Intel 6205, STA mode, 5GHz HT/40, HT20, 11a AP; 2GHz HT/20 AP This was pointed out to me by coworkers trying to use FreeBSD-HEAD in the office on their Thinkpad T420p laptops. TODO: * I don't like how the HT40 flags are configured - the whole interop/ protection config should be re-checked. Notably, I think curhtprotmode is 0 in a lot of cases, which means "no interoperability" and i think that's busted. Sponsored by: Norse Corp, Inc.
2015-07-07 03:51:29 +00:00
sc->rxon->flags |= htole32(iwn_get_rxon_ht_flags(sc, ic->ic_curchan));
DPRINTF(sc, IWN_DEBUG_RESET,
"%s: setting configuration; flags=0x%08x\n",
__func__, le32toh(sc->rxon->flags));
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
if (sc->sc_is_scanning)
device_printf(sc->sc_dev,
"%s: is_scanning set, before RXON\n",
__func__);
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
error = iwn_cmd(sc, IWN_CMD_RXON, sc->rxon, sc->rxonsz, 0);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (error != 0) {
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
device_printf(sc->sc_dev, "%s: RXON command failed\n",
__func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
return error;
}
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_add_broadcast_node(sc, 0)) != 0) {
device_printf(sc->sc_dev, "%s: could not add broadcast node\n",
__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
}
/* Configuration has changed, set TX power accordingly. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
if ((error = ops->set_txpower(sc, ic->ic_curchan, 0)) != 0) {
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
device_printf(sc->sc_dev, "%s: could not set TX power\n",
__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
}
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_set_critical_temp(sc)) != 0) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
device_printf(sc->sc_dev,
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
"%s: could not set critical temperature\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Set power saving level to CAM during initialization. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_set_pslevel(sc, 0, 0, 0)) != 0) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
device_printf(sc->sc_dev,
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
"%s: could not set power saving level\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return 0;
}
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
static uint16_t
iwn_get_active_dwell_time(struct iwn_softc *sc,
struct ieee80211_channel *c, uint8_t n_probes)
{
/* No channel? Default to 2GHz settings */
if (c == NULL || IEEE80211_IS_CHAN_2GHZ(c)) {
return (IWN_ACTIVE_DWELL_TIME_2GHZ +
IWN_ACTIVE_DWELL_FACTOR_2GHZ * (n_probes + 1));
}
/* 5GHz dwell time */
return (IWN_ACTIVE_DWELL_TIME_5GHZ +
IWN_ACTIVE_DWELL_FACTOR_5GHZ * (n_probes + 1));
}
/*
* Limit the total dwell time to 85% of the beacon interval.
*
* Returns the dwell time in milliseconds.
*/
static uint16_t
iwn_limit_dwell(struct iwn_softc *sc, uint16_t dwell_time)
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
struct ieee80211vap *vap = NULL;
int bintval = 0;
/* bintval is in TU (1.024mS) */
if (! TAILQ_EMPTY(&ic->ic_vaps)) {
vap = TAILQ_FIRST(&ic->ic_vaps);
bintval = vap->iv_bss->ni_intval;
}
/*
* If it's non-zero, we should calculate the minimum of
* it and the DWELL_BASE.
*
* XXX Yes, the math should take into account that bintval
* is 1.024mS, not 1mS..
*/
if (bintval > 0) {
DPRINTF(sc, IWN_DEBUG_SCAN,
"%s: bintval=%d\n",
__func__,
bintval);
return (MIN(IWN_PASSIVE_DWELL_BASE, ((bintval * 85) / 100)));
}
/* No association context? Default */
return (IWN_PASSIVE_DWELL_BASE);
}
static uint16_t
iwn_get_passive_dwell_time(struct iwn_softc *sc, struct ieee80211_channel *c)
{
uint16_t passive;
if (c == NULL || IEEE80211_IS_CHAN_2GHZ(c)) {
passive = IWN_PASSIVE_DWELL_BASE + IWN_PASSIVE_DWELL_TIME_2GHZ;
} else {
passive = IWN_PASSIVE_DWELL_BASE + IWN_PASSIVE_DWELL_TIME_5GHZ;
}
/* Clamp to the beacon interval if we're associated */
return (iwn_limit_dwell(sc, passive));
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Refactor out the scan id and scan vap as part of the scan work. Make the scan state optional - we'll obviously need a vap, but we now won't require the scan state. the only thing the scan state is needed for is to check for the list of SSIDs to scan - which we can now just plain ignore by passing in NULL as the scan state pointer. Tested: * Intel 5100 (STA)
2013-12-07 08:32:15 +00:00
iwn_scan(struct iwn_softc *sc, struct ieee80211vap *vap,
struct ieee80211_scan_state *ss, struct ieee80211_channel *c)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Refactor out the scan id and scan vap as part of the scan work. Make the scan state optional - we'll obviously need a vap, but we now won't require the scan state. the only thing the scan state is needed for is to check for the list of SSIDs to scan - which we can now just plain ignore by passing in NULL as the scan state pointer. Tested: * Intel 5100 (STA)
2013-12-07 08:32:15 +00:00
struct ieee80211_node *ni = vap->iv_bss;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_scan_hdr *hdr;
struct iwn_cmd_data *tx;
struct iwn_scan_essid *essid;
struct iwn_scan_chan *chan;
struct ieee80211_frame *wh;
struct ieee80211_rateset *rs;
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
uint8_t *buf, *frm;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
uint16_t rxchain;
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
uint8_t txant;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
int buflen, error;
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
int is_active;
uint16_t dwell_active, dwell_passive;
uint32_t extra, scan_service_time;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
/*
* We are absolutely not allowed to send a scan command when another
* scan command is pending.
*/
if (sc->sc_is_scanning) {
device_printf(sc->sc_dev, "%s: called whilst scanning!\n",
__func__);
return (EAGAIN);
}
Refactor out the scan channel to be assigned early on in iwn_scan() rather than it all being a mess of 'c' and 'ic->ic_curchan'. Tested: * Intel 5100 (STA)
2013-12-07 08:20:24 +00:00
/* Assign the scan channel */
c = ic->ic_curchan;
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon = &sc->rx_on[IWN_RXON_BSS_CTX];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
buf = malloc(IWN_SCAN_MAXSZ, M_DEVBUF, M_NOWAIT | M_ZERO);
if (buf == NULL) {
device_printf(sc->sc_dev,
"%s: could not allocate buffer for scan command\n",
__func__);
return ENOMEM;
}
hdr = (struct iwn_scan_hdr *)buf;
/*
* Move to the next channel if no frames are received within 10ms
* after sending the probe request.
*/
hdr->quiet_time = htole16(10); /* timeout in milliseconds */
hdr->quiet_threshold = htole16(1); /* min # of packets */
Fix (I think!) the scan timeouts on the intel NICs. This field needs to be (a) set, and (b) greater than the other timeouts (passive, active, maxquiet, etc.) It also is in microseconds, not milliseconds. I hope this will fix the scan hangs that people are seeing. Obtained from: Linux iwlwifi
2013-11-14 07:27:00 +00:00
/*
* Max needs to be greater than active and passive and quiet!
* It's also in microseconds!
*/
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
hdr->max_svc = htole32(250 * 1024);
/*
* Reset scan: interval=100
* Normal scan: interval=becaon interval
* suspend_time: 100 (TU)
*
*/
extra = (100 /* suspend_time */ / 100 /* beacon interval */) << 22;
//scan_service_time = extra | ((100 /* susp */ % 100 /* int */) * 1024);
scan_service_time = (4 << 22) | (100 * 1024); /* Hardcode for now! */
hdr->pause_svc = htole32(scan_service_time);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Select antennas for scanning. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
rxchain =
IWN_RXCHAIN_VALID(sc->rxchainmask) |
IWN_RXCHAIN_FORCE_MIMO_SEL(sc->rxchainmask) |
IWN_RXCHAIN_DRIVER_FORCE;
Refactor out the scan channel to be assigned early on in iwn_scan() rather than it all being a mess of 'c' and 'ic->ic_curchan'. Tested: * Intel 5100 (STA)
2013-12-07 08:20:24 +00:00
if (IEEE80211_IS_CHAN_A(c) &&
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
sc->hw_type == IWN_HW_REV_TYPE_4965) {
/* Ant A must be avoided in 5GHz because of an HW bug. */
The firmware of 4965 series adapters seems to die while trying to send probe requests at 1Mbps while being associated on a 5GHz channel. Sending those at 6Mbps does work, so use that instead during a background scan. This workaround allows us to re-enable background scan support for the 4965 adapters. Also, just enabling one antenna on 5GHz results in better reception of beacons: test 00:26:5a:c6:14:1a 40 54M -71:-95 200 E WME HTCAP ATH vs test 00:26:5a:c6:14:1a 40 54M -92:-95 200 E WME HTCAP ATH Due to roam:rssi thresholds set to 7 by default it might have been impossible to associate to that network. While here use IEEE80211_IS_CHAN_5GHZ() to determine the band.
2011-06-04 11:43:09 +00:00
rxchain |= IWN_RXCHAIN_FORCE_SEL(IWN_ANT_B);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
} else /* Use all available RX antennas. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
rxchain |= IWN_RXCHAIN_FORCE_SEL(sc->rxchainmask);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
hdr->rxchain = htole16(rxchain);
hdr->filter = htole32(IWN_FILTER_MULTICAST | IWN_FILTER_BEACON);
tx = (struct iwn_cmd_data *)(hdr + 1);
tx->flags = htole32(IWN_TX_AUTO_SEQ);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
tx->id = sc->broadcast_id;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
tx->lifetime = htole32(IWN_LIFETIME_INFINITE);
Refactor out the scan channel to be assigned early on in iwn_scan() rather than it all being a mess of 'c' and 'ic->ic_curchan'. Tested: * Intel 5100 (STA)
2013-12-07 08:20:24 +00:00
if (IEEE80211_IS_CHAN_5GHZ(c)) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Send probe requests at 6Mbps. */
Prepare for transmitting frames at MCS rates: - instead of calling iwn_plcp_signal() for every frame, map the expected value directly within wn->ridx - concat plcp, rflags and xrflags, there is no clean byte boundary within the flags, for example the antenna setting uses bit 6, 7 and 8 - there is still need for a custom rate to plcp mapping, as those expected by the hardware are not conform to the std
2011-05-08 11:49:50 +00:00
tx->rate = htole32(0xd);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
rs = &ic->ic_sup_rates[IEEE80211_MODE_11A];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
} else {
hdr->flags = htole32(IWN_RXON_24GHZ | IWN_RXON_AUTO);
The firmware of 4965 series adapters seems to die while trying to send probe requests at 1Mbps while being associated on a 5GHz channel. Sending those at 6Mbps does work, so use that instead during a background scan. This workaround allows us to re-enable background scan support for the 4965 adapters. Also, just enabling one antenna on 5GHz results in better reception of beacons: test 00:26:5a:c6:14:1a 40 54M -71:-95 200 E WME HTCAP ATH vs test 00:26:5a:c6:14:1a 40 54M -92:-95 200 E WME HTCAP ATH Due to roam:rssi thresholds set to 7 by default it might have been impossible to associate to that network. While here use IEEE80211_IS_CHAN_5GHZ() to determine the band.
2011-06-04 11:43:09 +00:00
if (sc->hw_type == IWN_HW_REV_TYPE_4965 &&
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->associd && sc->rxon->chan > 14)
The firmware of 4965 series adapters seems to die while trying to send probe requests at 1Mbps while being associated on a 5GHz channel. Sending those at 6Mbps does work, so use that instead during a background scan. This workaround allows us to re-enable background scan support for the 4965 adapters. Also, just enabling one antenna on 5GHz results in better reception of beacons: test 00:26:5a:c6:14:1a 40 54M -71:-95 200 E WME HTCAP ATH vs test 00:26:5a:c6:14:1a 40 54M -92:-95 200 E WME HTCAP ATH Due to roam:rssi thresholds set to 7 by default it might have been impossible to associate to that network. While here use IEEE80211_IS_CHAN_5GHZ() to determine the band.
2011-06-04 11:43:09 +00:00
tx->rate = htole32(0xd);
else {
/* Send probe requests at 1Mbps. */
tx->rate = htole32(10 | IWN_RFLAG_CCK);
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
rs = &ic->ic_sup_rates[IEEE80211_MODE_11G];
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
/* Use the first valid TX antenna. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
txant = IWN_LSB(sc->txchainmask);
Prepare for transmitting frames at MCS rates: - instead of calling iwn_plcp_signal() for every frame, map the expected value directly within wn->ridx - concat plcp, rflags and xrflags, there is no clean byte boundary within the flags, for example the antenna setting uses bit 6, 7 and 8 - there is still need for a custom rate to plcp mapping, as those expected by the hardware are not conform to the std
2011-05-08 11:49:50 +00:00
tx->rate |= htole32(IWN_RFLAG_ANT(txant));
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
/*
* Only do active scanning if we're announcing a probe request
* for a given SSID (or more, if we ever add it to the driver.)
*/
is_active = 0;
/*
* If we're scanning for a specific SSID, add it to the command.
Refactor out the scan id and scan vap as part of the scan work. Make the scan state optional - we'll obviously need a vap, but we now won't require the scan state. the only thing the scan state is needed for is to check for the list of SSIDs to scan - which we can now just plain ignore by passing in NULL as the scan state pointer. Tested: * Intel 5100 (STA)
2013-12-07 08:32:15 +00:00
*
* XXX maybe look at adding support for scanning multiple SSIDs?
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
*/
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
essid = (struct iwn_scan_essid *)(tx + 1);
Refactor out the scan id and scan vap as part of the scan work. Make the scan state optional - we'll obviously need a vap, but we now won't require the scan state. the only thing the scan state is needed for is to check for the list of SSIDs to scan - which we can now just plain ignore by passing in NULL as the scan state pointer. Tested: * Intel 5100 (STA)
2013-12-07 08:32:15 +00:00
if (ss != NULL) {
if (ss->ss_ssid[0].len != 0) {
essid[0].id = IEEE80211_ELEMID_SSID;
essid[0].len = ss->ss_ssid[0].len;
memcpy(essid[0].data, ss->ss_ssid[0].ssid, ss->ss_ssid[0].len);
}
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
Refactor out the scan id and scan vap as part of the scan work. Make the scan state optional - we'll obviously need a vap, but we now won't require the scan state. the only thing the scan state is needed for is to check for the list of SSIDs to scan - which we can now just plain ignore by passing in NULL as the scan state pointer. Tested: * Intel 5100 (STA)
2013-12-07 08:32:15 +00:00
DPRINTF(sc, IWN_DEBUG_SCAN, "%s: ssid_len=%d, ssid=%*s\n",
__func__,
ss->ss_ssid[0].len,
ss->ss_ssid[0].len,
ss->ss_ssid[0].ssid);
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
Refactor out the scan id and scan vap as part of the scan work. Make the scan state optional - we'll obviously need a vap, but we now won't require the scan state. the only thing the scan state is needed for is to check for the list of SSIDs to scan - which we can now just plain ignore by passing in NULL as the scan state pointer. Tested: * Intel 5100 (STA)
2013-12-07 08:32:15 +00:00
if (ss->ss_nssid > 0)
is_active = 1;
}
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Build a probe request frame. Most of the following code is a
* copy & paste of what is done in net80211.
*/
wh = (struct ieee80211_frame *)(essid + 20);
wh->i_fc[0] = IEEE80211_FC0_VERSION_0 | IEEE80211_FC0_TYPE_MGT |
IEEE80211_FC0_SUBTYPE_PROBE_REQ;
wh->i_fc[1] = IEEE80211_FC1_DIR_NODS;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
IEEE80211_ADDR_COPY(wh->i_addr1, vap->iv_ifp->if_broadcastaddr);
IEEE80211_ADDR_COPY(wh->i_addr2, IF_LLADDR(vap->iv_ifp));
IEEE80211_ADDR_COPY(wh->i_addr3, vap->iv_ifp->if_broadcastaddr);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*(uint16_t *)&wh->i_dur[0] = 0; /* filled by HW */
*(uint16_t *)&wh->i_seq[0] = 0; /* filled by HW */
frm = (uint8_t *)(wh + 1);
Reuse net80211 code: - IWN_TXOP_TO_US is equal to IEEE80211_TXOP_TO_US - use IEEE80211_DUR_TU - ieee80211_add_rates/ieee80211_add_xrates are public, use em - copied ieee80211_add_ssid it is not public
2011-04-14 17:26:13 +00:00
frm = ieee80211_add_ssid(frm, NULL, 0);
frm = ieee80211_add_rates(frm, rs);
if (rs->rs_nrates > IEEE80211_RATE_SIZE)
frm = ieee80211_add_xrates(frm, rs);
Add HT capabilities to probe requests.
2011-05-08 11:03:16 +00:00
if (ic->ic_htcaps & IEEE80211_HTC_HT)
frm = ieee80211_add_htcap(frm, ni);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Set length of probe request. */
tx->len = htole16(frm - (uint8_t *)wh);
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
/*
* If active scanning is requested but a certain channel is
* marked passive, we can do active scanning if we detect
* transmissions.
*
* There is an issue with some firmware versions that triggers
* a sysassert on a "good CRC threshold" of zero (== disabled),
* on a radar channel even though this means that we should NOT
* send probes.
*
* The "good CRC threshold" is the number of frames that we
* need to receive during our dwell time on a channel before
* sending out probes -- setting this to a huge value will
* mean we never reach it, but at the same time work around
* the aforementioned issue. Thus use IWL_GOOD_CRC_TH_NEVER
* here instead of IWL_GOOD_CRC_TH_DISABLED.
*
* This was fixed in later versions along with some other
* scan changes, and the threshold behaves as a flag in those
* versions.
*/
/*
* If we're doing active scanning, set the crc_threshold
* to a suitable value. This is different to active veruss
* passive scanning depending upon the channel flags; the
* firmware will obey that particular check for us.
*/
if (sc->tlv_feature_flags & IWN_UCODE_TLV_FLAGS_NEWSCAN)
hdr->crc_threshold = is_active ?
IWN_GOOD_CRC_TH_DEFAULT : IWN_GOOD_CRC_TH_DISABLED;
else
hdr->crc_threshold = is_active ?
IWN_GOOD_CRC_TH_DEFAULT : IWN_GOOD_CRC_TH_NEVER;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
chan = (struct iwn_scan_chan *)frm;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
chan->chan = htole16(ieee80211_chan2ieee(ic, c));
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
chan->flags = 0;
if (ss->ss_nssid > 0)
chan->flags |= htole32(IWN_CHAN_NPBREQS(1));
chan->dsp_gain = 0x6e;
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
/*
* Set the passive/active flag depending upon the channel mode.
* XXX TODO: take the is_active flag into account as well?
*/
if (c->ic_flags & IEEE80211_CHAN_PASSIVE)
chan->flags |= htole32(IWN_CHAN_PASSIVE);
else
chan->flags |= htole32(IWN_CHAN_ACTIVE);
/*
* Calculate the active/passive dwell times.
*/
dwell_active = iwn_get_active_dwell_time(sc, c, ss->ss_nssid);
dwell_passive = iwn_get_passive_dwell_time(sc, c);
/* Make sure they're valid */
if (dwell_passive <= dwell_active)
dwell_passive = dwell_active + 1;
chan->active = htole16(dwell_active);
chan->passive = htole16(dwell_passive);
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
if (IEEE80211_IS_CHAN_5GHZ(c))
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
chan->rf_gain = 0x3b;
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
else
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
chan->rf_gain = 0x28;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
DPRINTF(sc, IWN_DEBUG_STATE,
"%s: chan %u flags 0x%x rf_gain 0x%x "
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
"dsp_gain 0x%x active %d passive %d scan_svc_time %d crc 0x%x "
"isactive=%d numssid=%d\n", __func__,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
chan->chan, chan->flags, chan->rf_gain, chan->dsp_gain,
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
dwell_active, dwell_passive, scan_service_time,
hdr->crc_threshold, is_active, ss->ss_nssid);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
hdr->nchan++;
chan++;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
buflen = (uint8_t *)chan - buf;
hdr->len = htole16(buflen);
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
if (sc->sc_is_scanning) {
device_printf(sc->sc_dev,
"%s: called with is_scanning set!\n",
__func__);
}
sc->sc_is_scanning = 1;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DPRINTF(sc, IWN_DEBUG_STATE, "sending scan command nchan=%d\n",
hdr->nchan);
error = iwn_cmd(sc, IWN_CMD_SCAN, buf, buflen, 1);
free(buf, M_DEVBUF);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
iwn_auth(struct iwn_softc *sc, struct ieee80211vap *vap)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct ieee80211_node *ni = vap->iv_bss;
int error;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon = &sc->rx_on[IWN_RXON_BSS_CTX];
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Update adapter configuration. */
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
IEEE80211_ADDR_COPY(sc->rxon->bssid, ni->ni_bssid);
sc->rxon->chan = ieee80211_chan2ieee(ic, ni->ni_chan);
sc->rxon->flags = htole32(IWN_RXON_TSF | IWN_RXON_CTS_TO_SELF);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (IEEE80211_IS_CHAN_2GHZ(ni->ni_chan))
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->flags |= htole32(IWN_RXON_AUTO | IWN_RXON_24GHZ);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (ic->ic_flags & IEEE80211_F_SHSLOT)
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->flags |= htole32(IWN_RXON_SHSLOT);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (ic->ic_flags & IEEE80211_F_SHPREAMBLE)
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->flags |= htole32(IWN_RXON_SHPREAMBLE);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (IEEE80211_IS_CHAN_A(ni->ni_chan)) {
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->cck_mask = 0;
sc->rxon->ofdm_mask = 0x15;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
} else if (IEEE80211_IS_CHAN_B(ni->ni_chan)) {
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->cck_mask = 0x03;
sc->rxon->ofdm_mask = 0;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
} else {
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Assume 802.11b/g. */
Add some initial support for the Intel 6235. Tested: * Intel 5100 * Intel 6235 Obtained from: mav, others
2013-12-09 03:40:02 +00:00
sc->rxon->cck_mask = 0x03;
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->ofdm_mask = 0x15;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Attempt to make 5GHz HT/40 work on the 6xxx series NICs. The 6205 (Taylor Peak) in the Lenovo X230 works fine in 5GHz 11a and 11n HT20, but not 11n HT40. The NIC goes RX deaf the moment HT40 is configured. It's so RX deaf that it doesn't even hear beacons and the firmware sends "BEACON MISS" events. That's pretty deaf. I tried configuring up the HT40 flags in monitor mode and it worked - so I assumed that doing the transition from 20 -> 40MHz channel configuration when going auth->assoc (ie, after the NIC has been partially configured) is a problem. So for now, let's just always set them if they're available. Tested: * Intel 5300, STA mode, 5GHz HT/40 AP; 2GHz HT/20 AP * Intel 6205, STA mode, 5GHz HT/40, HT20, 11a AP; 2GHz HT/20 AP This was pointed out to me by coworkers trying to use FreeBSD-HEAD in the office on their Thinkpad T420p laptops. TODO: * I don't like how the HT40 flags are configured - the whole interop/ protection config should be re-checked. Notably, I think curhtprotmode is 0 in a lot of cases, which means "no interoperability" and i think that's busted. Sponsored by: Norse Corp, Inc.
2015-07-07 03:51:29 +00:00
/* try HT */
sc->rxon->flags |= htole32(iwn_get_rxon_ht_flags(sc, ic->ic_curchan));
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
DPRINTF(sc, IWN_DEBUG_STATE, "rxon chan %d flags %x cck %x ofdm %x\n",
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->chan, sc->rxon->flags, sc->rxon->cck_mask,
sc->rxon->ofdm_mask);
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
if (sc->sc_is_scanning)
device_printf(sc->sc_dev,
"%s: is_scanning set, before RXON\n",
__func__);
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
error = iwn_cmd(sc, IWN_CMD_RXON, sc->rxon, sc->rxonsz, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
device_printf(sc->sc_dev, "%s: RXON command failed, error %d\n",
__func__, error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Configuration has changed, set TX power accordingly. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
if ((error = ops->set_txpower(sc, ni->ni_chan, 1)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(sc->sc_dev,
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
"%s: could not set TX power, error %d\n", __func__, error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
* Reconfiguring RXON clears the firmware nodes table so we must
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
* add the broadcast node again.
*/
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_add_broadcast_node(sc, 1)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(sc->sc_dev,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"%s: could not add broadcast node, error %d\n", __func__,
error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return 0;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
iwn_run(struct iwn_softc *sc, struct ieee80211vap *vap)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
struct ieee80211_node *ni = vap->iv_bss;
struct iwn_node_info node;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
int error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon = &sc->rx_on[IWN_RXON_BSS_CTX];
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (ic->ic_opmode == IEEE80211_M_MONITOR) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Link LED blinks while monitoring. */
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
iwn_set_led(sc, IWN_LED_LINK, 5, 5);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return 0;
}
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_set_timing(sc, ni)) != 0) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
device_printf(sc->sc_dev,
"%s: could not set timing, error %d\n", __func__, error);
return error;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Update adapter configuration. */
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
IEEE80211_ADDR_COPY(sc->rxon->bssid, ni->ni_bssid);
sc->rxon->associd = htole16(IEEE80211_AID(ni->ni_associd));
sc->rxon->chan = ieee80211_chan2ieee(ic, ni->ni_chan);
sc->rxon->flags = htole32(IWN_RXON_TSF | IWN_RXON_CTS_TO_SELF);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (IEEE80211_IS_CHAN_2GHZ(ni->ni_chan))
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->flags |= htole32(IWN_RXON_AUTO | IWN_RXON_24GHZ);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (ic->ic_flags & IEEE80211_F_SHSLOT)
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->flags |= htole32(IWN_RXON_SHSLOT);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (ic->ic_flags & IEEE80211_F_SHPREAMBLE)
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->flags |= htole32(IWN_RXON_SHPREAMBLE);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (IEEE80211_IS_CHAN_A(ni->ni_chan)) {
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->cck_mask = 0;
sc->rxon->ofdm_mask = 0x15;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
} else if (IEEE80211_IS_CHAN_B(ni->ni_chan)) {
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->cck_mask = 0x03;
sc->rxon->ofdm_mask = 0;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
} else {
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Assume 802.11b/g. */
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->cck_mask = 0x0f;
sc->rxon->ofdm_mask = 0x15;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Attempt to make 5GHz HT/40 work on the 6xxx series NICs. The 6205 (Taylor Peak) in the Lenovo X230 works fine in 5GHz 11a and 11n HT20, but not 11n HT40. The NIC goes RX deaf the moment HT40 is configured. It's so RX deaf that it doesn't even hear beacons and the firmware sends "BEACON MISS" events. That's pretty deaf. I tried configuring up the HT40 flags in monitor mode and it worked - so I assumed that doing the transition from 20 -> 40MHz channel configuration when going auth->assoc (ie, after the NIC has been partially configured) is a problem. So for now, let's just always set them if they're available. Tested: * Intel 5300, STA mode, 5GHz HT/40 AP; 2GHz HT/20 AP * Intel 6205, STA mode, 5GHz HT/40, HT20, 11a AP; 2GHz HT/20 AP This was pointed out to me by coworkers trying to use FreeBSD-HEAD in the office on their Thinkpad T420p laptops. TODO: * I don't like how the HT40 flags are configured - the whole interop/ protection config should be re-checked. Notably, I think curhtprotmode is 0 in a lot of cases, which means "no interoperability" and i think that's busted. Sponsored by: Norse Corp, Inc.
2015-07-07 03:51:29 +00:00
/* try HT */
sc->rxon->flags |= htole32(iwn_get_rxon_ht_flags(sc, ni->ni_chan));
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
sc->rxon->filter |= htole32(IWN_FILTER_BSS);
Attempt to make 5GHz HT/40 work on the 6xxx series NICs. The 6205 (Taylor Peak) in the Lenovo X230 works fine in 5GHz 11a and 11n HT20, but not 11n HT40. The NIC goes RX deaf the moment HT40 is configured. It's so RX deaf that it doesn't even hear beacons and the firmware sends "BEACON MISS" events. That's pretty deaf. I tried configuring up the HT40 flags in monitor mode and it worked - so I assumed that doing the transition from 20 -> 40MHz channel configuration when going auth->assoc (ie, after the NIC has been partially configured) is a problem. So for now, let's just always set them if they're available. Tested: * Intel 5300, STA mode, 5GHz HT/40 AP; 2GHz HT/20 AP * Intel 6205, STA mode, 5GHz HT/40, HT20, 11a AP; 2GHz HT/20 AP This was pointed out to me by coworkers trying to use FreeBSD-HEAD in the office on their Thinkpad T420p laptops. TODO: * I don't like how the HT40 flags are configured - the whole interop/ protection config should be re-checked. Notably, I think curhtprotmode is 0 in a lot of cases, which means "no interoperability" and i think that's busted. Sponsored by: Norse Corp, Inc.
2015-07-07 03:51:29 +00:00
DPRINTF(sc, IWN_DEBUG_STATE, "rxon chan %d flags %x, curhtprotmode=%d\n",
sc->rxon->chan, le32toh(sc->rxon->flags), ic->ic_curhtprotmode);
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
if (sc->sc_is_scanning)
device_printf(sc->sc_dev,
"%s: is_scanning set, before RXON\n",
__func__);
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
error = iwn_cmd(sc, IWN_CMD_RXON, sc->rxon, sc->rxonsz, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"%s: could not update configuration, error %d\n", __func__,
error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Configuration has changed, set TX power accordingly. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
if ((error = ops->set_txpower(sc, ni->ni_chan, 1)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(sc->sc_dev,
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
"%s: could not set TX power, error %d\n", __func__, error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Instead of hardcoding TX rates and using that to fill the retry table use the neogotiated ni_rates instead.
2011-04-16 12:42:54 +00:00
/* Fake a join to initialize the TX rate. */
((struct iwn_node *)ni)->id = IWN_ID_BSS;
iwn_newassoc(ni, 1);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Add BSS node. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
memset(&node, 0, sizeof node);
IEEE80211_ADDR_COPY(node.macaddr, ni->ni_macaddr);
node.id = IWN_ID_BSS;
Notify firmware about various HT parameters once associated.
2011-05-08 12:11:20 +00:00
if (IEEE80211_IS_CHAN_HT(ni->ni_chan)) {
switch (ni->ni_htcap & IEEE80211_HTCAP_SMPS) {
case IEEE80211_HTCAP_SMPS_ENA:
node.htflags |= htole32(IWN_SMPS_MIMO_DIS);
break;
case IEEE80211_HTCAP_SMPS_DYNAMIC:
node.htflags |= htole32(IWN_SMPS_MIMO_PROT);
break;
}
node.htflags |= htole32(IWN_AMDPU_SIZE_FACTOR(3) |
IWN_AMDPU_DENSITY(5)); /* 4us */
if (IEEE80211_IS_CHAN_HT40(ni->ni_chan))
node.htflags |= htole32(IWN_NODE_HT40);
}
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
DPRINTF(sc, IWN_DEBUG_STATE, "%s: adding BSS node\n", __func__);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
error = ops->add_node(sc, &node, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(sc->sc_dev,
"%s: could not add BSS node, error %d\n", __func__, error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
DPRINTF(sc, IWN_DEBUG_STATE, "%s: setting link quality for node %d\n",
__func__, node.id);
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_set_link_quality(sc, ni)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(sc->sc_dev,
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
"%s: could not setup link quality for node %d, error %d\n",
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
__func__, node.id, error);
return error;
}
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_init_sensitivity(sc)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(sc->sc_dev,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"%s: could not set sensitivity, error %d\n", __func__,
error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Start periodic calibration timer. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
sc->calib.state = IWN_CALIB_STATE_ASSOC;
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
sc->calib_cnt = 0;
callout_reset(&sc->calib_to, msecs_to_ticks(500), iwn_calib_timeout,
sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Link LED always on while associated. */
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_set_led(sc, IWN_LED_LINK, 0, 1);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return 0;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/*
* This function is called by upper layer when an ADDBA request is received
* from another STA and before the ADDBA response is sent.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
iwn_ampdu_rx_start(struct ieee80211_node *ni, struct ieee80211_rx_ampdu *rap,
int baparamset, int batimeout, int baseqctl)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
{
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
#define MS(_v, _f) (((_v) & _f) >> _f##_S)
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
struct iwn_softc *sc = ni->ni_ic->ic_softc;
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
struct iwn_node *wn = (void *)ni;
struct iwn_node_info node;
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
uint16_t ssn;
uint8_t tid;
int error;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
tid = MS(le16toh(baparamset), IEEE80211_BAPS_TID);
ssn = MS(le16toh(baseqctl), IEEE80211_BASEQ_START);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
memset(&node, 0, sizeof node);
node.id = wn->id;
node.control = IWN_NODE_UPDATE;
node.flags = IWN_FLAG_SET_ADDBA;
node.addba_tid = tid;
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
node.addba_ssn = htole16(ssn);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
DPRINTF(sc, IWN_DEBUG_RECV, "ADDBA RA=%d TID=%d SSN=%d\n",
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
wn->id, tid, ssn);
error = ops->add_node(sc, &node, 1);
if (error != 0)
return error;
return sc->sc_ampdu_rx_start(ni, rap, baparamset, batimeout, baseqctl);
#undef MS
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
/*
* This function is called by upper layer on teardown of an HT-immediate
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
* Block Ack agreement (eg. uppon receipt of a DELBA frame).
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
iwn_ampdu_rx_stop(struct ieee80211_node *ni, struct ieee80211_rx_ampdu *rap)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
{
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
struct ieee80211com *ic = ni->ni_ic;
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
struct iwn_softc *sc = ic->ic_softc;
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
struct iwn_node *wn = (void *)ni;
struct iwn_node_info node;
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
uint8_t tid;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
/* XXX: tid as an argument */
for (tid = 0; tid < WME_NUM_TID; tid++) {
if (&ni->ni_rx_ampdu[tid] == rap)
break;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
memset(&node, 0, sizeof node);
node.id = wn->id;
node.control = IWN_NODE_UPDATE;
node.flags = IWN_FLAG_SET_DELBA;
node.delba_tid = tid;
DPRINTF(sc, IWN_DEBUG_RECV, "DELBA RA=%d TID=%d\n", wn->id, tid);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
(void)ops->add_node(sc, &node, 1);
Add support for RX packet aggregation.
2011-05-08 11:58:23 +00:00
sc->sc_ampdu_rx_stop(ni, rap);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
static int
iwn_addba_request(struct ieee80211_node *ni, struct ieee80211_tx_ampdu *tap,
int dialogtoken, int baparamset, int batimeout)
{
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
struct iwn_softc *sc = ni->ni_ic->ic_softc;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
int qid;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
for (qid = sc->firstaggqueue; qid < sc->ntxqs; qid++) {
if (sc->qid2tap[qid] == NULL)
break;
}
if (qid == sc->ntxqs) {
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: not free aggregation queue\n",
__func__);
return 0;
}
tap->txa_private = malloc(sizeof(int), M_DEVBUF, M_NOWAIT);
if (tap->txa_private == NULL) {
device_printf(sc->sc_dev,
"%s: failed to alloc TX aggregation structure\n", __func__);
return 0;
}
sc->qid2tap[qid] = tap;
*(int *)tap->txa_private = qid;
return sc->sc_addba_request(ni, tap, dialogtoken, baparamset,
batimeout);
}
static int
iwn_addba_response(struct ieee80211_node *ni, struct ieee80211_tx_ampdu *tap,
int code, int baparamset, int batimeout)
{
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
struct iwn_softc *sc = ni->ni_ic->ic_softc;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
int qid = *(int *)tap->txa_private;
Migrate the net80211 TX aggregation state to be from per-AC to per-TID. TODO: * Test mwl(4) more thoroughly! Reviewed by: bschmidt (for iwn)
2012-04-15 20:29:39 +00:00
uint8_t tid = tap->txa_tid;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
int ret;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
if (code == IEEE80211_STATUS_SUCCESS) {
ni->ni_txseqs[tid] = tap->txa_start & 0xfff;
ret = iwn_ampdu_tx_start(ni->ni_ic, ni, tid);
if (ret != 1)
return ret;
} else {
sc->qid2tap[qid] = NULL;
free(tap->txa_private, M_DEVBUF);
tap->txa_private = NULL;
}
return sc->sc_addba_response(ni, tap, code, baparamset, batimeout);
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/*
* This function is called by upper layer when an ADDBA response is received
* from another STA.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_ampdu_tx_start(struct ieee80211com *ic, struct ieee80211_node *ni,
uint8_t tid)
{
Migrate the net80211 TX aggregation state to be from per-AC to per-TID. TODO: * Test mwl(4) more thoroughly! Reviewed by: bschmidt (for iwn)
2012-04-15 20:29:39 +00:00
struct ieee80211_tx_ampdu *tap = &ni->ni_tx_ampdu[tid];
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
struct iwn_softc *sc = ni->ni_ic->ic_softc;
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
struct iwn_node *wn = (void *)ni;
struct iwn_node_info node;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
int error, qid;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Enable TX for the specified RA/TID. */
wn->disable_tid &= ~(1 << tid);
memset(&node, 0, sizeof node);
node.id = wn->id;
node.control = IWN_NODE_UPDATE;
node.flags = IWN_FLAG_SET_DISABLE_TID;
node.disable_tid = htole16(wn->disable_tid);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
error = ops->add_node(sc, &node, 1);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (error != 0)
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
return 0;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
return 0;
qid = *(int *)tap->txa_private;
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
DPRINTF(sc, IWN_DEBUG_XMIT, "%s: ra=%d tid=%d ssn=%d qid=%d\n",
__func__, wn->id, tid, tap->txa_start, qid);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
ops->ampdu_tx_start(sc, ni, qid, tid, tap->txa_start & 0xfff);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_nic_unlock(sc);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
iwn_set_link_quality(sc, ni);
return 1;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
iwn_ampdu_tx_stop(struct ieee80211_node *ni, struct ieee80211_tx_ampdu *tap)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
{
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
struct iwn_softc *sc = ni->ni_ic->ic_softc;
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Migrate the net80211 TX aggregation state to be from per-AC to per-TID. TODO: * Test mwl(4) more thoroughly! Reviewed by: bschmidt (for iwn)
2012-04-15 20:29:39 +00:00
uint8_t tid = tap->txa_tid;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
int qid;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
sc->sc_addba_stop(ni, tap);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
if (tap->txa_private == NULL)
return;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
qid = *(int *)tap->txa_private;
We need to defer passing the DELBA request to the firmware until the aggr queue is empty or the firmware will go nuts. PR: kern/167806 Tested by: osa@, Brandon Gooch (earlier version), Bojan Petrovic (earlier version) MFC after: 3 days
2012-06-27 16:07:01 +00:00
if (sc->txq[qid].queued != 0)
return;
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if (iwn_nic_lock(sc) != 0)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
return;
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
ops->ampdu_tx_stop(sc, qid, tid, tap->txa_start & 0xfff);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_nic_unlock(sc);
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
sc->qid2tap[qid] = NULL;
free(tap->txa_private, M_DEVBUF);
tap->txa_private = NULL;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn4965_ampdu_tx_start(struct iwn_softc *sc, struct ieee80211_node *ni,
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
int qid, uint8_t tid, uint16_t ssn)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
{
struct iwn_node *wn = (void *)ni;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Stop TX scheduler while we're changing its configuration. */
iwn_prph_write(sc, IWN4965_SCHED_QUEUE_STATUS(qid),
IWN4965_TXQ_STATUS_CHGACT);
/* Assign RA/TID translation to the queue. */
iwn_mem_write_2(sc, sc->sched_base + IWN4965_SCHED_TRANS_TBL(qid),
wn->id << 4 | tid);
/* Enable chain-building mode for the queue. */
iwn_prph_setbits(sc, IWN4965_SCHED_QCHAIN_SEL, 1 << qid);
/* Set starting sequence number from the ADDBA request. */
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
sc->txq[qid].cur = sc->txq[qid].read = (ssn & 0xff);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_WRITE(sc, IWN_HBUS_TARG_WRPTR, qid << 8 | (ssn & 0xff));
iwn_prph_write(sc, IWN4965_SCHED_QUEUE_RDPTR(qid), ssn);
/* Set scheduler window size. */
iwn_mem_write(sc, sc->sched_base + IWN4965_SCHED_QUEUE_OFFSET(qid),
IWN_SCHED_WINSZ);
/* Set scheduler frame limit. */
iwn_mem_write(sc, sc->sched_base + IWN4965_SCHED_QUEUE_OFFSET(qid) + 4,
IWN_SCHED_LIMIT << 16);
/* Enable interrupts for the queue. */
iwn_prph_setbits(sc, IWN4965_SCHED_INTR_MASK, 1 << qid);
/* Mark the queue as active. */
iwn_prph_write(sc, IWN4965_SCHED_QUEUE_STATUS(qid),
IWN4965_TXQ_STATUS_ACTIVE | IWN4965_TXQ_STATUS_AGGR_ENA |
iwn_tid2fifo[tid] << 1);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
iwn4965_ampdu_tx_stop(struct iwn_softc *sc, int qid, uint8_t tid, uint16_t ssn)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
{
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Stop TX scheduler while we're changing its configuration. */
iwn_prph_write(sc, IWN4965_SCHED_QUEUE_STATUS(qid),
IWN4965_TXQ_STATUS_CHGACT);
/* Set starting sequence number from the ADDBA request. */
IWN_WRITE(sc, IWN_HBUS_TARG_WRPTR, qid << 8 | (ssn & 0xff));
iwn_prph_write(sc, IWN4965_SCHED_QUEUE_RDPTR(qid), ssn);
/* Disable interrupts for the queue. */
iwn_prph_clrbits(sc, IWN4965_SCHED_INTR_MASK, 1 << qid);
/* Mark the queue as inactive. */
iwn_prph_write(sc, IWN4965_SCHED_QUEUE_STATUS(qid),
IWN4965_TXQ_STATUS_INACTIVE | iwn_tid2fifo[tid] << 1);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn5000_ampdu_tx_start(struct iwn_softc *sc, struct ieee80211_node *ni,
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
int qid, uint8_t tid, uint16_t ssn)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
{
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
struct iwn_node *wn = (void *)ni;
/* Stop TX scheduler while we're changing its configuration. */
iwn_prph_write(sc, IWN5000_SCHED_QUEUE_STATUS(qid),
IWN5000_TXQ_STATUS_CHGACT);
/* Assign RA/TID translation to the queue. */
iwn_mem_write_2(sc, sc->sched_base + IWN5000_SCHED_TRANS_TBL(qid),
wn->id << 4 | tid);
/* Enable chain-building mode for the queue. */
iwn_prph_setbits(sc, IWN5000_SCHED_QCHAIN_SEL, 1 << qid);
/* Enable aggregation for the queue. */
iwn_prph_setbits(sc, IWN5000_SCHED_AGGR_SEL, 1 << qid);
/* Set starting sequence number from the ADDBA request. */
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
sc->txq[qid].cur = sc->txq[qid].read = (ssn & 0xff);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_WRITE(sc, IWN_HBUS_TARG_WRPTR, qid << 8 | (ssn & 0xff));
iwn_prph_write(sc, IWN5000_SCHED_QUEUE_RDPTR(qid), ssn);
/* Set scheduler window size and frame limit. */
iwn_mem_write(sc, sc->sched_base + IWN5000_SCHED_QUEUE_OFFSET(qid) + 4,
IWN_SCHED_LIMIT << 16 | IWN_SCHED_WINSZ);
/* Enable interrupts for the queue. */
iwn_prph_setbits(sc, IWN5000_SCHED_INTR_MASK, 1 << qid);
/* Mark the queue as active. */
iwn_prph_write(sc, IWN5000_SCHED_QUEUE_STATUS(qid),
IWN5000_TXQ_STATUS_ACTIVE | iwn_tid2fifo[tid]);
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Add support for TX packet aggregation.
2011-05-08 12:06:12 +00:00
iwn5000_ampdu_tx_stop(struct iwn_softc *sc, int qid, uint8_t tid, uint16_t ssn)
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
{
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Stop TX scheduler while we're changing its configuration. */
iwn_prph_write(sc, IWN5000_SCHED_QUEUE_STATUS(qid),
IWN5000_TXQ_STATUS_CHGACT);
/* Disable aggregation for the queue. */
iwn_prph_clrbits(sc, IWN5000_SCHED_AGGR_SEL, 1 << qid);
/* Set starting sequence number from the ADDBA request. */
IWN_WRITE(sc, IWN_HBUS_TARG_WRPTR, qid << 8 | (ssn & 0xff));
iwn_prph_write(sc, IWN5000_SCHED_QUEUE_RDPTR(qid), ssn);
/* Disable interrupts for the queue. */
iwn_prph_clrbits(sc, IWN5000_SCHED_INTR_MASK, 1 << qid);
/* Mark the queue as inactive. */
iwn_prph_write(sc, IWN5000_SCHED_QUEUE_STATUS(qid),
IWN5000_TXQ_STATUS_INACTIVE | iwn_tid2fifo[tid]);
}
Group functions used for initial calibration.
2010-09-19 12:39:04 +00:00
/*
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
* Query calibration tables from the initialization firmware. We do this
* only once at first boot. Called from a process context.
Rewrite parts of the calibration code which is run while bringing up the device: - unobscure some of the code by moving it into its own functions - get rid of some magic numbers - create similar structure as the reference driver has, this should make further syncs easier
2010-09-19 12:47:41 +00:00
*/
static int
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
iwn5000_query_calibration(struct iwn_softc *sc)
Rewrite parts of the calibration code which is run while bringing up the device: - unobscure some of the code by moving it into its own functions - get rid of some magic numbers - create similar structure as the reference driver has, this should make further syncs easier
2010-09-19 12:47:41 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn5000_calib_config cmd;
int error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
memset(&cmd, 0, sizeof cmd);
Begin fleshing out some code to handle tracking PLCP error rates in preparation for the scan based retune logic. The linux iwlwifi driver does a rescan (onto a non-active channel) to force an RF retune when the PLCP error rates exceed a certain threshold. * Add code to track HT PLCP rate errors; * Separate out the PLCP error count fetch and update so the delta can be used when checking for PLCP error rates; * Implement the PLCP error logic from iwlwifi; * For now, just print out whenever the error rate exceeds the threshold. The actual scan based retune will take a bit more effort; the scan command code right now assumes that a scan state is passed in. This does need to change to be more flexible (both for this and in preparation for scanning multiple channels at once.) Tested: * 5100 (STA mode) * 2200 (STA mode) * 2230 (STA mode)
2013-12-07 08:03:10 +00:00
cmd.ucode.once.enable = htole32(0xffffffff);
cmd.ucode.once.start = htole32(0xffffffff);
cmd.ucode.once.send = htole32(0xffffffff);
cmd.ucode.flags = htole32(0xffffffff);
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE, "%s: sending calibration query\n",
__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
error = iwn_cmd(sc, IWN5000_CMD_CALIB_CONFIG, &cmd, sizeof cmd, 0);
if (error != 0)
return error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Wait at most two seconds for calibration to complete. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (!(sc->sc_flags & IWN_FLAG_CALIB_DONE))
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
error = msleep(sc, &sc->sc_mtx, PCATCH, "iwncal", 2 * hz);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
return error;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
/*
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
* Send calibration results to the runtime firmware. These results were
* obtained on first boot from the initialization firmware.
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*/
Rewrite parts of the calibration code which is run while bringing up the device: - unobscure some of the code by moving it into its own functions - get rid of some magic numbers - create similar structure as the reference driver has, this should make further syncs easier
2010-09-19 12:47:41 +00:00
static int
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
iwn5000_send_calibration(struct iwn_softc *sc)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
{
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
int idx, error;
Rewrite parts of the calibration code which is run while bringing up the device: - unobscure some of the code by moving it into its own functions - get rid of some magic numbers - create similar structure as the reference driver has, this should make further syncs easier
2010-09-19 12:47:41 +00:00
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
for (idx = 0; idx < IWN5000_PHY_CALIB_MAX_RESULT; idx++) {
if (!(sc->base_params->calib_need & (1<<idx))) {
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"No need of calib %d\n",
idx);
continue; /* no need for this calib */
}
if (sc->calibcmd[idx].buf == NULL) {
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"Need calib idx : %d but no available data\n",
idx);
continue;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
"send calibration result idx=%d len=%d\n", idx,
sc->calibcmd[idx].len);
error = iwn_cmd(sc, IWN_CMD_PHY_CALIB, sc->calibcmd[idx].buf,
sc->calibcmd[idx].len, 0);
if (error != 0) {
device_printf(sc->sc_dev,
"%s: could not send calibration result, error %d\n",
__func__, error);
return error;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
return 0;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn5000_send_wimax_coex(struct iwn_softc *sc)
{
struct iwn5000_wimax_coex wimax;
Begin fleshing out some code to handle tracking PLCP error rates in preparation for the scan based retune logic. The linux iwlwifi driver does a rescan (onto a non-active channel) to force an RF retune when the PLCP error rates exceed a certain threshold. * Add code to track HT PLCP rate errors; * Separate out the PLCP error count fetch and update so the delta can be used when checking for PLCP error rates; * Implement the PLCP error logic from iwlwifi; * For now, just print out whenever the error rate exceeds the threshold. The actual scan based retune will take a bit more effort; the scan command code right now assumes that a scan state is passed in. This does need to change to be more flexible (both for this and in preparation for scanning multiple channels at once.) Tested: * 5100 (STA mode) * 2200 (STA mode) * 2230 (STA mode)
2013-12-07 08:03:10 +00:00
#if 0
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (sc->hw_type == IWN_HW_REV_TYPE_6050) {
/* Enable WiMAX coexistence for combo adapters. */
wimax.flags =
IWN_WIMAX_COEX_ASSOC_WA_UNMASK |
IWN_WIMAX_COEX_UNASSOC_WA_UNMASK |
IWN_WIMAX_COEX_STA_TABLE_VALID |
IWN_WIMAX_COEX_ENABLE;
memcpy(wimax.events, iwn6050_wimax_events,
sizeof iwn6050_wimax_events);
} else
#endif
{
/* Disable WiMAX coexistence. */
wimax.flags = 0;
memset(wimax.events, 0, sizeof wimax.events);
}
DPRINTF(sc, IWN_DEBUG_RESET, "%s: Configuring WiMAX coexistence\n",
__func__);
return iwn_cmd(sc, IWN5000_CMD_WIMAX_COEX, &wimax, sizeof wimax, 0);
}
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
static int
iwn5000_crystal_calib(struct iwn_softc *sc)
{
struct iwn5000_phy_calib_crystal cmd;
memset(&cmd, 0, sizeof cmd);
cmd.code = IWN5000_PHY_CALIB_CRYSTAL;
cmd.ngroups = 1;
cmd.isvalid = 1;
cmd.cap_pin[0] = le32toh(sc->eeprom_crystal) & 0xff;
cmd.cap_pin[1] = (le32toh(sc->eeprom_crystal) >> 16) & 0xff;
DPRINTF(sc, IWN_DEBUG_CALIBRATE, "sending crystal calibration %d, %d\n",
cmd.cap_pin[0], cmd.cap_pin[1]);
return iwn_cmd(sc, IWN_CMD_PHY_CALIB, &cmd, sizeof cmd, 0);
}
The 6005 series devices need additional temperature offset calibration as well as the IWN_GP_DRIVER_CALIB_VER6 bit set. Obtained from: OpenBSD
2011-04-15 20:35:15 +00:00
static int
iwn5000_temp_offset_calib(struct iwn_softc *sc)
{
struct iwn5000_phy_calib_temp_offset cmd;
memset(&cmd, 0, sizeof cmd);
cmd.code = IWN5000_PHY_CALIB_TEMP_OFFSET;
cmd.ngroups = 1;
cmd.isvalid = 1;
if (sc->eeprom_temp != 0)
cmd.offset = htole16(sc->eeprom_temp);
else
cmd.offset = htole16(IWN_DEFAULT_TEMP_OFFSET);
DPRINTF(sc, IWN_DEBUG_CALIBRATE, "setting radio sensor offset to %d\n",
le16toh(cmd.offset));
return iwn_cmd(sc, IWN_CMD_PHY_CALIB, &cmd, sizeof cmd, 0);
}
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
static int
iwn5000_temp_offset_calibv2(struct iwn_softc *sc)
{
struct iwn5000_phy_calib_temp_offsetv2 cmd;
memset(&cmd, 0, sizeof cmd);
cmd.code = IWN5000_PHY_CALIB_TEMP_OFFSET;
cmd.ngroups = 1;
cmd.isvalid = 1;
if (sc->eeprom_temp != 0) {
cmd.offset_low = htole16(sc->eeprom_temp);
cmd.offset_high = htole16(sc->eeprom_temp_high);
} else {
cmd.offset_low = htole16(IWN_DEFAULT_TEMP_OFFSET);
cmd.offset_high = htole16(IWN_DEFAULT_TEMP_OFFSET);
}
cmd.burnt_voltage_ref = htole16(sc->eeprom_voltage);
DPRINTF(sc, IWN_DEBUG_CALIBRATE,
"setting radio sensor low offset to %d, high offset to %d, voltage to %d\n",
le16toh(cmd.offset_low),
le16toh(cmd.offset_high),
le16toh(cmd.burnt_voltage_ref));
return iwn_cmd(sc, IWN_CMD_PHY_CALIB, &cmd, sizeof cmd, 0);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* This function is called after the runtime firmware notifies us of its
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
* readiness (called in a process context).
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_post_alive(struct iwn_softc *sc)
{
int error, qid;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
return error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Clear TX scheduler state in SRAM. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
sc->sched_base = iwn_prph_read(sc, IWN_SCHED_SRAM_ADDR);
iwn_mem_set_region_4(sc, sc->sched_base + IWN4965_SCHED_CTX_OFF, 0,
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN4965_SCHED_CTX_LEN / sizeof (uint32_t));
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Set physical address of TX scheduler rings (1KB aligned). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_prph_write(sc, IWN4965_SCHED_DRAM_ADDR, sc->sched_dma.paddr >> 10);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_SETBITS(sc, IWN_FH_TX_CHICKEN, IWN_FH_TX_CHICKEN_SCHED_RETRY);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Disable chain mode for all our 16 queues. */
iwn_prph_write(sc, IWN4965_SCHED_QCHAIN_SEL, 0);
for (qid = 0; qid < IWN4965_NTXQUEUES; qid++) {
iwn_prph_write(sc, IWN4965_SCHED_QUEUE_RDPTR(qid), 0);
IWN_WRITE(sc, IWN_HBUS_TARG_WRPTR, qid << 8 | 0);
/* Set scheduler window size. */
iwn_mem_write(sc, sc->sched_base +
IWN4965_SCHED_QUEUE_OFFSET(qid), IWN_SCHED_WINSZ);
/* Set scheduler frame limit. */
iwn_mem_write(sc, sc->sched_base +
IWN4965_SCHED_QUEUE_OFFSET(qid) + 4,
IWN_SCHED_LIMIT << 16);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Enable interrupts for all our 16 queues. */
iwn_prph_write(sc, IWN4965_SCHED_INTR_MASK, 0xffff);
/* Identify TX FIFO rings (0-7). */
iwn_prph_write(sc, IWN4965_SCHED_TXFACT, 0xff);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Mark TX rings (4 EDCA + cmd + 2 HCCA) as active. */
for (qid = 0; qid < 7; qid++) {
static uint8_t qid2fifo[] = { 3, 2, 1, 0, 4, 5, 6 };
iwn_prph_write(sc, IWN4965_SCHED_QUEUE_STATUS(qid),
IWN4965_TXQ_STATUS_ACTIVE | qid2fifo[qid] << 1);
}
iwn_nic_unlock(sc);
return 0;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* This function is called after the initialization or runtime firmware
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
* notifies us of its readiness (called in a process context).
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_post_alive(struct iwn_softc *sc)
{
int error, qid;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Switch to using ICT interrupt mode. */
iwn5000_ict_reset(sc);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
if ((error = iwn_nic_lock(sc)) != 0){
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s end in error\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Clear TX scheduler state in SRAM. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
sc->sched_base = iwn_prph_read(sc, IWN_SCHED_SRAM_ADDR);
iwn_mem_set_region_4(sc, sc->sched_base + IWN5000_SCHED_CTX_OFF, 0,
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN5000_SCHED_CTX_LEN / sizeof (uint32_t));
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Set physical address of TX scheduler rings (1KB aligned). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_prph_write(sc, IWN5000_SCHED_DRAM_ADDR, sc->sched_dma.paddr >> 10);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_SETBITS(sc, IWN_FH_TX_CHICKEN, IWN_FH_TX_CHICKEN_SCHED_RETRY);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Enable chain mode for all queues, except command queue. */
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->sc_flags & IWN_FLAG_PAN_SUPPORT)
iwn_prph_write(sc, IWN5000_SCHED_QCHAIN_SEL, 0xfffdf);
else
iwn_prph_write(sc, IWN5000_SCHED_QCHAIN_SEL, 0xfffef);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_prph_write(sc, IWN5000_SCHED_AGGR_SEL, 0);
for (qid = 0; qid < IWN5000_NTXQUEUES; qid++) {
iwn_prph_write(sc, IWN5000_SCHED_QUEUE_RDPTR(qid), 0);
IWN_WRITE(sc, IWN_HBUS_TARG_WRPTR, qid << 8 | 0);
iwn_mem_write(sc, sc->sched_base +
IWN5000_SCHED_QUEUE_OFFSET(qid), 0);
/* Set scheduler window size and frame limit. */
iwn_mem_write(sc, sc->sched_base +
IWN5000_SCHED_QUEUE_OFFSET(qid) + 4,
IWN_SCHED_LIMIT << 16 | IWN_SCHED_WINSZ);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Enable interrupts for all our 20 queues. */
iwn_prph_write(sc, IWN5000_SCHED_INTR_MASK, 0xfffff);
/* Identify TX FIFO rings (0-7). */
iwn_prph_write(sc, IWN5000_SCHED_TXFACT, 0xff);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Mark TX rings (4 EDCA + cmd + 2 HCCA) as active. */
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->sc_flags & IWN_FLAG_PAN_SUPPORT) {
/* Mark TX rings as active. */
for (qid = 0; qid < 11; qid++) {
static uint8_t qid2fifo[] = { 3, 2, 1, 0, 0, 4, 2, 5, 4, 7, 5 };
iwn_prph_write(sc, IWN5000_SCHED_QUEUE_STATUS(qid),
IWN5000_TXQ_STATUS_ACTIVE | qid2fifo[qid]);
}
} else {
/* Mark TX rings (4 EDCA + cmd + 2 HCCA) as active. */
for (qid = 0; qid < 7; qid++) {
static uint8_t qid2fifo[] = { 3, 2, 1, 0, 7, 5, 6 };
iwn_prph_write(sc, IWN5000_SCHED_QUEUE_STATUS(qid),
IWN5000_TXQ_STATUS_ACTIVE | qid2fifo[qid]);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_nic_unlock(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Configure WiMAX coexistence for combo adapters. */
error = iwn5000_send_wimax_coex(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
"%s: could not configure WiMAX coexistence, error %d\n",
__func__, error);
return error;
}
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
if (sc->hw_type != IWN_HW_REV_TYPE_5150) {
/* Perform crystal calibration. */
error = iwn5000_crystal_calib(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (error != 0) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
device_printf(sc->sc_dev,
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
"%s: crystal calibration failed, error %d\n",
__func__, error);
Rewrite parts of the calibration code which is run while bringing up the device: - unobscure some of the code by moving it into its own functions - get rid of some magic numbers - create similar structure as the reference driver has, this should make further syncs easier
2010-09-19 12:47:41 +00:00
return error;
}
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
}
if (!(sc->sc_flags & IWN_FLAG_CALIB_DONE)) {
/* Query calibration from the initialization firmware. */
if ((error = iwn5000_query_calibration(sc)) != 0) {
Rewrite parts of the calibration code which is run while bringing up the device: - unobscure some of the code by moving it into its own functions - get rid of some magic numbers - create similar structure as the reference driver has, this should make further syncs easier
2010-09-19 12:47:41 +00:00
device_printf(sc->sc_dev,
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
"%s: could not query calibration, error %d\n",
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
__func__, error);
return error;
}
/*
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
* We have the calibration results now, reboot with the
* runtime firmware (call ourselves recursively!)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*/
iwn_hw_stop(sc);
error = iwn_hw_init(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
} else {
Revert some of local calibration changes in favour of the OpenBSD implementation. This includes the fix required for the 6050 series devices.
2011-04-15 20:31:02 +00:00
/* Send calibration results to runtime firmware. */
error = iwn5000_send_calibration(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* The firmware boot code is small and is intended to be copied directly into
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
* the NIC internal memory (no DMA transfer).
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_load_bootcode(struct iwn_softc *sc, const uint8_t *ucode, int size)
{
int error, ntries;
size /= sizeof (uint32_t);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
/* Copy microcode image into NIC memory. */
iwn_prph_write_region_4(sc, IWN_BSM_SRAM_BASE,
(const uint32_t *)ucode, size);
iwn_prph_write(sc, IWN_BSM_WR_MEM_SRC, 0);
iwn_prph_write(sc, IWN_BSM_WR_MEM_DST, IWN_FW_TEXT_BASE);
iwn_prph_write(sc, IWN_BSM_WR_DWCOUNT, size);
/* Start boot load now. */
iwn_prph_write(sc, IWN_BSM_WR_CTRL, IWN_BSM_WR_CTRL_START);
/* Wait for transfer to complete. */
for (ntries = 0; ntries < 1000; ntries++) {
if (!(iwn_prph_read(sc, IWN_BSM_WR_CTRL) &
IWN_BSM_WR_CTRL_START))
break;
DELAY(10);
}
if (ntries == 1000) {
device_printf(sc->sc_dev, "%s: could not load boot firmware\n",
__func__);
iwn_nic_unlock(sc);
return ETIMEDOUT;
}
/* Enable boot after power up. */
iwn_prph_write(sc, IWN_BSM_WR_CTRL, IWN_BSM_WR_CTRL_START_EN);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_nic_unlock(sc);
return 0;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_load_firmware(struct iwn_softc *sc)
{
struct iwn_fw_info *fw = &sc->fw;
struct iwn_dma_info *dma = &sc->fw_dma;
int error;
/* Copy initialization sections into pre-allocated DMA-safe memory. */
memcpy(dma->vaddr, fw->init.data, fw->init.datasz);
Fixes for firmware handling: - there is a local variable for sc->fw_dma, use that instead - OpenBSD uses 5*hz to wait for firmware to be loaded - in case the firmware module contains invalid data, actually release it
2011-04-15 16:59:56 +00:00
bus_dmamap_sync(dma->tag, dma->map, BUS_DMASYNC_PREWRITE);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
memcpy(dma->vaddr + IWN4965_FW_DATA_MAXSZ,
fw->init.text, fw->init.textsz);
Fixes for firmware handling: - there is a local variable for sc->fw_dma, use that instead - OpenBSD uses 5*hz to wait for firmware to be loaded - in case the firmware module contains invalid data, actually release it
2011-04-15 16:59:56 +00:00
bus_dmamap_sync(dma->tag, dma->map, BUS_DMASYNC_PREWRITE);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Tell adapter where to find initialization sections. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
iwn_prph_write(sc, IWN_BSM_DRAM_DATA_ADDR, dma->paddr >> 4);
iwn_prph_write(sc, IWN_BSM_DRAM_DATA_SIZE, fw->init.datasz);
iwn_prph_write(sc, IWN_BSM_DRAM_TEXT_ADDR,
(dma->paddr + IWN4965_FW_DATA_MAXSZ) >> 4);
iwn_prph_write(sc, IWN_BSM_DRAM_TEXT_SIZE, fw->init.textsz);
iwn_nic_unlock(sc);
/* Load firmware boot code. */
error = iwn4965_load_bootcode(sc, fw->boot.text, fw->boot.textsz);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
device_printf(sc->sc_dev, "%s: could not load boot firmware\n",
__func__);
return error;
}
/* Now press "execute". */
IWN_WRITE(sc, IWN_RESET, 0);
/* Wait at most one second for first alive notification. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = msleep(sc, &sc->sc_mtx, PCATCH, "iwninit", hz)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(sc->sc_dev,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
"%s: timeout waiting for adapter to initialize, error %d\n",
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
__func__, error);
return error;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Retrieve current temperature for initial TX power calibration. */
sc->rawtemp = sc->ucode_info.temp[3].chan20MHz;
sc->temp = iwn4965_get_temperature(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Copy runtime sections into pre-allocated DMA-safe memory. */
memcpy(dma->vaddr, fw->main.data, fw->main.datasz);
Fixes for firmware handling: - there is a local variable for sc->fw_dma, use that instead - OpenBSD uses 5*hz to wait for firmware to be loaded - in case the firmware module contains invalid data, actually release it
2011-04-15 16:59:56 +00:00
bus_dmamap_sync(dma->tag, dma->map, BUS_DMASYNC_PREWRITE);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
memcpy(dma->vaddr + IWN4965_FW_DATA_MAXSZ,
fw->main.text, fw->main.textsz);
Fixes for firmware handling: - there is a local variable for sc->fw_dma, use that instead - OpenBSD uses 5*hz to wait for firmware to be loaded - in case the firmware module contains invalid data, actually release it
2011-04-15 16:59:56 +00:00
bus_dmamap_sync(dma->tag, dma->map, BUS_DMASYNC_PREWRITE);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Tell adapter where to find runtime sections. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
iwn_prph_write(sc, IWN_BSM_DRAM_DATA_ADDR, dma->paddr >> 4);
iwn_prph_write(sc, IWN_BSM_DRAM_DATA_SIZE, fw->main.datasz);
iwn_prph_write(sc, IWN_BSM_DRAM_TEXT_ADDR,
(dma->paddr + IWN4965_FW_DATA_MAXSZ) >> 4);
iwn_prph_write(sc, IWN_BSM_DRAM_TEXT_SIZE,
IWN_FW_UPDATED | fw->main.textsz);
iwn_nic_unlock(sc);
return 0;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_load_firmware_section(struct iwn_softc *sc, uint32_t dst,
const uint8_t *section, int size)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_dma_info *dma = &sc->fw_dma;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
int error;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Copy firmware section into pre-allocated DMA-safe memory. */
memcpy(dma->vaddr, section, size);
Fixes for firmware handling: - there is a local variable for sc->fw_dma, use that instead - OpenBSD uses 5*hz to wait for firmware to be loaded - in case the firmware module contains invalid data, actually release it
2011-04-15 16:59:56 +00:00
bus_dmamap_sync(dma->tag, dma->map, BUS_DMASYNC_PREWRITE);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
IWN_WRITE(sc, IWN_FH_TX_CONFIG(IWN_SRVC_DMACHNL),
IWN_FH_TX_CONFIG_DMA_PAUSE);
IWN_WRITE(sc, IWN_FH_SRAM_ADDR(IWN_SRVC_DMACHNL), dst);
IWN_WRITE(sc, IWN_FH_TFBD_CTRL0(IWN_SRVC_DMACHNL),
IWN_LOADDR(dma->paddr));
IWN_WRITE(sc, IWN_FH_TFBD_CTRL1(IWN_SRVC_DMACHNL),
IWN_HIADDR(dma->paddr) << 28 | size);
IWN_WRITE(sc, IWN_FH_TXBUF_STATUS(IWN_SRVC_DMACHNL),
IWN_FH_TXBUF_STATUS_TBNUM(1) |
IWN_FH_TXBUF_STATUS_TBIDX(1) |
IWN_FH_TXBUF_STATUS_TFBD_VALID);
/* Kick Flow Handler to start DMA transfer. */
IWN_WRITE(sc, IWN_FH_TX_CONFIG(IWN_SRVC_DMACHNL),
IWN_FH_TX_CONFIG_DMA_ENA | IWN_FH_TX_CONFIG_CIRQ_HOST_ENDTFD);
iwn_nic_unlock(sc);
/* Wait at most five seconds for FH DMA transfer to complete. */
Fixes for firmware handling: - there is a local variable for sc->fw_dma, use that instead - OpenBSD uses 5*hz to wait for firmware to be loaded - in case the firmware module contains invalid data, actually release it
2011-04-15 16:59:56 +00:00
return msleep(sc, &sc->sc_mtx, PCATCH, "iwninit", 5 * hz);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_load_firmware(struct iwn_softc *sc)
{
struct iwn_fw_part *fw;
int error;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Load the initialization firmware on first boot only. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
fw = (sc->sc_flags & IWN_FLAG_CALIB_DONE) ?
&sc->fw.main : &sc->fw.init;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
error = iwn5000_load_firmware_section(sc, IWN_FW_TEXT_BASE,
fw->text, fw->textsz);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
"%s: could not load firmware %s section, error %d\n",
__func__, ".text", error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
error = iwn5000_load_firmware_section(sc, IWN_FW_DATA_BASE,
fw->data, fw->datasz);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
"%s: could not load firmware %s section, error %d\n",
__func__, ".data", error);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return error;
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Now press "execute". */
IWN_WRITE(sc, IWN_RESET, 0);
return 0;
}
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
/*
* Extract text and data sections from a legacy firmware image.
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
iwn_read_firmware_leg(struct iwn_softc *sc, struct iwn_fw_info *fw)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
{
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
const uint32_t *ptr;
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
size_t hdrlen = 24;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
uint32_t rev;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Fixes for firmware handling: - there is a local variable for sc->fw_dma, use that instead - OpenBSD uses 5*hz to wait for firmware to be loaded - in case the firmware module contains invalid data, actually release it
2011-04-15 16:59:56 +00:00
ptr = (const uint32_t *)fw->data;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
rev = le32toh(*ptr++);
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
sc->ucode_rev = rev;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Check firmware API version. */
if (IWN_FW_API(rev) <= 1) {
device_printf(sc->sc_dev,
"%s: bad firmware, need API version >=2\n", __func__);
return EINVAL;
}
if (IWN_FW_API(rev) >= 3) {
/* Skip build number (version 2 header). */
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
hdrlen += 4;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
ptr++;
}
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
if (fw->size < hdrlen) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(sc->sc_dev, "%s: firmware too short: %zu bytes\n",
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
__func__, fw->size);
return EINVAL;
}
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
fw->main.textsz = le32toh(*ptr++);
fw->main.datasz = le32toh(*ptr++);
fw->init.textsz = le32toh(*ptr++);
fw->init.datasz = le32toh(*ptr++);
fw->boot.textsz = le32toh(*ptr++);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Check that all firmware sections fit. */
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
if (fw->size < hdrlen + fw->main.textsz + fw->main.datasz +
fw->init.textsz + fw->init.datasz + fw->boot.textsz) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(sc->sc_dev, "%s: firmware too short: %zu bytes\n",
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
__func__, fw->size);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return EINVAL;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Get pointers to firmware sections. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
fw->main.text = (const uint8_t *)ptr;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
fw->main.data = fw->main.text + fw->main.textsz;
fw->init.text = fw->main.data + fw->main.datasz;
fw->init.data = fw->init.text + fw->init.textsz;
fw->boot.text = fw->init.data + fw->init.datasz;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return 0;
}
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
/*
* Extract text and data sections from a TLV firmware image.
*/
Fixes for firmware handling: - there is a local variable for sc->fw_dma, use that instead - OpenBSD uses 5*hz to wait for firmware to be loaded - in case the firmware module contains invalid data, actually release it
2011-04-15 16:59:56 +00:00
static int
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
iwn_read_firmware_tlv(struct iwn_softc *sc, struct iwn_fw_info *fw,
uint16_t alt)
{
const struct iwn_fw_tlv_hdr *hdr;
const struct iwn_fw_tlv *tlv;
const uint8_t *ptr, *end;
uint64_t altmask;
Pull some features out of the firmware: - If a ENH_SENS TLV section exit the firmware is capable of doing enhanced sensitivity calibration. - Newer devices/firmwares have more calibration commands therefore hardcoding the noise gain/reset commands no longer works. It is supposed to use the next index after the newest calibration type support. Read the command index of the TLV section if available.
2011-04-19 19:47:41 +00:00
uint32_t len, tmp;
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
if (fw->size < sizeof (*hdr)) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(sc->sc_dev, "%s: firmware too short: %zu bytes\n",
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
__func__, fw->size);
return EINVAL;
}
hdr = (const struct iwn_fw_tlv_hdr *)fw->data;
if (hdr->signature != htole32(IWN_FW_SIGNATURE)) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(sc->sc_dev, "%s: bad firmware signature 0x%08x\n",
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
__func__, le32toh(hdr->signature));
return EINVAL;
}
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
DPRINTF(sc, IWN_DEBUG_RESET, "FW: \"%.64s\", build 0x%x\n", hdr->descr,
le32toh(hdr->build));
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
sc->ucode_rev = le32toh(hdr->rev);
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
/*
* Select the closest supported alternative that is less than
* or equal to the specified one.
*/
altmask = le64toh(hdr->altmask);
while (alt > 0 && !(altmask & (1ULL << alt)))
alt--; /* Downgrade. */
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
DPRINTF(sc, IWN_DEBUG_RESET, "using alternative %d\n", alt);
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
ptr = (const uint8_t *)(hdr + 1);
end = (const uint8_t *)(fw->data + fw->size);
/* Parse type-length-value fields. */
while (ptr + sizeof (*tlv) <= end) {
tlv = (const struct iwn_fw_tlv *)ptr;
len = le32toh(tlv->len);
ptr += sizeof (*tlv);
if (ptr + len > end) {
device_printf(sc->sc_dev,
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
"%s: firmware too short: %zu bytes\n", __func__,
fw->size);
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
return EINVAL;
}
/* Skip other alternatives. */
if (tlv->alt != 0 && tlv->alt != htole16(alt))
goto next;
switch (le16toh(tlv->type)) {
case IWN_FW_TLV_MAIN_TEXT:
fw->main.text = ptr;
fw->main.textsz = len;
break;
case IWN_FW_TLV_MAIN_DATA:
fw->main.data = ptr;
fw->main.datasz = len;
break;
case IWN_FW_TLV_INIT_TEXT:
fw->init.text = ptr;
fw->init.textsz = len;
break;
case IWN_FW_TLV_INIT_DATA:
fw->init.data = ptr;
fw->init.datasz = len;
break;
case IWN_FW_TLV_BOOT_TEXT:
fw->boot.text = ptr;
fw->boot.textsz = len;
break;
Pull some features out of the firmware: - If a ENH_SENS TLV section exit the firmware is capable of doing enhanced sensitivity calibration. - Newer devices/firmwares have more calibration commands therefore hardcoding the noise gain/reset commands no longer works. It is supposed to use the next index after the newest calibration type support. Read the command index of the TLV section if available.
2011-04-19 19:47:41 +00:00
case IWN_FW_TLV_ENH_SENS:
if (!len)
sc->sc_flags |= IWN_FLAG_ENH_SENS;
break;
case IWN_FW_TLV_PHY_CALIB:
Use the correct endian-ness accessor for this TLV field. (It's coming from firmware and thus it's defined as little-endian.)
2013-11-29 22:35:24 +00:00
tmp = le32toh(*ptr);
Pull some features out of the firmware: - If a ENH_SENS TLV section exit the firmware is capable of doing enhanced sensitivity calibration. - Newer devices/firmwares have more calibration commands therefore hardcoding the noise gain/reset commands no longer works. It is supposed to use the next index after the newest calibration type support. Read the command index of the TLV section if available.
2011-04-19 19:47:41 +00:00
if (tmp < 253) {
sc->reset_noise_gain = tmp;
sc->noise_gain = tmp + 1;
}
break;
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
case IWN_FW_TLV_PAN:
sc->sc_flags |= IWN_FLAG_PAN_SUPPORT;
DPRINTF(sc, IWN_DEBUG_RESET,
"PAN Support found: %d\n", 1);
break;
Oops - fix bad indent. Sorry!
2013-12-02 03:43:37 +00:00
case IWN_FW_TLV_FLAGS:
Add some sanity checks to the TLV fetch. Obtained from: Linux iwlwifi
2013-12-02 03:42:39 +00:00
if (len < sizeof(uint32_t))
break;
if (len % sizeof(uint32_t))
break;
sc->tlv_feature_flags = le32toh(*ptr);
DPRINTF(sc, IWN_DEBUG_RESET,
"%s: feature: 0x%08x\n",
__func__,
sc->tlv_feature_flags);
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
break;
case IWN_FW_TLV_PBREQ_MAXLEN:
case IWN_FW_TLV_RUNT_EVTLOG_PTR:
case IWN_FW_TLV_RUNT_EVTLOG_SIZE:
case IWN_FW_TLV_RUNT_ERRLOG_PTR:
case IWN_FW_TLV_INIT_EVTLOG_PTR:
case IWN_FW_TLV_INIT_EVTLOG_SIZE:
case IWN_FW_TLV_INIT_ERRLOG_PTR:
case IWN_FW_TLV_WOWLAN_INST:
case IWN_FW_TLV_WOWLAN_DATA:
DPRINTF(sc, IWN_DEBUG_RESET,
iwn, wlan: fix typos Fxi tow typos Obtained from: DragonFlyBSD MFC after: 3 days
2015-04-10 20:55:17 +00:00
"TLV type %d recognized but not handled\n",
Prepare for the PAN (personal area network) support for iwn(4). * Break out the single, static RX context into a pointer, and .. * .. extend it to two RX contexts - a default and a PAN context. Whilst here, add a few extra fields in preparation for further iwn(4) work. Tested: * Intel 4965, STA mode - same level of stability * Intel 5100, STA mode - no change Submitted by: Cedric Gross <cg@gross.info>
2013-08-11 01:57:54 +00:00
le16toh(tlv->type));
break;
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
default:
DPRINTF(sc, IWN_DEBUG_RESET,
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
"TLV type %d not handled\n", le16toh(tlv->type));
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
break;
}
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
next: /* TLV fields are 32-bit aligned. */
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
ptr += (len + 3) & ~3;
}
return 0;
}
static int
iwn_read_firmware(struct iwn_softc *sc)
{
struct iwn_fw_info *fw = &sc->fw;
int error;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
IWN_UNLOCK(sc);
memset(fw, 0, sizeof (*fw));
/* Read firmware image from filesystem. */
sc->fw_fp = firmware_get(sc->fwname);
if (sc->fw_fp == NULL) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(sc->sc_dev, "%s: could not read firmware %s\n",
__func__, sc->fwname);
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
IWN_LOCK(sc);
return EINVAL;
}
IWN_LOCK(sc);
fw->size = sc->fw_fp->datasize;
fw->data = (const uint8_t *)sc->fw_fp->data;
if (fw->size < sizeof (uint32_t)) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(sc->sc_dev, "%s: firmware too short: %zu bytes\n",
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
__func__, fw->size);
Fixes for firmware handling: - there is a local variable for sc->fw_dma, use that instead - OpenBSD uses 5*hz to wait for firmware to be loaded - in case the firmware module contains invalid data, actually release it
2011-04-15 16:59:56 +00:00
firmware_put(sc->fw_fp, FIRMWARE_UNLOAD);
sc->fw_fp = NULL;
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
return EINVAL;
}
/* Retrieve text and data sections. */
if (*(const uint32_t *)fw->data != 0) /* Legacy image. */
error = iwn_read_firmware_leg(sc, fw);
else
error = iwn_read_firmware_tlv(sc, fw, 1);
if (error != 0) {
device_printf(sc->sc_dev,
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
"%s: could not read firmware sections, error %d\n",
__func__, error);
Fixes for firmware handling: - there is a local variable for sc->fw_dma, use that instead - OpenBSD uses 5*hz to wait for firmware to be loaded - in case the firmware module contains invalid data, actually release it
2011-04-15 16:59:56 +00:00
firmware_put(sc->fw_fp, FIRMWARE_UNLOAD);
sc->fw_fp = NULL;
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
return error;
}
Fix antenna configuration, microcode version checks and rate selection in preparation for the 5300 3x3 NIC. During this particular adventure, I did indeed discover that a whole swath of things made little to no sense. Those included, and are fixed here: * A lot of the antenna configuration bits assume the NIC has two receive chains. That's blatantly untrue for NICs that don't. * There was some disconnect between the antenna configuration when forming a PLCP rate DWORD (which includes the transmit antenna configuration), separate to the link quality antenna configuration. So now there's helper functions to return which antenna configurations to use and those are used wherever an antenna config is required. * The 5300 does up to three stream TX/RX (so MCS0->23), however the link quality table has only 16 slots. This means all of the rate entries are .. well, dual-stream rates. If this is the case, the "last MIMO" parameter can't be 16 or it panics the firmware. Set it to 15. * .. and since yes it has 16 slots, it only would try retransmitting from MCS8->MCS23, which can be quite .. terrible. Hard-code the last two retry slots to be the lowest configured rate. * I noticed some transmit configuration command stuff is different based on firmware API version, so I lifted that code from Linux. * Add / augment some more logging to make it easier to capture this stuff. Now, 3x3 is still terrible because the link quality configuration is plainly not good enough. I'll have to think about that. However, the original goal of this - 3x3 operation on the Intel 5300 NIC - actually worked. There are also rate control bugs in the way this driver handles notifying the net80211 rate control code when AMPDU is enabled. It always steps the rate up to the maximum rate possible - and this eventually ends in much sadness. I'll fix that later. As a side note - 2GHz HT40 now works on all the NICs I have tested. As a second side note - this exposed some bad 3x3 behaviour in the ath(4) rate control code where it starts off at a 3-stream rate and doesn't downgrade quickly enough. This makes the initial dhcp exchange take a long time. I'll fix the ath(4) rate code to start at a low fixed 1x1 MCS rate and step up if everything works out. Tested: * Intel 2200 * Intel 2230 * Intel 5300 * Intel 5100 * Intel 6205 * Intel 100 TODO: * Test the other NICs more thoroughly! Thank you to Michael Kosarev <russiane39@gmail.com> for donating the Intel 5300 NIC and pestering me about it since last year to try and make it all work.
2014-08-28 03:18:27 +00:00
device_printf(sc->sc_dev, "%s: ucode rev=0x%08x\n", __func__, sc->ucode_rev);
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
/* Make sure text and data sections fit in hardware memory. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
if (fw->main.textsz > sc->fw_text_maxsz ||
fw->main.datasz > sc->fw_data_maxsz ||
fw->init.textsz > sc->fw_text_maxsz ||
fw->init.datasz > sc->fw_data_maxsz ||
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
fw->boot.textsz > IWN_FW_BOOT_TEXT_MAXSZ ||
(fw->boot.textsz & 3) != 0) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(sc->sc_dev, "%s: firmware sections too large\n",
__func__);
Fixes for firmware handling: - there is a local variable for sc->fw_dma, use that instead - OpenBSD uses 5*hz to wait for firmware to be loaded - in case the firmware module contains invalid data, actually release it
2011-04-15 16:59:56 +00:00
firmware_put(sc->fw_fp, FIRMWARE_UNLOAD);
sc->fw_fp = NULL;
Add support for firmware images in "type-length-value" format. Obtained from: OpenBSD MFC after: 2 weeks
2010-07-15 10:37:49 +00:00
return EINVAL;
}
/* We can proceed with loading the firmware. */
return 0;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_clock_wait(struct iwn_softc *sc)
{
int ntries;
/* Set "initialization complete" bit. */
IWN_SETBITS(sc, IWN_GP_CNTRL, IWN_GP_CNTRL_INIT_DONE);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Wait for clock stabilization. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
for (ntries = 0; ntries < 2500; ntries++) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (IWN_READ(sc, IWN_GP_CNTRL) & IWN_GP_CNTRL_MAC_CLOCK_READY)
return 0;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
DELAY(10);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
device_printf(sc->sc_dev,
"%s: timeout waiting for clock stabilization\n", __func__);
return ETIMEDOUT;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_apm_init(struct iwn_softc *sc)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Rename some stuff in favour of the OpenBSD names: - prefer EDCA over WME - qid for a TXQ ID - reg for register values
2011-04-16 13:40:32 +00:00
uint32_t reg;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
int error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Disable L0s exit timer (NMI bug workaround). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_SETBITS(sc, IWN_GIO_CHICKEN, IWN_GIO_CHICKEN_DIS_L0S_TIMER);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Don't wait for ICH L0s (ICH bug workaround). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_SETBITS(sc, IWN_GIO_CHICKEN, IWN_GIO_CHICKEN_L1A_NO_L0S_RX);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Set FH wait threshold to max (HW bug under stress workaround). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_SETBITS(sc, IWN_DBG_HPET_MEM, 0xffff0000);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Enable HAP INTA to move adapter from L1a to L0s. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_SETBITS(sc, IWN_HW_IF_CONFIG, IWN_HW_IF_CONFIG_HAP_WAKE_L1A);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Retrieve PCIe Active State Power Management (ASPM). */
Rename some stuff in favour of the OpenBSD names: - prefer EDCA over WME - qid for a TXQ ID - reg for register values
2011-04-16 13:40:32 +00:00
reg = pci_read_config(sc->sc_dev, sc->sc_cap_off + 0x10, 1);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Workaround for HW instability in PCIe L0->L0s->L1 transition. */
Rename some stuff in favour of the OpenBSD names: - prefer EDCA over WME - qid for a TXQ ID - reg for register values
2011-04-16 13:40:32 +00:00
if (reg & 0x02) /* L1 Entry enabled. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_SETBITS(sc, IWN_GIO, IWN_GIO_L0S_ENA);
else
IWN_CLRBITS(sc, IWN_GIO, IWN_GIO_L0S_ENA);
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->pll_cfg_val)
IWN_SETBITS(sc, IWN_ANA_PLL, sc->base_params->pll_cfg_val);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Wait for clock stabilization before accessing prph. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_clock_wait(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (sc->hw_type == IWN_HW_REV_TYPE_4965) {
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Enable DMA and BSM (Bootstrap State Machine). */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_prph_write(sc, IWN_APMG_CLK_EN,
IWN_APMG_CLK_CTRL_DMA_CLK_RQT |
IWN_APMG_CLK_CTRL_BSM_CLK_RQT);
} else {
/* Enable DMA. */
iwn_prph_write(sc, IWN_APMG_CLK_EN,
IWN_APMG_CLK_CTRL_DMA_CLK_RQT);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DELAY(20);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Disable L1-Active. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_prph_setbits(sc, IWN_APMG_PCI_STT, IWN_APMG_PCI_STT_L1A_DIS);
iwn_nic_unlock(sc);
return 0;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_apm_stop_master(struct iwn_softc *sc)
{
int ntries;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Stop busmaster DMA activity. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_SETBITS(sc, IWN_RESET, IWN_RESET_STOP_MASTER);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
for (ntries = 0; ntries < 100; ntries++) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (IWN_READ(sc, IWN_RESET) & IWN_RESET_MASTER_DISABLED)
return;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
DELAY(10);
}
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
device_printf(sc->sc_dev, "%s: timeout waiting for master\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_apm_stop(struct iwn_softc *sc)
{
iwn_apm_stop_master(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Reset the entire device. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_SETBITS(sc, IWN_RESET, IWN_RESET_SW);
DELAY(10);
/* Clear "initialization complete" bit. */
IWN_CLRBITS(sc, IWN_GP_CNTRL, IWN_GP_CNTRL_INIT_DONE);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn4965_nic_config(struct iwn_softc *sc)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (IWN_RFCFG_TYPE(sc->rfcfg) == 1) {
/*
* I don't believe this to be correct but this is what the
* vendor driver is doing. Probably the bits should not be
* shifted in IWN_RFCFG_*.
*/
IWN_SETBITS(sc, IWN_HW_IF_CONFIG,
IWN_RFCFG_TYPE(sc->rfcfg) |
IWN_RFCFG_STEP(sc->rfcfg) |
IWN_RFCFG_DASH(sc->rfcfg));
}
IWN_SETBITS(sc, IWN_HW_IF_CONFIG,
IWN_HW_IF_CONFIG_RADIO_SI | IWN_HW_IF_CONFIG_MAC_SI);
return 0;
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn5000_nic_config(struct iwn_softc *sc)
{
uint32_t tmp;
int error;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (IWN_RFCFG_TYPE(sc->rfcfg) < 3) {
IWN_SETBITS(sc, IWN_HW_IF_CONFIG,
IWN_RFCFG_TYPE(sc->rfcfg) |
IWN_RFCFG_STEP(sc->rfcfg) |
IWN_RFCFG_DASH(sc->rfcfg));
}
IWN_SETBITS(sc, IWN_HW_IF_CONFIG,
IWN_HW_IF_CONFIG_RADIO_SI | IWN_HW_IF_CONFIG_MAC_SI);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
iwn_prph_setbits(sc, IWN_APMG_PS, IWN_APMG_PS_EARLY_PWROFF_DIS);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (sc->hw_type == IWN_HW_REV_TYPE_1000) {
/*
* Select first Switching Voltage Regulator (1.32V) to
* solve a stability issue related to noisy DC2DC line
* in the silicon of 1000 Series.
*/
tmp = iwn_prph_read(sc, IWN_APMG_DIGITAL_SVR);
tmp &= ~IWN_APMG_DIGITAL_SVR_VOLTAGE_MASK;
tmp |= IWN_APMG_DIGITAL_SVR_VOLTAGE_1_32;
iwn_prph_write(sc, IWN_APMG_DIGITAL_SVR, tmp);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_nic_unlock(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
if (sc->sc_flags & IWN_FLAG_INTERNAL_PA) {
/* Use internal power amplifier only. */
IWN_WRITE(sc, IWN_GP_DRIVER, IWN_GP_DRIVER_RADIO_2X2_IPA);
* Rename bluetooth coexistence flags, no binary change. * Enable DC calibration and crystal calibration on Centrino Advanced-N 6250 parts. * Workaround for a HW bug (does not affect 4965AGN) that may sporadically affect latency under some rare circumstances. From a similar commit to iwlwifi. * Update sensitivity settings for 5000 series to workaround a performance bug in the DSP (1000 is not affected so we keep the old values for 5000). * Update sensitivity settings for 6000 series. * Set differential gains on 6250 too (but use a 1.0 factor, not 1.5). * Init OFDM sensitivity with min value (which depends on the chip) instead of hardcoding it to 90. * Read calibration version from ROM and set IWN_GP_DRIVER_CALIB_VER6 bit on 6x50 if version >= 6. Approved by: rpaulo (mentor) Obtained from: OpenBSD MFC after: 2 weeks
2010-04-10 06:58:24 +00:00
}
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->additional_nic_config && sc->calib_ver >= 6) {
Fix some small whitespace nits. MFC after: 3 days
2010-07-15 08:05:20 +00:00
/* Indicate that ROM calibration version is >=6. */
IWN_SETBITS(sc, IWN_GP_DRIVER, IWN_GP_DRIVER_CALIB_VER6);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
}
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->additional_gp_drv_bit)
IWN_SETBITS(sc, IWN_GP_DRIVER,
sc->base_params->additional_gp_drv_bit);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return 0;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/*
* Take NIC ownership over Intel Active Management Technology (AMT).
*/
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_hw_prepare(struct iwn_softc *sc)
{
int ntries;
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Check if hardware is ready. */
IWN_SETBITS(sc, IWN_HW_IF_CONFIG, IWN_HW_IF_CONFIG_NIC_READY);
for (ntries = 0; ntries < 5; ntries++) {
if (IWN_READ(sc, IWN_HW_IF_CONFIG) &
IWN_HW_IF_CONFIG_NIC_READY)
return 0;
DELAY(10);
}
/* Hardware not ready, force into ready state. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_SETBITS(sc, IWN_HW_IF_CONFIG, IWN_HW_IF_CONFIG_PREPARE);
for (ntries = 0; ntries < 15000; ntries++) {
if (!(IWN_READ(sc, IWN_HW_IF_CONFIG) &
IWN_HW_IF_CONFIG_PREPARE_DONE))
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
break;
DELAY(10);
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (ntries == 15000)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
return ETIMEDOUT;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Hardware should be ready now. */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_SETBITS(sc, IWN_HW_IF_CONFIG, IWN_HW_IF_CONFIG_NIC_READY);
for (ntries = 0; ntries < 5; ntries++) {
if (IWN_READ(sc, IWN_HW_IF_CONFIG) &
IWN_HW_IF_CONFIG_NIC_READY)
return 0;
DELAY(10);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return ETIMEDOUT;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static int
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_hw_init(struct iwn_softc *sc)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
struct iwn_ops *ops = &sc->ops;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
int error, chnl, qid;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Clear pending interrupts. */
IWN_WRITE(sc, IWN_INT, 0xffffffff);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_apm_init(sc)) != 0) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
device_printf(sc->sc_dev,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"%s: could not power ON adapter, error %d\n", __func__,
error);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Select VMAIN power source. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
iwn_prph_clrbits(sc, IWN_APMG_PS, IWN_APMG_PS_PWR_SRC_MASK);
iwn_nic_unlock(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Perform adapter-specific initialization. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
if ((error = ops->nic_config(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Initialize RX ring. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
IWN_WRITE(sc, IWN_FH_RX_CONFIG, 0);
IWN_WRITE(sc, IWN_FH_RX_WPTR, 0);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Set physical address of RX ring (256-byte aligned). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_FH_RX_BASE, sc->rxq.desc_dma.paddr >> 8);
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Set physical address of RX status (16-byte aligned). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_FH_STATUS_WPTR, sc->rxq.stat_dma.paddr >> 4);
/* Enable RX. */
IWN_WRITE(sc, IWN_FH_RX_CONFIG,
IWN_FH_RX_CONFIG_ENA |
IWN_FH_RX_CONFIG_IGN_RXF_EMPTY | /* HW bug workaround */
IWN_FH_RX_CONFIG_IRQ_DST_HOST |
IWN_FH_RX_CONFIG_SINGLE_FRAME |
IWN_FH_RX_CONFIG_RB_TIMEOUT(0) |
IWN_FH_RX_CONFIG_NRBD(IWN_RX_RING_COUNT_LOG));
iwn_nic_unlock(sc);
IWN_WRITE(sc, IWN_FH_RX_WPTR, (IWN_RX_RING_COUNT - 1) & ~7);
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_nic_lock(sc)) != 0)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Initialize TX scheduler. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
iwn_prph_write(sc, sc->sched_txfact_addr, 0);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Set physical address of "keep warm" page (16-byte aligned). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_FH_KW_ADDR, sc->kw_dma.paddr >> 4);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Initialize TX rings. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
for (qid = 0; qid < sc->ntxqs; qid++) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
struct iwn_tx_ring *txq = &sc->txq[qid];
Sync comments with OpenBSD.
2011-04-16 14:38:57 +00:00
/* Set physical address of TX ring (256-byte aligned). */
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_FH_CBBC_QUEUE(qid),
txq->desc_dma.paddr >> 8);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_nic_unlock(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Enable DMA channels. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
for (chnl = 0; chnl < sc->ndmachnls; chnl++) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_FH_TX_CONFIG(chnl),
IWN_FH_TX_CONFIG_DMA_ENA |
IWN_FH_TX_CONFIG_DMA_CREDIT_ENA);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Clear "radio off" and "commands blocked" bits. */
IWN_WRITE(sc, IWN_UCODE_GP1_CLR, IWN_UCODE_GP1_RFKILL);
IWN_WRITE(sc, IWN_UCODE_GP1_CLR, IWN_UCODE_GP1_CMD_BLOCKED);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Clear pending interrupts. */
IWN_WRITE(sc, IWN_INT, 0xffffffff);
/* Enable interrupt coalescing. */
IWN_WRITE(sc, IWN_INT_COALESCING, 512 / 8);
/* Enable interrupts. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_WRITE(sc, IWN_INT_MASK, sc->int_mask);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* _Really_ make sure "radio off" bit is cleared! */
IWN_WRITE(sc, IWN_UCODE_GP1_CLR, IWN_UCODE_GP1_RFKILL);
IWN_WRITE(sc, IWN_UCODE_GP1_CLR, IWN_UCODE_GP1_RFKILL);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Add some new features: - 6000 series devices need enhanced sensitivity calibration. - 6000 series devices need a different setting for the shadow reg. - set the IWN_FLAG_HAS_11N bit if the EEPROM says the device has 11n support. Obtained from: OpenBSD
2011-04-16 14:56:13 +00:00
/* Enable shadow registers. */
Commit over some work to prepare the iwn(4) driver for further chipset support. * Extend the hardware base_params structure to include a bunch of hardware flags indicating what is and isn't supported. * Convert a bunch of the initial hardware configuration conditionals to consult the base_params structure. * Add new calibration code for temperature calibration for the Centrino 2xxx series NICs. * Add new bluetooth coexistence code for Centrino 2xxx series NICs. * For NICs that support PAN (personal area networking), use a different transmit queue and command queue setup, in preparation for said PAN support. * Extend the calibration array in iwn_softc to include enough space for the new calibration types. Tested (by myself, if not mentioned): * Intel 4965 * Intel 5100 * Intel 6150 * Intel 2230 * Intel 2200 (eadler) * Intel 1030 * Intel 6200 * Intel 6230 * Intel 6250 * Intel 6150 * Intel 100 What doesn't work: * Intel 6235 - fails in calibration at startup TODO: * Testing on Intel 53xx series hardware Submitted by: Cedric Gross <cg@cgross.info>
2013-11-12 05:58:23 +00:00
if (sc->base_params->shadow_reg_enable)
Add some new features: - 6000 series devices need enhanced sensitivity calibration. - 6000 series devices need a different setting for the shadow reg. - set the IWN_FLAG_HAS_11N bit if the EEPROM says the device has 11n support. Obtained from: OpenBSD
2011-04-16 14:56:13 +00:00
IWN_SETBITS(sc, IWN_SHADOW_REG_CTRL, 0x800fffff);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
if ((error = ops->load_firmware(sc)) != 0) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
device_printf(sc->sc_dev,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"%s: could not load firmware, error %d\n", __func__,
error);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return error;
}
/* Wait at most one second for firmware alive notification. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = msleep(sc, &sc->sc_mtx, PCATCH, "iwninit", hz)) != 0) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
device_printf(sc->sc_dev,
"%s: timeout waiting for adapter to initialize, error %d\n",
__func__, error);
return error;
}
/* Do post-firmware initialization. */
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
return ops->post_alive(sc);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_hw_stop(struct iwn_softc *sc)
{
int chnl, qid, ntries;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_RESET, IWN_RESET_NEVO);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Disable interrupts. */
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
IWN_WRITE(sc, IWN_INT_MASK, 0);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_INT, 0xffffffff);
IWN_WRITE(sc, IWN_FH_INT, 0xffffffff);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
sc->sc_flags &= ~IWN_FLAG_USE_ICT;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Make sure we no longer hold the NIC lock. */
iwn_nic_unlock(sc);
/* Stop TX scheduler. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
iwn_prph_write(sc, sc->sched_txfact_addr, 0);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Stop all DMA channels. */
if (iwn_nic_lock(sc) == 0) {
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
for (chnl = 0; chnl < sc->ndmachnls; chnl++) {
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_WRITE(sc, IWN_FH_TX_CONFIG(chnl), 0);
for (ntries = 0; ntries < 200; ntries++) {
Don't timeout when stopping DMA channels. Obtained from: OpenBSD
2011-04-15 16:50:37 +00:00
if (IWN_READ(sc, IWN_FH_TX_STATUS) &
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_FH_TX_STATUS_IDLE(chnl))
break;
DELAY(10);
}
}
iwn_nic_unlock(sc);
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Stop RX ring. */
iwn_reset_rx_ring(sc, &sc->rxq);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Reset all TX rings. */
Bring over the HAL/OPS changes, instead of two const structs it is now slightly more dynamic. Obtained from: OpenBSD
2011-04-16 14:51:46 +00:00
for (qid = 0; qid < sc->ntxqs; qid++)
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_reset_tx_ring(sc, &sc->txq[qid]);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (iwn_nic_lock(sc) == 0) {
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_prph_write(sc, IWN_APMG_CLK_DIS,
IWN_APMG_CLK_CTRL_DMA_CLK_RQT);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_nic_unlock(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
DELAY(5);
/* Power OFF adapter. */
iwn_apm_stop(sc);
}
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
static void
iwn_radio_on(void *arg0, int pending)
{
struct iwn_softc *sc = arg0;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
if (vap != NULL) {
iwn_init(sc);
ieee80211_init(vap);
}
}
static void
iwn_radio_off(void *arg0, int pending)
{
struct iwn_softc *sc = arg0;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Shuffle code around a bit. Mostly to group functional connected things, others to get the same order as the OpenBSD code.
2011-04-16 14:36:43 +00:00
iwn_stop(sc);
if (vap != NULL)
ieee80211_stop(vap);
/* Enable interrupts to get RF toggle notification. */
IWN_LOCK(sc);
IWN_WRITE(sc, IWN_INT, 0xffffffff);
IWN_WRITE(sc, IWN_INT_MASK, sc->int_mask);
IWN_UNLOCK(sc);
}
Make iwn(4) able to get itself back into working condition after "fatal firmware error" happens. Previously it was neccessary to reset it manually, using "/etc/rc.d/netif restart". Approved by: adrian@ MFC after: 1 month Sponsored by: The FreeBSD Foundation
2014-05-22 15:29:25 +00:00
static void
iwn_panicked(void *arg0, int pending)
{
struct iwn_softc *sc = arg0;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Make iwn(4) able to get itself back into working condition after "fatal firmware error" happens. Previously it was neccessary to reset it manually, using "/etc/rc.d/netif restart". Approved by: adrian@ MFC after: 1 month Sponsored by: The FreeBSD Foundation
2014-05-22 15:29:25 +00:00
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
int error;
if (vap == NULL) {
printf("%s: null vap\n", __func__);
return;
}
device_printf(sc->sc_dev, "%s: controller panicked, iv_state = %d; "
"resetting...\n", __func__, vap->iv_state);
Do the iwn(4) panic reinitialisation under IWN_LOCK(). I've checked each of the functions being called and there's either a _locked version or it's supposed to be called with IWN_LOCK() held.
2014-08-03 03:51:33 +00:00
IWN_LOCK(sc);
iwn_stop_locked(sc);
iwn_init_locked(sc);
Make iwn(4) able to get itself back into working condition after "fatal firmware error" happens. Previously it was neccessary to reset it manually, using "/etc/rc.d/netif restart". Approved by: adrian@ MFC after: 1 month Sponsored by: The FreeBSD Foundation
2014-05-22 15:29:25 +00:00
if (vap->iv_state >= IEEE80211_S_AUTH &&
(error = iwn_auth(sc, vap)) != 0) {
device_printf(sc->sc_dev,
"%s: could not move to auth state\n", __func__);
}
if (vap->iv_state >= IEEE80211_S_RUN &&
(error = iwn_run(sc, vap)) != 0) {
device_printf(sc->sc_dev,
"%s: could not move to run state\n", __func__);
}
Do the iwn(4) panic reinitialisation under IWN_LOCK(). I've checked each of the functions being called and there's either a _locked version or it's supposed to be called with IWN_LOCK() held.
2014-08-03 03:51:33 +00:00
IWN_UNLOCK(sc);
Make iwn(4) able to get itself back into working condition after "fatal firmware error" happens. Previously it was neccessary to reset it manually, using "/etc/rc.d/netif restart". Approved by: adrian@ MFC after: 1 month Sponsored by: The FreeBSD Foundation
2014-05-22 15:29:25 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_init_locked(struct iwn_softc *sc)
{
int error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s begin\n", __func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
IWN_LOCK_ASSERT(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
sc->sc_flags |= IWN_FLAG_RUNNING;
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_hw_prepare(sc)) != 0) {
Sync debug and error messages with OpenBSD.
2011-04-16 14:38:01 +00:00
device_printf(sc->sc_dev, "%s: hardware not ready, error %d\n",
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
__func__, error);
goto fail;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
/* Initialize interrupt mask to default value. */
sc->int_mask = IWN_INT_MASK_DEF;
sc->sc_flags &= ~IWN_FLAG_USE_ICT;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Check that the radio is not disabled by hardware switch. */
if (!(IWN_READ(sc, IWN_GP_CNTRL) & IWN_GP_CNTRL_RFKILL)) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(sc->sc_dev,
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
"radio is disabled by hardware switch\n");
/* Enable interrupts to get RF toggle notifications. */
IWN_WRITE(sc, IWN_INT, 0xffffffff);
IWN_WRITE(sc, IWN_INT_MASK, sc->int_mask);
return;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Read firmware images from the filesystem. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_read_firmware(sc)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(sc->sc_dev,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"%s: could not read firmware, error %d\n", __func__,
error);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
goto fail;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Initialize hardware and upload firmware. */
error = iwn_hw_init(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
firmware_put(sc->fw_fp, FIRMWARE_UNLOAD);
sc->fw_fp = NULL;
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
if (error != 0) {
device_printf(sc->sc_dev,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"%s: could not initialize hardware, error %d\n", __func__,
error);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
goto fail;
}
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Configure adapter now that it is ready. */
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
if ((error = iwn_config(sc)) != 0) {
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
device_printf(sc->sc_dev,
Just a whitespace sync, some more style(9) conform then others.
2011-04-16 14:40:42 +00:00
"%s: could not configure device, error %d\n", __func__,
error);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
goto fail;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
callout_reset(&sc->watchdog_to, hz, iwn_watchdog, sc);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end\n",__func__);
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
return;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
fail:
sc->sc_flags &= ~IWN_FLAG_RUNNING;
iwn_stop_locked(sc);
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->%s: end in error\n",__func__);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
iwn_init(struct iwn_softc *sc)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
IWN_LOCK(sc);
iwn_init_locked(sc);
IWN_UNLOCK(sc);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
if (sc->sc_flags & IWN_FLAG_RUNNING)
ieee80211_start_all(&sc->sc_ic);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_stop_locked(struct iwn_softc *sc)
{
IWN_LOCK_ASSERT(sc);
Overhaul the iwn(4) scan infrastructure to be slightly more "correct" for these chipsets. * Correctly set the active/passive flag in the scan request - this is NOT a "is the channel active|passive"; it's to do with whether we have an SSID to actively scan for or not. The firmware takes care of the active/passive setup of the channel. * Calculate the active/passive dwell time based on the beacon interval and the channel mode, rather than using a hard coded value. * For now, hardcode the scan service_time. It's defined as: 31:22 - number of beacon intervals to come back onto the home channel for; 0:21 - time (microseconds) to come back onto the home channel for. When doing an active scan when the NIC is active (whether we're associated or not - it only matters if we've setup the NIC to a destination or not) this determines how much time to stay on the home channel for when scanning. We can tune this based on the amount of active traffic. For now it's 4 beacon intervals and 100 microseconds. * Fix the "good crc threshold" setting. It differs based on the NIC firmware. Some older firmware required a workaround; the later firmware instead treats the field as a flag. * Enforce that we are not sending a scan command if one is already pending. Any time this is done is a bug and it absolutely needs to be fixed - so be very loud. * Add the SCAN flag to a few debug messages that are scan related but only occuring under STATE. Now, this does get noisy when you're scanning in an actively busy 2GHz network as the firmware (for reason I don't quite yet understand) seems hell bent on staying on some passive channels longer than it should. However, it should eventually recover and complete the scan. This is a work in progress; please let me know if things get stuck or if things improve! Tested: * intel centrino 2200 * intel centrino 2230 * intel 6200 * intel 5100 * intel 4965 (gets upset, but that's a known issue) Obtained from: linux iwlwifi
2013-12-02 03:59:45 +00:00
sc->sc_is_scanning = 0;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
sc->sc_tx_timer = 0;
Split up watchdog and calibration callout. This allows us to use different timing on both and to remove some monitor mode specific hacks (which has no calibration).
2011-04-15 20:17:52 +00:00
callout_stop(&sc->watchdog_to);
callout_stop(&sc->calib_to);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
sc->sc_flags &= ~IWN_FLAG_RUNNING;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
/* Power OFF hardware. */
iwn_hw_stop(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
Declare all functions as static. Approved by: rpaulo (mentor) MFC after: 2 weeks
2010-04-11 13:51:20 +00:00
static void
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
iwn_stop(struct iwn_softc *sc)
{
IWN_LOCK(sc);
iwn_stop_locked(sc);
IWN_UNLOCK(sc);
}
/*
* Callback from net80211 to start a scan.
*/
static void
iwn_scan_start(struct ieee80211com *ic)
{
Migrate ifp->if_softc -> ic->ic_softc. Tested: * Lenovo T400 (Intel 5300) * make universe
2015-08-17 23:35:31 +00:00
struct iwn_softc *sc = ic->ic_softc;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
IWN_LOCK(sc);
/* make the link LED blink while we're scanning */
iwn_set_led(sc, IWN_LED_LINK, 20, 2);
IWN_UNLOCK(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
/*
* Callback from net80211 to terminate a scan.
*/
static void
iwn_scan_end(struct ieee80211com *ic)
{
Migrate ifp->if_softc -> ic->ic_softc. Tested: * Lenovo T400 (Intel 5300) * make universe
2015-08-17 23:35:31 +00:00
struct iwn_softc *sc = ic->ic_softc;
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
IWN_LOCK(sc);
if (vap->iv_state == IEEE80211_S_RUN) {
/* Set link LED to ON status if we are associated */
iwn_set_led(sc, IWN_LED_LINK, 0, 1);
}
IWN_UNLOCK(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
/*
* Callback from net80211 to force a channel change.
*/
static void
iwn_set_channel(struct ieee80211com *ic)
{
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
const struct ieee80211_channel *c = ic->ic_curchan;
Migrate ifp->if_softc -> ic->ic_softc. Tested: * Lenovo T400 (Intel 5300) * make universe
2015-08-17 23:35:31 +00:00
struct iwn_softc *sc = ic->ic_softc;
Manually set the channel when using monitor mode - the firmware doesn't select it automatically. Submitted by: nox Reviewed by: bschmidt Approved by: re PR: kern/160815
2011-09-20 04:30:23 +00:00
int error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
IWN_LOCK(sc);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
sc->sc_rxtap.wr_chan_freq = htole16(c->ic_freq);
sc->sc_rxtap.wr_chan_flags = htole16(c->ic_flags);
sc->sc_txtap.wt_chan_freq = htole16(c->ic_freq);
sc->sc_txtap.wt_chan_flags = htole16(c->ic_flags);
Manually set the channel when using monitor mode - the firmware doesn't select it automatically. Submitted by: nox Reviewed by: bschmidt Approved by: re PR: kern/160815
2011-09-20 04:30:23 +00:00
/*
* Only need to set the channel in Monitor mode. AP scanning and auth
* are already taken care of by their respective firmware commands.
*/
if (ic->ic_opmode == IEEE80211_M_MONITOR) {
error = iwn_config(sc);
if (error != 0)
device_printf(sc->sc_dev,
"%s: error %d settting channel\n", __func__, error);
}
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
IWN_UNLOCK(sc);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
/*
* Callback from net80211 to start scanning of the current channel.
*/
static void
iwn_scan_curchan(struct ieee80211_scan_state *ss, unsigned long maxdwell)
{
struct ieee80211vap *vap = ss->ss_vap;
Add a channel parameter to iwn_scan(). This is in preparation for being able to use iwn_scan() to do an off channel scan to reset the RF tuning. It should be a no-op. Tested: * Intel 5100 (STA)
2013-12-07 08:25:24 +00:00
struct ieee80211com *ic = vap->iv_ic;
Store softc in ic_softc and access it without using struct ifnet. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 15:09:17 +00:00
struct iwn_softc *sc = ic->ic_softc;
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
int error;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
IWN_LOCK(sc);
Refactor out the scan id and scan vap as part of the scan work. Make the scan state optional - we'll obviously need a vap, but we now won't require the scan state. the only thing the scan state is needed for is to check for the list of SSIDs to scan - which we can now just plain ignore by passing in NULL as the scan state pointer. Tested: * Intel 5100 (STA)
2013-12-07 08:32:15 +00:00
error = iwn_scan(sc, vap, ss, ic->ic_curchan);
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
IWN_UNLOCK(sc);
if (error != 0)
ieee80211_cancel_scan(vap);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
/*
* Callback from net80211 to handle the minimum dwell time being met.
* The intent is to terminate the scan but we just let the firmware
* notify us when it's finished as we have no safe way to abort it.
*/
static void
iwn_scan_mindwell(struct ieee80211_scan_state *ss)
{
/* NB: don't try to abort scan; wait for firmware to finish */
}
static void
Updated iwn(4) driver supporting the newer series, 5000, 5150 and 5300. Submitted by: Bernhard Schmidt <bschmidt at techwires.net>
2009-10-23 22:04:18 +00:00
iwn_hw_reset(void *arg0, int pending)
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
{
struct iwn_softc *sc = arg0;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
Implement some function tracing. Submitted by: Cedric GROSS <cg@cgross.info>
2013-07-27 14:58:23 +00:00
DPRINTF(sc, IWN_DEBUG_TRACE, "->Doing %s\n", __func__);
iwn(4) update. Notable changes: * new firmware * untested support for 1000 and 6000 series * bgscan support * remove unnecessary RXON changes * allow setting of country/regdomain by enforcing channel flags read from the EEPROM * suspend/resume fixes * RF kill switch fixes * LED adjustments * several bus_dma*() related fixes * addressed some LORs * many other bug fixes Submitted by: Bernhard Schmidt <bschmidt at techwires.net> Obtained from: Brandon Gooch <jamesbrandongooch at gmail dot com> (LED related changes), Benjamin Kaduk <kaduk at mit dot edu> (LOR fixes), OpenBSD
2009-12-29 19:47:34 +00:00
iwn_stop(sc);
Create a taskqueue for each wireless interface which provides a serialised sleepable context for net80211 driver callbacks. This removes the need for USB and firmware based drivers to roll their own code to defer the chip programming for state changes, scan requests, channel changes and mcast/promisc updates. When a driver callback completes the hardware state is now guaranteed to have been updated and is in sync with net80211 layer. This nukes around 1300 lines of code from the wireless device drivers making them more readable and less race prone. The net80211 layer has been updated as follows - all state/channel changes are serialised on the taskqueue. - ieee80211_new_state() always queues and can now be called from any context - scanning runs from a single taskq function and executes to completion. driver callbacks are synchronous so the channel, phy mode and rx filters are guaranteed to be set in hardware before probe request frames are transmitted. Help and contributions from Sam Leffler. Reviewed by: sam
2009-05-02 15:14:18 +00:00
iwn_init(sc);
ieee80211_notify_radio(ic, 1);
Intel 4965 wireless driver (derived from openbsd driver of the same name)
2008-04-29 21:36:17 +00:00
}
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
#ifdef IWN_DEBUG
#define IWN_DESC(x) case x: return #x
/*
Fixing a typo. Approved by: sbruno (mentor, implicit)
2013-08-04 19:54:47 +00:00
* Translate CSR code to string
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
*/
static char *iwn_get_csr_string(int csr)
{
switch (csr) {
IWN_DESC(IWN_HW_IF_CONFIG);
IWN_DESC(IWN_INT_COALESCING);
IWN_DESC(IWN_INT);
IWN_DESC(IWN_INT_MASK);
IWN_DESC(IWN_FH_INT);
IWN_DESC(IWN_GPIO_IN);
IWN_DESC(IWN_RESET);
IWN_DESC(IWN_GP_CNTRL);
IWN_DESC(IWN_HW_REV);
IWN_DESC(IWN_EEPROM);
IWN_DESC(IWN_EEPROM_GP);
IWN_DESC(IWN_OTP_GP);
IWN_DESC(IWN_GIO);
IWN_DESC(IWN_GP_UCODE);
IWN_DESC(IWN_GP_DRIVER);
IWN_DESC(IWN_UCODE_GP1);
IWN_DESC(IWN_UCODE_GP2);
IWN_DESC(IWN_LED);
IWN_DESC(IWN_DRAM_INT_TBL);
IWN_DESC(IWN_GIO_CHICKEN);
IWN_DESC(IWN_ANA_PLL);
IWN_DESC(IWN_HW_REV_WA);
IWN_DESC(IWN_DBG_HPET_MEM);
default:
return "UNKNOWN CSR";
}
}
/*
* This function print firmware register
*/
static void
iwn_debug_register(struct iwn_softc *sc)
{
int i;
static const uint32_t csr_tbl[] = {
IWN_HW_IF_CONFIG,
IWN_INT_COALESCING,
IWN_INT,
IWN_INT_MASK,
IWN_FH_INT,
IWN_GPIO_IN,
IWN_RESET,
IWN_GP_CNTRL,
IWN_HW_REV,
IWN_EEPROM,
IWN_EEPROM_GP,
IWN_OTP_GP,
IWN_GIO,
IWN_GP_UCODE,
IWN_GP_DRIVER,
IWN_UCODE_GP1,
IWN_UCODE_GP2,
IWN_LED,
IWN_DRAM_INT_TBL,
IWN_GIO_CHICKEN,
IWN_ANA_PLL,
IWN_HW_REV_WA,
IWN_DBG_HPET_MEM,
};
DPRINTF(sc, IWN_DEBUG_REGISTER,
"CSR values: (2nd byte of IWN_INT_COALESCING is IWN_INT_PERIODIC)%s",
"\n");
Various iwn(4) fixes. * simplify channel logic for determining RF gain setting in scan setup * don't set TX timer on error * free node references for unsent frames on device stop * set maxfrags to IWN_MAX_SCATTER-1 (first segment is used by TX command) * add missing IWN_UNLOCK() from interrupt path when the hardware disappears. * pass control frames to host * nitems() instead of local macro Tested: * Intel 5100, STA mode PR: kern/196264 Submitted by: Andriy Voskoboinyk <s3erios@gmail.com>
2015-05-16 05:59:25 +00:00
for (i = 0; i < nitems(csr_tbl); i++){
iwn(4) debugging improvements. * Add in some new register debugging under IWN_DEBUG_REGISTER * Make IWN_DEBUG an option now for building. I'll chase this up with a commit to 'options' soon. Submitted by: Cedric GROSS <cg@cgross.info>
2013-08-01 21:45:30 +00:00
DPRINTF(sc, IWN_DEBUG_REGISTER," %10s: 0x%08x ",
iwn_get_csr_string(csr_tbl[i]), IWN_READ(sc, csr_tbl[i]));
if ((i+1) % 3 == 0)
DPRINTF(sc, IWN_DEBUG_REGISTER,"%s","\n");
}
DPRINTF(sc, IWN_DEBUG_REGISTER,"%s","\n");
}
#endif
Migrate the stats API for iwn(4) into a cdev ioctl, rather than tying into the vap. This allows for possible hardware interaction without needing a vap configured.
2015-08-30 21:54:33 +00:00
Reference in New Issue Copy Permalink