d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
4046851367
freebsd-dev/sys/dev/usb/wlan/if_rsu.c

3743 lines
95 KiB
C
Raw Normal View History

Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* $OpenBSD: if_rsu.c,v 1.17 2013/04/15 09:23:01 mglocker Exp $ */
/*-
* Copyright (c) 2010 Damien Bergamini <damien.bergamini@free.fr>
*
* 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.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
/*
* Driver for Realtek RTL8188SU/RTL8191SU/RTL8192SU.
*
* TODO:
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
* o tx a-mpdu
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
* o hostap / ibss / mesh
Flip on 11n by default; update TODO items.
2015-09-21 02:32:11 +00:00
* o power-save operation
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
*/
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
include opt_wlan.h . Tsk adrian.
2015-09-28 01:16:44 +00:00
#include "opt_wlan.h"
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
#include <sys/param.h>
#include <sys/endian.h>
#include <sys/sockio.h>
These files were getting sys/malloc.h and vm/uma.h with header pollution via sys/mbuf.h
2016-02-01 17:41:21 +00:00
#include <sys/malloc.h>
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
#include <sys/mbuf.h>
#include <sys/kernel.h>
#include <sys/socket.h>
#include <sys/systm.h>
#include <sys/conf.h>
#include <sys/bus.h>
#include <sys/rman.h>
#include <sys/firmware.h>
#include <sys/module.h>
#include <machine/bus.h>
#include <machine/resource.h>
#include <net/bpf.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>
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
#include <net/if_arp.h>
#include <net/if_dl.h>
#include <net/if_media.h>
#include <net/if_types.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
#include <netinet/if_ether.h>
#include <netinet/ip.h>
#include <net80211/ieee80211_var.h>
#include <net80211/ieee80211_regdomain.h>
#include <net80211/ieee80211_radiotap.h>
#include <dev/usb/usb.h>
#include <dev/usb/usbdi.h>
#include "usbdevs.h"
rsu(4): trim code for Rx rate calculation. Include ridx <-> rate conversion functions from rtwn(4) + reuse already calculated value for ieee80211_radiotap(9). Tested with Asus USB-N10, STA mode.
2017-11-02 00:17:52 +00:00
#include <dev/rtwn/if_rtwn_ridx.h> /* XXX */
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
#include <dev/usb/wlan/if_rsureg.h>
rsu(4): trim code for Rx rate calculation. Include ridx <-> rate conversion functions from rtwn(4) + reuse already calculated value for ieee80211_radiotap(9). Tested with Asus USB-N10, STA mode.
2017-11-02 00:17:52 +00:00
#define RSU_RATE_IS_CCK RTWN_RATE_IS_CCK
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
#ifdef USB_DEBUG
static int rsu_debug = 0;
SYSCTL_NODE(_hw_usb, OID_AUTO, rsu, CTLFLAG_RW, 0, "USB rsu");
Make a bunch of USB debug SYSCTLs tunable, so that their value(s) can be set before the USB device(s) are probed.
2015-01-05 15:04:17 +00:00
SYSCTL_INT(_hw_usb_rsu, OID_AUTO, debug, CTLFLAG_RWTUN, &rsu_debug, 0,
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
"Debug level");
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
#define RSU_DPRINTF(_sc, _flg, ...) \
do \
if (((_flg) == (RSU_DEBUG_ANY)) || (rsu_debug & (_flg))) \
device_printf((_sc)->sc_dev, __VA_ARGS__); \
while (0)
#else
#define RSU_DPRINTF(_sc, _flg, ...)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
#endif
Flip on 11n by default; update TODO items.
2015-09-21 02:32:11 +00:00
static int rsu_enable_11n = 1;
Add initial 11n support to if_rsu. * Add a tunable to enable 11n if it's available, so to not anger people who upgrade. kenv hw.usb.rsu.enable_11n=1 before inserting the device. * Add initial 11n htconfig bits; * Enable 40MHz mode if it's available; * Add 11n channels; * Set 11n bits in the firmware. It works for RX; I haven't tested TX aggregation just yet. However the firmware doesn't do RX re-ordering, so I have to tie it into the net80211 A-MPDU RX reorder path before I flip this on by default. I've verified that I'm indeed actually seeing MCS 0->7 rates being received. I haven't dug into whether it's actually transmitting 11n rates; I'll dig into that later.
2015-09-18 04:12:11 +00:00
TUNABLE_INT("hw.usb.rsu.enable_11n", &rsu_enable_11n);
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
#define RSU_DEBUG_ANY 0xffffffff
#define RSU_DEBUG_TX 0x00000001
#define RSU_DEBUG_RX 0x00000002
#define RSU_DEBUG_RESET 0x00000004
#define RSU_DEBUG_CALIB 0x00000008
#define RSU_DEBUG_STATE 0x00000010
#define RSU_DEBUG_SCAN 0x00000020
#define RSU_DEBUG_FWCMD 0x00000040
#define RSU_DEBUG_TXDONE 0x00000080
#define RSU_DEBUG_FW 0x00000100
#define RSU_DEBUG_FWDBG 0x00000200
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
#define RSU_DEBUG_AMPDU 0x00000400
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
#define RSU_DEBUG_KEY 0x00000800
rsu: use bitmap for all debug messages. - Replace all remaining DPRINTF(N)'s with RSU_DPRINTF. - Add new RSU_DEBUG_USB flag to track error codes returned by usbd_do_request_flags(). - Improve few messages.
2016-12-10 20:19:57 +00:00
#define RSU_DEBUG_USB 0x00001000
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static const STRUCT_USB_HOST_ID rsu_devs[] = {
#define RSU_HT_NOT_SUPPORTED 0
#define RSU_HT_SUPPORTED 1
#define RSU_DEV_HT(v,p) { USB_VPI(USB_VENDOR_##v, USB_PRODUCT_##v##_##p, \
RSU_HT_SUPPORTED) }
#define RSU_DEV(v,p) { USB_VPI(USB_VENDOR_##v, USB_PRODUCT_##v##_##p, \
RSU_HT_NOT_SUPPORTED) }
RSU_DEV(ASUS, RTL8192SU),
RSU_DEV(AZUREWAVE, RTL8192SU_4),
RSU_DEV_HT(ACCTON, RTL8192SU),
RSU_DEV_HT(ASUS, USBN10),
RSU_DEV_HT(AZUREWAVE, RTL8192SU_1),
RSU_DEV_HT(AZUREWAVE, RTL8192SU_2),
RSU_DEV_HT(AZUREWAVE, RTL8192SU_3),
RSU_DEV_HT(AZUREWAVE, RTL8192SU_5),
RSU_DEV_HT(BELKIN, RTL8192SU_1),
RSU_DEV_HT(BELKIN, RTL8192SU_2),
RSU_DEV_HT(BELKIN, RTL8192SU_3),
RSU_DEV_HT(CONCEPTRONIC2, RTL8192SU_1),
RSU_DEV_HT(CONCEPTRONIC2, RTL8192SU_2),
RSU_DEV_HT(CONCEPTRONIC2, RTL8192SU_3),
RSU_DEV_HT(COREGA, RTL8192SU),
RSU_DEV_HT(DLINK2, DWA131A1),
RSU_DEV_HT(DLINK2, RTL8192SU_1),
RSU_DEV_HT(DLINK2, RTL8192SU_2),
RSU_DEV_HT(EDIMAX, RTL8192SU_1),
RSU_DEV_HT(EDIMAX, RTL8192SU_2),
Use the real product name for Edimax rsu(4) id.
2013-11-12 02:34:48 +00:00
RSU_DEV_HT(EDIMAX, EW7622UMN),
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_DEV_HT(GUILLEMOT, HWGUN54),
RSU_DEV_HT(GUILLEMOT, HWNUM300),
RSU_DEV_HT(HAWKING, RTL8192SU_1),
RSU_DEV_HT(HAWKING, RTL8192SU_2),
RSU_DEV_HT(PLANEX2, GWUSNANO),
RSU_DEV_HT(REALTEK, RTL8171),
RSU_DEV_HT(REALTEK, RTL8172),
RSU_DEV_HT(REALTEK, RTL8173),
RSU_DEV_HT(REALTEK, RTL8174),
RSU_DEV_HT(REALTEK, RTL8192SU),
RSU_DEV_HT(REALTEK, RTL8712),
RSU_DEV_HT(REALTEK, RTL8713),
RSU_DEV_HT(SENAO, RTL8192SU_1),
RSU_DEV_HT(SENAO, RTL8192SU_2),
RSU_DEV_HT(SITECOMEU, WL349V1),
RSU_DEV_HT(SITECOMEU, WL353),
RSU_DEV_HT(SWEEX2, LW154),
Add Trendnet TEW-646UBH. Submitted by: bolshakov_1 at yahoo dot com
2014-11-19 05:38:45 +00:00
RSU_DEV_HT(TRENDNET, TEW646UBH),
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
#undef RSU_DEV_HT
#undef RSU_DEV
};
static device_probe_t rsu_match;
static device_attach_t rsu_attach;
static device_detach_t rsu_detach;
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
static usb_callback_t rsu_bulk_tx_callback_be_bk;
static usb_callback_t rsu_bulk_tx_callback_vi_vo;
Use the H2C endpoint for sending firmware commands, rather than the voice data queues. This is similar to the openbsd and rtlwifi/r92su drivers. Note: this driver still assumes it's a 4-endpoint device; I'll enforce that in a follow-up commit.
2015-09-17 03:19:09 +00:00
static usb_callback_t rsu_bulk_tx_callback_h2c;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static usb_callback_t rsu_bulk_rx_callback;
static usb_error_t rsu_do_request(struct rsu_softc *,
struct usb_device_request *, void *);
static struct ieee80211vap *
rsu_vap_create(struct ieee80211com *, const char name[],
int, enum ieee80211_opmode, int, const uint8_t bssid[],
const uint8_t mac[]);
static void rsu_vap_delete(struct ieee80211vap *);
static void rsu_scan_start(struct ieee80211com *);
static void rsu_scan_end(struct ieee80211com *);
urtwn, rtwn, rsu: switch to ieee80211_add_channel_list_2ghz(). - Use device's channel list instead of default one (from ieee80211_init_channels()); adds 12 - 14 2GHz channels. - Add ic_getradiocaps() method.
2016-05-26 16:39:11 +00:00
static void rsu_getradiocaps(struct ieee80211com *, int, int *,
struct ieee80211_channel[]);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static void rsu_set_channel(struct ieee80211com *);
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
static void rsu_scan_curchan(struct ieee80211_scan_state *, unsigned long);
static void rsu_scan_mindwell(struct ieee80211_scan_state *);
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
static void rsu_update_promisc(struct ieee80211com *);
rsu: add support for hardware multicast filter setup. The algorithm is the same as in rtwn(4). Tested with Asus USB-N10 (STA) + RTL8188CUS (AP).
2016-11-27 18:06:03 +00:00
static uint8_t rsu_get_multi_pos(const uint8_t[]);
static void rsu_set_multi(struct rsu_softc *);
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 rsu_update_mcast(struct ieee80211com *);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static int rsu_alloc_rx_list(struct rsu_softc *);
static void rsu_free_rx_list(struct rsu_softc *);
static int rsu_alloc_tx_list(struct rsu_softc *);
static void rsu_free_tx_list(struct rsu_softc *);
static void rsu_free_list(struct rsu_softc *, struct rsu_data [], int);
static struct rsu_data *_rsu_getbuf(struct rsu_softc *);
static struct rsu_data *rsu_getbuf(struct rsu_softc *);
Refactor out the tx buffer free code into a routine, rsu_freebuf(). This makes it easier to add more transmit buffers, have different buffer pools for things, etc.
2015-09-18 07:26:34 +00:00
static void rsu_freebuf(struct rsu_softc *, struct rsu_data *);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static int rsu_write_region_1(struct rsu_softc *, uint16_t, uint8_t *,
int);
static void rsu_write_1(struct rsu_softc *, uint16_t, uint8_t);
static void rsu_write_2(struct rsu_softc *, uint16_t, uint16_t);
static void rsu_write_4(struct rsu_softc *, uint16_t, uint32_t);
static int rsu_read_region_1(struct rsu_softc *, uint16_t, uint8_t *,
int);
static uint8_t rsu_read_1(struct rsu_softc *, uint16_t);
static uint16_t rsu_read_2(struct rsu_softc *, uint16_t);
static uint32_t rsu_read_4(struct rsu_softc *, uint16_t);
static int rsu_fw_iocmd(struct rsu_softc *, uint32_t);
static uint8_t rsu_efuse_read_1(struct rsu_softc *, uint16_t);
static int rsu_read_rom(struct rsu_softc *);
static int rsu_fw_cmd(struct rsu_softc *, uint8_t, void *, int);
static void rsu_calib_task(void *, int);
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
static void rsu_tx_task(void *, int);
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
static void rsu_set_led(struct rsu_softc *, int);
static int rsu_monitor_newstate(struct ieee80211vap *,
enum ieee80211_state, int);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static int rsu_newstate(struct ieee80211vap *, enum ieee80211_state, int);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
static int rsu_key_alloc(struct ieee80211vap *, struct ieee80211_key *,
ieee80211_keyix *, ieee80211_keyix *);
static int rsu_process_key(struct ieee80211vap *,
const struct ieee80211_key *, int);
static int rsu_key_set(struct ieee80211vap *,
const struct ieee80211_key *);
static int rsu_key_delete(struct ieee80211vap *,
const struct ieee80211_key *);
static int rsu_cam_read(struct rsu_softc *, uint8_t, uint32_t *);
static void rsu_cam_write(struct rsu_softc *, uint8_t, uint32_t);
static int rsu_key_check(struct rsu_softc *, ieee80211_keyix, int);
static uint8_t rsu_crypto_mode(struct rsu_softc *, u_int, int);
static int rsu_set_key_group(struct rsu_softc *,
const struct ieee80211_key *);
static int rsu_set_key_pair(struct rsu_softc *,
const struct ieee80211_key *);
static int rsu_reinit_static_keys(struct rsu_softc *);
static int rsu_delete_key(struct rsu_softc *sc, ieee80211_keyix);
static void rsu_delete_key_pair_cb(void *, int);
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
static int rsu_site_survey(struct rsu_softc *,
struct ieee80211_scan_ssid *);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static int rsu_join_bss(struct rsu_softc *, struct ieee80211_node *);
static int rsu_disconnect(struct rsu_softc *);
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
static int rsu_hwrssi_to_rssi(struct rsu_softc *, int hw_rssi);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static void rsu_event_survey(struct rsu_softc *, uint8_t *, int);
static void rsu_event_join_bss(struct rsu_softc *, uint8_t *, int);
static void rsu_rx_event(struct rsu_softc *, uint8_t, uint8_t *, int);
static void rsu_rx_multi_event(struct rsu_softc *, uint8_t *, int);
static int8_t rsu_get_rssi(struct rsu_softc *, int, void *);
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
static struct mbuf * rsu_rx_copy_to_mbuf(struct rsu_softc *,
struct r92s_rx_stat *, int);
rsu: add TSF field into Rx radiotap. Tested with Asus USB-N10.
2016-12-10 13:30:16 +00:00
static uint32_t rsu_get_tsf_low(struct rsu_softc *);
static uint32_t rsu_get_tsf_high(struct rsu_softc *);
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
static struct ieee80211_node * rsu_rx_frame(struct rsu_softc *, struct mbuf *);
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
static struct mbuf * rsu_rx_multi_frame(struct rsu_softc *, uint8_t *, int);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static struct mbuf *
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
rsu_rxeof(struct usb_xfer *, struct rsu_data *);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static void rsu_txeof(struct usb_xfer *, struct rsu_data *);
static int rsu_raw_xmit(struct ieee80211_node *, struct mbuf *,
const struct ieee80211_bpf_params *);
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
static void rsu_rxfilter_init(struct rsu_softc *);
static void rsu_rxfilter_set(struct rsu_softc *, uint32_t, uint32_t);
static void rsu_rxfilter_refresh(struct rsu_softc *);
rsu: various initialization fixes. - Do not ignore initialization errors; call ieee80211_stop() when initialization failed. - Use usb_pause_mtx() instead of DELAY() while waiting for firmware loading; this fixes system freeze during firmware startup. - Do not execute rsu_stop() when device is powered off; fixes 'unknown board type (rfconfig=0xff)' error when the device is reattached. Tested with Asus USB-N10.
2016-12-10 22:31:49 +00:00
static int rsu_init(struct rsu_softc *);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static int rsu_tx_start(struct rsu_softc *, struct ieee80211_node *,
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
struct mbuf *, struct rsu_data *);
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 rsu_transmit(struct ieee80211com *, struct mbuf *);
static void rsu_start(struct rsu_softc *);
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
static void _rsu_start(struct rsu_softc *);
rsu: fix and enable Rx TCP checksum offloading. Tested with Asus USB-N10, STA mode.
2016-12-11 17:15:25 +00:00
static int rsu_ioctl_net(struct ieee80211com *, u_long, 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 void rsu_parent(struct ieee80211com *);
static void rsu_stop(struct rsu_softc *);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
static void rsu_ms_delay(struct rsu_softc *, int);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static device_method_t rsu_methods[] = {
DEVMETHOD(device_probe, rsu_match),
DEVMETHOD(device_attach, rsu_attach),
DEVMETHOD(device_detach, rsu_detach),
DEVMETHOD_END
};
static driver_t rsu_driver = {
.name = "rsu",
.methods = rsu_methods,
.size = sizeof(struct rsu_softc)
};
static devclass_t rsu_devclass;
DRIVER_MODULE(rsu, uhub, rsu_driver, rsu_devclass, NULL, 0);
MODULE_DEPEND(rsu, wlan, 1, 1, 1);
MODULE_DEPEND(rsu, usb, 1, 1, 1);
MODULE_DEPEND(rsu, firmware, 1, 1, 1);
MODULE_VERSION(rsu, 1);
Create a USB_PNP_INFO and use it to export the existing PNP tables. Some drivers needed some slight re-arrangement of declarations to accommodate this. Change the USB pnp tables slightly to allow better compatibility with the system by moving linux driver info from start of each entry to the end. All other PNP tables in the system have the per-device flags and such at the end of the elements rather that at the beginning. Differential Review: https://reviews.freebsd.org/D3458
2015-12-11 05:28:00 +00:00
USB_PNP_HOST_INFO(rsu_devs);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
urtwn, rtwn, rsu: switch to ieee80211_add_channel_list_2ghz(). - Use device's channel list instead of default one (from ieee80211_init_channels()); adds 12 - 14 2GHz channels. - Add ic_getradiocaps() method.
2016-05-26 16:39:11 +00:00
static const uint8_t rsu_chan_2ghz[] =
{ 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 };
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
static uint8_t rsu_wme_ac_xfer_map[4] = {
[WME_AC_BE] = RSU_BULK_TX_BE_BK,
[WME_AC_BK] = RSU_BULK_TX_BE_BK,
[WME_AC_VI] = RSU_BULK_TX_VI_VO,
[WME_AC_VO] = RSU_BULK_TX_VI_VO,
};
Use the H2C endpoint for sending firmware commands, rather than the voice data queues. This is similar to the openbsd and rtlwifi/r92su drivers. Note: this driver still assumes it's a 4-endpoint device; I'll enforce that in a follow-up commit.
2015-09-17 03:19:09 +00:00
/* XXX hard-coded */
#define RSU_H2C_ENDPOINT 3
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static const struct usb_config rsu_config[RSU_N_TRANSFER] = {
[RSU_BULK_RX] = {
.type = UE_BULK,
.endpoint = UE_ADDR_ANY,
.direction = UE_DIR_IN,
.bufsize = RSU_RXBUFSZ,
.flags = {
.pipe_bof = 1,
.short_xfer_ok = 1
},
.callback = rsu_bulk_rx_callback
},
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
[RSU_BULK_TX_BE_BK] = {
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
.type = UE_BULK,
.endpoint = 0x06,
.direction = UE_DIR_OUT,
.bufsize = RSU_TXBUFSZ,
.flags = {
.ext_buffer = 1,
.pipe_bof = 1,
.force_short_xfer = 1
},
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
.callback = rsu_bulk_tx_callback_be_bk,
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
.timeout = RSU_TX_TIMEOUT
},
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
[RSU_BULK_TX_VI_VO] = {
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
.type = UE_BULK,
.endpoint = 0x04,
.direction = UE_DIR_OUT,
.bufsize = RSU_TXBUFSZ,
.flags = {
.ext_buffer = 1,
.pipe_bof = 1,
.force_short_xfer = 1
},
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
.callback = rsu_bulk_tx_callback_vi_vo,
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
.timeout = RSU_TX_TIMEOUT
},
Use the H2C endpoint for sending firmware commands, rather than the voice data queues. This is similar to the openbsd and rtlwifi/r92su drivers. Note: this driver still assumes it's a 4-endpoint device; I'll enforce that in a follow-up commit.
2015-09-17 03:19:09 +00:00
[RSU_BULK_TX_H2C] = {
.type = UE_BULK,
.endpoint = 0x0d,
.direction = UE_DIR_OUT,
.bufsize = RSU_TXBUFSZ,
.flags = {
.ext_buffer = 1,
.pipe_bof = 1,
.short_xfer_ok = 1
},
.callback = rsu_bulk_tx_callback_h2c,
.timeout = RSU_TX_TIMEOUT
},
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
};
static int
rsu_match(device_t self)
{
struct usb_attach_arg *uaa = device_get_ivars(self);
if (uaa->usb_mode != USB_MODE_HOST ||
uaa->info.bIfaceIndex != 0 ||
uaa->info.bConfigIndex != 0)
return (ENXIO);
return (usbd_lookup_id_by_uaa(rsu_devs, sizeof(rsu_devs), uaa));
}
Disable mgmt frame sending in if_rsu. The firmware in this NIC sends management frames. So far I'm not sure which ones it handles and which ones it doesn't handle - but this is what openbsd does. The association messages are handled by the firmware; the key negotiation for 802.1x and WPA are done as raw frames, not management frames. This successfully allows it to associate to my home networks whereas it didn't work beforehand. Tested: * RTL8712, cut 3, STA mode TODO: * The firmware does send a join response with a status code; that should be logged in a more obvious way to assist with debugging. Ie, the firmware is the thing that is saying "couldn't join, sorry!", not net80211.
2015-09-13 19:17:26 +00:00
static int
rsu_send_mgmt(struct ieee80211_node *ni, int type, int arg)
{
return (ENOTSUP);
}
Add 11n and QoS methods. The firmware takes care of ampdu tx/rx (except for RX reordering, grr), QoS/WME and other bits/pieces. So they're stubs, just in case.
2015-09-17 03:13:01 +00:00
static void
rsu_update_chw(struct ieee80211com *ic)
{
}
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
/*
* notification from net80211 that it'd like to do A-MPDU on the given TID.
*
* Note: this actually hangs traffic at the present moment, so don't use it.
* The firmware debug does indiciate it's sending and establishing a TX AMPDU
* session, but then no traffic flows.
*/
Add 11n and QoS methods. The firmware takes care of ampdu tx/rx (except for RX reordering, grr), QoS/WME and other bits/pieces. So they're stubs, just in case.
2015-09-17 03:13:01 +00:00
static int
rsu_ampdu_enable(struct ieee80211_node *ni, struct ieee80211_tx_ampdu *tap)
{
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
#if 0
struct rsu_softc *sc = ni->ni_ic->ic_softc;
struct r92s_add_ba_req req;
Add 11n and QoS methods. The firmware takes care of ampdu tx/rx (except for RX reordering, grr), QoS/WME and other bits/pieces. So they're stubs, just in case.
2015-09-17 03:13:01 +00:00
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
/* Don't enable if it's requested or running */
if (IEEE80211_AMPDU_REQUESTED(tap))
return (0);
if (IEEE80211_AMPDU_RUNNING(tap))
return (0);
/* We've decided to send addba; so send it */
req.tid = htole32(tap->txa_tid);
/* Attempt net80211 state */
if (ieee80211_ampdu_tx_request_ext(ni, tap->txa_tid) != 1)
return (0);
/* Send the firmware command */
RSU_DPRINTF(sc, RSU_DEBUG_AMPDU, "%s: establishing AMPDU TX for TID %d\n",
__func__,
tap->txa_tid);
RSU_LOCK(sc);
if (rsu_fw_cmd(sc, R92S_CMD_ADDBA_REQ, &req, sizeof(req)) != 1) {
RSU_UNLOCK(sc);
/* Mark failure */
(void) ieee80211_ampdu_tx_request_active_ext(ni, tap->txa_tid, 0);
return (0);
}
RSU_UNLOCK(sc);
/* Mark success; we don't get any further notifications */
(void) ieee80211_ampdu_tx_request_active_ext(ni, tap->txa_tid, 1);
#endif
/* Return 0, we're driving this ourselves */
Add 11n and QoS methods. The firmware takes care of ampdu tx/rx (except for RX reordering, grr), QoS/WME and other bits/pieces. So they're stubs, just in case.
2015-09-17 03:13:01 +00:00
return (0);
}
static int
rsu_wme_update(struct ieee80211com *ic)
{
/* Firmware handles this; not our problem */
return (0);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static int
rsu_attach(device_t self)
{
struct usb_attach_arg *uaa = device_get_ivars(self);
struct rsu_softc *sc = device_get_softc(self);
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;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
int error;
net80211 drivers: fix ieee80211_init_channels() usage Fix out-of-bounds read (all) / write (11n capable) for drivers that are using ieee80211_init_channels() to initialize channel list. Tested with: * RTL8188EU, STA mode. * RTL8188CUS, STA mode. * WUSB54GC, HOSTAP mode. Approved by: adrian (mentor) MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D4818
2016-01-07 18:41:03 +00:00
uint8_t iface_index;
Prepare for 11n - get the number of endpoints and whether 11n is available.
2015-09-17 03:08:02 +00:00
struct usb_interface *iface;
rsu(4): Add support for 1T2R and 2T2R NICs. This logic is mostly crimed from the reference driver and the linux r92su driver. I verified that it (a) worked on the rsu hardware I have, and (b) did traffic testing whilst watching what ath(4) sent as a hostap. It successfully sent MCS8..15 rates (which requires 2-stream reception) as well as MCS0..7 (which is 1-stream.) Tested: * RTL8712, 1T1R NIC, MCS rates 0..7. * RTL8712, 1T2R NIC, MCS rates 0..15 TODO: * Find a 2T2R NIC!
2015-09-29 06:56:00 +00:00
const char *rft;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
device_set_usb_desc(self);
sc->sc_udev = uaa->device;
sc->sc_dev = self;
rsu: fix and enable Rx TCP checksum offloading. Tested with Asus USB-N10, STA mode.
2016-12-11 17:15:25 +00:00
sc->sc_rx_checksum_enable = 1;
Add initial 11n support to if_rsu. * Add a tunable to enable 11n if it's available, so to not anger people who upgrade. kenv hw.usb.rsu.enable_11n=1 before inserting the device. * Add initial 11n htconfig bits; * Enable 40MHz mode if it's available; * Add 11n channels; * Set 11n bits in the firmware. It works for RX; I haven't tested TX aggregation just yet. However the firmware doesn't do RX re-ordering, so I have to tie it into the net80211 A-MPDU RX reorder path before I flip this on by default. I've verified that I'm indeed actually seeing MCS 0->7 rates being received. I haven't dug into whether it's actually transmitting 11n rates; I'll dig into that later.
2015-09-18 04:12:11 +00:00
if (rsu_enable_11n)
sc->sc_ht = !! (USB_GET_DRIVER_INFO(uaa) & RSU_HT_SUPPORTED);
Prepare for 11n - get the number of endpoints and whether 11n is available.
2015-09-17 03:08:02 +00:00
/* Get number of endpoints */
iface = usbd_get_iface(sc->sc_udev, 0);
sc->sc_nendpoints = iface->idesc->bNumEndpoints;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
Program the firmware setup stuff with the current hardware setup: * Do 1T1R for now, until we read the config out of ROM and use it. * Disable turbo mode, I dunno what this is, but the linux drivers have this disabled. * Set the firmware endpoints to what we read from USB. Tested: * RTL8712 cut 3, STA mode
2015-09-17 03:42:18 +00:00
/* Endpoints are hard-coded for now, so enforce 4-endpoint only */
if (sc->sc_nendpoints != 4) {
device_printf(sc->sc_dev,
"the driver currently only supports 4-endpoint devices\n");
return (ENXIO);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
mtx_init(&sc->sc_mtx, device_get_nameunit(self), MTX_NETWORK_LOCK,
MTX_DEF);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
RSU_DELKEY_BMAP_LOCK_INIT(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
TIMEOUT_TASK_INIT(taskqueue_thread, &sc->calib_task, 0,
rsu_calib_task, sc);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
TASK_INIT(&sc->del_key_task, 0, rsu_delete_key_pair_cb, sc);
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
TASK_INIT(&sc->tx_task, 0, rsu_tx_task, 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
mbufq_init(&sc->sc_snd, ifqmaxlen);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
- Stop transfers when RSU init fails. - Make sure TX/RX lists don't leak and are only allocated once. - Fix off-by one transfer index computation. - Give firmware loading more time. MFC after: 3 days
2014-05-22 06:28:09 +00:00
/* Allocate Tx/Rx buffers. */
error = rsu_alloc_rx_list(sc);
if (error != 0) {
device_printf(sc->sc_dev, "could not allocate Rx buffers\n");
goto fail_usb;
}
error = rsu_alloc_tx_list(sc);
if (error != 0) {
device_printf(sc->sc_dev, "could not allocate Tx buffers\n");
rsu_free_rx_list(sc);
goto fail_usb;
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
iface_index = 0;
error = usbd_transfer_setup(uaa->device, &iface_index, sc->sc_xfer,
rsu_config, RSU_N_TRANSFER, sc, &sc->sc_mtx);
if (error) {
device_printf(sc->sc_dev,
"could not allocate USB transfers, err=%s\n",
usbd_errstr(error));
Cleanup the allocations when the attachment fails.
2013-07-31 06:05:34 +00:00
goto fail_usb;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
RSU_LOCK(sc);
/* Read chip revision. */
sc->cut = MS(rsu_read_4(sc, R92S_PMC_FSM), R92S_PMC_FSM_CUT);
if (sc->cut != 3)
sc->cut = (sc->cut >> 1) + 1;
error = rsu_read_rom(sc);
Make sure an error case exits unlocked. Submitted by: Dmitry Luhtionov <dmitryluhtionov@gmail.com> MFC after: 1 week
2015-01-03 11:04:17 +00:00
RSU_UNLOCK(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (error != 0) {
device_printf(self, "could not read ROM\n");
Cleanup the allocations when the attachment fails.
2013-07-31 06:05:34 +00:00
goto fail_rom;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
rsu(4): Add support for 1T2R and 2T2R NICs. This logic is mostly crimed from the reference driver and the linux r92su driver. I verified that it (a) worked on the rsu hardware I have, and (b) did traffic testing whilst watching what ath(4) sent as a hostap. It successfully sent MCS8..15 rates (which requires 2-stream reception) as well as MCS0..7 (which is 1-stream.) Tested: * RTL8712, 1T1R NIC, MCS rates 0..7. * RTL8712, 1T2R NIC, MCS rates 0..15 TODO: * Find a 2T2R NIC!
2015-09-29 06:56:00 +00:00
/* Figure out TX/RX streams */
switch (sc->rom[84]) {
case 0x0:
sc->sc_rftype = RTL8712_RFCONFIG_1T1R;
sc->sc_nrxstream = 1;
sc->sc_ntxstream = 1;
rft = "1T1R";
break;
case 0x1:
sc->sc_rftype = RTL8712_RFCONFIG_1T2R;
sc->sc_nrxstream = 2;
sc->sc_ntxstream = 1;
rft = "1T2R";
break;
case 0x2:
sc->sc_rftype = RTL8712_RFCONFIG_2T2R;
sc->sc_nrxstream = 2;
sc->sc_ntxstream = 2;
rft = "2T2R";
break;
[rsu] add support for the "green" rsu NICs. They're still a 1T2R NIC, so reuse the same rfconfig and nstream configuration. Submitted by: Idwer Vollering <vidwer@gmail.com>
2017-01-09 21:46:24 +00:00
case 0x3: /* "green" NIC */
sc->sc_rftype = RTL8712_RFCONFIG_1T2R;
sc->sc_nrxstream = 2;
sc->sc_ntxstream = 1;
rft = "1T2R ('green')";
break;
rsu(4): Add support for 1T2R and 2T2R NICs. This logic is mostly crimed from the reference driver and the linux r92su driver. I verified that it (a) worked on the rsu hardware I have, and (b) did traffic testing whilst watching what ath(4) sent as a hostap. It successfully sent MCS8..15 rates (which requires 2-stream reception) as well as MCS0..7 (which is 1-stream.) Tested: * RTL8712, 1T1R NIC, MCS rates 0..7. * RTL8712, 1T2R NIC, MCS rates 0..15 TODO: * Find a 2T2R NIC!
2015-09-29 06:56:00 +00:00
default:
device_printf(sc->sc_dev,
"%s: unknown board type (rfconfig=0x%02x)\n",
__func__,
sc->rom[84]);
goto fail_rom;
}
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(ic->ic_macaddr, &sc->rom[0x12]);
rsu(4): Add support for 1T2R and 2T2R NICs. This logic is mostly crimed from the reference driver and the linux r92su driver. I verified that it (a) worked on the rsu hardware I have, and (b) did traffic testing whilst watching what ath(4) sent as a hostap. It successfully sent MCS8..15 rates (which requires 2-stream reception) as well as MCS0..7 (which is 1-stream.) Tested: * RTL8712, 1T1R NIC, MCS rates 0..7. * RTL8712, 1T2R NIC, MCS rates 0..15 TODO: * Find a 2T2R NIC!
2015-09-29 06:56:00 +00:00
device_printf(self, "MAC/BB RTL8712 cut %d %s\n", sc->cut, rft);
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
Set ic_softc in all 802.11 drivers. Not required right now, but will be used quite soon. Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-05-25 18:50:26 +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(self);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ic->ic_phytype = IEEE80211_T_OFDM; /* Not only, but not used. */
ic->ic_opmode = IEEE80211_M_STA; /* Default to BSS mode. */
/* Set device capabilities. */
ic->ic_caps =
IEEE80211_C_STA | /* station mode */
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
IEEE80211_C_MONITOR | /* monitor mode supported */
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
#if 0
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
IEEE80211_C_BGSCAN | /* Background scan. */
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
#endif
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
IEEE80211_C_SHPREAMBLE | /* Short preamble supported. */
Bring over the QoS logic from the Linux r92su driver. * the tx descriptor TID is priority, not TID. * the tx descriptor queue id mapping is separate from the TID/priority; rather than just "BE". TODO: * go and re-re-re-verify the queue mappings; the linux and openbsd mappings aren't exactly the same. I need to verify all of this before I try to flip on 11n RX.
2015-09-17 04:45:29 +00:00
IEEE80211_C_WME | /* WME/QoS */
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
IEEE80211_C_SHSLOT | /* Short slot time supported. */
IEEE80211_C_WPA; /* WPA/RSN. */
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
ic->ic_cryptocaps =
IEEE80211_CRYPTO_WEP |
IEEE80211_CRYPTO_TKIP |
IEEE80211_CRYPTO_AES_CCM;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Check if HT support is present. */
Add initial 11n support to if_rsu. * Add a tunable to enable 11n if it's available, so to not anger people who upgrade. kenv hw.usb.rsu.enable_11n=1 before inserting the device. * Add initial 11n htconfig bits; * Enable 40MHz mode if it's available; * Add 11n channels; * Set 11n bits in the firmware. It works for RX; I haven't tested TX aggregation just yet. However the firmware doesn't do RX re-ordering, so I have to tie it into the net80211 A-MPDU RX reorder path before I flip this on by default. I've verified that I'm indeed actually seeing MCS 0->7 rates being received. I haven't dug into whether it's actually transmitting 11n rates; I'll dig into that later.
2015-09-18 04:12:11 +00:00
if (sc->sc_ht) {
device_printf(sc->sc_dev, "%s: enabling 11n\n", __func__);
/* Enable basic HT */
ic->ic_htcaps = IEEE80211_HTC_HT |
[rsu] We don't do A-MPDU transmit right now, so don't bother registering for it.
2016-04-06 00:41:06 +00:00
#if 0
Add initial 11n support to if_rsu. * Add a tunable to enable 11n if it's available, so to not anger people who upgrade. kenv hw.usb.rsu.enable_11n=1 before inserting the device. * Add initial 11n htconfig bits; * Enable 40MHz mode if it's available; * Add 11n channels; * Set 11n bits in the firmware. It works for RX; I haven't tested TX aggregation just yet. However the firmware doesn't do RX re-ordering, so I have to tie it into the net80211 A-MPDU RX reorder path before I flip this on by default. I've verified that I'm indeed actually seeing MCS 0->7 rates being received. I haven't dug into whether it's actually transmitting 11n rates; I'll dig into that later.
2015-09-18 04:12:11 +00:00
IEEE80211_HTC_AMPDU |
[rsu] We don't do A-MPDU transmit right now, so don't bother registering for it.
2016-04-06 00:41:06 +00:00
#endif
Add initial 11n support to if_rsu. * Add a tunable to enable 11n if it's available, so to not anger people who upgrade. kenv hw.usb.rsu.enable_11n=1 before inserting the device. * Add initial 11n htconfig bits; * Enable 40MHz mode if it's available; * Add 11n channels; * Set 11n bits in the firmware. It works for RX; I haven't tested TX aggregation just yet. However the firmware doesn't do RX re-ordering, so I have to tie it into the net80211 A-MPDU RX reorder path before I flip this on by default. I've verified that I'm indeed actually seeing MCS 0->7 rates being received. I haven't dug into whether it's actually transmitting 11n rates; I'll dig into that later.
2015-09-18 04:12:11 +00:00
IEEE80211_HTC_AMSDU |
IEEE80211_HTCAP_MAXAMSDU_3839 |
IEEE80211_HTCAP_SMPS_OFF;
ic->ic_htcaps |= IEEE80211_HTCAP_CHWIDTH40;
/* set number of spatial streams */
rsu(4): Add support for 1T2R and 2T2R NICs. This logic is mostly crimed from the reference driver and the linux r92su driver. I verified that it (a) worked on the rsu hardware I have, and (b) did traffic testing whilst watching what ath(4) sent as a hostap. It successfully sent MCS8..15 rates (which requires 2-stream reception) as well as MCS0..7 (which is 1-stream.) Tested: * RTL8712, 1T1R NIC, MCS rates 0..7. * RTL8712, 1T2R NIC, MCS rates 0..15 TODO: * Find a 2T2R NIC!
2015-09-29 06:56:00 +00:00
ic->ic_txstream = sc->sc_ntxstream;
ic->ic_rxstream = sc->sc_nrxstream;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
ic->ic_flags_ext |= IEEE80211_FEXT_SCAN_OFFLOAD;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
urtwn, rtwn, rsu: switch to ieee80211_add_channel_list_2ghz(). - Use device's channel list instead of default one (from ieee80211_init_channels()); adds 12 - 14 2GHz channels. - Add ic_getradiocaps() method.
2016-05-26 16:39:11 +00:00
rsu_getradiocaps(ic, IEEE80211_CHAN_MAX, &ic->ic_nchans,
ic->ic_channels);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +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);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ic->ic_raw_xmit = rsu_raw_xmit;
ic->ic_scan_start = rsu_scan_start;
ic->ic_scan_end = rsu_scan_end;
urtwn, rtwn, rsu: switch to ieee80211_add_channel_list_2ghz(). - Use device's channel list instead of default one (from ieee80211_init_channels()); adds 12 - 14 2GHz channels. - Add ic_getradiocaps() method.
2016-05-26 16:39:11 +00:00
ic->ic_getradiocaps = rsu_getradiocaps;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ic->ic_set_channel = rsu_set_channel;
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
ic->ic_scan_curchan = rsu_scan_curchan;
ic->ic_scan_mindwell = rsu_scan_mindwell;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ic->ic_vap_create = rsu_vap_create;
ic->ic_vap_delete = rsu_vap_delete;
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
ic->ic_update_promisc = rsu_update_promisc;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ic->ic_update_mcast = rsu_update_mcast;
rsu: fix and enable Rx TCP checksum offloading. Tested with Asus USB-N10, STA mode.
2016-12-11 17:15:25 +00:00
ic->ic_ioctl = rsu_ioctl_net;
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_parent = rsu_parent;
ic->ic_transmit = rsu_transmit;
Disable mgmt frame sending in if_rsu. The firmware in this NIC sends management frames. So far I'm not sure which ones it handles and which ones it doesn't handle - but this is what openbsd does. The association messages are handled by the firmware; the key negotiation for 802.1x and WPA are done as raw frames, not management frames. This successfully allows it to associate to my home networks whereas it didn't work beforehand. Tested: * RTL8712, cut 3, STA mode TODO: * The firmware does send a join response with a status code; that should be logged in a more obvious way to assist with debugging. Ie, the firmware is the thing that is saying "couldn't join, sorry!", not net80211.
2015-09-13 19:17:26 +00:00
ic->ic_send_mgmt = rsu_send_mgmt;
Add 11n and QoS methods. The firmware takes care of ampdu tx/rx (except for RX reordering, grr), QoS/WME and other bits/pieces. So they're stubs, just in case.
2015-09-17 03:13:01 +00:00
ic->ic_update_chw = rsu_update_chw;
ic->ic_ampdu_enable = rsu_ampdu_enable;
ic->ic_wme.wme_update = rsu_wme_update;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ieee80211_radiotap_attach(ic, &sc->sc_txtap.wt_ihdr,
sizeof(sc->sc_txtap), RSU_TX_RADIOTAP_PRESENT,
&sc->sc_rxtap.wr_ihdr, sizeof(sc->sc_rxtap),
RSU_RX_RADIOTAP_PRESENT);
if (bootverbose)
ieee80211_announce(ic);
return (0);
Cleanup the allocations when the attachment fails.
2013-07-31 06:05:34 +00:00
fail_rom:
usbd_transfer_unsetup(sc->sc_xfer, RSU_N_TRANSFER);
fail_usb:
mtx_destroy(&sc->sc_mtx);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
return (ENXIO);
}
static int
rsu_detach(device_t self)
{
struct rsu_softc *sc = device_get_softc(self);
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;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +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
rsu_stop(sc);
Free the TX/RX list buffers /before/ tearing down net80211 state. Otherwise buffers in the RX queue get freed with their parent vap being gone and you end up with a juicy kernel panic.
2015-09-26 22:20:30 +00:00
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
usbd_transfer_unsetup(sc->sc_xfer, RSU_N_TRANSFER);
Drain the mbuf queue upon rsu_stop(). Correctly (I hope!) remove net80211 references before doing so. Just doing a dumb mbufq drain isn't enough. If enough traffic occurs and the mbuf queue fills up then transmit stalls (which I'm not fixing in this commit!) but then the mbuf queue stays full until the driver is removed. There's also the net80211 node refcounting leak. This just ensures that during rsu_stop and detach the mbuf queue is purged (and references!) so the queue-full situation can be recovered from.
2015-09-21 02:12:01 +00:00
Free the TX/RX list buffers /before/ tearing down net80211 state. Otherwise buffers in the RX queue get freed with their parent vap being gone and you end up with a juicy kernel panic.
2015-09-26 22:20:30 +00:00
/*
* Free buffers /before/ we detach from net80211, else node
* references to destroyed vaps will lead to a panic.
*/
/* Free Tx/Rx buffers. */
RSU_LOCK(sc);
rsu_free_tx_list(sc);
rsu_free_rx_list(sc);
RSU_UNLOCK(sc);
Drain the mbuf queue upon rsu_stop(). Correctly (I hope!) remove net80211 references before doing so. Just doing a dumb mbufq drain isn't enough. If enough traffic occurs and the mbuf queue fills up then transmit stalls (which I'm not fixing in this commit!) but then the mbuf queue stays full until the driver is removed. There's also the net80211 node refcounting leak. This just ensures that during rsu_stop and detach the mbuf queue is purged (and references!) so the queue-full situation can be recovered from.
2015-09-21 02:12:01 +00:00
/* Frames are freed; detach from net80211 */
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ieee80211_ifdetach(ic);
taskqueue_drain_timeout(taskqueue_thread, &sc->calib_task);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
taskqueue_drain(taskqueue_thread, &sc->del_key_task);
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
taskqueue_drain(taskqueue_thread, &sc->tx_task);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
RSU_DELKEY_BMAP_LOCK_DESTROY(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
mtx_destroy(&sc->sc_mtx);
return (0);
}
static usb_error_t
rsu_do_request(struct rsu_softc *sc, struct usb_device_request *req,
void *data)
{
usb_error_t err;
int ntries = 10;
RSU_ASSERT_LOCKED(sc);
while (ntries--) {
err = usbd_do_request_flags(sc->sc_udev, &sc->sc_mtx,
req, data, 0, NULL, 250 /* ms */);
Make sure detach code is executed in all cases. This fixes a panic when debugging is enabled. Reported by: Idwer Vollering <vidwer@gmail.com> MFC after: 3 days
2014-05-20 12:22:53 +00:00
if (err == 0 || err == USB_ERR_NOT_CONFIGURED)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
break;
rsu: use bitmap for all debug messages. - Replace all remaining DPRINTF(N)'s with RSU_DPRINTF. - Add new RSU_DEBUG_USB flag to track error codes returned by usbd_do_request_flags(). - Improve few messages.
2016-12-10 20:19:57 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_USB,
"Control request failed, %s (retries left: %d)\n",
usbd_errstr(err), ntries);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 10);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
return (err);
}
static struct ieee80211vap *
rsu_vap_create(struct ieee80211com *ic, const char name[IFNAMSIZ], int unit,
enum ieee80211_opmode opmode, int flags,
const uint8_t bssid[IEEE80211_ADDR_LEN],
const uint8_t mac[IEEE80211_ADDR_LEN])
{
rsu: fix and enable Rx TCP checksum offloading. Tested with Asus USB-N10, STA mode.
2016-12-11 17:15:25 +00:00
struct rsu_softc *sc = ic->ic_softc;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct rsu_vap *uvp;
struct ieee80211vap *vap;
rsu: fix and enable Rx TCP checksum offloading. Tested with Asus USB-N10, STA mode.
2016-12-11 17:15:25 +00:00
struct ifnet *ifp;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (!TAILQ_EMPTY(&ic->ic_vaps)) /* only one at a time */
return (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
uvp = malloc(sizeof(struct rsu_vap), M_80211_VAP, M_WAITOK | M_ZERO);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
vap = &uvp->vap;
Check for "ieee80211_vap_setup()" failure in all USB WLAN drivers. MFC after: 1 week
2013-11-06 12:57:01 +00:00
if (ieee80211_vap_setup(ic, vap, name, unit, opmode,
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
flags, bssid) != 0) {
Check for "ieee80211_vap_setup()" failure in all USB WLAN drivers. MFC after: 1 week
2013-11-06 12:57:01 +00:00
/* out of memory */
free(uvp, M_80211_VAP);
return (NULL);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu: fix and enable Rx TCP checksum offloading. Tested with Asus USB-N10, STA mode.
2016-12-11 17:15:25 +00:00
ifp = vap->iv_ifp;
ifp->if_capabilities = IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6;
RSU_LOCK(sc);
if (sc->sc_rx_checksum_enable)
ifp->if_capenable |= IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6;
RSU_UNLOCK(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* override state transition machine */
uvp->newstate = vap->iv_newstate;
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
if (opmode == IEEE80211_M_MONITOR)
vap->iv_newstate = rsu_monitor_newstate;
else
vap->iv_newstate = rsu_newstate;
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
vap->iv_key_alloc = rsu_key_alloc;
vap->iv_key_set = rsu_key_set;
vap->iv_key_delete = rsu_key_delete;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
Set AMPDU density/size parameters during vap creation. Inspired from: Linux r92su
2015-09-18 05:59:15 +00:00
/* Limits from the r92su driver */
vap->iv_ampdu_density = IEEE80211_HTCAP_MPDUDENSITY_16;
vap->iv_ampdu_rxmax = IEEE80211_HTCAP_MAXRXAMPDU_32K;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* complete setup */
ieee80211_vap_attach(vap, ieee80211_media_change,
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_media_status, mac);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ic->ic_opmode = opmode;
return (vap);
}
static void
rsu_vap_delete(struct ieee80211vap *vap)
{
struct rsu_vap *uvp = RSU_VAP(vap);
ieee80211_vap_detach(vap);
free(uvp, M_80211_VAP);
}
static void
rsu_scan_start(struct ieee80211com *ic)
{
Further conversions from ifp->if_softc -> ic_softc.
2015-08-20 05:13:54 +00:00
struct rsu_softc *sc = ic->ic_softc;
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
struct ieee80211_scan_state *ss = ic->ic_scan;
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
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;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Scanning is done by the firmware. */
RSU_LOCK(sc);
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
sc->sc_active_scan = !!(ss->ss_flags & IEEE80211_SCAN_ACTIVE);
/* XXX TODO: force awake if in network-sleep? */
error = rsu_site_survey(sc, ss->ss_nssid > 0 ? &ss->ss_ssid[0] : NULL);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_UNLOCK(sc);
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
if (error != 0) {
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
device_printf(sc->sc_dev,
"could not send site survey command\n");
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
ieee80211_cancel_scan(vap);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
static void
rsu_scan_end(struct ieee80211com *ic)
{
/* Nothing to do here. */
}
urtwn, rtwn, rsu: switch to ieee80211_add_channel_list_2ghz(). - Use device's channel list instead of default one (from ieee80211_init_channels()); adds 12 - 14 2GHz channels. - Add ic_getradiocaps() method.
2016-05-26 16:39:11 +00:00
static void
rsu_getradiocaps(struct ieee80211com *ic,
int maxchans, int *nchans, struct ieee80211_channel chans[])
{
struct rsu_softc *sc = ic->ic_softc;
uint8_t bands[IEEE80211_MODE_BYTES];
/* Set supported .11b and .11g rates. */
memset(bands, 0, sizeof(bands));
setbit(bands, IEEE80211_MODE_11B);
setbit(bands, IEEE80211_MODE_11G);
if (sc->sc_ht)
setbit(bands, IEEE80211_MODE_11NG);
ieee80211_add_channel_list_2ghz(chans, maxchans, nchans,
rsu: restore 40Mhz channel support. MFC after: 4 days.
2017-01-02 16:58:55 +00:00
rsu_chan_2ghz, nitems(rsu_chan_2ghz), bands,
(ic->ic_htcaps & IEEE80211_HTCAP_CHWIDTH40) != 0);
urtwn, rtwn, rsu: switch to ieee80211_add_channel_list_2ghz(). - Use device's channel list instead of default one (from ieee80211_init_channels()); adds 12 - 14 2GHz channels. - Add ic_getradiocaps() method.
2016-05-26 16:39:11 +00:00
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static void
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
rsu_set_channel(struct ieee80211com *ic)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
struct rsu_softc *sc = ic->ic_softc;
/*
* 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) {
struct r92s_set_channel cmd;
int error;
cmd.channel = IEEE80211_CHAN2IEEE(ic->ic_curchan);
RSU_LOCK(sc);
error = rsu_fw_cmd(sc, R92S_CMD_SET_CHANNEL, &cmd,
sizeof(cmd));
if (error != 0) {
device_printf(sc->sc_dev,
"%s: error %d setting channel\n", __func__,
error);
}
RSU_UNLOCK(sc);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
static void
rsu_scan_curchan(struct ieee80211_scan_state *ss, unsigned long maxdwell)
{
/* Scan is done in rsu_scan_start(). */
}
/**
* Called by the net80211 framework to indicate
* the minimum dwell time has been met, terminate the scan.
* We don't actually terminate the scan as the firmware will notify
* us when it's finished and we have no way to interrupt it.
*/
static void
rsu_scan_mindwell(struct ieee80211_scan_state *ss)
{
/* NB: don't try to abort scan; wait for firmware to finish */
}
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
static void
rsu_update_promisc(struct ieee80211com *ic)
{
struct rsu_softc *sc = ic->ic_softc;
RSU_LOCK(sc);
if (sc->sc_running)
rsu_rxfilter_refresh(sc);
RSU_UNLOCK(sc);
}
rsu: add support for hardware multicast filter setup. The algorithm is the same as in rtwn(4). Tested with Asus USB-N10 (STA) + RTL8188CUS (AP).
2016-11-27 18:06:03 +00:00
/*
* The same as rtwn_get_multi_pos() / rtwn_set_multi().
*/
static uint8_t
rsu_get_multi_pos(const uint8_t maddr[])
{
uint64_t mask = 0x00004d101df481b4;
uint8_t pos = 0x27; /* initial value */
int i, j;
for (i = 0; i < IEEE80211_ADDR_LEN; i++)
for (j = (i == 0) ? 1 : 0; j < 8; j++)
if ((maddr[i] >> j) & 1)
pos ^= (mask >> (i * 8 + j - 1));
pos &= 0x3f;
return (pos);
}
static void
rsu_set_multi(struct rsu_softc *sc)
{
struct ieee80211com *ic = &sc->sc_ic;
uint32_t mfilt[2];
RSU_ASSERT_LOCKED(sc);
/* general structure was copied from ath(4). */
if (ic->ic_allmulti == 0) {
struct ieee80211vap *vap;
struct ifnet *ifp;
struct ifmultiaddr *ifma;
/*
* Merge multicast addresses to form the hardware filter.
*/
mfilt[0] = mfilt[1] = 0;
TAILQ_FOREACH(vap, &ic->ic_vaps, iv_next) {
ifp = vap->iv_ifp;
if_maddr_rlock(ifp);
TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
caddr_t dl;
uint8_t pos;
dl = LLADDR((struct sockaddr_dl *)
ifma->ifma_addr);
pos = rsu_get_multi_pos(dl);
mfilt[pos / 32] |= (1 << (pos % 32));
}
if_maddr_runlock(ifp);
}
} else
mfilt[0] = mfilt[1] = ~0;
rsu_write_4(sc, R92S_MAR + 0, mfilt[0]);
rsu_write_4(sc, R92S_MAR + 4, mfilt[1]);
RSU_DPRINTF(sc, RSU_DEBUG_STATE, "%s: MC filter %08x:%08x\n",
__func__, mfilt[0], mfilt[1]);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +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
rsu_update_mcast(struct ieee80211com *ic)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
rsu: add support for hardware multicast filter setup. The algorithm is the same as in rtwn(4). Tested with Asus USB-N10 (STA) + RTL8188CUS (AP).
2016-11-27 18:06:03 +00:00
struct rsu_softc *sc = ic->ic_softc;
RSU_LOCK(sc);
if (sc->sc_running)
rsu_set_multi(sc);
RSU_UNLOCK(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
static int
rsu_alloc_list(struct rsu_softc *sc, struct rsu_data data[],
int ndata, int maxsz)
{
int i, error;
for (i = 0; i < ndata; i++) {
struct rsu_data *dp = &data[i];
dp->sc = sc;
dp->m = NULL;
dp->buf = malloc(maxsz, M_USBDEV, M_NOWAIT);
if (dp->buf == NULL) {
device_printf(sc->sc_dev,
"could not allocate buffer\n");
error = ENOMEM;
goto fail;
}
dp->ni = NULL;
}
return (0);
fail:
rsu_free_list(sc, data, ndata);
return (error);
}
static int
rsu_alloc_rx_list(struct rsu_softc *sc)
{
int error, i;
error = rsu_alloc_list(sc, sc->sc_rx, RSU_RX_LIST_COUNT,
RSU_RXBUFSZ);
if (error != 0)
return (error);
STAILQ_INIT(&sc->sc_rx_active);
STAILQ_INIT(&sc->sc_rx_inactive);
for (i = 0; i < RSU_RX_LIST_COUNT; i++)
STAILQ_INSERT_HEAD(&sc->sc_rx_inactive, &sc->sc_rx[i], next);
return (0);
}
static int
rsu_alloc_tx_list(struct rsu_softc *sc)
{
int error, i;
error = rsu_alloc_list(sc, sc->sc_tx, RSU_TX_LIST_COUNT,
RSU_TXBUFSZ);
if (error != 0)
return (error);
STAILQ_INIT(&sc->sc_tx_inactive);
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
for (i = 0; i != RSU_N_TRANSFER; i++) {
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
STAILQ_INIT(&sc->sc_tx_active[i]);
STAILQ_INIT(&sc->sc_tx_pending[i]);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
for (i = 0; i < RSU_TX_LIST_COUNT; i++) {
STAILQ_INSERT_HEAD(&sc->sc_tx_inactive, &sc->sc_tx[i], next);
}
return (0);
}
static void
rsu_free_tx_list(struct rsu_softc *sc)
{
- Stop transfers when RSU init fails. - Make sure TX/RX lists don't leak and are only allocated once. - Fix off-by one transfer index computation. - Give firmware loading more time. MFC after: 3 days
2014-05-22 06:28:09 +00:00
int i;
/* prevent further allocations from TX list(s) */
STAILQ_INIT(&sc->sc_tx_inactive);
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
for (i = 0; i != RSU_N_TRANSFER; i++) {
- Stop transfers when RSU init fails. - Make sure TX/RX lists don't leak and are only allocated once. - Fix off-by one transfer index computation. - Give firmware loading more time. MFC after: 3 days
2014-05-22 06:28:09 +00:00
STAILQ_INIT(&sc->sc_tx_active[i]);
STAILQ_INIT(&sc->sc_tx_pending[i]);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu_free_list(sc, sc->sc_tx, RSU_TX_LIST_COUNT);
}
static void
rsu_free_rx_list(struct rsu_softc *sc)
{
- Stop transfers when RSU init fails. - Make sure TX/RX lists don't leak and are only allocated once. - Fix off-by one transfer index computation. - Give firmware loading more time. MFC after: 3 days
2014-05-22 06:28:09 +00:00
/* prevent further allocations from RX list(s) */
STAILQ_INIT(&sc->sc_rx_inactive);
STAILQ_INIT(&sc->sc_rx_active);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu_free_list(sc, sc->sc_rx, RSU_RX_LIST_COUNT);
}
static void
rsu_free_list(struct rsu_softc *sc, struct rsu_data data[], int ndata)
{
int i;
for (i = 0; i < ndata; i++) {
struct rsu_data *dp = &data[i];
if (dp->buf != NULL) {
free(dp->buf, M_USBDEV);
dp->buf = NULL;
}
if (dp->ni != NULL) {
ieee80211_free_node(dp->ni);
dp->ni = NULL;
}
}
}
static struct rsu_data *
_rsu_getbuf(struct rsu_softc *sc)
{
struct rsu_data *bf;
bf = STAILQ_FIRST(&sc->sc_tx_inactive);
if (bf != NULL)
STAILQ_REMOVE_HEAD(&sc->sc_tx_inactive, next);
else
bf = NULL;
Add in a temporary (hah!) workaround for net80211 scanning versus NIC requirements. Don't start the opmode and join path until a pending survey is finished. This seems to reliably fix things. Ideally I'd just finish off the net80211 pluggable scan stuff and implement the methods here so if_rsu can just drive the scan machinery. However, that's a .. later thing. Whilst here, remove the getbuf debugging; it's okay to run out of transmit buffers under load; it however isn't okay to not be able to send commands. I'll fix that later.
2015-09-18 07:55:33 +00:00
return (bf);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
static struct rsu_data *
rsu_getbuf(struct rsu_softc *sc)
{
struct rsu_data *bf;
RSU_ASSERT_LOCKED(sc);
bf = _rsu_getbuf(sc);
Drain the mbuf queue upon rsu_stop(). Correctly (I hope!) remove net80211 references before doing so. Just doing a dumb mbufq drain isn't enough. If enough traffic occurs and the mbuf queue fills up then transmit stalls (which I'm not fixing in this commit!) but then the mbuf queue stays full until the driver is removed. There's also the net80211 node refcounting leak. This just ensures that during rsu_stop and detach the mbuf queue is purged (and references!) so the queue-full situation can be recovered from.
2015-09-21 02:12:01 +00:00
if (bf == NULL) {
RSU_DPRINTF(sc, RSU_DEBUG_TX, "%s: no buffers\n", __func__);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
return (bf);
}
Refactor out the tx buffer free code into a routine, rsu_freebuf(). This makes it easier to add more transmit buffers, have different buffer pools for things, etc.
2015-09-18 07:26:34 +00:00
static void
rsu_freebuf(struct rsu_softc *sc, struct rsu_data *bf)
{
RSU_ASSERT_LOCKED(sc);
STAILQ_INSERT_TAIL(&sc->sc_tx_inactive, bf, next);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static int
rsu_write_region_1(struct rsu_softc *sc, uint16_t addr, uint8_t *buf,
int len)
{
usb_device_request_t req;
req.bmRequestType = UT_WRITE_VENDOR_DEVICE;
req.bRequest = R92S_REQ_REGS;
USETW(req.wValue, addr);
USETW(req.wIndex, 0);
USETW(req.wLength, len);
return (rsu_do_request(sc, &req, buf));
}
static void
rsu_write_1(struct rsu_softc *sc, uint16_t addr, uint8_t val)
{
rsu_write_region_1(sc, addr, &val, 1);
}
static void
rsu_write_2(struct rsu_softc *sc, uint16_t addr, uint16_t val)
{
val = htole16(val);
rsu_write_region_1(sc, addr, (uint8_t *)&val, 2);
}
static void
rsu_write_4(struct rsu_softc *sc, uint16_t addr, uint32_t val)
{
val = htole32(val);
rsu_write_region_1(sc, addr, (uint8_t *)&val, 4);
}
static int
rsu_read_region_1(struct rsu_softc *sc, uint16_t addr, uint8_t *buf,
int len)
{
usb_device_request_t req;
req.bmRequestType = UT_READ_VENDOR_DEVICE;
req.bRequest = R92S_REQ_REGS;
USETW(req.wValue, addr);
USETW(req.wIndex, 0);
USETW(req.wLength, len);
return (rsu_do_request(sc, &req, buf));
}
static uint8_t
rsu_read_1(struct rsu_softc *sc, uint16_t addr)
{
uint8_t val;
if (rsu_read_region_1(sc, addr, &val, 1) != 0)
return (0xff);
return (val);
}
static uint16_t
rsu_read_2(struct rsu_softc *sc, uint16_t addr)
{
uint16_t val;
if (rsu_read_region_1(sc, addr, (uint8_t *)&val, 2) != 0)
return (0xffff);
return (le16toh(val));
}
static uint32_t
rsu_read_4(struct rsu_softc *sc, uint16_t addr)
{
uint32_t val;
if (rsu_read_region_1(sc, addr, (uint8_t *)&val, 4) != 0)
return (0xffffffff);
return (le32toh(val));
}
static int
rsu_fw_iocmd(struct rsu_softc *sc, uint32_t iocmd)
{
int ntries;
rsu_write_4(sc, R92S_IOCMD_CTRL, iocmd);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
for (ntries = 0; ntries < 50; ntries++) {
if (rsu_read_4(sc, R92S_IOCMD_CTRL) == 0)
return (0);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
return (ETIMEDOUT);
}
static uint8_t
rsu_efuse_read_1(struct rsu_softc *sc, uint16_t addr)
{
uint32_t reg;
int ntries;
reg = rsu_read_4(sc, R92S_EFUSE_CTRL);
reg = RW(reg, R92S_EFUSE_CTRL_ADDR, addr);
reg &= ~R92S_EFUSE_CTRL_VALID;
rsu_write_4(sc, R92S_EFUSE_CTRL, reg);
/* Wait for read operation to complete. */
for (ntries = 0; ntries < 100; ntries++) {
reg = rsu_read_4(sc, R92S_EFUSE_CTRL);
if (reg & R92S_EFUSE_CTRL_VALID)
return (MS(reg, R92S_EFUSE_CTRL_DATA));
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
device_printf(sc->sc_dev,
"could not read efuse byte at address 0x%x\n", addr);
return (0xff);
}
static int
rsu_read_rom(struct rsu_softc *sc)
{
uint8_t *rom = sc->rom;
uint16_t addr = 0;
uint32_t reg;
uint8_t off, msk;
int i;
/* Make sure that ROM type is eFuse and that autoload succeeded. */
reg = rsu_read_1(sc, R92S_EE_9346CR);
if ((reg & (R92S_9356SEL | R92S_EEPROM_EN)) != R92S_EEPROM_EN)
return (EIO);
/* Turn on 2.5V to prevent eFuse leakage. */
reg = rsu_read_1(sc, R92S_EFUSE_TEST + 3);
rsu_write_1(sc, R92S_EFUSE_TEST + 3, reg | 0x80);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu_write_1(sc, R92S_EFUSE_TEST + 3, reg & ~0x80);
/* Read full ROM image. */
memset(&sc->rom, 0xff, sizeof(sc->rom));
while (addr < 512) {
reg = rsu_efuse_read_1(sc, addr);
if (reg == 0xff)
break;
addr++;
off = reg >> 4;
msk = reg & 0xf;
for (i = 0; i < 4; i++) {
if (msk & (1 << i))
continue;
rom[off * 8 + i * 2 + 0] =
rsu_efuse_read_1(sc, addr);
addr++;
rom[off * 8 + i * 2 + 1] =
rsu_efuse_read_1(sc, addr);
addr++;
}
}
#ifdef USB_DEBUG
rsu: use bitmap for all debug messages. - Replace all remaining DPRINTF(N)'s with RSU_DPRINTF. - Add new RSU_DEBUG_USB flag to track error codes returned by usbd_do_request_flags(). - Improve few messages.
2016-12-10 20:19:57 +00:00
if (rsu_debug & RSU_DEBUG_RESET) {
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Dump ROM content. */
printf("\n");
for (i = 0; i < sizeof(sc->rom); i++)
printf("%02x:", rom[i]);
printf("\n");
}
#endif
return (0);
}
static int
rsu_fw_cmd(struct rsu_softc *sc, uint8_t code, void *buf, int len)
{
Use the H2C endpoint for sending firmware commands, rather than the voice data queues. This is similar to the openbsd and rtlwifi/r92su drivers. Note: this driver still assumes it's a 4-endpoint device; I'll enforce that in a follow-up commit.
2015-09-17 03:19:09 +00:00
const uint8_t which = RSU_H2C_ENDPOINT;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct rsu_data *data;
struct r92s_tx_desc *txd;
struct r92s_fw_cmd_hdr *cmd;
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
int cmdsz;
int xferlen;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
Add in a temporary (hah!) workaround for net80211 scanning versus NIC requirements. Don't start the opmode and join path until a pending survey is finished. This seems to reliably fix things. Ideally I'd just finish off the net80211 pluggable scan stuff and implement the methods here so if_rsu can just drive the scan machinery. However, that's a .. later thing. Whilst here, remove the getbuf debugging; it's okay to run out of transmit buffers under load; it however isn't okay to not be able to send commands. I'll fix that later.
2015-09-18 07:55:33 +00:00
RSU_ASSERT_LOCKED(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
data = rsu_getbuf(sc);
if (data == NULL)
return (ENOMEM);
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
/* Blank the entire payload, just to be safe */
memset(data->buf, '\0', RSU_TXBUFSZ);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Round-up command length to a multiple of 8 bytes. */
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
/* XXX TODO: is this required? */
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
cmdsz = (len + 7) & ~7;
xferlen = sizeof(*txd) + sizeof(*cmd) + cmdsz;
KASSERT(xferlen <= RSU_TXBUFSZ, ("%s: invalid length", __func__));
memset(data->buf, 0, xferlen);
/* Setup Tx descriptor. */
txd = (struct r92s_tx_desc *)data->buf;
txd->txdw0 = htole32(
SM(R92S_TXDW0_OFFSET, sizeof(*txd)) |
SM(R92S_TXDW0_PKTLEN, sizeof(*cmd) + cmdsz) |
R92S_TXDW0_OWN | R92S_TXDW0_FSG | R92S_TXDW0_LSG);
txd->txdw1 = htole32(SM(R92S_TXDW1_QSEL, R92S_TXDW1_QSEL_H2C));
/* Setup command header. */
cmd = (struct r92s_fw_cmd_hdr *)&txd[1];
cmd->len = htole16(cmdsz);
cmd->code = code;
cmd->seq = sc->cmd_seq;
sc->cmd_seq = (sc->cmd_seq + 1) & 0x7f;
/* Copy command payload. */
memcpy(&cmd[1], buf, len);
* fiddle with some more of the debugging output * yes, when a "sta disconnect" message comes through we should, like, disconnect things. We're not currently generating beacon miss messages, and net80211 isn't disconnecting things via software beacon miss receive. Tested: * RTL8712, cut 3, STA mode
2015-09-13 04:12:51 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_TX | RSU_DEBUG_FWCMD,
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
"%s: Tx cmd code=0x%x len=0x%x\n",
__func__, code, cmdsz);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
data->buflen = xferlen;
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
STAILQ_INSERT_TAIL(&sc->sc_tx_pending[which], data, next);
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
usbd_transfer_start(sc->sc_xfer[which]);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
return (0);
}
/* ARGSUSED */
static void
rsu_calib_task(void *arg, int pending __unused)
{
struct rsu_softc *sc = arg;
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
#ifdef notyet
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
uint32_t reg;
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
#endif
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_CALIB, "%s: running calibration task\n",
__func__);
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_LOCK(sc);
#ifdef notyet
/* Read WPS PBC status. */
rsu_write_1(sc, R92S_MAC_PINMUX_CTRL,
R92S_GPIOMUX_EN | SM(R92S_GPIOSEL_GPIO, R92S_GPIOSEL_GPIO_JTAG));
rsu_write_1(sc, R92S_GPIO_IO_SEL,
rsu_read_1(sc, R92S_GPIO_IO_SEL) & ~R92S_GPIO_WPS);
reg = rsu_read_1(sc, R92S_GPIO_CTRL);
if (reg != 0xff && (reg & R92S_GPIO_WPS))
rsu: use bitmap for all debug messages. - Replace all remaining DPRINTF(N)'s with RSU_DPRINTF. - Add new RSU_DEBUG_USB flag to track error codes returned by usbd_do_request_flags(). - Improve few messages.
2016-12-10 20:19:57 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_CALIB, "WPS PBC is pushed\n");
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
#endif
/* Read current signal level. */
if (rsu_fw_iocmd(sc, 0xf4000001) == 0) {
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
sc->sc_currssi = rsu_read_4(sc, R92S_IOCMD_DATA);
RSU_DPRINTF(sc, RSU_DEBUG_CALIB, "%s: RSSI=%d (%d)\n",
__func__, sc->sc_currssi,
rsu_hwrssi_to_rssi(sc, sc->sc_currssi));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
if (sc->sc_calibrating)
taskqueue_enqueue_timeout(taskqueue_thread, &sc->calib_task, hz);
RSU_UNLOCK(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
static void
rsu_tx_task(void *arg, int pending __unused)
{
struct rsu_softc *sc = arg;
RSU_LOCK(sc);
_rsu_start(sc);
RSU_UNLOCK(sc);
}
Send a power command to the firmware to shut down the radio as well during rsu_stop().
2015-09-22 05:48:51 +00:00
#define RSU_PWR_UNKNOWN 0x0
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
#define RSU_PWR_ACTIVE 0x1
#define RSU_PWR_OFF 0x2
#define RSU_PWR_SLEEP 0x3
/*
* Set the current power state.
*
* The rtlwifi code doesn't do this so aggressively; it
* waits for an idle period after association with
* no traffic before doing this.
*
* For now - it's on in all states except RUN, and
* in RUN it'll transition to allow sleep.
*/
struct r92s_pwr_cmd {
uint8_t mode;
uint8_t smart_ps;
uint8_t bcn_pass_time;
};
static int
rsu_set_fw_power_state(struct rsu_softc *sc, int state)
{
struct r92s_set_pwr_mode cmd;
//struct r92s_pwr_cmd cmd;
int error;
RSU_ASSERT_LOCKED(sc);
Send a power command to the firmware to shut down the radio as well during rsu_stop().
2015-09-22 05:48:51 +00:00
/* only change state if required */
if (sc->sc_curpwrstate == state)
return (0);
memset(&cmd, 0, sizeof(cmd));
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
switch (state) {
case RSU_PWR_ACTIVE:
/* Force the hardware awake */
rsu_write_1(sc, R92S_USB_HRPWM,
R92S_USB_HRPWM_PS_ST_ACTIVE | R92S_USB_HRPWM_PS_ALL_ON);
cmd.mode = R92S_PS_MODE_ACTIVE;
break;
case RSU_PWR_SLEEP:
cmd.mode = R92S_PS_MODE_DTIM; /* XXX configurable? */
cmd.smart_ps = 1; /* XXX 2 if doing p2p */
cmd.bcn_pass_time = 5; /* in 100mS usb.c, linux/rtlwifi */
break;
Send a power command to the firmware to shut down the radio as well during rsu_stop().
2015-09-22 05:48:51 +00:00
case RSU_PWR_OFF:
cmd.mode = R92S_PS_MODE_RADIOOFF;
break;
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
default:
device_printf(sc->sc_dev, "%s: unknown ps mode (%d)\n",
__func__,
state);
return (ENXIO);
}
RSU_DPRINTF(sc, RSU_DEBUG_RESET,
"%s: setting ps mode to %d (mode %d)\n",
__func__, state, cmd.mode);
error = rsu_fw_cmd(sc, R92S_CMD_SET_PWR_MODE, &cmd, sizeof(cmd));
Send a power command to the firmware to shut down the radio as well during rsu_stop().
2015-09-22 05:48:51 +00:00
if (error == 0)
sc->sc_curpwrstate = state;
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
return (error);
}
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
static void
rsu_set_led(struct rsu_softc *sc, int on)
{
rsu_write_1(sc, R92S_LEDCFG,
(rsu_read_1(sc, R92S_LEDCFG) & 0xf0) | (!on << 3));
}
static int
rsu_monitor_newstate(struct ieee80211vap *vap, enum ieee80211_state nstate,
int arg)
{
struct ieee80211com *ic = vap->iv_ic;
struct rsu_softc *sc = ic->ic_softc;
struct rsu_vap *uvp = RSU_VAP(vap);
if (vap->iv_state != nstate) {
IEEE80211_UNLOCK(ic);
RSU_LOCK(sc);
switch (nstate) {
case IEEE80211_S_INIT:
sc->sc_vap_is_running = 0;
rsu_set_led(sc, 0);
break;
case IEEE80211_S_RUN:
sc->sc_vap_is_running = 1;
rsu_set_led(sc, 1);
break;
default:
/* NOTREACHED */
break;
}
rsu_rxfilter_refresh(sc);
RSU_UNLOCK(sc);
IEEE80211_LOCK(ic);
}
return (uvp->newstate(vap, nstate, arg));
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static int
rsu_newstate(struct ieee80211vap *vap, enum ieee80211_state nstate, int arg)
{
struct rsu_vap *uvp = RSU_VAP(vap);
struct ieee80211com *ic = vap->iv_ic;
Further conversions from ifp->if_softc -> ic_softc.
2015-08-20 05:13:54 +00:00
struct rsu_softc *sc = ic->ic_softc;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct ieee80211_node *ni;
struct ieee80211_rateset *rs;
enum ieee80211_state ostate;
int error, startcal = 0;
ostate = vap->iv_state;
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_STATE, "%s: %s -> %s\n",
__func__,
ieee80211_state_name[ostate],
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ieee80211_state_name[nstate]);
IEEE80211_UNLOCK(ic);
if (ostate == IEEE80211_S_RUN) {
RSU_LOCK(sc);
/* Stop calibration. */
sc->sc_calibrating = 0;
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
/* Pause Tx for AC queues. */
rsu_write_1(sc, R92S_TXPAUSE, R92S_TXPAUSE_AC);
usb_pause_mtx(&sc->sc_mtx, USB_MS_TO_TICKS(10));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_UNLOCK(sc);
taskqueue_drain_timeout(taskqueue_thread, &sc->calib_task);
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
taskqueue_drain(taskqueue_thread, &sc->tx_task);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_LOCK(sc);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
/* Disassociate from our current BSS. */
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu_disconnect(sc);
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
usb_pause_mtx(&sc->sc_mtx, USB_MS_TO_TICKS(10));
/* Refresh Rx filter (may be modified by firmware). */
sc->sc_vap_is_running = 0;
rsu_rxfilter_refresh(sc);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
/* Reinstall static keys. */
if (sc->sc_running)
rsu_reinit_static_keys(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
} else
RSU_LOCK(sc);
switch (nstate) {
case IEEE80211_S_INIT:
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
(void) rsu_set_fw_power_state(sc, RSU_PWR_ACTIVE);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
break;
case IEEE80211_S_AUTH:
ni = ieee80211_ref_node(vap->iv_bss);
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
(void) rsu_set_fw_power_state(sc, RSU_PWR_ACTIVE);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
error = rsu_join_bss(sc, ni);
ieee80211_free_node(ni);
if (error != 0) {
device_printf(sc->sc_dev,
"could not send join command\n");
}
break;
case IEEE80211_S_RUN:
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
/* Flush all AC queues. */
rsu_write_1(sc, R92S_TXPAUSE, 0);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ni = ieee80211_ref_node(vap->iv_bss);
rs = &ni->ni_rates;
/* Indicate highest supported rate. */
ni->ni_txrate = rs->rs_rates[rs->rs_nrates - 1];
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
(void) rsu_set_fw_power_state(sc, RSU_PWR_SLEEP);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ieee80211_free_node(ni);
startcal = 1;
break;
default:
break;
}
rsu: do not restart calibration task when going out of RUN state. Clear 'sc_calibrating' flag and stop calibration task when interface is not associated; this fixes possible panic after detach. Reported and tested by: hselasky Reviewed by: adrian MFC after: 6 days
2016-09-25 19:13:07 +00:00
if (startcal != 0) {
sc->sc_calibrating = 1;
/* Start periodic calibration. */
taskqueue_enqueue_timeout(taskqueue_thread, &sc->calib_task,
hz);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_UNLOCK(sc);
IEEE80211_LOCK(ic);
return (uvp->newstate(vap, nstate, arg));
}
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
static int
rsu_key_alloc(struct ieee80211vap *vap, struct ieee80211_key *k,
ieee80211_keyix *keyix, ieee80211_keyix *rxkeyix)
{
struct rsu_softc *sc = vap->iv_ic->ic_softc;
int is_checked = 0;
if (&vap->iv_nw_keys[0] <= k &&
k < &vap->iv_nw_keys[IEEE80211_WEP_NKID]) {
net80211 + drivers: convert to ieee80211_crypto_get_key_wepidx(). Proposed by: adrian
2016-12-07 22:16:07 +00:00
*keyix = ieee80211_crypto_get_key_wepidx(vap, k);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
} else {
if (vap->iv_opmode != IEEE80211_M_STA) {
*keyix = 0;
/* TODO: obtain keyix from node id */
is_checked = 1;
k->wk_flags |= IEEE80211_KEY_SWCRYPT;
} else
*keyix = R92S_MACID_BSS;
}
if (!is_checked) {
RSU_LOCK(sc);
if (isset(sc->keys_bmap, *keyix)) {
device_printf(sc->sc_dev,
"%s: key slot %d is already used!\n",
__func__, *keyix);
RSU_UNLOCK(sc);
return (0);
}
setbit(sc->keys_bmap, *keyix);
RSU_UNLOCK(sc);
}
*rxkeyix = *keyix;
return (1);
}
static int
rsu_process_key(struct ieee80211vap *vap, const struct ieee80211_key *k,
int set)
{
struct rsu_softc *sc = vap->iv_ic->ic_softc;
int ret;
if (k->wk_flags & IEEE80211_KEY_SWCRYPT) {
/* Not for us. */
return (1);
}
/* Handle group keys. */
if (&vap->iv_nw_keys[0] <= k &&
k < &vap->iv_nw_keys[IEEE80211_WEP_NKID]) {
KASSERT(k->wk_keyix < nitems(sc->group_keys),
rsu: fix printf format specifiers.
2016-12-06 06:12:01 +00:00
("keyix %u > %zu\n", k->wk_keyix, nitems(sc->group_keys)));
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
RSU_LOCK(sc);
sc->group_keys[k->wk_keyix] = (set ? k : NULL);
if (!sc->sc_running) {
/* Static keys will be set during device startup. */
RSU_UNLOCK(sc);
return (1);
}
if (set)
ret = rsu_set_key_group(sc, k);
else
ret = rsu_delete_key(sc, k->wk_keyix);
RSU_UNLOCK(sc);
return (!ret);
}
if (set) {
/* wait for pending key removal */
taskqueue_drain(taskqueue_thread, &sc->del_key_task);
RSU_LOCK(sc);
ret = rsu_set_key_pair(sc, k);
RSU_UNLOCK(sc);
} else {
RSU_DELKEY_BMAP_LOCK(sc);
setbit(sc->free_keys_bmap, k->wk_keyix);
RSU_DELKEY_BMAP_UNLOCK(sc);
/* workaround ieee80211_node_delucastkey() locking */
taskqueue_enqueue(taskqueue_thread, &sc->del_key_task);
ret = 0; /* fake success */
}
return (!ret);
}
static int
rsu_key_set(struct ieee80211vap *vap, const struct ieee80211_key *k)
{
return (rsu_process_key(vap, k, 1));
}
static int
rsu_key_delete(struct ieee80211vap *vap, const struct ieee80211_key *k)
{
return (rsu_process_key(vap, k, 0));
}
static int
rsu_cam_read(struct rsu_softc *sc, uint8_t addr, uint32_t *val)
{
int ntries;
rsu_write_4(sc, R92S_CAMCMD,
R92S_CAMCMD_POLLING | SM(R92S_CAMCMD_ADDR, addr));
for (ntries = 0; ntries < 10; ntries++) {
if (!(rsu_read_4(sc, R92S_CAMCMD) & R92S_CAMCMD_POLLING))
break;
usb_pause_mtx(&sc->sc_mtx, USB_MS_TO_TICKS(1));
}
if (ntries == 10) {
device_printf(sc->sc_dev,
"%s: cannot read CAM entry at address %02X\n",
__func__, addr);
return (ETIMEDOUT);
}
*val = rsu_read_4(sc, R92S_CAMREAD);
return (0);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static void
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
rsu_cam_write(struct rsu_softc *sc, uint8_t addr, uint32_t data)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
rsu_write_4(sc, R92S_CAMWRITE, data);
rsu_write_4(sc, R92S_CAMCMD,
R92S_CAMCMD_POLLING | R92S_CAMCMD_WRITE |
SM(R92S_CAMCMD_ADDR, addr));
}
static int
rsu_key_check(struct rsu_softc *sc, ieee80211_keyix keyix, int is_valid)
{
uint32_t val;
int error, ntries;
for (ntries = 0; ntries < 20; ntries++) {
usb_pause_mtx(&sc->sc_mtx, USB_MS_TO_TICKS(1));
error = rsu_cam_read(sc, R92S_CAM_CTL0(keyix), &val);
if (error != 0) {
device_printf(sc->sc_dev,
"%s: cannot check key status!\n", __func__);
return (error);
}
if (((val & R92S_CAM_VALID) == 0) ^ is_valid)
break;
}
if (ntries == 20) {
device_printf(sc->sc_dev,
"%s: key %d is %s marked as valid, rejecting request\n",
__func__, keyix, is_valid ? "not" : "still");
return (EIO);
}
return (0);
}
/*
* Map net80211 cipher to RTL8712 security mode.
*/
static uint8_t
rsu_crypto_mode(struct rsu_softc *sc, u_int cipher, int keylen)
{
switch (cipher) {
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
case IEEE80211_CIPHER_WEP:
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
return keylen < 8 ? R92S_KEY_ALGO_WEP40 : R92S_KEY_ALGO_WEP104;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
case IEEE80211_CIPHER_TKIP:
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
return R92S_KEY_ALGO_TKIP;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
case IEEE80211_CIPHER_AES_CCM:
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
return R92S_KEY_ALGO_AES;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
default:
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
device_printf(sc->sc_dev, "unknown cipher %d\n", cipher);
return R92S_KEY_ALGO_INVALID;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
}
static int
rsu_set_key_group(struct rsu_softc *sc, const struct ieee80211_key *k)
{
struct r92s_fw_cmd_set_key key;
uint8_t algo;
int error;
RSU_ASSERT_LOCKED(sc);
/* Map net80211 cipher to HW crypto algorithm. */
algo = rsu_crypto_mode(sc, k->wk_cipher->ic_cipher, k->wk_keylen);
if (algo == R92S_KEY_ALGO_INVALID)
return (EINVAL);
memset(&key, 0, sizeof(key));
key.algo = algo;
key.cam_id = k->wk_keyix;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
key.grpkey = (k->wk_flags & IEEE80211_KEY_GROUP) != 0;
memcpy(key.key, k->wk_key, MIN(k->wk_keylen, sizeof(key.key)));
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_KEY | RSU_DEBUG_FWCMD,
"%s: keyix %u, group %u, algo %u/%u, flags %04X, len %u, "
"macaddr %s\n", __func__, key.cam_id, key.grpkey,
k->wk_cipher->ic_cipher, key.algo, k->wk_flags, k->wk_keylen,
ether_sprintf(k->wk_macaddr));
error = rsu_fw_cmd(sc, R92S_CMD_SET_KEY, &key, sizeof(key));
if (error != 0) {
device_printf(sc->sc_dev,
"%s: cannot send firmware command, error %d\n",
__func__, error);
return (error);
}
return (rsu_key_check(sc, k->wk_keyix, 1));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
static int
rsu_set_key_pair(struct rsu_softc *sc, const struct ieee80211_key *k)
{
struct r92s_fw_cmd_set_key_mac key;
uint8_t algo;
int error;
RSU_ASSERT_LOCKED(sc);
if (!sc->sc_running)
return (ESHUTDOWN);
/* Map net80211 cipher to HW crypto algorithm. */
algo = rsu_crypto_mode(sc, k->wk_cipher->ic_cipher, k->wk_keylen);
if (algo == R92S_KEY_ALGO_INVALID)
return (EINVAL);
memset(&key, 0, sizeof(key));
key.algo = algo;
memcpy(key.macaddr, k->wk_macaddr, sizeof(key.macaddr));
memcpy(key.key, k->wk_key, MIN(k->wk_keylen, sizeof(key.key)));
RSU_DPRINTF(sc, RSU_DEBUG_KEY | RSU_DEBUG_FWCMD,
"%s: keyix %u, algo %u/%u, flags %04X, len %u, macaddr %s\n",
__func__, k->wk_keyix, k->wk_cipher->ic_cipher, key.algo,
k->wk_flags, k->wk_keylen, ether_sprintf(key.macaddr));
error = rsu_fw_cmd(sc, R92S_CMD_SET_STA_KEY, &key, sizeof(key));
if (error != 0) {
device_printf(sc->sc_dev,
"%s: cannot send firmware command, error %d\n",
__func__, error);
return (error);
}
return (rsu_key_check(sc, k->wk_keyix, 1));
}
static int
rsu_reinit_static_keys(struct rsu_softc *sc)
{
int i, error;
for (i = 0; i < nitems(sc->group_keys); i++) {
if (sc->group_keys[i] != NULL) {
error = rsu_set_key_group(sc, sc->group_keys[i]);
if (error != 0) {
device_printf(sc->sc_dev,
"%s: failed to set static key %d, "
"error %d\n", __func__, i, error);
return (error);
}
}
}
return (0);
}
static int
rsu_delete_key(struct rsu_softc *sc, ieee80211_keyix keyix)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
struct r92s_fw_cmd_set_key key;
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
uint32_t val;
int error;
RSU_ASSERT_LOCKED(sc);
if (!sc->sc_running)
return (0);
/* check if it was automatically removed by firmware */
error = rsu_cam_read(sc, R92S_CAM_CTL0(keyix), &val);
if (error == 0 && (val & R92S_CAM_VALID) == 0) {
RSU_DPRINTF(sc, RSU_DEBUG_KEY,
"%s: key %u does not exist\n", __func__, keyix);
clrbit(sc->keys_bmap, keyix);
return (0);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
memset(&key, 0, sizeof(key));
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
key.cam_id = keyix;
RSU_DPRINTF(sc, RSU_DEBUG_KEY | RSU_DEBUG_FWCMD,
"%s: removing key %u\n", __func__, key.cam_id);
error = rsu_fw_cmd(sc, R92S_CMD_SET_KEY, &key, sizeof(key));
if (error != 0) {
device_printf(sc->sc_dev,
"%s: cannot send firmware command, error %d\n",
__func__, error);
goto finish;
}
usb_pause_mtx(&sc->sc_mtx, USB_MS_TO_TICKS(5));
/*
* Clear 'valid' bit manually (cannot be done via firmware command).
* Used for key check + when firmware command cannot be sent.
*/
finish:
rsu_cam_write(sc, R92S_CAM_CTL0(keyix), 0);
clrbit(sc->keys_bmap, keyix);
return (rsu_key_check(sc, keyix, 0));
}
static void
rsu_delete_key_pair_cb(void *arg, int pending __unused)
{
struct rsu_softc *sc = arg;
int i;
RSU_DELKEY_BMAP_LOCK(sc);
for (i = IEEE80211_WEP_NKID; i < R92S_CAM_ENTRY_LIMIT; i++) {
if (isset(sc->free_keys_bmap, i)) {
RSU_DELKEY_BMAP_UNLOCK(sc);
RSU_LOCK(sc);
RSU_DPRINTF(sc, RSU_DEBUG_KEY,
"%s: calling rsu_delete_key() with keyix = %d\n",
__func__, i);
(void) rsu_delete_key(sc, i);
RSU_UNLOCK(sc);
RSU_DELKEY_BMAP_LOCK(sc);
clrbit(sc->free_keys_bmap, i);
/* bmap can be changed */
i = IEEE80211_WEP_NKID - 1;
continue;
}
}
RSU_DELKEY_BMAP_UNLOCK(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
static int
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
rsu_site_survey(struct rsu_softc *sc, struct ieee80211_scan_ssid *ssid)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
struct r92s_fw_cmd_sitesurvey cmd;
Add in a temporary (hah!) workaround for net80211 scanning versus NIC requirements. Don't start the opmode and join path until a pending survey is finished. This seems to reliably fix things. Ideally I'd just finish off the net80211 pluggable scan stuff and implement the methods here so if_rsu can just drive the scan machinery. However, that's a .. later thing. Whilst here, remove the getbuf debugging; it's okay to run out of transmit buffers under load; it however isn't okay to not be able to send commands. I'll fix that later.
2015-09-18 07:55:33 +00:00
RSU_ASSERT_LOCKED(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
memset(&cmd, 0, sizeof(cmd));
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
/* TODO: passive channels? */
if (sc->sc_active_scan)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
cmd.active = htole32(1);
cmd.limit = htole32(48);
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
if (ssid != NULL) {
sc->sc_extra_scan = 1;
cmd.ssidlen = htole32(ssid->len);
memcpy(cmd.ssid, ssid->ssid, ssid->len);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
#ifdef USB_DEBUG
if (rsu_debug & (RSU_DEBUG_SCAN | RSU_DEBUG_FWCMD)) {
device_printf(sc->sc_dev,
"sending site survey command, active %d",
le32toh(cmd.active));
if (ssid != NULL) {
printf(", ssid: ");
ieee80211_print_essid(cmd.ssid, le32toh(cmd.ssidlen));
}
printf("\n");
Add in a temporary (hah!) workaround for net80211 scanning versus NIC requirements. Don't start the opmode and join path until a pending survey is finished. This seems to reliably fix things. Ideally I'd just finish off the net80211 pluggable scan stuff and implement the methods here so if_rsu can just drive the scan machinery. However, that's a .. later thing. Whilst here, remove the getbuf debugging; it's okay to run out of transmit buffers under load; it however isn't okay to not be able to send commands. I'll fix that later.
2015-09-18 07:55:33 +00:00
}
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
#endif
return (rsu_fw_cmd(sc, R92S_CMD_SITE_SURVEY, &cmd, sizeof(cmd)));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
static int
rsu_join_bss(struct rsu_softc *sc, 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
struct ieee80211com *ic = &sc->sc_ic;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct ieee80211vap *vap = ni->ni_vap;
struct ndis_wlan_bssid_ex *bss;
struct ndis_802_11_fixed_ies *fixed;
struct r92s_fw_cmd_auth auth;
More RSU fixes: - Properly align temporary buffer to 32-bit. - Add an extra parenthesis to make expression clear. - Range check the association ID received from hardware. MFC after: 1 week
2014-06-04 16:58:35 +00:00
uint8_t buf[sizeof(*bss) + 128] __aligned(4);
uint8_t *frm;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
uint8_t opmode;
int error;
Add in a temporary (hah!) workaround for net80211 scanning versus NIC requirements. Don't start the opmode and join path until a pending survey is finished. This seems to reliably fix things. Ideally I'd just finish off the net80211 pluggable scan stuff and implement the methods here so if_rsu can just drive the scan machinery. However, that's a .. later thing. Whilst here, remove the getbuf debugging; it's okay to run out of transmit buffers under load; it however isn't okay to not be able to send commands. I'll fix that later.
2015-09-18 07:55:33 +00:00
RSU_ASSERT_LOCKED(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Let the FW decide the opmode based on the capinfo field. */
opmode = NDIS802_11AUTOUNKNOWN;
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RESET,
"%s: setting operating mode to %d\n",
__func__, opmode);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
error = rsu_fw_cmd(sc, R92S_CMD_SET_OPMODE, &opmode, sizeof(opmode));
if (error != 0)
return (error);
memset(&auth, 0, sizeof(auth));
if (vap->iv_flags & IEEE80211_F_WPA) {
auth.mode = R92S_AUTHMODE_WPA;
More RSU fixes: - Properly align temporary buffer to 32-bit. - Add an extra parenthesis to make expression clear. - Range check the association ID received from hardware. MFC after: 1 week
2014-06-04 16:58:35 +00:00
auth.dot1x = (ni->ni_authmode == IEEE80211_AUTH_8021X);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
} else
auth.mode = R92S_AUTHMODE_OPEN;
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RESET,
"%s: setting auth mode to %d\n",
__func__, auth.mode);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
error = rsu_fw_cmd(sc, R92S_CMD_SET_AUTH, &auth, sizeof(auth));
if (error != 0)
return (error);
memset(buf, 0, sizeof(buf));
bss = (struct ndis_wlan_bssid_ex *)buf;
IEEE80211_ADDR_COPY(bss->macaddr, ni->ni_bssid);
bss->ssid.ssidlen = htole32(ni->ni_esslen);
memcpy(bss->ssid.ssid, ni->ni_essid, ni->ni_esslen);
if (vap->iv_flags & (IEEE80211_F_PRIVACY | IEEE80211_F_WPA))
bss->privacy = htole32(1);
bss->rssi = htole32(ni->ni_avgrssi);
if (ic->ic_curmode == IEEE80211_MODE_11B)
bss->networktype = htole32(NDIS802_11DS);
else
bss->networktype = htole32(NDIS802_11OFDM24);
bss->config.len = htole32(sizeof(bss->config));
bss->config.bintval = htole32(ni->ni_intval);
bss->config.dsconfig = htole32(ieee80211_chan2ieee(ic, ni->ni_chan));
bss->inframode = htole32(NDIS802_11INFRASTRUCTURE);
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
/* XXX verify how this is supposed to look! */
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
memcpy(bss->supprates, ni->ni_rates.rs_rates,
ni->ni_rates.rs_nrates);
/* Write the fixed fields of the beacon frame. */
fixed = (struct ndis_802_11_fixed_ies *)&bss[1];
memcpy(&fixed->tstamp, ni->ni_tstamp.data, 8);
fixed->bintval = htole16(ni->ni_intval);
fixed->capabilities = htole16(ni->ni_capinfo);
/* Write IEs to be included in the association request. */
frm = (uint8_t *)&fixed[1];
frm = ieee80211_add_rsn(frm, vap);
frm = ieee80211_add_wpa(frm, vap);
frm = ieee80211_add_qos(frm, ni);
Add initial 11n support to if_rsu. * Add a tunable to enable 11n if it's available, so to not anger people who upgrade. kenv hw.usb.rsu.enable_11n=1 before inserting the device. * Add initial 11n htconfig bits; * Enable 40MHz mode if it's available; * Add 11n channels; * Set 11n bits in the firmware. It works for RX; I haven't tested TX aggregation just yet. However the firmware doesn't do RX re-ordering, so I have to tie it into the net80211 A-MPDU RX reorder path before I flip this on by default. I've verified that I'm indeed actually seeing MCS 0->7 rates being received. I haven't dug into whether it's actually transmitting 11n rates; I'll dig into that later.
2015-09-18 04:12:11 +00:00
if ((ic->ic_flags & IEEE80211_F_WME) &&
(ni->ni_ies.wme_ie != NULL))
frm = ieee80211_add_wme_info(frm, &ic->ic_wme);
* Add 11n HT40 support - i needed to send both HTINFO and HTCAP fields in the join message so the firmware would pick it up. * Strip out the direct hardware fiddling for 40MHz mode - the firmware we're using doesn't require it (the rtl8712su firmware does; it is less 'fullmac' than what we're using.) * Fix the mbuf handling during errors - rsu_tx shouldn't free mbufs; it's up to the caller to do so. This brings it in line with what other drivers do or should be doing. Tested: * RTL8712, HT40 channel, STA mode (during this commit)
2015-09-26 07:25:53 +00:00
if (ni->ni_flags & IEEE80211_NODE_HT) {
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
frm = ieee80211_add_htcap(frm, ni);
* Add 11n HT40 support - i needed to send both HTINFO and HTCAP fields in the join message so the firmware would pick it up. * Strip out the direct hardware fiddling for 40MHz mode - the firmware we're using doesn't require it (the rtl8712su firmware does; it is less 'fullmac' than what we're using.) * Fix the mbuf handling during errors - rsu_tx shouldn't free mbufs; it's up to the caller to do so. This brings it in line with what other drivers do or should be doing. Tested: * RTL8712, HT40 channel, STA mode (during this commit)
2015-09-26 07:25:53 +00:00
frm = ieee80211_add_htinfo(frm, ni);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
bss->ieslen = htole32(frm - (uint8_t *)fixed);
bss->len = htole32(((frm - buf) + 3) & ~3);
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RESET | RSU_DEBUG_FWCMD,
"%s: sending join bss command to %s chan %d\n",
__func__,
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ether_sprintf(bss->macaddr), le32toh(bss->config.dsconfig));
return (rsu_fw_cmd(sc, R92S_CMD_JOIN_BSS, buf, sizeof(buf)));
}
static int
rsu_disconnect(struct rsu_softc *sc)
{
uint32_t zero = 0; /* :-) */
/* Disassociate from our current BSS. */
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_STATE | RSU_DEBUG_FWCMD,
"%s: sending disconnect command\n", __func__);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
return (rsu_fw_cmd(sc, R92S_CMD_DISCONNECT, &zero, sizeof(zero)));
}
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
/*
* Map the hardware provided RSSI value to a signal level.
* For the most part it's just something we divide by and cap
* so it doesn't overflow the representation by net80211.
*/
static int
rsu_hwrssi_to_rssi(struct rsu_softc *sc, int hw_rssi)
{
int v;
if (hw_rssi == 0)
return (0);
v = hw_rssi >> 4;
if (v > 80)
v = 80;
return (v);
}
Properly range check length of parsed information elements in RSU driver. Found by: Ilja van Sprundel <ivansprundel@ioactive.com> MFC after: 3 days Sponsored by: Mellanox Technologies
2017-07-30 16:45:28 +00:00
CTASSERT(MCLBYTES > sizeof(struct ieee80211_frame));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static void
rsu_event_survey(struct rsu_softc *sc, uint8_t *buf, int 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
struct ieee80211com *ic = &sc->sc_ic;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct ieee80211_frame *wh;
struct ndis_wlan_bssid_ex *bss;
Replace the scan event input path hack with the new rx-stats based method. This allows for arbitrary channel info to be placed in the input call rather than the totally gross hack of overriding ic_curchan. Without this I'm sure ic_curchan setting was racing with the scan code setting the channel itself..
2015-09-15 03:01:40 +00:00
struct ieee80211_rx_stats rxs;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct mbuf *m;
Properly range check length of parsed information elements in RSU driver. Found by: Ilja van Sprundel <ivansprundel@ioactive.com> MFC after: 3 days Sponsored by: Mellanox Technologies
2017-07-30 16:45:28 +00:00
uint32_t ieslen;
uint32_t pktlen;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (__predict_false(len < sizeof(*bss)))
return;
bss = (struct ndis_wlan_bssid_ex *)buf;
Properly range check length of parsed information elements in RSU driver. Found by: Ilja van Sprundel <ivansprundel@ioactive.com> MFC after: 3 days Sponsored by: Mellanox Technologies
2017-07-30 16:45:28 +00:00
ieslen = le32toh(bss->ieslen);
/* range check length of information element */
if (__predict_false(ieslen > (uint32_t)(len - sizeof(*bss))))
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
return;
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_SCAN,
"%s: found BSS %s: len=%d chan=%d inframode=%d "
Add RSSI logging to the event survey results.
2015-09-13 05:22:20 +00:00
"networktype=%d privacy=%d, RSSI=%d\n",
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
__func__,
Properly range check length of parsed information elements in RSU driver. Found by: Ilja van Sprundel <ivansprundel@ioactive.com> MFC after: 3 days Sponsored by: Mellanox Technologies
2017-07-30 16:45:28 +00:00
ether_sprintf(bss->macaddr), ieslen,
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
le32toh(bss->config.dsconfig), le32toh(bss->inframode),
Add RSSI logging to the event survey results.
2015-09-13 05:22:20 +00:00
le32toh(bss->networktype), le32toh(bss->privacy),
le32toh(bss->rssi));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Build a fake beacon frame to let net80211 do all the parsing. */
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
/* XXX TODO: just call the new scan API methods! */
Properly range check length of parsed information elements in RSU driver. Found by: Ilja van Sprundel <ivansprundel@ioactive.com> MFC after: 3 days Sponsored by: Mellanox Technologies
2017-07-30 16:45:28 +00:00
if (__predict_false(ieslen > (size_t)(MCLBYTES - sizeof(*wh))))
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
return;
Properly range check length of parsed information elements in RSU driver. Found by: Ilja van Sprundel <ivansprundel@ioactive.com> MFC after: 3 days Sponsored by: Mellanox Technologies
2017-07-30 16:45:28 +00:00
pktlen = sizeof(*wh) + ieslen;
Use m_get2() instead of m_getcl(). Spotted by: glebius
2014-01-10 14:47:20 +00:00
m = m_get2(pktlen, M_NOWAIT, MT_DATA, M_PKTHDR);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (__predict_false(m == NULL))
return;
wh = mtod(m, struct ieee80211_frame *);
wh->i_fc[0] = IEEE80211_FC0_VERSION_0 | IEEE80211_FC0_TYPE_MGT |
IEEE80211_FC0_SUBTYPE_BEACON;
wh->i_fc[1] = IEEE80211_FC1_DIR_NODS;
Fix external compiler warning(s). Avoid pointer dereferencing.
2013-12-04 12:07:46 +00:00
USETW(wh->i_dur, 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
IEEE80211_ADDR_COPY(wh->i_addr1, ieee80211broadcastaddr);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
IEEE80211_ADDR_COPY(wh->i_addr2, bss->macaddr);
IEEE80211_ADDR_COPY(wh->i_addr3, bss->macaddr);
*(uint16_t *)wh->i_seq = 0;
Properly range check length of parsed information elements in RSU driver. Found by: Ilja van Sprundel <ivansprundel@ioactive.com> MFC after: 3 days Sponsored by: Mellanox Technologies
2017-07-30 16:45:28 +00:00
memcpy(&wh[1], (uint8_t *)&bss[1], ieslen);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Finalize mbuf. */
m->m_pkthdr.len = m->m_len = pktlen;
Replace the scan event input path hack with the new rx-stats based method. This allows for arbitrary channel info to be placed in the input call rather than the totally gross hack of overriding ic_curchan. Without this I'm sure ic_curchan setting was racing with the scan code setting the channel itself..
2015-09-15 03:01:40 +00:00
/* Set channel flags for input path */
bzero(&rxs, sizeof(rxs));
rxs.r_flags |= IEEE80211_R_IEEE | IEEE80211_R_FREQ;
rxs.r_flags |= IEEE80211_R_NF | IEEE80211_R_RSSI;
rxs.c_ieee = le32toh(bss->config.dsconfig);
rxs.c_freq = ieee80211_ieee2mhz(rxs.c_ieee, IEEE80211_CHAN_2GHZ);
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
/* This is a number from 0..100; so let's just divide it down a bit */
[net80211] extend the ieee80211_rx_stats struct to include more information. There are a variety of more interesting RX statistics that we should keep track of but we don't. This is a starting point for adding more information. Specifically: * now the RX rate information and some of the packet status is passed up; * The 32 bit or 64 bit TSF is passed up; * the PHY mode is passed up; * the "I'm decap'ed AMSDU!" state is passed up; * number of RX chains is bumped to 4. This is all mostly a placeholder for getting the data into the RX status before we pass it up to net80211 - unfortunately we don't yet enforce that drivers provide it, nor do we pass the provided info back up the stack so anyone can use the data. We're going to need to use some of this data moving forward. Notably, now that some hardware can do AMSDU decap for us (the intel iwm driver can do it when we flip it on; the ath10k port I'm doing does it for us) then we need to pass it up through the stack so the duplicate RX sequence numbers and crypto/IV details don't cause the packet to be dropped and/or counted against a replay counter. It's also the beginning of being able to do more interesting node accounting in net80211. Specifically, once drivers start populating per-packet rate information, AMPDU information, timestamps, etc, we can start providing histograms of rate-versus-RSSI, account for receive time spent per node and other such interesting things. (Note: I'm also hoping to include ranging and RTT information for future chipset support; and it's likely going to include it in this kind of fashion.)
2016-10-08 01:12:29 +00:00
rxs.c_rssi = le32toh(bss->rssi) / 2;
rxs.c_nf = -96;
net80211: convert all ieee80211_input_mimo*() consumers to ieee80211_add_rx_params() + drop last (ieee80211_rx_stats) parameter Note: there is an additional check for ieee80211_get_rx_params() return value (which does not exist in the original diff). Reviewed by: adrian Differential Revision: https://reviews.freebsd.org/D8207
2016-10-12 20:50:13 +00:00
if (ieee80211_add_rx_params(m, &rxs) == 0)
return;
Replace the scan event input path hack with the new rx-stats based method. This allows for arbitrary channel info to be placed in the input call rather than the totally gross hack of overriding ic_curchan. Without this I'm sure ic_curchan setting was racing with the scan code setting the channel itself..
2015-09-15 03:01:40 +00:00
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* XXX avoid a LOR */
RSU_UNLOCK(sc);
net80211: convert all ieee80211_input_mimo*() consumers to ieee80211_add_rx_params() + drop last (ieee80211_rx_stats) parameter Note: there is an additional check for ieee80211_get_rx_params() return value (which does not exist in the original diff). Reviewed by: adrian Differential Revision: https://reviews.freebsd.org/D8207
2016-10-12 20:50:13 +00:00
ieee80211_input_mimo_all(ic, m);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_LOCK(sc);
}
static void
rsu_event_join_bss(struct rsu_softc *sc, uint8_t *buf, int 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
struct ieee80211com *ic = &sc->sc_ic;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
struct ieee80211_node *ni = vap->iv_bss;
struct r92s_event_join_bss *rsp;
More RSU fixes: - Properly align temporary buffer to 32-bit. - Add an extra parenthesis to make expression clear. - Range check the association ID received from hardware. MFC after: 1 week
2014-06-04 16:58:35 +00:00
uint32_t tmp;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
int res;
if (__predict_false(len < sizeof(*rsp)))
return;
rsp = (struct r92s_event_join_bss *)buf;
res = (int)le32toh(rsp->join_res);
* fiddle with some more of the debugging output * yes, when a "sta disconnect" message comes through we should, like, disconnect things. We're not currently generating beacon miss messages, and net80211 isn't disconnecting things via software beacon miss receive. Tested: * RTL8712, cut 3, STA mode
2015-09-13 04:12:51 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_STATE | RSU_DEBUG_FWCMD,
"%s: Rx join BSS event len=%d res=%d\n",
__func__, len, res);
Add in a temporary (hah!) workaround for net80211 scanning versus NIC requirements. Don't start the opmode and join path until a pending survey is finished. This seems to reliably fix things. Ideally I'd just finish off the net80211 pluggable scan stuff and implement the methods here so if_rsu can just drive the scan machinery. However, that's a .. later thing. Whilst here, remove the getbuf debugging; it's okay to run out of transmit buffers under load; it however isn't okay to not be able to send commands. I'll fix that later.
2015-09-18 07:55:33 +00:00
/*
* XXX Don't do this; there's likely a better way to tell
* the caller we failed.
*/
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (res <= 0) {
RSU_UNLOCK(sc);
ieee80211_new_state(vap, IEEE80211_S_SCAN, -1);
RSU_LOCK(sc);
return;
}
Add in a temporary (hah!) workaround for net80211 scanning versus NIC requirements. Don't start the opmode and join path until a pending survey is finished. This seems to reliably fix things. Ideally I'd just finish off the net80211 pluggable scan stuff and implement the methods here so if_rsu can just drive the scan machinery. However, that's a .. later thing. Whilst here, remove the getbuf debugging; it's okay to run out of transmit buffers under load; it however isn't okay to not be able to send commands. I'll fix that later.
2015-09-18 07:55:33 +00:00
More RSU fixes: - Properly align temporary buffer to 32-bit. - Add an extra parenthesis to make expression clear. - Range check the association ID received from hardware. MFC after: 1 week
2014-06-04 16:58:35 +00:00
tmp = le32toh(rsp->associd);
if (tmp >= vap->iv_max_aid) {
rsu: use bitmap for all debug messages. - Replace all remaining DPRINTF(N)'s with RSU_DPRINTF. - Add new RSU_DEBUG_USB flag to track error codes returned by usbd_do_request_flags(). - Improve few messages.
2016-12-10 20:19:57 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_ANY, "Assoc ID overflow\n");
More RSU fixes: - Properly align temporary buffer to 32-bit. - Add an extra parenthesis to make expression clear. - Range check the association ID received from hardware. MFC after: 1 week
2014-06-04 16:58:35 +00:00
tmp = 1;
}
* fiddle with some more of the debugging output * yes, when a "sta disconnect" message comes through we should, like, disconnect things. We're not currently generating beacon miss messages, and net80211 isn't disconnecting things via software beacon miss receive. Tested: * RTL8712, cut 3, STA mode
2015-09-13 04:12:51 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_STATE | RSU_DEBUG_FWCMD,
"%s: associated with %s associd=%d\n",
__func__, ether_sprintf(rsp->bss.macaddr), tmp);
/* XXX is this required? What's the top two bits for again? */
More RSU fixes: - Properly align temporary buffer to 32-bit. - Add an extra parenthesis to make expression clear. - Range check the association ID received from hardware. MFC after: 1 week
2014-06-04 16:58:35 +00:00
ni->ni_associd = tmp | 0xc000;
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
/* Refresh Rx filter (was changed by firmware). */
sc->sc_vap_is_running = 1;
rsu_rxfilter_refresh(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_UNLOCK(sc);
ieee80211_new_state(vap, IEEE80211_S_RUN,
IEEE80211_FC0_SUBTYPE_ASSOC_RESP);
RSU_LOCK(sc);
}
Add a very hacked up station only A-MPDU negotiation path. This is enough to set things up; there are still lots of retransmits seen but it's enough to get things working.
2015-09-18 05:03:01 +00:00
static void
rsu_event_addba_req_report(struct rsu_softc *sc, uint8_t *buf, int len)
{
struct ieee80211com *ic = &sc->sc_ic;
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
struct r92s_add_ba_event *ba = (void *) buf;
struct ieee80211_node *ni;
if (len < sizeof(*ba)) {
device_printf(sc->sc_dev, "%s: short read (%d)\n", __func__, len);
return;
}
if (vap == NULL)
return;
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_AMPDU, "%s: mac=%s, tid=%d, ssn=%d\n",
Add a very hacked up station only A-MPDU negotiation path. This is enough to set things up; there are still lots of retransmits seen but it's enough to get things working.
2015-09-18 05:03:01 +00:00
__func__,
ether_sprintf(ba->mac_addr),
(int) ba->tid,
(int) le16toh(ba->ssn));
/* XXX do node lookup; this is STA specific */
ni = ieee80211_ref_node(vap->iv_bss);
ieee80211_ampdu_rx_start_ext(ni, ba->tid, le16toh(ba->ssn) >> 4, 32);
ieee80211_free_node(ni);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static void
rsu_rx_event(struct rsu_softc *sc, uint8_t code, uint8_t *buf, int 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
struct ieee80211com *ic = &sc->sc_ic;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
* fiddle with some more of the debugging output * yes, when a "sta disconnect" message comes through we should, like, disconnect things. We're not currently generating beacon miss messages, and net80211 isn't disconnecting things via software beacon miss receive. Tested: * RTL8712, cut 3, STA mode
2015-09-13 04:12:51 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RX | RSU_DEBUG_FWCMD,
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
"%s: Rx event code=%d len=%d\n", __func__, code, len);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
switch (code) {
case R92S_EVT_SURVEY:
Add initial 11n support to if_rsu. * Add a tunable to enable 11n if it's available, so to not anger people who upgrade. kenv hw.usb.rsu.enable_11n=1 before inserting the device. * Add initial 11n htconfig bits; * Enable 40MHz mode if it's available; * Add 11n channels; * Set 11n bits in the firmware. It works for RX; I haven't tested TX aggregation just yet. However the firmware doesn't do RX re-ordering, so I have to tie it into the net80211 A-MPDU RX reorder path before I flip this on by default. I've verified that I'm indeed actually seeing MCS 0->7 rates being received. I haven't dug into whether it's actually transmitting 11n rates; I'll dig into that later.
2015-09-18 04:12:11 +00:00
rsu_event_survey(sc, buf, len);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
break;
case R92S_EVT_SURVEY_DONE:
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_SCAN,
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
"%s: %s scan done, found %d BSS\n",
__func__, sc->sc_extra_scan ? "direct" : "broadcast",
le32toh(*(uint32_t *)buf));
if (sc->sc_extra_scan == 1) {
/* Send broadcast probe request. */
sc->sc_extra_scan = 0;
if (vap != NULL && rsu_site_survey(sc, NULL) != 0) {
RSU_UNLOCK(sc);
ieee80211_cancel_scan(vap);
RSU_LOCK(sc);
}
break;
}
if (vap != NULL) {
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_UNLOCK(sc);
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
ieee80211_scan_done(vap);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_LOCK(sc);
}
break;
case R92S_EVT_JOIN_BSS:
if (vap->iv_state == IEEE80211_S_AUTH)
rsu_event_join_bss(sc, buf, len);
break;
case R92S_EVT_DEL_STA:
* fiddle with some more of the debugging output * yes, when a "sta disconnect" message comes through we should, like, disconnect things. We're not currently generating beacon miss messages, and net80211 isn't disconnecting things via software beacon miss receive. Tested: * RTL8712, cut 3, STA mode
2015-09-13 04:12:51 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_FWCMD | RSU_DEBUG_STATE,
"%s: disassociated from %s\n", __func__,
ether_sprintf(buf));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (vap->iv_state == IEEE80211_S_RUN &&
IEEE80211_ADDR_EQ(vap->iv_bss->ni_bssid, buf)) {
RSU_UNLOCK(sc);
ieee80211_new_state(vap, IEEE80211_S_SCAN, -1);
RSU_LOCK(sc);
}
break;
case R92S_EVT_WPS_PBC:
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RX | RSU_DEBUG_FWCMD,
"%s: WPS PBC pushed.\n", __func__);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
break;
case R92S_EVT_FWDBG:
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
buf[60] = '\0';
RSU_DPRINTF(sc, RSU_DEBUG_FWDBG, "FWDBG: %s\n", (char *)buf);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
break;
Add a very hacked up station only A-MPDU negotiation path. This is enough to set things up; there are still lots of retransmits seen but it's enough to get things working.
2015-09-18 05:03:01 +00:00
case R92S_EVT_ADDBA_REQ_REPORT:
rsu_event_addba_req_report(sc, buf, len);
break;
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
default:
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
device_printf(sc->sc_dev, "%s: unhandled code (%d)\n", __func__, code);
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
break;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
}
static void
rsu_rx_multi_event(struct rsu_softc *sc, uint8_t *buf, int len)
{
struct r92s_fw_cmd_hdr *cmd;
int cmdsz;
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RX, "%s: Rx events len=%d\n", __func__, len);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Skip Rx status. */
buf += sizeof(struct r92s_rx_stat);
len -= sizeof(struct r92s_rx_stat);
/* Process all events. */
for (;;) {
/* Check that command header fits. */
if (__predict_false(len < sizeof(*cmd)))
break;
cmd = (struct r92s_fw_cmd_hdr *)buf;
/* Check that command payload fits. */
cmdsz = le16toh(cmd->len);
if (__predict_false(len < sizeof(*cmd) + cmdsz))
break;
/* Process firmware event. */
rsu_rx_event(sc, cmd->code, (uint8_t *)&cmd[1], cmdsz);
if (!(cmd->seq & R92S_FW_CMD_MORE))
break;
buf += sizeof(*cmd) + cmdsz;
len -= sizeof(*cmd) + cmdsz;
}
}
static int8_t
rsu_get_rssi(struct rsu_softc *sc, int rate, void *physt)
{
static const int8_t cckoff[] = { 14, -2, -20, -40 };
struct r92s_rx_phystat *phy;
struct r92s_rx_cck *cck;
uint8_t rpt;
int8_t rssi;
if (rate <= 3) {
cck = (struct r92s_rx_cck *)physt;
rpt = (cck->agc_rpt >> 6) & 0x3;
rssi = cck->agc_rpt & 0x3e;
rssi = cckoff[rpt] - rssi;
} else { /* OFDM/HT. */
phy = (struct r92s_rx_phystat *)physt;
rssi = ((le32toh(phy->phydw1) >> 1) & 0x7f) - 106;
}
return (rssi);
}
static struct mbuf *
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
rsu_rx_copy_to_mbuf(struct rsu_softc *sc, struct r92s_rx_stat *stat,
int totlen)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +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;
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
struct mbuf *m;
uint32_t rxdw0;
int pktlen;
rxdw0 = le32toh(stat->rxdw0);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
if (__predict_false(rxdw0 & (R92S_RXDW0_CRCERR | R92S_RXDW0_ICVERR))) {
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RX,
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
"%s: RX flags error (%s)\n", __func__,
rxdw0 & R92S_RXDW0_CRCERR ? "CRC" : "ICV");
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
goto fail;
}
pktlen = MS(rxdw0, R92S_RXDW0_PKTLEN);
if (__predict_false(pktlen < sizeof (struct ieee80211_frame_ack))) {
RSU_DPRINTF(sc, RSU_DEBUG_RX,
"%s: frame is too short: %d\n", __func__, pktlen);
goto fail;
}
m = m_get2(totlen, M_NOWAIT, MT_DATA, M_PKTHDR);
if (__predict_false(m == NULL)) {
rsu: use bitmap for all debug messages. - Replace all remaining DPRINTF(N)'s with RSU_DPRINTF. - Add new RSU_DEBUG_USB flag to track error codes returned by usbd_do_request_flags(). - Improve few messages.
2016-12-10 20:19:57 +00:00
device_printf(sc->sc_dev,
"%s: could not allocate RX mbuf, totlen %d\n",
__func__, totlen);
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
goto fail;
}
/* Finalize mbuf. */
memcpy(mtod(m, uint8_t *), (uint8_t *)stat, totlen);
m->m_pkthdr.len = m->m_len = totlen;
return (m);
fail:
counter_u64_add(ic->ic_ierrors, 1);
return (NULL);
}
rsu: add TSF field into Rx radiotap. Tested with Asus USB-N10.
2016-12-10 13:30:16 +00:00
static uint32_t
rsu_get_tsf_low(struct rsu_softc *sc)
{
return (rsu_read_4(sc, R92S_TSFTR));
}
static uint32_t
rsu_get_tsf_high(struct rsu_softc *sc)
{
return (rsu_read_4(sc, R92S_TSFTR + 4));
}
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
static struct ieee80211_node *
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
rsu_rx_frame(struct rsu_softc *sc, struct mbuf *m)
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
{
struct ieee80211com *ic = &sc->sc_ic;
struct ieee80211_frame_min *wh;
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
struct ieee80211_rx_stats rxs;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct r92s_rx_stat *stat;
uint32_t rxdw0, rxdw3;
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
uint8_t cipher, rate;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
int infosz;
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
int rssi;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
stat = mtod(m, struct r92s_rx_stat *);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rxdw0 = le32toh(stat->rxdw0);
rxdw3 = le32toh(stat->rxdw3);
rate = MS(rxdw3, R92S_RXDW3_RATE);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
cipher = MS(rxdw0, R92S_RXDW0_CIPHER);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
infosz = MS(rxdw0, R92S_RXDW0_INFOSZ) * 8;
/* Get RSSI from PHY status descriptor if present. */
rsu: fix RSSI reporting, partially revert r288414. - Append RCR_APP_PHYSTS bit after firmware loading - otherwise firmware will reset the register and this modification will be lost. (without it Rx PHY descriptor section will contain garbage). - Check if R92S_RXDW0_PHYST bit is set (like it is done in rtwn(4)) - even if infosz is non-zero the section may not contain anything useful. - In case, if descriptor is absent (A-MPDU?) use last calibrated RSSI (rtwn(4) uses RSSI from the previous (sub)frame; probably, this approach should be used here too). Tested with Asus USB-N10, STA mode.
2016-12-04 21:40:49 +00:00
if (infosz != 0 && (rxdw0 & R92S_RXDW0_PHYST))
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
rssi = rsu_get_rssi(sc, rate, &stat[1]);
rsu: fix RSSI reporting, partially revert r288414. - Append RCR_APP_PHYSTS bit after firmware loading - otherwise firmware will reset the register and this modification will be lost. (without it Rx PHY descriptor section will contain garbage). - Check if R92S_RXDW0_PHYST bit is set (like it is done in rtwn(4)) - even if infosz is non-zero the section may not contain anything useful. - In case, if descriptor is absent (A-MPDU?) use last calibrated RSSI (rtwn(4) uses RSSI from the previous (sub)frame; probably, this approach should be used here too). Tested with Asus USB-N10, STA mode.
2016-12-04 21:40:49 +00:00
else {
/* Cheat and get the last calibrated RSSI */
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
rssi = rsu_hwrssi_to_rssi(sc, sc->sc_currssi);
rsu: fix RSSI reporting, partially revert r288414. - Append RCR_APP_PHYSTS bit after firmware loading - otherwise firmware will reset the register and this modification will be lost. (without it Rx PHY descriptor section will contain garbage). - Check if R92S_RXDW0_PHYST bit is set (like it is done in rtwn(4)) - even if infosz is non-zero the section may not contain anything useful. - In case, if descriptor is absent (A-MPDU?) use last calibrated RSSI (rtwn(4) uses RSSI from the previous (sub)frame; probably, this approach should be used here too). Tested with Asus USB-N10, STA mode.
2016-12-04 21:40:49 +00:00
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
/* Hardware does Rx TCP checksum offload. */
rsu: fix and enable Rx TCP checksum offloading. Tested with Asus USB-N10, STA mode.
2016-12-11 17:15:25 +00:00
/*
* This flag can be set for some other
* (e.g., EAPOL) frame types, so don't rely on it.
*/
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
if (rxdw3 & R92S_RXDW3_TCPCHKVALID) {
rsu: fix and enable Rx TCP checksum offloading. Tested with Asus USB-N10, STA mode.
2016-12-11 17:15:25 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RX,
"%s: TCP/IP checksums: %schecked / %schecked\n",
__func__,
(rxdw3 & R92S_RXDW3_TCPCHKRPT) ? "" : "not ",
(rxdw3 & R92S_RXDW3_IPCHKRPT) ? "" : "not ");
/*
* 'IP header checksum valid' bit will not be set if
* the frame was not checked / has incorrect checksum /
* does not have checksum (IPv6).
*
* NB: if DF bit is not set then frame will not be checked.
*/
if (rxdw3 & R92S_RXDW3_IPCHKRPT) {
m->m_pkthdr.csum_flags = CSUM_IP_CHECKED;
m->m_pkthdr.csum_flags |= CSUM_IP_VALID;
}
/*
* This is independent of the above check.
*/
if (rxdw3 & R92S_RXDW3_TCPCHKRPT) {
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
m->m_pkthdr.csum_flags |= CSUM_DATA_VALID;
rsu: fix and enable Rx TCP checksum offloading. Tested with Asus USB-N10, STA mode.
2016-12-11 17:15:25 +00:00
m->m_pkthdr.csum_flags |= CSUM_PSEUDO_HDR;
m->m_pkthdr.csum_data = 0xffff;
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
/* RX flags */
/* Set channel flags for input path */
bzero(&rxs, sizeof(rxs));
/* normal RSSI */
rxs.r_flags |= IEEE80211_R_NF | IEEE80211_R_RSSI;
rxs.c_rssi = rssi;
rxs.c_nf = -96;
/* Rate */
rsu(4): trim code for Rx rate calculation. Include ridx <-> rate conversion functions from rtwn(4) + reuse already calculated value for ieee80211_radiotap(9). Tested with Asus USB-N10, STA mode.
2017-11-02 00:17:52 +00:00
if (rate < 12) {
rxs.c_rate = ridx2rate[rate];
if (RSU_RATE_IS_CCK(rate))
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
rxs.c_pktflags |= IEEE80211_RX_F_CCK;
rsu(4): trim code for Rx rate calculation. Include ridx <-> rate conversion functions from rtwn(4) + reuse already calculated value for ieee80211_radiotap(9). Tested with Asus USB-N10, STA mode.
2017-11-02 00:17:52 +00:00
else
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
rxs.c_pktflags |= IEEE80211_RX_F_OFDM;
rsu(4): trim code for Rx rate calculation. Include ridx <-> rate conversion functions from rtwn(4) + reuse already calculated value for ieee80211_radiotap(9). Tested with Asus USB-N10, STA mode.
2017-11-02 00:17:52 +00:00
} else {
rxs.c_rate = IEEE80211_RATE_MCS | (rate - 12);
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
rxs.c_pktflags |= IEEE80211_RX_F_HT;
}
rsu(4): trim code for Rx rate calculation. Include ridx <-> rate conversion functions from rtwn(4) + reuse already calculated value for ieee80211_radiotap(9). Tested with Asus USB-N10, STA mode.
2017-11-02 00:17:52 +00:00
if (ieee80211_radiotap_active(ic)) {
struct rsu_rx_radiotap_header *tap = &sc->sc_rxtap;
/* Map HW rate index to 802.11 rate. */
tap->wr_flags = 0; /* TODO */
tap->wr_tsft = rsu_get_tsf_high(sc);
if (le32toh(stat->tsf_low) > rsu_get_tsf_low(sc))
tap->wr_tsft--;
tap->wr_tsft = (uint64_t)htole32(tap->wr_tsft) << 32;
tap->wr_tsft += stat->tsf_low;
tap->wr_rate = rxs.c_rate;
tap->wr_dbm_antsignal = rssi;
tap->wr_chan_freq = htole16(ic->ic_curchan->ic_freq);
tap->wr_chan_flags = htole16(ic->ic_curchan->ic_flags);
};
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
(void) ieee80211_add_rx_params(m, &rxs);
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
/* Drop descriptor. */
m_adj(m, sizeof(*stat) + infosz);
wh = mtod(m, struct ieee80211_frame_min *);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
if ((wh->i_fc[1] & IEEE80211_FC1_PROTECTED) &&
cipher != R92S_KEY_ALGO_NONE) {
m->m_flags |= M_WEP;
}
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RX,
"%s: Rx frame len %d, rate %d, infosz %d\n",
__func__, m->m_len, rate, infosz);
if (m->m_len >= sizeof(*wh))
return (ieee80211_find_rxnode(ic, wh));
return (NULL);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
static struct mbuf *
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
rsu_rx_multi_frame(struct rsu_softc *sc, uint8_t *buf, int len)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
struct r92s_rx_stat *stat;
uint32_t rxdw0;
int totlen, pktlen, infosz, npkts;
struct mbuf *m, *m0 = NULL, *prevm = NULL;
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
/*
* don't pass packets to the ieee80211 framework if the driver isn't
* RUNNING.
*/
if (!sc->sc_running)
return (NULL);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Get the number of encapsulated frames. */
stat = (struct r92s_rx_stat *)buf;
npkts = MS(le32toh(stat->rxdw2), R92S_RXDW2_PKTCNT);
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RX,
"%s: Rx %d frames in one chunk\n", __func__, npkts);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Process all of them. */
while (npkts-- > 0) {
if (__predict_false(len < sizeof(*stat)))
break;
stat = (struct r92s_rx_stat *)buf;
rxdw0 = le32toh(stat->rxdw0);
pktlen = MS(rxdw0, R92S_RXDW0_PKTLEN);
if (__predict_false(pktlen == 0))
break;
infosz = MS(rxdw0, R92S_RXDW0_INFOSZ) * 8;
/* Make sure everything fits in xfer. */
totlen = sizeof(*stat) + infosz + pktlen;
if (__predict_false(totlen > len))
break;
/* Process 802.11 frame. */
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
m = rsu_rx_copy_to_mbuf(sc, stat, totlen);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (m0 == NULL)
m0 = m;
if (prevm == NULL)
prevm = m;
else {
prevm->m_next = m;
prevm = m;
}
/* Next chunk is 128-byte aligned. */
totlen = (totlen + 127) & ~127;
buf += totlen;
len -= totlen;
}
return (m0);
}
static struct mbuf *
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
rsu_rxeof(struct usb_xfer *xfer, struct rsu_data *data)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
struct rsu_softc *sc = data->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;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct r92s_rx_stat *stat;
int len;
usbd_xfer_status(xfer, &len, NULL, NULL, NULL);
if (__predict_false(len < sizeof(*stat))) {
rsu: use bitmap for all debug messages. - Replace all remaining DPRINTF(N)'s with RSU_DPRINTF. - Add new RSU_DEBUG_USB flag to track error codes returned by usbd_do_request_flags(). - Improve few messages.
2016-12-10 20:19:57 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RX, "xfer too short %d\n", 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);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
return (NULL);
}
/* Determine if it is a firmware C2H event or an 802.11 frame. */
stat = (struct r92s_rx_stat *)data->buf;
if ((le32toh(stat->rxdw1) & 0x1ff) == 0x1ff) {
rsu_rx_multi_event(sc, data->buf, len);
/* No packets to process. */
return (NULL);
} else
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
return (rsu_rx_multi_frame(sc, data->buf, len));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
static void
rsu_bulk_rx_callback(struct usb_xfer *xfer, usb_error_t error)
{
struct rsu_softc *sc = usbd_xfer_softc(xfer);
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;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct ieee80211_node *ni;
struct mbuf *m = NULL, *next;
struct rsu_data *data;
RSU_ASSERT_LOCKED(sc);
switch (USB_GET_STATE(xfer)) {
case USB_ST_TRANSFERRED:
data = STAILQ_FIRST(&sc->sc_rx_active);
if (data == NULL)
goto tr_setup;
STAILQ_REMOVE_HEAD(&sc->sc_rx_active, next);
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
m = rsu_rxeof(xfer, data);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
STAILQ_INSERT_TAIL(&sc->sc_rx_inactive, data, next);
/* FALLTHROUGH */
case USB_ST_SETUP:
tr_setup:
data = STAILQ_FIRST(&sc->sc_rx_inactive);
if (data == NULL) {
KASSERT(m == NULL, ("mbuf isn't NULL"));
return;
}
STAILQ_REMOVE_HEAD(&sc->sc_rx_inactive, next);
STAILQ_INSERT_TAIL(&sc->sc_rx_active, data, next);
usbd_xfer_set_frame_data(xfer, 0, data->buf,
usbd_xfer_max_len(xfer));
usbd_transfer_submit(xfer);
/*
* To avoid LOR we should unlock our private mutex here to call
* ieee80211_input() because here is at the end of a USB
* callback and safe to unlock.
*/
while (m != NULL) {
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
next = m->m_next;
m->m_next = NULL;
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
ni = rsu_rx_frame(sc, m);
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
RSU_UNLOCK(sc);
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (ni != NULL) {
Add a very hacked up station only A-MPDU negotiation path. This is enough to set things up; there are still lots of retransmits seen but it's enough to get things working.
2015-09-18 05:03:01 +00:00
if (ni->ni_flags & IEEE80211_NODE_HT)
m->m_flags |= M_AMPDU;
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
(void)ieee80211_input_mimo(ni, m);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ieee80211_free_node(ni);
} else
[rsu] convert rsu to use the ieee80211_rx_stats struct to pass up RSSI, PHY and rate information. I don't yet know which RX descriptor bits map to shortgi, long-gi, short-preamble, long-preamble, STBC, LDPC, HT40, etc - so I can't easily add those just yet. There's apparently no per-frame RX RSSI information exposed so we also just use the results from the previous calibration task. This also tidies up how the per-mbuf RSSI is pushed into the frame - now that it's attached to the mbuf via rx_stats, we don't have to do any silly hijinx to get it out of the frame processing path. Tested: * RTL8712, 1x1 cut 3, STA mode
2016-12-22 21:01:56 +00:00
(void)ieee80211_input_mimo_all(ic, m);
rsu: fix frame processing in the Rx path (similar to r292207). - Fill in Rx radiotap header correctly (for every packet in a chain; not once per chain). - Fix rate / flags fields in Rx radiotap. - Add debug messages for discarded frames. - Pass received control (< sizeof(struct ieee80211_frame)) frames to net80211 (if allowed by device filter; cannot happen yet). Tested with Asus USB-N10. Differential Revision: https://reviews.freebsd.org/D5723
2016-12-03 16:02:53 +00:00
RSU_LOCK(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
m = next;
}
break;
default:
/* needs it to the inactive queue due to a error. */
data = STAILQ_FIRST(&sc->sc_rx_active);
if (data != NULL) {
STAILQ_REMOVE_HEAD(&sc->sc_rx_active, next);
STAILQ_INSERT_TAIL(&sc->sc_rx_inactive, data, next);
}
if (error != USB_ERR_CANCELLED) {
usbd_xfer_set_stall(xfer);
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);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
goto tr_setup;
}
break;
}
}
static void
rsu_txeof(struct usb_xfer *xfer, struct rsu_data *data)
{
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
#ifdef USB_DEBUG
struct rsu_softc *sc = usbd_xfer_softc(xfer);
#endif
RSU_DPRINTF(sc, RSU_DEBUG_TXDONE, "%s: called; data=%p\n",
__func__,
data);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (data->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
/* XXX status? */
ieee80211_tx_complete(data->ni, data->m, 0);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
data->m = NULL;
data->ni = NULL;
}
}
static void
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
rsu_bulk_tx_callback_sub(struct usb_xfer *xfer, usb_error_t error,
uint8_t which)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
struct rsu_softc *sc = usbd_xfer_softc(xfer);
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;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct rsu_data *data;
RSU_ASSERT_LOCKED(sc);
switch (USB_GET_STATE(xfer)) {
case USB_ST_TRANSFERRED:
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
data = STAILQ_FIRST(&sc->sc_tx_active[which]);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (data == NULL)
goto tr_setup;
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_TXDONE, "%s: transfer done %p\n",
__func__, data);
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
STAILQ_REMOVE_HEAD(&sc->sc_tx_active[which], next);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu_txeof(xfer, data);
Refactor out the tx buffer free code into a routine, rsu_freebuf(). This makes it easier to add more transmit buffers, have different buffer pools for things, etc.
2015-09-18 07:26:34 +00:00
rsu_freebuf(sc, data);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* FALLTHROUGH */
case USB_ST_SETUP:
tr_setup:
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
data = STAILQ_FIRST(&sc->sc_tx_pending[which]);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (data == NULL) {
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_TXDONE,
"%s: empty pending queue sc %p\n", __func__, sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
return;
}
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
STAILQ_REMOVE_HEAD(&sc->sc_tx_pending[which], next);
STAILQ_INSERT_TAIL(&sc->sc_tx_active[which], data, next);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
usbd_xfer_set_frame_data(xfer, 0, data->buf, data->buflen);
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_TXDONE,
"%s: submitting transfer %p\n",
__func__,
data);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
usbd_transfer_submit(xfer);
break;
default:
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
data = STAILQ_FIRST(&sc->sc_tx_active[which]);
if (data != NULL) {
STAILQ_REMOVE_HEAD(&sc->sc_tx_active[which], next);
rsu_txeof(xfer, data);
Refactor out the tx buffer free code into a routine, rsu_freebuf(). This makes it easier to add more transmit buffers, have different buffer pools for things, etc.
2015-09-18 07:26:34 +00:00
rsu_freebuf(sc, data);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +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
counter_u64_add(ic->ic_oerrors, 1);
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (error != USB_ERR_CANCELLED) {
usbd_xfer_set_stall(xfer);
goto tr_setup;
}
break;
}
modify the rssi logic a bit to actually return a useful rssi. The fullmac firmware doesn't seem to populate a useful rssi indicator in the RX descriptor, so if one plotted said values, they'd basically look like garbage. The reference driver implements a "get current rssi" firmware command which I guess is really meant for station operation only (as hostap operation would need rssi per station, not a single firmware read.) So: * populate sc_currssi during each calibration run; * use this in the RX path instead of trying to reconstruct the RSSI value and passing it around as a pointer; * do up a quick hack to map the rssi hardware value to some useful signal level; * the survey results provide an RSSI value between 0..100, so just do another quick hack to map it into some usefulish signal level; * supply a faked noise floor - I haven't yet found how to pull it out of the firmware. The scan results and the station RSSI information is now more useful for indicating signal strength / distance.
2015-09-30 05:19:16 +00:00
/*
* XXX TODO: if the queue is low, flush out FF TX frames.
* Remember to unlock the driver for now; net80211 doesn't
* defer it for us.
*/
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
static void
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
rsu_bulk_tx_callback_be_bk(struct usb_xfer *xfer, usb_error_t error)
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
{
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
struct rsu_softc *sc = usbd_xfer_softc(xfer);
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
rsu_bulk_tx_callback_sub(xfer, error, RSU_BULK_TX_BE_BK);
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
/* This kicks the TX taskqueue */
rsu_start(sc);
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
}
static void
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
rsu_bulk_tx_callback_vi_vo(struct usb_xfer *xfer, usb_error_t error)
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
{
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
struct rsu_softc *sc = usbd_xfer_softc(xfer);
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
rsu_bulk_tx_callback_sub(xfer, error, RSU_BULK_TX_VI_VO);
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
/* This kicks the TX taskqueue */
rsu_start(sc);
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
}
Use the H2C endpoint for sending firmware commands, rather than the voice data queues. This is similar to the openbsd and rtlwifi/r92su drivers. Note: this driver still assumes it's a 4-endpoint device; I'll enforce that in a follow-up commit.
2015-09-17 03:19:09 +00:00
static void
rsu_bulk_tx_callback_h2c(struct usb_xfer *xfer, usb_error_t error)
{
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
struct rsu_softc *sc = usbd_xfer_softc(xfer);
Use the H2C endpoint for sending firmware commands, rather than the voice data queues. This is similar to the openbsd and rtlwifi/r92su drivers. Note: this driver still assumes it's a 4-endpoint device; I'll enforce that in a follow-up commit.
2015-09-17 03:19:09 +00:00
rsu_bulk_tx_callback_sub(xfer, error, RSU_BULK_TX_H2C);
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
/* This kicks the TX taskqueue */
rsu_start(sc);
Use the H2C endpoint for sending firmware commands, rather than the voice data queues. This is similar to the openbsd and rtlwifi/r92su drivers. Note: this driver still assumes it's a 4-endpoint device; I'll enforce that in a follow-up commit.
2015-09-17 03:19:09 +00:00
}
* Add 11n HT40 support - i needed to send both HTINFO and HTCAP fields in the join message so the firmware would pick it up. * Strip out the direct hardware fiddling for 40MHz mode - the firmware we're using doesn't require it (the rtl8712su firmware does; it is less 'fullmac' than what we're using.) * Fix the mbuf handling during errors - rsu_tx shouldn't free mbufs; it's up to the caller to do so. This brings it in line with what other drivers do or should be doing. Tested: * RTL8712, HT40 channel, STA mode (during this commit)
2015-09-26 07:25:53 +00:00
/*
* Transmit the given frame.
*
* This doesn't free the node or mbuf upon failure.
*/
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static int
rsu_tx_start(struct rsu_softc *sc, struct ieee80211_node *ni,
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
struct mbuf *m0, struct rsu_data *data)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +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;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct ieee80211vap *vap = ni->ni_vap;
struct ieee80211_frame *wh;
struct ieee80211_key *k = NULL;
struct r92s_tx_desc *txd;
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
uint8_t type, cipher;
Bring over the QoS logic from the Linux r92su driver. * the tx descriptor TID is priority, not TID. * the tx descriptor queue id mapping is separate from the TID/priority; rather than just "BE". TODO: * go and re-re-re-verify the queue mappings; the linux and openbsd mappings aren't exactly the same. I need to verify all of this before I try to flip on 11n RX.
2015-09-17 04:45:29 +00:00
int prio = 0;
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
uint8_t which;
int hasqos;
int xferlen;
Bring over the QoS logic from the Linux r92su driver. * the tx descriptor TID is priority, not TID. * the tx descriptor queue id mapping is separate from the TID/priority; rather than just "BE". TODO: * go and re-re-re-verify the queue mappings; the linux and openbsd mappings aren't exactly the same. I need to verify all of this before I try to flip on 11n RX.
2015-09-17 04:45:29 +00:00
int qid;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_ASSERT_LOCKED(sc);
wh = mtod(m0, struct ieee80211_frame *);
type = wh->i_fc[0] & IEEE80211_FC0_TYPE_MASK;
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_TX, "%s: data=%p, m=%p\n",
__func__, data, m0);
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) {
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
k = ieee80211_crypto_encap(ni, m0);
if (k == NULL) {
device_printf(sc->sc_dev,
"ieee80211_crypto_encap returns NULL.\n");
/* XXX we don't expect the fragmented frames */
return (ENOBUFS);
}
wh = mtod(m0, struct ieee80211_frame *);
}
Bring over the QoS logic from the Linux r92su driver. * the tx descriptor TID is priority, not TID. * the tx descriptor queue id mapping is separate from the TID/priority; rather than just "BE". TODO: * go and re-re-re-verify the queue mappings; the linux and openbsd mappings aren't exactly the same. I need to verify all of this before I try to flip on 11n RX.
2015-09-17 04:45:29 +00:00
/* If we have QoS then use it */
/* XXX TODO: mbuf WME/PRI versus TID? */
if (IEEE80211_QOS_HAS_SEQ(wh)) {
/* Has QoS */
prio = M_WME_GETAC(m0);
which = rsu_wme_ac_xfer_map[prio];
hasqos = 1;
} else {
/* Non-QoS TID */
/* XXX TODO: tid=0 for non-qos TID? */
which = rsu_wme_ac_xfer_map[WME_AC_BE];
hasqos = 0;
prio = 0;
}
qid = rsu_ac2qid[prio];
#if 0
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
switch (type) {
case IEEE80211_FC0_TYPE_CTL:
case IEEE80211_FC0_TYPE_MGT:
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
which = rsu_wme_ac_xfer_map[WME_AC_VO];
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
break;
default:
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
which = rsu_wme_ac_xfer_map[M_WME_GETAC(m0)];
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
break;
}
hasqos = 0;
Bring over the QoS logic from the Linux r92su driver. * the tx descriptor TID is priority, not TID. * the tx descriptor queue id mapping is separate from the TID/priority; rather than just "BE". TODO: * go and re-re-re-verify the queue mappings; the linux and openbsd mappings aren't exactly the same. I need to verify all of this before I try to flip on 11n RX.
2015-09-17 04:45:29 +00:00
#endif
Add initial 11n support to if_rsu. * Add a tunable to enable 11n if it's available, so to not anger people who upgrade. kenv hw.usb.rsu.enable_11n=1 before inserting the device. * Add initial 11n htconfig bits; * Enable 40MHz mode if it's available; * Add 11n channels; * Set 11n bits in the firmware. It works for RX; I haven't tested TX aggregation just yet. However the firmware doesn't do RX re-ordering, so I have to tie it into the net80211 A-MPDU RX reorder path before I flip this on by default. I've verified that I'm indeed actually seeing MCS 0->7 rates being received. I haven't dug into whether it's actually transmitting 11n rates; I'll dig into that later.
2015-09-18 04:12:11 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_TX, "%s: pri=%d, which=%d, hasqos=%d\n",
__func__,
prio,
which,
hasqos);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Fill Tx descriptor. */
txd = (struct r92s_tx_desc *)data->buf;
memset(txd, 0, sizeof(*txd));
txd->txdw0 |= htole32(
SM(R92S_TXDW0_PKTLEN, m0->m_pkthdr.len) |
SM(R92S_TXDW0_OFFSET, sizeof(*txd)) |
R92S_TXDW0_OWN | R92S_TXDW0_FSG | R92S_TXDW0_LSG);
txd->txdw1 |= htole32(
Bring over the QoS logic from the Linux r92su driver. * the tx descriptor TID is priority, not TID. * the tx descriptor queue id mapping is separate from the TID/priority; rather than just "BE". TODO: * go and re-re-re-verify the queue mappings; the linux and openbsd mappings aren't exactly the same. I need to verify all of this before I try to flip on 11n RX.
2015-09-17 04:45:29 +00:00
SM(R92S_TXDW1_MACID, R92S_MACID_BSS) | SM(R92S_TXDW1_QSEL, qid));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (!hasqos)
txd->txdw1 |= htole32(R92S_TXDW1_NONQOS);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
if (k != NULL && !(k->wk_flags & IEEE80211_KEY_SWENCRYPT)) {
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
switch (k->wk_cipher->ic_cipher) {
case IEEE80211_CIPHER_WEP:
cipher = R92S_TXDW1_CIPHER_WEP;
break;
case IEEE80211_CIPHER_TKIP:
cipher = R92S_TXDW1_CIPHER_TKIP;
break;
case IEEE80211_CIPHER_AES_CCM:
cipher = R92S_TXDW1_CIPHER_AES;
break;
default:
cipher = R92S_TXDW1_CIPHER_NONE;
}
txd->txdw1 |= htole32(
SM(R92S_TXDW1_CIPHER, cipher) |
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
SM(R92S_TXDW1_KEYIDX, k->wk_keyix));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
Add initial 11n support to if_rsu. * Add a tunable to enable 11n if it's available, so to not anger people who upgrade. kenv hw.usb.rsu.enable_11n=1 before inserting the device. * Add initial 11n htconfig bits; * Enable 40MHz mode if it's available; * Add 11n channels; * Set 11n bits in the firmware. It works for RX; I haven't tested TX aggregation just yet. However the firmware doesn't do RX re-ordering, so I have to tie it into the net80211 A-MPDU RX reorder path before I flip this on by default. I've verified that I'm indeed actually seeing MCS 0->7 rates being received. I haven't dug into whether it's actually transmitting 11n rates; I'll dig into that later.
2015-09-18 04:12:11 +00:00
/* XXX todo: set AGGEN bit if appropriate? */
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
txd->txdw2 |= htole32(R92S_TXDW2_BK);
if (IEEE80211_IS_MULTICAST(wh->i_addr1))
txd->txdw2 |= htole32(R92S_TXDW2_BMCAST);
/*
* Firmware will use and increment the sequence number for the
Bring over the QoS logic from the Linux r92su driver. * the tx descriptor TID is priority, not TID. * the tx descriptor queue id mapping is separate from the TID/priority; rather than just "BE". TODO: * go and re-re-re-verify the queue mappings; the linux and openbsd mappings aren't exactly the same. I need to verify all of this before I try to flip on 11n RX.
2015-09-17 04:45:29 +00:00
* specified priority.
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
*/
Bring over the QoS logic from the Linux r92su driver. * the tx descriptor TID is priority, not TID. * the tx descriptor queue id mapping is separate from the TID/priority; rather than just "BE". TODO: * go and re-re-re-verify the queue mappings; the linux and openbsd mappings aren't exactly the same. I need to verify all of this before I try to flip on 11n RX.
2015-09-17 04:45:29 +00:00
txd->txdw3 |= htole32(SM(R92S_TXDW3_SEQ, prio));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (ieee80211_radiotap_active_vap(vap)) {
struct rsu_tx_radiotap_header *tap = &sc->sc_txtap;
tap->wt_flags = 0;
tap->wt_chan_freq = htole16(ic->ic_curchan->ic_freq);
tap->wt_chan_flags = htole16(ic->ic_curchan->ic_flags);
ieee80211_radiotap_tx(vap, m0);
}
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
xferlen = sizeof(*txd) + m0->m_pkthdr.len;
m_copydata(m0, 0, m0->m_pkthdr.len, (caddr_t)&txd[1]);
data->buflen = xferlen;
data->ni = ni;
data->m = m0;
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
STAILQ_INSERT_TAIL(&sc->sc_tx_pending[which], data, next);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
/* start transfer, if any */
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
usbd_transfer_start(sc->sc_xfer[which]);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
return (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
static int
rsu_transmit(struct ieee80211com *ic, struct mbuf *m)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +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 rsu_softc *sc = ic->ic_softc;
int error;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_LOCK(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_running) {
RSU_UNLOCK(sc);
return (ENXIO);
}
* Add 11n HT40 support - i needed to send both HTINFO and HTCAP fields in the join message so the firmware would pick it up. * Strip out the direct hardware fiddling for 40MHz mode - the firmware we're using doesn't require it (the rtl8712su firmware does; it is less 'fullmac' than what we're using.) * Fix the mbuf handling during errors - rsu_tx shouldn't free mbufs; it's up to the caller to do so. This brings it in line with what other drivers do or should be doing. Tested: * RTL8712, HT40 channel, STA mode (during this commit)
2015-09-26 07:25:53 +00:00
/*
* XXX TODO: ensure that we treat 'm' as a list of frames
* to transmit!
*/
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
error = mbufq_enqueue(&sc->sc_snd, m);
if (error) {
Drain the mbuf queue upon rsu_stop(). Correctly (I hope!) remove net80211 references before doing so. Just doing a dumb mbufq drain isn't enough. If enough traffic occurs and the mbuf queue fills up then transmit stalls (which I'm not fixing in this commit!) but then the mbuf queue stays full until the driver is removed. There's also the net80211 node refcounting leak. This just ensures that during rsu_stop and detach the mbuf queue is purged (and references!) so the queue-full situation can be recovered from.
2015-09-21 02:12:01 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_TX,
"%s: mbufq_enable: failed (%d)\n",
__func__,
error);
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
RSU_UNLOCK(sc);
return (error);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_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
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
/* This kicks the TX taskqueue */
rsu_start(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
return (0);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
Drain the mbuf queue upon rsu_stop(). Correctly (I hope!) remove net80211 references before doing so. Just doing a dumb mbufq drain isn't enough. If enough traffic occurs and the mbuf queue fills up then transmit stalls (which I'm not fixing in this commit!) but then the mbuf queue stays full until the driver is removed. There's also the net80211 node refcounting leak. This just ensures that during rsu_stop and detach the mbuf queue is purged (and references!) so the queue-full situation can be recovered from.
2015-09-21 02:12:01 +00:00
static void
rsu_drain_mbufq(struct rsu_softc *sc)
{
struct mbuf *m;
struct ieee80211_node *ni;
RSU_ASSERT_LOCKED(sc);
while ((m = mbufq_dequeue(&sc->sc_snd)) != NULL) {
ni = (struct ieee80211_node *)m->m_pkthdr.rcvif;
m->m_pkthdr.rcvif = NULL;
ieee80211_free_node(ni);
m_freem(m);
}
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static void
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
_rsu_start(struct rsu_softc *sc)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
struct ieee80211_node *ni;
struct rsu_data *bf;
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
struct mbuf *m;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_ASSERT_LOCKED(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
while ((m = mbufq_dequeue(&sc->sc_snd)) != NULL) {
bf = rsu_getbuf(sc);
if (bf == NULL) {
Drain the mbuf queue upon rsu_stop(). Correctly (I hope!) remove net80211 references before doing so. Just doing a dumb mbufq drain isn't enough. If enough traffic occurs and the mbuf queue fills up then transmit stalls (which I'm not fixing in this commit!) but then the mbuf queue stays full until the driver is removed. There's also the net80211 node refcounting leak. This just ensures that during rsu_stop and detach the mbuf queue is purged (and references!) so the queue-full situation can be recovered from.
2015-09-21 02:12:01 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_TX,
"%s: failed to get buffer\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
mbufq_prepend(&sc->sc_snd, m);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
break;
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
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ni = (struct ieee80211_node *)m->m_pkthdr.rcvif;
m->m_pkthdr.rcvif = 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
if (rsu_tx_start(sc, ni, m, bf) != 0) {
Drain the mbuf queue upon rsu_stop(). Correctly (I hope!) remove net80211 references before doing so. Just doing a dumb mbufq drain isn't enough. If enough traffic occurs and the mbuf queue fills up then transmit stalls (which I'm not fixing in this commit!) but then the mbuf queue stays full until the driver is removed. There's also the net80211 node refcounting leak. This just ensures that during rsu_stop and detach the mbuf queue is purged (and references!) so the queue-full situation can be recovered from.
2015-09-21 02:12:01 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_TX,
"%s: failed to transmit\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_inc_counter(ni->ni_vap->iv_ifp,
IFCOUNTER_OERRORS, 1);
Refactor out the tx buffer free code into a routine, rsu_freebuf(). This makes it easier to add more transmit buffers, have different buffer pools for things, etc.
2015-09-18 07:26:34 +00:00
rsu_freebuf(sc, bf);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
ieee80211_free_node(ni);
* Add 11n HT40 support - i needed to send both HTINFO and HTCAP fields in the join message so the firmware would pick it up. * Strip out the direct hardware fiddling for 40MHz mode - the firmware we're using doesn't require it (the rtl8712su firmware does; it is less 'fullmac' than what we're using.) * Fix the mbuf handling during errors - rsu_tx shouldn't free mbufs; it's up to the caller to do so. This brings it in line with what other drivers do or should be doing. Tested: * RTL8712, HT40 channel, STA mode (during this commit)
2015-09-26 07:25:53 +00:00
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
break;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
}
}
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
static void
rsu_start(struct rsu_softc *sc)
{
taskqueue_enqueue(taskqueue_thread, &sc->tx_task);
}
rsu: fix and enable Rx TCP checksum offloading. Tested with Asus USB-N10, STA mode.
2016-12-11 17:15:25 +00:00
static int
rsu_ioctl_net(struct ieee80211com *ic, u_long cmd, void *data)
{
struct rsu_softc *sc = ic->ic_softc;
struct ifreq *ifr = (struct ifreq *)data;
int error;
error = 0;
switch (cmd) {
case SIOCSIFCAP:
{
struct ieee80211vap *vap;
int rxmask;
rxmask = ifr->ifr_reqcap & (IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6);
RSU_LOCK(sc);
/* Both RXCSUM bits must be set (or unset). */
if (sc->sc_rx_checksum_enable &&
rxmask != (IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6)) {
rxmask = 0;
sc->sc_rx_checksum_enable = 0;
rsu_rxfilter_set(sc, R92S_RCR_TCP_OFFLD_EN, 0);
} else if (!sc->sc_rx_checksum_enable && rxmask != 0) {
rxmask = IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6;
sc->sc_rx_checksum_enable = 1;
rsu_rxfilter_set(sc, 0, R92S_RCR_TCP_OFFLD_EN);
} else {
/* Nothing to do. */
RSU_UNLOCK(sc);
break;
}
RSU_UNLOCK(sc);
IEEE80211_LOCK(ic); /* XXX */
TAILQ_FOREACH(vap, &ic->ic_vaps, iv_next) {
struct ifnet *ifp = vap->iv_ifp;
ifp->if_capenable &=
~(IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6);
ifp->if_capenable |= rxmask;
}
IEEE80211_UNLOCK(ic);
break;
}
default:
error = ENOTTY; /* for net80211 */
break;
}
return (error);
}
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
rsu_parent(struct ieee80211com *ic)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
Further conversions from ifp->if_softc -> ic_softc.
2015-08-20 05:13:54 +00:00
struct rsu_softc *sc = ic->ic_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
if (ic->ic_nrunning > 0) {
rsu: various initialization fixes. - Do not ignore initialization errors; call ieee80211_stop() when initialization failed. - Use usb_pause_mtx() instead of DELAY() while waiting for firmware loading; this fixes system freeze during firmware startup. - Do not execute rsu_stop() when device is powered off; fixes 'unknown board type (rfconfig=0xff)' error when the device is reattached. Tested with Asus USB-N10.
2016-12-10 22:31:49 +00:00
if (rsu_init(sc) == 0)
ieee80211_start_all(ic);
else {
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
if (vap != NULL)
ieee80211_stop(vap);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
rsu: various initialization fixes. - Do not ignore initialization errors; call ieee80211_stop() when initialization failed. - Use usb_pause_mtx() instead of DELAY() while waiting for firmware loading; this fixes system freeze during firmware startup. - Do not execute rsu_stop() when device is powered off; fixes 'unknown board type (rfconfig=0xff)' error when the device is reattached. Tested with Asus USB-N10.
2016-12-10 22:31:49 +00:00
} else
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
rsu_stop(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
/*
* Power on sequence for A-cut adapters.
*/
static void
rsu_power_on_acut(struct rsu_softc *sc)
{
uint32_t reg;
rsu_write_1(sc, R92S_SPS0_CTRL + 1, 0x53);
rsu_write_1(sc, R92S_SPS0_CTRL + 0, 0x57);
/* Enable AFE macro block's bandgap and Mbias. */
rsu_write_1(sc, R92S_AFE_MISC,
rsu_read_1(sc, R92S_AFE_MISC) |
R92S_AFE_MISC_BGEN | R92S_AFE_MISC_MBEN);
/* Enable LDOA15 block. */
rsu_write_1(sc, R92S_LDOA15_CTRL,
rsu_read_1(sc, R92S_LDOA15_CTRL) | R92S_LDA15_EN);
rsu_write_1(sc, R92S_SPS1_CTRL,
rsu_read_1(sc, R92S_SPS1_CTRL) | R92S_SPS1_LDEN);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 2000);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Enable switch regulator block. */
rsu_write_1(sc, R92S_SPS1_CTRL,
rsu_read_1(sc, R92S_SPS1_CTRL) | R92S_SPS1_SWEN);
rsu_write_4(sc, R92S_SPS1_CTRL, 0x00a7b267);
rsu_write_1(sc, R92S_SYS_ISO_CTRL + 1,
rsu_read_1(sc, R92S_SYS_ISO_CTRL + 1) | 0x08);
rsu_write_1(sc, R92S_SYS_FUNC_EN + 1,
rsu_read_1(sc, R92S_SYS_FUNC_EN + 1) | 0x20);
rsu_write_1(sc, R92S_SYS_ISO_CTRL + 1,
rsu_read_1(sc, R92S_SYS_ISO_CTRL + 1) & ~0x90);
/* Enable AFE clock. */
rsu_write_1(sc, R92S_AFE_XTAL_CTRL + 1,
rsu_read_1(sc, R92S_AFE_XTAL_CTRL + 1) & ~0x04);
/* Enable AFE PLL macro block. */
rsu_write_1(sc, R92S_AFE_PLL_CTRL,
rsu_read_1(sc, R92S_AFE_PLL_CTRL) | 0x11);
/* Attach AFE PLL to MACTOP/BB. */
rsu_write_1(sc, R92S_SYS_ISO_CTRL,
rsu_read_1(sc, R92S_SYS_ISO_CTRL) & ~0x11);
/* Switch to 40MHz clock instead of 80MHz. */
rsu_write_2(sc, R92S_SYS_CLKR,
rsu_read_2(sc, R92S_SYS_CLKR) & ~R92S_SYS_CLKSEL);
/* Enable MAC clock. */
rsu_write_2(sc, R92S_SYS_CLKR,
rsu_read_2(sc, R92S_SYS_CLKR) |
R92S_MAC_CLK_EN | R92S_SYS_CLK_EN);
rsu_write_1(sc, R92S_PMC_FSM, 0x02);
/* Enable digital core and IOREG R/W. */
rsu_write_1(sc, R92S_SYS_FUNC_EN + 1,
rsu_read_1(sc, R92S_SYS_FUNC_EN + 1) | 0x08);
rsu_write_1(sc, R92S_SYS_FUNC_EN + 1,
rsu_read_1(sc, R92S_SYS_FUNC_EN + 1) | 0x80);
/* Switch the control path to firmware. */
reg = rsu_read_2(sc, R92S_SYS_CLKR);
reg = (reg & ~R92S_SWHW_SEL) | R92S_FWHW_SEL;
rsu_write_2(sc, R92S_SYS_CLKR, reg);
rsu_write_2(sc, R92S_CR, 0x37fc);
/* Fix USB RX FIFO issue. */
rsu_write_1(sc, 0xfe5c,
rsu_read_1(sc, 0xfe5c) | 0x80);
rsu_write_1(sc, 0x00ab,
rsu_read_1(sc, 0x00ab) | 0xc0);
rsu_write_1(sc, R92S_SYS_CLKR,
rsu_read_1(sc, R92S_SYS_CLKR) & ~R92S_SYS_CPU_CLKSEL);
}
/*
* Power on sequence for B-cut and C-cut adapters.
*/
static void
rsu_power_on_bcut(struct rsu_softc *sc)
{
uint32_t reg;
int ntries;
/* Prevent eFuse leakage. */
rsu_write_1(sc, 0x37, 0xb0);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 10);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu_write_1(sc, 0x37, 0x30);
/* Switch the control path to hardware. */
reg = rsu_read_2(sc, R92S_SYS_CLKR);
if (reg & R92S_FWHW_SEL) {
rsu_write_2(sc, R92S_SYS_CLKR,
reg & ~(R92S_SWHW_SEL | R92S_FWHW_SEL));
}
rsu_write_1(sc, R92S_SYS_FUNC_EN + 1,
rsu_read_1(sc, R92S_SYS_FUNC_EN + 1) & ~0x8c);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu_write_1(sc, R92S_SPS0_CTRL + 1, 0x53);
rsu_write_1(sc, R92S_SPS0_CTRL + 0, 0x57);
reg = rsu_read_1(sc, R92S_AFE_MISC);
rsu_write_1(sc, R92S_AFE_MISC, reg | R92S_AFE_MISC_BGEN);
rsu_write_1(sc, R92S_AFE_MISC, reg | R92S_AFE_MISC_BGEN |
R92S_AFE_MISC_MBEN | R92S_AFE_MISC_I32_EN);
/* Enable PLL. */
rsu_write_1(sc, R92S_LDOA15_CTRL,
rsu_read_1(sc, R92S_LDOA15_CTRL) | R92S_LDA15_EN);
rsu_write_1(sc, R92S_LDOV12D_CTRL,
rsu_read_1(sc, R92S_LDOV12D_CTRL) | R92S_LDV12_EN);
rsu_write_1(sc, R92S_SYS_ISO_CTRL + 1,
rsu_read_1(sc, R92S_SYS_ISO_CTRL + 1) | 0x08);
rsu_write_1(sc, R92S_SYS_FUNC_EN + 1,
rsu_read_1(sc, R92S_SYS_FUNC_EN + 1) | 0x20);
/* Support 64KB IMEM. */
rsu_write_1(sc, R92S_SYS_ISO_CTRL + 1,
rsu_read_1(sc, R92S_SYS_ISO_CTRL + 1) & ~0x97);
/* Enable AFE clock. */
rsu_write_1(sc, R92S_AFE_XTAL_CTRL + 1,
rsu_read_1(sc, R92S_AFE_XTAL_CTRL + 1) & ~0x04);
/* Enable AFE PLL macro block. */
reg = rsu_read_1(sc, R92S_AFE_PLL_CTRL);
rsu_write_1(sc, R92S_AFE_PLL_CTRL, reg | 0x11);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu_write_1(sc, R92S_AFE_PLL_CTRL, reg | 0x51);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu_write_1(sc, R92S_AFE_PLL_CTRL, reg | 0x11);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Attach AFE PLL to MACTOP/BB. */
rsu_write_1(sc, R92S_SYS_ISO_CTRL,
rsu_read_1(sc, R92S_SYS_ISO_CTRL) & ~0x11);
/* Switch to 40MHz clock. */
rsu_write_1(sc, R92S_SYS_CLKR, 0x00);
/* Disable CPU clock and 80MHz SSC. */
rsu_write_1(sc, R92S_SYS_CLKR,
rsu_read_1(sc, R92S_SYS_CLKR) | 0xa0);
/* Enable MAC clock. */
rsu_write_2(sc, R92S_SYS_CLKR,
rsu_read_2(sc, R92S_SYS_CLKR) |
R92S_MAC_CLK_EN | R92S_SYS_CLK_EN);
rsu_write_1(sc, R92S_PMC_FSM, 0x02);
/* Enable digital core and IOREG R/W. */
rsu_write_1(sc, R92S_SYS_FUNC_EN + 1,
rsu_read_1(sc, R92S_SYS_FUNC_EN + 1) | 0x08);
rsu_write_1(sc, R92S_SYS_FUNC_EN + 1,
rsu_read_1(sc, R92S_SYS_FUNC_EN + 1) | 0x80);
/* Switch the control path to firmware. */
reg = rsu_read_2(sc, R92S_SYS_CLKR);
reg = (reg & ~R92S_SWHW_SEL) | R92S_FWHW_SEL;
rsu_write_2(sc, R92S_SYS_CLKR, reg);
rsu_write_2(sc, R92S_CR, 0x37fc);
/* Fix USB RX FIFO issue. */
rsu_write_1(sc, 0xfe5c,
rsu_read_1(sc, 0xfe5c) | 0x80);
rsu_write_1(sc, R92S_SYS_CLKR,
rsu_read_1(sc, R92S_SYS_CLKR) & ~R92S_SYS_CPU_CLKSEL);
rsu_write_1(sc, 0xfe1c, 0x80);
/* Make sure TxDMA is ready to download firmware. */
for (ntries = 0; ntries < 20; ntries++) {
reg = rsu_read_1(sc, R92S_TCR);
if ((reg & (R92S_TCR_IMEM_CHK_RPT | R92S_TCR_EMEM_CHK_RPT)) ==
(R92S_TCR_IMEM_CHK_RPT | R92S_TCR_EMEM_CHK_RPT))
break;
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
if (ntries == 20) {
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RESET | RSU_DEBUG_TX,
"%s: TxDMA is not ready\n",
__func__);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Reset TxDMA. */
reg = rsu_read_1(sc, R92S_CR);
rsu_write_1(sc, R92S_CR, reg & ~R92S_CR_TXDMA_EN);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu_write_1(sc, R92S_CR, reg | R92S_CR_TXDMA_EN);
}
}
static void
rsu_power_off(struct rsu_softc *sc)
{
/* Turn RF off. */
rsu_write_1(sc, R92S_RF_CTRL, 0x00);
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 5);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Turn MAC off. */
/* Switch control path. */
rsu_write_1(sc, R92S_SYS_CLKR + 1, 0x38);
/* Reset MACTOP. */
rsu_write_1(sc, R92S_SYS_FUNC_EN + 1, 0x70);
rsu_write_1(sc, R92S_PMC_FSM, 0x06);
rsu_write_1(sc, R92S_SYS_ISO_CTRL + 0, 0xf9);
rsu_write_1(sc, R92S_SYS_ISO_CTRL + 1, 0xe8);
/* Disable AFE PLL. */
rsu_write_1(sc, R92S_AFE_PLL_CTRL, 0x00);
/* Disable A15V. */
rsu_write_1(sc, R92S_LDOA15_CTRL, 0x54);
/* Disable eFuse 1.2V. */
rsu_write_1(sc, R92S_SYS_FUNC_EN + 1, 0x50);
rsu_write_1(sc, R92S_LDOV12D_CTRL, 0x24);
/* Enable AFE macro block's bandgap and Mbias. */
rsu_write_1(sc, R92S_AFE_MISC, 0x30);
/* Disable 1.6V LDO. */
rsu_write_1(sc, R92S_SPS0_CTRL + 0, 0x56);
rsu_write_1(sc, R92S_SPS0_CTRL + 1, 0x43);
Send a power command to the firmware to shut down the radio as well during rsu_stop().
2015-09-22 05:48:51 +00:00
/* Firmware - tell it to switch things off */
(void) rsu_set_fw_power_state(sc, RSU_PWR_OFF);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
static int
Unbreak sparc64 LINT. Need to fix this correctly at some point in the future.
2013-07-30 16:17:30 +00:00
rsu_fw_loadsection(struct rsu_softc *sc, const uint8_t *buf, int len)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
const uint8_t which = rsu_wme_ac_xfer_map[WME_AC_VO];
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct rsu_data *data;
struct r92s_tx_desc *txd;
int mlen;
while (len > 0) {
data = rsu_getbuf(sc);
if (data == NULL)
return (ENOMEM);
txd = (struct r92s_tx_desc *)data->buf;
memset(txd, 0, sizeof(*txd));
if (len <= RSU_TXBUFSZ - sizeof(*txd)) {
/* Last chunk. */
txd->txdw0 |= htole32(R92S_TXDW0_LINIP);
mlen = len;
} else
mlen = RSU_TXBUFSZ - sizeof(*txd);
txd->txdw0 |= htole32(SM(R92S_TXDW0_PKTLEN, mlen));
memcpy(&txd[1], buf, mlen);
data->buflen = sizeof(*txd) + mlen;
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_TX | RSU_DEBUG_FW | RSU_DEBUG_RESET,
"%s: starting transfer %p\n",
__func__, data);
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
STAILQ_INSERT_TAIL(&sc->sc_tx_pending[which], data, next);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
buf += mlen;
len -= mlen;
}
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
usbd_transfer_start(sc->sc_xfer[which]);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
return (0);
}
Properly range check length of parsed information elements in RSU driver. Found by: Ilja van Sprundel <ivansprundel@ioactive.com> MFC after: 3 days Sponsored by: Mellanox Technologies
2017-07-30 16:45:28 +00:00
CTASSERT(sizeof(size_t) >= sizeof(uint32_t));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
static int
rsu_load_firmware(struct rsu_softc *sc)
{
Unbreak sparc64 LINT. Need to fix this correctly at some point in the future.
2013-07-30 16:17:30 +00:00
const struct r92s_fw_hdr *hdr;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct r92s_fw_priv *dmem;
Add initial 11n support to if_rsu. * Add a tunable to enable 11n if it's available, so to not anger people who upgrade. kenv hw.usb.rsu.enable_11n=1 before inserting the device. * Add initial 11n htconfig bits; * Enable 40MHz mode if it's available; * Add 11n channels; * Set 11n bits in the firmware. It works for RX; I haven't tested TX aggregation just yet. However the firmware doesn't do RX re-ordering, so I have to tie it into the net80211 A-MPDU RX reorder path before I flip this on by default. I've verified that I'm indeed actually seeing MCS 0->7 rates being received. I haven't dug into whether it's actually transmitting 11n rates; I'll dig into that later.
2015-09-18 04:12:11 +00:00
struct ieee80211com *ic = &sc->sc_ic;
Unbreak sparc64 LINT. Need to fix this correctly at some point in the future.
2013-07-30 16:17:30 +00:00
const uint8_t *imem, *emem;
Properly range check length of parsed information elements in RSU driver. Found by: Ilja van Sprundel <ivansprundel@ioactive.com> MFC after: 3 days Sponsored by: Mellanox Technologies
2017-07-30 16:45:28 +00:00
uint32_t imemsz, ememsz;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
const struct firmware *fw;
size_t size;
uint32_t reg;
int ntries, error;
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
if (rsu_read_1(sc, R92S_TCR) & R92S_TCR_FWRDY) {
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_ANY,
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
"%s: Firmware already loaded\n",
__func__);
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
return (0);
}
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_UNLOCK(sc);
/* Read firmware image from the filesystem. */
if ((fw = firmware_get("rsu-rtl8712fw")) == NULL) {
device_printf(sc->sc_dev,
"%s: failed load firmware of file rsu-rtl8712fw\n",
__func__);
RSU_LOCK(sc);
return (ENXIO);
}
RSU_LOCK(sc);
size = fw->datasize;
if (size < sizeof(*hdr)) {
device_printf(sc->sc_dev, "firmware too short\n");
error = EINVAL;
goto fail;
}
Unbreak sparc64 LINT. Need to fix this correctly at some point in the future.
2013-07-30 16:17:30 +00:00
hdr = (const struct r92s_fw_hdr *)fw->data;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (hdr->signature != htole16(0x8712) &&
hdr->signature != htole16(0x8192)) {
device_printf(sc->sc_dev,
"invalid firmware signature 0x%x\n",
le16toh(hdr->signature));
error = EINVAL;
goto fail;
}
rsu: use bitmap for all debug messages. - Replace all remaining DPRINTF(N)'s with RSU_DPRINTF. - Add new RSU_DEBUG_USB flag to track error codes returned by usbd_do_request_flags(). - Improve few messages.
2016-12-10 20:19:57 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_FW, "FW V%d %02x-%02x %02x:%02x\n",
le16toh(hdr->version), hdr->month, hdr->day, hdr->hour,
hdr->minute);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Make sure that driver and firmware are in sync. */
if (hdr->privsz != htole32(sizeof(*dmem))) {
device_printf(sc->sc_dev, "unsupported firmware image\n");
error = EINVAL;
goto fail;
}
/* Get FW sections sizes. */
imemsz = le32toh(hdr->imemsz);
ememsz = le32toh(hdr->sramsz);
/* Check that all FW sections fit in image. */
Properly range check length of parsed information elements in RSU driver. Found by: Ilja van Sprundel <ivansprundel@ioactive.com> MFC after: 3 days Sponsored by: Mellanox Technologies
2017-07-30 16:45:28 +00:00
if (imemsz > (size_t)(size - sizeof(*hdr)) ||
ememsz > (size_t)(size - sizeof(*hdr) - imemsz)) {
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
device_printf(sc->sc_dev, "firmware too short\n");
error = EINVAL;
goto fail;
}
Unbreak sparc64 LINT. Need to fix this correctly at some point in the future.
2013-07-30 16:17:30 +00:00
imem = (const uint8_t *)&hdr[1];
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
emem = imem + imemsz;
/* Load IMEM section. */
error = rsu_fw_loadsection(sc, imem, imemsz);
if (error != 0) {
device_printf(sc->sc_dev,
"could not load firmware section %s\n", "IMEM");
goto fail;
}
/* Wait for load to complete. */
- Stop transfers when RSU init fails. - Make sure TX/RX lists don't leak and are only allocated once. - Fix off-by one transfer index computation. - Give firmware loading more time. MFC after: 3 days
2014-05-22 06:28:09 +00:00
for (ntries = 0; ntries != 50; ntries++) {
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 10);
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
reg = rsu_read_1(sc, R92S_TCR);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (reg & R92S_TCR_IMEM_CODE_DONE)
break;
}
- Stop transfers when RSU init fails. - Make sure TX/RX lists don't leak and are only allocated once. - Fix off-by one transfer index computation. - Give firmware loading more time. MFC after: 3 days
2014-05-22 06:28:09 +00:00
if (ntries == 50) {
device_printf(sc->sc_dev, "timeout waiting for IMEM transfer\n");
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
error = ETIMEDOUT;
goto fail;
}
/* Load EMEM section. */
error = rsu_fw_loadsection(sc, emem, ememsz);
if (error != 0) {
device_printf(sc->sc_dev,
"could not load firmware section %s\n", "EMEM");
goto fail;
}
/* Wait for load to complete. */
- Give firmware loading more time. MFC after: 3 days
2014-05-22 12:01:43 +00:00
for (ntries = 0; ntries != 50; ntries++) {
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 10);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
reg = rsu_read_2(sc, R92S_TCR);
if (reg & R92S_TCR_EMEM_CODE_DONE)
break;
}
- Give firmware loading more time. MFC after: 3 days
2014-05-22 12:01:43 +00:00
if (ntries == 50) {
- Stop transfers when RSU init fails. - Make sure TX/RX lists don't leak and are only allocated once. - Fix off-by one transfer index computation. - Give firmware loading more time. MFC after: 3 days
2014-05-22 06:28:09 +00:00
device_printf(sc->sc_dev, "timeout waiting for EMEM transfer\n");
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
error = ETIMEDOUT;
goto fail;
}
/* Enable CPU. */
rsu_write_1(sc, R92S_SYS_CLKR,
rsu_read_1(sc, R92S_SYS_CLKR) | R92S_SYS_CPU_CLKSEL);
if (!(rsu_read_1(sc, R92S_SYS_CLKR) & R92S_SYS_CPU_CLKSEL)) {
device_printf(sc->sc_dev, "could not enable system clock\n");
error = EIO;
goto fail;
}
rsu_write_2(sc, R92S_SYS_FUNC_EN,
rsu_read_2(sc, R92S_SYS_FUNC_EN) | R92S_FEN_CPUEN);
if (!(rsu_read_2(sc, R92S_SYS_FUNC_EN) & R92S_FEN_CPUEN)) {
device_printf(sc->sc_dev,
"could not enable microcontroller\n");
error = EIO;
goto fail;
}
/* Wait for CPU to initialize. */
for (ntries = 0; ntries < 100; ntries++) {
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
if (rsu_read_1(sc, R92S_TCR) & R92S_TCR_IMEM_RDY)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
break;
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
if (ntries == 100) {
device_printf(sc->sc_dev,
"timeout waiting for microcontroller\n");
error = ETIMEDOUT;
goto fail;
}
/* Update DMEM section before loading. */
Unbreak sparc64 LINT. Need to fix this correctly at some point in the future.
2013-07-30 16:17:30 +00:00
dmem = __DECONST(struct r92s_fw_priv *, &hdr->priv);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
memset(dmem, 0, sizeof(*dmem));
dmem->hci_sel = R92S_HCI_SEL_USB | R92S_HCI_SEL_8172;
Program the firmware setup stuff with the current hardware setup: * Do 1T1R for now, until we read the config out of ROM and use it. * Disable turbo mode, I dunno what this is, but the linux drivers have this disabled. * Set the firmware endpoints to what we read from USB. Tested: * RTL8712 cut 3, STA mode
2015-09-17 03:42:18 +00:00
dmem->nendpoints = sc->sc_nendpoints;
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
dmem->chip_version = sc->cut;
rsu(4): Add support for 1T2R and 2T2R NICs. This logic is mostly crimed from the reference driver and the linux r92su driver. I verified that it (a) worked on the rsu hardware I have, and (b) did traffic testing whilst watching what ath(4) sent as a hostap. It successfully sent MCS8..15 rates (which requires 2-stream reception) as well as MCS0..7 (which is 1-stream.) Tested: * RTL8712, 1T1R NIC, MCS rates 0..7. * RTL8712, 1T2R NIC, MCS rates 0..15 TODO: * Find a 2T2R NIC!
2015-09-29 06:56:00 +00:00
dmem->rf_config = sc->sc_rftype;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
dmem->vcs_type = R92S_VCS_TYPE_AUTO;
dmem->vcs_mode = R92S_VCS_MODE_RTS_CTS;
Program the firmware setup stuff with the current hardware setup: * Do 1T1R for now, until we read the config out of ROM and use it. * Disable turbo mode, I dunno what this is, but the linux drivers have this disabled. * Set the firmware endpoints to what we read from USB. Tested: * RTL8712 cut 3, STA mode
2015-09-17 03:42:18 +00:00
dmem->turbo_mode = 0;
Add initial 11n support to if_rsu. * Add a tunable to enable 11n if it's available, so to not anger people who upgrade. kenv hw.usb.rsu.enable_11n=1 before inserting the device. * Add initial 11n htconfig bits; * Enable 40MHz mode if it's available; * Add 11n channels; * Set 11n bits in the firmware. It works for RX; I haven't tested TX aggregation just yet. However the firmware doesn't do RX re-ordering, so I have to tie it into the net80211 A-MPDU RX reorder path before I flip this on by default. I've verified that I'm indeed actually seeing MCS 0->7 rates being received. I haven't dug into whether it's actually transmitting 11n rates; I'll dig into that later.
2015-09-18 04:12:11 +00:00
dmem->bw40_en = !! (ic->ic_htcaps & IEEE80211_HTCAP_CHWIDTH40);
dmem->amsdu2ampdu_en = !! (sc->sc_ht);
dmem->ampdu_en = !! (sc->sc_ht);
dmem->agg_offload = !! (sc->sc_ht);
.. oops, flip on QoS.
2015-09-17 07:04:15 +00:00
dmem->qos_en = 1;
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
dmem->ps_offload = 1;
dmem->lowpower_mode = 1; /* XXX TODO: configurable? */
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Load DMEM section. */
error = rsu_fw_loadsection(sc, (uint8_t *)dmem, sizeof(*dmem));
if (error != 0) {
device_printf(sc->sc_dev,
"could not load firmware section %s\n", "DMEM");
goto fail;
}
/* Wait for load to complete. */
for (ntries = 0; ntries < 100; ntries++) {
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
if (rsu_read_1(sc, R92S_TCR) & R92S_TCR_DMEM_CODE_DONE)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
break;
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
if (ntries == 100) {
device_printf(sc->sc_dev, "timeout waiting for %s transfer\n",
"DMEM");
error = ETIMEDOUT;
goto fail;
}
/* Wait for firmware readiness. */
for (ntries = 0; ntries < 60; ntries++) {
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
if (!(rsu_read_1(sc, R92S_TCR) & R92S_TCR_FWRDY))
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
break;
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(sc, 1);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
if (ntries == 60) {
device_printf(sc->sc_dev,
"timeout waiting for firmware readiness\n");
error = ETIMEDOUT;
goto fail;
}
fail:
firmware_put(fw, FIRMWARE_UNLOAD);
return (error);
}
static int
rsu_raw_xmit(struct ieee80211_node *ni, struct mbuf *m,
const struct ieee80211_bpf_params *params)
{
struct ieee80211com *ic = ni->ni_ic;
Further conversions from ifp->if_softc -> ic_softc.
2015-08-20 05:13:54 +00:00
struct rsu_softc *sc = ic->ic_softc;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
struct rsu_data *bf;
/* prevent management frames from being sent if we're not ready */
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_running) {
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
m_freem(m);
return (ENETDOWN);
}
RSU_LOCK(sc);
bf = rsu_getbuf(sc);
if (bf == NULL) {
m_freem(m);
RSU_UNLOCK(sc);
return (ENOBUFS);
}
- Split transmit queue into one for each type. Apparently there will be a race when using a single active queue for all transmit types. - Last argument of usb_pause_mtx() is ticks and not milliseconds. - Remove unused watchdog. - Remove some unused fields from the RSU softc structure. - Workaround usbd_transfer_start() recursion from inside of completion callback. MFC after: 3 days
2014-05-21 16:52:55 +00:00
if (rsu_tx_start(sc, ni, m, bf) != 0) {
* Add 11n HT40 support - i needed to send both HTINFO and HTCAP fields in the join message so the firmware would pick it up. * Strip out the direct hardware fiddling for 40MHz mode - the firmware we're using doesn't require it (the rtl8712su firmware does; it is less 'fullmac' than what we're using.) * Fix the mbuf handling during errors - rsu_tx shouldn't free mbufs; it's up to the caller to do so. This brings it in line with what other drivers do or should be doing. Tested: * RTL8712, HT40 channel, STA mode (during this commit)
2015-09-26 07:25:53 +00:00
m_freem(m);
Refactor out the tx buffer free code into a routine, rsu_freebuf(). This makes it easier to add more transmit buffers, have different buffer pools for things, etc.
2015-09-18 07:26:34 +00:00
rsu_freebuf(sc, bf);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
RSU_UNLOCK(sc);
return (EIO);
}
RSU_UNLOCK(sc);
return (0);
}
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
static void
rsu_rxfilter_init(struct rsu_softc *sc)
{
uint32_t reg;
RSU_ASSERT_LOCKED(sc);
/* Setup multicast filter. */
rsu_set_multi(sc);
/* Adjust Rx filter. */
reg = rsu_read_4(sc, R92S_RCR);
reg &= ~R92S_RCR_AICV;
reg |= R92S_RCR_APP_PHYSTS;
rsu: fix and enable Rx TCP checksum offloading. Tested with Asus USB-N10, STA mode.
2016-12-11 17:15:25 +00:00
if (sc->sc_rx_checksum_enable)
reg |= R92S_RCR_TCP_OFFLD_EN;
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
rsu_write_4(sc, R92S_RCR, reg);
/* Update dynamic Rx filter parts. */
rsu_rxfilter_refresh(sc);
}
static void
rsu_rxfilter_set(struct rsu_softc *sc, uint32_t clear, uint32_t set)
{
/* NB: firmware can touch this register too. */
rsu_write_4(sc, R92S_RCR,
(rsu_read_4(sc, R92S_RCR) & ~clear) | set);
}
static void
rsu_rxfilter_refresh(struct rsu_softc *sc)
{
struct ieee80211com *ic = &sc->sc_ic;
uint32_t mask_all, mask_min;
RSU_ASSERT_LOCKED(sc);
/* NB: RCR_AMF / RXFLTMAP_MGT are used by firmware. */
mask_all = R92S_RCR_ACF | R92S_RCR_AAP;
mask_min = R92S_RCR_APM;
if (sc->sc_vap_is_running)
mask_min |= R92S_RCR_CBSSID;
else
mask_all |= R92S_RCR_ADF;
if (ic->ic_opmode == IEEE80211_M_MONITOR) {
uint16_t rxfltmap;
if (sc->sc_vap_is_running)
rxfltmap = 0;
else
rxfltmap = R92S_RXFLTMAP_MGT_DEF;
rsu_write_2(sc, R92S_RXFLTMAP_MGT, rxfltmap);
}
if (ic->ic_promisc == 0 && ic->ic_opmode != IEEE80211_M_MONITOR)
rsu_rxfilter_set(sc, mask_all, mask_min);
else
rsu_rxfilter_set(sc, mask_min, mask_all);
}
rsu: various initialization fixes. - Do not ignore initialization errors; call ieee80211_stop() when initialization failed. - Use usb_pause_mtx() instead of DELAY() while waiting for firmware loading; this fixes system freeze during firmware startup. - Do not execute rsu_stop() when device is powered off; fixes 'unknown board type (rfconfig=0xff)' error when the device is reattached. Tested with Asus USB-N10.
2016-12-10 22:31:49 +00:00
static int
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
rsu_init(struct rsu_softc *sc)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +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;
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
uint8_t macaddr[IEEE80211_ADDR_LEN];
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
int error;
- Stop transfers when RSU init fails. - Make sure TX/RX lists don't leak and are only allocated once. - Fix off-by one transfer index computation. - Give firmware loading more time. MFC after: 3 days
2014-05-22 06:28:09 +00:00
int i;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
rsu: various initialization fixes. - Do not ignore initialization errors; call ieee80211_stop() when initialization failed. - Use usb_pause_mtx() instead of DELAY() while waiting for firmware loading; this fixes system freeze during firmware startup. - Do not execute rsu_stop() when device is powered off; fixes 'unknown board type (rfconfig=0xff)' error when the device is reattached. Tested with Asus USB-N10.
2016-12-10 22:31:49 +00:00
RSU_LOCK(sc);
if (sc->sc_running) {
RSU_UNLOCK(sc);
return (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
Drain the mbuf queue upon rsu_stop(). Correctly (I hope!) remove net80211 references before doing so. Just doing a dumb mbufq drain isn't enough. If enough traffic occurs and the mbuf queue fills up then transmit stalls (which I'm not fixing in this commit!) but then the mbuf queue stays full until the driver is removed. There's also the net80211 node refcounting leak. This just ensures that during rsu_stop and detach the mbuf queue is purged (and references!) so the queue-full situation can be recovered from.
2015-09-21 02:12:01 +00:00
/* Ensure the mbuf queue is drained */
rsu_drain_mbufq(sc);
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
/* Reset power management state. */
rsu_write_1(sc, R92S_USB_HRPWM, 0);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Power on adapter. */
if (sc->cut == 1)
rsu_power_on_acut(sc);
else
rsu_power_on_bcut(sc);
Fixes for the RSU driver: - The R92S_TCR register is an 8-bit register. Don't access it like a 16-bit register. - Disable parsing the delete station event, due to many false events. - Ensure that there is only one transfer queue for each endpoint, so that packets transmitted don't get out of order. MFC after: 1 week
2014-06-04 09:18:13 +00:00
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Load firmware. */
error = rsu_load_firmware(sc);
if (error != 0)
goto fail;
rsu_write_4(sc, R92S_CR,
rsu_read_4(sc, R92S_CR) & ~0xff000000);
/* Use 128 bytes pages. */
rsu_write_1(sc, 0x00b5,
rsu_read_1(sc, 0x00b5) | 0x01);
/* Enable USB Rx aggregation. */
rsu_write_1(sc, 0x00bd,
rsu_read_1(sc, 0x00bd) | 0x80);
/* Set USB Rx aggregation threshold. */
rsu_write_1(sc, 0x00d9, 0x01);
/* Set USB Rx aggregation timeout (1.7ms/4). */
rsu_write_1(sc, 0xfe5b, 0x04);
/* Fix USB Rx FIFO issue. */
rsu_write_1(sc, 0xfe5c,
rsu_read_1(sc, 0xfe5c) | 0x80);
/* Set MAC address. */
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(macaddr, vap ? vap->iv_myaddr : ic->ic_macaddr);
rsu_write_region_1(sc, R92S_MACID, macaddr, IEEE80211_ADDR_LEN);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
- The DELAY() should not be used in USB drivers. - The usb_pause_mtx() function takes ticks and not milliseconds as last argument. MFC after: 3 days
2014-05-20 15:47:37 +00:00
/* It really takes 1.5 seconds for the firmware to boot: */
rsu: various initialization fixes. - Do not ignore initialization errors; call ieee80211_stop() when initialization failed. - Use usb_pause_mtx() instead of DELAY() while waiting for firmware loading; this fixes system freeze during firmware startup. - Do not execute rsu_stop() when device is powered off; fixes 'unknown board type (rfconfig=0xff)' error when the device is reattached. Tested with Asus USB-N10.
2016-12-10 22:31:49 +00:00
usb_pause_mtx(&sc->sc_mtx, USB_MS_TO_TICKS(2000));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if_rsu debug fixes: * use an ath/iwn style debug bitmap - it's still global rather than per-device, but it's better than debug levels * disable bgscan - it just makes things unstable/unpredictable for now. Tested: * if_rsu - RTL8712 cut 3, STA mode
2015-09-12 23:10:34 +00:00
RSU_DPRINTF(sc, RSU_DEBUG_RESET, "%s: setting MAC address to %s\n",
__func__,
ether_sprintf(macaddr));
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
error = rsu_fw_cmd(sc, R92S_CMD_SET_MAC_ADDRESS, macaddr,
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
IEEE80211_ADDR_LEN);
if (error != 0) {
device_printf(sc->sc_dev, "could not set MAC address\n");
goto fail;
}
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
/* Initialize Rx filter. */
rsu_rxfilter_init(sc);
rsu: add support for hardware multicast filter setup. The algorithm is the same as in rtwn(4). Tested with Asus USB-N10 (STA) + RTL8188CUS (AP).
2016-11-27 18:06:03 +00:00
After reading the linux driver, document what this bit is doing (doing a full NIC awake.)
2015-09-13 04:41:13 +00:00
/* Set PS mode fully active */
Begin fleshing out basic power-on / power-off and A-MPDU TX support. * Add a new method to control NIC poweron / network-sleep / power off; * Add in A-MPDU TX negotiation support, but comment it out because it does break TX traffic; * blank out the tx buffer before sending a firmware message, just in case; * go into network-sleep once associated; TODO: * figure out why ampdu negotiation isn't working and breaking TX traffic, then enable it.
2015-09-22 02:57:18 +00:00
error = rsu_set_fw_power_state(sc, RSU_PWR_ACTIVE);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
if (error != 0) {
device_printf(sc->sc_dev, "could not set PS mode\n");
goto fail;
}
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
/* Install static keys (if any). */
error = rsu_reinit_static_keys(sc);
if (error != 0)
goto fail;
rsu: various scanning fixes. - Set IEEE80211_FEXT_SCAN_OFFLOAD flag; firmware can send null data frames when associated. - Check IEEE80211_SCAN_ACTIVE scan flag instead of IEEE80211_F_ASCAN ic flag; the last is never set since r170530. - Eliminate software scan (net80211) <-> site_survey (driver) race: * override ic_scan_curchan and ic_scan_mindwell pointers so net80211 will not try to finish scanning automatically; * inform net80211 about current status via ieee80211_cancel_scan() and ieee80211_scan_done(); * remove corresponding workaround from rsu_join_bss(). Now the driver can associate to an AP with hidden SSID. Tested with Asus USB-N10.
2016-11-27 12:03:34 +00:00
sc->sc_extra_scan = 0;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
usbd_transfer_start(sc->sc_xfer[RSU_BULK_RX]);
/* We're ready to go. */
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_running = 1;
rsu: various initialization fixes. - Do not ignore initialization errors; call ieee80211_stop() when initialization failed. - Use usb_pause_mtx() instead of DELAY() while waiting for firmware loading; this fixes system freeze during firmware startup. - Do not execute rsu_stop() when device is powered off; fixes 'unknown board type (rfconfig=0xff)' error when the device is reattached. Tested with Asus USB-N10.
2016-12-10 22:31:49 +00:00
RSU_UNLOCK(sc);
return (0);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
fail:
- Stop transfers when RSU init fails. - Make sure TX/RX lists don't leak and are only allocated once. - Fix off-by one transfer index computation. - Give firmware loading more time. MFC after: 3 days
2014-05-22 06:28:09 +00:00
/* Need to stop all failed transfers, if any */
for (i = 0; i != RSU_N_TRANSFER; i++)
usbd_transfer_stop(sc->sc_xfer[i]);
rsu: various initialization fixes. - Do not ignore initialization errors; call ieee80211_stop() when initialization failed. - Use usb_pause_mtx() instead of DELAY() while waiting for firmware loading; this fixes system freeze during firmware startup. - Do not execute rsu_stop() when device is powered off; fixes 'unknown board type (rfconfig=0xff)' error when the device is reattached. Tested with Asus USB-N10.
2016-12-10 22:31:49 +00:00
RSU_UNLOCK(sc);
return (error);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +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
rsu_stop(struct rsu_softc *sc)
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
{
int i;
rsu: various initialization fixes. - Do not ignore initialization errors; call ieee80211_stop() when initialization failed. - Use usb_pause_mtx() instead of DELAY() while waiting for firmware loading; this fixes system freeze during firmware startup. - Do not execute rsu_stop() when device is powered off; fixes 'unknown board type (rfconfig=0xff)' error when the device is reattached. Tested with Asus USB-N10.
2016-12-10 22:31:49 +00:00
RSU_LOCK(sc);
if (!sc->sc_running) {
RSU_UNLOCK(sc);
return;
}
Send a power command to the firmware to shut down the radio as well during rsu_stop().
2015-09-22 05:48:51 +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_running = 0;
rsu: add promiscuous mode support. - Add partial promiscuous mode support (no management frames; they cannot be received by the firmware and net80211 at the same time). - Add monitor mode support (all frames). Tested with Asus, USB-N10.
2016-12-10 18:47:13 +00:00
sc->sc_vap_is_running = 0;
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
sc->sc_calibrating = 0;
taskqueue_cancel_timeout(taskqueue_thread, &sc->calib_task, NULL);
Convert if_rsu to use a deferred transmit task rather than using rsu_start() to do it directly. Ensure that we re-queue starting transmit upon TX completion. This solves two issues: * It stops tx stalls - before this, if the transmit path filled the mbuf queue then it'd never start another transmit. * It enforces ordering - this is very required for 802.11n which requires frames to be transmitted in the order they're queued. Since everything remotely involved in USB has an unlock/thing/relock pattern with that mutex, the only way to guarantee TX ordering is to 100% defer it into a separate thread. This now survives an iperf test and gets a reliable 30mbit/sec.
2015-09-21 02:30:22 +00:00
taskqueue_cancel(taskqueue_thread, &sc->tx_task, NULL);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
/* Power off adapter. */
rsu_power_off(sc);
rsu: add hardware crypto support (WEP, TKIP and CCMP). This change includes firmware commands for key setup + some additional checking via CAMREAD / CAMWRITE registers. Nothing (except rsu_delete_key() for pairwise keys) is deferred; to ensure that things are done in order rsu_set_key() will wait until key deletion task will be finished. Tested with Asus USB-N10 (all ciphers). Differences from initial (reviewed) patch: - Pause AC queues before disassociation - since CMD_DISCONNECT clears crypto state all pending frames must be processed / dropped before it. - Check sc_running flag before trying to set static keys. - Clear key index from bitmap even when firmware command fails (it will be invalidated via CAMWRITE anyway). Reviewed by: adrian, kevlo Tested by: kevlo Differential Revision: https://reviews.freebsd.org/D8706
2016-12-06 00:13:49 +00:00
/*
* CAM is not accessible after shutdown;
* all entries are marked (by firmware?) as invalid.
*/
memset(sc->free_keys_bmap, 0, sizeof(sc->free_keys_bmap));
memset(sc->keys_bmap, 0, sizeof(sc->keys_bmap));
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
for (i = 0; i < RSU_N_TRANSFER; i++)
usbd_transfer_stop(sc->sc_xfer[i]);
Drain the mbuf queue upon rsu_stop(). Correctly (I hope!) remove net80211 references before doing so. Just doing a dumb mbufq drain isn't enough. If enough traffic occurs and the mbuf queue fills up then transmit stalls (which I'm not fixing in this commit!) but then the mbuf queue stays full until the driver is removed. There's also the net80211 node refcounting leak. This just ensures that during rsu_stop and detach the mbuf queue is purged (and references!) so the queue-full situation can be recovered from.
2015-09-21 02:12:01 +00:00
/* Ensure the mbuf queue is drained */
rsu_drain_mbufq(sc);
rsu: various initialization fixes. - Do not ignore initialization errors; call ieee80211_stop() when initialization failed. - Use usb_pause_mtx() instead of DELAY() while waiting for firmware loading; this fixes system freeze during firmware startup. - Do not execute rsu_stop() when device is powered off; fixes 'unknown board type (rfconfig=0xff)' error when the device is reattached. Tested with Asus USB-N10.
2016-12-10 22:31:49 +00:00
RSU_UNLOCK(sc);
Import OpenBSD's rsu(4) WLAN driver. Support chipsets are the Realtek RTL8188SU, RTL8191SU, and RTL8192SU. Many thanks to Idwer Vollering for porting/writing the man page and for testing. Reviewed by: adrian, hselasky Obtained from: OpenBSD Tested by: kevlo, Idwer Vollering <vidwer at gmail.com>
2013-07-30 02:07:57 +00:00
}
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
/*
* Note: usb_pause_mtx() actually releases the mutex before calling pause(),
* which breaks any kind of driver serialisation.
*/
- The DELAY() should not be used in USB drivers. - The usb_pause_mtx() function takes ticks and not milliseconds as last argument. MFC after: 3 days
2014-05-20 15:47:37 +00:00
static void
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
rsu_ms_delay(struct rsu_softc *sc, int ms)
- The DELAY() should not be used in USB drivers. - The usb_pause_mtx() function takes ticks and not milliseconds as last argument. MFC after: 3 days
2014-05-20 15:47:37 +00:00
{
Use DELAY() rather than usb_pause_mtx() - the latter releases the lock before waiting, which prevents the lock from really acting like a hardware serialiser. Sigh.
2015-09-17 03:01:19 +00:00
//usb_pause_mtx(&sc->sc_mtx, hz / 1000);
DELAY(ms * 1000);
- The DELAY() should not be used in USB drivers. - The usb_pause_mtx() function takes ticks and not milliseconds as last argument. MFC after: 3 days
2014-05-20 15:47:37 +00:00
}
Reference in New Issue Copy Permalink