freebsd-nq/sys/dev/usb/wlan/if_uath.c

2880 lines
77 KiB
C
Raw Normal View History

/*-
* SPDX-License-Identifier: (BSD-2-Clause-FreeBSD AND BSD-1-Clause)
*
* Copyright (c) 2006 Sam Leffler, Errno Consulting
* Copyright (c) 2008-2009 Weongyo Jeong <weongyo@freebsd.org>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer,
* without modification.
* 2. Redistributions in binary form must reproduce at minimum a disclaimer
* similar to the "NO WARRANTY" disclaimer below ("Disclaimer") and any
* redistribution must be conditioned upon including a substantially
* similar Disclaimer requirement for further binary redistribution.
*
* NO WARRANTY
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF NONINFRINGEMENT, MERCHANTIBILITY
* AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL
* THE COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR SPECIAL, EXEMPLARY,
* OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
* IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
* THE POSSIBILITY OF SUCH DAMAGES.
*/
/*
* This driver is distantly derived from a driver of the same name
* by Damien Bergamini. The original copyright is included below:
*
* Copyright (c) 2006
* 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 Atheros AR5523 USB parts.
*
* The driver requires firmware to be loaded into the device. This
* is done on device discovery from a user application (uathload)
* that is launched by devd when a device with suitable product ID
* is recognized. Once firmware has been loaded the device will
* reset the USB port and re-attach with the original product ID+1
* and this driver will be attached. The firmware is licensed for
* general use (royalty free) and may be incorporated in products.
* Note that the firmware normally packaged with the NDIS drivers
* for these devices does not work in this way and so does not work
* with this driver.
*/
#include "opt_wlan.h"
#include <sys/param.h>
#include <sys/sockio.h>
#include <sys/sysctl.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/mbuf.h>
#include <sys/kernel.h>
#include <sys/socket.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/module.h>
#include <sys/bus.h>
#include <sys/endian.h>
#include <sys/kdb.h>
#include <net/bpf.h>
#include <net/if.h>
#include <net/if_var.h>
#include <net/if_arp.h>
#include <net/ethernet.h>
#include <net/if_dl.h>
#include <net/if_media.h>
#include <net/if_types.h>
#ifdef INET
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
#include <netinet/if_ether.h>
#include <netinet/ip.h>
#endif
#include <net80211/ieee80211_var.h>
#include <net80211/ieee80211_input.h>
#include <net80211/ieee80211_regdomain.h>
#include <net80211/ieee80211_radiotap.h>
#include <dev/usb/usb.h>
#include <dev/usb/usbdi.h>
#include "usbdevs.h"
#include <dev/usb/wlan/if_uathreg.h>
#include <dev/usb/wlan/if_uathvar.h>
static SYSCTL_NODE(_hw_usb, OID_AUTO, uath, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
"USB Atheros");
static int uath_countrycode = CTRY_DEFAULT; /* country code */
SYSCTL_INT(_hw_usb_uath, OID_AUTO, countrycode, CTLFLAG_RWTUN, &uath_countrycode,
0, "country code");
static int uath_regdomain = 0; /* regulatory domain */
SYSCTL_INT(_hw_usb_uath, OID_AUTO, regdomain, CTLFLAG_RD, &uath_regdomain,
0, "regulatory domain");
#ifdef UATH_DEBUG
int uath_debug = 0;
SYSCTL_INT(_hw_usb_uath, OID_AUTO, debug, CTLFLAG_RWTUN, &uath_debug, 0,
"uath debug level");
enum {
UATH_DEBUG_XMIT = 0x00000001, /* basic xmit operation */
UATH_DEBUG_XMIT_DUMP = 0x00000002, /* xmit dump */
UATH_DEBUG_RECV = 0x00000004, /* basic recv operation */
UATH_DEBUG_TX_PROC = 0x00000008, /* tx ISR proc */
UATH_DEBUG_RX_PROC = 0x00000010, /* rx ISR proc */
UATH_DEBUG_RECV_ALL = 0x00000020, /* trace all frames (beacons) */
UATH_DEBUG_INIT = 0x00000040, /* initialization of dev */
UATH_DEBUG_DEVCAP = 0x00000080, /* dev caps */
UATH_DEBUG_CMDS = 0x00000100, /* commands */
UATH_DEBUG_CMDS_DUMP = 0x00000200, /* command buffer dump */
UATH_DEBUG_RESET = 0x00000400, /* reset processing */
UATH_DEBUG_STATE = 0x00000800, /* 802.11 state transitions */
UATH_DEBUG_MULTICAST = 0x00001000, /* multicast */
UATH_DEBUG_WME = 0x00002000, /* WME */
UATH_DEBUG_CHANNEL = 0x00004000, /* channel */
UATH_DEBUG_RATES = 0x00008000, /* rates */
UATH_DEBUG_CRYPTO = 0x00010000, /* crypto */
UATH_DEBUG_LED = 0x00020000, /* LED */
UATH_DEBUG_ANY = 0xffffffff
};
#define DPRINTF(sc, m, fmt, ...) do { \
if (sc->sc_debug & (m)) \
printf(fmt, __VA_ARGS__); \
} while (0)
#else
#define DPRINTF(sc, m, fmt, ...) do { \
(void) sc; \
} while (0)
#endif
/* recognized device vendors/products */
static const STRUCT_USB_HOST_ID uath_devs[] = {
#define UATH_DEV(v,p) { USB_VP(USB_VENDOR_##v, USB_PRODUCT_##v##_##p) }
UATH_DEV(ACCTON, SMCWUSBTG2),
UATH_DEV(ATHEROS, AR5523),
UATH_DEV(ATHEROS2, AR5523_1),
UATH_DEV(ATHEROS2, AR5523_2),
UATH_DEV(ATHEROS2, AR5523_3),
UATH_DEV(CONCEPTRONIC, AR5523_1),
UATH_DEV(CONCEPTRONIC, AR5523_2),
UATH_DEV(DLINK, DWLAG122),
UATH_DEV(DLINK, DWLAG132),
UATH_DEV(DLINK, DWLG132),
UATH_DEV(DLINK2, DWA120),
UATH_DEV(GIGASET, AR5523),
UATH_DEV(GIGASET, SMCWUSBTG),
UATH_DEV(GLOBALSUN, AR5523_1),
UATH_DEV(GLOBALSUN, AR5523_2),
UATH_DEV(NETGEAR, WG111U),
UATH_DEV(NETGEAR3, WG111T),
UATH_DEV(NETGEAR3, WPN111),
2010-01-29 02:32:46 +00:00
UATH_DEV(NETGEAR3, WPN111_2),
UATH_DEV(UMEDIA, TEW444UBEU),
UATH_DEV(UMEDIA, AR5523_2),
UATH_DEV(WISTRONNEWEB, AR5523_1),
UATH_DEV(WISTRONNEWEB, AR5523_2),
UATH_DEV(ZCOM, AR5523)
#undef UATH_DEV
};
static usb_callback_t uath_intr_rx_callback;
static usb_callback_t uath_intr_tx_callback;
static usb_callback_t uath_bulk_rx_callback;
static usb_callback_t uath_bulk_tx_callback;
static const struct usb_config uath_usbconfig[UATH_N_XFERS] = {
[UATH_INTR_RX] = {
.type = UE_BULK,
.endpoint = 0x1,
.direction = UE_DIR_IN,
.bufsize = UATH_MAX_CMDSZ,
.flags = {
.pipe_bof = 1,
.short_xfer_ok = 1
},
.callback = uath_intr_rx_callback
},
[UATH_INTR_TX] = {
.type = UE_BULK,
.endpoint = 0x1,
.direction = UE_DIR_OUT,
.bufsize = UATH_MAX_CMDSZ * UATH_CMD_LIST_COUNT,
.flags = {
.force_short_xfer = 1,
.pipe_bof = 1,
},
.callback = uath_intr_tx_callback,
.timeout = UATH_CMD_TIMEOUT
},
[UATH_BULK_RX] = {
.type = UE_BULK,
.endpoint = 0x2,
.direction = UE_DIR_IN,
.bufsize = MCLBYTES,
.flags = {
.ext_buffer = 1,
.pipe_bof = 1,
.short_xfer_ok = 1
},
.callback = uath_bulk_rx_callback
},
[UATH_BULK_TX] = {
.type = UE_BULK,
.endpoint = 0x2,
.direction = UE_DIR_OUT,
.bufsize = UATH_MAX_TXBUFSZ * UATH_TX_DATA_LIST_COUNT,
.flags = {
.force_short_xfer = 1,
.pipe_bof = 1
},
.callback = uath_bulk_tx_callback,
.timeout = UATH_DATA_TIMEOUT
}
};
static struct ieee80211vap *uath_vap_create(struct ieee80211com *,
const char [IFNAMSIZ], int, enum ieee80211_opmode, int,
const uint8_t [IEEE80211_ADDR_LEN],
const uint8_t [IEEE80211_ADDR_LEN]);
static void uath_vap_delete(struct ieee80211vap *);
static int uath_alloc_cmd_list(struct uath_softc *, struct uath_cmd []);
static void uath_free_cmd_list(struct uath_softc *, struct uath_cmd []);
static int uath_host_available(struct uath_softc *);
static int uath_get_capability(struct uath_softc *, uint32_t, uint32_t *);
static int uath_get_devcap(struct uath_softc *);
static struct uath_cmd *
uath_get_cmdbuf(struct uath_softc *);
static int uath_cmd_read(struct uath_softc *, uint32_t, const void *,
int, void *, int, int);
static int uath_cmd_write(struct uath_softc *, uint32_t, const void *,
int, int);
static void uath_stat(void *);
#ifdef UATH_DEBUG
static void uath_dump_cmd(const uint8_t *, int, char);
static const char *
uath_codename(int);
#endif
static int uath_get_devstatus(struct uath_softc *,
uint8_t macaddr[IEEE80211_ADDR_LEN]);
static int uath_get_status(struct uath_softc *, uint32_t, void *, int);
static int uath_alloc_rx_data_list(struct uath_softc *);
static int uath_alloc_tx_data_list(struct uath_softc *);
static void uath_free_rx_data_list(struct uath_softc *);
static void uath_free_tx_data_list(struct uath_softc *);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
static int uath_init(struct uath_softc *);
static void uath_stop(struct uath_softc *);
static void uath_parent(struct ieee80211com *);
static int uath_transmit(struct ieee80211com *, struct mbuf *);
static void uath_start(struct uath_softc *);
static int uath_raw_xmit(struct ieee80211_node *, struct mbuf *,
const struct ieee80211_bpf_params *);
static void uath_scan_start(struct ieee80211com *);
static void uath_scan_end(struct ieee80211com *);
static void uath_set_channel(struct ieee80211com *);
static void uath_update_mcast(struct ieee80211com *);
static void uath_update_promisc(struct ieee80211com *);
static int uath_config(struct uath_softc *, uint32_t, uint32_t);
static int uath_config_multi(struct uath_softc *, uint32_t, const void *,
int);
static int uath_switch_channel(struct uath_softc *,
struct ieee80211_channel *);
static int uath_set_rxfilter(struct uath_softc *, uint32_t, uint32_t);
static void uath_watchdog(void *);
static void uath_abort_xfers(struct uath_softc *);
static int uath_dataflush(struct uath_softc *);
static int uath_cmdflush(struct uath_softc *);
static int uath_flush(struct uath_softc *);
static int uath_set_ledstate(struct uath_softc *, int);
static int uath_set_chan(struct uath_softc *, struct ieee80211_channel *);
static int uath_reset_tx_queues(struct uath_softc *);
static int uath_wme_init(struct uath_softc *);
static struct uath_data *
uath_getbuf(struct uath_softc *);
static int uath_newstate(struct ieee80211vap *, enum ieee80211_state,
int);
static int uath_set_key(struct uath_softc *,
const struct ieee80211_key *, int);
static int uath_set_keys(struct uath_softc *, struct ieee80211vap *);
static void uath_sysctl_node(struct uath_softc *);
static int
uath_match(device_t dev)
{
struct usb_attach_arg *uaa = device_get_ivars(dev);
if (uaa->usb_mode != USB_MODE_HOST)
return (ENXIO);
if (uaa->info.bConfigIndex != UATH_CONFIG_INDEX)
return (ENXIO);
if (uaa->info.bIfaceIndex != UATH_IFACE_INDEX)
return (ENXIO);
return (usbd_lookup_id_by_uaa(uath_devs, sizeof(uath_devs), uaa));
}
static int
uath_attach(device_t dev)
{
struct uath_softc *sc = device_get_softc(dev);
struct usb_attach_arg *uaa = device_get_ivars(dev);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
uint8_t bands[IEEE80211_MODE_BYTES];
uint8_t iface_index = UATH_IFACE_INDEX; /* XXX */
usb_error_t error;
sc->sc_dev = dev;
sc->sc_udev = uaa->device;
#ifdef UATH_DEBUG
sc->sc_debug = uath_debug;
#endif
device_set_usb_desc(dev);
/*
* Only post-firmware devices here.
*/
mtx_init(&sc->sc_mtx, device_get_nameunit(sc->sc_dev), MTX_NETWORK_LOCK,
MTX_DEF);
callout_init(&sc->stat_ch, 0);
callout_init_mtx(&sc->watchdog_ch, &sc->sc_mtx, 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
mbufq_init(&sc->sc_snd, ifqmaxlen);
error = usbd_transfer_setup(uaa->device, &iface_index, sc->sc_xfer,
uath_usbconfig, UATH_N_XFERS, sc, &sc->sc_mtx);
if (error) {
device_printf(dev, "could not allocate USB transfers, "
"err=%s\n", usbd_errstr(error));
goto fail;
}
sc->sc_cmd_dma_buf =
usbd_xfer_get_frame_buffer(sc->sc_xfer[UATH_INTR_TX], 0);
sc->sc_tx_dma_buf =
usbd_xfer_get_frame_buffer(sc->sc_xfer[UATH_BULK_TX], 0);
/*
* Setup buffers for firmware commands.
*/
error = uath_alloc_cmd_list(sc, sc->sc_cmd);
if (error != 0) {
device_printf(sc->sc_dev,
"could not allocate Tx command list\n");
goto fail1;
}
/*
* We're now ready to send+receive firmware commands.
*/
UATH_LOCK(sc);
error = uath_host_available(sc);
if (error != 0) {
device_printf(sc->sc_dev, "could not initialize adapter\n");
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
goto fail2;
}
error = uath_get_devcap(sc);
if (error != 0) {
device_printf(sc->sc_dev,
"could not get device capabilities\n");
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
goto fail2;
}
UATH_UNLOCK(sc);
/* Create device sysctl node. */
uath_sysctl_node(sc);
UATH_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
error = uath_get_devstatus(sc, ic->ic_macaddr);
if (error != 0) {
device_printf(sc->sc_dev, "could not get device status\n");
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
goto fail2;
}
/*
* Allocate xfers for Rx/Tx data pipes.
*/
error = uath_alloc_rx_data_list(sc);
if (error != 0) {
device_printf(sc->sc_dev, "could not allocate Rx data list\n");
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
goto fail2;
}
error = uath_alloc_tx_data_list(sc);
if (error != 0) {
device_printf(sc->sc_dev, "could not allocate Tx data list\n");
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
goto fail2;
}
UATH_UNLOCK(sc);
ic->ic_softc = sc;
ic->ic_name = device_get_nameunit(dev);
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 */
IEEE80211_C_MONITOR | /* monitor mode supported */
IEEE80211_C_TXPMGT | /* tx power management */
IEEE80211_C_SHPREAMBLE | /* short preamble supported */
IEEE80211_C_SHSLOT | /* short slot time supported */
IEEE80211_C_WPA | /* 802.11i */
IEEE80211_C_BGSCAN | /* capable of bg scanning */
IEEE80211_C_TXFRAG; /* handle tx frags */
/* put a regulatory domain to reveal informations. */
uath_regdomain = sc->sc_devcap.regDomain;
memset(bands, 0, sizeof(bands));
setbit(bands, IEEE80211_MODE_11B);
setbit(bands, IEEE80211_MODE_11G);
if ((sc->sc_devcap.analog5GhzRevision & 0xf0) == 0x30)
setbit(bands, IEEE80211_MODE_11A);
/* XXX turbo */
ieee80211_init_channels(ic, NULL, bands);
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);
ic->ic_raw_xmit = uath_raw_xmit;
ic->ic_scan_start = uath_scan_start;
ic->ic_scan_end = uath_scan_end;
ic->ic_set_channel = uath_set_channel;
ic->ic_vap_create = uath_vap_create;
ic->ic_vap_delete = uath_vap_delete;
ic->ic_update_mcast = uath_update_mcast;
ic->ic_update_promisc = uath_update_promisc;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ic->ic_transmit = uath_transmit;
ic->ic_parent = uath_parent;
ieee80211_radiotap_attach(ic,
&sc->sc_txtap.wt_ihdr, sizeof(sc->sc_txtap),
UATH_TX_RADIOTAP_PRESENT,
&sc->sc_rxtap.wr_ihdr, sizeof(sc->sc_rxtap),
UATH_RX_RADIOTAP_PRESENT);
if (bootverbose)
ieee80211_announce(ic);
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
fail2: UATH_UNLOCK(sc);
uath_free_cmd_list(sc, sc->sc_cmd);
fail1: usbd_transfer_unsetup(sc->sc_xfer, UATH_N_XFERS);
fail:
return (error);
}
static int
uath_detach(device_t dev)
{
struct uath_softc *sc = device_get_softc(dev);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
unsigned int x;
/*
* Prevent further allocations from RX/TX/CMD
* data lists and ioctls
*/
UATH_LOCK(sc);
sc->sc_flags |= UATH_FLAG_INVALID;
STAILQ_INIT(&sc->sc_rx_active);
STAILQ_INIT(&sc->sc_rx_inactive);
STAILQ_INIT(&sc->sc_tx_active);
STAILQ_INIT(&sc->sc_tx_inactive);
STAILQ_INIT(&sc->sc_tx_pending);
STAILQ_INIT(&sc->sc_cmd_active);
STAILQ_INIT(&sc->sc_cmd_pending);
STAILQ_INIT(&sc->sc_cmd_waiting);
STAILQ_INIT(&sc->sc_cmd_inactive);
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
uath_stop(sc);
UATH_UNLOCK(sc);
callout_drain(&sc->stat_ch);
callout_drain(&sc->watchdog_ch);
/* drain USB transfers */
for (x = 0; x != UATH_N_XFERS; x++)
usbd_transfer_drain(sc->sc_xfer[x]);
/* free data buffers */
UATH_LOCK(sc);
uath_free_rx_data_list(sc);
uath_free_tx_data_list(sc);
uath_free_cmd_list(sc, sc->sc_cmd);
UATH_UNLOCK(sc);
/* free USB transfers and some data buffers */
usbd_transfer_unsetup(sc->sc_xfer, UATH_N_XFERS);
ieee80211_ifdetach(ic);
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_drain(&sc->sc_snd);
mtx_destroy(&sc->sc_mtx);
return (0);
}
static void
uath_free_cmd_list(struct uath_softc *sc, struct uath_cmd cmds[])
{
int i;
for (i = 0; i != UATH_CMD_LIST_COUNT; i++)
cmds[i].buf = NULL;
}
static int
uath_alloc_cmd_list(struct uath_softc *sc, struct uath_cmd cmds[])
{
int i;
STAILQ_INIT(&sc->sc_cmd_active);
STAILQ_INIT(&sc->sc_cmd_pending);
STAILQ_INIT(&sc->sc_cmd_waiting);
STAILQ_INIT(&sc->sc_cmd_inactive);
for (i = 0; i != UATH_CMD_LIST_COUNT; i++) {
struct uath_cmd *cmd = &cmds[i];
cmd->sc = sc; /* backpointer for callbacks */
cmd->msgid = i;
cmd->buf = ((uint8_t *)sc->sc_cmd_dma_buf) +
(i * UATH_MAX_CMDSZ);
STAILQ_INSERT_TAIL(&sc->sc_cmd_inactive, cmd, next);
UATH_STAT_INC(sc, st_cmd_inactive);
}
return (0);
}
static int
uath_host_available(struct uath_softc *sc)
{
struct uath_cmd_host_available setup;
UATH_ASSERT_LOCKED(sc);
/* inform target the host is available */
setup.sw_ver_major = htobe32(ATH_SW_VER_MAJOR);
setup.sw_ver_minor = htobe32(ATH_SW_VER_MINOR);
setup.sw_ver_patch = htobe32(ATH_SW_VER_PATCH);
setup.sw_ver_build = htobe32(ATH_SW_VER_BUILD);
return uath_cmd_read(sc, WDCMSG_HOST_AVAILABLE,
&setup, sizeof setup, NULL, 0, 0);
}
#ifdef UATH_DEBUG
static void
uath_dump_cmd(const uint8_t *buf, int len, char prefix)
{
const char *sep = "";
int i;
for (i = 0; i < len; i++) {
if ((i % 16) == 0) {
printf("%s%c ", sep, prefix);
sep = "\n";
}
else if ((i % 4) == 0)
printf(" ");
printf("%02x", buf[i]);
}
printf("\n");
}
static const char *
uath_codename(int code)
{
static const char *names[] = {
"0x00",
"HOST_AVAILABLE",
"BIND",
"TARGET_RESET",
"TARGET_GET_CAPABILITY",
"TARGET_SET_CONFIG",
"TARGET_GET_STATUS",
"TARGET_GET_STATS",
"TARGET_START",
"TARGET_STOP",
"TARGET_ENABLE",
"TARGET_DISABLE",
"CREATE_CONNECTION",
"UPDATE_CONNECT_ATTR",
"DELETE_CONNECT",
"SEND",
"FLUSH",
"STATS_UPDATE",
"BMISS",
"DEVICE_AVAIL",
"SEND_COMPLETE",
"DATA_AVAIL",
"SET_PWR_MODE",
"BMISS_ACK",
"SET_LED_STEADY",
"SET_LED_BLINK",
"SETUP_BEACON_DESC",
"BEACON_INIT",
"RESET_KEY_CACHE",
"RESET_KEY_CACHE_ENTRY",
"SET_KEY_CACHE_ENTRY",
"SET_DECOMP_MASK",
"SET_REGULATORY_DOMAIN",
"SET_LED_STATE",
"WRITE_ASSOCID",
"SET_STA_BEACON_TIMERS",
"GET_TSF",
"RESET_TSF",
"SET_ADHOC_MODE",
"SET_BASIC_RATE",
"MIB_CONTROL",
"GET_CHANNEL_DATA",
"GET_CUR_RSSI",
"SET_ANTENNA_SWITCH",
"0x2c", "0x2d", "0x2e",
"USE_SHORT_SLOT_TIME",
"SET_POWER_MODE",
"SETUP_PSPOLL_DESC",
"SET_RX_MULTICAST_FILTER",
"RX_FILTER",
"PER_CALIBRATION",
"RESET",
"DISABLE",
"PHY_DISABLE",
"SET_TX_POWER_LIMIT",
"SET_TX_QUEUE_PARAMS",
"SETUP_TX_QUEUE",
"RELEASE_TX_QUEUE",
};
static char buf[8];
if (code < nitems(names))
return names[code];
if (code == WDCMSG_SET_DEFAULT_KEY)
return "SET_DEFAULT_KEY";
snprintf(buf, sizeof(buf), "0x%02x", code);
return buf;
}
#endif
/*
* Low-level function to send read or write commands to the firmware.
*/
static int
uath_cmdsend(struct uath_softc *sc, uint32_t code, const void *idata, int ilen,
void *odata, int olen, int flags)
{
struct uath_cmd_hdr *hdr;
struct uath_cmd *cmd;
int error;
UATH_ASSERT_LOCKED(sc);
/* grab a xfer */
cmd = uath_get_cmdbuf(sc);
if (cmd == NULL) {
device_printf(sc->sc_dev, "%s: empty inactive queue\n",
__func__);
return (ENOBUFS);
}
cmd->flags = flags;
/* always bulk-out a multiple of 4 bytes */
cmd->buflen = roundup2(sizeof(struct uath_cmd_hdr) + ilen, 4);
hdr = (struct uath_cmd_hdr *)cmd->buf;
memset(hdr, 0, sizeof(struct uath_cmd_hdr));
hdr->len = htobe32(cmd->buflen);
hdr->code = htobe32(code);
hdr->msgid = cmd->msgid; /* don't care about endianness */
hdr->magic = htobe32((cmd->flags & UATH_CMD_FLAG_MAGIC) ? 1 << 24 : 0);
memcpy((uint8_t *)(hdr + 1), idata, ilen);
#ifdef UATH_DEBUG
if (sc->sc_debug & UATH_DEBUG_CMDS) {
printf("%s: send %s [flags 0x%x] olen %d\n",
__func__, uath_codename(code), cmd->flags, olen);
if (sc->sc_debug & UATH_DEBUG_CMDS_DUMP)
uath_dump_cmd(cmd->buf, cmd->buflen, '+');
}
#endif
cmd->odata = odata;
KASSERT(odata == NULL ||
olen < UATH_MAX_CMDSZ - sizeof(*hdr) + sizeof(uint32_t),
("odata %p olen %u", odata, olen));
cmd->olen = olen;
STAILQ_INSERT_TAIL(&sc->sc_cmd_pending, cmd, next);
UATH_STAT_INC(sc, st_cmd_pending);
usbd_transfer_start(sc->sc_xfer[UATH_INTR_TX]);
if (cmd->flags & UATH_CMD_FLAG_READ) {
usbd_transfer_start(sc->sc_xfer[UATH_INTR_RX]);
/* wait at most two seconds for command reply */
error = mtx_sleep(cmd, &sc->sc_mtx, 0, "uathcmd", 2 * hz);
cmd->odata = NULL; /* in case reply comes too late */
if (error != 0) {
device_printf(sc->sc_dev, "timeout waiting for reply "
"to cmd 0x%x (%u)\n", code, code);
} else if (cmd->olen != olen) {
device_printf(sc->sc_dev, "unexpected reply data count "
"to cmd 0x%x (%u), got %u, expected %u\n",
code, code, cmd->olen, olen);
error = EINVAL;
}
return (error);
}
return (0);
}
static int
uath_cmd_read(struct uath_softc *sc, uint32_t code, const void *idata,
int ilen, void *odata, int olen, int flags)
{
flags |= UATH_CMD_FLAG_READ;
return uath_cmdsend(sc, code, idata, ilen, odata, olen, flags);
}
static int
uath_cmd_write(struct uath_softc *sc, uint32_t code, const void *data, int len,
int flags)
{
flags &= ~UATH_CMD_FLAG_READ;
return uath_cmdsend(sc, code, data, len, NULL, 0, flags);
}
static struct uath_cmd *
uath_get_cmdbuf(struct uath_softc *sc)
{
struct uath_cmd *uc;
UATH_ASSERT_LOCKED(sc);
uc = STAILQ_FIRST(&sc->sc_cmd_inactive);
if (uc != NULL) {
STAILQ_REMOVE_HEAD(&sc->sc_cmd_inactive, next);
UATH_STAT_DEC(sc, st_cmd_inactive);
} else
uc = NULL;
if (uc == NULL)
DPRINTF(sc, UATH_DEBUG_XMIT, "%s: %s\n", __func__,
"out of command xmit buffers");
return (uc);
}
/*
* This function is called periodically (every second) when associated to
* query device statistics.
*/
static void
uath_stat(void *arg)
{
struct uath_softc *sc = arg;
int error;
UATH_LOCK(sc);
/*
* Send request for statistics asynchronously. The timer will be
* restarted when we'll get the stats notification.
*/
error = uath_cmd_write(sc, WDCMSG_TARGET_GET_STATS, NULL, 0,
UATH_CMD_FLAG_ASYNC);
if (error != 0) {
device_printf(sc->sc_dev,
"could not query stats, error %d\n", error);
}
UATH_UNLOCK(sc);
}
static int
uath_get_capability(struct uath_softc *sc, uint32_t cap, uint32_t *val)
{
int error;
cap = htobe32(cap);
error = uath_cmd_read(sc, WDCMSG_TARGET_GET_CAPABILITY,
&cap, sizeof cap, val, sizeof(uint32_t), UATH_CMD_FLAG_MAGIC);
if (error != 0) {
device_printf(sc->sc_dev, "could not read capability %u\n",
be32toh(cap));
return (error);
}
*val = be32toh(*val);
return (error);
}
static int
uath_get_devcap(struct uath_softc *sc)
{
#define GETCAP(x, v) do { \
error = uath_get_capability(sc, x, &v); \
if (error != 0) \
return (error); \
DPRINTF(sc, UATH_DEBUG_DEVCAP, \
"%s: %s=0x%08x\n", __func__, #x, v); \
} while (0)
struct uath_devcap *cap = &sc->sc_devcap;
int error;
/* collect device capabilities */
GETCAP(CAP_TARGET_VERSION, cap->targetVersion);
GETCAP(CAP_TARGET_REVISION, cap->targetRevision);
GETCAP(CAP_MAC_VERSION, cap->macVersion);
GETCAP(CAP_MAC_REVISION, cap->macRevision);
GETCAP(CAP_PHY_REVISION, cap->phyRevision);
GETCAP(CAP_ANALOG_5GHz_REVISION, cap->analog5GhzRevision);
GETCAP(CAP_ANALOG_2GHz_REVISION, cap->analog2GhzRevision);
GETCAP(CAP_REG_DOMAIN, cap->regDomain);
GETCAP(CAP_REG_CAP_BITS, cap->regCapBits);
#if 0
/* NB: not supported in rev 1.5 */
GETCAP(CAP_COUNTRY_CODE, cap->countryCode);
#endif
GETCAP(CAP_WIRELESS_MODES, cap->wirelessModes);
GETCAP(CAP_CHAN_SPREAD_SUPPORT, cap->chanSpreadSupport);
GETCAP(CAP_COMPRESS_SUPPORT, cap->compressSupport);
GETCAP(CAP_BURST_SUPPORT, cap->burstSupport);
GETCAP(CAP_FAST_FRAMES_SUPPORT, cap->fastFramesSupport);
GETCAP(CAP_CHAP_TUNING_SUPPORT, cap->chapTuningSupport);
GETCAP(CAP_TURBOG_SUPPORT, cap->turboGSupport);
GETCAP(CAP_TURBO_PRIME_SUPPORT, cap->turboPrimeSupport);
GETCAP(CAP_DEVICE_TYPE, cap->deviceType);
GETCAP(CAP_WME_SUPPORT, cap->wmeSupport);
GETCAP(CAP_TOTAL_QUEUES, cap->numTxQueues);
GETCAP(CAP_CONNECTION_ID_MAX, cap->connectionIdMax);
GETCAP(CAP_LOW_5GHZ_CHAN, cap->low5GhzChan);
GETCAP(CAP_HIGH_5GHZ_CHAN, cap->high5GhzChan);
GETCAP(CAP_LOW_2GHZ_CHAN, cap->low2GhzChan);
GETCAP(CAP_HIGH_2GHZ_CHAN, cap->high2GhzChan);
GETCAP(CAP_TWICE_ANTENNAGAIN_5G, cap->twiceAntennaGain5G);
GETCAP(CAP_TWICE_ANTENNAGAIN_2G, cap->twiceAntennaGain2G);
GETCAP(CAP_CIPHER_AES_CCM, cap->supportCipherAES_CCM);
GETCAP(CAP_CIPHER_TKIP, cap->supportCipherTKIP);
GETCAP(CAP_MIC_TKIP, cap->supportMicTKIP);
cap->supportCipherWEP = 1; /* NB: always available */
return (0);
}
static int
uath_get_devstatus(struct uath_softc *sc, uint8_t macaddr[IEEE80211_ADDR_LEN])
{
int error;
/* retrieve MAC address */
error = uath_get_status(sc, ST_MAC_ADDR, macaddr, IEEE80211_ADDR_LEN);
if (error != 0) {
device_printf(sc->sc_dev, "could not read MAC address\n");
return (error);
}
error = uath_get_status(sc, ST_SERIAL_NUMBER,
&sc->sc_serial[0], sizeof(sc->sc_serial));
if (error != 0) {
device_printf(sc->sc_dev,
"could not read device serial number\n");
return (error);
}
return (0);
}
static int
uath_get_status(struct uath_softc *sc, uint32_t which, void *odata, int olen)
{
int error;
which = htobe32(which);
error = uath_cmd_read(sc, WDCMSG_TARGET_GET_STATUS,
&which, sizeof(which), odata, olen, UATH_CMD_FLAG_MAGIC);
if (error != 0)
device_printf(sc->sc_dev,
"could not read EEPROM offset 0x%02x\n", be32toh(which));
return (error);
}
static void
uath_free_data_list(struct uath_softc *sc, struct uath_data data[], int ndata,
int fillmbuf)
{
int i;
for (i = 0; i < ndata; i++) {
struct uath_data *dp = &data[i];
if (fillmbuf == 1) {
if (dp->m != NULL) {
m_freem(dp->m);
dp->m = NULL;
dp->buf = NULL;
}
} else {
dp->buf = NULL;
}
if (dp->ni != NULL) {
ieee80211_free_node(dp->ni);
dp->ni = NULL;
}
}
}
static int
uath_alloc_data_list(struct uath_softc *sc, struct uath_data data[],
int ndata, int maxsz, void *dma_buf)
{
int i, error;
for (i = 0; i < ndata; i++) {
struct uath_data *dp = &data[i];
dp->sc = sc;
if (dma_buf == NULL) {
/* XXX check maxsz */
dp->m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
if (dp->m == NULL) {
device_printf(sc->sc_dev,
"could not allocate rx mbuf\n");
error = ENOMEM;
goto fail;
}
dp->buf = mtod(dp->m, uint8_t *);
} else {
dp->m = NULL;
dp->buf = ((uint8_t *)dma_buf) + (i * maxsz);
}
dp->ni = NULL;
}
return (0);
fail: uath_free_data_list(sc, data, ndata, 1 /* free mbufs */);
return (error);
}
static int
uath_alloc_rx_data_list(struct uath_softc *sc)
{
int error, i;
/* XXX is it enough to store the RX packet with MCLBYTES bytes? */
error = uath_alloc_data_list(sc,
sc->sc_rx, UATH_RX_DATA_LIST_COUNT, MCLBYTES,
NULL /* setup mbufs */);
if (error != 0)
return (error);
STAILQ_INIT(&sc->sc_rx_active);
STAILQ_INIT(&sc->sc_rx_inactive);
for (i = 0; i < UATH_RX_DATA_LIST_COUNT; i++) {
STAILQ_INSERT_HEAD(&sc->sc_rx_inactive, &sc->sc_rx[i],
next);
UATH_STAT_INC(sc, st_rx_inactive);
}
return (0);
}
static int
uath_alloc_tx_data_list(struct uath_softc *sc)
{
int error, i;
error = uath_alloc_data_list(sc,
sc->sc_tx, UATH_TX_DATA_LIST_COUNT, UATH_MAX_TXBUFSZ,
sc->sc_tx_dma_buf);
if (error != 0)
return (error);
STAILQ_INIT(&sc->sc_tx_active);
STAILQ_INIT(&sc->sc_tx_inactive);
STAILQ_INIT(&sc->sc_tx_pending);
for (i = 0; i < UATH_TX_DATA_LIST_COUNT; i++) {
STAILQ_INSERT_HEAD(&sc->sc_tx_inactive, &sc->sc_tx[i],
next);
UATH_STAT_INC(sc, st_tx_inactive);
}
return (0);
}
static void
uath_free_rx_data_list(struct uath_softc *sc)
{
uath_free_data_list(sc, sc->sc_rx, UATH_RX_DATA_LIST_COUNT,
1 /* free mbufs */);
}
static void
uath_free_tx_data_list(struct uath_softc *sc)
{
uath_free_data_list(sc, sc->sc_tx, UATH_TX_DATA_LIST_COUNT,
0 /* no mbufs */);
}
static struct ieee80211vap *
uath_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])
{
struct uath_vap *uvp;
struct ieee80211vap *vap;
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 uath_vap), M_80211_VAP, M_WAITOK | M_ZERO);
vap = &uvp->vap;
/* enable s/w bmiss handling for sta mode */
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 | IEEE80211_CLONE_NOBEACONS, bssid) != 0) {
/* out of memory */
free(uvp, M_80211_VAP);
return (NULL);
}
/* override state transition machine */
uvp->newstate = vap->iv_newstate;
vap->iv_newstate = uath_newstate;
/* 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);
ic->ic_opmode = opmode;
return (vap);
}
static void
uath_vap_delete(struct ieee80211vap *vap)
{
struct uath_vap *uvp = UATH_VAP(vap);
ieee80211_vap_detach(vap);
free(uvp, M_80211_VAP);
}
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
uath_init(struct uath_softc *sc)
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
uint32_t val;
int error;
UATH_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
if (sc->sc_flags & UATH_FLAG_INITDONE)
uath_stop(sc);
/* reset variables */
sc->sc_intrx_nextnum = sc->sc_msgid = 0;
val = htobe32(0);
uath_cmd_write(sc, WDCMSG_BIND, &val, sizeof val, 0);
/* 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
uath_config_multi(sc, CFG_MAC_ADDR,
vap ? vap->iv_myaddr : ic->ic_macaddr, IEEE80211_ADDR_LEN);
/* XXX honor net80211 state */
uath_config(sc, CFG_RATE_CONTROL_ENABLE, 0x00000001);
uath_config(sc, CFG_DIVERSITY_CTL, 0x00000001);
uath_config(sc, CFG_ABOLT, 0x0000003f);
uath_config(sc, CFG_WME_ENABLED, 0x00000001);
uath_config(sc, CFG_SERVICE_TYPE, 1);
uath_config(sc, CFG_TP_SCALE, 0x00000000);
uath_config(sc, CFG_TPC_HALF_DBM5, 0x0000003c);
uath_config(sc, CFG_TPC_HALF_DBM2, 0x0000003c);
uath_config(sc, CFG_OVERRD_TX_POWER, 0x00000000);
uath_config(sc, CFG_GMODE_PROTECTION, 0x00000000);
uath_config(sc, CFG_GMODE_PROTECT_RATE_INDEX, 0x00000003);
uath_config(sc, CFG_PROTECTION_TYPE, 0x00000000);
uath_config(sc, CFG_MODE_CTS, 0x00000002);
error = uath_cmd_read(sc, WDCMSG_TARGET_START, NULL, 0,
&val, sizeof(val), UATH_CMD_FLAG_MAGIC);
if (error) {
device_printf(sc->sc_dev,
"could not start target, error %d\n", error);
goto fail;
}
DPRINTF(sc, UATH_DEBUG_INIT, "%s returns handle: 0x%x\n",
uath_codename(WDCMSG_TARGET_START), be32toh(val));
/* set default channel */
error = uath_switch_channel(sc, ic->ic_curchan);
if (error) {
device_printf(sc->sc_dev,
"could not switch channel, error %d\n", error);
goto fail;
}
val = htobe32(TARGET_DEVICE_AWAKE);
uath_cmd_write(sc, WDCMSG_SET_PWR_MODE, &val, sizeof val, 0);
/* XXX? check */
uath_cmd_write(sc, WDCMSG_RESET_KEY_CACHE, NULL, 0, 0);
usbd_transfer_start(sc->sc_xfer[UATH_BULK_RX]);
/* enable Rx */
uath_set_rxfilter(sc, 0x0, UATH_FILTER_OP_INIT);
uath_set_rxfilter(sc,
UATH_FILTER_RX_UCAST | UATH_FILTER_RX_MCAST |
UATH_FILTER_RX_BCAST | UATH_FILTER_RX_BEACON,
UATH_FILTER_OP_SET);
sc->sc_flags |= UATH_FLAG_INITDONE;
callout_reset(&sc->watchdog_ch, hz, uath_watchdog, sc);
return (0);
fail:
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
uath_stop(sc);
return (error);
}
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
uath_stop(struct uath_softc *sc)
{
UATH_ASSERT_LOCKED(sc);
sc->sc_flags &= ~UATH_FLAG_INITDONE;
callout_stop(&sc->stat_ch);
callout_stop(&sc->watchdog_ch);
sc->sc_tx_timer = 0;
/* abort pending transmits */
uath_abort_xfers(sc);
/* flush data & control requests into the target */
(void)uath_flush(sc);
/* set a LED status to the disconnected. */
uath_set_ledstate(sc, 0);
/* stop the target */
uath_cmd_write(sc, WDCMSG_TARGET_STOP, NULL, 0, 0);
}
static int
uath_config(struct uath_softc *sc, uint32_t reg, uint32_t val)
{
struct uath_write_mac write;
int error;
write.reg = htobe32(reg);
write.len = htobe32(0); /* 0 = single write */
*(uint32_t *)write.data = htobe32(val);
error = uath_cmd_write(sc, WDCMSG_TARGET_SET_CONFIG, &write,
3 * sizeof (uint32_t), 0);
if (error != 0) {
device_printf(sc->sc_dev, "could not write register 0x%02x\n",
reg);
}
return (error);
}
static int
uath_config_multi(struct uath_softc *sc, uint32_t reg, const void *data,
int len)
{
struct uath_write_mac write;
int error;
write.reg = htobe32(reg);
write.len = htobe32(len);
bcopy(data, write.data, len);
/* properly handle the case where len is zero (reset) */
error = uath_cmd_write(sc, WDCMSG_TARGET_SET_CONFIG, &write,
(len == 0) ? sizeof (uint32_t) : 2 * sizeof (uint32_t) + len, 0);
if (error != 0) {
device_printf(sc->sc_dev,
"could not write %d bytes to register 0x%02x\n", len, reg);
}
return (error);
}
static int
uath_switch_channel(struct uath_softc *sc, struct ieee80211_channel *c)
{
int error;
UATH_ASSERT_LOCKED(sc);
/* set radio frequency */
error = uath_set_chan(sc, c);
if (error) {
device_printf(sc->sc_dev,
"could not set channel, error %d\n", error);
goto failed;
}
/* reset Tx rings */
error = uath_reset_tx_queues(sc);
if (error) {
device_printf(sc->sc_dev,
"could not reset Tx queues, error %d\n", error);
goto failed;
}
/* set Tx rings WME properties */
error = uath_wme_init(sc);
if (error) {
device_printf(sc->sc_dev,
"could not init Tx queues, error %d\n", error);
goto failed;
}
error = uath_set_ledstate(sc, 0);
if (error) {
device_printf(sc->sc_dev,
"could not set led state, error %d\n", error);
goto failed;
}
error = uath_flush(sc);
if (error) {
device_printf(sc->sc_dev,
"could not flush pipes, error %d\n", error);
goto failed;
}
failed:
return (error);
}
static int
uath_set_rxfilter(struct uath_softc *sc, uint32_t bits, uint32_t op)
{
struct uath_cmd_rx_filter rxfilter;
rxfilter.bits = htobe32(bits);
rxfilter.op = htobe32(op);
DPRINTF(sc, UATH_DEBUG_RECV | UATH_DEBUG_RECV_ALL,
"setting Rx filter=0x%x flags=0x%x\n", bits, op);
return uath_cmd_write(sc, WDCMSG_RX_FILTER, &rxfilter,
sizeof rxfilter, 0);
}
static void
uath_watchdog(void *arg)
{
struct uath_softc *sc = arg;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
if (sc->sc_tx_timer > 0) {
if (--sc->sc_tx_timer == 0) {
device_printf(sc->sc_dev, "device timeout\n");
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);
ieee80211_restart_all(ic);
return;
}
callout_reset(&sc->watchdog_ch, hz, uath_watchdog, sc);
}
}
static void
uath_abort_xfers(struct uath_softc *sc)
{
int i;
UATH_ASSERT_LOCKED(sc);
/* abort any pending transfers */
for (i = 0; i < UATH_N_XFERS; i++)
usbd_transfer_stop(sc->sc_xfer[i]);
}
static int
uath_flush(struct uath_softc *sc)
{
int error;
error = uath_dataflush(sc);
if (error != 0)
goto failed;
error = uath_cmdflush(sc);
if (error != 0)
goto failed;
failed:
return (error);
}
static int
uath_cmdflush(struct uath_softc *sc)
{
return uath_cmd_write(sc, WDCMSG_FLUSH, NULL, 0, 0);
}
static int
uath_dataflush(struct uath_softc *sc)
{
struct uath_data *data;
struct uath_chunk *chunk;
struct uath_tx_desc *desc;
UATH_ASSERT_LOCKED(sc);
data = uath_getbuf(sc);
if (data == NULL)
return (ENOBUFS);
data->buflen = sizeof(struct uath_chunk) + sizeof(struct uath_tx_desc);
data->m = NULL;
data->ni = NULL;
chunk = (struct uath_chunk *)data->buf;
desc = (struct uath_tx_desc *)(chunk + 1);
/* one chunk only */
chunk->seqnum = 0;
chunk->flags = UATH_CFLAGS_FINAL;
chunk->length = htobe16(sizeof (struct uath_tx_desc));
memset(desc, 0, sizeof(struct uath_tx_desc));
desc->msglen = htobe32(sizeof(struct uath_tx_desc));
desc->msgid = (sc->sc_msgid++) + 1; /* don't care about endianness */
desc->type = htobe32(WDCMSG_FLUSH);
desc->txqid = htobe32(0);
desc->connid = htobe32(0);
desc->flags = htobe32(0);
#ifdef UATH_DEBUG
if (sc->sc_debug & UATH_DEBUG_CMDS) {
DPRINTF(sc, UATH_DEBUG_RESET, "send flush ix %d\n",
desc->msgid);
if (sc->sc_debug & UATH_DEBUG_CMDS_DUMP)
uath_dump_cmd(data->buf, data->buflen, '+');
}
#endif
STAILQ_INSERT_TAIL(&sc->sc_tx_pending, data, next);
UATH_STAT_INC(sc, st_tx_pending);
sc->sc_tx_timer = 5;
usbd_transfer_start(sc->sc_xfer[UATH_BULK_TX]);
return (0);
}
static struct uath_data *
_uath_getbuf(struct uath_softc *sc)
{
struct uath_data *bf;
bf = STAILQ_FIRST(&sc->sc_tx_inactive);
if (bf != NULL) {
STAILQ_REMOVE_HEAD(&sc->sc_tx_inactive, next);
UATH_STAT_DEC(sc, st_tx_inactive);
} else
bf = NULL;
if (bf == NULL)
DPRINTF(sc, UATH_DEBUG_XMIT, "%s: %s\n", __func__,
"out of xmit buffers");
return (bf);
}
static struct uath_data *
uath_getbuf(struct uath_softc *sc)
{
struct uath_data *bf;
UATH_ASSERT_LOCKED(sc);
bf = _uath_getbuf(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 (bf == NULL)
DPRINTF(sc, UATH_DEBUG_XMIT, "%s: stop queue\n", __func__);
return (bf);
}
static int
uath_set_ledstate(struct uath_softc *sc, int connected)
{
DPRINTF(sc, UATH_DEBUG_LED,
"set led state %sconnected\n", connected ? "" : "!");
connected = htobe32(connected);
return uath_cmd_write(sc, WDCMSG_SET_LED_STATE,
&connected, sizeof connected, 0);
}
static int
uath_set_chan(struct uath_softc *sc, struct ieee80211_channel *c)
{
#ifdef UATH_DEBUG
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;
#endif
struct uath_cmd_reset reset;
memset(&reset, 0, sizeof(reset));
if (IEEE80211_IS_CHAN_2GHZ(c))
reset.flags |= htobe32(UATH_CHAN_2GHZ);
if (IEEE80211_IS_CHAN_5GHZ(c))
reset.flags |= htobe32(UATH_CHAN_5GHZ);
/* NB: 11g =>'s 11b so don't specify both OFDM and CCK */
if (IEEE80211_IS_CHAN_OFDM(c))
reset.flags |= htobe32(UATH_CHAN_OFDM);
else if (IEEE80211_IS_CHAN_CCK(c))
reset.flags |= htobe32(UATH_CHAN_CCK);
/* turbo can be used in either 2GHz or 5GHz */
if (c->ic_flags & IEEE80211_CHAN_TURBO)
reset.flags |= htobe32(UATH_CHAN_TURBO);
reset.freq = htobe32(c->ic_freq);
reset.maxrdpower = htobe32(50); /* XXX */
reset.channelchange = htobe32(1);
reset.keeprccontent = htobe32(0);
DPRINTF(sc, UATH_DEBUG_CHANNEL, "set channel %d, flags 0x%x freq %u\n",
ieee80211_chan2ieee(ic, c),
be32toh(reset.flags), be32toh(reset.freq));
return uath_cmd_write(sc, WDCMSG_RESET, &reset, sizeof reset, 0);
}
static int
uath_reset_tx_queues(struct uath_softc *sc)
{
int ac, error;
DPRINTF(sc, UATH_DEBUG_RESET, "%s: reset Tx queues\n", __func__);
for (ac = 0; ac < 4; ac++) {
const uint32_t qid = htobe32(ac);
error = uath_cmd_write(sc, WDCMSG_RELEASE_TX_QUEUE, &qid,
sizeof qid, 0);
if (error != 0)
break;
}
return (error);
}
static int
uath_wme_init(struct uath_softc *sc)
{
/* XXX get from net80211 */
static const struct uath_wme_settings uath_wme_11g[4] = {
{ 7, 4, 10, 0, 0 }, /* Background */
{ 3, 4, 10, 0, 0 }, /* Best-Effort */
{ 3, 3, 4, 26, 0 }, /* Video */
{ 2, 2, 3, 47, 0 } /* Voice */
};
struct uath_cmd_txq_setup qinfo;
int ac, error;
DPRINTF(sc, UATH_DEBUG_WME, "%s: setup Tx queues\n", __func__);
for (ac = 0; ac < 4; ac++) {
qinfo.qid = htobe32(ac);
qinfo.len = htobe32(sizeof(qinfo.attr));
qinfo.attr.priority = htobe32(ac); /* XXX */
qinfo.attr.aifs = htobe32(uath_wme_11g[ac].aifsn);
qinfo.attr.logcwmin = htobe32(uath_wme_11g[ac].logcwmin);
qinfo.attr.logcwmax = htobe32(uath_wme_11g[ac].logcwmax);
qinfo.attr.bursttime = htobe32(IEEE80211_TXOP_TO_US(
uath_wme_11g[ac].txop));
qinfo.attr.mode = htobe32(uath_wme_11g[ac].acm);/*XXX? */
qinfo.attr.qflags = htobe32(1); /* XXX? */
error = uath_cmd_write(sc, WDCMSG_SETUP_TX_QUEUE, &qinfo,
sizeof qinfo, 0);
if (error != 0)
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
uath_parent(struct ieee80211com *ic)
{
struct uath_softc *sc = ic->ic_softc;
int startall = 0;
UATH_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_flags & UATH_FLAG_INVALID) {
UATH_UNLOCK(sc);
return;
}
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
if (ic->ic_nrunning > 0) {
if (!(sc->sc_flags & UATH_FLAG_INITDONE)) {
uath_init(sc);
startall = 1;
}
} else if (sc->sc_flags & UATH_FLAG_INITDONE)
uath_stop(sc);
UATH_UNLOCK(sc);
if (startall)
ieee80211_start_all(ic);
}
static int
uath_tx_start(struct uath_softc *sc, struct mbuf *m0, struct ieee80211_node *ni,
struct uath_data *data)
{
struct ieee80211vap *vap = ni->ni_vap;
struct uath_chunk *chunk;
struct uath_tx_desc *desc;
const struct ieee80211_frame *wh;
struct ieee80211_key *k;
int framelen, msglen;
UATH_ASSERT_LOCKED(sc);
data->ni = ni;
data->m = m0;
chunk = (struct uath_chunk *)data->buf;
desc = (struct uath_tx_desc *)(chunk + 1);
if (ieee80211_radiotap_active_vap(vap)) {
struct uath_tx_radiotap_header *tap = &sc->sc_txtap;
tap->wt_flags = 0;
if (m0->m_flags & M_FRAG)
tap->wt_flags |= IEEE80211_RADIOTAP_F_FRAG;
ieee80211_radiotap_tx(vap, m0);
}
wh = mtod(m0, struct ieee80211_frame *);
if (wh->i_fc[1] & IEEE80211_FC1_PROTECTED) {
k = ieee80211_crypto_encap(ni, m0);
if (k == NULL) {
m_freem(m0);
return (ENOBUFS);
}
/* packet header may have moved, reset our local pointer */
wh = mtod(m0, struct ieee80211_frame *);
}
m_copydata(m0, 0, m0->m_pkthdr.len, (uint8_t *)(desc + 1));
framelen = m0->m_pkthdr.len + IEEE80211_CRC_LEN;
msglen = framelen + sizeof (struct uath_tx_desc);
data->buflen = msglen + sizeof (struct uath_chunk);
/* one chunk only for now */
chunk->seqnum = sc->sc_seqnum++;
chunk->flags = (m0->m_flags & M_FRAG) ? 0 : UATH_CFLAGS_FINAL;
if (m0->m_flags & M_LASTFRAG)
chunk->flags |= UATH_CFLAGS_FINAL;
chunk->flags = UATH_CFLAGS_FINAL;
chunk->length = htobe16(msglen);
/* fill Tx descriptor */
desc->msglen = htobe32(msglen);
/* NB: to get UATH_TX_NOTIFY reply, `msgid' must be larger than 0 */
desc->msgid = (sc->sc_msgid++) + 1; /* don't care about endianness */
desc->type = htobe32(WDCMSG_SEND);
switch (wh->i_fc[0] & IEEE80211_FC0_TYPE_MASK) {
case IEEE80211_FC0_TYPE_CTL:
case IEEE80211_FC0_TYPE_MGT:
/* NB: force all management frames to highest queue */
if (ni->ni_flags & IEEE80211_NODE_QOS) {
/* NB: force all management frames to highest queue */
desc->txqid = htobe32(WME_AC_VO | UATH_TXQID_MINRATE);
} else
desc->txqid = htobe32(WME_AC_BE | UATH_TXQID_MINRATE);
break;
case IEEE80211_FC0_TYPE_DATA:
/* XXX multicast frames should honor mcastrate */
desc->txqid = htobe32(M_WME_GETAC(m0));
break;
default:
device_printf(sc->sc_dev, "bogus frame type 0x%x (%s)\n",
wh->i_fc[0] & IEEE80211_FC0_TYPE_MASK, __func__);
m_freem(m0);
return (EIO);
}
if (vap->iv_state == IEEE80211_S_AUTH ||
vap->iv_state == IEEE80211_S_ASSOC ||
vap->iv_state == IEEE80211_S_RUN)
desc->connid = htobe32(UATH_ID_BSS);
else
desc->connid = htobe32(UATH_ID_INVALID);
desc->flags = htobe32(0 /* no UATH_TX_NOTIFY */);
desc->buflen = htobe32(m0->m_pkthdr.len);
#ifdef UATH_DEBUG
DPRINTF(sc, UATH_DEBUG_XMIT,
"send frame ix %u framelen %d msglen %d connid 0x%x txqid 0x%x\n",
desc->msgid, framelen, msglen, be32toh(desc->connid),
be32toh(desc->txqid));
if (sc->sc_debug & UATH_DEBUG_XMIT_DUMP)
uath_dump_cmd(data->buf, data->buflen, '+');
#endif
STAILQ_INSERT_TAIL(&sc->sc_tx_pending, data, next);
UATH_STAT_INC(sc, st_tx_pending);
usbd_transfer_start(sc->sc_xfer[UATH_BULK_TX]);
return (0);
}
/*
* Cleanup driver resources when we run out of buffers while processing
* fragments; return the tx buffers allocated and drop node references.
*/
static void
uath_txfrag_cleanup(struct uath_softc *sc,
uath_datahead *frags, struct ieee80211_node *ni)
{
struct uath_data *bf, *next;
UATH_ASSERT_LOCKED(sc);
STAILQ_FOREACH_SAFE(bf, frags, next, next) {
/* NB: bf assumed clean */
STAILQ_REMOVE_HEAD(frags, next);
STAILQ_INSERT_HEAD(&sc->sc_tx_inactive, bf, next);
UATH_STAT_INC(sc, st_tx_inactive);
ieee80211_node_decref(ni);
}
}
/*
* Setup xmit of a fragmented frame. Allocate a buffer for each frag and bump
* the node reference count to reflect the held reference to be setup by
* uath_tx_start.
*/
static int
uath_txfrag_setup(struct uath_softc *sc, uath_datahead *frags,
struct mbuf *m0, struct ieee80211_node *ni)
{
struct mbuf *m;
struct uath_data *bf;
UATH_ASSERT_LOCKED(sc);
for (m = m0->m_nextpkt; m != NULL; m = m->m_nextpkt) {
bf = uath_getbuf(sc);
if (bf == NULL) { /* out of buffers, cleanup */
uath_txfrag_cleanup(sc, frags, ni);
break;
}
ieee80211_node_incref(ni);
STAILQ_INSERT_TAIL(frags, bf, next);
}
return !STAILQ_EMPTY(frags);
}
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
uath_transmit(struct ieee80211com *ic, struct mbuf *m)
{
struct uath_softc *sc = ic->ic_softc;
int error;
UATH_LOCK(sc);
if ((sc->sc_flags & UATH_FLAG_INITDONE) == 0) {
UATH_UNLOCK(sc);
return (ENXIO);
}
error = mbufq_enqueue(&sc->sc_snd, m);
if (error) {
UATH_UNLOCK(sc);
return (error);
}
uath_start(sc);
UATH_UNLOCK(sc);
return (0);
}
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
uath_start(struct uath_softc *sc)
{
struct uath_data *bf;
struct ieee80211_node *ni;
struct mbuf *m, *next;
uath_datahead frags;
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
UATH_ASSERT_LOCKED(sc);
if ((sc->sc_flags & UATH_FLAG_INITDONE) == 0 ||
(sc->sc_flags & UATH_FLAG_INVALID))
return;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
while ((m = mbufq_dequeue(&sc->sc_snd)) != NULL) {
bf = uath_getbuf(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 (bf == NULL) {
mbufq_prepend(&sc->sc_snd, m);
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
ni = (struct ieee80211_node *)m->m_pkthdr.rcvif;
m->m_pkthdr.rcvif = NULL;
/*
* Check for fragmentation. If this frame has been broken up
* verify we have enough buffers to send all the fragments
* so all go out or none...
*/
STAILQ_INIT(&frags);
if ((m->m_flags & M_FRAG) &&
!uath_txfrag_setup(sc, &frags, m, ni)) {
DPRINTF(sc, UATH_DEBUG_XMIT,
"%s: out of txfrag buffers\n", __func__);
ieee80211_free_mbuf(m);
goto bad;
}
sc->sc_seqnum = 0;
nextfrag:
/*
* Pass the frame to the h/w for transmission.
* Fragmented frames have each frag chained together
* with m_nextpkt. We know there are sufficient uath_data's
* to send all the frags because of work done by
* uath_txfrag_setup.
*/
next = m->m_nextpkt;
if (uath_tx_start(sc, m, ni, bf) != 0) {
bad:
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);
reclaim:
STAILQ_INSERT_HEAD(&sc->sc_tx_inactive, bf, next);
UATH_STAT_INC(sc, st_tx_inactive);
uath_txfrag_cleanup(sc, &frags, ni);
ieee80211_free_node(ni);
continue;
}
if (next != NULL) {
/*
* Beware of state changing between frags.
XXX check sta power-save state?
*/
if (ni->ni_vap->iv_state != IEEE80211_S_RUN) {
DPRINTF(sc, UATH_DEBUG_XMIT,
"%s: flush fragmented packet, state %s\n",
__func__,
ieee80211_state_name[ni->ni_vap->iv_state]);
ieee80211_free_mbuf(next);
goto reclaim;
}
m = next;
bf = STAILQ_FIRST(&frags);
KASSERT(bf != NULL, ("no buf for txfrag"));
STAILQ_REMOVE_HEAD(&frags, next);
goto nextfrag;
}
sc->sc_tx_timer = 5;
}
}
static int
uath_raw_xmit(struct ieee80211_node *ni, struct mbuf *m,
const struct ieee80211_bpf_params *params)
{
struct ieee80211com *ic = ni->ni_ic;
struct uath_data *bf;
struct uath_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
UATH_LOCK(sc);
/* prevent management frames from being sent if we're not ready */
if ((sc->sc_flags & UATH_FLAG_INVALID) ||
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
!(sc->sc_flags & UATH_FLAG_INITDONE)) {
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
UATH_UNLOCK(sc);
return (ENETDOWN);
}
/* grab a TX buffer */
bf = uath_getbuf(sc);
if (bf == NULL) {
m_freem(m);
UATH_UNLOCK(sc);
return (ENOBUFS);
}
sc->sc_seqnum = 0;
if (uath_tx_start(sc, m, ni, bf) != 0) {
STAILQ_INSERT_HEAD(&sc->sc_tx_inactive, bf, next);
UATH_STAT_INC(sc, st_tx_inactive);
UATH_UNLOCK(sc);
return (EIO);
}
UATH_UNLOCK(sc);
sc->sc_tx_timer = 5;
return (0);
}
static void
uath_scan_start(struct ieee80211com *ic)
{
/* do nothing */
}
static void
uath_scan_end(struct ieee80211com *ic)
{
/* do nothing */
}
static void
uath_set_channel(struct ieee80211com *ic)
{
struct uath_softc *sc = ic->ic_softc;
UATH_LOCK(sc);
if ((sc->sc_flags & UATH_FLAG_INVALID) ||
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
(sc->sc_flags & UATH_FLAG_INITDONE) == 0) {
UATH_UNLOCK(sc);
return;
}
(void)uath_switch_channel(sc, ic->ic_curchan);
UATH_UNLOCK(sc);
}
static int
uath_set_rxmulti_filter(struct uath_softc *sc)
{
/* XXX broken */
return (0);
}
static void
uath_update_mcast(struct ieee80211com *ic)
{
struct uath_softc *sc = ic->ic_softc;
UATH_LOCK(sc);
if ((sc->sc_flags & UATH_FLAG_INVALID) ||
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
(sc->sc_flags & UATH_FLAG_INITDONE) == 0) {
UATH_UNLOCK(sc);
return;
}
/*
* this is for avoiding the race condition when we're try to
* connect to the AP with WPA.
*/
if (sc->sc_flags & UATH_FLAG_INITDONE)
(void)uath_set_rxmulti_filter(sc);
UATH_UNLOCK(sc);
}
static void
uath_update_promisc(struct ieee80211com *ic)
{
struct uath_softc *sc = ic->ic_softc;
UATH_LOCK(sc);
if ((sc->sc_flags & UATH_FLAG_INVALID) ||
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
(sc->sc_flags & UATH_FLAG_INITDONE) == 0) {
UATH_UNLOCK(sc);
return;
}
if (sc->sc_flags & UATH_FLAG_INITDONE) {
uath_set_rxfilter(sc,
UATH_FILTER_RX_UCAST | UATH_FILTER_RX_MCAST |
UATH_FILTER_RX_BCAST | UATH_FILTER_RX_BEACON |
UATH_FILTER_RX_PROM, UATH_FILTER_OP_SET);
}
UATH_UNLOCK(sc);
}
static int
uath_create_connection(struct uath_softc *sc, uint32_t connid)
{
const struct ieee80211_rateset *rs;
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);
struct ieee80211_node *ni;
struct uath_cmd_create_connection create;
ni = ieee80211_ref_node(vap->iv_bss);
memset(&create, 0, sizeof(create));
create.connid = htobe32(connid);
create.bssid = htobe32(0);
/* XXX packed or not? */
create.size = htobe32(sizeof(struct uath_cmd_rateset));
rs = &ni->ni_rates;
create.connattr.rateset.length = rs->rs_nrates;
bcopy(rs->rs_rates, &create.connattr.rateset.set[0],
rs->rs_nrates);
/* XXX turbo */
if (IEEE80211_IS_CHAN_A(ni->ni_chan))
create.connattr.wlanmode = htobe32(WLAN_MODE_11a);
else if (IEEE80211_IS_CHAN_ANYG(ni->ni_chan))
create.connattr.wlanmode = htobe32(WLAN_MODE_11g);
else
create.connattr.wlanmode = htobe32(WLAN_MODE_11b);
ieee80211_free_node(ni);
return uath_cmd_write(sc, WDCMSG_CREATE_CONNECTION, &create,
sizeof create, 0);
}
static int
uath_set_rates(struct uath_softc *sc, const struct ieee80211_rateset *rs)
{
struct uath_cmd_rates rates;
memset(&rates, 0, sizeof(rates));
rates.connid = htobe32(UATH_ID_BSS); /* XXX */
rates.size = htobe32(sizeof(struct uath_cmd_rateset));
/* XXX bounds check rs->rs_nrates */
rates.rateset.length = rs->rs_nrates;
bcopy(rs->rs_rates, &rates.rateset.set[0], rs->rs_nrates);
DPRINTF(sc, UATH_DEBUG_RATES,
"setting supported rates nrates=%d\n", rs->rs_nrates);
return uath_cmd_write(sc, WDCMSG_SET_BASIC_RATE,
&rates, sizeof rates, 0);
}
static int
uath_write_associd(struct uath_softc *sc)
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
struct ieee80211_node *ni;
struct uath_cmd_set_associd associd;
ni = ieee80211_ref_node(vap->iv_bss);
memset(&associd, 0, sizeof(associd));
associd.defaultrateix = htobe32(1); /* XXX */
associd.associd = htobe32(ni->ni_associd);
associd.timoffset = htobe32(0x3b); /* XXX */
IEEE80211_ADDR_COPY(associd.bssid, ni->ni_bssid);
ieee80211_free_node(ni);
return uath_cmd_write(sc, WDCMSG_WRITE_ASSOCID, &associd,
sizeof associd, 0);
}
static int
uath_set_ledsteady(struct uath_softc *sc, int lednum, int ledmode)
{
struct uath_cmd_ledsteady led;
led.lednum = htobe32(lednum);
led.ledmode = htobe32(ledmode);
DPRINTF(sc, UATH_DEBUG_LED, "set %s led %s (steady)\n",
(lednum == UATH_LED_LINK) ? "link" : "activity",
ledmode ? "on" : "off");
return uath_cmd_write(sc, WDCMSG_SET_LED_STEADY, &led, sizeof led, 0);
}
static int
uath_set_ledblink(struct uath_softc *sc, int lednum, int ledmode,
int blinkrate, int slowmode)
{
struct uath_cmd_ledblink led;
led.lednum = htobe32(lednum);
led.ledmode = htobe32(ledmode);
led.blinkrate = htobe32(blinkrate);
led.slowmode = htobe32(slowmode);
DPRINTF(sc, UATH_DEBUG_LED, "set %s led %s (blink)\n",
(lednum == UATH_LED_LINK) ? "link" : "activity",
ledmode ? "on" : "off");
return uath_cmd_write(sc, WDCMSG_SET_LED_BLINK, &led, sizeof led, 0);
}
static int
uath_newstate(struct ieee80211vap *vap, enum ieee80211_state nstate, int arg)
{
enum ieee80211_state ostate = vap->iv_state;
int error;
struct ieee80211_node *ni;
struct ieee80211com *ic = vap->iv_ic;
struct uath_softc *sc = ic->ic_softc;
struct uath_vap *uvp = UATH_VAP(vap);
DPRINTF(sc, UATH_DEBUG_STATE,
"%s: %s -> %s\n", __func__, ieee80211_state_name[vap->iv_state],
ieee80211_state_name[nstate]);
IEEE80211_UNLOCK(ic);
UATH_LOCK(sc);
callout_stop(&sc->stat_ch);
callout_stop(&sc->watchdog_ch);
ni = ieee80211_ref_node(vap->iv_bss);
switch (nstate) {
case IEEE80211_S_INIT:
if (ostate == IEEE80211_S_RUN) {
/* turn link and activity LEDs off */
uath_set_ledstate(sc, 0);
}
break;
case IEEE80211_S_SCAN:
break;
case IEEE80211_S_AUTH:
/* XXX good place? set RTS threshold */
uath_config(sc, CFG_USER_RTS_THRESHOLD, vap->iv_rtsthreshold);
/* XXX bad place */
error = uath_set_keys(sc, vap);
if (error != 0) {
device_printf(sc->sc_dev,
"could not set crypto keys, error %d\n", error);
break;
}
if (uath_switch_channel(sc, ni->ni_chan) != 0) {
device_printf(sc->sc_dev, "could not switch channel\n");
break;
}
if (uath_create_connection(sc, UATH_ID_BSS) != 0) {
device_printf(sc->sc_dev,
"could not create connection\n");
break;
}
break;
case IEEE80211_S_ASSOC:
if (uath_set_rates(sc, &ni->ni_rates) != 0) {
device_printf(sc->sc_dev,
"could not set negotiated rate set\n");
break;
}
break;
case IEEE80211_S_RUN:
/* XXX monitor mode doesn't be tested */
if (ic->ic_opmode == IEEE80211_M_MONITOR) {
uath_set_ledstate(sc, 1);
break;
}
/*
* Tx rate is controlled by firmware, report the maximum
* negotiated rate in ifconfig output.
*/
ni->ni_txrate = ni->ni_rates.rs_rates[ni->ni_rates.rs_nrates-1];
if (uath_write_associd(sc) != 0) {
device_printf(sc->sc_dev,
"could not write association id\n");
break;
}
/* turn link LED on */
uath_set_ledsteady(sc, UATH_LED_LINK, UATH_LED_ON);
/* make activity LED blink */
uath_set_ledblink(sc, UATH_LED_ACTIVITY, UATH_LED_ON, 1, 2);
/* set state to associated */
uath_set_ledstate(sc, 1);
/* start statistics timer */
callout_reset(&sc->stat_ch, hz, uath_stat, sc);
break;
default:
break;
}
ieee80211_free_node(ni);
UATH_UNLOCK(sc);
IEEE80211_LOCK(ic);
return (uvp->newstate(vap, nstate, arg));
}
static int
uath_set_key(struct uath_softc *sc, const struct ieee80211_key *wk,
int index)
{
#if 0
struct uath_cmd_crypto crypto;
int i;
memset(&crypto, 0, sizeof(crypto));
crypto.keyidx = htobe32(index);
crypto.magic1 = htobe32(1);
crypto.size = htobe32(368);
crypto.mask = htobe32(0xffff);
crypto.flags = htobe32(0x80000068);
if (index != UATH_DEFAULT_KEY)
crypto.flags |= htobe32(index << 16);
memset(crypto.magic2, 0xff, sizeof(crypto.magic2));
/*
* Each byte of the key must be XOR'ed with 10101010 before being
* transmitted to the firmware.
*/
for (i = 0; i < wk->wk_keylen; i++)
crypto.key[i] = wk->wk_key[i] ^ 0xaa;
DPRINTF(sc, UATH_DEBUG_CRYPTO,
"setting crypto key index=%d len=%d\n", index, wk->wk_keylen);
return uath_cmd_write(sc, WDCMSG_SET_KEY_CACHE_ENTRY, &crypto,
sizeof crypto, 0);
#else
/* XXX support H/W cryto */
return (0);
#endif
}
static int
uath_set_keys(struct uath_softc *sc, struct ieee80211vap *vap)
{
int i, error;
error = 0;
for (i = 0; i < IEEE80211_WEP_NKID; i++) {
const struct ieee80211_key *wk = &vap->iv_nw_keys[i];
if (wk->wk_flags & (IEEE80211_KEY_XMIT|IEEE80211_KEY_RECV)) {
error = uath_set_key(sc, wk, i);
if (error)
return (error);
}
}
if (vap->iv_def_txkey != IEEE80211_KEYIX_NONE) {
error = uath_set_key(sc, &vap->iv_nw_keys[vap->iv_def_txkey],
UATH_DEFAULT_KEY);
}
return (error);
}
#define UATH_SYSCTL_STAT_ADD32(c, h, n, p, d) \
SYSCTL_ADD_UINT(c, h, OID_AUTO, n, CTLFLAG_RD, p, 0, d)
static void
uath_sysctl_node(struct uath_softc *sc)
{
struct sysctl_ctx_list *ctx;
struct sysctl_oid_list *child;
struct sysctl_oid *tree;
struct uath_stat *stats;
stats = &sc->sc_stat;
ctx = device_get_sysctl_ctx(sc->sc_dev);
child = SYSCTL_CHILDREN(device_get_sysctl_tree(sc->sc_dev));
tree = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, "stats",
CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "UATH statistics");
child = SYSCTL_CHILDREN(tree);
UATH_SYSCTL_STAT_ADD32(ctx, child, "badchunkseqnum",
&stats->st_badchunkseqnum, "Bad chunk sequence numbers");
UATH_SYSCTL_STAT_ADD32(ctx, child, "invalidlen", &stats->st_invalidlen,
"Invalid length");
UATH_SYSCTL_STAT_ADD32(ctx, child, "multichunk", &stats->st_multichunk,
"Multi chunks");
UATH_SYSCTL_STAT_ADD32(ctx, child, "toobigrxpkt",
&stats->st_toobigrxpkt, "Too big rx packets");
UATH_SYSCTL_STAT_ADD32(ctx, child, "stopinprogress",
&stats->st_stopinprogress, "Stop in progress");
UATH_SYSCTL_STAT_ADD32(ctx, child, "crcerrs", &stats->st_crcerr,
"CRC errors");
UATH_SYSCTL_STAT_ADD32(ctx, child, "phyerr", &stats->st_phyerr,
"PHY errors");
UATH_SYSCTL_STAT_ADD32(ctx, child, "decrypt_crcerr",
&stats->st_decrypt_crcerr, "Decryption CRC errors");
UATH_SYSCTL_STAT_ADD32(ctx, child, "decrypt_micerr",
&stats->st_decrypt_micerr, "Decryption Misc errors");
UATH_SYSCTL_STAT_ADD32(ctx, child, "decomperr", &stats->st_decomperr,
"Decomp errors");
UATH_SYSCTL_STAT_ADD32(ctx, child, "keyerr", &stats->st_keyerr,
"Key errors");
UATH_SYSCTL_STAT_ADD32(ctx, child, "err", &stats->st_err,
"Unknown errors");
UATH_SYSCTL_STAT_ADD32(ctx, child, "cmd_active",
&stats->st_cmd_active, "Active numbers in Command queue");
UATH_SYSCTL_STAT_ADD32(ctx, child, "cmd_inactive",
&stats->st_cmd_inactive, "Inactive numbers in Command queue");
UATH_SYSCTL_STAT_ADD32(ctx, child, "cmd_pending",
&stats->st_cmd_pending, "Pending numbers in Command queue");
UATH_SYSCTL_STAT_ADD32(ctx, child, "cmd_waiting",
&stats->st_cmd_waiting, "Waiting numbers in Command queue");
UATH_SYSCTL_STAT_ADD32(ctx, child, "rx_active",
&stats->st_rx_active, "Active numbers in RX queue");
UATH_SYSCTL_STAT_ADD32(ctx, child, "rx_inactive",
&stats->st_rx_inactive, "Inactive numbers in RX queue");
UATH_SYSCTL_STAT_ADD32(ctx, child, "tx_active",
&stats->st_tx_active, "Active numbers in TX queue");
UATH_SYSCTL_STAT_ADD32(ctx, child, "tx_inactive",
&stats->st_tx_inactive, "Inactive numbers in TX queue");
UATH_SYSCTL_STAT_ADD32(ctx, child, "tx_pending",
&stats->st_tx_pending, "Pending numbers in TX queue");
}
#undef UATH_SYSCTL_STAT_ADD32
CTASSERT(sizeof(u_int) >= sizeof(uint32_t));
static void
uath_cmdeof(struct uath_softc *sc, struct uath_cmd *cmd)
{
struct uath_cmd_hdr *hdr;
uint32_t dlen;
hdr = (struct uath_cmd_hdr *)cmd->buf;
/* NB: msgid is passed thru w/o byte swapping */
#ifdef UATH_DEBUG
if (sc->sc_debug & UATH_DEBUG_CMDS) {
uint32_t len = be32toh(hdr->len);
printf("%s: %s [ix %u] len %u status %u\n",
__func__, uath_codename(be32toh(hdr->code)),
hdr->msgid, len, be32toh(hdr->magic));
if (sc->sc_debug & UATH_DEBUG_CMDS_DUMP)
uath_dump_cmd(cmd->buf,
len > UATH_MAX_CMDSZ ? sizeof(*hdr) : len, '-');
}
#endif
hdr->code = be32toh(hdr->code);
hdr->len = be32toh(hdr->len);
hdr->magic = be32toh(hdr->magic); /* target status on return */
switch (hdr->code & 0xff) {
/* reply to a read command */
default:
DPRINTF(sc, UATH_DEBUG_RX_PROC | UATH_DEBUG_RECV_ALL,
"%s: code %d hdr len %u\n",
__func__, hdr->code & 0xff, hdr->len);
/*
* The first response from the target after the
* HOST_AVAILABLE has an invalid msgid so we must
* treat it specially.
*/
if (hdr->msgid < UATH_CMD_LIST_COUNT) {
uint32_t *rp = (uint32_t *)(hdr+1);
u_int olen;
if (sizeof(*hdr) > hdr->len ||
hdr->len >= UATH_MAX_CMDSZ) {
device_printf(sc->sc_dev,
"%s: invalid WDC msg length %u; "
"msg ignored\n", __func__, hdr->len);
return;
}
/*
* Calculate return/receive payload size; the
* first word, if present, always gives the
* number of bytes--unless it's 0 in which
* case a single 32-bit word should be present.
*/
dlen = hdr->len - sizeof(*hdr);
if (dlen >= sizeof(uint32_t)) {
olen = be32toh(rp[0]);
dlen -= sizeof(uint32_t);
if (olen == 0) {
/* convention is 0 =>'s one word */
olen = sizeof(uint32_t);
/* XXX KASSERT(olen == dlen ) */
}
} else
olen = 0;
if (cmd->odata != NULL) {
/* NB: cmd->olen validated in uath_cmd */
if (olen > (u_int)cmd->olen) {
/* XXX complain? */
device_printf(sc->sc_dev,
"%s: cmd 0x%x olen %u cmd olen %u\n",
__func__, hdr->code, olen,
cmd->olen);
olen = cmd->olen;
}
if (olen > dlen) {
/* XXX complain, shouldn't happen */
device_printf(sc->sc_dev,
"%s: cmd 0x%x olen %u dlen %u\n",
__func__, hdr->code, olen, dlen);
olen = dlen;
}
/* XXX have submitter do this */
/* copy answer into caller's supplied buffer */
bcopy(&rp[1], cmd->odata, olen);
cmd->olen = olen;
}
}
wakeup_one(cmd); /* wake up caller */
break;
case WDCMSG_TARGET_START:
if (hdr->msgid >= UATH_CMD_LIST_COUNT) {
/* XXX */
return;
}
dlen = hdr->len - sizeof(*hdr);
if (dlen != sizeof(uint32_t)) {
device_printf(sc->sc_dev,
"%s: dlen (%u) != %zu!\n",
__func__, dlen, sizeof(uint32_t));
return;
}
/* XXX have submitter do this */
/* copy answer into caller's supplied buffer */
bcopy(hdr+1, cmd->odata, sizeof(uint32_t));
cmd->olen = sizeof(uint32_t);
wakeup_one(cmd); /* wake up caller */
break;
case WDCMSG_SEND_COMPLETE:
/* this notification is sent when UATH_TX_NOTIFY is set */
DPRINTF(sc, UATH_DEBUG_RX_PROC | UATH_DEBUG_RECV_ALL,
"%s: received Tx notification\n", __func__);
break;
case WDCMSG_TARGET_GET_STATS:
DPRINTF(sc, UATH_DEBUG_RX_PROC | UATH_DEBUG_RECV_ALL,
"%s: received device statistics\n", __func__);
callout_reset(&sc->stat_ch, hz, uath_stat, sc);
break;
}
}
static void
uath_intr_rx_callback(struct usb_xfer *xfer, usb_error_t error)
{
struct uath_softc *sc = usbd_xfer_softc(xfer);
struct uath_cmd *cmd;
struct uath_cmd_hdr *hdr;
struct usb_page_cache *pc;
int actlen;
usbd_xfer_status(xfer, &actlen, NULL, NULL, NULL);
UATH_ASSERT_LOCKED(sc);
switch (USB_GET_STATE(xfer)) {
case USB_ST_TRANSFERRED:
cmd = STAILQ_FIRST(&sc->sc_cmd_waiting);
if (cmd == NULL)
goto setup;
STAILQ_REMOVE_HEAD(&sc->sc_cmd_waiting, next);
UATH_STAT_DEC(sc, st_cmd_waiting);
STAILQ_INSERT_TAIL(&sc->sc_cmd_inactive, cmd, next);
UATH_STAT_INC(sc, st_cmd_inactive);
if (actlen < sizeof(struct uath_cmd_hdr)) {
device_printf(sc->sc_dev,
"%s: short xfer error (actlen %d)\n",
__func__, actlen);
goto setup;
}
pc = usbd_xfer_get_frame(xfer, 0);
usbd_copy_out(pc, 0, cmd->buf, actlen);
hdr = (struct uath_cmd_hdr *)cmd->buf;
hdr->len = be32toh(hdr->len);
if (hdr->len > (uint32_t)actlen) {
device_printf(sc->sc_dev,
"%s: truncated xfer (len %u, actlen %d)\n",
__func__, hdr->len, actlen);
goto setup;
}
uath_cmdeof(sc, cmd);
case USB_ST_SETUP:
setup:
usbd_xfer_set_frame_len(xfer, 0, usbd_xfer_max_len(xfer));
usbd_transfer_submit(xfer);
break;
default:
if (error != USB_ERR_CANCELLED) {
usbd_xfer_set_stall(xfer);
goto setup;
}
break;
}
}
static void
uath_intr_tx_callback(struct usb_xfer *xfer, usb_error_t error)
{
struct uath_softc *sc = usbd_xfer_softc(xfer);
struct uath_cmd *cmd;
UATH_ASSERT_LOCKED(sc);
cmd = STAILQ_FIRST(&sc->sc_cmd_active);
if (cmd != NULL && USB_GET_STATE(xfer) != USB_ST_SETUP) {
STAILQ_REMOVE_HEAD(&sc->sc_cmd_active, next);
UATH_STAT_DEC(sc, st_cmd_active);
STAILQ_INSERT_TAIL((cmd->flags & UATH_CMD_FLAG_READ) ?
&sc->sc_cmd_waiting : &sc->sc_cmd_inactive, cmd, next);
if (cmd->flags & UATH_CMD_FLAG_READ)
UATH_STAT_INC(sc, st_cmd_waiting);
else
UATH_STAT_INC(sc, st_cmd_inactive);
}
switch (USB_GET_STATE(xfer)) {
case USB_ST_TRANSFERRED:
case USB_ST_SETUP:
setup:
cmd = STAILQ_FIRST(&sc->sc_cmd_pending);
if (cmd == NULL) {
DPRINTF(sc, UATH_DEBUG_XMIT, "%s: empty pending queue\n",
__func__);
return;
}
STAILQ_REMOVE_HEAD(&sc->sc_cmd_pending, next);
UATH_STAT_DEC(sc, st_cmd_pending);
STAILQ_INSERT_TAIL((cmd->flags & UATH_CMD_FLAG_ASYNC) ?
&sc->sc_cmd_inactive : &sc->sc_cmd_active, cmd, next);
if (cmd->flags & UATH_CMD_FLAG_ASYNC)
UATH_STAT_INC(sc, st_cmd_inactive);
else
UATH_STAT_INC(sc, st_cmd_active);
usbd_xfer_set_frame_data(xfer, 0, cmd->buf, cmd->buflen);
usbd_transfer_submit(xfer);
break;
default:
if (error != USB_ERR_CANCELLED) {
usbd_xfer_set_stall(xfer);
goto setup;
}
break;
}
}
static void
uath_update_rxstat(struct uath_softc *sc, uint32_t status)
{
switch (status) {
case UATH_STATUS_STOP_IN_PROGRESS:
UATH_STAT_INC(sc, st_stopinprogress);
break;
case UATH_STATUS_CRC_ERR:
UATH_STAT_INC(sc, st_crcerr);
break;
case UATH_STATUS_PHY_ERR:
UATH_STAT_INC(sc, st_phyerr);
break;
case UATH_STATUS_DECRYPT_CRC_ERR:
UATH_STAT_INC(sc, st_decrypt_crcerr);
break;
case UATH_STATUS_DECRYPT_MIC_ERR:
UATH_STAT_INC(sc, st_decrypt_micerr);
break;
case UATH_STATUS_DECOMP_ERR:
UATH_STAT_INC(sc, st_decomperr);
break;
case UATH_STATUS_KEY_ERR:
UATH_STAT_INC(sc, st_keyerr);
break;
case UATH_STATUS_ERR:
UATH_STAT_INC(sc, st_err);
break;
default:
break;
}
}
CTASSERT(UATH_MIN_RXBUFSZ >= sizeof(struct uath_chunk));
static struct mbuf *
uath_data_rxeof(struct usb_xfer *xfer, struct uath_data *data,
struct uath_rx_desc **pdesc)
{
struct uath_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;
struct uath_chunk *chunk;
struct uath_rx_desc *desc;
struct mbuf *m = data->m, *mnew, *mp;
uint16_t chunklen;
int actlen;
usbd_xfer_status(xfer, &actlen, NULL, NULL, NULL);
if (actlen < (int)UATH_MIN_RXBUFSZ) {
DPRINTF(sc, UATH_DEBUG_RECV | UATH_DEBUG_RECV_ALL,
"%s: wrong xfer size (len=%d)\n", __func__, actlen);
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);
return (NULL);
}
chunk = (struct uath_chunk *)data->buf;
chunklen = be16toh(chunk->length);
if (chunk->seqnum == 0 && chunk->flags == 0 && chunklen == 0) {
device_printf(sc->sc_dev, "%s: strange response\n", __func__);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
counter_u64_add(ic->ic_ierrors, 1);
UATH_RESET_INTRX(sc);
return (NULL);
}
if (chunklen > actlen) {
device_printf(sc->sc_dev,
"%s: invalid chunk length (len %u > actlen %d)\n",
__func__, chunklen, actlen);
counter_u64_add(ic->ic_ierrors, 1);
/* XXX cleanup? */
UATH_RESET_INTRX(sc);
return (NULL);
}
if (chunk->seqnum != sc->sc_intrx_nextnum) {
DPRINTF(sc, UATH_DEBUG_XMIT, "invalid seqnum %d, expected %d\n",
chunk->seqnum, sc->sc_intrx_nextnum);
UATH_STAT_INC(sc, st_badchunkseqnum);
if (sc->sc_intrx_head != NULL)
m_freem(sc->sc_intrx_head);
UATH_RESET_INTRX(sc);
return (NULL);
}
/* check multi-chunk frames */
if ((chunk->seqnum == 0 && !(chunk->flags & UATH_CFLAGS_FINAL)) ||
(chunk->seqnum != 0 && (chunk->flags & UATH_CFLAGS_FINAL)) ||
chunk->flags & UATH_CFLAGS_RXMSG)
UATH_STAT_INC(sc, st_multichunk);
if (chunk->flags & UATH_CFLAGS_FINAL) {
if (chunklen < sizeof(struct uath_rx_desc)) {
device_printf(sc->sc_dev,
"%s: invalid chunk length %d\n",
__func__, chunklen);
counter_u64_add(ic->ic_ierrors, 1);
if (sc->sc_intrx_head != NULL)
m_freem(sc->sc_intrx_head);
UATH_RESET_INTRX(sc);
return (NULL);
}
chunklen -= sizeof(struct uath_rx_desc);
}
if (chunklen > 0 &&
(!(chunk->flags & UATH_CFLAGS_FINAL) || !(chunk->seqnum == 0))) {
/* we should use intermediate RX buffer */
if (chunk->seqnum == 0)
UATH_RESET_INTRX(sc);
if ((sc->sc_intrx_len + sizeof(struct uath_rx_desc) +
chunklen) > UATH_MAX_INTRX_SIZE) {
UATH_STAT_INC(sc, st_invalidlen);
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);
if (sc->sc_intrx_head != NULL)
m_freem(sc->sc_intrx_head);
UATH_RESET_INTRX(sc);
return (NULL);
}
m->m_len = chunklen;
m->m_data += sizeof(struct uath_chunk);
if (sc->sc_intrx_head == NULL) {
sc->sc_intrx_head = m;
sc->sc_intrx_tail = m;
} else {
m->m_flags &= ~M_PKTHDR;
sc->sc_intrx_tail->m_next = m;
sc->sc_intrx_tail = m;
}
}
sc->sc_intrx_len += chunklen;
mnew = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
if (mnew == NULL) {
DPRINTF(sc, UATH_DEBUG_RECV | UATH_DEBUG_RECV_ALL,
"%s: can't get new mbuf, drop frame\n", __func__);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
counter_u64_add(ic->ic_ierrors, 1);
if (sc->sc_intrx_head != NULL)
m_freem(sc->sc_intrx_head);
UATH_RESET_INTRX(sc);
return (NULL);
}
data->m = mnew;
data->buf = mtod(mnew, uint8_t *);
/* if the frame is not final continue the transfer */
if (!(chunk->flags & UATH_CFLAGS_FINAL)) {
sc->sc_intrx_nextnum++;
UATH_RESET_INTRX(sc);
return (NULL);
}
/*
* if the frame is not set UATH_CFLAGS_RXMSG, then rx descriptor is
* located at the end, 32-bit aligned
*/
desc = (chunk->flags & UATH_CFLAGS_RXMSG) ?
(struct uath_rx_desc *)(chunk + 1) :
(struct uath_rx_desc *)(((uint8_t *)chunk) +
sizeof(struct uath_chunk) + be16toh(chunk->length) -
sizeof(struct uath_rx_desc));
if ((uint8_t *)chunk + actlen - sizeof(struct uath_rx_desc) <
(uint8_t *)desc) {
device_printf(sc->sc_dev,
"%s: wrong Rx descriptor pointer "
"(desc %p chunk %p actlen %d)\n",
__func__, desc, chunk, actlen);
counter_u64_add(ic->ic_ierrors, 1);
if (sc->sc_intrx_head != NULL)
m_freem(sc->sc_intrx_head);
UATH_RESET_INTRX(sc);
return (NULL);
}
*pdesc = desc;
DPRINTF(sc, UATH_DEBUG_RECV | UATH_DEBUG_RECV_ALL,
"%s: frame len %u code %u status %u rate %u antenna %u "
"rssi %d channel %u phyerror %u connix %u decrypterror %u "
"keycachemiss %u\n", __func__, be32toh(desc->framelen)
, be32toh(desc->code), be32toh(desc->status), be32toh(desc->rate)
, be32toh(desc->antenna), be32toh(desc->rssi), be32toh(desc->channel)
, be32toh(desc->phyerror), be32toh(desc->connix)
, be32toh(desc->decrypterror), be32toh(desc->keycachemiss));
if (be32toh(desc->len) > MCLBYTES) {
DPRINTF(sc, UATH_DEBUG_RECV | UATH_DEBUG_RECV_ALL,
"%s: bad descriptor (len=%d)\n", __func__,
be32toh(desc->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);
UATH_STAT_INC(sc, st_toobigrxpkt);
if (sc->sc_intrx_head != NULL)
m_freem(sc->sc_intrx_head);
UATH_RESET_INTRX(sc);
return (NULL);
}
uath_update_rxstat(sc, be32toh(desc->status));
/* finalize mbuf */
if (sc->sc_intrx_head == NULL) {
uint32_t framelen;
if (be32toh(desc->framelen) < UATH_RX_DUMMYSIZE) {
device_printf(sc->sc_dev,
"%s: framelen too small (%u)\n",
__func__, be32toh(desc->framelen));
counter_u64_add(ic->ic_ierrors, 1);
if (sc->sc_intrx_head != NULL)
m_freem(sc->sc_intrx_head);
UATH_RESET_INTRX(sc);
return (NULL);
}
framelen = be32toh(desc->framelen) - UATH_RX_DUMMYSIZE;
if (framelen > actlen - sizeof(struct uath_chunk) ||
framelen < sizeof(struct ieee80211_frame_ack)) {
device_printf(sc->sc_dev,
"%s: wrong frame length (%u, actlen %d)!\n",
__func__, framelen, actlen);
counter_u64_add(ic->ic_ierrors, 1);
if (sc->sc_intrx_head != NULL)
m_freem(sc->sc_intrx_head);
UATH_RESET_INTRX(sc);
return (NULL);
}
m->m_pkthdr.len = m->m_len = framelen;
m->m_data += sizeof(struct uath_chunk);
} else {
mp = sc->sc_intrx_head;
mp->m_flags |= M_PKTHDR;
mp->m_pkthdr.len = sc->sc_intrx_len;
m = mp;
}
/* there are a lot more fields in the RX descriptor */
if ((sc->sc_flags & UATH_FLAG_INVALID) == 0 &&
ieee80211_radiotap_active(ic)) {
struct uath_rx_radiotap_header *tap = &sc->sc_rxtap;
uint32_t tsf_hi = be32toh(desc->tstamp_high);
uint32_t tsf_lo = be32toh(desc->tstamp_low);
/* XXX only get low order 24bits of tsf from h/w */
tap->wr_tsf = htole64(((uint64_t)tsf_hi << 32) | tsf_lo);
tap->wr_flags = 0;
if (be32toh(desc->status) == UATH_STATUS_CRC_ERR)
tap->wr_flags |= IEEE80211_RADIOTAP_F_BADFCS;
/* XXX map other status to BADFCS? */
/* XXX ath h/w rate code, need to map */
tap->wr_rate = be32toh(desc->rate);
tap->wr_antenna = be32toh(desc->antenna);
tap->wr_antsignal = -95 + be32toh(desc->rssi);
tap->wr_antnoise = -95;
}
UATH_RESET_INTRX(sc);
return (m);
}
static void
uath_bulk_rx_callback(struct usb_xfer *xfer, usb_error_t error)
{
struct uath_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;
struct ieee80211_frame *wh;
struct ieee80211_node *ni;
struct epoch_tracker et;
struct mbuf *m = NULL;
struct uath_data *data;
struct uath_rx_desc *desc = NULL;
int8_t nf;
UATH_ASSERT_LOCKED(sc);
switch (USB_GET_STATE(xfer)) {
case USB_ST_TRANSFERRED:
data = STAILQ_FIRST(&sc->sc_rx_active);
if (data == NULL)
goto setup;
STAILQ_REMOVE_HEAD(&sc->sc_rx_active, next);
UATH_STAT_DEC(sc, st_rx_active);
m = uath_data_rxeof(xfer, data, &desc);
STAILQ_INSERT_TAIL(&sc->sc_rx_inactive, data, next);
UATH_STAT_INC(sc, st_rx_inactive);
/* FALLTHROUGH */
case USB_ST_SETUP:
setup:
data = STAILQ_FIRST(&sc->sc_rx_inactive);
if (data == NULL)
return;
STAILQ_REMOVE_HEAD(&sc->sc_rx_inactive, next);
UATH_STAT_DEC(sc, st_rx_inactive);
STAILQ_INSERT_TAIL(&sc->sc_rx_active, data, next);
UATH_STAT_INC(sc, st_rx_active);
usbd_xfer_set_frame_data(xfer, 0, data->buf, MCLBYTES);
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.
*/
if (sc->sc_flags & UATH_FLAG_INVALID) {
if (m != NULL)
m_freem(m);
return;
}
UATH_UNLOCK(sc);
if (m != NULL && desc != NULL) {
wh = mtod(m, struct ieee80211_frame *);
ni = ieee80211_find_rxnode(ic,
(struct ieee80211_frame_min *)wh);
nf = -95; /* XXX */
NET_EPOCH_ENTER(et);
if (ni != NULL) {
(void) ieee80211_input(ni, m,
(int)be32toh(desc->rssi), nf);
/* node is no longer needed */
ieee80211_free_node(ni);
} else
(void) ieee80211_input_all(ic, m,
(int)be32toh(desc->rssi), nf);
NET_EPOCH_EXIT(et);
m = NULL;
desc = NULL;
}
UATH_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
uath_start(sc);
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);
UATH_STAT_DEC(sc, st_rx_active);
STAILQ_INSERT_TAIL(&sc->sc_rx_inactive, data, next);
UATH_STAT_INC(sc, st_rx_inactive);
}
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);
goto setup;
}
break;
}
}
static void
uath_data_txeof(struct usb_xfer *xfer, struct uath_data *data)
{
struct uath_softc *sc = usbd_xfer_softc(xfer);
UATH_ASSERT_LOCKED(sc);
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);
data->m = NULL;
data->ni = NULL;
}
sc->sc_tx_timer = 0;
}
static void
uath_bulk_tx_callback(struct usb_xfer *xfer, usb_error_t error)
{
struct uath_softc *sc = usbd_xfer_softc(xfer);
struct uath_data *data;
UATH_ASSERT_LOCKED(sc);
switch (USB_GET_STATE(xfer)) {
case USB_ST_TRANSFERRED:
data = STAILQ_FIRST(&sc->sc_tx_active);
if (data == NULL)
goto setup;
STAILQ_REMOVE_HEAD(&sc->sc_tx_active, next);
UATH_STAT_DEC(sc, st_tx_active);
uath_data_txeof(xfer, data);
STAILQ_INSERT_TAIL(&sc->sc_tx_inactive, data, next);
UATH_STAT_INC(sc, st_tx_inactive);
/* FALLTHROUGH */
case USB_ST_SETUP:
setup:
data = STAILQ_FIRST(&sc->sc_tx_pending);
if (data == NULL) {
DPRINTF(sc, UATH_DEBUG_XMIT, "%s: empty pending queue\n",
__func__);
return;
}
STAILQ_REMOVE_HEAD(&sc->sc_tx_pending, next);
UATH_STAT_DEC(sc, st_tx_pending);
STAILQ_INSERT_TAIL(&sc->sc_tx_active, data, next);
UATH_STAT_INC(sc, st_tx_active);
usbd_xfer_set_frame_data(xfer, 0, data->buf, data->buflen);
usbd_transfer_submit(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
uath_start(sc);
break;
default:
data = STAILQ_FIRST(&sc->sc_tx_active);
if (data == NULL)
goto setup;
if (data->ni != 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_inc_counter(data->ni->ni_vap->iv_ifp,
IFCOUNTER_OERRORS, 1);
if ((sc->sc_flags & UATH_FLAG_INVALID) == 0)
ieee80211_free_node(data->ni);
data->ni = NULL;
}
if (error != USB_ERR_CANCELLED) {
usbd_xfer_set_stall(xfer);
goto setup;
}
break;
}
}
static device_method_t uath_methods[] = {
DEVMETHOD(device_probe, uath_match),
DEVMETHOD(device_attach, uath_attach),
DEVMETHOD(device_detach, uath_detach),
DEVMETHOD_END
};
static driver_t uath_driver = {
.name = "uath",
.methods = uath_methods,
.size = sizeof(struct uath_softc)
};
static devclass_t uath_devclass;
DRIVER_MODULE(uath, uhub, uath_driver, uath_devclass, NULL, 0);
MODULE_DEPEND(uath, wlan, 1, 1, 1);
MODULE_DEPEND(uath, usb, 1, 1, 1);
MODULE_VERSION(uath, 1);
USB_PNP_HOST_INFO(uath_devs);