freebsd-dev/sys/dev/ral/rt2860.c

4331 lines
114 KiB
C
Raw Normal View History

/*-
* Copyright (c) 2007-2010 Damien Bergamini <damien.bergamini@free.fr>
* Copyright (c) 2012 Bernhard Schmidt <bschmidt@FreeBSD.org>
*
* 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.
*
* $OpenBSD: rt2860.c,v 1.65 2010/10/23 14:24:54 damien Exp $
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
/*-
* Ralink Technology RT2860/RT3090/RT3390/RT3562/RT5390/RT5392 chipset driver
* http://www.ralinktech.com/
*/
#include <sys/param.h>
#include <sys/sysctl.h>
#include <sys/sockio.h>
#include <sys/mbuf.h>
#include <sys/kernel.h>
#include <sys/socket.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/module.h>
#include <sys/bus.h>
#include <sys/endian.h>
#include <sys/firmware.h>
#include <machine/bus.h>
#include <machine/resource.h>
#include <sys/rman.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>
#include <net80211/ieee80211_var.h>
#include <net80211/ieee80211_radiotap.h>
#include <net80211/ieee80211_regdomain.h>
#include <net80211/ieee80211_ratectl.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
#include <netinet/ip.h>
#include <netinet/if_ether.h>
#include <dev/ral/rt2860reg.h>
#include <dev/ral/rt2860var.h>
#define RAL_DEBUG
#ifdef RAL_DEBUG
#define DPRINTF(x) do { if (sc->sc_debug > 0) printf x; } while (0)
#define DPRINTFN(n, x) do { if (sc->sc_debug >= (n)) printf x; } while (0)
#else
#define DPRINTF(x)
#define DPRINTFN(n, x)
#endif
static struct ieee80211vap *rt2860_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 rt2860_vap_delete(struct ieee80211vap *);
static void rt2860_dma_map_addr(void *, bus_dma_segment_t *, int, int);
static int rt2860_alloc_tx_ring(struct rt2860_softc *,
struct rt2860_tx_ring *);
static void rt2860_reset_tx_ring(struct rt2860_softc *,
struct rt2860_tx_ring *);
static void rt2860_free_tx_ring(struct rt2860_softc *,
struct rt2860_tx_ring *);
static int rt2860_alloc_tx_pool(struct rt2860_softc *);
static void rt2860_free_tx_pool(struct rt2860_softc *);
static int rt2860_alloc_rx_ring(struct rt2860_softc *,
struct rt2860_rx_ring *);
static void rt2860_reset_rx_ring(struct rt2860_softc *,
struct rt2860_rx_ring *);
static void rt2860_free_rx_ring(struct rt2860_softc *,
struct rt2860_rx_ring *);
static void rt2860_updatestats(struct rt2860_softc *);
static void rt2860_newassoc(struct ieee80211_node *, int);
static void rt2860_node_free(struct ieee80211_node *);
#ifdef IEEE80211_HT
static int rt2860_ampdu_rx_start(struct ieee80211com *,
struct ieee80211_node *, uint8_t);
static void rt2860_ampdu_rx_stop(struct ieee80211com *,
struct ieee80211_node *, uint8_t);
#endif
static int rt2860_newstate(struct ieee80211vap *, enum ieee80211_state,
int);
static uint16_t rt3090_efuse_read_2(struct rt2860_softc *, uint16_t);
static uint16_t rt2860_eeprom_read_2(struct rt2860_softc *, uint16_t);
static void rt2860_intr_coherent(struct rt2860_softc *);
static void rt2860_drain_stats_fifo(struct rt2860_softc *);
static void rt2860_tx_intr(struct rt2860_softc *, int);
static void rt2860_rx_intr(struct rt2860_softc *);
static void rt2860_tbtt_intr(struct rt2860_softc *);
static void rt2860_gp_intr(struct rt2860_softc *);
static int rt2860_tx(struct rt2860_softc *, struct mbuf *,
struct ieee80211_node *);
static int rt2860_raw_xmit(struct ieee80211_node *, struct mbuf *,
const struct ieee80211_bpf_params *);
static int rt2860_tx_raw(struct rt2860_softc *, struct mbuf *,
struct ieee80211_node *,
const struct ieee80211_bpf_params *params);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
static int rt2860_transmit(struct ieee80211com *, struct mbuf *);
static void rt2860_start(struct rt2860_softc *);
static void rt2860_watchdog(void *);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
static void rt2860_parent(struct ieee80211com *);
static void rt2860_mcu_bbp_write(struct rt2860_softc *, uint8_t, uint8_t);
static uint8_t rt2860_mcu_bbp_read(struct rt2860_softc *, uint8_t);
static void rt2860_rf_write(struct rt2860_softc *, uint8_t, uint32_t);
static uint8_t rt3090_rf_read(struct rt2860_softc *, uint8_t);
static void rt3090_rf_write(struct rt2860_softc *, uint8_t, uint8_t);
static int rt2860_mcu_cmd(struct rt2860_softc *, uint8_t, uint16_t, int);
static void rt2860_enable_mrr(struct rt2860_softc *);
static void rt2860_set_txpreamble(struct rt2860_softc *);
static void rt2860_set_basicrates(struct rt2860_softc *,
const struct ieee80211_rateset *);
static void rt2860_scan_start(struct ieee80211com *);
static void rt2860_scan_end(struct ieee80211com *);
static void rt2860_set_channel(struct ieee80211com *);
static void rt2860_select_chan_group(struct rt2860_softc *, int);
static void rt2860_set_chan(struct rt2860_softc *, u_int);
static void rt3090_set_chan(struct rt2860_softc *, u_int);
static void rt5390_set_chan(struct rt2860_softc *, u_int);
static int rt3090_rf_init(struct rt2860_softc *);
static void rt5390_rf_init(struct rt2860_softc *);
static void rt3090_rf_wakeup(struct rt2860_softc *);
static void rt5390_rf_wakeup(struct rt2860_softc *);
static int rt3090_filter_calib(struct rt2860_softc *, uint8_t, uint8_t,
uint8_t *);
static void rt3090_rf_setup(struct rt2860_softc *);
static void rt2860_set_leds(struct rt2860_softc *, uint16_t);
static void rt2860_set_gp_timer(struct rt2860_softc *, int);
static void rt2860_set_bssid(struct rt2860_softc *, const uint8_t *);
static void rt2860_set_macaddr(struct rt2860_softc *, const uint8_t *);
static void rt2860_update_promisc(struct ieee80211com *);
static void rt2860_updateslot(struct ieee80211com *);
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 rt2860_updateprot(struct rt2860_softc *);
static int rt2860_updateedca(struct ieee80211com *);
#ifdef HW_CRYPTO
static int rt2860_set_key(struct ieee80211com *, struct ieee80211_node *,
struct ieee80211_key *);
static void rt2860_delete_key(struct ieee80211com *,
struct ieee80211_node *, struct ieee80211_key *);
#endif
static int8_t rt2860_rssi2dbm(struct rt2860_softc *, uint8_t, uint8_t);
static const char *rt2860_get_rf(uint8_t);
static int rt2860_read_eeprom(struct rt2860_softc *,
uint8_t macaddr[IEEE80211_ADDR_LEN]);
static int rt2860_bbp_init(struct rt2860_softc *);
static void rt5390_bbp_init(struct rt2860_softc *);
static int rt2860_txrx_enable(struct rt2860_softc *);
static void rt2860_init(void *);
static void rt2860_init_locked(struct rt2860_softc *);
static void rt2860_stop(void *);
static void rt2860_stop_locked(struct rt2860_softc *);
static int rt2860_load_microcode(struct rt2860_softc *);
#ifdef NOT_YET
static void rt2860_calib(struct rt2860_softc *);
#endif
static void rt3090_set_rx_antenna(struct rt2860_softc *, int);
static void rt2860_switch_chan(struct rt2860_softc *,
struct ieee80211_channel *);
static int rt2860_setup_beacon(struct rt2860_softc *,
struct ieee80211vap *);
static void rt2860_enable_tsf_sync(struct rt2860_softc *);
static const struct {
uint32_t reg;
uint32_t val;
} rt2860_def_mac[] = {
RT2860_DEF_MAC
};
static const struct {
uint8_t reg;
uint8_t val;
} rt2860_def_bbp[] = {
RT2860_DEF_BBP
}, rt5390_def_bbp[] = {
RT5390_DEF_BBP
};
static const struct rfprog {
uint8_t chan;
uint32_t r1, r2, r3, r4;
} rt2860_rf2850[] = {
RT2860_RF2850
};
struct {
uint8_t n, r, k;
} rt3090_freqs[] = {
RT3070_RF3052
};
static const struct {
uint8_t reg;
uint8_t val;
} rt3090_def_rf[] = {
RT3070_DEF_RF
}, rt5390_def_rf[] = {
RT5390_DEF_RF
}, rt5392_def_rf[] = {
RT5392_DEF_RF
};
int
rt2860_attach(device_t dev, int id)
{
struct rt2860_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;
uint32_t tmp;
int error, ntries, qid;
uint8_t bands;
sc->sc_dev = dev;
sc->sc_debug = 0;
mtx_init(&sc->sc_mtx, device_get_nameunit(dev), MTX_NETWORK_LOCK,
MTX_DEF | MTX_RECURSE);
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);
/* wait for NIC to initialize */
for (ntries = 0; ntries < 100; ntries++) {
tmp = RAL_READ(sc, RT2860_ASIC_VER_ID);
if (tmp != 0 && tmp != 0xffffffff)
break;
DELAY(10);
}
if (ntries == 100) {
device_printf(sc->sc_dev,
"timeout waiting for NIC to initialize\n");
error = EIO;
goto fail1;
}
sc->mac_ver = tmp >> 16;
sc->mac_rev = tmp & 0xffff;
if (sc->mac_ver != 0x2860 &&
(id == 0x0681 || id == 0x0781 || id == 0x1059))
sc->sc_flags |= RT2860_ADVANCED_PS;
/* retrieve RF rev. no and various other things from EEPROM */
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
rt2860_read_eeprom(sc, ic->ic_macaddr);
device_printf(sc->sc_dev, "MAC/BBP RT%X (rev 0x%04X), "
"RF %s (MIMO %dT%dR), address %6D\n",
sc->mac_ver, sc->mac_rev, rt2860_get_rf(sc->rf_rev),
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->ntxchains, sc->nrxchains, ic->ic_macaddr, ":");
/*
* Allocate Tx (4 EDCAs + HCCA + Mgt) and Rx rings.
*/
for (qid = 0; qid < 6; qid++) {
if ((error = rt2860_alloc_tx_ring(sc, &sc->txq[qid])) != 0) {
device_printf(sc->sc_dev,
"could not allocate Tx ring %d\n", qid);
goto fail2;
}
}
if ((error = rt2860_alloc_rx_ring(sc, &sc->rxq)) != 0) {
device_printf(sc->sc_dev, "could not allocate Rx ring\n");
goto fail2;
}
if ((error = rt2860_alloc_tx_pool(sc)) != 0) {
device_printf(sc->sc_dev, "could not allocate Tx pool\n");
goto fail3;
}
/* mgmt ring is broken on RT2860C, use EDCA AC VO ring instead */
sc->mgtqid = (sc->mac_ver == 0x2860 && sc->mac_rev == 0x0100) ?
WME_AC_VO : 5;
ic->ic_softc = sc;
ic->ic_name = device_get_nameunit(dev);
ic->ic_opmode = IEEE80211_M_STA;
ic->ic_phytype = IEEE80211_T_OFDM; /* not only, but not used */
/* set device capabilities */
ic->ic_caps =
IEEE80211_C_STA /* station mode */
| IEEE80211_C_IBSS /* ibss, nee adhoc, mode */
| IEEE80211_C_HOSTAP /* hostap mode */
| IEEE80211_C_MONITOR /* monitor mode */
| IEEE80211_C_AHDEMO /* adhoc demo mode */
| IEEE80211_C_WDS /* 4-address traffic works */
| IEEE80211_C_MBSS /* mesh point link mode */
| IEEE80211_C_SHPREAMBLE /* short preamble supported */
| IEEE80211_C_SHSLOT /* short slot time supported */
| IEEE80211_C_WPA /* capable of WPA1+WPA2 */
#if 0
| IEEE80211_C_BGSCAN /* capable of bg scanning */
#endif
| IEEE80211_C_WME /* 802.11e */
;
bands = 0;
setbit(&bands, IEEE80211_MODE_11B);
setbit(&bands, IEEE80211_MODE_11G);
if (sc->rf_rev == RT2860_RF_2750 || sc->rf_rev == RT2860_RF_2850)
setbit(&bands, IEEE80211_MODE_11A);
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_wme.wme_update = rt2860_updateedca;
ic->ic_scan_start = rt2860_scan_start;
ic->ic_scan_end = rt2860_scan_end;
ic->ic_set_channel = rt2860_set_channel;
ic->ic_updateslot = rt2860_updateslot;
ic->ic_update_promisc = rt2860_update_promisc;
ic->ic_raw_xmit = rt2860_raw_xmit;
sc->sc_node_free = ic->ic_node_free;
ic->ic_node_free = rt2860_node_free;
ic->ic_newassoc = rt2860_newassoc;
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 = rt2860_transmit;
ic->ic_parent = rt2860_parent;
ic->ic_vap_create = rt2860_vap_create;
ic->ic_vap_delete = rt2860_vap_delete;
ieee80211_radiotap_attach(ic,
&sc->sc_txtap.wt_ihdr, sizeof(sc->sc_txtap),
RT2860_TX_RADIOTAP_PRESENT,
&sc->sc_rxtap.wr_ihdr, sizeof(sc->sc_rxtap),
RT2860_RX_RADIOTAP_PRESENT);
#ifdef RAL_DEBUG
SYSCTL_ADD_INT(device_get_sysctl_ctx(dev),
SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO,
"debug", CTLFLAG_RW, &sc->sc_debug, 0, "debug msgs");
#endif
if (bootverbose)
ieee80211_announce(ic);
return 0;
fail3: rt2860_free_rx_ring(sc, &sc->rxq);
fail2: while (--qid >= 0)
rt2860_free_tx_ring(sc, &sc->txq[qid]);
fail1: mtx_destroy(&sc->sc_mtx);
return error;
}
int
rt2860_detach(void *xsc)
{
struct rt2860_softc *sc = xsc;
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;
int qid;
RAL_LOCK(sc);
rt2860_stop_locked(sc);
RAL_UNLOCK(sc);
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);
for (qid = 0; qid < 6; qid++)
rt2860_free_tx_ring(sc, &sc->txq[qid]);
rt2860_free_rx_ring(sc, &sc->rxq);
rt2860_free_tx_pool(sc);
mtx_destroy(&sc->sc_mtx);
return 0;
}
void
rt2860_shutdown(void *xsc)
{
struct rt2860_softc *sc = xsc;
rt2860_stop(sc);
}
void
rt2860_suspend(void *xsc)
{
struct rt2860_softc *sc = xsc;
rt2860_stop(sc);
}
void
rt2860_resume(void *xsc)
{
struct rt2860_softc *sc = xsc;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
if (sc->sc_ic.ic_nrunning > 0)
rt2860_init(sc);
}
static struct ieee80211vap *
rt2860_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])
{
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 rt2860_softc *sc = ic->ic_softc;
struct rt2860_vap *rvp;
struct ieee80211vap *vap;
switch (opmode) {
case IEEE80211_M_STA:
case IEEE80211_M_IBSS:
case IEEE80211_M_AHDEMO:
case IEEE80211_M_MONITOR:
case IEEE80211_M_HOSTAP:
case IEEE80211_M_MBSS:
/* XXXRP: TBD */
if (!TAILQ_EMPTY(&ic->ic_vaps)) {
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
device_printf(sc->sc_dev, "only 1 vap supported\n");
return NULL;
}
if (opmode == IEEE80211_M_STA)
flags |= IEEE80211_CLONE_NOBEACONS;
break;
case IEEE80211_M_WDS:
if (TAILQ_EMPTY(&ic->ic_vaps) ||
ic->ic_opmode != IEEE80211_M_HOSTAP) {
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
device_printf(sc->sc_dev,
"wds only supported in ap mode\n");
return NULL;
}
/*
* Silently remove any request for a unique
* bssid; WDS vap's always share the local
* mac address.
*/
flags &= ~IEEE80211_CLONE_BSSID;
break;
default:
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
device_printf(sc->sc_dev, "unknown opmode %d\n", opmode);
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
rvp = malloc(sizeof(struct rt2860_vap), M_80211_VAP, M_WAITOK | M_ZERO);
vap = &rvp->ral_vap;
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ieee80211_vap_setup(ic, vap, name, unit, opmode, flags, bssid);
/* override state transition machine */
rvp->ral_newstate = vap->iv_newstate;
vap->iv_newstate = rt2860_newstate;
#if 0
vap->iv_update_beacon = rt2860_beacon_update;
#endif
/* HW supports up to 255 STAs (0-254) in HostAP and IBSS modes */
vap->iv_max_aid = min(IEEE80211_AID_MAX, RT2860_WCID_MAX);
ieee80211_ratectl_init(vap);
/* complete setup */
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ieee80211_vap_attach(vap, ieee80211_media_change,
ieee80211_media_status, mac);
if (TAILQ_FIRST(&ic->ic_vaps) == vap)
ic->ic_opmode = opmode;
return vap;
}
static void
rt2860_vap_delete(struct ieee80211vap *vap)
{
struct rt2860_vap *rvp = RT2860_VAP(vap);
ieee80211_ratectl_deinit(vap);
ieee80211_vap_detach(vap);
free(rvp, M_80211_VAP);
}
static void
rt2860_dma_map_addr(void *arg, bus_dma_segment_t *segs, int nseg, int error)
{
if (error != 0)
return;
KASSERT(nseg == 1, ("too many DMA segments, %d should be 1", nseg));
*(bus_addr_t *)arg = segs[0].ds_addr;
}
static int
rt2860_alloc_tx_ring(struct rt2860_softc *sc, struct rt2860_tx_ring *ring)
{
int size, error;
size = RT2860_TX_RING_COUNT * sizeof (struct rt2860_txd);
error = bus_dma_tag_create(bus_get_dma_tag(sc->sc_dev), 16, 0,
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL,
size, 1, size, 0, NULL, NULL, &ring->desc_dmat);
if (error != 0) {
2015-10-29 04:21:34 +00:00
device_printf(sc->sc_dev, "could not create desc DMA tag\n");
goto fail;
}
error = bus_dmamem_alloc(ring->desc_dmat, (void **)&ring->txd,
BUS_DMA_NOWAIT | BUS_DMA_ZERO, &ring->desc_map);
if (error != 0) {
device_printf(sc->sc_dev, "could not allocate DMA memory\n");
goto fail;
}
error = bus_dmamap_load(ring->desc_dmat, ring->desc_map, ring->txd,
size, rt2860_dma_map_addr, &ring->paddr, 0);
if (error != 0) {
device_printf(sc->sc_dev, "could not load desc DMA map\n");
goto fail;
}
bus_dmamap_sync(ring->desc_dmat, ring->desc_map, BUS_DMASYNC_PREWRITE);
return 0;
fail: rt2860_free_tx_ring(sc, ring);
return error;
}
void
rt2860_reset_tx_ring(struct rt2860_softc *sc, struct rt2860_tx_ring *ring)
{
struct rt2860_tx_data *data;
int i;
for (i = 0; i < RT2860_TX_RING_COUNT; i++) {
if ((data = ring->data[i]) == NULL)
continue; /* nothing mapped in this slot */
if (data->m != NULL) {
bus_dmamap_sync(sc->txwi_dmat, data->map,
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(sc->txwi_dmat, data->map);
m_freem(data->m);
data->m = NULL;
}
if (data->ni != NULL) {
ieee80211_free_node(data->ni);
data->ni = NULL;
}
SLIST_INSERT_HEAD(&sc->data_pool, data, next);
ring->data[i] = NULL;
}
ring->queued = 0;
ring->cur = ring->next = 0;
}
void
rt2860_free_tx_ring(struct rt2860_softc *sc, struct rt2860_tx_ring *ring)
{
struct rt2860_tx_data *data;
int i;
if (ring->txd != NULL) {
bus_dmamap_sync(ring->desc_dmat, ring->desc_map,
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(ring->desc_dmat, ring->desc_map);
bus_dmamem_free(ring->desc_dmat, ring->txd, ring->desc_map);
}
if (ring->desc_dmat != NULL)
bus_dma_tag_destroy(ring->desc_dmat);
for (i = 0; i < RT2860_TX_RING_COUNT; i++) {
if ((data = ring->data[i]) == NULL)
continue; /* nothing mapped in this slot */
if (data->m != NULL) {
bus_dmamap_sync(sc->txwi_dmat, data->map,
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(sc->txwi_dmat, data->map);
m_freem(data->m);
}
if (data->ni != NULL)
ieee80211_free_node(data->ni);
SLIST_INSERT_HEAD(&sc->data_pool, data, next);
}
}
/*
* Allocate a pool of TX Wireless Information blocks.
*/
int
rt2860_alloc_tx_pool(struct rt2860_softc *sc)
{
caddr_t vaddr;
bus_addr_t paddr;
int i, size, error;
size = RT2860_TX_POOL_COUNT * RT2860_TXWI_DMASZ;
/* init data_pool early in case of failure.. */
SLIST_INIT(&sc->data_pool);
error = bus_dma_tag_create(bus_get_dma_tag(sc->sc_dev), 1, 0,
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL,
size, 1, size, 0, NULL, NULL, &sc->txwi_dmat);
if (error != 0) {
device_printf(sc->sc_dev, "could not create txwi DMA tag\n");
goto fail;
}
error = bus_dmamem_alloc(sc->txwi_dmat, (void **)&sc->txwi_vaddr,
BUS_DMA_NOWAIT | BUS_DMA_ZERO, &sc->txwi_map);
if (error != 0) {
device_printf(sc->sc_dev, "could not allocate DMA memory\n");
goto fail;
}
error = bus_dmamap_load(sc->txwi_dmat, sc->txwi_map,
sc->txwi_vaddr, size, rt2860_dma_map_addr, &paddr, 0);
if (error != 0) {
device_printf(sc->sc_dev, "could not load txwi DMA map\n");
goto fail;
}
bus_dmamap_sync(sc->txwi_dmat, sc->txwi_map, BUS_DMASYNC_PREWRITE);
vaddr = sc->txwi_vaddr;
for (i = 0; i < RT2860_TX_POOL_COUNT; i++) {
struct rt2860_tx_data *data = &sc->data[i];
error = bus_dmamap_create(sc->txwi_dmat, 0, &data->map);
if (error != 0) {
device_printf(sc->sc_dev, "could not create DMA map\n");
goto fail;
}
data->txwi = (struct rt2860_txwi *)vaddr;
data->paddr = paddr;
vaddr += RT2860_TXWI_DMASZ;
paddr += RT2860_TXWI_DMASZ;
SLIST_INSERT_HEAD(&sc->data_pool, data, next);
}
return 0;
fail: rt2860_free_tx_pool(sc);
return error;
}
void
rt2860_free_tx_pool(struct rt2860_softc *sc)
{
if (sc->txwi_vaddr != NULL) {
bus_dmamap_sync(sc->txwi_dmat, sc->txwi_map,
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(sc->txwi_dmat, sc->txwi_map);
bus_dmamem_free(sc->txwi_dmat, sc->txwi_vaddr, sc->txwi_map);
}
if (sc->txwi_dmat != NULL)
bus_dma_tag_destroy(sc->txwi_dmat);
while (!SLIST_EMPTY(&sc->data_pool)) {
struct rt2860_tx_data *data;
data = SLIST_FIRST(&sc->data_pool);
bus_dmamap_destroy(sc->txwi_dmat, data->map);
SLIST_REMOVE_HEAD(&sc->data_pool, next);
}
}
int
rt2860_alloc_rx_ring(struct rt2860_softc *sc, struct rt2860_rx_ring *ring)
{
bus_addr_t physaddr;
int i, size, error;
size = RT2860_RX_RING_COUNT * sizeof (struct rt2860_rxd);
error = bus_dma_tag_create(bus_get_dma_tag(sc->sc_dev), 16, 0,
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL,
size, 1, size, 0, NULL, NULL, &ring->desc_dmat);
if (error != 0) {
device_printf(sc->sc_dev, "could not create desc DMA tag\n");
goto fail;
}
error = bus_dmamem_alloc(ring->desc_dmat, (void **)&ring->rxd,
BUS_DMA_NOWAIT | BUS_DMA_ZERO, &ring->desc_map);
if (error != 0) {
device_printf(sc->sc_dev, "could not allocate DMA memory\n");
goto fail;
}
error = bus_dmamap_load(ring->desc_dmat, ring->desc_map, ring->rxd,
size, rt2860_dma_map_addr, &ring->paddr, 0);
if (error != 0) {
device_printf(sc->sc_dev, "could not load desc DMA map\n");
goto fail;
}
error = bus_dma_tag_create(bus_get_dma_tag(sc->sc_dev), 1, 0,
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL, MCLBYTES,
1, MCLBYTES, 0, NULL, NULL, &ring->data_dmat);
if (error != 0) {
device_printf(sc->sc_dev, "could not create data DMA tag\n");
goto fail;
}
for (i = 0; i < RT2860_RX_RING_COUNT; i++) {
struct rt2860_rx_data *data = &ring->data[i];
struct rt2860_rxd *rxd = &ring->rxd[i];
error = bus_dmamap_create(ring->data_dmat, 0, &data->map);
if (error != 0) {
device_printf(sc->sc_dev, "could not create DMA map\n");
goto fail;
}
data->m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
if (data->m == NULL) {
device_printf(sc->sc_dev,
"could not allocate rx mbuf\n");
error = ENOMEM;
goto fail;
}
error = bus_dmamap_load(ring->data_dmat, data->map,
mtod(data->m, void *), MCLBYTES, rt2860_dma_map_addr,
&physaddr, 0);
if (error != 0) {
device_printf(sc->sc_dev,
"could not load rx buf DMA map");
goto fail;
}
rxd->sdp0 = htole32(physaddr);
rxd->sdl0 = htole16(MCLBYTES);
}
bus_dmamap_sync(ring->desc_dmat, ring->desc_map, BUS_DMASYNC_PREWRITE);
return 0;
fail: rt2860_free_rx_ring(sc, ring);
return error;
}
void
rt2860_reset_rx_ring(struct rt2860_softc *sc, struct rt2860_rx_ring *ring)
{
int i;
for (i = 0; i < RT2860_RX_RING_COUNT; i++)
ring->rxd[i].sdl0 &= ~htole16(RT2860_RX_DDONE);
bus_dmamap_sync(ring->desc_dmat, ring->desc_map, BUS_DMASYNC_PREWRITE);
ring->cur = 0;
}
void
rt2860_free_rx_ring(struct rt2860_softc *sc, struct rt2860_rx_ring *ring)
{
int i;
if (ring->rxd != NULL) {
bus_dmamap_sync(ring->desc_dmat, ring->desc_map,
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(ring->desc_dmat, ring->desc_map);
bus_dmamem_free(ring->desc_dmat, ring->rxd, ring->desc_map);
}
if (ring->desc_dmat != NULL)
bus_dma_tag_destroy(ring->desc_dmat);
for (i = 0; i < RT2860_RX_RING_COUNT; i++) {
struct rt2860_rx_data *data = &ring->data[i];
if (data->m != NULL) {
bus_dmamap_sync(ring->data_dmat, data->map,
BUS_DMASYNC_POSTREAD);
bus_dmamap_unload(ring->data_dmat, data->map);
m_freem(data->m);
}
if (data->map != NULL)
bus_dmamap_destroy(ring->data_dmat, data->map);
}
if (ring->data_dmat != NULL)
bus_dma_tag_destroy(ring->data_dmat);
}
static void
rt2860_updatestats(struct rt2860_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;
/*
* In IBSS or HostAP modes (when the hardware sends beacons), the
* MAC can run into a livelock and start sending CTS-to-self frames
* like crazy if protection is enabled. Fortunately, we can detect
* when such a situation occurs and reset the MAC.
*/
if (ic->ic_curmode != IEEE80211_M_STA) {
/* check if we're in a livelock situation.. */
uint32_t tmp = RAL_READ(sc, RT2860_DEBUG);
if ((tmp & (1 << 29)) && (tmp & (1 << 7 | 1 << 5))) {
/* ..and reset MAC/BBP for a while.. */
DPRINTF(("CTS-to-self livelock detected\n"));
RAL_WRITE(sc, RT2860_MAC_SYS_CTRL, RT2860_MAC_SRST);
RAL_BARRIER_WRITE(sc);
DELAY(1);
RAL_WRITE(sc, RT2860_MAC_SYS_CTRL,
RT2860_MAC_RX_EN | RT2860_MAC_TX_EN);
}
}
}
static void
rt2860_newassoc(struct ieee80211_node *ni, int isnew)
{
struct ieee80211com *ic = ni->ni_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
struct rt2860_softc *sc = ic->ic_softc;
uint8_t wcid;
wcid = IEEE80211_AID(ni->ni_associd);
if (isnew && ni->ni_associd != 0) {
sc->wcid2ni[wcid] = ni;
/* init WCID table entry */
RAL_WRITE_REGION_1(sc, RT2860_WCID_ENTRY(wcid),
ni->ni_macaddr, IEEE80211_ADDR_LEN);
}
DPRINTF(("new assoc isnew=%d addr=%s WCID=%d\n",
isnew, ether_sprintf(ni->ni_macaddr), wcid));
}
static void
rt2860_node_free(struct ieee80211_node *ni)
{
struct ieee80211com *ic = ni->ni_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
struct rt2860_softc *sc = ic->ic_softc;
uint8_t wcid;
if (ni->ni_associd != 0) {
wcid = IEEE80211_AID(ni->ni_associd);
/* clear Rx WCID search table entry */
RAL_SET_REGION_4(sc, RT2860_WCID_ENTRY(wcid), 0, 2);
}
sc->sc_node_free(ni);
}
#ifdef IEEE80211_HT
static int
rt2860_ampdu_rx_start(struct ieee80211com *ic, struct ieee80211_node *ni,
uint8_t tid)
{
struct rt2860_softc *sc = ic->ic_softc;
uint8_t wcid = ((struct rt2860_node *)ni)->wcid;
uint32_t tmp;
/* update BA session mask */
tmp = RAL_READ(sc, RT2860_WCID_ENTRY(wcid) + 4);
tmp |= (1 << tid) << 16;
RAL_WRITE(sc, RT2860_WCID_ENTRY(wcid) + 4, tmp);
return 0;
}
static void
rt2860_ampdu_rx_stop(struct ieee80211com *ic, struct ieee80211_node *ni,
uint8_t tid)
{
struct rt2860_softc *sc = ic->ic_softc;
uint8_t wcid = ((struct rt2860_node *)ni)->wcid;
uint32_t tmp;
/* update BA session mask */
tmp = RAL_READ(sc, RT2860_WCID_ENTRY(wcid) + 4);
tmp &= ~((1 << tid) << 16);
RAL_WRITE(sc, RT2860_WCID_ENTRY(wcid) + 4, tmp);
}
#endif
int
rt2860_newstate(struct ieee80211vap *vap, enum ieee80211_state nstate, int arg)
{
struct rt2860_vap *rvp = RT2860_VAP(vap);
struct ieee80211com *ic = vap->iv_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
struct rt2860_softc *sc = ic->ic_softc;
uint32_t tmp;
int error;
if (vap->iv_state == IEEE80211_S_RUN) {
/* turn link LED off */
rt2860_set_leds(sc, RT2860_LED_RADIO);
}
if (nstate == IEEE80211_S_INIT && vap->iv_state == IEEE80211_S_RUN) {
/* abort TSF synchronization */
tmp = RAL_READ(sc, RT2860_BCN_TIME_CFG);
RAL_WRITE(sc, RT2860_BCN_TIME_CFG,
tmp & ~(RT2860_BCN_TX_EN | RT2860_TSF_TIMER_EN |
RT2860_TBTT_TIMER_EN));
}
rt2860_set_gp_timer(sc, 0);
error = rvp->ral_newstate(vap, nstate, arg);
if (error != 0)
return (error);
if (nstate == IEEE80211_S_RUN) {
struct ieee80211_node *ni = vap->iv_bss;
if (ic->ic_opmode != IEEE80211_M_MONITOR) {
rt2860_enable_mrr(sc);
rt2860_set_txpreamble(sc);
rt2860_set_basicrates(sc, &ni->ni_rates);
rt2860_set_bssid(sc, ni->ni_bssid);
}
if (vap->iv_opmode == IEEE80211_M_HOSTAP ||
vap->iv_opmode == IEEE80211_M_IBSS ||
vap->iv_opmode == IEEE80211_M_MBSS) {
error = rt2860_setup_beacon(sc, vap);
if (error != 0)
return error;
}
if (ic->ic_opmode != IEEE80211_M_MONITOR) {
rt2860_enable_tsf_sync(sc);
rt2860_set_gp_timer(sc, 500);
}
/* turn link LED on */
rt2860_set_leds(sc, RT2860_LED_RADIO |
(IEEE80211_IS_CHAN_2GHZ(ni->ni_chan) ?
RT2860_LED_LINK_2GHZ : RT2860_LED_LINK_5GHZ));
}
return error;
}
/* Read 16-bit from eFUSE ROM (>=RT3071 only.) */
static uint16_t
rt3090_efuse_read_2(struct rt2860_softc *sc, uint16_t addr)
{
uint32_t tmp;
uint16_t reg;
int ntries;
addr *= 2;
/*-
* Read one 16-byte block into registers EFUSE_DATA[0-3]:
* DATA0: F E D C
* DATA1: B A 9 8
* DATA2: 7 6 5 4
* DATA3: 3 2 1 0
*/
tmp = RAL_READ(sc, RT3070_EFUSE_CTRL);
tmp &= ~(RT3070_EFSROM_MODE_MASK | RT3070_EFSROM_AIN_MASK);
tmp |= (addr & ~0xf) << RT3070_EFSROM_AIN_SHIFT | RT3070_EFSROM_KICK;
RAL_WRITE(sc, RT3070_EFUSE_CTRL, tmp);
for (ntries = 0; ntries < 500; ntries++) {
tmp = RAL_READ(sc, RT3070_EFUSE_CTRL);
if (!(tmp & RT3070_EFSROM_KICK))
break;
DELAY(2);
}
if (ntries == 500)
return 0xffff;
if ((tmp & RT3070_EFUSE_AOUT_MASK) == RT3070_EFUSE_AOUT_MASK)
return 0xffff; /* address not found */
/* determine to which 32-bit register our 16-bit word belongs */
reg = RT3070_EFUSE_DATA3 - (addr & 0xc);
tmp = RAL_READ(sc, reg);
return (addr & 2) ? tmp >> 16 : tmp & 0xffff;
}
/*
* Read 16 bits at address 'addr' from the serial EEPROM (either 93C46,
* 93C66 or 93C86).
*/
static uint16_t
rt2860_eeprom_read_2(struct rt2860_softc *sc, uint16_t addr)
{
uint32_t tmp;
uint16_t val;
int n;
/* clock C once before the first command */
RT2860_EEPROM_CTL(sc, 0);
RT2860_EEPROM_CTL(sc, RT2860_S);
RT2860_EEPROM_CTL(sc, RT2860_S | RT2860_C);
RT2860_EEPROM_CTL(sc, RT2860_S);
/* write start bit (1) */
RT2860_EEPROM_CTL(sc, RT2860_S | RT2860_D);
RT2860_EEPROM_CTL(sc, RT2860_S | RT2860_D | RT2860_C);
/* write READ opcode (10) */
RT2860_EEPROM_CTL(sc, RT2860_S | RT2860_D);
RT2860_EEPROM_CTL(sc, RT2860_S | RT2860_D | RT2860_C);
RT2860_EEPROM_CTL(sc, RT2860_S);
RT2860_EEPROM_CTL(sc, RT2860_S | RT2860_C);
/* write address (A5-A0 or A7-A0) */
n = ((RAL_READ(sc, RT2860_PCI_EECTRL) & 0x30) == 0) ? 5 : 7;
for (; n >= 0; n--) {
RT2860_EEPROM_CTL(sc, RT2860_S |
(((addr >> n) & 1) << RT2860_SHIFT_D));
RT2860_EEPROM_CTL(sc, RT2860_S |
(((addr >> n) & 1) << RT2860_SHIFT_D) | RT2860_C);
}
RT2860_EEPROM_CTL(sc, RT2860_S);
/* read data Q15-Q0 */
val = 0;
for (n = 15; n >= 0; n--) {
RT2860_EEPROM_CTL(sc, RT2860_S | RT2860_C);
tmp = RAL_READ(sc, RT2860_PCI_EECTRL);
val |= ((tmp & RT2860_Q) >> RT2860_SHIFT_Q) << n;
RT2860_EEPROM_CTL(sc, RT2860_S);
}
RT2860_EEPROM_CTL(sc, 0);
/* clear Chip Select and clock C */
RT2860_EEPROM_CTL(sc, RT2860_S);
RT2860_EEPROM_CTL(sc, 0);
RT2860_EEPROM_CTL(sc, RT2860_C);
return val;
}
static __inline uint16_t
rt2860_srom_read(struct rt2860_softc *sc, uint8_t addr)
{
/* either eFUSE ROM or EEPROM */
return sc->sc_srom_read(sc, addr);
}
static void
rt2860_intr_coherent(struct rt2860_softc *sc)
{
uint32_t tmp;
/* DMA finds data coherent event when checking the DDONE bit */
DPRINTF(("Tx/Rx Coherent interrupt\n"));
/* restart DMA engine */
tmp = RAL_READ(sc, RT2860_WPDMA_GLO_CFG);
tmp &= ~(RT2860_TX_WB_DDONE | RT2860_RX_DMA_EN | RT2860_TX_DMA_EN);
RAL_WRITE(sc, RT2860_WPDMA_GLO_CFG, tmp);
(void)rt2860_txrx_enable(sc);
}
static void
rt2860_drain_stats_fifo(struct rt2860_softc *sc)
{
struct ieee80211_node *ni;
uint32_t stat;
int retrycnt;
uint8_t wcid, mcs, pid;
/* drain Tx status FIFO (maxsize = 16) */
while ((stat = RAL_READ(sc, RT2860_TX_STAT_FIFO)) & RT2860_TXQ_VLD) {
DPRINTFN(4, ("tx stat 0x%08x\n", stat));
wcid = (stat >> RT2860_TXQ_WCID_SHIFT) & 0xff;
ni = sc->wcid2ni[wcid];
/* if no ACK was requested, no feedback is available */
if (!(stat & RT2860_TXQ_ACKREQ) || wcid == 0xff || ni == NULL)
continue;
/* update per-STA AMRR stats */
if (stat & RT2860_TXQ_OK) {
/*
* Check if there were retries, ie if the Tx success
* rate is different from the requested rate. Note
* that it works only because we do not allow rate
* fallback from OFDM to CCK.
*/
mcs = (stat >> RT2860_TXQ_MCS_SHIFT) & 0x7f;
pid = (stat >> RT2860_TXQ_PID_SHIFT) & 0xf;
if (mcs + 1 != pid)
retrycnt = 1;
else
retrycnt = 0;
ieee80211_ratectl_tx_complete(ni->ni_vap, ni,
IEEE80211_RATECTL_TX_SUCCESS, &retrycnt, NULL);
} else {
ieee80211_ratectl_tx_complete(ni->ni_vap, ni,
IEEE80211_RATECTL_TX_FAILURE, &retrycnt, 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(ni->ni_vap->iv_ifp,
IFCOUNTER_OERRORS, 1);
}
}
}
static void
rt2860_tx_intr(struct rt2860_softc *sc, int qid)
{
struct rt2860_tx_ring *ring = &sc->txq[qid];
uint32_t hw;
rt2860_drain_stats_fifo(sc);
hw = RAL_READ(sc, RT2860_TX_DTX_IDX(qid));
while (ring->next != hw) {
struct rt2860_tx_data *data = ring->data[ring->next];
if (data != NULL) {
bus_dmamap_sync(sc->txwi_dmat, data->map,
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(sc->txwi_dmat, data->map);
if (data->m->m_flags & M_TXCB) {
ieee80211_process_callback(data->ni, data->m,
0);
}
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
ieee80211_tx_complete(data->ni, data->m, 0);
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
data->m = NULL;
SLIST_INSERT_HEAD(&sc->data_pool, data, next);
ring->data[ring->next] = NULL;
}
ring->queued--;
ring->next = (ring->next + 1) % RT2860_TX_RING_COUNT;
}
sc->sc_tx_timer = 0;
if (ring->queued < RT2860_TX_RING_COUNT)
sc->qfullmsk &= ~(1 << qid);
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
rt2860_start(sc);
}
/*
* Return the Rx chain with the highest RSSI for a given frame.
*/
static __inline uint8_t
rt2860_maxrssi_chain(struct rt2860_softc *sc, const struct rt2860_rxwi *rxwi)
{
uint8_t rxchain = 0;
if (sc->nrxchains > 1) {
if (rxwi->rssi[1] > rxwi->rssi[rxchain])
rxchain = 1;
if (sc->nrxchains > 2)
if (rxwi->rssi[2] > rxwi->rssi[rxchain])
rxchain = 2;
}
return rxchain;
}
static void
rt2860_rx_intr(struct rt2860_softc *sc)
{
struct rt2860_rx_radiotap_header *tap;
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 mbuf *m, *m1;
bus_addr_t physaddr;
uint32_t hw;
uint16_t phy;
uint8_t ant;
int8_t rssi, nf;
int error;
hw = RAL_READ(sc, RT2860_FS_DRX_IDX) & 0xfff;
while (sc->rxq.cur != hw) {
struct rt2860_rx_data *data = &sc->rxq.data[sc->rxq.cur];
struct rt2860_rxd *rxd = &sc->rxq.rxd[sc->rxq.cur];
struct rt2860_rxwi *rxwi;
bus_dmamap_sync(sc->rxq.desc_dmat, sc->rxq.desc_map,
BUS_DMASYNC_POSTREAD);
if (__predict_false(!(rxd->sdl0 & htole16(RT2860_RX_DDONE)))) {
DPRINTF(("RXD DDONE bit not set!\n"));
break; /* should not happen */
}
if (__predict_false(rxd->flags &
htole32(RT2860_RX_CRCERR | RT2860_RX_ICVERR))) {
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 skip;
}
#ifdef HW_CRYPTO
if (__predict_false(rxd->flags & htole32(RT2860_RX_MICERR))) {
/* report MIC failures to net80211 for TKIP */
ic->ic_stats.is_rx_locmicfail++;
ieee80211_michael_mic_failure(ic, 0/* XXX */);
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 skip;
}
#endif
m1 = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
if (__predict_false(m1 == 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
counter_u64_add(ic->ic_ierrors, 1);
goto skip;
}
bus_dmamap_sync(sc->rxq.data_dmat, data->map,
BUS_DMASYNC_POSTREAD);
bus_dmamap_unload(sc->rxq.data_dmat, data->map);
error = bus_dmamap_load(sc->rxq.data_dmat, data->map,
mtod(m1, void *), MCLBYTES, rt2860_dma_map_addr,
&physaddr, 0);
if (__predict_false(error != 0)) {
m_freem(m1);
/* try to reload the old mbuf */
error = bus_dmamap_load(sc->rxq.data_dmat, data->map,
mtod(data->m, void *), MCLBYTES,
rt2860_dma_map_addr, &physaddr, 0);
if (__predict_false(error != 0)) {
panic("%s: could not load old rx mbuf",
device_get_name(sc->sc_dev));
}
/* physical address may have changed */
rxd->sdp0 = htole32(physaddr);
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 skip;
}
/*
* New mbuf successfully loaded, update Rx ring and continue
* processing.
*/
m = data->m;
data->m = m1;
rxd->sdp0 = htole32(physaddr);
rxwi = mtod(m, struct rt2860_rxwi *);
/* finalize mbuf */
m->m_data = (caddr_t)(rxwi + 1);
m->m_pkthdr.len = m->m_len = le16toh(rxwi->len) & 0xfff;
wh = mtod(m, struct ieee80211_frame *);
#ifdef HW_CRYPTO
if (wh->i_fc[1] & IEEE80211_FC1_PROTECTED) {
/* frame is decrypted by hardware */
wh->i_fc[1] &= ~IEEE80211_FC1_PROTECTED;
}
#endif
/* HW may insert 2 padding bytes after 802.11 header */
if (rxd->flags & htole32(RT2860_RX_L2PAD)) {
u_int hdrlen = ieee80211_hdrsize(wh);
ovbcopy(wh, (caddr_t)wh + 2, hdrlen);
m->m_data += 2;
wh = mtod(m, struct ieee80211_frame *);
}
ant = rt2860_maxrssi_chain(sc, rxwi);
rssi = rt2860_rssi2dbm(sc, rxwi->rssi[ant], ant);
nf = RT2860_NOISE_FLOOR;
if (ieee80211_radiotap_active(ic)) {
tap = &sc->sc_rxtap;
tap->wr_flags = 0;
tap->wr_antenna = ant;
tap->wr_antsignal = nf + rssi;
tap->wr_antnoise = nf;
/* in case it can't be found below */
tap->wr_rate = 2;
phy = le16toh(rxwi->phy);
switch (phy & RT2860_PHY_MODE) {
case RT2860_PHY_CCK:
switch ((phy & RT2860_PHY_MCS) & ~RT2860_PHY_SHPRE) {
case 0: tap->wr_rate = 2; break;
case 1: tap->wr_rate = 4; break;
case 2: tap->wr_rate = 11; break;
case 3: tap->wr_rate = 22; break;
}
if (phy & RT2860_PHY_SHPRE)
tap->wr_flags |= IEEE80211_RADIOTAP_F_SHORTPRE;
break;
case RT2860_PHY_OFDM:
switch (phy & RT2860_PHY_MCS) {
case 0: tap->wr_rate = 12; break;
case 1: tap->wr_rate = 18; break;
case 2: tap->wr_rate = 24; break;
case 3: tap->wr_rate = 36; break;
case 4: tap->wr_rate = 48; break;
case 5: tap->wr_rate = 72; break;
case 6: tap->wr_rate = 96; break;
case 7: tap->wr_rate = 108; break;
}
break;
}
}
RAL_UNLOCK(sc);
wh = mtod(m, struct ieee80211_frame *);
/* send the frame to the 802.11 layer */
ni = ieee80211_find_rxnode(ic,
(struct ieee80211_frame_min *)wh);
if (ni != NULL) {
(void)ieee80211_input(ni, m, rssi - nf, nf);
ieee80211_free_node(ni);
} else
(void)ieee80211_input_all(ic, m, rssi - nf, nf);
RAL_LOCK(sc);
skip: rxd->sdl0 &= ~htole16(RT2860_RX_DDONE);
bus_dmamap_sync(sc->rxq.desc_dmat, sc->rxq.desc_map,
BUS_DMASYNC_PREWRITE);
sc->rxq.cur = (sc->rxq.cur + 1) % RT2860_RX_RING_COUNT;
}
/* tell HW what we have processed */
RAL_WRITE(sc, RT2860_RX_CALC_IDX,
(sc->rxq.cur - 1) % RT2860_RX_RING_COUNT);
}
static void
rt2860_tbtt_intr(struct rt2860_softc *sc)
{
#if 0
struct ieee80211com *ic = &sc->sc_ic;
#ifndef IEEE80211_STA_ONLY
if (ic->ic_opmode == IEEE80211_M_HOSTAP) {
/* one less beacon until next DTIM */
if (ic->ic_dtim_count == 0)
ic->ic_dtim_count = ic->ic_dtim_period - 1;
else
ic->ic_dtim_count--;
/* update dynamic parts of beacon */
rt2860_setup_beacon(sc);
/* flush buffered multicast frames */
if (ic->ic_dtim_count == 0)
ieee80211_notify_dtim(ic);
}
#endif
/* check if protection mode has changed */
if ((sc->sc_ic_flags ^ ic->ic_flags) & IEEE80211_F_USEPROT) {
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
rt2860_updateprot(sc);
sc->sc_ic_flags = ic->ic_flags;
}
#endif
}
static void
rt2860_gp_intr(struct rt2860_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);
DPRINTFN(2, ("GP timeout state=%d\n", vap->iv_state));
if (vap->iv_state == IEEE80211_S_RUN)
rt2860_updatestats(sc);
}
void
rt2860_intr(void *arg)
{
struct rt2860_softc *sc = arg;
uint32_t r;
RAL_LOCK(sc);
r = RAL_READ(sc, RT2860_INT_STATUS);
if (__predict_false(r == 0xffffffff)) {
RAL_UNLOCK(sc);
return; /* device likely went away */
}
if (r == 0) {
RAL_UNLOCK(sc);
return; /* not for us */
}
/* acknowledge interrupts */
RAL_WRITE(sc, RT2860_INT_STATUS, r);
if (r & RT2860_TX_RX_COHERENT)
rt2860_intr_coherent(sc);
if (r & RT2860_MAC_INT_2) /* TX status */
rt2860_drain_stats_fifo(sc);
if (r & RT2860_TX_DONE_INT5)
rt2860_tx_intr(sc, 5);
if (r & RT2860_RX_DONE_INT)
rt2860_rx_intr(sc);
if (r & RT2860_TX_DONE_INT4)
rt2860_tx_intr(sc, 4);
if (r & RT2860_TX_DONE_INT3)
rt2860_tx_intr(sc, 3);
if (r & RT2860_TX_DONE_INT2)
rt2860_tx_intr(sc, 2);
if (r & RT2860_TX_DONE_INT1)
rt2860_tx_intr(sc, 1);
if (r & RT2860_TX_DONE_INT0)
rt2860_tx_intr(sc, 0);
if (r & RT2860_MAC_INT_0) /* TBTT */
rt2860_tbtt_intr(sc);
if (r & RT2860_MAC_INT_3) /* Auto wakeup */
/* TBD wakeup */;
if (r & RT2860_MAC_INT_4) /* GP timer */
rt2860_gp_intr(sc);
RAL_UNLOCK(sc);
}
static int
rt2860_tx(struct rt2860_softc *sc, struct mbuf *m, struct ieee80211_node *ni)
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
struct ieee80211vap *vap = ni->ni_vap;
struct rt2860_tx_ring *ring;
struct rt2860_tx_data *data;
struct rt2860_txd *txd;
struct rt2860_txwi *txwi;
struct ieee80211_frame *wh;
const struct ieee80211_txparam *tp;
struct ieee80211_key *k;
struct mbuf *m1;
bus_dma_segment_t segs[RT2860_MAX_SCATTER];
bus_dma_segment_t *seg;
u_int hdrlen;
uint16_t qos, dur;
uint8_t type, qsel, mcs, pid, tid, qid;
int i, nsegs, ntxds, pad, rate, ridx, error;
/* the data pool contains at least one element, pick the first */
data = SLIST_FIRST(&sc->data_pool);
wh = mtod(m, struct ieee80211_frame *);
if (wh->i_fc[1] & IEEE80211_FC1_PROTECTED) {
k = ieee80211_crypto_encap(ni, m);
if (k == NULL) {
m_freem(m);
return ENOBUFS;
}
/* packet header may have moved, reset our local pointer */
wh = mtod(m, struct ieee80211_frame *);
}
hdrlen = ieee80211_anyhdrsize(wh);
type = wh->i_fc[0] & IEEE80211_FC0_TYPE_MASK;
tp = &vap->iv_txparms[ieee80211_chan2mode(ni->ni_chan)];
if (IEEE80211_IS_MULTICAST(wh->i_addr1)) {
rate = tp->mcastrate;
} else if (m->m_flags & M_EAPOL) {
rate = tp->mgmtrate;
} else if (tp->ucastrate != IEEE80211_FIXED_RATE_NONE) {
rate = tp->ucastrate;
} else {
(void) ieee80211_ratectl_rate(ni, NULL, 0);
rate = ni->ni_txrate;
}
rate &= IEEE80211_RATE_VAL;
qid = M_WME_GETAC(m);
if (IEEE80211_QOS_HAS_SEQ(wh)) {
qos = ((const struct ieee80211_qosframe *)wh)->i_qos[0];
tid = qos & IEEE80211_QOS_TID;
} else {
qos = 0;
tid = 0;
}
ring = &sc->txq[qid];
ridx = ieee80211_legacy_rate_lookup(ic->ic_rt, rate);
/* get MCS code from rate index */
mcs = rt2860_rates[ridx].mcs;
/* setup TX Wireless Information */
txwi = data->txwi;
txwi->flags = 0;
/* let HW generate seq numbers for non-QoS frames */
txwi->xflags = qos ? 0 : RT2860_TX_NSEQ;
if (type == IEEE80211_FC0_TYPE_DATA)
txwi->wcid = IEEE80211_AID(ni->ni_associd);
else
txwi->wcid = 0xff;
txwi->len = htole16(m->m_pkthdr.len);
if (rt2860_rates[ridx].phy == IEEE80211_T_DS) {
txwi->phy = htole16(RT2860_PHY_CCK);
if (ridx != RT2860_RIDX_CCK1 &&
(ic->ic_flags & IEEE80211_F_SHPREAMBLE))
mcs |= RT2860_PHY_SHPRE;
} else
txwi->phy = htole16(RT2860_PHY_OFDM);
txwi->phy |= htole16(mcs);
/*
* We store the MCS code into the driver-private PacketID field.
* The PacketID is latched into TX_STAT_FIFO when Tx completes so
* that we know at which initial rate the frame was transmitted.
* We add 1 to the MCS code because setting the PacketID field to
* 0 means that we don't want feedback in TX_STAT_FIFO.
*/
pid = (mcs + 1) & 0xf;
txwi->len |= htole16(pid << RT2860_TX_PID_SHIFT);
/* check if RTS/CTS or CTS-to-self protection is required */
if (!IEEE80211_IS_MULTICAST(wh->i_addr1) &&
(m->m_pkthdr.len + IEEE80211_CRC_LEN > vap->iv_rtsthreshold ||
((ic->ic_flags & IEEE80211_F_USEPROT) &&
rt2860_rates[ridx].phy == IEEE80211_T_OFDM)))
txwi->txop = RT2860_TX_TXOP_HT;
else
txwi->txop = RT2860_TX_TXOP_BACKOFF;
if (!IEEE80211_IS_MULTICAST(wh->i_addr1) &&
(!qos || (qos & IEEE80211_QOS_ACKPOLICY) !=
IEEE80211_QOS_ACKPOLICY_NOACK)) {
txwi->xflags |= RT2860_TX_ACK;
if (ic->ic_flags & IEEE80211_F_SHPREAMBLE)
dur = rt2860_rates[ridx].sp_ack_dur;
else
dur = rt2860_rates[ridx].lp_ack_dur;
*(uint16_t *)wh->i_dur = htole16(dur);
}
/* ask MAC to insert timestamp into probe responses */
if ((wh->i_fc[0] &
(IEEE80211_FC0_TYPE_MASK | IEEE80211_FC0_SUBTYPE_MASK)) ==
(IEEE80211_FC0_TYPE_MGT | IEEE80211_FC0_SUBTYPE_PROBE_RESP))
/* NOTE: beacons do not pass through tx_data() */
txwi->flags |= RT2860_TX_TS;
if (ieee80211_radiotap_active_vap(vap)) {
struct rt2860_tx_radiotap_header *tap = &sc->sc_txtap;
tap->wt_flags = 0;
tap->wt_rate = rate;
if (mcs & RT2860_PHY_SHPRE)
tap->wt_flags |= IEEE80211_RADIOTAP_F_SHORTPRE;
ieee80211_radiotap_tx(vap, m);
}
pad = (hdrlen + 3) & ~3;
/* copy and trim 802.11 header */
memcpy(txwi + 1, wh, hdrlen);
m_adj(m, hdrlen);
error = bus_dmamap_load_mbuf_sg(sc->txwi_dmat, data->map, m, segs,
&nsegs, 0);
if (__predict_false(error != 0 && error != EFBIG)) {
device_printf(sc->sc_dev, "can't map mbuf (error %d)\n",
error);
m_freem(m);
return error;
}
if (__predict_true(error == 0)) {
/* determine how many TXDs are required */
ntxds = 1 + (nsegs / 2);
if (ring->queued + ntxds >= RT2860_TX_RING_COUNT) {
/* not enough free TXDs, force mbuf defrag */
bus_dmamap_unload(sc->txwi_dmat, data->map);
error = EFBIG;
}
}
if (__predict_false(error != 0)) {
m1 = m_defrag(m, M_NOWAIT);
if (m1 == NULL) {
device_printf(sc->sc_dev,
"could not defragment mbuf\n");
m_freem(m);
return ENOBUFS;
}
m = m1;
error = bus_dmamap_load_mbuf_sg(sc->txwi_dmat, data->map, m,
segs, &nsegs, 0);
if (__predict_false(error != 0)) {
device_printf(sc->sc_dev, "can't map mbuf (error %d)\n",
error);
m_freem(m);
return error;
}
/* determine how many TXDs are now required */
ntxds = 1 + (nsegs / 2);
if (ring->queued + ntxds >= RT2860_TX_RING_COUNT) {
/* this is a hopeless case, drop the mbuf! */
bus_dmamap_unload(sc->txwi_dmat, data->map);
m_freem(m);
return ENOBUFS;
}
}
qsel = (qid < WME_NUM_AC) ? RT2860_TX_QSEL_EDCA : RT2860_TX_QSEL_MGMT;
/* first segment is TXWI + 802.11 header */
txd = &ring->txd[ring->cur];
txd->sdp0 = htole32(data->paddr);
txd->sdl0 = htole16(sizeof (struct rt2860_txwi) + pad);
txd->flags = qsel;
/* setup payload segments */
seg = &segs[0];
for (i = nsegs; i >= 2; i -= 2) {
txd->sdp1 = htole32(seg->ds_addr);
txd->sdl1 = htole16(seg->ds_len);
seg++;
ring->cur = (ring->cur + 1) % RT2860_TX_RING_COUNT;
/* grab a new Tx descriptor */
txd = &ring->txd[ring->cur];
txd->sdp0 = htole32(seg->ds_addr);
txd->sdl0 = htole16(seg->ds_len);
txd->flags = qsel;
seg++;
}
/* finalize last segment */
if (i > 0) {
txd->sdp1 = htole32(seg->ds_addr);
txd->sdl1 = htole16(seg->ds_len | RT2860_TX_LS1);
} else {
txd->sdl0 |= htole16(RT2860_TX_LS0);
txd->sdl1 = 0;
}
/* remove from the free pool and link it into the SW Tx slot */
SLIST_REMOVE_HEAD(&sc->data_pool, next);
data->m = m;
data->ni = ni;
ring->data[ring->cur] = data;
bus_dmamap_sync(sc->txwi_dmat, sc->txwi_map, BUS_DMASYNC_PREWRITE);
bus_dmamap_sync(sc->txwi_dmat, data->map, BUS_DMASYNC_PREWRITE);
bus_dmamap_sync(ring->desc_dmat, ring->desc_map, BUS_DMASYNC_PREWRITE);
DPRINTFN(4, ("sending frame qid=%d wcid=%d nsegs=%d ridx=%d\n",
qid, txwi->wcid, nsegs, ridx));
ring->cur = (ring->cur + 1) % RT2860_TX_RING_COUNT;
ring->queued += ntxds;
if (ring->queued >= RT2860_TX_RING_COUNT)
sc->qfullmsk |= 1 << qid;
/* kick Tx */
RAL_WRITE(sc, RT2860_TX_CTX_IDX(qid), ring->cur);
return 0;
}
static int
rt2860_raw_xmit(struct ieee80211_node *ni, struct mbuf *m,
const struct ieee80211_bpf_params *params)
{
struct ieee80211com *ic = ni->ni_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
struct rt2860_softc *sc = ic->ic_softc;
int error;
RAL_LOCK(sc);
/* prevent management frames from being sent if we're not ready */
2015-10-29 04:21:34 +00:00
if (!(sc->sc_flags & RT2860_RUNNING)) {
RAL_UNLOCK(sc);
m_freem(m);
return ENETDOWN;
}
if (params == NULL) {
/*
* Legacy path; interpret frame contents to decide
* precisely how to send the frame.
*/
error = rt2860_tx(sc, m, ni);
} else {
/*
* Caller supplied explicit parameters to use in
* sending the frame.
*/
error = rt2860_tx_raw(sc, m, ni, params);
}
sc->sc_tx_timer = 5;
RAL_UNLOCK(sc);
return error;
}
static int
rt2860_tx_raw(struct rt2860_softc *sc, struct mbuf *m,
struct ieee80211_node *ni, const struct ieee80211_bpf_params *params)
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
struct ieee80211vap *vap = ni->ni_vap;
struct rt2860_tx_ring *ring;
struct rt2860_tx_data *data;
struct rt2860_txd *txd;
struct rt2860_txwi *txwi;
struct ieee80211_frame *wh;
struct mbuf *m1;
bus_dma_segment_t segs[RT2860_MAX_SCATTER];
bus_dma_segment_t *seg;
u_int hdrlen;
uint16_t dur;
uint8_t type, qsel, mcs, pid, tid, qid;
int i, nsegs, ntxds, pad, rate, ridx, error;
/* the data pool contains at least one element, pick the first */
data = SLIST_FIRST(&sc->data_pool);
wh = mtod(m, struct ieee80211_frame *);
hdrlen = ieee80211_hdrsize(wh);
type = wh->i_fc[0] & IEEE80211_FC0_TYPE_MASK;
/* Choose a TX rate index. */
rate = params->ibp_rate0;
ridx = ieee80211_legacy_rate_lookup(ic->ic_rt,
rate & IEEE80211_RATE_VAL);
if (ridx == (uint8_t)-1) {
/* XXX fall back to mcast/mgmt rate? */
m_freem(m);
return EINVAL;
}
qid = params->ibp_pri & 3;
tid = 0;
ring = &sc->txq[qid];
/* get MCS code from rate index */
mcs = rt2860_rates[ridx].mcs;
/* setup TX Wireless Information */
txwi = data->txwi;
txwi->flags = 0;
/* let HW generate seq numbers for non-QoS frames */
txwi->xflags = params->ibp_pri & 3 ? 0 : RT2860_TX_NSEQ;
txwi->wcid = 0xff;
txwi->len = htole16(m->m_pkthdr.len);
if (rt2860_rates[ridx].phy == IEEE80211_T_DS) {
txwi->phy = htole16(RT2860_PHY_CCK);
if (ridx != RT2860_RIDX_CCK1 &&
(ic->ic_flags & IEEE80211_F_SHPREAMBLE))
mcs |= RT2860_PHY_SHPRE;
} else
txwi->phy = htole16(RT2860_PHY_OFDM);
txwi->phy |= htole16(mcs);
/*
* We store the MCS code into the driver-private PacketID field.
* The PacketID is latched into TX_STAT_FIFO when Tx completes so
* that we know at which initial rate the frame was transmitted.
* We add 1 to the MCS code because setting the PacketID field to
* 0 means that we don't want feedback in TX_STAT_FIFO.
*/
pid = (mcs + 1) & 0xf;
txwi->len |= htole16(pid << RT2860_TX_PID_SHIFT);
/* check if RTS/CTS or CTS-to-self protection is required */
if (params->ibp_flags & IEEE80211_BPF_RTS ||
params->ibp_flags & IEEE80211_BPF_CTS)
txwi->txop = RT2860_TX_TXOP_HT;
else
txwi->txop = RT2860_TX_TXOP_BACKOFF;
if ((params->ibp_flags & IEEE80211_BPF_NOACK) == 0) {
txwi->xflags |= RT2860_TX_ACK;
if (ic->ic_flags & IEEE80211_F_SHPREAMBLE)
dur = rt2860_rates[ridx].sp_ack_dur;
else
dur = rt2860_rates[ridx].lp_ack_dur;
*(uint16_t *)wh->i_dur = htole16(dur);
}
/* ask MAC to insert timestamp into probe responses */
if ((wh->i_fc[0] &
(IEEE80211_FC0_TYPE_MASK | IEEE80211_FC0_SUBTYPE_MASK)) ==
(IEEE80211_FC0_TYPE_MGT | IEEE80211_FC0_SUBTYPE_PROBE_RESP))
/* NOTE: beacons do not pass through tx_data() */
txwi->flags |= RT2860_TX_TS;
if (ieee80211_radiotap_active_vap(vap)) {
struct rt2860_tx_radiotap_header *tap = &sc->sc_txtap;
tap->wt_flags = 0;
tap->wt_rate = rate;
if (mcs & RT2860_PHY_SHPRE)
tap->wt_flags |= IEEE80211_RADIOTAP_F_SHORTPRE;
ieee80211_radiotap_tx(vap, m);
}
pad = (hdrlen + 3) & ~3;
/* copy and trim 802.11 header */
memcpy(txwi + 1, wh, hdrlen);
m_adj(m, hdrlen);
error = bus_dmamap_load_mbuf_sg(sc->txwi_dmat, data->map, m, segs,
&nsegs, 0);
if (__predict_false(error != 0 && error != EFBIG)) {
device_printf(sc->sc_dev, "can't map mbuf (error %d)\n",
error);
m_freem(m);
return error;
}
if (__predict_true(error == 0)) {
/* determine how many TXDs are required */
ntxds = 1 + (nsegs / 2);
if (ring->queued + ntxds >= RT2860_TX_RING_COUNT) {
/* not enough free TXDs, force mbuf defrag */
bus_dmamap_unload(sc->txwi_dmat, data->map);
error = EFBIG;
}
}
if (__predict_false(error != 0)) {
m1 = m_defrag(m, M_NOWAIT);
if (m1 == NULL) {
device_printf(sc->sc_dev,
"could not defragment mbuf\n");
m_freem(m);
return ENOBUFS;
}
m = m1;
error = bus_dmamap_load_mbuf_sg(sc->txwi_dmat, data->map, m,
segs, &nsegs, 0);
if (__predict_false(error != 0)) {
device_printf(sc->sc_dev, "can't map mbuf (error %d)\n",
error);
m_freem(m);
return error;
}
/* determine how many TXDs are now required */
ntxds = 1 + (nsegs / 2);
if (ring->queued + ntxds >= RT2860_TX_RING_COUNT) {
/* this is a hopeless case, drop the mbuf! */
bus_dmamap_unload(sc->txwi_dmat, data->map);
m_freem(m);
return ENOBUFS;
}
}
qsel = (qid < WME_NUM_AC) ? RT2860_TX_QSEL_EDCA : RT2860_TX_QSEL_MGMT;
/* first segment is TXWI + 802.11 header */
txd = &ring->txd[ring->cur];
txd->sdp0 = htole32(data->paddr);
txd->sdl0 = htole16(sizeof (struct rt2860_txwi) + pad);
txd->flags = qsel;
/* setup payload segments */
seg = &segs[0];
for (i = nsegs; i >= 2; i -= 2) {
txd->sdp1 = htole32(seg->ds_addr);
txd->sdl1 = htole16(seg->ds_len);
seg++;
ring->cur = (ring->cur + 1) % RT2860_TX_RING_COUNT;
/* grab a new Tx descriptor */
txd = &ring->txd[ring->cur];
txd->sdp0 = htole32(seg->ds_addr);
txd->sdl0 = htole16(seg->ds_len);
txd->flags = qsel;
seg++;
}
/* finalize last segment */
if (i > 0) {
txd->sdp1 = htole32(seg->ds_addr);
txd->sdl1 = htole16(seg->ds_len | RT2860_TX_LS1);
} else {
txd->sdl0 |= htole16(RT2860_TX_LS0);
txd->sdl1 = 0;
}
/* remove from the free pool and link it into the SW Tx slot */
SLIST_REMOVE_HEAD(&sc->data_pool, next);
data->m = m;
data->ni = ni;
ring->data[ring->cur] = data;
bus_dmamap_sync(sc->txwi_dmat, sc->txwi_map, BUS_DMASYNC_PREWRITE);
bus_dmamap_sync(sc->txwi_dmat, data->map, BUS_DMASYNC_PREWRITE);
bus_dmamap_sync(ring->desc_dmat, ring->desc_map, BUS_DMASYNC_PREWRITE);
DPRINTFN(4, ("sending frame qid=%d wcid=%d nsegs=%d ridx=%d\n",
qid, txwi->wcid, nsegs, ridx));
ring->cur = (ring->cur + 1) % RT2860_TX_RING_COUNT;
ring->queued += ntxds;
if (ring->queued >= RT2860_TX_RING_COUNT)
sc->qfullmsk |= 1 << qid;
/* kick Tx */
RAL_WRITE(sc, RT2860_TX_CTX_IDX(qid), ring->cur);
return 0;
}
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
static int
rt2860_transmit(struct ieee80211com *ic, struct mbuf *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
struct rt2860_softc *sc = ic->ic_softc;
int error;
RAL_LOCK(sc);
2015-10-29 04:21:34 +00:00
if ((sc->sc_flags & RT2860_RUNNING) == 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
RAL_UNLOCK(sc);
return (ENXIO);
}
error = mbufq_enqueue(&sc->sc_snd, m);
if (error) {
RAL_UNLOCK(sc);
return (error);
}
rt2860_start(sc);
RAL_UNLOCK(sc);
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
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
rt2860_start(struct rt2860_softc *sc)
{
struct ieee80211_node *ni;
struct mbuf *m;
RAL_LOCK_ASSERT(sc);
2015-10-29 04:21:34 +00:00
if ((sc->sc_flags & RT2860_RUNNING) == 0)
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 (!SLIST_EMPTY(&sc->data_pool) && sc->qfullmsk == 0 &&
(m = mbufq_dequeue(&sc->sc_snd)) != NULL) {
ni = (struct ieee80211_node *)m->m_pkthdr.rcvif;
if (rt2860_tx(sc, m, ni) != 0) {
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
if_inc_counter(ni->ni_vap->iv_ifp,
IFCOUNTER_OERRORS, 1);
ieee80211_free_node(ni);
continue;
}
sc->sc_tx_timer = 5;
}
}
static void
rt2860_watchdog(void *arg)
{
struct rt2860_softc *sc = arg;
RAL_LOCK_ASSERT(sc);
2015-10-29 04:21:34 +00:00
KASSERT(sc->sc_flags & RT2860_RUNNING, ("not running"));
if (sc->sc_invalid) /* card ejected */
return;
if (sc->sc_tx_timer > 0 && --sc->sc_tx_timer == 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
device_printf(sc->sc_dev, "device timeout\n");
rt2860_stop_locked(sc);
rt2860_init_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
counter_u64_add(sc->sc_ic.ic_oerrors, 1);
return;
}
callout_reset(&sc->watchdog_ch, hz, rt2860_watchdog, 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
static void
rt2860_parent(struct ieee80211com *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
struct rt2860_softc *sc = ic->ic_softc;
int startall = 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
RAL_LOCK(sc);
if (ic->ic_nrunning> 0) {
2015-10-29 04:21:34 +00:00
if (!(sc->sc_flags & RT2860_RUNNING)) {
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
rt2860_init_locked(sc);
startall = 1;
} else
rt2860_update_promisc(ic);
2015-10-29 04:21:34 +00:00
} else if (sc->sc_flags & RT2860_RUNNING)
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
rt2860_stop_locked(sc);
RAL_UNLOCK(sc);
if (startall)
ieee80211_start_all(ic);
}
/*
* Reading and writing from/to the BBP is different from RT2560 and RT2661.
* We access the BBP through the 8051 microcontroller unit which means that
* the microcode must be loaded first.
*/
void
rt2860_mcu_bbp_write(struct rt2860_softc *sc, uint8_t reg, uint8_t val)
{
int ntries;
for (ntries = 0; ntries < 100; ntries++) {
if (!(RAL_READ(sc, RT2860_H2M_BBPAGENT) & RT2860_BBP_CSR_KICK))
break;
DELAY(1);
}
if (ntries == 100) {
device_printf(sc->sc_dev,
"could not write to BBP through MCU\n");
return;
}
RAL_WRITE(sc, RT2860_H2M_BBPAGENT, RT2860_BBP_RW_PARALLEL |
RT2860_BBP_CSR_KICK | reg << 8 | val);
RAL_BARRIER_WRITE(sc);
rt2860_mcu_cmd(sc, RT2860_MCU_CMD_BBP, 0, 0);
DELAY(1000);
}
uint8_t
rt2860_mcu_bbp_read(struct rt2860_softc *sc, uint8_t reg)
{
uint32_t val;
int ntries;
for (ntries = 0; ntries < 100; ntries++) {
if (!(RAL_READ(sc, RT2860_H2M_BBPAGENT) & RT2860_BBP_CSR_KICK))
break;
DELAY(1);
}
if (ntries == 100) {
device_printf(sc->sc_dev,
"could not read from BBP through MCU\n");
return 0;
}
RAL_WRITE(sc, RT2860_H2M_BBPAGENT, RT2860_BBP_RW_PARALLEL |
RT2860_BBP_CSR_KICK | RT2860_BBP_CSR_READ | reg << 8);
RAL_BARRIER_WRITE(sc);
rt2860_mcu_cmd(sc, RT2860_MCU_CMD_BBP, 0, 0);
DELAY(1000);
for (ntries = 0; ntries < 100; ntries++) {
val = RAL_READ(sc, RT2860_H2M_BBPAGENT);
if (!(val & RT2860_BBP_CSR_KICK))
return val & 0xff;
DELAY(1);
}
device_printf(sc->sc_dev, "could not read from BBP through MCU\n");
return 0;
}
/*
* Write to one of the 4 programmable 24-bit RF registers.
*/
static void
rt2860_rf_write(struct rt2860_softc *sc, uint8_t reg, uint32_t val)
{
uint32_t tmp;
int ntries;
for (ntries = 0; ntries < 100; ntries++) {
if (!(RAL_READ(sc, RT2860_RF_CSR_CFG0) & RT2860_RF_REG_CTRL))
break;
DELAY(1);
}
if (ntries == 100) {
device_printf(sc->sc_dev, "could not write to RF\n");
return;
}
/* RF registers are 24-bit on the RT2860 */
tmp = RT2860_RF_REG_CTRL | 24 << RT2860_RF_REG_WIDTH_SHIFT |
(val & 0x3fffff) << 2 | (reg & 3);
RAL_WRITE(sc, RT2860_RF_CSR_CFG0, tmp);
}
static uint8_t
rt3090_rf_read(struct rt2860_softc *sc, uint8_t reg)
{
uint32_t tmp;
int ntries;
for (ntries = 0; ntries < 100; ntries++) {
if (!(RAL_READ(sc, RT3070_RF_CSR_CFG) & RT3070_RF_KICK))
break;
DELAY(1);
}
if (ntries == 100) {
device_printf(sc->sc_dev, "could not read RF register\n");
return 0xff;
}
tmp = RT3070_RF_KICK | reg << 8;
RAL_WRITE(sc, RT3070_RF_CSR_CFG, tmp);
for (ntries = 0; ntries < 100; ntries++) {
tmp = RAL_READ(sc, RT3070_RF_CSR_CFG);
if (!(tmp & RT3070_RF_KICK))
break;
DELAY(1);
}
if (ntries == 100) {
device_printf(sc->sc_dev, "could not read RF register\n");
return 0xff;
}
return tmp & 0xff;
}
void
rt3090_rf_write(struct rt2860_softc *sc, uint8_t reg, uint8_t val)
{
uint32_t tmp;
int ntries;
for (ntries = 0; ntries < 10; ntries++) {
if (!(RAL_READ(sc, RT3070_RF_CSR_CFG) & RT3070_RF_KICK))
break;
DELAY(10);
}
if (ntries == 10) {
device_printf(sc->sc_dev, "could not write to RF\n");
return;
}
tmp = RT3070_RF_WRITE | RT3070_RF_KICK | reg << 8 | val;
RAL_WRITE(sc, RT3070_RF_CSR_CFG, tmp);
}
/*
* Send a command to the 8051 microcontroller unit.
*/
int
rt2860_mcu_cmd(struct rt2860_softc *sc, uint8_t cmd, uint16_t arg, int wait)
{
int slot, ntries;
uint32_t tmp;
uint8_t cid;
for (ntries = 0; ntries < 100; ntries++) {
if (!(RAL_READ(sc, RT2860_H2M_MAILBOX) & RT2860_H2M_BUSY))
break;
DELAY(2);
}
if (ntries == 100)
return EIO;
cid = wait ? cmd : RT2860_TOKEN_NO_INTR;
RAL_WRITE(sc, RT2860_H2M_MAILBOX, RT2860_H2M_BUSY | cid << 16 | arg);
RAL_BARRIER_WRITE(sc);
RAL_WRITE(sc, RT2860_HOST_CMD, cmd);
if (!wait)
return 0;
/* wait for the command to complete */
for (ntries = 0; ntries < 200; ntries++) {
tmp = RAL_READ(sc, RT2860_H2M_MAILBOX_CID);
/* find the command slot */
for (slot = 0; slot < 4; slot++, tmp >>= 8)
if ((tmp & 0xff) == cid)
break;
if (slot < 4)
break;
DELAY(100);
}
if (ntries == 200) {
/* clear command and status */
RAL_WRITE(sc, RT2860_H2M_MAILBOX_STATUS, 0xffffffff);
RAL_WRITE(sc, RT2860_H2M_MAILBOX_CID, 0xffffffff);
return ETIMEDOUT;
}
/* get command status (1 means success) */
tmp = RAL_READ(sc, RT2860_H2M_MAILBOX_STATUS);
tmp = (tmp >> (slot * 8)) & 0xff;
DPRINTF(("MCU command=0x%02x slot=%d status=0x%02x\n",
cmd, slot, tmp));
/* clear command and status */
RAL_WRITE(sc, RT2860_H2M_MAILBOX_STATUS, 0xffffffff);
RAL_WRITE(sc, RT2860_H2M_MAILBOX_CID, 0xffffffff);
return (tmp == 1) ? 0 : EIO;
}
static void
rt2860_enable_mrr(struct rt2860_softc *sc)
{
#define CCK(mcs) (mcs)
#define OFDM(mcs) (1 << 3 | (mcs))
RAL_WRITE(sc, RT2860_LG_FBK_CFG0,
OFDM(6) << 28 | /* 54->48 */
OFDM(5) << 24 | /* 48->36 */
OFDM(4) << 20 | /* 36->24 */
OFDM(3) << 16 | /* 24->18 */
OFDM(2) << 12 | /* 18->12 */
OFDM(1) << 8 | /* 12-> 9 */
OFDM(0) << 4 | /* 9-> 6 */
OFDM(0)); /* 6-> 6 */
RAL_WRITE(sc, RT2860_LG_FBK_CFG1,
CCK(2) << 12 | /* 11->5.5 */
CCK(1) << 8 | /* 5.5-> 2 */
CCK(0) << 4 | /* 2-> 1 */
CCK(0)); /* 1-> 1 */
#undef OFDM
#undef CCK
}
static void
rt2860_set_txpreamble(struct rt2860_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;
uint32_t tmp;
tmp = RAL_READ(sc, RT2860_AUTO_RSP_CFG);
tmp &= ~RT2860_CCK_SHORT_EN;
if (ic->ic_flags & IEEE80211_F_SHPREAMBLE)
tmp |= RT2860_CCK_SHORT_EN;
RAL_WRITE(sc, RT2860_AUTO_RSP_CFG, tmp);
}
void
rt2860_set_basicrates(struct rt2860_softc *sc,
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;
uint32_t mask = 0;
uint8_t rate;
int i;
for (i = 0; i < rs->rs_nrates; i++) {
rate = rs->rs_rates[i];
if (!(rate & IEEE80211_RATE_BASIC))
continue;
mask |= 1 << ieee80211_legacy_rate_lookup(ic->ic_rt,
IEEE80211_RV(rate));
}
RAL_WRITE(sc, RT2860_LEGACY_BASIC_RATE, mask);
}
static void
rt2860_scan_start(struct ieee80211com *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
struct rt2860_softc *sc = ic->ic_softc;
uint32_t tmp;
tmp = RAL_READ(sc, RT2860_BCN_TIME_CFG);
RAL_WRITE(sc, RT2860_BCN_TIME_CFG,
tmp & ~(RT2860_BCN_TX_EN | RT2860_TSF_TIMER_EN |
RT2860_TBTT_TIMER_EN));
rt2860_set_gp_timer(sc, 0);
}
static void
rt2860_scan_end(struct ieee80211com *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
struct rt2860_softc *sc = ic->ic_softc;
struct ieee80211vap *vap = TAILQ_FIRST(&ic->ic_vaps);
if (vap->iv_state == IEEE80211_S_RUN) {
rt2860_enable_tsf_sync(sc);
rt2860_set_gp_timer(sc, 500);
}
}
static void
rt2860_set_channel(struct ieee80211com *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
struct rt2860_softc *sc = ic->ic_softc;
RAL_LOCK(sc);
rt2860_switch_chan(sc, ic->ic_curchan);
RAL_UNLOCK(sc);
}
static void
rt2860_select_chan_group(struct rt2860_softc *sc, int group)
{
uint32_t tmp;
uint8_t agc;
rt2860_mcu_bbp_write(sc, 62, 0x37 - sc->lna[group]);
rt2860_mcu_bbp_write(sc, 63, 0x37 - sc->lna[group]);
rt2860_mcu_bbp_write(sc, 64, 0x37 - sc->lna[group]);
rt2860_mcu_bbp_write(sc, 86, 0x00);
if (group == 0) {
if (sc->ext_2ghz_lna) {
rt2860_mcu_bbp_write(sc, 82, 0x62);
rt2860_mcu_bbp_write(sc, 75, 0x46);
} else {
rt2860_mcu_bbp_write(sc, 82, 0x84);
rt2860_mcu_bbp_write(sc, 75, 0x50);
}
} else {
if (sc->ext_5ghz_lna) {
rt2860_mcu_bbp_write(sc, 82, 0xf2);
rt2860_mcu_bbp_write(sc, 75, 0x46);
} else {
rt2860_mcu_bbp_write(sc, 82, 0xf2);
rt2860_mcu_bbp_write(sc, 75, 0x50);
}
}
tmp = RAL_READ(sc, RT2860_TX_BAND_CFG);
tmp &= ~(RT2860_5G_BAND_SEL_N | RT2860_5G_BAND_SEL_P);
tmp |= (group == 0) ? RT2860_5G_BAND_SEL_N : RT2860_5G_BAND_SEL_P;
RAL_WRITE(sc, RT2860_TX_BAND_CFG, tmp);
/* enable appropriate Power Amplifiers and Low Noise Amplifiers */
tmp = RT2860_RFTR_EN | RT2860_TRSW_EN | RT2860_LNA_PE0_EN;
if (sc->nrxchains > 1)
tmp |= RT2860_LNA_PE1_EN;
if (sc->mac_ver == 0x3593 && sc->nrxchains > 2)
tmp |= RT3593_LNA_PE2_EN;
if (group == 0) { /* 2GHz */
tmp |= RT2860_PA_PE_G0_EN;
if (sc->ntxchains > 1)
tmp |= RT2860_PA_PE_G1_EN;
if (sc->mac_ver == 0x3593 && sc->ntxchains > 2)
tmp |= RT3593_PA_PE_G2_EN;
} else { /* 5GHz */
tmp |= RT2860_PA_PE_A0_EN;
if (sc->ntxchains > 1)
tmp |= RT2860_PA_PE_A1_EN;
if (sc->mac_ver == 0x3593 && sc->ntxchains > 2)
tmp |= RT3593_PA_PE_A2_EN;
}
RAL_WRITE(sc, RT2860_TX_PIN_CFG, tmp);
if (sc->mac_ver == 0x3593) {
tmp = RAL_READ(sc, RT2860_GPIO_CTRL);
if (sc->sc_flags & RT2860_PCIE) {
tmp &= ~0x01010000;
if (group == 0)
tmp |= 0x00010000;
} else {
tmp &= ~0x00008080;
if (group == 0)
tmp |= 0x00000080;
}
tmp = (tmp & ~0x00001000) | 0x00000010;
RAL_WRITE(sc, RT2860_GPIO_CTRL, tmp);
}
/* set initial AGC value */
if (group == 0) { /* 2GHz band */
if (sc->mac_ver >= 0x3071)
agc = 0x1c + sc->lna[0] * 2;
else
agc = 0x2e + sc->lna[0];
} else { /* 5GHz band */
agc = 0x32 + (sc->lna[group] * 5) / 3;
}
rt2860_mcu_bbp_write(sc, 66, agc);
DELAY(1000);
}
static void
rt2860_set_chan(struct rt2860_softc *sc, u_int chan)
{
const struct rfprog *rfprog = rt2860_rf2850;
uint32_t r2, r3, r4;
int8_t txpow1, txpow2;
u_int i;
/* find the settings for this channel (we know it exists) */
for (i = 0; rfprog[i].chan != chan; i++);
r2 = rfprog[i].r2;
if (sc->ntxchains == 1)
r2 |= 1 << 12; /* 1T: disable Tx chain 2 */
if (sc->nrxchains == 1)
r2 |= 1 << 15 | 1 << 4; /* 1R: disable Rx chains 2 & 3 */
else if (sc->nrxchains == 2)
r2 |= 1 << 4; /* 2R: disable Rx chain 3 */
/* use Tx power values from EEPROM */
txpow1 = sc->txpow1[i];
txpow2 = sc->txpow2[i];
if (chan > 14) {
if (txpow1 >= 0)
txpow1 = txpow1 << 1 | 1;
else
txpow1 = (7 + txpow1) << 1;
if (txpow2 >= 0)
txpow2 = txpow2 << 1 | 1;
else
txpow2 = (7 + txpow2) << 1;
}
r3 = rfprog[i].r3 | txpow1 << 7;
r4 = rfprog[i].r4 | sc->freq << 13 | txpow2 << 4;
rt2860_rf_write(sc, RT2860_RF1, rfprog[i].r1);
rt2860_rf_write(sc, RT2860_RF2, r2);
rt2860_rf_write(sc, RT2860_RF3, r3);
rt2860_rf_write(sc, RT2860_RF4, r4);
DELAY(200);
rt2860_rf_write(sc, RT2860_RF1, rfprog[i].r1);
rt2860_rf_write(sc, RT2860_RF2, r2);
rt2860_rf_write(sc, RT2860_RF3, r3 | 1);
rt2860_rf_write(sc, RT2860_RF4, r4);
DELAY(200);
rt2860_rf_write(sc, RT2860_RF1, rfprog[i].r1);
rt2860_rf_write(sc, RT2860_RF2, r2);
rt2860_rf_write(sc, RT2860_RF3, r3);
rt2860_rf_write(sc, RT2860_RF4, r4);
}
static void
rt3090_set_chan(struct rt2860_softc *sc, u_int chan)
{
int8_t txpow1, txpow2;
uint8_t rf;
int i;
/* RT3090 is 2GHz only */
KASSERT(chan >= 1 && chan <= 14, ("chan %d not support", chan));
/* find the settings for this channel (we know it exists) */
for (i = 0; rt2860_rf2850[i].chan != chan; i++);
/* use Tx power values from EEPROM */
txpow1 = sc->txpow1[i];
txpow2 = sc->txpow2[i];
rt3090_rf_write(sc, 2, rt3090_freqs[i].n);
rf = rt3090_rf_read(sc, 3);
rf = (rf & ~0x0f) | rt3090_freqs[i].k;
rt3090_rf_write(sc, 3, rf);
rf = rt3090_rf_read(sc, 6);
rf = (rf & ~0x03) | rt3090_freqs[i].r;
rt3090_rf_write(sc, 6, rf);
/* set Tx0 power */
rf = rt3090_rf_read(sc, 12);
rf = (rf & ~0x1f) | txpow1;
rt3090_rf_write(sc, 12, rf);
/* set Tx1 power */
rf = rt3090_rf_read(sc, 13);
rf = (rf & ~0x1f) | txpow2;
rt3090_rf_write(sc, 13, rf);
rf = rt3090_rf_read(sc, 1);
rf &= ~0xfc;
if (sc->ntxchains == 1)
rf |= RT3070_TX1_PD | RT3070_TX2_PD;
else if (sc->ntxchains == 2)
rf |= RT3070_TX2_PD;
if (sc->nrxchains == 1)
rf |= RT3070_RX1_PD | RT3070_RX2_PD;
else if (sc->nrxchains == 2)
rf |= RT3070_RX2_PD;
rt3090_rf_write(sc, 1, rf);
/* set RF offset */
rf = rt3090_rf_read(sc, 23);
rf = (rf & ~0x7f) | sc->freq;
rt3090_rf_write(sc, 23, rf);
/* program RF filter */
rf = rt3090_rf_read(sc, 24); /* Tx */
rf = (rf & ~0x3f) | sc->rf24_20mhz;
rt3090_rf_write(sc, 24, rf);
rf = rt3090_rf_read(sc, 31); /* Rx */
rf = (rf & ~0x3f) | sc->rf24_20mhz;
rt3090_rf_write(sc, 31, rf);
/* enable RF tuning */
rf = rt3090_rf_read(sc, 7);
rt3090_rf_write(sc, 7, rf | RT3070_TUNE);
}
static void
rt5390_set_chan(struct rt2860_softc *sc, u_int chan)
{
uint8_t h20mhz, rf, tmp;
int8_t txpow1, txpow2;
int i;
/* RT5390 is 2GHz only */
KASSERT(chan >= 1 && chan <= 14, ("chan %d not support", chan));
/* find the settings for this channel (we know it exists) */
for (i = 0; rt2860_rf2850[i].chan != chan; i++);
/* use Tx power values from EEPROM */
txpow1 = sc->txpow1[i];
txpow2 = sc->txpow2[i];
rt3090_rf_write(sc, 8, rt3090_freqs[i].n);
rt3090_rf_write(sc, 9, rt3090_freqs[i].k & 0x0f);
rf = rt3090_rf_read(sc, 11);
rf = (rf & ~0x03) | (rt3090_freqs[i].r & 0x03);
rt3090_rf_write(sc, 11, rf);
rf = rt3090_rf_read(sc, 49);
rf = (rf & ~0x3f) | (txpow1 & 0x3f);
/* the valid range of the RF R49 is 0x00~0x27 */
if ((rf & 0x3f) > 0x27)
rf = (rf & ~0x3f) | 0x27;
rt3090_rf_write(sc, 49, rf);
if (sc->mac_ver == 0x5392) {
rf = rt3090_rf_read(sc, 50);
rf = (rf & ~0x3f) | (txpow2 & 0x3f);
/* the valid range of the RF R50 is 0x00~0x27 */
if ((rf & 0x3f) > 0x27)
rf = (rf & ~0x3f) | 0x27;
rt3090_rf_write(sc, 50, rf);
}
rf = rt3090_rf_read(sc, 1);
rf |= RT3070_RF_BLOCK | RT3070_PLL_PD | RT3070_RX0_PD | RT3070_TX0_PD;
if (sc->mac_ver == 0x5392)
rf |= RT3070_RX1_PD | RT3070_TX1_PD;
rt3090_rf_write(sc, 1, rf);
rf = rt3090_rf_read(sc, 2);
rt3090_rf_write(sc, 2, rf | RT3593_RESCAL);
DELAY(1000);
rt3090_rf_write(sc, 2, rf & ~RT3593_RESCAL);
rf = rt3090_rf_read(sc, 17);
tmp = rf;
rf = (rf & ~0x7f) | (sc->freq & 0x7f);
rf = MIN(rf, 0x5f);
if (tmp != rf)
rt2860_mcu_cmd(sc, 0x74, (tmp << 8 ) | rf, 0);
if (sc->mac_ver == 0x5390) {
if (chan <= 4)
rf = 0x73;
else if (chan >= 5 && chan <= 6)
rf = 0x63;
else if (chan >= 7 && chan <= 10)
rf = 0x53;
else
rf = 43;
rt3090_rf_write(sc, 55, rf);
if (chan == 1)
rf = 0x0c;
else if (chan == 2)
rf = 0x0b;
else if (chan == 3)
rf = 0x0a;
else if (chan >= 4 && chan <= 6)
rf = 0x09;
else if (chan >= 7 && chan <= 12)
rf = 0x08;
else if (chan == 13)
rf = 0x07;
else
rf = 0x06;
rt3090_rf_write(sc, 59, rf);
}
/* Tx/Rx h20M */
h20mhz = (sc->rf24_20mhz & 0x20) >> 5;
rf = rt3090_rf_read(sc, 30);
rf = (rf & ~0x06) | (h20mhz << 1) | (h20mhz << 2);
rt3090_rf_write(sc, 30, rf);
/* Rx BB filter VCM */
rf = rt3090_rf_read(sc, 30);
rf = (rf & ~0x18) | 0x10;
rt3090_rf_write(sc, 30, rf);
/* Initiate VCO calibration. */
rf = rt3090_rf_read(sc, 3);
rf |= RT3593_VCOCAL;
rt3090_rf_write(sc, 3, rf);
}
static int
rt3090_rf_init(struct rt2860_softc *sc)
{
uint32_t tmp;
uint8_t rf, bbp;
int i;
rf = rt3090_rf_read(sc, 30);
/* toggle RF R30 bit 7 */
rt3090_rf_write(sc, 30, rf | 0x80);
DELAY(1000);
rt3090_rf_write(sc, 30, rf & ~0x80);
tmp = RAL_READ(sc, RT3070_LDO_CFG0);
tmp &= ~0x1f000000;
if (sc->patch_dac && sc->mac_rev < 0x0211)
tmp |= 0x0d000000; /* 1.35V */
else
tmp |= 0x01000000; /* 1.2V */
RAL_WRITE(sc, RT3070_LDO_CFG0, tmp);
/* patch LNA_PE_G1 */
tmp = RAL_READ(sc, RT3070_GPIO_SWITCH);
RAL_WRITE(sc, RT3070_GPIO_SWITCH, tmp & ~0x20);
/* initialize RF registers to default value */
for (i = 0; i < nitems(rt3090_def_rf); i++) {
rt3090_rf_write(sc, rt3090_def_rf[i].reg,
rt3090_def_rf[i].val);
}
/* select 20MHz bandwidth */
rt3090_rf_write(sc, 31, 0x14);
rf = rt3090_rf_read(sc, 6);
rt3090_rf_write(sc, 6, rf | 0x40);
if (sc->mac_ver != 0x3593) {
/* calibrate filter for 20MHz bandwidth */
sc->rf24_20mhz = 0x1f; /* default value */
rt3090_filter_calib(sc, 0x07, 0x16, &sc->rf24_20mhz);
/* select 40MHz bandwidth */
bbp = rt2860_mcu_bbp_read(sc, 4);
rt2860_mcu_bbp_write(sc, 4, (bbp & ~0x08) | 0x10);
rf = rt3090_rf_read(sc, 31);
rt3090_rf_write(sc, 31, rf | 0x20);
/* calibrate filter for 40MHz bandwidth */
sc->rf24_40mhz = 0x2f; /* default value */
rt3090_filter_calib(sc, 0x27, 0x19, &sc->rf24_40mhz);
/* go back to 20MHz bandwidth */
bbp = rt2860_mcu_bbp_read(sc, 4);
rt2860_mcu_bbp_write(sc, 4, bbp & ~0x18);
}
if (sc->mac_rev < 0x0211)
rt3090_rf_write(sc, 27, 0x03);
tmp = RAL_READ(sc, RT3070_OPT_14);
RAL_WRITE(sc, RT3070_OPT_14, tmp | 1);
if (sc->rf_rev == RT3070_RF_3020)
rt3090_set_rx_antenna(sc, 0);
bbp = rt2860_mcu_bbp_read(sc, 138);
if (sc->mac_ver == 0x3593) {
if (sc->ntxchains == 1)
bbp |= 0x60; /* turn off DAC1 and DAC2 */
else if (sc->ntxchains == 2)
bbp |= 0x40; /* turn off DAC2 */
if (sc->nrxchains == 1)
bbp &= ~0x06; /* turn off ADC1 and ADC2 */
else if (sc->nrxchains == 2)
bbp &= ~0x04; /* turn off ADC2 */
} else {
if (sc->ntxchains == 1)
bbp |= 0x20; /* turn off DAC1 */
if (sc->nrxchains == 1)
bbp &= ~0x02; /* turn off ADC1 */
}
rt2860_mcu_bbp_write(sc, 138, bbp);
rf = rt3090_rf_read(sc, 1);
rf &= ~(RT3070_RX0_PD | RT3070_TX0_PD);
rf |= RT3070_RF_BLOCK | RT3070_RX1_PD | RT3070_TX1_PD;
rt3090_rf_write(sc, 1, rf);
rf = rt3090_rf_read(sc, 15);
rt3090_rf_write(sc, 15, rf & ~RT3070_TX_LO2);
rf = rt3090_rf_read(sc, 17);
rf &= ~RT3070_TX_LO1;
if (sc->mac_rev >= 0x0211 && !sc->ext_2ghz_lna)
rf |= 0x20; /* fix for long range Rx issue */
if (sc->txmixgain_2ghz >= 2)
rf = (rf & ~0x7) | sc->txmixgain_2ghz;
rt3090_rf_write(sc, 17, rf);
rf = rt3090_rf_read(sc, 20);
rt3090_rf_write(sc, 20, rf & ~RT3070_RX_LO1);
rf = rt3090_rf_read(sc, 21);
rt3090_rf_write(sc, 21, rf & ~RT3070_RX_LO2);
return (0);
}
static void
rt5390_rf_init(struct rt2860_softc *sc)
{
uint8_t rf, bbp;
int i;
rf = rt3090_rf_read(sc, 2);
/* Toggle RF R2 bit 7. */
rt3090_rf_write(sc, 2, rf | RT3593_RESCAL);
DELAY(1000);
rt3090_rf_write(sc, 2, rf & ~RT3593_RESCAL);
/* Initialize RF registers to default value. */
if (sc->mac_ver == 0x5392) {
for (i = 0; i < nitems(rt5392_def_rf); i++) {
rt3090_rf_write(sc, rt5392_def_rf[i].reg,
rt5392_def_rf[i].val);
}
} else {
for (i = 0; i < nitems(rt5390_def_rf); i++) {
rt3090_rf_write(sc, rt5390_def_rf[i].reg,
rt5390_def_rf[i].val);
}
}
sc->rf24_20mhz = 0x1f;
sc->rf24_40mhz = 0x2f;
if (sc->mac_rev < 0x0211)
rt3090_rf_write(sc, 27, 0x03);
/* Set led open drain enable. */
RAL_WRITE(sc, RT3070_OPT_14, RAL_READ(sc, RT3070_OPT_14) | 1);
RAL_WRITE(sc, RT2860_TX_SW_CFG1, 0);
RAL_WRITE(sc, RT2860_TX_SW_CFG2, 0);
if (sc->mac_ver == 0x5390)
rt3090_set_rx_antenna(sc, 0);
/* Patch RSSI inaccurate issue. */
rt2860_mcu_bbp_write(sc, 79, 0x13);
rt2860_mcu_bbp_write(sc, 80, 0x05);
rt2860_mcu_bbp_write(sc, 81, 0x33);
/* Enable DC filter. */
if (sc->mac_rev >= 0x0211)
rt2860_mcu_bbp_write(sc, 103, 0xc0);
bbp = rt2860_mcu_bbp_read(sc, 138);
if (sc->ntxchains == 1)
bbp |= 0x20; /* Turn off DAC1. */
if (sc->nrxchains == 1)
bbp &= ~0x02; /* Turn off ADC1. */
rt2860_mcu_bbp_write(sc, 138, bbp);
/* Enable RX LO1 and LO2. */
rt3090_rf_write(sc, 38, rt3090_rf_read(sc, 38) & ~RT5390_RX_LO1);
rt3090_rf_write(sc, 39, rt3090_rf_read(sc, 39) & ~RT5390_RX_LO2);
/* Avoid data lost and CRC error. */
rt2860_mcu_bbp_write(sc, 4,
rt2860_mcu_bbp_read(sc, 4) | RT5390_MAC_IF_CTRL);
rf = rt3090_rf_read(sc, 30);
rf = (rf & ~0x18) | 0x10;
rt3090_rf_write(sc, 30, rf);
}
static void
rt3090_rf_wakeup(struct rt2860_softc *sc)
{
uint32_t tmp;
uint8_t rf;
if (sc->mac_ver == 0x3593) {
/* enable VCO */
rf = rt3090_rf_read(sc, 1);
rt3090_rf_write(sc, 1, rf | RT3593_VCO);
/* initiate VCO calibration */
rf = rt3090_rf_read(sc, 3);
rt3090_rf_write(sc, 3, rf | RT3593_VCOCAL);
/* enable VCO bias current control */
rf = rt3090_rf_read(sc, 6);
rt3090_rf_write(sc, 6, rf | RT3593_VCO_IC);
/* initiate res calibration */
rf = rt3090_rf_read(sc, 2);
rt3090_rf_write(sc, 2, rf | RT3593_RESCAL);
/* set reference current control to 0.33 mA */
rf = rt3090_rf_read(sc, 22);
rf &= ~RT3593_CP_IC_MASK;
rf |= 1 << RT3593_CP_IC_SHIFT;
rt3090_rf_write(sc, 22, rf);
/* enable RX CTB */
rf = rt3090_rf_read(sc, 46);
rt3090_rf_write(sc, 46, rf | RT3593_RX_CTB);
rf = rt3090_rf_read(sc, 20);
rf &= ~(RT3593_LDO_RF_VC_MASK | RT3593_LDO_PLL_VC_MASK);
rt3090_rf_write(sc, 20, rf);
} else {
/* enable RF block */
rf = rt3090_rf_read(sc, 1);
rt3090_rf_write(sc, 1, rf | RT3070_RF_BLOCK);
/* enable VCO bias current control */
rf = rt3090_rf_read(sc, 7);
rt3090_rf_write(sc, 7, rf | 0x30);
rf = rt3090_rf_read(sc, 9);
rt3090_rf_write(sc, 9, rf | 0x0e);
/* enable RX CTB */
rf = rt3090_rf_read(sc, 21);
rt3090_rf_write(sc, 21, rf | RT3070_RX_CTB);
/* fix Tx to Rx IQ glitch by raising RF voltage */
rf = rt3090_rf_read(sc, 27);
rf &= ~0x77;
if (sc->mac_rev < 0x0211)
rf |= 0x03;
rt3090_rf_write(sc, 27, rf);
}
if (sc->patch_dac && sc->mac_rev < 0x0211) {
tmp = RAL_READ(sc, RT3070_LDO_CFG0);
tmp = (tmp & ~0x1f000000) | 0x0d000000;
RAL_WRITE(sc, RT3070_LDO_CFG0, tmp);
}
}
static void
rt5390_rf_wakeup(struct rt2860_softc *sc)
{
uint32_t tmp;
uint8_t rf;
rf = rt3090_rf_read(sc, 1);
rf |= RT3070_RF_BLOCK | RT3070_PLL_PD | RT3070_RX0_PD |
RT3070_TX0_PD;
if (sc->mac_ver == 0x5392)
rf |= RT3070_RX1_PD | RT3070_TX1_PD;
rt3090_rf_write(sc, 1, rf);
rf = rt3090_rf_read(sc, 6);
rf |= RT3593_VCO_IC | RT3593_VCOCAL;
if (sc->mac_ver == 0x5390)
rf &= ~RT3593_VCO_IC;
rt3090_rf_write(sc, 6, rf);
rt3090_rf_write(sc, 2, rt3090_rf_read(sc, 2) | RT3593_RESCAL);
rf = rt3090_rf_read(sc, 22);
rf = (rf & ~0xe0) | 0x20;
rt3090_rf_write(sc, 22, rf);
rt3090_rf_write(sc, 42, rt3090_rf_read(sc, 42) | RT5390_RX_CTB);
rt3090_rf_write(sc, 20, rt3090_rf_read(sc, 20) & ~0x77);
rt3090_rf_write(sc, 3, rt3090_rf_read(sc, 3) | RT3593_VCOCAL);
if (sc->patch_dac && sc->mac_rev < 0x0211) {
tmp = RAL_READ(sc, RT3070_LDO_CFG0);
tmp = (tmp & ~0x1f000000) | 0x0d000000;
RAL_WRITE(sc, RT3070_LDO_CFG0, tmp);
}
}
static int
rt3090_filter_calib(struct rt2860_softc *sc, uint8_t init, uint8_t target,
uint8_t *val)
{
uint8_t rf22, rf24;
uint8_t bbp55_pb, bbp55_sb, delta;
int ntries;
/* program filter */
rf24 = rt3090_rf_read(sc, 24);
rf24 = (rf24 & 0xc0) | init; /* initial filter value */
rt3090_rf_write(sc, 24, rf24);
/* enable baseband loopback mode */
rf22 = rt3090_rf_read(sc, 22);
rt3090_rf_write(sc, 22, rf22 | RT3070_BB_LOOPBACK);
/* set power and frequency of passband test tone */
rt2860_mcu_bbp_write(sc, 24, 0x00);
for (ntries = 0; ntries < 100; ntries++) {
/* transmit test tone */
rt2860_mcu_bbp_write(sc, 25, 0x90);
DELAY(1000);
/* read received power */
bbp55_pb = rt2860_mcu_bbp_read(sc, 55);
if (bbp55_pb != 0)
break;
}
if (ntries == 100)
return (ETIMEDOUT);
/* set power and frequency of stopband test tone */
rt2860_mcu_bbp_write(sc, 24, 0x06);
for (ntries = 0; ntries < 100; ntries++) {
/* transmit test tone */
rt2860_mcu_bbp_write(sc, 25, 0x90);
DELAY(1000);
/* read received power */
bbp55_sb = rt2860_mcu_bbp_read(sc, 55);
delta = bbp55_pb - bbp55_sb;
if (delta > target)
break;
/* reprogram filter */
rf24++;
rt3090_rf_write(sc, 24, rf24);
}
if (ntries < 100) {
if (rf24 != init)
rf24--; /* backtrack */
*val = rf24;
rt3090_rf_write(sc, 24, rf24);
}
/* restore initial state */
rt2860_mcu_bbp_write(sc, 24, 0x00);
/* disable baseband loopback mode */
rf22 = rt3090_rf_read(sc, 22);
rt3090_rf_write(sc, 22, rf22 & ~RT3070_BB_LOOPBACK);
return (0);
}
static void
rt3090_rf_setup(struct rt2860_softc *sc)
{
uint8_t bbp;
int i;
if (sc->mac_rev >= 0x0211) {
/* enable DC filter */
rt2860_mcu_bbp_write(sc, 103, 0xc0);
/* improve power consumption */
bbp = rt2860_mcu_bbp_read(sc, 31);
rt2860_mcu_bbp_write(sc, 31, bbp & ~0x03);
}
RAL_WRITE(sc, RT2860_TX_SW_CFG1, 0);
if (sc->mac_rev < 0x0211) {
RAL_WRITE(sc, RT2860_TX_SW_CFG2,
sc->patch_dac ? 0x2c : 0x0f);
} else
RAL_WRITE(sc, RT2860_TX_SW_CFG2, 0);
/* initialize RF registers from ROM */
if (sc->mac_ver < 0x5390) {
for (i = 0; i < 10; i++) {
if (sc->rf[i].reg == 0 || sc->rf[i].reg == 0xff)
continue;
rt3090_rf_write(sc, sc->rf[i].reg, sc->rf[i].val);
}
}
}
static void
rt2860_set_leds(struct rt2860_softc *sc, uint16_t which)
{
rt2860_mcu_cmd(sc, RT2860_MCU_CMD_LEDS,
which | (sc->leds & 0x7f), 0);
}
/*
* Hardware has a general-purpose programmable timer interrupt that can
* periodically raise MAC_INT_4.
*/
static void
rt2860_set_gp_timer(struct rt2860_softc *sc, int ms)
{
uint32_t tmp;
/* disable GP timer before reprogramming it */
tmp = RAL_READ(sc, RT2860_INT_TIMER_EN);
RAL_WRITE(sc, RT2860_INT_TIMER_EN, tmp & ~RT2860_GP_TIMER_EN);
if (ms == 0)
return;
tmp = RAL_READ(sc, RT2860_INT_TIMER_CFG);
ms *= 16; /* Unit: 64us */
tmp = (tmp & 0xffff) | ms << RT2860_GP_TIMER_SHIFT;
RAL_WRITE(sc, RT2860_INT_TIMER_CFG, tmp);
/* enable GP timer */
tmp = RAL_READ(sc, RT2860_INT_TIMER_EN);
RAL_WRITE(sc, RT2860_INT_TIMER_EN, tmp | RT2860_GP_TIMER_EN);
}
static void
rt2860_set_bssid(struct rt2860_softc *sc, const uint8_t *bssid)
{
RAL_WRITE(sc, RT2860_MAC_BSSID_DW0,
bssid[0] | bssid[1] << 8 | bssid[2] << 16 | bssid[3] << 24);
RAL_WRITE(sc, RT2860_MAC_BSSID_DW1,
bssid[4] | bssid[5] << 8);
}
static void
rt2860_set_macaddr(struct rt2860_softc *sc, const uint8_t *addr)
{
RAL_WRITE(sc, RT2860_MAC_ADDR_DW0,
addr[0] | addr[1] << 8 | addr[2] << 16 | addr[3] << 24);
RAL_WRITE(sc, RT2860_MAC_ADDR_DW1,
addr[4] | addr[5] << 8 | 0xff << 16);
}
static void
rt2860_updateslot(struct ieee80211com *ic)
{
struct rt2860_softc *sc = ic->ic_softc;
uint32_t tmp;
tmp = RAL_READ(sc, RT2860_BKOFF_SLOT_CFG);
tmp &= ~0xff;
tmp |= (ic->ic_flags & IEEE80211_F_SHSLOT) ? 9 : 20;
RAL_WRITE(sc, RT2860_BKOFF_SLOT_CFG, tmp);
}
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
rt2860_updateprot(struct rt2860_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;
uint32_t tmp;
tmp = RT2860_RTSTH_EN | RT2860_PROT_NAV_SHORT | RT2860_TXOP_ALLOW_ALL;
/* setup protection frame rate (MCS code) */
tmp |= IEEE80211_IS_CHAN_5GHZ(ic->ic_curchan) ?
rt2860_rates[RT2860_RIDX_OFDM6].mcs :
rt2860_rates[RT2860_RIDX_CCK11].mcs;
/* CCK frames don't require protection */
RAL_WRITE(sc, RT2860_CCK_PROT_CFG, tmp);
if (ic->ic_flags & IEEE80211_F_USEPROT) {
if (ic->ic_protmode == IEEE80211_PROT_RTSCTS)
tmp |= RT2860_PROT_CTRL_RTS_CTS;
else if (ic->ic_protmode == IEEE80211_PROT_CTSONLY)
tmp |= RT2860_PROT_CTRL_CTS;
}
RAL_WRITE(sc, RT2860_OFDM_PROT_CFG, tmp);
}
static void
rt2860_update_promisc(struct ieee80211com *ic)
{
struct rt2860_softc *sc = ic->ic_softc;
uint32_t tmp;
tmp = RAL_READ(sc, RT2860_RX_FILTR_CFG);
tmp &= ~RT2860_DROP_NOT_MYBSS;
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_promisc == 0)
tmp |= RT2860_DROP_NOT_MYBSS;
RAL_WRITE(sc, RT2860_RX_FILTR_CFG, tmp);
}
static int
rt2860_updateedca(struct ieee80211com *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
struct rt2860_softc *sc = ic->ic_softc;
const struct wmeParams *wmep;
int aci;
wmep = ic->ic_wme.wme_chanParams.cap_wmeParams;
/* update MAC TX configuration registers */
for (aci = 0; aci < WME_NUM_AC; aci++) {
RAL_WRITE(sc, RT2860_EDCA_AC_CFG(aci),
wmep[aci].wmep_logcwmax << 16 |
wmep[aci].wmep_logcwmin << 12 |
wmep[aci].wmep_aifsn << 8 |
wmep[aci].wmep_txopLimit);
}
/* update SCH/DMA registers too */
RAL_WRITE(sc, RT2860_WMM_AIFSN_CFG,
wmep[WME_AC_VO].wmep_aifsn << 12 |
wmep[WME_AC_VI].wmep_aifsn << 8 |
wmep[WME_AC_BK].wmep_aifsn << 4 |
wmep[WME_AC_BE].wmep_aifsn);
RAL_WRITE(sc, RT2860_WMM_CWMIN_CFG,
wmep[WME_AC_VO].wmep_logcwmin << 12 |
wmep[WME_AC_VI].wmep_logcwmin << 8 |
wmep[WME_AC_BK].wmep_logcwmin << 4 |
wmep[WME_AC_BE].wmep_logcwmin);
RAL_WRITE(sc, RT2860_WMM_CWMAX_CFG,
wmep[WME_AC_VO].wmep_logcwmax << 12 |
wmep[WME_AC_VI].wmep_logcwmax << 8 |
wmep[WME_AC_BK].wmep_logcwmax << 4 |
wmep[WME_AC_BE].wmep_logcwmax);
RAL_WRITE(sc, RT2860_WMM_TXOP0_CFG,
wmep[WME_AC_BK].wmep_txopLimit << 16 |
wmep[WME_AC_BE].wmep_txopLimit);
RAL_WRITE(sc, RT2860_WMM_TXOP1_CFG,
wmep[WME_AC_VO].wmep_txopLimit << 16 |
wmep[WME_AC_VI].wmep_txopLimit);
return 0;
}
#ifdef HW_CRYPTO
static int
rt2860_set_key(struct ieee80211com *ic, struct ieee80211_node *ni,
struct ieee80211_key *k)
{
struct rt2860_softc *sc = ic->ic_softc;
bus_size_t base;
uint32_t attr;
uint8_t mode, wcid, iv[8];
/* defer setting of WEP keys until interface is brought up */
if ((ic->ic_if.if_flags & (IFF_UP | IFF_RUNNING)) !=
(IFF_UP | IFF_RUNNING))
return 0;
/* map net80211 cipher to RT2860 security mode */
switch (k->k_cipher) {
case IEEE80211_CIPHER_WEP40:
mode = RT2860_MODE_WEP40;
break;
case IEEE80211_CIPHER_WEP104:
mode = RT2860_MODE_WEP104;
break;
case IEEE80211_CIPHER_TKIP:
mode = RT2860_MODE_TKIP;
break;
case IEEE80211_CIPHER_CCMP:
mode = RT2860_MODE_AES_CCMP;
break;
default:
return EINVAL;
}
if (k->k_flags & IEEE80211_KEY_GROUP) {
wcid = 0; /* NB: update WCID0 for group keys */
base = RT2860_SKEY(0, k->k_id);
} else {
wcid = ((struct rt2860_node *)ni)->wcid;
base = RT2860_PKEY(wcid);
}
if (k->k_cipher == IEEE80211_CIPHER_TKIP) {
RAL_WRITE_REGION_1(sc, base, k->k_key, 16);
#ifndef IEEE80211_STA_ONLY
if (ic->ic_opmode == IEEE80211_M_HOSTAP) {
RAL_WRITE_REGION_1(sc, base + 16, &k->k_key[16], 8);
RAL_WRITE_REGION_1(sc, base + 24, &k->k_key[24], 8);
} else
#endif
{
RAL_WRITE_REGION_1(sc, base + 16, &k->k_key[24], 8);
RAL_WRITE_REGION_1(sc, base + 24, &k->k_key[16], 8);
}
} else
RAL_WRITE_REGION_1(sc, base, k->k_key, k->k_len);
if (!(k->k_flags & IEEE80211_KEY_GROUP) ||
(k->k_flags & IEEE80211_KEY_TX)) {
/* set initial packet number in IV+EIV */
if (k->k_cipher == IEEE80211_CIPHER_WEP40 ||
k->k_cipher == IEEE80211_CIPHER_WEP104) {
uint32_t val = arc4random();
/* skip weak IVs from Fluhrer/Mantin/Shamir */
if (val >= 0x03ff00 && (val & 0xf8ff00) == 0x00ff00)
val += 0x000100;
iv[0] = val;
iv[1] = val >> 8;
iv[2] = val >> 16;
iv[3] = k->k_id << 6;
iv[4] = iv[5] = iv[6] = iv[7] = 0;
} else {
if (k->k_cipher == IEEE80211_CIPHER_TKIP) {
iv[0] = k->k_tsc >> 8;
iv[1] = (iv[0] | 0x20) & 0x7f;
iv[2] = k->k_tsc;
} else /* CCMP */ {
iv[0] = k->k_tsc;
iv[1] = k->k_tsc >> 8;
iv[2] = 0;
}
iv[3] = k->k_id << 6 | IEEE80211_WEP_EXTIV;
iv[4] = k->k_tsc >> 16;
iv[5] = k->k_tsc >> 24;
iv[6] = k->k_tsc >> 32;
iv[7] = k->k_tsc >> 40;
}
RAL_WRITE_REGION_1(sc, RT2860_IVEIV(wcid), iv, 8);
}
if (k->k_flags & IEEE80211_KEY_GROUP) {
/* install group key */
attr = RAL_READ(sc, RT2860_SKEY_MODE_0_7);
attr &= ~(0xf << (k->k_id * 4));
attr |= mode << (k->k_id * 4);
RAL_WRITE(sc, RT2860_SKEY_MODE_0_7, attr);
} else {
/* install pairwise key */
attr = RAL_READ(sc, RT2860_WCID_ATTR(wcid));
attr = (attr & ~0xf) | (mode << 1) | RT2860_RX_PKEY_EN;
RAL_WRITE(sc, RT2860_WCID_ATTR(wcid), attr);
}
return 0;
}
static void
rt2860_delete_key(struct ieee80211com *ic, struct ieee80211_node *ni,
struct ieee80211_key *k)
{
struct rt2860_softc *sc = ic->ic_softc;
uint32_t attr;
uint8_t wcid;
if (k->k_flags & IEEE80211_KEY_GROUP) {
/* remove group key */
attr = RAL_READ(sc, RT2860_SKEY_MODE_0_7);
attr &= ~(0xf << (k->k_id * 4));
RAL_WRITE(sc, RT2860_SKEY_MODE_0_7, attr);
} else {
/* remove pairwise key */
wcid = ((struct rt2860_node *)ni)->wcid;
attr = RAL_READ(sc, RT2860_WCID_ATTR(wcid));
attr &= ~0xf;
RAL_WRITE(sc, RT2860_WCID_ATTR(wcid), attr);
}
}
#endif
static int8_t
rt2860_rssi2dbm(struct rt2860_softc *sc, uint8_t rssi, uint8_t rxchain)
{
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_channel *c = ic->ic_curchan;
int delta;
if (IEEE80211_IS_CHAN_5GHZ(c)) {
u_int chan = ieee80211_chan2ieee(ic, c);
delta = sc->rssi_5ghz[rxchain];
/* determine channel group */
if (chan <= 64)
delta -= sc->lna[1];
else if (chan <= 128)
delta -= sc->lna[2];
else
delta -= sc->lna[3];
} else
delta = sc->rssi_2ghz[rxchain] - sc->lna[0];
return -12 - delta - rssi;
}
/*
* Add `delta' (signed) to each 4-bit sub-word of a 32-bit word.
* Used to adjust per-rate Tx power registers.
*/
static __inline uint32_t
b4inc(uint32_t b32, int8_t delta)
{
int8_t i, b4;
for (i = 0; i < 8; i++) {
b4 = b32 & 0xf;
b4 += delta;
if (b4 < 0)
b4 = 0;
else if (b4 > 0xf)
b4 = 0xf;
b32 = b32 >> 4 | b4 << 28;
}
return b32;
}
static const char *
rt2860_get_rf(uint8_t rev)
{
switch (rev) {
case RT2860_RF_2820: return "RT2820";
case RT2860_RF_2850: return "RT2850";
case RT2860_RF_2720: return "RT2720";
case RT2860_RF_2750: return "RT2750";
case RT3070_RF_3020: return "RT3020";
case RT3070_RF_2020: return "RT2020";
case RT3070_RF_3021: return "RT3021";
case RT3070_RF_3022: return "RT3022";
case RT3070_RF_3052: return "RT3052";
case RT3070_RF_3320: return "RT3320";
case RT3070_RF_3053: return "RT3053";
case RT5390_RF_5390: return "RT5390";
default: return "unknown";
}
}
static int
rt2860_read_eeprom(struct rt2860_softc *sc, uint8_t macaddr[IEEE80211_ADDR_LEN])
{
int8_t delta_2ghz, delta_5ghz;
uint32_t tmp;
uint16_t val;
int ridx, ant, i;
/* check whether the ROM is eFUSE ROM or EEPROM */
sc->sc_srom_read = rt2860_eeprom_read_2;
if (sc->mac_ver >= 0x3071) {
tmp = RAL_READ(sc, RT3070_EFUSE_CTRL);
DPRINTF(("EFUSE_CTRL=0x%08x\n", tmp));
if (tmp & RT3070_SEL_EFUSE)
sc->sc_srom_read = rt3090_efuse_read_2;
}
/* read EEPROM version */
val = rt2860_srom_read(sc, RT2860_EEPROM_VERSION);
DPRINTF(("EEPROM rev=%d, FAE=%d\n", val >> 8, val & 0xff));
/* read MAC address */
val = rt2860_srom_read(sc, RT2860_EEPROM_MAC01);
macaddr[0] = val & 0xff;
macaddr[1] = val >> 8;
val = rt2860_srom_read(sc, RT2860_EEPROM_MAC23);
macaddr[2] = val & 0xff;
macaddr[3] = val >> 8;
val = rt2860_srom_read(sc, RT2860_EEPROM_MAC45);
macaddr[4] = val & 0xff;
macaddr[5] = val >> 8;
/* read country code */
val = rt2860_srom_read(sc, RT2860_EEPROM_COUNTRY);
DPRINTF(("EEPROM region code=0x%04x\n", val));
/* read vendor BBP settings */
for (i = 0; i < 8; i++) {
val = rt2860_srom_read(sc, RT2860_EEPROM_BBP_BASE + i);
sc->bbp[i].val = val & 0xff;
sc->bbp[i].reg = val >> 8;
DPRINTF(("BBP%d=0x%02x\n", sc->bbp[i].reg, sc->bbp[i].val));
}
if (sc->mac_ver >= 0x3071) {
/* read vendor RF settings */
for (i = 0; i < 10; i++) {
val = rt2860_srom_read(sc, RT3071_EEPROM_RF_BASE + i);
sc->rf[i].val = val & 0xff;
sc->rf[i].reg = val >> 8;
DPRINTF(("RF%d=0x%02x\n", sc->rf[i].reg,
sc->rf[i].val));
}
}
/* read RF frequency offset from EEPROM */
val = rt2860_srom_read(sc, RT2860_EEPROM_FREQ_LEDS);
sc->freq = ((val & 0xff) != 0xff) ? val & 0xff : 0;
DPRINTF(("EEPROM freq offset %d\n", sc->freq & 0xff));
if ((val >> 8) != 0xff) {
/* read LEDs operating mode */
sc->leds = val >> 8;
sc->led[0] = rt2860_srom_read(sc, RT2860_EEPROM_LED1);
sc->led[1] = rt2860_srom_read(sc, RT2860_EEPROM_LED2);
sc->led[2] = rt2860_srom_read(sc, RT2860_EEPROM_LED3);
} else {
/* broken EEPROM, use default settings */
sc->leds = 0x01;
sc->led[0] = 0x5555;
sc->led[1] = 0x2221;
sc->led[2] = 0xa9f8;
}
DPRINTF(("EEPROM LED mode=0x%02x, LEDs=0x%04x/0x%04x/0x%04x\n",
sc->leds, sc->led[0], sc->led[1], sc->led[2]));
/* read RF information */
val = rt2860_srom_read(sc, RT2860_EEPROM_ANTENNA);
if (val == 0xffff) {
DPRINTF(("invalid EEPROM antenna info, using default\n"));
if (sc->mac_ver >= 0x5390) {
/* default to RF5390 */
sc->rf_rev = RT5390_RF_5390;
sc->ntxchains = (sc->mac_ver == 0x5392) ? 2 : 1;
sc->nrxchains = (sc->mac_ver == 0x5392) ? 2 : 1;
} else if (sc->mac_ver == 0x3593) {
/* default to RF3053 3T3R */
sc->rf_rev = RT3070_RF_3053;
sc->ntxchains = 3;
sc->nrxchains = 3;
} else if (sc->mac_ver >= 0x3071) {
/* default to RF3020 1T1R */
sc->rf_rev = RT3070_RF_3020;
sc->ntxchains = 1;
sc->nrxchains = 1;
} else {
/* default to RF2820 1T2R */
sc->rf_rev = RT2860_RF_2820;
sc->ntxchains = 1;
sc->nrxchains = 2;
}
} else {
sc->rf_rev = (val >> 8) & 0xf;
if (sc->mac_ver >= 0x5390) {
sc->ntxchains = (sc->mac_ver == 0x5392) ? 2 : 1;
sc->nrxchains = (sc->mac_ver == 0x5392) ? 2 : 1;
} else {
sc->ntxchains = (val >> 4) & 0xf;
sc->nrxchains = val & 0xf;
}
}
DPRINTF(("EEPROM RF rev=0x%02x chains=%dT%dR\n",
sc->rf_rev, sc->ntxchains, sc->nrxchains));
/* check if RF supports automatic Tx access gain control */
val = rt2860_srom_read(sc, RT2860_EEPROM_CONFIG);
DPRINTF(("EEPROM CFG 0x%04x\n", val));
/* check if driver should patch the DAC issue */
if ((val >> 8) != 0xff)
sc->patch_dac = (val >> 15) & 1;
if ((val & 0xff) != 0xff) {
sc->ext_5ghz_lna = (val >> 3) & 1;
sc->ext_2ghz_lna = (val >> 2) & 1;
/* check if RF supports automatic Tx access gain control */
sc->calib_2ghz = sc->calib_5ghz = 0; /* XXX (val >> 1) & 1 */;
/* check if we have a hardware radio switch */
sc->rfswitch = val & 1;
}
if (sc->sc_flags & RT2860_ADVANCED_PS) {
/* read PCIe power save level */
val = rt2860_srom_read(sc, RT2860_EEPROM_PCIE_PSLEVEL);
if ((val & 0xff) != 0xff) {
sc->pslevel = val & 0x3;
val = rt2860_srom_read(sc, RT2860_EEPROM_REV);
if ((val & 0xff80) != 0x9280)
sc->pslevel = MIN(sc->pslevel, 1);
DPRINTF(("EEPROM PCIe PS Level=%d\n", sc->pslevel));
}
}
/* read power settings for 2GHz channels */
for (i = 0; i < 14; i += 2) {
val = rt2860_srom_read(sc,
RT2860_EEPROM_PWR2GHZ_BASE1 + i / 2);
sc->txpow1[i + 0] = (int8_t)(val & 0xff);
sc->txpow1[i + 1] = (int8_t)(val >> 8);
if (sc->mac_ver != 0x5390) {
val = rt2860_srom_read(sc,
RT2860_EEPROM_PWR2GHZ_BASE2 + i / 2);
sc->txpow2[i + 0] = (int8_t)(val & 0xff);
sc->txpow2[i + 1] = (int8_t)(val >> 8);
}
}
/* fix broken Tx power entries */
for (i = 0; i < 14; i++) {
if (sc->txpow1[i] < 0 ||
sc->txpow1[i] > ((sc->mac_ver >= 0x5390) ? 39 : 31))
sc->txpow1[i] = 5;
if (sc->mac_ver != 0x5390) {
if (sc->txpow2[i] < 0 ||
sc->txpow2[i] > ((sc->mac_ver == 0x5392) ? 39 : 31))
sc->txpow2[i] = 5;
}
DPRINTF(("chan %d: power1=%d, power2=%d\n",
rt2860_rf2850[i].chan, sc->txpow1[i], sc->txpow2[i]));
}
/* read power settings for 5GHz channels */
for (i = 0; i < 40; i += 2) {
val = rt2860_srom_read(sc,
RT2860_EEPROM_PWR5GHZ_BASE1 + i / 2);
sc->txpow1[i + 14] = (int8_t)(val & 0xff);
sc->txpow1[i + 15] = (int8_t)(val >> 8);
val = rt2860_srom_read(sc,
RT2860_EEPROM_PWR5GHZ_BASE2 + i / 2);
sc->txpow2[i + 14] = (int8_t)(val & 0xff);
sc->txpow2[i + 15] = (int8_t)(val >> 8);
}
/* fix broken Tx power entries */
for (i = 0; i < 40; i++) {
if (sc->txpow1[14 + i] < -7 || sc->txpow1[14 + i] > 15)
sc->txpow1[14 + i] = 5;
if (sc->txpow2[14 + i] < -7 || sc->txpow2[14 + i] > 15)
sc->txpow2[14 + i] = 5;
DPRINTF(("chan %d: power1=%d, power2=%d\n",
rt2860_rf2850[14 + i].chan, sc->txpow1[14 + i],
sc->txpow2[14 + i]));
}
/* read Tx power compensation for each Tx rate */
val = rt2860_srom_read(sc, RT2860_EEPROM_DELTAPWR);
delta_2ghz = delta_5ghz = 0;
if ((val & 0xff) != 0xff && (val & 0x80)) {
delta_2ghz = val & 0xf;
if (!(val & 0x40)) /* negative number */
delta_2ghz = -delta_2ghz;
}
val >>= 8;
if ((val & 0xff) != 0xff && (val & 0x80)) {
delta_5ghz = val & 0xf;
if (!(val & 0x40)) /* negative number */
delta_5ghz = -delta_5ghz;
}
DPRINTF(("power compensation=%d (2GHz), %d (5GHz)\n",
delta_2ghz, delta_5ghz));
for (ridx = 0; ridx < 5; ridx++) {
uint32_t reg;
val = rt2860_srom_read(sc, RT2860_EEPROM_RPWR + ridx * 2);
reg = val;
val = rt2860_srom_read(sc, RT2860_EEPROM_RPWR + ridx * 2 + 1);
reg |= (uint32_t)val << 16;
sc->txpow20mhz[ridx] = reg;
sc->txpow40mhz_2ghz[ridx] = b4inc(reg, delta_2ghz);
sc->txpow40mhz_5ghz[ridx] = b4inc(reg, delta_5ghz);
DPRINTF(("ridx %d: power 20MHz=0x%08x, 40MHz/2GHz=0x%08x, "
"40MHz/5GHz=0x%08x\n", ridx, sc->txpow20mhz[ridx],
sc->txpow40mhz_2ghz[ridx], sc->txpow40mhz_5ghz[ridx]));
}
/* read factory-calibrated samples for temperature compensation */
val = rt2860_srom_read(sc, RT2860_EEPROM_TSSI1_2GHZ);
sc->tssi_2ghz[0] = val & 0xff; /* [-4] */
sc->tssi_2ghz[1] = val >> 8; /* [-3] */
val = rt2860_srom_read(sc, RT2860_EEPROM_TSSI2_2GHZ);
sc->tssi_2ghz[2] = val & 0xff; /* [-2] */
sc->tssi_2ghz[3] = val >> 8; /* [-1] */
val = rt2860_srom_read(sc, RT2860_EEPROM_TSSI3_2GHZ);
sc->tssi_2ghz[4] = val & 0xff; /* [+0] */
sc->tssi_2ghz[5] = val >> 8; /* [+1] */
val = rt2860_srom_read(sc, RT2860_EEPROM_TSSI4_2GHZ);
sc->tssi_2ghz[6] = val & 0xff; /* [+2] */
sc->tssi_2ghz[7] = val >> 8; /* [+3] */
val = rt2860_srom_read(sc, RT2860_EEPROM_TSSI5_2GHZ);
sc->tssi_2ghz[8] = val & 0xff; /* [+4] */
sc->step_2ghz = val >> 8;
DPRINTF(("TSSI 2GHz: 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x "
"0x%02x 0x%02x step=%d\n", sc->tssi_2ghz[0], sc->tssi_2ghz[1],
sc->tssi_2ghz[2], sc->tssi_2ghz[3], sc->tssi_2ghz[4],
sc->tssi_2ghz[5], sc->tssi_2ghz[6], sc->tssi_2ghz[7],
sc->tssi_2ghz[8], sc->step_2ghz));
/* check that ref value is correct, otherwise disable calibration */
if (sc->tssi_2ghz[4] == 0xff)
sc->calib_2ghz = 0;
val = rt2860_srom_read(sc, RT2860_EEPROM_TSSI1_5GHZ);
sc->tssi_5ghz[0] = val & 0xff; /* [-4] */
sc->tssi_5ghz[1] = val >> 8; /* [-3] */
val = rt2860_srom_read(sc, RT2860_EEPROM_TSSI2_5GHZ);
sc->tssi_5ghz[2] = val & 0xff; /* [-2] */
sc->tssi_5ghz[3] = val >> 8; /* [-1] */
val = rt2860_srom_read(sc, RT2860_EEPROM_TSSI3_5GHZ);
sc->tssi_5ghz[4] = val & 0xff; /* [+0] */
sc->tssi_5ghz[5] = val >> 8; /* [+1] */
val = rt2860_srom_read(sc, RT2860_EEPROM_TSSI4_5GHZ);
sc->tssi_5ghz[6] = val & 0xff; /* [+2] */
sc->tssi_5ghz[7] = val >> 8; /* [+3] */
val = rt2860_srom_read(sc, RT2860_EEPROM_TSSI5_5GHZ);
sc->tssi_5ghz[8] = val & 0xff; /* [+4] */
sc->step_5ghz = val >> 8;
DPRINTF(("TSSI 5GHz: 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x "
"0x%02x 0x%02x step=%d\n", sc->tssi_5ghz[0], sc->tssi_5ghz[1],
sc->tssi_5ghz[2], sc->tssi_5ghz[3], sc->tssi_5ghz[4],
sc->tssi_5ghz[5], sc->tssi_5ghz[6], sc->tssi_5ghz[7],
sc->tssi_5ghz[8], sc->step_5ghz));
/* check that ref value is correct, otherwise disable calibration */
if (sc->tssi_5ghz[4] == 0xff)
sc->calib_5ghz = 0;
/* read RSSI offsets and LNA gains from EEPROM */
val = rt2860_srom_read(sc, RT2860_EEPROM_RSSI1_2GHZ);
sc->rssi_2ghz[0] = val & 0xff; /* Ant A */
sc->rssi_2ghz[1] = val >> 8; /* Ant B */
val = rt2860_srom_read(sc, RT2860_EEPROM_RSSI2_2GHZ);
if (sc->mac_ver >= 0x3071) {
/*
* On RT3090 chips (limited to 2 Rx chains), this ROM
* field contains the Tx mixer gain for the 2GHz band.
*/
if ((val & 0xff) != 0xff)
sc->txmixgain_2ghz = val & 0x7;
DPRINTF(("tx mixer gain=%u (2GHz)\n", sc->txmixgain_2ghz));
} else
sc->rssi_2ghz[2] = val & 0xff; /* Ant C */
sc->lna[2] = val >> 8; /* channel group 2 */
val = rt2860_srom_read(sc, RT2860_EEPROM_RSSI1_5GHZ);
sc->rssi_5ghz[0] = val & 0xff; /* Ant A */
sc->rssi_5ghz[1] = val >> 8; /* Ant B */
val = rt2860_srom_read(sc, RT2860_EEPROM_RSSI2_5GHZ);
sc->rssi_5ghz[2] = val & 0xff; /* Ant C */
sc->lna[3] = val >> 8; /* channel group 3 */
val = rt2860_srom_read(sc, RT2860_EEPROM_LNA);
if (sc->mac_ver >= 0x3071)
sc->lna[0] = RT3090_DEF_LNA;
else /* channel group 0 */
sc->lna[0] = val & 0xff;
sc->lna[1] = val >> 8; /* channel group 1 */
/* fix broken 5GHz LNA entries */
if (sc->lna[2] == 0 || sc->lna[2] == 0xff) {
DPRINTF(("invalid LNA for channel group %d\n", 2));
sc->lna[2] = sc->lna[1];
}
if (sc->lna[3] == 0 || sc->lna[3] == 0xff) {
DPRINTF(("invalid LNA for channel group %d\n", 3));
sc->lna[3] = sc->lna[1];
}
/* fix broken RSSI offset entries */
for (ant = 0; ant < 3; ant++) {
if (sc->rssi_2ghz[ant] < -10 || sc->rssi_2ghz[ant] > 10) {
DPRINTF(("invalid RSSI%d offset: %d (2GHz)\n",
ant + 1, sc->rssi_2ghz[ant]));
sc->rssi_2ghz[ant] = 0;
}
if (sc->rssi_5ghz[ant] < -10 || sc->rssi_5ghz[ant] > 10) {
DPRINTF(("invalid RSSI%d offset: %d (5GHz)\n",
ant + 1, sc->rssi_5ghz[ant]));
sc->rssi_5ghz[ant] = 0;
}
}
return 0;
}
static int
rt2860_bbp_init(struct rt2860_softc *sc)
{
int i, ntries;
/* wait for BBP to wake up */
for (ntries = 0; ntries < 20; ntries++) {
uint8_t bbp0 = rt2860_mcu_bbp_read(sc, 0);
if (bbp0 != 0 && bbp0 != 0xff)
break;
}
if (ntries == 20) {
device_printf(sc->sc_dev,
"timeout waiting for BBP to wake up\n");
return (ETIMEDOUT);
}
/* initialize BBP registers to default values */
if (sc->mac_ver >= 0x5390)
rt5390_bbp_init(sc);
else {
for (i = 0; i < nitems(rt2860_def_bbp); i++) {
rt2860_mcu_bbp_write(sc, rt2860_def_bbp[i].reg,
rt2860_def_bbp[i].val);
}
}
/* fix BBP84 for RT2860E */
if (sc->mac_ver == 0x2860 && sc->mac_rev != 0x0101)
rt2860_mcu_bbp_write(sc, 84, 0x19);
if (sc->mac_ver >= 0x3071) {
rt2860_mcu_bbp_write(sc, 79, 0x13);
rt2860_mcu_bbp_write(sc, 80, 0x05);
rt2860_mcu_bbp_write(sc, 81, 0x33);
} else if (sc->mac_ver == 0x2860 && sc->mac_rev == 0x0100) {
rt2860_mcu_bbp_write(sc, 69, 0x16);
rt2860_mcu_bbp_write(sc, 73, 0x12);
}
return 0;
}
static void
rt5390_bbp_init(struct rt2860_softc *sc)
{
uint8_t bbp;
int i;
/* Apply maximum likelihood detection for 2 stream case. */
if (sc->nrxchains > 1) {
bbp = rt2860_mcu_bbp_read(sc, 105);
rt2860_mcu_bbp_write(sc, 105, bbp | RT5390_MLD);
}
/* Avoid data lost and CRC error. */
bbp = rt2860_mcu_bbp_read(sc, 4);
rt2860_mcu_bbp_write(sc, 4, bbp | RT5390_MAC_IF_CTRL);
for (i = 0; i < nitems(rt5390_def_bbp); i++) {
rt2860_mcu_bbp_write(sc, rt5390_def_bbp[i].reg,
rt5390_def_bbp[i].val);
}
if (sc->mac_ver == 0x5392) {
rt2860_mcu_bbp_write(sc, 84, 0x9a);
rt2860_mcu_bbp_write(sc, 95, 0x9a);
rt2860_mcu_bbp_write(sc, 98, 0x12);
rt2860_mcu_bbp_write(sc, 106, 0x05);
rt2860_mcu_bbp_write(sc, 134, 0xd0);
rt2860_mcu_bbp_write(sc, 135, 0xf6);
}
bbp = rt2860_mcu_bbp_read(sc, 152);
rt2860_mcu_bbp_write(sc, 152, bbp | 0x80);
/* Disable hardware antenna diversity. */
if (sc->mac_ver == 0x5390)
rt2860_mcu_bbp_write(sc, 154, 0);
}
static int
rt2860_txrx_enable(struct rt2860_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;
uint32_t tmp;
int ntries;
/* enable Tx/Rx DMA engine */
RAL_WRITE(sc, RT2860_MAC_SYS_CTRL, RT2860_MAC_TX_EN);
RAL_BARRIER_READ_WRITE(sc);
for (ntries = 0; ntries < 200; ntries++) {
tmp = RAL_READ(sc, RT2860_WPDMA_GLO_CFG);
if ((tmp & (RT2860_TX_DMA_BUSY | RT2860_RX_DMA_BUSY)) == 0)
break;
DELAY(1000);
}
if (ntries == 200) {
device_printf(sc->sc_dev, "timeout waiting for DMA engine\n");
return ETIMEDOUT;
}
DELAY(50);
tmp |= RT2860_RX_DMA_EN | RT2860_TX_DMA_EN |
RT2860_WPDMA_BT_SIZE64 << RT2860_WPDMA_BT_SIZE_SHIFT;
RAL_WRITE(sc, RT2860_WPDMA_GLO_CFG, tmp);
/* set Rx filter */
tmp = RT2860_DROP_CRC_ERR | RT2860_DROP_PHY_ERR;
if (ic->ic_opmode != IEEE80211_M_MONITOR) {
tmp |= RT2860_DROP_UC_NOME | RT2860_DROP_DUPL |
RT2860_DROP_CTS | RT2860_DROP_BA | RT2860_DROP_ACK |
RT2860_DROP_VER_ERR | RT2860_DROP_CTRL_RSV |
RT2860_DROP_CFACK | RT2860_DROP_CFEND;
if (ic->ic_opmode == IEEE80211_M_STA)
tmp |= RT2860_DROP_RTS | RT2860_DROP_PSPOLL;
}
RAL_WRITE(sc, RT2860_RX_FILTR_CFG, tmp);
RAL_WRITE(sc, RT2860_MAC_SYS_CTRL,
RT2860_MAC_RX_EN | RT2860_MAC_TX_EN);
return 0;
}
static void
rt2860_init(void *arg)
{
struct rt2860_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;
RAL_LOCK(sc);
rt2860_init_locked(sc);
RAL_UNLOCK(sc);
2015-10-29 04:21:34 +00:00
if (sc->sc_flags & RT2860_RUNNING)
ieee80211_start_all(ic);
}
static void
rt2860_init_locked(struct rt2860_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 tmp;
uint8_t bbp1, bbp3;
int i, qid, ridx, ntries, error;
RAL_LOCK_ASSERT(sc);
if (sc->rfswitch) {
/* hardware has a radio switch on GPIO pin 2 */
if (!(RAL_READ(sc, RT2860_GPIO_CTRL) & (1 << 2))) {
device_printf(sc->sc_dev,
"radio is disabled by hardware switch\n");
#ifdef notyet
rt2860_stop_locked(sc);
return;
#endif
}
}
RAL_WRITE(sc, RT2860_PWR_PIN_CFG, RT2860_IO_RA_PE);
/* disable DMA */
tmp = RAL_READ(sc, RT2860_WPDMA_GLO_CFG);
tmp &= 0xff0;
RAL_WRITE(sc, RT2860_WPDMA_GLO_CFG, tmp);
/* PBF hardware reset */
RAL_WRITE(sc, RT2860_SYS_CTRL, 0xe1f);
RAL_BARRIER_WRITE(sc);
RAL_WRITE(sc, RT2860_SYS_CTRL, 0xe00);
if ((error = rt2860_load_microcode(sc)) != 0) {
device_printf(sc->sc_dev, "could not load 8051 microcode\n");
rt2860_stop_locked(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
rt2860_set_macaddr(sc, vap ? vap->iv_myaddr : ic->ic_macaddr);
/* init Tx power for all Tx rates (from EEPROM) */
for (ridx = 0; ridx < 5; ridx++) {
if (sc->txpow20mhz[ridx] == 0xffffffff)
continue;
RAL_WRITE(sc, RT2860_TX_PWR_CFG(ridx), sc->txpow20mhz[ridx]);
}
for (ntries = 0; ntries < 100; ntries++) {
tmp = RAL_READ(sc, RT2860_WPDMA_GLO_CFG);
if ((tmp & (RT2860_TX_DMA_BUSY | RT2860_RX_DMA_BUSY)) == 0)
break;
DELAY(1000);
}
if (ntries == 100) {
device_printf(sc->sc_dev, "timeout waiting for DMA engine\n");
rt2860_stop_locked(sc);
return;
}
tmp &= 0xff0;
RAL_WRITE(sc, RT2860_WPDMA_GLO_CFG, tmp);
/* reset Rx ring and all 6 Tx rings */
RAL_WRITE(sc, RT2860_WPDMA_RST_IDX, 0x1003f);
/* PBF hardware reset */
RAL_WRITE(sc, RT2860_SYS_CTRL, 0xe1f);
RAL_BARRIER_WRITE(sc);
RAL_WRITE(sc, RT2860_SYS_CTRL, 0xe00);
RAL_WRITE(sc, RT2860_PWR_PIN_CFG, RT2860_IO_RA_PE | RT2860_IO_RF_PE);
RAL_WRITE(sc, RT2860_MAC_SYS_CTRL, RT2860_BBP_HRST | RT2860_MAC_SRST);
RAL_BARRIER_WRITE(sc);
RAL_WRITE(sc, RT2860_MAC_SYS_CTRL, 0);
for (i = 0; i < nitems(rt2860_def_mac); i++)
RAL_WRITE(sc, rt2860_def_mac[i].reg, rt2860_def_mac[i].val);
if (sc->mac_ver >= 0x5390)
RAL_WRITE(sc, RT2860_TX_SW_CFG0, 0x00000404);
else if (sc->mac_ver >= 0x3071) {
/* set delay of PA_PE assertion to 1us (unit of 0.25us) */
RAL_WRITE(sc, RT2860_TX_SW_CFG0,
4 << RT2860_DLY_PAPE_EN_SHIFT);
}
if (!(RAL_READ(sc, RT2860_PCI_CFG) & RT2860_PCI_CFG_PCI)) {
sc->sc_flags |= RT2860_PCIE;
/* PCIe has different clock cycle count than PCI */
tmp = RAL_READ(sc, RT2860_US_CYC_CNT);
tmp = (tmp & ~0xff) | 0x7d;
RAL_WRITE(sc, RT2860_US_CYC_CNT, tmp);
}
/* wait while MAC is busy */
for (ntries = 0; ntries < 100; ntries++) {
if (!(RAL_READ(sc, RT2860_MAC_STATUS_REG) &
(RT2860_RX_STATUS_BUSY | RT2860_TX_STATUS_BUSY)))
break;
DELAY(1000);
}
if (ntries == 100) {
device_printf(sc->sc_dev, "timeout waiting for MAC\n");
rt2860_stop_locked(sc);
return;
}
/* clear Host to MCU mailbox */
RAL_WRITE(sc, RT2860_H2M_BBPAGENT, 0);
RAL_WRITE(sc, RT2860_H2M_MAILBOX, 0);
rt2860_mcu_cmd(sc, RT2860_MCU_CMD_RFRESET, 0, 0);
DELAY(1000);
if ((error = rt2860_bbp_init(sc)) != 0) {
rt2860_stop_locked(sc);
return;
}
/* clear RX WCID search table */
RAL_SET_REGION_4(sc, RT2860_WCID_ENTRY(0), 0, 512);
/* clear pairwise key table */
RAL_SET_REGION_4(sc, RT2860_PKEY(0), 0, 2048);
/* clear IV/EIV table */
RAL_SET_REGION_4(sc, RT2860_IVEIV(0), 0, 512);
/* clear WCID attribute table */
RAL_SET_REGION_4(sc, RT2860_WCID_ATTR(0), 0, 256);
/* clear shared key table */
RAL_SET_REGION_4(sc, RT2860_SKEY(0, 0), 0, 8 * 32);
/* clear shared key mode */
RAL_SET_REGION_4(sc, RT2860_SKEY_MODE_0_7, 0, 4);
/* init Tx rings (4 EDCAs + HCCA + Mgt) */
for (qid = 0; qid < 6; qid++) {
RAL_WRITE(sc, RT2860_TX_BASE_PTR(qid), sc->txq[qid].paddr);
RAL_WRITE(sc, RT2860_TX_MAX_CNT(qid), RT2860_TX_RING_COUNT);
RAL_WRITE(sc, RT2860_TX_CTX_IDX(qid), 0);
}
/* init Rx ring */
RAL_WRITE(sc, RT2860_RX_BASE_PTR, sc->rxq.paddr);
RAL_WRITE(sc, RT2860_RX_MAX_CNT, RT2860_RX_RING_COUNT);
RAL_WRITE(sc, RT2860_RX_CALC_IDX, RT2860_RX_RING_COUNT - 1);
/* setup maximum buffer sizes */
RAL_WRITE(sc, RT2860_MAX_LEN_CFG, 1 << 12 |
(MCLBYTES - sizeof (struct rt2860_rxwi) - 2));
for (ntries = 0; ntries < 100; ntries++) {
tmp = RAL_READ(sc, RT2860_WPDMA_GLO_CFG);
if ((tmp & (RT2860_TX_DMA_BUSY | RT2860_RX_DMA_BUSY)) == 0)
break;
DELAY(1000);
}
if (ntries == 100) {
device_printf(sc->sc_dev, "timeout waiting for DMA engine\n");
rt2860_stop_locked(sc);
return;
}
tmp &= 0xff0;
RAL_WRITE(sc, RT2860_WPDMA_GLO_CFG, tmp);
/* disable interrupts mitigation */
RAL_WRITE(sc, RT2860_DELAY_INT_CFG, 0);
/* write vendor-specific BBP values (from EEPROM) */
for (i = 0; i < 8; i++) {
if (sc->bbp[i].reg == 0 || sc->bbp[i].reg == 0xff)
continue;
rt2860_mcu_bbp_write(sc, sc->bbp[i].reg, sc->bbp[i].val);
}
/* select Main antenna for 1T1R devices */
if (sc->rf_rev == RT3070_RF_2020 ||
sc->rf_rev == RT3070_RF_3020 ||
sc->rf_rev == RT3070_RF_3320 ||
sc->mac_ver == 0x5390)
rt3090_set_rx_antenna(sc, 0);
/* send LEDs operating mode to microcontroller */
rt2860_mcu_cmd(sc, RT2860_MCU_CMD_LED1, sc->led[0], 0);
rt2860_mcu_cmd(sc, RT2860_MCU_CMD_LED2, sc->led[1], 0);
rt2860_mcu_cmd(sc, RT2860_MCU_CMD_LED3, sc->led[2], 0);
if (sc->mac_ver >= 0x5390)
rt5390_rf_init(sc);
else if (sc->mac_ver >= 0x3071) {
if ((error = rt3090_rf_init(sc)) != 0) {
rt2860_stop_locked(sc);
return;
}
}
rt2860_mcu_cmd(sc, RT2860_MCU_CMD_SLEEP, 0x02ff, 1);
rt2860_mcu_cmd(sc, RT2860_MCU_CMD_WAKEUP, 0, 1);
if (sc->mac_ver >= 0x5390)
rt5390_rf_wakeup(sc);
else if (sc->mac_ver >= 0x3071)
rt3090_rf_wakeup(sc);
/* disable non-existing Rx chains */
bbp3 = rt2860_mcu_bbp_read(sc, 3);
bbp3 &= ~(1 << 3 | 1 << 4);
if (sc->nrxchains == 2)
bbp3 |= 1 << 3;
else if (sc->nrxchains == 3)
bbp3 |= 1 << 4;
rt2860_mcu_bbp_write(sc, 3, bbp3);
/* disable non-existing Tx chains */
bbp1 = rt2860_mcu_bbp_read(sc, 1);
if (sc->ntxchains == 1)
bbp1 = (bbp1 & ~(1 << 3 | 1 << 4));
else if (sc->mac_ver == 0x3593 && sc->ntxchains == 2)
bbp1 = (bbp1 & ~(1 << 4)) | 1 << 3;
else if (sc->mac_ver == 0x3593 && sc->ntxchains == 3)
bbp1 = (bbp1 & ~(1 << 3)) | 1 << 4;
rt2860_mcu_bbp_write(sc, 1, bbp1);
if (sc->mac_ver >= 0x3071)
rt3090_rf_setup(sc);
/* select default channel */
rt2860_switch_chan(sc, ic->ic_curchan);
/* reset RF from MCU */
rt2860_mcu_cmd(sc, RT2860_MCU_CMD_RFRESET, 0, 0);
/* set RTS threshold */
tmp = RAL_READ(sc, RT2860_TX_RTS_CFG);
tmp &= ~0xffff00;
tmp |= IEEE80211_RTS_DEFAULT << 8;
RAL_WRITE(sc, RT2860_TX_RTS_CFG, tmp);
/* setup initial protection mode */
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
rt2860_updateprot(sc);
/* turn radio LED on */
rt2860_set_leds(sc, RT2860_LED_RADIO);
/* enable Tx/Rx DMA engine */
if ((error = rt2860_txrx_enable(sc)) != 0) {
rt2860_stop_locked(sc);
return;
}
/* clear pending interrupts */
RAL_WRITE(sc, RT2860_INT_STATUS, 0xffffffff);
/* enable interrupts */
RAL_WRITE(sc, RT2860_INT_MASK, 0x3fffc);
if (sc->sc_flags & RT2860_ADVANCED_PS)
rt2860_mcu_cmd(sc, RT2860_MCU_CMD_PSLEVEL, sc->pslevel, 0);
2015-10-29 04:21:34 +00:00
sc->sc_flags |= RT2860_RUNNING;
callout_reset(&sc->watchdog_ch, hz, rt2860_watchdog, sc);
}
static void
rt2860_stop(void *arg)
{
struct rt2860_softc *sc = arg;
RAL_LOCK(sc);
rt2860_stop_locked(sc);
RAL_UNLOCK(sc);
}
static void
rt2860_stop_locked(struct rt2860_softc *sc)
{
uint32_t tmp;
int qid;
2015-10-29 04:21:34 +00:00
if (sc->sc_flags & RT2860_RUNNING)
rt2860_set_leds(sc, 0); /* turn all LEDs off */
callout_stop(&sc->watchdog_ch);
sc->sc_tx_timer = 0;
2015-10-29 04:21:34 +00:00
sc->sc_flags &= ~RT2860_RUNNING;
/* disable interrupts */
RAL_WRITE(sc, RT2860_INT_MASK, 0);
/* disable GP timer */
rt2860_set_gp_timer(sc, 0);
/* disable Rx */
tmp = RAL_READ(sc, RT2860_MAC_SYS_CTRL);
tmp &= ~(RT2860_MAC_RX_EN | RT2860_MAC_TX_EN);
RAL_WRITE(sc, RT2860_MAC_SYS_CTRL, tmp);
/* reset adapter */
RAL_WRITE(sc, RT2860_MAC_SYS_CTRL, RT2860_BBP_HRST | RT2860_MAC_SRST);
RAL_BARRIER_WRITE(sc);
RAL_WRITE(sc, RT2860_MAC_SYS_CTRL, 0);
/* reset Tx and Rx rings (and reclaim TXWIs) */
sc->qfullmsk = 0;
for (qid = 0; qid < 6; qid++)
rt2860_reset_tx_ring(sc, &sc->txq[qid]);
rt2860_reset_rx_ring(sc, &sc->rxq);
}
int
rt2860_load_microcode(struct rt2860_softc *sc)
{
const struct firmware *fp;
int ntries, error;
RAL_LOCK_ASSERT(sc);
RAL_UNLOCK(sc);
fp = firmware_get("rt2860fw");
RAL_LOCK(sc);
if (fp == NULL) {
device_printf(sc->sc_dev,
"unable to receive rt2860fw firmware image\n");
return EINVAL;
}
/* set "host program ram write selection" bit */
RAL_WRITE(sc, RT2860_SYS_CTRL, RT2860_HST_PM_SEL);
/* write microcode image */
RAL_WRITE_REGION_1(sc, RT2860_FW_BASE, fp->data, fp->datasize);
/* kick microcontroller unit */
RAL_WRITE(sc, RT2860_SYS_CTRL, 0);
RAL_BARRIER_WRITE(sc);
RAL_WRITE(sc, RT2860_SYS_CTRL, RT2860_MCU_RESET);
RAL_WRITE(sc, RT2860_H2M_BBPAGENT, 0);
RAL_WRITE(sc, RT2860_H2M_MAILBOX, 0);
/* wait until microcontroller is ready */
RAL_BARRIER_READ_WRITE(sc);
for (ntries = 0; ntries < 1000; ntries++) {
if (RAL_READ(sc, RT2860_SYS_CTRL) & RT2860_MCU_READY)
break;
DELAY(1000);
}
if (ntries == 1000) {
device_printf(sc->sc_dev,
"timeout waiting for MCU to initialize\n");
error = ETIMEDOUT;
} else
error = 0;
firmware_put(fp, FIRMWARE_UNLOAD);
return error;
}
/*
* This function is called periodically to adjust Tx power based on
* temperature variation.
*/
#ifdef NOT_YET
static void
rt2860_calib(struct rt2860_softc *sc)
{
struct ieee80211com *ic = &sc->sc_ic;
const uint8_t *tssi;
uint8_t step, bbp49;
int8_t ridx, d;
/* read current temperature */
bbp49 = rt2860_mcu_bbp_read(sc, 49);
if (IEEE80211_IS_CHAN_2GHZ(ic->ic_bss->ni_chan)) {
tssi = &sc->tssi_2ghz[4];
step = sc->step_2ghz;
} else {
tssi = &sc->tssi_5ghz[4];
step = sc->step_5ghz;
}
if (bbp49 < tssi[0]) { /* lower than reference */
/* use higher Tx power than default */
for (d = 0; d > -4 && bbp49 <= tssi[d - 1]; d--);
} else if (bbp49 > tssi[0]) { /* greater than reference */
/* use lower Tx power than default */
for (d = 0; d < +4 && bbp49 >= tssi[d + 1]; d++);
} else {
/* use default Tx power */
d = 0;
}
d *= step;
DPRINTF(("BBP49=0x%02x, adjusting Tx power by %d\n", bbp49, d));
/* write adjusted Tx power values for each Tx rate */
for (ridx = 0; ridx < 5; ridx++) {
if (sc->txpow20mhz[ridx] == 0xffffffff)
continue;
RAL_WRITE(sc, RT2860_TX_PWR_CFG(ridx),
b4inc(sc->txpow20mhz[ridx], d));
}
}
#endif
static void
rt3090_set_rx_antenna(struct rt2860_softc *sc, int aux)
{
uint32_t tmp;
if (aux) {
if (sc->mac_ver == 0x5390) {
rt2860_mcu_bbp_write(sc, 152,
rt2860_mcu_bbp_read(sc, 152) & ~0x80);
} else {
tmp = RAL_READ(sc, RT2860_PCI_EECTRL);
RAL_WRITE(sc, RT2860_PCI_EECTRL, tmp & ~RT2860_C);
tmp = RAL_READ(sc, RT2860_GPIO_CTRL);
RAL_WRITE(sc, RT2860_GPIO_CTRL, (tmp & ~0x0808) | 0x08);
}
} else {
if (sc->mac_ver == 0x5390) {
rt2860_mcu_bbp_write(sc, 152,
rt2860_mcu_bbp_read(sc, 152) | 0x80);
} else {
tmp = RAL_READ(sc, RT2860_PCI_EECTRL);
RAL_WRITE(sc, RT2860_PCI_EECTRL, tmp | RT2860_C);
tmp = RAL_READ(sc, RT2860_GPIO_CTRL);
RAL_WRITE(sc, RT2860_GPIO_CTRL, tmp & ~0x0808);
}
}
}
static void
rt2860_switch_chan(struct rt2860_softc *sc, struct ieee80211_channel *c)
{
Replay r286410. Change KPI of how device drivers that provide wireless connectivity interact with the net80211 stack. Historical background: originally wireless devices created an interface, just like Ethernet devices do. Name of an interface matched the name of the driver that created. Later, wlan(4) layer was introduced, and the wlanX interfaces become the actual interface, leaving original ones as "a parent interface" of wlanX. Kernelwise, the KPI between net80211 layer and a driver became a mix of methods that pass a pointer to struct ifnet as identifier and methods that pass pointer to struct ieee80211com. From user point of view, the parent interface just hangs on in the ifconfig list, and user can't do anything useful with it. Now, the struct ifnet goes away. The struct ieee80211com is the only KPI between a device driver and net80211. Details: - The struct ieee80211com is embedded into drivers softc. - Packets are sent via new ic_transmit method, which is very much like the previous if_transmit. - Bringing parent up/down is done via new ic_parent method, which notifies driver about any changes: number of wlan(4) interfaces, number of them in promisc or allmulti state. - Device specific ioctls (if any) are received on new ic_ioctl method. - Packets/errors accounting are done by the stack. In certain cases, when driver experiences errors and can not attribute them to any specific interface, driver updates ic_oerrors or ic_ierrors counters. Details on interface configuration with new world order: - A sequence of commands needed to bring up wireless DOESN"T change. - /etc/rc.conf parameters DON'T change. - List of devices that can be used to create wlan(4) interfaces is now provided by net.wlan.devices sysctl. Most drivers in this change were converted by me, except of wpi(4), that was done by Andriy Voskoboinyk. Big thanks to Kevin Lo for testing changes to at least 8 drivers. Thanks to pluknet@, Oliver Hartmann, Olivier Cochard, gjb@, mmoll@, op@ and lev@, who also participated in testing. Reviewed by: adrian Sponsored by: Netflix Sponsored by: Nginx, Inc.
2015-08-27 08:56:39 +00:00
struct ieee80211com *ic = &sc->sc_ic;
u_int chan, group;
chan = ieee80211_chan2ieee(ic, c);
if (chan == 0 || chan == IEEE80211_CHAN_ANY)
return;
if (sc->mac_ver >= 0x5390)
rt5390_set_chan(sc, chan);
else if (sc->mac_ver >= 0x3071)
rt3090_set_chan(sc, chan);
else
rt2860_set_chan(sc, chan);
/* determine channel group */
if (chan <= 14)
group = 0;
else if (chan <= 64)
group = 1;
else if (chan <= 128)
group = 2;
else
group = 3;
/* XXX necessary only when group has changed! */
if (sc->mac_ver < 0x5390)
rt2860_select_chan_group(sc, group);
DELAY(1000);
}
static int
rt2860_setup_beacon(struct rt2860_softc *sc, struct ieee80211vap *vap)
{
struct ieee80211com *ic = vap->iv_ic;
struct rt2860_txwi txwi;
struct mbuf *m;
int ridx;
if ((m = ieee80211_beacon_alloc(vap->iv_bss)) == NULL)
return ENOBUFS;
memset(&txwi, 0, sizeof txwi);
txwi.wcid = 0xff;
txwi.len = htole16(m->m_pkthdr.len);
/* send beacons at the lowest available rate */
ridx = IEEE80211_IS_CHAN_5GHZ(ic->ic_bsschan) ?
RT2860_RIDX_OFDM6 : RT2860_RIDX_CCK1;
txwi.phy = htole16(rt2860_rates[ridx].mcs);
if (rt2860_rates[ridx].phy == IEEE80211_T_OFDM)
txwi.phy |= htole16(RT2860_PHY_OFDM);
txwi.txop = RT2860_TX_TXOP_HT;
txwi.flags = RT2860_TX_TS;
txwi.xflags = RT2860_TX_NSEQ;
RAL_WRITE_REGION_1(sc, RT2860_BCN_BASE(0),
(uint8_t *)&txwi, sizeof txwi);
RAL_WRITE_REGION_1(sc, RT2860_BCN_BASE(0) + sizeof txwi,
mtod(m, uint8_t *), m->m_pkthdr.len);
m_freem(m);
return 0;
}
static void
rt2860_enable_tsf_sync(struct rt2860_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 tmp;
tmp = RAL_READ(sc, RT2860_BCN_TIME_CFG);
tmp &= ~0x1fffff;
tmp |= vap->iv_bss->ni_intval * 16;
tmp |= RT2860_TSF_TIMER_EN | RT2860_TBTT_TIMER_EN;
if (vap->iv_opmode == IEEE80211_M_STA) {
/*
* Local TSF is always updated with remote TSF on beacon
* reception.
*/
tmp |= 1 << RT2860_TSF_SYNC_MODE_SHIFT;
}
else if (vap->iv_opmode == IEEE80211_M_IBSS ||
vap->iv_opmode == IEEE80211_M_MBSS) {
tmp |= RT2860_BCN_TX_EN;
/*
* Local TSF is updated with remote TSF on beacon reception
* only if the remote TSF is greater than local TSF.
*/
tmp |= 2 << RT2860_TSF_SYNC_MODE_SHIFT;
} else if (vap->iv_opmode == IEEE80211_M_HOSTAP) {
tmp |= RT2860_BCN_TX_EN;
/* SYNC with nobody */
tmp |= 3 << RT2860_TSF_SYNC_MODE_SHIFT;
}
RAL_WRITE(sc, RT2860_BCN_TIME_CFG, tmp);
}