linux/net/mac80211/tx.c

5808 lines
155 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/*
* Copyright 2002-2005, Instant802 Networks, Inc.
* Copyright 2005-2006, Devicescape Software, Inc.
* Copyright 2006-2007 Jiri Benc <jbenc@suse.cz>
* Copyright 2007 Johannes Berg <johannes@sipsolutions.net>
* Copyright 2013-2014 Intel Mobile Communications GmbH
* Copyright (C) 2018-2022 Intel Corporation
*
* Transmit and frame generation functions.
*/
#include <linux/kernel.h>
#include <linux/slab.h>
#include <linux/skbuff.h>
#include <linux/if_vlan.h>
#include <linux/etherdevice.h>
#include <linux/bitmap.h>
[MAC80211]: fix race conditions with keys During receive processing, we select the key long before using it and because there's no locking it is possible that we kfree() the key after having selected it but before using it for crypto operations. Obviously, this is bad. Secondly, during transmit processing, there are two possible races: We have a similar race between select_key() and using it for encryption, but we also have a race here between select_key() and hardware encryption (both when a key is removed.) This patch solves these issues by using RCU: when a key is to be freed, we first remove the pointer from the appropriate places (sdata->keys, sdata->default_key, sta->key) using rcu_assign_pointer() and then synchronize_rcu(). Then, we can safely kfree() the key and remove it from the hardware. There's a window here where the hardware may still be using it for decryption, but we can't work around that without having two hardware callbacks, one to disable the key for RX and one to disable it for TX; but the worst thing that will happen is that we receive a packet decrypted that we don't find a key for any more and then drop it. When we add a key, we first need to upload it to the hardware and then, using rcu_assign_pointer() again, link it into our structures. In the code using keys (TX/RX paths) we use rcu_dereference() to get the key and enclose the whole tx/rx section in a rcu_read_lock() ... rcu_read_unlock() block. Because we've uploaded the key to hardware before linking it into internal structures, we can guarantee that it is valid once get to into tx(). One possible race condition remains, however: when we have hardware acceleration enabled and the driver shuts down the queues, we end up queueing the frame. If now somebody removes the key, the key will be removed from hwaccel and then then driver will be asked to encrypt the frame with a key index that has been removed. Hence, drivers will need to be aware that the hw_key_index they are passed might not be under all circumstances. Most drivers will, however, simply ignore that condition and encrypt the frame with the selected key anyway, this only results in a frame being encrypted with a wrong key or dropped (rightfully) because the key was not valid. There isn't much we can do about it unless we want to walk the pending frame queue every time a key is removed and remove all frames that used it. This race condition, however, will most likely be solved once we add multiqueue support to mac80211 because then frames will be queued further up the stack instead of after being processed. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Acked-by: Michael Wu <flamingice@sourmilk.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-14 23:10:24 +08:00
#include <linux/rcupdate.h>
#include <linux/export.h>
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
#include <linux/timekeeping.h>
[NET]: Make the device list and device lookups per namespace. This patch makes most of the generic device layer network namespace safe. This patch makes dev_base_head a network namespace variable, and then it picks up a few associated variables. The functions: dev_getbyhwaddr dev_getfirsthwbytype dev_get_by_flags dev_get_by_name __dev_get_by_name dev_get_by_index __dev_get_by_index dev_ioctl dev_ethtool dev_load wireless_process_ioctl were modified to take a network namespace argument, and deal with it. vlan_ioctl_set and brioctl_set were modified so their hooks will receive a network namespace argument. So basically anthing in the core of the network stack that was affected to by the change of dev_base was modified to handle multiple network namespaces. The rest of the network stack was simply modified to explicitly use &init_net the initial network namespace. This can be fixed when those components of the network stack are modified to handle multiple network namespaces. For now the ifindex generator is left global. Fundametally ifindex numbers are per namespace, or else we will have corner case problems with migration when we get that far. At the same time there are assumptions in the network stack that the ifindex of a network device won't change. Making the ifindex number global seems a good compromise until the network stack can cope with ifindex changes when you change namespaces, and the like. Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-18 02:56:21 +08:00
#include <net/net_namespace.h>
#include <net/ieee80211_radiotap.h>
#include <net/cfg80211.h>
#include <net/mac80211.h>
#include <net/codel.h>
#include <net/codel_impl.h>
#include <asm/unaligned.h>
#include <net/fq_impl.h>
#include "ieee80211_i.h"
#include "driver-ops.h"
#include "led.h"
#include "mesh.h"
#include "wep.h"
#include "wpa.h"
#include "wme.h"
#include "rate.h"
/* misc utils */
static __le16 ieee80211_duration(struct ieee80211_tx_data *tx,
struct sk_buff *skb, int group_addr,
int next_frag_len)
{
int rate, mrate, erp, dur, i, shift = 0;
struct ieee80211_rate *txrate;
struct ieee80211_local *local = tx->local;
struct ieee80211_supported_band *sband;
struct ieee80211_hdr *hdr;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
struct ieee80211_chanctx_conf *chanctx_conf;
u32 rate_flags = 0;
/* assume HW handles this */
if (tx->rate.flags & (IEEE80211_TX_RC_MCS | IEEE80211_TX_RC_VHT_MCS))
return 0;
rcu_read_lock();
chanctx_conf = rcu_dereference(tx->sdata->vif.bss_conf.chanctx_conf);
if (chanctx_conf) {
shift = ieee80211_chandef_get_shift(&chanctx_conf->def);
rate_flags = ieee80211_chandef_rate_flags(&chanctx_conf->def);
}
rcu_read_unlock();
/* uh huh? */
if (WARN_ON_ONCE(tx->rate.idx < 0))
return 0;
sband = local->hw.wiphy->bands[info->band];
txrate = &sband->bitrates[tx->rate.idx];
erp = txrate->flags & IEEE80211_RATE_ERP_G;
/* device is expected to do this */
if (sband->band == NL80211_BAND_S1GHZ)
return 0;
/*
* data and mgmt (except PS Poll):
* - during CFP: 32768
* - during contention period:
* if addr1 is group address: 0
* if more fragments = 0 and addr1 is individual address: time to
* transmit one ACK plus SIFS
* if more fragments = 1 and addr1 is individual address: time to
* transmit next fragment plus 2 x ACK plus 3 x SIFS
*
* IEEE 802.11, 9.6:
* - control response frame (CTS or ACK) shall be transmitted using the
* same rate as the immediately previous frame in the frame exchange
* sequence, if this rate belongs to the PHY mandatory rates, or else
* at the highest possible rate belonging to the PHY rates in the
* BSSBasicRateSet
*/
hdr = (struct ieee80211_hdr *)skb->data;
if (ieee80211_is_ctl(hdr->frame_control)) {
/* TODO: These control frames are not currently sent by
* mac80211, but should they be implemented, this function
* needs to be updated to support duration field calculation.
*
* RTS: time needed to transmit pending data/mgmt frame plus
* one CTS frame plus one ACK frame plus 3 x SIFS
* CTS: duration of immediately previous RTS minus time
* required to transmit CTS and its SIFS
* ACK: 0 if immediately previous directed data/mgmt had
* more=0, with more=1 duration in ACK frame is duration
* from previous frame minus time needed to transmit ACK
* and its SIFS
* PS Poll: BIT(15) | BIT(14) | aid
*/
return 0;
}
/* data/mgmt */
if (0 /* FIX: data/mgmt during CFP */)
return cpu_to_le16(32768);
if (group_addr) /* Group address as the destination - no ACK */
return 0;
/* Individual destination address:
* IEEE 802.11, Ch. 9.6 (after IEEE 802.11g changes)
* CTS and ACK frames shall be transmitted using the highest rate in
* basic rate set that is less than or equal to the rate of the
* immediately previous frame and that is using the same modulation
* (CCK or OFDM). If no basic rate set matches with these requirements,
* the highest mandatory rate of the PHY that is less than or equal to
* the rate of the previous frame is used.
* Mandatory rates for IEEE 802.11g PHY: 1, 2, 5.5, 11, 6, 12, 24 Mbps
*/
rate = -1;
/* use lowest available if everything fails */
mrate = sband->bitrates[0].bitrate;
for (i = 0; i < sband->n_bitrates; i++) {
struct ieee80211_rate *r = &sband->bitrates[i];
if (r->bitrate > txrate->bitrate)
break;
if ((rate_flags & r->flags) != rate_flags)
continue;
if (tx->sdata->vif.bss_conf.basic_rates & BIT(i))
rate = DIV_ROUND_UP(r->bitrate, 1 << shift);
switch (sband->band) {
case NL80211_BAND_2GHZ:
case NL80211_BAND_LC: {
u32 flag;
if (tx->sdata->flags & IEEE80211_SDATA_OPERATING_GMODE)
flag = IEEE80211_RATE_MANDATORY_G;
else
flag = IEEE80211_RATE_MANDATORY_B;
if (r->flags & flag)
mrate = r->bitrate;
break;
}
case NL80211_BAND_5GHZ:
case NL80211_BAND_6GHZ:
if (r->flags & IEEE80211_RATE_MANDATORY_A)
mrate = r->bitrate;
break;
case NL80211_BAND_S1GHZ:
case NL80211_BAND_60GHZ:
/* TODO, for now fall through */
case NUM_NL80211_BANDS:
WARN_ON(1);
break;
}
}
if (rate == -1) {
/* No matching basic rate found; use highest suitable mandatory
* PHY rate */
rate = DIV_ROUND_UP(mrate, 1 << shift);
}
/* Don't calculate ACKs for QoS Frames with NoAck Policy set */
if (ieee80211_is_data_qos(hdr->frame_control) &&
*(ieee80211_get_qos_ctl(hdr)) & IEEE80211_QOS_CTL_ACK_POLICY_NOACK)
dur = 0;
else
/* Time needed to transmit ACK
* (10 bytes + 4-byte FCS = 112 bits) plus SIFS; rounded up
* to closest integer */
dur = ieee80211_frame_duration(sband->band, 10, rate, erp,
tx->sdata->vif.bss_conf.use_short_preamble,
shift);
if (next_frag_len) {
/* Frame is fragmented: duration increases with time needed to
* transmit next fragment plus ACK and 2 x SIFS. */
dur *= 2; /* ACK + SIFS */
/* next fragment */
dur += ieee80211_frame_duration(sband->band, next_frag_len,
txrate->bitrate, erp,
tx->sdata->vif.bss_conf.use_short_preamble,
shift);
}
return cpu_to_le16(dur);
}
/* tx handlers */
static ieee80211_tx_result debug_noinline
ieee80211_tx_h_dynamic_ps(struct ieee80211_tx_data *tx)
{
struct ieee80211_local *local = tx->local;
struct ieee80211_if_managed *ifmgd;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(tx->skb);
/* driver doesn't support power save */
if (!ieee80211_hw_check(&local->hw, SUPPORTS_PS))
return TX_CONTINUE;
/* hardware does dynamic power save */
if (ieee80211_hw_check(&local->hw, SUPPORTS_DYNAMIC_PS))
return TX_CONTINUE;
/* dynamic power save disabled */
if (local->hw.conf.dynamic_ps_timeout <= 0)
return TX_CONTINUE;
/* we are scanning, don't enable power save */
if (local->scanning)
return TX_CONTINUE;
if (!local->ps_sdata)
return TX_CONTINUE;
/* No point if we're going to suspend */
if (local->quiescing)
return TX_CONTINUE;
/* dynamic ps is supported only in managed mode */
if (tx->sdata->vif.type != NL80211_IFTYPE_STATION)
return TX_CONTINUE;
if (unlikely(info->flags & IEEE80211_TX_INTFL_OFFCHAN_TX_OK))
return TX_CONTINUE;
ifmgd = &tx->sdata->u.mgd;
/*
* Don't wakeup from power save if u-apsd is enabled, voip ac has
* u-apsd enabled and the frame is in voip class. This effectively
* means that even if all access categories have u-apsd enabled, in
* practise u-apsd is only used with the voip ac. This is a
* workaround for the case when received voip class packets do not
* have correct qos tag for some reason, due the network or the
* peer application.
*
* Note: ifmgd->uapsd_queues access is racy here. If the value is
* changed via debugfs, user needs to reassociate manually to have
* everything in sync.
*/
if ((ifmgd->flags & IEEE80211_STA_UAPSD_ENABLED) &&
(ifmgd->uapsd_queues & IEEE80211_WMM_IE_STA_QOSINFO_AC_VO) &&
skb_get_queue_mapping(tx->skb) == IEEE80211_AC_VO)
return TX_CONTINUE;
if (local->hw.conf.flags & IEEE80211_CONF_PS) {
ieee80211_stop_queues_by_reason(&local->hw,
IEEE80211_MAX_QUEUE_MAP,
IEEE80211_QUEUE_STOP_REASON_PS,
false);
ifmgd->flags &= ~IEEE80211_STA_NULLFUNC_ACKED;
ieee80211_queue_work(&local->hw,
&local->dynamic_ps_disable_work);
}
/* Don't restart the timer if we're not disassociated */
if (!ifmgd->associated)
return TX_CONTINUE;
mod_timer(&local->dynamic_ps_timer, jiffies +
msecs_to_jiffies(local->hw.conf.dynamic_ps_timeout));
return TX_CONTINUE;
}
static ieee80211_tx_result debug_noinline
ieee80211_tx_h_check_assoc(struct ieee80211_tx_data *tx)
{
struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)tx->skb->data;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(tx->skb);
bool assoc = false;
if (unlikely(info->flags & IEEE80211_TX_CTL_INJECTED))
return TX_CONTINUE;
if (unlikely(test_bit(SCAN_SW_SCANNING, &tx->local->scanning)) &&
test_bit(SDATA_STATE_OFFCHANNEL, &tx->sdata->state) &&
!ieee80211_is_probe_req(hdr->frame_control) &&
!ieee80211_is_any_nullfunc(hdr->frame_control))
/*
* When software scanning only nullfunc frames (to notify
* the sleep state to the AP) and probe requests (for the
* active scan) are allowed, all other frames should not be
* sent and we should not get here, but if we do
* nonetheless, drop them to avoid sending them
* off-channel. See the link below and
* ieee80211_start_scan() for more.
*
* http://article.gmane.org/gmane.linux.kernel.wireless.general/30089
*/
return TX_DROP;
if (tx->sdata->vif.type == NL80211_IFTYPE_OCB)
return TX_CONTINUE;
if (tx->flags & IEEE80211_TX_PS_BUFFERED)
return TX_CONTINUE;
if (tx->sta)
assoc = test_sta_flag(tx->sta, WLAN_STA_ASSOC);
if (likely(tx->flags & IEEE80211_TX_UNICAST)) {
if (unlikely(!assoc &&
ieee80211_is_data(hdr->frame_control))) {
#ifdef CONFIG_MAC80211_VERBOSE_DEBUG
sdata_info(tx->sdata,
"dropped data frame to not associated station %pM\n",
hdr->addr1);
#endif
I802_DEBUG_INC(tx->local->tx_handlers_drop_not_assoc);
return TX_DROP;
}
} else if (unlikely(ieee80211_is_data(hdr->frame_control) &&
ieee80211_vif_get_num_mcast_if(tx->sdata) == 0)) {
/*
* No associated STAs - no need to send multicast
* frames.
*/
return TX_DROP;
}
return TX_CONTINUE;
}
/* This function is called whenever the AP is about to exceed the maximum limit
* of buffered frames for power saving STAs. This situation should not really
* happen often during normal operation, so dropping the oldest buffered packet
* from each queue should be OK to make some room for new frames. */
static void purge_old_ps_buffers(struct ieee80211_local *local)
{
int total = 0, purged = 0;
struct sk_buff *skb;
struct ieee80211_sub_if_data *sdata;
struct sta_info *sta;
list_for_each_entry_rcu(sdata, &local->interfaces, list) {
struct ps_data *ps;
if (sdata->vif.type == NL80211_IFTYPE_AP)
ps = &sdata->u.ap.ps;
mac80211: mesh power save basics Add routines to - maintain a PS mode for each peer and a non-peer PS mode - indicate own PS mode in transmitted frames - track neighbor STAs power modes - buffer frames when neighbors are in PS mode - add TIM and Awake Window IE to beacons - release frames in Mesh Peer Service Periods Add local_pm to sta_info to represent the link-specific power mode at this station towards the remote station. When a peer link is established, use the default power mode stored in mesh config. Update the PS status if the peering status of a neighbor changes. Maintain a mesh power mode for non-peer mesh STAs. Set the non-peer power mode to active mode during peering. Authenticated mesh peering is currently not working when either node is configured to be in power save mode. Indicate the current power mode in transmitted frames. Use QoS Nulls to indicate mesh power mode transitions. For performance reasons, calls to the function setting the frame flags are placed in HWMP routing routines, as there the STA pointer is already available. Add peer_pm to sta_info to represent the peer's link-specific power mode towards the local station. Add nonpeer_pm to represent the peer's power mode towards all non-peer stations. Track power modes based on received frames. Add the ps_data structure to ieee80211_if_mesh (for TIM map, PS neighbor counter and group-addressed frame buffer). Set WLAN_STA_PS flag for STA in PS mode to use the unicast frame buffering routines in the tx path. Update num_sta_ps to buffer and release group-addressed frames after DTIM beacons. Announce the awake window duration in beacons if in light or deep sleep mode towards any peer or non-peer. Create a TIM IE similarly to AP mode and add it to mesh beacons. Parse received Awake Window IEs and check TIM IEs for buffered frames. Release frames towards peers in mesh Peer Service Periods. Use the corresponding trigger frames and monitor the MPSP status. Append a QoS Null as trigger frame if neccessary to properly end the MPSP. Currently, in HT channels MPSPs behave imperfectly and show large delay spikes and frame losses. Signed-off-by: Marco Porsch <marco@cozybit.com> Signed-off-by: Ivan Bezyazychnyy <ivan.bezyazychnyy@gmail.com> Signed-off-by: Mike Krinkin <krinkin.m.u@gmail.com> Signed-off-by: Max Filippov <jcmvbkbc@gmail.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2013-01-31 01:14:08 +08:00
else if (ieee80211_vif_is_mesh(&sdata->vif))
ps = &sdata->u.mesh.ps;
else
continue;
skb = skb_dequeue(&ps->bc_buf);
if (skb) {
purged++;
ieee80211_free_txskb(&local->hw, skb);
}
total += skb_queue_len(&ps->bc_buf);
}
/*
* Drop one frame from each station from the lowest-priority
* AC that has frames at all.
*/
list_for_each_entry_rcu(sta, &local->sta_list, list) {
int ac;
for (ac = IEEE80211_AC_BK; ac >= IEEE80211_AC_VO; ac--) {
skb = skb_dequeue(&sta->ps_tx_buf[ac]);
total += skb_queue_len(&sta->ps_tx_buf[ac]);
if (skb) {
purged++;
ieee80211_free_txskb(&local->hw, skb);
break;
}
}
}
local->total_ps_buffered = total;
ps_dbg_hw(&local->hw, "PS buffers full - purged %d frames\n", purged);
}
static ieee80211_tx_result
ieee80211_tx_h_multicast_ps_buf(struct ieee80211_tx_data *tx)
{
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(tx->skb);
struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)tx->skb->data;
struct ps_data *ps;
/*
* broadcast/multicast frame
*
mac80211: mesh power save basics Add routines to - maintain a PS mode for each peer and a non-peer PS mode - indicate own PS mode in transmitted frames - track neighbor STAs power modes - buffer frames when neighbors are in PS mode - add TIM and Awake Window IE to beacons - release frames in Mesh Peer Service Periods Add local_pm to sta_info to represent the link-specific power mode at this station towards the remote station. When a peer link is established, use the default power mode stored in mesh config. Update the PS status if the peering status of a neighbor changes. Maintain a mesh power mode for non-peer mesh STAs. Set the non-peer power mode to active mode during peering. Authenticated mesh peering is currently not working when either node is configured to be in power save mode. Indicate the current power mode in transmitted frames. Use QoS Nulls to indicate mesh power mode transitions. For performance reasons, calls to the function setting the frame flags are placed in HWMP routing routines, as there the STA pointer is already available. Add peer_pm to sta_info to represent the peer's link-specific power mode towards the local station. Add nonpeer_pm to represent the peer's power mode towards all non-peer stations. Track power modes based on received frames. Add the ps_data structure to ieee80211_if_mesh (for TIM map, PS neighbor counter and group-addressed frame buffer). Set WLAN_STA_PS flag for STA in PS mode to use the unicast frame buffering routines in the tx path. Update num_sta_ps to buffer and release group-addressed frames after DTIM beacons. Announce the awake window duration in beacons if in light or deep sleep mode towards any peer or non-peer. Create a TIM IE similarly to AP mode and add it to mesh beacons. Parse received Awake Window IEs and check TIM IEs for buffered frames. Release frames towards peers in mesh Peer Service Periods. Use the corresponding trigger frames and monitor the MPSP status. Append a QoS Null as trigger frame if neccessary to properly end the MPSP. Currently, in HT channels MPSPs behave imperfectly and show large delay spikes and frame losses. Signed-off-by: Marco Porsch <marco@cozybit.com> Signed-off-by: Ivan Bezyazychnyy <ivan.bezyazychnyy@gmail.com> Signed-off-by: Mike Krinkin <krinkin.m.u@gmail.com> Signed-off-by: Max Filippov <jcmvbkbc@gmail.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2013-01-31 01:14:08 +08:00
* If any of the associated/peer stations is in power save mode,
* the frame is buffered to be sent after DTIM beacon frame.
* This is done either by the hardware or us.
*/
mac80211: mesh power save basics Add routines to - maintain a PS mode for each peer and a non-peer PS mode - indicate own PS mode in transmitted frames - track neighbor STAs power modes - buffer frames when neighbors are in PS mode - add TIM and Awake Window IE to beacons - release frames in Mesh Peer Service Periods Add local_pm to sta_info to represent the link-specific power mode at this station towards the remote station. When a peer link is established, use the default power mode stored in mesh config. Update the PS status if the peering status of a neighbor changes. Maintain a mesh power mode for non-peer mesh STAs. Set the non-peer power mode to active mode during peering. Authenticated mesh peering is currently not working when either node is configured to be in power save mode. Indicate the current power mode in transmitted frames. Use QoS Nulls to indicate mesh power mode transitions. For performance reasons, calls to the function setting the frame flags are placed in HWMP routing routines, as there the STA pointer is already available. Add peer_pm to sta_info to represent the peer's link-specific power mode towards the local station. Add nonpeer_pm to represent the peer's power mode towards all non-peer stations. Track power modes based on received frames. Add the ps_data structure to ieee80211_if_mesh (for TIM map, PS neighbor counter and group-addressed frame buffer). Set WLAN_STA_PS flag for STA in PS mode to use the unicast frame buffering routines in the tx path. Update num_sta_ps to buffer and release group-addressed frames after DTIM beacons. Announce the awake window duration in beacons if in light or deep sleep mode towards any peer or non-peer. Create a TIM IE similarly to AP mode and add it to mesh beacons. Parse received Awake Window IEs and check TIM IEs for buffered frames. Release frames towards peers in mesh Peer Service Periods. Use the corresponding trigger frames and monitor the MPSP status. Append a QoS Null as trigger frame if neccessary to properly end the MPSP. Currently, in HT channels MPSPs behave imperfectly and show large delay spikes and frame losses. Signed-off-by: Marco Porsch <marco@cozybit.com> Signed-off-by: Ivan Bezyazychnyy <ivan.bezyazychnyy@gmail.com> Signed-off-by: Mike Krinkin <krinkin.m.u@gmail.com> Signed-off-by: Max Filippov <jcmvbkbc@gmail.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2013-01-31 01:14:08 +08:00
/* powersaving STAs currently only in AP/VLAN/mesh mode */
if (tx->sdata->vif.type == NL80211_IFTYPE_AP ||
tx->sdata->vif.type == NL80211_IFTYPE_AP_VLAN) {
if (!tx->sdata->bss)
return TX_CONTINUE;
ps = &tx->sdata->bss->ps;
mac80211: mesh power save basics Add routines to - maintain a PS mode for each peer and a non-peer PS mode - indicate own PS mode in transmitted frames - track neighbor STAs power modes - buffer frames when neighbors are in PS mode - add TIM and Awake Window IE to beacons - release frames in Mesh Peer Service Periods Add local_pm to sta_info to represent the link-specific power mode at this station towards the remote station. When a peer link is established, use the default power mode stored in mesh config. Update the PS status if the peering status of a neighbor changes. Maintain a mesh power mode for non-peer mesh STAs. Set the non-peer power mode to active mode during peering. Authenticated mesh peering is currently not working when either node is configured to be in power save mode. Indicate the current power mode in transmitted frames. Use QoS Nulls to indicate mesh power mode transitions. For performance reasons, calls to the function setting the frame flags are placed in HWMP routing routines, as there the STA pointer is already available. Add peer_pm to sta_info to represent the peer's link-specific power mode towards the local station. Add nonpeer_pm to represent the peer's power mode towards all non-peer stations. Track power modes based on received frames. Add the ps_data structure to ieee80211_if_mesh (for TIM map, PS neighbor counter and group-addressed frame buffer). Set WLAN_STA_PS flag for STA in PS mode to use the unicast frame buffering routines in the tx path. Update num_sta_ps to buffer and release group-addressed frames after DTIM beacons. Announce the awake window duration in beacons if in light or deep sleep mode towards any peer or non-peer. Create a TIM IE similarly to AP mode and add it to mesh beacons. Parse received Awake Window IEs and check TIM IEs for buffered frames. Release frames towards peers in mesh Peer Service Periods. Use the corresponding trigger frames and monitor the MPSP status. Append a QoS Null as trigger frame if neccessary to properly end the MPSP. Currently, in HT channels MPSPs behave imperfectly and show large delay spikes and frame losses. Signed-off-by: Marco Porsch <marco@cozybit.com> Signed-off-by: Ivan Bezyazychnyy <ivan.bezyazychnyy@gmail.com> Signed-off-by: Mike Krinkin <krinkin.m.u@gmail.com> Signed-off-by: Max Filippov <jcmvbkbc@gmail.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2013-01-31 01:14:08 +08:00
} else if (ieee80211_vif_is_mesh(&tx->sdata->vif)) {
ps = &tx->sdata->u.mesh.ps;
} else {
return TX_CONTINUE;
}
/* no buffering for ordered frames */
if (ieee80211_has_order(hdr->frame_control))
return TX_CONTINUE;
if (ieee80211_is_probe_req(hdr->frame_control))
return TX_CONTINUE;
if (ieee80211_hw_check(&tx->local->hw, QUEUE_CONTROL))
info->hw_queue = tx->sdata->vif.cab_queue;
/* no stations in PS mode and no buffered packets */
if (!atomic_read(&ps->num_sta_ps) && skb_queue_empty(&ps->bc_buf))
return TX_CONTINUE;
info->flags |= IEEE80211_TX_CTL_SEND_AFTER_DTIM;
/* device releases frame after DTIM beacon */
if (!ieee80211_hw_check(&tx->local->hw, HOST_BROADCAST_PS_BUFFERING))
return TX_CONTINUE;
/* buffered in mac80211 */
if (tx->local->total_ps_buffered >= TOTAL_MAX_TX_BUFFER)
purge_old_ps_buffers(tx->local);
if (skb_queue_len(&ps->bc_buf) >= AP_MAX_BC_BUFFER) {
ps_dbg(tx->sdata,
"BC TX buffer full - dropping the oldest frame\n");
ieee80211_free_txskb(&tx->local->hw, skb_dequeue(&ps->bc_buf));
} else
tx->local->total_ps_buffered++;
skb_queue_tail(&ps->bc_buf, tx->skb);
return TX_QUEUED;
}
static int ieee80211_use_mfp(__le16 fc, struct sta_info *sta,
struct sk_buff *skb)
{
if (!ieee80211_is_mgmt(fc))
return 0;
if (sta == NULL || !test_sta_flag(sta, WLAN_STA_MFP))
return 0;
if (!ieee80211_is_robust_mgmt_frame(skb))
return 0;
return 1;
}
static ieee80211_tx_result
ieee80211_tx_h_unicast_ps_buf(struct ieee80211_tx_data *tx)
{
struct sta_info *sta = tx->sta;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(tx->skb);
struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)tx->skb->data;
struct ieee80211_local *local = tx->local;
if (unlikely(!sta))
return TX_CONTINUE;
if (unlikely((test_sta_flag(sta, WLAN_STA_PS_STA) ||
test_sta_flag(sta, WLAN_STA_PS_DRIVER) ||
test_sta_flag(sta, WLAN_STA_PS_DELIVER)) &&
!(info->flags & IEEE80211_TX_CTL_NO_PS_BUFFER))) {
int ac = skb_get_queue_mapping(tx->skb);
if (ieee80211_is_mgmt(hdr->frame_control) &&
!ieee80211_is_bufferable_mmpdu(hdr->frame_control)) {
info->flags |= IEEE80211_TX_CTL_NO_PS_BUFFER;
return TX_CONTINUE;
}
ps_dbg(sta->sdata, "STA %pM aid %d: PS buffer for AC %d\n",
sta->sta.addr, sta->sta.aid, ac);
if (tx->local->total_ps_buffered >= TOTAL_MAX_TX_BUFFER)
purge_old_ps_buffers(tx->local);
mac80211: fix AP powersave TX vs. wakeup race There is a race between the TX path and the STA wakeup: while a station is sleeping, mac80211 buffers frames until it wakes up, then the frames are transmitted. However, the RX and TX path are concurrent, so the packet indicating wakeup can be processed while a packet is being transmitted. This can lead to a situation where the buffered frames list is emptied on the one side, while a frame is being added on the other side, as the station is still seen as sleeping in the TX path. As a result, the newly added frame will not be send anytime soon. It might be sent much later (and out of order) when the station goes to sleep and wakes up the next time. Additionally, it can lead to the crash below. Fix all this by synchronising both paths with a new lock. Both path are not fastpath since they handle PS situations. In a later patch we'll remove the extra skb queue locks to reduce locking overhead. BUG: unable to handle kernel NULL pointer dereference at 000000b0 IP: [<ff6f1791>] ieee80211_report_used_skb+0x11/0x3e0 [mac80211] *pde = 00000000 Oops: 0000 [#1] SMP DEBUG_PAGEALLOC EIP: 0060:[<ff6f1791>] EFLAGS: 00210282 CPU: 1 EIP is at ieee80211_report_used_skb+0x11/0x3e0 [mac80211] EAX: e5900da0 EBX: 00000000 ECX: 00000001 EDX: 00000000 ESI: e41d00c0 EDI: e5900da0 EBP: ebe458e4 ESP: ebe458b0 DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068 CR0: 8005003b CR2: 000000b0 CR3: 25a78000 CR4: 000407d0 DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000 DR6: ffff0ff0 DR7: 00000400 Process iperf (pid: 3934, ti=ebe44000 task=e757c0b0 task.ti=ebe44000) iwlwifi 0000:02:00.0: I iwl_pcie_enqueue_hcmd Sending command LQ_CMD (#4e), seq: 0x0903, 92 bytes at 3[3]:9 Stack: e403b32c ebe458c4 00200002 00200286 e403b338 ebe458cc c10960bb e5900da0 ff76a6ec ebe458d8 00000000 e41d00c0 e5900da0 ebe458f0 ff6f1b75 e403b210 ebe4598c ff723dc1 00000000 ff76a6ec e597c978 e403b758 00000002 00000002 Call Trace: [<ff6f1b75>] ieee80211_free_txskb+0x15/0x20 [mac80211] [<ff723dc1>] invoke_tx_handlers+0x1661/0x1780 [mac80211] [<ff7248a5>] ieee80211_tx+0x75/0x100 [mac80211] [<ff7249bf>] ieee80211_xmit+0x8f/0xc0 [mac80211] [<ff72550e>] ieee80211_subif_start_xmit+0x4fe/0xe20 [mac80211] [<c149ef70>] dev_hard_start_xmit+0x450/0x950 [<c14b9aa9>] sch_direct_xmit+0xa9/0x250 [<c14b9c9b>] __qdisc_run+0x4b/0x150 [<c149f732>] dev_queue_xmit+0x2c2/0xca0 Cc: stable@vger.kernel.org Reported-by: Yaara Rozenblum <yaara.rozenblum@intel.com> Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com> Reviewed-by: Stanislaw Gruszka <sgruszka@redhat.com> [reword commit log, use a separate lock] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-02-20 15:22:11 +08:00
/* sync with ieee80211_sta_ps_deliver_wakeup */
spin_lock(&sta->ps_lock);
/*
* STA woke up the meantime and all the frames on ps_tx_buf have
* been queued to pending queue. No reordering can happen, go
* ahead and Tx the packet.
*/
if (!test_sta_flag(sta, WLAN_STA_PS_STA) &&
!test_sta_flag(sta, WLAN_STA_PS_DRIVER) &&
!test_sta_flag(sta, WLAN_STA_PS_DELIVER)) {
mac80211: fix AP powersave TX vs. wakeup race There is a race between the TX path and the STA wakeup: while a station is sleeping, mac80211 buffers frames until it wakes up, then the frames are transmitted. However, the RX and TX path are concurrent, so the packet indicating wakeup can be processed while a packet is being transmitted. This can lead to a situation where the buffered frames list is emptied on the one side, while a frame is being added on the other side, as the station is still seen as sleeping in the TX path. As a result, the newly added frame will not be send anytime soon. It might be sent much later (and out of order) when the station goes to sleep and wakes up the next time. Additionally, it can lead to the crash below. Fix all this by synchronising both paths with a new lock. Both path are not fastpath since they handle PS situations. In a later patch we'll remove the extra skb queue locks to reduce locking overhead. BUG: unable to handle kernel NULL pointer dereference at 000000b0 IP: [<ff6f1791>] ieee80211_report_used_skb+0x11/0x3e0 [mac80211] *pde = 00000000 Oops: 0000 [#1] SMP DEBUG_PAGEALLOC EIP: 0060:[<ff6f1791>] EFLAGS: 00210282 CPU: 1 EIP is at ieee80211_report_used_skb+0x11/0x3e0 [mac80211] EAX: e5900da0 EBX: 00000000 ECX: 00000001 EDX: 00000000 ESI: e41d00c0 EDI: e5900da0 EBP: ebe458e4 ESP: ebe458b0 DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068 CR0: 8005003b CR2: 000000b0 CR3: 25a78000 CR4: 000407d0 DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000 DR6: ffff0ff0 DR7: 00000400 Process iperf (pid: 3934, ti=ebe44000 task=e757c0b0 task.ti=ebe44000) iwlwifi 0000:02:00.0: I iwl_pcie_enqueue_hcmd Sending command LQ_CMD (#4e), seq: 0x0903, 92 bytes at 3[3]:9 Stack: e403b32c ebe458c4 00200002 00200286 e403b338 ebe458cc c10960bb e5900da0 ff76a6ec ebe458d8 00000000 e41d00c0 e5900da0 ebe458f0 ff6f1b75 e403b210 ebe4598c ff723dc1 00000000 ff76a6ec e597c978 e403b758 00000002 00000002 Call Trace: [<ff6f1b75>] ieee80211_free_txskb+0x15/0x20 [mac80211] [<ff723dc1>] invoke_tx_handlers+0x1661/0x1780 [mac80211] [<ff7248a5>] ieee80211_tx+0x75/0x100 [mac80211] [<ff7249bf>] ieee80211_xmit+0x8f/0xc0 [mac80211] [<ff72550e>] ieee80211_subif_start_xmit+0x4fe/0xe20 [mac80211] [<c149ef70>] dev_hard_start_xmit+0x450/0x950 [<c14b9aa9>] sch_direct_xmit+0xa9/0x250 [<c14b9c9b>] __qdisc_run+0x4b/0x150 [<c149f732>] dev_queue_xmit+0x2c2/0xca0 Cc: stable@vger.kernel.org Reported-by: Yaara Rozenblum <yaara.rozenblum@intel.com> Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com> Reviewed-by: Stanislaw Gruszka <sgruszka@redhat.com> [reword commit log, use a separate lock] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-02-20 15:22:11 +08:00
spin_unlock(&sta->ps_lock);
return TX_CONTINUE;
}
if (skb_queue_len(&sta->ps_tx_buf[ac]) >= STA_MAX_TX_BUFFER) {
struct sk_buff *old = skb_dequeue(&sta->ps_tx_buf[ac]);
ps_dbg(tx->sdata,
"STA %pM TX buffer for AC %d full - dropping oldest frame\n",
sta->sta.addr, ac);
ieee80211_free_txskb(&local->hw, old);
} else
tx->local->total_ps_buffered++;
info->control.jiffies = jiffies;
info->control.vif = &tx->sdata->vif;
info->control.flags |= IEEE80211_TX_INTCFL_NEED_TXPROCESSING;
info->flags &= ~IEEE80211_TX_TEMPORARY_FLAGS;
skb_queue_tail(&sta->ps_tx_buf[ac], tx->skb);
mac80211: fix AP powersave TX vs. wakeup race There is a race between the TX path and the STA wakeup: while a station is sleeping, mac80211 buffers frames until it wakes up, then the frames are transmitted. However, the RX and TX path are concurrent, so the packet indicating wakeup can be processed while a packet is being transmitted. This can lead to a situation where the buffered frames list is emptied on the one side, while a frame is being added on the other side, as the station is still seen as sleeping in the TX path. As a result, the newly added frame will not be send anytime soon. It might be sent much later (and out of order) when the station goes to sleep and wakes up the next time. Additionally, it can lead to the crash below. Fix all this by synchronising both paths with a new lock. Both path are not fastpath since they handle PS situations. In a later patch we'll remove the extra skb queue locks to reduce locking overhead. BUG: unable to handle kernel NULL pointer dereference at 000000b0 IP: [<ff6f1791>] ieee80211_report_used_skb+0x11/0x3e0 [mac80211] *pde = 00000000 Oops: 0000 [#1] SMP DEBUG_PAGEALLOC EIP: 0060:[<ff6f1791>] EFLAGS: 00210282 CPU: 1 EIP is at ieee80211_report_used_skb+0x11/0x3e0 [mac80211] EAX: e5900da0 EBX: 00000000 ECX: 00000001 EDX: 00000000 ESI: e41d00c0 EDI: e5900da0 EBP: ebe458e4 ESP: ebe458b0 DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068 CR0: 8005003b CR2: 000000b0 CR3: 25a78000 CR4: 000407d0 DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000 DR6: ffff0ff0 DR7: 00000400 Process iperf (pid: 3934, ti=ebe44000 task=e757c0b0 task.ti=ebe44000) iwlwifi 0000:02:00.0: I iwl_pcie_enqueue_hcmd Sending command LQ_CMD (#4e), seq: 0x0903, 92 bytes at 3[3]:9 Stack: e403b32c ebe458c4 00200002 00200286 e403b338 ebe458cc c10960bb e5900da0 ff76a6ec ebe458d8 00000000 e41d00c0 e5900da0 ebe458f0 ff6f1b75 e403b210 ebe4598c ff723dc1 00000000 ff76a6ec e597c978 e403b758 00000002 00000002 Call Trace: [<ff6f1b75>] ieee80211_free_txskb+0x15/0x20 [mac80211] [<ff723dc1>] invoke_tx_handlers+0x1661/0x1780 [mac80211] [<ff7248a5>] ieee80211_tx+0x75/0x100 [mac80211] [<ff7249bf>] ieee80211_xmit+0x8f/0xc0 [mac80211] [<ff72550e>] ieee80211_subif_start_xmit+0x4fe/0xe20 [mac80211] [<c149ef70>] dev_hard_start_xmit+0x450/0x950 [<c14b9aa9>] sch_direct_xmit+0xa9/0x250 [<c14b9c9b>] __qdisc_run+0x4b/0x150 [<c149f732>] dev_queue_xmit+0x2c2/0xca0 Cc: stable@vger.kernel.org Reported-by: Yaara Rozenblum <yaara.rozenblum@intel.com> Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com> Reviewed-by: Stanislaw Gruszka <sgruszka@redhat.com> [reword commit log, use a separate lock] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-02-20 15:22:11 +08:00
spin_unlock(&sta->ps_lock);
if (!timer_pending(&local->sta_cleanup))
mod_timer(&local->sta_cleanup,
round_jiffies(jiffies +
STA_INFO_CLEANUP_INTERVAL));
/*
* We queued up some frames, so the TIM bit might
* need to be set, recalculate it.
*/
sta_info_recalc_tim(sta);
return TX_QUEUED;
} else if (unlikely(test_sta_flag(sta, WLAN_STA_PS_STA))) {
ps_dbg(tx->sdata,
"STA %pM in PS mode, but polling/in SP -> send frame\n",
sta->sta.addr);
}
return TX_CONTINUE;
}
static ieee80211_tx_result debug_noinline
ieee80211_tx_h_ps_buf(struct ieee80211_tx_data *tx)
{
if (unlikely(tx->flags & IEEE80211_TX_PS_BUFFERED))
return TX_CONTINUE;
if (tx->flags & IEEE80211_TX_UNICAST)
return ieee80211_tx_h_unicast_ps_buf(tx);
else
return ieee80211_tx_h_multicast_ps_buf(tx);
}
static ieee80211_tx_result debug_noinline
ieee80211_tx_h_check_control_port_protocol(struct ieee80211_tx_data *tx)
{
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(tx->skb);
if (unlikely(tx->sdata->control_port_protocol == tx->skb->protocol)) {
if (tx->sdata->control_port_no_encrypt)
info->flags |= IEEE80211_TX_INTFL_DONT_ENCRYPT;
info->control.flags |= IEEE80211_TX_CTRL_PORT_CTRL_PROTO;
mac80211: Send EAPOL frames at lowest rate The current minstrel_ht rate control behavior is somewhat optimistic in trying to find optimum TX rate. While this is usually fine for normal Data frames, there are cases where a more conservative set of retry parameters would be beneficial to make the connection more robust. EAPOL frames are critical to the authentication and especially the EAPOL-Key message 4/4 (the last message in the 4-way handshake) is important to get through to the AP. If that message is lost, the only recovery mechanism in many cases is to reassociate with the AP and start from scratch. This can often be avoided by trying to send the frame with more conservative rate and/or with more link layer retries. In most cases, minstrel_ht is currently using the initial EAPOL-Key frames for probing higher rates and this results in only five link layer transmission attempts (one at high(ish) MCS and four at MCS0). While this works with most APs, it looks like there are some deployed APs that may have issues with the EAPOL frames using HT MCS immediately after association. Similarly, there may be issues in cases where the signal strength or radio environment is not good enough to be able to get frames through even at couple of MCS 0 tries. The best approach for this would likely to be to reduce the TX rate for the last rate (3rd rate parameter in the set) to a low basic rate (say, 6 Mbps on 5 GHz and 2 or 5.5 Mbps on 2.4 GHz), but doing that cleanly requires some more effort. For now, we can start with a simple one-liner that forces the minimum rate to be used for EAPOL frames similarly how the TX rate is selected for the IEEE 802.11 Management frames. This does result in a small extra latency added to the cases where the AP would be able to receive the higher rate, but taken into account how small number of EAPOL frames are used, this is likely to be insignificant. A future optimization in the minstrel_ht design can also allow this patch to be reverted to get back to the more optimized initial TX rate. It should also be noted that many drivers that do not use minstrel as the rate control algorithm are already doing similar workarounds by forcing the lowest TX rate to be used for EAPOL frames. Cc: stable@vger.kernel.org Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Tested-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2015-02-26 21:50:50 +08:00
info->flags |= IEEE80211_TX_CTL_USE_MINRATE;
}
return TX_CONTINUE;
}
static ieee80211_tx_result debug_noinline
ieee80211_tx_h_select_key(struct ieee80211_tx_data *tx)
{
struct ieee80211_key *key;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(tx->skb);
struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)tx->skb->data;
[MAC80211]: fix race conditions with keys During receive processing, we select the key long before using it and because there's no locking it is possible that we kfree() the key after having selected it but before using it for crypto operations. Obviously, this is bad. Secondly, during transmit processing, there are two possible races: We have a similar race between select_key() and using it for encryption, but we also have a race here between select_key() and hardware encryption (both when a key is removed.) This patch solves these issues by using RCU: when a key is to be freed, we first remove the pointer from the appropriate places (sdata->keys, sdata->default_key, sta->key) using rcu_assign_pointer() and then synchronize_rcu(). Then, we can safely kfree() the key and remove it from the hardware. There's a window here where the hardware may still be using it for decryption, but we can't work around that without having two hardware callbacks, one to disable the key for RX and one to disable it for TX; but the worst thing that will happen is that we receive a packet decrypted that we don't find a key for any more and then drop it. When we add a key, we first need to upload it to the hardware and then, using rcu_assign_pointer() again, link it into our structures. In the code using keys (TX/RX paths) we use rcu_dereference() to get the key and enclose the whole tx/rx section in a rcu_read_lock() ... rcu_read_unlock() block. Because we've uploaded the key to hardware before linking it into internal structures, we can guarantee that it is valid once get to into tx(). One possible race condition remains, however: when we have hardware acceleration enabled and the driver shuts down the queues, we end up queueing the frame. If now somebody removes the key, the key will be removed from hwaccel and then then driver will be asked to encrypt the frame with a key index that has been removed. Hence, drivers will need to be aware that the hw_key_index they are passed might not be under all circumstances. Most drivers will, however, simply ignore that condition and encrypt the frame with the selected key anyway, this only results in a frame being encrypted with a wrong key or dropped (rightfully) because the key was not valid. There isn't much we can do about it unless we want to walk the pending frame queue every time a key is removed and remove all frames that used it. This race condition, however, will most likely be solved once we add multiqueue support to mac80211 because then frames will be queued further up the stack instead of after being processed. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Acked-by: Michael Wu <flamingice@sourmilk.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-14 23:10:24 +08:00
if (unlikely(info->flags & IEEE80211_TX_INTFL_DONT_ENCRYPT)) {
tx->key = NULL;
return TX_CONTINUE;
}
if (tx->sta &&
(key = rcu_dereference(tx->sta->ptk[tx->sta->ptk_idx])))
[MAC80211]: fix race conditions with keys During receive processing, we select the key long before using it and because there's no locking it is possible that we kfree() the key after having selected it but before using it for crypto operations. Obviously, this is bad. Secondly, during transmit processing, there are two possible races: We have a similar race between select_key() and using it for encryption, but we also have a race here between select_key() and hardware encryption (both when a key is removed.) This patch solves these issues by using RCU: when a key is to be freed, we first remove the pointer from the appropriate places (sdata->keys, sdata->default_key, sta->key) using rcu_assign_pointer() and then synchronize_rcu(). Then, we can safely kfree() the key and remove it from the hardware. There's a window here where the hardware may still be using it for decryption, but we can't work around that without having two hardware callbacks, one to disable the key for RX and one to disable it for TX; but the worst thing that will happen is that we receive a packet decrypted that we don't find a key for any more and then drop it. When we add a key, we first need to upload it to the hardware and then, using rcu_assign_pointer() again, link it into our structures. In the code using keys (TX/RX paths) we use rcu_dereference() to get the key and enclose the whole tx/rx section in a rcu_read_lock() ... rcu_read_unlock() block. Because we've uploaded the key to hardware before linking it into internal structures, we can guarantee that it is valid once get to into tx(). One possible race condition remains, however: when we have hardware acceleration enabled and the driver shuts down the queues, we end up queueing the frame. If now somebody removes the key, the key will be removed from hwaccel and then then driver will be asked to encrypt the frame with a key index that has been removed. Hence, drivers will need to be aware that the hw_key_index they are passed might not be under all circumstances. Most drivers will, however, simply ignore that condition and encrypt the frame with the selected key anyway, this only results in a frame being encrypted with a wrong key or dropped (rightfully) because the key was not valid. There isn't much we can do about it unless we want to walk the pending frame queue every time a key is removed and remove all frames that used it. This race condition, however, will most likely be solved once we add multiqueue support to mac80211 because then frames will be queued further up the stack instead of after being processed. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Acked-by: Michael Wu <flamingice@sourmilk.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-14 23:10:24 +08:00
tx->key = key;
else if (ieee80211_is_group_privacy_action(tx->skb) &&
(key = rcu_dereference(tx->sdata->deflink.default_multicast_key)))
tx->key = key;
else if (ieee80211_is_mgmt(hdr->frame_control) &&
is_multicast_ether_addr(hdr->addr1) &&
ieee80211_is_robust_mgmt_frame(tx->skb) &&
(key = rcu_dereference(tx->sdata->deflink.default_mgmt_key)))
tx->key = key;
else if (is_multicast_ether_addr(hdr->addr1) &&
(key = rcu_dereference(tx->sdata->deflink.default_multicast_key)))
tx->key = key;
else if (!is_multicast_ether_addr(hdr->addr1) &&
(key = rcu_dereference(tx->sdata->default_unicast_key)))
[MAC80211]: fix race conditions with keys During receive processing, we select the key long before using it and because there's no locking it is possible that we kfree() the key after having selected it but before using it for crypto operations. Obviously, this is bad. Secondly, during transmit processing, there are two possible races: We have a similar race between select_key() and using it for encryption, but we also have a race here between select_key() and hardware encryption (both when a key is removed.) This patch solves these issues by using RCU: when a key is to be freed, we first remove the pointer from the appropriate places (sdata->keys, sdata->default_key, sta->key) using rcu_assign_pointer() and then synchronize_rcu(). Then, we can safely kfree() the key and remove it from the hardware. There's a window here where the hardware may still be using it for decryption, but we can't work around that without having two hardware callbacks, one to disable the key for RX and one to disable it for TX; but the worst thing that will happen is that we receive a packet decrypted that we don't find a key for any more and then drop it. When we add a key, we first need to upload it to the hardware and then, using rcu_assign_pointer() again, link it into our structures. In the code using keys (TX/RX paths) we use rcu_dereference() to get the key and enclose the whole tx/rx section in a rcu_read_lock() ... rcu_read_unlock() block. Because we've uploaded the key to hardware before linking it into internal structures, we can guarantee that it is valid once get to into tx(). One possible race condition remains, however: when we have hardware acceleration enabled and the driver shuts down the queues, we end up queueing the frame. If now somebody removes the key, the key will be removed from hwaccel and then then driver will be asked to encrypt the frame with a key index that has been removed. Hence, drivers will need to be aware that the hw_key_index they are passed might not be under all circumstances. Most drivers will, however, simply ignore that condition and encrypt the frame with the selected key anyway, this only results in a frame being encrypted with a wrong key or dropped (rightfully) because the key was not valid. There isn't much we can do about it unless we want to walk the pending frame queue every time a key is removed and remove all frames that used it. This race condition, however, will most likely be solved once we add multiqueue support to mac80211 because then frames will be queued further up the stack instead of after being processed. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Acked-by: Michael Wu <flamingice@sourmilk.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-09-14 23:10:24 +08:00
tx->key = key;
else
tx->key = NULL;
if (tx->key) {
bool skip_hw = false;
/* TODO: add threshold stuff again */
switch (tx->key->conf.cipher) {
case WLAN_CIPHER_SUITE_WEP40:
case WLAN_CIPHER_SUITE_WEP104:
case WLAN_CIPHER_SUITE_TKIP:
if (!ieee80211_is_data_present(hdr->frame_control))
tx->key = NULL;
break;
case WLAN_CIPHER_SUITE_CCMP:
case WLAN_CIPHER_SUITE_CCMP_256:
case WLAN_CIPHER_SUITE_GCMP:
case WLAN_CIPHER_SUITE_GCMP_256:
if (!ieee80211_is_data_present(hdr->frame_control) &&
!ieee80211_use_mfp(hdr->frame_control, tx->sta,
tx->skb) &&
!ieee80211_is_group_privacy_action(tx->skb))
tx->key = NULL;
else
skip_hw = (tx->key->conf.flags &
IEEE80211_KEY_FLAG_SW_MGMT_TX) &&
ieee80211_is_mgmt(hdr->frame_control);
break;
case WLAN_CIPHER_SUITE_AES_CMAC:
case WLAN_CIPHER_SUITE_BIP_CMAC_256:
case WLAN_CIPHER_SUITE_BIP_GMAC_128:
case WLAN_CIPHER_SUITE_BIP_GMAC_256:
if (!ieee80211_is_mgmt(hdr->frame_control))
tx->key = NULL;
break;
}
if (unlikely(tx->key && tx->key->flags & KEY_FLAG_TAINTED &&
!ieee80211_is_deauth(hdr->frame_control)))
return TX_DROP;
if (!skip_hw && tx->key &&
tx->key->flags & KEY_FLAG_UPLOADED_TO_HARDWARE)
info->control.hw_key = &tx->key->conf;
} else if (ieee80211_is_data_present(hdr->frame_control) && tx->sta &&
test_sta_flag(tx->sta, WLAN_STA_USES_ENCRYPTION)) {
return TX_DROP;
}
return TX_CONTINUE;
}
static ieee80211_tx_result debug_noinline
ieee80211_tx_h_rate_ctrl(struct ieee80211_tx_data *tx)
{
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(tx->skb);
struct ieee80211_hdr *hdr = (void *)tx->skb->data;
struct ieee80211_supported_band *sband;
u32 len;
struct ieee80211_tx_rate_control txrc;
struct ieee80211_sta_rates *ratetbl = NULL;
bool encap = info->flags & IEEE80211_TX_CTL_HW_80211_ENCAP;
bool assoc = false;
memset(&txrc, 0, sizeof(txrc));
sband = tx->local->hw.wiphy->bands[info->band];
len = min_t(u32, tx->skb->len + FCS_LEN,
tx->local->hw.wiphy->frag_threshold);
/* set up the tx rate control struct we give the RC algo */
txrc.hw = &tx->local->hw;
txrc.sband = sband;
txrc.bss_conf = &tx->sdata->vif.bss_conf;
txrc.skb = tx->skb;
txrc.reported_rate.idx = -1;
txrc.rate_idx_mask = tx->sdata->rc_rateidx_mask[info->band];
if (tx->sdata->rc_has_mcs_mask[info->band])
txrc.rate_idx_mcs_mask =
tx->sdata->rc_rateidx_mcs_mask[info->band];
txrc.bss = (tx->sdata->vif.type == NL80211_IFTYPE_AP ||
tx->sdata->vif.type == NL80211_IFTYPE_MESH_POINT ||
tx->sdata->vif.type == NL80211_IFTYPE_ADHOC ||
tx->sdata->vif.type == NL80211_IFTYPE_OCB);
/* set up RTS protection if desired */
if (len > tx->local->hw.wiphy->rts_threshold) {
txrc.rts = true;
}
info->control.use_rts = txrc.rts;
info->control.use_cts_prot = tx->sdata->vif.bss_conf.use_cts_prot;
/*
* Use short preamble if the BSS can handle it, but not for
* management frames unless we know the receiver can handle
* that -- the management frame might be to a station that
* just wants a probe response.
*/
if (tx->sdata->vif.bss_conf.use_short_preamble &&
(ieee80211_is_tx_data(tx->skb) ||
(tx->sta && test_sta_flag(tx->sta, WLAN_STA_SHORT_PREAMBLE))))
txrc.short_preamble = true;
info->control.short_preamble = txrc.short_preamble;
/* don't ask rate control when rate already injected via radiotap */
if (info->control.flags & IEEE80211_TX_CTRL_RATE_INJECT)
return TX_CONTINUE;
if (tx->sta)
assoc = test_sta_flag(tx->sta, WLAN_STA_ASSOC);
/*
* Lets not bother rate control if we're associated and cannot
* talk to the sta. This should not happen.
*/
if (WARN(test_bit(SCAN_SW_SCANNING, &tx->local->scanning) && assoc &&
!rate_usable_index_exists(sband, &tx->sta->sta),
"%s: Dropped data frame as no usable bitrate found while "
"scanning and associated. Target station: "
"%pM on %d GHz band\n",
tx->sdata->name,
encap ? ((struct ethhdr *)hdr)->h_dest : hdr->addr1,
info->band ? 5 : 2))
return TX_DROP;
/*
* If we're associated with the sta at this point we know we can at
* least send the frame at the lowest bit rate.
*/
rate_control_get_rate(tx->sdata, tx->sta, &txrc);
if (tx->sta && !info->control.skip_table)
ratetbl = rcu_dereference(tx->sta->sta.rates);
if (unlikely(info->control.rates[0].idx < 0)) {
if (ratetbl) {
struct ieee80211_tx_rate rate = {
.idx = ratetbl->rate[0].idx,
.flags = ratetbl->rate[0].flags,
.count = ratetbl->rate[0].count
};
if (ratetbl->rate[0].idx < 0)
return TX_DROP;
tx->rate = rate;
} else {
return TX_DROP;
}
} else {
tx->rate = info->control.rates[0];
}
if (txrc.reported_rate.idx < 0) {
txrc.reported_rate = tx->rate;
if (tx->sta && ieee80211_is_tx_data(tx->skb))
mac80211: prepare sta handling for MLO support Currently in mac80211 each STA object is represented using sta_info datastructure with the associated STA specific information and drivers access ieee80211_sta part of it. With MLO (Multi Link Operation) support being added in 802.11be standard, though the association is logically with a single Multi Link capable STA, at the physical level communication can happen via different advertised links (uniquely identified by Channel, operating class, BSSID) and hence the need to handle multiple link STA parameters within a composite sta_info object called the MLD STA. The different link STA part of MLD STA are identified using the link address which can be same or different as the MLD STA address and unique link id based on the link vif. To support extension of such a model, the sta_info datastructure is modified to hold multiple link STA objects with link specific params currently within sta_info moved to this new structure. Similarly this is done for ieee80211_sta as well which will be accessed within mac80211 as well as by drivers, hence trivial driver changes are expected to support this. For current non MLO supported drivers, only one link STA is present and link information is accessed via 'deflink' member. For MLO drivers, we still need to define the APIs etc. to get the correct link ID and access the correct part of the station info. Currently in mac80211, all link STA info are accessed directly via deflink. These will be updated to access via link pointers indexed by link id with MLO support patches, with link id being 0 for non MLO supported cases. Except for couple of macro related changes, below spatch takes care of updating mac80211 and driver code to access to the link STA info via deflink. @ieee80211_sta@ struct ieee80211_sta *s; struct sta_info *si; identifier var = {supp_rates, ht_cap, vht_cap, he_cap, he_6ghz_capa, eht_cap, rx_nss, bandwidth, txpwr}; @@ ( s-> - var + deflink.var | si->sta. - var + deflink.var ) @sta_info@ struct sta_info *si; identifier var = {gtk, pcpu_rx_stats, rx_stats, rx_stats_avg, status_stats, tx_stats, cur_max_bandwidth}; @@ ( si-> - var + deflink.var ) Signed-off-by: Sriram R <quic_srirrama@quicinc.com> Link: https://lore.kernel.org/r/1649086883-13246-1-git-send-email-quic_srirrama@quicinc.com [remove MLO-drivers notes from commit message, not clear yet; run spatch] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2022-04-04 23:41:23 +08:00
tx->sta->deflink.tx_stats.last_rate = txrc.reported_rate;
} else if (tx->sta)
mac80211: prepare sta handling for MLO support Currently in mac80211 each STA object is represented using sta_info datastructure with the associated STA specific information and drivers access ieee80211_sta part of it. With MLO (Multi Link Operation) support being added in 802.11be standard, though the association is logically with a single Multi Link capable STA, at the physical level communication can happen via different advertised links (uniquely identified by Channel, operating class, BSSID) and hence the need to handle multiple link STA parameters within a composite sta_info object called the MLD STA. The different link STA part of MLD STA are identified using the link address which can be same or different as the MLD STA address and unique link id based on the link vif. To support extension of such a model, the sta_info datastructure is modified to hold multiple link STA objects with link specific params currently within sta_info moved to this new structure. Similarly this is done for ieee80211_sta as well which will be accessed within mac80211 as well as by drivers, hence trivial driver changes are expected to support this. For current non MLO supported drivers, only one link STA is present and link information is accessed via 'deflink' member. For MLO drivers, we still need to define the APIs etc. to get the correct link ID and access the correct part of the station info. Currently in mac80211, all link STA info are accessed directly via deflink. These will be updated to access via link pointers indexed by link id with MLO support patches, with link id being 0 for non MLO supported cases. Except for couple of macro related changes, below spatch takes care of updating mac80211 and driver code to access to the link STA info via deflink. @ieee80211_sta@ struct ieee80211_sta *s; struct sta_info *si; identifier var = {supp_rates, ht_cap, vht_cap, he_cap, he_6ghz_capa, eht_cap, rx_nss, bandwidth, txpwr}; @@ ( s-> - var + deflink.var | si->sta. - var + deflink.var ) @sta_info@ struct sta_info *si; identifier var = {gtk, pcpu_rx_stats, rx_stats, rx_stats_avg, status_stats, tx_stats, cur_max_bandwidth}; @@ ( si-> - var + deflink.var ) Signed-off-by: Sriram R <quic_srirrama@quicinc.com> Link: https://lore.kernel.org/r/1649086883-13246-1-git-send-email-quic_srirrama@quicinc.com [remove MLO-drivers notes from commit message, not clear yet; run spatch] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2022-04-04 23:41:23 +08:00
tx->sta->deflink.tx_stats.last_rate = txrc.reported_rate;
if (ratetbl)
return TX_CONTINUE;
if (unlikely(!info->control.rates[0].count))
info->control.rates[0].count = 1;
if (WARN_ON_ONCE((info->control.rates[0].count > 1) &&
(info->flags & IEEE80211_TX_CTL_NO_ACK)))
info->control.rates[0].count = 1;
return TX_CONTINUE;
}
static __le16 ieee80211_tx_next_seq(struct sta_info *sta, int tid)
{
u16 *seq = &sta->tid_seq[tid];
__le16 ret = cpu_to_le16(*seq);
/* Increase the sequence number. */
*seq = (*seq + 0x10) & IEEE80211_SCTL_SEQ;
return ret;
}
static ieee80211_tx_result debug_noinline
ieee80211_tx_h_sequence(struct ieee80211_tx_data *tx)
{
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(tx->skb);
struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)tx->skb->data;
int tid;
/*
* Packet injection may want to control the sequence
* number, if we have no matching interface then we
* neither assign one ourselves nor ask the driver to.
*/
if (unlikely(info->control.vif->type == NL80211_IFTYPE_MONITOR))
return TX_CONTINUE;
if (unlikely(ieee80211_is_ctl(hdr->frame_control)))
return TX_CONTINUE;
if (ieee80211_hdrlen(hdr->frame_control) < 24)
return TX_CONTINUE;
if (ieee80211_is_qos_nullfunc(hdr->frame_control))
return TX_CONTINUE;
if (info->control.flags & IEEE80211_TX_CTRL_NO_SEQNO)
return TX_CONTINUE;
/*
* Anything but QoS data that has a sequence number field
* (is long enough) gets a sequence number from the global
* counter. QoS data frames with a multicast destination
* also use the global counter (802.11-2012 9.3.2.10).
*/
if (!ieee80211_is_data_qos(hdr->frame_control) ||
is_multicast_ether_addr(hdr->addr1)) {
/* driver should assign sequence number */
info->flags |= IEEE80211_TX_CTL_ASSIGN_SEQ;
/* for pure STA mode without beacons, we can do it */
hdr->seq_ctrl = cpu_to_le16(tx->sdata->sequence_number);
tx->sdata->sequence_number += 0x10;
if (tx->sta)
mac80211: prepare sta handling for MLO support Currently in mac80211 each STA object is represented using sta_info datastructure with the associated STA specific information and drivers access ieee80211_sta part of it. With MLO (Multi Link Operation) support being added in 802.11be standard, though the association is logically with a single Multi Link capable STA, at the physical level communication can happen via different advertised links (uniquely identified by Channel, operating class, BSSID) and hence the need to handle multiple link STA parameters within a composite sta_info object called the MLD STA. The different link STA part of MLD STA are identified using the link address which can be same or different as the MLD STA address and unique link id based on the link vif. To support extension of such a model, the sta_info datastructure is modified to hold multiple link STA objects with link specific params currently within sta_info moved to this new structure. Similarly this is done for ieee80211_sta as well which will be accessed within mac80211 as well as by drivers, hence trivial driver changes are expected to support this. For current non MLO supported drivers, only one link STA is present and link information is accessed via 'deflink' member. For MLO drivers, we still need to define the APIs etc. to get the correct link ID and access the correct part of the station info. Currently in mac80211, all link STA info are accessed directly via deflink. These will be updated to access via link pointers indexed by link id with MLO support patches, with link id being 0 for non MLO supported cases. Except for couple of macro related changes, below spatch takes care of updating mac80211 and driver code to access to the link STA info via deflink. @ieee80211_sta@ struct ieee80211_sta *s; struct sta_info *si; identifier var = {supp_rates, ht_cap, vht_cap, he_cap, he_6ghz_capa, eht_cap, rx_nss, bandwidth, txpwr}; @@ ( s-> - var + deflink.var | si->sta. - var + deflink.var ) @sta_info@ struct sta_info *si; identifier var = {gtk, pcpu_rx_stats, rx_stats, rx_stats_avg, status_stats, tx_stats, cur_max_bandwidth}; @@ ( si-> - var + deflink.var ) Signed-off-by: Sriram R <quic_srirrama@quicinc.com> Link: https://lore.kernel.org/r/1649086883-13246-1-git-send-email-quic_srirrama@quicinc.com [remove MLO-drivers notes from commit message, not clear yet; run spatch] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2022-04-04 23:41:23 +08:00
tx->sta->deflink.tx_stats.msdu[IEEE80211_NUM_TIDS]++;
return TX_CONTINUE;
}
/*
* This should be true for injected/management frames only, for
* management frames we have set the IEEE80211_TX_CTL_ASSIGN_SEQ
* above since they are not QoS-data frames.
*/
if (!tx->sta)
return TX_CONTINUE;
/* include per-STA, per-TID sequence counter */
tid = ieee80211_get_tid(hdr);
mac80211: prepare sta handling for MLO support Currently in mac80211 each STA object is represented using sta_info datastructure with the associated STA specific information and drivers access ieee80211_sta part of it. With MLO (Multi Link Operation) support being added in 802.11be standard, though the association is logically with a single Multi Link capable STA, at the physical level communication can happen via different advertised links (uniquely identified by Channel, operating class, BSSID) and hence the need to handle multiple link STA parameters within a composite sta_info object called the MLD STA. The different link STA part of MLD STA are identified using the link address which can be same or different as the MLD STA address and unique link id based on the link vif. To support extension of such a model, the sta_info datastructure is modified to hold multiple link STA objects with link specific params currently within sta_info moved to this new structure. Similarly this is done for ieee80211_sta as well which will be accessed within mac80211 as well as by drivers, hence trivial driver changes are expected to support this. For current non MLO supported drivers, only one link STA is present and link information is accessed via 'deflink' member. For MLO drivers, we still need to define the APIs etc. to get the correct link ID and access the correct part of the station info. Currently in mac80211, all link STA info are accessed directly via deflink. These will be updated to access via link pointers indexed by link id with MLO support patches, with link id being 0 for non MLO supported cases. Except for couple of macro related changes, below spatch takes care of updating mac80211 and driver code to access to the link STA info via deflink. @ieee80211_sta@ struct ieee80211_sta *s; struct sta_info *si; identifier var = {supp_rates, ht_cap, vht_cap, he_cap, he_6ghz_capa, eht_cap, rx_nss, bandwidth, txpwr}; @@ ( s-> - var + deflink.var | si->sta. - var + deflink.var ) @sta_info@ struct sta_info *si; identifier var = {gtk, pcpu_rx_stats, rx_stats, rx_stats_avg, status_stats, tx_stats, cur_max_bandwidth}; @@ ( si-> - var + deflink.var ) Signed-off-by: Sriram R <quic_srirrama@quicinc.com> Link: https://lore.kernel.org/r/1649086883-13246-1-git-send-email-quic_srirrama@quicinc.com [remove MLO-drivers notes from commit message, not clear yet; run spatch] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2022-04-04 23:41:23 +08:00
tx->sta->deflink.tx_stats.msdu[tid]++;
hdr->seq_ctrl = ieee80211_tx_next_seq(tx->sta, tid);
return TX_CONTINUE;
}
static int ieee80211_fragment(struct ieee80211_tx_data *tx,
struct sk_buff *skb, int hdrlen,
int frag_threshold)
{
struct ieee80211_local *local = tx->local;
struct ieee80211_tx_info *info;
struct sk_buff *tmp;
int per_fragm = frag_threshold - hdrlen - FCS_LEN;
int pos = hdrlen + per_fragm;
int rem = skb->len - hdrlen - per_fragm;
if (WARN_ON(rem < 0))
return -EINVAL;
/* first fragment was already added to queue by caller */
while (rem) {
int fraglen = per_fragm;
if (fraglen > rem)
fraglen = rem;
rem -= fraglen;
tmp = dev_alloc_skb(local->tx_headroom +
frag_threshold +
IEEE80211_ENCRYPT_HEADROOM +
IEEE80211_ENCRYPT_TAILROOM);
if (!tmp)
return -ENOMEM;
__skb_queue_tail(&tx->skbs, tmp);
skb_reserve(tmp,
local->tx_headroom + IEEE80211_ENCRYPT_HEADROOM);
/* copy control information */
memcpy(tmp->cb, skb->cb, sizeof(tmp->cb));
info = IEEE80211_SKB_CB(tmp);
info->flags &= ~(IEEE80211_TX_CTL_CLEAR_PS_FILT |
IEEE80211_TX_CTL_FIRST_FRAGMENT);
if (rem)
info->flags |= IEEE80211_TX_CTL_MORE_FRAMES;
skb_copy_queue_mapping(tmp, skb);
tmp->priority = skb->priority;
tmp->dev = skb->dev;
/* copy header and data */
skb_put_data(tmp, skb->data, hdrlen);
skb_put_data(tmp, skb->data + pos, fraglen);
pos += fraglen;
}
/* adjust first fragment's length */
skb_trim(skb, hdrlen + per_fragm);
return 0;
}
static ieee80211_tx_result debug_noinline
ieee80211_tx_h_fragment(struct ieee80211_tx_data *tx)
{
struct sk_buff *skb = tx->skb;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
struct ieee80211_hdr *hdr = (void *)skb->data;
int frag_threshold = tx->local->hw.wiphy->frag_threshold;
int hdrlen;
int fragnum;
/* no matter what happens, tx->skb moves to tx->skbs */
__skb_queue_tail(&tx->skbs, skb);
tx->skb = NULL;
if (info->flags & IEEE80211_TX_CTL_DONTFRAG)
return TX_CONTINUE;
if (ieee80211_hw_check(&tx->local->hw, SUPPORTS_TX_FRAG))
return TX_CONTINUE;
/*
* Warn when submitting a fragmented A-MPDU frame and drop it.
* This scenario is handled in ieee80211_tx_prepare but extra
* caution taken here as fragmented ampdu may cause Tx stop.
*/
if (WARN_ON(info->flags & IEEE80211_TX_CTL_AMPDU))
return TX_DROP;
hdrlen = ieee80211_hdrlen(hdr->frame_control);
/* internal error, why isn't DONTFRAG set? */
if (WARN_ON(skb->len + FCS_LEN <= frag_threshold))
return TX_DROP;
/*
* Now fragment the frame. This will allocate all the fragments and
* chain them (using skb as the first fragment) to skb->next.
* During transmission, we will remove the successfully transmitted
* fragments from this list. When the low-level driver rejects one
* of the fragments then we will simply pretend to accept the skb
* but store it away as pending.
*/
if (ieee80211_fragment(tx, skb, hdrlen, frag_threshold))
return TX_DROP;
/* update duration/seq/flags of fragments */
fragnum = 0;
skb_queue_walk(&tx->skbs, skb) {
const __le16 morefrags = cpu_to_le16(IEEE80211_FCTL_MOREFRAGS);
hdr = (void *)skb->data;
info = IEEE80211_SKB_CB(skb);
if (!skb_queue_is_last(&tx->skbs, skb)) {
hdr->frame_control |= morefrags;
/*
* No multi-rate retries for fragmented frames, that
* would completely throw off the NAV at other STAs.
*/
info->control.rates[1].idx = -1;
info->control.rates[2].idx = -1;
info->control.rates[3].idx = -1;
BUILD_BUG_ON(IEEE80211_TX_MAX_RATES != 4);
info->flags &= ~IEEE80211_TX_CTL_RATE_CTRL_PROBE;
} else {
hdr->frame_control &= ~morefrags;
}
hdr->seq_ctrl |= cpu_to_le16(fragnum & IEEE80211_SCTL_FRAG);
fragnum++;
}
return TX_CONTINUE;
}
static ieee80211_tx_result debug_noinline
ieee80211_tx_h_stats(struct ieee80211_tx_data *tx)
{
struct sk_buff *skb;
int ac = -1;
if (!tx->sta)
return TX_CONTINUE;
skb_queue_walk(&tx->skbs, skb) {
ac = skb_get_queue_mapping(skb);
mac80211: prepare sta handling for MLO support Currently in mac80211 each STA object is represented using sta_info datastructure with the associated STA specific information and drivers access ieee80211_sta part of it. With MLO (Multi Link Operation) support being added in 802.11be standard, though the association is logically with a single Multi Link capable STA, at the physical level communication can happen via different advertised links (uniquely identified by Channel, operating class, BSSID) and hence the need to handle multiple link STA parameters within a composite sta_info object called the MLD STA. The different link STA part of MLD STA are identified using the link address which can be same or different as the MLD STA address and unique link id based on the link vif. To support extension of such a model, the sta_info datastructure is modified to hold multiple link STA objects with link specific params currently within sta_info moved to this new structure. Similarly this is done for ieee80211_sta as well which will be accessed within mac80211 as well as by drivers, hence trivial driver changes are expected to support this. For current non MLO supported drivers, only one link STA is present and link information is accessed via 'deflink' member. For MLO drivers, we still need to define the APIs etc. to get the correct link ID and access the correct part of the station info. Currently in mac80211, all link STA info are accessed directly via deflink. These will be updated to access via link pointers indexed by link id with MLO support patches, with link id being 0 for non MLO supported cases. Except for couple of macro related changes, below spatch takes care of updating mac80211 and driver code to access to the link STA info via deflink. @ieee80211_sta@ struct ieee80211_sta *s; struct sta_info *si; identifier var = {supp_rates, ht_cap, vht_cap, he_cap, he_6ghz_capa, eht_cap, rx_nss, bandwidth, txpwr}; @@ ( s-> - var + deflink.var | si->sta. - var + deflink.var ) @sta_info@ struct sta_info *si; identifier var = {gtk, pcpu_rx_stats, rx_stats, rx_stats_avg, status_stats, tx_stats, cur_max_bandwidth}; @@ ( si-> - var + deflink.var ) Signed-off-by: Sriram R <quic_srirrama@quicinc.com> Link: https://lore.kernel.org/r/1649086883-13246-1-git-send-email-quic_srirrama@quicinc.com [remove MLO-drivers notes from commit message, not clear yet; run spatch] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2022-04-04 23:41:23 +08:00
tx->sta->deflink.tx_stats.bytes[ac] += skb->len;
}
if (ac >= 0)
mac80211: prepare sta handling for MLO support Currently in mac80211 each STA object is represented using sta_info datastructure with the associated STA specific information and drivers access ieee80211_sta part of it. With MLO (Multi Link Operation) support being added in 802.11be standard, though the association is logically with a single Multi Link capable STA, at the physical level communication can happen via different advertised links (uniquely identified by Channel, operating class, BSSID) and hence the need to handle multiple link STA parameters within a composite sta_info object called the MLD STA. The different link STA part of MLD STA are identified using the link address which can be same or different as the MLD STA address and unique link id based on the link vif. To support extension of such a model, the sta_info datastructure is modified to hold multiple link STA objects with link specific params currently within sta_info moved to this new structure. Similarly this is done for ieee80211_sta as well which will be accessed within mac80211 as well as by drivers, hence trivial driver changes are expected to support this. For current non MLO supported drivers, only one link STA is present and link information is accessed via 'deflink' member. For MLO drivers, we still need to define the APIs etc. to get the correct link ID and access the correct part of the station info. Currently in mac80211, all link STA info are accessed directly via deflink. These will be updated to access via link pointers indexed by link id with MLO support patches, with link id being 0 for non MLO supported cases. Except for couple of macro related changes, below spatch takes care of updating mac80211 and driver code to access to the link STA info via deflink. @ieee80211_sta@ struct ieee80211_sta *s; struct sta_info *si; identifier var = {supp_rates, ht_cap, vht_cap, he_cap, he_6ghz_capa, eht_cap, rx_nss, bandwidth, txpwr}; @@ ( s-> - var + deflink.var | si->sta. - var + deflink.var ) @sta_info@ struct sta_info *si; identifier var = {gtk, pcpu_rx_stats, rx_stats, rx_stats_avg, status_stats, tx_stats, cur_max_bandwidth}; @@ ( si-> - var + deflink.var ) Signed-off-by: Sriram R <quic_srirrama@quicinc.com> Link: https://lore.kernel.org/r/1649086883-13246-1-git-send-email-quic_srirrama@quicinc.com [remove MLO-drivers notes from commit message, not clear yet; run spatch] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2022-04-04 23:41:23 +08:00
tx->sta->deflink.tx_stats.packets[ac]++;
return TX_CONTINUE;
}
static ieee80211_tx_result debug_noinline
ieee80211_tx_h_encrypt(struct ieee80211_tx_data *tx)
{
if (!tx->key)
return TX_CONTINUE;
switch (tx->key->conf.cipher) {
case WLAN_CIPHER_SUITE_WEP40:
case WLAN_CIPHER_SUITE_WEP104:
return ieee80211_crypto_wep_encrypt(tx);
case WLAN_CIPHER_SUITE_TKIP:
return ieee80211_crypto_tkip_encrypt(tx);
case WLAN_CIPHER_SUITE_CCMP:
return ieee80211_crypto_ccmp_encrypt(
tx, IEEE80211_CCMP_MIC_LEN);
case WLAN_CIPHER_SUITE_CCMP_256:
return ieee80211_crypto_ccmp_encrypt(
tx, IEEE80211_CCMP_256_MIC_LEN);
case WLAN_CIPHER_SUITE_AES_CMAC:
return ieee80211_crypto_aes_cmac_encrypt(tx);
case WLAN_CIPHER_SUITE_BIP_CMAC_256:
return ieee80211_crypto_aes_cmac_256_encrypt(tx);
case WLAN_CIPHER_SUITE_BIP_GMAC_128:
case WLAN_CIPHER_SUITE_BIP_GMAC_256:
return ieee80211_crypto_aes_gmac_encrypt(tx);
case WLAN_CIPHER_SUITE_GCMP:
case WLAN_CIPHER_SUITE_GCMP_256:
return ieee80211_crypto_gcmp_encrypt(tx);
}
return TX_DROP;
}
static ieee80211_tx_result debug_noinline
ieee80211_tx_h_calculate_duration(struct ieee80211_tx_data *tx)
{
struct sk_buff *skb;
struct ieee80211_hdr *hdr;
int next_len;
bool group_addr;
skb_queue_walk(&tx->skbs, skb) {
hdr = (void *) skb->data;
if (unlikely(ieee80211_is_pspoll(hdr->frame_control)))
break; /* must not overwrite AID */
if (!skb_queue_is_last(&tx->skbs, skb)) {
struct sk_buff *next = skb_queue_next(&tx->skbs, skb);
next_len = next->len;
} else
next_len = 0;
group_addr = is_multicast_ether_addr(hdr->addr1);
hdr->duration_id =
ieee80211_duration(tx, skb, group_addr, next_len);
}
return TX_CONTINUE;
}
/* actual transmit path */
static bool ieee80211_tx_prep_agg(struct ieee80211_tx_data *tx,
struct sk_buff *skb,
struct ieee80211_tx_info *info,
struct tid_ampdu_tx *tid_tx,
int tid)
{
bool queued = false;
bool reset_agg_timer = false;
struct sk_buff *purge_skb = NULL;
if (test_bit(HT_AGG_STATE_OPERATIONAL, &tid_tx->state)) {
info->flags |= IEEE80211_TX_CTL_AMPDU;
reset_agg_timer = true;
} else if (test_bit(HT_AGG_STATE_WANT_START, &tid_tx->state)) {
/*
* nothing -- this aggregation session is being started
* but that might still fail with the driver
*/
} else if (!tx->sta->sta.txq[tid]) {
spin_lock(&tx->sta->lock);
/*
* Need to re-check now, because we may get here
*
* 1) in the window during which the setup is actually
* already done, but not marked yet because not all
* packets are spliced over to the driver pending
* queue yet -- if this happened we acquire the lock
* either before or after the splice happens, but
* need to recheck which of these cases happened.
*
* 2) during session teardown, if the OPERATIONAL bit
* was cleared due to the teardown but the pointer
* hasn't been assigned NULL yet (or we loaded it
* before it was assigned) -- in this case it may
* now be NULL which means we should just let the
* packet pass through because splicing the frames
* back is already done.
*/
tid_tx = rcu_dereference_protected_tid_tx(tx->sta, tid);
if (!tid_tx) {
/* do nothing, let packet pass through */
} else if (test_bit(HT_AGG_STATE_OPERATIONAL, &tid_tx->state)) {
info->flags |= IEEE80211_TX_CTL_AMPDU;
reset_agg_timer = true;
} else {
queued = true;
if (info->flags & IEEE80211_TX_CTL_NO_PS_BUFFER) {
clear_sta_flag(tx->sta, WLAN_STA_SP);
ps_dbg(tx->sta->sdata,
"STA %pM aid %d: SP frame queued, close the SP w/o telling the peer\n",
tx->sta->sta.addr, tx->sta->sta.aid);
}
info->control.vif = &tx->sdata->vif;
info->control.flags |= IEEE80211_TX_INTCFL_NEED_TXPROCESSING;
info->flags &= ~IEEE80211_TX_TEMPORARY_FLAGS;
__skb_queue_tail(&tid_tx->pending, skb);
if (skb_queue_len(&tid_tx->pending) > STA_MAX_TX_BUFFER)
purge_skb = __skb_dequeue(&tid_tx->pending);
}
spin_unlock(&tx->sta->lock);
if (purge_skb)
ieee80211_free_txskb(&tx->local->hw, purge_skb);
}
/* reset session timer */
if (reset_agg_timer)
tid_tx->last_tx = jiffies;
return queued;
}
static void
ieee80211_aggr_check(struct ieee80211_sub_if_data *sdata,
struct sta_info *sta,
struct sk_buff *skb)
{
struct rate_control_ref *ref = sdata->local->rate_ctrl;
u16 tid;
if (!ref || !(ref->ops->capa & RATE_CTRL_CAPA_AMPDU_TRIGGER))
return;
mac80211: prepare sta handling for MLO support Currently in mac80211 each STA object is represented using sta_info datastructure with the associated STA specific information and drivers access ieee80211_sta part of it. With MLO (Multi Link Operation) support being added in 802.11be standard, though the association is logically with a single Multi Link capable STA, at the physical level communication can happen via different advertised links (uniquely identified by Channel, operating class, BSSID) and hence the need to handle multiple link STA parameters within a composite sta_info object called the MLD STA. The different link STA part of MLD STA are identified using the link address which can be same or different as the MLD STA address and unique link id based on the link vif. To support extension of such a model, the sta_info datastructure is modified to hold multiple link STA objects with link specific params currently within sta_info moved to this new structure. Similarly this is done for ieee80211_sta as well which will be accessed within mac80211 as well as by drivers, hence trivial driver changes are expected to support this. For current non MLO supported drivers, only one link STA is present and link information is accessed via 'deflink' member. For MLO drivers, we still need to define the APIs etc. to get the correct link ID and access the correct part of the station info. Currently in mac80211, all link STA info are accessed directly via deflink. These will be updated to access via link pointers indexed by link id with MLO support patches, with link id being 0 for non MLO supported cases. Except for couple of macro related changes, below spatch takes care of updating mac80211 and driver code to access to the link STA info via deflink. @ieee80211_sta@ struct ieee80211_sta *s; struct sta_info *si; identifier var = {supp_rates, ht_cap, vht_cap, he_cap, he_6ghz_capa, eht_cap, rx_nss, bandwidth, txpwr}; @@ ( s-> - var + deflink.var | si->sta. - var + deflink.var ) @sta_info@ struct sta_info *si; identifier var = {gtk, pcpu_rx_stats, rx_stats, rx_stats_avg, status_stats, tx_stats, cur_max_bandwidth}; @@ ( si-> - var + deflink.var ) Signed-off-by: Sriram R <quic_srirrama@quicinc.com> Link: https://lore.kernel.org/r/1649086883-13246-1-git-send-email-quic_srirrama@quicinc.com [remove MLO-drivers notes from commit message, not clear yet; run spatch] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2022-04-04 23:41:23 +08:00
if (!sta || !sta->sta.deflink.ht_cap.ht_supported ||
!sta->sta.wme || skb_get_queue_mapping(skb) == IEEE80211_AC_VO ||
skb->protocol == sdata->control_port_protocol)
return;
tid = skb->priority & IEEE80211_QOS_CTL_TID_MASK;
if (likely(sta->ampdu_mlme.tid_tx[tid]))
return;
ieee80211_start_tx_ba_session(&sta->sta, tid, 0);
}
/*
* initialises @tx
* pass %NULL for the station if unknown, a valid pointer if known
* or an ERR_PTR() if the station is known not to exist
*/
static ieee80211_tx_result
ieee80211_tx_prepare(struct ieee80211_sub_if_data *sdata,
struct ieee80211_tx_data *tx,
struct sta_info *sta, struct sk_buff *skb)
{
struct ieee80211_local *local = sdata->local;
struct ieee80211_hdr *hdr;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
bool aggr_check = false;
int tid;
memset(tx, 0, sizeof(*tx));
tx->skb = skb;
tx->local = local;
tx->sdata = sdata;
__skb_queue_head_init(&tx->skbs);
/*
* If this flag is set to true anywhere, and we get here,
* we are doing the needed processing, so remove the flag
* now.
*/
info->control.flags &= ~IEEE80211_TX_INTCFL_NEED_TXPROCESSING;
hdr = (struct ieee80211_hdr *) skb->data;
if (likely(sta)) {
if (!IS_ERR(sta))
tx->sta = sta;
} else {
if (sdata->vif.type == NL80211_IFTYPE_AP_VLAN) {
tx->sta = rcu_dereference(sdata->u.vlan.sta);
if (!tx->sta && sdata->wdev.use_4addr)
return TX_DROP;
} else if (tx->sdata->control_port_protocol == tx->skb->protocol) {
tx->sta = sta_info_get_bss(sdata, hdr->addr1);
}
if (!tx->sta && !is_multicast_ether_addr(hdr->addr1)) {
tx->sta = sta_info_get(sdata, hdr->addr1);
aggr_check = true;
}
}
if (tx->sta && ieee80211_is_data_qos(hdr->frame_control) &&
!ieee80211_is_qos_nullfunc(hdr->frame_control) &&
ieee80211_hw_check(&local->hw, AMPDU_AGGREGATION) &&
!ieee80211_hw_check(&local->hw, TX_AMPDU_SETUP_IN_HW)) {
struct tid_ampdu_tx *tid_tx;
tid = ieee80211_get_tid(hdr);
tid_tx = rcu_dereference(tx->sta->ampdu_mlme.tid_tx[tid]);
if (!tid_tx && aggr_check) {
ieee80211_aggr_check(sdata, tx->sta, skb);
tid_tx = rcu_dereference(tx->sta->ampdu_mlme.tid_tx[tid]);
}
if (tid_tx) {
bool queued;
queued = ieee80211_tx_prep_agg(tx, skb, info,
tid_tx, tid);
if (unlikely(queued))
return TX_QUEUED;
}
}
if (is_multicast_ether_addr(hdr->addr1)) {
tx->flags &= ~IEEE80211_TX_UNICAST;
info->flags |= IEEE80211_TX_CTL_NO_ACK;
} else
tx->flags |= IEEE80211_TX_UNICAST;
if (!(info->flags & IEEE80211_TX_CTL_DONTFRAG)) {
if (!(tx->flags & IEEE80211_TX_UNICAST) ||
skb->len + FCS_LEN <= local->hw.wiphy->frag_threshold ||
info->flags & IEEE80211_TX_CTL_AMPDU)
info->flags |= IEEE80211_TX_CTL_DONTFRAG;
}
if (!tx->sta)
info->flags |= IEEE80211_TX_CTL_CLEAR_PS_FILT;
else if (test_and_clear_sta_flag(tx->sta, WLAN_STA_CLEAR_PS_FILT)) {
info->flags |= IEEE80211_TX_CTL_CLEAR_PS_FILT;
ieee80211_check_fast_xmit(tx->sta);
}
info->flags |= IEEE80211_TX_CTL_FIRST_FRAGMENT;
return TX_CONTINUE;
}
static struct txq_info *ieee80211_get_txq(struct ieee80211_local *local,
struct ieee80211_vif *vif,
mac80211: prevent skb/txq mismatch Station structure is considered as not uploaded (to driver) until drv_sta_state() finishes. This call is however done after the structure is attached to mac80211 internal lists and hashes. This means mac80211 can lookup (and use) station structure before it is uploaded to a driver. If this happens (structure exists, but sta->uploaded is false) fast_tx path can still be taken. Deep in the fastpath call the sta->uploaded is checked against to derive "pubsta" argument for ieee80211_get_txq(). If sta->uploaded is false (and sta is actually non-NULL) ieee80211_get_txq() effectively downgraded to vif->txq. At first glance this may look innocent but coerces mac80211 into a state that is almost guaranteed (codel may drop offending skb) to crash because a station-oriented skb gets queued up on vif-oriented txq. The ieee80211_tx_dequeue() ends up looking at info->control.flags and tries to use txq->sta which in the fail case is NULL. It's probably pointless to pretend one can downgrade skb from sta-txq to vif-txq. Since downgrading unicast traffic to vif->txq must not be done there's no txq to put a frame on if sta->uploaded is false. Therefore the code is made to fall back to regular tx() op path if the described condition is hit. Only drivers using wake_tx_queue were affected. Example crash dump before fix: Unable to handle kernel paging request at virtual address ffffe26c PC is at ieee80211_tx_dequeue+0x204/0x690 [mac80211] [<bf4252a4>] (ieee80211_tx_dequeue [mac80211]) from [<bf4b1388>] (ath10k_mac_tx_push_txq+0x54/0x1c0 [ath10k_core]) [<bf4b1388>] (ath10k_mac_tx_push_txq [ath10k_core]) from [<bf4bdfbc>] (ath10k_htt_txrx_compl_task+0xd78/0x11d0 [ath10k_core]) [<bf4bdfbc>] (ath10k_htt_txrx_compl_task [ath10k_core]) [<bf51c5a4>] (ath10k_pci_napi_poll+0x54/0xe8 [ath10k_pci]) [<bf51c5a4>] (ath10k_pci_napi_poll [ath10k_pci]) from [<c0572e90>] (net_rx_action+0xac/0x160) Reported-by: Mohammed Shafi Shajakhan <mohammed@qti.qualcomm.com> Signed-off-by: Michal Kazior <michal.kazior@tieto.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-01-13 20:32:51 +08:00
struct sta_info *sta,
struct sk_buff *skb)
{
struct ieee80211_hdr *hdr = (struct ieee80211_hdr *) skb->data;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
struct ieee80211_txq *txq = NULL;
if ((info->flags & IEEE80211_TX_CTL_SEND_AFTER_DTIM) ||
(info->control.flags & IEEE80211_TX_CTRL_PS_RESPONSE))
return NULL;
if (!(info->flags & IEEE80211_TX_CTL_HW_80211_ENCAP) &&
unlikely(!ieee80211_is_data_present(hdr->frame_control))) {
if ((!ieee80211_is_mgmt(hdr->frame_control) ||
ieee80211_is_bufferable_mmpdu(hdr->frame_control) ||
vif->type == NL80211_IFTYPE_STATION) &&
sta && sta->uploaded) {
/*
* This will be NULL if the driver didn't set the
* opt-in hardware flag.
*/
txq = sta->sta.txq[IEEE80211_NUM_TIDS];
}
} else if (sta) {
u8 tid = skb->priority & IEEE80211_QOS_CTL_TID_MASK;
mac80211: prevent skb/txq mismatch Station structure is considered as not uploaded (to driver) until drv_sta_state() finishes. This call is however done after the structure is attached to mac80211 internal lists and hashes. This means mac80211 can lookup (and use) station structure before it is uploaded to a driver. If this happens (structure exists, but sta->uploaded is false) fast_tx path can still be taken. Deep in the fastpath call the sta->uploaded is checked against to derive "pubsta" argument for ieee80211_get_txq(). If sta->uploaded is false (and sta is actually non-NULL) ieee80211_get_txq() effectively downgraded to vif->txq. At first glance this may look innocent but coerces mac80211 into a state that is almost guaranteed (codel may drop offending skb) to crash because a station-oriented skb gets queued up on vif-oriented txq. The ieee80211_tx_dequeue() ends up looking at info->control.flags and tries to use txq->sta which in the fail case is NULL. It's probably pointless to pretend one can downgrade skb from sta-txq to vif-txq. Since downgrading unicast traffic to vif->txq must not be done there's no txq to put a frame on if sta->uploaded is false. Therefore the code is made to fall back to regular tx() op path if the described condition is hit. Only drivers using wake_tx_queue were affected. Example crash dump before fix: Unable to handle kernel paging request at virtual address ffffe26c PC is at ieee80211_tx_dequeue+0x204/0x690 [mac80211] [<bf4252a4>] (ieee80211_tx_dequeue [mac80211]) from [<bf4b1388>] (ath10k_mac_tx_push_txq+0x54/0x1c0 [ath10k_core]) [<bf4b1388>] (ath10k_mac_tx_push_txq [ath10k_core]) from [<bf4bdfbc>] (ath10k_htt_txrx_compl_task+0xd78/0x11d0 [ath10k_core]) [<bf4bdfbc>] (ath10k_htt_txrx_compl_task [ath10k_core]) [<bf51c5a4>] (ath10k_pci_napi_poll+0x54/0xe8 [ath10k_pci]) [<bf51c5a4>] (ath10k_pci_napi_poll [ath10k_pci]) from [<c0572e90>] (net_rx_action+0xac/0x160) Reported-by: Mohammed Shafi Shajakhan <mohammed@qti.qualcomm.com> Signed-off-by: Michal Kazior <michal.kazior@tieto.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-01-13 20:32:51 +08:00
if (!sta->uploaded)
return NULL;
txq = sta->sta.txq[tid];
} else if (vif) {
txq = vif->txq;
}
if (!txq)
return NULL;
return to_txq_info(txq);
}
static void ieee80211_set_skb_enqueue_time(struct sk_buff *skb)
{
IEEE80211_SKB_CB(skb)->control.enqueue_time = codel_get_time();
}
static u32 codel_skb_len_func(const struct sk_buff *skb)
{
return skb->len;
}
static codel_time_t codel_skb_time_func(const struct sk_buff *skb)
{
const struct ieee80211_tx_info *info;
info = (const struct ieee80211_tx_info *)skb->cb;
return info->control.enqueue_time;
}
static struct sk_buff *codel_dequeue_func(struct codel_vars *cvars,
void *ctx)
{
struct ieee80211_local *local;
struct txq_info *txqi;
struct fq *fq;
struct fq_flow *flow;
txqi = ctx;
local = vif_to_sdata(txqi->txq.vif)->local;
fq = &local->fq;
if (cvars == &txqi->def_cvars)
flow = &txqi->tin.default_flow;
else
flow = &fq->flows[cvars - local->cvars];
return fq_flow_dequeue(fq, flow);
}
static void codel_drop_func(struct sk_buff *skb,
void *ctx)
{
struct ieee80211_local *local;
struct ieee80211_hw *hw;
struct txq_info *txqi;
txqi = ctx;
local = vif_to_sdata(txqi->txq.vif)->local;
hw = &local->hw;
ieee80211_free_txskb(hw, skb);
}
static struct sk_buff *fq_tin_dequeue_func(struct fq *fq,
struct fq_tin *tin,
struct fq_flow *flow)
{
struct ieee80211_local *local;
struct txq_info *txqi;
struct codel_vars *cvars;
struct codel_params *cparams;
struct codel_stats *cstats;
local = container_of(fq, struct ieee80211_local, fq);
txqi = container_of(tin, struct txq_info, tin);
cstats = &txqi->cstats;
if (txqi->txq.sta) {
struct sta_info *sta = container_of(txqi->txq.sta,
struct sta_info, sta);
cparams = &sta->cparams;
} else {
cparams = &local->cparams;
}
if (flow == &tin->default_flow)
cvars = &txqi->def_cvars;
else
cvars = &local->cvars[flow - fq->flows];
return codel_dequeue(txqi,
&flow->backlog,
cparams,
cvars,
cstats,
codel_skb_len_func,
codel_skb_time_func,
codel_drop_func,
codel_dequeue_func);
}
static void fq_skb_free_func(struct fq *fq,
struct fq_tin *tin,
struct fq_flow *flow,
struct sk_buff *skb)
{
struct ieee80211_local *local;
local = container_of(fq, struct ieee80211_local, fq);
ieee80211_free_txskb(&local->hw, skb);
}
static void ieee80211_txq_enqueue(struct ieee80211_local *local,
struct txq_info *txqi,
struct sk_buff *skb)
{
struct fq *fq = &local->fq;
struct fq_tin *tin = &txqi->tin;
u32 flow_idx = fq_flow_idx(fq, skb);
ieee80211_set_skb_enqueue_time(skb);
spin_lock_bh(&fq->lock);
/*
* For management frames, don't really apply codel etc.,
* we don't want to apply any shaping or anything we just
* want to simplify the driver API by having them on the
* txqi.
*/
if (unlikely(txqi->txq.tid == IEEE80211_NUM_TIDS)) {
IEEE80211_SKB_CB(skb)->control.flags |=
IEEE80211_TX_INTCFL_NEED_TXPROCESSING;
__skb_queue_tail(&txqi->frags, skb);
} else {
fq_tin_enqueue(fq, tin, flow_idx, skb,
fq_skb_free_func);
}
spin_unlock_bh(&fq->lock);
}
static bool fq_vlan_filter_func(struct fq *fq, struct fq_tin *tin,
struct fq_flow *flow, struct sk_buff *skb,
void *data)
{
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
return info->control.vif == data;
}
void ieee80211_txq_remove_vlan(struct ieee80211_local *local,
struct ieee80211_sub_if_data *sdata)
{
struct fq *fq = &local->fq;
struct txq_info *txqi;
struct fq_tin *tin;
struct ieee80211_sub_if_data *ap;
if (WARN_ON(sdata->vif.type != NL80211_IFTYPE_AP_VLAN))
return;
ap = container_of(sdata->bss, struct ieee80211_sub_if_data, u.ap);
if (!ap->vif.txq)
return;
txqi = to_txq_info(ap->vif.txq);
tin = &txqi->tin;
spin_lock_bh(&fq->lock);
fq_tin_filter(fq, tin, fq_vlan_filter_func, &sdata->vif,
fq_skb_free_func);
spin_unlock_bh(&fq->lock);
}
void ieee80211_txq_init(struct ieee80211_sub_if_data *sdata,
struct sta_info *sta,
struct txq_info *txqi, int tid)
{
fq_tin_init(&txqi->tin);
codel_vars_init(&txqi->def_cvars);
codel_stats_init(&txqi->cstats);
__skb_queue_head_init(&txqi->frags);
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
RB_CLEAR_NODE(&txqi->schedule_order);
txqi->txq.vif = &sdata->vif;
if (!sta) {
sdata->vif.txq = &txqi->txq;
txqi->txq.tid = 0;
txqi->txq.ac = IEEE80211_AC_BE;
return;
}
if (tid == IEEE80211_NUM_TIDS) {
if (sdata->vif.type == NL80211_IFTYPE_STATION) {
/* Drivers need to opt in to the management MPDU TXQ */
if (!ieee80211_hw_check(&sdata->local->hw,
STA_MMPDU_TXQ))
return;
} else if (!ieee80211_hw_check(&sdata->local->hw,
BUFF_MMPDU_TXQ)) {
/* Drivers need to opt in to the bufferable MMPDU TXQ */
return;
}
txqi->txq.ac = IEEE80211_AC_VO;
} else {
txqi->txq.ac = ieee80211_ac_from_tid(tid);
}
txqi->txq.sta = &sta->sta;
txqi->txq.tid = tid;
sta->sta.txq[tid] = &txqi->txq;
}
void ieee80211_txq_purge(struct ieee80211_local *local,
struct txq_info *txqi)
{
struct fq *fq = &local->fq;
struct fq_tin *tin = &txqi->tin;
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
spin_lock_bh(&fq->lock);
fq_tin_reset(fq, tin, fq_skb_free_func);
ieee80211_purge_tx_queue(&local->hw, &txqi->frags);
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
spin_unlock_bh(&fq->lock);
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
ieee80211_unschedule_txq(&local->hw, &txqi->txq, true);
}
void ieee80211_txq_set_params(struct ieee80211_local *local)
{
if (local->hw.wiphy->txq_limit)
local->fq.limit = local->hw.wiphy->txq_limit;
else
local->hw.wiphy->txq_limit = local->fq.limit;
if (local->hw.wiphy->txq_memory_limit)
local->fq.memory_limit = local->hw.wiphy->txq_memory_limit;
else
local->hw.wiphy->txq_memory_limit = local->fq.memory_limit;
if (local->hw.wiphy->txq_quantum)
local->fq.quantum = local->hw.wiphy->txq_quantum;
else
local->hw.wiphy->txq_quantum = local->fq.quantum;
}
int ieee80211_txq_setup_flows(struct ieee80211_local *local)
{
struct fq *fq = &local->fq;
int ret;
int i;
bool supp_vht = false;
enum nl80211_band band;
if (!local->ops->wake_tx_queue)
return 0;
ret = fq_init(fq, 4096);
if (ret)
return ret;
/*
* If the hardware doesn't support VHT, it is safe to limit the maximum
* queue size. 4 Mbytes is 64 max-size aggregates in 802.11n.
*/
for (band = 0; band < NUM_NL80211_BANDS; band++) {
struct ieee80211_supported_band *sband;
sband = local->hw.wiphy->bands[band];
if (!sband)
continue;
supp_vht = supp_vht || sband->vht_cap.vht_supported;
}
if (!supp_vht)
fq->memory_limit = 4 << 20; /* 4 Mbytes */
codel_params_init(&local->cparams);
local->cparams.interval = MS2TIME(100);
local->cparams.target = MS2TIME(20);
local->cparams.ecn = true;
local->cvars = kcalloc(fq->flows_cnt, sizeof(local->cvars[0]),
GFP_KERNEL);
if (!local->cvars) {
mac80211: fix fq lockdep warnings Some lockdep assertions were not fulfilled and resulted in a kernel warning/call trace if driver used intermediate software queues (e.g. ath10k). Existing code sequences should've guaranteed safety but it's always good to be extra careful. The call trace could look like this: [ 237.335805] ------------[ cut here ]------------ [ 237.335852] WARNING: CPU: 3 PID: 1921 at include/net/fq_impl.h:22 fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.335855] Modules linked in: ath10k_pci(E-) ath10k_core(E) ath(E) mac80211(E) cfg80211(E) [ 237.335913] CPU: 3 PID: 1921 Comm: rmmod Tainted: G W E 4.7.0-rc4-wt-ath+ #1377 [ 237.335916] Hardware name: Hewlett-Packard HP ProBook 6540b/1722, BIOS 68CDD Ver. F.04 01/27/2010 [ 237.335918] 00200286 00200286 eff85dac c14151e2 f901574e 00000000 eff85de0 c1081075 [ 237.335928] c1ab91f0 00000003 00000781 f901574e 00000016 f8fbabad f8fbabad 00000016 [ 237.335938] eb24ff60 00000000 ef3886c0 eff85df4 c10810ba 00000009 00000000 00000000 [ 237.335948] Call Trace: [ 237.335953] [<c14151e2>] dump_stack+0x76/0xb4 [ 237.335957] [<c1081075>] __warn+0xe5/0x100 [ 237.336002] [<f8fbabad>] ? fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336046] [<f8fbabad>] ? fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336053] [<c10810ba>] warn_slowpath_null+0x2a/0x30 [ 237.336095] [<f8fbabad>] fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336137] [<f8fbc67a>] fq_flow_reset.constprop.56+0x2a/0x90 [mac80211] [ 237.336180] [<f8fbc79a>] fq_reset.constprop.59+0x2a/0x50 [mac80211] [ 237.336222] [<f8fc04e8>] ieee80211_txq_teardown_flows+0x38/0x40 [mac80211] [ 237.336258] [<f8f7c1a4>] ieee80211_unregister_hw+0xe4/0x120 [mac80211] [ 237.336275] [<f933f536>] ath10k_mac_unregister+0x16/0x50 [ath10k_core] [ 237.336292] [<f934592d>] ath10k_core_unregister+0x3d/0x90 [ath10k_core] [ 237.336301] [<f85f8836>] ath10k_pci_remove+0x36/0xa0 [ath10k_pci] [ 237.336307] [<c1470388>] pci_device_remove+0x38/0xb0 ... Fixes: 5caa328e3811 ("mac80211: implement codel on fair queuing flows") Fixes: fa962b92120b ("mac80211: implement fair queueing per txq") Tested-by: Kalle Valo <kvalo@qca.qualcomm.com> Reported-by: Kalle Valo <kvalo@qca.qualcomm.com> Signed-off-by: Michal Kazior <michal.kazior@tieto.com> Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
2016-06-29 20:00:34 +08:00
spin_lock_bh(&fq->lock);
fq_reset(fq, fq_skb_free_func);
mac80211: fix fq lockdep warnings Some lockdep assertions were not fulfilled and resulted in a kernel warning/call trace if driver used intermediate software queues (e.g. ath10k). Existing code sequences should've guaranteed safety but it's always good to be extra careful. The call trace could look like this: [ 237.335805] ------------[ cut here ]------------ [ 237.335852] WARNING: CPU: 3 PID: 1921 at include/net/fq_impl.h:22 fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.335855] Modules linked in: ath10k_pci(E-) ath10k_core(E) ath(E) mac80211(E) cfg80211(E) [ 237.335913] CPU: 3 PID: 1921 Comm: rmmod Tainted: G W E 4.7.0-rc4-wt-ath+ #1377 [ 237.335916] Hardware name: Hewlett-Packard HP ProBook 6540b/1722, BIOS 68CDD Ver. F.04 01/27/2010 [ 237.335918] 00200286 00200286 eff85dac c14151e2 f901574e 00000000 eff85de0 c1081075 [ 237.335928] c1ab91f0 00000003 00000781 f901574e 00000016 f8fbabad f8fbabad 00000016 [ 237.335938] eb24ff60 00000000 ef3886c0 eff85df4 c10810ba 00000009 00000000 00000000 [ 237.335948] Call Trace: [ 237.335953] [<c14151e2>] dump_stack+0x76/0xb4 [ 237.335957] [<c1081075>] __warn+0xe5/0x100 [ 237.336002] [<f8fbabad>] ? fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336046] [<f8fbabad>] ? fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336053] [<c10810ba>] warn_slowpath_null+0x2a/0x30 [ 237.336095] [<f8fbabad>] fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336137] [<f8fbc67a>] fq_flow_reset.constprop.56+0x2a/0x90 [mac80211] [ 237.336180] [<f8fbc79a>] fq_reset.constprop.59+0x2a/0x50 [mac80211] [ 237.336222] [<f8fc04e8>] ieee80211_txq_teardown_flows+0x38/0x40 [mac80211] [ 237.336258] [<f8f7c1a4>] ieee80211_unregister_hw+0xe4/0x120 [mac80211] [ 237.336275] [<f933f536>] ath10k_mac_unregister+0x16/0x50 [ath10k_core] [ 237.336292] [<f934592d>] ath10k_core_unregister+0x3d/0x90 [ath10k_core] [ 237.336301] [<f85f8836>] ath10k_pci_remove+0x36/0xa0 [ath10k_pci] [ 237.336307] [<c1470388>] pci_device_remove+0x38/0xb0 ... Fixes: 5caa328e3811 ("mac80211: implement codel on fair queuing flows") Fixes: fa962b92120b ("mac80211: implement fair queueing per txq") Tested-by: Kalle Valo <kvalo@qca.qualcomm.com> Reported-by: Kalle Valo <kvalo@qca.qualcomm.com> Signed-off-by: Michal Kazior <michal.kazior@tieto.com> Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
2016-06-29 20:00:34 +08:00
spin_unlock_bh(&fq->lock);
return -ENOMEM;
}
for (i = 0; i < fq->flows_cnt; i++)
codel_vars_init(&local->cvars[i]);
ieee80211_txq_set_params(local);
return 0;
}
void ieee80211_txq_teardown_flows(struct ieee80211_local *local)
{
struct fq *fq = &local->fq;
if (!local->ops->wake_tx_queue)
return;
kfree(local->cvars);
local->cvars = NULL;
mac80211: fix fq lockdep warnings Some lockdep assertions were not fulfilled and resulted in a kernel warning/call trace if driver used intermediate software queues (e.g. ath10k). Existing code sequences should've guaranteed safety but it's always good to be extra careful. The call trace could look like this: [ 237.335805] ------------[ cut here ]------------ [ 237.335852] WARNING: CPU: 3 PID: 1921 at include/net/fq_impl.h:22 fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.335855] Modules linked in: ath10k_pci(E-) ath10k_core(E) ath(E) mac80211(E) cfg80211(E) [ 237.335913] CPU: 3 PID: 1921 Comm: rmmod Tainted: G W E 4.7.0-rc4-wt-ath+ #1377 [ 237.335916] Hardware name: Hewlett-Packard HP ProBook 6540b/1722, BIOS 68CDD Ver. F.04 01/27/2010 [ 237.335918] 00200286 00200286 eff85dac c14151e2 f901574e 00000000 eff85de0 c1081075 [ 237.335928] c1ab91f0 00000003 00000781 f901574e 00000016 f8fbabad f8fbabad 00000016 [ 237.335938] eb24ff60 00000000 ef3886c0 eff85df4 c10810ba 00000009 00000000 00000000 [ 237.335948] Call Trace: [ 237.335953] [<c14151e2>] dump_stack+0x76/0xb4 [ 237.335957] [<c1081075>] __warn+0xe5/0x100 [ 237.336002] [<f8fbabad>] ? fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336046] [<f8fbabad>] ? fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336053] [<c10810ba>] warn_slowpath_null+0x2a/0x30 [ 237.336095] [<f8fbabad>] fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336137] [<f8fbc67a>] fq_flow_reset.constprop.56+0x2a/0x90 [mac80211] [ 237.336180] [<f8fbc79a>] fq_reset.constprop.59+0x2a/0x50 [mac80211] [ 237.336222] [<f8fc04e8>] ieee80211_txq_teardown_flows+0x38/0x40 [mac80211] [ 237.336258] [<f8f7c1a4>] ieee80211_unregister_hw+0xe4/0x120 [mac80211] [ 237.336275] [<f933f536>] ath10k_mac_unregister+0x16/0x50 [ath10k_core] [ 237.336292] [<f934592d>] ath10k_core_unregister+0x3d/0x90 [ath10k_core] [ 237.336301] [<f85f8836>] ath10k_pci_remove+0x36/0xa0 [ath10k_pci] [ 237.336307] [<c1470388>] pci_device_remove+0x38/0xb0 ... Fixes: 5caa328e3811 ("mac80211: implement codel on fair queuing flows") Fixes: fa962b92120b ("mac80211: implement fair queueing per txq") Tested-by: Kalle Valo <kvalo@qca.qualcomm.com> Reported-by: Kalle Valo <kvalo@qca.qualcomm.com> Signed-off-by: Michal Kazior <michal.kazior@tieto.com> Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
2016-06-29 20:00:34 +08:00
spin_lock_bh(&fq->lock);
fq_reset(fq, fq_skb_free_func);
mac80211: fix fq lockdep warnings Some lockdep assertions were not fulfilled and resulted in a kernel warning/call trace if driver used intermediate software queues (e.g. ath10k). Existing code sequences should've guaranteed safety but it's always good to be extra careful. The call trace could look like this: [ 237.335805] ------------[ cut here ]------------ [ 237.335852] WARNING: CPU: 3 PID: 1921 at include/net/fq_impl.h:22 fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.335855] Modules linked in: ath10k_pci(E-) ath10k_core(E) ath(E) mac80211(E) cfg80211(E) [ 237.335913] CPU: 3 PID: 1921 Comm: rmmod Tainted: G W E 4.7.0-rc4-wt-ath+ #1377 [ 237.335916] Hardware name: Hewlett-Packard HP ProBook 6540b/1722, BIOS 68CDD Ver. F.04 01/27/2010 [ 237.335918] 00200286 00200286 eff85dac c14151e2 f901574e 00000000 eff85de0 c1081075 [ 237.335928] c1ab91f0 00000003 00000781 f901574e 00000016 f8fbabad f8fbabad 00000016 [ 237.335938] eb24ff60 00000000 ef3886c0 eff85df4 c10810ba 00000009 00000000 00000000 [ 237.335948] Call Trace: [ 237.335953] [<c14151e2>] dump_stack+0x76/0xb4 [ 237.335957] [<c1081075>] __warn+0xe5/0x100 [ 237.336002] [<f8fbabad>] ? fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336046] [<f8fbabad>] ? fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336053] [<c10810ba>] warn_slowpath_null+0x2a/0x30 [ 237.336095] [<f8fbabad>] fq_flow_dequeue+0xed/0x140 [mac80211] [ 237.336137] [<f8fbc67a>] fq_flow_reset.constprop.56+0x2a/0x90 [mac80211] [ 237.336180] [<f8fbc79a>] fq_reset.constprop.59+0x2a/0x50 [mac80211] [ 237.336222] [<f8fc04e8>] ieee80211_txq_teardown_flows+0x38/0x40 [mac80211] [ 237.336258] [<f8f7c1a4>] ieee80211_unregister_hw+0xe4/0x120 [mac80211] [ 237.336275] [<f933f536>] ath10k_mac_unregister+0x16/0x50 [ath10k_core] [ 237.336292] [<f934592d>] ath10k_core_unregister+0x3d/0x90 [ath10k_core] [ 237.336301] [<f85f8836>] ath10k_pci_remove+0x36/0xa0 [ath10k_pci] [ 237.336307] [<c1470388>] pci_device_remove+0x38/0xb0 ... Fixes: 5caa328e3811 ("mac80211: implement codel on fair queuing flows") Fixes: fa962b92120b ("mac80211: implement fair queueing per txq") Tested-by: Kalle Valo <kvalo@qca.qualcomm.com> Reported-by: Kalle Valo <kvalo@qca.qualcomm.com> Signed-off-by: Michal Kazior <michal.kazior@tieto.com> Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
2016-06-29 20:00:34 +08:00
spin_unlock_bh(&fq->lock);
}
static bool ieee80211_queue_skb(struct ieee80211_local *local,
struct ieee80211_sub_if_data *sdata,
struct sta_info *sta,
struct sk_buff *skb)
{
struct ieee80211_vif *vif;
struct txq_info *txqi;
if (!local->ops->wake_tx_queue ||
sdata->vif.type == NL80211_IFTYPE_MONITOR)
return false;
if (sdata->vif.type == NL80211_IFTYPE_AP_VLAN)
sdata = container_of(sdata->bss,
struct ieee80211_sub_if_data, u.ap);
vif = &sdata->vif;
mac80211: prevent skb/txq mismatch Station structure is considered as not uploaded (to driver) until drv_sta_state() finishes. This call is however done after the structure is attached to mac80211 internal lists and hashes. This means mac80211 can lookup (and use) station structure before it is uploaded to a driver. If this happens (structure exists, but sta->uploaded is false) fast_tx path can still be taken. Deep in the fastpath call the sta->uploaded is checked against to derive "pubsta" argument for ieee80211_get_txq(). If sta->uploaded is false (and sta is actually non-NULL) ieee80211_get_txq() effectively downgraded to vif->txq. At first glance this may look innocent but coerces mac80211 into a state that is almost guaranteed (codel may drop offending skb) to crash because a station-oriented skb gets queued up on vif-oriented txq. The ieee80211_tx_dequeue() ends up looking at info->control.flags and tries to use txq->sta which in the fail case is NULL. It's probably pointless to pretend one can downgrade skb from sta-txq to vif-txq. Since downgrading unicast traffic to vif->txq must not be done there's no txq to put a frame on if sta->uploaded is false. Therefore the code is made to fall back to regular tx() op path if the described condition is hit. Only drivers using wake_tx_queue were affected. Example crash dump before fix: Unable to handle kernel paging request at virtual address ffffe26c PC is at ieee80211_tx_dequeue+0x204/0x690 [mac80211] [<bf4252a4>] (ieee80211_tx_dequeue [mac80211]) from [<bf4b1388>] (ath10k_mac_tx_push_txq+0x54/0x1c0 [ath10k_core]) [<bf4b1388>] (ath10k_mac_tx_push_txq [ath10k_core]) from [<bf4bdfbc>] (ath10k_htt_txrx_compl_task+0xd78/0x11d0 [ath10k_core]) [<bf4bdfbc>] (ath10k_htt_txrx_compl_task [ath10k_core]) [<bf51c5a4>] (ath10k_pci_napi_poll+0x54/0xe8 [ath10k_pci]) [<bf51c5a4>] (ath10k_pci_napi_poll [ath10k_pci]) from [<c0572e90>] (net_rx_action+0xac/0x160) Reported-by: Mohammed Shafi Shajakhan <mohammed@qti.qualcomm.com> Signed-off-by: Michal Kazior <michal.kazior@tieto.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-01-13 20:32:51 +08:00
txqi = ieee80211_get_txq(local, vif, sta, skb);
if (!txqi)
return false;
ieee80211_txq_enqueue(local, txqi, skb);
schedule_and_wake_txq(local, txqi);
return true;
}
static bool ieee80211_tx_frags(struct ieee80211_local *local,
struct ieee80211_vif *vif,
struct sta_info *sta,
struct sk_buff_head *skbs,
bool txpending)
{
struct ieee80211_tx_control control = {};
struct sk_buff *skb, *tmp;
unsigned long flags;
skb_queue_walk_safe(skbs, skb, tmp) {
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
int q = info->hw_queue;
#ifdef CONFIG_MAC80211_VERBOSE_DEBUG
if (WARN_ON_ONCE(q >= local->hw.queues)) {
__skb_unlink(skb, skbs);
ieee80211_free_txskb(&local->hw, skb);
continue;
}
#endif
spin_lock_irqsave(&local->queue_stop_reason_lock, flags);
if (local->queue_stop_reasons[q] ||
(!txpending && !skb_queue_empty(&local->pending[q]))) {
if (unlikely(info->flags &
IEEE80211_TX_INTFL_OFFCHAN_TX_OK)) {
if (local->queue_stop_reasons[q] &
~BIT(IEEE80211_QUEUE_STOP_REASON_OFFCHANNEL)) {
/*
* Drop off-channel frames if queues
* are stopped for any reason other
* than off-channel operation. Never
* queue them.
*/
spin_unlock_irqrestore(
&local->queue_stop_reason_lock,
flags);
ieee80211_purge_tx_queue(&local->hw,
skbs);
return true;
}
} else {
/*
* Since queue is stopped, queue up frames for
* later transmission from the tx-pending
* tasklet when the queue is woken again.
*/
if (txpending)
skb_queue_splice_init(skbs,
&local->pending[q]);
else
skb_queue_splice_tail_init(skbs,
&local->pending[q]);
spin_unlock_irqrestore(&local->queue_stop_reason_lock,
flags);
return false;
}
}
spin_unlock_irqrestore(&local->queue_stop_reason_lock, flags);
info->control.vif = vif;
control.sta = sta ? &sta->sta : NULL;
__skb_unlink(skb, skbs);
drv_tx(local, &control, skb);
}
return true;
}
/*
* Returns false if the frame couldn't be transmitted but was queued instead.
*/
static bool __ieee80211_tx(struct ieee80211_local *local,
struct sk_buff_head *skbs, struct sta_info *sta,
bool txpending)
{
struct ieee80211_tx_info *info;
struct ieee80211_sub_if_data *sdata;
struct ieee80211_vif *vif;
struct sk_buff *skb;
bool result;
if (WARN_ON(skb_queue_empty(skbs)))
return true;
skb = skb_peek(skbs);
info = IEEE80211_SKB_CB(skb);
sdata = vif_to_sdata(info->control.vif);
if (sta && !sta->uploaded)
sta = NULL;
switch (sdata->vif.type) {
case NL80211_IFTYPE_MONITOR:
if (sdata->u.mntr.flags & MONITOR_FLAG_ACTIVE) {
vif = &sdata->vif;
break;
}
sdata = rcu_dereference(local->monitor_sdata);
if (sdata) {
vif = &sdata->vif;
info->hw_queue =
vif->hw_queue[skb_get_queue_mapping(skb)];
} else if (ieee80211_hw_check(&local->hw, QUEUE_CONTROL)) {
ieee80211_purge_tx_queue(&local->hw, skbs);
return true;
} else
vif = NULL;
break;
case NL80211_IFTYPE_AP_VLAN:
sdata = container_of(sdata->bss,
struct ieee80211_sub_if_data, u.ap);
fallthrough;
default:
vif = &sdata->vif;
break;
}
result = ieee80211_tx_frags(local, vif, sta, skbs, txpending);
WARN_ON_ONCE(!skb_queue_empty(skbs));
return result;
}
/*
* Invoke TX handlers, return 0 on success and non-zero if the
* frame was dropped or queued.
*
* The handlers are split into an early and late part. The latter is everything
* that can be sensitive to reordering, and will be deferred to after packets
* are dequeued from the intermediate queues (when they are enabled).
*/
static int invoke_tx_handlers_early(struct ieee80211_tx_data *tx)
{
ieee80211_tx_result res = TX_DROP;
#define CALL_TXH(txh) \
do { \
res = txh(tx); \
if (res != TX_CONTINUE) \
goto txh_done; \
} while (0)
CALL_TXH(ieee80211_tx_h_dynamic_ps);
CALL_TXH(ieee80211_tx_h_check_assoc);
CALL_TXH(ieee80211_tx_h_ps_buf);
CALL_TXH(ieee80211_tx_h_check_control_port_protocol);
CALL_TXH(ieee80211_tx_h_select_key);
txh_done:
if (unlikely(res == TX_DROP)) {
I802_DEBUG_INC(tx->local->tx_handlers_drop);
if (tx->skb)
ieee80211_free_txskb(&tx->local->hw, tx->skb);
else
ieee80211_purge_tx_queue(&tx->local->hw, &tx->skbs);
return -1;
} else if (unlikely(res == TX_QUEUED)) {
I802_DEBUG_INC(tx->local->tx_handlers_queued);
return -1;
}
return 0;
}
/*
* Late handlers can be called while the sta lock is held. Handlers that can
* cause packets to be generated will cause deadlock!
*/
static int invoke_tx_handlers_late(struct ieee80211_tx_data *tx)
{
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(tx->skb);
ieee80211_tx_result res = TX_CONTINUE;
if (!ieee80211_hw_check(&tx->local->hw, HAS_RATE_CONTROL))
CALL_TXH(ieee80211_tx_h_rate_ctrl);
if (unlikely(info->flags & IEEE80211_TX_INTFL_RETRANSMISSION)) {
__skb_queue_tail(&tx->skbs, tx->skb);
tx->skb = NULL;
goto txh_done;
}
CALL_TXH(ieee80211_tx_h_michael_mic_add);
CALL_TXH(ieee80211_tx_h_sequence);
CALL_TXH(ieee80211_tx_h_fragment);
/* handlers after fragment must be aware of tx info fragmentation! */
CALL_TXH(ieee80211_tx_h_stats);
CALL_TXH(ieee80211_tx_h_encrypt);
if (!ieee80211_hw_check(&tx->local->hw, HAS_RATE_CONTROL))
mac80211: do not calc frame duration when using HW rate-control When rate-control is performed in HW, we cannot calculate frame duration as we do not have the skb transmission rate in SW. ieee80211_tx_h_calculate_duration() should only be called when ieee80211_tx_h_rate_ctrl() has been called before to initialize data in skb->cb. This doesn't happen for drivers with HW rate-control. Fixes the following warning when operating in AP-mode in a driver with HW rate-control. WARNING: at net/mac80211/tx.c:57 ieee80211_duration+0x54/0x1d8 [mac80211]() Modules linked in: wl1271_sdio wl1271 firmware_class crc7 mac80211 cfg80211 [<c0046090>] (unwind_backtrace+0x0/0x124) from [<c0064c10>] (warn_slowpath_common+0x4c/0x64) [<c0064c10>] (warn_slowpath_common+0x4c/0x64) from [<c0064c40>] (warn_slowpath_null+0x18/0x1c) [<c0064c40>] (warn_slowpath_null+0x18/0x1c) from [<bf040e34>] (ieee80211_duration+0x54/0x1d8 [mac80211]) [<bf040e34>] (ieee80211_duration+0x54/0x1d8 [mac80211]) from [<bf04200c>] (invoke_tx_handlers+0xfa0/0x1088 [mac80211]) [<bf04200c>] (invoke_tx_handlers+0xfa0/0x1088 [mac80211]) from [<bf042178>] (ieee80211_tx+0x84/0x248 [mac80211]) [<bf042178>] (ieee80211_tx+0x84/0x248 [mac80211]) from [<bf042f44>] (ieee80211_tx_pending+0x12c/0x278 [mac80211]) [<bf042f44>] (ieee80211_tx_pending+0x12c/0x278 [mac80211]) from [<c0069a9c>] (tasklet_action+0x68/0xbc) [<c0069a9c>] (tasklet_action+0x68/0xbc) from [<c006a044>] (__do_softirq+0x84/0x114) [<c006a044>] (__do_softirq+0x84/0x114) from [<c006a1b8>] (do_softirq+0x48/0x54) [<c006a1b8>] (do_softirq+0x48/0x54) from [<c006a4f8>] (local_bh_enable+0x98/0xcc) [<c006a4f8>] (local_bh_enable+0x98/0xcc) from [<bf074e60>] (wl1271_rx+0x2e8/0x3a4 [wl1271]) [<bf074e60>] (wl1271_rx+0x2e8/0x3a4 [wl1271]) from [<bf071ae4>] (wl1271_irq_work+0x230/0x310 [wl1271]) [<bf071ae4>] (wl1271_irq_work+0x230/0x310 [wl1271]) from [<c0076864>] (process_one_work+0x208/0x350) [<c0076864>] (process_one_work+0x208/0x350) from [<c0076e14>] (worker_thread+0x1cc/0x300) [<c0076e14>] (worker_thread+0x1cc/0x300) from [<c007bb88>] (kthread+0x84/0x8c) [<c007bb88>] (kthread+0x84/0x8c) from [<c0041494>] (kernel_thread_exit+0x0/0x8) Signed-off-by: Arik Nemtsov <arik@wizery.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
2011-02-01 04:29:12 +08:00
CALL_TXH(ieee80211_tx_h_calculate_duration);
#undef CALL_TXH
txh_done:
if (unlikely(res == TX_DROP)) {
I802_DEBUG_INC(tx->local->tx_handlers_drop);
if (tx->skb)
ieee80211_free_txskb(&tx->local->hw, tx->skb);
else
ieee80211_purge_tx_queue(&tx->local->hw, &tx->skbs);
return -1;
} else if (unlikely(res == TX_QUEUED)) {
I802_DEBUG_INC(tx->local->tx_handlers_queued);
return -1;
}
return 0;
}
static int invoke_tx_handlers(struct ieee80211_tx_data *tx)
{
int r = invoke_tx_handlers_early(tx);
if (r)
return r;
return invoke_tx_handlers_late(tx);
}
bool ieee80211_tx_prepare_skb(struct ieee80211_hw *hw,
struct ieee80211_vif *vif, struct sk_buff *skb,
int band, struct ieee80211_sta **sta)
{
struct ieee80211_sub_if_data *sdata = vif_to_sdata(vif);
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
struct ieee80211_tx_data tx;
struct sk_buff *skb2;
if (ieee80211_tx_prepare(sdata, &tx, NULL, skb) == TX_DROP)
return false;
info->band = band;
info->control.vif = vif;
info->hw_queue = vif->hw_queue[skb_get_queue_mapping(skb)];
if (invoke_tx_handlers(&tx))
return false;
if (sta) {
if (tx.sta)
*sta = &tx.sta->sta;
else
*sta = NULL;
}
/* this function isn't suitable for fragmented data frames */
skb2 = __skb_dequeue(&tx.skbs);
if (WARN_ON(skb2 != skb || !skb_queue_empty(&tx.skbs))) {
ieee80211_free_txskb(hw, skb2);
ieee80211_purge_tx_queue(hw, &tx.skbs);
return false;
}
return true;
}
EXPORT_SYMBOL(ieee80211_tx_prepare_skb);
/*
* Returns false if the frame couldn't be transmitted but was queued instead.
*/
static bool ieee80211_tx(struct ieee80211_sub_if_data *sdata,
struct sta_info *sta, struct sk_buff *skb,
bool txpending)
{
struct ieee80211_local *local = sdata->local;
struct ieee80211_tx_data tx;
ieee80211_tx_result res_prepare;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
bool result = true;
if (unlikely(skb->len < 10)) {
dev_kfree_skb(skb);
return true;
}
/* initialises tx */
res_prepare = ieee80211_tx_prepare(sdata, &tx, sta, skb);
if (unlikely(res_prepare == TX_DROP)) {
ieee80211_free_txskb(&local->hw, skb);
return true;
} else if (unlikely(res_prepare == TX_QUEUED)) {
return true;
}
/* set up hw_queue value early */
if (!(info->flags & IEEE80211_TX_CTL_TX_OFFCHAN) ||
!ieee80211_hw_check(&local->hw, QUEUE_CONTROL))
info->hw_queue =
sdata->vif.hw_queue[skb_get_queue_mapping(skb)];
if (invoke_tx_handlers_early(&tx))
return true;
if (ieee80211_queue_skb(local, sdata, tx.sta, tx.skb))
return true;
if (!invoke_tx_handlers_late(&tx))
result = __ieee80211_tx(local, &tx.skbs, tx.sta, txpending);
return result;
}
/* device xmit handlers */
enum ieee80211_encrypt {
ENCRYPT_NO,
ENCRYPT_MGMT,
ENCRYPT_DATA,
};
static int ieee80211_skb_resize(struct ieee80211_sub_if_data *sdata,
struct sk_buff *skb,
int head_need,
enum ieee80211_encrypt encrypt)
{
struct ieee80211_local *local = sdata->local;
bool enc_tailroom;
int tail_need = 0;
enc_tailroom = encrypt == ENCRYPT_MGMT ||
(encrypt == ENCRYPT_DATA &&
sdata->crypto_tx_tailroom_needed_cnt);
if (enc_tailroom) {
tail_need = IEEE80211_ENCRYPT_TAILROOM;
tail_need -= skb_tailroom(skb);
tail_need = max_t(int, tail_need, 0);
}
if (skb_cloned(skb) &&
(!ieee80211_hw_check(&local->hw, SUPPORTS_CLONED_SKBS) ||
!skb_clone_writable(skb, ETH_HLEN) || enc_tailroom))
I802_DEBUG_INC(local->tx_expand_skb_head_cloned);
else if (head_need || tail_need)
I802_DEBUG_INC(local->tx_expand_skb_head);
else
return 0;
if (pskb_expand_head(skb, head_need, tail_need, GFP_ATOMIC)) {
wiphy_debug(local->hw.wiphy,
"failed to reallocate TX buffer\n");
return -ENOMEM;
}
return 0;
}
void ieee80211_xmit(struct ieee80211_sub_if_data *sdata,
struct sta_info *sta, struct sk_buff *skb)
{
struct ieee80211_local *local = sdata->local;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
struct ieee80211_hdr *hdr = (struct ieee80211_hdr *) skb->data;
int headroom;
enum ieee80211_encrypt encrypt;
if (info->flags & IEEE80211_TX_INTFL_DONT_ENCRYPT)
encrypt = ENCRYPT_NO;
else if (ieee80211_is_mgmt(hdr->frame_control))
encrypt = ENCRYPT_MGMT;
else
encrypt = ENCRYPT_DATA;
headroom = local->tx_headroom;
if (encrypt != ENCRYPT_NO)
headroom += IEEE80211_ENCRYPT_HEADROOM;
headroom -= skb_headroom(skb);
headroom = max_t(int, 0, headroom);
if (ieee80211_skb_resize(sdata, skb, headroom, encrypt)) {
ieee80211_free_txskb(&local->hw, skb);
return;
}
/* reload after potential resize */
hdr = (struct ieee80211_hdr *) skb->data;
info->control.vif = &sdata->vif;
mac80211: mesh power save basics Add routines to - maintain a PS mode for each peer and a non-peer PS mode - indicate own PS mode in transmitted frames - track neighbor STAs power modes - buffer frames when neighbors are in PS mode - add TIM and Awake Window IE to beacons - release frames in Mesh Peer Service Periods Add local_pm to sta_info to represent the link-specific power mode at this station towards the remote station. When a peer link is established, use the default power mode stored in mesh config. Update the PS status if the peering status of a neighbor changes. Maintain a mesh power mode for non-peer mesh STAs. Set the non-peer power mode to active mode during peering. Authenticated mesh peering is currently not working when either node is configured to be in power save mode. Indicate the current power mode in transmitted frames. Use QoS Nulls to indicate mesh power mode transitions. For performance reasons, calls to the function setting the frame flags are placed in HWMP routing routines, as there the STA pointer is already available. Add peer_pm to sta_info to represent the peer's link-specific power mode towards the local station. Add nonpeer_pm to represent the peer's power mode towards all non-peer stations. Track power modes based on received frames. Add the ps_data structure to ieee80211_if_mesh (for TIM map, PS neighbor counter and group-addressed frame buffer). Set WLAN_STA_PS flag for STA in PS mode to use the unicast frame buffering routines in the tx path. Update num_sta_ps to buffer and release group-addressed frames after DTIM beacons. Announce the awake window duration in beacons if in light or deep sleep mode towards any peer or non-peer. Create a TIM IE similarly to AP mode and add it to mesh beacons. Parse received Awake Window IEs and check TIM IEs for buffered frames. Release frames towards peers in mesh Peer Service Periods. Use the corresponding trigger frames and monitor the MPSP status. Append a QoS Null as trigger frame if neccessary to properly end the MPSP. Currently, in HT channels MPSPs behave imperfectly and show large delay spikes and frame losses. Signed-off-by: Marco Porsch <marco@cozybit.com> Signed-off-by: Ivan Bezyazychnyy <ivan.bezyazychnyy@gmail.com> Signed-off-by: Mike Krinkin <krinkin.m.u@gmail.com> Signed-off-by: Max Filippov <jcmvbkbc@gmail.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2013-01-31 01:14:08 +08:00
if (ieee80211_vif_is_mesh(&sdata->vif)) {
if (ieee80211_is_data(hdr->frame_control) &&
is_unicast_ether_addr(hdr->addr1)) {
if (mesh_nexthop_resolve(sdata, skb))
mac80211: mesh power save basics Add routines to - maintain a PS mode for each peer and a non-peer PS mode - indicate own PS mode in transmitted frames - track neighbor STAs power modes - buffer frames when neighbors are in PS mode - add TIM and Awake Window IE to beacons - release frames in Mesh Peer Service Periods Add local_pm to sta_info to represent the link-specific power mode at this station towards the remote station. When a peer link is established, use the default power mode stored in mesh config. Update the PS status if the peering status of a neighbor changes. Maintain a mesh power mode for non-peer mesh STAs. Set the non-peer power mode to active mode during peering. Authenticated mesh peering is currently not working when either node is configured to be in power save mode. Indicate the current power mode in transmitted frames. Use QoS Nulls to indicate mesh power mode transitions. For performance reasons, calls to the function setting the frame flags are placed in HWMP routing routines, as there the STA pointer is already available. Add peer_pm to sta_info to represent the peer's link-specific power mode towards the local station. Add nonpeer_pm to represent the peer's power mode towards all non-peer stations. Track power modes based on received frames. Add the ps_data structure to ieee80211_if_mesh (for TIM map, PS neighbor counter and group-addressed frame buffer). Set WLAN_STA_PS flag for STA in PS mode to use the unicast frame buffering routines in the tx path. Update num_sta_ps to buffer and release group-addressed frames after DTIM beacons. Announce the awake window duration in beacons if in light or deep sleep mode towards any peer or non-peer. Create a TIM IE similarly to AP mode and add it to mesh beacons. Parse received Awake Window IEs and check TIM IEs for buffered frames. Release frames towards peers in mesh Peer Service Periods. Use the corresponding trigger frames and monitor the MPSP status. Append a QoS Null as trigger frame if neccessary to properly end the MPSP. Currently, in HT channels MPSPs behave imperfectly and show large delay spikes and frame losses. Signed-off-by: Marco Porsch <marco@cozybit.com> Signed-off-by: Ivan Bezyazychnyy <ivan.bezyazychnyy@gmail.com> Signed-off-by: Mike Krinkin <krinkin.m.u@gmail.com> Signed-off-by: Max Filippov <jcmvbkbc@gmail.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2013-01-31 01:14:08 +08:00
return; /* skb queued: don't free */
} else {
ieee80211_mps_set_frame_flags(sdata, NULL, hdr);
}
}
ieee80211_set_qos_hdr(sdata, skb);
ieee80211_tx(sdata, sta, skb, false);
}
mac80211: Fix NULL ptr deref for injected rate info The commit cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") moved the code to validate the radiotap header from ieee80211_monitor_start_xmit to ieee80211_parse_tx_radiotap. This made is possible to share more code with the new Tx queue selection code for injected frames. But at the same time, it now required the call of ieee80211_parse_tx_radiotap at the beginning of functions which wanted to handle the radiotap header. And this broke the rate parser for radiotap header parser. The radiotap parser for rates is operating most of the time only on the data in the actual radiotap header. But for the 802.11a/b/g rates, it must also know the selected band from the chandef information. But this information is only written to the ieee80211_tx_info at the end of the ieee80211_monitor_start_xmit - long after ieee80211_parse_tx_radiotap was already called. The info->band information was therefore always 0 (NL80211_BAND_2GHZ) when the parser code tried to access it. For a 5GHz only device, injecting a frame with 802.11a rates would cause a NULL pointer dereference because local->hw.wiphy->bands[NL80211_BAND_2GHZ] would most likely have been NULL when the radiotap parser searched for the correct rate index of the driver. Cc: stable@vger.kernel.org Reported-by: Ben Greear <greearb@candelatech.com> Fixes: cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") Signed-off-by: Mathy Vanhoef <Mathy.Vanhoef@kuleuven.be> [sven@narfation.org: added commit message] Signed-off-by: Sven Eckelmann <sven@narfation.org> Link: https://lore.kernel.org/r/20210530133226.40587-1-sven@narfation.org Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-05-30 21:32:26 +08:00
static bool ieee80211_validate_radiotap_len(struct sk_buff *skb)
{
struct ieee80211_radiotap_header *rthdr =
(struct ieee80211_radiotap_header *)skb->data;
/* check for not even having the fixed radiotap header part */
if (unlikely(skb->len < sizeof(struct ieee80211_radiotap_header)))
return false; /* too short to be possibly valid */
/* is it a header version we can trust to find length from? */
if (unlikely(rthdr->it_version))
return false; /* only version 0 is supported */
/* does the skb contain enough to deliver on the alleged length? */
if (unlikely(skb->len < ieee80211_get_radiotap_len(skb->data)))
return false; /* skb too short for claimed rt header extent */
return true;
}
bool ieee80211_parse_tx_radiotap(struct sk_buff *skb,
struct net_device *dev)
{
struct ieee80211_local *local = wdev_priv(dev->ieee80211_ptr);
struct ieee80211_radiotap_iterator iterator;
struct ieee80211_radiotap_header *rthdr =
(struct ieee80211_radiotap_header *) skb->data;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
int ret = ieee80211_radiotap_iterator_init(&iterator, rthdr, skb->len,
NULL);
u16 txflags;
u16 rate = 0;
bool rate_found = false;
u8 rate_retries = 0;
u16 rate_flags = 0;
u8 mcs_known, mcs_flags, mcs_bw;
u16 vht_known;
u8 vht_mcs = 0, vht_nss = 0;
int i;
mac80211: Fix NULL ptr deref for injected rate info The commit cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") moved the code to validate the radiotap header from ieee80211_monitor_start_xmit to ieee80211_parse_tx_radiotap. This made is possible to share more code with the new Tx queue selection code for injected frames. But at the same time, it now required the call of ieee80211_parse_tx_radiotap at the beginning of functions which wanted to handle the radiotap header. And this broke the rate parser for radiotap header parser. The radiotap parser for rates is operating most of the time only on the data in the actual radiotap header. But for the 802.11a/b/g rates, it must also know the selected band from the chandef information. But this information is only written to the ieee80211_tx_info at the end of the ieee80211_monitor_start_xmit - long after ieee80211_parse_tx_radiotap was already called. The info->band information was therefore always 0 (NL80211_BAND_2GHZ) when the parser code tried to access it. For a 5GHz only device, injecting a frame with 802.11a rates would cause a NULL pointer dereference because local->hw.wiphy->bands[NL80211_BAND_2GHZ] would most likely have been NULL when the radiotap parser searched for the correct rate index of the driver. Cc: stable@vger.kernel.org Reported-by: Ben Greear <greearb@candelatech.com> Fixes: cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") Signed-off-by: Mathy Vanhoef <Mathy.Vanhoef@kuleuven.be> [sven@narfation.org: added commit message] Signed-off-by: Sven Eckelmann <sven@narfation.org> Link: https://lore.kernel.org/r/20210530133226.40587-1-sven@narfation.org Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-05-30 21:32:26 +08:00
if (!ieee80211_validate_radiotap_len(skb))
return false;
info->flags |= IEEE80211_TX_INTFL_DONT_ENCRYPT |
IEEE80211_TX_CTL_DONTFRAG;
/*
* for every radiotap entry that is present
* (ieee80211_radiotap_iterator_next returns -ENOENT when no more
* entries present, or -EINVAL on error)
*/
while (!ret) {
ret = ieee80211_radiotap_iterator_next(&iterator);
if (ret)
continue;
/* see if this argument is something we can use */
switch (iterator.this_arg_index) {
/*
* You must take care when dereferencing iterator.this_arg
* for multibyte types... the pointer is not aligned. Use
* get_unaligned((type *)iterator.this_arg) to dereference
* iterator.this_arg for type "type" safely on all arches.
*/
case IEEE80211_RADIOTAP_FLAGS:
if (*iterator.this_arg & IEEE80211_RADIOTAP_F_FCS) {
/*
* this indicates that the skb we have been
* handed has the 32-bit FCS CRC at the end...
* we should react to that by snipping it off
* because it will be recomputed and added
* on transmission
*/
if (skb->len < (iterator._max_length + FCS_LEN))
return false;
skb_trim(skb, skb->len - FCS_LEN);
}
if (*iterator.this_arg & IEEE80211_RADIOTAP_F_WEP)
info->flags &= ~IEEE80211_TX_INTFL_DONT_ENCRYPT;
if (*iterator.this_arg & IEEE80211_RADIOTAP_F_FRAG)
info->flags &= ~IEEE80211_TX_CTL_DONTFRAG;
break;
case IEEE80211_RADIOTAP_TX_FLAGS:
txflags = get_unaligned_le16(iterator.this_arg);
if (txflags & IEEE80211_RADIOTAP_F_TX_NOACK)
info->flags |= IEEE80211_TX_CTL_NO_ACK;
if (txflags & IEEE80211_RADIOTAP_F_TX_NOSEQNO)
info->control.flags |= IEEE80211_TX_CTRL_NO_SEQNO;
if (txflags & IEEE80211_RADIOTAP_F_TX_ORDER)
info->control.flags |=
IEEE80211_TX_CTRL_DONT_REORDER;
break;
case IEEE80211_RADIOTAP_RATE:
rate = *iterator.this_arg;
rate_flags = 0;
rate_found = true;
break;
case IEEE80211_RADIOTAP_DATA_RETRIES:
rate_retries = *iterator.this_arg;
break;
case IEEE80211_RADIOTAP_MCS:
mcs_known = iterator.this_arg[0];
mcs_flags = iterator.this_arg[1];
if (!(mcs_known & IEEE80211_RADIOTAP_MCS_HAVE_MCS))
break;
rate_found = true;
rate = iterator.this_arg[2];
rate_flags = IEEE80211_TX_RC_MCS;
if (mcs_known & IEEE80211_RADIOTAP_MCS_HAVE_GI &&
mcs_flags & IEEE80211_RADIOTAP_MCS_SGI)
rate_flags |= IEEE80211_TX_RC_SHORT_GI;
mcs_bw = mcs_flags & IEEE80211_RADIOTAP_MCS_BW_MASK;
if (mcs_known & IEEE80211_RADIOTAP_MCS_HAVE_BW &&
mcs_bw == IEEE80211_RADIOTAP_MCS_BW_40)
rate_flags |= IEEE80211_TX_RC_40_MHZ_WIDTH;
if (mcs_known & IEEE80211_RADIOTAP_MCS_HAVE_FEC &&
mcs_flags & IEEE80211_RADIOTAP_MCS_FEC_LDPC)
info->flags |= IEEE80211_TX_CTL_LDPC;
if (mcs_known & IEEE80211_RADIOTAP_MCS_HAVE_STBC) {
u8 stbc = u8_get_bits(mcs_flags,
IEEE80211_RADIOTAP_MCS_STBC_MASK);
info->flags |=
u32_encode_bits(stbc,
IEEE80211_TX_CTL_STBC);
}
break;
case IEEE80211_RADIOTAP_VHT:
vht_known = get_unaligned_le16(iterator.this_arg);
rate_found = true;
rate_flags = IEEE80211_TX_RC_VHT_MCS;
if ((vht_known & IEEE80211_RADIOTAP_VHT_KNOWN_GI) &&
(iterator.this_arg[2] &
IEEE80211_RADIOTAP_VHT_FLAG_SGI))
rate_flags |= IEEE80211_TX_RC_SHORT_GI;
if (vht_known &
IEEE80211_RADIOTAP_VHT_KNOWN_BANDWIDTH) {
if (iterator.this_arg[3] == 1)
rate_flags |=
IEEE80211_TX_RC_40_MHZ_WIDTH;
else if (iterator.this_arg[3] == 4)
rate_flags |=
IEEE80211_TX_RC_80_MHZ_WIDTH;
else if (iterator.this_arg[3] == 11)
rate_flags |=
IEEE80211_TX_RC_160_MHZ_WIDTH;
}
vht_mcs = iterator.this_arg[4] >> 4;
mac80211: limit injected vht mcs/nss in ieee80211_parse_tx_radiotap Limit max values for vht mcs and nss in ieee80211_parse_tx_radiotap routine in order to fix the following warning reported by syzbot: WARNING: CPU: 0 PID: 10717 at include/net/mac80211.h:989 ieee80211_rate_set_vht include/net/mac80211.h:989 [inline] WARNING: CPU: 0 PID: 10717 at include/net/mac80211.h:989 ieee80211_parse_tx_radiotap+0x101e/0x12d0 net/mac80211/tx.c:2244 Modules linked in: CPU: 0 PID: 10717 Comm: syz-executor.5 Not tainted 5.14.0-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:ieee80211_rate_set_vht include/net/mac80211.h:989 [inline] RIP: 0010:ieee80211_parse_tx_radiotap+0x101e/0x12d0 net/mac80211/tx.c:2244 RSP: 0018:ffffc9000186f3e8 EFLAGS: 00010216 RAX: 0000000000000618 RBX: ffff88804ef76500 RCX: ffffc900143a5000 RDX: 0000000000040000 RSI: ffffffff888f478e RDI: 0000000000000003 RBP: 00000000ffffffff R08: 0000000000000000 R09: 0000000000000100 R10: ffffffff888f46f9 R11: 0000000000000000 R12: 00000000fffffff8 R13: ffff88804ef7653c R14: 0000000000000001 R15: 0000000000000004 FS: 00007fbf5718f700(0000) GS:ffff8880b9c00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000001b2de23000 CR3: 000000006a671000 CR4: 00000000001506f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000600 Call Trace: ieee80211_monitor_select_queue+0xa6/0x250 net/mac80211/iface.c:740 netdev_core_pick_tx+0x169/0x2e0 net/core/dev.c:4089 __dev_queue_xmit+0x6f9/0x3710 net/core/dev.c:4165 __bpf_tx_skb net/core/filter.c:2114 [inline] __bpf_redirect_no_mac net/core/filter.c:2139 [inline] __bpf_redirect+0x5ba/0xd20 net/core/filter.c:2162 ____bpf_clone_redirect net/core/filter.c:2429 [inline] bpf_clone_redirect+0x2ae/0x420 net/core/filter.c:2401 bpf_prog_eeb6f53a69e5c6a2+0x59/0x234 bpf_dispatcher_nop_func include/linux/bpf.h:717 [inline] __bpf_prog_run include/linux/filter.h:624 [inline] bpf_prog_run include/linux/filter.h:631 [inline] bpf_test_run+0x381/0xa30 net/bpf/test_run.c:119 bpf_prog_test_run_skb+0xb84/0x1ee0 net/bpf/test_run.c:663 bpf_prog_test_run kernel/bpf/syscall.c:3307 [inline] __sys_bpf+0x2137/0x5df0 kernel/bpf/syscall.c:4605 __do_sys_bpf kernel/bpf/syscall.c:4691 [inline] __se_sys_bpf kernel/bpf/syscall.c:4689 [inline] __x64_sys_bpf+0x75/0xb0 kernel/bpf/syscall.c:4689 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x4665f9 Reported-by: syzbot+0196ac871673f0c20f68@syzkaller.appspotmail.com Fixes: 646e76bb5daf4 ("mac80211: parse VHT info in injected frames") Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Link: https://lore.kernel.org/r/c26c3f02dcb38ab63b2f2534cb463d95ee81bb13.1632141760.git.lorenzo@kernel.org Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-09-20 20:45:22 +08:00
if (vht_mcs > 11)
vht_mcs = 0;
vht_nss = iterator.this_arg[4] & 0xF;
mac80211: limit injected vht mcs/nss in ieee80211_parse_tx_radiotap Limit max values for vht mcs and nss in ieee80211_parse_tx_radiotap routine in order to fix the following warning reported by syzbot: WARNING: CPU: 0 PID: 10717 at include/net/mac80211.h:989 ieee80211_rate_set_vht include/net/mac80211.h:989 [inline] WARNING: CPU: 0 PID: 10717 at include/net/mac80211.h:989 ieee80211_parse_tx_radiotap+0x101e/0x12d0 net/mac80211/tx.c:2244 Modules linked in: CPU: 0 PID: 10717 Comm: syz-executor.5 Not tainted 5.14.0-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:ieee80211_rate_set_vht include/net/mac80211.h:989 [inline] RIP: 0010:ieee80211_parse_tx_radiotap+0x101e/0x12d0 net/mac80211/tx.c:2244 RSP: 0018:ffffc9000186f3e8 EFLAGS: 00010216 RAX: 0000000000000618 RBX: ffff88804ef76500 RCX: ffffc900143a5000 RDX: 0000000000040000 RSI: ffffffff888f478e RDI: 0000000000000003 RBP: 00000000ffffffff R08: 0000000000000000 R09: 0000000000000100 R10: ffffffff888f46f9 R11: 0000000000000000 R12: 00000000fffffff8 R13: ffff88804ef7653c R14: 0000000000000001 R15: 0000000000000004 FS: 00007fbf5718f700(0000) GS:ffff8880b9c00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000001b2de23000 CR3: 000000006a671000 CR4: 00000000001506f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000600 Call Trace: ieee80211_monitor_select_queue+0xa6/0x250 net/mac80211/iface.c:740 netdev_core_pick_tx+0x169/0x2e0 net/core/dev.c:4089 __dev_queue_xmit+0x6f9/0x3710 net/core/dev.c:4165 __bpf_tx_skb net/core/filter.c:2114 [inline] __bpf_redirect_no_mac net/core/filter.c:2139 [inline] __bpf_redirect+0x5ba/0xd20 net/core/filter.c:2162 ____bpf_clone_redirect net/core/filter.c:2429 [inline] bpf_clone_redirect+0x2ae/0x420 net/core/filter.c:2401 bpf_prog_eeb6f53a69e5c6a2+0x59/0x234 bpf_dispatcher_nop_func include/linux/bpf.h:717 [inline] __bpf_prog_run include/linux/filter.h:624 [inline] bpf_prog_run include/linux/filter.h:631 [inline] bpf_test_run+0x381/0xa30 net/bpf/test_run.c:119 bpf_prog_test_run_skb+0xb84/0x1ee0 net/bpf/test_run.c:663 bpf_prog_test_run kernel/bpf/syscall.c:3307 [inline] __sys_bpf+0x2137/0x5df0 kernel/bpf/syscall.c:4605 __do_sys_bpf kernel/bpf/syscall.c:4691 [inline] __se_sys_bpf kernel/bpf/syscall.c:4689 [inline] __x64_sys_bpf+0x75/0xb0 kernel/bpf/syscall.c:4689 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x4665f9 Reported-by: syzbot+0196ac871673f0c20f68@syzkaller.appspotmail.com Fixes: 646e76bb5daf4 ("mac80211: parse VHT info in injected frames") Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Link: https://lore.kernel.org/r/c26c3f02dcb38ab63b2f2534cb463d95ee81bb13.1632141760.git.lorenzo@kernel.org Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-09-20 20:45:22 +08:00
if (!vht_nss || vht_nss > 8)
vht_nss = 1;
break;
/*
* Please update the file
* Documentation/networking/mac80211-injection.rst
* when parsing new fields here.
*/
default:
break;
}
}
if (ret != -ENOENT) /* ie, if we didn't simply run out of fields */
return false;
if (rate_found) {
mac80211: Fix NULL ptr deref for injected rate info The commit cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") moved the code to validate the radiotap header from ieee80211_monitor_start_xmit to ieee80211_parse_tx_radiotap. This made is possible to share more code with the new Tx queue selection code for injected frames. But at the same time, it now required the call of ieee80211_parse_tx_radiotap at the beginning of functions which wanted to handle the radiotap header. And this broke the rate parser for radiotap header parser. The radiotap parser for rates is operating most of the time only on the data in the actual radiotap header. But for the 802.11a/b/g rates, it must also know the selected band from the chandef information. But this information is only written to the ieee80211_tx_info at the end of the ieee80211_monitor_start_xmit - long after ieee80211_parse_tx_radiotap was already called. The info->band information was therefore always 0 (NL80211_BAND_2GHZ) when the parser code tried to access it. For a 5GHz only device, injecting a frame with 802.11a rates would cause a NULL pointer dereference because local->hw.wiphy->bands[NL80211_BAND_2GHZ] would most likely have been NULL when the radiotap parser searched for the correct rate index of the driver. Cc: stable@vger.kernel.org Reported-by: Ben Greear <greearb@candelatech.com> Fixes: cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") Signed-off-by: Mathy Vanhoef <Mathy.Vanhoef@kuleuven.be> [sven@narfation.org: added commit message] Signed-off-by: Sven Eckelmann <sven@narfation.org> Link: https://lore.kernel.org/r/20210530133226.40587-1-sven@narfation.org Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-05-30 21:32:26 +08:00
struct ieee80211_supported_band *sband =
local->hw.wiphy->bands[info->band];
info->control.flags |= IEEE80211_TX_CTRL_RATE_INJECT;
for (i = 0; i < IEEE80211_TX_MAX_RATES; i++) {
info->control.rates[i].idx = -1;
info->control.rates[i].flags = 0;
info->control.rates[i].count = 0;
}
if (rate_flags & IEEE80211_TX_RC_MCS) {
info->control.rates[0].idx = rate;
} else if (rate_flags & IEEE80211_TX_RC_VHT_MCS) {
ieee80211_rate_set_vht(info->control.rates, vht_mcs,
vht_nss);
mac80211: Fix NULL ptr deref for injected rate info The commit cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") moved the code to validate the radiotap header from ieee80211_monitor_start_xmit to ieee80211_parse_tx_radiotap. This made is possible to share more code with the new Tx queue selection code for injected frames. But at the same time, it now required the call of ieee80211_parse_tx_radiotap at the beginning of functions which wanted to handle the radiotap header. And this broke the rate parser for radiotap header parser. The radiotap parser for rates is operating most of the time only on the data in the actual radiotap header. But for the 802.11a/b/g rates, it must also know the selected band from the chandef information. But this information is only written to the ieee80211_tx_info at the end of the ieee80211_monitor_start_xmit - long after ieee80211_parse_tx_radiotap was already called. The info->band information was therefore always 0 (NL80211_BAND_2GHZ) when the parser code tried to access it. For a 5GHz only device, injecting a frame with 802.11a rates would cause a NULL pointer dereference because local->hw.wiphy->bands[NL80211_BAND_2GHZ] would most likely have been NULL when the radiotap parser searched for the correct rate index of the driver. Cc: stable@vger.kernel.org Reported-by: Ben Greear <greearb@candelatech.com> Fixes: cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") Signed-off-by: Mathy Vanhoef <Mathy.Vanhoef@kuleuven.be> [sven@narfation.org: added commit message] Signed-off-by: Sven Eckelmann <sven@narfation.org> Link: https://lore.kernel.org/r/20210530133226.40587-1-sven@narfation.org Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-05-30 21:32:26 +08:00
} else if (sband) {
for (i = 0; i < sband->n_bitrates; i++) {
if (rate * 5 != sband->bitrates[i].bitrate)
continue;
info->control.rates[0].idx = i;
break;
}
}
if (info->control.rates[0].idx < 0)
info->control.flags &= ~IEEE80211_TX_CTRL_RATE_INJECT;
info->control.rates[0].flags = rate_flags;
info->control.rates[0].count = min_t(u8, rate_retries + 1,
local->hw.max_rate_tries);
}
return true;
}
netdev_tx_t ieee80211_monitor_start_xmit(struct sk_buff *skb,
struct net_device *dev)
{
struct ieee80211_local *local = wdev_priv(dev->ieee80211_ptr);
struct ieee80211_chanctx_conf *chanctx_conf;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
struct ieee80211_hdr *hdr;
struct ieee80211_sub_if_data *tmp_sdata, *sdata;
struct cfg80211_chan_def *chandef;
u16 len_rthdr;
int hdrlen;
memset(info, 0, sizeof(*info));
info->flags = IEEE80211_TX_CTL_REQ_TX_STATUS |
IEEE80211_TX_CTL_INJECTED;
mac80211: Fix NULL ptr deref for injected rate info The commit cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") moved the code to validate the radiotap header from ieee80211_monitor_start_xmit to ieee80211_parse_tx_radiotap. This made is possible to share more code with the new Tx queue selection code for injected frames. But at the same time, it now required the call of ieee80211_parse_tx_radiotap at the beginning of functions which wanted to handle the radiotap header. And this broke the rate parser for radiotap header parser. The radiotap parser for rates is operating most of the time only on the data in the actual radiotap header. But for the 802.11a/b/g rates, it must also know the selected band from the chandef information. But this information is only written to the ieee80211_tx_info at the end of the ieee80211_monitor_start_xmit - long after ieee80211_parse_tx_radiotap was already called. The info->band information was therefore always 0 (NL80211_BAND_2GHZ) when the parser code tried to access it. For a 5GHz only device, injecting a frame with 802.11a rates would cause a NULL pointer dereference because local->hw.wiphy->bands[NL80211_BAND_2GHZ] would most likely have been NULL when the radiotap parser searched for the correct rate index of the driver. Cc: stable@vger.kernel.org Reported-by: Ben Greear <greearb@candelatech.com> Fixes: cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") Signed-off-by: Mathy Vanhoef <Mathy.Vanhoef@kuleuven.be> [sven@narfation.org: added commit message] Signed-off-by: Sven Eckelmann <sven@narfation.org> Link: https://lore.kernel.org/r/20210530133226.40587-1-sven@narfation.org Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-05-30 21:32:26 +08:00
/* Sanity-check the length of the radiotap header */
if (!ieee80211_validate_radiotap_len(skb))
goto fail;
/* we now know there is a radiotap header with a length we can use */
len_rthdr = ieee80211_get_radiotap_len(skb->data);
/*
* fix up the pointers accounting for the radiotap
* header still being in there. We are being given
* a precooked IEEE80211 header so no need for
* normal processing
*/
skb_set_mac_header(skb, len_rthdr);
/*
* these are just fixed to the end of the rt area since we
* don't have any better information and at this point, nobody cares
*/
skb_set_network_header(skb, len_rthdr);
skb_set_transport_header(skb, len_rthdr);
if (skb->len < len_rthdr + 2)
goto fail;
hdr = (struct ieee80211_hdr *)(skb->data + len_rthdr);
hdrlen = ieee80211_hdrlen(hdr->frame_control);
if (skb->len < len_rthdr + hdrlen)
goto fail;
/*
* Initialize skb->protocol if the injected frame is a data frame
* carrying a rfc1042 header
*/
if (ieee80211_is_data(hdr->frame_control) &&
skb->len >= len_rthdr + hdrlen + sizeof(rfc1042_header) + 2) {
u8 *payload = (u8 *)hdr + hdrlen;
if (ether_addr_equal(payload, rfc1042_header))
skb->protocol = cpu_to_be16((payload[6] << 8) |
payload[7]);
}
rcu_read_lock();
/*
* We process outgoing injected frames that have a local address
* we handle as though they are non-injected frames.
* This code here isn't entirely correct, the local MAC address
* isn't always enough to find the interface to use; for proper
* VLAN support we have an nl80211-based mechanism.
*
* This is necessary, for example, for old hostapd versions that
* don't use nl80211-based management TX/RX.
*/
sdata = IEEE80211_DEV_TO_SUB_IF(dev);
list_for_each_entry_rcu(tmp_sdata, &local->interfaces, list) {
if (!ieee80211_sdata_running(tmp_sdata))
continue;
if (tmp_sdata->vif.type == NL80211_IFTYPE_MONITOR ||
tmp_sdata->vif.type == NL80211_IFTYPE_AP_VLAN)
continue;
if (ether_addr_equal(tmp_sdata->vif.addr, hdr->addr2)) {
sdata = tmp_sdata;
break;
}
}
chanctx_conf = rcu_dereference(sdata->vif.bss_conf.chanctx_conf);
if (!chanctx_conf) {
tmp_sdata = rcu_dereference(local->monitor_sdata);
if (tmp_sdata)
chanctx_conf =
rcu_dereference(tmp_sdata->vif.bss_conf.chanctx_conf);
}
if (chanctx_conf)
chandef = &chanctx_conf->def;
else if (!local->use_chanctx)
chandef = &local->_oper_chandef;
else
goto fail_rcu;
/*
* Frame injection is not allowed if beaconing is not allowed
* or if we need radar detection. Beaconing is usually not allowed when
* the mode or operation (Adhoc, AP, Mesh) does not support DFS.
* Passive scan is also used in world regulatory domains where
* your country is not known and as such it should be treated as
* NO TX unless the channel is explicitly allowed in which case
* your current regulatory domain would not have the passive scan
* flag.
*
* Since AP mode uses monitor interfaces to inject/TX management
* frames we can make AP mode the exception to this rule once it
* supports radar detection as its implementation can deal with
* radar detection by itself. We can do that later by adding a
* monitor flag interfaces used for AP support.
*/
if (!cfg80211_reg_can_beacon(local->hw.wiphy, chandef,
sdata->vif.type))
goto fail_rcu;
info->band = chandef->chan->band;
/* Initialize skb->priority according to frame type and TID class,
* with respect to the sub interface that the frame will actually
* be transmitted on. If the DONT_REORDER flag is set, the original
* skb-priority is preserved to assure frames injected with this
* flag are not reordered relative to each other.
*/
ieee80211_select_queue_80211(sdata, skb, hdr);
skb_set_queue_mapping(skb, ieee80211_ac_from_tid(skb->priority));
mac80211: Fix NULL ptr deref for injected rate info The commit cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") moved the code to validate the radiotap header from ieee80211_monitor_start_xmit to ieee80211_parse_tx_radiotap. This made is possible to share more code with the new Tx queue selection code for injected frames. But at the same time, it now required the call of ieee80211_parse_tx_radiotap at the beginning of functions which wanted to handle the radiotap header. And this broke the rate parser for radiotap header parser. The radiotap parser for rates is operating most of the time only on the data in the actual radiotap header. But for the 802.11a/b/g rates, it must also know the selected band from the chandef information. But this information is only written to the ieee80211_tx_info at the end of the ieee80211_monitor_start_xmit - long after ieee80211_parse_tx_radiotap was already called. The info->band information was therefore always 0 (NL80211_BAND_2GHZ) when the parser code tried to access it. For a 5GHz only device, injecting a frame with 802.11a rates would cause a NULL pointer dereference because local->hw.wiphy->bands[NL80211_BAND_2GHZ] would most likely have been NULL when the radiotap parser searched for the correct rate index of the driver. Cc: stable@vger.kernel.org Reported-by: Ben Greear <greearb@candelatech.com> Fixes: cb17ed29a7a5 ("mac80211: parse radiotap header when selecting Tx queue") Signed-off-by: Mathy Vanhoef <Mathy.Vanhoef@kuleuven.be> [sven@narfation.org: added commit message] Signed-off-by: Sven Eckelmann <sven@narfation.org> Link: https://lore.kernel.org/r/20210530133226.40587-1-sven@narfation.org Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-05-30 21:32:26 +08:00
/*
* Process the radiotap header. This will now take into account the
* selected chandef above to accurately set injection rates and
* retransmissions.
*/
if (!ieee80211_parse_tx_radiotap(skb, dev))
goto fail_rcu;
/* remove the injection radiotap header */
skb_pull(skb, len_rthdr);
ieee80211_xmit(sdata, NULL, skb);
rcu_read_unlock();
return NETDEV_TX_OK;
fail_rcu:
rcu_read_unlock();
fail:
dev_kfree_skb(skb);
return NETDEV_TX_OK; /* meaning, we dealt with the skb */
}
static inline bool ieee80211_is_tdls_setup(struct sk_buff *skb)
{
u16 ethertype = (skb->data[12] << 8) | skb->data[13];
return ethertype == ETH_P_TDLS &&
skb->len > 14 &&
skb->data[14] == WLAN_TDLS_SNAP_RFTYPE;
}
int ieee80211_lookup_ra_sta(struct ieee80211_sub_if_data *sdata,
struct sk_buff *skb,
struct sta_info **sta_out)
{
struct sta_info *sta;
switch (sdata->vif.type) {
case NL80211_IFTYPE_AP_VLAN:
sta = rcu_dereference(sdata->u.vlan.sta);
if (sta) {
*sta_out = sta;
return 0;
} else if (sdata->wdev.use_4addr) {
return -ENOLINK;
}
fallthrough;
case NL80211_IFTYPE_AP:
case NL80211_IFTYPE_OCB:
case NL80211_IFTYPE_ADHOC:
if (is_multicast_ether_addr(skb->data)) {
*sta_out = ERR_PTR(-ENOENT);
return 0;
}
sta = sta_info_get_bss(sdata, skb->data);
break;
#ifdef CONFIG_MAC80211_MESH
case NL80211_IFTYPE_MESH_POINT:
/* determined much later */
*sta_out = NULL;
return 0;
#endif
case NL80211_IFTYPE_STATION:
if (sdata->wdev.wiphy->flags & WIPHY_FLAG_SUPPORTS_TDLS) {
sta = sta_info_get(sdata, skb->data);
if (sta && test_sta_flag(sta, WLAN_STA_TDLS_PEER)) {
if (test_sta_flag(sta,
WLAN_STA_TDLS_PEER_AUTH)) {
*sta_out = sta;
return 0;
}
/*
* TDLS link during setup - throw out frames to
* peer. Allow TDLS-setup frames to unauthorized
* peers for the special case of a link teardown
* after a TDLS sta is removed due to being
* unreachable.
*/
if (!ieee80211_is_tdls_setup(skb))
return -EINVAL;
}
}
sta = sta_info_get(sdata, sdata->deflink.u.mgd.bssid);
if (!sta)
return -ENOLINK;
break;
default:
return -EINVAL;
}
*sta_out = sta ?: ERR_PTR(-ENOENT);
return 0;
}
static u16 ieee80211_store_ack_skb(struct ieee80211_local *local,
struct sk_buff *skb,
u32 *info_flags,
u64 *cookie)
{
struct sk_buff *ack_skb;
u16 info_id = 0;
if (skb->sk)
ack_skb = skb_clone_sk(skb);
else
ack_skb = skb_clone(skb, GFP_ATOMIC);
if (ack_skb) {
unsigned long flags;
int id;
spin_lock_irqsave(&local->ack_status_lock, flags);
id = idr_alloc(&local->ack_status_frames, ack_skb,
1, 0x2000, GFP_ATOMIC);
spin_unlock_irqrestore(&local->ack_status_lock, flags);
if (id >= 0) {
info_id = id;
*info_flags |= IEEE80211_TX_CTL_REQ_TX_STATUS;
if (cookie) {
*cookie = ieee80211_mgmt_tx_cookie(local);
IEEE80211_SKB_CB(ack_skb)->ack.cookie = *cookie;
}
} else {
kfree_skb(ack_skb);
}
}
return info_id;
}
/**
* ieee80211_build_hdr - build 802.11 header in the given frame
* @sdata: virtual interface to build the header for
* @skb: the skb to build the header in
* @info_flags: skb flags to set
* @sta: the station pointer
* @ctrl_flags: info control flags to set
* @cookie: cookie pointer to fill (if not %NULL)
*
* This function takes the skb with 802.3 header and reformats the header to
* the appropriate IEEE 802.11 header based on which interface the packet is
* being transmitted on.
*
* Note that this function also takes care of the TX status request and
* potential unsharing of the SKB - this needs to be interleaved with the
* header building.
*
* The function requires the read-side RCU lock held
*
* Returns: the (possibly reallocated) skb or an ERR_PTR() code
*/
static struct sk_buff *ieee80211_build_hdr(struct ieee80211_sub_if_data *sdata,
struct sk_buff *skb, u32 info_flags,
struct sta_info *sta, u32 ctrl_flags,
u64 *cookie)
{
struct ieee80211_local *local = sdata->local;
struct ieee80211_tx_info *info;
int head_need;
u16 ethertype, hdrlen, meshhdrlen = 0;
__le16 fc;
struct ieee80211_hdr hdr;
struct ieee80211s_hdr mesh_hdr __maybe_unused;
struct mesh_path __maybe_unused *mppath = NULL, *mpath = NULL;
const u8 *encaps_data;
int encaps_len, skip_header_bytes;
bool wme_sta = false, authorized = false;
bool tdls_peer;
bool multicast;
u16 info_id = 0;
struct ieee80211_chanctx_conf *chanctx_conf;
struct ieee80211_sub_if_data *ap_sdata;
enum nl80211_band band;
int ret;
if (IS_ERR(sta))
sta = NULL;
mac80211: debugfs option to force TX status frames At Technical University of Munich we use MAC 802.11 TX status frames to perform several measurements in MAC 802.11 setups. With ath based drivers this was possible until commit d94a461d7a7df6 ("ath9k: use ieee80211_tx_status_noskb where possible") as the driver ignored the IEEE80211_TX_CTL_REQ_TX_STATUS flag and always delivered tx_status frames. Since that commit, this behavior was changed and the driver now adheres to IEEE80211_TX_CTL_REQ_TX_STATUS. Due to performance reasons, IEEE80211_TX_CTL_REQ_TX_STATUS is not set for data frames from interfaces in managed mode. Hence, frames that are sent from a managed mode interface do never deliver tx_status frames. This remains true even if a monitor mode interface (the measurement interface) is added to the same ieee80211 physical device. Thus, there is no possibility for receiving tx_status frames for frames sent on an interface in managed mode, if the driver adheres to IEEE80211_TX_CTL_REQ_TX_STATUS. In order to force delivery of tx_status frames for research and debugging purposes, implement a debugfs option force_tx_status for ieee80211 physical devices. When this option is set for a physical device, IEEE80211_TX_CTL_REQ_TX_STATUS is enabled in all packets sent from that device. This option can be set via /sys/kernel/debug/ieee80211/<dev>/force_tx_status. The default is disabled. Co-developed-by: Charlie Groh <ga58taw@mytum.de> Signed-off-by: Charlie Groh <ga58taw@mytum.de> Signed-off-by: Julius Niedworok <julius.n@gmx.net> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-03-29 04:01:06 +08:00
#ifdef CONFIG_MAC80211_DEBUGFS
if (local->force_tx_status)
info_flags |= IEEE80211_TX_CTL_REQ_TX_STATUS;
#endif
/* convert Ethernet header to proper 802.11 header (based on
* operation mode) */
ethertype = (skb->data[12] << 8) | skb->data[13];
fc = cpu_to_le16(IEEE80211_FTYPE_DATA | IEEE80211_STYPE_DATA);
switch (sdata->vif.type) {
case NL80211_IFTYPE_AP_VLAN:
if (sdata->wdev.use_4addr) {
fc |= cpu_to_le16(IEEE80211_FCTL_FROMDS | IEEE80211_FCTL_TODS);
/* RA TA DA SA */
memcpy(hdr.addr1, sta->sta.addr, ETH_ALEN);
memcpy(hdr.addr2, sdata->vif.addr, ETH_ALEN);
memcpy(hdr.addr3, skb->data, ETH_ALEN);
memcpy(hdr.addr4, skb->data + ETH_ALEN, ETH_ALEN);
hdrlen = 30;
authorized = test_sta_flag(sta, WLAN_STA_AUTHORIZED);
wme_sta = sta->sta.wme;
}
ap_sdata = container_of(sdata->bss, struct ieee80211_sub_if_data,
u.ap);
chanctx_conf = rcu_dereference(ap_sdata->vif.bss_conf.chanctx_conf);
if (!chanctx_conf) {
ret = -ENOTCONN;
goto free;
}
band = chanctx_conf->def.chan->band;
if (sdata->wdev.use_4addr)
break;
fallthrough;
case NL80211_IFTYPE_AP:
if (sdata->vif.type == NL80211_IFTYPE_AP)
chanctx_conf = rcu_dereference(sdata->vif.bss_conf.chanctx_conf);
if (!chanctx_conf) {
ret = -ENOTCONN;
goto free;
}
fc |= cpu_to_le16(IEEE80211_FCTL_FROMDS);
/* DA BSSID SA */
memcpy(hdr.addr1, skb->data, ETH_ALEN);
memcpy(hdr.addr2, sdata->vif.addr, ETH_ALEN);
memcpy(hdr.addr3, skb->data + ETH_ALEN, ETH_ALEN);
hdrlen = 24;
band = chanctx_conf->def.chan->band;
break;
#ifdef CONFIG_MAC80211_MESH
case NL80211_IFTYPE_MESH_POINT:
if (!is_multicast_ether_addr(skb->data)) {
struct sta_info *next_hop;
bool mpp_lookup = true;
mpath = mesh_path_lookup(sdata, skb->data);
if (mpath) {
mpp_lookup = false;
next_hop = rcu_dereference(mpath->next_hop);
if (!next_hop ||
!(mpath->flags & (MESH_PATH_ACTIVE |
MESH_PATH_RESOLVING)))
mpp_lookup = true;
}
if (mpp_lookup) {
mppath = mpp_path_lookup(sdata, skb->data);
if (mppath)
mppath->exp_time = jiffies;
}
if (mppath && mpath)
mesh_path_del(sdata, mpath->dst);
}
/*
* Use address extension if it is a packet from
* another interface or if we know the destination
* is being proxied by a portal (i.e. portal address
* differs from proxied address)
*/
if (ether_addr_equal(sdata->vif.addr, skb->data + ETH_ALEN) &&
!(mppath && !ether_addr_equal(mppath->mpp, skb->data))) {
hdrlen = ieee80211_fill_mesh_addresses(&hdr, &fc,
skb->data, skb->data + ETH_ALEN);
meshhdrlen = ieee80211_new_mesh_header(sdata, &mesh_hdr,
NULL, NULL);
} else {
/* DS -> MBSS (802.11-2012 13.11.3.3).
* For unicast with unknown forwarding information,
* destination might be in the MBSS or if that fails
* forwarded to another mesh gate. In either case
* resolution will be handled in ieee80211_xmit(), so
* leave the original DA. This also works for mcast */
const u8 *mesh_da = skb->data;
if (mppath)
mesh_da = mppath->mpp;
else if (mpath)
mesh_da = mpath->dst;
hdrlen = ieee80211_fill_mesh_addresses(&hdr, &fc,
mesh_da, sdata->vif.addr);
if (is_multicast_ether_addr(mesh_da))
/* DA TA mSA AE:SA */
meshhdrlen = ieee80211_new_mesh_header(
sdata, &mesh_hdr,
skb->data + ETH_ALEN, NULL);
else
/* RA TA mDA mSA AE:DA SA */
meshhdrlen = ieee80211_new_mesh_header(
sdata, &mesh_hdr, skb->data,
skb->data + ETH_ALEN);
}
chanctx_conf = rcu_dereference(sdata->vif.bss_conf.chanctx_conf);
if (!chanctx_conf) {
ret = -ENOTCONN;
goto free;
}
band = chanctx_conf->def.chan->band;
/* For injected frames, fill RA right away as nexthop lookup
* will be skipped.
*/
if ((ctrl_flags & IEEE80211_TX_CTRL_SKIP_MPATH_LOOKUP) &&
is_zero_ether_addr(hdr.addr1))
memcpy(hdr.addr1, skb->data, ETH_ALEN);
break;
#endif
case NL80211_IFTYPE_STATION:
/* we already did checks when looking up the RA STA */
tdls_peer = test_sta_flag(sta, WLAN_STA_TDLS_PEER);
if (tdls_peer) {
/* DA SA BSSID */
memcpy(hdr.addr1, skb->data, ETH_ALEN);
memcpy(hdr.addr2, skb->data + ETH_ALEN, ETH_ALEN);
memcpy(hdr.addr3, sdata->deflink.u.mgd.bssid, ETH_ALEN);
hdrlen = 24;
} else if (sdata->u.mgd.use_4addr &&
cpu_to_be16(ethertype) != sdata->control_port_protocol) {
fc |= cpu_to_le16(IEEE80211_FCTL_FROMDS |
IEEE80211_FCTL_TODS);
/* RA TA DA SA */
memcpy(hdr.addr1, sdata->deflink.u.mgd.bssid, ETH_ALEN);
memcpy(hdr.addr2, sdata->vif.addr, ETH_ALEN);
memcpy(hdr.addr3, skb->data, ETH_ALEN);
memcpy(hdr.addr4, skb->data + ETH_ALEN, ETH_ALEN);
hdrlen = 30;
} else {
fc |= cpu_to_le16(IEEE80211_FCTL_TODS);
/* BSSID SA DA */
memcpy(hdr.addr1, sdata->deflink.u.mgd.bssid, ETH_ALEN);
memcpy(hdr.addr2, skb->data + ETH_ALEN, ETH_ALEN);
memcpy(hdr.addr3, skb->data, ETH_ALEN);
hdrlen = 24;
}
chanctx_conf = rcu_dereference(sdata->vif.bss_conf.chanctx_conf);
if (!chanctx_conf) {
ret = -ENOTCONN;
goto free;
}
band = chanctx_conf->def.chan->band;
break;
case NL80211_IFTYPE_OCB:
/* DA SA BSSID */
memcpy(hdr.addr1, skb->data, ETH_ALEN);
memcpy(hdr.addr2, skb->data + ETH_ALEN, ETH_ALEN);
eth_broadcast_addr(hdr.addr3);
hdrlen = 24;
chanctx_conf = rcu_dereference(sdata->vif.bss_conf.chanctx_conf);
if (!chanctx_conf) {
ret = -ENOTCONN;
goto free;
}
band = chanctx_conf->def.chan->band;
break;
case NL80211_IFTYPE_ADHOC:
/* DA SA BSSID */
memcpy(hdr.addr1, skb->data, ETH_ALEN);
memcpy(hdr.addr2, skb->data + ETH_ALEN, ETH_ALEN);
memcpy(hdr.addr3, sdata->u.ibss.bssid, ETH_ALEN);
hdrlen = 24;
chanctx_conf = rcu_dereference(sdata->vif.bss_conf.chanctx_conf);
if (!chanctx_conf) {
ret = -ENOTCONN;
goto free;
}
band = chanctx_conf->def.chan->band;
break;
default:
ret = -EINVAL;
goto free;
}
multicast = is_multicast_ether_addr(hdr.addr1);
/* sta is always NULL for mesh */
if (sta) {
authorized = test_sta_flag(sta, WLAN_STA_AUTHORIZED);
wme_sta = sta->sta.wme;
} else if (ieee80211_vif_is_mesh(&sdata->vif)) {
/* For mesh, the use of the QoS header is mandatory */
wme_sta = true;
}
/* receiver does QoS (which also means we do) use it */
if (wme_sta) {
fc |= cpu_to_le16(IEEE80211_STYPE_QOS_DATA);
hdrlen += 2;
}
/*
* Drop unicast frames to unauthorised stations unless they are
* EAPOL frames from the local station.
*/
if (unlikely(!ieee80211_vif_is_mesh(&sdata->vif) &&
(sdata->vif.type != NL80211_IFTYPE_OCB) &&
!multicast && !authorized &&
(cpu_to_be16(ethertype) != sdata->control_port_protocol ||
!ether_addr_equal(sdata->vif.addr, skb->data + ETH_ALEN)))) {
#ifdef CONFIG_MAC80211_VERBOSE_DEBUG
net_info_ratelimited("%s: dropped frame to %pM (unauthorized port)\n",
sdata->name, hdr.addr1);
#endif
I802_DEBUG_INC(local->tx_handlers_drop_unauth_port);
ret = -EPERM;
goto free;
}
if (unlikely(!multicast && ((skb->sk &&
skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS) ||
ctrl_flags & IEEE80211_TX_CTL_REQ_TX_STATUS)))
info_id = ieee80211_store_ack_skb(local, skb, &info_flags,
cookie);
mac80211: Fix BUG in pskb_expand_head when transmitting shared skbs mac80211 doesn't handle shared skbs correctly at the moment. As a result a possible resize can trigger a BUG in pskb_expand_head. [ 676.030000] Kernel bug detected[#1]: [ 676.030000] Cpu 0 [ 676.030000] $ 0 : 00000000 00000000 819662ff 00000002 [ 676.030000] $ 4 : 81966200 00000020 00000000 00000020 [ 676.030000] $ 8 : 819662e0 800043c0 00000002 00020000 [ 676.030000] $12 : 3b9aca00 00000000 00000000 00470000 [ 676.030000] $16 : 80ea2000 00000000 00000000 00000000 [ 676.030000] $20 : 818aa200 80ea2018 80ea2000 00000008 [ 676.030000] $24 : 00000002 800ace5c [ 676.030000] $28 : 8199a000 8199bd20 81938f88 80f180d4 [ 676.030000] Hi : 0000026e [ 676.030000] Lo : 0000757e [ 676.030000] epc : 801245e4 pskb_expand_head+0x44/0x1d8 [ 676.030000] Not tainted [ 676.030000] ra : 80f180d4 ieee80211_skb_resize+0xb0/0x114 [mac80211] [ 676.030000] Status: 1000a403 KERNEL EXL IE [ 676.030000] Cause : 10800024 [ 676.030000] PrId : 0001964c (MIPS 24Kc) [ 676.030000] Modules linked in: mac80211_hwsim rt2800lib rt2x00soc rt2x00pci rt2x00lib mac80211 crc_itu_t crc_ccitt cfg80211 compat arc4 aes_generic deflate ecb cbc [last unloaded: rt2800pci] [ 676.030000] Process kpktgend_0 (pid: 97, threadinfo=8199a000, task=81879f48, tls=00000000) [ 676.030000] Stack : ffffffff 00000000 00000000 00000014 00000004 80ea2000 00000000 00000000 [ 676.030000] 818aa200 80f180d4 ffffffff 0000000a 81879f78 81879f48 81879f48 00000018 [ 676.030000] 81966246 80ea2000 818432e0 80f1a420 80203050 81814d98 00000001 81879f48 [ 676.030000] 81879f48 00000018 81966246 818432e0 0000001a 8199bdd4 0000001c 80f1b72c [ 676.030000] 80203020 8001292c 80ef4aa2 7f10b55d 801ab5b8 81879f48 00000188 80005c90 [ 676.030000] ... [ 676.030000] Call Trace: [ 676.030000] [<801245e4>] pskb_expand_head+0x44/0x1d8 [ 676.030000] [<80f180d4>] ieee80211_skb_resize+0xb0/0x114 [mac80211] [ 676.030000] [<80f1a420>] ieee80211_xmit+0x150/0x22c [mac80211] [ 676.030000] [<80f1b72c>] ieee80211_subif_start_xmit+0x6f4/0x73c [mac80211] [ 676.030000] [<8014361c>] pktgen_thread_worker+0xfac/0x16f8 [ 676.030000] [<8002ebe8>] kthread+0x7c/0x88 [ 676.030000] [<80008e0c>] kernel_thread_helper+0x10/0x18 [ 676.030000] [ 676.030000] [ 676.030000] Code: 24020001 10620005 2502001f <0200000d> 0804917a 00000000 2502001f 00441023 00531021 Fix this by making a local copy of shared skbs prior to mangeling them. To avoid copying the skb unnecessarily move the skb_copy call below the checks that don't need write access to the skb. Also, move the assignment of nh_pos and h_pos below the skb_copy to point to the correct skb. It would be possible to avoid another resize of the copied skb by using skb_copy_expand instead of skb_copy but that would make the patch more complex. Also, shared skbs are a corner case right now, so the resize shouldn't matter much. Cc: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com> Cc: stable@kernel.org Signed-off-by: John W. Linville <linville@tuxdriver.com>
2010-12-03 01:44:09 +08:00
/*
* If the skb is shared we need to obtain our own copy.
*/
if (skb_shared(skb)) {
struct sk_buff *tmp_skb = skb;
/* can't happen -- skb is a clone if info_id != 0 */
WARN_ON(info_id);
skb = skb_clone(skb, GFP_ATOMIC);
mac80211: Fix BUG in pskb_expand_head when transmitting shared skbs mac80211 doesn't handle shared skbs correctly at the moment. As a result a possible resize can trigger a BUG in pskb_expand_head. [ 676.030000] Kernel bug detected[#1]: [ 676.030000] Cpu 0 [ 676.030000] $ 0 : 00000000 00000000 819662ff 00000002 [ 676.030000] $ 4 : 81966200 00000020 00000000 00000020 [ 676.030000] $ 8 : 819662e0 800043c0 00000002 00020000 [ 676.030000] $12 : 3b9aca00 00000000 00000000 00470000 [ 676.030000] $16 : 80ea2000 00000000 00000000 00000000 [ 676.030000] $20 : 818aa200 80ea2018 80ea2000 00000008 [ 676.030000] $24 : 00000002 800ace5c [ 676.030000] $28 : 8199a000 8199bd20 81938f88 80f180d4 [ 676.030000] Hi : 0000026e [ 676.030000] Lo : 0000757e [ 676.030000] epc : 801245e4 pskb_expand_head+0x44/0x1d8 [ 676.030000] Not tainted [ 676.030000] ra : 80f180d4 ieee80211_skb_resize+0xb0/0x114 [mac80211] [ 676.030000] Status: 1000a403 KERNEL EXL IE [ 676.030000] Cause : 10800024 [ 676.030000] PrId : 0001964c (MIPS 24Kc) [ 676.030000] Modules linked in: mac80211_hwsim rt2800lib rt2x00soc rt2x00pci rt2x00lib mac80211 crc_itu_t crc_ccitt cfg80211 compat arc4 aes_generic deflate ecb cbc [last unloaded: rt2800pci] [ 676.030000] Process kpktgend_0 (pid: 97, threadinfo=8199a000, task=81879f48, tls=00000000) [ 676.030000] Stack : ffffffff 00000000 00000000 00000014 00000004 80ea2000 00000000 00000000 [ 676.030000] 818aa200 80f180d4 ffffffff 0000000a 81879f78 81879f48 81879f48 00000018 [ 676.030000] 81966246 80ea2000 818432e0 80f1a420 80203050 81814d98 00000001 81879f48 [ 676.030000] 81879f48 00000018 81966246 818432e0 0000001a 8199bdd4 0000001c 80f1b72c [ 676.030000] 80203020 8001292c 80ef4aa2 7f10b55d 801ab5b8 81879f48 00000188 80005c90 [ 676.030000] ... [ 676.030000] Call Trace: [ 676.030000] [<801245e4>] pskb_expand_head+0x44/0x1d8 [ 676.030000] [<80f180d4>] ieee80211_skb_resize+0xb0/0x114 [mac80211] [ 676.030000] [<80f1a420>] ieee80211_xmit+0x150/0x22c [mac80211] [ 676.030000] [<80f1b72c>] ieee80211_subif_start_xmit+0x6f4/0x73c [mac80211] [ 676.030000] [<8014361c>] pktgen_thread_worker+0xfac/0x16f8 [ 676.030000] [<8002ebe8>] kthread+0x7c/0x88 [ 676.030000] [<80008e0c>] kernel_thread_helper+0x10/0x18 [ 676.030000] [ 676.030000] [ 676.030000] Code: 24020001 10620005 2502001f <0200000d> 0804917a 00000000 2502001f 00441023 00531021 Fix this by making a local copy of shared skbs prior to mangeling them. To avoid copying the skb unnecessarily move the skb_copy call below the checks that don't need write access to the skb. Also, move the assignment of nh_pos and h_pos below the skb_copy to point to the correct skb. It would be possible to avoid another resize of the copied skb by using skb_copy_expand instead of skb_copy but that would make the patch more complex. Also, shared skbs are a corner case right now, so the resize shouldn't matter much. Cc: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com> Cc: stable@kernel.org Signed-off-by: John W. Linville <linville@tuxdriver.com>
2010-12-03 01:44:09 +08:00
kfree_skb(tmp_skb);
if (!skb) {
ret = -ENOMEM;
goto free;
}
mac80211: Fix BUG in pskb_expand_head when transmitting shared skbs mac80211 doesn't handle shared skbs correctly at the moment. As a result a possible resize can trigger a BUG in pskb_expand_head. [ 676.030000] Kernel bug detected[#1]: [ 676.030000] Cpu 0 [ 676.030000] $ 0 : 00000000 00000000 819662ff 00000002 [ 676.030000] $ 4 : 81966200 00000020 00000000 00000020 [ 676.030000] $ 8 : 819662e0 800043c0 00000002 00020000 [ 676.030000] $12 : 3b9aca00 00000000 00000000 00470000 [ 676.030000] $16 : 80ea2000 00000000 00000000 00000000 [ 676.030000] $20 : 818aa200 80ea2018 80ea2000 00000008 [ 676.030000] $24 : 00000002 800ace5c [ 676.030000] $28 : 8199a000 8199bd20 81938f88 80f180d4 [ 676.030000] Hi : 0000026e [ 676.030000] Lo : 0000757e [ 676.030000] epc : 801245e4 pskb_expand_head+0x44/0x1d8 [ 676.030000] Not tainted [ 676.030000] ra : 80f180d4 ieee80211_skb_resize+0xb0/0x114 [mac80211] [ 676.030000] Status: 1000a403 KERNEL EXL IE [ 676.030000] Cause : 10800024 [ 676.030000] PrId : 0001964c (MIPS 24Kc) [ 676.030000] Modules linked in: mac80211_hwsim rt2800lib rt2x00soc rt2x00pci rt2x00lib mac80211 crc_itu_t crc_ccitt cfg80211 compat arc4 aes_generic deflate ecb cbc [last unloaded: rt2800pci] [ 676.030000] Process kpktgend_0 (pid: 97, threadinfo=8199a000, task=81879f48, tls=00000000) [ 676.030000] Stack : ffffffff 00000000 00000000 00000014 00000004 80ea2000 00000000 00000000 [ 676.030000] 818aa200 80f180d4 ffffffff 0000000a 81879f78 81879f48 81879f48 00000018 [ 676.030000] 81966246 80ea2000 818432e0 80f1a420 80203050 81814d98 00000001 81879f48 [ 676.030000] 81879f48 00000018 81966246 818432e0 0000001a 8199bdd4 0000001c 80f1b72c [ 676.030000] 80203020 8001292c 80ef4aa2 7f10b55d 801ab5b8 81879f48 00000188 80005c90 [ 676.030000] ... [ 676.030000] Call Trace: [ 676.030000] [<801245e4>] pskb_expand_head+0x44/0x1d8 [ 676.030000] [<80f180d4>] ieee80211_skb_resize+0xb0/0x114 [mac80211] [ 676.030000] [<80f1a420>] ieee80211_xmit+0x150/0x22c [mac80211] [ 676.030000] [<80f1b72c>] ieee80211_subif_start_xmit+0x6f4/0x73c [mac80211] [ 676.030000] [<8014361c>] pktgen_thread_worker+0xfac/0x16f8 [ 676.030000] [<8002ebe8>] kthread+0x7c/0x88 [ 676.030000] [<80008e0c>] kernel_thread_helper+0x10/0x18 [ 676.030000] [ 676.030000] [ 676.030000] Code: 24020001 10620005 2502001f <0200000d> 0804917a 00000000 2502001f 00441023 00531021 Fix this by making a local copy of shared skbs prior to mangeling them. To avoid copying the skb unnecessarily move the skb_copy call below the checks that don't need write access to the skb. Also, move the assignment of nh_pos and h_pos below the skb_copy to point to the correct skb. It would be possible to avoid another resize of the copied skb by using skb_copy_expand instead of skb_copy but that would make the patch more complex. Also, shared skbs are a corner case right now, so the resize shouldn't matter much. Cc: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com> Cc: stable@kernel.org Signed-off-by: John W. Linville <linville@tuxdriver.com>
2010-12-03 01:44:09 +08:00
}
hdr.frame_control = fc;
hdr.duration_id = 0;
hdr.seq_ctrl = 0;
skip_header_bytes = ETH_HLEN;
if (ethertype == ETH_P_AARP || ethertype == ETH_P_IPX) {
encaps_data = bridge_tunnel_header;
encaps_len = sizeof(bridge_tunnel_header);
skip_header_bytes -= 2;
} else if (ethertype >= ETH_P_802_3_MIN) {
encaps_data = rfc1042_header;
encaps_len = sizeof(rfc1042_header);
skip_header_bytes -= 2;
} else {
encaps_data = NULL;
encaps_len = 0;
}
skb_pull(skb, skip_header_bytes);
head_need = hdrlen + encaps_len + meshhdrlen - skb_headroom(skb);
/*
* So we need to modify the skb header and hence need a copy of
* that. The head_need variable above doesn't, so far, include
* the needed header space that we don't need right away. If we
* can, then we don't reallocate right now but only after the
* frame arrives at the master device (if it does...)
*
* If we cannot, however, then we will reallocate to include all
* the ever needed space. Also, if we need to reallocate it anyway,
* make it big enough for everything we may ever need.
*/
if (head_need > 0 || skb_cloned(skb)) {
head_need += IEEE80211_ENCRYPT_HEADROOM;
head_need += local->tx_headroom;
head_need = max_t(int, 0, head_need);
if (ieee80211_skb_resize(sdata, skb, head_need, ENCRYPT_DATA)) {
ieee80211_free_txskb(&local->hw, skb);
skb = NULL;
return ERR_PTR(-ENOMEM);
}
}
if (encaps_data)
memcpy(skb_push(skb, encaps_len), encaps_data, encaps_len);
#ifdef CONFIG_MAC80211_MESH
if (meshhdrlen > 0)
memcpy(skb_push(skb, meshhdrlen), &mesh_hdr, meshhdrlen);
#endif
if (ieee80211_is_data_qos(fc)) {
__le16 *qos_control;
qos_control = skb_push(skb, 2);
memcpy(skb_push(skb, hdrlen - 2), &hdr, hdrlen - 2);
/*
* Maybe we could actually set some fields here, for now just
* initialise to zero to indicate no special operation.
*/
*qos_control = 0;
} else
memcpy(skb_push(skb, hdrlen), &hdr, hdrlen);
skb_reset_mac_header(skb);
info = IEEE80211_SKB_CB(skb);
memset(info, 0, sizeof(*info));
info->flags = info_flags;
info->ack_frame_id = info_id;
info->band = band;
info->control.flags = ctrl_flags;
return skb;
free:
kfree_skb(skb);
return ERR_PTR(ret);
}
/*
* fast-xmit overview
*
* The core idea of this fast-xmit is to remove per-packet checks by checking
* them out of band. ieee80211_check_fast_xmit() implements the out-of-band
* checks that are needed to get the sta->fast_tx pointer assigned, after which
* much less work can be done per packet. For example, fragmentation must be
* disabled or the fast_tx pointer will not be set. All the conditions are seen
* in the code here.
*
* Once assigned, the fast_tx data structure also caches the per-packet 802.11
* header and other data to aid packet processing in ieee80211_xmit_fast().
*
* The most difficult part of this is that when any of these assumptions
* change, an external trigger (i.e. a call to ieee80211_clear_fast_xmit(),
* ieee80211_check_fast_xmit() or friends) is required to reset the data,
* since the per-packet code no longer checks the conditions. This is reflected
* by the calls to these functions throughout the rest of the code, and must be
* maintained if any of the TX path checks change.
*/
void ieee80211_check_fast_xmit(struct sta_info *sta)
{
struct ieee80211_fast_tx build = {}, *fast_tx = NULL, *old;
struct ieee80211_local *local = sta->local;
struct ieee80211_sub_if_data *sdata = sta->sdata;
struct ieee80211_hdr *hdr = (void *)build.hdr;
struct ieee80211_chanctx_conf *chanctx_conf;
__le16 fc;
if (!ieee80211_hw_check(&local->hw, SUPPORT_FAST_XMIT))
return;
/* Locking here protects both the pointer itself, and against concurrent
* invocations winning data access races to, e.g., the key pointer that
* is used.
* Without it, the invocation of this function right after the key
* pointer changes wouldn't be sufficient, as another CPU could access
* the pointer, then stall, and then do the cache update after the CPU
* that invalidated the key.
* With the locking, such scenarios cannot happen as the check for the
* key and the fast-tx assignment are done atomically, so the CPU that
* modifies the key will either wait or other one will see the key
* cleared/changed already.
*/
spin_lock_bh(&sta->lock);
if (ieee80211_hw_check(&local->hw, SUPPORTS_PS) &&
!ieee80211_hw_check(&local->hw, SUPPORTS_DYNAMIC_PS) &&
sdata->vif.type == NL80211_IFTYPE_STATION)
goto out;
if (!test_sta_flag(sta, WLAN_STA_AUTHORIZED))
goto out;
if (test_sta_flag(sta, WLAN_STA_PS_STA) ||
test_sta_flag(sta, WLAN_STA_PS_DRIVER) ||
test_sta_flag(sta, WLAN_STA_PS_DELIVER) ||
test_sta_flag(sta, WLAN_STA_CLEAR_PS_FILT))
goto out;
if (sdata->noack_map)
goto out;
/* fast-xmit doesn't handle fragmentation at all */
if (local->hw.wiphy->frag_threshold != (u32)-1 &&
!ieee80211_hw_check(&local->hw, SUPPORTS_TX_FRAG))
goto out;
rcu_read_lock();
chanctx_conf = rcu_dereference(sdata->vif.bss_conf.chanctx_conf);
if (!chanctx_conf) {
rcu_read_unlock();
goto out;
}
build.band = chanctx_conf->def.chan->band;
rcu_read_unlock();
fc = cpu_to_le16(IEEE80211_FTYPE_DATA | IEEE80211_STYPE_DATA);
switch (sdata->vif.type) {
case NL80211_IFTYPE_ADHOC:
/* DA SA BSSID */
build.da_offs = offsetof(struct ieee80211_hdr, addr1);
build.sa_offs = offsetof(struct ieee80211_hdr, addr2);
memcpy(hdr->addr3, sdata->u.ibss.bssid, ETH_ALEN);
build.hdr_len = 24;
break;
case NL80211_IFTYPE_STATION:
if (test_sta_flag(sta, WLAN_STA_TDLS_PEER)) {
/* DA SA BSSID */
build.da_offs = offsetof(struct ieee80211_hdr, addr1);
build.sa_offs = offsetof(struct ieee80211_hdr, addr2);
memcpy(hdr->addr3, sdata->deflink.u.mgd.bssid, ETH_ALEN);
build.hdr_len = 24;
break;
}
if (sdata->u.mgd.use_4addr) {
/* non-regular ethertype cannot use the fastpath */
fc |= cpu_to_le16(IEEE80211_FCTL_FROMDS |
IEEE80211_FCTL_TODS);
/* RA TA DA SA */
memcpy(hdr->addr1, sdata->deflink.u.mgd.bssid, ETH_ALEN);
memcpy(hdr->addr2, sdata->vif.addr, ETH_ALEN);
build.da_offs = offsetof(struct ieee80211_hdr, addr3);
build.sa_offs = offsetof(struct ieee80211_hdr, addr4);
build.hdr_len = 30;
break;
}
fc |= cpu_to_le16(IEEE80211_FCTL_TODS);
/* BSSID SA DA */
memcpy(hdr->addr1, sdata->deflink.u.mgd.bssid, ETH_ALEN);
build.da_offs = offsetof(struct ieee80211_hdr, addr3);
build.sa_offs = offsetof(struct ieee80211_hdr, addr2);
build.hdr_len = 24;
break;
case NL80211_IFTYPE_AP_VLAN:
if (sdata->wdev.use_4addr) {
fc |= cpu_to_le16(IEEE80211_FCTL_FROMDS |
IEEE80211_FCTL_TODS);
/* RA TA DA SA */
memcpy(hdr->addr1, sta->sta.addr, ETH_ALEN);
memcpy(hdr->addr2, sdata->vif.addr, ETH_ALEN);
build.da_offs = offsetof(struct ieee80211_hdr, addr3);
build.sa_offs = offsetof(struct ieee80211_hdr, addr4);
build.hdr_len = 30;
break;
}
fallthrough;
case NL80211_IFTYPE_AP:
fc |= cpu_to_le16(IEEE80211_FCTL_FROMDS);
/* DA BSSID SA */
build.da_offs = offsetof(struct ieee80211_hdr, addr1);
memcpy(hdr->addr2, sdata->vif.addr, ETH_ALEN);
build.sa_offs = offsetof(struct ieee80211_hdr, addr3);
build.hdr_len = 24;
break;
default:
/* not handled on fast-xmit */
goto out;
}
if (sta->sta.wme) {
build.hdr_len += 2;
fc |= cpu_to_le16(IEEE80211_STYPE_QOS_DATA);
}
/* We store the key here so there's no point in using rcu_dereference()
* but that's fine because the code that changes the pointers will call
* this function after doing so. For a single CPU that would be enough,
* for multiple see the comment above.
*/
build.key = rcu_access_pointer(sta->ptk[sta->ptk_idx]);
if (!build.key)
build.key = rcu_access_pointer(sdata->default_unicast_key);
if (build.key) {
bool gen_iv, iv_spc, mmic;
gen_iv = build.key->conf.flags & IEEE80211_KEY_FLAG_GENERATE_IV;
iv_spc = build.key->conf.flags & IEEE80211_KEY_FLAG_PUT_IV_SPACE;
mmic = build.key->conf.flags &
(IEEE80211_KEY_FLAG_GENERATE_MMIC |
IEEE80211_KEY_FLAG_PUT_MIC_SPACE);
/* don't handle software crypto */
if (!(build.key->flags & KEY_FLAG_UPLOADED_TO_HARDWARE))
goto out;
mac80211: Fix PTK rekey freezes and clear text leak Rekeying PTK keys without "Extended Key ID for Individually Addressed Frames" did use a procedure not suitable to replace in-use keys and could caused the following issues: 1) Freeze caused by incoming frames: If the local STA installed the key prior to the remote STA we still had the old key active in the hardware when mac80211 switched over to the new key. Therefore there was a window where the card could hand over frames decoded with the old key to mac80211 and bump the new PN (IV) value to an incorrect high number. When it happened the local replay detection silently started to drop all frames sent with the new key. 2) Freeze caused by outgoing frames: If mac80211 was providing the PN (IV) and handed over a clear text frame for encryption to the hardware prior to a key change the driver/card could have processed the queued frame after switching to the new key. This bumped the PN value on the remote STA to an incorrect high number, tricking the remote STA to discard all frames we sent later. 3) Freeze caused by RX aggregation reorder buffer: An aggregation session started with the old key and ending after the switch to the new key also bumped the PN to an incorrect high number, freezing the connection quite similar to 1). 4) Freeze caused by repeating lost frames in an aggregation session: A driver could repeat a lost frame and encrypt it with the new key while in a TX aggregation session without updating the PN for the new key. This also could freeze connections similar to 2). 5) Clear text leak: Removing encryption offload from the card cleared the encryption offload flag only after the card had deleted the key and we did not stop TX during the rekey. The driver/card could therefore get unencrypted frames from mac80211 while no longer be instructed to encrypt them. To prevent those issues the key install logic has been changed: - Mac80211 divers known to be able to rekey PTK0 keys have to set @NL80211_EXT_FEATURE_CAN_REPLACE_PTK0, - mac80211 stops queuing frames depending on the key during the replace - the key is first replaced in the hardware and after that in mac80211 - and mac80211 stops/blocks new aggregation sessions during the rekey. For drivers not setting @NL80211_EXT_FEATURE_CAN_REPLACE_PTK0 the user space must avoid PTK rekeys if "Extended Key ID for Individually Addressed Frames" is not being used. Rekeys for mac80211 drivers without this flag will generate a warning and use an extra call to ieee80211_flush_queues() to both highlight and try to prevent the issues with not updated drivers. The core of the fix changes the key install procedure from: - atomic switch over to the new key in mac80211 - remove the old key in the hardware (stops encryption offloading, fall back to software encryption with a potential clear text packet leak in between) - delete the inactive old key in mac80211 - enable hardware encryption offloading for the new key to: - if it's a PTK mark the old key as tainted to drop TX frames with the outgoing key - replace the key in hardware with the new one - atomic switch over to the new (not marked as tainted) key in mac80211 (which also resumes TX) - delete the inactive old key in mac80211 With the new sequence the hardware will be unable to decrypt frames encrypted with the old key prior to switching to the new key in mac80211 and thus prevent PNs from packets decrypted with the old key to be accounted against the new key. For that to work the drivers have to provide a clear boundary. Mac80211 drivers setting @NL80211_EXT_FEATURE_CAN_REPLACE_PTK0 confirm to provide it and mac80211 will then be able to correctly rekey in-use PTK keys with those drivers. The mac80211 requirements for drivers to set the flag have been added to the "Hardware crypto acceleration" documentation section. It drills down to: The drivers must not hand over frames decrypted with the old key to mac80211 once the call to set_key() with %DISABLE_KEY has been completed. It's allowed to either drop or continue to use the old key for any outgoing frames which are already in the queues, but it must not send out any of them unencrypted or encrypted with the new key. Even with the new boundary in place aggregation sessions with the reorder buffer are problematic: RX aggregation session started prior and completed after the rekey could still dump frames received with the old key at mac80211 after it switched over to the new key. This is side stepped by stopping all (RX and TX) aggregation sessions when replacing a PTK key and hardware key offloading. Stopping TX aggregation sessions avoids the need to get the PNs (IVs) updated in frames prepared for the old key and (re)transmitted after the switch to the new key. As a bonus it improves the compatibility when the remote STA is not handling rekeys as it should. When using software crypto aggregation sessions are not stopped. Mac80211 won't be able to decode the dangerous frames and discard them without special handling. Signed-off-by: Alexander Wetzel <alexander@wetzel-home.de> [trim overly long rekey warning] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-08-31 21:00:38 +08:00
/* Key is being removed */
if (build.key->flags & KEY_FLAG_TAINTED)
goto out;
switch (build.key->conf.cipher) {
case WLAN_CIPHER_SUITE_CCMP:
case WLAN_CIPHER_SUITE_CCMP_256:
if (gen_iv)
build.pn_offs = build.hdr_len;
if (gen_iv || iv_spc)
build.hdr_len += IEEE80211_CCMP_HDR_LEN;
break;
case WLAN_CIPHER_SUITE_GCMP:
case WLAN_CIPHER_SUITE_GCMP_256:
if (gen_iv)
build.pn_offs = build.hdr_len;
if (gen_iv || iv_spc)
build.hdr_len += IEEE80211_GCMP_HDR_LEN;
break;
case WLAN_CIPHER_SUITE_TKIP:
/* cannot handle MMIC or IV generation in xmit-fast */
if (mmic || gen_iv)
goto out;
if (iv_spc)
build.hdr_len += IEEE80211_TKIP_IV_LEN;
break;
case WLAN_CIPHER_SUITE_WEP40:
case WLAN_CIPHER_SUITE_WEP104:
/* cannot handle IV generation in fast-xmit */
if (gen_iv)
goto out;
if (iv_spc)
build.hdr_len += IEEE80211_WEP_IV_LEN;
break;
case WLAN_CIPHER_SUITE_AES_CMAC:
case WLAN_CIPHER_SUITE_BIP_CMAC_256:
case WLAN_CIPHER_SUITE_BIP_GMAC_128:
case WLAN_CIPHER_SUITE_BIP_GMAC_256:
WARN(1,
"management cipher suite 0x%x enabled for data\n",
build.key->conf.cipher);
goto out;
default:
/* we don't know how to generate IVs for this at all */
if (WARN_ON(gen_iv))
goto out;
}
fc |= cpu_to_le16(IEEE80211_FCTL_PROTECTED);
}
hdr->frame_control = fc;
memcpy(build.hdr + build.hdr_len,
rfc1042_header, sizeof(rfc1042_header));
build.hdr_len += sizeof(rfc1042_header);
fast_tx = kmemdup(&build, sizeof(build), GFP_ATOMIC);
/* if the kmemdup fails, continue w/o fast_tx */
out:
/* we might have raced against another call to this function */
old = rcu_dereference_protected(sta->fast_tx,
lockdep_is_held(&sta->lock));
rcu_assign_pointer(sta->fast_tx, fast_tx);
if (old)
kfree_rcu(old, rcu_head);
spin_unlock_bh(&sta->lock);
}
void ieee80211_check_fast_xmit_all(struct ieee80211_local *local)
{
struct sta_info *sta;
rcu_read_lock();
list_for_each_entry_rcu(sta, &local->sta_list, list)
ieee80211_check_fast_xmit(sta);
rcu_read_unlock();
}
void ieee80211_check_fast_xmit_iface(struct ieee80211_sub_if_data *sdata)
{
struct ieee80211_local *local = sdata->local;
struct sta_info *sta;
rcu_read_lock();
list_for_each_entry_rcu(sta, &local->sta_list, list) {
if (sdata != sta->sdata &&
(!sta->sdata->bss || sta->sdata->bss != sdata->bss))
continue;
ieee80211_check_fast_xmit(sta);
}
rcu_read_unlock();
}
void ieee80211_clear_fast_xmit(struct sta_info *sta)
{
struct ieee80211_fast_tx *fast_tx;
spin_lock_bh(&sta->lock);
fast_tx = rcu_dereference_protected(sta->fast_tx,
lockdep_is_held(&sta->lock));
RCU_INIT_POINTER(sta->fast_tx, NULL);
spin_unlock_bh(&sta->lock);
if (fast_tx)
kfree_rcu(fast_tx, rcu_head);
}
static bool ieee80211_amsdu_realloc_pad(struct ieee80211_local *local,
struct sk_buff *skb, int headroom)
{
if (skb_headroom(skb) < headroom) {
I802_DEBUG_INC(local->tx_expand_skb_head);
if (pskb_expand_head(skb, headroom, 0, GFP_ATOMIC)) {
wiphy_debug(local->hw.wiphy,
"failed to reallocate TX buffer\n");
return false;
}
}
return true;
}
static bool ieee80211_amsdu_prepare_head(struct ieee80211_sub_if_data *sdata,
struct ieee80211_fast_tx *fast_tx,
struct sk_buff *skb)
{
struct ieee80211_local *local = sdata->local;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
struct ieee80211_hdr *hdr;
struct ethhdr *amsdu_hdr;
int hdr_len = fast_tx->hdr_len - sizeof(rfc1042_header);
int subframe_len = skb->len - hdr_len;
void *data;
u8 *qc, *h_80211_src, *h_80211_dst;
const u8 *bssid;
if (info->flags & IEEE80211_TX_CTL_RATE_CTRL_PROBE)
return false;
if (info->control.flags & IEEE80211_TX_CTRL_AMSDU)
return true;
if (!ieee80211_amsdu_realloc_pad(local, skb,
sizeof(*amsdu_hdr) +
local->hw.extra_tx_headroom))
return false;
data = skb_push(skb, sizeof(*amsdu_hdr));
memmove(data, data + sizeof(*amsdu_hdr), hdr_len);
hdr = data;
amsdu_hdr = data + hdr_len;
/* h_80211_src/dst is addr* field within hdr */
h_80211_src = data + fast_tx->sa_offs;
h_80211_dst = data + fast_tx->da_offs;
amsdu_hdr->h_proto = cpu_to_be16(subframe_len);
ether_addr_copy(amsdu_hdr->h_source, h_80211_src);
ether_addr_copy(amsdu_hdr->h_dest, h_80211_dst);
/* according to IEEE 802.11-2012 8.3.2 table 8-19, the outer SA/DA
* fields needs to be changed to BSSID for A-MSDU frames depending
* on FromDS/ToDS values.
*/
switch (sdata->vif.type) {
case NL80211_IFTYPE_STATION:
bssid = sdata->deflink.u.mgd.bssid;
break;
case NL80211_IFTYPE_AP:
case NL80211_IFTYPE_AP_VLAN:
bssid = sdata->vif.addr;
break;
default:
bssid = NULL;
}
if (bssid && ieee80211_has_fromds(hdr->frame_control))
ether_addr_copy(h_80211_src, bssid);
if (bssid && ieee80211_has_tods(hdr->frame_control))
ether_addr_copy(h_80211_dst, bssid);
qc = ieee80211_get_qos_ctl(hdr);
*qc |= IEEE80211_QOS_CTL_A_MSDU_PRESENT;
info->control.flags |= IEEE80211_TX_CTRL_AMSDU;
return true;
}
static bool ieee80211_amsdu_aggregate(struct ieee80211_sub_if_data *sdata,
struct sta_info *sta,
struct ieee80211_fast_tx *fast_tx,
struct sk_buff *skb)
{
struct ieee80211_local *local = sdata->local;
struct fq *fq = &local->fq;
struct fq_tin *tin;
struct fq_flow *flow;
u8 tid = skb->priority & IEEE80211_QOS_CTL_TAG1D_MASK;
struct ieee80211_txq *txq = sta->sta.txq[tid];
struct txq_info *txqi;
struct sk_buff **frag_tail, *head;
int subframe_len = skb->len - ETH_ALEN;
u8 max_subframes = sta->sta.max_amsdu_subframes;
int max_frags = local->hw.max_tx_fragments;
int max_amsdu_len = sta->sta.max_amsdu_len;
int orig_truesize;
u32 flow_idx;
__be16 len;
void *data;
bool ret = false;
unsigned int orig_len;
int n = 2, nfrags, pad = 0;
u16 hdrlen;
if (!ieee80211_hw_check(&local->hw, TX_AMSDU))
return false;
if (sdata->vif.offload_flags & IEEE80211_OFFLOAD_ENCAP_ENABLED)
return false;
if (skb_is_gso(skb))
return false;
if (!txq)
return false;
txqi = to_txq_info(txq);
if (test_bit(IEEE80211_TXQ_NO_AMSDU, &txqi->flags))
return false;
if (sta->sta.max_rc_amsdu_len)
max_amsdu_len = min_t(int, max_amsdu_len,
sta->sta.max_rc_amsdu_len);
if (sta->sta.max_tid_amsdu_len[tid])
max_amsdu_len = min_t(int, max_amsdu_len,
sta->sta.max_tid_amsdu_len[tid]);
flow_idx = fq_flow_idx(fq, skb);
spin_lock_bh(&fq->lock);
/* TODO: Ideally aggregation should be done on dequeue to remain
* responsive to environment changes.
*/
tin = &txqi->tin;
flow = fq_flow_classify(fq, tin, flow_idx, skb);
head = skb_peek_tail(&flow->queue);
if (!head || skb_is_gso(head))
goto out;
orig_truesize = head->truesize;
orig_len = head->len;
if (skb->len + head->len > max_amsdu_len)
goto out;
nfrags = 1 + skb_shinfo(skb)->nr_frags;
nfrags += 1 + skb_shinfo(head)->nr_frags;
frag_tail = &skb_shinfo(head)->frag_list;
while (*frag_tail) {
nfrags += 1 + skb_shinfo(*frag_tail)->nr_frags;
frag_tail = &(*frag_tail)->next;
n++;
}
if (max_subframes && n > max_subframes)
goto out;
if (max_frags && nfrags > max_frags)
goto out;
if (!drv_can_aggregate_in_amsdu(local, head, skb))
goto out;
if (!ieee80211_amsdu_prepare_head(sdata, fast_tx, head))
goto out;
/* If n == 2, the "while (*frag_tail)" loop above didn't execute
* and frag_tail should be &skb_shinfo(head)->frag_list.
* However, ieee80211_amsdu_prepare_head() can reallocate it.
* Reload frag_tail to have it pointing to the correct place.
*/
if (n == 2)
frag_tail = &skb_shinfo(head)->frag_list;
/*
* Pad out the previous subframe to a multiple of 4 by adding the
* padding to the next one, that's being added. Note that head->len
* is the length of the full A-MSDU, but that works since each time
* we add a new subframe we pad out the previous one to a multiple
* of 4 and thus it no longer matters in the next round.
*/
hdrlen = fast_tx->hdr_len - sizeof(rfc1042_header);
if ((head->len - hdrlen) & 3)
pad = 4 - ((head->len - hdrlen) & 3);
if (!ieee80211_amsdu_realloc_pad(local, skb, sizeof(rfc1042_header) +
2 + pad))
goto out_recalc;
ret = true;
data = skb_push(skb, ETH_ALEN + 2);
memmove(data, data + ETH_ALEN + 2, 2 * ETH_ALEN);
data += 2 * ETH_ALEN;
len = cpu_to_be16(subframe_len);
memcpy(data, &len, 2);
memcpy(data + 2, rfc1042_header, sizeof(rfc1042_header));
memset(skb_push(skb, pad), 0, pad);
head->len += skb->len;
head->data_len += skb->len;
*frag_tail = skb;
out_recalc:
fq->memory_usage += head->truesize - orig_truesize;
if (head->len != orig_len) {
flow->backlog += head->len - orig_len;
tin->backlog_bytes += head->len - orig_len;
}
out:
spin_unlock_bh(&fq->lock);
return ret;
}
/*
* Can be called while the sta lock is held. Anything that can cause packets to
* be generated will cause deadlock!
*/
static ieee80211_tx_result
ieee80211_xmit_fast_finish(struct ieee80211_sub_if_data *sdata,
struct sta_info *sta, u8 pn_offs,
struct ieee80211_key *key,
struct ieee80211_tx_data *tx)
{
struct sk_buff *skb = tx->skb;
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
struct ieee80211_hdr *hdr = (void *)skb->data;
u8 tid = IEEE80211_NUM_TIDS;
if (!ieee80211_hw_check(&tx->local->hw, HAS_RATE_CONTROL) &&
ieee80211_tx_h_rate_ctrl(tx) != TX_CONTINUE)
return TX_DROP;
if (key)
info->control.hw_key = &key->conf;
dev_sw_netstats_tx_add(skb->dev, 1, skb->len);
if (hdr->frame_control & cpu_to_le16(IEEE80211_STYPE_QOS_DATA)) {
tid = skb->priority & IEEE80211_QOS_CTL_TAG1D_MASK;
hdr->seq_ctrl = ieee80211_tx_next_seq(sta, tid);
} else {
info->flags |= IEEE80211_TX_CTL_ASSIGN_SEQ;
hdr->seq_ctrl = cpu_to_le16(sdata->sequence_number);
sdata->sequence_number += 0x10;
}
if (skb_shinfo(skb)->gso_size)
mac80211: prepare sta handling for MLO support Currently in mac80211 each STA object is represented using sta_info datastructure with the associated STA specific information and drivers access ieee80211_sta part of it. With MLO (Multi Link Operation) support being added in 802.11be standard, though the association is logically with a single Multi Link capable STA, at the physical level communication can happen via different advertised links (uniquely identified by Channel, operating class, BSSID) and hence the need to handle multiple link STA parameters within a composite sta_info object called the MLD STA. The different link STA part of MLD STA are identified using the link address which can be same or different as the MLD STA address and unique link id based on the link vif. To support extension of such a model, the sta_info datastructure is modified to hold multiple link STA objects with link specific params currently within sta_info moved to this new structure. Similarly this is done for ieee80211_sta as well which will be accessed within mac80211 as well as by drivers, hence trivial driver changes are expected to support this. For current non MLO supported drivers, only one link STA is present and link information is accessed via 'deflink' member. For MLO drivers, we still need to define the APIs etc. to get the correct link ID and access the correct part of the station info. Currently in mac80211, all link STA info are accessed directly via deflink. These will be updated to access via link pointers indexed by link id with MLO support patches, with link id being 0 for non MLO supported cases. Except for couple of macro related changes, below spatch takes care of updating mac80211 and driver code to access to the link STA info via deflink. @ieee80211_sta@ struct ieee80211_sta *s; struct sta_info *si; identifier var = {supp_rates, ht_cap, vht_cap, he_cap, he_6ghz_capa, eht_cap, rx_nss, bandwidth, txpwr}; @@ ( s-> - var + deflink.var | si->sta. - var + deflink.var ) @sta_info@ struct sta_info *si; identifier var = {gtk, pcpu_rx_stats, rx_stats, rx_stats_avg, status_stats, tx_stats, cur_max_bandwidth}; @@ ( si-> - var + deflink.var ) Signed-off-by: Sriram R <quic_srirrama@quicinc.com> Link: https://lore.kernel.org/r/1649086883-13246-1-git-send-email-quic_srirrama@quicinc.com [remove MLO-drivers notes from commit message, not clear yet; run spatch] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2022-04-04 23:41:23 +08:00
sta->deflink.tx_stats.msdu[tid] +=
DIV_ROUND_UP(skb->len, skb_shinfo(skb)->gso_size);
else
mac80211: prepare sta handling for MLO support Currently in mac80211 each STA object is represented using sta_info datastructure with the associated STA specific information and drivers access ieee80211_sta part of it. With MLO (Multi Link Operation) support being added in 802.11be standard, though the association is logically with a single Multi Link capable STA, at the physical level communication can happen via different advertised links (uniquely identified by Channel, operating class, BSSID) and hence the need to handle multiple link STA parameters within a composite sta_info object called the MLD STA. The different link STA part of MLD STA are identified using the link address which can be same or different as the MLD STA address and unique link id based on the link vif. To support extension of such a model, the sta_info datastructure is modified to hold multiple link STA objects with link specific params currently within sta_info moved to this new structure. Similarly this is done for ieee80211_sta as well which will be accessed within mac80211 as well as by drivers, hence trivial driver changes are expected to support this. For current non MLO supported drivers, only one link STA is present and link information is accessed via 'deflink' member. For MLO drivers, we still need to define the APIs etc. to get the correct link ID and access the correct part of the station info. Currently in mac80211, all link STA info are accessed directly via deflink. These will be updated to access via link pointers indexed by link id with MLO support patches, with link id being 0 for non MLO supported cases. Except for couple of macro related changes, below spatch takes care of updating mac80211 and driver code to access to the link STA info via deflink. @ieee80211_sta@ struct ieee80211_sta *s; struct sta_info *si; identifier var = {supp_rates, ht_cap, vht_cap, he_cap, he_6ghz_capa, eht_cap, rx_nss, bandwidth, txpwr}; @@ ( s-> - var + deflink.var | si->sta. - var + deflink.var ) @sta_info@ struct sta_info *si; identifier var = {gtk, pcpu_rx_stats, rx_stats, rx_stats_avg, status_stats, tx_stats, cur_max_bandwidth}; @@ ( si-> - var + deflink.var ) Signed-off-by: Sriram R <quic_srirrama@quicinc.com> Link: https://lore.kernel.org/r/1649086883-13246-1-git-send-email-quic_srirrama@quicinc.com [remove MLO-drivers notes from commit message, not clear yet; run spatch] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2022-04-04 23:41:23 +08:00
sta->deflink.tx_stats.msdu[tid]++;
info->hw_queue = sdata->vif.hw_queue[skb_get_queue_mapping(skb)];
/* statistics normally done by ieee80211_tx_h_stats (but that
* has to consider fragmentation, so is more complex)
*/
mac80211: prepare sta handling for MLO support Currently in mac80211 each STA object is represented using sta_info datastructure with the associated STA specific information and drivers access ieee80211_sta part of it. With MLO (Multi Link Operation) support being added in 802.11be standard, though the association is logically with a single Multi Link capable STA, at the physical level communication can happen via different advertised links (uniquely identified by Channel, operating class, BSSID) and hence the need to handle multiple link STA parameters within a composite sta_info object called the MLD STA. The different link STA part of MLD STA are identified using the link address which can be same or different as the MLD STA address and unique link id based on the link vif. To support extension of such a model, the sta_info datastructure is modified to hold multiple link STA objects with link specific params currently within sta_info moved to this new structure. Similarly this is done for ieee80211_sta as well which will be accessed within mac80211 as well as by drivers, hence trivial driver changes are expected to support this. For current non MLO supported drivers, only one link STA is present and link information is accessed via 'deflink' member. For MLO drivers, we still need to define the APIs etc. to get the correct link ID and access the correct part of the station info. Currently in mac80211, all link STA info are accessed directly via deflink. These will be updated to access via link pointers indexed by link id with MLO support patches, with link id being 0 for non MLO supported cases. Except for couple of macro related changes, below spatch takes care of updating mac80211 and driver code to access to the link STA info via deflink. @ieee80211_sta@ struct ieee80211_sta *s; struct sta_info *si; identifier var = {supp_rates, ht_cap, vht_cap, he_cap, he_6ghz_capa, eht_cap, rx_nss, bandwidth, txpwr}; @@ ( s-> - var + deflink.var | si->sta. - var + deflink.var ) @sta_info@ struct sta_info *si; identifier var = {gtk, pcpu_rx_stats, rx_stats, rx_stats_avg, status_stats, tx_stats, cur_max_bandwidth}; @@ ( si-> - var + deflink.var ) Signed-off-by: Sriram R <quic_srirrama@quicinc.com> Link: https://lore.kernel.org/r/1649086883-13246-1-git-send-email-quic_srirrama@quicinc.com [remove MLO-drivers notes from commit message, not clear yet; run spatch] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2022-04-04 23:41:23 +08:00
sta->deflink.tx_stats.bytes[skb_get_queue_mapping(skb)] += skb->len;
sta->deflink.tx_stats.packets[skb_get_queue_mapping(skb)]++;
if (pn_offs) {
u64 pn;
u8 *crypto_hdr = skb->data + pn_offs;
switch (key->conf.cipher) {
case WLAN_CIPHER_SUITE_CCMP:
case WLAN_CIPHER_SUITE_CCMP_256:
case WLAN_CIPHER_SUITE_GCMP:
case WLAN_CIPHER_SUITE_GCMP_256:
pn = atomic64_inc_return(&key->conf.tx_pn);
crypto_hdr[0] = pn;
crypto_hdr[1] = pn >> 8;
crypto_hdr[3] = 0x20 | (key->conf.keyidx << 6);
crypto_hdr[4] = pn >> 16;
crypto_hdr[5] = pn >> 24;
crypto_hdr[6] = pn >> 32;
crypto_hdr[7] = pn >> 40;
break;
}
}
return TX_CONTINUE;
}
static bool ieee80211_xmit_fast(struct ieee80211_sub_if_data *sdata,
struct sta_info *sta,
struct ieee80211_fast_tx *fast_tx,
struct sk_buff *skb)
{
struct ieee80211_local *local = sdata->local;
u16 ethertype = (skb->data[12] << 8) | skb->data[13];
int extra_head = fast_tx->hdr_len - (ETH_HLEN - 2);
int hw_headroom = sdata->local->hw.extra_tx_headroom;
struct ethhdr eth;
struct ieee80211_tx_info *info;
struct ieee80211_hdr *hdr = (void *)fast_tx->hdr;
struct ieee80211_tx_data tx;
ieee80211_tx_result r;
struct tid_ampdu_tx *tid_tx = NULL;
u8 tid = IEEE80211_NUM_TIDS;
/* control port protocol needs a lot of special handling */
if (cpu_to_be16(ethertype) == sdata->control_port_protocol)
return false;
/* only RFC 1042 SNAP */
if (ethertype < ETH_P_802_3_MIN)
return false;
/* don't handle TX status request here either */
if (skb->sk && skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS)
return false;
if (hdr->frame_control & cpu_to_le16(IEEE80211_STYPE_QOS_DATA)) {
tid = skb->priority & IEEE80211_QOS_CTL_TAG1D_MASK;
tid_tx = rcu_dereference(sta->ampdu_mlme.tid_tx[tid]);
if (tid_tx) {
if (!test_bit(HT_AGG_STATE_OPERATIONAL, &tid_tx->state))
return false;
if (tid_tx->timeout)
tid_tx->last_tx = jiffies;
}
}
/* after this point (skb is modified) we cannot return false */
if (skb_shared(skb)) {
struct sk_buff *tmp_skb = skb;
skb = skb_clone(skb, GFP_ATOMIC);
kfree_skb(tmp_skb);
if (!skb)
return true;
}
if ((hdr->frame_control & cpu_to_le16(IEEE80211_STYPE_QOS_DATA)) &&
ieee80211_amsdu_aggregate(sdata, sta, fast_tx, skb))
return true;
/* will not be crypto-handled beyond what we do here, so use false
* as the may-encrypt argument for the resize to not account for
* more room than we already have in 'extra_head'
*/
if (unlikely(ieee80211_skb_resize(sdata, skb,
max_t(int, extra_head + hw_headroom -
skb_headroom(skb), 0),
ENCRYPT_NO))) {
kfree_skb(skb);
return true;
}
memcpy(&eth, skb->data, ETH_HLEN - 2);
hdr = skb_push(skb, extra_head);
memcpy(skb->data, fast_tx->hdr, fast_tx->hdr_len);
memcpy(skb->data + fast_tx->da_offs, eth.h_dest, ETH_ALEN);
memcpy(skb->data + fast_tx->sa_offs, eth.h_source, ETH_ALEN);
info = IEEE80211_SKB_CB(skb);
memset(info, 0, sizeof(*info));
info->band = fast_tx->band;
info->control.vif = &sdata->vif;
info->flags = IEEE80211_TX_CTL_FIRST_FRAGMENT |
IEEE80211_TX_CTL_DONTFRAG |
(tid_tx ? IEEE80211_TX_CTL_AMPDU : 0);
info->control.flags = IEEE80211_TX_CTRL_FAST_XMIT;
mac80211: debugfs option to force TX status frames At Technical University of Munich we use MAC 802.11 TX status frames to perform several measurements in MAC 802.11 setups. With ath based drivers this was possible until commit d94a461d7a7df6 ("ath9k: use ieee80211_tx_status_noskb where possible") as the driver ignored the IEEE80211_TX_CTL_REQ_TX_STATUS flag and always delivered tx_status frames. Since that commit, this behavior was changed and the driver now adheres to IEEE80211_TX_CTL_REQ_TX_STATUS. Due to performance reasons, IEEE80211_TX_CTL_REQ_TX_STATUS is not set for data frames from interfaces in managed mode. Hence, frames that are sent from a managed mode interface do never deliver tx_status frames. This remains true even if a monitor mode interface (the measurement interface) is added to the same ieee80211 physical device. Thus, there is no possibility for receiving tx_status frames for frames sent on an interface in managed mode, if the driver adheres to IEEE80211_TX_CTL_REQ_TX_STATUS. In order to force delivery of tx_status frames for research and debugging purposes, implement a debugfs option force_tx_status for ieee80211 physical devices. When this option is set for a physical device, IEEE80211_TX_CTL_REQ_TX_STATUS is enabled in all packets sent from that device. This option can be set via /sys/kernel/debug/ieee80211/<dev>/force_tx_status. The default is disabled. Co-developed-by: Charlie Groh <ga58taw@mytum.de> Signed-off-by: Charlie Groh <ga58taw@mytum.de> Signed-off-by: Julius Niedworok <julius.n@gmx.net> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-03-29 04:01:06 +08:00
#ifdef CONFIG_MAC80211_DEBUGFS
if (local->force_tx_status)
info->flags |= IEEE80211_TX_CTL_REQ_TX_STATUS;
#endif
if (hdr->frame_control & cpu_to_le16(IEEE80211_STYPE_QOS_DATA)) {
tid = skb->priority & IEEE80211_QOS_CTL_TAG1D_MASK;
*ieee80211_get_qos_ctl(hdr) = tid;
}
__skb_queue_head_init(&tx.skbs);
tx.flags = IEEE80211_TX_UNICAST;
tx.local = local;
tx.sdata = sdata;
tx.sta = sta;
tx.key = fast_tx->key;
if (ieee80211_queue_skb(local, sdata, sta, skb))
return true;
tx.skb = skb;
r = ieee80211_xmit_fast_finish(sdata, sta, fast_tx->pn_offs,
fast_tx->key, &tx);
tx.skb = NULL;
if (r == TX_DROP) {
kfree_skb(skb);
return true;
}
if (sdata->vif.type == NL80211_IFTYPE_AP_VLAN)
sdata = container_of(sdata->bss,
struct ieee80211_sub_if_data, u.ap);
__skb_queue_tail(&tx.skbs, skb);
ieee80211_tx_frags(local, &sdata->vif, sta, &tx.skbs, false);
return true;
}
struct sk_buff *ieee80211_tx_dequeue(struct ieee80211_hw *hw,
struct ieee80211_txq *txq)
{
struct ieee80211_local *local = hw_to_local(hw);
struct txq_info *txqi = container_of(txq, struct txq_info, txq);
struct ieee80211_hdr *hdr;
struct sk_buff *skb = NULL;
struct fq *fq = &local->fq;
struct fq_tin *tin = &txqi->tin;
struct ieee80211_tx_info *info;
struct ieee80211_tx_data tx;
ieee80211_tx_result r;
mac80211: add stop/start logic for software TXQs Sometimes, it is required to stop the transmissions momentarily and resume it later; stopping the txqs becomes very critical in scenarios where the packet transmission has to be ceased completely. For example, during the hardware restart, during off channel operations, when initiating CSA(upon detecting a radar on the DFS channel), etc. The TX queue stop/start logic in mac80211 works well in stopping the TX when drivers make use of netdev queues, i.e, when Qdiscs in network layer take care of traffic scheduling. Since the devices implementing wake_tx_queue can run without Qdiscs, packets will be handed to mac80211 directly without queueing them in the netdev queues. Also, mac80211 does not invoke any of the netif_stop_*/netif_wake_* APIs if wake_tx_queue is implemented. Since the queues are not stopped in this case, transmissions can continue and this will impact negatively on the operation of the wireless device. For example, During hardware restart, we stop the netdev queues so that packets are not sent to the driver. Since ath10k implements wake_tx_queue, TX queues will not be stopped and packets might reach the hardware while it is restarting; this can make hardware unresponsive and the only possible option for recovery is to reboot the entire system. There is another problem to this, it is observed that the packets were sent on the DFS channel for a prolonged duration after radar detection impacting the channel closing time. We can still invoke netif stop/wake APIs when wake_tx_queue is implemented but this could lead to packet drops in network layer; adding stop/start logic for software TXQs in mac80211 instead makes more sense; the change proposed adds the same in mac80211. Signed-off-by: Manikanta Pubbisetty <mpubbise@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-07-11 02:42:53 +08:00
struct ieee80211_vif *vif = txq->vif;
WARN_ON_ONCE(softirq_count() == 0);
if (!ieee80211_txq_airtime_check(hw, txq))
return NULL;
begin:
spin_lock_bh(&fq->lock);
mac80211: add stop/start logic for software TXQs Sometimes, it is required to stop the transmissions momentarily and resume it later; stopping the txqs becomes very critical in scenarios where the packet transmission has to be ceased completely. For example, during the hardware restart, during off channel operations, when initiating CSA(upon detecting a radar on the DFS channel), etc. The TX queue stop/start logic in mac80211 works well in stopping the TX when drivers make use of netdev queues, i.e, when Qdiscs in network layer take care of traffic scheduling. Since the devices implementing wake_tx_queue can run without Qdiscs, packets will be handed to mac80211 directly without queueing them in the netdev queues. Also, mac80211 does not invoke any of the netif_stop_*/netif_wake_* APIs if wake_tx_queue is implemented. Since the queues are not stopped in this case, transmissions can continue and this will impact negatively on the operation of the wireless device. For example, During hardware restart, we stop the netdev queues so that packets are not sent to the driver. Since ath10k implements wake_tx_queue, TX queues will not be stopped and packets might reach the hardware while it is restarting; this can make hardware unresponsive and the only possible option for recovery is to reboot the entire system. There is another problem to this, it is observed that the packets were sent on the DFS channel for a prolonged duration after radar detection impacting the channel closing time. We can still invoke netif stop/wake APIs when wake_tx_queue is implemented but this could lead to packet drops in network layer; adding stop/start logic for software TXQs in mac80211 instead makes more sense; the change proposed adds the same in mac80211. Signed-off-by: Manikanta Pubbisetty <mpubbise@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-07-11 02:42:53 +08:00
if (test_bit(IEEE80211_TXQ_STOP, &txqi->flags) ||
test_bit(IEEE80211_TXQ_STOP_NETIF_TX, &txqi->flags))
goto out;
mac80211: fix TXQ AC confusion Normally, TXQs have txq->tid = tid; txq->ac = ieee80211_ac_from_tid(tid); However, the special management TXQ actually has txq->tid = IEEE80211_NUM_TIDS; // 16 txq->ac = IEEE80211_AC_VO; This makes sense, but ieee80211_ac_from_tid(16) is the same as ieee80211_ac_from_tid(0) which is just IEEE80211_AC_BE. Now, normally this is fine. However, if the netdev queues were stopped, then the code in ieee80211_tx_dequeue() will propagate the stop from the interface (vif->txqs_stopped[]) if the AC 2 (ieee80211_ac_from_tid(txq->tid)) is marked as stopped. On wake, however, __ieee80211_wake_txqs() will wake the TXQ if AC 0 (txq->ac) is woken up. If a driver stops all queues with ieee80211_stop_tx_queues() and then wakes them again with ieee80211_wake_tx_queues(), the ieee80211_wake_txqs() tasklet will run to resync queue and TXQ state. If all queues were woken, then what'll happen is that _ieee80211_wake_txqs() will run in order of HW queues 0-3, typically (and certainly for iwlwifi) corresponding to ACs 0-3, so it'll call __ieee80211_wake_txqs() for each AC in order 0-3. When __ieee80211_wake_txqs() is called for AC 0 (VO) that'll wake up the management TXQ (remember its tid is 16), and the driver's wake_tx_queue() will be called. That tries to get a frame, which will immediately *stop* the TXQ again, because now we check against AC 2, and AC 2 hasn't yet been marked as woken up again in sdata->vif.txqs_stopped[] since we're only in the __ieee80211_wake_txqs() call for AC 0. Thus, the management TXQ will never be started again. Fix this by checking txq->ac directly instead of calculating the AC as ieee80211_ac_from_tid(txq->tid). Fixes: adf8ed01e4fd ("mac80211: add an optional TXQ for other PS-buffered frames") Acked-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210323210500.bf4d50afea4a.I136ffde910486301f8818f5442e3c9bf8670a9c4@changeid Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-03-24 04:05:01 +08:00
if (vif->txqs_stopped[txq->ac]) {
mac80211: add stop/start logic for software TXQs Sometimes, it is required to stop the transmissions momentarily and resume it later; stopping the txqs becomes very critical in scenarios where the packet transmission has to be ceased completely. For example, during the hardware restart, during off channel operations, when initiating CSA(upon detecting a radar on the DFS channel), etc. The TX queue stop/start logic in mac80211 works well in stopping the TX when drivers make use of netdev queues, i.e, when Qdiscs in network layer take care of traffic scheduling. Since the devices implementing wake_tx_queue can run without Qdiscs, packets will be handed to mac80211 directly without queueing them in the netdev queues. Also, mac80211 does not invoke any of the netif_stop_*/netif_wake_* APIs if wake_tx_queue is implemented. Since the queues are not stopped in this case, transmissions can continue and this will impact negatively on the operation of the wireless device. For example, During hardware restart, we stop the netdev queues so that packets are not sent to the driver. Since ath10k implements wake_tx_queue, TX queues will not be stopped and packets might reach the hardware while it is restarting; this can make hardware unresponsive and the only possible option for recovery is to reboot the entire system. There is another problem to this, it is observed that the packets were sent on the DFS channel for a prolonged duration after radar detection impacting the channel closing time. We can still invoke netif stop/wake APIs when wake_tx_queue is implemented but this could lead to packet drops in network layer; adding stop/start logic for software TXQs in mac80211 instead makes more sense; the change proposed adds the same in mac80211. Signed-off-by: Manikanta Pubbisetty <mpubbise@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-07-11 02:42:53 +08:00
set_bit(IEEE80211_TXQ_STOP_NETIF_TX, &txqi->flags);
goto out;
}
/* Make sure fragments stay together. */
skb = __skb_dequeue(&txqi->frags);
if (unlikely(skb)) {
if (!(IEEE80211_SKB_CB(skb)->control.flags &
IEEE80211_TX_INTCFL_NEED_TXPROCESSING))
goto out;
IEEE80211_SKB_CB(skb)->control.flags &=
~IEEE80211_TX_INTCFL_NEED_TXPROCESSING;
} else {
skb = fq_tin_dequeue(fq, tin, fq_tin_dequeue_func);
}
if (!skb)
goto out;
spin_unlock_bh(&fq->lock);
hdr = (struct ieee80211_hdr *)skb->data;
info = IEEE80211_SKB_CB(skb);
memset(&tx, 0, sizeof(tx));
__skb_queue_head_init(&tx.skbs);
tx.local = local;
tx.skb = skb;
tx.sdata = vif_to_sdata(info->control.vif);
if (txq->sta) {
tx.sta = container_of(txq->sta, struct sta_info, sta);
/*
* Drop unicast frames to unauthorised stations unless they are
* injected frames or EAPOL frames from the local station.
*/
if (unlikely(!(info->flags & IEEE80211_TX_CTL_INJECTED) &&
ieee80211_is_data(hdr->frame_control) &&
!ieee80211_vif_is_mesh(&tx.sdata->vif) &&
tx.sdata->vif.type != NL80211_IFTYPE_OCB &&
!is_multicast_ether_addr(hdr->addr1) &&
!test_sta_flag(tx.sta, WLAN_STA_AUTHORIZED) &&
(!(info->control.flags &
IEEE80211_TX_CTRL_PORT_CTRL_PROTO) ||
!ether_addr_equal(tx.sdata->vif.addr,
hdr->addr2)))) {
I802_DEBUG_INC(local->tx_handlers_drop_unauth_port);
ieee80211_free_txskb(&local->hw, skb);
goto begin;
}
}
/*
* The key can be removed while the packet was queued, so need to call
* this here to get the current key.
*/
r = ieee80211_tx_h_select_key(&tx);
if (r != TX_CONTINUE) {
ieee80211_free_txskb(&local->hw, skb);
goto begin;
}
if (test_bit(IEEE80211_TXQ_AMPDU, &txqi->flags))
info->flags |= IEEE80211_TX_CTL_AMPDU;
else
info->flags &= ~IEEE80211_TX_CTL_AMPDU;
if (info->flags & IEEE80211_TX_CTL_HW_80211_ENCAP) {
if (!ieee80211_hw_check(&local->hw, HAS_RATE_CONTROL)) {
r = ieee80211_tx_h_rate_ctrl(&tx);
if (r != TX_CONTINUE) {
ieee80211_free_txskb(&local->hw, skb);
goto begin;
}
}
goto encap_out;
}
if (info->control.flags & IEEE80211_TX_CTRL_FAST_XMIT) {
struct sta_info *sta = container_of(txq->sta, struct sta_info,
sta);
u8 pn_offs = 0;
if (tx.key &&
(tx.key->conf.flags & IEEE80211_KEY_FLAG_GENERATE_IV))
pn_offs = ieee80211_hdrlen(hdr->frame_control);
r = ieee80211_xmit_fast_finish(sta->sdata, sta, pn_offs,
tx.key, &tx);
if (r != TX_CONTINUE) {
ieee80211_free_txskb(&local->hw, skb);
goto begin;
}
} else {
if (invoke_tx_handlers_late(&tx))
goto begin;
skb = __skb_dequeue(&tx.skbs);
if (!skb_queue_empty(&tx.skbs)) {
spin_lock_bh(&fq->lock);
skb_queue_splice_tail(&tx.skbs, &txqi->frags);
spin_unlock_bh(&fq->lock);
}
}
if (skb_has_frag_list(skb) &&
!ieee80211_hw_check(&local->hw, TX_FRAG_LIST)) {
if (skb_linearize(skb)) {
ieee80211_free_txskb(&local->hw, skb);
goto begin;
}
}
switch (tx.sdata->vif.type) {
case NL80211_IFTYPE_MONITOR:
if (tx.sdata->u.mntr.flags & MONITOR_FLAG_ACTIVE) {
vif = &tx.sdata->vif;
break;
}
tx.sdata = rcu_dereference(local->monitor_sdata);
if (tx.sdata) {
vif = &tx.sdata->vif;
info->hw_queue =
vif->hw_queue[skb_get_queue_mapping(skb)];
} else if (ieee80211_hw_check(&local->hw, QUEUE_CONTROL)) {
ieee80211_free_txskb(&local->hw, skb);
goto begin;
} else {
vif = NULL;
}
break;
case NL80211_IFTYPE_AP_VLAN:
tx.sdata = container_of(tx.sdata->bss,
struct ieee80211_sub_if_data, u.ap);
fallthrough;
default:
vif = &tx.sdata->vif;
break;
}
encap_out:
IEEE80211_SKB_CB(skb)->control.vif = vif;
if (vif &&
wiphy_ext_feature_isset(local->hw.wiphy, NL80211_EXT_FEATURE_AQL)) {
bool ampdu = txq->ac != IEEE80211_AC_VO;
u32 airtime;
airtime = ieee80211_calc_expected_tx_airtime(hw, vif, txq->sta,
skb->len, ampdu);
if (airtime) {
airtime = ieee80211_info_set_tx_time_est(info, airtime);
ieee80211_sta_update_pending_airtime(local, tx.sta,
txq->ac,
airtime,
false);
}
}
return skb;
mac80211: add stop/start logic for software TXQs Sometimes, it is required to stop the transmissions momentarily and resume it later; stopping the txqs becomes very critical in scenarios where the packet transmission has to be ceased completely. For example, during the hardware restart, during off channel operations, when initiating CSA(upon detecting a radar on the DFS channel), etc. The TX queue stop/start logic in mac80211 works well in stopping the TX when drivers make use of netdev queues, i.e, when Qdiscs in network layer take care of traffic scheduling. Since the devices implementing wake_tx_queue can run without Qdiscs, packets will be handed to mac80211 directly without queueing them in the netdev queues. Also, mac80211 does not invoke any of the netif_stop_*/netif_wake_* APIs if wake_tx_queue is implemented. Since the queues are not stopped in this case, transmissions can continue and this will impact negatively on the operation of the wireless device. For example, During hardware restart, we stop the netdev queues so that packets are not sent to the driver. Since ath10k implements wake_tx_queue, TX queues will not be stopped and packets might reach the hardware while it is restarting; this can make hardware unresponsive and the only possible option for recovery is to reboot the entire system. There is another problem to this, it is observed that the packets were sent on the DFS channel for a prolonged duration after radar detection impacting the channel closing time. We can still invoke netif stop/wake APIs when wake_tx_queue is implemented but this could lead to packet drops in network layer; adding stop/start logic for software TXQs in mac80211 instead makes more sense; the change proposed adds the same in mac80211. Signed-off-by: Manikanta Pubbisetty <mpubbise@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-07-11 02:42:53 +08:00
out:
spin_unlock_bh(&fq->lock);
return skb;
}
EXPORT_SYMBOL(ieee80211_tx_dequeue);
struct ieee80211_txq *ieee80211_next_txq(struct ieee80211_hw *hw, u8 ac)
{
struct ieee80211_local *local = hw_to_local(hw);
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
struct airtime_sched_info *air_sched;
u64 now = ktime_get_coarse_boottime_ns();
struct ieee80211_txq *ret = NULL;
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
struct airtime_info *air_info;
struct txq_info *txqi = NULL;
struct rb_node *node;
bool first = false;
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
air_sched = &local->airtime[ac];
spin_lock_bh(&air_sched->lock);
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
node = air_sched->schedule_pos;
begin:
if (!node) {
node = rb_first_cached(&air_sched->active_txqs);
first = true;
} else {
node = rb_next(node);
}
if (!node)
goto out;
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
txqi = container_of(node, struct txq_info, schedule_order);
air_info = to_airtime_info(&txqi->txq);
if (air_info->v_t > air_sched->v_t &&
(!first || !airtime_catchup_v_t(air_sched, air_info->v_t, now)))
goto out;
if (!ieee80211_txq_airtime_check(hw, &txqi->txq)) {
first = false;
goto begin;
mac80211: Implement Airtime-based Queue Limit (AQL) In order for the Fq_CoDel algorithm integrated in mac80211 layer to operate effectively to control excessive queueing latency, the CoDel algorithm requires an accurate measure of how long packets stays in the queue, AKA sojourn time. The sojourn time measured at the mac80211 layer doesn't include queueing latency in the lower layer (firmware/hardware) and CoDel expects lower layer to have a short queue. However, most 802.11ac chipsets offload tasks such TX aggregation to firmware or hardware, thus have a deep lower layer queue. Without a mechanism to control the lower layer queue size, packets only stay in mac80211 layer transiently before being sent to firmware queue. As a result, the sojourn time measured by CoDel in the mac80211 layer is almost always lower than the CoDel latency target, hence CoDel does little to control the latency, even when the lower layer queue causes excessive latency. The Byte Queue Limits (BQL) mechanism is commonly used to address the similar issue with wired network interface. However, this method cannot be applied directly to the wireless network interface. "Bytes" is not a suitable measure of queue depth in the wireless network, as the data rate can vary dramatically from station to station in the same network, from a few Mbps to over Gbps. This patch implements an Airtime-based Queue Limit (AQL) to make CoDel work effectively with wireless drivers that utilized firmware/hardware offloading. AQL allows each txq to release just enough packets to the lower layer to form 1-2 large aggregations to keep hardware fully utilized and retains the rest of the frames in mac80211 layer to be controlled by the CoDel algorithm. Signed-off-by: Kan Yan <kyan@google.com> [ Toke: Keep API to set pending airtime internal, fix nits in commit msg ] Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20191119060610.76681-4-kyan@google.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-11-19 14:06:09 +08:00
}
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
air_sched->schedule_pos = node;
air_sched->last_schedule_activity = now;
ret = &txqi->txq;
out:
spin_unlock_bh(&air_sched->lock);
return ret;
}
EXPORT_SYMBOL(ieee80211_next_txq);
static void __ieee80211_insert_txq(struct rb_root_cached *root,
struct txq_info *txqi)
{
struct rb_node **new = &root->rb_root.rb_node;
struct airtime_info *old_air, *new_air;
struct rb_node *parent = NULL;
struct txq_info *__txqi;
bool leftmost = true;
while (*new) {
parent = *new;
__txqi = rb_entry(parent, struct txq_info, schedule_order);
old_air = to_airtime_info(&__txqi->txq);
new_air = to_airtime_info(&txqi->txq);
if (new_air->v_t <= old_air->v_t) {
new = &parent->rb_left;
} else {
new = &parent->rb_right;
leftmost = false;
}
}
mac80211: Implement Airtime-based Queue Limit (AQL) In order for the Fq_CoDel algorithm integrated in mac80211 layer to operate effectively to control excessive queueing latency, the CoDel algorithm requires an accurate measure of how long packets stays in the queue, AKA sojourn time. The sojourn time measured at the mac80211 layer doesn't include queueing latency in the lower layer (firmware/hardware) and CoDel expects lower layer to have a short queue. However, most 802.11ac chipsets offload tasks such TX aggregation to firmware or hardware, thus have a deep lower layer queue. Without a mechanism to control the lower layer queue size, packets only stay in mac80211 layer transiently before being sent to firmware queue. As a result, the sojourn time measured by CoDel in the mac80211 layer is almost always lower than the CoDel latency target, hence CoDel does little to control the latency, even when the lower layer queue causes excessive latency. The Byte Queue Limits (BQL) mechanism is commonly used to address the similar issue with wired network interface. However, this method cannot be applied directly to the wireless network interface. "Bytes" is not a suitable measure of queue depth in the wireless network, as the data rate can vary dramatically from station to station in the same network, from a few Mbps to over Gbps. This patch implements an Airtime-based Queue Limit (AQL) to make CoDel work effectively with wireless drivers that utilized firmware/hardware offloading. AQL allows each txq to release just enough packets to the lower layer to form 1-2 large aggregations to keep hardware fully utilized and retains the rest of the frames in mac80211 layer to be controlled by the CoDel algorithm. Signed-off-by: Kan Yan <kyan@google.com> [ Toke: Keep API to set pending airtime internal, fix nits in commit msg ] Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20191119060610.76681-4-kyan@google.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-11-19 14:06:09 +08:00
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
rb_link_node(&txqi->schedule_order, parent, new);
rb_insert_color_cached(&txqi->schedule_order, root, leftmost);
}
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
void ieee80211_resort_txq(struct ieee80211_hw *hw,
struct ieee80211_txq *txq)
{
struct airtime_info *air_info = to_airtime_info(txq);
struct ieee80211_local *local = hw_to_local(hw);
struct txq_info *txqi = to_txq_info(txq);
struct airtime_sched_info *air_sched;
mac80211: Implement Airtime-based Queue Limit (AQL) In order for the Fq_CoDel algorithm integrated in mac80211 layer to operate effectively to control excessive queueing latency, the CoDel algorithm requires an accurate measure of how long packets stays in the queue, AKA sojourn time. The sojourn time measured at the mac80211 layer doesn't include queueing latency in the lower layer (firmware/hardware) and CoDel expects lower layer to have a short queue. However, most 802.11ac chipsets offload tasks such TX aggregation to firmware or hardware, thus have a deep lower layer queue. Without a mechanism to control the lower layer queue size, packets only stay in mac80211 layer transiently before being sent to firmware queue. As a result, the sojourn time measured by CoDel in the mac80211 layer is almost always lower than the CoDel latency target, hence CoDel does little to control the latency, even when the lower layer queue causes excessive latency. The Byte Queue Limits (BQL) mechanism is commonly used to address the similar issue with wired network interface. However, this method cannot be applied directly to the wireless network interface. "Bytes" is not a suitable measure of queue depth in the wireless network, as the data rate can vary dramatically from station to station in the same network, from a few Mbps to over Gbps. This patch implements an Airtime-based Queue Limit (AQL) to make CoDel work effectively with wireless drivers that utilized firmware/hardware offloading. AQL allows each txq to release just enough packets to the lower layer to form 1-2 large aggregations to keep hardware fully utilized and retains the rest of the frames in mac80211 layer to be controlled by the CoDel algorithm. Signed-off-by: Kan Yan <kyan@google.com> [ Toke: Keep API to set pending airtime internal, fix nits in commit msg ] Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20191119060610.76681-4-kyan@google.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-11-19 14:06:09 +08:00
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
air_sched = &local->airtime[txq->ac];
mac80211: Implement Airtime-based Queue Limit (AQL) In order for the Fq_CoDel algorithm integrated in mac80211 layer to operate effectively to control excessive queueing latency, the CoDel algorithm requires an accurate measure of how long packets stays in the queue, AKA sojourn time. The sojourn time measured at the mac80211 layer doesn't include queueing latency in the lower layer (firmware/hardware) and CoDel expects lower layer to have a short queue. However, most 802.11ac chipsets offload tasks such TX aggregation to firmware or hardware, thus have a deep lower layer queue. Without a mechanism to control the lower layer queue size, packets only stay in mac80211 layer transiently before being sent to firmware queue. As a result, the sojourn time measured by CoDel in the mac80211 layer is almost always lower than the CoDel latency target, hence CoDel does little to control the latency, even when the lower layer queue causes excessive latency. The Byte Queue Limits (BQL) mechanism is commonly used to address the similar issue with wired network interface. However, this method cannot be applied directly to the wireless network interface. "Bytes" is not a suitable measure of queue depth in the wireless network, as the data rate can vary dramatically from station to station in the same network, from a few Mbps to over Gbps. This patch implements an Airtime-based Queue Limit (AQL) to make CoDel work effectively with wireless drivers that utilized firmware/hardware offloading. AQL allows each txq to release just enough packets to the lower layer to form 1-2 large aggregations to keep hardware fully utilized and retains the rest of the frames in mac80211 layer to be controlled by the CoDel algorithm. Signed-off-by: Kan Yan <kyan@google.com> [ Toke: Keep API to set pending airtime internal, fix nits in commit msg ] Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20191119060610.76681-4-kyan@google.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-11-19 14:06:09 +08:00
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
lockdep_assert_held(&air_sched->lock);
if (!RB_EMPTY_NODE(&txqi->schedule_order)) {
struct airtime_info *a_prev = NULL, *a_next = NULL;
struct txq_info *t_prev, *t_next;
struct rb_node *n_prev, *n_next;
/* Erasing a node can cause an expensive rebalancing operation,
* so we check the previous and next nodes first and only remove
* and re-insert if the current node is not already in the
* correct position.
*/
if ((n_prev = rb_prev(&txqi->schedule_order)) != NULL) {
t_prev = container_of(n_prev, struct txq_info,
schedule_order);
a_prev = to_airtime_info(&t_prev->txq);
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
}
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
if ((n_next = rb_next(&txqi->schedule_order)) != NULL) {
t_next = container_of(n_next, struct txq_info,
schedule_order);
a_next = to_airtime_info(&t_next->txq);
}
if ((!a_prev || a_prev->v_t <= air_info->v_t) &&
(!a_next || a_next->v_t > air_info->v_t))
return;
if (air_sched->schedule_pos == &txqi->schedule_order)
air_sched->schedule_pos = n_prev;
rb_erase_cached(&txqi->schedule_order,
&air_sched->active_txqs);
RB_CLEAR_NODE(&txqi->schedule_order);
__ieee80211_insert_txq(&air_sched->active_txqs, txqi);
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
}
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
}
void ieee80211_update_airtime_weight(struct ieee80211_local *local,
struct airtime_sched_info *air_sched,
u64 now, bool force)
{
struct airtime_info *air_info, *tmp;
u64 weight_sum = 0;
if (unlikely(!now))
now = ktime_get_coarse_boottime_ns();
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
lockdep_assert_held(&air_sched->lock);
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
if (!force && (air_sched->last_weight_update <
now - AIRTIME_ACTIVE_DURATION))
return;
list_for_each_entry_safe(air_info, tmp,
&air_sched->active_list, list) {
if (airtime_is_active(air_info, now))
weight_sum += air_info->weight;
else
list_del_init(&air_info->list);
}
airtime_weight_sum_set(air_sched, weight_sum);
air_sched->last_weight_update = now;
}
void ieee80211_schedule_txq(struct ieee80211_hw *hw,
struct ieee80211_txq *txq)
__acquires(txq_lock) __releases(txq_lock)
{
struct ieee80211_local *local = hw_to_local(hw);
struct txq_info *txqi = to_txq_info(txq);
struct airtime_sched_info *air_sched;
u64 now = ktime_get_coarse_boottime_ns();
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
struct airtime_info *air_info;
u8 ac = txq->ac;
bool was_active;
air_sched = &local->airtime[ac];
air_info = to_airtime_info(txq);
spin_lock_bh(&air_sched->lock);
was_active = airtime_is_active(air_info, now);
airtime_set_active(air_sched, air_info, now);
if (!RB_EMPTY_NODE(&txqi->schedule_order))
goto out;
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
/* If the station has been inactive for a while, catch up its v_t so it
* doesn't get indefinite priority; see comment above the definition of
* AIRTIME_MAX_BEHIND.
*/
if ((!was_active && air_info->v_t < air_sched->v_t) ||
air_info->v_t < air_sched->v_t - AIRTIME_MAX_BEHIND)
air_info->v_t = air_sched->v_t;
ieee80211_update_airtime_weight(local, air_sched, now, !was_active);
__ieee80211_insert_txq(&air_sched->active_txqs, txqi);
out:
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
spin_unlock_bh(&air_sched->lock);
}
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
EXPORT_SYMBOL(ieee80211_schedule_txq);
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
static void __ieee80211_unschedule_txq(struct ieee80211_hw *hw,
struct ieee80211_txq *txq,
bool purge)
{
struct ieee80211_local *local = hw_to_local(hw);
struct txq_info *txqi = to_txq_info(txq);
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
struct airtime_sched_info *air_sched;
struct airtime_info *air_info;
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
air_sched = &local->airtime[txq->ac];
air_info = to_airtime_info(&txqi->txq);
lockdep_assert_held(&air_sched->lock);
if (purge) {
list_del_init(&air_info->list);
ieee80211_update_airtime_weight(local, air_sched, 0, true);
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
}
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
if (RB_EMPTY_NODE(&txqi->schedule_order))
return;
if (air_sched->schedule_pos == &txqi->schedule_order)
air_sched->schedule_pos = rb_prev(&txqi->schedule_order);
if (!purge)
airtime_set_active(air_sched, air_info,
ktime_get_coarse_boottime_ns());
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
rb_erase_cached(&txqi->schedule_order,
&air_sched->active_txqs);
RB_CLEAR_NODE(&txqi->schedule_order);
}
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
void ieee80211_unschedule_txq(struct ieee80211_hw *hw,
struct ieee80211_txq *txq,
bool purge)
__acquires(txq_lock) __releases(txq_lock)
{
struct ieee80211_local *local = hw_to_local(hw);
spin_lock_bh(&local->airtime[txq->ac].lock);
__ieee80211_unschedule_txq(hw, txq, purge);
spin_unlock_bh(&local->airtime[txq->ac].lock);
}
void ieee80211_return_txq(struct ieee80211_hw *hw,
struct ieee80211_txq *txq, bool force)
{
struct ieee80211_local *local = hw_to_local(hw);
struct txq_info *txqi = to_txq_info(txq);
spin_lock_bh(&local->airtime[txq->ac].lock);
if (!RB_EMPTY_NODE(&txqi->schedule_order) && !force &&
!txq_has_queue(txq))
__ieee80211_unschedule_txq(hw, txq, false);
spin_unlock_bh(&local->airtime[txq->ac].lock);
}
EXPORT_SYMBOL(ieee80211_return_txq);
DEFINE_STATIC_KEY_FALSE(aql_disable);
mac80211: Implement Airtime-based Queue Limit (AQL) In order for the Fq_CoDel algorithm integrated in mac80211 layer to operate effectively to control excessive queueing latency, the CoDel algorithm requires an accurate measure of how long packets stays in the queue, AKA sojourn time. The sojourn time measured at the mac80211 layer doesn't include queueing latency in the lower layer (firmware/hardware) and CoDel expects lower layer to have a short queue. However, most 802.11ac chipsets offload tasks such TX aggregation to firmware or hardware, thus have a deep lower layer queue. Without a mechanism to control the lower layer queue size, packets only stay in mac80211 layer transiently before being sent to firmware queue. As a result, the sojourn time measured by CoDel in the mac80211 layer is almost always lower than the CoDel latency target, hence CoDel does little to control the latency, even when the lower layer queue causes excessive latency. The Byte Queue Limits (BQL) mechanism is commonly used to address the similar issue with wired network interface. However, this method cannot be applied directly to the wireless network interface. "Bytes" is not a suitable measure of queue depth in the wireless network, as the data rate can vary dramatically from station to station in the same network, from a few Mbps to over Gbps. This patch implements an Airtime-based Queue Limit (AQL) to make CoDel work effectively with wireless drivers that utilized firmware/hardware offloading. AQL allows each txq to release just enough packets to the lower layer to form 1-2 large aggregations to keep hardware fully utilized and retains the rest of the frames in mac80211 layer to be controlled by the CoDel algorithm. Signed-off-by: Kan Yan <kyan@google.com> [ Toke: Keep API to set pending airtime internal, fix nits in commit msg ] Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20191119060610.76681-4-kyan@google.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-11-19 14:06:09 +08:00
bool ieee80211_txq_airtime_check(struct ieee80211_hw *hw,
struct ieee80211_txq *txq)
{
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
struct airtime_info *air_info = to_airtime_info(txq);
mac80211: Implement Airtime-based Queue Limit (AQL) In order for the Fq_CoDel algorithm integrated in mac80211 layer to operate effectively to control excessive queueing latency, the CoDel algorithm requires an accurate measure of how long packets stays in the queue, AKA sojourn time. The sojourn time measured at the mac80211 layer doesn't include queueing latency in the lower layer (firmware/hardware) and CoDel expects lower layer to have a short queue. However, most 802.11ac chipsets offload tasks such TX aggregation to firmware or hardware, thus have a deep lower layer queue. Without a mechanism to control the lower layer queue size, packets only stay in mac80211 layer transiently before being sent to firmware queue. As a result, the sojourn time measured by CoDel in the mac80211 layer is almost always lower than the CoDel latency target, hence CoDel does little to control the latency, even when the lower layer queue causes excessive latency. The Byte Queue Limits (BQL) mechanism is commonly used to address the similar issue with wired network interface. However, this method cannot be applied directly to the wireless network interface. "Bytes" is not a suitable measure of queue depth in the wireless network, as the data rate can vary dramatically from station to station in the same network, from a few Mbps to over Gbps. This patch implements an Airtime-based Queue Limit (AQL) to make CoDel work effectively with wireless drivers that utilized firmware/hardware offloading. AQL allows each txq to release just enough packets to the lower layer to form 1-2 large aggregations to keep hardware fully utilized and retains the rest of the frames in mac80211 layer to be controlled by the CoDel algorithm. Signed-off-by: Kan Yan <kyan@google.com> [ Toke: Keep API to set pending airtime internal, fix nits in commit msg ] Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20191119060610.76681-4-kyan@google.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-11-19 14:06:09 +08:00
struct ieee80211_local *local = hw_to_local(hw);
if (!wiphy_ext_feature_isset(local->hw.wiphy, NL80211_EXT_FEATURE_AQL))
mac80211: Implement Airtime-based Queue Limit (AQL) In order for the Fq_CoDel algorithm integrated in mac80211 layer to operate effectively to control excessive queueing latency, the CoDel algorithm requires an accurate measure of how long packets stays in the queue, AKA sojourn time. The sojourn time measured at the mac80211 layer doesn't include queueing latency in the lower layer (firmware/hardware) and CoDel expects lower layer to have a short queue. However, most 802.11ac chipsets offload tasks such TX aggregation to firmware or hardware, thus have a deep lower layer queue. Without a mechanism to control the lower layer queue size, packets only stay in mac80211 layer transiently before being sent to firmware queue. As a result, the sojourn time measured by CoDel in the mac80211 layer is almost always lower than the CoDel latency target, hence CoDel does little to control the latency, even when the lower layer queue causes excessive latency. The Byte Queue Limits (BQL) mechanism is commonly used to address the similar issue with wired network interface. However, this method cannot be applied directly to the wireless network interface. "Bytes" is not a suitable measure of queue depth in the wireless network, as the data rate can vary dramatically from station to station in the same network, from a few Mbps to over Gbps. This patch implements an Airtime-based Queue Limit (AQL) to make CoDel work effectively with wireless drivers that utilized firmware/hardware offloading. AQL allows each txq to release just enough packets to the lower layer to form 1-2 large aggregations to keep hardware fully utilized and retains the rest of the frames in mac80211 layer to be controlled by the CoDel algorithm. Signed-off-by: Kan Yan <kyan@google.com> [ Toke: Keep API to set pending airtime internal, fix nits in commit msg ] Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20191119060610.76681-4-kyan@google.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-11-19 14:06:09 +08:00
return true;
if (static_branch_unlikely(&aql_disable))
return true;
mac80211: Implement Airtime-based Queue Limit (AQL) In order for the Fq_CoDel algorithm integrated in mac80211 layer to operate effectively to control excessive queueing latency, the CoDel algorithm requires an accurate measure of how long packets stays in the queue, AKA sojourn time. The sojourn time measured at the mac80211 layer doesn't include queueing latency in the lower layer (firmware/hardware) and CoDel expects lower layer to have a short queue. However, most 802.11ac chipsets offload tasks such TX aggregation to firmware or hardware, thus have a deep lower layer queue. Without a mechanism to control the lower layer queue size, packets only stay in mac80211 layer transiently before being sent to firmware queue. As a result, the sojourn time measured by CoDel in the mac80211 layer is almost always lower than the CoDel latency target, hence CoDel does little to control the latency, even when the lower layer queue causes excessive latency. The Byte Queue Limits (BQL) mechanism is commonly used to address the similar issue with wired network interface. However, this method cannot be applied directly to the wireless network interface. "Bytes" is not a suitable measure of queue depth in the wireless network, as the data rate can vary dramatically from station to station in the same network, from a few Mbps to over Gbps. This patch implements an Airtime-based Queue Limit (AQL) to make CoDel work effectively with wireless drivers that utilized firmware/hardware offloading. AQL allows each txq to release just enough packets to the lower layer to form 1-2 large aggregations to keep hardware fully utilized and retains the rest of the frames in mac80211 layer to be controlled by the CoDel algorithm. Signed-off-by: Kan Yan <kyan@google.com> [ Toke: Keep API to set pending airtime internal, fix nits in commit msg ] Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20191119060610.76681-4-kyan@google.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-11-19 14:06:09 +08:00
if (!txq->sta)
return true;
if (unlikely(txq->tid == IEEE80211_NUM_TIDS))
return true;
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
if (atomic_read(&air_info->aql_tx_pending) < air_info->aql_limit_low)
mac80211: Implement Airtime-based Queue Limit (AQL) In order for the Fq_CoDel algorithm integrated in mac80211 layer to operate effectively to control excessive queueing latency, the CoDel algorithm requires an accurate measure of how long packets stays in the queue, AKA sojourn time. The sojourn time measured at the mac80211 layer doesn't include queueing latency in the lower layer (firmware/hardware) and CoDel expects lower layer to have a short queue. However, most 802.11ac chipsets offload tasks such TX aggregation to firmware or hardware, thus have a deep lower layer queue. Without a mechanism to control the lower layer queue size, packets only stay in mac80211 layer transiently before being sent to firmware queue. As a result, the sojourn time measured by CoDel in the mac80211 layer is almost always lower than the CoDel latency target, hence CoDel does little to control the latency, even when the lower layer queue causes excessive latency. The Byte Queue Limits (BQL) mechanism is commonly used to address the similar issue with wired network interface. However, this method cannot be applied directly to the wireless network interface. "Bytes" is not a suitable measure of queue depth in the wireless network, as the data rate can vary dramatically from station to station in the same network, from a few Mbps to over Gbps. This patch implements an Airtime-based Queue Limit (AQL) to make CoDel work effectively with wireless drivers that utilized firmware/hardware offloading. AQL allows each txq to release just enough packets to the lower layer to form 1-2 large aggregations to keep hardware fully utilized and retains the rest of the frames in mac80211 layer to be controlled by the CoDel algorithm. Signed-off-by: Kan Yan <kyan@google.com> [ Toke: Keep API to set pending airtime internal, fix nits in commit msg ] Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20191119060610.76681-4-kyan@google.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-11-19 14:06:09 +08:00
return true;
if (atomic_read(&local->aql_total_pending_airtime) <
local->aql_threshold &&
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
atomic_read(&air_info->aql_tx_pending) < air_info->aql_limit_high)
mac80211: Implement Airtime-based Queue Limit (AQL) In order for the Fq_CoDel algorithm integrated in mac80211 layer to operate effectively to control excessive queueing latency, the CoDel algorithm requires an accurate measure of how long packets stays in the queue, AKA sojourn time. The sojourn time measured at the mac80211 layer doesn't include queueing latency in the lower layer (firmware/hardware) and CoDel expects lower layer to have a short queue. However, most 802.11ac chipsets offload tasks such TX aggregation to firmware or hardware, thus have a deep lower layer queue. Without a mechanism to control the lower layer queue size, packets only stay in mac80211 layer transiently before being sent to firmware queue. As a result, the sojourn time measured by CoDel in the mac80211 layer is almost always lower than the CoDel latency target, hence CoDel does little to control the latency, even when the lower layer queue causes excessive latency. The Byte Queue Limits (BQL) mechanism is commonly used to address the similar issue with wired network interface. However, this method cannot be applied directly to the wireless network interface. "Bytes" is not a suitable measure of queue depth in the wireless network, as the data rate can vary dramatically from station to station in the same network, from a few Mbps to over Gbps. This patch implements an Airtime-based Queue Limit (AQL) to make CoDel work effectively with wireless drivers that utilized firmware/hardware offloading. AQL allows each txq to release just enough packets to the lower layer to form 1-2 large aggregations to keep hardware fully utilized and retains the rest of the frames in mac80211 layer to be controlled by the CoDel algorithm. Signed-off-by: Kan Yan <kyan@google.com> [ Toke: Keep API to set pending airtime internal, fix nits in commit msg ] Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20191119060610.76681-4-kyan@google.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2019-11-19 14:06:09 +08:00
return true;
return false;
}
EXPORT_SYMBOL(ieee80211_txq_airtime_check);
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
bool ieee80211_txq_may_transmit(struct ieee80211_hw *hw,
struct ieee80211_txq *txq)
{
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
struct txq_info *first_txqi = NULL, *txqi = to_txq_info(txq);
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
struct ieee80211_local *local = hw_to_local(hw);
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
struct airtime_sched_info *air_sched;
struct airtime_info *air_info;
struct rb_node *node = NULL;
bool ret = false;
u64 now;
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
if (!ieee80211_txq_airtime_check(hw, txq))
return false;
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
air_sched = &local->airtime[txq->ac];
spin_lock_bh(&air_sched->lock);
if (RB_EMPTY_NODE(&txqi->schedule_order))
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
goto out;
now = ktime_get_coarse_boottime_ns();
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
/* Like in ieee80211_next_txq(), make sure the first station in the
* scheduling order is eligible for transmission to avoid starvation.
*/
node = rb_first_cached(&air_sched->active_txqs);
if (node) {
first_txqi = container_of(node, struct txq_info,
schedule_order);
air_info = to_airtime_info(&first_txqi->txq);
if (air_sched->v_t < air_info->v_t)
airtime_catchup_v_t(air_sched, air_info->v_t, now);
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
}
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
air_info = to_airtime_info(&txqi->txq);
if (air_info->v_t <= air_sched->v_t) {
air_sched->last_schedule_activity = now;
ret = true;
}
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
out:
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
spin_unlock_bh(&air_sched->lock);
return ret;
mac80211: Add airtime accounting and scheduling to TXQs This adds airtime accounting and scheduling to the mac80211 TXQ scheduler. A new callback, ieee80211_sta_register_airtime(), is added that drivers can call to report airtime usage for stations. When airtime information is present, mac80211 will schedule TXQs (through ieee80211_next_txq()) in a way that enforces airtime fairness between active stations. This scheduling works the same way as the ath9k in-driver airtime fairness scheduling. If no airtime usage is reported by the driver, the scheduler will default to round-robin scheduling. For drivers that don't control TXQ scheduling in software, a new API function, ieee80211_txq_may_transmit(), is added which the driver can use to check if the TXQ is eligible for transmission, or should be throttled to enforce fairness. Calls to this function must also be enclosed in ieee80211_txq_schedule_{start,end}() calls to ensure proper locking. The API ieee80211_txq_may_transmit() also ensures that TXQ list will be aligned aginst driver's own round-robin scheduler list. i.e it rotates the TXQ list till it makes the requested node becomes the first entry in TXQ list. Thus both the TXQ list and driver's list are in sync. Co-developed-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Louie Lu <git@louie.lu> [added debugfs write op to reset airtime counter] Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk> Signed-off-by: Rajkumar Manoharan <rmanohar@codeaurora.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2018-12-19 09:02:08 +08:00
}
EXPORT_SYMBOL(ieee80211_txq_may_transmit);
void ieee80211_txq_schedule_start(struct ieee80211_hw *hw, u8 ac)
{
struct ieee80211_local *local = hw_to_local(hw);
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
struct airtime_sched_info *air_sched = &local->airtime[ac];
mac80211: Switch to a virtual time-based airtime scheduler This switches the airtime scheduler in mac80211 to use a virtual time-based scheduler instead of the round-robin scheduler used before. This has a couple of advantages: - No need to sync up the round-robin scheduler in firmware/hardware with the round-robin airtime scheduler. - If several stations are eligible for transmission we can schedule both of them; no need to hard-block the scheduling rotation until the head of the queue has used up its quantum. - The check of whether a station is eligible for transmission becomes simpler (in ieee80211_txq_may_transmit()). The drawback is that scheduling becomes slightly more expensive, as we need to maintain an rbtree of TXQs sorted by virtual time. This means that ieee80211_register_airtime() becomes O(logN) in the number of currently scheduled TXQs because it can change the order of the scheduled stations. We mitigate this overhead by only resorting when a station changes position in the tree, and hopefully N rarely grows too big (it's only TXQs currently backlogged, not all associated stations), so it shouldn't be too big of an issue. To prevent divisions in the fast path, we maintain both station sums and pre-computed reciprocals of the sums. This turns the fast-path operation into a multiplication, with divisions only happening as the number of active stations change (to re-compute the current sum of all active station weights). To prevent this re-computation of the reciprocal from happening too frequently, we use a time-based notion of station activity, instead of updating the weight every time a station gets scheduled or de-scheduled. As queues can oscillate between empty and occupied quite frequently, this can significantly cut down on the number of re-computations. It also has the added benefit of making the station airtime calculation independent on whether the queue happened to have drained at the time an airtime value was accounted. Co-developed-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Yibo Zhao <yiboz@codeaurora.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://lore.kernel.org/r/20210623134755.235545-1-toke@redhat.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2021-06-23 21:47:55 +08:00
spin_lock_bh(&air_sched->lock);
air_sched->schedule_pos = NULL;
spin_unlock_bh(&air_sched->lock);
}
EXPORT_SYMBOL(ieee80211_txq_schedule_start);
void __ieee80211_subif_start_xmit(struct sk_buff *skb,
struct net_device *dev,
u32 info_flags,
u32 ctrl_flags,
u64 *cookie)
{
struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev);
struct ieee80211_local *local = sdata->local;
struct sta_info *sta;
struct sk_buff *next;
int len = skb->len;
if (unlikely(skb->len < ETH_HLEN)) {
kfree_skb(skb);
return;
}
rcu_read_lock();
if (ieee80211_lookup_ra_sta(sdata, skb, &sta))
goto out_free;
if (IS_ERR(sta))
sta = NULL;
if (local->ops->wake_tx_queue) {
u16 queue = __ieee80211_select_queue(sdata, sta, skb);
skb_set_queue_mapping(skb, queue);
skb_get_hash(skb);
}
ieee80211_aggr_check(sdata, sta, skb);
sk_pacing_shift_update(skb->sk, sdata->local->hw.tx_sk_pacing_shift);
if (sta) {
struct ieee80211_fast_tx *fast_tx;
fast_tx = rcu_dereference(sta->fast_tx);
if (fast_tx &&
ieee80211_xmit_fast(sdata, sta, fast_tx, skb))
goto out;
}
if (skb_is_gso(skb)) {
struct sk_buff *segs;
segs = skb_gso_segment(skb, 0);
if (IS_ERR(segs)) {
goto out_free;
} else if (segs) {
consume_skb(skb);
skb = segs;
}
} else {
/* we cannot process non-linear frames on this path */
if (skb_linearize(skb))
goto out_free;
/* the frame could be fragmented, software-encrypted, and other
* things so we cannot really handle checksum offload with it -
* fix it up in software before we handle anything else.
*/
if (skb->ip_summed == CHECKSUM_PARTIAL) {
skb_set_transport_header(skb,
skb_checksum_start_offset(skb));
if (skb_checksum_help(skb))
goto out_free;
}
}
skb_list_walk_safe(skb, skb, next) {
skb_mark_not_on_list(skb);
if (skb->protocol == sdata->control_port_protocol)
ctrl_flags |= IEEE80211_TX_CTRL_SKIP_MPATH_LOOKUP;
skb = ieee80211_build_hdr(sdata, skb, info_flags,
sta, ctrl_flags, cookie);
if (IS_ERR(skb)) {
kfree_skb_list(next);
goto out;
}
dev_sw_netstats_tx_add(dev, 1, skb->len);
ieee80211_xmit(sdata, sta, skb);
}
goto out;
out_free:
kfree_skb(skb);
len = 0;
out:
if (len)
ieee80211_tpt_led_trig_tx(local, len);
rcu_read_unlock();
}
static int ieee80211_change_da(struct sk_buff *skb, struct sta_info *sta)
{
struct ethhdr *eth;
int err;
err = skb_ensure_writable(skb, ETH_HLEN);
if (unlikely(err))
return err;
eth = (void *)skb->data;
ether_addr_copy(eth->h_dest, sta->sta.addr);
return 0;
}
static bool ieee80211_multicast_to_unicast(struct sk_buff *skb,
struct net_device *dev)
{
struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev);
const struct ethhdr *eth = (void *)skb->data;
const struct vlan_ethhdr *ethvlan = (void *)skb->data;
__be16 ethertype;
if (likely(!is_multicast_ether_addr(eth->h_dest)))
return false;
switch (sdata->vif.type) {
case NL80211_IFTYPE_AP_VLAN:
if (sdata->u.vlan.sta)
return false;
if (sdata->wdev.use_4addr)
return false;
fallthrough;
case NL80211_IFTYPE_AP:
/* check runtime toggle for this bss */
if (!sdata->bss->multicast_to_unicast)
return false;
break;
default:
return false;
}
/* multicast to unicast conversion only for some payload */
ethertype = eth->h_proto;
if (ethertype == htons(ETH_P_8021Q) && skb->len >= VLAN_ETH_HLEN)
ethertype = ethvlan->h_vlan_encapsulated_proto;
switch (ethertype) {
case htons(ETH_P_ARP):
case htons(ETH_P_IP):
case htons(ETH_P_IPV6):
break;
default:
return false;
}
return true;
}
static void
ieee80211_convert_to_unicast(struct sk_buff *skb, struct net_device *dev,
struct sk_buff_head *queue)
{
struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev);
struct ieee80211_local *local = sdata->local;
const struct ethhdr *eth = (struct ethhdr *)skb->data;
struct sta_info *sta, *first = NULL;
struct sk_buff *cloned_skb;
rcu_read_lock();
list_for_each_entry_rcu(sta, &local->sta_list, list) {
if (sdata != sta->sdata)
/* AP-VLAN mismatch */
continue;
if (unlikely(ether_addr_equal(eth->h_source, sta->sta.addr)))
/* do not send back to source */
continue;
if (!first) {
first = sta;
continue;
}
cloned_skb = skb_clone(skb, GFP_ATOMIC);
if (!cloned_skb)
goto multicast;
if (unlikely(ieee80211_change_da(cloned_skb, sta))) {
dev_kfree_skb(cloned_skb);
goto multicast;
}
__skb_queue_tail(queue, cloned_skb);
}
if (likely(first)) {
if (unlikely(ieee80211_change_da(skb, first)))
goto multicast;
__skb_queue_tail(queue, skb);
} else {
/* no STA connected, drop */
kfree_skb(skb);
skb = NULL;
}
goto out;
multicast:
__skb_queue_purge(queue);
__skb_queue_tail(queue, skb);
out:
rcu_read_unlock();
}
/**
* ieee80211_subif_start_xmit - netif start_xmit function for 802.3 vifs
* @skb: packet to be sent
* @dev: incoming interface
*
* On failure skb will be freed.
*/
netdev_tx_t ieee80211_subif_start_xmit(struct sk_buff *skb,
struct net_device *dev)
{
if (unlikely(ieee80211_multicast_to_unicast(skb, dev))) {
struct sk_buff_head queue;
__skb_queue_head_init(&queue);
ieee80211_convert_to_unicast(skb, dev, &queue);
while ((skb = __skb_dequeue(&queue)))
__ieee80211_subif_start_xmit(skb, dev, 0, 0, NULL);
} else {
__ieee80211_subif_start_xmit(skb, dev, 0, 0, NULL);
}
return NETDEV_TX_OK;
}
static bool ieee80211_tx_8023(struct ieee80211_sub_if_data *sdata,
struct sk_buff *skb, struct sta_info *sta,
bool txpending)
{
struct ieee80211_local *local = sdata->local;
struct ieee80211_tx_control control = {};
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
struct ieee80211_sta *pubsta = NULL;
unsigned long flags;
int q = info->hw_queue;
if (sta)
sk_pacing_shift_update(skb->sk, local->hw.tx_sk_pacing_shift);
ieee80211_tpt_led_trig_tx(local, skb->len);
if (ieee80211_queue_skb(local, sdata, sta, skb))
return true;
spin_lock_irqsave(&local->queue_stop_reason_lock, flags);
if (local->queue_stop_reasons[q] ||
(!txpending && !skb_queue_empty(&local->pending[q]))) {
if (txpending)
skb_queue_head(&local->pending[q], skb);
else
skb_queue_tail(&local->pending[q], skb);
spin_unlock_irqrestore(&local->queue_stop_reason_lock, flags);
return false;
}
spin_unlock_irqrestore(&local->queue_stop_reason_lock, flags);
if (sta && sta->uploaded)
pubsta = &sta->sta;
control.sta = pubsta;
drv_tx(local, &control, skb);
return true;
}
static void ieee80211_8023_xmit(struct ieee80211_sub_if_data *sdata,
struct net_device *dev, struct sta_info *sta,
struct ieee80211_key *key, struct sk_buff *skb)
{
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
struct ieee80211_local *local = sdata->local;
struct tid_ampdu_tx *tid_tx;
u8 tid;
if (local->ops->wake_tx_queue) {
u16 queue = __ieee80211_select_queue(sdata, sta, skb);
skb_set_queue_mapping(skb, queue);
skb_get_hash(skb);
}
if (unlikely(test_bit(SCAN_SW_SCANNING, &local->scanning)) &&
test_bit(SDATA_STATE_OFFCHANNEL, &sdata->state))
goto out_free;
memset(info, 0, sizeof(*info));
ieee80211_aggr_check(sdata, sta, skb);
tid = skb->priority & IEEE80211_QOS_CTL_TAG1D_MASK;
tid_tx = rcu_dereference(sta->ampdu_mlme.tid_tx[tid]);
if (tid_tx) {
if (!test_bit(HT_AGG_STATE_OPERATIONAL, &tid_tx->state)) {
/* fall back to non-offload slow path */
__ieee80211_subif_start_xmit(skb, dev, 0, 0, NULL);
return;
}
info->flags |= IEEE80211_TX_CTL_AMPDU;
if (tid_tx->timeout)
tid_tx->last_tx = jiffies;
}
if (unlikely(skb->sk &&
skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS))
info->ack_frame_id = ieee80211_store_ack_skb(local, skb,
&info->flags, NULL);
info->hw_queue = sdata->vif.hw_queue[skb_get_queue_mapping(skb)];
dev_sw_netstats_tx_add(dev, 1, skb->len);
mac80211: prepare sta handling for MLO support Currently in mac80211 each STA object is represented using sta_info datastructure with the associated STA specific information and drivers access ieee80211_sta part of it. With MLO (Multi Link Operation) support being added in 802.11be standard, though the association is logically with a single Multi Link capable STA, at the physical level communication can happen via different advertised links (uniquely identified by Channel, operating class, BSSID) and hence the need to handle multiple link STA parameters within a composite sta_info object called the MLD STA. The different link STA part of MLD STA are identified using the link address which can be same or different as the MLD STA address and unique link id based on the link vif. To support extension of such a model, the sta_info datastructure is modified to hold multiple link STA objects with link specific params currently within sta_info moved to this new structure. Similarly this is done for ieee80211_sta as well which will be accessed within mac80211 as well as by drivers, hence trivial driver changes are expected to support this. For current non MLO supported drivers, only one link STA is present and link information is accessed via 'deflink' member. For MLO drivers, we still need to define the APIs etc. to get the correct link ID and access the correct part of the station info. Currently in mac80211, all link STA info are accessed directly via deflink. These will be updated to access via link pointers indexed by link id with MLO support patches, with link id being 0 for non MLO supported cases. Except for couple of macro related changes, below spatch takes care of updating mac80211 and driver code to access to the link STA info via deflink. @ieee80211_sta@ struct ieee80211_sta *s; struct sta_info *si; identifier var = {supp_rates, ht_cap, vht_cap, he_cap, he_6ghz_capa, eht_cap, rx_nss, bandwidth, txpwr}; @@ ( s-> - var + deflink.var | si->sta. - var + deflink.var ) @sta_info@ struct sta_info *si; identifier var = {gtk, pcpu_rx_stats, rx_stats, rx_stats_avg, status_stats, tx_stats, cur_max_bandwidth}; @@ ( si-> - var + deflink.var ) Signed-off-by: Sriram R <quic_srirrama@quicinc.com> Link: https://lore.kernel.org/r/1649086883-13246-1-git-send-email-quic_srirrama@quicinc.com [remove MLO-drivers notes from commit message, not clear yet; run spatch] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2022-04-04 23:41:23 +08:00
sta->deflink.tx_stats.bytes[skb_get_queue_mapping(skb)] += skb->len;
sta->deflink.tx_stats.packets[skb_get_queue_mapping(skb)]++;
if (sdata->vif.type == NL80211_IFTYPE_AP_VLAN)
sdata = container_of(sdata->bss,
struct ieee80211_sub_if_data, u.ap);
info->flags |= IEEE80211_TX_CTL_HW_80211_ENCAP;
info->control.vif = &sdata->vif;
if (key)
info->control.hw_key = &key->conf;
ieee80211_tx_8023(sdata, skb, sta, false);
return;
out_free:
kfree_skb(skb);
}
netdev_tx_t ieee80211_subif_start_xmit_8023(struct sk_buff *skb,
struct net_device *dev)
{
struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev);
struct ethhdr *ehdr = (struct ethhdr *)skb->data;
struct ieee80211_key *key;
struct sta_info *sta;
if (unlikely(skb->len < ETH_HLEN)) {
kfree_skb(skb);
return NETDEV_TX_OK;
}
rcu_read_lock();
if (ieee80211_lookup_ra_sta(sdata, skb, &sta)) {
kfree_skb(skb);
goto out;
}
if (unlikely(IS_ERR_OR_NULL(sta) || !sta->uploaded ||
!test_sta_flag(sta, WLAN_STA_AUTHORIZED) ||
sdata->control_port_protocol == ehdr->h_proto))
goto skip_offload;
key = rcu_dereference(sta->ptk[sta->ptk_idx]);
if (!key)
key = rcu_dereference(sdata->default_unicast_key);
if (key && (!(key->flags & KEY_FLAG_UPLOADED_TO_HARDWARE) ||
key->conf.cipher == WLAN_CIPHER_SUITE_TKIP))
goto skip_offload;
ieee80211_8023_xmit(sdata, dev, sta, key, skb);
goto out;
skip_offload:
ieee80211_subif_start_xmit(skb, dev);
out:
rcu_read_unlock();
return NETDEV_TX_OK;
}
struct sk_buff *
ieee80211_build_data_template(struct ieee80211_sub_if_data *sdata,
struct sk_buff *skb, u32 info_flags)
{
struct ieee80211_hdr *hdr;
struct ieee80211_tx_data tx = {
.local = sdata->local,
.sdata = sdata,
};
struct sta_info *sta;
rcu_read_lock();
if (ieee80211_lookup_ra_sta(sdata, skb, &sta)) {
kfree_skb(skb);
skb = ERR_PTR(-EINVAL);
goto out;
}
skb = ieee80211_build_hdr(sdata, skb, info_flags, sta, 0, NULL);
if (IS_ERR(skb))
goto out;
hdr = (void *)skb->data;
tx.sta = sta_info_get(sdata, hdr->addr1);
tx.skb = skb;
if (ieee80211_tx_h_select_key(&tx) != TX_CONTINUE) {
rcu_read_unlock();
kfree_skb(skb);
return ERR_PTR(-EINVAL);
}
out:
rcu_read_unlock();
return skb;
}
/*
* ieee80211_clear_tx_pending may not be called in a context where
* it is possible that it packets could come in again.
*/
void ieee80211_clear_tx_pending(struct ieee80211_local *local)
{
struct sk_buff *skb;
int i;
for (i = 0; i < local->hw.queues; i++) {
while ((skb = skb_dequeue(&local->pending[i])) != NULL)
ieee80211_free_txskb(&local->hw, skb);
}
}
/*
* Returns false if the frame couldn't be transmitted but was queued instead,
* which in this case means re-queued -- take as an indication to stop sending
* more pending frames.
*/
static bool ieee80211_tx_pending_skb(struct ieee80211_local *local,
struct sk_buff *skb)
{
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
struct ieee80211_sub_if_data *sdata;
struct sta_info *sta;
struct ieee80211_hdr *hdr;
bool result;
struct ieee80211_chanctx_conf *chanctx_conf;
sdata = vif_to_sdata(info->control.vif);
if (info->control.flags & IEEE80211_TX_INTCFL_NEED_TXPROCESSING) {
chanctx_conf = rcu_dereference(sdata->vif.bss_conf.chanctx_conf);
if (unlikely(!chanctx_conf)) {
dev_kfree_skb(skb);
return true;
}
info->band = chanctx_conf->def.chan->band;
result = ieee80211_tx(sdata, NULL, skb, true);
} else if (info->flags & IEEE80211_TX_CTL_HW_80211_ENCAP) {
if (ieee80211_lookup_ra_sta(sdata, skb, &sta)) {
dev_kfree_skb(skb);
return true;
}
if (IS_ERR(sta) || (sta && !sta->uploaded))
sta = NULL;
result = ieee80211_tx_8023(sdata, skb, sta, true);
} else {
struct sk_buff_head skbs;
__skb_queue_head_init(&skbs);
__skb_queue_tail(&skbs, skb);
hdr = (struct ieee80211_hdr *)skb->data;
sta = sta_info_get(sdata, hdr->addr1);
result = __ieee80211_tx(local, &skbs, sta, true);
}
return result;
}
/*
* Transmit all pending packets. Called from tasklet.
*/
void ieee80211_tx_pending(struct tasklet_struct *t)
{
struct ieee80211_local *local = from_tasklet(local, t,
tx_pending_tasklet);
unsigned long flags;
int i;
bool txok;
rcu_read_lock();
spin_lock_irqsave(&local->queue_stop_reason_lock, flags);
for (i = 0; i < local->hw.queues; i++) {
/*
* If queue is stopped by something other than due to pending
* frames, or we have no pending frames, proceed to next queue.
*/
if (local->queue_stop_reasons[i] ||
skb_queue_empty(&local->pending[i]))
continue;
while (!skb_queue_empty(&local->pending[i])) {
struct sk_buff *skb = __skb_dequeue(&local->pending[i]);
struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb);
if (WARN_ON(!info->control.vif)) {
ieee80211_free_txskb(&local->hw, skb);
continue;
}
spin_unlock_irqrestore(&local->queue_stop_reason_lock,
flags);
txok = ieee80211_tx_pending_skb(local, skb);
spin_lock_irqsave(&local->queue_stop_reason_lock,
flags);
if (!txok)
break;
}
if (skb_queue_empty(&local->pending[i]))
ieee80211_propagate_queue_wake(local, i);
}
spin_unlock_irqrestore(&local->queue_stop_reason_lock, flags);
rcu_read_unlock();
}
/* functions for drivers to get certain frames */
static void __ieee80211_beacon_add_tim(struct ieee80211_sub_if_data *sdata,
struct ps_data *ps, struct sk_buff *skb,
bool is_template)
{
u8 *pos, *tim;
int aid0 = 0;
int i, have_bits = 0, n1, n2;
/* Generate bitmap for TIM only if there are any STAs in power save
* mode. */
if (atomic_read(&ps->num_sta_ps) > 0)
/* in the hope that this is faster than
* checking byte-for-byte */
have_bits = !bitmap_empty((unsigned long *)ps->tim,
IEEE80211_MAX_AID+1);
if (!is_template) {
if (ps->dtim_count == 0)
ps->dtim_count = sdata->vif.bss_conf.dtim_period - 1;
else
ps->dtim_count--;
}
tim = pos = skb_put(skb, 6);
*pos++ = WLAN_EID_TIM;
*pos++ = 4;
*pos++ = ps->dtim_count;
*pos++ = sdata->vif.bss_conf.dtim_period;
if (ps->dtim_count == 0 && !skb_queue_empty(&ps->bc_buf))
aid0 = 1;
ps->dtim_bc_mc = aid0 == 1;
if (have_bits) {
/* Find largest even number N1 so that bits numbered 1 through
* (N1 x 8) - 1 in the bitmap are 0 and number N2 so that bits
* (N2 + 1) x 8 through 2007 are 0. */
n1 = 0;
for (i = 0; i < IEEE80211_MAX_TIM_LEN; i++) {
if (ps->tim[i]) {
n1 = i & 0xfe;
break;
}
}
n2 = n1;
for (i = IEEE80211_MAX_TIM_LEN - 1; i >= n1; i--) {
if (ps->tim[i]) {
n2 = i;
break;
}
}
/* Bitmap control */
*pos++ = n1 | aid0;
/* Part Virt Bitmap */
skb_put(skb, n2 - n1);
memcpy(pos, ps->tim + n1, n2 - n1 + 1);
tim[1] = n2 - n1 + 4;
} else {
*pos++ = aid0; /* Bitmap control */
*pos++ = 0; /* Part Virt Bitmap */
}
}
static int ieee80211_beacon_add_tim(struct ieee80211_sub_if_data *sdata,
struct ps_data *ps, struct sk_buff *skb,
bool is_template)
{
struct ieee80211_local *local = sdata->local;
/*
* Not very nice, but we want to allow the driver to call
* ieee80211_beacon_get() as a response to the set_tim()
* callback. That, however, is already invoked under the
* sta_lock to guarantee consistent and race-free update
* of the tim bitmap in mac80211 and the driver.
*/
if (local->tim_in_locked_section) {
__ieee80211_beacon_add_tim(sdata, ps, skb, is_template);
} else {
spin_lock_bh(&local->tim_lock);
__ieee80211_beacon_add_tim(sdata, ps, skb, is_template);
spin_unlock_bh(&local->tim_lock);
}
return 0;
}
static void ieee80211_set_beacon_cntdwn(struct ieee80211_sub_if_data *sdata,
struct beacon_data *beacon)
{
u8 *beacon_data, count, max_count = 1;
struct probe_resp *resp;
size_t beacon_data_len;
u16 *bcn_offsets;
int i;
switch (sdata->vif.type) {
case NL80211_IFTYPE_AP:
beacon_data = beacon->tail;
beacon_data_len = beacon->tail_len;
break;
case NL80211_IFTYPE_ADHOC:
beacon_data = beacon->head;
beacon_data_len = beacon->head_len;
break;
case NL80211_IFTYPE_MESH_POINT:
beacon_data = beacon->head;
beacon_data_len = beacon->head_len;
break;
default:
return;
}
rcu_read_lock();
resp = rcu_dereference(sdata->deflink.u.ap.probe_resp);
bcn_offsets = beacon->cntdwn_counter_offsets;
count = beacon->cntdwn_current_counter;
if (sdata->vif.bss_conf.csa_active)
max_count = IEEE80211_MAX_CNTDWN_COUNTERS_NUM;
for (i = 0; i < max_count; ++i) {
if (bcn_offsets[i]) {
if (WARN_ON_ONCE(bcn_offsets[i] >= beacon_data_len)) {
rcu_read_unlock();
return;
}
beacon_data[bcn_offsets[i]] = count;
}
if (sdata->vif.type == NL80211_IFTYPE_AP && resp) {
u16 *resp_offsets = resp->cntdwn_counter_offsets;
resp->data[resp_offsets[i]] = count;
}
}
rcu_read_unlock();
}
static u8 __ieee80211_beacon_update_cntdwn(struct beacon_data *beacon)
{
beacon->cntdwn_current_counter--;
/* the counter should never reach 0 */
WARN_ON_ONCE(!beacon->cntdwn_current_counter);
return beacon->cntdwn_current_counter;
}
u8 ieee80211_beacon_update_cntdwn(struct ieee80211_vif *vif)
{
struct ieee80211_sub_if_data *sdata = vif_to_sdata(vif);
struct beacon_data *beacon = NULL;
u8 count = 0;
rcu_read_lock();
if (sdata->vif.type == NL80211_IFTYPE_AP)
beacon = rcu_dereference(sdata->deflink.u.ap.beacon);
else if (sdata->vif.type == NL80211_IFTYPE_ADHOC)
beacon = rcu_dereference(sdata->u.ibss.presp);
else if (ieee80211_vif_is_mesh(&sdata->vif))
beacon = rcu_dereference(sdata->u.mesh.beacon);
if (!beacon)
goto unlock;
count = __ieee80211_beacon_update_cntdwn(beacon);
unlock:
rcu_read_unlock();
return count;
}
EXPORT_SYMBOL(ieee80211_beacon_update_cntdwn);
void ieee80211_beacon_set_cntdwn(struct ieee80211_vif *vif, u8 counter)
{
struct ieee80211_sub_if_data *sdata = vif_to_sdata(vif);
struct beacon_data *beacon = NULL;
rcu_read_lock();
if (sdata->vif.type == NL80211_IFTYPE_AP)
beacon = rcu_dereference(sdata->deflink.u.ap.beacon);
else if (sdata->vif.type == NL80211_IFTYPE_ADHOC)
beacon = rcu_dereference(sdata->u.ibss.presp);
else if (ieee80211_vif_is_mesh(&sdata->vif))
beacon = rcu_dereference(sdata->u.mesh.beacon);
if (!beacon)
goto unlock;
if (counter < beacon->cntdwn_current_counter)
beacon->cntdwn_current_counter = counter;
unlock:
rcu_read_unlock();
}
EXPORT_SYMBOL(ieee80211_beacon_set_cntdwn);
bool ieee80211_beacon_cntdwn_is_complete(struct ieee80211_vif *vif)
{
struct ieee80211_sub_if_data *sdata = vif_to_sdata(vif);
struct beacon_data *beacon = NULL;
u8 *beacon_data;
size_t beacon_data_len;
int ret = false;
if (!ieee80211_sdata_running(sdata))
return false;
rcu_read_lock();
if (vif->type == NL80211_IFTYPE_AP) {
beacon = rcu_dereference(sdata->deflink.u.ap.beacon);
if (WARN_ON(!beacon || !beacon->tail))
goto out;
beacon_data = beacon->tail;
beacon_data_len = beacon->tail_len;
} else if (vif->type == NL80211_IFTYPE_ADHOC) {
struct ieee80211_if_ibss *ifibss = &sdata->u.ibss;
beacon = rcu_dereference(ifibss->presp);
if (!beacon)
goto out;
beacon_data = beacon->head;
beacon_data_len = beacon->head_len;
} else if (vif->type == NL80211_IFTYPE_MESH_POINT) {
struct ieee80211_if_mesh *ifmsh = &sdata->u.mesh;
beacon = rcu_dereference(ifmsh->beacon);
if (!beacon)
goto out;
beacon_data = beacon->head;
beacon_data_len = beacon->head_len;
} else {
WARN_ON(1);
goto out;
}
if (!beacon->cntdwn_counter_offsets[0])
goto out;
if (WARN_ON_ONCE(beacon->cntdwn_counter_offsets[0] > beacon_data_len))
goto out;
if (beacon_data[beacon->cntdwn_counter_offsets[0]] == 1)
ret = true;
out:
rcu_read_unlock();
return ret;
}
EXPORT_SYMBOL(ieee80211_beacon_cntdwn_is_complete);
static int ieee80211_beacon_protect(struct sk_buff *skb,
struct ieee80211_local *local,
struct ieee80211_sub_if_data *sdata)
{
ieee80211_tx_result res;
struct ieee80211_tx_data tx;
struct sk_buff *check_skb;
memset(&tx, 0, sizeof(tx));
tx.key = rcu_dereference(sdata->deflink.default_beacon_key);
if (!tx.key)
return 0;
tx.local = local;
tx.sdata = sdata;
__skb_queue_head_init(&tx.skbs);
__skb_queue_tail(&tx.skbs, skb);
res = ieee80211_tx_h_encrypt(&tx);
check_skb = __skb_dequeue(&tx.skbs);
/* we may crash after this, but it'd be a bug in crypto */
WARN_ON(check_skb != skb);
if (WARN_ON_ONCE(res != TX_CONTINUE))
return -EINVAL;
return 0;
}
static void
ieee80211_beacon_get_finish(struct ieee80211_hw *hw,
struct ieee80211_vif *vif,
struct ieee80211_mutable_offsets *offs,
struct beacon_data *beacon,
struct sk_buff *skb,
struct ieee80211_chanctx_conf *chanctx_conf,
u16 csa_off_base)
{
struct ieee80211_local *local = hw_to_local(hw);
struct ieee80211_sub_if_data *sdata = vif_to_sdata(vif);
struct ieee80211_tx_info *info;
enum nl80211_band band;
struct ieee80211_tx_rate_control txrc;
/* CSA offsets */
if (offs && beacon) {
u16 i;
for (i = 0; i < IEEE80211_MAX_CNTDWN_COUNTERS_NUM; i++) {
u16 csa_off = beacon->cntdwn_counter_offsets[i];
if (!csa_off)
continue;
offs->cntdwn_counter_offs[i] = csa_off_base + csa_off;
}
}
band = chanctx_conf->def.chan->band;
info = IEEE80211_SKB_CB(skb);
info->flags |= IEEE80211_TX_INTFL_DONT_ENCRYPT;
info->flags |= IEEE80211_TX_CTL_NO_ACK;
info->band = band;
memset(&txrc, 0, sizeof(txrc));
txrc.hw = hw;
txrc.sband = local->hw.wiphy->bands[band];
txrc.bss_conf = &sdata->vif.bss_conf;
txrc.skb = skb;
txrc.reported_rate.idx = -1;
if (sdata->beacon_rate_set && sdata->beacon_rateidx_mask[band])
txrc.rate_idx_mask = sdata->beacon_rateidx_mask[band];
else
txrc.rate_idx_mask = sdata->rc_rateidx_mask[band];
txrc.bss = true;
rate_control_get_rate(sdata, NULL, &txrc);
info->control.vif = vif;
info->flags |= IEEE80211_TX_CTL_CLEAR_PS_FILT |
IEEE80211_TX_CTL_ASSIGN_SEQ |
IEEE80211_TX_CTL_FIRST_FRAGMENT;
}
static void
ieee80211_beacon_add_mbssid(struct sk_buff *skb, struct beacon_data *beacon)
{
int i;
if (!beacon->mbssid_ies)
return;
for (i = 0; i < beacon->mbssid_ies->cnt; i++)
skb_put_data(skb, beacon->mbssid_ies->elem[i].data,
beacon->mbssid_ies->elem[i].len);
}
static struct sk_buff *
ieee80211_beacon_get_ap(struct ieee80211_hw *hw,
struct ieee80211_vif *vif,
struct ieee80211_mutable_offsets *offs,
bool is_template,
struct beacon_data *beacon,
struct ieee80211_chanctx_conf *chanctx_conf)
{
struct ieee80211_local *local = hw_to_local(hw);
struct ieee80211_sub_if_data *sdata = vif_to_sdata(vif);
struct ieee80211_if_ap *ap = &sdata->u.ap;
struct sk_buff *skb = NULL;
u16 csa_off_base = 0;
int mbssid_len;
if (beacon->cntdwn_counter_offsets[0]) {
if (!is_template)
ieee80211_beacon_update_cntdwn(vif);
ieee80211_set_beacon_cntdwn(sdata, beacon);
}
/* headroom, head length,
* tail length, maximum TIM length and multiple BSSID length
*/
mbssid_len = ieee80211_get_mbssid_beacon_len(beacon->mbssid_ies);
skb = dev_alloc_skb(local->tx_headroom + beacon->head_len +
beacon->tail_len + 256 +
local->hw.extra_beacon_tailroom + mbssid_len);
if (!skb)
return NULL;
skb_reserve(skb, local->tx_headroom);
skb_put_data(skb, beacon->head, beacon->head_len);
ieee80211_beacon_add_tim(sdata, &ap->ps, skb, is_template);
if (offs) {
offs->tim_offset = beacon->head_len;
offs->tim_length = skb->len - beacon->head_len;
offs->cntdwn_counter_offs[0] = beacon->cntdwn_counter_offsets[0];
if (mbssid_len) {
ieee80211_beacon_add_mbssid(skb, beacon);
offs->mbssid_off = skb->len - mbssid_len;
}
/* for AP the csa offsets are from tail */
csa_off_base = skb->len;
}
if (beacon->tail)
skb_put_data(skb, beacon->tail, beacon->tail_len);
if (ieee80211_beacon_protect(skb, local, sdata) < 0)
return NULL;
ieee80211_beacon_get_finish(hw, vif, offs, beacon, skb, chanctx_conf,
csa_off_base);
return skb;
}
static struct sk_buff *
__ieee80211_beacon_get(struct ieee80211_hw *hw,
struct ieee80211_vif *vif,
struct ieee80211_mutable_offsets *offs,
bool is_template)
{
struct ieee80211_local *local = hw_to_local(hw);
struct beacon_data *beacon = NULL;
struct sk_buff *skb = NULL;
struct ieee80211_sub_if_data *sdata = NULL;
struct ieee80211_chanctx_conf *chanctx_conf;
rcu_read_lock();
sdata = vif_to_sdata(vif);
chanctx_conf = rcu_dereference(sdata->vif.bss_conf.chanctx_conf);
if (!ieee80211_sdata_running(sdata) || !chanctx_conf)
goto out;
if (offs)
memset(offs, 0, sizeof(*offs));
if (sdata->vif.type == NL80211_IFTYPE_AP) {
beacon = rcu_dereference(sdata->deflink.u.ap.beacon);
if (!beacon)
goto out;
skb = ieee80211_beacon_get_ap(hw, vif, offs, is_template,
beacon, chanctx_conf);
} else if (sdata->vif.type == NL80211_IFTYPE_ADHOC) {
struct ieee80211_if_ibss *ifibss = &sdata->u.ibss;
struct ieee80211_hdr *hdr;
beacon = rcu_dereference(ifibss->presp);
if (!beacon)
goto out;
if (beacon->cntdwn_counter_offsets[0]) {
if (!is_template)
__ieee80211_beacon_update_cntdwn(beacon);
ieee80211_set_beacon_cntdwn(sdata, beacon);
}
skb = dev_alloc_skb(local->tx_headroom + beacon->head_len +
local->hw.extra_beacon_tailroom);
if (!skb)
goto out;
skb_reserve(skb, local->tx_headroom);
skb_put_data(skb, beacon->head, beacon->head_len);
hdr = (struct ieee80211_hdr *) skb->data;
hdr->frame_control = cpu_to_le16(IEEE80211_FTYPE_MGMT |
IEEE80211_STYPE_BEACON);
ieee80211_beacon_get_finish(hw, vif, offs, beacon, skb,
chanctx_conf, 0);
} else if (ieee80211_vif_is_mesh(&sdata->vif)) {
struct ieee80211_if_mesh *ifmsh = &sdata->u.mesh;
beacon = rcu_dereference(ifmsh->beacon);
if (!beacon)
goto out;
if (beacon->cntdwn_counter_offsets[0]) {
if (!is_template)
/* TODO: For mesh csa_counter is in TU, so
* decrementing it by one isn't correct, but
* for now we leave it consistent with overall
* mac80211's behavior.
*/
__ieee80211_beacon_update_cntdwn(beacon);
ieee80211_set_beacon_cntdwn(sdata, beacon);
}
if (ifmsh->sync_ops)
ifmsh->sync_ops->adjust_tsf(sdata, beacon);
skb = dev_alloc_skb(local->tx_headroom +
beacon->head_len +
mac80211: mesh power save basics Add routines to - maintain a PS mode for each peer and a non-peer PS mode - indicate own PS mode in transmitted frames - track neighbor STAs power modes - buffer frames when neighbors are in PS mode - add TIM and Awake Window IE to beacons - release frames in Mesh Peer Service Periods Add local_pm to sta_info to represent the link-specific power mode at this station towards the remote station. When a peer link is established, use the default power mode stored in mesh config. Update the PS status if the peering status of a neighbor changes. Maintain a mesh power mode for non-peer mesh STAs. Set the non-peer power mode to active mode during peering. Authenticated mesh peering is currently not working when either node is configured to be in power save mode. Indicate the current power mode in transmitted frames. Use QoS Nulls to indicate mesh power mode transitions. For performance reasons, calls to the function setting the frame flags are placed in HWMP routing routines, as there the STA pointer is already available. Add peer_pm to sta_info to represent the peer's link-specific power mode towards the local station. Add nonpeer_pm to represent the peer's power mode towards all non-peer stations. Track power modes based on received frames. Add the ps_data structure to ieee80211_if_mesh (for TIM map, PS neighbor counter and group-addressed frame buffer). Set WLAN_STA_PS flag for STA in PS mode to use the unicast frame buffering routines in the tx path. Update num_sta_ps to buffer and release group-addressed frames after DTIM beacons. Announce the awake window duration in beacons if in light or deep sleep mode towards any peer or non-peer. Create a TIM IE similarly to AP mode and add it to mesh beacons. Parse received Awake Window IEs and check TIM IEs for buffered frames. Release frames towards peers in mesh Peer Service Periods. Use the corresponding trigger frames and monitor the MPSP status. Append a QoS Null as trigger frame if neccessary to properly end the MPSP. Currently, in HT channels MPSPs behave imperfectly and show large delay spikes and frame losses. Signed-off-by: Marco Porsch <marco@cozybit.com> Signed-off-by: Ivan Bezyazychnyy <ivan.bezyazychnyy@gmail.com> Signed-off-by: Mike Krinkin <krinkin.m.u@gmail.com> Signed-off-by: Max Filippov <jcmvbkbc@gmail.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2013-01-31 01:14:08 +08:00
256 + /* TIM IE */
beacon->tail_len +
local->hw.extra_beacon_tailroom);
if (!skb)
goto out;
skb_reserve(skb, local->tx_headroom);
skb_put_data(skb, beacon->head, beacon->head_len);
ieee80211_beacon_add_tim(sdata, &ifmsh->ps, skb, is_template);
if (offs) {
offs->tim_offset = beacon->head_len;
offs->tim_length = skb->len - beacon->head_len;
}
skb_put_data(skb, beacon->tail, beacon->tail_len);
ieee80211_beacon_get_finish(hw, vif, offs, beacon, skb,
chanctx_conf, 0);
} else {
WARN_ON(1);
goto out;
}
out:
rcu_read_unlock();
return skb;
}
struct sk_buff *
ieee80211_beacon_get_template(struct ieee80211_hw *hw,
struct ieee80211_vif *vif,
struct ieee80211_mutable_offsets *offs)
{
return __ieee80211_beacon_get(hw, vif, offs, true);
}
EXPORT_SYMBOL(ieee80211_beacon_get_template);
struct sk_buff *ieee80211_beacon_get_tim(struct ieee80211_hw *hw,
struct ieee80211_vif *vif,
u16 *tim_offset, u16 *tim_length)
{
struct ieee80211_mutable_offsets offs = {};
struct sk_buff *bcn = __ieee80211_beacon_get(hw, vif, &offs, false);
struct sk_buff *copy;
struct ieee80211_supported_band *sband;
int shift;
if (!bcn)
return bcn;
if (tim_offset)
*tim_offset = offs.tim_offset;
if (tim_length)
*tim_length = offs.tim_length;
if (ieee80211_hw_check(hw, BEACON_TX_STATUS) ||
!hw_to_local(hw)->monitors)
return bcn;
/* send a copy to monitor interfaces */
copy = skb_copy(bcn, GFP_ATOMIC);
if (!copy)
return bcn;
shift = ieee80211_vif_get_shift(vif);
mac80211: Fix possible sband related NULL pointer de-reference Existing API 'ieee80211_get_sdata_band' returns default 2 GHz band even if the channel context configuration is NULL. This crashes for chipsets which support 5 Ghz alone when it tries to access members of 'sband'. Channel context configuration can be NULL in multivif case and when channel switch is in progress (or) when it fails. Fix this by replacing the API 'ieee80211_get_sdata_band' with 'ieee80211_get_sband' which returns a NULL pointer for sband when the channel configuration is NULL. An example scenario is as below: In multivif mode (AP + STA) with drivers like ath10k, when we do a channel switch in the AP vif (which has a number of clients connected) and a STA vif which is connected to some other AP, when the channel switch in AP vif fails, while the STA vifs tries to connect to the other AP, there is a window where the channel context is NULL/invalid and this results in a crash while the clients connected to the AP vif tries to reconnect and this race is very similar to the one investigated by Michal in https://patchwork.kernel.org/patch/3788161/ and this does happens with hardware that supports 5Ghz alone after long hours of testing with continuous channel switch on the AP vif ieee80211 phy0: channel context reservation cannot be finalized because some interfaces aren't switching wlan0: failed to finalize CSA, disconnecting wlan0-1: deauthenticating from 8c:fd:f0:01:54:9c by local choice (Reason: 3=DEAUTH_LEAVING) WARNING: CPU: 1 PID: 19032 at net/mac80211/ieee80211_i.h:1013 sta_info_alloc+0x374/0x3fc [mac80211] [<bf77272c>] (sta_info_alloc [mac80211]) [<bf78776c>] (ieee80211_add_station [mac80211])) [<bf73cc50>] (nl80211_new_station [cfg80211]) Unable to handle kernel NULL pointer dereference at virtual address 00000014 pgd = d5f4c000 Internal error: Oops: 17 [#1] PREEMPT SMP ARM PC is at sta_info_alloc+0x380/0x3fc [mac80211] LR is at sta_info_alloc+0x37c/0x3fc [mac80211] [<bf772738>] (sta_info_alloc [mac80211]) [<bf78776c>] (ieee80211_add_station [mac80211]) [<bf73cc50>] (nl80211_new_station [cfg80211])) Cc: Michal Kazior <michal.kazior@tieto.com> Signed-off-by: Mohammed Shafi Shajakhan <mohammed@qti.qualcomm.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-04-27 15:15:38 +08:00
sband = ieee80211_get_sband(vif_to_sdata(vif));
if (!sband)
return bcn;
ieee80211_tx_monitor(hw_to_local(hw), copy, sband, 1, shift, false,
NULL);
return bcn;
}
EXPORT_SYMBOL(ieee80211_beacon_get_tim);
struct sk_buff *ieee80211_proberesp_get(struct ieee80211_hw *hw,
struct ieee80211_vif *vif)
{
struct sk_buff *skb = NULL;
struct probe_resp *presp = NULL;
struct ieee80211_hdr *hdr;
struct ieee80211_sub_if_data *sdata = vif_to_sdata(vif);
if (sdata->vif.type != NL80211_IFTYPE_AP)
return NULL;
rcu_read_lock();
presp = rcu_dereference(sdata->deflink.u.ap.probe_resp);
if (!presp)
goto out;
skb = dev_alloc_skb(presp->len);
if (!skb)
goto out;
skb_put_data(skb, presp->data, presp->len);
hdr = (struct ieee80211_hdr *) skb->data;
memset(hdr->addr1, 0, sizeof(hdr->addr1));
out:
rcu_read_unlock();
return skb;
}
EXPORT_SYMBOL(ieee80211_proberesp_get);
struct sk_buff *ieee80211_get_fils_discovery_tmpl(struct ieee80211_hw *hw,
struct ieee80211_vif *vif)
{
struct sk_buff *skb = NULL;
struct fils_discovery_data *tmpl = NULL;
struct ieee80211_sub_if_data *sdata = vif_to_sdata(vif);
if (sdata->vif.type != NL80211_IFTYPE_AP)
return NULL;
rcu_read_lock();
tmpl = rcu_dereference(sdata->deflink.u.ap.fils_discovery);
if (!tmpl) {
rcu_read_unlock();
return NULL;
}
skb = dev_alloc_skb(sdata->local->hw.extra_tx_headroom + tmpl->len);
if (skb) {
skb_reserve(skb, sdata->local->hw.extra_tx_headroom);
skb_put_data(skb, tmpl->data, tmpl->len);
}
rcu_read_unlock();
return skb;
}
EXPORT_SYMBOL(ieee80211_get_fils_discovery_tmpl);
struct sk_buff *
ieee80211_get_unsol_bcast_probe_resp_tmpl(struct ieee80211_hw *hw,
struct ieee80211_vif *vif)
{
struct sk_buff *skb = NULL;
struct unsol_bcast_probe_resp_data *tmpl = NULL;
struct ieee80211_sub_if_data *sdata = vif_to_sdata(vif);
if (sdata->vif.type != NL80211_IFTYPE_AP)
return NULL;
rcu_read_lock();
tmpl = rcu_dereference(sdata->deflink.u.ap.unsol_bcast_probe_resp);
if (!tmpl) {
rcu_read_unlock();
return NULL;
}
skb = dev_alloc_skb(sdata->local->hw.extra_tx_headroom + tmpl->len);
if (skb) {
skb_reserve(skb, sdata->local->hw.extra_tx_headroom);
skb_put_data(skb, tmpl->data, tmpl->len);
}
rcu_read_unlock();
return skb;
}
EXPORT_SYMBOL(ieee80211_get_unsol_bcast_probe_resp_tmpl);
struct sk_buff *ieee80211_pspoll_get(struct ieee80211_hw *hw,
struct ieee80211_vif *vif)
{
struct ieee80211_sub_if_data *sdata;
struct ieee80211_pspoll *pspoll;
struct ieee80211_local *local;
struct sk_buff *skb;
if (WARN_ON(vif->type != NL80211_IFTYPE_STATION))
return NULL;
sdata = vif_to_sdata(vif);
local = sdata->local;
skb = dev_alloc_skb(local->hw.extra_tx_headroom + sizeof(*pspoll));
if (!skb)
return NULL;
skb_reserve(skb, local->hw.extra_tx_headroom);
pspoll = skb_put_zero(skb, sizeof(*pspoll));
pspoll->frame_control = cpu_to_le16(IEEE80211_FTYPE_CTL |
IEEE80211_STYPE_PSPOLL);
pspoll->aid = cpu_to_le16(sdata->vif.cfg.aid);
/* aid in PS-Poll has its two MSBs each set to 1 */
pspoll->aid |= cpu_to_le16(1 << 15 | 1 << 14);
memcpy(pspoll->bssid, sdata->deflink.u.mgd.bssid, ETH_ALEN);
memcpy(pspoll->ta, vif->addr, ETH_ALEN);
return skb;
}
EXPORT_SYMBOL(ieee80211_pspoll_get);
struct sk_buff *ieee80211_nullfunc_get(struct ieee80211_hw *hw,
struct ieee80211_vif *vif,
bool qos_ok)
{
struct ieee80211_hdr_3addr *nullfunc;
struct ieee80211_sub_if_data *sdata;
struct ieee80211_local *local;
struct sk_buff *skb;
bool qos = false;
if (WARN_ON(vif->type != NL80211_IFTYPE_STATION))
return NULL;
sdata = vif_to_sdata(vif);
local = sdata->local;
if (qos_ok) {
struct sta_info *sta;
rcu_read_lock();
sta = sta_info_get(sdata, sdata->deflink.u.mgd.bssid);
qos = sta && sta->sta.wme;
rcu_read_unlock();
}
skb = dev_alloc_skb(local->hw.extra_tx_headroom +
sizeof(*nullfunc) + 2);
if (!skb)
return NULL;
skb_reserve(skb, local->hw.extra_tx_headroom);
nullfunc = skb_put_zero(skb, sizeof(*nullfunc));
nullfunc->frame_control = cpu_to_le16(IEEE80211_FTYPE_DATA |
IEEE80211_STYPE_NULLFUNC |
IEEE80211_FCTL_TODS);
if (qos) {
__le16 qoshdr = cpu_to_le16(7);
BUILD_BUG_ON((IEEE80211_STYPE_QOS_NULLFUNC |
IEEE80211_STYPE_NULLFUNC) !=
IEEE80211_STYPE_QOS_NULLFUNC);
nullfunc->frame_control |=
cpu_to_le16(IEEE80211_STYPE_QOS_NULLFUNC);
skb->priority = 7;
skb_set_queue_mapping(skb, IEEE80211_AC_VO);
skb_put_data(skb, &qoshdr, sizeof(qoshdr));
}
memcpy(nullfunc->addr1, sdata->deflink.u.mgd.bssid, ETH_ALEN);
memcpy(nullfunc->addr2, vif->addr, ETH_ALEN);
memcpy(nullfunc->addr3, sdata->deflink.u.mgd.bssid, ETH_ALEN);
return skb;
}
EXPORT_SYMBOL(ieee80211_nullfunc_get);
struct sk_buff *ieee80211_probereq_get(struct ieee80211_hw *hw,
const u8 *src_addr,
const u8 *ssid, size_t ssid_len,
size_t tailroom)
{
struct ieee80211_local *local = hw_to_local(hw);
struct ieee80211_hdr_3addr *hdr;
struct sk_buff *skb;
size_t ie_ssid_len;
u8 *pos;
ie_ssid_len = 2 + ssid_len;
skb = dev_alloc_skb(local->hw.extra_tx_headroom + sizeof(*hdr) +
ie_ssid_len + tailroom);
if (!skb)
return NULL;
skb_reserve(skb, local->hw.extra_tx_headroom);
hdr = skb_put_zero(skb, sizeof(*hdr));
hdr->frame_control = cpu_to_le16(IEEE80211_FTYPE_MGMT |
IEEE80211_STYPE_PROBE_REQ);
eth_broadcast_addr(hdr->addr1);
memcpy(hdr->addr2, src_addr, ETH_ALEN);
eth_broadcast_addr(hdr->addr3);
pos = skb_put(skb, ie_ssid_len);
*pos++ = WLAN_EID_SSID;
*pos++ = ssid_len;
if (ssid_len)
memcpy(pos, ssid, ssid_len);
pos += ssid_len;
return skb;
}
EXPORT_SYMBOL(ieee80211_probereq_get);
void ieee80211_rts_get(struct ieee80211_hw *hw, struct ieee80211_vif *vif,
const void *frame, size_t frame_len,
const struct ieee80211_tx_info *frame_txctl,
struct ieee80211_rts *rts)
{
const struct ieee80211_hdr *hdr = frame;
rts->frame_control =
cpu_to_le16(IEEE80211_FTYPE_CTL | IEEE80211_STYPE_RTS);
rts->duration = ieee80211_rts_duration(hw, vif, frame_len,
frame_txctl);
memcpy(rts->ra, hdr->addr1, sizeof(rts->ra));
memcpy(rts->ta, hdr->addr2, sizeof(rts->ta));
}
EXPORT_SYMBOL(ieee80211_rts_get);
void ieee80211_ctstoself_get(struct ieee80211_hw *hw, struct ieee80211_vif *vif,
const void *frame, size_t frame_len,
const struct ieee80211_tx_info *frame_txctl,
struct ieee80211_cts *cts)
{
const struct ieee80211_hdr *hdr = frame;
cts->frame_control =
cpu_to_le16(IEEE80211_FTYPE_CTL | IEEE80211_STYPE_CTS);
cts->duration = ieee80211_ctstoself_duration(hw, vif,
frame_len, frame_txctl);
memcpy(cts->ra, hdr->addr1, sizeof(cts->ra));
}
EXPORT_SYMBOL(ieee80211_ctstoself_get);
struct sk_buff *
ieee80211_get_buffered_bc(struct ieee80211_hw *hw,
struct ieee80211_vif *vif)
{
struct ieee80211_local *local = hw_to_local(hw);
struct sk_buff *skb = NULL;
struct ieee80211_tx_data tx;
struct ieee80211_sub_if_data *sdata;
struct ps_data *ps;
struct ieee80211_tx_info *info;
struct ieee80211_chanctx_conf *chanctx_conf;
sdata = vif_to_sdata(vif);
rcu_read_lock();
chanctx_conf = rcu_dereference(sdata->vif.bss_conf.chanctx_conf);
if (!chanctx_conf)
goto out;
if (sdata->vif.type == NL80211_IFTYPE_AP) {
struct beacon_data *beacon =
rcu_dereference(sdata->deflink.u.ap.beacon);
if (!beacon || !beacon->head)
goto out;
ps = &sdata->u.ap.ps;
mac80211: mesh power save basics Add routines to - maintain a PS mode for each peer and a non-peer PS mode - indicate own PS mode in transmitted frames - track neighbor STAs power modes - buffer frames when neighbors are in PS mode - add TIM and Awake Window IE to beacons - release frames in Mesh Peer Service Periods Add local_pm to sta_info to represent the link-specific power mode at this station towards the remote station. When a peer link is established, use the default power mode stored in mesh config. Update the PS status if the peering status of a neighbor changes. Maintain a mesh power mode for non-peer mesh STAs. Set the non-peer power mode to active mode during peering. Authenticated mesh peering is currently not working when either node is configured to be in power save mode. Indicate the current power mode in transmitted frames. Use QoS Nulls to indicate mesh power mode transitions. For performance reasons, calls to the function setting the frame flags are placed in HWMP routing routines, as there the STA pointer is already available. Add peer_pm to sta_info to represent the peer's link-specific power mode towards the local station. Add nonpeer_pm to represent the peer's power mode towards all non-peer stations. Track power modes based on received frames. Add the ps_data structure to ieee80211_if_mesh (for TIM map, PS neighbor counter and group-addressed frame buffer). Set WLAN_STA_PS flag for STA in PS mode to use the unicast frame buffering routines in the tx path. Update num_sta_ps to buffer and release group-addressed frames after DTIM beacons. Announce the awake window duration in beacons if in light or deep sleep mode towards any peer or non-peer. Create a TIM IE similarly to AP mode and add it to mesh beacons. Parse received Awake Window IEs and check TIM IEs for buffered frames. Release frames towards peers in mesh Peer Service Periods. Use the corresponding trigger frames and monitor the MPSP status. Append a QoS Null as trigger frame if neccessary to properly end the MPSP. Currently, in HT channels MPSPs behave imperfectly and show large delay spikes and frame losses. Signed-off-by: Marco Porsch <marco@cozybit.com> Signed-off-by: Ivan Bezyazychnyy <ivan.bezyazychnyy@gmail.com> Signed-off-by: Mike Krinkin <krinkin.m.u@gmail.com> Signed-off-by: Max Filippov <jcmvbkbc@gmail.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2013-01-31 01:14:08 +08:00
} else if (ieee80211_vif_is_mesh(&sdata->vif)) {
ps = &sdata->u.mesh.ps;
} else {
goto out;
}
if (ps->dtim_count != 0 || !ps->dtim_bc_mc)
goto out; /* send buffered bc/mc only after DTIM beacon */
while (1) {
skb = skb_dequeue(&ps->bc_buf);
if (!skb)
goto out;
local->total_ps_buffered--;
if (!skb_queue_empty(&ps->bc_buf) && skb->len >= 2) {
struct ieee80211_hdr *hdr =
(struct ieee80211_hdr *) skb->data;
/* more buffered multicast/broadcast frames ==> set
* MoreData flag in IEEE 802.11 header to inform PS
* STAs */
hdr->frame_control |=
cpu_to_le16(IEEE80211_FCTL_MOREDATA);
}
if (sdata->vif.type == NL80211_IFTYPE_AP)
sdata = IEEE80211_DEV_TO_SUB_IF(skb->dev);
if (!ieee80211_tx_prepare(sdata, &tx, NULL, skb))
break;
ieee80211_free_txskb(hw, skb);
}
info = IEEE80211_SKB_CB(skb);
tx.flags |= IEEE80211_TX_PS_BUFFERED;
info->band = chanctx_conf->def.chan->band;
if (invoke_tx_handlers(&tx))
skb = NULL;
out:
rcu_read_unlock();
return skb;
}
EXPORT_SYMBOL(ieee80211_get_buffered_bc);
int ieee80211_reserve_tid(struct ieee80211_sta *pubsta, u8 tid)
{
struct sta_info *sta = container_of(pubsta, struct sta_info, sta);
struct ieee80211_sub_if_data *sdata = sta->sdata;
struct ieee80211_local *local = sdata->local;
int ret;
u32 queues;
lockdep_assert_held(&local->sta_mtx);
/* only some cases are supported right now */
switch (sdata->vif.type) {
case NL80211_IFTYPE_STATION:
case NL80211_IFTYPE_AP:
case NL80211_IFTYPE_AP_VLAN:
break;
default:
WARN_ON(1);
return -EINVAL;
}
if (WARN_ON(tid >= IEEE80211_NUM_UPS))
return -EINVAL;
if (sta->reserved_tid == tid) {
ret = 0;
goto out;
}
if (sta->reserved_tid != IEEE80211_TID_UNRESERVED) {
sdata_err(sdata, "TID reservation already active\n");
ret = -EALREADY;
goto out;
}
ieee80211_stop_vif_queues(sdata->local, sdata,
IEEE80211_QUEUE_STOP_REASON_RESERVE_TID);
synchronize_net();
/* Tear down BA sessions so we stop aggregating on this TID */
if (ieee80211_hw_check(&local->hw, AMPDU_AGGREGATION)) {
set_sta_flag(sta, WLAN_STA_BLOCK_BA);
__ieee80211_stop_tx_ba_session(sta, tid,
AGG_STOP_LOCAL_REQUEST);
}
queues = BIT(sdata->vif.hw_queue[ieee802_1d_to_ac[tid]]);
__ieee80211_flush_queues(local, sdata, queues, false);
sta->reserved_tid = tid;
ieee80211_wake_vif_queues(local, sdata,
IEEE80211_QUEUE_STOP_REASON_RESERVE_TID);
if (ieee80211_hw_check(&local->hw, AMPDU_AGGREGATION))
clear_sta_flag(sta, WLAN_STA_BLOCK_BA);
ret = 0;
out:
return ret;
}
EXPORT_SYMBOL(ieee80211_reserve_tid);
void ieee80211_unreserve_tid(struct ieee80211_sta *pubsta, u8 tid)
{
struct sta_info *sta = container_of(pubsta, struct sta_info, sta);
struct ieee80211_sub_if_data *sdata = sta->sdata;
lockdep_assert_held(&sdata->local->sta_mtx);
/* only some cases are supported right now */
switch (sdata->vif.type) {
case NL80211_IFTYPE_STATION:
case NL80211_IFTYPE_AP:
case NL80211_IFTYPE_AP_VLAN:
break;
default:
WARN_ON(1);
return;
}
if (tid != sta->reserved_tid) {
sdata_err(sdata, "TID to unreserve (%d) isn't reserved\n", tid);
return;
}
sta->reserved_tid = IEEE80211_TID_UNRESERVED;
}
EXPORT_SYMBOL(ieee80211_unreserve_tid);
void __ieee80211_tx_skb_tid_band(struct ieee80211_sub_if_data *sdata,
struct sk_buff *skb, int tid,
enum nl80211_band band)
{
int ac = ieee80211_ac_from_tid(tid);
skb_reset_mac_header(skb);
skb_set_queue_mapping(skb, ac);
skb->priority = tid;
skb->dev = sdata->dev;
/*
* The other path calling ieee80211_xmit is from the tasklet,
* and while we can handle concurrent transmissions locking
* requirements are that we do not come into tx with bhs on.
*/
local_bh_disable();
IEEE80211_SKB_CB(skb)->band = band;
ieee80211_xmit(sdata, NULL, skb);
local_bh_enable();
}
int ieee80211_tx_control_port(struct wiphy *wiphy, struct net_device *dev,
const u8 *buf, size_t len,
const u8 *dest, __be16 proto, bool unencrypted,
u64 *cookie)
{
struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev);
struct ieee80211_local *local = sdata->local;
struct sta_info *sta;
struct sk_buff *skb;
struct ethhdr *ehdr;
u32 ctrl_flags = 0;
u32 flags = 0;
/* Only accept CONTROL_PORT_PROTOCOL configured in CONNECT/ASSOCIATE
* or Pre-Authentication
*/
if (proto != sdata->control_port_protocol &&
proto != cpu_to_be16(ETH_P_PREAUTH))
return -EINVAL;
if (proto == sdata->control_port_protocol)
ctrl_flags |= IEEE80211_TX_CTRL_PORT_CTRL_PROTO |
IEEE80211_TX_CTRL_SKIP_MPATH_LOOKUP;
if (unencrypted)
flags |= IEEE80211_TX_INTFL_DONT_ENCRYPT;
if (cookie)
ctrl_flags |= IEEE80211_TX_CTL_REQ_TX_STATUS;
flags |= IEEE80211_TX_INTFL_NL80211_FRAME_TX;
skb = dev_alloc_skb(local->hw.extra_tx_headroom +
sizeof(struct ethhdr) + len);
if (!skb)
return -ENOMEM;
skb_reserve(skb, local->hw.extra_tx_headroom + sizeof(struct ethhdr));
skb_put_data(skb, buf, len);
ehdr = skb_push(skb, sizeof(struct ethhdr));
memcpy(ehdr->h_dest, dest, ETH_ALEN);
memcpy(ehdr->h_source, sdata->vif.addr, ETH_ALEN);
ehdr->h_proto = proto;
skb->dev = dev;
skb->protocol = proto;
skb_reset_network_header(skb);
skb_reset_mac_header(skb);
/* update QoS header to prioritize control port frames if possible,
* priorization also happens for control port frames send over
* AF_PACKET
*/
rcu_read_lock();
if (ieee80211_lookup_ra_sta(sdata, skb, &sta) == 0 && !IS_ERR(sta)) {
u16 queue = __ieee80211_select_queue(sdata, sta, skb);
skb_set_queue_mapping(skb, queue);
skb_get_hash(skb);
}
rcu_read_unlock();
/* mutex lock is only needed for incrementing the cookie counter */
mutex_lock(&local->mtx);
local_bh_disable();
__ieee80211_subif_start_xmit(skb, skb->dev, flags, ctrl_flags, cookie);
local_bh_enable();
mutex_unlock(&local->mtx);
return 0;
}
int ieee80211_probe_mesh_link(struct wiphy *wiphy, struct net_device *dev,
const u8 *buf, size_t len)
{
struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev);
struct ieee80211_local *local = sdata->local;
struct sk_buff *skb;
skb = dev_alloc_skb(local->hw.extra_tx_headroom + len +
30 + /* header size */
18); /* 11s header size */
if (!skb)
return -ENOMEM;
skb_reserve(skb, local->hw.extra_tx_headroom);
skb_put_data(skb, buf, len);
skb->dev = dev;
skb->protocol = htons(ETH_P_802_3);
skb_reset_network_header(skb);
skb_reset_mac_header(skb);
local_bh_disable();
__ieee80211_subif_start_xmit(skb, skb->dev, 0,
IEEE80211_TX_CTRL_SKIP_MPATH_LOOKUP,
NULL);
local_bh_enable();
return 0;
}