[PATCH AUTOSEL 5.17 086/149] ath11k: Fix frames flush failure caused by deadlock

From: Sasha Levin
Date: Fri Apr 01 2022 - 10:40:20 EST

From: Baochen Qiang <quic_bqiang@xxxxxxxxxxx>

[ Upstream commit 261b07519518bd14cb168b287b17e1d195f8d0c8 ]

We are seeing below warnings:

kernel: [25393.301506] ath11k_pci 0000:01:00.0: failed to flush mgmt transmit queue 0
kernel: [25398.421509] ath11k_pci 0000:01:00.0: failed to flush mgmt transmit queue 0
kernel: [25398.421831] ath11k_pci 0000:01:00.0: dropping mgmt frame for vdev 0, is_started 0

this means ath11k fails to flush mgmt. frames because wmi_mgmt_tx_work
has no chance to run in 5 seconds.

By setting /proc/sys/kernel/hung_task_timeout_secs to 20 and increasing
ATH11K_FLUSH_TIMEOUT to 50 we get below warnings:

kernel: [ 120.763160] INFO: task wpa_supplicant:924 blocked for more than 20 seconds.
kernel: [ 120.763169] Not tainted 5.10.90 #12
kernel: [ 120.763177] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
kernel: [ 120.763186] task:wpa_supplicant state:D stack: 0 pid: 924 ppid: 1 flags:0x000043a0
kernel: [ 120.763201] Call Trace:
kernel: [ 120.763214] __schedule+0x785/0x12fa
kernel: [ 120.763224] ? lockdep_hardirqs_on_prepare+0xe2/0x1bb
kernel: [ 120.763242] schedule+0x7e/0xa1
kernel: [ 120.763253] schedule_timeout+0x98/0xfe
kernel: [ 120.763266] ? run_local_timers+0x4a/0x4a
kernel: [ 120.763291] ath11k_mac_flush_tx_complete+0x197/0x2b1 [ath11k 13c3a9bf37790f4ac8103b3decf7ab4008ac314a]
kernel: [ 120.763306] ? init_wait_entry+0x2e/0x2e
kernel: [ 120.763343] __ieee80211_flush_queues+0x167/0x21f [mac80211 335da900954f1c5ea7f1613d92088ce83342042c]
kernel: [ 120.763378] __ieee80211_recalc_idle+0x105/0x125 [mac80211 335da900954f1c5ea7f1613d92088ce83342042c]
kernel: [ 120.763411] ieee80211_recalc_idle+0x14/0x27 [mac80211 335da900954f1c5ea7f1613d92088ce83342042c]
kernel: [ 120.763441] ieee80211_free_chanctx+0x77/0xa2 [mac80211 335da900954f1c5ea7f1613d92088ce83342042c]
kernel: [ 120.763473] __ieee80211_vif_release_channel+0x100/0x131 [mac80211 335da900954f1c5ea7f1613d92088ce83342042c]
kernel: [ 120.763540] ieee80211_vif_release_channel+0x66/0x81 [mac80211 335da900954f1c5ea7f1613d92088ce83342042c]
kernel: [ 120.763572] ieee80211_destroy_auth_data+0xa3/0xe6 [mac80211 335da900954f1c5ea7f1613d92088ce83342042c]
kernel: [ 120.763612] ieee80211_mgd_deauth+0x178/0x29b [mac80211 335da900954f1c5ea7f1613d92088ce83342042c]
kernel: [ 120.763654] cfg80211_mlme_deauth+0x1a8/0x22c [cfg80211 8945aa5bc2af5f6972336665d8ad6f9c191ad5be]
kernel: [ 120.763697] nl80211_deauthenticate+0xfa/0x123 [cfg80211 8945aa5bc2af5f6972336665d8ad6f9c191ad5be]
kernel: [ 120.763715] genl_rcv_msg+0x392/0x3c2
kernel: [ 120.763750] ? nl80211_associate+0x432/0x432 [cfg80211 8945aa5bc2af5f6972336665d8ad6f9c191ad5be]
kernel: [ 120.763782] ? nl80211_associate+0x432/0x432 [cfg80211 8945aa5bc2af5f6972336665d8ad6f9c191ad5be]
kernel: [ 120.763802] ? genl_rcv+0x36/0x36
kernel: [ 120.763814] netlink_rcv_skb+0x89/0xf7
kernel: [ 120.763829] genl_rcv+0x28/0x36
kernel: [ 120.763840] netlink_unicast+0x179/0x24b
kernel: [ 120.763854] netlink_sendmsg+0x393/0x401
kernel: [ 120.763872] sock_sendmsg+0x72/0x76
kernel: [ 120.763886] ____sys_sendmsg+0x170/0x1e6
kernel: [ 120.763897] ? copy_msghdr_from_user+0x7a/0xa2
kernel: [ 120.763914] ___sys_sendmsg+0x95/0xd1
kernel: [ 120.763940] __sys_sendmsg+0x85/0xbf
kernel: [ 120.763956] do_syscall_64+0x43/0x55
kernel: [ 120.763966] entry_SYSCALL_64_after_hwframe+0x44/0xa9
kernel: [ 120.763977] RIP: 0033:0x79089f3fcc83
kernel: [ 120.763986] RSP: 002b:00007ffe604f0508 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
kernel: [ 120.763997] RAX: ffffffffffffffda RBX: 000059b40e987690 RCX: 000079089f3fcc83
kernel: [ 120.764006] RDX: 0000000000000000 RSI: 00007ffe604f0558 RDI: 0000000000000009
kernel: [ 120.764014] RBP: 00007ffe604f0540 R08: 0000000000000004 R09: 0000000000400000
kernel: [ 120.764023] R10: 00007ffe604f0638 R11: 0000000000000246 R12: 000059b40ea04980
kernel: [ 120.764032] R13: 00007ffe604f0638 R14: 000059b40e98c360 R15: 00007ffe604f0558
...
kernel: [ 120.765230] INFO: task kworker/u32:26:4239 blocked for more than 20 seconds.
kernel: [ 120.765238] Not tainted 5.10.90 #12
kernel: [ 120.765245] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
kernel: [ 120.765253] task:kworker/u32:26 state:D stack: 0 pid: 4239 ppid: 2 flags:0x00004080
kernel: [ 120.765284] Workqueue: phy0 ieee80211_iface_work [mac80211]
kernel: [ 120.765295] Call Trace:
kernel: [ 120.765306] __schedule+0x785/0x12fa
kernel: [ 120.765316] ? find_held_lock+0x3d/0xb2
kernel: [ 120.765331] schedule+0x7e/0xa1
kernel: [ 120.765340] schedule_preempt_disabled+0x15/0x1e
kernel: [ 120.765349] __mutex_lock_common+0x561/0xc0d
kernel: [ 120.765375] ? ieee80211_sta_work+0x3e/0x1232 [mac80211 335da900954f1c5ea7f1613d92088ce83342042c]
kernel: [ 120.765390] mutex_lock_nested+0x20/0x26
kernel: [ 120.765416] ieee80211_sta_work+0x3e/0x1232 [mac80211 335da900954f1c5ea7f1613d92088ce83342042c]
kernel: [ 120.765430] ? skb_dequeue+0x54/0x5e
kernel: [ 120.765456] ? ieee80211_iface_work+0x7b/0x339 [mac80211 335da900954f1c5ea7f1613d92088ce83342042c]
kernel: [ 120.765485] process_one_work+0x270/0x504
kernel: [ 120.765501] worker_thread+0x215/0x376
kernel: [ 120.765514] kthread+0x159/0x168
kernel: [ 120.765526] ? pr_cont_work+0x5b/0x5b
kernel: [ 120.765536] ? kthread_blkcg+0x31/0x31
kernel: [ 120.765550] ret_from_fork+0x22/0x30
...
kernel: [ 120.765867] Showing all locks held in the system:
...
kernel: [ 120.766164] 5 locks held by wpa_supplicant/924:
kernel: [ 120.766172] #0: ffffffffb1e63eb0 (cb_lock){++++}-{3:3}, at: genl_rcv+0x19/0x36
kernel: [ 120.766197] #1: ffffffffb1e5b1c8 (rtnl_mutex){+.+.}-{3:3}, at: nl80211_pre_doit+0x2a/0x15c [cfg80211]
kernel: [ 120.766238] #2: ffff99f08347cd08 (&wdev->mtx){+.+.}-{3:3}, at: nl80211_deauthenticate+0xde/0x123 [cfg80211]
kernel: [ 120.766279] #3: ffff99f09df12a48 (&local->mtx){+.+.}-{3:3}, at: ieee80211_destroy_auth_data+0x9b/0xe6 [mac80211]
kernel: [ 120.766321] #4: ffff99f09df12ce0 (&local->chanctx_mtx){+.+.}-{3:3}, at: ieee80211_vif_release_channel+0x5e/0x81 [mac80211]
...
kernel: [ 120.766585] 3 locks held by kworker/u32:26/4239:
kernel: [ 120.766593] #0: ffff99f04458f948 ((wq_completion)phy0){+.+.}-{0:0}, at: process_one_work+0x19a/0x504
kernel: [ 120.766621] #1: ffffbad54b3cfe50 ((work_completion)(&sdata->work)){+.+.}-{0:0}, at: process_one_work+0x1c0/0x504
kernel: [ 120.766649] #2: ffff99f08347cd08 (&wdev->mtx){+.+.}-{3:3}, at: ieee80211_sta_work+0x3e/0x1232 [mac80211]

With above info the issue is clear: First wmi_mgmt_tx_work is inserted
to local->workqueue after sdata->work inserted, then wpa_supplicant
acquires wdev->mtx in nl80211_deauthenticate and finally calls
ath11k_mac_op_flush where it waits all mgmt. frames to be sent out by
wmi_mgmt_tx_work. Meanwhile, sdata->work is blocked by wdev->mtx in
ieee80211_sta_work, as a result wmi_mgmt_tx_work has no chance to run.

Change to use ab->workqueue instead of local->workqueue to fix this issue.

Signed-off-by: Baochen Qiang <quic_bqiang@xxxxxxxxxxx>
Signed-off-by: Kalle Valo <quic_kvalo@xxxxxxxxxxx>
Link: https://lore.kernel.org/r/20220217084545.18844-1-quic_bqiang@xxxxxxxxxxx
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
drivers/net/wireless/ath/ath11k/mac.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/wireless/ath/ath11k/mac.c b/drivers/net/wireless/ath/ath11k/mac.c
index 07f499d5ec92..36be4ec969ad 100644
--- a/drivers/net/wireless/ath/ath11k/mac.c
+++ b/drivers/net/wireless/ath/ath11k/mac.c
@@ -5566,7 +5566,7 @@ static int ath11k_mac_mgmt_tx(struct ath11k *ar, struct sk_buff *skb,

skb_queue_tail(q, skb);
atomic_inc(&ar->num_pending_mgmt_tx);
- ieee80211_queue_work(ar->hw, &ar->wmi_mgmt_tx_work);
+ queue_work(ar->ab->workqueue, &ar->wmi_mgmt_tx_work);

return 0;
}
--
2.34.1