rtw88: avoid circular locking between local->iflist_mtx and rtwdev->mutex

Remove circular lock dependency by using atomic version of interfaces
iterate in watch_dog_work(), hence avoid taking local->iflist_mtx
(rtw_vif_watch_dog_iter() only update some data, it can be called from
atomic context). Fixes below LOCKDEP warning:

[ 1157.219415] ======================================================
[ 1157.225772] [ INFO: possible circular locking dependency detected ]
[ 1157.232150] 3.10.0-1043.el7.sgruszka1.x86_64.debug #1 Not tainted
[ 1157.238346] -------------------------------------------------------
[ 1157.244635] kworker/u4:2/14490 is trying to acquire lock:
[ 1157.250194]  (&rtwdev->mutex){+.+.+.}, at: [<ffffffffc098322b>] rtw_ops_config+0x2b/0x90 [rtw88]
[ 1157.259151]
but task is already holding lock:
[ 1157.265085]  (&local->iflist_mtx){+.+...}, at: [<ffffffffc0b8ab7a>] ieee80211_mgd_probe_ap.part.28+0xca/0x160 [mac80211]
[ 1157.276169]
which lock already depends on the new lock.

[ 1157.284488]
the existing dependency chain (in reverse order) is:
[ 1157.292101]
-> #2 (&local->iflist_mtx){+.+...}:
[ 1157.296919]        [<ffffffffbc741a29>] lock_acquire+0x99/0x1e0
[ 1157.302955]        [<ffffffffbce72793>] mutex_lock_nested+0x93/0x410
[ 1157.309416]        [<ffffffffc0b6038f>] ieee80211_iterate_interfaces+0x2f/0x60 [mac80211]
[ 1157.317730]        [<ffffffffc09811ab>] rtw_watch_dog_work+0xcb/0x130 [rtw88]
[ 1157.325003]        [<ffffffffbc6d77bc>] process_one_work+0x22c/0x720
[ 1157.331481]        [<ffffffffbc6d7dd6>] worker_thread+0x126/0x3b0
[ 1157.337589]        [<ffffffffbc6e107f>] kthread+0xef/0x100
[ 1157.343260]        [<ffffffffbce848b7>] ret_from_fork_nospec_end+0x0/0x39
[ 1157.350091]
-> #1 ((&(&rtwdev->watch_dog_work)->work)){+.+...}:
[ 1157.356314]        [<ffffffffbc741a29>] lock_acquire+0x99/0x1e0
[ 1157.362427]        [<ffffffffbc6d570b>] flush_work+0x5b/0x310
[ 1157.368287]        [<ffffffffbc6d740e>] __cancel_work_timer+0xae/0x170
[ 1157.374940]        [<ffffffffbc6d7583>] cancel_delayed_work_sync+0x13/0x20
[ 1157.381930]        [<ffffffffc0982b49>] rtw_core_stop+0x29/0x50 [rtw88]
[ 1157.388679]        [<ffffffffc098bee6>] rtw_enter_ips+0x16/0x20 [rtw88]
[ 1157.395428]        [<ffffffffc0983242>] rtw_ops_config+0x42/0x90 [rtw88]
[ 1157.402173]        [<ffffffffc0b13343>] ieee80211_hw_config+0xc3/0x680 [mac80211]
[ 1157.409854]        [<ffffffffc0b3925b>] ieee80211_do_open+0x69b/0x9c0 [mac80211]
[ 1157.417418]        [<ffffffffc0b395e9>] ieee80211_open+0x69/0x70 [mac80211]
[ 1157.424496]        [<ffffffffbcd03442>] __dev_open+0xe2/0x160
[ 1157.430356]        [<ffffffffbcd03773>] __dev_change_flags+0xa3/0x180
[ 1157.436922]        [<ffffffffbcd03879>] dev_change_flags+0x29/0x60
[ 1157.443224]        [<ffffffffbcda14c4>] devinet_ioctl+0x794/0x890
[ 1157.449331]        [<ffffffffbcda27b5>] inet_ioctl+0x75/0xa0
[ 1157.455087]        [<ffffffffbccd54eb>] sock_do_ioctl+0x2b/0x60
[ 1157.461178]        [<ffffffffbccd5753>] sock_ioctl+0x233/0x310
[ 1157.467109]        [<ffffffffbc8bd820>] do_vfs_ioctl+0x410/0x6c0
[ 1157.473233]        [<ffffffffbc8bdb71>] SyS_ioctl+0xa1/0xc0
[ 1157.478914]        [<ffffffffbce84a5e>] system_call_fastpath+0x25/0x2a
[ 1157.485569]
-> #0 (&rtwdev->mutex){+.+.+.}:
[ 1157.490022]        [<ffffffffbc7409d1>] __lock_acquire+0xec1/0x1630
[ 1157.496305]        [<ffffffffbc741a29>] lock_acquire+0x99/0x1e0
[ 1157.502413]        [<ffffffffbce72793>] mutex_lock_nested+0x93/0x410
[ 1157.508890]        [<ffffffffc098322b>] rtw_ops_config+0x2b/0x90 [rtw88]
[ 1157.515724]        [<ffffffffc0b13343>] ieee80211_hw_config+0xc3/0x680 [mac80211]
[ 1157.523370]        [<ffffffffc0b8a4ca>] ieee80211_recalc_ps.part.27+0x9a/0x180 [mac80211]
[ 1157.531685]        [<ffffffffc0b8abc5>] ieee80211_mgd_probe_ap.part.28+0x115/0x160 [mac80211]
[ 1157.540353]        [<ffffffffc0b8b40d>] ieee80211_beacon_connection_loss_work+0x4d/0x80 [mac80211]
[ 1157.549513]        [<ffffffffbc6d77bc>] process_one_work+0x22c/0x720
[ 1157.555886]        [<ffffffffbc6d7dd6>] worker_thread+0x126/0x3b0
[ 1157.562170]        [<ffffffffbc6e107f>] kthread+0xef/0x100
[ 1157.567765]        [<ffffffffbce848b7>] ret_from_fork_nospec_end+0x0/0x39
[ 1157.574579]
other info that might help us debug this:

[ 1157.582788] Chain exists of:
  &rtwdev->mutex --> (&(&rtwdev->watch_dog_work)->work) --> &local->iflist_mtx

[ 1157.593024]  Possible unsafe locking scenario:

[ 1157.599046]        CPU0                    CPU1
[ 1157.603653]        ----                    ----
[ 1157.608258]   lock(&local->iflist_mtx);
[ 1157.612180]                                lock((&(&rtwdev->watch_dog_work)->work));
[ 1157.620074]                                lock(&local->iflist_mtx);
[ 1157.626555]   lock(&rtwdev->mutex);
[ 1157.630124]
 *** DEADLOCK ***

[ 1157.636148] 4 locks held by kworker/u4:2/14490:
[ 1157.640755]  #0:  (%s#6){.+.+.+}, at: [<ffffffffbc6d774a>] process_one_work+0x1ba/0x720
[ 1157.648965]  #1:  ((&ifmgd->beacon_connection_loss_work)){+.+.+.}, at: [<ffffffffbc6d774a>] process_one_work+0x1ba/0x720
[ 1157.659950]  #2:  (&wdev->mtx){+.+.+.}, at: [<ffffffffc0b8aad5>] ieee80211_mgd_probe_ap.part.28+0x25/0x160 [mac80211]
[ 1157.670901]  #3:  (&local->iflist_mtx){+.+...}, at: [<ffffffffc0b8ab7a>] ieee80211_mgd_probe_ap.part.28+0xca/0x160 [mac80211]
[ 1157.682466]

Fixes: e3037485c68e ("rtw88: new Realtek 802.11ac driver")
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Acked-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

1 files changed, 2 insertions, 1 deletions

diff --git a/drivers/net/wireless/realtek/rtw88/main.c b/drivers/net/wireless/realtek/rtw88/main.c
index f447361f7573..b2dac4609138 100644
--- a/drivers/net/wireless/realtek/rtw88/main.c
+++ b/drivers/net/wireless/realtek/rtw88/main.c
@@ -162,7 +162,8 @@ static void rtw_watch_dog_work(struct work_struct *work)
 	rtwdev->stats.tx_cnt = 0;
 	rtwdev->stats.rx_cnt = 0;
 
-	rtw_iterate_vifs(rtwdev, rtw_vif_watch_dog_iter, &data);
+	/* use atomic version to avoid taking local->iflist_mtx mutex */
+	rtw_iterate_vifs_atomic(rtwdev, rtw_vif_watch_dog_iter, &data);
 
 	/* fw supports only one station associated to enter lps, if there are
 	 * more than two stations associated to the AP, then we can not enter