]> git.ipfire.org Git - thirdparty/linux.git/commit
leds: avoid flush_work in atomic context
authorPavel Machek <pavel@ucw.cz>
Sun, 26 May 2019 07:38:55 +0000 (09:38 +0200)
committerJacek Anaszewski <jacek.anaszewski@gmail.com>
Fri, 31 May 2019 20:29:14 +0000 (22:29 +0200)
commit8c0f693c6effbc3f42f77a9e81209af9af20910c
treedfefc946d53a23b3873a8e86155edbf63bdcaf57
parentcd6c84d8f0cdc911df435bb075ba22ce3c605b07
leds: avoid flush_work in atomic context

It turns out that various triggers use led_blink_setup() from atomic
context, so we can't do a flush_work there. Flush is still needed for
slow LEDs, but we can move it to sysfs code where it is safe.

    WARNING: inconsistent lock state
    5.2.0-rc1 #1 Tainted: G        W
    --------------------------------
    inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage.
    swapper/1/0 [HC0[0]:SC1[1]:HE1:SE0] takes:
    000000006e30541b
    ((work_completion)(&led_cdev->set_brightness_work)){+.?.}, at:
    +__flush_work+0x3b/0x38a
    {SOFTIRQ-ON-W} state was registered at:
      lock_acquire+0x146/0x1a1
     __flush_work+0x5b/0x38a
     flush_work+0xb/0xd
     led_blink_setup+0x1e/0xd3
     led_blink_set+0x3f/0x44
     tpt_trig_timer+0xdb/0x106
     ieee80211_mod_tpt_led_trig+0xed/0x112

Fixes: 0db37915d912 ("leds: avoid races with workqueue")
Signed-off-by: Pavel Machek <pavel@ucw.cz>
Tested-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Jacek Anaszewski <jacek.anaszewski@gmail.com>
drivers/leds/led-core.c
drivers/leds/trigger/ledtrig-timer.c