On Thu, 04 Feb 2021 18:27:24 +0100, Ranjani Sridharan wrote:
From: Sathyanarayana Nujella sathyanarayana.nujella@intel.com
During Coldboot stress tests, system encountered the following panic. Panic logs depicts rt5682_i2c_shutdown() happened first and then later jack detect handler workqueue function triggered. This situation causes panic as rt5682_i2c_shutdown() resets codec. Fix this panic by cancelling properly delayed work.
Panic log: [ 20.936124] sof_pci_shutdown [ 20.940248] snd_sof_device_shutdown [ 20.945023] snd_sof_shutdown [ 21.126849] rt5682_i2c_shutdown [ 21.286053] rt5682_jack_detect_handler [ 21.291235] BUG: kernel NULL pointer dereference, address: 000000000000037c [ 21.299302] #PF: supervisor read access in kernel mode [ 21.305254] #PF: error_code(0x0000) - not-present page [ 21.311218] PGD 0 P4D 0 [ 21.314155] Oops: 0000 [#1] PREEMPT SMP NOPTI [ 21.319206] CPU: 2 PID: 123 Comm: kworker/2:3 Tainted: G U 5.4.68 #10 [ 21.333687] ACPI: Preparing to enter system sleep state S5 [ 21.337669] Workqueue: events_power_efficient rt5682_jack_detect_handler [snd_soc_rt5682] [ 21.337671] RIP: 0010:rt5682_jack_detect_handler+0x6c/0x279 [snd_soc_rt5682]
Fixes: a50067d4f3c1d ('ASoC: rt5682: split i2c driver into separate module') Signed-off-by: Jairaj Arava jairaj.arava@intel.com Signed-off-by: Sathyanarayana Nujella sathyanarayana.nujella@intel.com Reviewed-by: Pierre-Louis Bossart pierre-louis.bossart@intel.com Reviewed-by: Shuming Fan shumingf@realtek.com Signed-off-by: Ranjani Sridharan ranjani.sridharan@linux.intel.com
sound/soc/codecs/rt5682-i2c.c | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/sound/soc/codecs/rt5682-i2c.c b/sound/soc/codecs/rt5682-i2c.c index 37d13120f5ba..6effd2611e3c 100644 --- a/sound/soc/codecs/rt5682-i2c.c +++ b/sound/soc/codecs/rt5682-i2c.c @@ -273,6 +273,8 @@ static void rt5682_i2c_shutdown(struct i2c_client *client) { struct rt5682_priv *rt5682 = i2c_get_clientdata(client);
- cancel_delayed_work_sync(&rt5682->jack_detect_work);
- rt5682_reset(rt5682);
}
The driver has two works. Doesn't another one need Canceling? Just to be sure...
thanks,
Takashi