Repository navigation
System cannot resume from S3 when playback on APL-nocodec. #592
Description
Summary:
System cannot resume from S3 when playback on APL-nocodec, and the system freeze caused no dmesg or logger info.
BTW, if there is no aplay, rtcwake works fine.
Test steps:
1. aplay -Dhw:0,0 -c2 -fS16_LE -vv -i music.wav # Terminal 1
2. sudo rtcwake -m mem -s 2 # Terminal 2
Test env:
sof master: 8051fe3
kernel sof-dev: e4fb585
tplg: sof-apl-nocodec.tplg
Activity
mengdonglin commentedon Jan 24, 2019
@markyang @stevyan could you help to bisect?
libinyang commentedon Jan 25, 2019
As Ranjani mentioned, #582 can fix it. Let's wait for 582 merged.
wenqingfu commentedon Jan 25, 2019
bi-sect always helps since it's such a new issue @markyang
ranj063 commentedon Jan 27, 2019
@wenqingfu I'm not sure this is a regression. This should never have worked before PR #582
wenqingfu commentedon Jan 27, 2019
it was green most time in the daily regression reports last week. Maybe this bug is on nocodec? or @markyang was testing something wrong ...
keyonjie commentedon Jan 28, 2019
I remember it worked when I reported this(thesofproject/sof#788) 25 days ago.
markyang commentedon Jan 28, 2019
The last working test env:
the daily build is here: http://otc-sof-sh-build-01.sh.intel.com/daily/20190122/
I am going to bisect if I'm free.
mengdonglin commentedon Jan 28, 2019
@markyang @stevyan Please help to verify if PR #582 can fix this issue.
According to thesofproject/sof#788 (comment), HDMI playback cannot continue after a S3 cycle on APL w/ pcm512 codec. And I believe S3 never worked on APL in nocodec mode.
DarrenxWang commentedon Jan 29, 2019
The CML has the same issue.
Test env:
sof cml-001-drop-stable: 1f5a965
kernel topic/cros-sof-v4.19: f69d618
tplg: sof-cml-rt5682.tplg
2 remaining items
emilchudzik commentedon Jan 29, 2019
I have repro using our test env to check D3/D0.
Issue redirect to Bartosz Kokoszko for deeper debugging.
@markyang could you attach dmesg log from your env?
markyang commentedon Jan 30, 2019
The last working daily build is here: http://otc-sof-sh-build-01.sh.intel.com/daily/20190122/
markyang commentedon Jan 30, 2019
@emilchudzik I don't know how to get the dmesg log when a device freezes, do you have any good idea?
stevyan commentedon Jan 30, 2019
Verify Result:
This issue can be fixed with PR#582 based on APL-nocodec.
Test Env:
sof master: 365b132
kernel sof-dev: pr582-8
tplg: sof-apl-nocodec.tplg
Reference Info:
libinyang commentedon Jan 30, 2019
OK. So let wait for #582 merging.
DarrenxWang commentedon Jan 30, 2019
We apply PR #582 to topic/cros-sof-v4.19 kernel branch. Now CML can suspend, but can’t resume.
markyang commentedon Jan 30, 2019
bi-sect results
revert c74579a(ASoC: intel: skl: Fix display power regression), then S3 works.
mengdonglin commentedon Jan 30, 2019
@markyang Good catch! Libin confirmed it's this patch c74579a caused regression on APL in no-codec mode and will debug it.
emilchudzik commentedon Jan 30, 2019
Is this revert applies also for CML?
markyang commentedon Feb 2, 2019
Summary:
This issue cannot be reproduced on APL-nocodec anymore.
Test env:
sof master: 29f7b31
kernel sof-dev: 95a0bc3
tplg: sof-apl-pcm512x.tplg
mengdonglin commentedon Feb 10, 2019
I think this issue is solved on APL by merging PR #582
@xun2z Please verify the issue for CML (WHL in I2S mode). If you still see the system needs a long time to resume from S3, please file a new issue for CML.
selftests/bpf: Add test for stale pkt range after scalar arithmetic