Repository navigation
[BUG][BYT-NOCODEC][BDW_WSB_RT286]Input/output error when simultaneous-playback-capture / multiple-pipeline-playback #3170
Description
Describe the bug
Input/output error when do simultaneous-playback-capture test
After error occured all the pipeline can work
To Reproduce
1."sudo reboot" to reboot system
2.cd sof-test
3.cd test-case
4.export TPLG=sof-byt-nocodec.tplg
5../simultaneous-playback-capture.sh -l 100
Reproduction Rate
1 round: failed at 13/100
Expected behavior
No error occured
Impact
Input/output error when do simultaneous-playback-capture test of aplay(0,0) and arecord (0,0)
ubuntu@jf-byt-mb-nocodec-1:~$ aplay -l
**** List of PLAYBACK Hardware Devices ****
card 0: sofnocodec [sof-nocodec], device 0: PCM (*) []
Subdevices: 1/1
Subdevice #0: subdevice #0
card 0: sofnocodec [sof-nocodec], device 1: PCM Deep Buffer (*) []
Subdevices: 1/1
Subdevice #0: subdevice #0
ubuntu@jf-byt-mb-nocodec-1:~$ arecord -l
**** List of CAPTURE Hardware Devices ****
card 0: sofnocodec [sof-nocodec], device 0: PCM (*) []
Subdevices: 1/1
Subdevice #0: subdevice #0
No error dmesg,no error sof error trace
Environment
Branch name and commit hash of the 2 repositories: sof (firmware/topology) and linux (kernel driver).
Kernel: {sof-dev fa7850de}
SOF: {master:318dc9f7}
Name of the topology file
Topology: {sof-byt-nocodec.tplg }
Name of the platform(s) on which the bug is observed.
Platform: {BYT MB with nocodec}
Activity
Liviali155 commentedon Jul 14, 2020
On BSW with onboard codec MAX98090 in I2S mode also has this issue with sof-dev(9eb3d58)+master(5564a90)
Failed at 39/50
dmesg0714.log
sof-logger0714.log
mengdonglin commentedon Jul 14, 2020
@slawblauciak sof-byt-nocodec.tplg is created from "sof-cht-nocodec.m4", updated in #3080 to remove SRC for Baytrail and CherryTrail by @plbossart
plbossart commentedon Jul 28, 2020
@Liviali155 can you retry with the SOF PR #3245
The symptoms of no dmesg error, no trace and the -EIO error seem completely aligned with my findings on those platforms.
Liviali155 commentedon Jul 29, 2020
After applied #3245,issue still can be reproduced on BSW with onboard codec MAX98090 in I2S mode and BYT MB with nocodec
dmesg0729.log
sof-logger0729.log
plbossart commentedon Jul 29, 2020
@Liviali155 can you retry with both #3245 and #3257
Somehow I have the feeling we have the same problem of not having enough memory resulting in some sort of underflow error. It's not clear e.g. why we have these repeated sof-logger messages in both issues
Seems to me like the firmware gets lost with bad pointers and can't recover in both #3170 and #3171
@mmaka1 @lgirdwood FYI
plbossart commentedon Jul 29, 2020
@Liviali155 Also wondering if 61a2c75 (' dma: dw: fix locking and calculations in dw_dma_get_data_size') has an impact on multiple pipelines. This changes the behavior for dmic, I am trying to see if reverting it can help. Can you also try on your side.
plbossart commentedon Jul 29, 2020
@Liviali155 Adding PR#3245, @3257 and reverting 61a2c75 seems to solve the issue for 100 iterations. branch here: https://github.com/plbossart/sof/tree/fix/multi-pipelines
I will launch a longer test on my side.
Liviali155 commentedon Jul 30, 2020
@plbossart Used https://github.com/plbossart/sof/tree/fix/multi-pipelines(commit:d9c6405a) to test,issue still can be reproduced on byt-nocodec , failed at 287/1000, seems the reproduce rate is lower than before
sof-logger0730.log
dmesg0730.log
plbossart commentedon Jul 30, 2020
Ack @Liviali155, same on my side. My first run worked for 100 iteration, but on the second a failure happened on iteration 9/1000.
We need to figure this one out, something's not right with scheduling/concurrency. @lgirdwood FYI
lgirdwood commentedon Aug 4, 2020
I don't know how to turn on mixer debug?
Mixers are used in all platforms, but here indeed the use of the mixer is different: it's the first element in a 'DAI' pipeline
lgirdwood commentedon Aug 5, 2020
Oh, I think individual debug is still blocked on UUID PR, Maybe just be easier to change trace_dbg to trace_err in mixer_copy().
Yep, but I no longer see a sof-byt-nocodec.m4 topology in master, so could your topology binary be stale here and potentially causing an issue ?
I do suspect that mixer is not correctly getting the state of it's sources/sinks and this results in 0 bytes to copy, but then I've no idea why does not complain about over/under runs....
29 remaining items
plbossart commentedon Jan 27, 2022
FWIW we seem to have an interrupt issue on Broadwell thesofproject/linux#3400
marc-hb commentedon Feb 14, 2022
Still happening in daily 10146?model=BDW_WSB_RT286&testcase=multiple-pipeline-capture-50
Start Time: 2022-02-11 22:27:26 UTC
Kernel Branch: topic/sof-dev
Kernel Commit: 98119478
SOF Branch: main
SOF Commit: b8954754f055
multiple-pipeline.sh: temporarily disable capture on BDW
mengdonglin commentedon Feb 22, 2022
As Broadwell (BDW) is a very old platform, we lower its priority and will not fix issues with multi-pipeline test cases on BDW.
marc-hb commentedon Feb 22, 2022
The test still failed yesterday in 10402?model=BDW_WSB_RT286&testcase=multiple-pipeline-capture-50, let's close this when thesofproject/sof-test#863 is merged so we stop testing this every day like someone is assigned to it.
We had only 6 distinct failures in 10402 and this was one of them.
XiaoyunWu6666 commentedon Feb 23, 2022
@marc-hb I think we can close this again since https://github.com/intel-innersource/drivers.audio.ci.sof-framework/pull/185 got merged and also see current test 10444
[console]
multiple-pipeline.sh: temporarily disable capture on BDW