[BUG] [CHT] Underflow/errors reported in multiple-playback CI tests. #3534

Activity

1. plbossart

   added

   bugSomething isn't working as expectedSomething isn't working as expected

   on Oct 19, 2020
2. plbossart commented on Oct 19, 2020

   

   plbossart

   on Oct 19, 2020

   MemberAuthor

   More actions

   Re-running the tests shows no issue reported (https://sof-ci.01.org/linuxpr/PR2500/build4703/devicetest/), so this is a transient error condition.
3. mengdonglin

   added

   CHTApplies to Cherry Trail platformApplies to Cherry Trail platform

   on Oct 20, 2020
4. xiulipan commented on Oct 20, 2020

   

   xiulipan

   on Oct 20, 2020

   Contributor

   More actions

   @plbossart will this be related with the IPC position update delay? We do see such write I/O error on legacy platforms(BYT/CHT/BDW) using IPC position update.
   If enable IPC position update for HDA platforms, we will also see such kind of I/O error.
5. lgirdwood commented on Oct 20, 2020

   

   lgirdwood

   on Oct 20, 2020

   Member

   More actions

   @plbossart @xiulipan we should not be reading back position via IPC but reading back from SRAM window on CHT
6. plbossart commented on Oct 20, 2020

   

   plbossart

   on Oct 20, 2020

   MemberAuthor

   More actions

   @lgirdwood can you double-check the linux driver and let me know if we are doing things the right way? This position report has been a problem for the longest time but only you and @keyonjie understand it.
7. lgirdwood commented on Oct 20, 2020

   

   lgirdwood

   on Oct 20, 2020

   Member

   More actions

   @plbossart pointer callback is not defined for BYT/CHT so it's using mailbox and NOT IPC method.

   static snd_pcm_uframes_t sof_pcm_pointer(struct snd_soc_component *component,
   					 struct snd_pcm_substream *substream)
   {
   	struct snd_soc_pcm_runtime *rtd = asoc_substream_to_rtd(substream);
   	struct snd_sof_dev *sdev = snd_soc_component_get_drvdata(component);
   	struct snd_sof_pcm *spcm;
   	snd_pcm_uframes_t host, dai;
   
   	/* nothing to do for BE */
   	if (rtd->dai_link->no_pcm)
   		return 0;
   
   	/* use dsp ops pointer callback directly if set */
   	if (sof_ops(sdev)->pcm_pointer)
   		return sof_ops(sdev)->pcm_pointer(sdev, substream);
   
   	spcm = snd_sof_find_spcm_dai(component, rtd);
   	if (!spcm)
   		return -EINVAL;
   
   	/* read position from DSP */
   	host = bytes_to_frames(substream->runtime,
   			       spcm->stream[substream->stream].posn.host_posn);
   	dai = bytes_to_frames(substream->runtime,
   			      spcm->stream[substream->stream].posn.dai_posn);
   
   	dev_vdbg(component->dev,
   		 "PCM: stream %d dir %d DMA position %lu DAI position %lu\n",
   		 spcm->pcm.pcm_id, substream->stream, host, dai);
   
   	return host;
   }
8. plbossart commented on Oct 20, 2020

   

   plbossart

   on Oct 20, 2020

   MemberAuthor

   More actions

   @plbossart pointer callback is not defined for BYT/CHT so it's using mailbox and NOT IPC method.

   ok, so now I have no idea why we only look at the position during an IPC and why we enabled the BATCH option to disable timer-based scheduling in PulseAudio.
9. lgirdwood commented on Oct 21, 2020

   

   lgirdwood

   on Oct 21, 2020

   Member

   More actions

   @plbossart pointer callback is not defined for BYT/CHT so it's using mailbox and NOT IPC method.

   ok, so now I have no idea why we only look at the position during an IPC and why we enabled the BATCH option to disable timer-based scheduling in PulseAudio.

   Might be worthwhile enabling the trace here in the kernel (as FW could be sending garbage at time or sending for wrong stream).
10. plbossart commented on Oct 21, 2020

    

    plbossart

    on Oct 21, 2020

    MemberAuthor

    More actions

    @lgirdwood the DMA trace is always enabled in our tests.
11. lgirdwood commented on Oct 21, 2020

    

    lgirdwood

    on Oct 21, 2020

    Member

    More actions

    @plbossart sorry I mean from above kernel code (which may be off as it's verbose)

    dev_vdbg(component->dev,
    		 "PCM: stream %d dir %d DMA position %lu DAI position %lu\n",
    		 spcm->pcm.pcm_id, substream->stream, host, dai);
12. mengdonglin

    added

    won't fixThis will not be worked on atm (e.g. a bug closed for lack of user request, hardware etc)This will not be worked on atm (e.g. a bug closed for lack of user request, hardware etc)

    on Jun 21, 2022
13. mengdonglin commented on Jun 21, 2022

    

    mengdonglin

    on Jun 21, 2022

    Collaborator

    More actions

    Move CHT platform to maintenance mode and SOF v2.2 will be the last upstream release with support for this platform. There will be no more FW bug fixing effort on this platform.
14. mengdonglin

    closed this as completedon Jun 21, 2022