You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I tested with glk-003-drop-stable branch(commit 22fa01d) on my LeafHill, arecord can run with HDA analog codec device, but all captured PCM data are 0s.
Update status:
Janusz tested the FW on windows with loop-back mode, found some noise in capture data.
I tested it on Linux, only zero data is gotten in FW and HOST. I checked the output of codec
with an analyzer, the data is sent by codec to DSP. So I think it is very possible that Host
driver sets some HDA registers wrongly
1. Codec sends PCM data to host on HDA bus validated by logic analyser.
2. Host only reads back PCM data as 0s.
So can you check
1. FW link DMA buffer contents, just add some trace to dump first samples in period on every 10th copy. This will show if FW is reading back 0s or not.
2. If FW reads 0s, please check pinmux settings for HDA Rx pin.
@mmaka1 and @tlauda any ideas at your end if FW reads 0s in link DMA buffer ?
I confirm 1 & 2. I also dump data in FW, all zero.
PR thesofproject/linux#194 is to fix this issue. @RanderWang Maybe you can use the magic words "fixes #433" in your PR to automatically close this issue when your PR is merged?
Activity
mengdonglin commentedon Sep 28, 2018
@RanderWang Could you help to share if capture can work on WHL after you enables HD-A on WHL? Thanks
RanderWang commentedon Sep 29, 2018
@mengdonglin yes, all zero. Thank you for your info and I don't have to debug kernel
RanderWang commentedon Oct 16, 2018
@lgirdwood
Update status:
Janusz tested the FW on windows with loop-back mode, found some noise in capture data.
I tested it on Linux, only zero data is gotten in FW and HOST. I checked the output of codec
with an analyzer, the data is sent by codec to DSP. So I think it is very possible that Host
driver sets some HDA registers wrongly
lgirdwood commentedon Oct 16, 2018
@RanderWang ok to confirm
So can you check
FW link DMA buffer contents, just add some trace to dump first samples in period on every 10th copy. This will show if FW is reading back 0s or not.
If FW reads 0s, please check pinmux settings for HDA Rx pin.
@mmaka1 and @tlauda any ideas at your end if FW reads 0s in link DMA buffer ?
RanderWang commentedon Oct 17, 2018
I confirm 1 & 2. I also dump data in FW, all zero.
RanderWang commentedon Oct 17, 2018
@lgirdwood
RanderWang commentedon Oct 17, 2018
I also checked PPLC registers with Janusz, all are correct. Now I ask Janusz for help to dump other registers on Windows.
mengdonglin commentedon Oct 19, 2018
PR thesofproject/linux#194 is to fix this issue.
@RanderWang Maybe you can use the magic words "fixes #433" in your PR to automatically close this issue when your PR is merged?
5 remaining items
mengdonglin commentedon Nov 9, 2018
@RanderWang Is the issue fixed or need any other PR?
RanderWang commentedon Nov 9, 2018
@mengdonglin Another PR
jocelyn-li commentedon Nov 24, 2018
thesofproject/linux#291 is to fix this issue.
jocelyn-li commentedon Dec 7, 2018
@RanderWang can this be closed as thesofproject/linux#291 was merged?
mengdonglin commentedon Dec 17, 2018
Closing this issue since it's fixed by thesofproject/linux#291.
mengdonglin commentedon Dec 17, 2018
Will revisit HDA capture on WHL when #700 is fixed.
RanderWang commentedon Mar 5, 2019
fixed