tplgtool2.py: show PCM names on graph output by marc-hb · Pull Request #1054 · thesofproject/sof-test

marc-hb · 2023-06-09T22:15:17Z

No description provided.

Fixed a quadruple """" that broke the parser in some editors. Minor rewording. Signed-off-by: Marc Herbert <marc.herbert@intel.com>

marc-hb · 2023-06-09T22:38:02Z

Seems to work great... for topology1. ~~There's no "PCMxP" in topology2~~. EDIT: hopefully fixed now, parsing host-copier.N

https://sof-ci.01.org/softestpr/PR1054/build417/devicetest/index.html

marc-hb · 2023-06-10T04:36:50Z

Besides host.copier.7.capture having no corresponding PCM in https://sof-ci.01.org/softestpr/PR1054/build420/devicetest/index.html?model=TGLU_RVP_NOCODEC_IPC4ZPH&testcase=verify-sof-firmware-load and in some MTL devices, the graphs looks much better with PCM names.

In any case it seems a bit weird to have to parse the widget name... is there no simpler widget->pcm connection in the topology?

EDIT: not really, see https://github.com/thesofproject/linux/blob/2776754a4cd984a390a17a0c96635beb2ad408d8/sound/soc/sof/topology.c#L1148

RanderWang

LGTM

plbossart · 2023-06-12T16:33:00Z

+        assert n_l <= 1, f"Unexpected number of PCMs matching pcm_id={pcm_id}: {n_l} ({pcm_names})"
+
+        return pcm_names[0] if n_l == 1 else f"No PCM with ID={pcm_id}"
+


IIRC @ranj063 already did some changes in the topology2 naming, not sure if these changes are based on top of the topology2 changes?

In addition, while the name of the PCM device is nice, the PCM device number is the most useful for testing. How difficult would it be to add the hw: ID?

IIRC @ranj063 already did some changes in the topology2 naming, not sure if these changes are based on top of the topology2 changes?

Are these changes merged yet? In which git repo?

This is sof-test: it's a different git repo so the tests are running with the latest of everything that was merged. I think. EDIT: as a sheer coincidence just discovered sof-test is tested with yesterday's SOF.

How difficult would it be to add the hw: ID?

Isn't that the pcm_id I'm matching against? If not then what is it? @ranj063 can you help explain? I'm a total topology noob sorry, just trying stuff and marvelling at how lucky I am to make the graphs look so much better with so little Python code.

IIRC @ranj063 already did some changes in the topology2 naming, not sure if these changes are based on top of the topology2 changes?

In addition, while the name of the PCM device is nice, the PCM device number is the most useful for testing. How difficult would it be to add the hw: ID?

@plbossart I think this is a good change on top of what I did with the naming. With a name like host-copier.1.playback you've already got the PCM ID i.e. 1 and this adds the name of the PCM next to it

ranj063 · 2023-06-15T21:00:22Z

Besides host.copier.7.capture having no corresponding PCM in https://sof-ci.01.org/softestpr/PR1054/build420/devicetest/index.html?model=TGLU_RVP_NOCODEC_IPC4ZPH&testcase=verify-sof-firmware-load and in some MTL devices, the graphs looks much better with PCM names.

In any case it seems a bit weird to have to parse the widget name... is there no simpler widget->pcm connection in the topology?

fixed now with thesofproject/sof#7747

marc-hb · 2023-06-16T06:25:54Z

+        if match is None:
+            # Don't crash the entire program when passed a wrong input;
+            # put an error message in the graph instead.
+            return f'No PCM or host-copier number found in "{wname}"'


From @ranj063 : this is a topology error, make it a hard fail.

marc-hb · 2023-07-06T02:33:42Z

I started running the new version manually on all .tplg files and I'm finding new errors:

topology/topology1/production/sof-tgl-nocodec.tplg
topology/topology1/production/sof-adl-nocodec.tplg
etc.

pcm=Port0;id=0;type=duplex;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=2;ch_max=2;
pcm=Port1;id=1;type=duplex;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=2;ch_max=2;
pcm=Port2;id=2;type=duplex;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=2;ch_max=2;
pcm=DMIC;id=10;type=capture;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=4;ch_max=4;
pcm=DMIC16kHz;id=11;type=capture;fmt=S16_LE;rate_min=16000;rate_max=16000;ch_min=4;ch_max=4;

AssertionError: Zero PCM with ID=3 found for widget=PCM3P

Also affects apl-nocodec in
https://sof-ci.01.org/softestpr/PR1054/build565/devicetest/index.html?model=APL_UP2_NOCODEC&testcase=verify-tplg-binary

pcm=DMIC;id=10;type=capture;cap_name=Highpass Capture 10;eq=EQIIR20.0;fmts=S16_LE S24_LE S32_LE;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=4;ch_max=4;rate=48000;channel=4;dev=hw:0,10;snd=/dev/snd/pcmC0D10c;
pcm=DMIC16kHz;id=11;type=capture;cap_name=Highpass Capture 11;eq=EQIIR21.0;fmts=S16_LE S24_LE S32_LE;fmt=S16_LE;rate_min=16000;rate_max=16000;ch_min=4;ch_max=4;rate=16000;channel=4;dev=hw:0,11;snd=/dev/snd/pcmC0D11c;
pcm=Port5;id=2;type=playback;cap_name=Playback 2;pga=PGA9.0 PGA5.0;fmts=S16_LE S24_LE S32_LE;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=2;ch_max=2;rate=48000;channel=2;dev=hw:0,2;snd=/dev/snd/pcmC0D2p;
pcm=Port5;id=2;type=capture;cap_name=Passthrough Capture 2;pga=PGA6.0;fmts=S16_LE S24_LE S32_LE;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=2;ch_max=2;rate=48000;channel=2;dev=hw:0,2;snd=/dev/snd/pcmC0D2c;
pcm=Port5 Deep Buffer;id=3;type=playback;cap_name=Playback 3;pga=PGA11.0 PGA5.0;fmts=S16_LE S24_LE S32_LE;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=2;ch_max=2;rate=48000;channel=2;dev=hw:0,3;snd=/dev/snd/pcmC0D3p;


topology/topology1/production/sof-ehl-rt5660-nohdmi.tplg --- 

pcm=Headset;id=0;type=duplex;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=2;ch_max=2;
pcm=DMIC;id=1;type=capture;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=4;ch_max=4;
pcm=DMIC16kHz;id=2;type=capture;fmt=S16_LE;rate_min=16000;rate_max=16000;ch_min=2;ch_max=2;

AssertionError: Zero PCM with ID=3 found for widget=PCM3C

EDIT: the assert interruption was the most likely reason why the graph below was incomplete. Ignore this graph!

~~Weird~~: without interruption: https://sof-ci.01.org/softestpr/PR1075/build549/devicetest/index.html?model=APL_UP2_NOCODEC&testcase=verify-tplg-binary

pcm=DMIC;id=10;type=capture;cap_name=Highpass Capture 10;eq=EQIIR20.0;fmts=S16_LE S24_LE S32_LE;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=4;ch_max=4;rate=48000;channel=4;dev=hw:0,10;snd=/dev/snd/pcmC0D10c;
pcm=DMIC16kHz;id=11;type=capture;cap_name=Highpass Capture 11;eq=EQIIR21.0;fmts=S16_LE S24_LE S32_LE;fmt=S16_LE;rate_min=16000;rate_max=16000;ch_min=4;ch_max=4;rate=16000;channel=4;dev=hw:0,11;snd=/dev/snd/pcmC0D11c;
pcm=Port5;id=2;type=playback;cap_name=Playback 2;pga=PGA9.0 PGA5.0;fmts=S16_LE S24_LE S32_LE;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=2;ch_max=2;rate=48000;channel=2;dev=hw:0,2;snd=/dev/snd/pcmC0D2p;
pcm=Port5;id=2;type=capture;cap_name=Passthrough Capture 2;pga=PGA6.0;fmts=S16_LE S24_LE S32_LE;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=2;ch_max=2;rate=48000;channel=2;dev=hw:0,2;snd=/dev/snd/pcmC0D2c;
pcm=Port5 Deep Buffer;id=3;type=playback;cap_name=Playback 3;pga=PGA11.0 PGA5.0;fmts=S16_LE S24_LE S32_LE;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=2;ch_max=2;rate=48000;channel=2;dev=hw:0,3;snd=/dev/snd/pcmC0D3p;

marc-hb · 2023-07-06T02:38:12Z

topology/topology2/sof-ace-tplg/sof-mtl-max98357a-rt5682.tplg

pcm=EchoRef;id=27;type=capture;fmt=S16_LE;rate_min=48000;rate_max=48000;ch_min=2;ch_max=2;
pcm=DMIC0 RTC AEC;id=27;type=capture;fmt=S32_LE;rate_min=48000;rate_max=48000;ch_min=4;ch_max=4;
pcm=DMIC0 RTC AEC NS;id=28;type=capture;fmt=S32_LE;rate_min=48000;rate_max=48000;ch_min=4;ch_max=4;


AssertionError: More than one PCM has pcm_id=27: 2 (['EchoRef', 'DMIC0 RTC AEC'])

RanderWang

LGTM

The guessing can stop. Signed-off-by: Marc Herbert <marc.herbert@intel.com>

The graph is still produced but now there's something that can be used for testing. Signed-off-by: Marc Herbert <marc.herbert@intel.com>

marc-hb · 2023-07-06T20:32:30Z

graph still missing in https://sof-ci.01.org/softestpr/PR1054/build568/devicetest/index.html?model=APL_UP2_NOCODEC&testcase=verify-tplg-binary, DUT-env.py in the CI framework assumes there's no output in case of error. Working on that.

marc-hb · 2023-07-06T23:46:11Z

SOFCI TEST

EDIT: https://sof-ci.01.org/softestpr/PR1054/build576/devicetest/index.html shows the APL_UP_NOCODEC graph with the missing pcm_id

marc-hb · 2023-07-07T00:51:19Z

For a while I naively thought I could quickly fix the problematic topologies above - and then merge this. I have neither enough time nor knowledge for that. So the most recent and... perfect-at-last revision does a "soft fail": it still produces the graph even when a PCM id is not found. That's what it used to do before this PR and graph is probably useful for troubleshooting anyway so I think it's the best design after all. An error message is printed on stdout and placed in the graph too, see example in https://sof-ci.01.org/softestpr/PR1054/build576/devicetest/index.html?model=APL_UP2_NOCODEC&testcase=verify-tplg-binary

There's also a non-zero exit status.

MTL https://sof-ci.01.org/softestpr/PR1054/build577/devicetest/index.html and stable-v2.2 have the small and usual amount of suspend/resume failures; nothing else

https://sof-ci.01.org/softestpr/PR1054/build575/devicetest/index.html?model=TGLU_RVP_SDW_IPC4ZPH&testcase=multiple-pause-resume-5 is known. Everything else green there.

RanderWang

LGTM

marc-hb · 2023-07-07T14:40:03Z

thanks @RanderWang for your patience through the error-handling back and forth!

tplgtool2.py: minor docstring fixes

b77dc13

Fixed a quadruple """" that broke the parser in some editors. Minor rewording. Signed-off-by: Marc Herbert <marc.herbert@intel.com>

marc-hb requested review from RanderWang, fredoh9, jsarha, paulstelian97 and ranj063 June 9, 2023 22:38

marc-hb force-pushed the PCM-names branch from 3eb2a5e to 235d70e Compare June 9, 2023 23:55

marc-hb marked this pull request as ready for review June 10, 2023 00:23

marc-hb requested a review from a team as a code owner June 10, 2023 00:23

marc-hb requested review from kv2019i and plbossart June 10, 2023 04:38

RanderWang previously approved these changes Jun 12, 2023

View reviewed changes

plbossart reviewed Jun 12, 2023

View reviewed changes

marc-hb marked this pull request as draft June 12, 2023 19:15

marc-hb requested a review from juimonen June 12, 2023 20:48

marc-hb marked this pull request as ready for review June 13, 2023 17:22

ranj063 previously approved these changes Jun 15, 2023

View reviewed changes

marc-hb marked this pull request as draft June 15, 2023 21:52

marc-hb commented Jun 16, 2023

View reviewed changes

plbossart previously approved these changes Jun 16, 2023

View reviewed changes

marc-hb mentioned this pull request Jun 22, 2023

[BUG] Remove buffer core ID from topology viewer #925

Open

marc-hb dismissed stale reviews from plbossart, ranj063, and RanderWang via 902e261 July 6, 2023 01:56

marc-hb force-pushed the PCM-names branch from 235d70e to 902e261 Compare July 6, 2023 01:56

marc-hb marked this pull request as ready for review July 6, 2023 02:45

RanderWang previously approved these changes Jul 6, 2023

View reviewed changes

marc-hb added 2 commits July 6, 2023 18:29

tplgtool2.py: show PCM names on graph output

5553e2f

The guessing can stop. Signed-off-by: Marc Herbert <marc.herbert@intel.com>

tplgtool2.py: use the number of graph.errors as an exit code

4d4163d

The graph is still produced but now there's something that can be used for testing. Signed-off-by: Marc Herbert <marc.herbert@intel.com>

marc-hb dismissed RanderWang’s stale review via 4d4163d July 6, 2023 18:30

marc-hb force-pushed the PCM-names branch from 902e261 to 4d4163d Compare July 6, 2023 18:30

This comment was marked as outdated.

Sign in to view

RanderWang approved these changes Jul 7, 2023

View reviewed changes

marc-hb merged commit eb9755e into thesofproject:main Jul 7, 2023

marc-hb deleted the PCM-names branch July 7, 2023 14:40

		assert n_l <= 1, f"Unexpected number of PCMs matching pcm_id={pcm_id}: {n_l} ({pcm_names})"

		return pcm_names[0] if n_l == 1 else f"No PCM with ID={pcm_id}"

Conversation

marc-hb commented Jun 9, 2023

Uh oh!

marc-hb commented Jun 9, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

marc-hb commented Jun 10, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

RanderWang left a comment

Choose a reason for hiding this comment

Uh oh!

plbossart Jun 12, 2023

Choose a reason for hiding this comment

Uh oh!

marc-hb Jun 12, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ranj063 Jun 15, 2023

Choose a reason for hiding this comment

Uh oh!

ranj063 commented Jun 15, 2023

Uh oh!

marc-hb Jun 16, 2023

Choose a reason for hiding this comment

Uh oh!

marc-hb commented Jul 6, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

marc-hb commented Jul 6, 2023

Uh oh!

RanderWang left a comment

Choose a reason for hiding this comment

Uh oh!

marc-hb commented Jul 6, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment was marked as outdated.

This comment was marked as outdated.

marc-hb commented Jul 6, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

marc-hb commented Jul 7, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

RanderWang left a comment

Choose a reason for hiding this comment

Uh oh!

marc-hb commented Jul 7, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

marc-hb commented Jun 9, 2023 •

edited

Loading

marc-hb commented Jun 10, 2023 •

edited

Loading

marc-hb Jun 12, 2023 •

edited

Loading

marc-hb commented Jul 6, 2023 •

edited

Loading

marc-hb commented Jul 6, 2023 •

edited

Loading

marc-hb commented Jul 6, 2023 •

edited

Loading

marc-hb commented Jul 7, 2023 •

edited

Loading