remove SRC for Baytrail and CherryTrail #3080

plbossart · 2020-06-22T16:41:42Z

remove SRC since it's just too big for the internal memories.
We keep the mixer w/ a regular PCM and a 'deep buffer' one (limited to 10ms periods due to memory sizes).

Tested on CHT w/ rt5640 and max98090 (Cyan Chromebook)

plbossart · 2020-06-22T16:50:36Z

Tentatively Fixes #3063

singalsu

But please fix the missing PCM. I don' think we have other way to test mixer component.

tools/topology/sof-byt-codec.m4

marc-hb · 2020-06-23T15:55:20Z

hence RFC status.

Someone showed me it's now possible to switch any PR back to "Draft" status. Github drafts:

cannot be merged
appear in a different color
can be filtered out in queries with draft:false or -is:draft (forgot which one sorry)

I've been using github drafts since the feature was created and there's no drawback whatsoever. Let's use drafts.

Also:

drafts do not spam CODEOWNERS automatically (you can still add anyone manually)

... so I always start every PR as a draft in case CI tells me I missed something, this saves a lot of email.

https://github.blog/2019-02-14-introducing-draft-pull-requests/
https://github.blog/changelog/2020-04-08-convert-pull-request-to-draft/

Off-topic sorry.

Add separate pipeline starting with a mixer (similar to Baytrail switch matrix and SKL closed-source firmware) Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

Add a new pipeline definition to avoid mucking with either pipe-low-latency or pipe-media. The code is 90% similar to pipeline-volume-playback but includes a scheduling component. Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

remove the low-latency pipeline and pcm-media. Instead, we use 2 two pipe-host-volume-playback with different periods connected to the same mixer-dai. Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

lgirdwood

This all looks good, but it looks like CI is attempting a hw params that are no longer supported ?

lgirdwood · 2020-06-29T19:56:09Z

This all looks good, but it looks like CI is attempting a hw params that are no longer supported ?

The same params are actually working on boards with a codec, but I dont see any immediate difference between the topologies.

plbossart · 2020-06-29T20:58:21Z

This all looks good, but it looks like CI is attempting a hw params that are no longer supported ?

The same params are actually working on boards with a codec, but I dont see any immediate difference between the topologies.

kernel: [  177.961849]  PCM: ASoC: no BE found for SSP2.OUT
kernel: [  177.962840]  PCM: ASoC: no backend DAIs enabled for PCM

Likely a mismatch between topology and machine driver. Will test, maybe I went too fast on the nocodec support

Additional work needed to make the BE name and index configuration, as well as LBM quirk. We should be using a single topology here. Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

Same parameters, so let's use the same template Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

No one uses this topology so let's remove it. Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

Only the SSP settings need to be different. Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

We really don't care about SRC on those platforms and the memory footprint is too large anyways. Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

plbossart · 2020-06-29T22:33:45Z

This all looks good, but it looks like CI is attempting a hw params that are no longer supported ?

The same params are actually working on boards with a codec, but I dont see any immediate difference between the topologies.
kernel: [  177.961849]  PCM: ASoC: no BE found for SSP2.OUT
kernel: [  177.962840]  PCM: ASoC: no backend DAIs enabled for PCM
Likely a mismatch between topology and machine driver. Will test, maybe I went too fast on the nocodec support

What I thought, stupid edit mistake

 DAI_ADD(sof/pipe-mixer-dai-playback.m4,
-       1, SSP, SSP_NUM, SSP2-Codec,
+       1, SSP, SSP_NUM, NoCodec-2,

Fixed now, let's see how CI likes it.

aiChaoSONG · 2020-06-30T02:44:48Z

SOFCI TEST

xiulipan · 2020-06-30T03:19:02Z

Now the .TEXT size is 0x12ea7 on CHT, 0x12d27 on BYT

https://sof-ci.01.org/sofpr/PR3080/build6448/build/cht_xcc.txt

writing module 0 sof-cht

	Totals	Start		End		Size
	TEXT	0xff2c0000	0xff2d2ea7	0x12ea7
	DATA	0xff300010	0xff328000	0x27ff0
	BSS	0xff3023f0	0xff306258	0x3e68

https://sof-ci.01.org/sofpr/PR3080/build6448/build/byt_xcc.txt

writing module 0 sof-byt

	Totals	Start		End		Size
	TEXT	0xff2c0000	0xff2d2d27	0x12d27
	DATA	0xff300010	0xff328000	0x27ff0
	BSS	0xff302310	0xff306178	0x3e68

aiChaoSONG

Verified by CI

lgirdwood · 2020-06-30T10:04:24Z

Jenkins BAT fails sound clean, could be startup DUT or Test equipment clock delta appearing as a higher frequency that settles quickly.
@zrombel any idea why CI would fail for ICL when this is unrelated.

aiChaoSONG · 2020-07-07T06:42:49Z

@lgirdwood Can we get this merged? The ICL error obviously has nothing to do with this PR.

lgirdwood · 2020-07-07T08:23:06Z

@aiChaoSONG it was a CI server issue, it's been rerun and is now good..

marc-hb · 2020-07-08T13:31:41Z

Note there's a new, unrelated regression on these same platforms now: #3146

marc-hb · 2020-07-11T00:09:32Z

I think the wrong _defconfig files have been used for the last 3- months: #2711 (comment)

marc-hb · 2020-07-14T22:53:00Z

PR #3167 just got merged and it turns off CONFIG_OPTIMIZE_FOR_SIZE again for baytrail (and maybe for others)

plbossart · 2020-07-14T22:56:59Z

PR #3167 just got merged and it turns off CONFIG_OPTIMIZE_FOR_SIZE again for baytrail (and maybe for others)

Are you saying PR #3167 is good or bad? Can't follow your point

marc-hb · 2020-07-14T23:14:43Z

Are you saying PR #3167 is good or bad?

I'm saying neither; I'm merely trying to mitigate any future .TEXT size confusion.

plbossart requested review from ranj063 and singalsu June 22, 2020 16:41

plbossart requested review from dbaluta, jajanusz, lbetlej, lgirdwood, mmaka1, tlauda and xiulipan as code owners June 22, 2020 16:41

singalsu approved these changes Jun 22, 2020

View reviewed changes

lgirdwood requested changes Jun 23, 2020

View reviewed changes

tools/topology/sof-byt-codec.m4 Outdated Show resolved Hide resolved

plbossart added 3 commits June 26, 2020 18:14

topology: add pipe-mixer-dai-playback

5a5aa06

Add separate pipeline starting with a mixer (similar to Baytrail switch matrix and SKL closed-source firmware) Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

plbossart force-pushed the fix/byt-cht-remove-SRC branch from 3d523a3 to 96b0629 Compare June 27, 2020 19:53

plbossart changed the title ~~[RFC] remove SRC for Baytrail and CherryTrail~~ remove SRC for Baytrail and CherryTrail Jun 27, 2020

plbossart requested review from lgirdwood and singalsu June 27, 2020 20:49

lgirdwood reviewed Jun 29, 2020

View reviewed changes

plbossart added 5 commits June 29, 2020 17:30

topology: align cht-nocodec and byt-codec topologies

545da4c

Additional work needed to make the BE name and index configuration, as well as LBM quirk. We should be using a single topology here. Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

topology: cht-rt5682: align with byt-codec topology

75038eb

Same parameters, so let's use the same template Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

topology: remove cht-src-50kHz-pcm512x

18d437c

No one uses this topology so let's remove it. Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

topology: cht-max98090: align with byt-nocodec

04c1973

Only the SSP settings need to be different. Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

SRC: remove from Baytrail/Cherrytrail configurations

f795271

We really don't care about SRC on those platforms and the memory footprint is too large anyways. Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

plbossart force-pushed the fix/byt-cht-remove-SRC branch from 96b0629 to f795271 Compare June 29, 2020 22:35

plbossart requested a review from lgirdwood June 29, 2020 22:35

xiulipan approved these changes Jun 30, 2020

View reviewed changes

lgirdwood approved these changes Jun 30, 2020

View reviewed changes

aiChaoSONG approved these changes Jun 30, 2020

View reviewed changes

lgirdwood merged commit 7ef7313 into thesofproject:master Jul 7, 2020

This was referenced Jul 14, 2020

[BUG][BYT-NOCODEC]Input/output error when aplay via (0,1) using format s24_le/s32_le #3171

Closed

[BUG][BYT-NOCODEC][BDW_WSB_RT286]Input/output error when simultaneous-playback-capture / multiple-pipeline-playback #3170

Closed

marc-hb mentioned this pull request Jul 15, 2020

[BUG][CI][CHT][BYT] Build failure on CHT: .text will not fit in region sof_text_start #3063

Closed

remove SRC for Baytrail and CherryTrail #3080

remove SRC for Baytrail and CherryTrail #3080

Uh oh!

Conversation

plbossart commented Jun 22, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

plbossart commented Jun 22, 2020

Uh oh!

singalsu left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

marc-hb commented Jun 23, 2020

Uh oh!

lgirdwood left a comment

Choose a reason for hiding this comment

Uh oh!

lgirdwood commented Jun 29, 2020

Uh oh!

plbossart commented Jun 29, 2020

Uh oh!

plbossart commented Jun 29, 2020

Uh oh!

aiChaoSONG commented Jun 30, 2020

Uh oh!

xiulipan commented Jun 30, 2020

Uh oh!

aiChaoSONG left a comment • 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!

lgirdwood commented Jun 30, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

aiChaoSONG commented Jul 7, 2020

Uh oh!

lgirdwood commented Jul 7, 2020

Uh oh!

marc-hb commented Jul 8, 2020

Uh oh!

marc-hb commented Jul 11, 2020

Uh oh!

marc-hb commented Jul 14, 2020

Uh oh!

plbossart commented Jul 14, 2020

Uh oh!

marc-hb commented Jul 14, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

plbossart commented Jun 22, 2020 •

edited

Loading

aiChaoSONG left a comment •

edited

Loading

lgirdwood commented Jun 30, 2020 •

edited

Loading