[Sound-open-firmware] Status of Braswell

[Curtis Malainey]

Thanks Pierre,

I am working off the CML branch right now as we are using the same kernel.
I think I am seeing the DMA issues you are referring to.

 CORE  LEVEL      COMP_ID                TIMESTAMP            DELTA
            FILE_NAME    CONTENT
    0      1         WAIT          36873616.250000  36873616.000000
   src/lib/wait.c:118    ewt
    0      1          DMA          36873622.291667         6.041667
src/drivers/dw/dma.c:478    dw_dma_stop() error: dma 1 channel 0
timeout
    0      1          DMA          66669445.312500  29795824.000000
src/drivers/dw/dma.c:309    dw_dma_start() error: dma 1 channel 0 not
ready ena 0x1 status 0x3
    0      1         PIPE          66669456.406250        11.093750
src/audio/pipeline.c:808    pipeline_copy() error: ret = -16,
start->comp.id = 9, dir = 1
    0      1         PIPE 1.10     66669465.833333         9.427083
src/audio/pipeline.c:920    pipeline_xrun_recover()
    0      1         COMP          66669474.322917         8.489583
rc/audio/component.c:197    comp_set_state() error: wrong state = 5,
COMP_TRIGGER_PREPARE
    0      1         PIPE          66669481.614583         7.291667
src/audio/pipeline.c:411    pipeline_prepare() error: ret =
-22,dev->comp.id = 1
    0      1         PIPE 1.10     66669490.052083         8.437500
src/audio/pipeline.c:927    pipeline_xrun_recover() error:
pipeline_prepare() failed, ret = -22
    0      1         PIPE 1.10     66669497.135417         7.083333
src/audio/pipeline.c:998    pipeline_task(): xrun recover failed!
pipeline will be stopped!
    0      1         WAIT          67675418.072917   1005920.937500
   src/lib/wait.c:118    ewt
    0      1          DMA          67675424.166667         6.093750
src/drivers/dw/dma.c:478    dw_dma_stop() error: dma 1 channel 0
timeout
    0      1          DMA          75861656.458333   8186232.500000
src/drivers/dw/dma.c:309    dw_dma_start() error: dma 1 channel 0 not
ready ena 0x1 status 0x3
    0      1         PIPE          75861667.552083        11.093750
src/audio/pipeline.c:808    pipeline_copy() error: ret = -16,
start->comp.id = 9, dir = 1
    0      1         PIPE 1.10     75861676.979167         9.427083
src/audio/pipeline.c:920    pipeline_xrun_recover()
    0      1         COMP          75861685.468750         8.489583
rc/audio/component.c:197    comp_set_state() error: wrong state = 5,
COMP_TRIGGER_PREPARE
    0      1         PIPE          75861692.760417         7.291667
src/audio/pipeline.c:411    pipeline_prepare() error: ret =
-22,dev->comp.id = 1
    0      1         PIPE 1.10     75861701.197917         8.437500
src/audio/pipeline.c:927    pipeline_xrun_recover() error:
pipeline_prepare() failed, ret = -22
    0      1         PIPE 1.10     75861708.229167         7.031250
src/audio/pipeline.c:998    pipeline_task(): xrun recover failed!
pipeline will be stopped!
    0      1         WAIT          76871284.687500   1009576.437500
   src/lib/wait.c:118    ewt
    0      1          DMA          76871290.781250         6.093750
src/drivers/dw/dma.c:478    dw_dma_stop() error: dma 1 channel 0
timeout

I also was unable to load the topology until I removed the media
pipline since the sname on the pipeline was not being loaded and
therefore the pipeline refused to load. We are upreving the driver
here soon and will report back if that fixes anything. Thanks for the
heads up.

Curtis

On Mon, Aug 26, 2019 at 12:02 PM Pierre-Louis Bossart
<pierre-louis.bossart at linux.intel.com> wrote:
>
> Hi Curtis,
>
> > I am just curious what the status of Braswell devices are right now.
> > If the status is unknown that is fine. I may be doing some
> > experimenting with them in the near future. Any known issues or
> > caveats would be appreciated. :)
>
> braswell is essentially cherrytrail which is very nearly identical to
> baytrail except for an oscillator frequency. I tested SOF on Cyan a
> number of times in the past, so all the plumbing should already be there.
> The only caveat is that there is quite a bit of DMA rework at the moment
> so there might be some re-testing required if you take the latest
> development branch.
> Also our module load/reload tests failed, but it's more of a dependency
> with the max98080/TI aux chip that's not well handled than SOF proper.
> -Pierre