SoundWire: rt711-sdca: transfers timed out on component remove/probe #3650
Comments
|
@shumingfan @bardliao I am not sure what happens here first, I don't understand why the component remove plays with the regcache. static void rt711_sdca_remove(struct snd_soc_component *component)
{
struct rt711_sdca_priv *rt711 = snd_soc_component_get_drvdata(component);
regcache_cache_only(rt711->regmap, true);
regcache_cache_only(rt711->mbq_regmap, true);
}Why is this necessary? unfortunately, removing this doesn't seem to solve the problem. I am starting to wonder if there are device timing restrictions here, or something that needs to be turned on before accessing the vendor-specific registers? |
|
8b527e9 suggests removing those regcache changes, but that doesn't fix the issue. |
|
Also need thesofproject/sof#5813 with the HDMI commented out @bardliao |
|
@plbossart I tested it on SKU 0A5E and no issue found. BTW, thesofproject/sof#5813 seems not work for me. I modified it as below. --- a/tools/topology/topology1/sof-icl-rt711-rt1308-rt715-hdmi.m4
+++ b/tools/topology/topology1/sof-icl-rt711-rt1308-rt715-hdmi.m4
@@ -32,7 +32,7 @@ ifdef(`MIC_LINK',`',
`define(MIC_LINK, `3')')
# uncomment to remove HDMI support
-#define(NOHDMI, `1')
+define(NOHDMI, `1')
# UAJ ID: 0, 1
# AMP ID: 2, 3 (if EXT_AMP_REF defined)
I have to remove the HDMI part manually. I also tried with SKU 0A32, but somehow snd_soc_sof_sdw is already in used after bootup. So I can't run the test on my SKU 0A32 device. |
|
sorry @bardliao, yes the HDMI part has to be removed manually in the file as you did. This is just for tests, I don't want to have a new topology maintained here, just a quick hook to remove HDMI in one shot. For the tests, I removed pulseaudio, otherwise the card is marked as in use. |
No, sorry, @plbossart I didn't make it clear. I meant the change didn't work for me. I have to remove the HDMI related code from sof-icl-rt711-rt1308-rt715-hdmi.m4 manually. Like and so on.
I can run the test and found no issue on SKU 0A32, too. |
|
@bardliao I forgot to push a local change, now fixed with thesofproject/sof@ef9883d |
It works now :) |
|
@bardliao can you reproduce the issue with the timeout? |
|
@plbossart If I apply 861d91f and d2e9a6d only, I will see below error. |
|
@bardliao the error above happens when you remove the rt1316 driver, right? That's what is reported already in #3531 and is fixed by the commit e850e40 ('soundwire: revisit driver bind/unbind and callbacks'). However I have not seen the error what sequence did you use to see it? EDIT: sorry, this was just a dev_dbg message |
|
@bardliao @shumingfan I can reproduce the error, even with #3657 applied on top. The test has to be run in a loop though, it doesn't happen the first time. while true; do bash ./test.sh; done |
|
@plbossart I run the test.sh in loop with #3642. I got no error in more than one hour. |
|
@plbossart I also run the test.sh on Dell SKU 0A5D (non-SDCA) with #3642. The 'timed out' messages doesn't happen. |
|
The test fails only on the SDCA skew @bardliao @shumingfan, it's related to the rt711-sdca codec driver. |
|
Cannot reproduce this with the changes in #3642 |
Make sure that the bus and codecs are pm_runtime active when the card is registered/created. This avoid timeouts when accessing registers. BugLink: thesofproject#3651 BugLink: thesofproject#3650 Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Make sure that the bus and codecs are pm_runtime active when the card is registered/created. This avoid timeouts when accessing registers. BugLink: thesofproject#3651 BugLink: thesofproject#3650 Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Make sure that the bus and codecs are pm_runtime active when the card is registered/created. This avoid timeouts when accessing registers. BugLink: thesofproject#3651 BugLink: thesofproject#3650 Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Make sure that the bus and codecs are pm_runtime active when the card is registered/created. This avoid timeouts when accessing registers. BugLink: #3651 BugLink: #3650 Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Rander Wang <rander.wang@intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com> Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Make sure that the bus and codecs are pm_runtime active when the card is registered/created. This avoid timeouts when accessing registers. BugLink: #3651 BugLink: #3650 Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Rander Wang <rander.wang@intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com> Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
In bind/unbind driver tests, the transfers can fail with a timeout during .set_jack_detect
The text was updated successfully, but these errors were encountered: