kernel oops during sof_bootloop.sh test #1466
Comments
|
This seems to happen during a pm_runtime resume operation |
|
same problem on HDAudio+DMIC machine... Gah. |
|
@plbossart a long shot but can you please try if this helps? diff --git a/sound/soc/soc-core.c b/sound/soc/soc-core.c
index 55014e7ae0d8..317fc0e0d54c 100644
--- a/sound/soc/soc-core.c
+++ b/sound/soc/soc-core.c
@@ -311,7 +311,6 @@ static int snd_soc_rtdcom_add(struct snd_soc_pcm_runtime *rtd,
return -ENOMEM;
rtdcom->component = component;
- INIT_LIST_HEAD(&rtdcom->list);
/*
* When rtd was freed, created rtdcom here will be
@@ -430,6 +429,8 @@ static void soc_free_pcm_runtime(struct snd_soc_pcm_runtime *rtd)
* soc_new_pcm_runtime()
*/
device_unregister(rtd->dev);
+
+ INIT_LIST_HEAD(&rtd->component_list);
}
static struct snd_soc_pcm_runtime *soc_new_pcm_runtime( |
|
The error is "root-caused' to "ASoC: soc-core: add soc_unbind_dai_link()" the code moved the pcm runtime cleanup from soc_cleanup_card_resources() to snd_soc_remove_dai_link(), but the latter is called both from soc_cleanup_card_resources() and topology. I have no idea why things are done in this way... |
@plbossart Im guessing my change didnt help then. Anyway, my theory is that the problem is caused because snd_soc_remove_dai_link() called when the topology is removed unregisters the rtd->dev and later on when the pcm->private_free() gets called, it runs into errors because both the rtd device has been already unregistered and the component removed too. |
|
Error not seen on Environment |
|
closed |
Just saw this on SoundWire device, this may be related to recent ALSA/soc-core changes?
The text was updated successfully, but these errors were encountered: