On 10/19/22 08:56, Maxime Ripard wrote: > Hi, > > On Thu, Sep 22, 2022 at 10:58:46PM +0800, Zhang Qilong wrote: >> Using the newest pm_runtime_resume_and_get is more appropriate >> for simplifing code here. >> >> Signed-off-by: Zhang Qilong <zhangqilong3@xxxxxxxxxx> >> --- >> sound/soc/soc-component.c | 6 ++---- >> 1 file changed, 2 insertions(+), 4 deletions(-) >> >> diff --git a/sound/soc/soc-component.c b/sound/soc/soc-component.c >> index e12f8244242b..659b9ade4158 100644 >> --- a/sound/soc/soc-component.c >> +++ b/sound/soc/soc-component.c >> @@ -1213,11 +1213,9 @@ int snd_soc_pcm_component_pm_runtime_get(struct snd_soc_pcm_runtime *rtd, >> int i; >> >> for_each_rtd_components(rtd, i, component) { >> - int ret = pm_runtime_get_sync(component->dev); >> - if (ret < 0 && ret != -EACCES) { >> - pm_runtime_put_noidle(component->dev); >> + int ret = pm_runtime_resume_and_get(component->dev); >> + if (ret < 0 && ret != -EACCES) >> return soc_component_ret(component, ret); >> - } >> /* mark stream if succeeded */ >> soc_component_mark_push(component, stream, pm); > > This creates an issue on the RaspberryPi4 on 6.1-rc1. > > At boot time, we now have a bunch of: > [ 35.536496] hdmi-audio-codec hdmi-audio-codec.1.auto: Runtime PM usage count underflow! > > They were bisected back to that commit, and reverting it makes it go > away. > > I think this is due to the fact that the function here used to call > pm_runtime_put_noidle() only if there was an error, and that error > wasn't EACCES. However, pm_runtime_resume_and_get() will call > pm_runtime_put_noidle() for any error. > > Thus, if our __pm_runtime_resume() call return EACCES, we used to keep > the our reference but we will now drop it. So I guess we should just > revert it, possibly with a comment? This is already reverted, see patch from Peter: [PATCH] Revert "ASoC: soc-component: using pm_runtime_resume_and_get instead of pm_runtime_get_sync"