Stop timer in the 'hw_free' callback instead of the 'close' callback
since we want the timer to be stopped before the DMA buffer is released.
Otherwise, it could trigger a kernel panic in some circumstances, for
instance when the DMA buffer is already released but the timer callback
is still running.

Signed-off-by: Ivan Orlov <ivan.orlov0322@gmail.com>
---
V1 -> V2:
- Remove useless NULLing of v_iter->substream. It will be released in
'close' callback.

 sound/drivers/pcmtest.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/sound/drivers/pcmtest.c b/sound/drivers/pcmtest.c
index b59b78a09224..8a78fa56106f 100644
--- a/sound/drivers/pcmtest.c
+++ b/sound/drivers/pcmtest.c
@@ -396,8 +396,6 @@ static int snd_pcmtst_pcm_close(struct snd_pcm_substream *substream)
 {
 	struct pcmtst_buf_iter *v_iter = substream->runtime->private_data;
 
-	timer_shutdown_sync(&v_iter->timer_instance);
-	v_iter->substream = NULL;
 	playback_capture_test = !v_iter->is_buf_corrupted;
 	kfree(v_iter);
 	return 0;
@@ -499,6 +497,10 @@ static int snd_pcmtst_pcm_hw_params(struct snd_pcm_substream *substream,
 
 static int snd_pcmtst_pcm_hw_free(struct snd_pcm_substream *substream)
 {
+	struct pcmtst_buf_iter *v_iter = substream->runtime->private_data;
+
+	timer_shutdown_sync(&v_iter->timer_instance);
+
 	return 0;
 }
 
-- 
2.34.1

On Wed, 06 Dec 2023 16:10:46 +0100,
Ivan Orlov wrote:
> 
> Stop timer in the 'hw_free' callback instead of the 'close' callback
> since we want the timer to be stopped before the DMA buffer is released.
> Otherwise, it could trigger a kernel panic in some circumstances, for
> instance when the DMA buffer is already released but the timer callback
> is still running.

You can't call timer_shutdown_sync() at hw_free.  The PCM stream is
still there and you can re-setup via hw_params without closing.  But,
after timer_shutdown_sync(), the timer instance must not be used any
longer.

A more proper way would be to call timer_delete() (no sync) at trigger
STOP, then call timer_delete_sync() at sync_stop op in addition.  This
assures the immediate stop and the sync before changing to another PCM
state.

timer_shutdown_sync() can be kept at the close op to be sure.


thanks,

Takashi

> 
> Signed-off-by: Ivan Orlov <ivan.orlov0322@gmail.com>
> ---
> V1 -> V2:
> - Remove useless NULLing of v_iter->substream. It will be released in
> 'close' callback.
> 
>  sound/drivers/pcmtest.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/sound/drivers/pcmtest.c b/sound/drivers/pcmtest.c
> index b59b78a09224..8a78fa56106f 100644
> --- a/sound/drivers/pcmtest.c
> +++ b/sound/drivers/pcmtest.c
> @@ -396,8 +396,6 @@ static int snd_pcmtst_pcm_close(struct snd_pcm_substream *substream)
>  {
>  	struct pcmtst_buf_iter *v_iter = substream->runtime->private_data;
>  
> -	timer_shutdown_sync(&v_iter->timer_instance);
> -	v_iter->substream = NULL;
>  	playback_capture_test = !v_iter->is_buf_corrupted;
>  	kfree(v_iter);
>  	return 0;
> @@ -499,6 +497,10 @@ static int snd_pcmtst_pcm_hw_params(struct snd_pcm_substream *substream,
>  
>  static int snd_pcmtst_pcm_hw_free(struct snd_pcm_substream *substream)
>  {
> +	struct pcmtst_buf_iter *v_iter = substream->runtime->private_data;
> +
> +	timer_shutdown_sync(&v_iter->timer_instance);
> +
>  	return 0;
>  }
>  
> -- 
> 2.34.1
>

On 12/6/23 16:59, Takashi Iwai wrote:
> On Wed, 06 Dec 2023 16:10:46 +0100,
> Ivan Orlov wrote:
>>
>> Stop timer in the 'hw_free' callback instead of the 'close' callback
>> since we want the timer to be stopped before the DMA buffer is released.
>> Otherwise, it could trigger a kernel panic in some circumstances, for
>> instance when the DMA buffer is already released but the timer callback
>> is still running.
> 
> You can't call timer_shutdown_sync() at hw_free.  The PCM stream is
> still there and you can re-setup via hw_params without closing.  But,
> after timer_shutdown_sync(), the timer instance must not be used any
> longer.
> 
> A more proper way would be to call timer_delete() (no sync) at trigger
> STOP, then call timer_delete_sync() at sync_stop op in addition.  This
> assures the immediate stop and the sync before changing to another PCM
> state.
> 
> timer_shutdown_sync() can be kept at the close op to be sure.
> 
> 
> thanks,
> 
> Takashi
> 

Hi Takashi,

Thanks a lot for the review, sounds reasonable, I'll make suggested 
changes and send the version 3 :)

--
Kind regards,
Ivan Orlov