ALSA: echoaudio: Fix memory leak in snd_echo_resume()
diff mbox series

Message ID 20200810061500.9329-1-dinghao.liu@zju.edu.cn
State New
Headers show
Series
  • ALSA: echoaudio: Fix memory leak in snd_echo_resume()
Related show

Commit Message

Dinghao Liu Aug. 10, 2020, 6:14 a.m. UTC
When restore_dsp_rettings() fails, chip should be freed
just like when init_hw() and request_irq() fails.

Fixes: 47b5d028fdce8 ("ALSA: Echoaudio - Add suspend support #2")
Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn>
---
 sound/pci/echoaudio/echoaudio.c | 1 +
 1 file changed, 1 insertion(+)

Comments

Takashi Iwai Aug. 10, 2020, 6:54 a.m. UTC | #1
On Mon, 10 Aug 2020 08:14:58 +0200,
Dinghao Liu wrote:
> 
> When restore_dsp_rettings() fails, chip should be freed
> just like when init_hw() and request_irq() fails.
> 
> Fixes: 47b5d028fdce8 ("ALSA: Echoaudio - Add suspend support #2")
> Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn>

Actually it's rather wrong to call snd_echo_free() in the resume error
path.  This may lead to an Oops at the next time the system goes to
resume.  That said, we should drop snd_echo_free() calls in other
error paths in snd_echo_resume() instead.


thanks,

Takashi


> ---
>  sound/pci/echoaudio/echoaudio.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/sound/pci/echoaudio/echoaudio.c b/sound/pci/echoaudio/echoaudio.c
> index 6aeb99aa2414..2aa183fe5dc1 100644
> --- a/sound/pci/echoaudio/echoaudio.c
> +++ b/sound/pci/echoaudio/echoaudio.c
> @@ -2226,6 +2226,7 @@ static int snd_echo_resume(struct device *dev)
>  	chip->pipe_alloc_mask = pipe_alloc_mask;
>  	if (err < 0) {
>  		kfree(commpage_bak);
> +		snd_echo_free(chip);
>  		return err;
>  	}
>  
> -- 
> 2.17.1
>
Dinghao Liu Aug. 13, 2020, 6:44 a.m. UTC | #2
> On Mon, 10 Aug 2020 08:14:58 +0200,
> Dinghao Liu wrote:
> > 
> > When restore_dsp_rettings() fails, chip should be freed
> > just like when init_hw() and request_irq() fails.
> > 
> > Fixes: 47b5d028fdce8 ("ALSA: Echoaudio - Add suspend support #2")
> > Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn>
> 
> Actually it's rather wrong to call snd_echo_free() in the resume error
> path.  This may lead to an Oops at the next time the system goes to
> resume.  That said, we should drop snd_echo_free() calls in other
> error paths in snd_echo_resume() instead.
> 

Thank you for your advice! I will send a new patch to fix this soon.

Regards,
Dinghao

Patch
diff mbox series

diff --git a/sound/pci/echoaudio/echoaudio.c b/sound/pci/echoaudio/echoaudio.c
index 6aeb99aa2414..2aa183fe5dc1 100644
--- a/sound/pci/echoaudio/echoaudio.c
+++ b/sound/pci/echoaudio/echoaudio.c
@@ -2226,6 +2226,7 @@  static int snd_echo_resume(struct device *dev)
 	chip->pipe_alloc_mask = pipe_alloc_mask;
 	if (err < 0) {
 		kfree(commpage_bak);
+		snd_echo_free(chip);
 		return err;
 	}