At Mon, 19 Jan 2015 10:42:56 +0100, Jiri Slaby wrote:

There is a race between timer and process contexts. Process context does not disable irqs, so when a timer ticks inside process' critical section, the system can deadlock. Fix this by a traditional _irqsave variant of spin_lock.

Signed-off-by: Jiri Slaby jslaby@suse.cz Cc: Jaroslav Kysela perex@perex.cz Cc: Takashi Iwai tiwai@suse.de Cc: alsa-devel@alsa-project.org

---

sound/drivers/dummy.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/sound/drivers/dummy.c b/sound/drivers/dummy.c index 6d6bf4093583..18a7cfc99078 100644 --- a/sound/drivers/dummy.c +++ b/sound/drivers/dummy.c @@ -271,10 +271,13 @@ static void dummy_systimer_update(struct dummy_systimer_pcm *dpcm) static int dummy_systimer_start(struct snd_pcm_substream *substream) { struct dummy_systimer_pcm *dpcm = substream->runtime->private_data;

• spin_lock(&dpcm->lock);

• unsigned long flags;
• spin_lock_irqsave(&dpcm->lock, flags); dpcm->base_time = jiffies; dummy_systimer_rearm(dpcm);

• spin_unlock(&dpcm->lock);

• spin_unlock_irqrestore(&dpcm->lock, flags);

This looks correct, but...

return 0; }

@@ -322,12 +325,14 @@ static snd_pcm_uframes_t dummy_systimer_pointer(struct snd_pcm_substream *substream) { struct dummy_systimer_pcm *dpcm = substream->runtime->private_data;

• unsigned long flags; snd_pcm_uframes_t pos;

• spin_lock(&dpcm->lock);

• spin_lock_irqsave(&dpcm->lock, flags); dummy_systimer_update(dpcm); pos = dpcm->frac_pos / HZ;

• spin_unlock(&dpcm->lock);

• spin_unlock_irqrestore(&dpcm->lock, flags);

This chunk is superfluous. The pointer callback is guaranteed to be called in the irq-disabled context.

thanks,

Takashi