p, Jeeja <jeeja...@intel.com>;
> alsa-de...@alsa-project.org; intel-gfx@lists.freedesktop.org; # v4 . 7+
> <sta...@vger.kernel.org>
> Subject: Re: [Intel-gfx] [PATCH] ALSA: hda: Use loop counter for
> hdac_wait_for_cmd_dmas() timeout
>
> On Thu, May 04, 2017 at 12:25:
On Thu, May 04, 2017 at 12:25:26PM +0200, Takashi Iwai wrote:
> On Thu, 04 May 2017 12:18:29 +0200,
> Chris Wilson wrote:
> >
> > hdac_wait_for_cmd_dmas() uses a jiffie timeout to ensure that we do not
> > wait forever for stuck hardware. However, it is called from an
> > irq-disabled context
On Thu, 04 May 2017 12:30:32 +0200,
Vinod Koul wrote:
>
> On Thu, May 04, 2017 at 12:25:26PM +0200, Takashi Iwai wrote:
> > On Thu, 04 May 2017 12:18:29 +0200,
> > Chris Wilson wrote:
> > >
> > > hdac_wait_for_cmd_dmas() uses a jiffie timeout to ensure that we do not
> > > wait forever for stuck
On Thu, May 04, 2017 at 12:25:26PM +0200, Takashi Iwai wrote:
> On Thu, 04 May 2017 12:18:29 +0200,
> Chris Wilson wrote:
> >
> > hdac_wait_for_cmd_dmas() uses a jiffie timeout to ensure that we do not
> > wait forever for stuck hardware. However, it is called from an
> > irq-disabled context
On Thu, 04 May 2017 12:18:29 +0200,
Chris Wilson wrote:
>
> hdac_wait_for_cmd_dmas() uses a jiffie timeout to ensure that we do not
> wait forever for stuck hardware. However, it is called from an
> irq-disabled context which prevents jiffie from advancing and so the
> loop doesn't terminate if
hdac_wait_for_cmd_dmas() uses a jiffie timeout to ensure that we do not
wait forever for stuck hardware. However, it is called from an
irq-disabled context which prevents jiffie from advancing and so the
loop doesn't terminate if the hardware fails. This can then cause NMI
watchdog warnings, such