LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
* [PATCH 2/2] ASoC: wm8996: ensure lower bounds of 1 for timeout
@ 2015-03-08 10:02 Nicholas Mc Guire
  2015-03-09 14:27 ` Charles Keepax
  0 siblings, 1 reply; 2+ messages in thread
From: Nicholas Mc Guire @ 2015-03-08 10:02 UTC (permalink / raw)
  To: Liam Girdwood
  Cc: Mark Brown, Jaroslav Kysela, Takashi Iwai, Lars-Peter Clausen,
	Charles Keepax, Tomi Valkeinen, Xiubo Li, Sachin Kamat, Axel Lin,
	abdoulaye berthe, patches, alsa-devel, linux-kernel,
	Nicholas Mc Guire

wait_for_completion_timeout can be called with timeout == 0 due to
msecs_to_jiffies(2) == 1 for HZ < 1000 and usecs_to_jiffies(300) == 1
for all reasonable values of HZ, thus the following timeout /= 2; sets
timeout to 0. This patch simply adds a lower-bounds of 1.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
---

line numbers from wm9666.c linux-next -next-20150306

2101      /* Wait for the FLL to lock, using the interrupt if possible */
2102      if (Fref > 1000000)
2103              timeout = usecs_to_jiffies(300);
2104      else
2105              timeout = msecs_to_jiffies(2);
2106 
2107      /* Allow substantially longer if we've actually got the IRQ, poll
2108       * at a slightly higher rate if we don't.
2109       */
2110      if (i2c->irq)
2111              timeout *= 10;
2112      else
2113              timeout /= 2;
2114
2115      ret = wait_for_completion_timeout(&wm8996->fll_lock,
2116                                        timeout);

The call-path here would then be: 
 wait_for_completion_timeout(...,timeout)
  -> wait_for_common action==schedule_timeout
   -> schedule_timeout(timeout)
      ...
      expire = timeout + jiffies;

so expire == jiffies here and the timer would expire immediately
for systems with HZ<1000 due to msecs_to_jiffies(2) == 1 and the
following timeout /= 2 -> timeout == 0, which is probably a bit 
more that a "slightly higher rate".

This was only compile tested for exynos_defconfig + CONFIG_COMPILE_TEST=y,
SND_SOC_ALL_CODECS=m (implies CONFIG_SND_SOC_WM8996=m)

Patch is against 4.0-rc2 linux-next (localversion-next is -next-20150306)

 sound/soc/codecs/wm8996.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/sound/soc/codecs/wm8996.c b/sound/soc/codecs/wm8996.c
index 24d9705..5ded102 100644
--- a/sound/soc/codecs/wm8996.c
+++ b/sound/soc/codecs/wm8996.c
@@ -2110,7 +2110,8 @@ static int wm8996_set_fll(struct snd_soc_codec *codec, int fll_id, int source,
 	if (i2c->irq)
 		timeout *= 10;
 	else
-		timeout /= 2;
+		/* ensure timeout of atleast 1 jiffies */
+		timeout = timeout/2 ? : 1;
 
 	for (retry = 0; retry < 10; retry++) {
 		time_left = wait_for_completion_timeout(&wm8996->fll_lock,
-- 
1.7.10.4


^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [PATCH 2/2] ASoC: wm8996: ensure lower bounds of 1 for timeout
  2015-03-08 10:02 [PATCH 2/2] ASoC: wm8996: ensure lower bounds of 1 for timeout Nicholas Mc Guire
@ 2015-03-09 14:27 ` Charles Keepax
  0 siblings, 0 replies; 2+ messages in thread
From: Charles Keepax @ 2015-03-09 14:27 UTC (permalink / raw)
  To: Nicholas Mc Guire
  Cc: Liam Girdwood, Mark Brown, patches, alsa-devel, linux-kernel

On Sun, Mar 08, 2015 at 06:02:38AM -0400, Nicholas Mc Guire wrote:
> wait_for_completion_timeout can be called with timeout == 0 due to
> msecs_to_jiffies(2) == 1 for HZ < 1000 and usecs_to_jiffies(300) == 1
> for all reasonable values of HZ, thus the following timeout /= 2; sets
> timeout to 0. This patch simply adds a lower-bounds of 1.
> 
> Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
> ---

Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>

Thanks,
Charles

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2015-03-09 14:27 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-03-08 10:02 [PATCH 2/2] ASoC: wm8996: ensure lower bounds of 1 for timeout Nicholas Mc Guire
2015-03-09 14:27 ` Charles Keepax

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).