[sound] mpd loops on current -next

[sound] mpd loops on current -next

[Jiri Slaby]

Hi,

I found a regression in current -next caused by
c44941e05900fe85f6091456f9e313b446a97ae9
(PCM midlevel: Do not update hw_ptr_jiffies when hw_ptr is not changed)

When mpd is playing, it loops with a pattern such as:
16:05:02.762544 ioctl(14, 0x4122, 0x3)  = 0
16:05:02.762565 ioctl(14, 0x4122, 0)    = 0
16:05:02.762587 ioctl(14, 0x4122, 0x1)  = 0
16:05:02.762609 ioctl(14, 0x4122, 0x3)  = 0
16:05:02.762630 ioctl(14, 0x4122, 0)    = 0
16:05:02.762652 ioctl(14, 0x4122, 0)    = 0
16:05:02.762674 poll([{fd=13, events=POLLIN|POLLERR|POLLNVAL}], 1, -1) =
1 ([{fd=13, revents=POLLIN}])
16:05:02.762698 ioctl(14, 0x4122, 0x3)  = 0
16:05:02.762720 ioctl(14, 0x4122, 0)    = 0
16:05:02.762742 ioctl(14, 0x4122, 0x1)  = 0
16:05:02.762765 ioctl(14, 0x4122, 0x3)  = 0
16:05:02.762787 ioctl(14, 0x4122, 0)    = 0
16:05:02.762809 ioctl(14, 0x4122, 0)    = 0

Reverting the patch makes the problem disappear.

Re: [sound] mpd loops on current -next

[Jaroslav Kysela]

On Sat, 6 Jun 2009, Jiri Slaby wrote:

> Hi,
>
> I found a regression in current -next caused by
> c44941e05900fe85f6091456f9e313b446a97ae9
> (PCM midlevel: Do not update hw_ptr_jiffies when hw_ptr is not changed)

Oops. It seems like typo. Could you try this patch, please?

diff --git a/sound/core/pcm_lib.c b/sound/core/pcm_lib.c
index dd9126b..bf34603 100644
--- a/sound/core/pcm_lib.c
+++ b/sound/core/pcm_lib.c
@@ -371,7 +371,7 @@ int snd_pcm_update_hw_ptr(struct snd_pcm_substream *substream)
  	    runtime->silence_size > 0)
  		snd_pcm_playback_silence(substream, new_hw_ptr);

-	if (runtime->status->hw_ptr != new_hw_ptr)
+	if (runtime->status->hw_ptr == new_hw_ptr)
  		return 0;

  	runtime->hw_ptr_base = hw_base;

 					Thanks,
 						Jaroslav

-----
Jaroslav Kysela <perex@perex.cz>
Linux Kernel Sound Maintainer
ALSA Project, Red Hat, Inc.

Re: [sound] mpd loops on current -next

[Jiri Slaby]

On 06/06/2009 05:52 PM, Jaroslav Kysela wrote:
> On Sat, 6 Jun 2009, Jiri Slaby wrote:
>> I found a regression in current -next caused by
>> c44941e05900fe85f6091456f9e313b446a97ae9
>> (PCM midlevel: Do not update hw_ptr_jiffies when hw_ptr is not changed)
> 
> Oops. It seems like typo. Could you try this patch, please?

Yup, looks like that did the job.

Thanks.

Re: [sound] mpd loops on current -next

[Takashi Iwai]

At Sat, 6 Jun 2009 17:52:44 +0200 (CEST),
Jaroslav Kysela wrote:
> 
> On Sat, 6 Jun 2009, Jiri Slaby wrote:
> 
> > Hi,
> >
> > I found a regression in current -next caused by
> > c44941e05900fe85f6091456f9e313b446a97ae9
> > (PCM midlevel: Do not update hw_ptr_jiffies when hw_ptr is not changed)
> 
> Oops. It seems like typo. Could you try this patch, please?

Ah, that's the cause of the bug with PA on the recent driver.
I didn't experience the problem because I myself don't use PA :)

Now merged to sound git tree.


thanks,

Takashi

> 
> diff --git a/sound/core/pcm_lib.c b/sound/core/pcm_lib.c
> index dd9126b..bf34603 100644
> --- a/sound/core/pcm_lib.c
> +++ b/sound/core/pcm_lib.c
> @@ -371,7 +371,7 @@ int snd_pcm_update_hw_ptr(struct snd_pcm_substream *substream)
>   	    runtime->silence_size > 0)
>   		snd_pcm_playback_silence(substream, new_hw_ptr);
> 
> -	if (runtime->status->hw_ptr != new_hw_ptr)
> +	if (runtime->status->hw_ptr == new_hw_ptr)
>   		return 0;
> 
>   	runtime->hw_ptr_base = hw_base;
> 
>  					Thanks,
>  						Jaroslav
> 
> -----
> Jaroslav Kysela <perex@perex.cz>
> Linux Kernel Sound Maintainer
> ALSA Project, Red Hat, Inc.
> 

Re: [sound] mpd loops on current -next

[Jiri Slaby]

On 06/06/2009 06:37 PM, Takashi Iwai wrote:
>>> I found a regression in current -next caused by
>>> c44941e05900fe85f6091456f9e313b446a97ae9
>>> (PCM midlevel: Do not update hw_ptr_jiffies when hw_ptr is not changed)
>>
>> Oops. It seems like typo. Could you try this patch, please?
> 
> Ah, that's the cause of the bug with PA on the recent driver.
> I didn't experience the problem because I myself don't use PA :)

FWIW me neither...

Re: [sound] mpd loops on current -next

[Takashi Iwai]

At Sat, 06 Jun 2009 18:37:11 +0200,
I wrote:
> 
> At Sat, 6 Jun 2009 17:52:44 +0200 (CEST),
> Jaroslav Kysela wrote:
> > 
> > On Sat, 6 Jun 2009, Jiri Slaby wrote:
> > 
> > > Hi,
> > >
> > > I found a regression in current -next caused by
> > > c44941e05900fe85f6091456f9e313b446a97ae9
> > > (PCM midlevel: Do not update hw_ptr_jiffies when hw_ptr is not changed)
> > 
> > Oops. It seems like typo. Could you try this patch, please?
> 
> Ah, that's the cause of the bug with PA on the recent driver.
> I didn't experience the problem because I myself don't use PA :)

I found another bug with that change.  The fix is attached below.


Takashi

===
>From ab1863fc9bc18c806338564124b1e5e7e3ef53d1 Mon Sep 17 00:00:00 2001
From: Takashi Iwai <tiwai@suse.de>
Date: Sun, 7 Jun 2009 12:09:17 +0200
Subject: [PATCH] ALSA: pcm - Fix update of runtime->hw_ptr_interrupt

The commit 13f040f9e55d41e92e485389123654971e03b819 made another
regression, the missing update of runtime->hw_ptr_interrupt.
Since this field is only checked in snd_pcmupdate__hw_ptr_interrupt(),
not in snd_pcm_update_hw_ptr(), it must be updated before the hw_ptr
change check.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
---
 sound/core/pcm_lib.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/sound/core/pcm_lib.c b/sound/core/pcm_lib.c
index bf34603..adb306f 100644
--- a/sound/core/pcm_lib.c
+++ b/sound/core/pcm_lib.c
@@ -299,6 +299,8 @@ static int snd_pcm_update_hw_ptr_interrupt(struct snd_pcm_substream *substream)
 		hw_ptr_interrupt =
 			new_hw_ptr - new_hw_ptr % runtime->period_size;
 	}
+	runtime->hw_ptr_interrupt = hw_ptr_interrupt;
+
 	if (substream->stream == SNDRV_PCM_STREAM_PLAYBACK &&
 	    runtime->silence_size > 0)
 		snd_pcm_playback_silence(substream, new_hw_ptr);
@@ -309,7 +311,6 @@ static int snd_pcm_update_hw_ptr_interrupt(struct snd_pcm_substream *substream)
 	runtime->hw_ptr_base = hw_base;
 	runtime->status->hw_ptr = new_hw_ptr;
 	runtime->hw_ptr_jiffies = jiffies;
-	runtime->hw_ptr_interrupt = hw_ptr_interrupt;
 	if (runtime->tstamp_mode == SNDRV_PCM_TSTAMP_ENABLE)
 		snd_pcm_gettime(runtime, (struct timespec *)&runtime->status->tstamp);
 
-- 
1.6.3.2