[alsa-devel] Something wrong with snd_pcm_avail_update
Heya!
There seems to be something wrong with snd_pcm_update_avail() for mmap in recent 2.6.26 kernels:
Quoting Alexander Gavrilov:
<snip> Fedora 9 with kernel 2.6.26.3-29 exhibits a bug where the driver decrements its hw.ptr under certain stress conditions, or at least it appears so from debugging.
It causes alsa-lib to give absurdly huge avail numbers to pulseaudio, which crashes it. This patch adds a work-around to avoid such unfriendly behavior until the bug is fixed.
As this is not a real fix, it displays a message every time it is activated. </snip>
https://bugzilla.redhat.com/show_bug.cgi?id=462200
(see the patch attached to that bug report)
Takashi, Jaroslav, do you know anything about this?
Lennart
At Wed, 24 Sep 2008 02:19:54 +0200, Lennart Poettering wrote:
Heya!
There seems to be something wrong with snd_pcm_update_avail() for mmap in recent 2.6.26 kernels:
Quoting Alexander Gavrilov:
<snip> Fedora 9 with kernel 2.6.26.3-29 exhibits a bug where the driver decrements its hw.ptr under certain stress conditions, or at least it appears so from debugging.
It causes alsa-lib to give absurdly huge avail numbers to pulseaudio, which crashes it. This patch adds a work-around to avoid such unfriendly behavior until the bug is fixed.
As this is not a real fix, it displays a message every time it is activated.
</snip>
https://bugzilla.redhat.com/show_bug.cgi?id=462200
(see the patch attached to that bug report)
Takashi, Jaroslav, do you know anything about this?
The DMA position handling on some HD-audio hardware is unstable. This could be a reason.
The latest 2.6.27-rc* kernel has a few workarounds for such. Give it a try.
Takashi
On Wed, 24 Sep 2008, Lennart Poettering wrote:
Heya!
There seems to be something wrong with snd_pcm_update_avail() for mmap in recent 2.6.26 kernels:
Quoting Alexander Gavrilov:
<snip> Fedora 9 with kernel 2.6.26.3-29 exhibits a bug where the driver decrements its hw.ptr under certain stress conditions, or at least it appears so from debugging.
It causes alsa-lib to give absurdly huge avail numbers to pulseaudio, which crashes it. This patch adds a work-around to avoid such unfriendly behavior until the bug is fixed.
As this is not a real fix, it displays a message every time it is activated.
</snip>
https://bugzilla.redhat.com/show_bug.cgi?id=462200
(see the patch attached to that bug report)
Takashi, Jaroslav, do you know anything about this?
If it's HDA hardware, the user might try to test the recent linux kernels or the ALSA driver compiled from tarball available on www.alsa-project.org. The hw position routines were improved recently for the HDA driver.
Jaroslav
----- Jaroslav Kysela perex@perex.cz Linux Kernel Sound Maintainer ALSA Project, Red Hat, Inc.
participants (3)
-
Jaroslav Kysela -
Lennart Poettering -
Takashi Iwai