Re: [Alsa-user] Audio drop-out with ac3 on iec958 during heavy SATA io
Brought to you by:
perex
From: Lee R. <rlr...@jo...> - 2006-09-10 17:50:57
|
On Sun, 2006-09-10 at 14:24 +0200, Dominique Dumont wrote: > Takashi Iwai <ti...@su...> writes: > > > Then check whether XRUN occurs at next. Make sure that you compiled > > the driver with debug option (--with-debug=full configure option), > > then do the following as root: > > > > # echo 2 > /proc/asound/card0/pcm2p/xrun_debug > > > > then play the ac3 file. You'll get stacktraces in kernel messages if > > XRUN is detected by the sound driver. It's not perfect but an easy > > way. > > Since alsa site is down, I've tried to get XRUN traces with the > kernel. I've enabled SND_DEBUG and CONFIG_SND_PCM_XRUN_DEBUG. > > Note that I got a few XRUN while I got a *lot* of ac3 ... err > ... hiccups (for lack of a better word). > > Here's what I got (with your patch) with a recompiled 2.6.17 kernel: > (I started ac3dec, then md5sum) > > Sep 10 13:55:57 gandalf kernel: XRUN: pcmC0D2p > Sep 10 13:55:57 gandalf kernel: <f0987cdd> snd_pcm_period_elapsed+0x1ef/0x27d [snd_pcm] <f0b060ad> snd_emu10k1_fx8010_interrupt+0x37/0x57 [snd_emu10k1] > Sep 10 13:55:57 gandalf kernel: <f0affda5> snd_emu10k1_interrupt+0x2ad/0x390 [snd_emu10k1] <b0123e04> do_timer+0x3a/0x2e2 I see your kernel uses the ACPI PM timer for timing. Is it a dual core AMD by any chance? If you use an SMP kernel, do you get different results with a UP kernel? Lee |