various drivers get underruns

[chepioq]

I use Fedora and I have this message in /var/log/messages

[root@Dominique ~]# grep pulse /var/log/messages
Nov 23 15:05:47 localhost pulseaudio[3137]: module-alsa-sink.c: Increasing wakeup watermark to 34,01 ms
Nov 23 15:08:55 localhost pulseaudio[3137]: module-alsa-sink.c: Increasing wakeup watermark to 68,03 ms
Nov 23 15:25:54 localhost pulseaudio[3137]: module-alsa-sink.c: Increasing wakeup watermark to 34,01 ms
Nov 23 15:29:42 localhost pulseaudio[3137]: module-alsa-sink.c: Increasing wakeup watermark to 34,01 ms
Nov 23 15:39:33 localhost pulseaudio[3137]: module-alsa-sink.c: Increasing wakeup watermark to 34,01 ms
Nov 23 16:27:38 localhost pulseaudio[3137]: module-alsa-sink.c: Increasing wakeup watermark to 34,01 ms
Nov 23 17:03:19 localhost pulseaudio[3137]: module-alsa-sink.c: Increasing wakeup watermark to 36,64 ms
Nov 23 19:48:50 localhost pulseaudio[3137]: module-alsa-sink.c: Increasing wakeup watermark to 34,01 ms
Nov 23 20:22:07 localhost pulseaudio[3156]: module-alsa-sink.c: Increasing wakeup watermark to 40,00 ms
Nov 23 20:28:31 localhost pulseaudio[3156]: module-alsa-sink.c: Increasing wakeup watermark to 34,01 ms
Nov 23 21:36:17 localhost pulseaudio[3156]: module-alsa-sink.c: Increasing wakeup watermark to 34,01 ms
Nov 23 21:43:31 localhost pulseaudio[3156]: module-alsa-sink.c: Increasing wakeup watermark to 3,99 ms
Nov 24 05:33:26 localhost pulseaudio[3096]: module-alsa-sink.c: Increasing wakeup watermark to 40,00 ms
Nov 24 06:40:13 localhost pulseaudio[3096]: module-alsa-sink.c: Increasing wakeup watermark to 34,01 ms
Nov 24 06:41:14 localhost pulseaudio[3096]: module-alsa-sink.c: Increasing wakeup watermark to 36,64 ms
Nov 24 06:47:39 localhost pulseaudio[3096]: module-alsa-sink.c: Increasing wakeup watermark to 36,64 ms
Nov 24 07:28:40 localhost pulseaudio[3096]: module-alsa-sink.c: Increasing wakeup watermark to 36,64 ms
Nov 24 07:29:47 localhost pulseaudio[3096]: module-alsa-sink.c: Increasing wakeup watermark to 34,01 ms
Nov 24 08:15:51 localhost pulseaudio[3067]: module-alsa-sink.c: Increasing wakeup watermark to 40,00 ms
Nov 24 08:23:23 localhost pulseaudio[3067]: module-alsa-sink.c: Increasing wakeup watermark to 80,00 ms
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: Increasing wakeup watermark to 36,64 ms
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: Increasing wakeup watermark to 73,29 ms
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 09:32:35 localhost pulseaudio[3067]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 11:08:05 localhost pulseaudio[3142]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 11:14:55 localhost pulseaudio[3142]: module-alsa-sink.c: Increasing wakeup watermark to 5,99 ms
Nov 24 11:14:57 localhost pulseaudio[3142]: module-alsa-sink.c: Increasing wakeup watermark to 11,97 ms
Nov 24 11:15:20 localhost pulseaudio[6848]: pid.c: Daemon already running.
Nov 24 11:16:09 localhost pulseaudio[3142]: module-alsa-sink.c: Increasing wakeup watermark to 23,95 ms
Nov 24 11:16:19 localhost pulseaudio[3142]: module-alsa-sink.c: Increasing wakeup watermark to 47,89 ms
Nov 24 11:16:40 localhost pulseaudio[7987]: pid.c: Daemon already running.
Nov 24 11:18:14 localhost pulseaudio[3142]: module-alsa-sink.c: Increasing wakeup watermark to 95,78 ms
Nov 24 11:18:44 localhost pulseaudio[9410]: pid.c: Daemon already running.
Nov 24 11:26:49 localhost pulseaudio[3142]: module-alsa-sink.c: Increasing wakeup watermark to 191,56 ms
Nov 24 11:26:59 localhost pulseaudio[3142]: alsa-util.c: snd_pcm_avail_update() returned a value that is exceptionally large: 349728 bytes (1982 ms) Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 11:26:59 localhost pulseaudio[3142]: module-alsa-sink.c: Increasing wakeup watermark to 368,53 ms
Nov 24 11:52:06 localhost pulseaudio[3139]: module-alsa-sink.c: Increasing wakeup watermark to 40,00 ms
Nov 24 11:52:15 localhost pulseaudio[3139]: module-alsa-sink.c: Increasing wakeup watermark to 80,00 ms
Nov 24 11:52:38 localhost pulseaudio[12264]: pid.c: Daemon already running.
Nov 24 12:06:57 localhost pulseaudio[3139]: module-alsa-sink.c: Increasing wakeup watermark to 160,00 ms
Nov 24 12:11:24 localhost pulseaudio[3139]: alsa-util.c: snd_pcm_avail_update() returned a value that is exceptionally large: 378208 bytes (2144 ms) Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 12:11:24 localhost pulseaudio[3139]: module-alsa-sink.c: Increasing wakeup watermark to 320,00 ms
Nov 24 12:13:21 localhost pulseaudio[3143]: module-alsa-sink.c: Increasing wakeup watermark to 40,00 ms
Nov 24 12:14:18 localhost pulseaudio[3143]: module-alsa-sink.c: Increasing wakeup watermark to 80,00 ms
Nov 24 12:14:45 localhost pulseaudio[3143]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 12:17:26 localhost pulseaudio[3143]: module-alsa-sink.c: Increasing wakeup watermark to 5,99 ms
Nov 24 12:17:36 localhost pulseaudio[3143]: module-alsa-sink.c: Increasing wakeup watermark to 11,97 ms
Nov 24 12:17:48 localhost pulseaudio[3143]: module-alsa-sink.c: Increasing wakeup watermark to 17,01 ms
Nov 24 14:07:02 localhost pulseaudio[3143]: module-alsa-sink.c: Increasing wakeup watermark to 34,01 ms
Nov 24 16:18:52 localhost pulseaudio[3085]: module-alsa-sink.c: Increasing wakeup watermark to 40,00 ms
Nov 24 16:19:02 localhost pulseaudio[3085]: alsa-util.c: snd_pcm_avail_update() returned a value that is exceptionally large: 360128 bytes (2041 ms) Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.
Nov 24 16:19:02 localhost pulseaudio[3085]: module-alsa-sink.c: Increasing wakeup watermark to 80,00 ms
Nov 24 16:19:03 localhost pulseaudio[3085]: module-alsa-sink.c: Increasing wakeup watermark to 160,00 ms
[root@Dominique ~]# 

I don't know if it is important but I report...

[abompard]

I confirm this problem, on Fedora 10, with PulseAudio 0.9.13. When the following line occurs :

module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSAdriver bug. Please report this issue to the PulseAudio developers.

There's a hiccup in the playback.

[rsingh]

I also confirm the bug in Fedora 10... keeps coming in the messages.

module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSAdriver bug. Please report this issue to the PulseAudio developers.

I have a Dell Inspiron Laptop i945GM Chipset Internal Sigmatel Soundcard

[ralfs]

I also confirm the bug in Fedora 10:

pulseaudio[2573]: module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.

I have: 00:11.5 Multimedia audio controller: VIA Technologies, Inc. VT8233/A/8235/8237 AC97 Audio Controller (rev 60)

pulseaudio-0.9.13-6.fc10.i386 alsa-plugins-pulseaudio-1.0.18-1.rc3.fc10.i386

[atenrok]

same story here. Sound playback skips for several times a second and this message appears in logfile.

module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers.

Nforce4-based motherboard with built-in Nvidia CK804 audiocontroller.

[bleargh]

I can also confirm the bug in Fedora 10. Same error message, same stuttering. Really annoying, especially considering that it was touted as "glitch-free audio"... stuttering repeatedly throughout playback doesn't sound glitch free. :(

Stutters bad enough here that it basically makes music playback pointless; stutters for ~5-10s at least six or seven times in the course of a ~4min song.

I'm using an Asus K8n-DL motherboard, which has an on-board nVidia CK804 audio controller.

[hmmsjan]

Confirm the problem, also FC10

Mobo: Asus P4P800

Sound cards: Sound blaster PCI module snd_ens1371

Motherboard Intel,module snd_intel8x0

/sbin/lspci: 02:0b.0 Multimedia audio controller: Ensoniq 5880 AudioPCI (rev 02) 00:1f.5 Multimedia audio controller: Intel Corporation 82801EB/ER (ICH5/ICH5R) AC'97 Audio Controller (rev 02,

The on-board card is OK, the Sound blaster is unusable. Worked fine with pulseaudio in Fedora 8.

From /var/log/messages:

Dec 6 22:52:42 maarten pulseaudio[3122]: alsa-util.c: Device front:1 doesn't support 44100 Hz, changed to 44099 Hz.

Dec 6 22:55:42 maarten pulseaudio[3122]: module-alsa-sink.c: Increasing wakeup watermark to 160.00 ms Dec 6 22:55:45 maarten pulseaudio[3122]: module-alsa-sink.c: Increasing wakeup watermark to 207.01 ms

Bypassing pulseaudio and directing mplayer to ALSA with "mplayer -ao alsa:device=hw=1.0" works correctly. Pity, because I like the pulseaudio card switching and volume sliders for each application.

[henry]

Hi,

here the problem occurs when

default-sample-channels = 4

is configured in the file

/etc/pulse/daemon.conf

Fedora 10, SB0400 Audigy2 Value, snd_emu10k1, pulseaudio.x86_64 0.9.13-6.fc10 alsa-plugins-pulseaudio.x86_64 1.0.18-1.rc3.fc10

[hmmsjan]

Hi,

Have default-sample-channels=2 here, changing did not help.

I grabbed the pulseaudio source RPM from Fedora 8, extracted it and compiled with

./configure --prefix=/pa2; make; make install.

/pa2/bin/pulseaudio -D

export LD_CONFIG_PATH=/pa2/lib

mplayer -ao pulse nicemusicfromkco_amsterdam.mp3

Some error messages appear:

shm.c: Invalid shared memory segment size alsa-util.c: Device front:1 doesn't support 44100 Hz, changed to 44099 Hz. shm.c: Invalid shared memory segment size pstream.c: Failed to import memory block.

but.... works perfect out-of-the-box with only 10% of CPU-load of to the FC10 version...

Kernel drivers and alsa-lib belong to Fedora 10, so I think there are some changes in pulseaudio which my aged ens1371 chip does not like :(

Good Luck....

[hmmsjan]

Hi,

I've seen in an article that there has been a fundamental change in pulseaudio between FC8 and FC10, from interrupt driven to timer-driven. Always learned that interrupt is the way to go when driving hardware optimally, but there are sensible counterarguments in this case.... This may explain the problems with my low-end card, but apparently I'm not the only one. Seen in another article that you can go back to the old mode by adding "tsched=0" to the hal-detect module load in default.pa. In my case, after changing that, I've a few seconds sound and then a crash of the pulseaudio daemon, so this really did not help me.... Tried in mean time the 9.11 version of pulseaudio, could be a bit better than 9.13, but also stuttering. Only the FC8 version is working perfectly.

Kind Regards,

[lennart]

Apparently some ac97 and creative drivers are broken and cause this misbehaviour. Sorry for that. Still waiting for fixes from upstream ALSA.

[LoboPops]

My sound wasn't working at all after installing last night. Was working fine with F9. Got the same message as ticket and then it got much worse.Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:15:35 localhost pulseaudio[2865]: asyncq.c: q overrun, queuing locally Dec 18 22:17:03 localhost pulseaudio[2865]: cpulimit.c: Recevied request to terminate due to CPU overload. Dec 18 22:17:03 localhost pulseaudio[2865]: asyncmsgq.c: Assertion 'pa_asyncq_push(a->asyncq, &i, TRUE) == 0' failed at pulsecore/asyncmsgq.c:164, function pa_asyncmsgq_send(). Aborting.

[zak_neutron]

Hi

I can confirm this too - I too have thousands of

"the module-alsa-sink.c: ALSA woke us up to write new data to the device, but there was actually nothing to write! Most likely this is an ALSA driver bug. Please report this issue to the PulseAudio developers."

in my /var/log/messages

I am on F10

PA - 0.9.13

I have Dell Creative Labs Sound blaster soundcard - EMU10K1X.

I have had intermittent stuttering on Firefly media server playback on F10 client machines

[stevefalco]

I too am getting audio breakups, along with the messages others have reported (module-alsa-sink.c: ALSA woke us...).

The interesting thing in my case is that I have the problem with Amarok, but not with RhythmBox?. I.e. Amarok is unlistenable, but RhythmBox? plays flawlessly. Not sure what underlying players each app is using, but it apparently makes a big difference.

System is Fedora 10, x86_64, pulseaudio-0.9.13-6.fc10.x86_64. Soundcard is SB Audigy 2 ZS [SB0350] EMU10K1_Audigy.

[stevefalco]

As suggested in "12/15/08 07:46:09 changed by hmmsjan", I tried building the F8 version of PulseAudio, and this cured my Amarok player stuttering problem. I should also slightly revise my previous report. It turns out I should have said "Rhythmbox plays almost flawlessly" - under high load, it too stuttered, but to a much lesser degree than Amarok.

In message "12/16/08 16:21:25 changed by lennart", there was mention of an upstream ALSA bug. Does anyone know the ALSA bug number - I'd like to follow its progress.

[Authenticated User]

I was able to reduce this issue on Fedora 10 to a minimum (on the following hardware/software) by peforming the actions described below.

lspci|grep -i audio; uname -a; rpm -q pulseaudio

00:04.0 Multimedia audio controller: nVidia Corporation CK804 AC'97 Audio Controller (rev a2)
Linux cp2.local 2.6.27.9-159.fc10.x86_64 #1 SMP Tue Dec 16 14:47:52 EST 2008 x86_64 x86_64 x86_64 GNU/Linux
pulseaudio-0.9.13-6.fc10.x86_64

1. Add the user(s) to the "pulse-rt" group

useradd -G pulse-rt <user>

2. Add the following (from ALSA bug report for this issue) to /etc/modprobe.conf

options snd-ac97-codec power_save=0
options snd-intel8x0 index=0 ac97_clock=48000

3. Add the following (determined using the output reported for the sound card by running pulseaudio -v) to /etc/pulse/daemon.conf

default-fragments = 4
default-fragment-size-msec = 1920
default-sample-rate = 48000
no-cpu-limit = yes

4. Install the following ALSA updates

rpm -Uvh http://kojipkgs.fedoraproject.org/packages/alsa-lib/1.0.18/7.fc10/x86_64/alsa-lib-1.0.18-7.fc10.x86_64.rpm
rpm -Uvh http://kojipkgs.fedoraproject.org/packages/alsa-plugins/1.0.18/2.fc10/x86_64/alsa-plugins-pulseaudio-1.0.18-2.fc10.x86_64.rpm

[greg0ire]

Hi, I'm trying to apply your solution, but I can't install rpm file http://kojipkgs.fedoraproject.org/packages/alsa-lib/1.0.18/7.fc10/i386/alsa-lib-1.0.18-7.fc10.i386.rpm I get this dependancy message, which insinuates I'm trying to install... alsa-lib-devel... It seems this RPM file is wrong, isn't it? Dépendance manquante : alsa-lib = 1.0.18-6.rc3.fc10 est requis par le paquetage alsa-lib-devel-1.0.18-6.rc3.fc10.i386

[chepioq]

Hello greg0ire you must install the alsa-lib-devel-1.0.18-7.fc10.i386.rpm package because you have alsa-lib-devel-1.0.18-6.rc3.fc10.i386...

[greg0ire]

Thank you, I finally managed to install this rpm thanks to your advice, but /var/log/messages is still flooded with these messages...
I am not sure I put the good values at step 3, I did not know how to find them in pulseaudio -v out...

[Authenticated User]

When running pulseaudio -v, under the snd_pcm_dump() section, I used "rate" for the default-sample-rate (and also in the ac97_clock setting in the /etc/modprobe.conf step); "period_time" as the basis for calculating default-fragment-size-msec (I found no value in setting this higher than pulseaudio's 2 second max); the default-fragments was trial and error with values ranging from 1 to 64.

[patrys]

Disclaimer: I'm not using Fedora but PLD Linux. It seemed pointless to create a bug for every distro.

I'm getting a lot of:

W: protocol-native.c: Failed to push data into queue
E: asyncq.c: q overrun, queuing locally
E: asyncmsgq.c: Assertion 'pa_asyncq_push(a->asyncq, &i, TRUE) == 0' failed at pulsecore/asyncmsgq.c:164, function pa_asyncmsgq_send(). Aborting.

The first two can flood the logs for quite a long time before the third one appears usually taking down an application or two.

$ rpm -q pulseaudio alsa-lib alsa-plugins-pulse gstreamer gstreamer-pulseaudio
pulseaudio-0.9.14-1.i686
alsa-lib-1.0.18-1.i686
alsa-plugins-pulse-1.0.18-1.i686
gstreamer-0.10.22-3.i686
gstreamer-pulseaudio-0.10.11-1.i686

Using the per-user PA daemon (not the global RT one as some people above).

I'm a developer so I will gladly help with debugging or testing patches, just point me in the right direction.

[Authenticated User]

The issue has been resolved on my specific equipment with the combination of the latest kernel, ALSA and pulseaudio components. Here's my output from alsa-info and a list of the specific versions of ALSA and pulseaudio packages that are working for me.

lspci|grep -i audio; uname -a; rpm -q pulseaudio

00:04.0 Multimedia audio controller: nVidia Corporation CK804 AC'97 Audio Controller (rev a2)
Linux cp2.local 2.6.27.12-170.2.5.fc10.x86_64 #1 SMP Wed Jan 21 01:33:24 EST 2009 x86_64 x86_64 x86_64 GNU/Linux
pulseaudio-0.9.14-2.fc10.x86_64

rpm -q pulseaudio alsa-firmware alsa-lib alsa-plugins-pulseaudio alsa-utils alsa-tools

pulseaudio-0.9.14-2.fc10.x86_64
alsa-firmware-1.0.17-1.fc10.noarch
alsa-lib-1.0.19-1.fc10.x86_64
alsa-plugins-pulseaudio-1.0.18-2.fc10.x86_64
alsa-utils-1.0.19-1.fc10.x86_64
alsa-tools-1.0.17-1.fc10.x86_64

The steps to minimize the issue that I undertook in reply 16 are still active; however, after the latest versions of ALSA and pulseaudio were installed, I still heard an occasional dropout when the automatically detected tsched=1 option was active, so in /etc/pulse/default.pa I added tsched=0 to the module-hal-detect line and haven't heard a dropout since.

load-module module-hal-detect tsched=0

The occasional "ALSA woke us up to write new data to the device, but there was actually nothing to write!" message still shows in /var/log/messages, but the corresponding audio drop is not. Thank you to both the pulseaudio and ALSA teams for your hard work and dedication; your efforts are greatly appreciated.

[Authenticated User]

Looks like I was too quick to congratulate, even so, my thanks for the effort involved remains steadfast. The issue remains in some circumstances, for example in emulators like MAME and UAE, a continuous stream of "I: module-alsa-sink.c: Underrun!" makes the sound unbearable with tsched=0 but tolerable with tsched=1; conversely, music plays with no dropouts in Audacious when tsched=0 is used in combination with the pulseaudio plugin set with a 960ms buffer.

[timj]

Not sure if this is actually the same issue or not as I don't get underruns, but I do get glitches during playback (sound just like a record skipping) on Fedora 10 with an ICE1712 card (Audiophile 24/96) and:

pulseaudio-0.9.13-6.fc10 alsa-lib-1.0.19-1.fc10

I get these kind of things in the PA log when the glitches occur:

N: module-alsa-sink.c: Increasing wakeup watermark to 40.00 ms N: module-alsa-sink.c: Increasing wakeup watermark to 71.27 ms

but it never seems to really end up at a happy place with no glitches.

Setting tsched=0 seems to do the trick, with no glitches right from the start. I didn't change any of the other parameters (sample rate etc.).

[hmmsjan]

Dear All,

I tried the version for Fedora 11, (pulseaudio-0.9.14-2.fc10.i386), but no improvement. Tried some debugging with my limited knowledge, but the only thing I can find is a suddenly irregular, too high value of snd_pcm_avail() coming without any reason.

pulseaudio -vvv 2>&1 | egrep '(pcm_mmap_begin|Underrun!|commit)'

D: alsa-util.c: Starting with pcm_mmap_begin at 190.554000 with 15539 available

D: module-alsa-sink.c: mmap-commit 12541 3843

D: module-alsa-sink.c: mmap-commit returned 3843

D: alsa-util.c: Starting with pcm_mmap_begin at 1.103000 with 11760 available

D: module-alsa-sink.c: mmap-commit 0 4565

D: module-alsa-sink.c: mmap-commit returned 4565

D: alsa-util.c: Starting with pcm_mmap_begin at 140.762000 with 13411 available

D: module-alsa-sink.c: mmap-commit 4565 6280

D: module-alsa-sink.c: mmap-commit returned 6280

D: alsa-util.c: Starting with pcm_mmap_begin at 190.409000 with 15531 available

D: module-alsa-sink.c: mmap-commit 10845 5539

D: module-alsa-sink.c: mmap-commit returned 5539

D: alsa-util.c: Starting with pcm_mmap_begin at 1.077000 with 10048 available

D: module-alsa-sink.c: mmap-commit 0 2869

D: module-alsa-sink.c: mmap-commit returned 2869

D: alsa-util.c: Starting with pcm_mmap_begin at 141.564000 with 13427 available

D: module-alsa-sink.c: mmap-commit 2869 6296

D: module-alsa-sink.c: mmap-commit returned 6296 I: module-alsa-sink.c: Underrun!

D: alsa-util.c: Starting with pcm_mmap_begin at 3.341000 with 23667 available

D: module-alsa-sink.c: mmap-commit 9165 7219

D: module-alsa-sink.c: mmap-commit returned 7219

If I can help with further debugging, please send private mail, with my old soundcard the problem appears within a few minutes.... Alsa-lib: alsa-lib-1.0.19-1.fc10.i686

P.S. With tsched=0, sound is find despite of many "Alsa woke up...." message, but pulseaudio frequently crashes upon startup with:

E: source.c: Assertion 'PA_SOURCE_IS_OPENED(s->thread_info.state)' failed at pulsecore/source.c:428, function pa_source_post(). Aborting.

Kind Regards, H.Janssen Alkmaar

[lennart]

Please follow what is suggested at the end of wiki:BrokenSoundDrivers

[lennart]

BTW, the pa_asyncq_ issue has been fixed a while back in git. My last comment only applies to the original issue explained in the bug report.

Guys, you'd do me a great favour if you wouldn't hijack unrelated bug reports. This gets really difficult to follow then. Please understand that I will ignore all comments that are unrleated to the original bug report. If in doubt, create a new bug report and let us merge them instead of hijacking unrelated issues. Thank you.