Thread: [Alsa-user] USB audio on SL10.0
Brought to you by:
perex
From: Alfons A. <fon...@al...> - 2006-01-30 15:51:47
|
Hello, This weekend I installed SL10.0 on my Thinkpad. Everything runs fine except for USB audio (EDIROL UA-2). I had it working perfectly with -p 256 -n 3 on SL9.2, but now even -p 1024 gives problems, in two ways. All this with jack 0.99.61 (same one as I used on 9.2 - it's newer than the version supplied with 10.0 !) in RT mode, using the hw device. - Switching workspaces in WM gives the occasional xrun. - Exactly every 40 seconds something happens that triggers a burst of timeouts in jack. I've killed every non-essential daemon I could find, but the problem remains. Any hints ? TIA, -- FA |
From: Clemens L. <cl...@la...> - 2006-01-30 17:40:23
|
Alfons Adriaensen wrote: > This weekend I installed SL10.0 on my Thinkpad. Everything runs fine > except for USB audio (EDIROL UA-2). I had it working perfectly with > -p 256 -n 3 on SL9.2, but now even -p 1024 gives problems, in two > ways. What are the ALSA versions used by these? > - Switching workspaces in WM gives the occasional xrun. What graphics driver are you using? Does disabling 3D acceleration help? > - Exactly every 40 seconds something happens that triggers > a burst of timeouts in jack. This might be your BIOS going into SMM to check for powersaving stuff. Try enabling or disabling ACPI. Regards, Clemens |
From: fons a. <fon...@sk...> - 2006-01-30 23:55:19
|
On Mon, Jan 30, 2006 at 06:39:57PM +0100, Clemens Ladisch wrote: > Alfons Adriaensen wrote: > > This weekend I installed SL10.0 on my Thinkpad. Everything runs fine > > except for USB audio (EDIROL UA-2). I had it working perfectly with > > -p 256 -n 3 on SL9.2, but now even -p 1024 gives problems, in two > > ways. > > What are the ALSA versions used by these? 1.0.9b > > - Switching workspaces in WM gives the occasional xrun. > > What graphics driver are you using? > Does disabling 3D acceleration help? Same as before on SL9.2. Graphics HW is ATI RV250Lf, no 3D. I didn't see that problem again today, even if tried quite hard. > > - Exactly every 40 seconds something happens that triggers > > a burst of timeouts in jack. > > This might be your BIOS going into SMM to check for powersaving stuff. > Try enabling or disabling ACPI. It *is* related to ACPI - starting the kernel with acpi=off removes the problem - and of course some desirable features :-( This was not required with 9.2, and I did *days* of live recording with that setup without a single incident, so I don't think it's the BIOS. I killed everything related to acpi - no effect. There seems to be a new process [kacpid] that I can't remember having seen before. It can't be killed, and I can only assume the problem is somewhere inside that one. I tried all sorts of BIOS interrupt assignments, nothing helped. So maybe the only solution is a patched kernel... Now, probably because I'm using -n 3 on USB, all of this happens without any effect on the audio, but qjackctl doesn't like the shower of messages, and also ardour's cpu load goes way up for some seconds after the timeout events, so it's not a healthy situation. -- FA |
From: Lee R. <rlr...@jo...> - 2006-01-31 00:05:42
|
On Tue, 2006-01-31 at 01:01 +0100, fons adriaensen wrote: > > What are the ALSA versions used by these? > > 1.0.9b Any chance you can try the latest ALSA version and/or kernel to rule out a bug that has since been fixed? Lee |
From: fons a. <fon...@sk...> - 2006-01-31 01:00:24
|
On Mon, Jan 30, 2006 at 07:05:34PM -0500, Lee Revell wrote: > On Tue, 2006-01-31 at 01:01 +0100, fons adriaensen wrote: > > > What are the ALSA versions used by these? > > > > 1.0.9b > > Any chance you can try the latest ALSA version and/or kernel to rule out > a bug that has since been fixed? Things have worked perfectly with a series of older versions on the same hardware. If there is documented evidence that a bug that explains the particular problem I have has been introduced in the version I use, and has been corrected again in later versions, then I'd consider upgrading the ALSA code. As a random try without even a hint if will help or not - no way. It would only complicate the situation. -- FA |
From: Lee R. <rlr...@jo...> - 2006-01-31 01:30:49
|
On Tue, 2006-01-31 at 02:06 +0100, fons adriaensen wrote: > On Mon, Jan 30, 2006 at 07:05:34PM -0500, Lee Revell wrote: > > On Tue, 2006-01-31 at 01:01 +0100, fons adriaensen wrote: > > > > What are the ALSA versions used by these? > > > > > > 1.0.9b > > > > Any chance you can try the latest ALSA version and/or kernel to rule out > > a bug that has since been fixed? > > Things have worked perfectly with a series of older versions on the same > hardware. If there is documented evidence that a bug that explains the > particular problem I have has been introduced in the version I use, and > has been corrected again in later versions, then I'd consider upgrading > the ALSA code. As a random try without even a hint if will help or not - > no way. It would only complicate the situation. > Well you could try just upgrading the kernel and leaving userspace alone, that would be trivial to revert, and would rule out a problem at the kernel USB layer... Lee |
From: Alfons A. <fon...@al...> - 2006-01-31 10:05:11
|
On Mon, Jan 30, 2006 at 08:30:42PM -0500, Lee Revell wrote: > Well you could try just upgrading the kernel and leaving userspace > alone, that would be trivial to revert, and would rule out a problem at > the kernel USB layer... That could well be the next step. -- FA |
From: Clemens L. <cl...@la...> - 2006-01-31 08:18:09
|
fons adriaensen wrote: > On Mon, Jan 30, 2006 at 06:39:57PM +0100, Clemens Ladisch wrote: > > Alfons Adriaensen wrote: > > > This weekend I installed SL10.0 on my Thinkpad. Everything runs fine > > > except for USB audio (EDIROL UA-2). I had it working perfectly with > > > -p 256 -n 3 on SL9.2, but now even -p 1024 gives problems, in two > > > ways. > > > > What are the ALSA versions used by these? > > 1.0.9b You could try 1.0.11rc3, but your problem doesn't seem to be USB related. (You could try your internal sound card, just to be sure.) > > > - Exactly every 40 seconds something happens that triggers > > > a burst of timeouts in jack. > > > > This might be your BIOS going into SMM to check for powersaving stuff. > > Try enabling or disabling ACPI. > > It *is* related to ACPI - starting the kernel with acpi=off > removes the problem - and of course some desirable features :-( > This was not required with 9.2, Slack 9.2 used a 2.4 kernel. Did that one have your desirable features? > and I did *days* of live recording with that setup without a single > incident, so I don't think it's the BIOS. ACPI specifies how the kernel executes things defined by the BIOS. I guess the old kernel just didn't have much ACPI support. Regards, Clemens |
From: Alfons A. <fon...@al...> - 2006-01-31 09:49:37
|
On Tue, Jan 31, 2006 at 09:17:45AM +0100, Clemens Ladisch wrote: > You could try 1.0.11rc3, but your problem doesn't seem to be USB > related. > > (You could try your internal sound card, just to be sure.) No problem at all with that one (intel8x0). > > > > - Exactly every 40 seconds something happens that triggers > > > > a burst of timeouts in jack. > > > > > > This might be your BIOS going into SMM to check for powersaving stuff. > > > Try enabling or disabling ACPI. > > > > It *is* related to ACPI - starting the kernel with acpi=off > > removes the problem - and of course some desirable features :-( > > This was not required with 9.2, > > Slack 9.2 used a 2.4 kernel. Did that one have your desirable features? > > ACPI specifies how the kernel executes things defined by the BIOS. > I guess the old kernel just didn't have much ACPI support. ??? SL here means SuSE Linux, not Slack - sorry about the confusion... SL9.2 was a 2.6.8 kernel. Full ACPI support, worked a charm. SL10.0 uses 2.6.13. I haven't noticed any recent problems with snd_usb_audio on the LAD or ALSA lists, and sound has always worked OOTB with all SuSE versions I ever used. So I felt quite confident about this upgrade. -- FA |
From: Clemens L. <cl...@la...> - 2006-01-31 16:34:59
|
Alfons Adriaensen wrote: > On Tue, Jan 31, 2006 at 09:17:45AM +0100, Clemens Ladisch wrote: > > You could try 1.0.11rc3, but your problem doesn't seem to be USB > > related. > > > > (You could try your internal sound card, just to be sure.) > > No problem at all with that one (intel8x0). How small can you make Jack's period size? > SL9.2 was a 2.6.8 kernel. Full ACPI support, worked a charm. SL10.0 > uses 2.6.13. I haven't noticed any recent problems with snd_usb_audio > on the LAD or ALSA lists, snd_usb_audio is broken in both 2.6.14 and 2.6.15 (2.6.15.2 or ALSA 1.0.11rc3 are OK). The latest ALSA's USB driver has many enhancements, it should be worth a try. HTH Clemens |
From: fons a. <fon...@sk...> - 2006-02-02 00:38:34
|
On Tue, Jan 31, 2006 at 05:34:39PM +0100, Clemens Ladisch wrote: > snd_usb_audio is broken in both 2.6.14 and 2.6.15 (2.6.15.2 or ALSA > 1.0.11rc3 are OK). I installed 2.6.15.2 today - the problem remains. Added the rt-preempt patch (the latest one applied with a few small shifts), and modified IRQ thread prios in all imaginable combinations - still the same result. Only completely disabling acpi on the kernel command line seems to help. Apparently something has changed in the acpi system since 2.6.8 where it worked perfectly for me. I'm now seriously considering going back to SL9.2 :-( -- FA |
From: Lee R. <rlr...@jo...> - 2006-02-02 01:30:55
|
On Thu, 2006-02-02 at 01:44 +0100, fons adriaensen wrote: > On Tue, Jan 31, 2006 at 05:34:39PM +0100, Clemens Ladisch wrote: > > > snd_usb_audio is broken in both 2.6.14 and 2.6.15 (2.6.15.2 or ALSA > > 1.0.11rc3 are OK). > > I installed 2.6.15.2 today - the problem remains. Added the > rt-preempt patch (the latest one applied with a few small shifts), > and modified IRQ thread prios in all imaginable combinations - > still the same result. > > Only completely disabling acpi on the kernel command line seems > to help. > > Apparently something has changed in the acpi system since 2.6.8 > where it worked perfectly for me. > > I'm now seriously considering going back to SL9.2 :-( > If you have time to hunt it down, I'm sure the ACPI maintainers would like to hear about it, unfortunately it sounds like it will be a real pain... Lee |