Skip to contentSkip to conversation

ALSA problems with 2.6.18-rc3

DISCUSSION Andrew Benton 6 participants 19 messages Aug 8, 2006 – Aug 10, 2006

lore
Share & tools
lore
#0 lore raw

Hello World,
I use the optical digital output from my computer and normally it works
well. However, with a 2.6.18-rc* kernel I have to use alsamixer to get
any output from it. I tweak the levels till it's working and use alsactl
store to save it. Normally, when I reboot it's enough to have a
bootscript run alsactl restore and it should work, but as I said, with
a 2.6.18-rc* kernel it doesn't. Every time I have to use alsamixer to
change the <IEC958 P> setting to A/D Conv and then back to AC-Link again
to get the diode to come on and start sending a signal to the amplifier.
And the problems don't stop there. At first it sounds fine but after a few minutes it starts to develop a harsh, tinny, metallic, distorted quality and it becomes unlistenable.

According to alsamixer v1.0.11, the card is an Intel ICH5 and the chip is an Analog Devices AD1981B

Andy

#1 lore raw
Quoted 13 lines · Andrew Benton

Hello World,
I use the optical digital output from my computer and normally it works well. However, with a 2.6.18-rc* kernel I have to use alsamixer to get any output from it. I tweak the levels till it's working and use alsactl store to save it. Normally, when I reboot it's enough to have a bootscript run alsactl restore and it should work, but as I said, with a 2.6.18-rc* kernel it doesn't. Every time I have to use alsamixer to change the <IEC958 P> setting to A/D Conv and then back to AC-Link again to get the diode to come on and start sending a signal to the amplifier. And the problems don't stop there. At first it sounds fine but after a few minutes it starts to develop a harsh, tinny, metallic, distorted quality and it becomes unlistenable.

Quoted 2 lines

According to alsamixer v1.0.11, the card is an Intel ICH5 and the chip is an Analog Devices AD1981B

Please try to identify the change that introduced the regression. What was the last kernel/ALSA version that worked correctly?

Lee

#2 lore raw
Quoted 2 lines · Lee Revell

Please try to identify the change that introduced the regression. What was the last kernel/ALSA version that worked correctly?

The change happened between 2.6.17 and 2.6.18-rc1. Specifically, 2.6.17-git4 works and 2.6.17-git5 doesn't.

Andy

#3 lore raw
Quoted 6 lines · Andrew Benton
Quoted 2 lines · Lee Revell

Please try to identify the change that introduced the regression. What was the last kernel/ALSA version that worked correctly?

The change happened between 2.6.17 and 2.6.18-rc1. Specifically, 2.6.17-git4 works and 2.6.17-git5 doesn't.

Takashi-san,

Does this help at all? Many users are reporting that sound broke with 2.6.18-rc*.

Lee

#4 lore raw
Quoted 5 lines · Lee Revell
Quoted 4 lines · Andrew Benton
Quoted 2 lines · Lee Revell

Please try to identify the change that introduced the regression. What was the last kernel/ALSA version that worked correctly?

Quoted 7 lines
Quoted 2 lines

The change happened between 2.6.17 and 2.6.18-rc1. Specifically, 2.6.17-git4 works and 2.6.17-git5 doesn't.

Takashi-san,

Does this help at all? Many users are reporting that sound broke with 2.6.18-rc*.

Fwiw, 2.6.18-rc1 works on arm (pxa2xx). Maybe this is related to another subsystem not found on arm.

Liam

#5 lore raw
In reply to Lee Revell
Quoted 8 lines · Lee Revell
Quoted 6 lines · Andrew Benton
Quoted 2 lines · Lee Revell

Please try to identify the change that introduced the regression. What was the last kernel/ALSA version that worked correctly?

The change happened between 2.6.17 and 2.6.18-rc1. Specifically, 2.6.17-git4 works and 2.6.17-git5 doesn't.

Quoted 7 lines

Takashi-san,

Does this help at all? Many users are reporting that sound broke with 2.6.18-rc*.

Lee

Takashi-san's suggestion earlier today of running an "alsactl -F restore" seems to have fixed all those diffs right up, I now have good sound with an emu10k1 using an audigy 2 as card-0, running kernel-2.6.18-rc4.

Quoted 5 lines

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 7 lines
Cheers, Gene
People having trouble with vz bouncing email to me should add the word
'online' between the 'verizon', and the dot which bypasses vz's
stupid bounce rules.  I do use spamassassin too. :-)
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2006 by Maurice Eugene Heskett, all rights reserved.
#6 lore raw
Quoted 11 lines · Gene Heskett
Quoted 10 lines · Lee Revell
Quoted 6 lines · Andrew Benton
Quoted 2 lines · Lee Revell

Please try to identify the change that introduced the regression. What was the last kernel/ALSA version that worked correctly?

The change happened between 2.6.17 and 2.6.18-rc1. Specifically, 2.6.17-git4 works and 2.6.17-git5 doesn't.

Takashi-san,

Quoted 8 lines
Quoted 5 lines

Does this help at all? Many users are reporting that sound broke with 2.6.18-rc*.

Lee

Takashi-san's suggestion earlier today of running an "alsactl -F restore" seems to have fixed all those diffs right up, I now have good sound with an emu10k1 using an audigy 2 as card-0, running kernel-2.6.18-rc4.

Distros should probably be using this as a default. Otherwise, simply adding a new mixer control will cause restoring mixer settings to fail.

Lee

#7 lore raw

On Wednesday 09 August 2006 16:41, Lee Revell wrote:
[...]

Quoted 12 lines
Quoted 11 lines
Quoted 6 lines

Takashi-san,

Does this help at all? Many users are reporting that sound broke with 2.6.18-rc*.

Lee

Takashi-san's suggestion earlier today of running an "alsactl -F restore" seems to have fixed all those diffs right up, I now have good sound with an emu10k1 using an audigy 2 as card-0, running kernel-2.6.18-rc4.

Quoted 4 lines

Distros should probably be using this as a default. Otherwise, simply adding a new mixer control will cause restoring mixer settings to fail.

Lee

I already have the 'alsactl restore' in my rc.local. Would there be any harm in just adding the -F to that invocation, or will that just restore it to a 'default' condition always. Seems like it would, canceling anything you have done & then did an 'alsactl store' to save..

Signature · 7 lines
Cheers, Gene
People having trouble with vz bouncing email to me should add the word
'online' between the 'verizon', and the dot which bypasses vz's
stupid bounce rules.  I do use spamassassin too. :-)
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2006 by Maurice Eugene Heskett, all rights reserved.
#8 lore raw
Quoted 17 lines · Gene Heskett

On Wednesday 09 August 2006 16:41, Lee Revell wrote:
[...]

Quoted 15 lines
Quoted 11 lines
Quoted 6 lines

Takashi-san,

Does this help at all? Many users are reporting that sound broke with 2.6.18-rc*.

Lee

Takashi-san's suggestion earlier today of running an "alsactl -F restore" seems to have fixed all those diffs right up, I now have good sound with an emu10k1 using an audigy 2 as card-0, running kernel-2.6.18-rc4.

Distros should probably be using this as a default. Otherwise, simply adding a new mixer control will cause restoring mixer settings to fail.

Quoted 7 lines
Quoted 1 line

Lee

I already have the 'alsactl restore' in my rc.local. Would there be any harm in just adding the -F to that invocation, or will that just restore it to a 'default' condition always. Seems like it would, canceling anything you have done & then did an 'alsactl store' to save..

That's what I was suggesting - just add -F to the alsactl restore in your init script. It won't restore it to a default state - the only difference is that it will do a better job restoring your mixer state if new controls are added by a driver update.

alsactl --help:

-F,--force try to restore the matching controls as much as
possible

Lee

#9 lore raw
Quoted 1 line · Lee Revell

On Wed, 2006-08-09 at 16:51 -0400, Gene Heskett wrote:

[snip]

Quoted 10 lines
Quoted 4 lines

I already have the 'alsactl restore' in my rc.local. Would there be any harm in just adding the -F to that invocation, or will that just restore it to a 'default' condition always. Seems like it would, canceling anything you have done & then did an 'alsactl store' to save..

That's what I was suggesting - just add -F to the alsactl restore in your init script. It won't restore it to a default state - the only difference is that it will do a better job restoring your mixer state if new controls are added by a driver update.

Quoted 4 lines

alsactl --help:

-F,--force try to restore the matching controls as much as
possible

I assume there are drawbacks to such an option, since whatever method is used to "force" the control may make a mistake if similarly named controls are renamed.

Personally, I think the correct approach would be to have more sensible default values. Having the External Amplifier default off when it cripples analogue output on emu10k1, and has no effect on digital output, seems rather weird. Also, I never really understood the rationale for the "all zeros" mixer default. Why not 50%?

Signature · 5 lines
Cheers,
Alistair.

Final year Computer Science undergraduate.
1F2 55 South Clerk Street, Edinburgh, UK.
#10 lore raw
Quoted 18 lines · Alistair John Strachan
Quoted 1 line · Lee Revell

On Wed, 2006-08-09 at 16:51 -0400, Gene Heskett wrote:

[snip]

Quoted 10 lines
Quoted 4 lines

I already have the 'alsactl restore' in my rc.local. Would there be any harm in just adding the -F to that invocation, or will that just restore it to a 'default' condition always. Seems like it would, canceling anything you have done & then did an 'alsactl store' to save..

That's what I was suggesting - just add -F to the alsactl restore in your init script. It won't restore it to a default state - the only difference is that it will do a better job restoring your mixer state if new controls are added by a driver update.

Quoted 4 lines

alsactl --help:

-F,--force try to restore the matching controls as much as
possible

Quoted 8 lines

I assume there are drawbacks to such an option, since whatever method is used to "force" the control may make a mistake if similarly named controls are renamed.

Personally, I think the correct approach would be to have more sensible default values. Having the External Amplifier default off when it cripples analogue output on emu10k1, and has no effect on digital output, seems rather weird.

It's impossible to predict the effect of some mixer controls across the wide range of hardware that ALSA supports. What makes sound work on one machine is likely to break it on another.

Quoted 3 lines

Also, I never really understood the rationale for the "all zeros"
mixer default. Why not 50%?

This is policy and policy belongs in userspace. Distros are free to ship with any default mixer settings they choose.

Lee

#11 lore raw
Quoted 26 lines · Lee Revell
Quoted 25 lines · Alistair John Strachan
Quoted 1 line · Lee Revell

On Wed, 2006-08-09 at 16:51 -0400, Gene Heskett wrote:

[snip]

Quoted 11 lines
Quoted 5 lines

I already have the 'alsactl restore' in my rc.local. Would there be any harm in just adding the -F to that invocation, or will that just restore it to a 'default' condition always. Seems like it would, canceling anything you have done & then did an 'alsactl store' to save..

That's what I was suggesting - just add -F to the alsactl restore in your init script. It won't restore it to a default state - the only difference is that it will do a better job restoring your mixer state if new controls are added by a driver update.

Quoted 4 lines

alsactl --help:

-F,--force try to restore the matching controls as much as
possible

I assume there are drawbacks to such an option, since whatever method is used to "force" the control may make a mistake if similarly named controls are renamed.

Quoted 8 lines
Quoted 4 lines

Personally, I think the correct approach would be to have more sensible default values. Having the External Amplifier default off when it cripples analogue output on emu10k1, and has no effect on digital output, seems rather weird.

It's impossible to predict the effect of some mixer controls across the wide range of hardware that ALSA supports. What makes sound work on one machine is likely to break it on another.

However, ALSA _has_ defaults for these controls, which I believe are usually "off" or "zero". All I'm suggesting is that these defaults are plainly suboptimal for emu10k1, and probably other cards to which this statement simply does not apply. Shipping defaults is one thing, but shipping useless defaults is quite another. We have policy all over the kernel for providing "sane defaults" e.g. filesystem mount options.

Quoted 5 lines
Quoted 2 lines

Also, I never really understood the rationale for the "all zeros"
mixer default. Why not 50%?

This is policy and policy belongs in userspace. Distros are free to ship with any default mixer settings they choose.

True, I suppose if the mixers aren't even vaguely compatible across hardware, "zero" is your best bet. I still see people reporting bugs (such as this) where the distro has failed them. -F is not a complete solution.

Signature · 5 lines
Cheers,
Alistair.

Final year Computer Science undergraduate.
1F2 55 South Clerk Street, Edinburgh, UK.
#12 lore raw
Quoted 6 lines · Alistair John Strachan
Quoted 1 line

It's impossible to predict the effect of some mixer controls across

the

Quoted 1 line

wide range of hardware that ALSA supports. What makes sound work on

one

Quoted 1 line

machine is likely to break it on another.

Quoted 7 lines

However, ALSA _has_ defaults for these controls, which I believe are usually "off" or "zero". All I'm suggesting is that these defaults are plainly suboptimal for emu10k1, and probably other cards to which this statement simply does not apply. Shipping defaults is one thing, but shipping useless defaults is quite another. We have policy all over the kernel for providing "sane defaults" e.g. filesystem mount options.

I think muted is a sane default - the only sane default. Otherwise you could damage speakers or hearing. Also many devices will be noisier if unused inputs are enabled. Is it really that hard for users to unmute the mixer or for distros to create their own config?

Also, analog output on my emu10k1 works perfectly with "External Amplifier" disabled.

Lee

#13 lore raw
Quoted 1 line · Lee Revell

On Wed, 2006-08-09 at 23:07 +0100, Alistair John Strachan wrote:

[snip]

Quoted 8 lines
Quoted 7 lines

However, ALSA _has_ defaults for these controls, which I believe are usually "off" or "zero". All I'm suggesting is that these defaults are plainly suboptimal for emu10k1, and probably other cards to which this statement simply does not apply. Shipping defaults is one thing, but shipping useless defaults is quite another. We have policy all over the kernel for providing "sane defaults" e.g. filesystem mount options.

Quoted 4 lines

I think muted is a sane default - the only sane default. Otherwise you could damage speakers or hearing. Also many devices will be noisier if unused inputs are enabled. Is it really that hard for users to unmute the mixer or for distros to create their own config?

I think this argument is mostly deflated by the fact that other multimedia-centric operating systems do not ship with controls muted. If hardware is detected, all sliders are set to 50%. I think this is superior.

As you pointed out, it probably is best left to distros. I just can't help thinking that there might be some scope for improving the default state of boolean tunables, even if sliders are left at zero.

Quoted 2 lines

Also, analog output on my emu10k1 works perfectly with "External Amplifier" disabled.

Interesting, evidence in your favour. It does not work here. It also did not work on the AC'97 in my HP NC6000, which also has External something-or-other and which also defaulted to "off".

Took some figuring out. :-)

Signature · 5 lines
Cheers,
Alistair.

Final year Computer Science undergraduate.
1F2 55 South Clerk Street, Edinburgh, UK.
#14 lore raw
Quoted 4 lines · Alistair John Strachan

I think this argument is mostly deflated by the fact that other multimedia-centric operating systems do not ship with controls muted. If hardware is detected, all sliders are set to 50%. I think this is superior.

The developers audio drivers for other operating systems have full access to the documentation of all the hardware they support, so it's possible for them to establish "working" defaults for every device.

Lee

#15 lore raw
In reply to Alistair John Strachan

At Wed, 9 Aug 2006 23:28:47 +0100,

Quoted 7 lines · Alistair John Strachan
Quoted 3 lines

Also, analog output on my emu10k1 works perfectly with "External Amplifier" disabled.

Interesting, evidence in your favour. It does not work here. It also did not work on the AC'97 in my HP NC6000, which also has External something-or-other and which also defaulted to "off".

The external amplifier is on as default in the driver level. There must be something else that took off.

Takashi

#16 lore raw
In reply to Lee Revell
Quoted 33 lines · Lee Revell
Quoted 19 lines · Gene Heskett

On Wednesday 09 August 2006 16:41, Lee Revell wrote:
[...]

Quoted 16 lines
Quoted 11 lines
Quoted 6 lines

Takashi-san,

Does this help at all? Many users are reporting that sound broke with 2.6.18-rc*.

Lee

Takashi-san's suggestion earlier today of running an "alsactl -F restore" seems to have fixed all those diffs right up, I now have good sound with an emu10k1 using an audigy 2 as card-0, running kernel-2.6.18-rc4.

Distros should probably be using this as a default. Otherwise, simply adding a new mixer control will cause restoring mixer settings to fail.

Quoted 7 lines
Quoted 1 line

Lee

I already have the 'alsactl restore' in my rc.local. Would there be any harm in just adding the -F to that invocation, or will that just restore it to a 'default' condition always. Seems like it would, canceling anything you have done & then did an 'alsactl store' to save..

That's what I was suggesting - just add -F to the alsactl restore in your init script. It won't restore it to a default state - the only difference is that it will do a better job restoring your mixer state if new controls are added by a driver update.

Quoted 6 lines

alsactl --help:

-F,--force try to restore the matching controls as much as
possible

Lee

Great, Lee, thanks. That also gave me a good excuse to expand the operating $PATH for rc.local, recent heyu changes seem to have killed my restoration of the config that existed after cron made the last change in config.

Signature · 13 lines
Cheers, Gene
People having trouble with vz bouncing email to me should add the word
'online' between the 'verizon', and the dot which bypasses vz's
stupid bounce rules.  I do use spamassassin too. :-)
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2006 by Maurice Eugene Heskett, all rights reserved.

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
#17 lore raw
In reply to Lee Revell

At Wed, 09 Aug 2006 12:35:32 -0400,

Quoted 6 lines · Lee Revell
Quoted 4 lines · Andrew Benton
Quoted 2 lines · Lee Revell

Please try to identify the change that introduced the regression. What was the last kernel/ALSA version that worked correctly?

Quoted 7 lines
Quoted 2 lines

The change happened between 2.6.17 and 2.6.18-rc1. Specifically, 2.6.17-git4 works and 2.6.17-git5 doesn't.

Takashi-san,

Does this help at all? Many users are reporting that sound broke with 2.6.18-rc*.

Did you get so many bugreports as 2.6.18 regression? AFAIK, I had two on LKML, and both of them are no serious problem.

Takashi

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642

#18 lore raw
In reply to Andrew Benton

At Tue, 08 Aug 2006 21:28:21 +0100,

Quoted 14 lines · Andrew Benton

Hello World,
I use the optical digital output from my computer and normally it works well. However, with a 2.6.18-rc* kernel I have to use alsamixer to get any output from it. I tweak the levels till it's working and use alsactl store to save it. Normally, when I reboot it's enough to have a bootscript run alsactl restore and it should work, but as I said, with a 2.6.18-rc* kernel it doesn't. Every time I have to use alsamixer to change the <IEC958 P> setting to A/D Conv and then back to AC-Link again to get the diode to come on and start sending a signal to the amplifier. And the problems don't stop there. At first it sounds fine but after a few minutes it starts to develop a harsh, tinny, metallic, distorted quality and it becomes unlistenable.

Quoted 2 lines

According to alsamixer v1.0.11, the card is an Intel ICH5 and the chip is an Analog Devices AD1981B

There is no influential change in intel8x0 and ac97 stuff between 2.6.17 and 2.6.18. If it worked with 2.6.17, it must be something else.

Takashi