Skip to contentSkip to conversation

another in kernel alsa update that breaks backward compatibilty?

DISCUSSION Gene Heskett 7 participants 16 messages Aug 9, 2006

lore
Share & tools
lore
#0 lore raw

Greetings;

The old fart is back again. :)

I've just done a divide and conquer on kernel versions, and have found that while I DO have a kde audio signon for kernels 2.6.18-rc1-rc3-rc4, I do not have any other functioning audio, including the kde sound effects I normally get. xmms and tvtime are mute, as are the firefox plugins to play videos from the network. 2.6.17.8 and below works great yet.

So whats the fix?

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.
#1 lore raw
Quoted 4 lines · On Wed, 09 Aug 2006 11:40:02 -0400 Gene Heskett <[email protected]>…

Greetings;

The old fart is back again. :)

Quoted 8 lines

I've just done a divide and conquer on kernel versions, and have found that while I DO have a kde audio signon for kernels 2.6.18-rc1-rc3-rc4, I do not have any other functioning audio, including the kde sound effects I normally get. xmms and tvtime are mute, as are the firefox plugins to play videos from the network. 2.6.17.8 and below works great yet.

So whats the fix?

Demand stable ABI.

Signature · 1 line
Visit my http://appsfromscratch.berlios.de/ open source project.
#2 lore raw
Quoted 13 lines · Sergei Steshenko
Quoted 10 lines · On Wed, 09 Aug 2006 11:40:02 -0400 Gene Heskett <[email protected]>…

Greetings;

The old fart is back again. :)

I've just done a divide and conquer on kernel versions, and have found that while I DO have a kde audio signon for kernels 2.6.18-rc1-rc3-rc4, I do not have any other functioning audio, including the kde sound effects I normally get. xmms and tvtime are mute, as are the firefox plugins to play videos from the network. 2.6.17.8 and below works great yet.

Quoted 6 lines
Quoted 2 lines

So whats the fix?

Demand stable ABI.

Stable ABI has nothing to do with this, it's called "a bug".

Lee

#3 lore raw
In reply to Sergei Steshenko

hi,

Quoted 15 lines · Sergei Steshenko
Quoted 10 lines · On Wed, 09 Aug 2006 11:40:02 -0400 Gene Heskett <[email protected]>…

Greetings;

The old fart is back again. :)

I've just done a divide and conquer on kernel versions, and have found that while I DO have a kde audio signon for kernels 2.6.18-rc1-rc3-rc4, I do not have any other functioning audio, including the kde sound effects I normally get. xmms and tvtime are mute, as are the firefox plugins to play videos from the network. 2.6.17.8 and below works great yet.

Quoted 10 lines
Quoted 4 lines

So whats the fix?

Demand stable ABI.

sorry for the noise, but it's been a while now since i began reading mails from this list, and i must admit i don't always (never?) see the point of such messages...
if you can help me understand, i'll be very happy to get something more detailed from you...

thanks,

-- Ben

#4 lore raw
Quoted 10 lines · Benoit Fouet
Quoted 5 lines

Demand stable ABI.

sorry for the noise, but it's been a while now since i began reading mails from this list, and i must admit i don't always (never?) see the point of such messages...
if you can help me understand, i'll be very happy to get something more detailed from you...

Documentation/stable_api_nonsense.txt

Sam

#5 lore raw
Quoted 6 lines · On Wed, 9 Aug 2006 18:00:43 +0200 Sam Ravnborg <[email protected]> wrote:
Quoted 5 lines · Benoit Fouet
Quoted 5 lines

Demand stable ABI.

Quoted 9 lines
Quoted 5 lines

sorry for the noise, but it's been a while now since i began reading mails from this list, and i must admit i don't always (never?) see the point of such messages...
if you can help me understand, i'll be very happy to get something more detailed from you...

Documentation/stable_api_nonsense.txt

Sam

I love senselessness and technical incompetence of the document.

As I was taught at school, to prove that a statement is wrong one has to prove that it is wrong once.

Regardless of what the document says stable ABI can be achieved today - run a chosen Linux kernel version + chosen ALSA version under XEN or similar, and assign sound card to these (chosen Linux kernel version + chosen ALSA version).

Redirect sound ('ncat' + friends) to this (chosen Linux kernel version + chosen ALSA version) from your kernel in which developers refuse to ensure stable ABI.

Because of the chosen (kernel+ALSA) you have stable ABI regardless of what Documentation/stable_api_nonsense.txt says and ALSA + kernel developers think.

--Sergei.

Signature · 1 line
Visit my http://appsfromscratch.berlios.de/ open source project.
#6 lore raw
Quoted 21 lines · Sergei Steshenko
Quoted 6 lines · On Wed, 9 Aug 2006 18:00:43 +0200 Sam Ravnborg <[email protected]> wrote:
Quoted 5 lines · Benoit Fouet
Quoted 5 lines

Demand stable ABI.

Quoted 9 lines
Quoted 5 lines

sorry for the noise, but it's been a while now since i began reading mails from this list, and i must admit i don't always (never?) see the point of such messages...
if you can help me understand, i'll be very happy to get something more detailed from you...

Documentation/stable_api_nonsense.txt

Sam

I love senselessness and technical incompetence of the document.

Quoted 7 lines

As I was taught at school, to prove that a statement is wrong one has to prove that it is wrong once.

Regardless of what the document says stable ABI can be achieved today - run a chosen Linux kernel version + chosen ALSA version under XEN or similar, and assign sound card to these (chosen Linux kernel version + chosen ALSA version).

Your focus is on user<->kernel API. The document is about kernel internal API. Thats two different things and it was not clear if the first poster was pointing at one or the other.

user<->kernel API is supposed to be stable - if ALSA breaks that then ALSA has something to fix.

Sam

#7 lore raw
In reply to Sergei Steshenko
Quoted 16 lines · Sergei Steshenko
Quoted 13 lines · On Wed, 9 Aug 2006 18:00:43 +0200 Sam Ravnborg <[email protected]> wrote:
Quoted 10 lines · Benoit Fouet
Quoted 5 lines

Demand stable ABI.

sorry for the noise, but it's been a while now since i began reading mails from this list, and i must admit i don't always (never?) see the point of such messages...
if you can help me understand, i'll be very happy to get something more detailed from you...

Documentation/stable_api_nonsense.txt

Quoted 8 lines
Quoted 2 lines

Sam

I love senselessness and technical incompetence of the document.

As I was taught at school, to prove that a statement is wrong one has to prove that it is wrong once.

Yep, the only trick is that you need a valid proof ;)

Quoted 5 lines

Regardless of what the document says stable ABI can be achieved today - run a chosen Linux kernel version + chosen ALSA version under XEN or similar, and assign sound card to these (chosen Linux kernel version + chosen ALSA version).

Quoted 8 lines

Redirect sound ('ncat' + friends) to this (chosen Linux kernel version + chosen ALSA version) from your kernel in which developers refuse to ensure stable ABI.

Because of the chosen (kernel+ALSA) you have stable ABI regardless of what Documentation/stable_api_nonsense.txt says and ALSA + kernel developers think.

You are confused. By your logic you do not need XEN at all - just take a kernel version + alsa and never change/update it - and viola! "stable" ABI.

Signature · 1 line
Dmitry
#8 lore raw
Quoted 37 lines · On Wed, 9 Aug 2006 12:36:23 -0400 "Dmitry Torokhov"…
Quoted 16 lines · Sergei Steshenko
Quoted 13 lines · On Wed, 9 Aug 2006 18:00:43 +0200 Sam Ravnborg <[email protected]> wrote:
Quoted 10 lines · Benoit Fouet
Quoted 5 lines

Demand stable ABI.

sorry for the noise, but it's been a while now since i began reading mails from this list, and i must admit i don't always (never?) see the point of such messages...
if you can help me understand, i'll be very happy to get something more detailed from you...

Documentation/stable_api_nonsense.txt

Quoted 8 lines
Quoted 2 lines

Sam

I love senselessness and technical incompetence of the document.

As I was taught at school, to prove that a statement is wrong one has to prove that it is wrong once.

Yep, the only trick is that you need a valid proof ;)

Quoted 9 lines

Regardless of what the document says stable ABI can be achieved today - run a chosen Linux kernel version + chosen ALSA version under XEN or similar, and assign sound card to these (chosen Linux kernel version + chosen ALSA version).

Redirect sound ('ncat' + friends) to this (chosen Linux kernel version + chosen ALSA version) from your kernel in which developers refuse to ensure stable ABI.

Quoted 9 lines
Quoted 4 lines

Because of the chosen (kernel+ALSA) you have stable ABI regardless of what Documentation/stable_api_nonsense.txt says and ALSA + kernel developers think.

You are confused. By your logic you do not need XEN at all - just take a kernel version + alsa and never change/update it - and viola! "stable" ABI.

I simply described how one ABI (ALSA <-> kernel in this case) can be stabilized, while new non-ALSA related features (and potentially unstable ABI) can still be had.

If computer has enough resources, practically every ABI can be stabilized (if desired) this way - as long as the ABI is PCI slot related.

That is, I can, for example, stabilize ALSA-kernel interface choosing (ALSA 1.0.11 + kernel 2.6.17) and I can stabilize TV card interface using (whatever v4l + kernel 2.6.18), etc,

--Sergei.

Signature · 1 line
Visit my http://appsfromscratch.berlios.de/ open source project.
#9 lore raw
Quoted 8 lines · Sergei Steshenko
Quoted 5 lines · On Wed, 9 Aug 2006 12:36:23 -0400 "Dmitry Torokhov"…

You are confused. By your logic you do not need XEN at all - just take a kernel version + alsa and never change/update it - and viola! "stable" ABI.

Quoted 8 lines

I simply described how one ABI (ALSA <-> kernel in this case) can be stabilized, while new non-ALSA related features (and potentially unstable ABI) can still be had.

If computer has enough resources, practically every ABI can be stabilized (if desired) this way - as long as the ABI is PCI slot related.

And in extreme case once you "stablizie" everything you end up with a system that is not upgradeable at all.

Quoted 4 lines

That is, I can, for example, stabilize ALSA-kernel interface choosing (ALSA 1.0.11 + kernel 2.6.17) and I can stabilize TV card interface using (whatever v4l + kernel 2.6.18), etc,

But you are not stabilizing ABI, you are freezing a subsystem. Stable ABI does not mean that bugs do not get fixed and new hardware support is not being addeed, as in your case.

Signature · 1 line
Dmitry
#10 lore raw
Quoted 21 lines · On Wed, 9 Aug 2006 13:54:29 -0400 "Dmitry Torokhov"…
Quoted 8 lines · Sergei Steshenko
Quoted 5 lines · On Wed, 9 Aug 2006 12:36:23 -0400 "Dmitry Torokhov"…

You are confused. By your logic you do not need XEN at all - just take a kernel version + alsa and never change/update it - and viola! "stable" ABI.

Quoted 8 lines

I simply described how one ABI (ALSA <-> kernel in this case) can be stabilized, while new non-ALSA related features (and potentially unstable ABI) can still be had.

If computer has enough resources, practically every ABI can be stabilized (if desired) this way - as long as the ABI is PCI slot related.

And in extreme case once you "stablizie" everything you end up with a system that is not upgradeable at all.

Quoted 9 lines
Quoted 4 lines

That is, I can, for example, stabilize ALSA-kernel interface choosing (ALSA 1.0.11 + kernel 2.6.17) and I can stabilize TV card interface using (whatever v4l + kernel 2.6.18), etc,

But you are not stabilizing ABI, you are freezing a subsystem. Stable ABI does not mean that bugs do not get fixed and new hardware support is not being addeed, as in your case.

I did stabilize ABI - I can be using the same (bit to bit) audio driver regardless of changes in the kernel not related to ALSA.

I can consider this whole ugly and clumsy construct as a "super" kernel in which by construction nothing changes in the audio part.

That, I as end user don't care what developers break in the non-audio part of this kernel - for me audio part is stable.

I let the non-audio part evolve while the audio part remains the same - even at binary level.

--Sergei.

Signature · 1 line
Visit my http://appsfromscratch.berlios.de/ open source project.
#11 lore raw
In reply to Sergei Steshenko
Quoted 14 lines · Sergei Steshenko

On Wed, 09 Aug 2006 11:40:02 -0400

Quoted 11 lines · Gene Heskett

Greetings;

The old fart is back again. :)

I've just done a divide and conquer on kernel versions, and have found that while I DO have a kde audio signon for kernels 2.6.18-rc1-rc3-rc4, I do not have any other functioning audio, including the kde sound effects I normally get. xmms and tvtime are mute, as are the firefox plugins to play videos from the network. 2.6.17.8 and below works great yet.

Quoted 3 lines
Quoted 1 line

So whats the fix?

Demand stable ABI.

It does not appear to be so. And ATM booted to 18-rc1, I didn't see an error message when rc.local made a call of "[root@coyote gene]# alsactl restore
alsactl: set_control:894: warning: name mismatch (Mic Boost (+20dB)/Mic Boost (+20dB) Switch) for control #45
alsactl: set_control:896: warning: index mismatch (0/0) for control #45
alsactl: set_control:898: failed to obtain info for control #45 (Operation not permitted)
[root@coyote gene]#

But as you can see, the error was there nontheless. I've seen this or a very similar error for .18-rc3 and .18-rc4.

This walks and qwacks like the alsa interface has been diddled, again. But since it KNOWS what hardware its running, in this case an audigy 2, not Value, so why was apparently working code broken and then commited to the kernel tree?

Humm, because of my use of 2 independant audio channels here, I'm forced to use kamix in order to address both systems. And I found it! Somehow, that card managed to get its "external amplifier" 'led' turned on, which apparently kills the normal line outs that drive my speakers. The display also looks slightly busier, like another slider has been added?

Now to check the newer kernels again after doing an 'alsactl store' from the cli. Once I've done that, the above error is not repeated.

And, even kookier, is that after doing the alsactl store, the external amplifier button no longer effects it. Anybody got any migrain medicine?

And tvtime now has a voice too, goodie.. :)

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.
#12 lore raw
Quoted 4 lines · Gene Heskett

This walks and qwacks like the alsa interface has been diddled, again. But since it KNOWS what hardware its running, in this case an audigy 2, not Value, so why was apparently working code broken and then commited to the kernel tree?

It's a bug. The interface was not changed intentionally. ALSA policy for at least a year has been "no incompatible changes".

Lee

#13 lore raw
In reply to Gene Heskett

At Wed, 09 Aug 2006 12:07:26 -0400,

Quoted 20 lines · Gene Heskett
Quoted 14 lines · Sergei Steshenko

On Wed, 09 Aug 2006 11:40:02 -0400

Quoted 11 lines · Gene Heskett

Greetings;

The old fart is back again. :)

I've just done a divide and conquer on kernel versions, and have found that while I DO have a kde audio signon for kernels 2.6.18-rc1-rc3-rc4, I do not have any other functioning audio, including the kde sound effects I normally get. xmms and tvtime are mute, as are the firefox plugins to play videos from the network. 2.6.17.8 and below works great yet.

Quoted 3 lines
Quoted 1 line

So whats the fix?

Demand stable ABI.

Quoted 9 lines

It does not appear to be so. And ATM booted to 18-rc1, I didn't see an error message when rc.local made a call of "[root@coyote gene]# alsactl restore
alsactl: set_control:894: warning: name mismatch (Mic Boost (+20dB)/Mic Boost (+20dB) Switch) for control #45
alsactl: set_control:896: warning: index mismatch (0/0) for control #45
alsactl: set_control:898: failed to obtain info for control #45 (Operation not permitted)
[root@coyote gene]#

This kind of problems can be fixed by calling "alsactl restore" with -F option. The warnings may be still there but the values are handled better.

The behavior with -F could be default, but it's kept so just for the compatibility reason.

Takashi

#14 lore raw
Quoted 32 lines · Takashi Iwai

At Wed, 09 Aug 2006 12:07:26 -0400,

Quoted 19 lines · Gene Heskett
Quoted 14 lines · Sergei Steshenko

On Wed, 09 Aug 2006 11:40:02 -0400

Quoted 11 lines · Gene Heskett

Greetings;

The old fart is back again. :)

I've just done a divide and conquer on kernel versions, and have found that while I DO have a kde audio signon for kernels 2.6.18-rc1-rc3-rc4, I do not have any other functioning audio, including the kde sound effects I normally get. xmms and tvtime are mute, as are the firefox plugins to play videos from the network. 2.6.17.8 and below works great yet.

Quoted 3 lines
Quoted 1 line

So whats the fix?

Demand stable ABI.

Quoted 9 lines

It does not appear to be so. And ATM booted to 18-rc1, I didn't see an error message when rc.local made a call of "[root@coyote gene]# alsactl restore
alsactl: set_control:894: warning: name mismatch (Mic Boost (+20dB)/Mic Boost (+20dB) Switch) for control #45
alsactl: set_control:896: warning: index mismatch (0/0) for control #45
alsactl: set_control:898: failed to obtain info for control #45 (Operation not permitted)
[root@coyote gene]#

Quoted 8 lines

This kind of problems can be fixed by calling "alsactl restore" with -F option. The warnings may be still there but the values are handled better.

The behavior with -F could be default, but it's kept so just for the compatibility reason.

Humm, and that fixed it for 2.6.18-rc4. Gawd, I'd sure like to get some of that stuff, it must be great stuff. Goes off scratching head & letting imagination work overtime.

Quoted 1 line

Takashi

Many thanks Takashi.

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.
#15 lore raw
In reply to Gene Heskett
Quoted 44 lines · Gene Heskett
Quoted 14 lines · Sergei Steshenko

On Wed, 09 Aug 2006 11:40:02 -0400

Quoted 11 lines · Gene Heskett

Greetings;

The old fart is back again. :)

I've just done a divide and conquer on kernel versions, and have found that while I DO have a kde audio signon for kernels 2.6.18-rc1-rc3-rc4, I do not have any other functioning audio, including the kde sound effects I normally get. xmms and tvtime are mute, as are the firefox plugins to play videos from the network. 2.6.17.8 and below works great yet.

Quoted 3 lines
Quoted 1 line

So whats the fix?

Demand stable ABI.

It does not appear to be so. And ATM booted to 18-rc1, I didn't see an error message when rc.local made a call of "[root@coyote gene]# alsactl restore
alsactl: set_control:894: warning: name mismatch (Mic Boost (+20dB)/Mic Boost (+20dB) Switch) for control #45
alsactl: set_control:896: warning: index mismatch (0/0) for control #45
alsactl: set_control:898: failed to obtain info for control #45 (Operation not permitted)
[root@coyote gene]#

But as you can see, the error was there nontheless. I've seen this or a very similar error for .18-rc3 and .18-rc4.

This walks and qwacks like the alsa interface has been diddled, again. But since it KNOWS what hardware its running, in this case an audigy 2, not Value, so why was apparently working code broken and then commited to the kernel tree?

Humm, because of my use of 2 independant audio channels here, I'm forced to use kamix in order to address both systems. And I found it! Somehow, that card managed to get its "external amplifier" 'led' turned on, which apparently kills the normal line outs that drive my speakers. The display also looks slightly busier, like another slider has been added?

Quoted 7 lines

Now to check the newer kernels again after doing an 'alsactl store' from the cli. Once I've done that, the above error is not repeated.

And, even kookier, is that after doing the alsactl store, the external amplifier button no longer effects it. Anybody got any migrain medicine?

And tvtime now has a voice too, goodie.. :)

But, once booted to 2.6.18-rc4, the audio except for the kde signon splash, is well and truely muted for all other usage.

I haven't built -rc2 yet, I'll try that next. Alsamixer cannot coax a peep out of it now.

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.