Fw: PATCH: creating devices for multiple sound cards

Fw: PATCH: creating devices for multiple sound cards

[Nerijus Baliunas]

Hello,

Why this patch was not applied (I checked 2.4.7)?

------ Forwarded message ------
From: Chris Rankin <rankinc@pacbell.net>
Date: Sat, 14 Apr 2001 09:44:35 +0200
Subject: PATCH: creating devices for multiple sound cards
To: linux-sound@vger.kernel.org

Hi,
I have just installed a new sound card in my machine, and have left
the built-in sound chip enabled despite the motherboard manufacturer's
warning of Bad Things. The result is that I now have two sound devices
in my machine - an ICH (i810_audio) and a Soundblaster Live! The Live!
is device 0, of course. Anyway, I am using the ALSA drivers because
the native i810_audio module caused a kernel panic (NULL pointer
reference in an interrupt handler, I believe), and I have noticed that
the ALSA drivers have difficulty creating OSS device nodes for both
cards. I have traced the problem to the register_sound_special()
function within Linux's soundcore.o module. Basically, ALSA uses this
function to create OSS devices, but this function was never written
with more than one sound device in mind. Until now.

A few points about this patch:

1. I have renamed "sequencer2" to "music" because of a potential naming
conflict between devices 14,8 ("music" on card 0) and 14,33 ("sequencer"
on card 2).

2. I have renamed device "midi00" to "midi". "midi01" now becomes "midi1"
as well, although I would also point out that this is now consistent with
the behaviour of the register_sound_midi() function.

Having applied this patch, ALSA gives me two mixer devices in /dev/sound,
and I no longer get messages like these in my logs:

Apr 12 01:36:54 (none) kernel: devfs: devfs_register(): device already registered: "unknown"

Apr 12 01:40:04 (none) kernel: devfs: devfs_register(): device already registered: "mixer"
Apr 12 01:40:04 (none) kernel: devfs: devfs_register(): device already registered: "dsp"

Cheers,
Chris

--- linux-2.4.3/drivers/sound/sound_core.c.orig        Fri Mar 16 23:00:44 2001
+++ linux-2.4.3/drivers/sound/sound_core.c        Sat Apr 14 00:13:18 2001
@@ -227,9 +227,10 @@
  
 int register_sound_special(struct file_operations *fops, int unit)
 {
-        char *name;
+        const char *name;
+        const int minor = (unit & 15);

-        switch (unit) {
+        switch (minor) {
             case 0:
                 name = "mixer";
                 break;
@@ -237,7 +238,7 @@
                 name = "sequencer";
                 break;
             case 2:
-                name = "midi00";
+                name = "midi";
                 break;
             case 3:
                 name = "dsp";
@@ -255,7 +256,7 @@
                 name = "unknown7";
                 break;
             case 8:
-                name = "sequencer2";
+                name = "music";
                 break;
             case 9:
                 name = "dmmidi";
@@ -279,7 +280,7 @@
                 name = "unknown";
                 break;
         }
-        return sound_insert_unit(&chains[unit&15], fops, -1, unit, unit+1,
+        return sound_insert_unit(&chains[minor], fops, -1, minor, minor+128,
                                  name, S_IRUSR | S_IWUSR);
 }
  

To unsubscribe from this list: send the line "unsubscribe linux-sound" in
the body of a message to majordomo@vger.kernel.org

-------- End of message -------

-
To unsubscribe from this list: send the line "unsubscribe linux-sound" in
the body of a message to majordomo@vger.kernel.org

Fw: PATCH: creating devices for multiple sound cards

[Nerijus Baliunas]

Hello,

Why this patch was not applied (I checked 2.4.7)?

------ Forwarded message ------
From: Chris Rankin <rankinc@pacbell.net>
Date: Sat, 14 Apr 2001 09:44:35 +0200
Subject: PATCH: creating devices for multiple sound cards
To: linux-sound@vger.kernel.org

Hi,
I have just installed a new sound card in my machine, and have left
the built-in sound chip enabled despite the motherboard manufacturer's
warning of Bad Things. The result is that I now have two sound devices
in my machine - an ICH (i810_audio) and a Soundblaster Live! The Live!
is device 0, of course. Anyway, I am using the ALSA drivers because
the native i810_audio module caused a kernel panic (NULL pointer
reference in an interrupt handler, I believe), and I have noticed that
the ALSA drivers have difficulty creating OSS device nodes for both
cards. I have traced the problem to the register_sound_special()
function within Linux's soundcore.o module. Basically, ALSA uses this
function to create OSS devices, but this function was never written
with more than one sound device in mind. Until now.

A few points about this patch:

1. I have renamed "sequencer2" to "music" because of a potential naming
conflict between devices 14,8 ("music" on card 0) and 14,33 ("sequencer"
on card 2).

2. I have renamed device "midi00" to "midi". "midi01" now becomes "midi1"
as well, although I would also point out that this is now consistent with
the behaviour of the register_sound_midi() function.

Having applied this patch, ALSA gives me two mixer devices in /dev/sound,
and I no longer get messages like these in my logs:

Apr 12 01:36:54 (none) kernel: devfs: devfs_register(): device already registered: "unknown"

Apr 12 01:40:04 (none) kernel: devfs: devfs_register(): device already registered: "mixer"
Apr 12 01:40:04 (none) kernel: devfs: devfs_register(): device already registered: "dsp"

Cheers,
Chris

--- linux-2.4.3/drivers/sound/sound_core.c.orig        Fri Mar 16 23:00:44 2001
+++ linux-2.4.3/drivers/sound/sound_core.c        Sat Apr 14 00:13:18 2001
@@ -227,9 +227,10 @@
  
 int register_sound_special(struct file_operations *fops, int unit)
 {
-        char *name;
+        const char *name;
+        const int minor = (unit & 15);

-        switch (unit) {
+        switch (minor) {
             case 0:
                 name = "mixer";
                 break;
@@ -237,7 +238,7 @@
                 name = "sequencer";
                 break;
             case 2:
-                name = "midi00";
+                name = "midi";
                 break;
             case 3:
                 name = "dsp";
@@ -255,7 +256,7 @@
                 name = "unknown7";
                 break;
             case 8:
-                name = "sequencer2";
+                name = "music";
                 break;
             case 9:
                 name = "dmmidi";
@@ -279,7 +280,7 @@
                 name = "unknown";
                 break;
         }
-        return sound_insert_unit(&chains[unit&15], fops, -1, unit, unit+1,
+        return sound_insert_unit(&chains[minor], fops, -1, minor, minor+128,
                                  name, S_IRUSR | S_IWUSR);
 }
  

To unsubscribe from this list: send the line "unsubscribe linux-sound" in
the body of a message to majordomo@vger.kernel.org

-------- End of message -------

Re: Fw: PATCH: creating devices for multiple sound cards

[Chris Rankin]

Hi,

I saw your email on linux-kernel; actually I'm glad
that patch *wasn't* applied because I think it was
flawed. I have recently posted a newer version to
linux-sound and sound-list@redhat.com in the hope that
the 2.4 soundcore maintainer would pick it up.

Who *is* the soundcore maintainer for 2.4? The
MAINTAINERS file says that Alan Cox is the 2.2
maintainer only.

I'll email the latest patch to linux-kernel as well,
since linux-sound appears to be dead these days.

Cheers,
Chris


__________________________________________________
Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
http://phonecard.yahoo.com/

Re: Fw: PATCH: creating devices for multiple sound cards

[Bill Pringlemeir]

>>>>> "Zach" == Zach Brown <zab@zabbo.net> writes:
[snip]
 Zach> doing many other things.  Hopefully the kernel side of ALSA
 Zach> will be acceptable for inclusion in 2.5.  At least it has an
 Zach> army of people actively maintaining it.

If ALSA will replace OSS, then does it make it somewhat futile to add
things to the current set of sound drivers?  I was going to look at
the SBLive driver.  I have been side tracked by an xterm bug; it seems
to have bad handling of utmp on Linux.

regards,
Bill Pringlemeir.

Re: Fw: PATCH: creating devices for multiple sound cards

[Alan Cox]

> If ALSA will replace OSS, then does it make it somewhat futile to add
> things to the current set of sound drivers?  I was going to look at
> the SBLive driver.  I have been side tracked by an xterm bug; it seems
> to have bad handling of utmp on Linux.

Most folks I know are using utmpter for that - it avoids xterm being setuid
setgid or anything else that you really dont want to make it.

As to the sblive - well its probably 2 years away from 2.6 and much of the 
work is being shared - the same bugs turn up in both ALSA and OSS quite
often and several OSS drivers are based off ALSA ones

Re: Fw: PATCH: creating devices for multiple sound cards

[Chris Rankin]

ALSA won't replace OSS overnight, and so a small
amount of OSS-maintenance can't possibly hurt. For
example, I use the OSS version of the emu10k1 driver
for DVD sound because the OSS one is realtime-capable
(and so provides better synchronisation) but the ALSA
one isn't yet.

Besides, the API I've tried to fix is actually *used*
by ALSA to provide OSS-emulation devices, except that
it doesn't work on a devfs-based machine with multiple
sound cards because all the device names clash.

Cheers,
Chris

--- Bill Pringlemeir <bpringle@sympatico.ca> wrote:
> >>>>> "Zach" == Zach Brown <zab@zabbo.net> writes:
> [snip]
>  Zach> doing many other things.  Hopefully the
> kernel side of ALSA
>  Zach> will be acceptable for inclusion in 2.5.  At
> least it has an
>  Zach> army of people actively maintaining it.
> 
> If ALSA will replace OSS, then does it make it
> somewhat futile to add
> things to the current set of sound drivers?  I was
> going to look at
> the SBLive driver.  I have been side tracked by an
> xterm bug; it seems
> to have bad handling of utmp on Linux.
> 
> regards,
> Bill Pringlemeir.
> 


__________________________________________________
Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
http://phonecard.yahoo.com/

Re: Fw: PATCH: creating devices for multiple sound cards

[Zach Brown]

> If ALSA will replace OSS, then does it make it somewhat futile to add
> things to the current set of sound drivers?  

Not at all.  As Alan said, its going to be quite a while before people
are seriously using 2.[56].  Until then its OSS for the majority of
users.

- z