Bug 469132 – >=sys-kernel/gentoo-sources-3.5 - ALSA snd-powermac I2C Regression on PPC and PPC64.

• Product: Gentoo Linux
• Status: RESOLVED TEST-REQUEST

[Description GrowlTiger 2013-05-09 02:50:29 UTC]

Created attachment 347790 [details]
emerge --info

ALSA snd-powermac I2C Regression in kernel >3.5.0-rc6+ PPC

ALSA module snd-powermac fails to load on recent kernel versions running on various PPC chips. This makes ALSA unusable for those kernels.

From dmesg:

[   10.537194] i2c i2c-2: PMac Keywest Audio: attach_adapter method is deprecated 
[   10.537198] i2c i2c-2: Please use another way to instantiate your i2c_client 
[   10.537212] i2c i2c-2: Failed to register i2c client keywest at 0x35 (-16)

Using git-bisect I ascertained where the bug first appeared.

The bug is not present in kernel versions before 3.5.0-rc6+, and all subsequent kernels have the problem.

Here is the commit info from the bisect:

Bisecting: 0 revisions left to test after this (roughly 0 steps) 
[3a3dd0186f619b74e61e6f29dddcaf59af7d3cac] i2c/powermac: Improve detection of devices from device-tree 
3a3dd0186f619b74e61e6f29dddcaf59af7d3cac is the first bad commit 
commit 3a3dd0186f619b74e61e6f29dddcaf59af7d3cac 
Author: Benjamin Herrenschmidt <[e-mail]> 
Date:   Mon Jun 18 12:00:50 2012 +1000 

    i2c/powermac: Improve detection of devices from device-tree 
    
    This patch adds a number of workarounds for broken Apple device-trees 
    mostly around sound chips. It handles creating the missing audio codec 
    devices and works around various issues with missing addresses or 
    missing compatible properties. 
    
    Signed-off-by: Benjamin Herrenschmidt <[e-mail]> 

:040000 040000 849b9ecd549338cf00f024c15aaa01d5217e1806 90e49b5143a97c43b4f07040b56dc6002ad51f9b M   drivers

I've duplicated this bug on several PPC G4 machines, including:

PowerBook 12 (PowerBook6,1) 12" Aluminum PowerBook G4 867 MHz PowerPC 7455 processor originally released January 7, 2003, model number A1010, order number M8760LL/A

# cat /proc/cpuinfo
processor   : 0 
cpu      : 7455, altivec supported 
clock      : 867.000000MHz 
revision   : 3.3 (pvr 8001 0303) 
bogomips   : 108.20 
timebase   : 33280414 
platform   : PowerMac 
model      : PowerBook6,1 
machine      : PowerBook6,1 
motherboard   : PowerBook6,1 MacRISC3 Power Macintosh 
detected as   : 287 (PowerBook G4 12") 
pmac flags   : 0000001a 
L2 cache   : 256K unified 
pmac-generation   : NewWorld 
Memory      : 1152 MB


iMac G4 (PowerMac6,1) 17" iMac G4 1.25GHz G4 PowerPC 7457 processor originally released Sep 8, 2003, model number M6498, order number M9168LL/A
# cat /proc/cpuinfo
processor	: 0
cpu		: 7455, altivec supported
clock		: 1249.999995MHz
revision	: 3.3 (pvr 8001 0303)
bogomips	: 83.20
timebase	: 41600571
platform	: PowerMac
model		: PowerMac6,1
machine		: PowerMac6,1
motherboard	: PowerMac6,1 MacRISC3 Power Macintosh
detected as	: 287 (Flat panel iMac)
pmac flags	: 00000010
L2 cache	: 256K unified
pmac-generation	: NewWorld
Memory		: 512 MB

Any help would be much appreciated. Thank You.

[Comment 1 GrowlTiger 2013-05-09 03:01:03 UTC]

The kernel tree I'm using is gentoo-sources

[Comment 2 Tom Wijsman (TomWij) (RETIRED) 2013-05-09 04:06:18 UTC]

I have no idea from a first look at the commit, you may be able to try unapplying some of its changes to get an idea of where the fault in the commit lies. See the URL field of this bug for the commit.

[Comment 3 Norayr Chilingarian 2013-05-20 20:11:49 UTC]

I have similar issues on ibook g4 https://bugs.gentoo.org/show_bug.cgi?id=460580

[Comment 4 Tom Wijsman (TomWij) (RETIRED) 2013-07-19 10:07:36 UTC]

Did you try something? Let us know if you need help.

[Comment 5 GrowlTiger 2013-08-11 22:46:17 UTC]

(In reply to Tom Wijsman (TomWij) from comment #4)
> Did you try something? Let us know if you need help.

Hi Tom, yes, I'd like some help. Would you recommend just removing a line or two in a trial-and-error fashion? I don't see anything that reaches out to me and says "bug."

Thank You,
Bill

[Comment 6 Tom Wijsman (TomWij) (RETIRED) 2013-08-12 21:51:36 UTC]

(In reply to GrowlTiger from comment #5)
> Hi Tom, yes, I'd like some help. Would you recommend just removing a line or
> two in a trial-and-error fashion? I don't see anything that reaches out to
> me and says "bug."

It's quite a commit, over 100 lines; I'll try to read over it soon, let's see...

[Comment 7 Tom Wijsman (TomWij) (RETIRED) 2013-10-14 15:39:25 UTC]

Created attachment 360882 [details]
drivers/i2c/busses/i2c-powermac.c commit log

Hmm, sorry, the commit is too big for an out-stander to understand what really is going on and what the possible issue might be.

==> Can you try =sys-kernel/git-sources-3.12-rc5? <==

It contains a commit that might outline the problem better:

commit 874e955b171f0505e2371c1ca80a2188efdc217d
Author: Wolfram Sang <[e-mail]>
Date:   Thu Jul 11 22:02:03 2013 +0100

    i2c: powermac: fix return path on error
    
    We want to bail out immediately if i2c_add_adapter failed and not try to
    register child nodes with a nilled adapter structure.
    
    Signed-off-by: Wolfram Sang <[e-mail]>

And possibly some other fixes (see attachment).

[Comment 8 Tom Wijsman (TomWij) (RETIRED) 2013-10-14 15:40:36 UTC]

(Ignore the second half of that attachment, apparently hit CTRL+V twice)