Boot failure with 2.6.21 with snd-maestro3
When snd-maestro3 (ESS Maestro 3 driver) is enabled, kernel 2.6.21 fails to boot - it seems to stop at various points depending on what other modules are enabled. Disabling snd-maestro3 fixes all issues. Moreover, with everything else being same, using maestro3.c from 2.6.20 kernel and using it in 2.6.21 kernel also works - i.e., system boots fine. I will try to isolate the issue and follow up.
Giri
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
Quoted 6 lines · Giridhar Pemmasani
When snd-maestro3 (ESS Maestro 3 driver) is enabled, kernel 2.6.21 fails to boot - it seems to stop at various points depending on what other modules are enabled. Disabling snd-maestro3 fixes all issues. Moreover, with everything else being same, using
maestro3.cfrom 2.6.20 kernel and using it in 2.6.21 kernel also works - i.e., system boots fine. I will try to isolate the issue and follow up.
Thanks for your report.
There were only the two attached patches between 2.6.20 and 2.6.21 to this driver:
- maestro3: add request_firmware()
- maestro3 - Use quirk list helper function
2.6.21 + 2.6.20 maestro3.c works for you.
Does 2.6.21 + 2.6.20 maestro3.c + patch-maestro3-1 work?
Quoted 1 line
Giri
cu
Adrian
Signature · 4 lines
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seedcommit 81d7724a8ee84693befbd60d730199ffb3988f29
Author: Clemens Ladisch <[email protected]> Date: Mon Nov 6 09:26:41 2006 +0100
[ALSA] maestro3: add request_firmware()
Load the ASSP codes using request_firmware(), if possible, instead of using the built-in blobs.
sound/pci/Kconfig +1 -0
@@ -587,6 +587,7 @@ config SND_KORG1212 config SND_MAESTRO3 tristate "ESS Allegro/Maestro3" depends on SND + select FW_LOADER select SND_AC97_CODEC help Say Y here to include support for soundcards based on ESS Maestro 3
sound/pci/maestro3.c +73 -6
141 lines not on this page · raw message
commit 1061eeb44493176eb1d12b47d619e61c428c4395
Author: Takashi Iwai <[email protected]> Date: Fri Nov 24 15:36:46 2006 +0100
[ALSA] maestro3 - Use quirk list helper function
Clean up maestro3 amp and GPIO quirks using snd_pci_quirk_lookup().
sound/pci/maestro3.c +127 -167
361 lines not on this page · raw message
Hi,
Quoted 6 lines · Giridhar Pemmasani
When snd-maestro3 (ESS Maestro 3 driver) is enabled, kernel 2.6.21 fails to boot - it seems to stop at various points depending on what other modules are enabled. Disabling snd-maestro3 fixes all issues. Moreover, with everything else being same, using
maestro3.cfrom 2.6.20 kernel and using it in 2.6.21 kernel also works - i.e., system boots fine. I will try to isolate the issue and follow up.
Please revert this patches
http://www.stardust.webpages.pl/files/patches/alsa_revert/
and test kernel after each revert.
Takashi, I'll add this to known regressions for -stable team.
Quoted 2 lines
Giri
Regards,
Michal
Signature · 3 lines
Michal K. K. Piotrowski Kernel Monkeys (http://kernel.wikidot.com/start)
Quoted 13 lines · --- Adrian Bunk
Quoted 1 line · Giridhar Pemmasani
When snd-maestro3 (ESS Maestro 3 driver) is enabled, kernel 2.6.21 fails
to
Quoted 1 line
boot - it seems to stop at various points depending on what other modules
are
Quoted 1 line
enabled. Disabling snd-maestro3 fixes all issues. Moreover, with
everything
Quoted 1 line
else being same, using
maestro3.cfrom 2.6.20 kernel and using it in2.6.21
Quoted 1 line
kernel also works - i.e., system boots fine. I will try to isolate the
issue
Quoted 1 line
and follow up.
Quoted 8 lines
There were only the two attached patches between 2.6.20 and 2.6.21 to this driver:
- maestro3: addrequest_firmware()
- maestro3 - Use quirk list helper function2.6.21 + 2.6.20
maestro3.cworks for you.Does 2.6.21 + 2.6.20
maestro3.c+ patch-maestro3-1 work?
2.6.21 with 2.6.20's maestro3.c with patch-maestro3-1 exhibited the same issue. However, I identified the cause: maestro3 driver in 2.6.21 (even just patch-maestro3-1) tries to load firmware file from user space. But each time request_firmware is called it takes about 1 minute to timeout (because I don't have the firmware installed?). If I let the system take its time, even with maestro3.c from 2.6.21, it boots after timing out twice. So after about 2 minutes of no activity, system does boot. Following are trace messages I added to identify the issue:
[ 17.004000] input: TPPS/2 IBM TrackPoint as /class/input/input5 [ 70.904000] sound/pci/maestro3.c (snd_m3_create:2763): -2 [ 70.920000] sound/pci/maestro3.c (snd_m3_create:2774): -2 [ 130.940000] sound/pci/maestro3.c (snd_m3_create:2777): -2 [ 130.956000] sound/pci/maestro3.c (snd_m3_create:2788): -2 [ 130.972000] sound/pci/maestro3.c (snd_m3_create:2794): 0 [ 130.992000] sound/pci/maestro3.c (snd_m3_create:2797): 0 [ 131.012000] sound/pci/maestro3.c (snd_m3_create:2802): 0 [ 131.044000] sound/pci/maestro3.c (snd_m3_create:2804): 0 [ 131.580000] sound/pci/maestro3.c (snd_m3_create:2807): 0 [ 131.596000] sound/pci/maestro3.c (snd_m3_create:2810): 0 [ 131.612000] sound/pci/maestro3.c (snd_m3_create:2814): 0 [ 131.628000] sound/pci/maestro3.c (snd_m3_create:2821): 0 [ 132.364000] device-mapper: ioctl: 4.11.0-ioctl (2006-10-12) initialised:
Giri
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
At Sun, 29 Apr 2007 14:09:29 -0700 (PDT),
Quoted 23 lines · Giridhar Pemmasani
Quoted 20 lines · --- Adrian Bunk
Quoted 1 line · Giridhar Pemmasani
When snd-maestro3 (ESS Maestro 3 driver) is enabled, kernel 2.6.21 fails
to
Quoted 1 line
boot - it seems to stop at various points depending on what other modules
are
Quoted 1 line
enabled. Disabling snd-maestro3 fixes all issues. Moreover, with
everything
Quoted 1 line
else being same, using
maestro3.cfrom 2.6.20 kernel and using it in2.6.21
Quoted 1 line
kernel also works - i.e., system boots fine. I will try to isolate the
issue
Quoted 1 line
and follow up.
There were only the two attached patches between 2.6.20 and 2.6.21 to this driver:
- maestro3: addrequest_firmware()
- maestro3 - Use quirk list helper function2.6.21 + 2.6.20
maestro3.cworks for you.
Quoted 7 lines
Quoted 1 line
Does 2.6.21 + 2.6.20
maestro3.c+ patch-maestro3-1 work?2.6.21 with 2.6.20's
maestro3.cwith patch-maestro3-1 exhibited the same issue. However, I identified the cause: maestro3 driver in 2.6.21 (even just patch-maestro3-1) tries to load firmware file from user space. But each timerequest_firmwareis called it takes about 1 minute to timeout (because I don't have the firmware installed?).
Yes, some drivers (e.g. maestro3, ymfpci) require now external firmwares. Try to install them from the latest alsa-firmware package.
I already added new Kconfigs to choose whether external or in-kernel firmware for 2.6.22. Anything else should be as it was. (The replacement with quirk list helper doesn't change the functionality.)
Takashi