[SOLVED] Kernel 2.6.30 panic

xyrico · 2009-06-11 22:46:35

I just upgraded to the new kernel in hopes of getting better video performance on my Intel 945, but it refuses to boot. I get the message

Kernel panic - not syncing: stack protector: Kernel stack is corrupted in: f817aa2e

which comes up during 'Loading UDev uevents' while booting. The fallback also gives the same error. Never had anything like this happen before, so I guess I'll have to downgrade for now. I'm still curious as to what could be wrong, though, or if I'm not the only one having this problem. I'm running a [testing] system on an older Acer 9410z laptop, if that's of any help.

Last edited by xyrico (2009-06-13 21:50:25)

anrxc · 2009-06-11 23:35:50

You are probably using acer-wmi? It's broken! I just sent a bug report, and I'm waiting a reply.

Boot from some rescue/live cd, and blacklist acer_wmi in rc.conf, if you still want to use 2.6.30 (not much point, loosing extra keys, rfkill support etc.)

Edit: I have to dissapoint you on the Intel thing too, performance is crap.

Last edited by anrxc (2009-06-11 23:37:18)

xyrico · 2009-06-12 00:10:29

Thanks for the info. Looks like I have good reasons to stick with 2.6.29 for now, then.

Army · 2009-06-12 08:54:31

Just try out the method anrxc mentioned, blacklisting acer_wmi might work. You do this by adding

!acer_wmi

to the modules line in /etc/rc.conf. The ! is important, it means "do not load!!" Then you might be able to stay with 2.6.30

xyrico · 2009-06-12 10:13:10

I thought about doing that, but after checking to see what other modules depended on acer_wmi

lsmod | grep acer_wmi
acer_wmi               15456  0 
wmi                     6248  1 acer_wmi
rfkill                  9456  5 rfkill_input,acer_wmi,b43
led_class               3416  3 acer_wmi,sdhci,b43

I (EDIT: assumed) that blacklisting it would prevent rfkill and, in turn, b43 from loading. Since wireless is the only way I can get a connection on this laptop, I don't see an alternative. Unless I'm missing something or I find an alternative module to enable me to use my wireless card, I'll probably just have to wait it out.

On the other hand, maybe somebody else could use this info to avoid a hassle if the culprit is indeed acer_wmi. Won't hurt to try, anyway, so I'll give it a shot.

Last edited by xyrico (2009-06-12 10:27:55)

bangoskank · 2009-06-12 10:21:48

xyrico wrote:

I thought about doing that, but after checking to see what other modules depended on acer_wmi
lsmod | grep acer_wmi
acer_wmi               15456  0 
wmi                     6248  1 acer_wmi
rfkill                  9456  5 rfkill_input,acer_wmi,b43
led_class               3416  3 acer_wmi,sdhci,b43
I realized that blacklisting it would prevent rfkill and, in turn, b43 from loading. Since wireless is the only way I can get a connection on this laptop, I don't see an alternative. Unless I'm missing something or I find an alternative module to enable me to use my wireless card, I'll probably just have to wait it out.

Actually, these are reverse-dependencies -- note the "Used by" heading when lsmod is run. Removing acer_wmi shouldn't be a problem at all.

xyrico · 2009-06-12 10:36:33

Oh, wow, duh. Really, though, thanks for setting me straight. It's an important distinction and also explains why I'm able to post this while running 2.6.30 without acer_wmi

Blacklisting it seems to have done the trick. In fact, all my function keys and everything else works fine so far. Plus, I gained 2 seconds on the gtkperf test (down to 12s) so at least it's doing slightly better graphics-wise.

Thanks for the help and encouragement

Last edited by xyrico (2009-06-12 10:40:17)

anrxc · 2009-06-12 13:24:55

This is a bug in the wmi driver (which acer-wmi uses), a patch is waiting to go upstream.

combuster · 2009-06-12 14:14:57

There are no problems with dell_wmi on my box... It's x86_64 though...

skottish · 2009-06-13 00:33:28

Army wrote:

Just try out the method anrxc mentioned, blacklisting acer_wmi might work. You do this by adding
!acer_wmi
to the modules line in /etc/rc.conf. The ! is important, it means "do not load!!" Then you might be able to stay with 2.6.30

Army, you rule. I feel deranged and dirty if I don't have testing enabled. Thanks for this.

anrxc · 2009-06-13 12:26:52

Tobias included the patch in the 3rd build of the kernel package. It's fixed now.

Army · 2009-06-13 12:40:31

skottish wrote:

Army, you rule. I feel deranged and dirty if I don't have testing enabled. Thanks for this.

I just described how to handle such a situation.

skottish · 2009-06-13 13:47:09

The 2.6.30-3 release fixes the problem.

*** Thanks Tobias ***

xyrico · 2009-06-13 21:49:49

Works great now, thanks everyone.

Arch Linux

#1 2009-06-11 22:46:35

[SOLVED] Kernel 2.6.30 panic

#2 2009-06-11 23:35:50

Re: [SOLVED] Kernel 2.6.30 panic

#3 2009-06-12 00:10:29

Re: [SOLVED] Kernel 2.6.30 panic

#4 2009-06-12 08:54:31

Re: [SOLVED] Kernel 2.6.30 panic

#5 2009-06-12 10:13:10

Re: [SOLVED] Kernel 2.6.30 panic

#6 2009-06-12 10:21:48

Re: [SOLVED] Kernel 2.6.30 panic

#7 2009-06-12 10:36:33

Re: [SOLVED] Kernel 2.6.30 panic

#8 2009-06-12 13:24:55

Re: [SOLVED] Kernel 2.6.30 panic

#9 2009-06-12 14:14:57

Re: [SOLVED] Kernel 2.6.30 panic

#10 2009-06-13 00:33:28

Re: [SOLVED] Kernel 2.6.30 panic

#11 2009-06-13 12:26:52

Re: [SOLVED] Kernel 2.6.30 panic

#12 2009-06-13 12:40:31

Re: [SOLVED] Kernel 2.6.30 panic

#13 2009-06-13 13:47:09

Re: [SOLVED] Kernel 2.6.30 panic

#14 2009-06-13 21:49:49

Re: [SOLVED] Kernel 2.6.30 panic

Board footer