Red Hat Bugzilla – Bug 444544
[cpuidle] Frequent lockups
Last modified: 2008-05-02 18:59:17 EDT
Description of problem: System occasionally does a soft-lockup, most frequently
during terminal I/O but occasionally while idle (in screensaver). Network
interface remains up (pingable), and occasionally I can ssh in from other
systems and perform some activities but attempts to "su" always hang. Please
find attached the (partial) output of "dmesg" from one of those logins.
Here's the output of "uname -a":
Linux tremor.dachte.org 2.6.25-8.fc9.x86_64 #1 SMP Wed Apr 23 03:20:41 EDT 2008
x86_64 x86_64 x86_64 GNU/Linux
I am using the Fedora 9 prerelease.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
Just using the system eventually causes the problem to manifest
Soft hang - gui becomes completely nonresponsive, attempts to switch to a VT
also fail. Network interface remains up, sometimes incoming ssh works.
Created attachment 304055 [details]
partial dmesg output
I should mention that this laptop has seen every version of Fedora from 5
onwards and I haven't had this problem before.
Try adding "processor.max_cstate=1" to the kernel boot options.
Let's leave this open for a bit longer, but after adding that yesterday I
haven't yet had another lockup, which is encouraging. I'll add another comment
if it happens again.
This appears to fix the problem. Should I make a note of it that I should always
add this to my system configurations with fedora from now on, or will this be
changed into a default?
this will burn you a lot of power though; it basically disables a lot of power
saving that is going on...
This looks like a dupe of 444059
I'm not sure if that's the case - my laptop is set never to suspend while
plugged in, and this problem happened primarily when my laptop was plugged in.
the suspend comment was a misnomer. It later happened to me also after just
booting up and idling. It looks like we both have the same ATI chipset,
(Turion64 cpu too maybe?) and the backtrace is pretty much identical.
As it so happens, we have the same cpu and chipset, according to lspci and
Unrelated: I wouldn't be surprised if we have to do the same tricks to make the
system basically usable (like disabling the trackpad due to i8042 wonkiness)....
So I guess this indeed a dupe.
*** This bug has been marked as a duplicate of 444059 ***