The module snd-pcsp makes the machine unusable because it makes it freeze after pressing a backspace,tab, or anything that would cause a system beep. After trying to understand why using acpi=off on kernel line it would work normally, i decided to blacklist this module along with pcspkr and it solved the problem. I don't see a reason why this would interact with acpi, but nevertheless solving my problem (better be without beeps than without acpi on laptop). This is a problem that was affecting me quite some time ago (i used gentoo and had my kernel compiled without it, and when decided to change most livecd's would crash right after boot). (just a note : The kernel doesn't panic and no message is displayed, just freezes ). The sound driver i'm using is intel-hda...
how about boot option pci=noacpi?
Booting with pci=noacpi still leads to a kernel freeze.
Will you please try the following boot options? a. idle=poll b. processor.max_cstate=1 (the processor should be compiled as built-in kernel modules) c. nolapic_timer Thanks.
a. still leads to crash. b. have to make the processor built-in so didn't tried it yet. c. The machine loads the module and doesn't freeze. Output of dmesg: [ 308.214884] PC-Speaker initialization failed. [ 308.242874] pcspkr: probe of pcspkr failed with error -5 [ 312.823307] input: PC Speaker as /class/input/input11 [ 312.850815] PCSP: Timer resolution is not sufficient (4000250nS) [ 312.850820] PCSP: Make sure you have HPET and ACPI enabled. [ 312.850826] PC-Speaker initialization failed. [ 312.874821] pcspkr: probe of pcspkr failed with error -5 I'm currently using Debian Lenny (Linux unknown 2.6.26-2-686 #1 SMP Thu May 28 15:39:35 UTC 2009 i686 GNU/Linux) . I can compile a kernel with any proposed changes you may wish to try. Just the processor for now? Or more proposed changes cross your mind?
> This is a problem that was affecting me quite some time ago clearing regression flag, since it is unclear if there is an old version of linux w/o this problem. > 2.6.26-2-686 Can you reproduce this problem using an upstream kernel, eg 2.6.30? If no, then this should be a bug against debian rather than the upstream kernel. what do you see when you run: grep . /sys/devices/system/clocksource/*/* and cat /proc/timer_list to show us what timers are available on the system?
Since i had used always a compiled kernel (without the module) , i didn't noticed from when it started so i can't precise, only when i had to get a livecd to run on machine it would hang (archlinux,debian,etc) so i can assume it was a problem with stock kernel. I'd happily suggest major distro's to either blacklist this module or change kernel versions in their livecd's, so less people run into this problem and sucessfully install a distro without an hack. So good work with 2.6.30, it simply worked out of box.
> with 2.6.30, it simply worked out of box. okay thanks, lets close this bug then.