Random System Freeze
 

I'm running Fedora 12 with GNOME on an old Toshiba Satellite A15. From time to time I get a system freeze, where everything stops working besides the touch pad (I can move the cursor, just can't click on anything). When I boot up, I see an error message that it skips past before I can read anything, although I have spotted the snippets 'pci' 'error' and what appears to be the 0x at the start of a memory address. When I try and shut down the system (going either by command line or the System->Shut Down part), I get to the part where it says "Unmounting filesystems..." and then it stops right there. When I boot it back up, I get an error that says "A crash in package kernel has been detected." The same happens when I boot up after the hard freeze. It seems more frequent when I have a heavy load on system memory, but for the most part it seems random. I've seen it crash after 5 minutes with only Pidgin or gedit open, I've seen it run for weeks with over 100 Firefox tabs, Pidgin, 3 documents, and GIMP running, but it seems more likely in the second scenario. It also seems to have a chance of freezing again within 5 minutes of rebooting (even if I don't touch anything after login). I ran Fedora 8 and Fedora 10 on this computer just fine, with most of the same system settings (copied my /home folder for each upgrade). I have a dual-boot system (XP/Fedora), with a third partition for data and a 200MB partition for the boot loader. The partition with Fedora is encrypted, and the crashing while trying to unmount the filesystem implies to me that might be the problem, but whatever it is, it wasn't in Fedora 8 or 10 (at least as far as I could tell).

Luckily, Linux will tell you the problem outright!

Open up a Terminal and type in "dmesg" and post the contents here. (It is long, but the more the merrier!)

Also, a log file at '/var/log/Xorg.0.log' may assist in diagnosing the issue, if its X that going crazy. Please post this one as an attachment because it is a monster.

I decided to attach both, as the output of dmesg filled up beyond what my terminal window would hold, and a quick glance at the document put it at pretty close to twice the size of the Xorg log file you asked for, and the file size is over 10KB larger. This seems to have taken me from annoyed with my computer, to somewhat worried about my computer. I just realized I forgot to mention the wireless card in the PCMCIA slot. It's an old D-Link AirPlusG, model series DWL-G630. Having looked through the dmesg output, my guess is the problem is somewhere near line 600, as some of the stuff in that area looks similar to what I'm seeing on boot-up.

Now, let's just hope I added the attachments right...

Well, I have had a lot to do today, but I took a look at your files finally, and while I can not be 100% in my diagnosis, I hope that this will suffice.

I noticed in your Xorg that the kernel and X WTF's trying to figure out your display resolution. We can tell X what your resolution is; via the file located at '/etc/X11/Xorg.conf'.
Append this somewhere in the file (doesn't really matter too much), with your correct resolution. If the display option is already there append it under the 24 header.

(Quoting the Arch Linux Wiki :))
The kernel also just did an FSCK of the FS, and repaired a few things. Perhaps this may have cause problems.

I noticed you have a Toshiba laptop. There is an LQ thread about these and, wham!, your dmesg says it all:

Perhaps does this system instability happen when you change the brightness? We can turn ACPI off, but only if you don't care about your fan never going, power saving features, and perhaps battery problems. (I don't think you'll want this...)

Another thing, the kernel makes quite some fuss about APIC. This is a fairly major function, but I did some Googling and looked at your dmesg and realized the kernel only sees 1 processor. Perhaps we can fix some problems by appending the "noapic" kernel parameter to you grub.conf (GRUB2), or menu.lst (GRUB '1'). This may well be safe enough to do w/out really doing too much because you only have 1 CPU. (or so Linux says...)
Quoting the kernel :) ;

Another case is that you have two CPUS, but your BIOS is messing stuff up. The kernel reported:
You could enable the parameter, the same way as above for disabling it, but instead appending "lapic"

It looks to me like the Linux kernel is having a lot of trouble with APIC. The laptop also has a firmware bug with brightness. Try the suggestions above, and have a look at the Toshiba's BIOS and see where APIC is disables\d, or if it has the parameter. I believe this will solve your problem :)

Remember that the Kernel often will say what the problem is.

I hope this information resolves your issue. :)

Well, I spent a while testing things out, and it seems to still be there. I couldn't find the /etc/X11/Xorg.conf file, or anything similar, except for /usr/share/man/man5/xorg.conf.5.gz. I was using find as root, I can't remember the exact syntax, as that was a few days ago.

I did do the noapic thing, but the only difference that seemed to make was to make the screen go black when it froze instead of just freezing. I went into the BIOS, but I couldn't find anything related to APIC there. From what you said, that seems like the probable culprit, but I'm stuck on how to fix it.

Hardware-wise, what I know of in here is a 2.0 GHz P4 single-core CPU, 768 MB of RAM, a 160 GB Seagate hard drive (replacement for the original), and the D-Link PCMICA wireless card. The exact model is the Toshiba Satellite A15-S1692, at least according to the sticker on the bottom of the laptop.

At this point, I don't know what to do about it. It's starting to depress me (and it crashed right before I posted my reply, so I had to start over).