Just a question: does anyone ever faced a problem while doing a "pacman -Syu"?
It happened several times on my box that this command froze completly my system, just after having retrieved the database for the "current" repository.
When the problem occurs, my system is nearly completly suspended. In this situation, I can't even peform a "ps -ef". It starts to display some processes and then it hangs before finishing.
Then I try to reboot, but the shutdown hangs too while trying to terminate still running processes.
Each time after a crash and a reboot, I can see from /var/log/errors the following entry:
Mar 15 18:11:05 pokemon kernel: Unable to handle kernel NULL pointer dereference at virtual address 0000 0003 Mar 15 18:11:05 pokemon kernel: *pde = 00000000 Mar 15 18:40:35 pokemon kernel: *pde = 00000000
Hum... this log is not very helpful!
Does someone already experienced the same problem?
Right now, my only hypothesis is that the problem can be linked to the libtar library used by pacman, but I'm not really sure of it.
The point is that pacman uses the 1.2.5 release, and the latest available one is now the 1.2.11, which corrects lots of bugs. So, I will made a try with it.
Anyway, I've instrumented my pacman binary with lots of "printf" everywhere in the "sync" code (especially around calls to libtar functions) to try to find out which is the instruction that generates that evil crash...
Unfortunately it does not dump a core file, it only hangs... so it's quite difficult to investigate
It occurs about once a week. It happens since the very first day I installed Arch (4 months ago). At that time, I was thinking the problem was with my kernel or something with my configuration... but I have to admit that, to date, my system never crashed, excepted during "pacman -Syu" operations...
well i never have had problems of this sort. i have had problems with changes in gnome package dependencies causing pacman unable to upgrade my system but this is due to my custom lprng packages conflicting with ANY cups dependency. all i usually have to do is rebuild the trouble maker package with lprng support.
I am not your friend
I have the same problem
the HW config is IBM Netvista Celeron 1300
I checked dmesg and it said that it panic/ed in kjournal
then I rebuild the kernel / 2.4.20 with ac2 patch (PKGBUILD needs some tweeking)
and it seems more stable now.
It is strange that I install Arch 0.4 the day before on some other PC and it is really stable.
That PC is Intel P4 2.4 (intel board)
and I wanted to mention that install also seg/faulted many times before succed with a bare bone BASE packages.
segfaulted in what circumstances? you should file a detailed bug. i should mentione though "stepping back" in the install procedure will cause troubles.
I am not your friend