On Wed, Jul 6, 2011 at 12:27 PM, Dale <rdalek1...@gmail.com> wrote: > Jesús J. Guerrero Botella wrote: >> >> Dale, random hard-lockups are only due to hardware or kerne, it can't >> be otherwisel (drivers count as part of kernel). The fact that >> compilation doesn't lock your system only means that the thing >> (whatever it is) is not bount to intensive I/O operations and/or high >> cpu loads. >> >> Openldap itself can't hard lock up anything if the kernel doesn't give >> it permissions to do so (kernel bug) or if the hardware is not faulty. >> Same goes for tray apps. >> >> > > OK. I tested this and it doesn't help any. I tried three different > kernels, all of which I have used in the past with no problems, and get the > same thing. I have tried reinstalling nvidia-drivers and a different > versions of nvidia-drivers, same thing. So, either previously working > kernels are now broke, nvidia which was working fine just a few days ago > with no recent updates here just broke or just maybe it is something else we > have yet to figure out yet. I also ran memtest for HOURS with not one > problem reported. > > Given I have tried the above, do you still think it is kernel, nvidia or > hardware? I'm about to run tests on the drive now. I suspect it is going > to show no problems as well. > > This is also the reason I keep old kernels laying around: > > root@fireball / # ls -al /boot/bzImage* > -rw-r--r-- 1 root root 4257840 Mar 21 18:39 /boot/bzImage-2.6.38-1 > -rw-r--r-- 1 root root 4480304 Mar 22 12:00 /boot/bzImage-2.6.38-2 > -rw-r--r-- 1 root root 4493872 Mar 25 13:02 /boot/bzImage-2.6.38-3 > -rw-r--r-- 1 root root 4496336 Mar 29 03:25 /boot/bzImage-2.6.38-r1-1 > -rw-r--r-- 1 root root 4454480 Apr 7 19:13 /boot/bzImage-2.6.38-r1-2 > -rw-r--r-- 1 root root 4451760 May 3 02:16 /boot/bzImage-2.6.38-r3-1 > -rw-r--r-- 1 root root 4451536 May 12 06:12 /boot/bzImage-2.6.38-r5-1 > root@fireball / # > > I did try .39 but it had issues. I got rid of those. > > Dale >
If I had to guess I'd say, since this followed a power failure where the machine was live and operating (if I've understood the thread through a quick scan) that some file on disk has gotten corrupted and it's that corruption that's causing the problem. You've checked memory. Let's assume that te processor and MB weren't damaged by this event. If that's the case - and unfortunately I don't know of any way to ensure it hasn't as it requires one to have a bit-accurate image of the machine before the power failure - there's probably no way to eliminate this as a possibility short of an emerge -e @world. It's not where I'd start. I'd probably look for core dump files or very carefully do experiment s trying to isolate exactly what part of KDE is firing off the problem. re-emerging the NVidia driver is a no-brainer as it takes no more than 1-2 minutes to test things. Rebuilding the machine is certainly more involved. If you have lots of disk space you might rsync the whole machine to a new partition to do the work, then using something other than KDE which doesn't crash rebuild the copy from a chroot which leaves the machine usable while the rebuild is going on. None of this sounds like fun... - Mark
