https://bugzilla.novell.com/show_bug.cgi?id=350980 User robin.listas@telefonica.net added comment https://bugzilla.novell.com/show_bug.cgi?id=350980#c53 --- Comment #53 from Carlos Robinson <robin.listas@telefonica.net> 2008-11-25 08:40:03 MST --- (In reply to comment #52 from Thomas Renninger)
It happens only with certain suse versions Ok, I think I got it, sorry about that.
Don't worry, this bug has a long story.
Let's try to limit this to as less patches as possible, then try out the ones that look guilty.
This 11.0 kernel was the first which did not work: 2.6.25.18-0.2-pae
Correct.
Did you do upates before, can you remember the last working kernel?
I did all the updates offered by Yast Online Update. I don't remember which one was the previous kernel, but it has to be the previous one in the update server. Perhaps the info is in y2logRPM - yes, here: nimrodel:/var/log/YaST2 # grep kernel-pae- y2logRPM 2008-09-09 22:53:16 kernel-pae-2.6.25.5-1.1.i586.rpm installed ok 2008-09-10 11:44:16 kernel-pae-2.6.25.11-0.1.i586.rpm installed ok 2008-09-11 20:56:28 kernel-pae-2.6.25.16-0.1.i586.rpm installed ok 2008-10-30 02:14:24 kernel-pae-2.6.25.18-0.2.i586.rpm installed ok So, this kernel was installed on 2008-10-30, but I didn't notice the problem earlier because I had a lot of network activity (from 08-11-01 to 08-11-11), and this produces interrupts which keep awake the machine. And grepping for the string "failed after 300 secs" in the warn log, which is the tell-tale, I first see it on Nov 19 15:19:40 I also see I installed a bunch of things on Nov 12, but nothing on the 19th, or at least nothing I can think interesting. I could attach if you want the list with dates, produced by: rpm -q -a --queryformat "%{INSTALLTIME}\t%{INSTALLTIME:day} \ %{BUILDTIME:day} %-30{NAME}\t%15{VERSION}-%-7{RELEASE}\t%{arch} \ %25{PACKAGER}\n" | sort | cut --fields="2-"
For 10.3 all kernels (you tested) were broken.
Yes, as far as I remember. At least, we couldn't find anything that worked. I played recompiling kernels, too. Most of it is logged in this Bugzilla and related ones. Maybe later I can dig out the numbers, the history of this problem.
10.2 and before worked well.
Right.
This is all about i386-pae kernels or did you also try x86_64 or i386 without -pae support kernels?
Well, 10.2 and 10.3 were non pae, pae started to be shipped with 11.0, afaik. This machine is 32 bits, so x86_64 is out of the question. Another detail I remembered: [Bug 426592] clocksource 'acpi_pm' looses time. So I'm forcing the TSC clocksource; I think I have been doing that since 10.3 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.