On Wed, 26 Jun 2002, Miguel Beccari wrote:
> This is not my first post about kernel crashing (see old posts as: > spontaneous reboots). > > However now I have some more information to provide you. So I hope it > could help to determinate the problem that many people have. > > First: the hardware > > cpu > : 740/750 > temperature : 39 C (uncalibrated) > clock > : 300MHz > revision > : 2.2 (pvr 0008 0202) > bogomips > : 598.01 > machine > : PowerMac1,1 > motherboard > : PowerMac1,1 MacRISC Power Macintosh > L2 cache : 512K unified > memory > : 96MB > pmac-generation > : NewWorld > > Second: the kernel > > Linux serverg3 2.4.18-6.1mdk #1 lun giu 24 07:53:06 CEST 2002 ppc > > Third: the facts > > I upgraded from 8.0 to 8.2 PPC hoping that the crash/reboot problem will > solve: well, it still persists. > > One or 2 times I could see the error on the monitor (it was months I was > trying to see it, because when the crash happens the machine reboots > by itself and no messages in the logs). > > The error said something like: > "Machine check in kernel mode > Caused by (from SRR1=41000): Transfer error > " > ... could not read ... > > "Altivec error" > > .... could not read... > > > So it seems to be something about altivec. > > > Yesterday I recompiled the kernel *without* altivec support (my G3 does > not have altivec). > > Today the new crash with the same error. > > My next step is to leave mdk kernel and recompile the unpatched one from > kernel.org > > > Has someone an idea about this ? > > Note that the crash is rare: it can crash in 24 hours or in 25 days. > That's the problem. The console message doesn't reveal a lot, and the there doesn't seem to be any pattern at all to the reboots. I have reboots here too, but I am positive in my case that it is power line related, as they generally occur at the same time as a local power event (dryer kicks on, etc.) You'd think a laptop would be fairly immune to such things, byt my experience shows otherwise. I've also read bits on the MOL list about failure to service the pmu under heavy load, which will cause a power down, but they said it was definitely a power down, not a reboot. You'd probably be better off with BenH's kernel than kernel.org if you want to explore other kernel options. I don't know how fast Ben's work gets into the main kernel tree. The last few Mandrake iterations have not built out of the box on PPC. Stew Benedict -- MandrakeSoft PPC FAQ: http://www.linux-mandrake.com/en/ppcFAQ.php3 IRC: irc.openproject.net #cooker-ppc
