Jon Thackray writes: > maximilian attems writes: > > On Fri, 14 Jan 2005, Jon Thackray wrote: > > > > Content-Description: message body text > > > maximilian attems writes: > > > > On Thu, 13 Jan 2005, Jon Thackray wrote: > > > > > > > > > System has Intel 865 MB, P4/3400, 2Gb RAM, five discs including > > > > > one SCSI and one SATA. It worked fine with the 2.6.5 kernel, but the > > > > > 2.6.8 kernel won't boot. > > > > > > > > did you try booting with append "acpi=off"? > > > > > > No. I don't know what apic is. How should I modified the grub menu.lst > > > to do that? I can probably reboot the machine in a few days to try > > > that. I've attached a successful 2.6.5 boot dmesg. > > > > apic != acpi > > you'll find more about in kernel sources in Documentation/i386/IO-APIC.txt > > > > doesn't grub(8) has a nice man, i don't use it but something like this > > should work (make shure to have your correct root partition there): > > kernel /boot/vmlinuz-2.6.8.1-3-686 root=/dev/hda1 ro acpi=off > > Thanks, I'll give that a try. I've now tried this. The 2.6.8 kernel boots to completion. However, something has changed between 2.6.5 and 2.6.8 regarding the handling of SATA discs. In 2.6.5, the scsi subsystem gets initialised before all the drives are mounted. This allows my SATA disc to be assigned to /dev/sda, and also means the fsck stage works (even if all it does is check the journal counter). Under 2.6.8, the scsi subsystem doesn't get started, and so the SATA disc doesn't get assigned to /dev/sda, and the fsck stage drops me to a prompt offering me single user mode. I attach dmesg outputs demonstrating this. The interesting part of the 2.6.5 log starts at line 173, where it says "SCSI subsystem initialized". I'm going to continue to use 2.6.8 as I wish to see if the SATA corruptions under ext3 that I reported with the 2.6.5 kernel have gone away.
Attachment:
20050101152318-2.6.8
Description: 2.6.8 dmesg
Attachment:
20050101152326-2.6.5
Description: 2.6.5 dmesg