Re: AW: kernel compile on RS6000 7043-260
On Friday 26 August 2005 11:23, Ralph wrote:
> Hi Pat,
> thanks for Your fast reply, but that isn't it.
> I did exactly what You wrote before, so the sym53c8xx
> and scsi disk IS in config flagged with "Y".
> When I compare the top of the logs, there's already
> something different. The original kernel loads ramdisk first
> and later finds the cramfs and mounts it.
> The new kernel does a chrp-boot...
> Must really have to do with initrd/cramfs/ramdisk.
Can you send your config? I've got a couple of 7043-260s, and could try
out your kernel on one of mine. The thing I noticed is that it doesn't
seem to be trying to init the scsi controller on the machine in the
'newer' config you have listed. I'll have to do a bit of digging
around, but I'll try to get my kernel config to send you.
> BTW I just want to build my own kernel to get rid of
> the IDE support in the original kernel, since the "noprobe"
> kernel argument doesn't work and this adds about 2 minutes
> boot time for probing something, that isn't really there.
> Maybe noprobe is fixed in a newer kernel, but as far as I know,
> 2.6.11 is the last 32bit-smp-kernel. And although I'm on a 64-bit
> machine, I'd like to use the 32-bit environment for compatibility
> reasons to an older ppc-machine here.
FYI, I've only run a 64-bit kernel on POWER3 machines like the 260; the
64-bit kernel runs 32-bit userland code just fine, you can use a 64-bit
kernel to run the same exact software that's running on a 32-bit
Purdue University Research Computing --- http://www.rcac.purdue.edu/
The Computer Refuge --- http://computer-refuge.org