Is the panic with the root filesystem on IDE/SCSI or the Booter loading the kernel from SCSI or IDE?
It doesn't matter if I boot Mac OS from either the IDE or SCSI, nor whether I load the kernel from IDE or SCSI - only whether IDE or SCSI is used for the filesystem root.
Is this in current and 5.1? Is it a hard 128MB line (ie does 160M work)
Current works with the full amount, and 160 megs works, so it's some amount between 160 and 196 megs of memory on netbsd-5 and 5.1 kernels.
Are the soft errors in the dmesg unrelated timewise to the panic? Does SCSI then IDE still work reliably under current?
I haven't had an opportunity to test a current kernel booted with the SCSI root long enough to see if that's happy yet. I'll do that when a rather large rsync is done.
With regards to the soft errors, they happen regardless of the drive and with no ill effects. I ran this machine for more than a year with a 750 gig IDE drive which gave tons of soft errors but with no data corruption; the drive tested just fine when I connected it to another machine and used it extensively, and the SMART data showed nothing out of the ordinary. A guess might be that the soft errors are interrupted PIO transfers which have to be retransferred or something like that - a problem on the Quadra's end, not the drive's.
Thanks, John