On Tue, Mar 30, 2004 at 04:23:33PM +0200, Havard Eidnes wrote:

> which I suspect is still the abort() routine in the HBA driver.
> ...yep, 0x882e1064 is wd33c93_abort() with the next one in line
> being wd33c93_timeout().  Is there any chance that this part of
> the code could be made more robust against passing of "strange"
> arguments, or is this something which is supposed to absolutely
> never happen?

This shouldn't happen.  Taking a quick look at the codepath, the pointer to
acb should _never_ contain junk.

> sd0: no disk label

This is okay, pre-install.

> sq0: Unexpected interrupt!

This is normal on an IP20.  The seeq/hpc combo doesn't handle "early interrupts"
well.

> sd0(wdsc0:0:2:0): wdsc0: timed out; asr=0x00 [acb 0x889157b0 (flags 0x11, dleft 0)], <state 1, nexus 0x0, resid 10000, msg(q 0,o 100)>trap: TLB miss (load or instr. fetch) in kernel mode
> status=0xf802, cause=0x30000008, epc=0x882e10c8, vaddr=0x8

Okay, not a null pointer per se; something is scribbling on memory that
shouldn't be touched.  Would be interesting to find out why SCSI writes are
timing out.  Bad termination?

-- Chris
	GPG key FEB9DE7F (91AF 4534 4529 4BCC 31A5  938E 023E EEFB FEB9 DE7F)