On Fri, Jun 29, 2001 at 11:59:06PM +0100, Matthew Theobalds wrote:
> Pretty sure it was a panic since it said:
> 
> 	panic: scsipi_execute_xs
> 
> And then went into the debugger.

Yep. Figured that out after I went and read your PR. Meant to edit
that bit of my message after doing so and before sending, but I
didn't get back to it.

> Would a crash dump be the output when one types "c" in the debugger and 
> lots of Hexdecimal numbers scroll on the screen?

Um, no a crash dump involves panic(9) having caused the kernel to
write its memory state out to your primary swap partition when it
panicked, then savecore having copied the data out of swap (before
it's been added by swapctl(8), of course) and into a file under
/var/crash.

If you've got one of those there (or, better yet, if you can blow
away what's there, then provoke the panic again and get one the
circumstances of which are clear), then reply to the PR saying
you've got it. (Maybe don't attach it to the email though... I'm not
sure how accepting of really large binary MIME attachments, but I
doubt it would exactly be appreciated. ;^>)

> In that case, it would be possible, but difficult, since it would 
> require typing them all out to another machine. Is this what you meant?

Um, you could do that too, much more easily by connecting a serial
console to ttya (your modem port) and then logging the output of
whatever you choose to do in the debugger to a file on the machine
running the terminal emulator. (ZTerm on a mac, minicom on another
Unix machine, whatever.)

> I'd definitely agree, that almost occured to me, since scsi seems to be 
> mentioned a lot in the panic and it occurs at about the time one would 
> expect a SCSI system to be shutdown.
> 
> Should I add words to this effect to the PR, or do you think they'll 
> already have realised that?

Check the PR in the database. I've already done so.

http://www.netbsd.org/cgi-bin/query-pr-single.pl?number=13334

Cheers...

-- 
       ~ g r @ eclipsed.net