Date: Mon, 28 Apr 2003 15:06:12 +1000 From: "Johny Mattsson (EPA)" <Johny.Mattsson@ericsson.com.au> To: "'current@freebsd.org'" <current@freebsd.org> Subject: RE: kernel paniced: now what? - IDEA Message-ID: <CF3A6B018CD5844BA8C3FC7E059725BFE28EE3@eaubrnt018.epa.ericsson.se>
next in thread | raw e-mail | index | archive | help
Hi all,
Seeing as this is a question that pops up rather frequently ("it panicked -
now what?"), I thought I'd put up an idea for discussion:
I suspect a lot of you people have or have had exposure to products such as
Cisco and/or NetScreen. One of the really nifty things with those (and
probably others) is the command "show tech" (or "get tech" in NetScreen's
case), which provides a rather complete debug dump including state
information, statistics, configuration data, and the like, which can be used
quite efficiently to pinpoint problems.
How about introducing such a command to the kernel debugger? That way even a
non-kernel-hacker such as myself would have a quite straight forward way of
generating a good quality bug report. Say for example it was to include the
stack trace, parameter printouts/dereferences for each of the frames, mutex
states, memory stats, and some other useful goodies. Of course, if you don't
have a serial terminal attached, it'll probably be too much info to copy by
hand, but that's a separate issue.
I'd love to back this idea up with patches, but I'm not a kernel hacker, and
have no exposure to debugger internals either. If what I'm suggesting is
infeasible, I'll take your word for it, but I thought I'd put the question
out there at least.
Comments? Would the resulting advantages compensate for the time invested in
adding such a feature?
Regards,
/Johny
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CF3A6B018CD5844BA8C3FC7E059725BFE28EE3>