> What as? I have both klogd and syslogd running on my system, and have
> had a few oops on various occasions, but have NEVER seen anything in
> /var/log/messages relating to them...
Oopses would be reported as kern.*. This can provide a lot of cruft if you
have:
kern.* /var/log/kernel.log
in /etc/syslog.conf; this can be resolved by replacing kern.* with
kern.info. (Oopses aren't reported at level info, but syslog logs info and
higher unless you specify =info to get just level info.)
> > But if klogd or syslogd are not running then the Oops never gets
> > written to disk. Alas this occurs all too often.
Or if your /etc/syslog.conf has not been told to log kern.*, or if the
logfile it's written to is written non-synchronously and your machine
immediately crashes, or if the panic takes out the filesystem driver for
/var/log's partition, or if the disk driver panics, or ...
While it's somewhat offtopic to this particular thread, I think there ought
to be some kind of way to stash the panic message *somewhere* other than
syslog. The last ten kernel panics I've encountered have either locked up
my machine hard, happened while I'm in X[1], or involved spontaneous
reboots.
Of course, what to do with a kernel panic raises several problems. SPARCs,
at least, have support from their ROM for doing kernel coredumps, and no
real mode vs. protected mode problems. Other architectures may have
similar features, but alas, I've only ever used SPARC and Intel. (Well, I
guess a Sun 3/60 kinda counts as M68K, but it's got similar PROM support,
and isn't very M68K-like at all.)
It must be possible to do *something* - heck, even NT manages to do a crash
dump *into an NTFS filesystem*. And while I don't necessarily want a dump
of my 128 megabytes of RAM into / (which wouldn't fit, anyway), logging the
Oops somewhere would be nice.
Unfortunately, I'm not much of a kernel hacker, and my low-level PC
knowledge has gradually become obsolete over the last five years or so[2],
and I don't really have many ideas on what it might be possible to do in
the event of a kernel panic. Maybe a kernel command-line option to blat
panics to a dedicated floppy drive? I have a 1.2M 5 1/4" drive I've not
used in maybe two years, I'd be happy for panics to be written to it...
Anyway, enough rambling. I'm sure y'all have more important things to
attend to.
Happy New Year!
--Jeff
[1]I realize that if I used fbdev, I'd still see the panic. However, my
CLGD5446 isn't currently supported (AFAIK) by fbdev, and it's slow enough
without having to go through the VESA layer, too.)
[2]I went straight from an XT running DOS to a 486SX/25 running Linux
(after about a month of Windows) in 1996, and my low-level hardware
knowledge wasn't even up to date then.
-- Jeffrey Sean Connell | Networking/Telecommunications Engineer, GXC ankh@canuck.gen.nz | PGP key at http://www.canuck.gen.nz/~ankh/pgpkey.html ---------------------+-------------------------------------------------------
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/