Re: crashinfo issue in HEAD

From: Gavin Atkinson <gavin_at_FreeBSD.org>
Date: Fri, 11 Sep 2009 17:04:08 +0100
On Fri, 2009-09-11 at 17:41 +0300, Andriy Gapon wrote:
> on 11/09/2009 17:26 Gavin Atkinson said the following:
> > On Fri, 2009-09-11 at 16:47 +0300, Andriy Gapon wrote:
> >> I had a panic (provoked via sw watchdog, nothing exciting) and after reboot I
> >> noticed that couple of processes dumped core, apparently while running as part of
> >> crashinfo procedure.
> >>
> >> savecore: reboot after panic: watchdog timeout
> >> savecore: writing core to vmcore.0
> >> kernel: pid 725 (ps), uid 0: exited on signal 11 (core dumped)
> >> kernel: pid 742 (fstat), uid 0: exited on signal 11
> >>
> >> core.0.txt accordingly has:
> >> ------------------------------------------------------------------------
> >> ps -axl
> >> Segmentation fault (core dumped)
> > 
> > You need at least r196990 (which has not yet been MFC'd) and
> > http://people.freebsd.org/~gavin/PRs/137890.2.diff .  Hopefully both of
> > these will appear in 8.x.
> 
> I mentioned HEAD in the subject, but didn't say specifically that this is very
> recent HEAD - r197046. And amd64. Sorry for withholding such important details.
> So that patch is not in HEAD yet? I will try it then, thanks!

Hmm, in that case this is probably a different bug to the ones already
found, and that patch won't help.  Can you recompile ps and libkvm with
debugging symbols, run it under gdb and get a backtrace?  Your original
stack trace shows kvm_getprocs() calling _kvm_freeprocs(), however the
code doesn't appear to ever do this: indeed, nothing in libkvm calls
_kvm_freeprocs() directly.

Gavin
Received on Fri Sep 11 2009 - 14:04:12 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:55 UTC