Re: Nagios SIGSEGV on FreeBSD 8

On Tue, Sep 22, 2009 at 07:43:10PM -0500, Scott Lambert wrote:
> Bcc:ing port to close the thread there.
> 
> On Wed, Sep 23, 2009 at 12:58:01AM +0200, Stefan Bethke wrote:
> > Am 22.09.2009 um 23:29 schrieb Scott Lambert:
> > >I've posted this to FreeBSD-ports and Nagios-Users without a nibble.
> > >
> > >I've been running a FreeBSD 8-BETA2 server for DNS on a network I
> > >recently took over.  No problems.  We needed to get Nagios running on
> > >that network to watch all the hosts in RFC 1918 space.  Taking the
> > >easy route, I just installed the Nagios 3.0.6 port on this 8-BETA2
> > >box.
> > >
> > >Nagios runs great until an acknowledged down host (with
> > >acknowledgment comment) comes back up.  Nagios exits on a SIGSEGV.
> > >It seems to only happen when we have retention data (retention.dat)
> > >showing the host down.  If we just restart Nagios without removing
> > >the retention.dat file, it exists on SIGSEGV the next time it tries
> > >to mark the host up.  I upgraded to the nagios-devel (Nagios 3.1.2)
> > >port and we have the same problem.
> >
> > I'v discovered this bug in Nagios some weeks ago, but posting to
> > nagios-devel did not elicit a response.  The problem is a use-after-
> > free bug in the code that deals with comments (or it was in my case
> > anyway).
> >
> > The bug leads to a SEGV because -current (including all 8.0 betas) has
> > malloc flags set that make free write a canary value to the just-freed
> > block.  If you move up to rc1, or set the "j" flag, nagios appears to
> > work.
> >
> > If you need references to the threads on freebsd-ports and nagios-
> > devel, let me know.
> 
> "sudo ln -s 'j' /etc/malloc.conf" seems to have done the trick.  Nice,
> easy, fix (or bandaid?).

It's a bandaid. The bug still exists in Nagios and should certainly be
fixed there.

> I don't know how I failed to find your thread before.
> 
> I found your thread on nagios-devel:
> 
> http://archive.netbsd.se/?ml=nagios-devel&a=2009-07&t=11196612
> 
> Thank you for taking the time to figure this out in the first place and
> to reply to my message.

The maintainer of the nagios port is usually very responsive to things.
If a patch is available for this it may make sense to apply it to our
port.

-- WXS