You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Randy Terbush <ra...@zyzzyva.com> on 1997/03/22 21:30:04 UTC
Re: another SIGSEGV
Yes, I am seeing them. I thought it was some of my hacks. It
started occuring _before_ the proxy patches, and happens in a
server that does _not_ have the proxy code compiled in.
I'm setting up a test machine on FreeBSD 2.2 and will see if I can
track this down.
> I'm getting SIGSEGV after lingering close calls here on FreeBSD 2.1.7.
> >From the error log:
>
> [Thu Mar 20 04:06:37 1997] shutdown: Invalid argument
> [Thu Mar 20 04:06:37 1997] - lingering_close
> [Thu Mar 20 04:06:38 1997] shutdown: Invalid argument
> [Thu Mar 20 04:06:38 1997] - lingering_close
> [Thu Mar 20 04:06:40 1997] httpd: caught SIGSEGV, dumping core
> [Thu Mar 20 04:09:12 1997] shutdown: Invalid argument
> [Thu Mar 20 04:09:12 1997] - lingering_close
>
> etc. I'm checking to see if it's related to the latest proxy patch.
> Is anyone else seeing this? I've got a report it happens on Next too.
> Disabling lingering close stops it. It doesn't appear to happen after
> every lingering close, though. Might not even be related, but the error
> message (Invalid argument) is suspicious.
> --
> chuck
> Chuck Murcko
> The Topsail Group, West Chester PA USA
> chuck@topsail.org
Re: another SIGSEGV
Posted by Chuck Murcko <ch...@topsail.org>.
Chuck Murcko wrote:
>
> Well, the SIGSEGVs happen independently of the shutdown() errors, but
> for some reason I'm not getting core dumps here. I'll check into what to
> change in the OS config to get them turned on. It seems the default
> changed from on to off from 2.1.5 to 2.1.7.
>
> Oddly (and seemingly unrelatedly), the invalid argument to shutdown is
> always 0x0a on FreeBSD 2.1.7.
>
OK, the SIGSEGVs appear to be a memory leak somewhere. Setting
RlimitMEM max
greatly delays the onset of 'em. I'll build a version here with a
debugging malloc lib and see what I find.
--
chuck
Chuck Murcko
The Topsail Group, West Chester PA USA
chuck@topsail.org
Re: another SIGSEGV
Posted by Chuck Murcko <ch...@topsail.org>.
Well, the SIGSEGVs happen independently of the shutdown() errors, but
for some reason I'm not getting core dumps here. I'll check into what to
change in the OS config to get them turned on. It seems the default
changed from on to off from 2.1.5 to 2.1.7.
Oddly (and seemingly unrelatedly), the invalid argument to shutdown is
always 0x0a on FreeBSD 2.1.7.
Randy Terbush wrote:
>
> Yes, I am seeing them. I thought it was some of my hacks. It
> started occuring _before_ the proxy patches, and happens in a
> server that does _not_ have the proxy code compiled in.
>
> I'm setting up a test machine on FreeBSD 2.2 and will see if I can
> track this down.
>
> > I'm getting SIGSEGV after lingering close calls here on FreeBSD 2.1.7.
> > >From the error log:
> >
> > [Thu Mar 20 04:06:37 1997] shutdown: Invalid argument
> > [Thu Mar 20 04:06:37 1997] - lingering_close
> > [Thu Mar 20 04:06:38 1997] shutdown: Invalid argument
> > [Thu Mar 20 04:06:38 1997] - lingering_close
> > [Thu Mar 20 04:06:40 1997] httpd: caught SIGSEGV, dumping core
> > [Thu Mar 20 04:09:12 1997] shutdown: Invalid argument
> > [Thu Mar 20 04:09:12 1997] - lingering_close
> >
> > etc. I'm checking to see if it's related to the latest proxy patch.
> > Is anyone else seeing this? I've got a report it happens on Next too.
> > Disabling lingering close stops it. It doesn't appear to happen after
> > every lingering close, though. Might not even be related, but the error
> > message (Invalid argument) is suspicious.
--
chuck
Chuck Murcko
The Topsail Group, West Chester PA USA
chuck@topsail.org