You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Marc Slemko <ma...@worldgate.com> on 1997/09/12 17:04:39 UTC
Apache Freezing (select: (listen): Bad address) (fwd)
I still don't buy this, but FWIW...
He is claiming that since the kernel doesn't like what Apache is passing
it then Apache must be broken, since it couldn't every be Linux. It does
happen with 1.2, Linux is the only OS it happens on. That doesn't mean it
can't be a bug in Apache, but...
---------- Forwarded message ----------
>Path: pulp.ucs.ualberta.ca!rover.ucs.ualberta.ca!news.bc.net!logbridge.uoregon.edu!newsfeed.internetmci.com!newsfeed.internetmci.com!195.99.66.215!news-feed1.eu.concert.net!news-peer.bt.net!btnet!newsfeed.cableol.net!news.cymru.net!not-for-mail
>From: alan@snowcrash.cymru.net (Alan Cox)
>Newsgroups: comp.infosystems.www.servers.unix
>Subject: Re: Apache Freezing (select: (listen): Bad address)
>Date: 12 Sep 1997 13:18:30 +0100
>Organization: CymruNET
>Lines: 20
>Message-ID: <5v...@snowcrash.cymru.net>
>References: <5v...@news1.tor.acc.ca> <5v...@scanner.worldgate.com>
>NNTP-Posting-Host: snowcrash.cymru.net
>Xref: pulp.ucs.ualberta.ca comp.infosystems.www.servers.unix:37054
In article <5v...@scanner.worldgate.com>,
Marc Slemko <ma...@znep.com> wrote:
>Appears to be a bug in the Linux kernel. It gets confused. It is
>possible this is a bug in Apache, but unlikely since it doesn't
>happen on other platforms and Apache really doesn't do much that could
>possibly cause it.
Apache passes the kernel an invalid user space memory address. I've verified
and traced the problem. It _is_ an apache bug not a Linux bug. I've also
never seen it occur with Apache 1.2, so I guess someone has an obscure
memory scribble in Apache 1.1
Alan
--
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Wouldn't it be nice if all the spam went directly to //
postmaster@uu.net security@uu.net fraud@uu.net spam-complaints@uu.net //
postmaster@localhost acctspay@uu.net and help@uunet.uu.net =============''=INUX
Re: Apache Freezing (select: (listen): Bad address) (fwd)
Posted by Dean Gaudet <dg...@arctic.org>.
On Sat, 13 Sep 1997, Ben Laurie wrote:
> OTOH, if listenmaxfd got stomped on, it could have a bad effect, I'd
> guess.
I looked at this too, and it just truncates listenmaxfd to OPEN_MAX so
shouldn't generate an EFAULT based on that.
Dean
Re: Apache Freezing (select: (listen): Bad address) (fwd)
Posted by Ben Laurie <be...@algroup.co.uk>.
Marc Slemko wrote:
>
> On Fri, 12 Sep 1997, Ben Laurie wrote:
>
> > Marc Slemko wrote:
> > >
> > > I still don't buy this, but FWIW...
> > >
> > > He is claiming that since the kernel doesn't like what Apache is passing
> > > it then Apache must be broken, since it couldn't every be Linux. It does
> > > happen with 1.2, Linux is the only OS it happens on. That doesn't mean it
> > > can't be a bug in Apache, but...
> >
> > If Apache is passing an invalid address, then it is Apache. Of course,
> > the fact that the kernel says it is invalid is not proof that it is
> > invalid - but that may not be what he means when he says he's verified
> > and traced it.
>
> Apache doesn't pass addresses. It passes a fd to a listen() call.
Err, no. This is a select() call we are talking about. But since it is
passing the address of a stack-based variable, some pretty serious (and
strange) scribbling would have to have happened for it to be Apache's
fault.
OTOH, if listenmaxfd got stomped on, it could have a bad effect, I'd
guess.
Cheers,
Ben.
--
Ben Laurie |Phone: +44 (181) 994 6435|Apache Group member
Freelance Consultant |Fax: +44 (181) 994 6472|http://www.apache.org
and Technical Director|Email: ben@algroup.co.uk |Apache-SSL author
A.L. Digital Ltd, |http://www.algroup.co.uk/Apache-SSL
London, England. |"Apache: TDG" http://www.ora.com/catalog/apache
Re: Apache Freezing (select: (listen): Bad address) (fwd)
Posted by Marc Slemko <ma...@worldgate.com>.
On Fri, 12 Sep 1997, Ben Laurie wrote:
> Marc Slemko wrote:
> >
> > I still don't buy this, but FWIW...
> >
> > He is claiming that since the kernel doesn't like what Apache is passing
> > it then Apache must be broken, since it couldn't every be Linux. It does
> > happen with 1.2, Linux is the only OS it happens on. That doesn't mean it
> > can't be a bug in Apache, but...
>
> If Apache is passing an invalid address, then it is Apache. Of course,
> the fact that the kernel says it is invalid is not proof that it is
> invalid - but that may not be what he means when he says he's verified
> and traced it.
Apache doesn't pass addresses. It passes a fd to a listen() call. The fd
it gets is from other system and library calls. If they give a fd with
bogus information, Apache can not be faulted for passing it. If you look
at what Apache does with that fd, there is very little. It is possible,
but unlikely, that Apache is scribbling over the memory associated with
it.
Re: Apache Freezing (select: (listen): Bad address) (fwd)
Posted by Ben Laurie <be...@algroup.co.uk>.
Marc Slemko wrote:
>
> I still don't buy this, but FWIW...
>
> He is claiming that since the kernel doesn't like what Apache is passing
> it then Apache must be broken, since it couldn't every be Linux. It does
> happen with 1.2, Linux is the only OS it happens on. That doesn't mean it
> can't be a bug in Apache, but...
If Apache is passing an invalid address, then it is Apache. Of course,
the fact that the kernel says it is invalid is not proof that it is
invalid - but that may not be what he means when he says he's verified
and traced it.
Cheers,
Ben.
--
Ben Laurie |Phone: +44 (181) 994 6435|Apache Group member
Freelance Consultant |Fax: +44 (181) 994 6472|http://www.apache.org
and Technical Director|Email: ben@algroup.co.uk |Apache-SSL author
A.L. Digital Ltd, |http://www.algroup.co.uk/Apache-SSL
London, England. |"Apache: TDG" http://www.ora.com/catalog/apache