[732] in linux-net channel archive

home help back first fref pref prev next nref lref last post

Re: malloc() woes?

daemon@ATHENA.MIT.EDU (Andrew Walker)
Wed Jul 19 22:11:17 1995

From: andy@keo.kvaerner.no (Andrew Walker)
To: scottr@msn.fullfeed.com
Date: Wed, 19 Jul 1995 11:57:44 +0200 (MET DST)
Cc: linux-gcc@vger.rutgers.edu, linux-net@vger.rutgers.edu
In-Reply-To: <199507190649.IAA14872@london.keo.kvaerner.no> from "Andrew Walker" at Jul 19, 95 08:49:22 am

I wrote (in my ignorance):
> 
> Hi,
> 
> I too have a problem with code using malloc(), in libc 5.2.3. I assumed the
> problem was in the nfs server code (nfs-server-2.2beta2) until I read this posting.
> Now I'm not so sure.
> 
> I get segmentation faults in malloc when re-initializing the authority stuff
> from /etc/exports, but before that I also get some wierd effects mounting file
> systems from my Linux box.
> 
> I have a very simple /etc/exports, which contains just:
> 
> 	/usr/local	(rw)
> 
> which should export /usr/local as read-write to any clients that want it. When
> I mount it from a Sun 3/80 (SunOS 4.1.1) I see the directory structure, but the
> permisions are real screwy. I can create, but not modify files in a tmp directory
> (permisions drwxrwxrwt). Users can't even create files in directories that they own.
> 
> If /etc/exports is changed to:
> 
> 	/usr/local	fulcrum(rw)
> 
> where fulcrum is the name of the client then the permissions are ok.

Ok, ok. Forget this bit. I read the code - anonymous exports default to
'all_squash'. I realise what was going on here (even if I disagree with
it as a default).

> If I send a SIGHUP to rpc.nfsd or rpc.mountd to re-read /etc/exports they both
> die with SIGSEGV in malloc, called from auth_init() doing a re-initialize.
> 
> I'll try to debug things a bit tonight, if I get time.

This bit still holds true. I compiled SunOS versions of rpc.nfsd and rpc.mountd,
and tested them a bit. There don't seem to be any problems, which points at the
Linux libc/malloc() - but then how strict is the SunOS libc/malloc() ??????

I think I'll try purify....

-Andy

-- 
Andy Walker                              Kvaerner Engineering a.s.
Andrew.Walker@keo.kvaerner.no            P.O. Box 222, N-1324 Lysaker, Norway

   ......if the answer isn't violence, neither is it silence......


home help back first fref pref prev next nref lref last post