Re: mountd segfault on itanium2

[Garrick Staples <garrick@usc.edu>]

[-- Attachment #1: Type: text/plain, Size: 1651 bytes --]

On Mon, May 03, 2004 at 08:35:46AM -0500, Jose R. Santos alleged:
> On 04/30/04 22:07:30, Garrick Staples wrote:
> > I have an ugly work-around that seems to be working.  It seems that 2.6 has a
> > new nfs interface for userspace.  By forcing mountd to use the older 2.4
> > interface, it doesn't segfault anymore.  So something in the new code paths is
> > broken.
> > 
> > In support/nfs/cachio.c:
> > int
> > check_new_cache(void)
> > {
> >         struct stat stb;
> > 
> >         return 0;  /* DISABLE NEW 2.6 INTERFACE */
> > 
> >         return  (stat("/proc/fs/nfs/filehandle", &stb) == 0) ||
> >                 (stat("/proc/fs/nfsd/filehandle", &stb) == 0);
> > }
> 
> If you want to use the old syscall interface, all you need to do is make 
> sure nfsd is not mounted in /proc/fs/nfsd.

Thank you.  That's much easier.

 
> > Am I losing any functionality by doing this?  I can't actually find any
> > problems.
> 
> I think there are no functionality losses but I have seen issues with using
> the syscall interface on large memory systems.  The syscall mount and 
> unmount the nfsdfs for every syscall.  This will cause the inode and dentry
> for this filesystem to be flush, but in order to do that it needs to walk 
> through all inode and dentry caches under lock.  If you have a system with
> 16GB of ram, mounting and unmounting can take a really long time.
> 
> -JRS

Both machines have 8GB.  I definitly see the slower mounting, but it's not
prohibitive.  And slow mounting is better than a segfault.


-- 
Garrick Staples, Linux/HPCC Administrator
University of Southern California

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]