Re: IRIX NFS server + Linux client == odd

From: Marcelo E. Magallon (marcelo.magallon@bigfoot.com)
Date: Mon May 15 2000 - 15:45:48 EST


>> Doug McNaught <doug@mcnaught.org> writes:

> Trond Myklebust <trond.myklebust@fys.uio.no> writes:
>
> > I suspect that the problem is a truncation error: IRIX uses 64-bit
> > cookies doesn't it? On NFSv2, you can only pass 32-bit cookies, hence
> > what is probably going on is that the hashing routine used by the
> > server is generating 2 identical cookies.
>
> From my recollection, Irix has a '32bitclients' export option that may
> work around this issue.

The option is there...

Trond is right:

     32bitclients
               Causes the server to mask off the high order 32 bits of
               directory cookies in NFS version 3 directory operations. This
               option may be required when clients run 32-bit operating
               systems that assume the entire cookie is contained in 32 bits
               and reject responses containing version 3 cookies with high
               bits on. IRIX 5.3 and Solaris 2.5 are examples of 32-bit
               operating systems with this behavior, which produces error
               messages like "Cannot decode response" on directory operations.
               XFS filesystems on the server can generate cookies with high
               bits on. Exporting filesystems with the 32bitclients option
               causes these bits to be masked and prevents error messages.

If I'm reading that right IRIX just chops the high bits off...

Calling getdents on IRIX (IRIX's trace won't give me the same
information as strace, and strace does not run on IRIX) I get the same
results on IRIX (as NFS3 client) as Linux... (that's d_ino, d_off,
d_reclen and d_name):

     3522927 473709853 56 946308814.179258_73.reggae:2,S
     3522897 473709853 56 946308812.179258_42.reggae:2,S

With the filesystem mounted as NFS2 (vers=2 passed to mount) I get
exactly the same result...

???

                                        Marcelo

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Mon May 15 2000 - 21:00:26 EST