In article <slrn3vfsehd30l.rk.martijnb@atlas.ipv6.stack.nl>,
Martijn van Buul  <martijnb@atlas.ipv6.stack.nl> wrote:
>Hello;
>
>I'm having a bit of a problem with mounting NFS shares exported by a Linux
>(2.4.30) with both a 4.0.99 as well as a a 4.0_BETA machine. Everything
>works fine if I don't specify any mount options, but things go wrong
>if I use NFS over TCP (I'm doing an NFS mount over a wireless link; using
>nfs-over-tcp has proven to be helpful in this case)
>
>It would almost seem like NFS requests are cut short. Large directories 
>appear empty, file access fails, except for small files. Files smaller than
>3897 bytes work, larger files fail with a "RPC struct is bad". The
>silly thing is that this happens only one way - while I can copy larger files
>from my NetBSD client to the Linux server without problem. There are no
>warnings to be found in dmesg or /var/log/messages on either client or
>server.
>
>Restricting the max. NFS read- and write data sizes to 3072 seems to "fix" 
>the problem, but I can't help but think that this can't be right, and that
>this merely camouflages the problem. Using UDP works right away.
>
>Any ideas what might be wrong? I tested this on both my (NetBSD) desktop,
>as well as my laptop, using both wireless as well as wired connection, in
>an attempt to exclude the possible influence of ill-behaving switches..
>

I would send-pr...

christos