NFS (a few questions)

Steven N. Hirsch (
Wed, 9 Jul 1997 18:43:16 -0400 (EDT)

On Wed, 9 Jul 1997, Olaf Kirch wrote:

> Harald Koenig wrote:
> May I also request that people put `NFS' somewhere in the subject
> when they report a problem to linux-kernel that they suspect is
> nfs-related? Quite frequently, I scan the lists only by subject, so
> threads like this one usually escape me.

Ok, done. I have been beating on the "new" nfs client and server code for
several months. However, on all kernels newer than 2.1.36, long nfs read
or write operations reliably lock the network layer for minutes at a time.
The log and screen get filled with messages about "..unable to get buffer
space" (from the network layer?).

On 2.1.36 it's fairly stable (and fast!) with only the occasional "mmap
i_node ring corrupted" message (WHAT in blazes does that mean??).

Most annoying problem for me is that I cannot seem to patch the nfs code
into any of the 2.0.x kernels I've hacked at! I've tried 2.0.28 and
2.0.30. Even after fixing some failed patches (context changed slightly)
and adding references in the entry.S (I believe I did this correctly, but
I'm not 100% positive), it reports service unavailable when exportfs is
run on the new kernel...

Would it be possible for you to re-generate the diffs against one of the
later 2.0.x kernels? Specifically, the symbol table and entry point
patches were not to be found in the latest linux-nfs distribution.

Many thanks!