Re: NFS: deterministic oops in nfs_file_closeh

Ion Badulescu (ionut@moisil.cs.columbia.edu)
Mon, 24 Aug 1998 01:06:50 -0400

On Sun, 23 Aug 1998 19:04:45 EDT, Alan Cox wrote:
>
>> What do you think about the non-caching problem?
>> NFS-files are cached until the files are closed. Seems that the close
>> invalidates all caching.
>
> Be nice to fix but Im worried right now about the data losing and other
> unpleasant bugs not a performance tweak 8(

If past experience with 2.0 is of any value, nfs_delete_inode is called too
early in the process so it's not the right place to totally destroy the
inode. Some of the code in that function should probably go to nfs_clear_inode
(which 2.0 didn't have unfortunately).

If I get some time this coming week, I might have a look at the problem.

BTW, performance tweak or not, NFS (at least over UDP) is basically unusable
right now, I get something like 10k/s write & read performance to a Solaris
server. NFS/TCP is much better, but it has other problems, although your last
patch is definitely a major step in the right direction.

Ion

-- 
  It is better to keep your mouth shut and be thought a fool,
            than to open it and remove all doubt.


-
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.altern.org/andrebalsa/doc/lkml-faq.html