On Tue, 2012-03-06 at 16:24 +0000, Adamson, Andy wrote:
> With "NFS: Properly handle the case where the delegation is revoked" patch added to Trond's origin/testign branch, my python test that removes the (delegation) stateid on a READ, and returns NFS4ERR_BAD_STATEID passes as the client recovers by testing and freeing the stateid, and then sending an OPEN with a CLAIM_NULL.
>
> -->Andy
>
>
> BTW: this patch cleans up some stateid compares and the second chunk is needed for the origin/testing branch to compile. You've probably already found this…..
Yes. In order to make the patch apply to the stable kernels, I actually
inserted it before the stateid changes in the 'devel' branch, and then
fixed the stateid patches up instead. The current 'testing' branch
should have a working version of both.
BTW: with this patch, the "NFSv4.1: Fix the checking of the stateid when
returning a delegation" is not needed as a stable fix, since we no
longer use nfs_async_inode_return_delegation() in state recovery so I've
removed the Cc: stable from that patch.
--
Trond Myklebust
Linux NFS client maintainer
NetApp
Trond.Myklebust@xxxxxxxxxx
www.netapp.com
��.n��������+%������w��{.n�����{��w��� jg��� �����ݢj����G���� ���j:+v���w�m������w�������h�����٥
