On Thu, Jul 3, 2014 at 12:11 PM, Christoph Hellwig <hch@xxxxxxxxxxxxx> wrote:
> On Thu, Jul 03, 2014 at 11:32:08AM -0400, J. Bruce Fields wrote:
>> > > > + struct nfs4_client *found;
>> > > > +
>> > > > + if (cstate->clp) {
>> > > > + found = cstate->clp;
>> > > > + if (!same_clid(&found->cl_clientid, clid))
>> > > > + return nfserr_stale_clientid;
>> > >
>> > > That's new behavior, isn't it?
>> > >
>> >
>> > Yeah, I suppose it is, but it's hard to understand a valid use-case for
>> > sending multiple ops in a compound with different clientids. Certainly
>> > no well-behaved client would do that, would it? (Hmm, that might be an
>> > interesting pynfs test, come to think of it).
>>
>> We could ask for a clarification in 3530bis if there's not already
>> something there that clearly forbids this, but I'm not sure if it's even
>> worth it.
>
> Or just handle it and be done with me. After all we'd just need to put
> the existing client, and store the new one in the cstate.
It is definitely not allowed in NFSv4.1. There is a special place in
hell^W^W^Werrorcode NFS4ERR_SEQUENCE_POS that tells you that your
compound is insane.
--
Trond Myklebust
Linux NFS client maintainer, PrimaryData
trond.myklebust@xxxxxxxxxxxxxxx
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
