[PATCH v1 1/2] NFSD: Do not cache failed SEQUENCE operations

[Chuck Lever <cel@kernel.org>]

From: Chuck Lever <chuck.lever@oracle.com>

RFC 8881 Section 2.10.6.4 says:

> If sa_cachethis or csa_cachethis is TRUE, then the replier MUST
> cache a reply except if an error is returned by the SEQUENCE or
> CB_SEQUENCE operation (see Section 2.10.6.1.2).

This text also appears in RFC 5661. Further, Section 2.10.6.1.2
says:

> Any time SEQUENCE or CB_SEQUENCE returns an error, the sequence ID
> of the slot MUST NOT change. The replier MUST NOT modify the reply
> cache entry for the slot whenever an error is returned from
> SEQUENCE or CB_SEQUENCE.

NFSD caches the result of solo SEQUENCE operations, but rtm's
reproducer sends two operations in the failing COMPOUND. NFSD should
not attempt to cache the reply.

The logic in nfsd4_is_solo_sequence() was incorrect: it checked the
current operation index, not the total count of operations in the
COMPOUND. If the SEQUENCE operation, which is always operation 1,
fails in a multi-operation compound, resp->opcnt is always 1. Thus
when a SEQUENCE operation fails, nfsd4_is_solo_sequence() always
returns true.

Reported-by: <rtm@csail.mit.edu>
Closes: https://lore.kernel.org/linux-nfs/c3628d57-94ae-48cf-8c9e-49087a28cec9@oracle.com/T/#t
Fixes: 468de9e54a90 ("nfsd41: expand solo sequence check")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 fs/nfsd/xdr4.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/nfsd/xdr4.h b/fs/nfsd/xdr4.h
index ee0570cbdd9e..d1837a10b0c2 100644
--- a/fs/nfsd/xdr4.h
+++ b/fs/nfsd/xdr4.h
@@ -926,7 +926,8 @@ struct nfsd4_compoundres {
 static inline bool nfsd4_is_solo_sequence(struct nfsd4_compoundres *resp)
 {
 	struct nfsd4_compoundargs *args = resp->rqstp->rq_argp;
-	return resp->opcnt == 1 && args->ops[0].opnum == OP_SEQUENCE;
+
+	return args->opcnt == 1 && args->ops[0].opnum == OP_SEQUENCE;
 }
 
 /*
-- 
2.51.0