Return-Path: Received: from fieldses.org ([174.143.236.118]:60155 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932194Ab1INLp1 (ORCPT ); Wed, 14 Sep 2011 07:45:27 -0400 From: "J. Bruce Fields" To: linux-nfs@vger.kernel.org Cc: "J. Bruce Fields" Subject: [PATCH 03/25] nfsd4: extend state lock over seqid replay logic Date: Wed, 14 Sep 2011 07:44:59 -0400 Message-Id: <1316000721-3289-4-git-send-email-bfields@redhat.com> In-Reply-To: <1316000721-3289-1-git-send-email-bfields@redhat.com> References: <1316000721-3289-1-git-send-email-bfields@redhat.com> To: linux-nfs@vger.kernel.org Sender: linux-nfs-owner@vger.kernel.org List-ID: Content-Type: text/plain MIME-Version: 1.0 There are currently a couple races in the seqid replay code: a retransmission could come while we're still encoding the original reply, or a new seqid-mutating call could come as we're encoding a replay. So, extend the state lock over the encoding (both encoding of a replayed reply and caching of the original encoded reply). I really hate doing this, and previously added the stateowner reference-counting code to avoid it (which was insufficient)--but I don't see a less complicated alternative at the moment. Signed-off-by: J. Bruce Fields --- fs/nfsd/nfs4proc.c | 5 +++-- fs/nfsd/nfs4state.c | 12 ++++++++---- 2 files changed, 11 insertions(+), 6 deletions(-) diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c index 50bae74..50063a8 100644 --- a/fs/nfsd/nfs4proc.c +++ b/fs/nfsd/nfs4proc.c @@ -408,8 +408,8 @@ out: if (open->op_stateowner) { nfs4_get_stateowner(open->op_stateowner); cstate->replay_owner = open->op_stateowner; - } - nfs4_unlock_state(); + } else + nfs4_unlock_state(); return status; } @@ -1227,6 +1227,7 @@ encode_op: be32_to_cpu(status)); if (cstate->replay_owner) { + nfs4_unlock_state(); nfs4_put_stateowner(cstate->replay_owner); cstate->replay_owner = NULL; } diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index bc1a9db..6cf729a 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -3501,7 +3501,8 @@ nfsd4_open_confirm(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, nfsd4_create_clid_dir(sop->so_client); out: - nfs4_unlock_state(); + if (!cstate->replay_owner) + nfs4_unlock_state(); return status; } @@ -3568,7 +3569,8 @@ nfsd4_open_downgrade(struct svc_rqst *rqstp, memcpy(&od->od_stateid, &stp->st_stateid, sizeof(stateid_t)); status = nfs_ok; out: - nfs4_unlock_state(); + if (!cstate->replay_owner) + nfs4_unlock_state(); return status; } @@ -3609,7 +3611,8 @@ nfsd4_close(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, if (list_empty(&so->so_stateids)) move_to_close_lru(so); out: - nfs4_unlock_state(); + if (!cstate->replay_owner) + nfs4_unlock_state(); return status; } @@ -4071,7 +4074,8 @@ nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate, out: if (status && lock->lk_is_new && lock_sop) release_lockowner(lock_sop); - nfs4_unlock_state(); + if (!cstate->replay_owner) + nfs4_unlock_state(); return status; } -- 1.7.4.1