NFSv4: Ensure delegation recall and byte range lock removal don't conflict

Add a mutex to the struct nfs4_state_owner to ensure that delegation
recall doesn't conflict with byte range lock removal.

Note that we nest the new mutex _outside_ the state manager reclaim
protection (nfsi->rwsem) in order to avoid deadlocks.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

4 files changed, 16 insertions, 2 deletions

diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c
index 2542cdaa1116..6390a4b5fee7 100644
--- a/fs/nfs/delegation.c
+++ b/fs/nfs/delegation.c
@@ -71,8 +71,10 @@ static int nfs_delegation_claim_locks(struct nfs_open_context *ctx, struct nfs4_
         int status = 0;
 
         if (inode->i_flock == NULL)
-                goto out;
+                return 0;
 
+        if (inode->i_flock == NULL)
+                goto out;
         /* Protect inode->i_flock using the file locks lock */
         lock_flocks();
         for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) {
@@ -113,12 +115,15 @@ again:
                 get_nfs_open_context(ctx);
                 spin_unlock(&inode->i_lock);
                 sp = state->owner;
+                /* Block nfs4_proc_unlck */
+                mutex_lock(&sp->so_delegreturn_mutex);
                 seq = raw_seqcount_begin(&sp->so_reclaim_seqcount);
                 err = nfs4_open_delegation_recall(ctx, state, stateid);
                 if (!err)
                         err = nfs_delegation_claim_locks(ctx, state);
                 if (!err && read_seqcount_retry(&sp->so_reclaim_seqcount, seq))
                         err = -EAGAIN;
+                mutex_unlock(&sp->so_delegreturn_mutex);
                 put_nfs_open_context(ctx);
                 if (err != 0)
                         return err;
diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index b12b73472020..944c9a5c1039 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -93,6 +93,7 @@ struct nfs4_state_owner {
         struct list_head     so_states;
         struct nfs_seqid_counter so_seqid;
         seqcount_t           so_reclaim_seqcount;
+        struct mutex         so_delegreturn_mutex;
 };
 
 enum {
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index f7e05ade5572..d51227371c67 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -4485,7 +4485,9 @@ static struct rpc_task *nfs4_do_unlck(struct file_lock *fl,
 
 static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
 {
-        struct nfs_inode *nfsi = NFS_I(state->inode);
+        struct inode *inode = state->inode;
+        struct nfs4_state_owner *sp = state->owner;
+        struct nfs_inode *nfsi = NFS_I(inode);
         struct nfs_seqid *seqid;
         struct nfs4_lock_state *lsp;
         struct rpc_task *task;
@@ -4495,12 +4497,17 @@ static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *
         status = nfs4_set_lock_state(state, request);
         /* Unlock _before_ we do the RPC call */
         request->fl_flags |= FL_EXISTS;
+        /* Exclude nfs_delegation_claim_locks() */
+        mutex_lock(&sp->so_delegreturn_mutex);
+        /* Exclude nfs4_reclaim_open_stateid() - note nesting! */
         down_read(&nfsi->rwsem);
         if (do_vfs_lock(request->fl_file, request) == -ENOENT) {
                 up_read(&nfsi->rwsem);
+                mutex_unlock(&sp->so_delegreturn_mutex);
                 goto out;
         }
         up_read(&nfsi->rwsem);
+        mutex_unlock(&sp->so_delegreturn_mutex);
         if (status != 0)
                 goto out;
         /* Is this a delegated lock? */
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index fff97228cdec..6ace365c6334 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -519,6 +519,7 @@ nfs4_alloc_state_owner(struct nfs_server *server,
         atomic_set(&sp->so_count, 1);
         INIT_LIST_HEAD(&sp->so_lru);
         seqcount_init(&sp->so_reclaim_seqcount);
+        mutex_init(&sp->so_delegreturn_mutex);
         return sp;
 }