NFSv4: Ensure that file unlock requests don't conflict with state recovery

The unlock path is currently failing to take the nfs_client->cl_sem read
lock, and hence the recovery path may see locks disappear from underneath
it.
Also ensure that it takes the nfs_inode->rwsem read lock so that it there
is no conflict with delegation recalls.

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

2 files changed, 20 insertions, 10 deletions

diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 660c5dcfb0a5..aec4e47c462d 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -3273,6 +3273,8 @@ 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_client *clp = state->owner->so_client;
+        struct nfs_inode *nfsi = NFS_I(state->inode);
         struct nfs_seqid *seqid;
         struct nfs4_lock_state *lsp;
         struct rpc_task *task;
@@ -3282,8 +3284,15 @@ 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;
-        if (do_vfs_lock(request->fl_file, request) == -ENOENT)
+        down_read(&clp->cl_sem);
+        down_read(&nfsi->rwsem);
+        if (do_vfs_lock(request->fl_file, request) == -ENOENT) {
+                up_read(&nfsi->rwsem);
+                up_read(&clp->cl_sem);
                 goto out;
+        }
+        up_read(&nfsi->rwsem);
+        up_read(&clp->cl_sem);
         if (status != 0)
                 goto out;
         /* Is this a delegated lock? */
@@ -3510,6 +3519,7 @@ static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request
 static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 {
         struct nfs_client *clp = state->owner->so_client;
+        struct nfs_inode *nfsi = NFS_I(state->inode);
         unsigned char fl_flags = request->fl_flags;
         int status;
 
@@ -3522,18 +3532,13 @@ static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock
         if (status < 0)
                 goto out;
         down_read(&clp->cl_sem);
+        down_read(&nfsi->rwsem);
         if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
-                struct nfs_inode *nfsi = NFS_I(state->inode);
                 /* Yes: cache locks! */
-                down_read(&nfsi->rwsem);
                 /* ...but avoid races with delegation recall... */
-                if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
-                        request->fl_flags = fl_flags & ~FL_SLEEP;
-                        status = do_vfs_lock(request->fl_file, request);
-                        up_read(&nfsi->rwsem);
-                        goto out_unlock;
-                }
-                up_read(&nfsi->rwsem);
+                request->fl_flags = fl_flags & ~FL_SLEEP;
+                status = do_vfs_lock(request->fl_file, request);
+                goto out_unlock;
         }
         status = _nfs4_do_setlk(state, cmd, request, 0);
         if (status != 0)
@@ -3543,6 +3548,7 @@ static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock
         if (do_vfs_lock(request->fl_file, request) < 0)
                 printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n", __func__);
 out_unlock:
+        up_read(&nfsi->rwsem);
         up_read(&clp->cl_sem);
 out:
         request->fl_flags = fl_flags;
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index beade5570b56..16c9fbdf97b4 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -849,9 +849,11 @@ static int nfs4_state_mark_reclaim_nograce(struct nfs_client *clp, struct nfs4_s
 static int nfs4_reclaim_locks(struct nfs4_state *state, const struct nfs4_state_recovery_ops *ops)
 {
         struct inode *inode = state->inode;
+        struct nfs_inode *nfsi = NFS_I(inode);
         struct file_lock *fl;
         int status = 0;
 
+        down_write(&nfsi->rwsem);
         for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) {
                 if (!(fl->fl_flags & (FL_POSIX|FL_FLOCK)))
                         continue;
@@ -874,8 +876,10 @@ static int nfs4_reclaim_locks(struct nfs4_state *state, const struct nfs4_state_
                                 goto out_err;
                 }
         }
+        up_write(&nfsi->rwsem);
         return 0;
 out_err:
+        up_write(&nfsi->rwsem);
         return status;
 }