Skip to contentSkip to conversation

lockd: Mostly remove BKL from the server

PATCH In review Bryan Schumaker 3 participants 5 messages Sep 21, 2010 – Sep 22, 2010

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 2/2 lockd: Mostly remove BKL from the server

lockd: Mostly remove BKL from the server

This patch removes all but one call to lock_kernel() from the server.

---

fs/lockd/svc4proc.c +0 -2
index 031c656..a336e83 100644
@@ -230,9 +230,7 @@ static void nlm4svc_callback_exit(struct rpc_task *task, void *data)
 
 static void nlm4svc_callback_release(void *data)
 {
-	lock_kernel();
 	nlm_release_call(data);
-	unlock_kernel();
 }
 
 static const struct rpc_call_ops nlm4svc_callback_ops = {
fs/lockd/svclock.c +21 -10
index 84055d3..6f1ef00 100644
@@ -52,12 +52,13 @@ static const struct rpc_call_ops nlmsvc_grant_ops;
  * The list of blocked locks to retry
  */
 static LIST_HEAD(nlm_blocked);
+static DEFINE_SPINLOCK(nlm_blocked_lock);
 
 /*
  * Insert a blocked lock into the global list
  */
 static void
-nlmsvc_insert_block(struct nlm_block *block, unsigned long when)
+nlmsvc_insert_block_locked(struct nlm_block *block, unsigned long when)
 {
 	struct nlm_block *b;
 	struct list_head *pos;
@@ -87,6 +88,13 @@ nlmsvc_insert_block(struct nlm_block *block, unsigned long when)
 	block->b_when = when;
 }
 
+static void nlmsvc_insert_block(struct nlm_block *block, unsigned long when)
+{
+	spin_lock(&nlm_blocked_lock);
+	nlmsvc_insert_block_locked(block, when);
+	spin_unlock(&nlm_blocked_lock);
+}
+
 /*
  * Remove a block from the global list
  */
@@ -94,7 +102,9 @@ static inline void
 nlmsvc_remove_block(struct nlm_block *block)
 {
 	if (!list_empty(&block->b_list)) {
+		spin_lock(&nlm_blocked_lock);
 		list_del_init(&block->b_list);
+		spin_unlock(&nlm_blocked_lock);
 		nlmsvc_release_block(block);
 	}
 }
@@ -651,7 +661,7 @@ static int nlmsvc_grant_deferred(struct file_lock *fl, struct file_lock *conf,
 	struct nlm_block *block;
 	int rc = -ENOENT;
 
-	lock_kernel();
+	spin_lock(&nlm_blocked_lock);
 	list_for_each_entry(block, &nlm_blocked, b_list) {
 		if (nlm_compare_locks(&block->b_call->a_args.lock.fl, fl)) {
 			dprintk("lockd: nlmsvc_notify_blocked block %p flags %d\n",
@@ -665,13 +675,13 @@ static int nlmsvc_grant_deferred(struct file_lock *fl, struct file_lock *conf,
 			} else if (result == 0)
 				block->b_granted = 1;
 
-			nlmsvc_insert_block(block, 0);
+			nlmsvc_insert_block_locked(block, 0);
 			svc_wake_up(block->b_daemon);
 			rc = 0;
 			break;
 		}
 	}
-	unlock_kernel();
+	spin_unlock(&nlm_blocked_lock);
 	if (rc == -ENOENT)
 		printk(KERN_WARNING "lockd: grant for unknown block\n");
 	return rc;
@@ -803,7 +813,7 @@ static void nlmsvc_grant_callback(struct rpc_task *task, void *data)
 
 	dprintk("lockd: GRANT_MSG RPC callback\n");
 
-	lock_kernel();
+	spin_lock(&nlm_blocked_lock);
 	/* if the block is not on a list at this point then it has
 	 * been invalidated. Don't try to requeue it.
 	 *
@@ -825,19 +835,20 @@ static void nlmsvc_grant_callback(struct rpc_task *task, void *data)
 		/* Call was successful, now wait for client callback */
 		timeout = 60 * HZ;
 	}
-	nlmsvc_insert_block(block, timeout);
+	nlmsvc_insert_block_locked(block, timeout);
 	svc_wake_up(block->b_daemon);
 out:
-	unlock_kernel();
+	spin_unlock(&nlm_blocked_lock);
 }
 
+/*
+ * FIXME: nlmsvc_release_block() grabs a mutex.  This is not allowed for an
+ * .rpc_release rpc_call_op
+ */
 static void nlmsvc_grant_release(void *data)
 {
 	struct nlm_rqst		*call = data;
-
-	lock_kernel();
 	nlmsvc_release_block(call->a_block);
-	unlock_kernel();
 }
 
 static const struct rpc_call_ops nlmsvc_grant_ops = {
fs/lockd/svcproc.c +0 -2
index 0f2ab74..c3069f3 100644
@@ -260,9 +260,7 @@ static void nlmsvc_callback_exit(struct rpc_task *task, void *data)
 
 static void nlmsvc_callback_release(void *data)
 {
-	lock_kernel();
 	nlm_release_call(data);
-	unlock_kernel();
 }
 
 static const struct rpc_call_ops nlmsvc_callback_ops = {
#1 lore raw
Quoted 3 lines · Bryan Schumaker

lockd: Mostly remove BKL from the server

This patch removes all but one call to lock_kernel() from the server.

Thanks. Does this need to go in any particular order with respect to the other BKL patches, or should I just apply this whenever?

--b.

Quoted 138 lines

---

fs/lockd/svc4proc.c +0 -2
index 031c656..a336e83 100644
@@ -230,9 +230,7 @@ static void nlm4svc_callback_exit(struct rpc_task *task, void *data)
 
 static void nlm4svc_callback_release(void *data)
 {
-	lock_kernel();
 	nlm_release_call(data);
-	unlock_kernel();
 }
 
 static const struct rpc_call_ops nlm4svc_callback_ops = {
fs/lockd/svclock.c +21 -10
index 84055d3..6f1ef00 100644
@@ -52,12 +52,13 @@ static const struct rpc_call_ops nlmsvc_grant_ops;
  * The list of blocked locks to retry
  */
 static LIST_HEAD(nlm_blocked);
+static DEFINE_SPINLOCK(nlm_blocked_lock);
 
 /*
  * Insert a blocked lock into the global list
  */
 static void
-nlmsvc_insert_block(struct nlm_block *block, unsigned long when)
+nlmsvc_insert_block_locked(struct nlm_block *block, unsigned long when)
 {
 	struct nlm_block *b;
 	struct list_head *pos;
@@ -87,6 +88,13 @@ nlmsvc_insert_block(struct nlm_block *block, unsigned long when)
 	block->b_when = when;
 }
 
+static void nlmsvc_insert_block(struct nlm_block *block, unsigned long when)
+{
+	spin_lock(&nlm_blocked_lock);
+	nlmsvc_insert_block_locked(block, when);
+	spin_unlock(&nlm_blocked_lock);
+}
+
 /*
  * Remove a block from the global list
  */
@@ -94,7 +102,9 @@ static inline void
 nlmsvc_remove_block(struct nlm_block *block)
 {
 	if (!list_empty(&block->b_list)) {
+		spin_lock(&nlm_blocked_lock);
 		list_del_init(&block->b_list);
+		spin_unlock(&nlm_blocked_lock);
 		nlmsvc_release_block(block);
 	}
 }
@@ -651,7 +661,7 @@ static int nlmsvc_grant_deferred(struct file_lock *fl, struct file_lock *conf,
 	struct nlm_block *block;
 	int rc = -ENOENT;
 
-	lock_kernel();
+	spin_lock(&nlm_blocked_lock);
 	list_for_each_entry(block, &nlm_blocked, b_list) {
 		if (nlm_compare_locks(&block->b_call->a_args.lock.fl, fl)) {
 			dprintk("lockd: nlmsvc_notify_blocked block %p flags %d\n",
@@ -665,13 +675,13 @@ static int nlmsvc_grant_deferred(struct file_lock *fl, struct file_lock *conf,
 			} else if (result == 0)
 				block->b_granted = 1;
 
-			nlmsvc_insert_block(block, 0);
+			nlmsvc_insert_block_locked(block, 0);
 			svc_wake_up(block->b_daemon);
 			rc = 0;
 			break;
 		}
 	}
-	unlock_kernel();
+	spin_unlock(&nlm_blocked_lock);
 	if (rc == -ENOENT)
 		printk(KERN_WARNING "lockd: grant for unknown block\n");
 	return rc;
@@ -803,7 +813,7 @@ static void nlmsvc_grant_callback(struct rpc_task *task, void *data)
 
 	dprintk("lockd: GRANT_MSG RPC callback\n");
 
-	lock_kernel();
+	spin_lock(&nlm_blocked_lock);
 	/* if the block is not on a list at this point then it has
 	 * been invalidated. Don't try to requeue it.
 	 *
@@ -825,19 +835,20 @@ static void nlmsvc_grant_callback(struct rpc_task *task, void *data)
 		/* Call was successful, now wait for client callback */
 		timeout = 60 * HZ;
 	}
-	nlmsvc_insert_block(block, timeout);
+	nlmsvc_insert_block_locked(block, timeout);
 	svc_wake_up(block->b_daemon);
 out:
-	unlock_kernel();
+	spin_unlock(&nlm_blocked_lock);
 }
 
+/*
+ * FIXME: nlmsvc_release_block() grabs a mutex.  This is not allowed for an
+ * .rpc_release rpc_call_op
+ */
 static void nlmsvc_grant_release(void *data)
 {
 	struct nlm_rqst		*call = data;
-
-	lock_kernel();
 	nlmsvc_release_block(call->a_block);
-	unlock_kernel();
 }
 
 static const struct rpc_call_ops nlmsvc_grant_ops = {
fs/lockd/svcproc.c +0 -3
index 0f2ab74..c3069f3 100644
@@ -260,9 +260,7 @@ static void nlmsvc_callback_exit(struct rpc_task *task, void *data)
 
 static void nlmsvc_callback_release(void *data)
 {
-	lock_kernel();
 	nlm_release_call(data);
-	unlock_kernel();
 }
 
 static const struct rpc_call_ops nlmsvc_callback_ops = {
--

To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#2 lore raw
Quoted 9 lines · J. Bruce Fields
Quoted 3 lines · Bryan Schumaker

lockd: Mostly remove BKL from the server

This patch removes all but one call to lock_kernel() from the server.

Thanks. Does this need to go in any particular order with respect to the other BKL patches, or should I just apply this whenever?

--b.

It doesn't use lock_flocks(), so you should be able to apply it whenever.

-Bryan

#3 lore raw
Quoted 4 lines · Bryan Schumaker
Quoted 3 lines · J. Bruce Fields
Quoted 2 lines · Bryan Schumaker

lockd: Mostly remove BKL from the server

Quoted 8 lines
Quoted 6 lines
Quoted 1 line

This patch removes all but one call to lock_kernel() from the server.

Thanks. Does this need to go in any particular order with respect to the other BKL patches, or should I just apply this whenever?

--b.

It doesn't use lock_flocks(), so you should be able to apply it whenever.

Note that a patch defining lock_flocks() in terms of the BKL went into Linus' tree last night, so we no longer have to worry about that dependency.

Cheers
Trond

#4 lore raw
In reply to Bryan Schumaker
Quoted 1 line · Bryan Schumaker

lockd: Mostly remove BKL from the server

I wish I had better file-locking tests: testing of cross-node gfs2 conflicts would be especially useful. This all looks right, though. Applying.

--b.

Quoted 136 lines

This patch removes all but one call to lock_kernel() from the server.

---

fs/lockd/svc4proc.c +0 -2
index 031c656..a336e83 100644
@@ -230,9 +230,7 @@ static void nlm4svc_callback_exit(struct rpc_task *task, void *data)
 
 static void nlm4svc_callback_release(void *data)
 {
-	lock_kernel();
 	nlm_release_call(data);
-	unlock_kernel();
 }
 
 static const struct rpc_call_ops nlm4svc_callback_ops = {
fs/lockd/svclock.c +21 -10
index 84055d3..6f1ef00 100644
@@ -52,12 +52,13 @@ static const struct rpc_call_ops nlmsvc_grant_ops;
  * The list of blocked locks to retry
  */
 static LIST_HEAD(nlm_blocked);
+static DEFINE_SPINLOCK(nlm_blocked_lock);
 
 /*
  * Insert a blocked lock into the global list
  */
 static void
-nlmsvc_insert_block(struct nlm_block *block, unsigned long when)
+nlmsvc_insert_block_locked(struct nlm_block *block, unsigned long when)
 {
 	struct nlm_block *b;
 	struct list_head *pos;
@@ -87,6 +88,13 @@ nlmsvc_insert_block(struct nlm_block *block, unsigned long when)
 	block->b_when = when;
 }
 
+static void nlmsvc_insert_block(struct nlm_block *block, unsigned long when)
+{
+	spin_lock(&nlm_blocked_lock);
+	nlmsvc_insert_block_locked(block, when);
+	spin_unlock(&nlm_blocked_lock);
+}
+
 /*
  * Remove a block from the global list
  */
@@ -94,7 +102,9 @@ static inline void
 nlmsvc_remove_block(struct nlm_block *block)
 {
 	if (!list_empty(&block->b_list)) {
+		spin_lock(&nlm_blocked_lock);
 		list_del_init(&block->b_list);
+		spin_unlock(&nlm_blocked_lock);
 		nlmsvc_release_block(block);
 	}
 }
@@ -651,7 +661,7 @@ static int nlmsvc_grant_deferred(struct file_lock *fl, struct file_lock *conf,
 	struct nlm_block *block;
 	int rc = -ENOENT;
 
-	lock_kernel();
+	spin_lock(&nlm_blocked_lock);
 	list_for_each_entry(block, &nlm_blocked, b_list) {
 		if (nlm_compare_locks(&block->b_call->a_args.lock.fl, fl)) {
 			dprintk("lockd: nlmsvc_notify_blocked block %p flags %d\n",
@@ -665,13 +675,13 @@ static int nlmsvc_grant_deferred(struct file_lock *fl, struct file_lock *conf,
 			} else if (result == 0)
 				block->b_granted = 1;
 
-			nlmsvc_insert_block(block, 0);
+			nlmsvc_insert_block_locked(block, 0);
 			svc_wake_up(block->b_daemon);
 			rc = 0;
 			break;
 		}
 	}
-	unlock_kernel();
+	spin_unlock(&nlm_blocked_lock);
 	if (rc == -ENOENT)
 		printk(KERN_WARNING "lockd: grant for unknown block\n");
 	return rc;
@@ -803,7 +813,7 @@ static void nlmsvc_grant_callback(struct rpc_task *task, void *data)
 
 	dprintk("lockd: GRANT_MSG RPC callback\n");
 
-	lock_kernel();
+	spin_lock(&nlm_blocked_lock);
 	/* if the block is not on a list at this point then it has
 	 * been invalidated. Don't try to requeue it.
 	 *
@@ -825,19 +835,20 @@ static void nlmsvc_grant_callback(struct rpc_task *task, void *data)
 		/* Call was successful, now wait for client callback */
 		timeout = 60 * HZ;
 	}
-	nlmsvc_insert_block(block, timeout);
+	nlmsvc_insert_block_locked(block, timeout);
 	svc_wake_up(block->b_daemon);
 out:
-	unlock_kernel();
+	spin_unlock(&nlm_blocked_lock);
 }
 
+/*
+ * FIXME: nlmsvc_release_block() grabs a mutex.  This is not allowed for an
+ * .rpc_release rpc_call_op
+ */
 static void nlmsvc_grant_release(void *data)
 {
 	struct nlm_rqst		*call = data;
-
-	lock_kernel();
 	nlmsvc_release_block(call->a_block);
-	unlock_kernel();
 }
 
 static const struct rpc_call_ops nlmsvc_grant_ops = {
fs/lockd/svcproc.c +0 -2
index 0f2ab74..c3069f3 100644
@@ -260,9 +260,7 @@ static void nlmsvc_callback_exit(struct rpc_task *task, void *data)
 
 static void nlmsvc_callback_release(void *data)
 {
-	lock_kernel();
 	nlm_release_call(data);
-	unlock_kernel();
 }
 
 static const struct rpc_call_ops nlmsvc_callback_ops = {