Skip to contentSkip to conversation

dm: do not call dm_sync_table() when creating new devices

PATCH In review Hannes Reinecke 3 participants 3 messages Nov 5, 2014 Device mapper

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 dm: do not call dm_sync_table() when creating new devices

When creating new devices dm_sync_table() calls synchronize_rcu_expedited(), causing _all_ pending RCU pointers to be flushed. This causes a latency overhead especially noticeable when creating lots of devices.
And all of this is pointless as there are no old maps to be disconnected, and hence no stale pointers which would need to be cleared up.

---
 drivers/md/dm.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)
drivers/md/dm.c +6 -3
index 58f3927..2c2c6cf 100644
@@ -2341,7 +2341,8 @@ static struct dm_table *__bind(struct mapped_device *md, struct dm_table *t,
 		set_bit(DMF_MERGE_IS_OPTIONAL, &md->flags);
 	else
 		clear_bit(DMF_MERGE_IS_OPTIONAL, &md->flags);
-	dm_sync_table(md);
+	if (old_map)
+		dm_sync_table(md);
 
 	return old_map;
 }
@@ -2782,7 +2783,8 @@ int dm_suspend(struct mapped_device *md, unsigned suspend_flags)
 	 * flush_workqueue(md->wq).
 	 */
 	set_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags);
-	synchronize_srcu(&md->io_barrier);
+	if (map)
+		synchronize_srcu(&md->io_barrier);
 
 	/*
 	 * Stop md->queue before flushing md->wq in case request-based
@@ -2802,7 +2804,8 @@ int dm_suspend(struct mapped_device *md, unsigned suspend_flags)
 
 	if (noflush)
 		clear_bit(DMF_NOFLUSH_SUSPENDING, &md->flags);
-	synchronize_srcu(&md->io_barrier);
+	if (map)
+		synchronize_srcu(&md->io_barrier);
 
 	/* were we interrupted ? */
 	if (r < 0) {
Signature · 1 line
1.8.5.2
#1 lore raw
Quoted 10 lines · Hannes Reinecke

When creating new devices dm_sync_table() calls synchronize_rcu_expedited(), causing _all_ pending RCU pointers to be flushed. This causes a latency overhead especially noticeable when creating lots of devices.
And all of this is pointless as there are no old maps to be disconnected, and hence no stale pointers which would need to be cleared up.

Looks good, I'll get it staged for 3.19 (will need to be rebased on the linux-dm.git 'dm-for-3.19' branch given the suspend related refactoring there, but I'll take care of it as needed).

maintainer Reviewed-by on v1 1/1
#2 lore raw
In reply to Hannes Reinecke
Quoted 10 lines · Hannes Reinecke

When creating new devices dm_sync_table() calls synchronize_rcu_expedited(), causing _all_ pending RCU pointers to be flushed. This causes a latency overhead especially noticeable when creating lots of devices.
And all of this is pointless as there are no old maps to be disconnected, and hence no stale pointers which would need to be cleared up.

Quoted 46 lines
---
 drivers/md/dm.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)
drivers/md/dm.c +6 -3
index 58f3927..2c2c6cf 100644
@@ -2341,7 +2341,8 @@ static struct dm_table *__bind(struct mapped_device *md, struct dm_table *t,
 		set_bit(DMF_MERGE_IS_OPTIONAL, &md->flags);
 	else
 		clear_bit(DMF_MERGE_IS_OPTIONAL, &md->flags);
-	dm_sync_table(md);
+	if (old_map)
+		dm_sync_table(md);
 
 	return old_map;
 }
@@ -2782,7 +2783,8 @@ int dm_suspend(struct mapped_device *md, unsigned suspend_flags)
 	 * flush_workqueue(md->wq).
 	 */
 	set_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags);
-	synchronize_srcu(&md->io_barrier);
+	if (map)
+		synchronize_srcu(&md->io_barrier);
 
 	/*
 	 * Stop md->queue before flushing md->wq in case request-based
@@ -2802,7 +2804,8 @@ int dm_suspend(struct mapped_device *md, unsigned suspend_flags)
 
 	if (noflush)
 		clear_bit(DMF_NOFLUSH_SUSPENDING, &md->flags);
-	synchronize_srcu(&md->io_barrier);
+	if (map)
+		synchronize_srcu(&md->io_barrier);
 
 	/* were we interrupted ? */
 	if (r < 0) {
Signature · 6 lines
1.8.5.2

--
dm-devel mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/dm-devel