dm: ignore merge_bvec for snapshots when safe

Add a new flag DMF_MERGE_IS_OPTIONAL to struct mapped_device to indicate
whether the device can accept bios larger than the size its merge
function returns.  When set, use this to send large bios to snapshots
which can split them if necessary.  Snapshot I/O may be significantly
fragmented and this approach seems to improve peformance.

Before the patch, dm_set_device_limits restricted bio size to page size
if the underlying device had a merge function and the target didn't
provide a merge function.  After the patch, dm_set_device_limits
restricts bio size to page size if the underlying device has a merge
function, doesn't have DMF_MERGE_IS_OPTIONAL flag and the target doesn't
provide a merge function.

The snapshot target can't provide a merge function because when the merge
function is called, it is impossible to determine where the bio will be
remapped.  Previously this led us to impose a 4k limit, which we can
now remove if the snapshot store is located on a device without a merge
function.  Together with another patch for optimizing full chunk writes,
it improves performance from 29MB/s to 40MB/s when writing to the
filesystem on snapshot store.

If the snapshot store is placed on a non-dm device with a merge function
(such as md-raid), device mapper still limits all bios to page size.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>

diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index d58b200..03516c7 100644 (file)
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -540,8 +540,7 @@ int dm_set_device_limits(struct dm_target *ti, struct dm_dev *dev,
         * If not we'll force DM to use PAGE_SIZE or
         * smaller I/O, just to be safe.
         */
-
-       if (q->merge_bvec_fn && !ti->type->merge)
+       if (dm_queue_merge_is_compulsory(q) && !ti->type->merge)
                blk_limits_max_hw_sectors(limits,
                                          (unsigned int) (PAGE_SIZE >> 9));
        return 0;

diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index aeb0fa1..1000eaf 100644 (file)
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -111,6 +111,7 @@ EXPORT_SYMBOL_GPL(dm_get_rq_mapinfo);
 #define DMF_FREEING 3
 #define DMF_DELETING 4
 #define DMF_NOFLUSH_SUSPENDING 5
+#define DMF_MERGE_IS_OPTIONAL 6
 
 /*
  * Work processed by per-device workqueue.
@@ -1992,6 +1993,59 @@ static void __set_size(struct mapped_device *md, sector_t size)
        i_size_write(md->bdev->bd_inode, (loff_t)size << SECTOR_SHIFT);
 }
 
+/*
+ * Return 1 if the queue has a compulsory merge_bvec_fn function.
+ *
+ * If this function returns 0, then the device is either a non-dm
+ * device without a merge_bvec_fn, or it is a dm device that is
+ * able to split any bios it receives that are too big.
+ */
+int dm_queue_merge_is_compulsory(struct request_queue *q)
+{
+       struct mapped_device *dev_md;
+
+       if (!q->merge_bvec_fn)
+               return 0;
+
+       if (q->make_request_fn == dm_request) {
+               dev_md = q->queuedata;
+               if (test_bit(DMF_MERGE_IS_OPTIONAL, &dev_md->flags))
+                       return 0;
+       }
+
+       return 1;
+}
+
+static int dm_device_merge_is_compulsory(struct dm_target *ti,
+                                        struct dm_dev *dev, sector_t start,
+                                        sector_t len, void *data)
+{
+       struct block_device *bdev = dev->bdev;
+       struct request_queue *q = bdev_get_queue(bdev);
+
+       return dm_queue_merge_is_compulsory(q);
+}
+
+/*
+ * Return 1 if it is acceptable to ignore merge_bvec_fn based
+ * on the properties of the underlying devices.
+ */
+static int dm_table_merge_is_optional(struct dm_table *table)
+{
+       unsigned i = 0;
+       struct dm_target *ti;
+
+       while (i < dm_table_get_num_targets(table)) {
+               ti = dm_table_get_target(table, i++);
+
+               if (ti->type->iterate_devices &&
+                   ti->type->iterate_devices(ti, dm_device_merge_is_compulsory, NULL))
+                       return 0;
+       }
+
+       return 1;
+}
+
 /*
  * Returns old map, which caller must destroy.
  */
@@ -2002,6 +2056,7 @@ static struct dm_table *__bind(struct mapped_device *md, struct dm_table *t,
        struct request_queue *q = md->queue;
        sector_t size;
        unsigned long flags;
+       int merge_is_optional;
 
        size = dm_table_get_size(t);
 
@@ -2027,10 +2082,16 @@ static struct dm_table *__bind(struct mapped_device *md, struct dm_table *t,
 
        __bind_mempools(md, t);
 
+       merge_is_optional = dm_table_merge_is_optional(t);
+
        write_lock_irqsave(&md->map_lock, flags);
        old_map = md->map;
        md->map = t;
        dm_table_set_restrictions(t, q, limits);
+       if (merge_is_optional)
+               set_bit(DMF_MERGE_IS_OPTIONAL, &md->flags);
+       else
+               clear_bit(DMF_MERGE_IS_OPTIONAL, &md->flags);
        write_unlock_irqrestore(&md->map_lock, flags);
 
        return old_map;

diff --git a/drivers/md/dm.h b/drivers/md/dm.h
index 1aaf167..6745dbd 100644 (file)
--- a/drivers/md/dm.h
+++ b/drivers/md/dm.h
@@ -66,6 +66,8 @@ int dm_table_alloc_md_mempools(struct dm_table *t);
 void dm_table_free_md_mempools(struct dm_table *t);
 struct dm_md_mempools *dm_table_get_md_mempools(struct dm_table *t);
 
+int dm_queue_merge_is_compulsory(struct request_queue *q);
+
 void dm_lock_md_type(struct mapped_device *md);
 void dm_unlock_md_type(struct mapped_device *md);
 void dm_set_md_type(struct mapped_device *md, unsigned type);