Btrfs: Add support for duplicate blocks on a single spindle

Signed-off-by: Chris Mason <chris.mason@oracle.com>

4 files changed, 37 insertions, 9 deletions

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 0a207861472..72deae63ec2 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -419,6 +419,7 @@ struct btrfs_csum_item {
 #define BTRFS_BLOCK_GROUP_METADATA (1 << 2)
 #define BTRFS_BLOCK_GROUP_RAID0    (1 << 3)
 #define BTRFS_BLOCK_GROUP_RAID1    (1 << 4)
+#define BTRFS_BLOCK_GROUP_DUP	   (1 << 5)
 
 
 struct btrfs_block_group_item {
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index ff75ad58676..42522232fde 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -913,9 +913,11 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 
 	fs_info->generation = btrfs_super_generation(disk_super) + 1;
 	if (btrfs_super_num_devices(disk_super) > 0) {
-		fs_info->data_alloc_profile = BTRFS_BLOCK_GROUP_RAID0;
-		fs_info->metadata_alloc_profile = BTRFS_BLOCK_GROUP_RAID1;
-		fs_info->system_alloc_profile = BTRFS_BLOCK_GROUP_RAID0;
+		fs_info->data_alloc_profile = BTRFS_BLOCK_GROUP_RAID0 |
+			BTRFS_BLOCK_GROUP_RAID1;
+		fs_info->metadata_alloc_profile = BTRFS_BLOCK_GROUP_RAID1 |
+			BTRFS_BLOCK_GROUP_DUP;
+		fs_info->system_alloc_profile = fs_info->metadata_alloc_profile;
 	}
 	mutex_unlock(&fs_info->fs_mutex);
 	return tree_root;
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 4ab98d8b73f..1885ec4280c 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -231,7 +231,7 @@ again:
 		if (start + num  > total_fs_bytes)
 			goto new_group;
 		if (!block_group_bits(cache, data)) {
-			printk("block group bits don't match %Lu %Lu\n", cache->flags, data);
+			printk("block group bits don't match %Lu %d\n", cache->flags, data);
 		}
 		*start_ret = start;
 		return 0;
@@ -1048,7 +1048,8 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags,
 static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
 {
 	u64 extra_flags = flags & (BTRFS_BLOCK_GROUP_RAID0 |
-				   BTRFS_BLOCK_GROUP_RAID1);
+				   BTRFS_BLOCK_GROUP_RAID1 |
+				   BTRFS_BLOCK_GROUP_DUP);
 	if (extra_flags) {
 		if (flags & BTRFS_BLOCK_GROUP_DATA)
 			fs_info->avail_data_alloc_bits |= extra_flags;
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index bc3c0b97588..b9294e3c05f 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -627,6 +627,7 @@ int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 	struct extent_map *em;
 	u64 physical;
 	u64 calc_size = 1024 * 1024 * 1024;
+	u64 min_free = calc_size;
 	u64 avail;
 	u64 max_avail = 0;
 	int num_stripes = 1;
@@ -641,6 +642,8 @@ int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 
 	if (type & (BTRFS_BLOCK_GROUP_RAID0))
 		num_stripes = btrfs_super_num_devices(&info->super_copy);
+	if (type & (BTRFS_BLOCK_GROUP_DUP))
+		num_stripes = 2;
 	if (type & (BTRFS_BLOCK_GROUP_RAID1)) {
 		num_stripes = min_t(u64, 2,
 				  btrfs_super_num_devices(&info->super_copy));
@@ -649,16 +652,23 @@ again:
 	INIT_LIST_HEAD(&private_devs);
 	cur = dev_list->next;
 	index = 0;
+
+	if (type & BTRFS_BLOCK_GROUP_DUP)
+		min_free = calc_size * 2;
+
 	/* build a private list of devices we will allocate from */
 	while(index < num_stripes) {
 		device = list_entry(cur, struct btrfs_device, dev_list);
+
 		avail = device->total_bytes - device->bytes_used;
 		cur = cur->next;
 		if (avail > max_avail)
 			max_avail = avail;
-		if (avail >= calc_size) {
+		if (avail >= min_free) {
 			list_move_tail(&device->dev_list, &private_devs);
 			index++;
+			if (type & BTRFS_BLOCK_GROUP_DUP)
+				index++;
 		}
 		if (cur == dev_list)
 			break;
@@ -689,17 +699,22 @@ again:
 
 	stripes = &chunk->stripe;
 
-	if (type & BTRFS_BLOCK_GROUP_RAID1)
+	if (type & (BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_DUP))
 		*num_bytes = calc_size;
 	else
 		*num_bytes = calc_size * num_stripes;
 
 	index = 0;
+printk("new chunk type %Lu start %Lu size %Lu\n", type, key.objectid, *num_bytes);
 	while(index < num_stripes) {
 		BUG_ON(list_empty(&private_devs));
 		cur = private_devs.next;
 		device = list_entry(cur, struct btrfs_device, dev_list);
-		list_move_tail(&device->dev_list, dev_list);
+
+		/* loop over this device again if we're doing a dup group */
+		if (!(type & BTRFS_BLOCK_GROUP_DUP) ||
+		    (index == num_stripes - 1))
+			list_move_tail(&device->dev_list, dev_list);
 
 		ret = btrfs_alloc_dev_extent(trans, device,
 					     key.objectid,
@@ -839,6 +854,14 @@ int btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
 			}
 			*total_devs = 1;
 		}
+	} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
+		if (rw == WRITE) {
+			*total_devs = map->num_stripes;
+			stripe_index = dev_nr;
+		} else {
+			stripe_index = 0;
+			*total_devs = 1;
+		}
 	} else {
 		/*
 		 * after this do_div call, stripe_nr is the number of stripes
@@ -851,7 +874,8 @@ int btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
 	*phys = map->stripes[stripe_index].physical + stripe_offset +
 		stripe_nr * map->stripe_len;
 
-	if (map->type & (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1)) {
+	if (map->type & (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1 |
+			 BTRFS_BLOCK_GROUP_DUP)) {
 		/* we limit the length of each bio to what fits in a stripe */
 		*length = min_t(u64, em->len - offset,
 			      map->stripe_len - stripe_offset);