Commit 0647bf56 authored by Wang Shilong's avatar Wang Shilong Committed by Chris Mason
Browse files

Btrfs: improve forever loop when doing balance relocation 



We hit a forever loop when doing balance relocation,the reason
is that we firstly reserve 4M(node size is 16k).and within transaction
we will try to add extra reservation for snapshot roots,this will
return -EAGAIN if there has been a thread flushing space to reserve
space.We will do this again and again with filesystem becoming nearly
full.

If the above '-EAGAIN' case happens, we try to refill reservation more
outsize of transaction, and this will return eariler in enospc case,however,
this dosen't really hurt because it makes no sense doing balance relocation
with the filesystem nearly full.

Miao Xie helped a lot to track this issue, thanks.

Signed-off-by: default avatarWang Shilong <wangsl.fnst@cn.fujitsu.com>
Signed-off-by: default avatarMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: default avatarJosef Bacik <jbacik@fb.com>
Signed-off-by: default avatarChris Mason <clm@fb.com>
parent 6126e3ca

fs/btrfs/relocation.c

+36 −38
Original line number Original line Diff line number Diff line
@@ -94,6 +94,7 @@ struct backref_edge { 




#define LOWER	0

#define LOWER	0

#define UPPER	1

#define UPPER	1

#define RELOCATION_RESERVED_NODES	256





struct backref_cache {

struct backref_cache {

	/* red black tree of all backref nodes in the cache */

	/* red black tree of all backref nodes in the cache */
@@ -176,6 +177,8 @@ struct reloc_control { 
	u64 merging_rsv_size;

	u64 merging_rsv_size;

	/* size of relocated tree nodes */

	/* size of relocated tree nodes */

	u64 nodes_relocated;

	u64 nodes_relocated;

	/* reserved size for block group relocation*/

	u64 reserved_bytes;





	u64 search_start;

	u64 search_start;

	u64 extents_found;

	u64 extents_found;
@@ -184,7 +187,6 @@ struct reloc_control { 
	unsigned int create_reloc_tree:1;

	unsigned int create_reloc_tree:1;

	unsigned int merge_reloc_tree:1;

	unsigned int merge_reloc_tree:1;

	unsigned int found_file_extent:1;

	unsigned int found_file_extent:1;

	unsigned int commit_transaction:1;

};

};





/* stages of data relocation */

/* stages of data relocation */
@@ -2590,28 +2592,36 @@ static int reserve_metadata_space(struct btrfs_trans_handle *trans, 
	struct btrfs_root *root = rc->extent_root;

	struct btrfs_root *root = rc->extent_root;

	u64 num_bytes;

	u64 num_bytes;

	int ret;

	int ret;

	u64 tmp;





	num_bytes = calcu_metadata_size(rc, node, 1) * 2;

	num_bytes = calcu_metadata_size(rc, node, 1) * 2;





	trans->block_rsv = rc->block_rsv;

	trans->block_rsv = rc->block_rsv;

	ret = btrfs_block_rsv_add(root, rc->block_rsv, num_bytes,

	rc->reserved_bytes += num_bytes;

	ret = btrfs_block_rsv_refill(root, rc->block_rsv, num_bytes,

				BTRFS_RESERVE_FLUSH_ALL);

				BTRFS_RESERVE_FLUSH_ALL);

	if (ret) {

	if (ret) {

		if (ret == -EAGAIN)

		if (ret == -EAGAIN) {

			rc->commit_transaction = 1;

			tmp = rc->extent_root->nodesize *

				RELOCATION_RESERVED_NODES;

			while (tmp <= rc->reserved_bytes)

				tmp <<= 1;

			/*

			 * only one thread can access block_rsv at this point,

			 * so we don't need hold lock to protect block_rsv.

			 * we expand more reservation size here to allow enough

			 * space for relocation and we will return eailer in

			 * enospc case.

			 */

			rc->block_rsv->size = tmp + rc->extent_root->nodesize *

					      RELOCATION_RESERVED_NODES;

		}

		return ret;

		return ret;

	}

	}





	return 0;

	return 0;

}

}





static void release_metadata_space(struct reloc_control *rc,

				   struct backref_node *node)

{

	u64 num_bytes = calcu_metadata_size(rc, node, 0) * 2;

	btrfs_block_rsv_release(rc->extent_root, rc->block_rsv, num_bytes);

}



/*

/*

 * relocate a block tree, and then update pointers in upper level

 * relocate a block tree, and then update pointers in upper level

 * blocks that reference the block to point to the new location.

 * blocks that reference the block to point to the new location.
@@ -2898,7 +2908,6 @@ static int relocate_tree_block(struct btrfs_trans_handle *trans, 
				struct btrfs_path *path)

				struct btrfs_path *path)

{

{

	struct btrfs_root *root;

	struct btrfs_root *root;

	int release = 0;

	int ret = 0;

	int ret = 0;





	if (!node)

	if (!node)
@@ -2915,7 +2924,6 @@ static int relocate_tree_block(struct btrfs_trans_handle *trans, 
		ret = reserve_metadata_space(trans, rc, node);

		ret = reserve_metadata_space(trans, rc, node);

		if (ret)

		if (ret)

			goto out;

			goto out;

		release = 1;

	}

	}





	if (root) {

	if (root) {
@@ -2940,11 +2948,8 @@ static int relocate_tree_block(struct btrfs_trans_handle *trans, 
		ret = do_relocation(trans, rc, node, key, path, 1);

		ret = do_relocation(trans, rc, node, key, path, 1);

	}

	}

out:

out:

	if (ret || node->level == 0 || node->cowonly) {

	if (ret || node->level == 0 || node->cowonly)

		if (release)

			release_metadata_space(rc, node);

		remove_backref_node(&rc->backref_cache, node);

		remove_backref_node(&rc->backref_cache, node);

	}

	return ret;

	return ret;

}

}
@@ -3867,29 +3872,20 @@ static noinline_for_stack 
int prepare_to_relocate(struct reloc_control *rc)

int prepare_to_relocate(struct reloc_control *rc)

{

{

	struct btrfs_trans_handle *trans;

	struct btrfs_trans_handle *trans;

	int ret;





	rc->block_rsv = btrfs_alloc_block_rsv(rc->extent_root,

	rc->block_rsv = btrfs_alloc_block_rsv(rc->extent_root,

					      BTRFS_BLOCK_RSV_TEMP);

					      BTRFS_BLOCK_RSV_TEMP);

	if (!rc->block_rsv)

	if (!rc->block_rsv)

		return -ENOMEM;

		return -ENOMEM;





	/*

	 * reserve some space for creating reloc trees.

	 * btrfs_init_reloc_root will use them when there

	 * is no reservation in transaction handle.

	 */

	ret = btrfs_block_rsv_add(rc->extent_root, rc->block_rsv,

				  rc->extent_root->nodesize * 256,

				  BTRFS_RESERVE_FLUSH_ALL);

	if (ret)

		return ret;



	memset(&rc->cluster, 0, sizeof(rc->cluster));

	memset(&rc->cluster, 0, sizeof(rc->cluster));

	rc->search_start = rc->block_group->key.objectid;

	rc->search_start = rc->block_group->key.objectid;

	rc->extents_found = 0;

	rc->extents_found = 0;

	rc->nodes_relocated = 0;

	rc->nodes_relocated = 0;

	rc->merging_rsv_size = 0;

	rc->merging_rsv_size = 0;

	rc->reserved_bytes = 0;

	rc->block_rsv->size = rc->extent_root->nodesize *

			      RELOCATION_RESERVED_NODES;





	rc->create_reloc_tree = 1;

	rc->create_reloc_tree = 1;

	set_reloc_control(rc);

	set_reloc_control(rc);
@@ -3933,6 +3929,14 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc) 
	}

	}





	while (1) {

	while (1) {

		rc->reserved_bytes = 0;

		ret = btrfs_block_rsv_refill(rc->extent_root,

					rc->block_rsv, rc->block_rsv->size,

					BTRFS_RESERVE_FLUSH_ALL);

		if (ret) {

			err = ret;

			break;

		}

		progress++;

		progress++;

		trans = btrfs_start_transaction(rc->extent_root, 0);

		trans = btrfs_start_transaction(rc->extent_root, 0);

		if (IS_ERR(trans)) {

		if (IS_ERR(trans)) {
@@ -4020,14 +4024,8 @@ restart: 
			}

			}

		}

		}





		if (rc->commit_transaction) {

			rc->commit_transaction = 0;

			ret = btrfs_commit_transaction(trans, rc->extent_root);

			BUG_ON(ret);

		} else {

		btrfs_end_transaction_throttle(trans, rc->extent_root);

		btrfs_end_transaction_throttle(trans, rc->extent_root);

		btrfs_btree_balance_dirty(rc->extent_root);

		btrfs_btree_balance_dirty(rc->extent_root);

		}

		trans = NULL;

		trans = NULL;





		if (rc->stage == MOVE_DATA_EXTENTS &&

		if (rc->stage == MOVE_DATA_EXTENTS &&

CRA Git | Maintained and supported by SUSTech CRA and CCSE