1 files changed, 70 insertions, 132 deletions

diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 8ff31f81d870..1c591d6eae58 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -74,8 +74,9 @@ enum {
 	RESERVE_ALLOC_NO_ACCOUNT = 2,
 };
 
-static int update_block_group(struct btrfs_root *root,
-			      u64 bytenr, u64 num_bytes, int alloc);
+static int update_block_group(struct btrfs_trans_handle *trans,
+			      struct btrfs_root *root, u64 bytenr,
+			      u64 num_bytes, int alloc);
 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root,
 				u64 bytenr, u64 num_bytes, u64 parent,
@@ -1889,8 +1890,8 @@ static int btrfs_issue_discard(struct block_device *bdev,
 	return blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_NOFS, 0);
 }
 
-static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
-				u64 num_bytes, u64 *actual_bytes)
+int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
+			 u64 num_bytes, u64 *actual_bytes)
 {
 	int ret;
 	u64 discarded_bytes = 0;
@@ -3139,9 +3140,11 @@ static int write_one_cache_group(struct btrfs_trans_handle *trans,
 	struct extent_buffer *leaf;
 
 	ret = btrfs_search_slot(trans, extent_root, &cache->key, path, 0, 1);
-	if (ret < 0)
+	if (ret) {
+		if (ret > 0)
+			ret = -ENOENT;
 		goto fail;
-	BUG_ON(ret); /* Corruption */
+	}
 
 	leaf = path->nodes[0];
 	bi = btrfs_item_ptr_offset(leaf, path->slots[0]);
@@ -3149,11 +3152,9 @@ static int write_one_cache_group(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_release_path(path);
 fail:
-	if (ret) {
+	if (ret)
 		btrfs_abort_transaction(trans, root, ret);
-		return ret;
-	}
-	return 0;
+	return ret;
 
 }
 
@@ -3315,120 +3316,42 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
 				   struct btrfs_root *root)
 {
 	struct btrfs_block_group_cache *cache;
-	int err = 0;
+	struct btrfs_transaction *cur_trans = trans->transaction;
+	int ret = 0;
 	struct btrfs_path *path;
-	u64 last = 0;
+
+	if (list_empty(&cur_trans->dirty_bgs))
+		return 0;
 
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
 
-again:
-	while (1) {
-		cache = btrfs_lookup_first_block_group(root->fs_info, last);
-		while (cache) {
-			if (cache->disk_cache_state == BTRFS_DC_CLEAR)
-				break;
-			cache = next_block_group(root, cache);
-		}
-		if (!cache) {
-			if (last == 0)
-				break;
-			last = 0;
-			continue;
-		}
-		err = cache_save_setup(cache, trans, path);
-		last = cache->key.objectid + cache->key.offset;
-		btrfs_put_block_group(cache);
-	}
-
-	while (1) {
-		if (last == 0) {
-			err = btrfs_run_delayed_refs(trans, root,
-						     (unsigned long)-1);
-			if (err) /* File system offline */
-				goto out;
-		}
-
-		cache = btrfs_lookup_first_block_group(root->fs_info, last);
-		while (cache) {
-			if (cache->disk_cache_state == BTRFS_DC_CLEAR) {
-				btrfs_put_block_group(cache);
-				goto again;
-			}
-
-			if (cache->dirty)
-				break;
-			cache = next_block_group(root, cache);
-		}
-		if (!cache) {
-			if (last == 0)
-				break;
-			last = 0;
-			continue;
-		}
-
-		if (cache->disk_cache_state == BTRFS_DC_SETUP)
-			cache->disk_cache_state = BTRFS_DC_NEED_WRITE;
-		cache->dirty = 0;
-		last = cache->key.objectid + cache->key.offset;
-
-		err = write_one_cache_group(trans, root, path, cache);
-		btrfs_put_block_group(cache);
-		if (err) /* File system offline */
-			goto out;
-	}
-
-	while (1) {
-		/*
-		 * I don't think this is needed since we're just marking our
-		 * preallocated extent as written, but just in case it can't
-		 * hurt.
-		 */
-		if (last == 0) {
-			err = btrfs_run_delayed_refs(trans, root,
-						     (unsigned long)-1);
-			if (err) /* File system offline */
-				goto out;
-		}
-
-		cache = btrfs_lookup_first_block_group(root->fs_info, last);
-		while (cache) {
-			/*
-			 * Really this shouldn't happen, but it could if we
-			 * couldn't write the entire preallocated extent and
-			 * splitting the extent resulted in a new block.
-			 */
-			if (cache->dirty) {
-				btrfs_put_block_group(cache);
-				goto again;
-			}
-			if (cache->disk_cache_state == BTRFS_DC_NEED_WRITE)
-				break;
-			cache = next_block_group(root, cache);
-		}
-		if (!cache) {
-			if (last == 0)
-				break;
-			last = 0;
-			continue;
-		}
-
-		err = btrfs_write_out_cache(root, trans, cache, path);
-
-		/*
-		 * If we didn't have an error then the cache state is still
-		 * NEED_WRITE, so we can set it to WRITTEN.
-		 */
-		if (!err && cache->disk_cache_state == BTRFS_DC_NEED_WRITE)
-			cache->disk_cache_state = BTRFS_DC_WRITTEN;
-		last = cache->key.objectid + cache->key.offset;
+	/*
+	 * We don't need the lock here since we are protected by the transaction
+	 * commit.  We want to do the cache_save_setup first and then run the
+	 * delayed refs to make sure we have the best chance at doing this all
+	 * in one shot.
+	 */
+	while (!list_empty(&cur_trans->dirty_bgs)) {
+		cache = list_first_entry(&cur_trans->dirty_bgs,
+					 struct btrfs_block_group_cache,
+					 dirty_list);
+		list_del_init(&cache->dirty_list);
+		if (cache->disk_cache_state == BTRFS_DC_CLEAR)
+			cache_save_setup(cache, trans, path);
+		if (!ret)
+			ret = btrfs_run_delayed_refs(trans, root,
+						     (unsigned long) -1);
+		if (!ret && cache->disk_cache_state == BTRFS_DC_SETUP)
+			btrfs_write_out_cache(root, trans, cache, path);
+		if (!ret)
+			ret = write_one_cache_group(trans, root, path, cache);
 		btrfs_put_block_group(cache);
 	}
-out:
 
 	btrfs_free_path(path);
-	return err;
+	return ret;
 }
 
 int btrfs_extent_readonly(struct btrfs_root *root, u64 bytenr)
@@ -5375,8 +5298,9 @@ void btrfs_delalloc_release_space(struct inode *inode, u64 num_bytes)
 	btrfs_free_reserved_data_space(inode, num_bytes);
 }
 
-static int update_block_group(struct btrfs_root *root,
-			      u64 bytenr, u64 num_bytes, int alloc)
+static int update_block_group(struct btrfs_trans_handle *trans,
+			      struct btrfs_root *root, u64 bytenr,
+			      u64 num_bytes, int alloc)
 {
 	struct btrfs_block_group_cache *cache = NULL;
 	struct btrfs_fs_info *info = root->fs_info;
@@ -5414,6 +5338,14 @@ static int update_block_group(struct btrfs_root *root,
 		if (!alloc && cache->cached == BTRFS_CACHE_NO)
 			cache_block_group(cache, 1);
 
+		spin_lock(&trans->transaction->dirty_bgs_lock);
+		if (list_empty(&cache->dirty_list)) {
+			list_add_tail(&cache->dirty_list,
+				      &trans->transaction->dirty_bgs);
+			btrfs_get_block_group(cache);
+		}
+		spin_unlock(&trans->transaction->dirty_bgs_lock);
+
 		byte_in_group = bytenr - cache->key.objectid;
 		WARN_ON(byte_in_group > cache->key.offset);
 
@@ -5424,7 +5356,6 @@ static int update_block_group(struct btrfs_root *root,
 		    cache->disk_cache_state < BTRFS_DC_CLEAR)
 			cache->disk_cache_state = BTRFS_DC_CLEAR;
 
-		cache->dirty = 1;
 		old_val = btrfs_block_group_used(&cache->item);
 		num_bytes = min(total, cache->key.offset - byte_in_group);
 		if (alloc) {
@@ -5727,7 +5658,8 @@ void btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans,
 	update_global_block_rsv(fs_info);
 }
 
-static int unpin_extent_range(struct btrfs_root *root, u64 start, u64 end)
+static int unpin_extent_range(struct btrfs_root *root, u64 start, u64 end,
+			      const bool return_free_space)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_block_group_cache *cache = NULL;
@@ -5751,7 +5683,8 @@ static int unpin_extent_range(struct btrfs_root *root, u64 start, u64 end)
 
 		if (start < cache->last_byte_to_unpin) {
 			len = min(len, cache->last_byte_to_unpin - start);
-			btrfs_add_free_space(cache, start, len);
+			if (return_free_space)
+				btrfs_add_free_space(cache, start, len);
 		}
 
 		start += len;
@@ -5815,7 +5748,7 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
 						   end + 1 - start, NULL);
 
 		clear_extent_dirty(unpin, start, end, GFP_NOFS);
-		unpin_extent_range(root, start, end);
+		unpin_extent_range(root, start, end, true);
 		cond_resched();
 	}
 
@@ -6101,7 +6034,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 			}
 		}
 
-		ret = update_block_group(root, bytenr, num_bytes, 0);
+		ret = update_block_group(trans, root, bytenr, num_bytes, 0);
 		if (ret) {
 			btrfs_abort_transaction(trans, extent_root, ret);
 			goto out;
@@ -7061,7 +6994,7 @@ static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
 	if (ret)
 		return ret;
 
-	ret = update_block_group(root, ins->objectid, ins->offset, 1);
+	ret = update_block_group(trans, root, ins->objectid, ins->offset, 1);
 	if (ret) { /* -ENOENT, logic error */
 		btrfs_err(fs_info, "update block group failed for %llu %llu",
 			ins->objectid, ins->offset);
@@ -7150,7 +7083,8 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 			return ret;
 	}
 
-	ret = update_block_group(root, ins->objectid, root->nodesize, 1);
+	ret = update_block_group(trans, root, ins->objectid, root->nodesize,
+				 1);
 	if (ret) { /* -ENOENT, logic error */
 		btrfs_err(fs_info, "update block group failed for %llu %llu",
 			ins->objectid, ins->offset);
@@ -8871,6 +8805,7 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 				       cache_node);
 		rb_erase(&block_group->cache_node,
 			 &info->block_group_cache_tree);
+		RB_CLEAR_NODE(&block_group->cache_node);
 		spin_unlock(&info->block_group_cache_lock);
 
 		down_write(&block_group->space_info->groups_sem);
@@ -9001,6 +8936,7 @@ btrfs_create_block_group_cache(struct btrfs_root *root, u64 start, u64 size)
 	INIT_LIST_HEAD(&cache->cluster_list);
 	INIT_LIST_HEAD(&cache->bg_list);
 	INIT_LIST_HEAD(&cache->ro_list);
+	INIT_LIST_HEAD(&cache->dirty_list);
 	btrfs_init_free_space_ctl(cache);
 	atomic_set(&cache->trimming, 0);
 
@@ -9064,9 +9000,8 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 			 * b) Setting 'dirty flag' makes sure that we flush
 			 *    the new space cache info onto disk.
 			 */
-			cache->disk_cache_state = BTRFS_DC_CLEAR;
 			if (btrfs_test_opt(root, SPACE_CACHE))
-				cache->dirty = 1;
+				cache->disk_cache_state = BTRFS_DC_CLEAR;
 		}
 
 		read_extent_buffer(leaf, &cache->item,
@@ -9129,6 +9064,7 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 			spin_lock(&info->block_group_cache_lock);
 			rb_erase(&cache->cache_node,
 				 &info->block_group_cache_tree);
+			RB_CLEAR_NODE(&cache->cache_node);
 			spin_unlock(&info->block_group_cache_lock);
 			btrfs_put_block_group(cache);
 			goto error;
@@ -9270,6 +9206,7 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 		spin_lock(&root->fs_info->block_group_cache_lock);
 		rb_erase(&cache->cache_node,
 			 &root->fs_info->block_group_cache_tree);
+		RB_CLEAR_NODE(&cache->cache_node);
 		spin_unlock(&root->fs_info->block_group_cache_lock);
 		btrfs_put_block_group(cache);
 		return ret;
@@ -9455,6 +9392,13 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 		}
 	}
 
+	spin_lock(&trans->transaction->dirty_bgs_lock);
+	if (!list_empty(&block_group->dirty_list)) {
+		list_del_init(&block_group->dirty_list);
+		btrfs_put_block_group(block_group);
+	}
+	spin_unlock(&trans->transaction->dirty_bgs_lock);
+
 	btrfs_remove_free_space_cache(block_group);
 
 	spin_lock(&block_group->space_info->lock);
@@ -9689,13 +9633,7 @@ out:
 
 int btrfs_error_unpin_extent_range(struct btrfs_root *root, u64 start, u64 end)
 {
-	return unpin_extent_range(root, start, end);
-}
-
-int btrfs_error_discard_extent(struct btrfs_root *root, u64 bytenr,
-			       u64 num_bytes, u64 *actual_bytes)
-{
-	return btrfs_discard_extent(root, bytenr, num_bytes, actual_bytes);
+	return unpin_extent_range(root, start, end, false);
 }
 
 int btrfs_trim_fs(struct btrfs_root *root, struct fstrim_range *range)