1 files changed, 139 insertions, 120 deletions

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 82324c327a50..6e5dc68ff661 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -504,13 +504,20 @@ static void fragment_free_space(struct btrfs_block_group *block_group)
 #endif
 
 /*
- * This is only called by btrfs_cache_block_group, since we could have freed
- * extents we need to check the pinned_extents for any extents that can't be
- * used yet since their free space will be released as soon as the transaction
- * commits.
+ * Add a free space range to the in memory free space cache of a block group.
+ * This checks if the range contains super block locations and any such
+ * locations are not added to the free space cache.
+ *
+ * @block_group:      The target block group.
+ * @start:            Start offset of the range.
+ * @end:              End offset of the range (exclusive).
+ * @total_added_ret:  Optional pointer to return the total amount of space
+ *                    added to the block group's free space cache.
+ *
+ * Returns 0 on success or < 0 on error.
  */
-int add_new_free_space(struct btrfs_block_group *block_group, u64 start, u64 end,
-		       u64 *total_added_ret)
+int btrfs_add_new_free_space(struct btrfs_block_group *block_group, u64 start,
+			     u64 end, u64 *total_added_ret)
 {
 	struct btrfs_fs_info *info = block_group->fs_info;
 	u64 extent_start, extent_end, size;
@@ -520,11 +527,10 @@ int add_new_free_space(struct btrfs_block_group *block_group, u64 start, u64 end
 		*total_added_ret = 0;
 
 	while (start < end) {
-		ret = find_first_extent_bit(&info->excluded_extents, start,
-					    &extent_start, &extent_end,
-					    EXTENT_DIRTY | EXTENT_UPTODATE,
-					    NULL);
-		if (ret)
+		if (!find_first_extent_bit(&info->excluded_extents, start,
+					   &extent_start, &extent_end,
+					   EXTENT_DIRTY | EXTENT_UPTODATE,
+					   NULL))
 			break;
 
 		if (extent_start <= start) {
@@ -799,8 +805,8 @@ next:
 		    key.type == BTRFS_METADATA_ITEM_KEY) {
 			u64 space_added;
 
-			ret = add_new_free_space(block_group, last, key.objectid,
-						 &space_added);
+			ret = btrfs_add_new_free_space(block_group, last,
+						       key.objectid, &space_added);
 			if (ret)
 				goto out;
 			total_found += space_added;
@@ -821,14 +827,20 @@ next:
 		path->slots[0]++;
 	}
 
-	ret = add_new_free_space(block_group, last,
-				 block_group->start + block_group->length,
-				 NULL);
+	ret = btrfs_add_new_free_space(block_group, last,
+				       block_group->start + block_group->length,
+				       NULL);
 out:
 	btrfs_free_path(path);
 	return ret;
 }
 
+static inline void btrfs_free_excluded_extents(const struct btrfs_block_group *bg)
+{
+	clear_extent_bits(&bg->fs_info->excluded_extents, bg->start,
+			  bg->start + bg->length - 1, EXTENT_UPTODATE);
+}
+
 static noinline void caching_thread(struct btrfs_work *work)
 {
 	struct btrfs_block_group *block_group;
@@ -923,7 +935,7 @@ int btrfs_cache_block_group(struct btrfs_block_group *cache, bool wait)
 	caching_ctl->block_group = cache;
 	refcount_set(&caching_ctl->count, 2);
 	atomic_set(&caching_ctl->progress, 0);
-	btrfs_init_work(&caching_ctl->work, caching_thread, NULL, NULL);
+	btrfs_init_work(&caching_ctl->work, caching_thread, NULL);
 
 	spin_lock(&cache->lock);
 	if (cache->cached != BTRFS_CACHE_NO) {
@@ -1274,7 +1286,7 @@ out:
 	/* Once for the lookup reference */
 	btrfs_put_block_group(block_group);
 	if (remove_rsv)
-		btrfs_delayed_refs_rsv_release(fs_info, 1);
+		btrfs_dec_delayed_refs_rsv_bg_updates(fs_info);
 	btrfs_free_path(path);
 	return ret;
 }
@@ -2098,8 +2110,9 @@ static int exclude_super_stripes(struct btrfs_block_group *cache)
 	if (cache->start < BTRFS_SUPER_INFO_OFFSET) {
 		stripe_len = BTRFS_SUPER_INFO_OFFSET - cache->start;
 		cache->bytes_super += stripe_len;
-		ret = btrfs_add_excluded_extent(fs_info, cache->start,
-						stripe_len);
+		ret = set_extent_bit(&fs_info->excluded_extents, cache->start,
+				     cache->start + stripe_len - 1,
+				     EXTENT_UPTODATE, NULL);
 		if (ret)
 			return ret;
 	}
@@ -2125,8 +2138,9 @@ static int exclude_super_stripes(struct btrfs_block_group *cache)
 				cache->start + cache->length - logical[nr]);
 
 			cache->bytes_super += len;
-			ret = btrfs_add_excluded_extent(fs_info, logical[nr],
-							len);
+			ret = set_extent_bit(&fs_info->excluded_extents, logical[nr],
+					     logical[nr] + len - 1,
+					     EXTENT_UPTODATE, NULL);
 			if (ret) {
 				kfree(logical);
 				return ret;
@@ -2319,8 +2333,8 @@ static int read_one_block_group(struct btrfs_fs_info *info,
 		btrfs_free_excluded_extents(cache);
 	} else if (cache->used == 0) {
 		cache->cached = BTRFS_CACHE_FINISHED;
-		ret = add_new_free_space(cache, cache->start,
-					 cache->start + cache->length, NULL);
+		ret = btrfs_add_new_free_space(cache, cache->start,
+					       cache->start + cache->length, NULL);
 		btrfs_free_excluded_extents(cache);
 		if (ret)
 			goto error;
@@ -2587,7 +2601,7 @@ static int insert_dev_extent(struct btrfs_trans_handle *trans,
 	btrfs_set_dev_extent_chunk_offset(leaf, extent, chunk_offset);
 
 	btrfs_set_dev_extent_length(leaf, extent, num_bytes);
-	btrfs_mark_buffer_dirty(leaf);
+	btrfs_mark_buffer_dirty(trans, leaf);
 out:
 	btrfs_free_path(path);
 	return ret;
@@ -2695,7 +2709,7 @@ void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans)
 
 		/* Already aborted the transaction if it failed. */
 next:
-		btrfs_delayed_refs_rsv_release(fs_info, 1);
+		btrfs_dec_delayed_refs_rsv_bg_inserts(fs_info);
 		list_del_init(&block_group->bg_list);
 		clear_bit(BLOCK_GROUP_FLAG_NEW, &block_group->runtime_flags);
 	}
@@ -2767,7 +2781,7 @@ struct btrfs_block_group *btrfs_make_block_group(struct btrfs_trans_handle *tran
 		return ERR_PTR(ret);
 	}
 
-	ret = add_new_free_space(cache, chunk_offset, chunk_offset + size, NULL);
+	ret = btrfs_add_new_free_space(cache, chunk_offset, chunk_offset + size, NULL);
 	btrfs_free_excluded_extents(cache);
 	if (ret) {
 		btrfs_put_block_group(cache);
@@ -2805,8 +2819,7 @@ struct btrfs_block_group *btrfs_make_block_group(struct btrfs_trans_handle *tran
 #endif
 
 	list_add_tail(&cache->bg_list, &trans->new_bgs);
-	trans->delayed_ref_updates++;
-	btrfs_update_delayed_refs_rsv(trans);
+	btrfs_inc_delayed_refs_rsv_bg_inserts(fs_info);
 
 	set_avail_alloc_bits(fs_info, type);
 	return cache;
@@ -3011,11 +3024,19 @@ static int update_block_group_item(struct btrfs_trans_handle *trans,
 						   cache->global_root_id);
 	btrfs_set_stack_block_group_flags(&bgi, cache->flags);
 	write_extent_buffer(leaf, &bgi, bi, sizeof(bgi));
-	btrfs_mark_buffer_dirty(leaf);
+	btrfs_mark_buffer_dirty(trans, leaf);
 fail:
 	btrfs_release_path(path);
-	/* We didn't update the block group item, need to revert @commit_used. */
-	if (ret < 0) {
+	/*
+	 * We didn't update the block group item, need to revert commit_used
+	 * unless the block group item didn't exist yet - this is to prevent a
+	 * race with a concurrent insertion of the block group item, with
+	 * insert_block_group_item(), that happened just after we attempted to
+	 * update. In that case we would reset commit_used to 0 just after the
+	 * insertion set it to a value greater than 0 - if the block group later
+	 * becomes with 0 used bytes, we would incorrectly skip its update.
+	 */
+	if (ret < 0 && ret != -ENOENT) {
 		spin_lock(&cache->lock);
 		cache->commit_used = old_commit_used;
 		spin_unlock(&cache->lock);
@@ -3029,7 +3050,6 @@ static int cache_save_setup(struct btrfs_block_group *block_group,
 			    struct btrfs_path *path)
 {
 	struct btrfs_fs_info *fs_info = block_group->fs_info;
-	struct btrfs_root *root = fs_info->tree_root;
 	struct inode *inode = NULL;
 	struct extent_changeset *data_reserved = NULL;
 	u64 alloc_hint = 0;
@@ -3081,7 +3101,7 @@ again:
 	 * time.
 	 */
 	BTRFS_I(inode)->generation = 0;
-	ret = btrfs_update_inode(trans, root, BTRFS_I(inode));
+	ret = btrfs_update_inode(trans, BTRFS_I(inode));
 	if (ret) {
 		/*
 		 * So theoretically we could recover from this, simply set the
@@ -3348,7 +3368,7 @@ again:
 		if (should_put)
 			btrfs_put_block_group(cache);
 		if (drop_reserve)
-			btrfs_delayed_refs_rsv_release(fs_info, 1);
+			btrfs_dec_delayed_refs_rsv_bg_updates(fs_info);
 		/*
 		 * Avoid blocking other tasks for too long. It might even save
 		 * us from writing caches for block groups that are going to be
@@ -3452,8 +3472,7 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
 		cache_save_setup(cache, trans, path);
 
 		if (!ret)
-			ret = btrfs_run_delayed_refs(trans,
-						     (unsigned long) -1);
+			ret = btrfs_run_delayed_refs(trans, U64_MAX);
 
 		if (!ret && cache->disk_cache_state == BTRFS_DC_SETUP) {
 			cache->io_ctl.inode = NULL;
@@ -3496,7 +3515,7 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
 		/* If its not on the io list, we need to put the block group */
 		if (should_put)
 			btrfs_put_block_group(cache);
-		btrfs_delayed_refs_rsv_release(fs_info, 1);
+		btrfs_dec_delayed_refs_rsv_bg_updates(fs_info);
 		spin_lock(&cur_trans->dirty_bgs_lock);
 	}
 	spin_unlock(&cur_trans->dirty_bgs_lock);
@@ -3521,12 +3540,12 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 			     u64 bytenr, u64 num_bytes, bool alloc)
 {
 	struct btrfs_fs_info *info = trans->fs_info;
-	struct btrfs_block_group *cache = NULL;
-	u64 total = num_bytes;
+	struct btrfs_space_info *space_info;
+	struct btrfs_block_group *cache;
 	u64 old_val;
-	u64 byte_in_group;
+	bool reclaim = false;
+	bool bg_already_dirty = true;
 	int factor;
-	int ret = 0;
 
 	/* Block accounting for super block */
 	spin_lock(&info->delalloc_root_lock);
@@ -3538,97 +3557,86 @@ int btrfs_update_block_group(struct btrfs_trans_handle *trans,
 	btrfs_set_super_bytes_used(info->super_copy, old_val);
 	spin_unlock(&info->delalloc_root_lock);
 
-	while (total) {
-		struct btrfs_space_info *space_info;
-		bool reclaim = false;
-
-		cache = btrfs_lookup_block_group(info, bytenr);
-		if (!cache) {
-			ret = -ENOENT;
-			break;
-		}
-		space_info = cache->space_info;
-		factor = btrfs_bg_type_to_factor(cache->flags);
+	cache = btrfs_lookup_block_group(info, bytenr);
+	if (!cache)
+		return -ENOENT;
 
-		/*
-		 * If this block group has free space cache written out, we
-		 * need to make sure to load it if we are removing space.  This
-		 * is because we need the unpinning stage to actually add the
-		 * space back to the block group, otherwise we will leak space.
-		 */
-		if (!alloc && !btrfs_block_group_done(cache))
-			btrfs_cache_block_group(cache, true);
+	/* An extent can not span multiple block groups. */
+	ASSERT(bytenr + num_bytes <= cache->start + cache->length);
 
-		byte_in_group = bytenr - cache->start;
-		WARN_ON(byte_in_group > cache->length);
+	space_info = cache->space_info;
+	factor = btrfs_bg_type_to_factor(cache->flags);
 
-		spin_lock(&space_info->lock);
-		spin_lock(&cache->lock);
+	/*
+	 * If this block group has free space cache written out, we need to make
+	 * sure to load it if we are removing space.  This is because we need
+	 * the unpinning stage to actually add the space back to the block group,
+	 * otherwise we will leak space.
+	 */
+	if (!alloc && !btrfs_block_group_done(cache))
+		btrfs_cache_block_group(cache, true);
 
-		if (btrfs_test_opt(info, SPACE_CACHE) &&
-		    cache->disk_cache_state < BTRFS_DC_CLEAR)
-			cache->disk_cache_state = BTRFS_DC_CLEAR;
+	spin_lock(&space_info->lock);
+	spin_lock(&cache->lock);
 
-		old_val = cache->used;
-		num_bytes = min(total, cache->length - byte_in_group);
-		if (alloc) {
-			old_val += num_bytes;
-			cache->used = old_val;
-			cache->reserved -= num_bytes;
-			space_info->bytes_reserved -= num_bytes;
-			space_info->bytes_used += num_bytes;
-			space_info->disk_used += num_bytes * factor;
-			spin_unlock(&cache->lock);
-			spin_unlock(&space_info->lock);
-		} else {
-			old_val -= num_bytes;
-			cache->used = old_val;
-			cache->pinned += num_bytes;
-			btrfs_space_info_update_bytes_pinned(info, space_info,
-							     num_bytes);
-			space_info->bytes_used -= num_bytes;
-			space_info->disk_used -= num_bytes * factor;
+	if (btrfs_test_opt(info, SPACE_CACHE) &&
+	    cache->disk_cache_state < BTRFS_DC_CLEAR)
+		cache->disk_cache_state = BTRFS_DC_CLEAR;
 
-			reclaim = should_reclaim_block_group(cache, num_bytes);
+	old_val = cache->used;
+	if (alloc) {
+		old_val += num_bytes;
+		cache->used = old_val;
+		cache->reserved -= num_bytes;
+		space_info->bytes_reserved -= num_bytes;
+		space_info->bytes_used += num_bytes;
+		space_info->disk_used += num_bytes * factor;
+		spin_unlock(&cache->lock);
+		spin_unlock(&space_info->lock);
+	} else {
+		old_val -= num_bytes;
+		cache->used = old_val;
+		cache->pinned += num_bytes;
+		btrfs_space_info_update_bytes_pinned(info, space_info, num_bytes);
+		space_info->bytes_used -= num_bytes;
+		space_info->disk_used -= num_bytes * factor;
 
-			spin_unlock(&cache->lock);
-			spin_unlock(&space_info->lock);
+		reclaim = should_reclaim_block_group(cache, num_bytes);
 
-			set_extent_bit(&trans->transaction->pinned_extents,
-				       bytenr, bytenr + num_bytes - 1,
-				       EXTENT_DIRTY, NULL);
-		}
+		spin_unlock(&cache->lock);
+		spin_unlock(&space_info->lock);
 
-		spin_lock(&trans->transaction->dirty_bgs_lock);
-		if (list_empty(&cache->dirty_list)) {
-			list_add_tail(&cache->dirty_list,
-				      &trans->transaction->dirty_bgs);
-			trans->delayed_ref_updates++;
-			btrfs_get_block_group(cache);
-		}
-		spin_unlock(&trans->transaction->dirty_bgs_lock);
+		set_extent_bit(&trans->transaction->pinned_extents, bytenr,
+			       bytenr + num_bytes - 1, EXTENT_DIRTY, NULL);
+	}
 
-		/*
-		 * No longer have used bytes in this block group, queue it for
-		 * deletion. We do this after adding the block group to the
-		 * dirty list to avoid races between cleaner kthread and space
-		 * cache writeout.
-		 */
-		if (!alloc && old_val == 0) {
-			if (!btrfs_test_opt(info, DISCARD_ASYNC))
-				btrfs_mark_bg_unused(cache);
-		} else if (!alloc && reclaim) {
-			btrfs_mark_bg_to_reclaim(cache);
-		}
+	spin_lock(&trans->transaction->dirty_bgs_lock);
+	if (list_empty(&cache->dirty_list)) {
+		list_add_tail(&cache->dirty_list, &trans->transaction->dirty_bgs);
+		bg_already_dirty = false;
+		btrfs_get_block_group(cache);
+	}
+	spin_unlock(&trans->transaction->dirty_bgs_lock);
 
-		btrfs_put_block_group(cache);
-		total -= num_bytes;
-		bytenr += num_bytes;
+	/*
+	 * No longer have used bytes in this block group, queue it for deletion.
+	 * We do this after adding the block group to the dirty list to avoid
+	 * races between cleaner kthread and space cache writeout.
+	 */
+	if (!alloc && old_val == 0) {
+		if (!btrfs_test_opt(info, DISCARD_ASYNC))
+			btrfs_mark_bg_unused(cache);
+	} else if (!alloc && reclaim) {
+		btrfs_mark_bg_to_reclaim(cache);
 	}
 
+	btrfs_put_block_group(cache);
+
 	/* Modified block groups are accounted for in the delayed_refs_rsv. */
-	btrfs_update_delayed_refs_rsv(trans);
-	return ret;
+	if (!bg_already_dirty)
+		btrfs_inc_delayed_refs_rsv_bg_updates(info);
+
+	return 0;
 }
 
 /*
@@ -4075,7 +4083,7 @@ int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
 
 	if (IS_ERR(ret_bg)) {
 		ret = PTR_ERR(ret_bg);
-	} else if (from_extent_allocation) {
+	} else if (from_extent_allocation && (flags & BTRFS_BLOCK_GROUP_DATA)) {
 		/*
 		 * New block group is likely to be used soon. Try to activate
 		 * it now. Failure is OK for now.
@@ -4273,6 +4281,17 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 	struct btrfs_caching_control *caching_ctl;
 	struct rb_node *n;
 
+	if (btrfs_is_zoned(info)) {
+		if (info->active_meta_bg) {
+			btrfs_put_block_group(info->active_meta_bg);
+			info->active_meta_bg = NULL;
+		}
+		if (info->active_system_bg) {
+			btrfs_put_block_group(info->active_system_bg);
+			info->active_system_bg = NULL;
+		}
+	}
+
 	write_lock(&info->block_group_cache_lock);
 	while (!list_empty(&info->caching_block_groups)) {
 		caching_ctl = list_entry(info->caching_block_groups.next,