[PATCH 00/15] Migrate the block group code into it's own file

[PATCH 00/15] Migrate the block group code into it's own file

[Josef Bacik]

This is the rebased set of the much larger group of patches I sent last month.
The first 10 patches are already merged, these just didn't apply cleanly.  I
went through and applied each one, deleted and re-copied anything that didn't
merge cleanly, and compiled between each patch to make sure everything was
kosher.  This series just moves code around with the goal of making
extent-tree.c smaller.  I made no other changes other than moving code around, I
want to keep cleanups a separate thing _after_ we move the code around.  Thanks,

Josef

[PATCH 01/15] btrfs: migrate the block group caching code

[Josef Bacik]

We can now just copy it over to block-group.c.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.c | 463 +++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/block-group.h |   2 +
 fs/btrfs/ctree.h       |   2 -
 fs/btrfs/extent-tree.c | 459 ----------------------------------------
 4 files changed, 465 insertions(+), 461 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 1f3afa0b42ba..e8c29045e1b9 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -3,6 +3,9 @@
 #include "ctree.h"
 #include "block-group.h"
 #include "space-info.h"
+#include "disk-io.h"
+#include "free-space-cache.h"
+#include "free-space-tree.h"
 
 void btrfs_get_block_group(struct btrfs_block_group_cache *cache)
 {
@@ -200,3 +203,463 @@ void btrfs_wait_block_group_reservations(struct btrfs_block_group_cache *bg)
 
 	wait_var_event(&bg->reservations, !atomic_read(&bg->reservations));
 }
+
+struct btrfs_caching_control *btrfs_get_caching_control(
+		struct btrfs_block_group_cache *cache)
+{
+	struct btrfs_caching_control *ctl;
+
+	spin_lock(&cache->lock);
+	if (!cache->caching_ctl) {
+		spin_unlock(&cache->lock);
+		return NULL;
+	}
+
+	ctl = cache->caching_ctl;
+	refcount_inc(&ctl->count);
+	spin_unlock(&cache->lock);
+	return ctl;
+}
+
+void btrfs_put_caching_control(struct btrfs_caching_control *ctl)
+{
+	if (refcount_dec_and_test(&ctl->count))
+		kfree(ctl);
+}
+
+/*
+ * when we wait for progress in the block group caching, its because
+ * our allocation attempt failed at least once.  So, we must sleep
+ * and let some progress happen before we try again.
+ *
+ * This function will sleep at least once waiting for new free space to
+ * show up, and then it will check the block group free space numbers
+ * for our min num_bytes.  Another option is to have it go ahead
+ * and look in the rbtree for a free extent of a given size, but this
+ * is a good start.
+ *
+ * Callers of this must check if cache->cached == BTRFS_CACHE_ERROR before using
+ * any of the information in this block group.
+ */
+void
+btrfs_wait_block_group_cache_progress(struct btrfs_block_group_cache *cache,
+				      u64 num_bytes)
+{
+	struct btrfs_caching_control *caching_ctl;
+
+	caching_ctl = btrfs_get_caching_control(cache);
+	if (!caching_ctl)
+		return;
+
+	wait_event(caching_ctl->wait, btrfs_block_group_cache_done(cache) ||
+		   (cache->free_space_ctl->free_space >= num_bytes));
+
+	btrfs_put_caching_control(caching_ctl);
+}
+
+int btrfs_wait_block_group_cache_done(struct btrfs_block_group_cache *cache)
+{
+	struct btrfs_caching_control *caching_ctl;
+	int ret = 0;
+
+	caching_ctl = btrfs_get_caching_control(cache);
+	if (!caching_ctl)
+		return (cache->cached == BTRFS_CACHE_ERROR) ? -EIO : 0;
+
+	wait_event(caching_ctl->wait, btrfs_block_group_cache_done(cache));
+	if (cache->cached == BTRFS_CACHE_ERROR)
+		ret = -EIO;
+	btrfs_put_caching_control(caching_ctl);
+	return ret;
+}
+
+#ifdef CONFIG_BTRFS_DEBUG
+void btrfs_fragment_free_space(struct btrfs_block_group_cache *block_group)
+{
+	struct btrfs_fs_info *fs_info = block_group->fs_info;
+	u64 start = block_group->key.objectid;
+	u64 len = block_group->key.offset;
+	u64 chunk = block_group->flags & BTRFS_BLOCK_GROUP_METADATA ?
+		fs_info->nodesize : fs_info->sectorsize;
+	u64 step = chunk << 1;
+
+	while (len > chunk) {
+		btrfs_remove_free_space(block_group, start, chunk);
+		start += step;
+		if (len < step)
+			len = 0;
+		else
+			len -= step;
+	}
+}
+#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.
+ */
+u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
+		       u64 start, u64 end)
+{
+	struct btrfs_fs_info *info = block_group->fs_info;
+	u64 extent_start, extent_end, size, total_added = 0;
+	int ret;
+
+	while (start < end) {
+		ret = find_first_extent_bit(info->pinned_extents, start,
+					    &extent_start, &extent_end,
+					    EXTENT_DIRTY | EXTENT_UPTODATE,
+					    NULL);
+		if (ret)
+			break;
+
+		if (extent_start <= start) {
+			start = extent_end + 1;
+		} else if (extent_start > start && extent_start < end) {
+			size = extent_start - start;
+			total_added += size;
+			ret = btrfs_add_free_space(block_group, start,
+						   size);
+			BUG_ON(ret); /* -ENOMEM or logic error */
+			start = extent_end + 1;
+		} else {
+			break;
+		}
+	}
+
+	if (start < end) {
+		size = end - start;
+		total_added += size;
+		ret = btrfs_add_free_space(block_group, start, size);
+		BUG_ON(ret); /* -ENOMEM or logic error */
+	}
+
+	return total_added;
+}
+
+static int load_extent_tree_free(struct btrfs_caching_control *caching_ctl)
+{
+	struct btrfs_block_group_cache *block_group = caching_ctl->block_group;
+	struct btrfs_fs_info *fs_info = block_group->fs_info;
+	struct btrfs_root *extent_root = fs_info->extent_root;
+	struct btrfs_path *path;
+	struct extent_buffer *leaf;
+	struct btrfs_key key;
+	u64 total_found = 0;
+	u64 last = 0;
+	u32 nritems;
+	int ret;
+	bool wakeup = true;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	last = max_t(u64, block_group->key.objectid, BTRFS_SUPER_INFO_OFFSET);
+
+#ifdef CONFIG_BTRFS_DEBUG
+	/*
+	 * If we're fragmenting we don't want to make anybody think we can
+	 * allocate from this block group until we've had a chance to fragment
+	 * the free space.
+	 */
+	if (btrfs_should_fragment_free_space(block_group))
+		wakeup = false;
+#endif
+	/*
+	 * We don't want to deadlock with somebody trying to allocate a new
+	 * extent for the extent root while also trying to search the extent
+	 * root to add free space.  So we skip locking and search the commit
+	 * root, since its read-only
+	 */
+	path->skip_locking = 1;
+	path->search_commit_root = 1;
+	path->reada = READA_FORWARD;
+
+	key.objectid = last;
+	key.offset = 0;
+	key.type = BTRFS_EXTENT_ITEM_KEY;
+
+next:
+	ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
+	if (ret < 0)
+		goto out;
+
+	leaf = path->nodes[0];
+	nritems = btrfs_header_nritems(leaf);
+
+	while (1) {
+		if (btrfs_fs_closing(fs_info) > 1) {
+			last = (u64)-1;
+			break;
+		}
+
+		if (path->slots[0] < nritems) {
+			btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+		} else {
+			ret = btrfs_find_next_key(extent_root, path, &key, 0,
+						  0);
+			if (ret)
+				break;
+
+			if (need_resched() ||
+			    rwsem_is_contended(&fs_info->commit_root_sem)) {
+				if (wakeup)
+					caching_ctl->progress = last;
+				btrfs_release_path(path);
+				up_read(&fs_info->commit_root_sem);
+				mutex_unlock(&caching_ctl->mutex);
+				cond_resched();
+				mutex_lock(&caching_ctl->mutex);
+				down_read(&fs_info->commit_root_sem);
+				goto next;
+			}
+
+			ret = btrfs_next_leaf(extent_root, path);
+			if (ret < 0)
+				goto out;
+			if (ret)
+				break;
+			leaf = path->nodes[0];
+			nritems = btrfs_header_nritems(leaf);
+			continue;
+		}
+
+		if (key.objectid < last) {
+			key.objectid = last;
+			key.offset = 0;
+			key.type = BTRFS_EXTENT_ITEM_KEY;
+
+			if (wakeup)
+				caching_ctl->progress = last;
+			btrfs_release_path(path);
+			goto next;
+		}
+
+		if (key.objectid < block_group->key.objectid) {
+			path->slots[0]++;
+			continue;
+		}
+
+		if (key.objectid >= block_group->key.objectid +
+		    block_group->key.offset)
+			break;
+
+		if (key.type == BTRFS_EXTENT_ITEM_KEY ||
+		    key.type == BTRFS_METADATA_ITEM_KEY) {
+			total_found += add_new_free_space(block_group, last,
+							  key.objectid);
+			if (key.type == BTRFS_METADATA_ITEM_KEY)
+				last = key.objectid +
+					fs_info->nodesize;
+			else
+				last = key.objectid + key.offset;
+
+			if (total_found > CACHING_CTL_WAKE_UP) {
+				total_found = 0;
+				if (wakeup)
+					wake_up(&caching_ctl->wait);
+			}
+		}
+		path->slots[0]++;
+	}
+	ret = 0;
+
+	total_found += add_new_free_space(block_group, last,
+					  block_group->key.objectid +
+					  block_group->key.offset);
+	caching_ctl->progress = (u64)-1;
+
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
+static noinline void caching_thread(struct btrfs_work *work)
+{
+	struct btrfs_block_group_cache *block_group;
+	struct btrfs_fs_info *fs_info;
+	struct btrfs_caching_control *caching_ctl;
+	int ret;
+
+	caching_ctl = container_of(work, struct btrfs_caching_control, work);
+	block_group = caching_ctl->block_group;
+	fs_info = block_group->fs_info;
+
+	mutex_lock(&caching_ctl->mutex);
+	down_read(&fs_info->commit_root_sem);
+
+	if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE))
+		ret = load_free_space_tree(caching_ctl);
+	else
+		ret = load_extent_tree_free(caching_ctl);
+
+	spin_lock(&block_group->lock);
+	block_group->caching_ctl = NULL;
+	block_group->cached = ret ? BTRFS_CACHE_ERROR : BTRFS_CACHE_FINISHED;
+	spin_unlock(&block_group->lock);
+
+#ifdef CONFIG_BTRFS_DEBUG
+	if (btrfs_should_fragment_free_space(block_group)) {
+		u64 bytes_used;
+
+		spin_lock(&block_group->space_info->lock);
+		spin_lock(&block_group->lock);
+		bytes_used = block_group->key.offset -
+			btrfs_block_group_used(&block_group->item);
+		block_group->space_info->bytes_used += bytes_used >> 1;
+		spin_unlock(&block_group->lock);
+		spin_unlock(&block_group->space_info->lock);
+		btrfs_fragment_free_space(block_group);
+	}
+#endif
+
+	caching_ctl->progress = (u64)-1;
+
+	up_read(&fs_info->commit_root_sem);
+	btrfs_free_excluded_extents(block_group);
+	mutex_unlock(&caching_ctl->mutex);
+
+	wake_up(&caching_ctl->wait);
+
+	btrfs_put_caching_control(caching_ctl);
+	btrfs_put_block_group(block_group);
+}
+
+int btrfs_cache_block_group(struct btrfs_block_group_cache *cache,
+			    int load_cache_only)
+{
+	DEFINE_WAIT(wait);
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	struct btrfs_caching_control *caching_ctl;
+	int ret = 0;
+
+	caching_ctl = kzalloc(sizeof(*caching_ctl), GFP_NOFS);
+	if (!caching_ctl)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&caching_ctl->list);
+	mutex_init(&caching_ctl->mutex);
+	init_waitqueue_head(&caching_ctl->wait);
+	caching_ctl->block_group = cache;
+	caching_ctl->progress = cache->key.objectid;
+	refcount_set(&caching_ctl->count, 1);
+	btrfs_init_work(&caching_ctl->work, btrfs_cache_helper,
+			caching_thread, NULL, NULL);
+
+	spin_lock(&cache->lock);
+	/*
+	 * This should be a rare occasion, but this could happen I think in the
+	 * case where one thread starts to load the space cache info, and then
+	 * some other thread starts a transaction commit which tries to do an
+	 * allocation while the other thread is still loading the space cache
+	 * info.  The previous loop should have kept us from choosing this block
+	 * group, but if we've moved to the state where we will wait on caching
+	 * block groups we need to first check if we're doing a fast load here,
+	 * so we can wait for it to finish, otherwise we could end up allocating
+	 * from a block group who's cache gets evicted for one reason or
+	 * another.
+	 */
+	while (cache->cached == BTRFS_CACHE_FAST) {
+		struct btrfs_caching_control *ctl;
+
+		ctl = cache->caching_ctl;
+		refcount_inc(&ctl->count);
+		prepare_to_wait(&ctl->wait, &wait, TASK_UNINTERRUPTIBLE);
+		spin_unlock(&cache->lock);
+
+		schedule();
+
+		finish_wait(&ctl->wait, &wait);
+		btrfs_put_caching_control(ctl);
+		spin_lock(&cache->lock);
+	}
+
+	if (cache->cached != BTRFS_CACHE_NO) {
+		spin_unlock(&cache->lock);
+		kfree(caching_ctl);
+		return 0;
+	}
+	WARN_ON(cache->caching_ctl);
+	cache->caching_ctl = caching_ctl;
+	cache->cached = BTRFS_CACHE_FAST;
+	spin_unlock(&cache->lock);
+
+	if (btrfs_test_opt(fs_info, SPACE_CACHE)) {
+		mutex_lock(&caching_ctl->mutex);
+		ret = load_free_space_cache(cache);
+
+		spin_lock(&cache->lock);
+		if (ret == 1) {
+			cache->caching_ctl = NULL;
+			cache->cached = BTRFS_CACHE_FINISHED;
+			cache->last_byte_to_unpin = (u64)-1;
+			caching_ctl->progress = (u64)-1;
+		} else {
+			if (load_cache_only) {
+				cache->caching_ctl = NULL;
+				cache->cached = BTRFS_CACHE_NO;
+			} else {
+				cache->cached = BTRFS_CACHE_STARTED;
+				cache->has_caching_ctl = 1;
+			}
+		}
+		spin_unlock(&cache->lock);
+#ifdef CONFIG_BTRFS_DEBUG
+		if (ret == 1 &&
+		    btrfs_should_fragment_free_space(cache)) {
+			u64 bytes_used;
+
+			spin_lock(&cache->space_info->lock);
+			spin_lock(&cache->lock);
+			bytes_used = cache->key.offset -
+				btrfs_block_group_used(&cache->item);
+			cache->space_info->bytes_used += bytes_used >> 1;
+			spin_unlock(&cache->lock);
+			spin_unlock(&cache->space_info->lock);
+			btrfs_fragment_free_space(cache);
+		}
+#endif
+		mutex_unlock(&caching_ctl->mutex);
+
+		wake_up(&caching_ctl->wait);
+		if (ret == 1) {
+			btrfs_put_caching_control(caching_ctl);
+			btrfs_free_excluded_extents(cache);
+			return 0;
+		}
+	} else {
+		/*
+		 * We're either using the free space tree or no caching at all.
+		 * Set cached to the appropriate value and wakeup any waiters.
+		 */
+		spin_lock(&cache->lock);
+		if (load_cache_only) {
+			cache->caching_ctl = NULL;
+			cache->cached = BTRFS_CACHE_NO;
+		} else {
+			cache->cached = BTRFS_CACHE_STARTED;
+			cache->has_caching_ctl = 1;
+		}
+		spin_unlock(&cache->lock);
+		wake_up(&caching_ctl->wait);
+	}
+
+	if (load_cache_only) {
+		btrfs_put_caching_control(caching_ctl);
+		return 0;
+	}
+
+	down_write(&fs_info->commit_root_sem);
+	refcount_inc(&caching_ctl->count);
+	list_add_tail(&caching_ctl->list, &fs_info->caching_block_groups);
+	up_write(&fs_info->commit_root_sem);
+
+	btrfs_get_block_group(cache);
+
+	btrfs_queue_work(fs_info->caching_workers, &caching_ctl->work);
+
+	return ret;
+}
diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index ef101fd52158..80b388ece277 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -174,6 +174,8 @@ int btrfs_cache_block_group(struct btrfs_block_group_cache *cache,
 void btrfs_put_caching_control(struct btrfs_caching_control *ctl);
 struct btrfs_caching_control *btrfs_get_caching_control(
 		struct btrfs_block_group_cache *cache);
+u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
+		       u64 start, u64 end);
 
 static inline int btrfs_block_group_cache_done(
 		struct btrfs_block_group_cache *cache)
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 357316173d84..28c90916a8ae 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2618,8 +2618,6 @@ int btrfs_start_write_no_snapshotting(struct btrfs_root *root);
 void btrfs_end_write_no_snapshotting(struct btrfs_root *root);
 void btrfs_wait_for_snapshot_creation(struct btrfs_root *root);
 void check_system_chunk(struct btrfs_trans_handle *trans, const u64 type);
-u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
-		       u64 start, u64 end);
 void btrfs_mark_bg_unused(struct btrfs_block_group_cache *bg);
 
 /* ctree.c */
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index ad15d05e256b..90348105991d 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -181,420 +181,6 @@ static int exclude_super_stripes(struct btrfs_block_group_cache *cache)
 	return 0;
 }
 
-struct btrfs_caching_control *btrfs_get_caching_control(
-		struct btrfs_block_group_cache *cache)
-{
-	struct btrfs_caching_control *ctl;
-
-	spin_lock(&cache->lock);
-	if (!cache->caching_ctl) {
-		spin_unlock(&cache->lock);
-		return NULL;
-	}
-
-	ctl = cache->caching_ctl;
-	refcount_inc(&ctl->count);
-	spin_unlock(&cache->lock);
-	return ctl;
-}
-
-void btrfs_put_caching_control(struct btrfs_caching_control *ctl)
-{
-	if (refcount_dec_and_test(&ctl->count))
-		kfree(ctl);
-}
-
-#ifdef CONFIG_BTRFS_DEBUG
-void btrfs_fragment_free_space(struct btrfs_block_group_cache *block_group)
-{
-	struct btrfs_fs_info *fs_info = block_group->fs_info;
-	u64 start = block_group->key.objectid;
-	u64 len = block_group->key.offset;
-	u64 chunk = block_group->flags & BTRFS_BLOCK_GROUP_METADATA ?
-		fs_info->nodesize : fs_info->sectorsize;
-	u64 step = chunk << 1;
-
-	while (len > chunk) {
-		btrfs_remove_free_space(block_group, start, chunk);
-		start += step;
-		if (len < step)
-			len = 0;
-		else
-			len -= step;
-	}
-}
-#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.
- */
-u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
-		       u64 start, u64 end)
-{
-	struct btrfs_fs_info *info = block_group->fs_info;
-	u64 extent_start, extent_end, size, total_added = 0;
-	int ret;
-
-	while (start < end) {
-		ret = find_first_extent_bit(info->pinned_extents, start,
-					    &extent_start, &extent_end,
-					    EXTENT_DIRTY | EXTENT_UPTODATE,
-					    NULL);
-		if (ret)
-			break;
-
-		if (extent_start <= start) {
-			start = extent_end + 1;
-		} else if (extent_start > start && extent_start < end) {
-			size = extent_start - start;
-			total_added += size;
-			ret = btrfs_add_free_space(block_group, start,
-						   size);
-			BUG_ON(ret); /* -ENOMEM or logic error */
-			start = extent_end + 1;
-		} else {
-			break;
-		}
-	}
-
-	if (start < end) {
-		size = end - start;
-		total_added += size;
-		ret = btrfs_add_free_space(block_group, start, size);
-		BUG_ON(ret); /* -ENOMEM or logic error */
-	}
-
-	return total_added;
-}
-
-static int load_extent_tree_free(struct btrfs_caching_control *caching_ctl)
-{
-	struct btrfs_block_group_cache *block_group = caching_ctl->block_group;
-	struct btrfs_fs_info *fs_info = block_group->fs_info;
-	struct btrfs_root *extent_root = fs_info->extent_root;
-	struct btrfs_path *path;
-	struct extent_buffer *leaf;
-	struct btrfs_key key;
-	u64 total_found = 0;
-	u64 last = 0;
-	u32 nritems;
-	int ret;
-	bool wakeup = true;
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
-	last = max_t(u64, block_group->key.objectid, BTRFS_SUPER_INFO_OFFSET);
-
-#ifdef CONFIG_BTRFS_DEBUG
-	/*
-	 * If we're fragmenting we don't want to make anybody think we can
-	 * allocate from this block group until we've had a chance to fragment
-	 * the free space.
-	 */
-	if (btrfs_should_fragment_free_space(block_group))
-		wakeup = false;
-#endif
-	/*
-	 * We don't want to deadlock with somebody trying to allocate a new
-	 * extent for the extent root while also trying to search the extent
-	 * root to add free space.  So we skip locking and search the commit
-	 * root, since its read-only
-	 */
-	path->skip_locking = 1;
-	path->search_commit_root = 1;
-	path->reada = READA_FORWARD;
-
-	key.objectid = last;
-	key.offset = 0;
-	key.type = BTRFS_EXTENT_ITEM_KEY;
-
-next:
-	ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
-	if (ret < 0)
-		goto out;
-
-	leaf = path->nodes[0];
-	nritems = btrfs_header_nritems(leaf);
-
-	while (1) {
-		if (btrfs_fs_closing(fs_info) > 1) {
-			last = (u64)-1;
-			break;
-		}
-
-		if (path->slots[0] < nritems) {
-			btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
-		} else {
-			ret = btrfs_find_next_key(extent_root, path, &key, 0, 0);
-			if (ret)
-				break;
-
-			if (need_resched() ||
-			    rwsem_is_contended(&fs_info->commit_root_sem)) {
-				if (wakeup)
-					caching_ctl->progress = last;
-				btrfs_release_path(path);
-				up_read(&fs_info->commit_root_sem);
-				mutex_unlock(&caching_ctl->mutex);
-				cond_resched();
-				mutex_lock(&caching_ctl->mutex);
-				down_read(&fs_info->commit_root_sem);
-				goto next;
-			}
-
-			ret = btrfs_next_leaf(extent_root, path);
-			if (ret < 0)
-				goto out;
-			if (ret)
-				break;
-			leaf = path->nodes[0];
-			nritems = btrfs_header_nritems(leaf);
-			continue;
-		}
-
-		if (key.objectid < last) {
-			key.objectid = last;
-			key.offset = 0;
-			key.type = BTRFS_EXTENT_ITEM_KEY;
-
-			if (wakeup)
-				caching_ctl->progress = last;
-			btrfs_release_path(path);
-			goto next;
-		}
-
-		if (key.objectid < block_group->key.objectid) {
-			path->slots[0]++;
-			continue;
-		}
-
-		if (key.objectid >= block_group->key.objectid +
-		    block_group->key.offset)
-			break;
-
-		if (key.type == BTRFS_EXTENT_ITEM_KEY ||
-		    key.type == BTRFS_METADATA_ITEM_KEY) {
-			total_found += add_new_free_space(block_group, last,
-							  key.objectid);
-			if (key.type == BTRFS_METADATA_ITEM_KEY)
-				last = key.objectid +
-					fs_info->nodesize;
-			else
-				last = key.objectid + key.offset;
-
-			if (total_found > CACHING_CTL_WAKE_UP) {
-				total_found = 0;
-				if (wakeup)
-					wake_up(&caching_ctl->wait);
-			}
-		}
-		path->slots[0]++;
-	}
-	ret = 0;
-
-	total_found += add_new_free_space(block_group, last,
-					  block_group->key.objectid +
-					  block_group->key.offset);
-	caching_ctl->progress = (u64)-1;
-
-out:
-	btrfs_free_path(path);
-	return ret;
-}
-
-static noinline void caching_thread(struct btrfs_work *work)
-{
-	struct btrfs_block_group_cache *block_group;
-	struct btrfs_fs_info *fs_info;
-	struct btrfs_caching_control *caching_ctl;
-	int ret;
-
-	caching_ctl = container_of(work, struct btrfs_caching_control, work);
-	block_group = caching_ctl->block_group;
-	fs_info = block_group->fs_info;
-
-	mutex_lock(&caching_ctl->mutex);
-	down_read(&fs_info->commit_root_sem);
-
-	if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE))
-		ret = load_free_space_tree(caching_ctl);
-	else
-		ret = load_extent_tree_free(caching_ctl);
-
-	spin_lock(&block_group->lock);
-	block_group->caching_ctl = NULL;
-	block_group->cached = ret ? BTRFS_CACHE_ERROR : BTRFS_CACHE_FINISHED;
-	spin_unlock(&block_group->lock);
-
-#ifdef CONFIG_BTRFS_DEBUG
-	if (btrfs_should_fragment_free_space(block_group)) {
-		u64 bytes_used;
-
-		spin_lock(&block_group->space_info->lock);
-		spin_lock(&block_group->lock);
-		bytes_used = block_group->key.offset -
-			btrfs_block_group_used(&block_group->item);
-		block_group->space_info->bytes_used += bytes_used >> 1;
-		spin_unlock(&block_group->lock);
-		spin_unlock(&block_group->space_info->lock);
-		btrfs_fragment_free_space(block_group);
-	}
-#endif
-
-	caching_ctl->progress = (u64)-1;
-
-	up_read(&fs_info->commit_root_sem);
-	btrfs_free_excluded_extents(block_group);
-	mutex_unlock(&caching_ctl->mutex);
-
-	wake_up(&caching_ctl->wait);
-
-	btrfs_put_caching_control(caching_ctl);
-	btrfs_put_block_group(block_group);
-}
-
-int btrfs_cache_block_group(struct btrfs_block_group_cache *cache,
-			    int load_cache_only)
-{
-	DEFINE_WAIT(wait);
-	struct btrfs_fs_info *fs_info = cache->fs_info;
-	struct btrfs_caching_control *caching_ctl;
-	int ret = 0;
-
-	caching_ctl = kzalloc(sizeof(*caching_ctl), GFP_NOFS);
-	if (!caching_ctl)
-		return -ENOMEM;
-
-	INIT_LIST_HEAD(&caching_ctl->list);
-	mutex_init(&caching_ctl->mutex);
-	init_waitqueue_head(&caching_ctl->wait);
-	caching_ctl->block_group = cache;
-	caching_ctl->progress = cache->key.objectid;
-	refcount_set(&caching_ctl->count, 1);
-	btrfs_init_work(&caching_ctl->work, btrfs_cache_helper,
-			caching_thread, NULL, NULL);
-
-	spin_lock(&cache->lock);
-	/*
-	 * This should be a rare occasion, but this could happen I think in the
-	 * case where one thread starts to load the space cache info, and then
-	 * some other thread starts a transaction commit which tries to do an
-	 * allocation while the other thread is still loading the space cache
-	 * info.  The previous loop should have kept us from choosing this block
-	 * group, but if we've moved to the state where we will wait on caching
-	 * block groups we need to first check if we're doing a fast load here,
-	 * so we can wait for it to finish, otherwise we could end up allocating
-	 * from a block group who's cache gets evicted for one reason or
-	 * another.
-	 */
-	while (cache->cached == BTRFS_CACHE_FAST) {
-		struct btrfs_caching_control *ctl;
-
-		ctl = cache->caching_ctl;
-		refcount_inc(&ctl->count);
-		prepare_to_wait(&ctl->wait, &wait, TASK_UNINTERRUPTIBLE);
-		spin_unlock(&cache->lock);
-
-		schedule();
-
-		finish_wait(&ctl->wait, &wait);
-		btrfs_put_caching_control(ctl);
-		spin_lock(&cache->lock);
-	}
-
-	if (cache->cached != BTRFS_CACHE_NO) {
-		spin_unlock(&cache->lock);
-		kfree(caching_ctl);
-		return 0;
-	}
-	WARN_ON(cache->caching_ctl);
-	cache->caching_ctl = caching_ctl;
-	cache->cached = BTRFS_CACHE_FAST;
-	spin_unlock(&cache->lock);
-
-	if (btrfs_test_opt(fs_info, SPACE_CACHE)) {
-		mutex_lock(&caching_ctl->mutex);
-		ret = load_free_space_cache(cache);
-
-		spin_lock(&cache->lock);
-		if (ret == 1) {
-			cache->caching_ctl = NULL;
-			cache->cached = BTRFS_CACHE_FINISHED;
-			cache->last_byte_to_unpin = (u64)-1;
-			caching_ctl->progress = (u64)-1;
-		} else {
-			if (load_cache_only) {
-				cache->caching_ctl = NULL;
-				cache->cached = BTRFS_CACHE_NO;
-			} else {
-				cache->cached = BTRFS_CACHE_STARTED;
-				cache->has_caching_ctl = 1;
-			}
-		}
-		spin_unlock(&cache->lock);
-#ifdef CONFIG_BTRFS_DEBUG
-		if (ret == 1 &&
-		    btrfs_should_fragment_free_space(cache)) {
-			u64 bytes_used;
-
-			spin_lock(&cache->space_info->lock);
-			spin_lock(&cache->lock);
-			bytes_used = cache->key.offset -
-				btrfs_block_group_used(&cache->item);
-			cache->space_info->bytes_used += bytes_used >> 1;
-			spin_unlock(&cache->lock);
-			spin_unlock(&cache->space_info->lock);
-			btrfs_fragment_free_space(cache);
-		}
-#endif
-		mutex_unlock(&caching_ctl->mutex);
-
-		wake_up(&caching_ctl->wait);
-		if (ret == 1) {
-			btrfs_put_caching_control(caching_ctl);
-			btrfs_free_excluded_extents(cache);
-			return 0;
-		}
-	} else {
-		/*
-		 * We're either using the free space tree or no caching at all.
-		 * Set cached to the appropriate value and wakeup any waiters.
-		 */
-		spin_lock(&cache->lock);
-		if (load_cache_only) {
-			cache->caching_ctl = NULL;
-			cache->cached = BTRFS_CACHE_NO;
-		} else {
-			cache->cached = BTRFS_CACHE_STARTED;
-			cache->has_caching_ctl = 1;
-		}
-		spin_unlock(&cache->lock);
-		wake_up(&caching_ctl->wait);
-	}
-
-	if (load_cache_only) {
-		btrfs_put_caching_control(caching_ctl);
-		return 0;
-	}
-
-	down_write(&fs_info->commit_root_sem);
-	refcount_inc(&caching_ctl->count);
-	list_add_tail(&caching_ctl->list, &fs_info->caching_block_groups);
-	up_write(&fs_info->commit_root_sem);
-
-	btrfs_get_block_group(cache);
-
-	btrfs_queue_work(fs_info->caching_workers, &caching_ctl->work);
-
-	return ret;
-}
-
-
 static u64 generic_ref_to_space_flags(struct btrfs_ref *ref)
 {
 	if (ref->type == BTRFS_REF_METADATA) {
@@ -5056,51 +4642,6 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_ref *ref)
 	return ret;
 }
 
-/*
- * when we wait for progress in the block group caching, its because
- * our allocation attempt failed at least once.  So, we must sleep
- * and let some progress happen before we try again.
- *
- * This function will sleep at least once waiting for new free space to
- * show up, and then it will check the block group free space numbers
- * for our min num_bytes.  Another option is to have it go ahead
- * and look in the rbtree for a free extent of a given size, but this
- * is a good start.
- *
- * Callers of this must check if cache->cached == BTRFS_CACHE_ERROR before using
- * any of the information in this block group.
- */
-void btrfs_wait_block_group_cache_progress(struct btrfs_block_group_cache *cache,
-					   u64 num_bytes)
-{
-	struct btrfs_caching_control *caching_ctl;
-
-	caching_ctl = btrfs_get_caching_control(cache);
-	if (!caching_ctl)
-		return;
-
-	wait_event(caching_ctl->wait, btrfs_block_group_cache_done(cache) ||
-		   (cache->free_space_ctl->free_space >= num_bytes));
-
-	btrfs_put_caching_control(caching_ctl);
-}
-
-int btrfs_wait_block_group_cache_done(struct btrfs_block_group_cache *cache)
-{
-	struct btrfs_caching_control *caching_ctl;
-	int ret = 0;
-
-	caching_ctl = btrfs_get_caching_control(cache);
-	if (!caching_ctl)
-		return (cache->cached == BTRFS_CACHE_ERROR) ? -EIO : 0;
-
-	wait_event(caching_ctl->wait, btrfs_block_group_cache_done(cache));
-	if (cache->cached == BTRFS_CACHE_ERROR)
-		ret = -EIO;
-	btrfs_put_caching_control(caching_ctl);
-	return ret;
-}
-
 enum btrfs_loop_type {
 	LOOP_CACHING_NOWAIT,
 	LOOP_CACHING_WAIT,
-- 
2.21.0

[PATCH 02/15] btrfs: temporarily export inc_block_group_ro

[Josef Bacik]

This is used in a few logical parts of the block group code, temporarily
export it so we can move things in pieces.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.h |  2 ++
 fs/btrfs/extent-tree.c | 15 ++++++++-------
 2 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index 80b388ece277..26c5bf876737 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -185,4 +185,6 @@ static inline int btrfs_block_group_cache_done(
 		cache->cached == BTRFS_CACHE_ERROR;
 }
 
+int __btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache,
+			       int force);
 #endif /* BTRFS_BLOCK_GROUP_H */
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 90348105991d..54dc910eb6c8 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -6836,7 +6836,8 @@ static u64 update_block_group_flags(struct btrfs_fs_info *fs_info, u64 flags)
  * data in this block group. That check should be done by relocation routine,
  * not this function.
  */
-static int inc_block_group_ro(struct btrfs_block_group_cache *cache, int force)
+int __btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache,
+			       int force)
 {
 	struct btrfs_space_info *sinfo = cache->space_info;
 	u64 num_bytes;
@@ -6946,14 +6947,14 @@ int btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache)
 			goto out;
 	}
 
-	ret = inc_block_group_ro(cache, 0);
+	ret = __btrfs_inc_block_group_ro(cache, 0);
 	if (!ret)
 		goto out;
 	alloc_flags = get_alloc_profile(fs_info, cache->space_info->flags);
 	ret = btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
 	if (ret < 0)
 		goto out;
-	ret = inc_block_group_ro(cache, 0);
+	ret = __btrfs_inc_block_group_ro(cache, 0);
 out:
 	if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM) {
 		alloc_flags = update_block_group_flags(fs_info, cache->flags);
@@ -7486,7 +7487,7 @@ int btrfs_read_block_groups(struct btrfs_fs_info *info)
 
 		set_avail_alloc_bits(info, cache->flags);
 		if (btrfs_chunk_readonly(info, cache->key.objectid)) {
-			inc_block_group_ro(cache, 1);
+			__btrfs_inc_block_group_ro(cache, 1);
 		} else if (btrfs_block_group_used(&cache->item) == 0) {
 			ASSERT(list_empty(&cache->bg_list));
 			btrfs_mark_bg_unused(cache);
@@ -7507,11 +7508,11 @@ int btrfs_read_block_groups(struct btrfs_fs_info *info)
 		list_for_each_entry(cache,
 				&space_info->block_groups[BTRFS_RAID_RAID0],
 				list)
-			inc_block_group_ro(cache, 1);
+			__btrfs_inc_block_group_ro(cache, 1);
 		list_for_each_entry(cache,
 				&space_info->block_groups[BTRFS_RAID_SINGLE],
 				list)
-			inc_block_group_ro(cache, 1);
+			__btrfs_inc_block_group_ro(cache, 1);
 	}
 
 	btrfs_add_raid_kobjects(info);
@@ -8051,7 +8052,7 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		spin_unlock(&block_group->lock);
 
 		/* We don't want to force the issue, only flip if it's ok. */
-		ret = inc_block_group_ro(block_group, 0);
+		ret = __btrfs_inc_block_group_ro(block_group, 0);
 		up_write(&space_info->groups_sem);
 		if (ret < 0) {
 			ret = 0;
-- 
2.21.0

[PATCH 03/15] btrfs: migrate the block group removal code

[Josef Bacik]

This is the removal code and the unused bgs code.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.c | 541 +++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/block-group.h |   7 +
 fs/btrfs/ctree.h       |   7 -
 fs/btrfs/extent-tree.c | 537 ----------------------------------------
 4 files changed, 548 insertions(+), 544 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index e8c29045e1b9..0bc51bfe4904 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -6,6 +6,10 @@
 #include "disk-io.h"
 #include "free-space-cache.h"
 #include "free-space-tree.h"
+#include "disk-io.h"
+#include "volumes.h"
+#include "transaction.h"
+#include "ref-verify.h"
 
 void btrfs_get_block_group(struct btrfs_block_group_cache *cache)
 {
@@ -663,3 +667,540 @@ int btrfs_cache_block_group(struct btrfs_block_group_cache *cache,
 
 	return ret;
 }
+
+static void clear_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
+{
+	u64 extra_flags = chunk_to_extended(flags) &
+				BTRFS_EXTENDED_PROFILE_MASK;
+
+	write_seqlock(&fs_info->profiles_lock);
+	if (flags & BTRFS_BLOCK_GROUP_DATA)
+		fs_info->avail_data_alloc_bits &= ~extra_flags;
+	if (flags & BTRFS_BLOCK_GROUP_METADATA)
+		fs_info->avail_metadata_alloc_bits &= ~extra_flags;
+	if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
+		fs_info->avail_system_alloc_bits &= ~extra_flags;
+	write_sequnlock(&fs_info->profiles_lock);
+}
+
+/*
+ * Clear incompat bits for the following feature(s):
+ *
+ * - RAID56 - in case there's neither RAID5 nor RAID6 profile block group
+ *            in the whole filesystem
+ */
+static void clear_incompat_bg_bits(struct btrfs_fs_info *fs_info, u64 flags)
+{
+	if (flags & BTRFS_BLOCK_GROUP_RAID56_MASK) {
+		struct list_head *head = &fs_info->space_info;
+		struct btrfs_space_info *sinfo;
+
+		list_for_each_entry_rcu(sinfo, head, list) {
+			bool found = false;
+
+			down_read(&sinfo->groups_sem);
+			if (!list_empty(&sinfo->block_groups[BTRFS_RAID_RAID5]))
+				found = true;
+			if (!list_empty(&sinfo->block_groups[BTRFS_RAID_RAID6]))
+				found = true;
+			up_read(&sinfo->groups_sem);
+
+			if (found)
+				return;
+		}
+		btrfs_clear_fs_incompat(fs_info, RAID56);
+	}
+}
+
+int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
+			     u64 group_start, struct extent_map *em)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_root *root = fs_info->extent_root;
+	struct btrfs_path *path;
+	struct btrfs_block_group_cache *block_group;
+	struct btrfs_free_cluster *cluster;
+	struct btrfs_root *tree_root = fs_info->tree_root;
+	struct btrfs_key key;
+	struct inode *inode;
+	struct kobject *kobj = NULL;
+	int ret;
+	int index;
+	int factor;
+	struct btrfs_caching_control *caching_ctl = NULL;
+	bool remove_em;
+	bool remove_rsv = false;
+
+	block_group = btrfs_lookup_block_group(fs_info, group_start);
+	BUG_ON(!block_group);
+	BUG_ON(!block_group->ro);
+
+	trace_btrfs_remove_block_group(block_group);
+	/*
+	 * Free the reserved super bytes from this block group before
+	 * remove it.
+	 */
+	btrfs_free_excluded_extents(block_group);
+	btrfs_free_ref_tree_range(fs_info, block_group->key.objectid,
+				  block_group->key.offset);
+
+	memcpy(&key, &block_group->key, sizeof(key));
+	index = btrfs_bg_flags_to_raid_index(block_group->flags);
+	factor = btrfs_bg_type_to_factor(block_group->flags);
+
+	/* make sure this block group isn't part of an allocation cluster */
+	cluster = &fs_info->data_alloc_cluster;
+	spin_lock(&cluster->refill_lock);
+	btrfs_return_cluster_to_free_space(block_group, cluster);
+	spin_unlock(&cluster->refill_lock);
+
+	/*
+	 * make sure this block group isn't part of a metadata
+	 * allocation cluster
+	 */
+	cluster = &fs_info->meta_alloc_cluster;
+	spin_lock(&cluster->refill_lock);
+	btrfs_return_cluster_to_free_space(block_group, cluster);
+	spin_unlock(&cluster->refill_lock);
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	/*
+	 * get the inode first so any iput calls done for the io_list
+	 * aren't the final iput (no unlinks allowed now)
+	 */
+	inode = lookup_free_space_inode(block_group, path);
+
+	mutex_lock(&trans->transaction->cache_write_mutex);
+	/*
+	 * Make sure our free space cache IO is done before removing the
+	 * free space inode
+	 */
+	spin_lock(&trans->transaction->dirty_bgs_lock);
+	if (!list_empty(&block_group->io_list)) {
+		list_del_init(&block_group->io_list);
+
+		WARN_ON(!IS_ERR(inode) && inode != block_group->io_ctl.inode);
+
+		spin_unlock(&trans->transaction->dirty_bgs_lock);
+		btrfs_wait_cache_io(trans, block_group, path);
+		btrfs_put_block_group(block_group);
+		spin_lock(&trans->transaction->dirty_bgs_lock);
+	}
+
+	if (!list_empty(&block_group->dirty_list)) {
+		list_del_init(&block_group->dirty_list);
+		remove_rsv = true;
+		btrfs_put_block_group(block_group);
+	}
+	spin_unlock(&trans->transaction->dirty_bgs_lock);
+	mutex_unlock(&trans->transaction->cache_write_mutex);
+
+	if (!IS_ERR(inode)) {
+		ret = btrfs_orphan_add(trans, BTRFS_I(inode));
+		if (ret) {
+			btrfs_add_delayed_iput(inode);
+			goto out;
+		}
+		clear_nlink(inode);
+		/* One for the block groups ref */
+		spin_lock(&block_group->lock);
+		if (block_group->iref) {
+			block_group->iref = 0;
+			block_group->inode = NULL;
+			spin_unlock(&block_group->lock);
+			iput(inode);
+		} else {
+			spin_unlock(&block_group->lock);
+		}
+		/* One for our lookup ref */
+		btrfs_add_delayed_iput(inode);
+	}
+
+	key.objectid = BTRFS_FREE_SPACE_OBJECTID;
+	key.offset = block_group->key.objectid;
+	key.type = 0;
+
+	ret = btrfs_search_slot(trans, tree_root, &key, path, -1, 1);
+	if (ret < 0)
+		goto out;
+	if (ret > 0)
+		btrfs_release_path(path);
+	if (ret == 0) {
+		ret = btrfs_del_item(trans, tree_root, path);
+		if (ret)
+			goto out;
+		btrfs_release_path(path);
+	}
+
+	spin_lock(&fs_info->block_group_cache_lock);
+	rb_erase(&block_group->cache_node,
+		 &fs_info->block_group_cache_tree);
+	RB_CLEAR_NODE(&block_group->cache_node);
+
+	if (fs_info->first_logical_byte == block_group->key.objectid)
+		fs_info->first_logical_byte = (u64)-1;
+	spin_unlock(&fs_info->block_group_cache_lock);
+
+	down_write(&block_group->space_info->groups_sem);
+	/*
+	 * we must use list_del_init so people can check to see if they
+	 * are still on the list after taking the semaphore
+	 */
+	list_del_init(&block_group->list);
+	if (list_empty(&block_group->space_info->block_groups[index])) {
+		kobj = block_group->space_info->block_group_kobjs[index];
+		block_group->space_info->block_group_kobjs[index] = NULL;
+		clear_avail_alloc_bits(fs_info, block_group->flags);
+	}
+	up_write(&block_group->space_info->groups_sem);
+	clear_incompat_bg_bits(fs_info, block_group->flags);
+	if (kobj) {
+		kobject_del(kobj);
+		kobject_put(kobj);
+	}
+
+	if (block_group->has_caching_ctl)
+		caching_ctl = btrfs_get_caching_control(block_group);
+	if (block_group->cached == BTRFS_CACHE_STARTED)
+		btrfs_wait_block_group_cache_done(block_group);
+	if (block_group->has_caching_ctl) {
+		down_write(&fs_info->commit_root_sem);
+		if (!caching_ctl) {
+			struct btrfs_caching_control *ctl;
+
+			list_for_each_entry(ctl,
+				    &fs_info->caching_block_groups, list)
+				if (ctl->block_group == block_group) {
+					caching_ctl = ctl;
+					refcount_inc(&caching_ctl->count);
+					break;
+				}
+		}
+		if (caching_ctl)
+			list_del_init(&caching_ctl->list);
+		up_write(&fs_info->commit_root_sem);
+		if (caching_ctl) {
+			/* Once for the caching bgs list and once for us. */
+			btrfs_put_caching_control(caching_ctl);
+			btrfs_put_caching_control(caching_ctl);
+		}
+	}
+
+	spin_lock(&trans->transaction->dirty_bgs_lock);
+	WARN_ON(!list_empty(&block_group->dirty_list));
+	WARN_ON(!list_empty(&block_group->io_list));
+	spin_unlock(&trans->transaction->dirty_bgs_lock);
+
+	btrfs_remove_free_space_cache(block_group);
+
+	spin_lock(&block_group->space_info->lock);
+	list_del_init(&block_group->ro_list);
+
+	if (btrfs_test_opt(fs_info, ENOSPC_DEBUG)) {
+		WARN_ON(block_group->space_info->total_bytes
+			< block_group->key.offset);
+		WARN_ON(block_group->space_info->bytes_readonly
+			< block_group->key.offset);
+		WARN_ON(block_group->space_info->disk_total
+			< block_group->key.offset * factor);
+	}
+	block_group->space_info->total_bytes -= block_group->key.offset;
+	block_group->space_info->bytes_readonly -= block_group->key.offset;
+	block_group->space_info->disk_total -= block_group->key.offset * factor;
+
+	spin_unlock(&block_group->space_info->lock);
+
+	memcpy(&key, &block_group->key, sizeof(key));
+
+	mutex_lock(&fs_info->chunk_mutex);
+	spin_lock(&block_group->lock);
+	block_group->removed = 1;
+	/*
+	 * At this point trimming can't start on this block group, because we
+	 * removed the block group from the tree fs_info->block_group_cache_tree
+	 * so no one can't find it anymore and even if someone already got this
+	 * block group before we removed it from the rbtree, they have already
+	 * incremented block_group->trimming - if they didn't, they won't find
+	 * any free space entries because we already removed them all when we
+	 * called btrfs_remove_free_space_cache().
+	 *
+	 * And we must not remove the extent map from the fs_info->mapping_tree
+	 * to prevent the same logical address range and physical device space
+	 * ranges from being reused for a new block group. This is because our
+	 * fs trim operation (btrfs_trim_fs() / btrfs_ioctl_fitrim()) is
+	 * completely transactionless, so while it is trimming a range the
+	 * currently running transaction might finish and a new one start,
+	 * allowing for new block groups to be created that can reuse the same
+	 * physical device locations unless we take this special care.
+	 *
+	 * There may also be an implicit trim operation if the file system
+	 * is mounted with -odiscard. The same protections must remain
+	 * in place until the extents have been discarded completely when
+	 * the transaction commit has completed.
+	 */
+	remove_em = (atomic_read(&block_group->trimming) == 0);
+	spin_unlock(&block_group->lock);
+
+	mutex_unlock(&fs_info->chunk_mutex);
+
+	ret = remove_block_group_free_space(trans, block_group);
+	if (ret)
+		goto out;
+
+	btrfs_put_block_group(block_group);
+	btrfs_put_block_group(block_group);
+
+	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+	if (ret > 0)
+		ret = -EIO;
+	if (ret < 0)
+		goto out;
+
+	ret = btrfs_del_item(trans, root, path);
+	if (ret)
+		goto out;
+
+	if (remove_em) {
+		struct extent_map_tree *em_tree;
+
+		em_tree = &fs_info->mapping_tree;
+		write_lock(&em_tree->lock);
+		remove_extent_mapping(em_tree, em);
+		write_unlock(&em_tree->lock);
+		/* once for the tree */
+		free_extent_map(em);
+	}
+out:
+	if (remove_rsv)
+		btrfs_delayed_refs_rsv_release(fs_info, 1);
+	btrfs_free_path(path);
+	return ret;
+}
+
+struct btrfs_trans_handle *
+btrfs_start_trans_remove_block_group(struct btrfs_fs_info *fs_info,
+				     const u64 chunk_offset)
+{
+	struct extent_map_tree *em_tree = &fs_info->mapping_tree;
+	struct extent_map *em;
+	struct map_lookup *map;
+	unsigned int num_items;
+
+	read_lock(&em_tree->lock);
+	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
+	read_unlock(&em_tree->lock);
+	ASSERT(em && em->start == chunk_offset);
+
+	/*
+	 * We need to reserve 3 + N units from the metadata space info in order
+	 * to remove a block group (done at btrfs_remove_chunk() and at
+	 * btrfs_remove_block_group()), which are used for:
+	 *
+	 * 1 unit for adding the free space inode's orphan (located in the tree
+	 * of tree roots).
+	 * 1 unit for deleting the block group item (located in the extent
+	 * tree).
+	 * 1 unit for deleting the free space item (located in tree of tree
+	 * roots).
+	 * N units for deleting N device extent items corresponding to each
+	 * stripe (located in the device tree).
+	 *
+	 * In order to remove a block group we also need to reserve units in the
+	 * system space info in order to update the chunk tree (update one or
+	 * more device items and remove one chunk item), but this is done at
+	 * btrfs_remove_chunk() through a call to check_system_chunk().
+	 */
+	map = em->map_lookup;
+	num_items = 3 + map->num_stripes;
+	free_extent_map(em);
+
+	return btrfs_start_transaction_fallback_global_rsv(fs_info->extent_root,
+							   num_items, 1);
+}
+
+/*
+ * Process the unused_bgs list and remove any that don't have any allocated
+ * space inside of them.
+ */
+void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_block_group_cache *block_group;
+	struct btrfs_space_info *space_info;
+	struct btrfs_trans_handle *trans;
+	int ret = 0;
+
+	if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))
+		return;
+
+	spin_lock(&fs_info->unused_bgs_lock);
+	while (!list_empty(&fs_info->unused_bgs)) {
+		u64 start, end;
+		int trimming;
+
+		block_group = list_first_entry(&fs_info->unused_bgs,
+					       struct btrfs_block_group_cache,
+					       bg_list);
+		list_del_init(&block_group->bg_list);
+
+		space_info = block_group->space_info;
+
+		if (ret || btrfs_mixed_space_info(space_info)) {
+			btrfs_put_block_group(block_group);
+			continue;
+		}
+		spin_unlock(&fs_info->unused_bgs_lock);
+
+		mutex_lock(&fs_info->delete_unused_bgs_mutex);
+
+		/* Don't want to race with allocators so take the groups_sem */
+		down_write(&space_info->groups_sem);
+		spin_lock(&block_group->lock);
+		if (block_group->reserved || block_group->pinned ||
+		    btrfs_block_group_used(&block_group->item) ||
+		    block_group->ro ||
+		    list_is_singular(&block_group->list)) {
+			/*
+			 * We want to bail if we made new allocations or have
+			 * outstanding allocations in this block group.  We do
+			 * the ro check in case balance is currently acting on
+			 * this block group.
+			 */
+			trace_btrfs_skip_unused_block_group(block_group);
+			spin_unlock(&block_group->lock);
+			up_write(&space_info->groups_sem);
+			goto next;
+		}
+		spin_unlock(&block_group->lock);
+
+		/* We don't want to force the issue, only flip if it's ok. */
+		ret = __btrfs_inc_block_group_ro(block_group, 0);
+		up_write(&space_info->groups_sem);
+		if (ret < 0) {
+			ret = 0;
+			goto next;
+		}
+
+		/*
+		 * Want to do this before we do anything else so we can recover
+		 * properly if we fail to join the transaction.
+		 */
+		trans = btrfs_start_trans_remove_block_group(fs_info,
+						     block_group->key.objectid);
+		if (IS_ERR(trans)) {
+			btrfs_dec_block_group_ro(block_group);
+			ret = PTR_ERR(trans);
+			goto next;
+		}
+
+		/*
+		 * We could have pending pinned extents for this block group,
+		 * just delete them, we don't care about them anymore.
+		 */
+		start = block_group->key.objectid;
+		end = start + block_group->key.offset - 1;
+		/*
+		 * Hold the unused_bg_unpin_mutex lock to avoid racing with
+		 * btrfs_finish_extent_commit(). If we are at transaction N,
+		 * another task might be running finish_extent_commit() for the
+		 * previous transaction N - 1, and have seen a range belonging
+		 * to the block group in freed_extents[] before we were able to
+		 * clear the whole block group range from freed_extents[]. This
+		 * means that task can lookup for the block group after we
+		 * unpinned it from freed_extents[] and removed it, leading to
+		 * a BUG_ON() at btrfs_unpin_extent_range().
+		 */
+		mutex_lock(&fs_info->unused_bg_unpin_mutex);
+		ret = clear_extent_bits(&fs_info->freed_extents[0], start, end,
+				  EXTENT_DIRTY);
+		if (ret) {
+			mutex_unlock(&fs_info->unused_bg_unpin_mutex);
+			btrfs_dec_block_group_ro(block_group);
+			goto end_trans;
+		}
+		ret = clear_extent_bits(&fs_info->freed_extents[1], start, end,
+				  EXTENT_DIRTY);
+		if (ret) {
+			mutex_unlock(&fs_info->unused_bg_unpin_mutex);
+			btrfs_dec_block_group_ro(block_group);
+			goto end_trans;
+		}
+		mutex_unlock(&fs_info->unused_bg_unpin_mutex);
+
+		/* Reset pinned so btrfs_put_block_group doesn't complain */
+		spin_lock(&space_info->lock);
+		spin_lock(&block_group->lock);
+
+		btrfs_space_info_update_bytes_pinned(fs_info, space_info,
+						     -block_group->pinned);
+		space_info->bytes_readonly += block_group->pinned;
+		percpu_counter_add_batch(&space_info->total_bytes_pinned,
+				   -block_group->pinned,
+				   BTRFS_TOTAL_BYTES_PINNED_BATCH);
+		block_group->pinned = 0;
+
+		spin_unlock(&block_group->lock);
+		spin_unlock(&space_info->lock);
+
+		/* DISCARD can flip during remount */
+		trimming = btrfs_test_opt(fs_info, DISCARD);
+
+		/* Implicit trim during transaction commit. */
+		if (trimming)
+			btrfs_get_block_group_trimming(block_group);
+
+		/*
+		 * Btrfs_remove_chunk will abort the transaction if things go
+		 * horribly wrong.
+		 */
+		ret = btrfs_remove_chunk(trans, block_group->key.objectid);
+
+		if (ret) {
+			if (trimming)
+				btrfs_put_block_group_trimming(block_group);
+			goto end_trans;
+		}
+
+		/*
+		 * If we're not mounted with -odiscard, we can just forget
+		 * about this block group. Otherwise we'll need to wait
+		 * until transaction commit to do the actual discard.
+		 */
+		if (trimming) {
+			spin_lock(&fs_info->unused_bgs_lock);
+			/*
+			 * A concurrent scrub might have added us to the list
+			 * fs_info->unused_bgs, so use a list_move operation
+			 * to add the block group to the deleted_bgs list.
+			 */
+			list_move(&block_group->bg_list,
+				  &trans->transaction->deleted_bgs);
+			spin_unlock(&fs_info->unused_bgs_lock);
+			btrfs_get_block_group(block_group);
+		}
+end_trans:
+		btrfs_end_transaction(trans);
+next:
+		mutex_unlock(&fs_info->delete_unused_bgs_mutex);
+		btrfs_put_block_group(block_group);
+		spin_lock(&fs_info->unused_bgs_lock);
+	}
+	spin_unlock(&fs_info->unused_bgs_lock);
+}
+
+void btrfs_mark_bg_unused(struct btrfs_block_group_cache *bg)
+{
+	struct btrfs_fs_info *fs_info = bg->fs_info;
+
+	spin_lock(&fs_info->unused_bgs_lock);
+	if (list_empty(&bg->bg_list)) {
+		btrfs_get_block_group(bg);
+		trace_btrfs_add_unused_block_group(bg);
+		list_add_tail(&bg->bg_list, &fs_info->unused_bgs);
+	}
+	spin_unlock(&fs_info->unused_bgs_lock);
+}
diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index 26c5bf876737..5ee2ea695ab0 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -176,6 +176,13 @@ struct btrfs_caching_control *btrfs_get_caching_control(
 		struct btrfs_block_group_cache *cache);
 u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
 		       u64 start, u64 end);
+struct btrfs_trans_handle *btrfs_start_trans_remove_block_group(
+				struct btrfs_fs_info *fs_info,
+				const u64 chunk_offset);
+int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
+			     u64 group_start, struct extent_map *em);
+void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info);
+void btrfs_mark_bg_unused(struct btrfs_block_group_cache *bg);
 
 static inline int btrfs_block_group_cache_done(
 		struct btrfs_block_group_cache *cache)
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 28c90916a8ae..266c3cb5f3c9 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2530,12 +2530,6 @@ int btrfs_read_block_groups(struct btrfs_fs_info *info);
 int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 			   u64 bytes_used, u64 type, u64 chunk_offset,
 			   u64 size);
-struct btrfs_trans_handle *btrfs_start_trans_remove_block_group(
-				struct btrfs_fs_info *fs_info,
-				const u64 chunk_offset);
-int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
-			     u64 group_start, struct extent_map *em);
-void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info);
 void btrfs_get_block_group_trimming(struct btrfs_block_group_cache *cache);
 void btrfs_put_block_group_trimming(struct btrfs_block_group_cache *cache);
 void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans);
@@ -2618,7 +2612,6 @@ int btrfs_start_write_no_snapshotting(struct btrfs_root *root);
 void btrfs_end_write_no_snapshotting(struct btrfs_root *root);
 void btrfs_wait_for_snapshot_creation(struct btrfs_root *root);
 void check_system_chunk(struct btrfs_trans_handle *trans, const u64 type);
-void btrfs_mark_bg_unused(struct btrfs_block_group_cache *bg);
 
 /* ctree.c */
 int btrfs_bin_search(struct extent_buffer *eb, const struct btrfs_key *key,
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 54dc910eb6c8..d28e736fdef2 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -7642,530 +7642,6 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, u64 bytes_used,
 	return 0;
 }
 
-static void clear_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
-{
-	u64 extra_flags = chunk_to_extended(flags) &
-				BTRFS_EXTENDED_PROFILE_MASK;
-
-	write_seqlock(&fs_info->profiles_lock);
-	if (flags & BTRFS_BLOCK_GROUP_DATA)
-		fs_info->avail_data_alloc_bits &= ~extra_flags;
-	if (flags & BTRFS_BLOCK_GROUP_METADATA)
-		fs_info->avail_metadata_alloc_bits &= ~extra_flags;
-	if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
-		fs_info->avail_system_alloc_bits &= ~extra_flags;
-	write_sequnlock(&fs_info->profiles_lock);
-}
-
-/*
- * Clear incompat bits for the following feature(s):
- *
- * - RAID56 - in case there's neither RAID5 nor RAID6 profile block group
- *            in the whole filesystem
- */
-static void clear_incompat_bg_bits(struct btrfs_fs_info *fs_info, u64 flags)
-{
-	if (flags & BTRFS_BLOCK_GROUP_RAID56_MASK) {
-		struct list_head *head = &fs_info->space_info;
-		struct btrfs_space_info *sinfo;
-
-		list_for_each_entry_rcu(sinfo, head, list) {
-			bool found = false;
-
-			down_read(&sinfo->groups_sem);
-			if (!list_empty(&sinfo->block_groups[BTRFS_RAID_RAID5]))
-				found = true;
-			if (!list_empty(&sinfo->block_groups[BTRFS_RAID_RAID6]))
-				found = true;
-			up_read(&sinfo->groups_sem);
-
-			if (found)
-				return;
-		}
-		btrfs_clear_fs_incompat(fs_info, RAID56);
-	}
-}
-
-int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
-			     u64 group_start, struct extent_map *em)
-{
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_root *root = fs_info->extent_root;
-	struct btrfs_path *path;
-	struct btrfs_block_group_cache *block_group;
-	struct btrfs_free_cluster *cluster;
-	struct btrfs_root *tree_root = fs_info->tree_root;
-	struct btrfs_key key;
-	struct inode *inode;
-	struct kobject *kobj = NULL;
-	int ret;
-	int index;
-	int factor;
-	struct btrfs_caching_control *caching_ctl = NULL;
-	bool remove_em;
-	bool remove_rsv = false;
-
-	block_group = btrfs_lookup_block_group(fs_info, group_start);
-	BUG_ON(!block_group);
-	BUG_ON(!block_group->ro);
-
-	trace_btrfs_remove_block_group(block_group);
-	/*
-	 * Free the reserved super bytes from this block group before
-	 * remove it.
-	 */
-	btrfs_free_excluded_extents(block_group);
-	btrfs_free_ref_tree_range(fs_info, block_group->key.objectid,
-				  block_group->key.offset);
-
-	memcpy(&key, &block_group->key, sizeof(key));
-	index = btrfs_bg_flags_to_raid_index(block_group->flags);
-	factor = btrfs_bg_type_to_factor(block_group->flags);
-
-	/* make sure this block group isn't part of an allocation cluster */
-	cluster = &fs_info->data_alloc_cluster;
-	spin_lock(&cluster->refill_lock);
-	btrfs_return_cluster_to_free_space(block_group, cluster);
-	spin_unlock(&cluster->refill_lock);
-
-	/*
-	 * make sure this block group isn't part of a metadata
-	 * allocation cluster
-	 */
-	cluster = &fs_info->meta_alloc_cluster;
-	spin_lock(&cluster->refill_lock);
-	btrfs_return_cluster_to_free_space(block_group, cluster);
-	spin_unlock(&cluster->refill_lock);
-
-	path = btrfs_alloc_path();
-	if (!path) {
-		ret = -ENOMEM;
-		goto out;
-	}
-
-	/*
-	 * get the inode first so any iput calls done for the io_list
-	 * aren't the final iput (no unlinks allowed now)
-	 */
-	inode = lookup_free_space_inode(block_group, path);
-
-	mutex_lock(&trans->transaction->cache_write_mutex);
-	/*
-	 * Make sure our free space cache IO is done before removing the
-	 * free space inode
-	 */
-	spin_lock(&trans->transaction->dirty_bgs_lock);
-	if (!list_empty(&block_group->io_list)) {
-		list_del_init(&block_group->io_list);
-
-		WARN_ON(!IS_ERR(inode) && inode != block_group->io_ctl.inode);
-
-		spin_unlock(&trans->transaction->dirty_bgs_lock);
-		btrfs_wait_cache_io(trans, block_group, path);
-		btrfs_put_block_group(block_group);
-		spin_lock(&trans->transaction->dirty_bgs_lock);
-	}
-
-	if (!list_empty(&block_group->dirty_list)) {
-		list_del_init(&block_group->dirty_list);
-		remove_rsv = true;
-		btrfs_put_block_group(block_group);
-	}
-	spin_unlock(&trans->transaction->dirty_bgs_lock);
-	mutex_unlock(&trans->transaction->cache_write_mutex);
-
-	if (!IS_ERR(inode)) {
-		ret = btrfs_orphan_add(trans, BTRFS_I(inode));
-		if (ret) {
-			btrfs_add_delayed_iput(inode);
-			goto out;
-		}
-		clear_nlink(inode);
-		/* One for the block groups ref */
-		spin_lock(&block_group->lock);
-		if (block_group->iref) {
-			block_group->iref = 0;
-			block_group->inode = NULL;
-			spin_unlock(&block_group->lock);
-			iput(inode);
-		} else {
-			spin_unlock(&block_group->lock);
-		}
-		/* One for our lookup ref */
-		btrfs_add_delayed_iput(inode);
-	}
-
-	key.objectid = BTRFS_FREE_SPACE_OBJECTID;
-	key.offset = block_group->key.objectid;
-	key.type = 0;
-
-	ret = btrfs_search_slot(trans, tree_root, &key, path, -1, 1);
-	if (ret < 0)
-		goto out;
-	if (ret > 0)
-		btrfs_release_path(path);
-	if (ret == 0) {
-		ret = btrfs_del_item(trans, tree_root, path);
-		if (ret)
-			goto out;
-		btrfs_release_path(path);
-	}
-
-	spin_lock(&fs_info->block_group_cache_lock);
-	rb_erase(&block_group->cache_node,
-		 &fs_info->block_group_cache_tree);
-	RB_CLEAR_NODE(&block_group->cache_node);
-
-	if (fs_info->first_logical_byte == block_group->key.objectid)
-		fs_info->first_logical_byte = (u64)-1;
-	spin_unlock(&fs_info->block_group_cache_lock);
-
-	down_write(&block_group->space_info->groups_sem);
-	/*
-	 * we must use list_del_init so people can check to see if they
-	 * are still on the list after taking the semaphore
-	 */
-	list_del_init(&block_group->list);
-	if (list_empty(&block_group->space_info->block_groups[index])) {
-		kobj = block_group->space_info->block_group_kobjs[index];
-		block_group->space_info->block_group_kobjs[index] = NULL;
-		clear_avail_alloc_bits(fs_info, block_group->flags);
-	}
-	up_write(&block_group->space_info->groups_sem);
-	clear_incompat_bg_bits(fs_info, block_group->flags);
-	if (kobj) {
-		kobject_del(kobj);
-		kobject_put(kobj);
-	}
-
-	if (block_group->has_caching_ctl)
-		caching_ctl = btrfs_get_caching_control(block_group);
-	if (block_group->cached == BTRFS_CACHE_STARTED)
-		btrfs_wait_block_group_cache_done(block_group);
-	if (block_group->has_caching_ctl) {
-		down_write(&fs_info->commit_root_sem);
-		if (!caching_ctl) {
-			struct btrfs_caching_control *ctl;
-
-			list_for_each_entry(ctl,
-				    &fs_info->caching_block_groups, list)
-				if (ctl->block_group == block_group) {
-					caching_ctl = ctl;
-					refcount_inc(&caching_ctl->count);
-					break;
-				}
-		}
-		if (caching_ctl)
-			list_del_init(&caching_ctl->list);
-		up_write(&fs_info->commit_root_sem);
-		if (caching_ctl) {
-			/* Once for the caching bgs list and once for us. */
-			btrfs_put_caching_control(caching_ctl);
-			btrfs_put_caching_control(caching_ctl);
-		}
-	}
-
-	spin_lock(&trans->transaction->dirty_bgs_lock);
-	WARN_ON(!list_empty(&block_group->dirty_list));
-	WARN_ON(!list_empty(&block_group->io_list));
-	spin_unlock(&trans->transaction->dirty_bgs_lock);
-
-	btrfs_remove_free_space_cache(block_group);
-
-	spin_lock(&block_group->space_info->lock);
-	list_del_init(&block_group->ro_list);
-
-	if (btrfs_test_opt(fs_info, ENOSPC_DEBUG)) {
-		WARN_ON(block_group->space_info->total_bytes
-			< block_group->key.offset);
-		WARN_ON(block_group->space_info->bytes_readonly
-			< block_group->key.offset);
-		WARN_ON(block_group->space_info->disk_total
-			< block_group->key.offset * factor);
-	}
-	block_group->space_info->total_bytes -= block_group->key.offset;
-	block_group->space_info->bytes_readonly -= block_group->key.offset;
-	block_group->space_info->disk_total -= block_group->key.offset * factor;
-
-	spin_unlock(&block_group->space_info->lock);
-
-	memcpy(&key, &block_group->key, sizeof(key));
-
-	mutex_lock(&fs_info->chunk_mutex);
-	spin_lock(&block_group->lock);
-	block_group->removed = 1;
-	/*
-	 * At this point trimming can't start on this block group, because we
-	 * removed the block group from the tree fs_info->block_group_cache_tree
-	 * so no one can't find it anymore and even if someone already got this
-	 * block group before we removed it from the rbtree, they have already
-	 * incremented block_group->trimming - if they didn't, they won't find
-	 * any free space entries because we already removed them all when we
-	 * called btrfs_remove_free_space_cache().
-	 *
-	 * And we must not remove the extent map from the fs_info->mapping_tree
-	 * to prevent the same logical address range and physical device space
-	 * ranges from being reused for a new block group. This is because our
-	 * fs trim operation (btrfs_trim_fs() / btrfs_ioctl_fitrim()) is
-	 * completely transactionless, so while it is trimming a range the
-	 * currently running transaction might finish and a new one start,
-	 * allowing for new block groups to be created that can reuse the same
-	 * physical device locations unless we take this special care.
-	 *
-	 * There may also be an implicit trim operation if the file system
-	 * is mounted with -odiscard. The same protections must remain
-	 * in place until the extents have been discarded completely when
-	 * the transaction commit has completed.
-	 */
-	remove_em = (atomic_read(&block_group->trimming) == 0);
-	spin_unlock(&block_group->lock);
-
-	mutex_unlock(&fs_info->chunk_mutex);
-
-	ret = remove_block_group_free_space(trans, block_group);
-	if (ret)
-		goto out;
-
-	btrfs_put_block_group(block_group);
-	btrfs_put_block_group(block_group);
-
-	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
-	if (ret > 0)
-		ret = -EIO;
-	if (ret < 0)
-		goto out;
-
-	ret = btrfs_del_item(trans, root, path);
-	if (ret)
-		goto out;
-
-	if (remove_em) {
-		struct extent_map_tree *em_tree;
-
-		em_tree = &fs_info->mapping_tree;
-		write_lock(&em_tree->lock);
-		remove_extent_mapping(em_tree, em);
-		write_unlock(&em_tree->lock);
-		/* once for the tree */
-		free_extent_map(em);
-	}
-out:
-	if (remove_rsv)
-		btrfs_delayed_refs_rsv_release(fs_info, 1);
-	btrfs_free_path(path);
-	return ret;
-}
-
-struct btrfs_trans_handle *
-btrfs_start_trans_remove_block_group(struct btrfs_fs_info *fs_info,
-				     const u64 chunk_offset)
-{
-	struct extent_map_tree *em_tree = &fs_info->mapping_tree;
-	struct extent_map *em;
-	struct map_lookup *map;
-	unsigned int num_items;
-
-	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
-	read_unlock(&em_tree->lock);
-	ASSERT(em && em->start == chunk_offset);
-
-	/*
-	 * We need to reserve 3 + N units from the metadata space info in order
-	 * to remove a block group (done at btrfs_remove_chunk() and at
-	 * btrfs_remove_block_group()), which are used for:
-	 *
-	 * 1 unit for adding the free space inode's orphan (located in the tree
-	 * of tree roots).
-	 * 1 unit for deleting the block group item (located in the extent
-	 * tree).
-	 * 1 unit for deleting the free space item (located in tree of tree
-	 * roots).
-	 * N units for deleting N device extent items corresponding to each
-	 * stripe (located in the device tree).
-	 *
-	 * In order to remove a block group we also need to reserve units in the
-	 * system space info in order to update the chunk tree (update one or
-	 * more device items and remove one chunk item), but this is done at
-	 * btrfs_remove_chunk() through a call to check_system_chunk().
-	 */
-	map = em->map_lookup;
-	num_items = 3 + map->num_stripes;
-	free_extent_map(em);
-
-	return btrfs_start_transaction_fallback_global_rsv(fs_info->extent_root,
-							   num_items, 1);
-}
-
-/*
- * Process the unused_bgs list and remove any that don't have any allocated
- * space inside of them.
- */
-void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
-{
-	struct btrfs_block_group_cache *block_group;
-	struct btrfs_space_info *space_info;
-	struct btrfs_trans_handle *trans;
-	int ret = 0;
-
-	if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))
-		return;
-
-	spin_lock(&fs_info->unused_bgs_lock);
-	while (!list_empty(&fs_info->unused_bgs)) {
-		u64 start, end;
-		int trimming;
-
-		block_group = list_first_entry(&fs_info->unused_bgs,
-					       struct btrfs_block_group_cache,
-					       bg_list);
-		list_del_init(&block_group->bg_list);
-
-		space_info = block_group->space_info;
-
-		if (ret || btrfs_mixed_space_info(space_info)) {
-			btrfs_put_block_group(block_group);
-			continue;
-		}
-		spin_unlock(&fs_info->unused_bgs_lock);
-
-		mutex_lock(&fs_info->delete_unused_bgs_mutex);
-
-		/* Don't want to race with allocators so take the groups_sem */
-		down_write(&space_info->groups_sem);
-		spin_lock(&block_group->lock);
-		if (block_group->reserved || block_group->pinned ||
-		    btrfs_block_group_used(&block_group->item) ||
-		    block_group->ro ||
-		    list_is_singular(&block_group->list)) {
-			/*
-			 * We want to bail if we made new allocations or have
-			 * outstanding allocations in this block group.  We do
-			 * the ro check in case balance is currently acting on
-			 * this block group.
-			 */
-			trace_btrfs_skip_unused_block_group(block_group);
-			spin_unlock(&block_group->lock);
-			up_write(&space_info->groups_sem);
-			goto next;
-		}
-		spin_unlock(&block_group->lock);
-
-		/* We don't want to force the issue, only flip if it's ok. */
-		ret = __btrfs_inc_block_group_ro(block_group, 0);
-		up_write(&space_info->groups_sem);
-		if (ret < 0) {
-			ret = 0;
-			goto next;
-		}
-
-		/*
-		 * Want to do this before we do anything else so we can recover
-		 * properly if we fail to join the transaction.
-		 */
-		trans = btrfs_start_trans_remove_block_group(fs_info,
-						     block_group->key.objectid);
-		if (IS_ERR(trans)) {
-			btrfs_dec_block_group_ro(block_group);
-			ret = PTR_ERR(trans);
-			goto next;
-		}
-
-		/*
-		 * We could have pending pinned extents for this block group,
-		 * just delete them, we don't care about them anymore.
-		 */
-		start = block_group->key.objectid;
-		end = start + block_group->key.offset - 1;
-		/*
-		 * Hold the unused_bg_unpin_mutex lock to avoid racing with
-		 * btrfs_finish_extent_commit(). If we are at transaction N,
-		 * another task might be running finish_extent_commit() for the
-		 * previous transaction N - 1, and have seen a range belonging
-		 * to the block group in freed_extents[] before we were able to
-		 * clear the whole block group range from freed_extents[]. This
-		 * means that task can lookup for the block group after we
-		 * unpinned it from freed_extents[] and removed it, leading to
-		 * a BUG_ON() at btrfs_unpin_extent_range().
-		 */
-		mutex_lock(&fs_info->unused_bg_unpin_mutex);
-		ret = clear_extent_bits(&fs_info->freed_extents[0], start, end,
-				  EXTENT_DIRTY);
-		if (ret) {
-			mutex_unlock(&fs_info->unused_bg_unpin_mutex);
-			btrfs_dec_block_group_ro(block_group);
-			goto end_trans;
-		}
-		ret = clear_extent_bits(&fs_info->freed_extents[1], start, end,
-				  EXTENT_DIRTY);
-		if (ret) {
-			mutex_unlock(&fs_info->unused_bg_unpin_mutex);
-			btrfs_dec_block_group_ro(block_group);
-			goto end_trans;
-		}
-		mutex_unlock(&fs_info->unused_bg_unpin_mutex);
-
-		/* Reset pinned so btrfs_put_block_group doesn't complain */
-		spin_lock(&space_info->lock);
-		spin_lock(&block_group->lock);
-
-		btrfs_space_info_update_bytes_pinned(fs_info, space_info,
-						     -block_group->pinned);
-		space_info->bytes_readonly += block_group->pinned;
-		percpu_counter_add_batch(&space_info->total_bytes_pinned,
-				   -block_group->pinned,
-				   BTRFS_TOTAL_BYTES_PINNED_BATCH);
-		block_group->pinned = 0;
-
-		spin_unlock(&block_group->lock);
-		spin_unlock(&space_info->lock);
-
-		/* DISCARD can flip during remount */
-		trimming = btrfs_test_opt(fs_info, DISCARD);
-
-		/* Implicit trim during transaction commit. */
-		if (trimming)
-			btrfs_get_block_group_trimming(block_group);
-
-		/*
-		 * Btrfs_remove_chunk will abort the transaction if things go
-		 * horribly wrong.
-		 */
-		ret = btrfs_remove_chunk(trans, block_group->key.objectid);
-
-		if (ret) {
-			if (trimming)
-				btrfs_put_block_group_trimming(block_group);
-			goto end_trans;
-		}
-
-		/*
-		 * If we're not mounted with -odiscard, we can just forget
-		 * about this block group. Otherwise we'll need to wait
-		 * until transaction commit to do the actual discard.
-		 */
-		if (trimming) {
-			spin_lock(&fs_info->unused_bgs_lock);
-			/*
-			 * A concurrent scrub might have added us to the list
-			 * fs_info->unused_bgs, so use a list_move operation
-			 * to add the block group to the deleted_bgs list.
-			 */
-			list_move(&block_group->bg_list,
-				  &trans->transaction->deleted_bgs);
-			spin_unlock(&fs_info->unused_bgs_lock);
-			btrfs_get_block_group(block_group);
-		}
-end_trans:
-		btrfs_end_transaction(trans);
-next:
-		mutex_unlock(&fs_info->delete_unused_bgs_mutex);
-		btrfs_put_block_group(block_group);
-		spin_lock(&fs_info->unused_bgs_lock);
-	}
-	spin_unlock(&fs_info->unused_bgs_lock);
-}
-
 int btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info,
 				   u64 start, u64 end)
 {
@@ -8370,16 +7846,3 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
  * operations while snapshotting is ongoing and that cause the snapshot to be
  * inconsistent (writes followed by expanding truncates for example).
  */
-
-void btrfs_mark_bg_unused(struct btrfs_block_group_cache *bg)
-{
-	struct btrfs_fs_info *fs_info = bg->fs_info;
-
-	spin_lock(&fs_info->unused_bgs_lock);
-	if (list_empty(&bg->bg_list)) {
-		btrfs_get_block_group(bg);
-		trace_btrfs_add_unused_block_group(bg);
-		list_add_tail(&bg->bg_list, &fs_info->unused_bgs);
-	}
-	spin_unlock(&fs_info->unused_bgs_lock);
-}
-- 
2.21.0

[PATCH 04/15] btrfs: export get_alloc_profile

[Josef Bacik]

This gets used directly by a bunch of the block group code, export it to
make it easier to move things around.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/ctree.h       |  1 +
 fs/btrfs/extent-tree.c | 16 ++++++++--------
 2 files changed, 9 insertions(+), 8 deletions(-)

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 266c3cb5f3c9..460f2c53e428 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2467,6 +2467,7 @@ static inline u64 btrfs_calc_trunc_metadata_size(struct btrfs_fs_info *fs_info,
 int btrfs_add_excluded_extent(struct btrfs_fs_info *fs_info,
 			      u64 start, u64 num_bytes);
 void btrfs_free_excluded_extents(struct btrfs_block_group_cache *cache);
+u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags);
 int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 			   unsigned long count);
 void btrfs_cleanup_ref_head_accounting(struct btrfs_fs_info *fs_info,
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index d28e736fdef2..791a2c43a1c0 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -3255,7 +3255,7 @@ static u64 btrfs_reduce_alloc_profile(struct btrfs_fs_info *fs_info, u64 flags)
 	return extended_to_chunk(flags | allowed);
 }
 
-static u64 get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags)
+u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags)
 {
 	unsigned seq;
 	u64 flags;
@@ -3288,23 +3288,23 @@ static u64 get_alloc_profile_by_root(struct btrfs_root *root, int data)
 	else
 		flags = BTRFS_BLOCK_GROUP_METADATA;
 
-	ret = get_alloc_profile(fs_info, flags);
+	ret = btrfs_get_alloc_profile(fs_info, flags);
 	return ret;
 }
 
 u64 btrfs_data_alloc_profile(struct btrfs_fs_info *fs_info)
 {
-	return get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_DATA);
+	return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_DATA);
 }
 
 u64 btrfs_metadata_alloc_profile(struct btrfs_fs_info *fs_info)
 {
-	return get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_METADATA);
+	return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_METADATA);
 }
 
 u64 btrfs_system_alloc_profile(struct btrfs_fs_info *fs_info)
 {
-	return get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
+	return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
 }
 
 static void force_metadata_allocation(struct btrfs_fs_info *info)
@@ -6950,7 +6950,7 @@ int btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache)
 	ret = __btrfs_inc_block_group_ro(cache, 0);
 	if (!ret)
 		goto out;
-	alloc_flags = get_alloc_profile(fs_info, cache->space_info->flags);
+	alloc_flags = btrfs_get_alloc_profile(fs_info, cache->space_info->flags);
 	ret = btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
 	if (ret < 0)
 		goto out;
@@ -6970,7 +6970,7 @@ int btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache)
 
 int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, u64 type)
 {
-	u64 alloc_flags = get_alloc_profile(trans->fs_info, type);
+	u64 alloc_flags = btrfs_get_alloc_profile(trans->fs_info, type);
 
 	return btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
 }
@@ -7495,7 +7495,7 @@ int btrfs_read_block_groups(struct btrfs_fs_info *info)
 	}
 
 	list_for_each_entry_rcu(space_info, &info->space_info, list) {
-		if (!(get_alloc_profile(info, space_info->flags) &
+		if (!(btrfs_get_alloc_profile(info, space_info->flags) &
 		      (BTRFS_BLOCK_GROUP_RAID10 |
 		       BTRFS_BLOCK_GROUP_RAID1_MASK |
 		       BTRFS_BLOCK_GROUP_RAID56_MASK |
-- 
2.21.0

[PATCH 05/15] btrfs: migrate the block group read/creation code

[Josef Bacik]

All of the prep work has been done so we can now cleanly move this chunk
over.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.c | 598 +++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/block-group.h |   5 +
 fs/btrfs/ctree.h       |   5 -
 fs/btrfs/extent-tree.c | 596 ----------------------------------------
 4 files changed, 603 insertions(+), 601 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 0bc51bfe4904..c0ca7d60e5f4 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -10,6 +10,8 @@
 #include "volumes.h"
 #include "transaction.h"
 #include "ref-verify.h"
+#include "sysfs.h"
+#include "tree-log.h"
 
 void btrfs_get_block_group(struct btrfs_block_group_cache *cache)
 {
@@ -36,6 +38,46 @@ void btrfs_put_block_group(struct btrfs_block_group_cache *cache)
 	}
 }
 
+/*
+ * this adds the block group to the fs_info rb tree for the block group
+ * cache
+ */
+static int btrfs_add_block_group_cache(struct btrfs_fs_info *info,
+				struct btrfs_block_group_cache *block_group)
+{
+	struct rb_node **p;
+	struct rb_node *parent = NULL;
+	struct btrfs_block_group_cache *cache;
+
+	spin_lock(&info->block_group_cache_lock);
+	p = &info->block_group_cache_tree.rb_node;
+
+	while (*p) {
+		parent = *p;
+		cache = rb_entry(parent, struct btrfs_block_group_cache,
+				 cache_node);
+		if (block_group->key.objectid < cache->key.objectid) {
+			p = &(*p)->rb_left;
+		} else if (block_group->key.objectid > cache->key.objectid) {
+			p = &(*p)->rb_right;
+		} else {
+			spin_unlock(&info->block_group_cache_lock);
+			return -EEXIST;
+		}
+	}
+
+	rb_link_node(&block_group->cache_node, parent, p);
+	rb_insert_color(&block_group->cache_node,
+			&info->block_group_cache_tree);
+
+	if (info->first_logical_byte > block_group->key.objectid)
+		info->first_logical_byte = block_group->key.objectid;
+
+	spin_unlock(&info->block_group_cache_lock);
+
+	return 0;
+}
+
 /*
  * This will return the block group at or after bytenr if contains is 0, else
  * it will return the block group that contains the bytenr
@@ -1204,3 +1246,559 @@ void btrfs_mark_bg_unused(struct btrfs_block_group_cache *bg)
 	}
 	spin_unlock(&fs_info->unused_bgs_lock);
 }
+
+static int find_first_block_group(struct btrfs_fs_info *fs_info,
+				  struct btrfs_path *path,
+				  struct btrfs_key *key)
+{
+	struct btrfs_root *root = fs_info->extent_root;
+	int ret = 0;
+	struct btrfs_key found_key;
+	struct extent_buffer *leaf;
+	struct btrfs_block_group_item bg;
+	u64 flags;
+	int slot;
+
+	ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
+	if (ret < 0)
+		goto out;
+
+	while (1) {
+		slot = path->slots[0];
+		leaf = path->nodes[0];
+		if (slot >= btrfs_header_nritems(leaf)) {
+			ret = btrfs_next_leaf(root, path);
+			if (ret == 0)
+				continue;
+			if (ret < 0)
+				goto out;
+			break;
+		}
+		btrfs_item_key_to_cpu(leaf, &found_key, slot);
+
+		if (found_key.objectid >= key->objectid &&
+		    found_key.type == BTRFS_BLOCK_GROUP_ITEM_KEY) {
+			struct extent_map_tree *em_tree;
+			struct extent_map *em;
+
+			em_tree = &root->fs_info->mapping_tree;
+			read_lock(&em_tree->lock);
+			em = lookup_extent_mapping(em_tree, found_key.objectid,
+						   found_key.offset);
+			read_unlock(&em_tree->lock);
+			if (!em) {
+				btrfs_err(fs_info,
+			"logical %llu len %llu found bg but no related chunk",
+					  found_key.objectid, found_key.offset);
+				ret = -ENOENT;
+			} else if (em->start != found_key.objectid ||
+				   em->len != found_key.offset) {
+				btrfs_err(fs_info,
+		"block group %llu len %llu mismatch with chunk %llu len %llu",
+					  found_key.objectid, found_key.offset,
+					  em->start, em->len);
+				ret = -EUCLEAN;
+			} else {
+				read_extent_buffer(leaf, &bg,
+					btrfs_item_ptr_offset(leaf, slot),
+					sizeof(bg));
+				flags = btrfs_block_group_flags(&bg) &
+					BTRFS_BLOCK_GROUP_TYPE_MASK;
+
+				if (flags != (em->map_lookup->type &
+					      BTRFS_BLOCK_GROUP_TYPE_MASK)) {
+					btrfs_err(fs_info,
+"block group %llu len %llu type flags 0x%llx mismatch with chunk type flags 0x%llx",
+						found_key.objectid,
+						found_key.offset, flags,
+						(BTRFS_BLOCK_GROUP_TYPE_MASK &
+						 em->map_lookup->type));
+					ret = -EUCLEAN;
+				} else {
+					ret = 0;
+				}
+			}
+			free_extent_map(em);
+			goto out;
+		}
+		path->slots[0]++;
+	}
+out:
+	return ret;
+}
+
+static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
+{
+	u64 extra_flags = chunk_to_extended(flags) &
+				BTRFS_EXTENDED_PROFILE_MASK;
+
+	write_seqlock(&fs_info->profiles_lock);
+	if (flags & BTRFS_BLOCK_GROUP_DATA)
+		fs_info->avail_data_alloc_bits |= extra_flags;
+	if (flags & BTRFS_BLOCK_GROUP_METADATA)
+		fs_info->avail_metadata_alloc_bits |= extra_flags;
+	if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
+		fs_info->avail_system_alloc_bits |= extra_flags;
+	write_sequnlock(&fs_info->profiles_lock);
+}
+
+static int exclude_super_stripes(struct btrfs_block_group_cache *cache)
+{
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	u64 bytenr;
+	u64 *logical;
+	int stripe_len;
+	int i, nr, ret;
+
+	if (cache->key.objectid < BTRFS_SUPER_INFO_OFFSET) {
+		stripe_len = BTRFS_SUPER_INFO_OFFSET - cache->key.objectid;
+		cache->bytes_super += stripe_len;
+		ret = btrfs_add_excluded_extent(fs_info, cache->key.objectid,
+						stripe_len);
+		if (ret)
+			return ret;
+	}
+
+	for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
+		bytenr = btrfs_sb_offset(i);
+		ret = btrfs_rmap_block(fs_info, cache->key.objectid,
+				       bytenr, &logical, &nr, &stripe_len);
+		if (ret)
+			return ret;
+
+		while (nr--) {
+			u64 start, len;
+
+			if (logical[nr] > cache->key.objectid +
+			    cache->key.offset)
+				continue;
+
+			if (logical[nr] + stripe_len <= cache->key.objectid)
+				continue;
+
+			start = logical[nr];
+			if (start < cache->key.objectid) {
+				start = cache->key.objectid;
+				len = (logical[nr] + stripe_len) - start;
+			} else {
+				len = min_t(u64, stripe_len,
+					    cache->key.objectid +
+					    cache->key.offset - start);
+			}
+
+			cache->bytes_super += len;
+			ret = btrfs_add_excluded_extent(fs_info, start, len);
+			if (ret) {
+				kfree(logical);
+				return ret;
+			}
+		}
+
+		kfree(logical);
+	}
+	return 0;
+}
+
+static void link_block_group(struct btrfs_block_group_cache *cache)
+{
+	struct btrfs_space_info *space_info = cache->space_info;
+	int index = btrfs_bg_flags_to_raid_index(cache->flags);
+	bool first = false;
+
+	down_write(&space_info->groups_sem);
+	if (list_empty(&space_info->block_groups[index]))
+		first = true;
+	list_add_tail(&cache->list, &space_info->block_groups[index]);
+	up_write(&space_info->groups_sem);
+
+	if (first)
+		btrfs_sysfs_add_block_group_type(cache);
+}
+
+static struct btrfs_block_group_cache *
+btrfs_create_block_group_cache(struct btrfs_fs_info *fs_info,
+			       u64 start, u64 size)
+{
+	struct btrfs_block_group_cache *cache;
+
+	cache = kzalloc(sizeof(*cache), GFP_NOFS);
+	if (!cache)
+		return NULL;
+
+	cache->free_space_ctl = kzalloc(sizeof(*cache->free_space_ctl),
+					GFP_NOFS);
+	if (!cache->free_space_ctl) {
+		kfree(cache);
+		return NULL;
+	}
+
+	cache->key.objectid = start;
+	cache->key.offset = size;
+	cache->key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
+
+	cache->fs_info = fs_info;
+	cache->full_stripe_len = btrfs_full_stripe_len(fs_info, start);
+	set_free_space_tree_thresholds(cache);
+
+	atomic_set(&cache->count, 1);
+	spin_lock_init(&cache->lock);
+	init_rwsem(&cache->data_rwsem);
+	INIT_LIST_HEAD(&cache->list);
+	INIT_LIST_HEAD(&cache->cluster_list);
+	INIT_LIST_HEAD(&cache->bg_list);
+	INIT_LIST_HEAD(&cache->ro_list);
+	INIT_LIST_HEAD(&cache->dirty_list);
+	INIT_LIST_HEAD(&cache->io_list);
+	btrfs_init_free_space_ctl(cache);
+	atomic_set(&cache->trimming, 0);
+	mutex_init(&cache->free_space_lock);
+	btrfs_init_full_stripe_locks_tree(&cache->full_stripe_locks_root);
+
+	return cache;
+}
+
+
+/*
+ * Iterate all chunks and verify that each of them has the corresponding block
+ * group
+ */
+static int check_chunk_block_group_mappings(struct btrfs_fs_info *fs_info)
+{
+	struct extent_map_tree *map_tree = &fs_info->mapping_tree;
+	struct extent_map *em;
+	struct btrfs_block_group_cache *bg;
+	u64 start = 0;
+	int ret = 0;
+
+	while (1) {
+		read_lock(&map_tree->lock);
+		/*
+		 * lookup_extent_mapping will return the first extent map
+		 * intersecting the range, so setting @len to 1 is enough to
+		 * get the first chunk.
+		 */
+		em = lookup_extent_mapping(map_tree, start, 1);
+		read_unlock(&map_tree->lock);
+		if (!em)
+			break;
+
+		bg = btrfs_lookup_block_group(fs_info, em->start);
+		if (!bg) {
+			btrfs_err(fs_info,
+	"chunk start=%llu len=%llu doesn't have corresponding block group",
+				     em->start, em->len);
+			ret = -EUCLEAN;
+			free_extent_map(em);
+			break;
+		}
+		if (bg->key.objectid != em->start ||
+		    bg->key.offset != em->len ||
+		    (bg->flags & BTRFS_BLOCK_GROUP_TYPE_MASK) !=
+		    (em->map_lookup->type & BTRFS_BLOCK_GROUP_TYPE_MASK)) {
+			btrfs_err(fs_info,
+"chunk start=%llu len=%llu flags=0x%llx doesn't match block group start=%llu len=%llu flags=0x%llx",
+				em->start, em->len,
+				em->map_lookup->type & BTRFS_BLOCK_GROUP_TYPE_MASK,
+				bg->key.objectid, bg->key.offset,
+				bg->flags & BTRFS_BLOCK_GROUP_TYPE_MASK);
+			ret = -EUCLEAN;
+			free_extent_map(em);
+			btrfs_put_block_group(bg);
+			break;
+		}
+		start = em->start + em->len;
+		free_extent_map(em);
+		btrfs_put_block_group(bg);
+	}
+	return ret;
+}
+
+int btrfs_read_block_groups(struct btrfs_fs_info *info)
+{
+	struct btrfs_path *path;
+	int ret;
+	struct btrfs_block_group_cache *cache;
+	struct btrfs_space_info *space_info;
+	struct btrfs_key key;
+	struct btrfs_key found_key;
+	struct extent_buffer *leaf;
+	int need_clear = 0;
+	u64 cache_gen;
+	u64 feature;
+	int mixed;
+
+	feature = btrfs_super_incompat_flags(info->super_copy);
+	mixed = !!(feature & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS);
+
+	key.objectid = 0;
+	key.offset = 0;
+	key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+	path->reada = READA_FORWARD;
+
+	cache_gen = btrfs_super_cache_generation(info->super_copy);
+	if (btrfs_test_opt(info, SPACE_CACHE) &&
+	    btrfs_super_generation(info->super_copy) != cache_gen)
+		need_clear = 1;
+	if (btrfs_test_opt(info, CLEAR_CACHE))
+		need_clear = 1;
+
+	while (1) {
+		ret = find_first_block_group(info, path, &key);
+		if (ret > 0)
+			break;
+		if (ret != 0)
+			goto error;
+
+		leaf = path->nodes[0];
+		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
+
+		cache = btrfs_create_block_group_cache(info, found_key.objectid,
+						       found_key.offset);
+		if (!cache) {
+			ret = -ENOMEM;
+			goto error;
+		}
+
+		if (need_clear) {
+			/*
+			 * When we mount with old space cache, we need to
+			 * set BTRFS_DC_CLEAR and set dirty flag.
+			 *
+			 * a) Setting 'BTRFS_DC_CLEAR' makes sure that we
+			 *    truncate the old free space cache inode and
+			 *    setup a new one.
+			 * b) Setting 'dirty flag' makes sure that we flush
+			 *    the new space cache info onto disk.
+			 */
+			if (btrfs_test_opt(info, SPACE_CACHE))
+				cache->disk_cache_state = BTRFS_DC_CLEAR;
+		}
+
+		read_extent_buffer(leaf, &cache->item,
+				   btrfs_item_ptr_offset(leaf, path->slots[0]),
+				   sizeof(cache->item));
+		cache->flags = btrfs_block_group_flags(&cache->item);
+		if (!mixed &&
+		    ((cache->flags & BTRFS_BLOCK_GROUP_METADATA) &&
+		    (cache->flags & BTRFS_BLOCK_GROUP_DATA))) {
+			btrfs_err(info,
+"bg %llu is a mixed block group but filesystem hasn't enabled mixed block groups",
+				  cache->key.objectid);
+			ret = -EINVAL;
+			goto error;
+		}
+
+		key.objectid = found_key.objectid + found_key.offset;
+		btrfs_release_path(path);
+
+		/*
+		 * We need to exclude the super stripes now so that the space
+		 * info has super bytes accounted for, otherwise we'll think
+		 * we have more space than we actually do.
+		 */
+		ret = exclude_super_stripes(cache);
+		if (ret) {
+			/*
+			 * We may have excluded something, so call this just in
+			 * case.
+			 */
+			btrfs_free_excluded_extents(cache);
+			btrfs_put_block_group(cache);
+			goto error;
+		}
+
+		/*
+		 * check for two cases, either we are full, and therefore
+		 * don't need to bother with the caching work since we won't
+		 * find any space, or we are empty, and we can just add all
+		 * the space in and be done with it.  This saves us _a_lot_ of
+		 * time, particularly in the full case.
+		 */
+		if (found_key.offset == btrfs_block_group_used(&cache->item)) {
+			cache->last_byte_to_unpin = (u64)-1;
+			cache->cached = BTRFS_CACHE_FINISHED;
+			btrfs_free_excluded_extents(cache);
+		} else if (btrfs_block_group_used(&cache->item) == 0) {
+			cache->last_byte_to_unpin = (u64)-1;
+			cache->cached = BTRFS_CACHE_FINISHED;
+			add_new_free_space(cache, found_key.objectid,
+					   found_key.objectid +
+					   found_key.offset);
+			btrfs_free_excluded_extents(cache);
+		}
+
+		ret = btrfs_add_block_group_cache(info, cache);
+		if (ret) {
+			btrfs_remove_free_space_cache(cache);
+			btrfs_put_block_group(cache);
+			goto error;
+		}
+
+		trace_btrfs_add_block_group(info, cache, 0);
+		btrfs_update_space_info(info, cache->flags, found_key.offset,
+					btrfs_block_group_used(&cache->item),
+					cache->bytes_super, &space_info);
+
+		cache->space_info = space_info;
+
+		link_block_group(cache);
+
+		set_avail_alloc_bits(info, cache->flags);
+		if (btrfs_chunk_readonly(info, cache->key.objectid)) {
+			__btrfs_inc_block_group_ro(cache, 1);
+		} else if (btrfs_block_group_used(&cache->item) == 0) {
+			ASSERT(list_empty(&cache->bg_list));
+			btrfs_mark_bg_unused(cache);
+		}
+	}
+
+	list_for_each_entry_rcu(space_info, &info->space_info, list) {
+		if (!(btrfs_get_alloc_profile(info, space_info->flags) &
+		      (BTRFS_BLOCK_GROUP_RAID10 |
+		       BTRFS_BLOCK_GROUP_RAID1_MASK |
+		       BTRFS_BLOCK_GROUP_RAID56_MASK |
+		       BTRFS_BLOCK_GROUP_DUP)))
+			continue;
+		/*
+		 * avoid allocating from un-mirrored block group if there are
+		 * mirrored block groups.
+		 */
+		list_for_each_entry(cache,
+				&space_info->block_groups[BTRFS_RAID_RAID0],
+				list)
+			__btrfs_inc_block_group_ro(cache, 1);
+		list_for_each_entry(cache,
+				&space_info->block_groups[BTRFS_RAID_SINGLE],
+				list)
+			__btrfs_inc_block_group_ro(cache, 1);
+	}
+
+	btrfs_add_raid_kobjects(info);
+	btrfs_init_global_block_rsv(info);
+	ret = check_chunk_block_group_mappings(info);
+error:
+	btrfs_free_path(path);
+	return ret;
+}
+
+void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_block_group_cache *block_group;
+	struct btrfs_root *extent_root = fs_info->extent_root;
+	struct btrfs_block_group_item item;
+	struct btrfs_key key;
+	int ret = 0;
+
+	if (!trans->can_flush_pending_bgs)
+		return;
+
+	while (!list_empty(&trans->new_bgs)) {
+		block_group = list_first_entry(&trans->new_bgs,
+					       struct btrfs_block_group_cache,
+					       bg_list);
+		if (ret)
+			goto next;
+
+		spin_lock(&block_group->lock);
+		memcpy(&item, &block_group->item, sizeof(item));
+		memcpy(&key, &block_group->key, sizeof(key));
+		spin_unlock(&block_group->lock);
+
+		ret = btrfs_insert_item(trans, extent_root, &key, &item,
+					sizeof(item));
+		if (ret)
+			btrfs_abort_transaction(trans, ret);
+		ret = btrfs_finish_chunk_alloc(trans, key.objectid, key.offset);
+		if (ret)
+			btrfs_abort_transaction(trans, ret);
+		add_block_group_free_space(trans, block_group);
+		/* already aborted the transaction if it failed. */
+next:
+		btrfs_delayed_refs_rsv_release(fs_info, 1);
+		list_del_init(&block_group->bg_list);
+	}
+	btrfs_trans_release_chunk_metadata(trans);
+}
+
+int btrfs_make_block_group(struct btrfs_trans_handle *trans, u64 bytes_used,
+			   u64 type, u64 chunk_offset, u64 size)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_block_group_cache *cache;
+	int ret;
+
+	btrfs_set_log_full_commit(trans);
+
+	cache = btrfs_create_block_group_cache(fs_info, chunk_offset, size);
+	if (!cache)
+		return -ENOMEM;
+
+	btrfs_set_block_group_used(&cache->item, bytes_used);
+	btrfs_set_block_group_chunk_objectid(&cache->item,
+					     BTRFS_FIRST_CHUNK_TREE_OBJECTID);
+	btrfs_set_block_group_flags(&cache->item, type);
+
+	cache->flags = type;
+	cache->last_byte_to_unpin = (u64)-1;
+	cache->cached = BTRFS_CACHE_FINISHED;
+	cache->needs_free_space = 1;
+	ret = exclude_super_stripes(cache);
+	if (ret) {
+		/*
+		 * We may have excluded something, so call this just in
+		 * case.
+		 */
+		btrfs_free_excluded_extents(cache);
+		btrfs_put_block_group(cache);
+		return ret;
+	}
+
+	add_new_free_space(cache, chunk_offset, chunk_offset + size);
+
+	btrfs_free_excluded_extents(cache);
+
+#ifdef CONFIG_BTRFS_DEBUG
+	if (btrfs_should_fragment_free_space(cache)) {
+		u64 new_bytes_used = size - bytes_used;
+
+		bytes_used += new_bytes_used >> 1;
+		btrfs_fragment_free_space(cache);
+	}
+#endif
+	/*
+	 * Ensure the corresponding space_info object is created and
+	 * assigned to our block group. We want our bg to be added to the rbtree
+	 * with its ->space_info set.
+	 */
+	cache->space_info = btrfs_find_space_info(fs_info, cache->flags);
+	ASSERT(cache->space_info);
+
+	ret = btrfs_add_block_group_cache(fs_info, cache);
+	if (ret) {
+		btrfs_remove_free_space_cache(cache);
+		btrfs_put_block_group(cache);
+		return ret;
+	}
+
+	/*
+	 * Now that our block group has its ->space_info set and is inserted in
+	 * the rbtree, update the space info's counters.
+	 */
+	trace_btrfs_add_block_group(fs_info, cache, 1);
+	btrfs_update_space_info(fs_info, cache->flags, size, bytes_used,
+				cache->bytes_super, &cache->space_info);
+	btrfs_update_global_block_rsv(fs_info);
+
+	link_block_group(cache);
+
+	list_add_tail(&cache->bg_list, &trans->new_bgs);
+	trans->delayed_ref_updates++;
+	btrfs_update_delayed_refs_rsv(trans);
+
+	set_avail_alloc_bits(fs_info, type);
+	return 0;
+}
diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index 5ee2ea695ab0..2745b99e5ae2 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -183,6 +183,11 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 			     u64 group_start, struct extent_map *em);
 void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info);
 void btrfs_mark_bg_unused(struct btrfs_block_group_cache *bg);
+int btrfs_read_block_groups(struct btrfs_fs_info *info);
+int btrfs_make_block_group(struct btrfs_trans_handle *trans,
+			   u64 bytes_used, u64 type, u64 chunk_offset,
+			   u64 size);
+void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans);
 
 static inline int btrfs_block_group_cache_done(
 		struct btrfs_block_group_cache *cache)
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 460f2c53e428..1cfe7adddc97 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2527,13 +2527,8 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans);
 int btrfs_setup_space_cache(struct btrfs_trans_handle *trans);
 int btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr);
 int btrfs_free_block_groups(struct btrfs_fs_info *info);
-int btrfs_read_block_groups(struct btrfs_fs_info *info);
-int btrfs_make_block_group(struct btrfs_trans_handle *trans,
-			   u64 bytes_used, u64 type, u64 chunk_offset,
-			   u64 size);
 void btrfs_get_block_group_trimming(struct btrfs_block_group_cache *cache);
 void btrfs_put_block_group_trimming(struct btrfs_block_group_cache *cache);
-void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans);
 u64 btrfs_data_alloc_profile(struct btrfs_fs_info *fs_info);
 u64 btrfs_metadata_alloc_profile(struct btrfs_fs_info *fs_info);
 u64 btrfs_system_alloc_profile(struct btrfs_fs_info *fs_info);
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 791a2c43a1c0..37556ce36d68 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -59,46 +59,6 @@ static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
 	return (cache->flags & bits) == bits;
 }
 
-/*
- * this adds the block group to the fs_info rb tree for the block group
- * cache
- */
-static int btrfs_add_block_group_cache(struct btrfs_fs_info *info,
-				struct btrfs_block_group_cache *block_group)
-{
-	struct rb_node **p;
-	struct rb_node *parent = NULL;
-	struct btrfs_block_group_cache *cache;
-
-	spin_lock(&info->block_group_cache_lock);
-	p = &info->block_group_cache_tree.rb_node;
-
-	while (*p) {
-		parent = *p;
-		cache = rb_entry(parent, struct btrfs_block_group_cache,
-				 cache_node);
-		if (block_group->key.objectid < cache->key.objectid) {
-			p = &(*p)->rb_left;
-		} else if (block_group->key.objectid > cache->key.objectid) {
-			p = &(*p)->rb_right;
-		} else {
-			spin_unlock(&info->block_group_cache_lock);
-			return -EEXIST;
-		}
-	}
-
-	rb_link_node(&block_group->cache_node, parent, p);
-	rb_insert_color(&block_group->cache_node,
-			&info->block_group_cache_tree);
-
-	if (info->first_logical_byte > block_group->key.objectid)
-		info->first_logical_byte = block_group->key.objectid;
-
-	spin_unlock(&info->block_group_cache_lock);
-
-	return 0;
-}
-
 int btrfs_add_excluded_extent(struct btrfs_fs_info *fs_info,
 			      u64 start, u64 num_bytes)
 {
@@ -124,63 +84,6 @@ void btrfs_free_excluded_extents(struct btrfs_block_group_cache *cache)
 			  start, end, EXTENT_UPTODATE);
 }
 
-static int exclude_super_stripes(struct btrfs_block_group_cache *cache)
-{
-	struct btrfs_fs_info *fs_info = cache->fs_info;
-	u64 bytenr;
-	u64 *logical;
-	int stripe_len;
-	int i, nr, ret;
-
-	if (cache->key.objectid < BTRFS_SUPER_INFO_OFFSET) {
-		stripe_len = BTRFS_SUPER_INFO_OFFSET - cache->key.objectid;
-		cache->bytes_super += stripe_len;
-		ret = btrfs_add_excluded_extent(fs_info, cache->key.objectid,
-						stripe_len);
-		if (ret)
-			return ret;
-	}
-
-	for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
-		bytenr = btrfs_sb_offset(i);
-		ret = btrfs_rmap_block(fs_info, cache->key.objectid,
-				       bytenr, &logical, &nr, &stripe_len);
-		if (ret)
-			return ret;
-
-		while (nr--) {
-			u64 start, len;
-
-			if (logical[nr] > cache->key.objectid +
-			    cache->key.offset)
-				continue;
-
-			if (logical[nr] + stripe_len <= cache->key.objectid)
-				continue;
-
-			start = logical[nr];
-			if (start < cache->key.objectid) {
-				start = cache->key.objectid;
-				len = (logical[nr] + stripe_len) - start;
-			} else {
-				len = min_t(u64, stripe_len,
-					    cache->key.objectid +
-					    cache->key.offset - start);
-			}
-
-			cache->bytes_super += len;
-			ret = btrfs_add_excluded_extent(fs_info, start, len);
-			if (ret) {
-				kfree(logical);
-				return ret;
-			}
-		}
-
-		kfree(logical);
-	}
-	return 0;
-}
-
 static u64 generic_ref_to_space_flags(struct btrfs_ref *ref)
 {
 	if (ref->type == BTRFS_REF_METADATA) {
@@ -3160,21 +3063,6 @@ int btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr)
 	return readonly;
 }
 
-static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
-{
-	u64 extra_flags = chunk_to_extended(flags) &
-				BTRFS_EXTENDED_PROFILE_MASK;
-
-	write_seqlock(&fs_info->profiles_lock);
-	if (flags & BTRFS_BLOCK_GROUP_DATA)
-		fs_info->avail_data_alloc_bits |= extra_flags;
-	if (flags & BTRFS_BLOCK_GROUP_METADATA)
-		fs_info->avail_metadata_alloc_bits |= extra_flags;
-	if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
-		fs_info->avail_system_alloc_bits |= extra_flags;
-	write_sequnlock(&fs_info->profiles_lock);
-}
-
 /*
  * returns target flags in extended format or 0 if restripe for this
  * chunk_type is not in progress
@@ -7030,86 +6918,6 @@ void btrfs_dec_block_group_ro(struct btrfs_block_group_cache *cache)
 	spin_unlock(&sinfo->lock);
 }
 
-static int find_first_block_group(struct btrfs_fs_info *fs_info,
-				  struct btrfs_path *path,
-				  struct btrfs_key *key)
-{
-	struct btrfs_root *root = fs_info->extent_root;
-	int ret = 0;
-	struct btrfs_key found_key;
-	struct extent_buffer *leaf;
-	struct btrfs_block_group_item bg;
-	u64 flags;
-	int slot;
-
-	ret = btrfs_search_slot(NULL, root, key, path, 0, 0);
-	if (ret < 0)
-		goto out;
-
-	while (1) {
-		slot = path->slots[0];
-		leaf = path->nodes[0];
-		if (slot >= btrfs_header_nritems(leaf)) {
-			ret = btrfs_next_leaf(root, path);
-			if (ret == 0)
-				continue;
-			if (ret < 0)
-				goto out;
-			break;
-		}
-		btrfs_item_key_to_cpu(leaf, &found_key, slot);
-
-		if (found_key.objectid >= key->objectid &&
-		    found_key.type == BTRFS_BLOCK_GROUP_ITEM_KEY) {
-			struct extent_map_tree *em_tree;
-			struct extent_map *em;
-
-			em_tree = &root->fs_info->mapping_tree;
-			read_lock(&em_tree->lock);
-			em = lookup_extent_mapping(em_tree, found_key.objectid,
-						   found_key.offset);
-			read_unlock(&em_tree->lock);
-			if (!em) {
-				btrfs_err(fs_info,
-			"logical %llu len %llu found bg but no related chunk",
-					  found_key.objectid, found_key.offset);
-				ret = -ENOENT;
-			} else if (em->start != found_key.objectid ||
-				   em->len != found_key.offset) {
-				btrfs_err(fs_info,
-		"block group %llu len %llu mismatch with chunk %llu len %llu",
-					  found_key.objectid, found_key.offset,
-					  em->start, em->len);
-				ret = -EUCLEAN;
-			} else {
-				read_extent_buffer(leaf, &bg,
-					btrfs_item_ptr_offset(leaf, slot),
-					sizeof(bg));
-				flags = btrfs_block_group_flags(&bg) &
-					BTRFS_BLOCK_GROUP_TYPE_MASK;
-
-				if (flags != (em->map_lookup->type &
-					      BTRFS_BLOCK_GROUP_TYPE_MASK)) {
-					btrfs_err(fs_info,
-"block group %llu len %llu type flags 0x%llx mismatch with chunk type flags 0x%llx",
-						found_key.objectid,
-						found_key.offset, flags,
-						(BTRFS_BLOCK_GROUP_TYPE_MASK &
-						 em->map_lookup->type));
-					ret = -EUCLEAN;
-				} else {
-					ret = 0;
-				}
-			}
-			free_extent_map(em);
-			goto out;
-		}
-		path->slots[0]++;
-	}
-out:
-	return ret;
-}
-
 void btrfs_put_block_group_cache(struct btrfs_fs_info *info)
 {
 	struct btrfs_block_group_cache *block_group;
@@ -7238,410 +7046,6 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 	return 0;
 }
 
-static void link_block_group(struct btrfs_block_group_cache *cache)
-{
-	struct btrfs_space_info *space_info = cache->space_info;
-	int index = btrfs_bg_flags_to_raid_index(cache->flags);
-	bool first = false;
-
-	down_write(&space_info->groups_sem);
-	if (list_empty(&space_info->block_groups[index]))
-		first = true;
-	list_add_tail(&cache->list, &space_info->block_groups[index]);
-	up_write(&space_info->groups_sem);
-
-	if (first)
-		btrfs_sysfs_add_block_group_type(cache);
-}
-
-static struct btrfs_block_group_cache *
-btrfs_create_block_group_cache(struct btrfs_fs_info *fs_info,
-			       u64 start, u64 size)
-{
-	struct btrfs_block_group_cache *cache;
-
-	cache = kzalloc(sizeof(*cache), GFP_NOFS);
-	if (!cache)
-		return NULL;
-
-	cache->free_space_ctl = kzalloc(sizeof(*cache->free_space_ctl),
-					GFP_NOFS);
-	if (!cache->free_space_ctl) {
-		kfree(cache);
-		return NULL;
-	}
-
-	cache->key.objectid = start;
-	cache->key.offset = size;
-	cache->key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
-
-	cache->fs_info = fs_info;
-	cache->full_stripe_len = btrfs_full_stripe_len(fs_info, start);
-	set_free_space_tree_thresholds(cache);
-
-	atomic_set(&cache->count, 1);
-	spin_lock_init(&cache->lock);
-	init_rwsem(&cache->data_rwsem);
-	INIT_LIST_HEAD(&cache->list);
-	INIT_LIST_HEAD(&cache->cluster_list);
-	INIT_LIST_HEAD(&cache->bg_list);
-	INIT_LIST_HEAD(&cache->ro_list);
-	INIT_LIST_HEAD(&cache->dirty_list);
-	INIT_LIST_HEAD(&cache->io_list);
-	btrfs_init_free_space_ctl(cache);
-	atomic_set(&cache->trimming, 0);
-	mutex_init(&cache->free_space_lock);
-	btrfs_init_full_stripe_locks_tree(&cache->full_stripe_locks_root);
-
-	return cache;
-}
-
-
-/*
- * Iterate all chunks and verify that each of them has the corresponding block
- * group
- */
-static int check_chunk_block_group_mappings(struct btrfs_fs_info *fs_info)
-{
-	struct extent_map_tree *map_tree = &fs_info->mapping_tree;
-	struct extent_map *em;
-	struct btrfs_block_group_cache *bg;
-	u64 start = 0;
-	int ret = 0;
-
-	while (1) {
-		read_lock(&map_tree->lock);
-		/*
-		 * lookup_extent_mapping will return the first extent map
-		 * intersecting the range, so setting @len to 1 is enough to
-		 * get the first chunk.
-		 */
-		em = lookup_extent_mapping(map_tree, start, 1);
-		read_unlock(&map_tree->lock);
-		if (!em)
-			break;
-
-		bg = btrfs_lookup_block_group(fs_info, em->start);
-		if (!bg) {
-			btrfs_err(fs_info,
-	"chunk start=%llu len=%llu doesn't have corresponding block group",
-				     em->start, em->len);
-			ret = -EUCLEAN;
-			free_extent_map(em);
-			break;
-		}
-		if (bg->key.objectid != em->start ||
-		    bg->key.offset != em->len ||
-		    (bg->flags & BTRFS_BLOCK_GROUP_TYPE_MASK) !=
-		    (em->map_lookup->type & BTRFS_BLOCK_GROUP_TYPE_MASK)) {
-			btrfs_err(fs_info,
-"chunk start=%llu len=%llu flags=0x%llx doesn't match block group start=%llu len=%llu flags=0x%llx",
-				em->start, em->len,
-				em->map_lookup->type & BTRFS_BLOCK_GROUP_TYPE_MASK,
-				bg->key.objectid, bg->key.offset,
-				bg->flags & BTRFS_BLOCK_GROUP_TYPE_MASK);
-			ret = -EUCLEAN;
-			free_extent_map(em);
-			btrfs_put_block_group(bg);
-			break;
-		}
-		start = em->start + em->len;
-		free_extent_map(em);
-		btrfs_put_block_group(bg);
-	}
-	return ret;
-}
-
-int btrfs_read_block_groups(struct btrfs_fs_info *info)
-{
-	struct btrfs_path *path;
-	int ret;
-	struct btrfs_block_group_cache *cache;
-	struct btrfs_space_info *space_info;
-	struct btrfs_key key;
-	struct btrfs_key found_key;
-	struct extent_buffer *leaf;
-	int need_clear = 0;
-	u64 cache_gen;
-	u64 feature;
-	int mixed;
-
-	feature = btrfs_super_incompat_flags(info->super_copy);
-	mixed = !!(feature & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS);
-
-	key.objectid = 0;
-	key.offset = 0;
-	key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-	path->reada = READA_FORWARD;
-
-	cache_gen = btrfs_super_cache_generation(info->super_copy);
-	if (btrfs_test_opt(info, SPACE_CACHE) &&
-	    btrfs_super_generation(info->super_copy) != cache_gen)
-		need_clear = 1;
-	if (btrfs_test_opt(info, CLEAR_CACHE))
-		need_clear = 1;
-
-	while (1) {
-		ret = find_first_block_group(info, path, &key);
-		if (ret > 0)
-			break;
-		if (ret != 0)
-			goto error;
-
-		leaf = path->nodes[0];
-		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
-
-		cache = btrfs_create_block_group_cache(info, found_key.objectid,
-						       found_key.offset);
-		if (!cache) {
-			ret = -ENOMEM;
-			goto error;
-		}
-
-		if (need_clear) {
-			/*
-			 * When we mount with old space cache, we need to
-			 * set BTRFS_DC_CLEAR and set dirty flag.
-			 *
-			 * a) Setting 'BTRFS_DC_CLEAR' makes sure that we
-			 *    truncate the old free space cache inode and
-			 *    setup a new one.
-			 * b) Setting 'dirty flag' makes sure that we flush
-			 *    the new space cache info onto disk.
-			 */
-			if (btrfs_test_opt(info, SPACE_CACHE))
-				cache->disk_cache_state = BTRFS_DC_CLEAR;
-		}
-
-		read_extent_buffer(leaf, &cache->item,
-				   btrfs_item_ptr_offset(leaf, path->slots[0]),
-				   sizeof(cache->item));
-		cache->flags = btrfs_block_group_flags(&cache->item);
-		if (!mixed &&
-		    ((cache->flags & BTRFS_BLOCK_GROUP_METADATA) &&
-		    (cache->flags & BTRFS_BLOCK_GROUP_DATA))) {
-			btrfs_err(info,
-"bg %llu is a mixed block group but filesystem hasn't enabled mixed block groups",
-				  cache->key.objectid);
-			ret = -EINVAL;
-			goto error;
-		}
-
-		key.objectid = found_key.objectid + found_key.offset;
-		btrfs_release_path(path);
-
-		/*
-		 * We need to exclude the super stripes now so that the space
-		 * info has super bytes accounted for, otherwise we'll think
-		 * we have more space than we actually do.
-		 */
-		ret = exclude_super_stripes(cache);
-		if (ret) {
-			/*
-			 * We may have excluded something, so call this just in
-			 * case.
-			 */
-			btrfs_free_excluded_extents(cache);
-			btrfs_put_block_group(cache);
-			goto error;
-		}
-
-		/*
-		 * check for two cases, either we are full, and therefore
-		 * don't need to bother with the caching work since we won't
-		 * find any space, or we are empty, and we can just add all
-		 * the space in and be done with it.  This saves us _a_lot_ of
-		 * time, particularly in the full case.
-		 */
-		if (found_key.offset == btrfs_block_group_used(&cache->item)) {
-			cache->last_byte_to_unpin = (u64)-1;
-			cache->cached = BTRFS_CACHE_FINISHED;
-			btrfs_free_excluded_extents(cache);
-		} else if (btrfs_block_group_used(&cache->item) == 0) {
-			cache->last_byte_to_unpin = (u64)-1;
-			cache->cached = BTRFS_CACHE_FINISHED;
-			add_new_free_space(cache, found_key.objectid,
-					   found_key.objectid +
-					   found_key.offset);
-			btrfs_free_excluded_extents(cache);
-		}
-
-		ret = btrfs_add_block_group_cache(info, cache);
-		if (ret) {
-			btrfs_remove_free_space_cache(cache);
-			btrfs_put_block_group(cache);
-			goto error;
-		}
-
-		trace_btrfs_add_block_group(info, cache, 0);
-		btrfs_update_space_info(info, cache->flags, found_key.offset,
-					btrfs_block_group_used(&cache->item),
-					cache->bytes_super, &space_info);
-
-		cache->space_info = space_info;
-
-		link_block_group(cache);
-
-		set_avail_alloc_bits(info, cache->flags);
-		if (btrfs_chunk_readonly(info, cache->key.objectid)) {
-			__btrfs_inc_block_group_ro(cache, 1);
-		} else if (btrfs_block_group_used(&cache->item) == 0) {
-			ASSERT(list_empty(&cache->bg_list));
-			btrfs_mark_bg_unused(cache);
-		}
-	}
-
-	list_for_each_entry_rcu(space_info, &info->space_info, list) {
-		if (!(btrfs_get_alloc_profile(info, space_info->flags) &
-		      (BTRFS_BLOCK_GROUP_RAID10 |
-		       BTRFS_BLOCK_GROUP_RAID1_MASK |
-		       BTRFS_BLOCK_GROUP_RAID56_MASK |
-		       BTRFS_BLOCK_GROUP_DUP)))
-			continue;
-		/*
-		 * avoid allocating from un-mirrored block group if there are
-		 * mirrored block groups.
-		 */
-		list_for_each_entry(cache,
-				&space_info->block_groups[BTRFS_RAID_RAID0],
-				list)
-			__btrfs_inc_block_group_ro(cache, 1);
-		list_for_each_entry(cache,
-				&space_info->block_groups[BTRFS_RAID_SINGLE],
-				list)
-			__btrfs_inc_block_group_ro(cache, 1);
-	}
-
-	btrfs_add_raid_kobjects(info);
-	btrfs_init_global_block_rsv(info);
-	ret = check_chunk_block_group_mappings(info);
-error:
-	btrfs_free_path(path);
-	return ret;
-}
-
-void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans)
-{
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_block_group_cache *block_group;
-	struct btrfs_root *extent_root = fs_info->extent_root;
-	struct btrfs_block_group_item item;
-	struct btrfs_key key;
-	int ret = 0;
-
-	if (!trans->can_flush_pending_bgs)
-		return;
-
-	while (!list_empty(&trans->new_bgs)) {
-		block_group = list_first_entry(&trans->new_bgs,
-					       struct btrfs_block_group_cache,
-					       bg_list);
-		if (ret)
-			goto next;
-
-		spin_lock(&block_group->lock);
-		memcpy(&item, &block_group->item, sizeof(item));
-		memcpy(&key, &block_group->key, sizeof(key));
-		spin_unlock(&block_group->lock);
-
-		ret = btrfs_insert_item(trans, extent_root, &key, &item,
-					sizeof(item));
-		if (ret)
-			btrfs_abort_transaction(trans, ret);
-		ret = btrfs_finish_chunk_alloc(trans, key.objectid, key.offset);
-		if (ret)
-			btrfs_abort_transaction(trans, ret);
-		add_block_group_free_space(trans, block_group);
-		/* already aborted the transaction if it failed. */
-next:
-		btrfs_delayed_refs_rsv_release(fs_info, 1);
-		list_del_init(&block_group->bg_list);
-	}
-	btrfs_trans_release_chunk_metadata(trans);
-}
-
-int btrfs_make_block_group(struct btrfs_trans_handle *trans, u64 bytes_used,
-			   u64 type, u64 chunk_offset, u64 size)
-{
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_block_group_cache *cache;
-	int ret;
-
-	btrfs_set_log_full_commit(trans);
-
-	cache = btrfs_create_block_group_cache(fs_info, chunk_offset, size);
-	if (!cache)
-		return -ENOMEM;
-
-	btrfs_set_block_group_used(&cache->item, bytes_used);
-	btrfs_set_block_group_chunk_objectid(&cache->item,
-					     BTRFS_FIRST_CHUNK_TREE_OBJECTID);
-	btrfs_set_block_group_flags(&cache->item, type);
-
-	cache->flags = type;
-	cache->last_byte_to_unpin = (u64)-1;
-	cache->cached = BTRFS_CACHE_FINISHED;
-	cache->needs_free_space = 1;
-	ret = exclude_super_stripes(cache);
-	if (ret) {
-		/*
-		 * We may have excluded something, so call this just in
-		 * case.
-		 */
-		btrfs_free_excluded_extents(cache);
-		btrfs_put_block_group(cache);
-		return ret;
-	}
-
-	add_new_free_space(cache, chunk_offset, chunk_offset + size);
-
-	btrfs_free_excluded_extents(cache);
-
-#ifdef CONFIG_BTRFS_DEBUG
-	if (btrfs_should_fragment_free_space(cache)) {
-		u64 new_bytes_used = size - bytes_used;
-
-		bytes_used += new_bytes_used >> 1;
-		btrfs_fragment_free_space(cache);
-	}
-#endif
-	/*
-	 * Ensure the corresponding space_info object is created and
-	 * assigned to our block group. We want our bg to be added to the rbtree
-	 * with its ->space_info set.
-	 */
-	cache->space_info = btrfs_find_space_info(fs_info, cache->flags);
-	ASSERT(cache->space_info);
-
-	ret = btrfs_add_block_group_cache(fs_info, cache);
-	if (ret) {
-		btrfs_remove_free_space_cache(cache);
-		btrfs_put_block_group(cache);
-		return ret;
-	}
-
-	/*
-	 * Now that our block group has its ->space_info set and is inserted in
-	 * the rbtree, update the space info's counters.
-	 */
-	trace_btrfs_add_block_group(fs_info, cache, 1);
-	btrfs_update_space_info(fs_info, cache->flags, size, bytes_used,
-				cache->bytes_super, &cache->space_info);
-	btrfs_update_global_block_rsv(fs_info);
-
-	link_block_group(cache);
-
-	list_add_tail(&cache->bg_list, &trans->new_bgs);
-	trans->delayed_ref_updates++;
-	btrfs_update_delayed_refs_rsv(trans);
-
-	set_avail_alloc_bits(fs_info, type);
-	return 0;
-}
-
 int btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info,
 				   u64 start, u64 end)
 {
-- 
2.21.0

[PATCH 06/15] btrfs: temporarily export btrfs_get_restripe_target

[Josef Bacik]

This gets used by a few different logical chunks of the block group
code, export it while we move things around.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.h | 2 ++
 fs/btrfs/extent-tree.c | 6 +++---
 2 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index 2745b99e5ae2..30bb722fce02 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -199,4 +199,6 @@ static inline int btrfs_block_group_cache_done(
 
 int __btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache,
 			       int force);
+u64 btrfs_get_restripe_target(struct btrfs_fs_info *fs_info, u64 flags);
+
 #endif /* BTRFS_BLOCK_GROUP_H */
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 37556ce36d68..412c93abe6a4 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -3069,7 +3069,7 @@ int btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr)
  *
  * should be called with balance_lock held
  */
-static u64 get_restripe_target(struct btrfs_fs_info *fs_info, u64 flags)
+u64 btrfs_get_restripe_target(struct btrfs_fs_info *fs_info, u64 flags)
 {
 	struct btrfs_balance_control *bctl = fs_info->balance_ctl;
 	u64 target = 0;
@@ -3110,7 +3110,7 @@ static u64 btrfs_reduce_alloc_profile(struct btrfs_fs_info *fs_info, u64 flags)
 	 * try to reduce to the target profile
 	 */
 	spin_lock(&fs_info->balance_lock);
-	target = get_restripe_target(fs_info, flags);
+	target = btrfs_get_restripe_target(fs_info, flags);
 	if (target) {
 		/* pick target profile only if it's already available */
 		if ((flags & target) & BTRFS_EXTENDED_PROFILE_MASK) {
@@ -6672,7 +6672,7 @@ static u64 update_block_group_flags(struct btrfs_fs_info *fs_info, u64 flags)
 	 * if restripe for this chunk_type is on pick target profile and
 	 * return, otherwise do the usual balance
 	 */
-	stripped = get_restripe_target(fs_info, flags);
+	stripped = btrfs_get_restripe_target(fs_info, flags);
 	if (stripped)
 		return extended_to_chunk(stripped);
 
-- 
2.21.0

[PATCH 07/15] btrfs: migrate inc/dec_block_group_ro code

[Josef Bacik]

This can easily be moved now.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.c | 213 +++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/block-group.h |   2 +
 fs/btrfs/ctree.h       |   2 -
 fs/btrfs/extent-tree.c | 213 -----------------------------------------
 4 files changed, 215 insertions(+), 215 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index c0ca7d60e5f4..3a0a1255aa1e 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -1802,3 +1802,216 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, u64 bytes_used,
 	set_avail_alloc_bits(fs_info, type);
 	return 0;
 }
+
+static u64 update_block_group_flags(struct btrfs_fs_info *fs_info, u64 flags)
+{
+	u64 num_devices;
+	u64 stripped;
+
+	/*
+	 * if restripe for this chunk_type is on pick target profile and
+	 * return, otherwise do the usual balance
+	 */
+	stripped = btrfs_get_restripe_target(fs_info, flags);
+	if (stripped)
+		return extended_to_chunk(stripped);
+
+	num_devices = fs_info->fs_devices->rw_devices;
+
+	stripped = BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID56_MASK |
+		BTRFS_BLOCK_GROUP_RAID1_MASK | BTRFS_BLOCK_GROUP_RAID10;
+
+	if (num_devices == 1) {
+		stripped |= BTRFS_BLOCK_GROUP_DUP;
+		stripped = flags & ~stripped;
+
+		/* turn raid0 into single device chunks */
+		if (flags & BTRFS_BLOCK_GROUP_RAID0)
+			return stripped;
+
+		/* turn mirroring into duplication */
+		if (flags & (BTRFS_BLOCK_GROUP_RAID1_MASK |
+			     BTRFS_BLOCK_GROUP_RAID10))
+			return stripped | BTRFS_BLOCK_GROUP_DUP;
+	} else {
+		/* they already had raid on here, just return */
+		if (flags & stripped)
+			return flags;
+
+		stripped |= BTRFS_BLOCK_GROUP_DUP;
+		stripped = flags & ~stripped;
+
+		/* switch duplicated blocks with raid1 */
+		if (flags & BTRFS_BLOCK_GROUP_DUP)
+			return stripped | BTRFS_BLOCK_GROUP_RAID1;
+
+		/* this is drive concat, leave it alone */
+	}
+
+	return flags;
+}
+
+/*
+ * Mark block group @cache read-only, so later write won't happen to block
+ * group @cache.
+ *
+ * If @force is not set, this function will only mark the block group readonly
+ * if we have enough free space (1M) in other metadata/system block groups.
+ * If @force is not set, this function will mark the block group readonly
+ * without checking free space.
+ *
+ * NOTE: This function doesn't care if other block groups can contain all the
+ * data in this block group. That check should be done by relocation routine,
+ * not this function.
+ */
+int __btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache,
+			       int force)
+{
+	struct btrfs_space_info *sinfo = cache->space_info;
+	u64 num_bytes;
+	u64 sinfo_used;
+	u64 min_allocable_bytes;
+	int ret = -ENOSPC;
+
+	/*
+	 * We need some metadata space and system metadata space for
+	 * allocating chunks in some corner cases until we force to set
+	 * it to be readonly.
+	 */
+	if ((sinfo->flags &
+	     (BTRFS_BLOCK_GROUP_SYSTEM | BTRFS_BLOCK_GROUP_METADATA)) &&
+	    !force)
+		min_allocable_bytes = SZ_1M;
+	else
+		min_allocable_bytes = 0;
+
+	spin_lock(&sinfo->lock);
+	spin_lock(&cache->lock);
+
+	if (cache->ro) {
+		cache->ro++;
+		ret = 0;
+		goto out;
+	}
+
+	num_bytes = cache->key.offset - cache->reserved - cache->pinned -
+		    cache->bytes_super - btrfs_block_group_used(&cache->item);
+	sinfo_used = btrfs_space_info_used(sinfo, true);
+
+	/*
+	 * sinfo_used + num_bytes should always <= sinfo->total_bytes.
+	 *
+	 * Here we make sure if we mark this bg RO, we still have enough
+	 * free space as buffer (if min_allocable_bytes is not 0).
+	 */
+	if (sinfo_used + num_bytes + min_allocable_bytes <=
+	    sinfo->total_bytes) {
+		sinfo->bytes_readonly += num_bytes;
+		cache->ro++;
+		list_add_tail(&cache->ro_list, &sinfo->ro_bgs);
+		ret = 0;
+	}
+out:
+	spin_unlock(&cache->lock);
+	spin_unlock(&sinfo->lock);
+	if (ret == -ENOSPC && btrfs_test_opt(cache->fs_info, ENOSPC_DEBUG)) {
+		btrfs_info(cache->fs_info,
+			"unable to make block group %llu ro",
+			cache->key.objectid);
+		btrfs_info(cache->fs_info,
+			"sinfo_used=%llu bg_num_bytes=%llu min_allocable=%llu",
+			sinfo_used, num_bytes, min_allocable_bytes);
+		btrfs_dump_space_info(cache->fs_info, cache->space_info, 0, 0);
+	}
+	return ret;
+}
+
+int btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache)
+
+{
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	struct btrfs_trans_handle *trans;
+	u64 alloc_flags;
+	int ret;
+
+again:
+	trans = btrfs_join_transaction(fs_info->extent_root);
+	if (IS_ERR(trans))
+		return PTR_ERR(trans);
+
+	/*
+	 * we're not allowed to set block groups readonly after the dirty
+	 * block groups cache has started writing.  If it already started,
+	 * back off and let this transaction commit
+	 */
+	mutex_lock(&fs_info->ro_block_group_mutex);
+	if (test_bit(BTRFS_TRANS_DIRTY_BG_RUN, &trans->transaction->flags)) {
+		u64 transid = trans->transid;
+
+		mutex_unlock(&fs_info->ro_block_group_mutex);
+		btrfs_end_transaction(trans);
+
+		ret = btrfs_wait_for_commit(fs_info, transid);
+		if (ret)
+			return ret;
+		goto again;
+	}
+
+	/*
+	 * if we are changing raid levels, try to allocate a corresponding
+	 * block group with the new raid level.
+	 */
+	alloc_flags = update_block_group_flags(fs_info, cache->flags);
+	if (alloc_flags != cache->flags) {
+		ret = btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
+		/*
+		 * ENOSPC is allowed here, we may have enough space
+		 * already allocated at the new raid level to
+		 * carry on
+		 */
+		if (ret == -ENOSPC)
+			ret = 0;
+		if (ret < 0)
+			goto out;
+	}
+
+	ret = __btrfs_inc_block_group_ro(cache, 0);
+	if (!ret)
+		goto out;
+	alloc_flags = btrfs_get_alloc_profile(fs_info, cache->space_info->flags);
+	ret = btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
+	if (ret < 0)
+		goto out;
+	ret = __btrfs_inc_block_group_ro(cache, 0);
+out:
+	if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM) {
+		alloc_flags = update_block_group_flags(fs_info, cache->flags);
+		mutex_lock(&fs_info->chunk_mutex);
+		check_system_chunk(trans, alloc_flags);
+		mutex_unlock(&fs_info->chunk_mutex);
+	}
+	mutex_unlock(&fs_info->ro_block_group_mutex);
+
+	btrfs_end_transaction(trans);
+	return ret;
+}
+
+void btrfs_dec_block_group_ro(struct btrfs_block_group_cache *cache)
+{
+	struct btrfs_space_info *sinfo = cache->space_info;
+	u64 num_bytes;
+
+	BUG_ON(!cache->ro);
+
+	spin_lock(&sinfo->lock);
+	spin_lock(&cache->lock);
+	if (!--cache->ro) {
+		num_bytes = cache->key.offset - cache->reserved -
+			    cache->pinned - cache->bytes_super -
+			    btrfs_block_group_used(&cache->item);
+		sinfo->bytes_readonly -= num_bytes;
+		list_del_init(&cache->ro_list);
+	}
+	spin_unlock(&cache->lock);
+	spin_unlock(&sinfo->lock);
+}
diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index 30bb722fce02..f8b9bcfac9e9 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -188,6 +188,8 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 			   u64 bytes_used, u64 type, u64 chunk_offset,
 			   u64 size);
 void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans);
+int btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache);
+void btrfs_dec_block_group_ro(struct btrfs_block_group_cache *cache);
 
 static inline int btrfs_block_group_cache_done(
 		struct btrfs_block_group_cache *cache)
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 1cfe7adddc97..65a9ee97bce5 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2590,8 +2590,6 @@ void btrfs_delalloc_release_extents(struct btrfs_inode *inode, u64 num_bytes,
 				    bool qgroup_free);
 
 int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes);
-int btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache);
-void btrfs_dec_block_group_ro(struct btrfs_block_group_cache *cache);
 void btrfs_put_block_group_cache(struct btrfs_fs_info *info);
 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
 int btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info,
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 412c93abe6a4..6d96b378a189 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -6663,199 +6663,6 @@ int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
-static u64 update_block_group_flags(struct btrfs_fs_info *fs_info, u64 flags)
-{
-	u64 num_devices;
-	u64 stripped;
-
-	/*
-	 * if restripe for this chunk_type is on pick target profile and
-	 * return, otherwise do the usual balance
-	 */
-	stripped = btrfs_get_restripe_target(fs_info, flags);
-	if (stripped)
-		return extended_to_chunk(stripped);
-
-	num_devices = fs_info->fs_devices->rw_devices;
-
-	stripped = BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID56_MASK |
-		BTRFS_BLOCK_GROUP_RAID1_MASK | BTRFS_BLOCK_GROUP_RAID10;
-
-	if (num_devices == 1) {
-		stripped |= BTRFS_BLOCK_GROUP_DUP;
-		stripped = flags & ~stripped;
-
-		/* turn raid0 into single device chunks */
-		if (flags & BTRFS_BLOCK_GROUP_RAID0)
-			return stripped;
-
-		/* turn mirroring into duplication */
-		if (flags & (BTRFS_BLOCK_GROUP_RAID1_MASK |
-			     BTRFS_BLOCK_GROUP_RAID10))
-			return stripped | BTRFS_BLOCK_GROUP_DUP;
-	} else {
-		/* they already had raid on here, just return */
-		if (flags & stripped)
-			return flags;
-
-		stripped |= BTRFS_BLOCK_GROUP_DUP;
-		stripped = flags & ~stripped;
-
-		/* switch duplicated blocks with raid1 */
-		if (flags & BTRFS_BLOCK_GROUP_DUP)
-			return stripped | BTRFS_BLOCK_GROUP_RAID1;
-
-		/* this is drive concat, leave it alone */
-	}
-
-	return flags;
-}
-
-/*
- * Mark block group @cache read-only, so later write won't happen to block
- * group @cache.
- *
- * If @force is not set, this function will only mark the block group readonly
- * if we have enough free space (1M) in other metadata/system block groups.
- * If @force is not set, this function will mark the block group readonly
- * without checking free space.
- *
- * NOTE: This function doesn't care if other block groups can contain all the
- * data in this block group. That check should be done by relocation routine,
- * not this function.
- */
-int __btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache,
-			       int force)
-{
-	struct btrfs_space_info *sinfo = cache->space_info;
-	u64 num_bytes;
-	u64 sinfo_used;
-	u64 min_allocable_bytes;
-	int ret = -ENOSPC;
-
-	/*
-	 * We need some metadata space and system metadata space for
-	 * allocating chunks in some corner cases until we force to set
-	 * it to be readonly.
-	 */
-	if ((sinfo->flags &
-	     (BTRFS_BLOCK_GROUP_SYSTEM | BTRFS_BLOCK_GROUP_METADATA)) &&
-	    !force)
-		min_allocable_bytes = SZ_1M;
-	else
-		min_allocable_bytes = 0;
-
-	spin_lock(&sinfo->lock);
-	spin_lock(&cache->lock);
-
-	if (cache->ro) {
-		cache->ro++;
-		ret = 0;
-		goto out;
-	}
-
-	num_bytes = cache->key.offset - cache->reserved - cache->pinned -
-		    cache->bytes_super - btrfs_block_group_used(&cache->item);
-	sinfo_used = btrfs_space_info_used(sinfo, true);
-
-	/*
-	 * sinfo_used + num_bytes should always <= sinfo->total_bytes.
-	 *
-	 * Here we make sure if we mark this bg RO, we still have enough
-	 * free space as buffer (if min_allocable_bytes is not 0).
-	 */
-	if (sinfo_used + num_bytes + min_allocable_bytes <=
-	    sinfo->total_bytes) {
-		sinfo->bytes_readonly += num_bytes;
-		cache->ro++;
-		list_add_tail(&cache->ro_list, &sinfo->ro_bgs);
-		ret = 0;
-	}
-out:
-	spin_unlock(&cache->lock);
-	spin_unlock(&sinfo->lock);
-	if (ret == -ENOSPC && btrfs_test_opt(cache->fs_info, ENOSPC_DEBUG)) {
-		btrfs_info(cache->fs_info,
-			"unable to make block group %llu ro",
-			cache->key.objectid);
-		btrfs_info(cache->fs_info,
-			"sinfo_used=%llu bg_num_bytes=%llu min_allocable=%llu",
-			sinfo_used, num_bytes, min_allocable_bytes);
-		btrfs_dump_space_info(cache->fs_info, cache->space_info, 0, 0);
-	}
-	return ret;
-}
-
-int btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache)
-
-{
-	struct btrfs_fs_info *fs_info = cache->fs_info;
-	struct btrfs_trans_handle *trans;
-	u64 alloc_flags;
-	int ret;
-
-again:
-	trans = btrfs_join_transaction(fs_info->extent_root);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
-
-	/*
-	 * we're not allowed to set block groups readonly after the dirty
-	 * block groups cache has started writing.  If it already started,
-	 * back off and let this transaction commit
-	 */
-	mutex_lock(&fs_info->ro_block_group_mutex);
-	if (test_bit(BTRFS_TRANS_DIRTY_BG_RUN, &trans->transaction->flags)) {
-		u64 transid = trans->transid;
-
-		mutex_unlock(&fs_info->ro_block_group_mutex);
-		btrfs_end_transaction(trans);
-
-		ret = btrfs_wait_for_commit(fs_info, transid);
-		if (ret)
-			return ret;
-		goto again;
-	}
-
-	/*
-	 * if we are changing raid levels, try to allocate a corresponding
-	 * block group with the new raid level.
-	 */
-	alloc_flags = update_block_group_flags(fs_info, cache->flags);
-	if (alloc_flags != cache->flags) {
-		ret = btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
-		/*
-		 * ENOSPC is allowed here, we may have enough space
-		 * already allocated at the new raid level to
-		 * carry on
-		 */
-		if (ret == -ENOSPC)
-			ret = 0;
-		if (ret < 0)
-			goto out;
-	}
-
-	ret = __btrfs_inc_block_group_ro(cache, 0);
-	if (!ret)
-		goto out;
-	alloc_flags = btrfs_get_alloc_profile(fs_info, cache->space_info->flags);
-	ret = btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
-	if (ret < 0)
-		goto out;
-	ret = __btrfs_inc_block_group_ro(cache, 0);
-out:
-	if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM) {
-		alloc_flags = update_block_group_flags(fs_info, cache->flags);
-		mutex_lock(&fs_info->chunk_mutex);
-		check_system_chunk(trans, alloc_flags);
-		mutex_unlock(&fs_info->chunk_mutex);
-	}
-	mutex_unlock(&fs_info->ro_block_group_mutex);
-
-	btrfs_end_transaction(trans);
-	return ret;
-}
-
 int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, u64 type)
 {
 	u64 alloc_flags = btrfs_get_alloc_profile(trans->fs_info, type);
@@ -6898,26 +6705,6 @@ u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo)
 	return free_bytes;
 }
 
-void btrfs_dec_block_group_ro(struct btrfs_block_group_cache *cache)
-{
-	struct btrfs_space_info *sinfo = cache->space_info;
-	u64 num_bytes;
-
-	BUG_ON(!cache->ro);
-
-	spin_lock(&sinfo->lock);
-	spin_lock(&cache->lock);
-	if (!--cache->ro) {
-		num_bytes = cache->key.offset - cache->reserved -
-			    cache->pinned - cache->bytes_super -
-			    btrfs_block_group_used(&cache->item);
-		sinfo->bytes_readonly -= num_bytes;
-		list_del_init(&cache->ro_list);
-	}
-	spin_unlock(&cache->lock);
-	spin_unlock(&sinfo->lock);
-}
-
 void btrfs_put_block_group_cache(struct btrfs_fs_info *info)
 {
 	struct btrfs_block_group_cache *block_group;
-- 
2.21.0

[PATCH 08/15] btrfs: migrate the dirty bg writeout code

[Josef Bacik]

This can be easily migrated over now.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.c | 520 +++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/block-group.h |   3 +
 fs/btrfs/ctree.h       |   3 -
 fs/btrfs/extent-tree.c | 518 ----------------------------------------
 4 files changed, 523 insertions(+), 521 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 3a0a1255aa1e..de9d5b3cdac6 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -12,6 +12,7 @@
 #include "ref-verify.h"
 #include "sysfs.h"
 #include "tree-log.h"
+#include "delalloc-space.h"
 
 void btrfs_get_block_group(struct btrfs_block_group_cache *cache)
 {
@@ -2015,3 +2016,522 @@ void btrfs_dec_block_group_ro(struct btrfs_block_group_cache *cache)
 	spin_unlock(&cache->lock);
 	spin_unlock(&sinfo->lock);
 }
+
+static int write_one_cache_group(struct btrfs_trans_handle *trans,
+				 struct btrfs_path *path,
+				 struct btrfs_block_group_cache *cache)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	int ret;
+	struct btrfs_root *extent_root = fs_info->extent_root;
+	unsigned long bi;
+	struct extent_buffer *leaf;
+
+	ret = btrfs_search_slot(trans, extent_root, &cache->key, path, 0, 1);
+	if (ret) {
+		if (ret > 0)
+			ret = -ENOENT;
+		goto fail;
+	}
+
+	leaf = path->nodes[0];
+	bi = btrfs_item_ptr_offset(leaf, path->slots[0]);
+	write_extent_buffer(leaf, &cache->item, bi, sizeof(cache->item));
+	btrfs_mark_buffer_dirty(leaf);
+fail:
+	btrfs_release_path(path);
+	return ret;
+
+}
+
+static int cache_save_setup(struct btrfs_block_group_cache *block_group,
+			    struct btrfs_trans_handle *trans,
+			    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;
+	int dcs = BTRFS_DC_ERROR;
+	u64 num_pages = 0;
+	int retries = 0;
+	int ret = 0;
+
+	/*
+	 * If this block group is smaller than 100 megs don't bother caching the
+	 * block group.
+	 */
+	if (block_group->key.offset < (100 * SZ_1M)) {
+		spin_lock(&block_group->lock);
+		block_group->disk_cache_state = BTRFS_DC_WRITTEN;
+		spin_unlock(&block_group->lock);
+		return 0;
+	}
+
+	if (trans->aborted)
+		return 0;
+again:
+	inode = lookup_free_space_inode(block_group, path);
+	if (IS_ERR(inode) && PTR_ERR(inode) != -ENOENT) {
+		ret = PTR_ERR(inode);
+		btrfs_release_path(path);
+		goto out;
+	}
+
+	if (IS_ERR(inode)) {
+		BUG_ON(retries);
+		retries++;
+
+		if (block_group->ro)
+			goto out_free;
+
+		ret = create_free_space_inode(trans, block_group, path);
+		if (ret)
+			goto out_free;
+		goto again;
+	}
+
+	/*
+	 * We want to set the generation to 0, that way if anything goes wrong
+	 * from here on out we know not to trust this cache when we load up next
+	 * time.
+	 */
+	BTRFS_I(inode)->generation = 0;
+	ret = btrfs_update_inode(trans, root, inode);
+	if (ret) {
+		/*
+		 * So theoretically we could recover from this, simply set the
+		 * super cache generation to 0 so we know to invalidate the
+		 * cache, but then we'd have to keep track of the block groups
+		 * that fail this way so we know we _have_ to reset this cache
+		 * before the next commit or risk reading stale cache.  So to
+		 * limit our exposure to horrible edge cases lets just abort the
+		 * transaction, this only happens in really bad situations
+		 * anyway.
+		 */
+		btrfs_abort_transaction(trans, ret);
+		goto out_put;
+	}
+	WARN_ON(ret);
+
+	/* We've already setup this transaction, go ahead and exit */
+	if (block_group->cache_generation == trans->transid &&
+	    i_size_read(inode)) {
+		dcs = BTRFS_DC_SETUP;
+		goto out_put;
+	}
+
+	if (i_size_read(inode) > 0) {
+		ret = btrfs_check_trunc_cache_free_space(fs_info,
+					&fs_info->global_block_rsv);
+		if (ret)
+			goto out_put;
+
+		ret = btrfs_truncate_free_space_cache(trans, NULL, inode);
+		if (ret)
+			goto out_put;
+	}
+
+	spin_lock(&block_group->lock);
+	if (block_group->cached != BTRFS_CACHE_FINISHED ||
+	    !btrfs_test_opt(fs_info, SPACE_CACHE)) {
+		/*
+		 * don't bother trying to write stuff out _if_
+		 * a) we're not cached,
+		 * b) we're with nospace_cache mount option,
+		 * c) we're with v2 space_cache (FREE_SPACE_TREE).
+		 */
+		dcs = BTRFS_DC_WRITTEN;
+		spin_unlock(&block_group->lock);
+		goto out_put;
+	}
+	spin_unlock(&block_group->lock);
+
+	/*
+	 * We hit an ENOSPC when setting up the cache in this transaction, just
+	 * skip doing the setup, we've already cleared the cache so we're safe.
+	 */
+	if (test_bit(BTRFS_TRANS_CACHE_ENOSPC, &trans->transaction->flags)) {
+		ret = -ENOSPC;
+		goto out_put;
+	}
+
+	/*
+	 * Try to preallocate enough space based on how big the block group is.
+	 * Keep in mind this has to include any pinned space which could end up
+	 * taking up quite a bit since it's not folded into the other space
+	 * cache.
+	 */
+	num_pages = div_u64(block_group->key.offset, SZ_256M);
+	if (!num_pages)
+		num_pages = 1;
+
+	num_pages *= 16;
+	num_pages *= PAGE_SIZE;
+
+	ret = btrfs_check_data_free_space(inode, &data_reserved, 0, num_pages);
+	if (ret)
+		goto out_put;
+
+	ret = btrfs_prealloc_file_range_trans(inode, trans, 0, 0, num_pages,
+					      num_pages, num_pages,
+					      &alloc_hint);
+	/*
+	 * Our cache requires contiguous chunks so that we don't modify a bunch
+	 * of metadata or split extents when writing the cache out, which means
+	 * we can enospc if we are heavily fragmented in addition to just normal
+	 * out of space conditions.  So if we hit this just skip setting up any
+	 * other block groups for this transaction, maybe we'll unpin enough
+	 * space the next time around.
+	 */
+	if (!ret)
+		dcs = BTRFS_DC_SETUP;
+	else if (ret == -ENOSPC)
+		set_bit(BTRFS_TRANS_CACHE_ENOSPC, &trans->transaction->flags);
+
+out_put:
+	iput(inode);
+out_free:
+	btrfs_release_path(path);
+out:
+	spin_lock(&block_group->lock);
+	if (!ret && dcs == BTRFS_DC_SETUP)
+		block_group->cache_generation = trans->transid;
+	block_group->disk_cache_state = dcs;
+	spin_unlock(&block_group->lock);
+
+	extent_changeset_free(data_reserved);
+	return ret;
+}
+
+int btrfs_setup_space_cache(struct btrfs_trans_handle *trans)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_block_group_cache *cache, *tmp;
+	struct btrfs_transaction *cur_trans = trans->transaction;
+	struct btrfs_path *path;
+
+	if (list_empty(&cur_trans->dirty_bgs) ||
+	    !btrfs_test_opt(fs_info, SPACE_CACHE))
+		return 0;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	/* Could add new block groups, use _safe just in case */
+	list_for_each_entry_safe(cache, tmp, &cur_trans->dirty_bgs,
+				 dirty_list) {
+		if (cache->disk_cache_state == BTRFS_DC_CLEAR)
+			cache_save_setup(cache, trans, path);
+	}
+
+	btrfs_free_path(path);
+	return 0;
+}
+
+/*
+ * transaction commit does final block group cache writeback during a
+ * critical section where nothing is allowed to change the FS.  This is
+ * required in order for the cache to actually match the block group,
+ * but can introduce a lot of latency into the commit.
+ *
+ * So, btrfs_start_dirty_block_groups is here to kick off block group
+ * cache IO.  There's a chance we'll have to redo some of it if the
+ * block group changes again during the commit, but it greatly reduces
+ * the commit latency by getting rid of the easy block groups while
+ * we're still allowing others to join the commit.
+ */
+int btrfs_start_dirty_block_groups(struct btrfs_trans_handle *trans)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_block_group_cache *cache;
+	struct btrfs_transaction *cur_trans = trans->transaction;
+	int ret = 0;
+	int should_put;
+	struct btrfs_path *path = NULL;
+	LIST_HEAD(dirty);
+	struct list_head *io = &cur_trans->io_bgs;
+	int num_started = 0;
+	int loops = 0;
+
+	spin_lock(&cur_trans->dirty_bgs_lock);
+	if (list_empty(&cur_trans->dirty_bgs)) {
+		spin_unlock(&cur_trans->dirty_bgs_lock);
+		return 0;
+	}
+	list_splice_init(&cur_trans->dirty_bgs, &dirty);
+	spin_unlock(&cur_trans->dirty_bgs_lock);
+
+again:
+	/*
+	 * make sure all the block groups on our dirty list actually
+	 * exist
+	 */
+	btrfs_create_pending_block_groups(trans);
+
+	if (!path) {
+		path = btrfs_alloc_path();
+		if (!path)
+			return -ENOMEM;
+	}
+
+	/*
+	 * cache_write_mutex is here only to save us from balance or automatic
+	 * removal of empty block groups deleting this block group while we are
+	 * writing out the cache
+	 */
+	mutex_lock(&trans->transaction->cache_write_mutex);
+	while (!list_empty(&dirty)) {
+		bool drop_reserve = true;
+
+		cache = list_first_entry(&dirty,
+					 struct btrfs_block_group_cache,
+					 dirty_list);
+		/*
+		 * this can happen if something re-dirties a block
+		 * group that is already under IO.  Just wait for it to
+		 * finish and then do it all again
+		 */
+		if (!list_empty(&cache->io_list)) {
+			list_del_init(&cache->io_list);
+			btrfs_wait_cache_io(trans, cache, path);
+			btrfs_put_block_group(cache);
+		}
+
+
+		/*
+		 * btrfs_wait_cache_io uses the cache->dirty_list to decide
+		 * if it should update the cache_state.  Don't delete
+		 * until after we wait.
+		 *
+		 * Since we're not running in the commit critical section
+		 * we need the dirty_bgs_lock to protect from update_block_group
+		 */
+		spin_lock(&cur_trans->dirty_bgs_lock);
+		list_del_init(&cache->dirty_list);
+		spin_unlock(&cur_trans->dirty_bgs_lock);
+
+		should_put = 1;
+
+		cache_save_setup(cache, trans, path);
+
+		if (cache->disk_cache_state == BTRFS_DC_SETUP) {
+			cache->io_ctl.inode = NULL;
+			ret = btrfs_write_out_cache(trans, cache, path);
+			if (ret == 0 && cache->io_ctl.inode) {
+				num_started++;
+				should_put = 0;
+
+				/*
+				 * The cache_write_mutex is protecting the
+				 * io_list, also refer to the definition of
+				 * btrfs_transaction::io_bgs for more details
+				 */
+				list_add_tail(&cache->io_list, io);
+			} else {
+				/*
+				 * if we failed to write the cache, the
+				 * generation will be bad and life goes on
+				 */
+				ret = 0;
+			}
+		}
+		if (!ret) {
+			ret = write_one_cache_group(trans, path, cache);
+			/*
+			 * Our block group might still be attached to the list
+			 * of new block groups in the transaction handle of some
+			 * other task (struct btrfs_trans_handle->new_bgs). This
+			 * means its block group item isn't yet in the extent
+			 * tree. If this happens ignore the error, as we will
+			 * try again later in the critical section of the
+			 * transaction commit.
+			 */
+			if (ret == -ENOENT) {
+				ret = 0;
+				spin_lock(&cur_trans->dirty_bgs_lock);
+				if (list_empty(&cache->dirty_list)) {
+					list_add_tail(&cache->dirty_list,
+						      &cur_trans->dirty_bgs);
+					btrfs_get_block_group(cache);
+					drop_reserve = false;
+				}
+				spin_unlock(&cur_trans->dirty_bgs_lock);
+			} else if (ret) {
+				btrfs_abort_transaction(trans, ret);
+			}
+		}
+
+		/* if it's not on the io list, we need to put the block group */
+		if (should_put)
+			btrfs_put_block_group(cache);
+		if (drop_reserve)
+			btrfs_delayed_refs_rsv_release(fs_info, 1);
+
+		if (ret)
+			break;
+
+		/*
+		 * Avoid blocking other tasks for too long. It might even save
+		 * us from writing caches for block groups that are going to be
+		 * removed.
+		 */
+		mutex_unlock(&trans->transaction->cache_write_mutex);
+		mutex_lock(&trans->transaction->cache_write_mutex);
+	}
+	mutex_unlock(&trans->transaction->cache_write_mutex);
+
+	/*
+	 * go through delayed refs for all the stuff we've just kicked off
+	 * and then loop back (just once)
+	 */
+	ret = btrfs_run_delayed_refs(trans, 0);
+	if (!ret && loops == 0) {
+		loops++;
+		spin_lock(&cur_trans->dirty_bgs_lock);
+		list_splice_init(&cur_trans->dirty_bgs, &dirty);
+		/*
+		 * dirty_bgs_lock protects us from concurrent block group
+		 * deletes too (not just cache_write_mutex).
+		 */
+		if (!list_empty(&dirty)) {
+			spin_unlock(&cur_trans->dirty_bgs_lock);
+			goto again;
+		}
+		spin_unlock(&cur_trans->dirty_bgs_lock);
+	} else if (ret < 0) {
+		btrfs_cleanup_dirty_bgs(cur_trans, fs_info);
+	}
+
+	btrfs_free_path(path);
+	return ret;
+}
+
+int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_block_group_cache *cache;
+	struct btrfs_transaction *cur_trans = trans->transaction;
+	int ret = 0;
+	int should_put;
+	struct btrfs_path *path;
+	struct list_head *io = &cur_trans->io_bgs;
+	int num_started = 0;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	/*
+	 * Even though we are in the critical section of the transaction commit,
+	 * we can still have concurrent tasks adding elements to this
+	 * transaction's list of dirty block groups. These tasks correspond to
+	 * endio free space workers started when writeback finishes for a
+	 * space cache, which run inode.c:btrfs_finish_ordered_io(), and can
+	 * allocate new block groups as a result of COWing nodes of the root
+	 * tree when updating the free space inode. The writeback for the space
+	 * caches is triggered by an earlier call to
+	 * btrfs_start_dirty_block_groups() and iterations of the following
+	 * loop.
+	 * Also 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.
+	 */
+	spin_lock(&cur_trans->dirty_bgs_lock);
+	while (!list_empty(&cur_trans->dirty_bgs)) {
+		cache = list_first_entry(&cur_trans->dirty_bgs,
+					 struct btrfs_block_group_cache,
+					 dirty_list);
+
+		/*
+		 * this can happen if cache_save_setup re-dirties a block
+		 * group that is already under IO.  Just wait for it to
+		 * finish and then do it all again
+		 */
+		if (!list_empty(&cache->io_list)) {
+			spin_unlock(&cur_trans->dirty_bgs_lock);
+			list_del_init(&cache->io_list);
+			btrfs_wait_cache_io(trans, cache, path);
+			btrfs_put_block_group(cache);
+			spin_lock(&cur_trans->dirty_bgs_lock);
+		}
+
+		/*
+		 * don't remove from the dirty list until after we've waited
+		 * on any pending IO
+		 */
+		list_del_init(&cache->dirty_list);
+		spin_unlock(&cur_trans->dirty_bgs_lock);
+		should_put = 1;
+
+		cache_save_setup(cache, trans, path);
+
+		if (!ret)
+			ret = btrfs_run_delayed_refs(trans,
+						     (unsigned long) -1);
+
+		if (!ret && cache->disk_cache_state == BTRFS_DC_SETUP) {
+			cache->io_ctl.inode = NULL;
+			ret = btrfs_write_out_cache(trans, cache, path);
+			if (ret == 0 && cache->io_ctl.inode) {
+				num_started++;
+				should_put = 0;
+				list_add_tail(&cache->io_list, io);
+			} else {
+				/*
+				 * if we failed to write the cache, the
+				 * generation will be bad and life goes on
+				 */
+				ret = 0;
+			}
+		}
+		if (!ret) {
+			ret = write_one_cache_group(trans, path, cache);
+			/*
+			 * One of the free space endio workers might have
+			 * created a new block group while updating a free space
+			 * cache's inode (at inode.c:btrfs_finish_ordered_io())
+			 * and hasn't released its transaction handle yet, in
+			 * which case the new block group is still attached to
+			 * its transaction handle and its creation has not
+			 * finished yet (no block group item in the extent tree
+			 * yet, etc). If this is the case, wait for all free
+			 * space endio workers to finish and retry. This is a
+			 * a very rare case so no need for a more efficient and
+			 * complex approach.
+			 */
+			if (ret == -ENOENT) {
+				wait_event(cur_trans->writer_wait,
+				   atomic_read(&cur_trans->num_writers) == 1);
+				ret = write_one_cache_group(trans, path, cache);
+			}
+			if (ret)
+				btrfs_abort_transaction(trans, ret);
+		}
+
+		/* 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);
+		spin_lock(&cur_trans->dirty_bgs_lock);
+	}
+	spin_unlock(&cur_trans->dirty_bgs_lock);
+
+	/*
+	 * Refer to the definition of io_bgs member for details why it's safe
+	 * to use it without any locking
+	 */
+	while (!list_empty(io)) {
+		cache = list_first_entry(io, struct btrfs_block_group_cache,
+					 io_list);
+		list_del_init(&cache->io_list);
+		btrfs_wait_cache_io(trans, cache, path);
+		btrfs_put_block_group(cache);
+	}
+
+	btrfs_free_path(path);
+	return ret;
+}
+
diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index f8b9bcfac9e9..f23da9d82525 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -190,6 +190,9 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans);
 int btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache);
 void btrfs_dec_block_group_ro(struct btrfs_block_group_cache *cache);
+int btrfs_start_dirty_block_groups(struct btrfs_trans_handle *trans);
+int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans);
+int btrfs_setup_space_cache(struct btrfs_trans_handle *trans);
 
 static inline int btrfs_block_group_cache_done(
 		struct btrfs_block_group_cache *cache)
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 65a9ee97bce5..2a41e09727c3 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2522,9 +2522,6 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans);
 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 			 struct btrfs_ref *generic_ref);
 
-int btrfs_start_dirty_block_groups(struct btrfs_trans_handle *trans);
-int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans);
-int btrfs_setup_space_cache(struct btrfs_trans_handle *trans);
 int btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr);
 int btrfs_free_block_groups(struct btrfs_fs_info *info);
 void btrfs_get_block_group_trimming(struct btrfs_block_group_cache *cache);
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 6d96b378a189..7f65958efc40 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2532,524 +2532,6 @@ int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 	return __btrfs_mod_ref(trans, root, buf, full_backref, 0);
 }
 
-static int write_one_cache_group(struct btrfs_trans_handle *trans,
-				 struct btrfs_path *path,
-				 struct btrfs_block_group_cache *cache)
-{
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-	int ret;
-	struct btrfs_root *extent_root = fs_info->extent_root;
-	unsigned long bi;
-	struct extent_buffer *leaf;
-
-	ret = btrfs_search_slot(trans, extent_root, &cache->key, path, 0, 1);
-	if (ret) {
-		if (ret > 0)
-			ret = -ENOENT;
-		goto fail;
-	}
-
-	leaf = path->nodes[0];
-	bi = btrfs_item_ptr_offset(leaf, path->slots[0]);
-	write_extent_buffer(leaf, &cache->item, bi, sizeof(cache->item));
-	btrfs_mark_buffer_dirty(leaf);
-fail:
-	btrfs_release_path(path);
-	return ret;
-
-}
-
-static int cache_save_setup(struct btrfs_block_group_cache *block_group,
-			    struct btrfs_trans_handle *trans,
-			    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;
-	int dcs = BTRFS_DC_ERROR;
-	u64 num_pages = 0;
-	int retries = 0;
-	int ret = 0;
-
-	/*
-	 * If this block group is smaller than 100 megs don't bother caching the
-	 * block group.
-	 */
-	if (block_group->key.offset < (100 * SZ_1M)) {
-		spin_lock(&block_group->lock);
-		block_group->disk_cache_state = BTRFS_DC_WRITTEN;
-		spin_unlock(&block_group->lock);
-		return 0;
-	}
-
-	if (trans->aborted)
-		return 0;
-again:
-	inode = lookup_free_space_inode(block_group, path);
-	if (IS_ERR(inode) && PTR_ERR(inode) != -ENOENT) {
-		ret = PTR_ERR(inode);
-		btrfs_release_path(path);
-		goto out;
-	}
-
-	if (IS_ERR(inode)) {
-		BUG_ON(retries);
-		retries++;
-
-		if (block_group->ro)
-			goto out_free;
-
-		ret = create_free_space_inode(trans, block_group, path);
-		if (ret)
-			goto out_free;
-		goto again;
-	}
-
-	/*
-	 * We want to set the generation to 0, that way if anything goes wrong
-	 * from here on out we know not to trust this cache when we load up next
-	 * time.
-	 */
-	BTRFS_I(inode)->generation = 0;
-	ret = btrfs_update_inode(trans, root, inode);
-	if (ret) {
-		/*
-		 * So theoretically we could recover from this, simply set the
-		 * super cache generation to 0 so we know to invalidate the
-		 * cache, but then we'd have to keep track of the block groups
-		 * that fail this way so we know we _have_ to reset this cache
-		 * before the next commit or risk reading stale cache.  So to
-		 * limit our exposure to horrible edge cases lets just abort the
-		 * transaction, this only happens in really bad situations
-		 * anyway.
-		 */
-		btrfs_abort_transaction(trans, ret);
-		goto out_put;
-	}
-	WARN_ON(ret);
-
-	/* We've already setup this transaction, go ahead and exit */
-	if (block_group->cache_generation == trans->transid &&
-	    i_size_read(inode)) {
-		dcs = BTRFS_DC_SETUP;
-		goto out_put;
-	}
-
-	if (i_size_read(inode) > 0) {
-		ret = btrfs_check_trunc_cache_free_space(fs_info,
-					&fs_info->global_block_rsv);
-		if (ret)
-			goto out_put;
-
-		ret = btrfs_truncate_free_space_cache(trans, NULL, inode);
-		if (ret)
-			goto out_put;
-	}
-
-	spin_lock(&block_group->lock);
-	if (block_group->cached != BTRFS_CACHE_FINISHED ||
-	    !btrfs_test_opt(fs_info, SPACE_CACHE)) {
-		/*
-		 * don't bother trying to write stuff out _if_
-		 * a) we're not cached,
-		 * b) we're with nospace_cache mount option,
-		 * c) we're with v2 space_cache (FREE_SPACE_TREE).
-		 */
-		dcs = BTRFS_DC_WRITTEN;
-		spin_unlock(&block_group->lock);
-		goto out_put;
-	}
-	spin_unlock(&block_group->lock);
-
-	/*
-	 * We hit an ENOSPC when setting up the cache in this transaction, just
-	 * skip doing the setup, we've already cleared the cache so we're safe.
-	 */
-	if (test_bit(BTRFS_TRANS_CACHE_ENOSPC, &trans->transaction->flags)) {
-		ret = -ENOSPC;
-		goto out_put;
-	}
-
-	/*
-	 * Try to preallocate enough space based on how big the block group is.
-	 * Keep in mind this has to include any pinned space which could end up
-	 * taking up quite a bit since it's not folded into the other space
-	 * cache.
-	 */
-	num_pages = div_u64(block_group->key.offset, SZ_256M);
-	if (!num_pages)
-		num_pages = 1;
-
-	num_pages *= 16;
-	num_pages *= PAGE_SIZE;
-
-	ret = btrfs_check_data_free_space(inode, &data_reserved, 0, num_pages);
-	if (ret)
-		goto out_put;
-
-	ret = btrfs_prealloc_file_range_trans(inode, trans, 0, 0, num_pages,
-					      num_pages, num_pages,
-					      &alloc_hint);
-	/*
-	 * Our cache requires contiguous chunks so that we don't modify a bunch
-	 * of metadata or split extents when writing the cache out, which means
-	 * we can enospc if we are heavily fragmented in addition to just normal
-	 * out of space conditions.  So if we hit this just skip setting up any
-	 * other block groups for this transaction, maybe we'll unpin enough
-	 * space the next time around.
-	 */
-	if (!ret)
-		dcs = BTRFS_DC_SETUP;
-	else if (ret == -ENOSPC)
-		set_bit(BTRFS_TRANS_CACHE_ENOSPC, &trans->transaction->flags);
-
-out_put:
-	iput(inode);
-out_free:
-	btrfs_release_path(path);
-out:
-	spin_lock(&block_group->lock);
-	if (!ret && dcs == BTRFS_DC_SETUP)
-		block_group->cache_generation = trans->transid;
-	block_group->disk_cache_state = dcs;
-	spin_unlock(&block_group->lock);
-
-	extent_changeset_free(data_reserved);
-	return ret;
-}
-
-int btrfs_setup_space_cache(struct btrfs_trans_handle *trans)
-{
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_block_group_cache *cache, *tmp;
-	struct btrfs_transaction *cur_trans = trans->transaction;
-	struct btrfs_path *path;
-
-	if (list_empty(&cur_trans->dirty_bgs) ||
-	    !btrfs_test_opt(fs_info, SPACE_CACHE))
-		return 0;
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
-	/* Could add new block groups, use _safe just in case */
-	list_for_each_entry_safe(cache, tmp, &cur_trans->dirty_bgs,
-				 dirty_list) {
-		if (cache->disk_cache_state == BTRFS_DC_CLEAR)
-			cache_save_setup(cache, trans, path);
-	}
-
-	btrfs_free_path(path);
-	return 0;
-}
-
-/*
- * transaction commit does final block group cache writeback during a
- * critical section where nothing is allowed to change the FS.  This is
- * required in order for the cache to actually match the block group,
- * but can introduce a lot of latency into the commit.
- *
- * So, btrfs_start_dirty_block_groups is here to kick off block group
- * cache IO.  There's a chance we'll have to redo some of it if the
- * block group changes again during the commit, but it greatly reduces
- * the commit latency by getting rid of the easy block groups while
- * we're still allowing others to join the commit.
- */
-int btrfs_start_dirty_block_groups(struct btrfs_trans_handle *trans)
-{
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_block_group_cache *cache;
-	struct btrfs_transaction *cur_trans = trans->transaction;
-	int ret = 0;
-	int should_put;
-	struct btrfs_path *path = NULL;
-	LIST_HEAD(dirty);
-	struct list_head *io = &cur_trans->io_bgs;
-	int num_started = 0;
-	int loops = 0;
-
-	spin_lock(&cur_trans->dirty_bgs_lock);
-	if (list_empty(&cur_trans->dirty_bgs)) {
-		spin_unlock(&cur_trans->dirty_bgs_lock);
-		return 0;
-	}
-	list_splice_init(&cur_trans->dirty_bgs, &dirty);
-	spin_unlock(&cur_trans->dirty_bgs_lock);
-
-again:
-	/*
-	 * make sure all the block groups on our dirty list actually
-	 * exist
-	 */
-	btrfs_create_pending_block_groups(trans);
-
-	if (!path) {
-		path = btrfs_alloc_path();
-		if (!path)
-			return -ENOMEM;
-	}
-
-	/*
-	 * cache_write_mutex is here only to save us from balance or automatic
-	 * removal of empty block groups deleting this block group while we are
-	 * writing out the cache
-	 */
-	mutex_lock(&trans->transaction->cache_write_mutex);
-	while (!list_empty(&dirty)) {
-		bool drop_reserve = true;
-
-		cache = list_first_entry(&dirty,
-					 struct btrfs_block_group_cache,
-					 dirty_list);
-		/*
-		 * this can happen if something re-dirties a block
-		 * group that is already under IO.  Just wait for it to
-		 * finish and then do it all again
-		 */
-		if (!list_empty(&cache->io_list)) {
-			list_del_init(&cache->io_list);
-			btrfs_wait_cache_io(trans, cache, path);
-			btrfs_put_block_group(cache);
-		}
-
-
-		/*
-		 * btrfs_wait_cache_io uses the cache->dirty_list to decide
-		 * if it should update the cache_state.  Don't delete
-		 * until after we wait.
-		 *
-		 * Since we're not running in the commit critical section
-		 * we need the dirty_bgs_lock to protect from update_block_group
-		 */
-		spin_lock(&cur_trans->dirty_bgs_lock);
-		list_del_init(&cache->dirty_list);
-		spin_unlock(&cur_trans->dirty_bgs_lock);
-
-		should_put = 1;
-
-		cache_save_setup(cache, trans, path);
-
-		if (cache->disk_cache_state == BTRFS_DC_SETUP) {
-			cache->io_ctl.inode = NULL;
-			ret = btrfs_write_out_cache(trans, cache, path);
-			if (ret == 0 && cache->io_ctl.inode) {
-				num_started++;
-				should_put = 0;
-
-				/*
-				 * The cache_write_mutex is protecting the
-				 * io_list, also refer to the definition of
-				 * btrfs_transaction::io_bgs for more details
-				 */
-				list_add_tail(&cache->io_list, io);
-			} else {
-				/*
-				 * if we failed to write the cache, the
-				 * generation will be bad and life goes on
-				 */
-				ret = 0;
-			}
-		}
-		if (!ret) {
-			ret = write_one_cache_group(trans, path, cache);
-			/*
-			 * Our block group might still be attached to the list
-			 * of new block groups in the transaction handle of some
-			 * other task (struct btrfs_trans_handle->new_bgs). This
-			 * means its block group item isn't yet in the extent
-			 * tree. If this happens ignore the error, as we will
-			 * try again later in the critical section of the
-			 * transaction commit.
-			 */
-			if (ret == -ENOENT) {
-				ret = 0;
-				spin_lock(&cur_trans->dirty_bgs_lock);
-				if (list_empty(&cache->dirty_list)) {
-					list_add_tail(&cache->dirty_list,
-						      &cur_trans->dirty_bgs);
-					btrfs_get_block_group(cache);
-					drop_reserve = false;
-				}
-				spin_unlock(&cur_trans->dirty_bgs_lock);
-			} else if (ret) {
-				btrfs_abort_transaction(trans, ret);
-			}
-		}
-
-		/* if it's not on the io list, we need to put the block group */
-		if (should_put)
-			btrfs_put_block_group(cache);
-		if (drop_reserve)
-			btrfs_delayed_refs_rsv_release(fs_info, 1);
-
-		if (ret)
-			break;
-
-		/*
-		 * Avoid blocking other tasks for too long. It might even save
-		 * us from writing caches for block groups that are going to be
-		 * removed.
-		 */
-		mutex_unlock(&trans->transaction->cache_write_mutex);
-		mutex_lock(&trans->transaction->cache_write_mutex);
-	}
-	mutex_unlock(&trans->transaction->cache_write_mutex);
-
-	/*
-	 * go through delayed refs for all the stuff we've just kicked off
-	 * and then loop back (just once)
-	 */
-	ret = btrfs_run_delayed_refs(trans, 0);
-	if (!ret && loops == 0) {
-		loops++;
-		spin_lock(&cur_trans->dirty_bgs_lock);
-		list_splice_init(&cur_trans->dirty_bgs, &dirty);
-		/*
-		 * dirty_bgs_lock protects us from concurrent block group
-		 * deletes too (not just cache_write_mutex).
-		 */
-		if (!list_empty(&dirty)) {
-			spin_unlock(&cur_trans->dirty_bgs_lock);
-			goto again;
-		}
-		spin_unlock(&cur_trans->dirty_bgs_lock);
-	} else if (ret < 0) {
-		btrfs_cleanup_dirty_bgs(cur_trans, fs_info);
-	}
-
-	btrfs_free_path(path);
-	return ret;
-}
-
-int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
-{
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_block_group_cache *cache;
-	struct btrfs_transaction *cur_trans = trans->transaction;
-	int ret = 0;
-	int should_put;
-	struct btrfs_path *path;
-	struct list_head *io = &cur_trans->io_bgs;
-	int num_started = 0;
-
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-
-	/*
-	 * Even though we are in the critical section of the transaction commit,
-	 * we can still have concurrent tasks adding elements to this
-	 * transaction's list of dirty block groups. These tasks correspond to
-	 * endio free space workers started when writeback finishes for a
-	 * space cache, which run inode.c:btrfs_finish_ordered_io(), and can
-	 * allocate new block groups as a result of COWing nodes of the root
-	 * tree when updating the free space inode. The writeback for the space
-	 * caches is triggered by an earlier call to
-	 * btrfs_start_dirty_block_groups() and iterations of the following
-	 * loop.
-	 * Also 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.
-	 */
-	spin_lock(&cur_trans->dirty_bgs_lock);
-	while (!list_empty(&cur_trans->dirty_bgs)) {
-		cache = list_first_entry(&cur_trans->dirty_bgs,
-					 struct btrfs_block_group_cache,
-					 dirty_list);
-
-		/*
-		 * this can happen if cache_save_setup re-dirties a block
-		 * group that is already under IO.  Just wait for it to
-		 * finish and then do it all again
-		 */
-		if (!list_empty(&cache->io_list)) {
-			spin_unlock(&cur_trans->dirty_bgs_lock);
-			list_del_init(&cache->io_list);
-			btrfs_wait_cache_io(trans, cache, path);
-			btrfs_put_block_group(cache);
-			spin_lock(&cur_trans->dirty_bgs_lock);
-		}
-
-		/*
-		 * don't remove from the dirty list until after we've waited
-		 * on any pending IO
-		 */
-		list_del_init(&cache->dirty_list);
-		spin_unlock(&cur_trans->dirty_bgs_lock);
-		should_put = 1;
-
-		cache_save_setup(cache, trans, path);
-
-		if (!ret)
-			ret = btrfs_run_delayed_refs(trans,
-						     (unsigned long) -1);
-
-		if (!ret && cache->disk_cache_state == BTRFS_DC_SETUP) {
-			cache->io_ctl.inode = NULL;
-			ret = btrfs_write_out_cache(trans, cache, path);
-			if (ret == 0 && cache->io_ctl.inode) {
-				num_started++;
-				should_put = 0;
-				list_add_tail(&cache->io_list, io);
-			} else {
-				/*
-				 * if we failed to write the cache, the
-				 * generation will be bad and life goes on
-				 */
-				ret = 0;
-			}
-		}
-		if (!ret) {
-			ret = write_one_cache_group(trans, path, cache);
-			/*
-			 * One of the free space endio workers might have
-			 * created a new block group while updating a free space
-			 * cache's inode (at inode.c:btrfs_finish_ordered_io())
-			 * and hasn't released its transaction handle yet, in
-			 * which case the new block group is still attached to
-			 * its transaction handle and its creation has not
-			 * finished yet (no block group item in the extent tree
-			 * yet, etc). If this is the case, wait for all free
-			 * space endio workers to finish and retry. This is a
-			 * a very rare case so no need for a more efficient and
-			 * complex approach.
-			 */
-			if (ret == -ENOENT) {
-				wait_event(cur_trans->writer_wait,
-				   atomic_read(&cur_trans->num_writers) == 1);
-				ret = write_one_cache_group(trans, path, cache);
-			}
-			if (ret)
-				btrfs_abort_transaction(trans, ret);
-		}
-
-		/* 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);
-		spin_lock(&cur_trans->dirty_bgs_lock);
-	}
-	spin_unlock(&cur_trans->dirty_bgs_lock);
-
-	/*
-	 * Refer to the definition of io_bgs member for details why it's safe
-	 * to use it without any locking
-	 */
-	while (!list_empty(io)) {
-		cache = list_first_entry(io, struct btrfs_block_group_cache,
-					 io_list);
-		list_del_init(&cache->io_list);
-		btrfs_wait_cache_io(trans, cache, path);
-		btrfs_put_block_group(cache);
-	}
-
-	btrfs_free_path(path);
-	return ret;
-}
-
 int btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr)
 {
 	struct btrfs_block_group_cache *block_group;
-- 
2.21.0

[PATCH 09/15] btrfs: export block group accounting helpers

[Josef Bacik]

Want to move these functions into block-group.c, so export them.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.h |  6 ++++++
 fs/btrfs/extent-tree.c | 21 ++++++++++-----------
 2 files changed, 16 insertions(+), 11 deletions(-)

diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index f23da9d82525..e17effab028f 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -193,6 +193,12 @@ void btrfs_dec_block_group_ro(struct btrfs_block_group_cache *cache);
 int btrfs_start_dirty_block_groups(struct btrfs_trans_handle *trans);
 int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans);
 int btrfs_setup_space_cache(struct btrfs_trans_handle *trans);
+int btrfs_update_block_group(struct btrfs_trans_handle *trans,
+			     u64 bytenr, u64 num_bytes, int alloc);
+int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache,
+			     u64 ram_bytes, u64 num_bytes, int delalloc);
+void btrfs_free_reserved_bytes(struct btrfs_block_group_cache *cache,
+			       u64 num_bytes, int delalloc);
 
 static inline int btrfs_block_group_cache_done(
 		struct btrfs_block_group_cache *cache)
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 7f65958efc40..6959debccd35 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2914,8 +2914,8 @@ int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
 	return ret;
 }
 
-static int update_block_group(struct btrfs_trans_handle *trans,
-			      u64 bytenr, u64 num_bytes, int alloc)
+int btrfs_update_block_group(struct btrfs_trans_handle *trans,
+			     u64 bytenr, u64 num_bytes, int alloc)
 {
 	struct btrfs_fs_info *info = trans->fs_info;
 	struct btrfs_block_group_cache *cache = NULL;
@@ -3215,8 +3215,8 @@ btrfs_inc_block_group_reservations(struct btrfs_block_group_cache *bg)
  * reservation and the block group has become read only we cannot make the
  * reservation and return -EAGAIN, otherwise this function always succeeds.
  */
-static int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache,
-				    u64 ram_bytes, u64 num_bytes, int delalloc)
+int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache,
+			     u64 ram_bytes, u64 num_bytes, int delalloc)
 {
 	struct btrfs_space_info *space_info = cache->space_info;
 	int ret = 0;
@@ -3249,9 +3249,8 @@ static int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache,
  * A and before transaction A commits you free that leaf, you call this with
  * reserve set to 0 in order to clear the reservation.
  */
-
-static void btrfs_free_reserved_bytes(struct btrfs_block_group_cache *cache,
-				      u64 num_bytes, int delalloc)
+void btrfs_free_reserved_bytes(struct btrfs_block_group_cache *cache,
+			       u64 num_bytes, int delalloc)
 {
 	struct btrfs_space_info *space_info = cache->space_info;
 
@@ -3824,7 +3823,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 			goto out;
 		}
 
-		ret = update_block_group(trans, bytenr, num_bytes, 0);
+		ret = btrfs_update_block_group(trans, bytenr, num_bytes, 0);
 		if (ret) {
 			btrfs_abort_transaction(trans, ret);
 			goto out;
@@ -4904,7 +4903,7 @@ static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
 	if (ret)
 		return ret;
 
-	ret = update_block_group(trans, ins->objectid, ins->offset, 1);
+	ret = btrfs_update_block_group(trans, 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);
@@ -4994,8 +4993,8 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 	if (ret)
 		return ret;
 
-	ret = update_block_group(trans, extent_key.objectid,
-				 fs_info->nodesize, 1);
+	ret = btrfs_update_block_group(trans, extent_key.objectid,
+				       fs_info->nodesize, 1);
 	if (ret) { /* -ENOENT, logic error */
 		btrfs_err(fs_info, "update block group failed for %llu %llu",
 			extent_key.objectid, extent_key.offset);
-- 
2.21.0

[PATCH 10/15] btrfs: migrate the block group space accounting helpers

[Josef Bacik]

We can now easily migrate this code as well.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.c | 173 +++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/extent-tree.c | 173 -----------------------------------------
 2 files changed, 173 insertions(+), 173 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index de9d5b3cdac6..5c2995d9b572 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -2535,3 +2535,176 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans)
 	return ret;
 }
 
+int btrfs_update_block_group(struct btrfs_trans_handle *trans,
+			     u64 bytenr, u64 num_bytes, int alloc)
+{
+	struct btrfs_fs_info *info = trans->fs_info;
+	struct btrfs_block_group_cache *cache = NULL;
+	u64 total = num_bytes;
+	u64 old_val;
+	u64 byte_in_group;
+	int factor;
+	int ret = 0;
+
+	/* block accounting for super block */
+	spin_lock(&info->delalloc_root_lock);
+	old_val = btrfs_super_bytes_used(info->super_copy);
+	if (alloc)
+		old_val += num_bytes;
+	else
+		old_val -= num_bytes;
+	btrfs_set_super_bytes_used(info->super_copy, old_val);
+	spin_unlock(&info->delalloc_root_lock);
+
+	while (total) {
+		cache = btrfs_lookup_block_group(info, bytenr);
+		if (!cache) {
+			ret = -ENOENT;
+			break;
+		}
+		factor = btrfs_bg_type_to_factor(cache->flags);
+
+		/*
+		 * 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 && cache->cached == BTRFS_CACHE_NO)
+			btrfs_cache_block_group(cache, 1);
+
+		byte_in_group = bytenr - cache->key.objectid;
+		WARN_ON(byte_in_group > cache->key.offset);
+
+		spin_lock(&cache->space_info->lock);
+		spin_lock(&cache->lock);
+
+		if (btrfs_test_opt(info, SPACE_CACHE) &&
+		    cache->disk_cache_state < BTRFS_DC_CLEAR)
+			cache->disk_cache_state = BTRFS_DC_CLEAR;
+
+		old_val = btrfs_block_group_used(&cache->item);
+		num_bytes = min(total, cache->key.offset - byte_in_group);
+		if (alloc) {
+			old_val += num_bytes;
+			btrfs_set_block_group_used(&cache->item, old_val);
+			cache->reserved -= num_bytes;
+			cache->space_info->bytes_reserved -= num_bytes;
+			cache->space_info->bytes_used += num_bytes;
+			cache->space_info->disk_used += num_bytes * factor;
+			spin_unlock(&cache->lock);
+			spin_unlock(&cache->space_info->lock);
+		} else {
+			old_val -= num_bytes;
+			btrfs_set_block_group_used(&cache->item, old_val);
+			cache->pinned += num_bytes;
+			btrfs_space_info_update_bytes_pinned(info,
+					cache->space_info, num_bytes);
+			cache->space_info->bytes_used -= num_bytes;
+			cache->space_info->disk_used -= num_bytes * factor;
+			spin_unlock(&cache->lock);
+			spin_unlock(&cache->space_info->lock);
+
+			trace_btrfs_space_reservation(info, "pinned",
+						      cache->space_info->flags,
+						      num_bytes, 1);
+			percpu_counter_add_batch(&cache->space_info->total_bytes_pinned,
+					   num_bytes,
+					   BTRFS_TOTAL_BYTES_PINNED_BATCH);
+			set_extent_dirty(info->pinned_extents,
+					 bytenr, bytenr + num_bytes - 1,
+					 GFP_NOFS | __GFP_NOFAIL);
+		}
+
+		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);
+
+		/*
+		 * 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)
+			btrfs_mark_bg_unused(cache);
+
+		btrfs_put_block_group(cache);
+		total -= num_bytes;
+		bytenr += num_bytes;
+	}
+
+	/* Modified block groups are accounted for in the delayed_refs_rsv. */
+	btrfs_update_delayed_refs_rsv(trans);
+	return ret;
+}
+
+/**
+ * btrfs_add_reserved_bytes - update the block_group and space info counters
+ * @cache:	The cache we are manipulating
+ * @ram_bytes:  The number of bytes of file content, and will be same to
+ *              @num_bytes except for the compress path.
+ * @num_bytes:	The number of bytes in question
+ * @delalloc:   The blocks are allocated for the delalloc write
+ *
+ * This is called by the allocator when it reserves space. If this is a
+ * reservation and the block group has become read only we cannot make the
+ * reservation and return -EAGAIN, otherwise this function always succeeds.
+ */
+int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache,
+			     u64 ram_bytes, u64 num_bytes, int delalloc)
+{
+	struct btrfs_space_info *space_info = cache->space_info;
+	int ret = 0;
+
+	spin_lock(&space_info->lock);
+	spin_lock(&cache->lock);
+	if (cache->ro) {
+		ret = -EAGAIN;
+	} else {
+		cache->reserved += num_bytes;
+		space_info->bytes_reserved += num_bytes;
+		btrfs_space_info_update_bytes_may_use(cache->fs_info,
+						      space_info, -ram_bytes);
+		if (delalloc)
+			cache->delalloc_bytes += num_bytes;
+	}
+	spin_unlock(&cache->lock);
+	spin_unlock(&space_info->lock);
+	return ret;
+}
+
+/**
+ * btrfs_free_reserved_bytes - update the block_group and space info counters
+ * @cache:      The cache we are manipulating
+ * @num_bytes:  The number of bytes in question
+ * @delalloc:   The blocks are allocated for the delalloc write
+ *
+ * This is called by somebody who is freeing space that was never actually used
+ * on disk.  For example if you reserve some space for a new leaf in transaction
+ * A and before transaction A commits you free that leaf, you call this with
+ * reserve set to 0 in order to clear the reservation.
+ */
+void btrfs_free_reserved_bytes(struct btrfs_block_group_cache *cache,
+			       u64 num_bytes, int delalloc)
+{
+	struct btrfs_space_info *space_info = cache->space_info;
+
+	spin_lock(&space_info->lock);
+	spin_lock(&cache->lock);
+	if (cache->ro)
+		space_info->bytes_readonly += num_bytes;
+	cache->reserved -= num_bytes;
+	space_info->bytes_reserved -= num_bytes;
+	space_info->max_extent_size = 0;
+
+	if (delalloc)
+		cache->delalloc_bytes -= num_bytes;
+	spin_unlock(&cache->lock);
+	spin_unlock(&space_info->lock);
+}
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 6959debccd35..7e694d4837ad 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2914,115 +2914,6 @@ int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
 	return ret;
 }
 
-int btrfs_update_block_group(struct btrfs_trans_handle *trans,
-			     u64 bytenr, u64 num_bytes, int alloc)
-{
-	struct btrfs_fs_info *info = trans->fs_info;
-	struct btrfs_block_group_cache *cache = NULL;
-	u64 total = num_bytes;
-	u64 old_val;
-	u64 byte_in_group;
-	int factor;
-	int ret = 0;
-
-	/* block accounting for super block */
-	spin_lock(&info->delalloc_root_lock);
-	old_val = btrfs_super_bytes_used(info->super_copy);
-	if (alloc)
-		old_val += num_bytes;
-	else
-		old_val -= num_bytes;
-	btrfs_set_super_bytes_used(info->super_copy, old_val);
-	spin_unlock(&info->delalloc_root_lock);
-
-	while (total) {
-		cache = btrfs_lookup_block_group(info, bytenr);
-		if (!cache) {
-			ret = -ENOENT;
-			break;
-		}
-		factor = btrfs_bg_type_to_factor(cache->flags);
-
-		/*
-		 * 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 && cache->cached == BTRFS_CACHE_NO)
-			btrfs_cache_block_group(cache, 1);
-
-		byte_in_group = bytenr - cache->key.objectid;
-		WARN_ON(byte_in_group > cache->key.offset);
-
-		spin_lock(&cache->space_info->lock);
-		spin_lock(&cache->lock);
-
-		if (btrfs_test_opt(info, SPACE_CACHE) &&
-		    cache->disk_cache_state < BTRFS_DC_CLEAR)
-			cache->disk_cache_state = BTRFS_DC_CLEAR;
-
-		old_val = btrfs_block_group_used(&cache->item);
-		num_bytes = min(total, cache->key.offset - byte_in_group);
-		if (alloc) {
-			old_val += num_bytes;
-			btrfs_set_block_group_used(&cache->item, old_val);
-			cache->reserved -= num_bytes;
-			cache->space_info->bytes_reserved -= num_bytes;
-			cache->space_info->bytes_used += num_bytes;
-			cache->space_info->disk_used += num_bytes * factor;
-			spin_unlock(&cache->lock);
-			spin_unlock(&cache->space_info->lock);
-		} else {
-			old_val -= num_bytes;
-			btrfs_set_block_group_used(&cache->item, old_val);
-			cache->pinned += num_bytes;
-			btrfs_space_info_update_bytes_pinned(info,
-					cache->space_info, num_bytes);
-			cache->space_info->bytes_used -= num_bytes;
-			cache->space_info->disk_used -= num_bytes * factor;
-			spin_unlock(&cache->lock);
-			spin_unlock(&cache->space_info->lock);
-
-			trace_btrfs_space_reservation(info, "pinned",
-						      cache->space_info->flags,
-						      num_bytes, 1);
-			percpu_counter_add_batch(&cache->space_info->total_bytes_pinned,
-					   num_bytes,
-					   BTRFS_TOTAL_BYTES_PINNED_BATCH);
-			set_extent_dirty(info->pinned_extents,
-					 bytenr, bytenr + num_bytes - 1,
-					 GFP_NOFS | __GFP_NOFAIL);
-		}
-
-		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);
-
-		/*
-		 * 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)
-			btrfs_mark_bg_unused(cache);
-
-		btrfs_put_block_group(cache);
-		total -= num_bytes;
-		bytenr += num_bytes;
-	}
-
-	/* Modified block groups are accounted for in the delayed_refs_rsv. */
-	btrfs_update_delayed_refs_rsv(trans);
-	return ret;
-}
-
 static u64 first_logical_byte(struct btrfs_fs_info *fs_info, u64 search_start)
 {
 	struct btrfs_block_group_cache *cache;
@@ -3203,70 +3094,6 @@ btrfs_inc_block_group_reservations(struct btrfs_block_group_cache *bg)
 	atomic_inc(&bg->reservations);
 }
 
-/**
- * btrfs_add_reserved_bytes - update the block_group and space info counters
- * @cache:	The cache we are manipulating
- * @ram_bytes:  The number of bytes of file content, and will be same to
- *              @num_bytes except for the compress path.
- * @num_bytes:	The number of bytes in question
- * @delalloc:   The blocks are allocated for the delalloc write
- *
- * This is called by the allocator when it reserves space. If this is a
- * reservation and the block group has become read only we cannot make the
- * reservation and return -EAGAIN, otherwise this function always succeeds.
- */
-int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache,
-			     u64 ram_bytes, u64 num_bytes, int delalloc)
-{
-	struct btrfs_space_info *space_info = cache->space_info;
-	int ret = 0;
-
-	spin_lock(&space_info->lock);
-	spin_lock(&cache->lock);
-	if (cache->ro) {
-		ret = -EAGAIN;
-	} else {
-		cache->reserved += num_bytes;
-		space_info->bytes_reserved += num_bytes;
-		btrfs_space_info_update_bytes_may_use(cache->fs_info,
-						      space_info, -ram_bytes);
-		if (delalloc)
-			cache->delalloc_bytes += num_bytes;
-	}
-	spin_unlock(&cache->lock);
-	spin_unlock(&space_info->lock);
-	return ret;
-}
-
-/**
- * btrfs_free_reserved_bytes - update the block_group and space info counters
- * @cache:      The cache we are manipulating
- * @num_bytes:  The number of bytes in question
- * @delalloc:   The blocks are allocated for the delalloc write
- *
- * This is called by somebody who is freeing space that was never actually used
- * on disk.  For example if you reserve some space for a new leaf in transaction
- * A and before transaction A commits you free that leaf, you call this with
- * reserve set to 0 in order to clear the reservation.
- */
-void btrfs_free_reserved_bytes(struct btrfs_block_group_cache *cache,
-			       u64 num_bytes, int delalloc)
-{
-	struct btrfs_space_info *space_info = cache->space_info;
-
-	spin_lock(&space_info->lock);
-	spin_lock(&cache->lock);
-	if (cache->ro)
-		space_info->bytes_readonly += num_bytes;
-	cache->reserved -= num_bytes;
-	space_info->bytes_reserved -= num_bytes;
-	space_info->max_extent_size = 0;
-
-	if (delalloc)
-		cache->delalloc_bytes -= num_bytes;
-	spin_unlock(&cache->lock);
-	spin_unlock(&space_info->lock);
-}
 void btrfs_prepare_extent_commit(struct btrfs_fs_info *fs_info)
 {
 	struct btrfs_caching_control *next;
-- 
2.21.0

[PATCH 11/15] btrfs: migrate the chunk allocation code

[Josef Bacik]

This feels more at home in block-group.c than in extent-tree.c.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.c    | 245 ++++++++++++++++++++++++++++++++++++++
 fs/btrfs/block-group.h    |  24 ++++
 fs/btrfs/ctree.h          |  24 ----
 fs/btrfs/delalloc-space.c |   1 +
 fs/btrfs/extent-tree.c    | 244 -------------------------------------
 5 files changed, 270 insertions(+), 268 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 5c2995d9b572..0221c8c6c7d4 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -13,6 +13,7 @@
 #include "sysfs.h"
 #include "tree-log.h"
 #include "delalloc-space.h"
+#include "math.h"
 
 void btrfs_get_block_group(struct btrfs_block_group_cache *cache)
 {
@@ -2708,3 +2709,247 @@ void btrfs_free_reserved_bytes(struct btrfs_block_group_cache *cache,
 	spin_unlock(&cache->lock);
 	spin_unlock(&space_info->lock);
 }
+
+static void force_metadata_allocation(struct btrfs_fs_info *info)
+{
+	struct list_head *head = &info->space_info;
+	struct btrfs_space_info *found;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(found, head, list) {
+		if (found->flags & BTRFS_BLOCK_GROUP_METADATA)
+			found->force_alloc = CHUNK_ALLOC_FORCE;
+	}
+	rcu_read_unlock();
+}
+
+static int should_alloc_chunk(struct btrfs_fs_info *fs_info,
+			      struct btrfs_space_info *sinfo, int force)
+{
+	u64 bytes_used = btrfs_space_info_used(sinfo, false);
+	u64 thresh;
+
+	if (force == CHUNK_ALLOC_FORCE)
+		return 1;
+
+	/*
+	 * in limited mode, we want to have some free space up to
+	 * about 1% of the FS size.
+	 */
+	if (force == CHUNK_ALLOC_LIMITED) {
+		thresh = btrfs_super_total_bytes(fs_info->super_copy);
+		thresh = max_t(u64, SZ_64M, div_factor_fine(thresh, 1));
+
+		if (sinfo->total_bytes - bytes_used < thresh)
+			return 1;
+	}
+
+	if (bytes_used + SZ_2M < div_factor(sinfo->total_bytes, 8))
+		return 0;
+	return 1;
+}
+
+static u64 get_profile_num_devs(struct btrfs_fs_info *fs_info, u64 type)
+{
+	u64 num_dev;
+
+	num_dev = btrfs_raid_array[btrfs_bg_flags_to_raid_index(type)].devs_max;
+	if (!num_dev)
+		num_dev = fs_info->fs_devices->rw_devices;
+
+	return num_dev;
+}
+
+/*
+ * If @is_allocation is true, reserve space in the system space info necessary
+ * for allocating a chunk, otherwise if it's false, reserve space necessary for
+ * removing a chunk.
+ */
+void check_system_chunk(struct btrfs_trans_handle *trans, u64 type)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_space_info *info;
+	u64 left;
+	u64 thresh;
+	int ret = 0;
+	u64 num_devs;
+
+	/*
+	 * Needed because we can end up allocating a system chunk and for an
+	 * atomic and race free space reservation in the chunk block reserve.
+	 */
+	lockdep_assert_held(&fs_info->chunk_mutex);
+
+	info = btrfs_find_space_info(fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
+	spin_lock(&info->lock);
+	left = info->total_bytes - btrfs_space_info_used(info, true);
+	spin_unlock(&info->lock);
+
+	num_devs = get_profile_num_devs(fs_info, type);
+
+	/* num_devs device items to update and 1 chunk item to add or remove */
+	thresh = btrfs_calc_trunc_metadata_size(fs_info, num_devs) +
+		btrfs_calc_trans_metadata_size(fs_info, 1);
+
+	if (left < thresh && btrfs_test_opt(fs_info, ENOSPC_DEBUG)) {
+		btrfs_info(fs_info, "left=%llu, need=%llu, flags=%llu",
+			   left, thresh, type);
+		btrfs_dump_space_info(fs_info, info, 0, 0);
+	}
+
+	if (left < thresh) {
+		u64 flags = btrfs_system_alloc_profile(fs_info);
+
+		/*
+		 * Ignore failure to create system chunk. We might end up not
+		 * needing it, as we might not need to COW all nodes/leafs from
+		 * the paths we visit in the chunk tree (they were already COWed
+		 * or created in the current transaction for example).
+		 */
+		ret = btrfs_alloc_chunk(trans, flags);
+	}
+
+	if (!ret) {
+		ret = btrfs_block_rsv_add(fs_info->chunk_root,
+					  &fs_info->chunk_block_rsv,
+					  thresh, BTRFS_RESERVE_NO_FLUSH);
+		if (!ret)
+			trans->chunk_bytes_reserved += thresh;
+	}
+}
+
+int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, u64 type)
+{
+	u64 alloc_flags = btrfs_get_alloc_profile(trans->fs_info, type);
+
+	return btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
+}
+
+/*
+ * If force is CHUNK_ALLOC_FORCE:
+ *    - return 1 if it successfully allocates a chunk,
+ *    - return errors including -ENOSPC otherwise.
+ * If force is NOT CHUNK_ALLOC_FORCE:
+ *    - return 0 if it doesn't need to allocate a new chunk,
+ *    - return 1 if it successfully allocates a chunk,
+ *    - return errors including -ENOSPC otherwise.
+ */
+int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
+		      enum btrfs_chunk_alloc_enum force)
+{
+	struct btrfs_fs_info *fs_info = trans->fs_info;
+	struct btrfs_space_info *space_info;
+	bool wait_for_alloc = false;
+	bool should_alloc = false;
+	int ret = 0;
+
+	/* Don't re-enter if we're already allocating a chunk */
+	if (trans->allocating_chunk)
+		return -ENOSPC;
+
+	space_info = btrfs_find_space_info(fs_info, flags);
+	ASSERT(space_info);
+
+	do {
+		spin_lock(&space_info->lock);
+		if (force < space_info->force_alloc)
+			force = space_info->force_alloc;
+		should_alloc = should_alloc_chunk(fs_info, space_info, force);
+		if (space_info->full) {
+			/* No more free physical space */
+			if (should_alloc)
+				ret = -ENOSPC;
+			else
+				ret = 0;
+			spin_unlock(&space_info->lock);
+			return ret;
+		} else if (!should_alloc) {
+			spin_unlock(&space_info->lock);
+			return 0;
+		} else if (space_info->chunk_alloc) {
+			/*
+			 * Someone is already allocating, so we need to block
+			 * until this someone is finished and then loop to
+			 * recheck if we should continue with our allocation
+			 * attempt.
+			 */
+			wait_for_alloc = true;
+			spin_unlock(&space_info->lock);
+			mutex_lock(&fs_info->chunk_mutex);
+			mutex_unlock(&fs_info->chunk_mutex);
+		} else {
+			/* Proceed with allocation */
+			space_info->chunk_alloc = 1;
+			wait_for_alloc = false;
+			spin_unlock(&space_info->lock);
+		}
+
+		cond_resched();
+	} while (wait_for_alloc);
+
+	mutex_lock(&fs_info->chunk_mutex);
+	trans->allocating_chunk = true;
+
+	/*
+	 * If we have mixed data/metadata chunks we want to make sure we keep
+	 * allocating mixed chunks instead of individual chunks.
+	 */
+	if (btrfs_mixed_space_info(space_info))
+		flags |= (BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA);
+
+	/*
+	 * if we're doing a data chunk, go ahead and make sure that
+	 * we keep a reasonable number of metadata chunks allocated in the
+	 * FS as well.
+	 */
+	if (flags & BTRFS_BLOCK_GROUP_DATA && fs_info->metadata_ratio) {
+		fs_info->data_chunk_allocations++;
+		if (!(fs_info->data_chunk_allocations %
+		      fs_info->metadata_ratio))
+			force_metadata_allocation(fs_info);
+	}
+
+	/*
+	 * Check if we have enough space in SYSTEM chunk because we may need
+	 * to update devices.
+	 */
+	check_system_chunk(trans, flags);
+
+	ret = btrfs_alloc_chunk(trans, flags);
+	trans->allocating_chunk = false;
+
+	spin_lock(&space_info->lock);
+	if (ret < 0) {
+		if (ret == -ENOSPC)
+			space_info->full = 1;
+		else
+			goto out;
+	} else {
+		ret = 1;
+		space_info->max_extent_size = 0;
+	}
+
+	space_info->force_alloc = CHUNK_ALLOC_NO_FORCE;
+out:
+	space_info->chunk_alloc = 0;
+	spin_unlock(&space_info->lock);
+	mutex_unlock(&fs_info->chunk_mutex);
+	/*
+	 * When we allocate a new chunk we reserve space in the chunk block
+	 * reserve to make sure we can COW nodes/leafs in the chunk tree or
+	 * add new nodes/leafs to it if we end up needing to do it when
+	 * inserting the chunk item and updating device items as part of the
+	 * second phase of chunk allocation, performed by
+	 * btrfs_finish_chunk_alloc(). So make sure we don't accumulate a
+	 * large number of new block groups to create in our transaction
+	 * handle's new_bgs list to avoid exhausting the chunk block reserve
+	 * in extreme cases - like having a single transaction create many new
+	 * block groups when starting to write out the free space caches of all
+	 * the block groups that were made dirty during the lifetime of the
+	 * transaction.
+	 */
+	if (trans->chunk_bytes_reserved >= (u64)SZ_2M)
+		btrfs_create_pending_block_groups(trans);
+
+	return ret;
+}
diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index e17effab028f..4b8ac8e9323c 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -10,6 +10,26 @@ enum btrfs_disk_cache_state {
 	BTRFS_DC_SETUP,
 };
 
+/*
+ * control flags for do_chunk_alloc's force field
+ * CHUNK_ALLOC_NO_FORCE means to only allocate a chunk
+ * if we really need one.
+ *
+ * CHUNK_ALLOC_LIMITED means to only try and allocate one
+ * if we have very few chunks already allocated.  This is
+ * used as part of the clustering code to help make sure
+ * we have a good pool of storage to cluster in, without
+ * filling the FS with empty chunks
+ *
+ * CHUNK_ALLOC_FORCE means it must try to allocate one
+ *
+ */
+enum btrfs_chunk_alloc_enum {
+	CHUNK_ALLOC_NO_FORCE,
+	CHUNK_ALLOC_LIMITED,
+	CHUNK_ALLOC_FORCE,
+};
+
 struct btrfs_caching_control {
 	struct list_head list;
 	struct mutex mutex;
@@ -199,6 +219,10 @@ int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache,
 			     u64 ram_bytes, u64 num_bytes, int delalloc);
 void btrfs_free_reserved_bytes(struct btrfs_block_group_cache *cache,
 			       u64 num_bytes, int delalloc);
+int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
+		      enum btrfs_chunk_alloc_enum force);
+int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, u64 type);
+void check_system_chunk(struct btrfs_trans_handle *trans, const u64 type);
 
 static inline int btrfs_block_group_cache_done(
 		struct btrfs_block_group_cache *cache)
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 2a41e09727c3..db56a1c81843 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2556,28 +2556,6 @@ enum btrfs_flush_state {
 	COMMIT_TRANS		=	10,
 };
 
-/*
- * control flags for do_chunk_alloc's force field
- * CHUNK_ALLOC_NO_FORCE means to only allocate a chunk
- * if we really need one.
- *
- * CHUNK_ALLOC_LIMITED means to only try and allocate one
- * if we have very few chunks already allocated.  This is
- * used as part of the clustering code to help make sure
- * we have a good pool of storage to cluster in, without
- * filling the FS with empty chunks
- *
- * CHUNK_ALLOC_FORCE means it must try to allocate one
- *
- */
-enum btrfs_chunk_alloc_enum {
-	CHUNK_ALLOC_NO_FORCE,
-	CHUNK_ALLOC_LIMITED,
-	CHUNK_ALLOC_FORCE,
-};
-
-int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
-		      enum btrfs_chunk_alloc_enum force);
 int btrfs_subvolume_reserve_metadata(struct btrfs_root *root,
 				     struct btrfs_block_rsv *rsv,
 				     int nitems, bool use_global_rsv);
@@ -2593,7 +2571,6 @@ int btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info,
 				   u64 start, u64 end);
 int btrfs_discard_extent(struct btrfs_fs_info *fs_info, u64 bytenr,
 			 u64 num_bytes, u64 *actual_bytes);
-int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, u64 type);
 int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range);
 
 int btrfs_init_space_info(struct btrfs_fs_info *fs_info);
@@ -2602,7 +2579,6 @@ int btrfs_delayed_refs_qgroup_accounting(struct btrfs_trans_handle *trans,
 int btrfs_start_write_no_snapshotting(struct btrfs_root *root);
 void btrfs_end_write_no_snapshotting(struct btrfs_root *root);
 void btrfs_wait_for_snapshot_creation(struct btrfs_root *root);
-void check_system_chunk(struct btrfs_trans_handle *trans, const u64 type);
 
 /* ctree.c */
 int btrfs_bin_search(struct extent_buffer *eb, const struct btrfs_key *key,
diff --git a/fs/btrfs/delalloc-space.c b/fs/btrfs/delalloc-space.c
index 17f7c0d38768..d2dfc201b2e1 100644
--- a/fs/btrfs/delalloc-space.c
+++ b/fs/btrfs/delalloc-space.c
@@ -7,6 +7,7 @@
 #include "space-info.h"
 #include "transaction.h"
 #include "qgroup.h"
+#include "block-group.h"
 
 int btrfs_alloc_data_chunk_ondemand(struct btrfs_inode *inode, u64 bytes)
 {
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 7e694d4837ad..a218f2187ebe 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2677,243 +2677,6 @@ u64 btrfs_system_alloc_profile(struct btrfs_fs_info *fs_info)
 	return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
 }
 
-static void force_metadata_allocation(struct btrfs_fs_info *info)
-{
-	struct list_head *head = &info->space_info;
-	struct btrfs_space_info *found;
-
-	rcu_read_lock();
-	list_for_each_entry_rcu(found, head, list) {
-		if (found->flags & BTRFS_BLOCK_GROUP_METADATA)
-			found->force_alloc = CHUNK_ALLOC_FORCE;
-	}
-	rcu_read_unlock();
-}
-
-static int should_alloc_chunk(struct btrfs_fs_info *fs_info,
-			      struct btrfs_space_info *sinfo, int force)
-{
-	u64 bytes_used = btrfs_space_info_used(sinfo, false);
-	u64 thresh;
-
-	if (force == CHUNK_ALLOC_FORCE)
-		return 1;
-
-	/*
-	 * in limited mode, we want to have some free space up to
-	 * about 1% of the FS size.
-	 */
-	if (force == CHUNK_ALLOC_LIMITED) {
-		thresh = btrfs_super_total_bytes(fs_info->super_copy);
-		thresh = max_t(u64, SZ_64M, div_factor_fine(thresh, 1));
-
-		if (sinfo->total_bytes - bytes_used < thresh)
-			return 1;
-	}
-
-	if (bytes_used + SZ_2M < div_factor(sinfo->total_bytes, 8))
-		return 0;
-	return 1;
-}
-
-static u64 get_profile_num_devs(struct btrfs_fs_info *fs_info, u64 type)
-{
-	u64 num_dev;
-
-	num_dev = btrfs_raid_array[btrfs_bg_flags_to_raid_index(type)].devs_max;
-	if (!num_dev)
-		num_dev = fs_info->fs_devices->rw_devices;
-
-	return num_dev;
-}
-
-/*
- * If @is_allocation is true, reserve space in the system space info necessary
- * for allocating a chunk, otherwise if it's false, reserve space necessary for
- * removing a chunk.
- */
-void check_system_chunk(struct btrfs_trans_handle *trans, u64 type)
-{
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_space_info *info;
-	u64 left;
-	u64 thresh;
-	int ret = 0;
-	u64 num_devs;
-
-	/*
-	 * Needed because we can end up allocating a system chunk and for an
-	 * atomic and race free space reservation in the chunk block reserve.
-	 */
-	lockdep_assert_held(&fs_info->chunk_mutex);
-
-	info = btrfs_find_space_info(fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
-	spin_lock(&info->lock);
-	left = info->total_bytes - btrfs_space_info_used(info, true);
-	spin_unlock(&info->lock);
-
-	num_devs = get_profile_num_devs(fs_info, type);
-
-	/* num_devs device items to update and 1 chunk item to add or remove */
-	thresh = btrfs_calc_trunc_metadata_size(fs_info, num_devs) +
-		btrfs_calc_trans_metadata_size(fs_info, 1);
-
-	if (left < thresh && btrfs_test_opt(fs_info, ENOSPC_DEBUG)) {
-		btrfs_info(fs_info, "left=%llu, need=%llu, flags=%llu",
-			   left, thresh, type);
-		btrfs_dump_space_info(fs_info, info, 0, 0);
-	}
-
-	if (left < thresh) {
-		u64 flags = btrfs_system_alloc_profile(fs_info);
-
-		/*
-		 * Ignore failure to create system chunk. We might end up not
-		 * needing it, as we might not need to COW all nodes/leafs from
-		 * the paths we visit in the chunk tree (they were already COWed
-		 * or created in the current transaction for example).
-		 */
-		ret = btrfs_alloc_chunk(trans, flags);
-	}
-
-	if (!ret) {
-		ret = btrfs_block_rsv_add(fs_info->chunk_root,
-					  &fs_info->chunk_block_rsv,
-					  thresh, BTRFS_RESERVE_NO_FLUSH);
-		if (!ret)
-			trans->chunk_bytes_reserved += thresh;
-	}
-}
-
-/*
- * If force is CHUNK_ALLOC_FORCE:
- *    - return 1 if it successfully allocates a chunk,
- *    - return errors including -ENOSPC otherwise.
- * If force is NOT CHUNK_ALLOC_FORCE:
- *    - return 0 if it doesn't need to allocate a new chunk,
- *    - return 1 if it successfully allocates a chunk,
- *    - return errors including -ENOSPC otherwise.
- */
-int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
-		      enum btrfs_chunk_alloc_enum force)
-{
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct btrfs_space_info *space_info;
-	bool wait_for_alloc = false;
-	bool should_alloc = false;
-	int ret = 0;
-
-	/* Don't re-enter if we're already allocating a chunk */
-	if (trans->allocating_chunk)
-		return -ENOSPC;
-
-	space_info = btrfs_find_space_info(fs_info, flags);
-	ASSERT(space_info);
-
-	do {
-		spin_lock(&space_info->lock);
-		if (force < space_info->force_alloc)
-			force = space_info->force_alloc;
-		should_alloc = should_alloc_chunk(fs_info, space_info, force);
-		if (space_info->full) {
-			/* No more free physical space */
-			if (should_alloc)
-				ret = -ENOSPC;
-			else
-				ret = 0;
-			spin_unlock(&space_info->lock);
-			return ret;
-		} else if (!should_alloc) {
-			spin_unlock(&space_info->lock);
-			return 0;
-		} else if (space_info->chunk_alloc) {
-			/*
-			 * Someone is already allocating, so we need to block
-			 * until this someone is finished and then loop to
-			 * recheck if we should continue with our allocation
-			 * attempt.
-			 */
-			wait_for_alloc = true;
-			spin_unlock(&space_info->lock);
-			mutex_lock(&fs_info->chunk_mutex);
-			mutex_unlock(&fs_info->chunk_mutex);
-		} else {
-			/* Proceed with allocation */
-			space_info->chunk_alloc = 1;
-			wait_for_alloc = false;
-			spin_unlock(&space_info->lock);
-		}
-
-		cond_resched();
-	} while (wait_for_alloc);
-
-	mutex_lock(&fs_info->chunk_mutex);
-	trans->allocating_chunk = true;
-
-	/*
-	 * If we have mixed data/metadata chunks we want to make sure we keep
-	 * allocating mixed chunks instead of individual chunks.
-	 */
-	if (btrfs_mixed_space_info(space_info))
-		flags |= (BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA);
-
-	/*
-	 * if we're doing a data chunk, go ahead and make sure that
-	 * we keep a reasonable number of metadata chunks allocated in the
-	 * FS as well.
-	 */
-	if (flags & BTRFS_BLOCK_GROUP_DATA && fs_info->metadata_ratio) {
-		fs_info->data_chunk_allocations++;
-		if (!(fs_info->data_chunk_allocations %
-		      fs_info->metadata_ratio))
-			force_metadata_allocation(fs_info);
-	}
-
-	/*
-	 * Check if we have enough space in SYSTEM chunk because we may need
-	 * to update devices.
-	 */
-	check_system_chunk(trans, flags);
-
-	ret = btrfs_alloc_chunk(trans, flags);
-	trans->allocating_chunk = false;
-
-	spin_lock(&space_info->lock);
-	if (ret < 0) {
-		if (ret == -ENOSPC)
-			space_info->full = 1;
-		else
-			goto out;
-	} else {
-		ret = 1;
-		space_info->max_extent_size = 0;
-	}
-
-	space_info->force_alloc = CHUNK_ALLOC_NO_FORCE;
-out:
-	space_info->chunk_alloc = 0;
-	spin_unlock(&space_info->lock);
-	mutex_unlock(&fs_info->chunk_mutex);
-	/*
-	 * When we allocate a new chunk we reserve space in the chunk block
-	 * reserve to make sure we can COW nodes/leafs in the chunk tree or
-	 * add new nodes/leafs to it if we end up needing to do it when
-	 * inserting the chunk item and updating device items as part of the
-	 * second phase of chunk allocation, performed by
-	 * btrfs_finish_chunk_alloc(). So make sure we don't accumulate a
-	 * large number of new block groups to create in our transaction
-	 * handle's new_bgs list to avoid exhausting the chunk block reserve
-	 * in extreme cases - like having a single transaction create many new
-	 * block groups when starting to write out the free space caches of all
-	 * the block groups that were made dirty during the lifetime of the
-	 * transaction.
-	 */
-	if (trans->chunk_bytes_reserved >= (u64)SZ_2M)
-		btrfs_create_pending_block_groups(trans);
-
-	return ret;
-}
-
 static u64 first_logical_byte(struct btrfs_fs_info *fs_info, u64 search_start)
 {
 	struct btrfs_block_group_cache *cache;
@@ -5971,13 +5734,6 @@ int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
-int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, u64 type)
-{
-	u64 alloc_flags = btrfs_get_alloc_profile(trans->fs_info, type);
-
-	return btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
-}
-
 /*
  * helper to account the unused space of all the readonly block group in the
  * space_info. takes mirrors into account.
-- 
2.21.0

[PATCH 12/15] btrfs: migrate the alloc_profile helpers

[Josef Bacik]

These feel more at home in block-group.c.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.c | 100 +++++++++++++++++++++++++++++++++++
 fs/btrfs/block-group.h |  16 ++++++
 fs/btrfs/ctree.h       |   4 --
 fs/btrfs/extent-tree.c | 115 -----------------------------------------
 4 files changed, 116 insertions(+), 119 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 0221c8c6c7d4..330f96134ffb 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -15,6 +15,106 @@
 #include "delalloc-space.h"
 #include "math.h"
 
+/*
+ * returns target flags in extended format or 0 if restripe for this
+ * chunk_type is not in progress
+ *
+ * should be called with balance_lock held
+ */
+u64 btrfs_get_restripe_target(struct btrfs_fs_info *fs_info, u64 flags)
+{
+	struct btrfs_balance_control *bctl = fs_info->balance_ctl;
+	u64 target = 0;
+
+	if (!bctl)
+		return 0;
+
+	if (flags & BTRFS_BLOCK_GROUP_DATA &&
+	    bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) {
+		target = BTRFS_BLOCK_GROUP_DATA | bctl->data.target;
+	} else if (flags & BTRFS_BLOCK_GROUP_SYSTEM &&
+		   bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
+		target = BTRFS_BLOCK_GROUP_SYSTEM | bctl->sys.target;
+	} else if (flags & BTRFS_BLOCK_GROUP_METADATA &&
+		   bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) {
+		target = BTRFS_BLOCK_GROUP_METADATA | bctl->meta.target;
+	}
+
+	return target;
+}
+
+/*
+ * @flags: available profiles in extended format (see ctree.h)
+ *
+ * Returns reduced profile in chunk format.  If profile changing is in
+ * progress (either running or paused) picks the target profile (if it's
+ * already available), otherwise falls back to plain reducing.
+ */
+static u64 btrfs_reduce_alloc_profile(struct btrfs_fs_info *fs_info, u64 flags)
+{
+	u64 num_devices = fs_info->fs_devices->rw_devices;
+	u64 target;
+	u64 raid_type;
+	u64 allowed = 0;
+
+	/*
+	 * see if restripe for this chunk_type is in progress, if so
+	 * try to reduce to the target profile
+	 */
+	spin_lock(&fs_info->balance_lock);
+	target = btrfs_get_restripe_target(fs_info, flags);
+	if (target) {
+		/* pick target profile only if it's already available */
+		if ((flags & target) & BTRFS_EXTENDED_PROFILE_MASK) {
+			spin_unlock(&fs_info->balance_lock);
+			return extended_to_chunk(target);
+		}
+	}
+	spin_unlock(&fs_info->balance_lock);
+
+	/* First, mask out the RAID levels which aren't possible */
+	for (raid_type = 0; raid_type < BTRFS_NR_RAID_TYPES; raid_type++) {
+		if (num_devices >= btrfs_raid_array[raid_type].devs_min)
+			allowed |= btrfs_raid_array[raid_type].bg_flag;
+	}
+	allowed &= flags;
+
+	if (allowed & BTRFS_BLOCK_GROUP_RAID6)
+		allowed = BTRFS_BLOCK_GROUP_RAID6;
+	else if (allowed & BTRFS_BLOCK_GROUP_RAID5)
+		allowed = BTRFS_BLOCK_GROUP_RAID5;
+	else if (allowed & BTRFS_BLOCK_GROUP_RAID10)
+		allowed = BTRFS_BLOCK_GROUP_RAID10;
+	else if (allowed & BTRFS_BLOCK_GROUP_RAID1)
+		allowed = BTRFS_BLOCK_GROUP_RAID1;
+	else if (allowed & BTRFS_BLOCK_GROUP_RAID0)
+		allowed = BTRFS_BLOCK_GROUP_RAID0;
+
+	flags &= ~BTRFS_BLOCK_GROUP_PROFILE_MASK;
+
+	return extended_to_chunk(flags | allowed);
+}
+
+u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags)
+{
+	unsigned seq;
+	u64 flags;
+
+	do {
+		flags = orig_flags;
+		seq = read_seqbegin(&fs_info->profiles_lock);
+
+		if (flags & BTRFS_BLOCK_GROUP_DATA)
+			flags |= fs_info->avail_data_alloc_bits;
+		else if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
+			flags |= fs_info->avail_system_alloc_bits;
+		else if (flags & BTRFS_BLOCK_GROUP_METADATA)
+			flags |= fs_info->avail_metadata_alloc_bits;
+	} while (read_seqretry(&fs_info->profiles_lock, seq));
+
+	return btrfs_reduce_alloc_profile(fs_info, flags);
+}
+
 void btrfs_get_block_group(struct btrfs_block_group_cache *cache)
 {
 	atomic_inc(&cache->count);
diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index 4b8ac8e9323c..ab5434ddd7f4 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -223,6 +223,22 @@ int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
 		      enum btrfs_chunk_alloc_enum force);
 int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, u64 type);
 void check_system_chunk(struct btrfs_trans_handle *trans, const u64 type);
+u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags);
+
+static inline u64 btrfs_data_alloc_profile(struct btrfs_fs_info *fs_info)
+{
+	return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_DATA);
+}
+
+static inline u64 btrfs_metadata_alloc_profile(struct btrfs_fs_info *fs_info)
+{
+	return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_METADATA);
+}
+
+static inline u64 btrfs_system_alloc_profile(struct btrfs_fs_info *fs_info)
+{
+	return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
+}
 
 static inline int btrfs_block_group_cache_done(
 		struct btrfs_block_group_cache *cache)
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index db56a1c81843..fab6e9792d23 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2467,7 +2467,6 @@ static inline u64 btrfs_calc_trunc_metadata_size(struct btrfs_fs_info *fs_info,
 int btrfs_add_excluded_extent(struct btrfs_fs_info *fs_info,
 			      u64 start, u64 num_bytes);
 void btrfs_free_excluded_extents(struct btrfs_block_group_cache *cache);
-u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags);
 int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 			   unsigned long count);
 void btrfs_cleanup_ref_head_accounting(struct btrfs_fs_info *fs_info,
@@ -2526,9 +2525,6 @@ int btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr);
 int btrfs_free_block_groups(struct btrfs_fs_info *info);
 void btrfs_get_block_group_trimming(struct btrfs_block_group_cache *cache);
 void btrfs_put_block_group_trimming(struct btrfs_block_group_cache *cache);
-u64 btrfs_data_alloc_profile(struct btrfs_fs_info *fs_info);
-u64 btrfs_metadata_alloc_profile(struct btrfs_fs_info *fs_info);
-u64 btrfs_system_alloc_profile(struct btrfs_fs_info *fs_info);
 void btrfs_clear_space_info_full(struct btrfs_fs_info *info);
 
 enum btrfs_reserve_flush_enum {
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index a218f2187ebe..40e4f2ce4952 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2545,106 +2545,6 @@ int btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr)
 	return readonly;
 }
 
-/*
- * returns target flags in extended format or 0 if restripe for this
- * chunk_type is not in progress
- *
- * should be called with balance_lock held
- */
-u64 btrfs_get_restripe_target(struct btrfs_fs_info *fs_info, u64 flags)
-{
-	struct btrfs_balance_control *bctl = fs_info->balance_ctl;
-	u64 target = 0;
-
-	if (!bctl)
-		return 0;
-
-	if (flags & BTRFS_BLOCK_GROUP_DATA &&
-	    bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) {
-		target = BTRFS_BLOCK_GROUP_DATA | bctl->data.target;
-	} else if (flags & BTRFS_BLOCK_GROUP_SYSTEM &&
-		   bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
-		target = BTRFS_BLOCK_GROUP_SYSTEM | bctl->sys.target;
-	} else if (flags & BTRFS_BLOCK_GROUP_METADATA &&
-		   bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) {
-		target = BTRFS_BLOCK_GROUP_METADATA | bctl->meta.target;
-	}
-
-	return target;
-}
-
-/*
- * @flags: available profiles in extended format (see ctree.h)
- *
- * Returns reduced profile in chunk format.  If profile changing is in
- * progress (either running or paused) picks the target profile (if it's
- * already available), otherwise falls back to plain reducing.
- */
-static u64 btrfs_reduce_alloc_profile(struct btrfs_fs_info *fs_info, u64 flags)
-{
-	u64 num_devices = fs_info->fs_devices->rw_devices;
-	u64 target;
-	u64 raid_type;
-	u64 allowed = 0;
-
-	/*
-	 * see if restripe for this chunk_type is in progress, if so
-	 * try to reduce to the target profile
-	 */
-	spin_lock(&fs_info->balance_lock);
-	target = btrfs_get_restripe_target(fs_info, flags);
-	if (target) {
-		/* pick target profile only if it's already available */
-		if ((flags & target) & BTRFS_EXTENDED_PROFILE_MASK) {
-			spin_unlock(&fs_info->balance_lock);
-			return extended_to_chunk(target);
-		}
-	}
-	spin_unlock(&fs_info->balance_lock);
-
-	/* First, mask out the RAID levels which aren't possible */
-	for (raid_type = 0; raid_type < BTRFS_NR_RAID_TYPES; raid_type++) {
-		if (num_devices >= btrfs_raid_array[raid_type].devs_min)
-			allowed |= btrfs_raid_array[raid_type].bg_flag;
-	}
-	allowed &= flags;
-
-	if (allowed & BTRFS_BLOCK_GROUP_RAID6)
-		allowed = BTRFS_BLOCK_GROUP_RAID6;
-	else if (allowed & BTRFS_BLOCK_GROUP_RAID5)
-		allowed = BTRFS_BLOCK_GROUP_RAID5;
-	else if (allowed & BTRFS_BLOCK_GROUP_RAID10)
-		allowed = BTRFS_BLOCK_GROUP_RAID10;
-	else if (allowed & BTRFS_BLOCK_GROUP_RAID1)
-		allowed = BTRFS_BLOCK_GROUP_RAID1;
-	else if (allowed & BTRFS_BLOCK_GROUP_RAID0)
-		allowed = BTRFS_BLOCK_GROUP_RAID0;
-
-	flags &= ~BTRFS_BLOCK_GROUP_PROFILE_MASK;
-
-	return extended_to_chunk(flags | allowed);
-}
-
-u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags)
-{
-	unsigned seq;
-	u64 flags;
-
-	do {
-		flags = orig_flags;
-		seq = read_seqbegin(&fs_info->profiles_lock);
-
-		if (flags & BTRFS_BLOCK_GROUP_DATA)
-			flags |= fs_info->avail_data_alloc_bits;
-		else if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
-			flags |= fs_info->avail_system_alloc_bits;
-		else if (flags & BTRFS_BLOCK_GROUP_METADATA)
-			flags |= fs_info->avail_metadata_alloc_bits;
-	} while (read_seqretry(&fs_info->profiles_lock, seq));
-
-	return btrfs_reduce_alloc_profile(fs_info, flags);
-}
-
 static u64 get_alloc_profile_by_root(struct btrfs_root *root, int data)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -2662,21 +2562,6 @@ static u64 get_alloc_profile_by_root(struct btrfs_root *root, int data)
 	return ret;
 }
 
-u64 btrfs_data_alloc_profile(struct btrfs_fs_info *fs_info)
-{
-	return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_DATA);
-}
-
-u64 btrfs_metadata_alloc_profile(struct btrfs_fs_info *fs_info)
-{
-	return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_METADATA);
-}
-
-u64 btrfs_system_alloc_profile(struct btrfs_fs_info *fs_info)
-{
-	return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
-}
-
 static u64 first_logical_byte(struct btrfs_fs_info *fs_info, u64 search_start)
 {
 	struct btrfs_block_group_cache *cache;
-- 
2.21.0

[PATCH 13/15] btrfs: migrate the block group cleanup code

[Josef Bacik]

This can now be easily migrated as well.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.c | 128 +++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/block-group.h |   2 +
 fs/btrfs/ctree.h       |   2 -
 fs/btrfs/extent-tree.c | 128 -----------------------------------------
 4 files changed, 130 insertions(+), 130 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 330f96134ffb..451dfd64ed36 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -3053,3 +3053,131 @@ int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
 
 	return ret;
 }
+
+void btrfs_put_block_group_cache(struct btrfs_fs_info *info)
+{
+	struct btrfs_block_group_cache *block_group;
+	u64 last = 0;
+
+	while (1) {
+		struct inode *inode;
+
+		block_group = btrfs_lookup_first_block_group(info, last);
+		while (block_group) {
+			btrfs_wait_block_group_cache_done(block_group);
+			spin_lock(&block_group->lock);
+			if (block_group->iref)
+				break;
+			spin_unlock(&block_group->lock);
+			block_group = btrfs_next_block_group(block_group);
+		}
+		if (!block_group) {
+			if (last == 0)
+				break;
+			last = 0;
+			continue;
+		}
+
+		inode = block_group->inode;
+		block_group->iref = 0;
+		block_group->inode = NULL;
+		spin_unlock(&block_group->lock);
+		ASSERT(block_group->io_ctl.inode == NULL);
+		iput(inode);
+		last = block_group->key.objectid + block_group->key.offset;
+		btrfs_put_block_group(block_group);
+	}
+}
+
+/*
+ * Must be called only after stopping all workers, since we could have block
+ * group caching kthreads running, and therefore they could race with us if we
+ * freed the block groups before stopping them.
+ */
+int btrfs_free_block_groups(struct btrfs_fs_info *info)
+{
+	struct btrfs_block_group_cache *block_group;
+	struct btrfs_space_info *space_info;
+	struct btrfs_caching_control *caching_ctl;
+	struct rb_node *n;
+
+	down_write(&info->commit_root_sem);
+	while (!list_empty(&info->caching_block_groups)) {
+		caching_ctl = list_entry(info->caching_block_groups.next,
+					 struct btrfs_caching_control, list);
+		list_del(&caching_ctl->list);
+		btrfs_put_caching_control(caching_ctl);
+	}
+	up_write(&info->commit_root_sem);
+
+	spin_lock(&info->unused_bgs_lock);
+	while (!list_empty(&info->unused_bgs)) {
+		block_group = list_first_entry(&info->unused_bgs,
+					       struct btrfs_block_group_cache,
+					       bg_list);
+		list_del_init(&block_group->bg_list);
+		btrfs_put_block_group(block_group);
+	}
+	spin_unlock(&info->unused_bgs_lock);
+
+	spin_lock(&info->block_group_cache_lock);
+	while ((n = rb_last(&info->block_group_cache_tree)) != NULL) {
+		block_group = rb_entry(n, struct btrfs_block_group_cache,
+				       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);
+		list_del(&block_group->list);
+		up_write(&block_group->space_info->groups_sem);
+
+		/*
+		 * We haven't cached this block group, which means we could
+		 * possibly have excluded extents on this block group.
+		 */
+		if (block_group->cached == BTRFS_CACHE_NO ||
+		    block_group->cached == BTRFS_CACHE_ERROR)
+			btrfs_free_excluded_extents(block_group);
+
+		btrfs_remove_free_space_cache(block_group);
+		ASSERT(block_group->cached != BTRFS_CACHE_STARTED);
+		ASSERT(list_empty(&block_group->dirty_list));
+		ASSERT(list_empty(&block_group->io_list));
+		ASSERT(list_empty(&block_group->bg_list));
+		ASSERT(atomic_read(&block_group->count) == 1);
+		btrfs_put_block_group(block_group);
+
+		spin_lock(&info->block_group_cache_lock);
+	}
+	spin_unlock(&info->block_group_cache_lock);
+
+	/* now that all the block groups are freed, go through and
+	 * free all the space_info structs.  This is only called during
+	 * the final stages of unmount, and so we know nobody is
+	 * using them.  We call synchronize_rcu() once before we start,
+	 * just to be on the safe side.
+	 */
+	synchronize_rcu();
+
+	btrfs_release_global_block_rsv(info);
+
+	while (!list_empty(&info->space_info)) {
+		space_info = list_entry(info->space_info.next,
+					struct btrfs_space_info,
+					list);
+
+		/*
+		 * Do not hide this behind enospc_debug, this is actually
+		 * important and indicates a real bug if this happens.
+		 */
+		if (WARN_ON(space_info->bytes_pinned > 0 ||
+			    space_info->bytes_reserved > 0 ||
+			    space_info->bytes_may_use > 0))
+			btrfs_dump_space_info(info, space_info, 0, 0);
+		list_del(&space_info->list);
+		btrfs_sysfs_remove_space_info(space_info);
+	}
+	return 0;
+}
diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index ab5434ddd7f4..2647c0aa76b8 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -224,6 +224,8 @@ int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
 int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, u64 type);
 void check_system_chunk(struct btrfs_trans_handle *trans, const u64 type);
 u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags);
+void btrfs_put_block_group_cache(struct btrfs_fs_info *info);
+int btrfs_free_block_groups(struct btrfs_fs_info *info);
 
 static inline u64 btrfs_data_alloc_profile(struct btrfs_fs_info *fs_info)
 {
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index fab6e9792d23..00c52b4a4dc6 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2522,7 +2522,6 @@ int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 			 struct btrfs_ref *generic_ref);
 
 int btrfs_extent_readonly(struct btrfs_fs_info *fs_info, u64 bytenr);
-int btrfs_free_block_groups(struct btrfs_fs_info *info);
 void btrfs_get_block_group_trimming(struct btrfs_block_group_cache *cache);
 void btrfs_put_block_group_trimming(struct btrfs_block_group_cache *cache);
 void btrfs_clear_space_info_full(struct btrfs_fs_info *info);
@@ -2561,7 +2560,6 @@ void btrfs_delalloc_release_extents(struct btrfs_inode *inode, u64 num_bytes,
 				    bool qgroup_free);
 
 int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes);
-void btrfs_put_block_group_cache(struct btrfs_fs_info *info);
 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
 int btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info,
 				   u64 start, u64 end);
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 40e4f2ce4952..f483345715b1 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -5654,134 +5654,6 @@ u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo)
 	return free_bytes;
 }
 
-void btrfs_put_block_group_cache(struct btrfs_fs_info *info)
-{
-	struct btrfs_block_group_cache *block_group;
-	u64 last = 0;
-
-	while (1) {
-		struct inode *inode;
-
-		block_group = btrfs_lookup_first_block_group(info, last);
-		while (block_group) {
-			btrfs_wait_block_group_cache_done(block_group);
-			spin_lock(&block_group->lock);
-			if (block_group->iref)
-				break;
-			spin_unlock(&block_group->lock);
-			block_group = btrfs_next_block_group(block_group);
-		}
-		if (!block_group) {
-			if (last == 0)
-				break;
-			last = 0;
-			continue;
-		}
-
-		inode = block_group->inode;
-		block_group->iref = 0;
-		block_group->inode = NULL;
-		spin_unlock(&block_group->lock);
-		ASSERT(block_group->io_ctl.inode == NULL);
-		iput(inode);
-		last = block_group->key.objectid + block_group->key.offset;
-		btrfs_put_block_group(block_group);
-	}
-}
-
-/*
- * Must be called only after stopping all workers, since we could have block
- * group caching kthreads running, and therefore they could race with us if we
- * freed the block groups before stopping them.
- */
-int btrfs_free_block_groups(struct btrfs_fs_info *info)
-{
-	struct btrfs_block_group_cache *block_group;
-	struct btrfs_space_info *space_info;
-	struct btrfs_caching_control *caching_ctl;
-	struct rb_node *n;
-
-	down_write(&info->commit_root_sem);
-	while (!list_empty(&info->caching_block_groups)) {
-		caching_ctl = list_entry(info->caching_block_groups.next,
-					 struct btrfs_caching_control, list);
-		list_del(&caching_ctl->list);
-		btrfs_put_caching_control(caching_ctl);
-	}
-	up_write(&info->commit_root_sem);
-
-	spin_lock(&info->unused_bgs_lock);
-	while (!list_empty(&info->unused_bgs)) {
-		block_group = list_first_entry(&info->unused_bgs,
-					       struct btrfs_block_group_cache,
-					       bg_list);
-		list_del_init(&block_group->bg_list);
-		btrfs_put_block_group(block_group);
-	}
-	spin_unlock(&info->unused_bgs_lock);
-
-	spin_lock(&info->block_group_cache_lock);
-	while ((n = rb_last(&info->block_group_cache_tree)) != NULL) {
-		block_group = rb_entry(n, struct btrfs_block_group_cache,
-				       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);
-		list_del(&block_group->list);
-		up_write(&block_group->space_info->groups_sem);
-
-		/*
-		 * We haven't cached this block group, which means we could
-		 * possibly have excluded extents on this block group.
-		 */
-		if (block_group->cached == BTRFS_CACHE_NO ||
-		    block_group->cached == BTRFS_CACHE_ERROR)
-			btrfs_free_excluded_extents(block_group);
-
-		btrfs_remove_free_space_cache(block_group);
-		ASSERT(block_group->cached != BTRFS_CACHE_STARTED);
-		ASSERT(list_empty(&block_group->dirty_list));
-		ASSERT(list_empty(&block_group->io_list));
-		ASSERT(list_empty(&block_group->bg_list));
-		ASSERT(atomic_read(&block_group->count) == 1);
-		btrfs_put_block_group(block_group);
-
-		spin_lock(&info->block_group_cache_lock);
-	}
-	spin_unlock(&info->block_group_cache_lock);
-
-	/* now that all the block groups are freed, go through and
-	 * free all the space_info structs.  This is only called during
-	 * the final stages of unmount, and so we know nobody is
-	 * using them.  We call synchronize_rcu() once before we start,
-	 * just to be on the safe side.
-	 */
-	synchronize_rcu();
-
-	btrfs_release_global_block_rsv(info);
-
-	while (!list_empty(&info->space_info)) {
-		space_info = list_entry(info->space_info.next,
-					struct btrfs_space_info,
-					list);
-
-		/*
-		 * Do not hide this behind enospc_debug, this is actually
-		 * important and indicates a real bug if this happens.
-		 */
-		if (WARN_ON(space_info->bytes_pinned > 0 ||
-			    space_info->bytes_reserved > 0 ||
-			    space_info->bytes_may_use > 0))
-			btrfs_dump_space_info(info, space_info, 0, 0);
-		list_del(&space_info->list);
-		btrfs_sysfs_remove_space_info(space_info);
-	}
-	return 0;
-}
-
 int btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info,
 				   u64 start, u64 end)
 {
-- 
2.21.0

[PATCH 14/15] btrfs: unexport the temporary exported functions

[Josef Bacik]

These were renamed and exported to facilitate logical migration of
different code chunks into block-group.c.  Now that all the users are in
one file go ahead and rename them back, move the code around, and make
them static.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/block-group.c | 176 ++++++++++++++++++++---------------------
 fs/btrfs/block-group.h |   6 --
 2 files changed, 88 insertions(+), 94 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 451dfd64ed36..0df939285702 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -21,7 +21,7 @@
  *
  * should be called with balance_lock held
  */
-u64 btrfs_get_restripe_target(struct btrfs_fs_info *fs_info, u64 flags)
+static u64 get_restripe_target(struct btrfs_fs_info *fs_info, u64 flags)
 {
 	struct btrfs_balance_control *bctl = fs_info->balance_ctl;
 	u64 target = 0;
@@ -62,7 +62,7 @@ static u64 btrfs_reduce_alloc_profile(struct btrfs_fs_info *fs_info, u64 flags)
 	 * try to reduce to the target profile
 	 */
 	spin_lock(&fs_info->balance_lock);
-	target = btrfs_get_restripe_target(fs_info, flags);
+	target = get_restripe_target(fs_info, flags);
 	if (target) {
 		/* pick target profile only if it's already available */
 		if ((flags & target) & BTRFS_EXTENDED_PROFILE_MASK) {
@@ -422,7 +422,7 @@ int btrfs_wait_block_group_cache_done(struct btrfs_block_group_cache *cache)
 }
 
 #ifdef CONFIG_BTRFS_DEBUG
-void btrfs_fragment_free_space(struct btrfs_block_group_cache *block_group)
+static void fragment_free_space(struct btrfs_block_group_cache *block_group)
 {
 	struct btrfs_fs_info *fs_info = block_group->fs_info;
 	u64 start = block_group->key.objectid;
@@ -660,7 +660,7 @@ static noinline void caching_thread(struct btrfs_work *work)
 		block_group->space_info->bytes_used += bytes_used >> 1;
 		spin_unlock(&block_group->lock);
 		spin_unlock(&block_group->space_info->lock);
-		btrfs_fragment_free_space(block_group);
+		fragment_free_space(block_group);
 	}
 #endif
 
@@ -767,7 +767,7 @@ int btrfs_cache_block_group(struct btrfs_block_group_cache *cache,
 			cache->space_info->bytes_used += bytes_used >> 1;
 			spin_unlock(&cache->lock);
 			spin_unlock(&cache->space_info->lock);
-			btrfs_fragment_free_space(cache);
+			fragment_free_space(cache);
 		}
 #endif
 		mutex_unlock(&caching_ctl->mutex);
@@ -1167,6 +1167,81 @@ btrfs_start_trans_remove_block_group(struct btrfs_fs_info *fs_info,
 							   num_items, 1);
 }
 
+/*
+ * Mark block group @cache read-only, so later write won't happen to block
+ * group @cache.
+ *
+ * If @force is not set, this function will only mark the block group readonly
+ * if we have enough free space (1M) in other metadata/system block groups.
+ * If @force is not set, this function will mark the block group readonly
+ * without checking free space.
+ *
+ * NOTE: This function doesn't care if other block groups can contain all the
+ * data in this block group. That check should be done by relocation routine,
+ * not this function.
+ */
+static int inc_block_group_ro(struct btrfs_block_group_cache *cache,
+			      int force)
+{
+	struct btrfs_space_info *sinfo = cache->space_info;
+	u64 num_bytes;
+	u64 sinfo_used;
+	u64 min_allocable_bytes;
+	int ret = -ENOSPC;
+
+	/*
+	 * We need some metadata space and system metadata space for
+	 * allocating chunks in some corner cases until we force to set
+	 * it to be readonly.
+	 */
+	if ((sinfo->flags &
+	     (BTRFS_BLOCK_GROUP_SYSTEM | BTRFS_BLOCK_GROUP_METADATA)) &&
+	    !force)
+		min_allocable_bytes = SZ_1M;
+	else
+		min_allocable_bytes = 0;
+
+	spin_lock(&sinfo->lock);
+	spin_lock(&cache->lock);
+
+	if (cache->ro) {
+		cache->ro++;
+		ret = 0;
+		goto out;
+	}
+
+	num_bytes = cache->key.offset - cache->reserved - cache->pinned -
+		    cache->bytes_super - btrfs_block_group_used(&cache->item);
+	sinfo_used = btrfs_space_info_used(sinfo, true);
+
+	/*
+	 * sinfo_used + num_bytes should always <= sinfo->total_bytes.
+	 *
+	 * Here we make sure if we mark this bg RO, we still have enough
+	 * free space as buffer (if min_allocable_bytes is not 0).
+	 */
+	if (sinfo_used + num_bytes + min_allocable_bytes <=
+	    sinfo->total_bytes) {
+		sinfo->bytes_readonly += num_bytes;
+		cache->ro++;
+		list_add_tail(&cache->ro_list, &sinfo->ro_bgs);
+		ret = 0;
+	}
+out:
+	spin_unlock(&cache->lock);
+	spin_unlock(&sinfo->lock);
+	if (ret == -ENOSPC && btrfs_test_opt(cache->fs_info, ENOSPC_DEBUG)) {
+		btrfs_info(cache->fs_info,
+			"unable to make block group %llu ro",
+			cache->key.objectid);
+		btrfs_info(cache->fs_info,
+			"sinfo_used=%llu bg_num_bytes=%llu min_allocable=%llu",
+			sinfo_used, num_bytes, min_allocable_bytes);
+		btrfs_dump_space_info(cache->fs_info, cache->space_info, 0, 0);
+	}
+	return ret;
+}
+
 /*
  * Process the unused_bgs list and remove any that don't have any allocated
  * space inside of them.
@@ -1222,7 +1297,7 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		spin_unlock(&block_group->lock);
 
 		/* We don't want to force the issue, only flip if it's ok. */
-		ret = __btrfs_inc_block_group_ro(block_group, 0);
+		ret = inc_block_group_ro(block_group, 0);
 		up_write(&space_info->groups_sem);
 		if (ret < 0) {
 			ret = 0;
@@ -1750,7 +1825,7 @@ int btrfs_read_block_groups(struct btrfs_fs_info *info)
 
 		set_avail_alloc_bits(info, cache->flags);
 		if (btrfs_chunk_readonly(info, cache->key.objectid)) {
-			__btrfs_inc_block_group_ro(cache, 1);
+			inc_block_group_ro(cache, 1);
 		} else if (btrfs_block_group_used(&cache->item) == 0) {
 			ASSERT(list_empty(&cache->bg_list));
 			btrfs_mark_bg_unused(cache);
@@ -1771,11 +1846,11 @@ int btrfs_read_block_groups(struct btrfs_fs_info *info)
 		list_for_each_entry(cache,
 				&space_info->block_groups[BTRFS_RAID_RAID0],
 				list)
-			__btrfs_inc_block_group_ro(cache, 1);
+			inc_block_group_ro(cache, 1);
 		list_for_each_entry(cache,
 				&space_info->block_groups[BTRFS_RAID_SINGLE],
 				list)
-			__btrfs_inc_block_group_ro(cache, 1);
+			inc_block_group_ro(cache, 1);
 	}
 
 	btrfs_add_raid_kobjects(info);
@@ -1868,7 +1943,7 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, u64 bytes_used,
 		u64 new_bytes_used = size - bytes_used;
 
 		bytes_used += new_bytes_used >> 1;
-		btrfs_fragment_free_space(cache);
+		fragment_free_space(cache);
 	}
 #endif
 	/*
@@ -1914,7 +1989,7 @@ static u64 update_block_group_flags(struct btrfs_fs_info *fs_info, u64 flags)
 	 * if restripe for this chunk_type is on pick target profile and
 	 * return, otherwise do the usual balance
 	 */
-	stripped = btrfs_get_restripe_target(fs_info, flags);
+	stripped = get_restripe_target(fs_info, flags);
 	if (stripped)
 		return extended_to_chunk(stripped);
 
@@ -1953,81 +2028,6 @@ static u64 update_block_group_flags(struct btrfs_fs_info *fs_info, u64 flags)
 	return flags;
 }
 
-/*
- * Mark block group @cache read-only, so later write won't happen to block
- * group @cache.
- *
- * If @force is not set, this function will only mark the block group readonly
- * if we have enough free space (1M) in other metadata/system block groups.
- * If @force is not set, this function will mark the block group readonly
- * without checking free space.
- *
- * NOTE: This function doesn't care if other block groups can contain all the
- * data in this block group. That check should be done by relocation routine,
- * not this function.
- */
-int __btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache,
-			       int force)
-{
-	struct btrfs_space_info *sinfo = cache->space_info;
-	u64 num_bytes;
-	u64 sinfo_used;
-	u64 min_allocable_bytes;
-	int ret = -ENOSPC;
-
-	/*
-	 * We need some metadata space and system metadata space for
-	 * allocating chunks in some corner cases until we force to set
-	 * it to be readonly.
-	 */
-	if ((sinfo->flags &
-	     (BTRFS_BLOCK_GROUP_SYSTEM | BTRFS_BLOCK_GROUP_METADATA)) &&
-	    !force)
-		min_allocable_bytes = SZ_1M;
-	else
-		min_allocable_bytes = 0;
-
-	spin_lock(&sinfo->lock);
-	spin_lock(&cache->lock);
-
-	if (cache->ro) {
-		cache->ro++;
-		ret = 0;
-		goto out;
-	}
-
-	num_bytes = cache->key.offset - cache->reserved - cache->pinned -
-		    cache->bytes_super - btrfs_block_group_used(&cache->item);
-	sinfo_used = btrfs_space_info_used(sinfo, true);
-
-	/*
-	 * sinfo_used + num_bytes should always <= sinfo->total_bytes.
-	 *
-	 * Here we make sure if we mark this bg RO, we still have enough
-	 * free space as buffer (if min_allocable_bytes is not 0).
-	 */
-	if (sinfo_used + num_bytes + min_allocable_bytes <=
-	    sinfo->total_bytes) {
-		sinfo->bytes_readonly += num_bytes;
-		cache->ro++;
-		list_add_tail(&cache->ro_list, &sinfo->ro_bgs);
-		ret = 0;
-	}
-out:
-	spin_unlock(&cache->lock);
-	spin_unlock(&sinfo->lock);
-	if (ret == -ENOSPC && btrfs_test_opt(cache->fs_info, ENOSPC_DEBUG)) {
-		btrfs_info(cache->fs_info,
-			"unable to make block group %llu ro",
-			cache->key.objectid);
-		btrfs_info(cache->fs_info,
-			"sinfo_used=%llu bg_num_bytes=%llu min_allocable=%llu",
-			sinfo_used, num_bytes, min_allocable_bytes);
-		btrfs_dump_space_info(cache->fs_info, cache->space_info, 0, 0);
-	}
-	return ret;
-}
-
 int btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache)
 
 {
@@ -2077,14 +2077,14 @@ int btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache)
 			goto out;
 	}
 
-	ret = __btrfs_inc_block_group_ro(cache, 0);
+	ret = inc_block_group_ro(cache, 0);
 	if (!ret)
 		goto out;
 	alloc_flags = btrfs_get_alloc_profile(fs_info, cache->space_info->flags);
 	ret = btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
 	if (ret < 0)
 		goto out;
-	ret = __btrfs_inc_block_group_ro(cache, 0);
+	ret = inc_block_group_ro(cache, 0);
 out:
 	if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM) {
 		alloc_flags = update_block_group_flags(fs_info, cache->flags);
diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index 2647c0aa76b8..6f4658b1ec94 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -169,7 +169,6 @@ static inline int btrfs_should_fragment_free_space(
 	       (btrfs_test_opt(fs_info, FRAGMENT_DATA) &&
 		block_group->flags &  BTRFS_BLOCK_GROUP_DATA);
 }
-void btrfs_fragment_free_space(struct btrfs_block_group_cache *block_group);
 #endif
 
 struct btrfs_block_group_cache *btrfs_lookup_first_block_group(
@@ -249,9 +248,4 @@ static inline int btrfs_block_group_cache_done(
 	return cache->cached == BTRFS_CACHE_FINISHED ||
 		cache->cached == BTRFS_CACHE_ERROR;
 }
-
-int __btrfs_inc_block_group_ro(struct btrfs_block_group_cache *cache,
-			       int force);
-u64 btrfs_get_restripe_target(struct btrfs_fs_info *fs_info, u64 flags);
-
 #endif /* BTRFS_BLOCK_GROUP_H */
-- 
2.21.0

[PATCH 15/15] btrfs: remove comment and leftover cruft

[Josef Bacik]

Commit "btrfs: convert snapshot/nocow exlcusion to drw lock" removed
this code, but didn't remove the comment or the definitions, do that
now.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
---
 fs/btrfs/ctree.h       | 3 ---
 fs/btrfs/extent-tree.c | 9 ---------
 2 files changed, 12 deletions(-)

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 00c52b4a4dc6..a247237b8804 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2570,9 +2570,6 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range);
 int btrfs_init_space_info(struct btrfs_fs_info *fs_info);
 int btrfs_delayed_refs_qgroup_accounting(struct btrfs_trans_handle *trans,
 					 struct btrfs_fs_info *fs_info);
-int btrfs_start_write_no_snapshotting(struct btrfs_root *root);
-void btrfs_end_write_no_snapshotting(struct btrfs_root *root);
-void btrfs_wait_for_snapshot_creation(struct btrfs_root *root);
 
 /* ctree.c */
 int btrfs_bin_search(struct extent_buffer *eb, const struct btrfs_key *key,
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index f483345715b1..b6bdbaaaaffd 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -5849,12 +5849,3 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 		return bg_ret;
 	return dev_ret;
 }
-
-/*
- * btrfs_{start,end}_write_no_snapshotting() are similar to
- * mnt_{want,drop}_write(), they are used to prevent some tasks from writing
- * data into the page cache through nocow before the subvolume is snapshoted,
- * but flush the data into disk after the snapshot creation, or to prevent
- * operations while snapshotting is ongoing and that cause the snapshot to be
- * inconsistent (writes followed by expanding truncates for example).
- */
-- 
2.21.0

Re: [PATCH 15/15] btrfs: remove comment and leftover cruft

[David Sterba]

On Tue, Aug 06, 2019 at 12:28:37PM -0400, Josef Bacik wrote:
> Commit "btrfs: convert snapshot/nocow exlcusion to drw lock" removed
> this code, but didn't remove the comment or the definitions, do that
> now.

The DRW lock patchset is in for-next just as a preview and for early
testing so any fixups should be folded to the original patchset.

Re: [PATCH 00/15] Migrate the block group code into it's own file

[David Sterba]

On Tue, Aug 06, 2019 at 12:28:22PM -0400, Josef Bacik wrote:
> This is the rebased set of the much larger group of patches I sent last month.
> The first 10 patches are already merged, these just didn't apply cleanly.  I
> went through and applied each one, deleted and re-copied anything that didn't
> merge cleanly, and compiled between each patch to make sure everything was
> kosher.  This series just moves code around with the goal of making
> extent-tree.c smaller.  I made no other changes other than moving code around, I
> want to keep cleanups a separate thing _after_ we move the code around.  Thanks,

I had the remaining part of the previous patchset ready so I only
checked that this one is the same. The export of btrfs_get_alloc_profile
is in the patch "btrfs: migrate the block group read/creation code" and
the 15/15 is from unrelated patchset.

There were again conflicts with patches merged recently (sysfs and
device stats), I made sure the changes are not lost but this again
proves that large code moves need to be scheduled to the end of devel
cycle.