Merge tag 'for-4.17-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux

Pull btrfs updates from David Sterba:
 "There are a several user visible changes, the rest is mostly invisible
  and continues to clean up the whole code base.

  User visible changes:
   - new mount option nossd_spread (pair for ssd_spread)

   - mount option subvolid will detect junk after the number and fail
     the mount

   - add message after cancelled device replace

   - direct module dependency on libcrc32, removed own crc wrappers

   - removed user space transaction ioctls

   - use lighter locking when reading /proc/self/mounts, RCU instead of
     mutex to avoid unnecessary contention

  Enhancements:
   - skip writeback of last page when truncating file to same size

   - send: do not issue unnecessary truncate operations

   - mount option token specifiers: use %u for unsigned values, more
     validation

   - selftests: more tree block validations

  qgroups:
   - preparatory work for splitting reservation types for data and
     metadata, this should allow for more accurate tracking and fix some
     issues with underflows or do further enhancements

   - split metadata reservations for started and joined transaction so
     they do not get mixed up and are accounted correctly at commit time

   - with the above, it's possible to revert patch that potentially
     deadlocks when trying to make more space by explicitly committing
     when the quota limit is hit

   - fix root item corruption when multiple same source snapshots are
     created with quota enabled

  RAID56:
   - make sure target is identical to source when raid56 rebuild fails
     after dev-replace

   - faster rebuild during scrub, batch by stripes and not
     block-by-block

   - make more use of cached data when rebuilding from a missing device

  Fixes:
   - null pointer deref when device replace target is missing

   - fix fsync after hole punching when using no-holes feature

   - fix lockdep splat when allocating percpu data with wrong GFP flags

  Cleanups, refactoring, core changes:
   - drop redunant parameters from various functions

   - kill and opencode trivial helpers

   - __cold/__exit function annotations

   - dead code removal

   - continued audit and documentation of memory barriers

   - error handling: handle removal from uuid tree

   - error handling: remove handling of impossible condtitons

   - more debugging or error messages

   - updated tracepoints

   - one VLA use removal (and one still left)"

* tag 'for-4.17-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux: (164 commits)
  btrfs: lift errors from add_extent_changeset to the callers
  Btrfs: print error messages when failing to read trees
  btrfs: user proper type for btrfs_mask_flags flags
  btrfs: split dev-replace locking helpers for read and write
  btrfs: remove stale comments about fs_mutex
  btrfs: use RCU in btrfs_show_devname for device list traversal
  btrfs: update barrier in should_cow_block
  btrfs: use lockdep_assert_held for mutexes
  btrfs: use lockdep_assert_held for spinlocks
  btrfs: Validate child tree block's level and first key
  btrfs: tests/qgroup: Fix wrong tree backref level
  Btrfs: fix copy_items() return value when logging an inode
  Btrfs: fix fsync after hole punching when using no-holes feature
  btrfs: use helper to set ulist aux from a qgroup
  Revert "btrfs: qgroups: Retry after commit on getting EDQUOT"
  btrfs: qgroup: Update trace events for metadata reservation
  btrfs: qgroup: Use root::qgroup_meta_rsv_* to record qgroup meta reserved space
  btrfs: delayed-inode: Use new qgroup meta rsv for delayed inode and item
  btrfs: qgroup: Use separate meta reservation type for delalloc
  btrfs: qgroup: Introduce function to convert META_PREALLOC into META_PERTRANS
  ...

1 files changed, 173 insertions, 144 deletions

diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index e0460d7b5622..e08d0d45af4f 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -27,7 +27,7 @@
 #include <linux/ratelimit.h>
 #include <linux/percpu_counter.h>
 #include <linux/lockdep.h>
-#include "hash.h"
+#include <linux/crc32c.h>
 #include "tree-log.h"
 #include "disk-io.h"
 #include "print-tree.h"
@@ -535,13 +535,11 @@ 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;
-	struct btrfs_root *extent_root;
 	int ret;
 
 	caching_ctl = container_of(work, struct btrfs_caching_control, work);
 	block_group = caching_ctl->block_group;
 	fs_info = block_group->fs_info;
-	extent_root = fs_info->extent_root;
 
 	mutex_lock(&caching_ctl->mutex);
 	down_read(&fs_info->commit_root_sem);
@@ -1203,11 +1201,11 @@ static u64 hash_extent_data_ref(u64 root_objectid, u64 owner, u64 offset)
 	__le64 lenum;
 
 	lenum = cpu_to_le64(root_objectid);
-	high_crc = btrfs_crc32c(high_crc, &lenum, sizeof(lenum));
+	high_crc = crc32c(high_crc, &lenum, sizeof(lenum));
 	lenum = cpu_to_le64(owner);
-	low_crc = btrfs_crc32c(low_crc, &lenum, sizeof(lenum));
+	low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
 	lenum = cpu_to_le64(offset);
-	low_crc = btrfs_crc32c(low_crc, &lenum, sizeof(lenum));
+	low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
 
 	return ((u64)high_crc << 31) ^ (u64)low_crc;
 }
@@ -2652,9 +2650,9 @@ static int cleanup_ref_head(struct btrfs_trans_handle *trans,
  * Returns -ENOMEM or -EIO on failure and will abort the transaction.
  */
 static noinline int __btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
-					     struct btrfs_fs_info *fs_info,
 					     unsigned long nr)
 {
+	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_delayed_ref_root *delayed_refs;
 	struct btrfs_delayed_ref_node *ref;
 	struct btrfs_delayed_ref_head *locked_ref = NULL;
@@ -2994,7 +2992,7 @@ static void delayed_ref_async_start(struct btrfs_work *work)
 	if (trans->transid > async->transid)
 		goto end;
 
-	ret = btrfs_run_delayed_refs(trans, fs_info, async->count);
+	ret = btrfs_run_delayed_refs(trans, async->count);
 	if (ret)
 		async->error = ret;
 end:
@@ -3053,8 +3051,9 @@ int btrfs_async_run_delayed_refs(struct btrfs_fs_info *fs_info,
  * Returns <0 on error and aborts the transaction
  */
 int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
-			   struct btrfs_fs_info *fs_info, unsigned long count)
+			   unsigned long count)
 {
+	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct rb_node *node;
 	struct btrfs_delayed_ref_root *delayed_refs;
 	struct btrfs_delayed_ref_head *head;
@@ -3078,7 +3077,7 @@ again:
 	delayed_refs->run_delayed_start = find_middle(&delayed_refs->root);
 #endif
 	trans->can_flush_pending_bgs = false;
-	ret = __btrfs_run_delayed_refs(trans, fs_info, count);
+	ret = __btrfs_run_delayed_refs(trans, count);
 	if (ret < 0) {
 		btrfs_abort_transaction(trans, ret);
 		return ret;
@@ -3086,7 +3085,7 @@ again:
 
 	if (run_all) {
 		if (!list_empty(&trans->new_bgs))
-			btrfs_create_pending_block_groups(trans, fs_info);
+			btrfs_create_pending_block_groups(trans);
 
 		spin_lock(&delayed_refs->lock);
 		node = rb_first(&delayed_refs->href_root);
@@ -3660,9 +3659,9 @@ int btrfs_setup_space_cache(struct btrfs_trans_handle *trans,
  * 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)
+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;
@@ -3686,7 +3685,7 @@ again:
 	 * make sure all the block groups on our dirty list actually
 	 * exist
 	 */
-	btrfs_create_pending_block_groups(trans, fs_info);
+	btrfs_create_pending_block_groups(trans);
 
 	if (!path) {
 		path = btrfs_alloc_path();
@@ -3741,8 +3740,9 @@ again:
 				should_put = 0;
 
 				/*
-				 * the cache_write_mutex is protecting
-				 * the io_list
+				 * 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 {
@@ -3800,7 +3800,7 @@ again:
 	 * 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, fs_info, 0);
+	ret = btrfs_run_delayed_refs(trans, 0);
 	if (!ret && loops == 0) {
 		loops++;
 		spin_lock(&cur_trans->dirty_bgs_lock);
@@ -3882,7 +3882,7 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
 		cache_save_setup(cache, trans, path);
 
 		if (!ret)
-			ret = btrfs_run_delayed_refs(trans, fs_info,
+			ret = btrfs_run_delayed_refs(trans,
 						     (unsigned long) -1);
 
 		if (!ret && cache->disk_cache_state == BTRFS_DC_SETUP) {
@@ -3934,6 +3934,10 @@ int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
 	}
 	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);
@@ -4332,8 +4336,7 @@ again:
 
 		/* commit the current transaction and try again */
 commit_trans:
-		if (need_commit &&
-		    !atomic_read(&fs_info->open_ioctl_trans)) {
+		if (need_commit) {
 			need_commit--;
 
 			if (need_commit > 0) {
@@ -4541,7 +4544,7 @@ void check_system_chunk(struct btrfs_trans_handle *trans,
 	 * Needed because we can end up allocating a system chunk and for an
 	 * atomic and race free space reservation in the chunk block reserve.
 	 */
-	ASSERT(mutex_is_locked(&fs_info->chunk_mutex));
+	lockdep_assert_held(&fs_info->chunk_mutex);
 
 	info = __find_space_info(fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
 	spin_lock(&info->lock);
@@ -4602,11 +4605,7 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 		return -ENOSPC;
 
 	space_info = __find_space_info(fs_info, flags);
-	if (!space_info) {
-		ret = create_space_info(fs_info, flags, &space_info);
-		if (ret)
-			return ret;
-	}
+	ASSERT(space_info);
 
 again:
 	spin_lock(&space_info->lock);
@@ -4705,7 +4704,7 @@ out:
 	 */
 	if (trans->can_flush_pending_bgs &&
 	    trans->chunk_bytes_reserved >= (u64)SZ_2M) {
-		btrfs_create_pending_block_groups(trans, fs_info);
+		btrfs_create_pending_block_groups(trans);
 		btrfs_trans_release_chunk_metadata(trans);
 	}
 	return ret;
@@ -4826,7 +4825,6 @@ static void shrink_delalloc(struct btrfs_fs_info *fs_info, u64 to_reclaim,
 	long time_left;
 	unsigned long nr_pages;
 	int loops;
-	enum btrfs_reserve_flush_enum flush;
 
 	/* Calc the number of the pages we need flush for space reservation */
 	items = calc_reclaim_items_nr(fs_info, to_reclaim);
@@ -4867,10 +4865,6 @@ static void shrink_delalloc(struct btrfs_fs_info *fs_info, u64 to_reclaim,
 			   atomic_read(&fs_info->async_delalloc_pages) <=
 			   (int)max_reclaim);
 skip_async:
-		if (!trans)
-			flush = BTRFS_RESERVE_FLUSH_ALL;
-		else
-			flush = BTRFS_RESERVE_NO_FLUSH;
 		spin_lock(&space_info->lock);
 		if (list_empty(&space_info->tickets) &&
 		    list_empty(&space_info->priority_tickets)) {
@@ -4993,7 +4987,7 @@ static void flush_space(struct btrfs_fs_info *fs_info,
 			ret = PTR_ERR(trans);
 			break;
 		}
-		ret = btrfs_run_delayed_items_nr(trans, fs_info, nr);
+		ret = btrfs_run_delayed_items_nr(trans, nr);
 		btrfs_end_transaction(trans);
 		break;
 	case FLUSH_DELALLOC:
@@ -5388,10 +5382,15 @@ static int reserve_metadata_bytes(struct btrfs_root *root,
 		    !block_rsv_use_bytes(global_rsv, orig_bytes))
 			ret = 0;
 	}
-	if (ret == -ENOSPC)
+	if (ret == -ENOSPC) {
 		trace_btrfs_space_reservation(fs_info, "space_info:enospc",
 					      block_rsv->space_info->flags,
 					      orig_bytes, 1);
+
+		if (btrfs_test_opt(fs_info, ENOSPC_DEBUG))
+			dump_space_info(fs_info, block_rsv->space_info,
+					orig_bytes, 0);
+	}
 	return ret;
 }
 
@@ -5760,6 +5759,9 @@ static int btrfs_inode_rsv_refill(struct btrfs_inode *inode,
 	if (num_bytes == 0)
 		return 0;
 
+	ret = btrfs_qgroup_reserve_meta_prealloc(root, num_bytes, true);
+	if (ret)
+		return ret;
 	ret = reserve_metadata_bytes(root, block_rsv, num_bytes, flush);
 	if (!ret) {
 		block_rsv_add_bytes(block_rsv, num_bytes, 0);
@@ -5772,11 +5774,15 @@ static int btrfs_inode_rsv_refill(struct btrfs_inode *inode,
 /**
  * btrfs_inode_rsv_release - release any excessive reservation.
  * @inode - the inode we need to release from.
+ * @qgroup_free - free or convert qgroup meta.
+ *   Unlike normal operation, qgroup meta reservation needs to know if we are
+ *   freeing qgroup reservation or just converting it into per-trans.  Normally
+ *   @qgroup_free is true for error handling, and false for normal release.
  *
  * This is the same as btrfs_block_rsv_release, except that it handles the
  * tracepoint for the reservation.
  */
-static void btrfs_inode_rsv_release(struct btrfs_inode *inode)
+static void btrfs_inode_rsv_release(struct btrfs_inode *inode, bool qgroup_free)
 {
 	struct btrfs_fs_info *fs_info = inode->root->fs_info;
 	struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
@@ -5792,6 +5798,10 @@ static void btrfs_inode_rsv_release(struct btrfs_inode *inode)
 	if (released > 0)
 		trace_btrfs_space_reservation(fs_info, "delalloc",
 					      btrfs_ino(inode), released, 0);
+	if (qgroup_free)
+		btrfs_qgroup_free_meta_prealloc(inode->root, released);
+	else
+		btrfs_qgroup_convert_reserved_meta(inode->root, released);
 }
 
 void btrfs_block_rsv_release(struct btrfs_fs_info *fs_info,
@@ -5892,24 +5902,6 @@ static void release_global_block_rsv(struct btrfs_fs_info *fs_info)
 	WARN_ON(fs_info->delayed_block_rsv.reserved > 0);
 }
 
-void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
-				  struct btrfs_fs_info *fs_info)
-{
-	if (!trans->block_rsv) {
-		ASSERT(!trans->bytes_reserved);
-		return;
-	}
-
-	if (!trans->bytes_reserved)
-		return;
-
-	ASSERT(trans->block_rsv == &fs_info->trans_block_rsv);
-	trace_btrfs_space_reservation(fs_info, "transaction",
-				      trans->transid, trans->bytes_reserved, 0);
-	btrfs_block_rsv_release(fs_info, trans->block_rsv,
-				trans->bytes_reserved);
-	trans->bytes_reserved = 0;
-}
 
 /*
  * To be called after all the new block groups attached to the transaction
@@ -5951,7 +5943,7 @@ int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans,
 	 */
 	u64 num_bytes = btrfs_calc_trans_metadata_size(fs_info, 1);
 
-	trace_btrfs_space_reservation(fs_info, "orphan", btrfs_ino(inode), 
+	trace_btrfs_space_reservation(fs_info, "orphan", btrfs_ino(inode),
 			num_bytes, 1);
 	return btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes, 1);
 }
@@ -5995,7 +5987,7 @@ int btrfs_subvolume_reserve_metadata(struct btrfs_root *root,
 	if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags)) {
 		/* One for parent inode, two for dir entries */
 		num_bytes = 3 * fs_info->nodesize;
-		ret = btrfs_qgroup_reserve_meta(root, num_bytes, true);
+		ret = btrfs_qgroup_reserve_meta_prealloc(root, num_bytes, true);
 		if (ret)
 			return ret;
 	} else {
@@ -6014,7 +6006,7 @@ int btrfs_subvolume_reserve_metadata(struct btrfs_root *root,
 		ret = btrfs_block_rsv_migrate(global_rsv, rsv, num_bytes, 1);
 
 	if (ret && *qgroup_reserved)
-		btrfs_qgroup_free_meta(root, *qgroup_reserved);
+		btrfs_qgroup_free_meta_prealloc(root, *qgroup_reserved);
 
 	return ret;
 }
@@ -6051,7 +6043,6 @@ static void btrfs_calculate_inode_block_rsv_size(struct btrfs_fs_info *fs_info,
 int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
-	struct btrfs_root *root = inode->root;
 	unsigned nr_extents;
 	enum btrfs_reserve_flush_enum flush = BTRFS_RESERVE_FLUSH_ALL;
 	int ret = 0;
@@ -6068,13 +6059,13 @@ int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
 	if (btrfs_is_free_space_inode(inode)) {
 		flush = BTRFS_RESERVE_NO_FLUSH;
 		delalloc_lock = false;
-	} else if (current->journal_info) {
-		flush = BTRFS_RESERVE_FLUSH_LIMIT;
-	}
+	} else {
+		if (current->journal_info)
+			flush = BTRFS_RESERVE_FLUSH_LIMIT;
 
-	if (flush != BTRFS_RESERVE_NO_FLUSH &&
-	    btrfs_transaction_in_commit(fs_info))
-		schedule_timeout(1);
+		if (btrfs_transaction_in_commit(fs_info))
+			schedule_timeout(1);
+	}
 
 	if (delalloc_lock)
 		mutex_lock(&inode->delalloc_mutex);
@@ -6089,19 +6080,9 @@ int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
 	btrfs_calculate_inode_block_rsv_size(fs_info, inode);
 	spin_unlock(&inode->lock);
 
-	if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags)) {
-		ret = btrfs_qgroup_reserve_meta(root,
-				nr_extents * fs_info->nodesize, true);
-		if (ret)
-			goto out_fail;
-	}
-
 	ret = btrfs_inode_rsv_refill(inode, flush);
-	if (unlikely(ret)) {
-		btrfs_qgroup_free_meta(root,
-				       nr_extents * fs_info->nodesize);
+	if (unlikely(ret))
 		goto out_fail;
-	}
 
 	if (delalloc_lock)
 		mutex_unlock(&inode->delalloc_mutex);
@@ -6115,7 +6096,7 @@ out_fail:
 	btrfs_calculate_inode_block_rsv_size(fs_info, inode);
 	spin_unlock(&inode->lock);
 
-	btrfs_inode_rsv_release(inode);
+	btrfs_inode_rsv_release(inode, true);
 	if (delalloc_lock)
 		mutex_unlock(&inode->delalloc_mutex);
 	return ret;
@@ -6125,12 +6106,14 @@ out_fail:
  * btrfs_delalloc_release_metadata - release a metadata reservation for an inode
  * @inode: the inode to release the reservation for.
  * @num_bytes: the number of bytes we are releasing.
+ * @qgroup_free: free qgroup reservation or convert it to per-trans reservation
  *
  * This will release the metadata reservation for an inode.  This can be called
  * once we complete IO for a given set of bytes to release their metadata
  * reservations, or on error for the same reason.
  */
-void btrfs_delalloc_release_metadata(struct btrfs_inode *inode, u64 num_bytes)
+void btrfs_delalloc_release_metadata(struct btrfs_inode *inode, u64 num_bytes,
+				     bool qgroup_free)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
 
@@ -6143,13 +6126,14 @@ void btrfs_delalloc_release_metadata(struct btrfs_inode *inode, u64 num_bytes)
 	if (btrfs_is_testing(fs_info))
 		return;
 
-	btrfs_inode_rsv_release(inode);
+	btrfs_inode_rsv_release(inode, qgroup_free);
 }
 
 /**
  * btrfs_delalloc_release_extents - release our outstanding_extents
  * @inode: the inode to balance the reservation for.
  * @num_bytes: the number of bytes we originally reserved with
+ * @qgroup_free: do we need to free qgroup meta reservation or convert them.
  *
  * When we reserve space we increase outstanding_extents for the extents we may
  * add.  Once we've set the range as delalloc or created our ordered extents we
@@ -6157,7 +6141,8 @@ void btrfs_delalloc_release_metadata(struct btrfs_inode *inode, u64 num_bytes)
  * temporarily tracked outstanding_extents.  This _must_ be used in conjunction
  * with btrfs_delalloc_reserve_metadata.
  */
-void btrfs_delalloc_release_extents(struct btrfs_inode *inode, u64 num_bytes)
+void btrfs_delalloc_release_extents(struct btrfs_inode *inode, u64 num_bytes,
+				    bool qgroup_free)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
 	unsigned num_extents;
@@ -6171,7 +6156,7 @@ void btrfs_delalloc_release_extents(struct btrfs_inode *inode, u64 num_bytes)
 	if (btrfs_is_testing(fs_info))
 		return;
 
-	btrfs_inode_rsv_release(inode);
+	btrfs_inode_rsv_release(inode, qgroup_free);
 }
 
 /**
@@ -6227,9 +6212,9 @@ int btrfs_delalloc_reserve_space(struct inode *inode,
  */
 void btrfs_delalloc_release_space(struct inode *inode,
 				  struct extent_changeset *reserved,
-				  u64 start, u64 len)
+				  u64 start, u64 len, bool qgroup_free)
 {
-	btrfs_delalloc_release_metadata(BTRFS_I(inode), len);
+	btrfs_delalloc_release_metadata(BTRFS_I(inode), len, qgroup_free);
 	btrfs_free_reserved_data_space(inode, reserved, start, len);
 }
 
@@ -6783,9 +6768,9 @@ static int unpin_extent_range(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
-int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
-			       struct btrfs_fs_info *fs_info)
+int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans)
 {
+	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_block_group_cache *block_group, *tmp;
 	struct list_head *deleted_bgs;
 	struct extent_io_tree *unpin;
@@ -7351,29 +7336,6 @@ wait_block_group_cache_done(struct btrfs_block_group_cache *cache)
 	return ret;
 }
 
-int __get_raid_index(u64 flags)
-{
-	if (flags & BTRFS_BLOCK_GROUP_RAID10)
-		return BTRFS_RAID_RAID10;
-	else if (flags & BTRFS_BLOCK_GROUP_RAID1)
-		return BTRFS_RAID_RAID1;
-	else if (flags & BTRFS_BLOCK_GROUP_DUP)
-		return BTRFS_RAID_DUP;
-	else if (flags & BTRFS_BLOCK_GROUP_RAID0)
-		return BTRFS_RAID_RAID0;
-	else if (flags & BTRFS_BLOCK_GROUP_RAID5)
-		return BTRFS_RAID_RAID5;
-	else if (flags & BTRFS_BLOCK_GROUP_RAID6)
-		return BTRFS_RAID_RAID6;
-
-	return BTRFS_RAID_SINGLE; /* BTRFS_BLOCK_GROUP_SINGLE */
-}
-
-int get_block_group_index(struct btrfs_block_group_cache *cache)
-{
-	return __get_raid_index(cache->flags);
-}
-
 static const char *btrfs_raid_type_names[BTRFS_NR_RAID_TYPES] = {
 	[BTRFS_RAID_RAID10]	= "raid10",
 	[BTRFS_RAID_RAID1]	= "raid1",
@@ -7488,7 +7450,7 @@ static noinline int find_free_extent(struct btrfs_fs_info *fs_info,
 	u64 empty_cluster = 0;
 	struct btrfs_space_info *space_info;
 	int loop = 0;
-	int index = __get_raid_index(flags);
+	int index = btrfs_bg_flags_to_raid_index(flags);
 	bool failed_cluster_refill = false;
 	bool failed_alloc = false;
 	bool use_cluster = true;
@@ -7574,7 +7536,8 @@ static noinline int find_free_extent(struct btrfs_fs_info *fs_info,
 				btrfs_put_block_group(block_group);
 				up_read(&space_info->groups_sem);
 			} else {
-				index = get_block_group_index(block_group);
+				index = btrfs_bg_flags_to_raid_index(
+						block_group->flags);
 				btrfs_lock_block_group(block_group, delalloc);
 				goto have_block_group;
 			}
@@ -7584,7 +7547,7 @@ static noinline int find_free_extent(struct btrfs_fs_info *fs_info,
 	}
 search:
 	have_caching_bg = false;
-	if (index == 0 || index == __get_raid_index(flags))
+	if (index == 0 || index == btrfs_bg_flags_to_raid_index(flags))
 		full_search = true;
 	down_read(&space_info->groups_sem);
 	list_for_each_entry(block_group, &space_info->block_groups[index],
@@ -7842,7 +7805,8 @@ checks:
 loop:
 		failed_cluster_refill = false;
 		failed_alloc = false;
-		BUG_ON(index != get_block_group_index(block_group));
+		BUG_ON(btrfs_bg_flags_to_raid_index(block_group->flags) !=
+		       index);
 		btrfs_release_block_group(block_group, delalloc);
 		cond_resched();
 	}
@@ -7996,6 +7960,51 @@ again:
 	up_read(&info->groups_sem);
 }
 
+/*
+ * btrfs_reserve_extent - entry point to the extent allocator. Tries to find a
+ *			  hole that is at least as big as @num_bytes.
+ *
+ * @root           -	The root that will contain this extent
+ *
+ * @ram_bytes      -	The amount of space in ram that @num_bytes take. This
+ *			is used for accounting purposes. This value differs
+ *			from @num_bytes only in the case of compressed extents.
+ *
+ * @num_bytes      -	Number of bytes to allocate on-disk.
+ *
+ * @min_alloc_size -	Indicates the minimum amount of space that the
+ *			allocator should try to satisfy. In some cases
+ *			@num_bytes may be larger than what is required and if
+ *			the filesystem is fragmented then allocation fails.
+ *			However, the presence of @min_alloc_size gives a
+ *			chance to try and satisfy the smaller allocation.
+ *
+ * @empty_size     -	A hint that you plan on doing more COW. This is the
+ *			size in bytes the allocator should try to find free
+ *			next to the block it returns.  This is just a hint and
+ *			may be ignored by the allocator.
+ *
+ * @hint_byte      -	Hint to the allocator to start searching above the byte
+ *			address passed. It might be ignored.
+ *
+ * @ins            -	This key is modified to record the found hole. It will
+ *			have the following values:
+ *			ins->objectid == start position
+ *			ins->flags = BTRFS_EXTENT_ITEM_KEY
+ *			ins->offset == the size of the hole.
+ *
+ * @is_data        -	Boolean flag indicating whether an extent is
+ *			allocated for data (true) or metadata (false)
+ *
+ * @delalloc       -	Boolean flag indicating whether this allocation is for
+ *			delalloc or not. If 'true' data_rwsem of block groups
+ *			is going to be acquired.
+ *
+ *
+ * Returns 0 when an allocation succeeded or < 0 when an error occurred. In
+ * case -ENOSPC is returned then @ins->offset will contain the size of the
+ * largest available hole the allocator managed to find.
+ */
 int btrfs_reserve_extent(struct btrfs_root *root, u64 ram_bytes,
 			 u64 num_bytes, u64 min_alloc_size,
 			 u64 empty_size, u64 hint_byte,
@@ -8699,6 +8708,7 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans,
 	u64 parent;
 	u32 blocksize;
 	struct btrfs_key key;
+	struct btrfs_key first_key;
 	struct extent_buffer *next;
 	int level = wc->level;
 	int reada = 0;
@@ -8719,6 +8729,8 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans,
 	}
 
 	bytenr = btrfs_node_blockptr(path->nodes[level], path->slots[level]);
+	btrfs_node_key_to_cpu(path->nodes[level], &first_key,
+			      path->slots[level]);
 	blocksize = fs_info->nodesize;
 
 	next = find_extent_buffer(fs_info, bytenr);
@@ -8783,7 +8795,8 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans,
 	if (!next) {
 		if (reada && level == 1)
 			reada_walk_down(trans, root, wc, path);
-		next = read_tree_block(fs_info, bytenr, generation);
+		next = read_tree_block(fs_info, bytenr, generation, level - 1,
+				       &first_key);
 		if (IS_ERR(next)) {
 			return PTR_ERR(next);
 		} else if (!extent_buffer_uptodate(next)) {
@@ -9648,7 +9661,7 @@ int btrfs_can_relocate(struct btrfs_fs_info *fs_info, u64 bytenr)
 	 */
 	target = get_restripe_target(fs_info, block_group->flags);
 	if (target) {
-		index = __get_raid_index(extended_to_chunk(target));
+		index = btrfs_bg_flags_to_raid_index(extended_to_chunk(target));
 	} else {
 		/*
 		 * this is just a balance, so if we were marked as full
@@ -9662,7 +9675,7 @@ int btrfs_can_relocate(struct btrfs_fs_info *fs_info, u64 bytenr)
 			goto out;
 		}
 
-		index = get_block_group_index(block_group);
+		index = btrfs_bg_flags_to_raid_index(block_group->flags);
 	}
 
 	if (index == BTRFS_RAID_RAID10) {
@@ -9911,10 +9924,40 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 	return 0;
 }
 
+/* link_block_group will queue up kobjects to add when we're reclaim-safe */
+void btrfs_add_raid_kobjects(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_space_info *space_info;
+	struct raid_kobject *rkobj;
+	LIST_HEAD(list);
+	int index;
+	int ret = 0;
+
+	spin_lock(&fs_info->pending_raid_kobjs_lock);
+	list_splice_init(&fs_info->pending_raid_kobjs, &list);
+	spin_unlock(&fs_info->pending_raid_kobjs_lock);
+
+	list_for_each_entry(rkobj, &list, list) {
+		space_info = __find_space_info(fs_info, rkobj->flags);
+		index = btrfs_bg_flags_to_raid_index(rkobj->flags);
+
+		ret = kobject_add(&rkobj->kobj, &space_info->kobj,
+				  "%s", get_raid_name(index));
+		if (ret) {
+			kobject_put(&rkobj->kobj);
+			break;
+		}
+	}
+	if (ret)
+		btrfs_warn(fs_info,
+			   "failed to add kobject for block cache, ignoring");
+}
+
 static void link_block_group(struct btrfs_block_group_cache *cache)
 {
 	struct btrfs_space_info *space_info = cache->space_info;
-	int index = get_block_group_index(cache);
+	struct btrfs_fs_info *fs_info = cache->fs_info;
+	int index = btrfs_bg_flags_to_raid_index(cache->flags);
 	bool first = false;
 
 	down_write(&space_info->groups_sem);
@@ -9924,27 +9967,20 @@ static void link_block_group(struct btrfs_block_group_cache *cache)
 	up_write(&space_info->groups_sem);
 
 	if (first) {
-		struct raid_kobject *rkobj;
-		int ret;
-
-		rkobj = kzalloc(sizeof(*rkobj), GFP_NOFS);
-		if (!rkobj)
-			goto out_err;
-		rkobj->raid_type = index;
-		kobject_init(&rkobj->kobj, &btrfs_raid_ktype);
-		ret = kobject_add(&rkobj->kobj, &space_info->kobj,
-				  "%s", get_raid_name(index));
-		if (ret) {
-			kobject_put(&rkobj->kobj);
-			goto out_err;
+		struct raid_kobject *rkobj = kzalloc(sizeof(*rkobj), GFP_NOFS);
+		if (!rkobj) {
+			btrfs_warn(cache->fs_info,
+				"couldn't alloc memory for raid level kobject");
+			return;
 		}
+		rkobj->flags = cache->flags;
+		kobject_init(&rkobj->kobj, &btrfs_raid_ktype);
+
+		spin_lock(&fs_info->pending_raid_kobjs_lock);
+		list_add_tail(&rkobj->list, &fs_info->pending_raid_kobjs);
+		spin_unlock(&fs_info->pending_raid_kobjs_lock);
 		space_info->block_group_kobjs[index] = &rkobj->kobj;
 	}
-
-	return;
-out_err:
-	btrfs_warn(cache->fs_info,
-		   "failed to add kobject for block cache, ignoring");
 }
 
 static struct btrfs_block_group_cache *
@@ -10160,6 +10196,7 @@ int btrfs_read_block_groups(struct btrfs_fs_info *info)
 			inc_block_group_ro(cache, 1);
 	}
 
+	btrfs_add_raid_kobjects(info);
 	init_global_block_rsv(info);
 	ret = 0;
 error:
@@ -10167,9 +10204,9 @@ error:
 	return ret;
 }
 
-void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans,
-				       struct btrfs_fs_info *fs_info)
+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, *tmp;
 	struct btrfs_root *extent_root = fs_info->extent_root;
 	struct btrfs_block_group_item item;
@@ -10254,15 +10291,7 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 	 * with its ->space_info set.
 	 */
 	cache->space_info = __find_space_info(fs_info, cache->flags);
-	if (!cache->space_info) {
-		ret = create_space_info(fs_info, cache->flags,
-				       &cache->space_info);
-		if (ret) {
-			btrfs_remove_free_space_cache(cache);
-			btrfs_put_block_group(cache);
-			return ret;
-		}
-	}
+	ASSERT(cache->space_info);
 
 	ret = btrfs_add_block_group_cache(fs_info, cache);
 	if (ret) {
@@ -10334,7 +10363,7 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 				  block_group->key.offset);
 
 	memcpy(&key, &block_group->key, sizeof(key));
-	index = get_block_group_index(block_group);
+	index = btrfs_bg_flags_to_raid_index(block_group->flags);
 	if (block_group->flags & (BTRFS_BLOCK_GROUP_DUP |
 				  BTRFS_BLOCK_GROUP_RAID1 |
 				  BTRFS_BLOCK_GROUP_RAID10))