Merge tag 'for-5.9-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux

Pull btrfs updates from David Sterba:
 "We don't have any big feature updates this time, there are lots of
  small enhacements or fixes. A highlight perhaps is the parallel fsync
  performance improvements, numbers below.

  Regarding the dio/iomap that was reverted last time, the required API
  changes are likely to land in the upcoming cycle, the btrfs part will
  be updated afterwards.

  User visible changes:

   - new mount option rescue= to group all recovery-related mount
     options so we don't have many specific options, currently
     introducing only aliases for existing options, future extensions
     are in development to allow read-only mount with partially damaged
     structures:
      - usebackuproot is an alias for rescue=usebackuproot
      - nologreplay is an alias for rescue=nologreplay

   - start deprecation of mount option inode_cache, removal scheduled to
     v5.11

   - removed deprecated mount options alloc_start and subvolrootid

   - device stats corruption counter gets incremented when a checksum
     mismatch is found

   - qgroup information exported in /sys/fs/btrfs/<UUID>/qgroups/<id>
     using sysfs

   - add link /sys/fs/btrfs/<UUID>/bdi pointing to the associated
     backing dev info

   - FS_INFO ioctl enhancements:
      - add flags to request/describe newly added items
      - new item: numeric checksum type and checksum size
      - new item: generation
      - new item: metadata_uuid

   - seed device: with one new read-write device added, print the new
     device information in /proc/mounts

   - balance: detect cancellation by Ctrl-C in existing cancellation
     points

  Performance improvements:

   - optimized versions of various helpers on little-endian
     architectures, where we don't have to do LE/BE conversion from
     on-disk format

   - tree-log/fsync optimizations leading to lower max latency reported
     by dbench, reduced by about 12%

   - all chunk tree leaves are prefetched at mount time, can improve
     mount time on large (terabyte-sized) filesystems

   - speed up parallel fsync of files with reflinked/deduped extents,
     with jobs 16 to 1024 the throughput gets improved roughly by 50% on
     average and runtime decreased roughly by 30% on average, notable
     outlier is 128 jobs with +121.2% on throughput and -54.6% runtime

   - another speed up of parallel fsync, reduce number of checksum tree
     lookups and contention, the improvements start to show up with 2
     tasks with +20% throughput and -16% runtime up to 64 with +200%
     throughput and -66% runtime

  Core:

   - umount-time qgroup leak checker

   - qgroups
      - add a way to unreserve partial range after failure, avoiding
        some EDQUOT errors
      - improved flushing logic when EDQUOT is hit

   - possible EINTR interruption caused by failed reservations after
     transaction start is better handled and documented

   - transaction abort errors are unified to EROFS in case it's not the
     original reason of abort or we don't have other way to determine
     the reason

  Fixes:

   - make truncate succeed on a NOCOW file even if data space is
     exhausted

   - fix cancelling balance on filesystem with exhausted metadata space

   - anon block device:
      - preallocate anon bdev when subvolume is created to report
        failure early
      - shorten time the anon bdev id is allocated
      - don't allocate anon bdev for internal roots

   - minor memory leak in ref-verify

   - refuse invalid combinations of compression and NOCOW file flags

   - lockdep fixes, updating the device locks

   - remove obsolete fallback logic for block group profile adjustments
     when switching from 1 to more devices, causing allocation of
     unwanted block groups

  Other cleanups, refactoring, simplifications:

   - conversions from struct inode to struct btrfs_inode in internal
     functions

   - removal of unused struct members"

* tag 'for-5.9-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux: (151 commits)
  btrfs: do not set the full sync flag on the inode during page release
  btrfs: release old extent maps during page release
  btrfs: fix race between page release and a fast fsync
  btrfs: open-code remount flag setting in btrfs_remount
  btrfs: if we're restriping, use the target restripe profile
  btrfs: don't adjust bg flags and use default allocation profiles
  btrfs: fix lockdep splat from btrfs_dump_space_info
  btrfs: move the chunk_mutex in btrfs_read_chunk_tree
  btrfs: open device without device_list_mutex
  btrfs: sysfs: use NOFS for device creation
  btrfs: return EROFS for BTRFS_FS_STATE_ERROR cases
  btrfs: document special case error codes for fs errors
  btrfs: don't WARN if we abort a transaction with EROFS
  btrfs: reduce contention on log trees when logging checksums
  btrfs: remove done label in writepage_delalloc
  btrfs: add comments for btrfs_reserve_flush_enum
  btrfs: relocation: review the call sites which can be interrupted by signal
  btrfs: avoid possible signal interruption of btrfs_drop_snapshot() on relocation tree
  btrfs: relocation: allow signal to cancel balance
  btrfs: raid56: remove out label in __raid56_parity_recover
  ...

Author: torvalds

fs/btrfs/block-group.c

@@ -65,11 +65,8 @@ static u64 btrfs_reduce_alloc_profile(struct btrfs_fs_info *fs_info, u64 flags)
 	spin_lock(&fs_info->balance_lock);
 	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) {
-			spin_unlock(&fs_info->balance_lock);
-			return extended_to_chunk(target);
-		}
+		spin_unlock(&fs_info->balance_lock);
+		return extended_to_chunk(target);
 	}
 	spin_unlock(&fs_info->balance_lock);
 
@@ -118,12 +115,12 @@ u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags)
 
 void btrfs_get_block_group(struct btrfs_block_group *cache)
 {
-	atomic_inc(&cache->count);
+	refcount_inc(&cache->refs);
 }
 
 void btrfs_put_block_group(struct btrfs_block_group *cache)
 {
-	if (atomic_dec_and_test(&cache->count)) {
+	if (refcount_dec_and_test(&cache->refs)) {
 		WARN_ON(cache->pinned > 0);
 		WARN_ON(cache->reserved > 0);
 
@@ -1111,7 +1108,6 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 	if (ret < 0)
 		goto out;
 
-	mutex_lock(&fs_info->chunk_mutex);
 	spin_lock(&block_group->lock);
 	block_group->removed = 1;
 	/*
@@ -1143,8 +1139,6 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 	remove_em = (atomic_read(&block_group->frozen) == 0);
 	spin_unlock(&block_group->lock);
 
-	mutex_unlock(&fs_info->chunk_mutex);
-
 	if (remove_em) {
 		struct extent_map_tree *em_tree;
 
@@ -1532,21 +1526,70 @@ void btrfs_mark_bg_unused(struct btrfs_block_group *bg)
 	spin_unlock(&fs_info->unused_bgs_lock);
 }
 
+static int read_bg_from_eb(struct btrfs_fs_info *fs_info, struct btrfs_key *key,
+			   struct btrfs_path *path)
+{
+	struct extent_map_tree *em_tree;
+	struct extent_map *em;
+	struct btrfs_block_group_item bg;
+	struct extent_buffer *leaf;
+	int slot;
+	u64 flags;
+	int ret = 0;
+
+	slot = path->slots[0];
+	leaf = path->nodes[0];
+
+	em_tree = &fs_info->mapping_tree;
+	read_lock(&em_tree->lock);
+	em = lookup_extent_mapping(em_tree, key->objectid, key->offset);
+	read_unlock(&em_tree->lock);
+	if (!em) {
+		btrfs_err(fs_info,
+			  "logical %llu len %llu found bg but no related chunk",
+			  key->objectid, key->offset);
+		return -ENOENT;
+	}
+
+	if (em->start != key->objectid || em->len != key->offset) {
+		btrfs_err(fs_info,
+			"block group %llu len %llu mismatch with chunk %llu len %llu",
+			key->objectid, key->offset, em->start, em->len);
+		ret = -EUCLEAN;
+		goto out_free_em;
+	}
+
+	read_extent_buffer(leaf, &bg, btrfs_item_ptr_offset(leaf, slot),
+			   sizeof(bg));
+	flags = btrfs_stack_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",
+			  key->objectid, key->offset, flags,
+			  (BTRFS_BLOCK_GROUP_TYPE_MASK & em->map_lookup->type));
+		ret = -EUCLEAN;
+	}
+
+out_free_em:
+	free_extent_map(em);
+	return ret;
+}
+
 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;
+	int ret;
 	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;
+		return ret;
 
 	while (1) {
 		slot = path->slots[0];
@@ -1563,49 +1606,10 @@ static int find_first_block_group(struct btrfs_fs_info *fs_info,
 
 		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_stack_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;
+			ret = read_bg_from_eb(fs_info, &found_key, path);
+			break;
 		}
+
 		path->slots[0]++;
 	}
 out:
@@ -1657,19 +1661,12 @@ int btrfs_rmap_block(struct btrfs_fs_info *fs_info, u64 chunk_start,
 		return -EIO;
 
 	map = em->map_lookup;
-	data_stripe_length = em->len;
+	data_stripe_length = em->orig_block_len;
 	io_stripe_size = map->stripe_len;
 
-	if (map->type & BTRFS_BLOCK_GROUP_RAID10)
-		data_stripe_length = div_u64(data_stripe_length,
-					     map->num_stripes / map->sub_stripes);
-	else if (map->type & BTRFS_BLOCK_GROUP_RAID0)
-		data_stripe_length = div_u64(data_stripe_length, map->num_stripes);
-	else if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
-		data_stripe_length = div_u64(data_stripe_length,
-					     nr_data_stripes(map));
+	/* For RAID5/6 adjust to a full IO stripe length */
+	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK)
 		io_stripe_size = map->stripe_len * nr_data_stripes(map);
-	}
 
 	buf = kcalloc(map->num_stripes, sizeof(u64), GFP_NOFS);
 	if (!buf) {
@@ -1748,25 +1745,12 @@ static int exclude_super_stripes(struct btrfs_block_group *cache)
 			return ret;
 
 		while (nr--) {
-			u64 start, len;
-
-			if (logical[nr] > cache->start + cache->length)
-				continue;
-
-			if (logical[nr] + stripe_len <= cache->start)
-				continue;
-
-			start = logical[nr];
-			if (start < cache->start) {
-				start = cache->start;
-				len = (logical[nr] + stripe_len) - start;
-			} else {
-				len = min_t(u64, stripe_len,
-					    cache->start + cache->length - start);
-			}
+			u64 len = min_t(u64, stripe_len,
+				cache->start + cache->length - logical[nr]);
 
 			cache->bytes_super += len;
-			ret = btrfs_add_excluded_extent(fs_info, start, len);
+			ret = btrfs_add_excluded_extent(fs_info, logical[nr],
+							len);
 			if (ret) {
 				kfree(logical);
 				return ret;
@@ -1818,7 +1802,7 @@ static struct btrfs_block_group *btrfs_create_block_group_cache(
 
 	cache->discard_index = BTRFS_DISCARD_INDEX_UNUSED;
 
-	atomic_set(&cache->count, 1);
+	refcount_set(&cache->refs, 1);
 	spin_lock_init(&cache->lock);
 	init_rwsem(&cache->data_rwsem);
 	INIT_LIST_HEAD(&cache->list);
@@ -2207,54 +2191,6 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, u64 bytes_used,
 	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 = 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 one block group RO, can be called several times for the same block
  * group.
@@ -2300,7 +2236,7 @@ int btrfs_inc_block_group_ro(struct btrfs_block_group *cache,
 		 * 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);
+		alloc_flags = btrfs_get_alloc_profile(fs_info, cache->flags);
 		if (alloc_flags != cache->flags) {
 			ret = btrfs_chunk_alloc(trans, alloc_flags,
 						CHUNK_ALLOC_FORCE);
@@ -2327,7 +2263,7 @@ int btrfs_inc_block_group_ro(struct btrfs_block_group *cache,
 	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);
+		alloc_flags = btrfs_get_alloc_profile(fs_info, cache->flags);
 		mutex_lock(&fs_info->chunk_mutex);
 		check_system_chunk(trans, alloc_flags);
 		mutex_unlock(&fs_info->chunk_mutex);
@@ -2521,7 +2457,8 @@ static int cache_save_setup(struct btrfs_block_group *block_group,
 	num_pages *= 16;
 	num_pages *= PAGE_SIZE;
 
-	ret = btrfs_check_data_free_space(inode, &data_reserved, 0, num_pages);
+	ret = btrfs_check_data_free_space(BTRFS_I(inode), &data_reserved, 0,
+					  num_pages);
 	if (ret)
 		goto out_put;
 
@@ -3392,7 +3329,7 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 		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);
+		ASSERT(refcount_read(&block_group->refs) == 1);
 		btrfs_put_block_group(block_group);
 
 		spin_lock(&info->block_group_cache_lock);
@@ -3447,15 +3384,13 @@ void btrfs_unfreeze_block_group(struct btrfs_block_group *block_group)
 	spin_unlock(&block_group->lock);
 
 	if (cleanup) {
-		mutex_lock(&fs_info->chunk_mutex);
 		em_tree = &fs_info->mapping_tree;
 		write_lock(&em_tree->lock);
 		em = lookup_extent_mapping(em_tree, block_group->start,
 					   1);
 		BUG_ON(!em); /* logic error, can't happen */
 		remove_extent_mapping(em_tree, em);
 		write_unlock(&em_tree->lock);
-		mutex_unlock(&fs_info->chunk_mutex);
 
 		/* once for us and once for the tree */
 		free_extent_map(em);

fs/btrfs/block-group.h

@@ -114,8 +114,7 @@ struct btrfs_block_group {
 	/* For block groups in the same raid type */
 	struct list_head list;
 
-	/* Usage count */
-	atomic_t count;
+	refcount_t refs;
 
 	/*
 	 * List of struct btrfs_free_clusters for this block group.

fs/btrfs/btrfs_inode.h

@@ -151,6 +151,17 @@ struct btrfs_inode {
 	 */
 	u64 last_unlink_trans;
 
+	/*
+	 * The id/generation of the last transaction where this inode was
+	 * either the source or the destination of a clone/dedupe operation.
+	 * Used when logging an inode to know if there are shared extents that
+	 * need special care when logging checksum items, to avoid duplicate
+	 * checksum items in a log (which can lead to a corruption where we end
+	 * up with missing checksum ranges after log replay).
+	 * Protected by the vfs inode lock.
+	 */
+	u64 last_reflink_trans;
+
 	/*
 	 * Number of bytes outstanding that are going to need csums.  This is
 	 * used in ENOSPC accounting.