Merge tag 'for-6.7-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux

Pull btrfs fixes from David Sterba:
  "Some fixes to quota accounting code, mostly around error handling and
   correctness:

   - free reserves on various error paths, after IO errors or
     transaction abort

   - don't clear reserved range at the folio release time, it'll be
     properly cleared after final write

   - fix integer overflow due to int used when passing around size of
     freed reservations

   - fix a regression in squota accounting that missed some cases with
     delayed refs"

* tag 'for-6.7-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: ensure releasing squota reserve on head refs
  btrfs: don't clear qgroup reserved bit in release_folio
  btrfs: free qgroup pertrans reserve on transaction abort
  btrfs: fix qgroup_free_reserved_data int overflow
  btrfs: free qgroup reserve when ORDERED_IOERR is set

fs/btrfs/delalloc-space.c

@@ -199,7 +199,7 @@ void btrfs_free_reserved_data_space(struct btrfs_inode *inode,
 	start = round_down(start, fs_info->sectorsize);
 
 	btrfs_free_reserved_data_space_noquota(fs_info, len);
-	btrfs_qgroup_free_data(inode, reserved, start, len);
+	btrfs_qgroup_free_data(inode, reserved, start, len, NULL);
 }
 
 /*

fs/btrfs/disk-io.c

@@ -4799,6 +4799,32 @@ void btrfs_cleanup_dirty_bgs(struct btrfs_transaction *cur_trans,
 	}
 }
 
+static void btrfs_free_all_qgroup_pertrans(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_root *gang[8];
+	int i;
+	int ret;
+
+	spin_lock(&fs_info->fs_roots_radix_lock);
+	while (1) {
+		ret = radix_tree_gang_lookup_tag(&fs_info->fs_roots_radix,
+						 (void **)gang, 0,
+						 ARRAY_SIZE(gang),
+						 BTRFS_ROOT_TRANS_TAG);
+		if (ret == 0)
+			break;
+		for (i = 0; i < ret; i++) {
+			struct btrfs_root *root = gang[i];
+
+			btrfs_qgroup_free_meta_all_pertrans(root);
+			radix_tree_tag_clear(&fs_info->fs_roots_radix,
+					(unsigned long)root->root_key.objectid,
+					BTRFS_ROOT_TRANS_TAG);
+		}
+	}
+	spin_unlock(&fs_info->fs_roots_radix_lock);
+}
+
 void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
 				   struct btrfs_fs_info *fs_info)
 {
@@ -4827,6 +4853,8 @@ void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
 				     EXTENT_DIRTY);
 	btrfs_destroy_pinned_extent(fs_info, &cur_trans->pinned_extents);
 
+	btrfs_free_all_qgroup_pertrans(fs_info);
+
 	cur_trans->state =TRANS_STATE_COMPLETED;
 	wake_up(&cur_trans->commit_wait);
 }

fs/btrfs/extent-tree.c

@@ -1547,6 +1547,23 @@ static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
+static void free_head_ref_squota_rsv(struct btrfs_fs_info *fs_info,
+				     struct btrfs_delayed_ref_head *href)
+{
+	u64 root = href->owning_root;
+
+	/*
+	 * Don't check must_insert_reserved, as this is called from contexts
+	 * where it has already been unset.
+	 */
+	if (btrfs_qgroup_mode(fs_info) != BTRFS_QGROUP_MODE_SIMPLE ||
+	    !href->is_data || !is_fstree(root))
+		return;
+
+	btrfs_qgroup_free_refroot(fs_info, root, href->reserved_bytes,
+				  BTRFS_QGROUP_RSV_DATA);
+}
+
 static int run_delayed_data_ref(struct btrfs_trans_handle *trans,
 				struct btrfs_delayed_ref_head *href,
 				struct btrfs_delayed_ref_node *node,
@@ -1569,7 +1586,6 @@ static int run_delayed_data_ref(struct btrfs_trans_handle *trans,
 		struct btrfs_squota_delta delta = {
 			.root = href->owning_root,
 			.num_bytes = node->num_bytes,
-			.rsv_bytes = href->reserved_bytes,
 			.is_data = true,
 			.is_inc	= true,
 			.generation = trans->transid,
@@ -1586,11 +1602,9 @@ static int run_delayed_data_ref(struct btrfs_trans_handle *trans,
 						 flags, ref->objectid,
 						 ref->offset, &key,
 						 node->ref_mod, href->owning_root);
+		free_head_ref_squota_rsv(trans->fs_info, href);
 		if (!ret)
 			ret = btrfs_record_squota_delta(trans->fs_info, &delta);
-		else
-			btrfs_qgroup_free_refroot(trans->fs_info, delta.root,
-						  delta.rsv_bytes, BTRFS_QGROUP_RSV_DATA);
 	} else if (node->action == BTRFS_ADD_DELAYED_REF) {
 		ret = __btrfs_inc_extent_ref(trans, node, parent, ref->root,
 					     ref->objectid, ref->offset,
@@ -1742,7 +1756,6 @@ static int run_delayed_tree_ref(struct btrfs_trans_handle *trans,
 		struct btrfs_squota_delta delta = {
 			.root = href->owning_root,
 			.num_bytes = fs_info->nodesize,
-			.rsv_bytes = 0,
 			.is_data = false,
 			.is_inc = true,
 			.generation = trans->transid,
@@ -1774,8 +1787,10 @@ static int run_one_delayed_ref(struct btrfs_trans_handle *trans,
 	int ret = 0;
 
 	if (TRANS_ABORTED(trans)) {
-		if (insert_reserved)
+		if (insert_reserved) {
 			btrfs_pin_extent(trans, node->bytenr, node->num_bytes, 1);
+			free_head_ref_squota_rsv(trans->fs_info, href);
+		}
 		return 0;
 	}
 
@@ -1871,6 +1886,8 @@ u64 btrfs_cleanup_ref_head_accounting(struct btrfs_fs_info *fs_info,
 				  struct btrfs_delayed_ref_root *delayed_refs,
 				  struct btrfs_delayed_ref_head *head)
 {
+	u64 ret = 0;
+
 	/*
 	 * We had csum deletions accounted for in our delayed refs rsv, we need
 	 * to drop the csum leaves for this update from our delayed_refs_rsv.
@@ -1885,14 +1902,13 @@ u64 btrfs_cleanup_ref_head_accounting(struct btrfs_fs_info *fs_info,
 
 		btrfs_delayed_refs_rsv_release(fs_info, 0, nr_csums);
 
-		return btrfs_calc_delayed_ref_csum_bytes(fs_info, nr_csums);
+		ret = btrfs_calc_delayed_ref_csum_bytes(fs_info, nr_csums);
 	}
-	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_SIMPLE &&
-	    head->must_insert_reserved && head->is_data)
-		btrfs_qgroup_free_refroot(fs_info, head->owning_root,
-					  head->reserved_bytes, BTRFS_QGROUP_RSV_DATA);
+	/* must_insert_reserved can be set only if we didn't run the head ref. */
+	if (head->must_insert_reserved)
+		free_head_ref_squota_rsv(fs_info, head);
 
-	return 0;
+	return ret;
 }
 
 static int cleanup_ref_head(struct btrfs_trans_handle *trans,
@@ -2033,6 +2049,12 @@ static int btrfs_run_delayed_refs_for_head(struct btrfs_trans_handle *trans,
 		 * spin lock.
 		 */
 		must_insert_reserved = locked_ref->must_insert_reserved;
+		/*
+		 * Unsetting this on the head ref relinquishes ownership of
+		 * the rsv_bytes, so it is critical that every possible code
+		 * path from here forward frees all reserves including qgroup
+		 * reserve.
+		 */
 		locked_ref->must_insert_reserved = false;
 
 		extent_op = locked_ref->extent_op;
@@ -3292,7 +3314,6 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 		struct btrfs_squota_delta delta = {
 			.root = delayed_ref_root,
 			.num_bytes = num_bytes,
-			.rsv_bytes = 0,
 			.is_data = is_data,
 			.is_inc = false,
 			.generation = btrfs_extent_generation(leaf, ei),
@@ -4937,7 +4958,6 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans,
 		.root = root_objectid,
 		.num_bytes = ins->offset,
 		.generation = trans->transid,
-		.rsv_bytes = 0,
 		.is_data = true,
 		.is_inc = true,
 	};

fs/btrfs/extent_io.c

@@ -2302,7 +2302,8 @@ static int try_release_extent_state(struct extent_io_tree *tree,
 		ret = 0;
 	} else {
 		u32 clear_bits = ~(EXTENT_LOCKED | EXTENT_NODATASUM |
-				   EXTENT_DELALLOC_NEW | EXTENT_CTLBITS);
+				   EXTENT_DELALLOC_NEW | EXTENT_CTLBITS |
+				   EXTENT_QGROUP_RESERVED);
 
 		/*
 		 * At this point we can safely clear everything except the

fs/btrfs/file.c

@@ -3192,7 +3192,7 @@ static long btrfs_fallocate(struct file *file, int mode,
 			qgroup_reserved -= range->len;
 		} else if (qgroup_reserved > 0) {
 			btrfs_qgroup_free_data(BTRFS_I(inode), data_reserved,
-					       range->start, range->len);
+					       range->start, range->len, NULL);
 			qgroup_reserved -= range->len;
 		}
 		list_del(&range->list);

fs/btrfs/inode.c

@@ -688,7 +688,7 @@ static noinline int cow_file_range_inline(struct btrfs_inode *inode, u64 size,
 	 * And at reserve time, it's always aligned to page size, so
 	 * just free one page here.
 	 */
-	btrfs_qgroup_free_data(inode, NULL, 0, PAGE_SIZE);
+	btrfs_qgroup_free_data(inode, NULL, 0, PAGE_SIZE, NULL);
 	btrfs_free_path(path);
 	btrfs_end_transaction(trans);
 	return ret;
@@ -5132,7 +5132,7 @@ static void evict_inode_truncate_pages(struct inode *inode)
 		 */
 		if (state_flags & EXTENT_DELALLOC)
 			btrfs_qgroup_free_data(BTRFS_I(inode), NULL, start,
-					       end - start + 1);
+					       end - start + 1, NULL);
 
 		clear_extent_bit(io_tree, start, end,
 				 EXTENT_CLEAR_ALL_BITS | EXTENT_DO_ACCOUNTING,
@@ -8059,7 +8059,7 @@ static void btrfs_invalidate_folio(struct folio *folio, size_t offset,
 		 *    reserved data space.
 		 *    Since the IO will never happen for this page.
 		 */
-		btrfs_qgroup_free_data(inode, NULL, cur, range_end + 1 - cur);
+		btrfs_qgroup_free_data(inode, NULL, cur, range_end + 1 - cur, NULL);
 		if (!inode_evicting) {
 			clear_extent_bit(tree, cur, range_end, EXTENT_LOCKED |
 				 EXTENT_DELALLOC | EXTENT_UPTODATE |
@@ -9491,7 +9491,7 @@ static struct btrfs_trans_handle *insert_prealloc_file_extent(
 	struct btrfs_path *path;
 	u64 start = ins->objectid;
 	u64 len = ins->offset;
-	int qgroup_released;
+	u64 qgroup_released = 0;
 	int ret;
 
 	memset(&stack_fi, 0, sizeof(stack_fi));
@@ -9504,9 +9504,9 @@ static struct btrfs_trans_handle *insert_prealloc_file_extent(
 	btrfs_set_stack_file_extent_compression(&stack_fi, BTRFS_COMPRESS_NONE);
 	/* Encryption and other encoding is reserved and all 0 */
 
-	qgroup_released = btrfs_qgroup_release_data(inode, file_offset, len);
-	if (qgroup_released < 0)
-		return ERR_PTR(qgroup_released);
+	ret = btrfs_qgroup_release_data(inode, file_offset, len, &qgroup_released);
+	if (ret < 0)
+		return ERR_PTR(ret);
 
 	if (trans) {
 		ret = insert_reserved_file_extent(trans, inode,
@@ -10401,7 +10401,7 @@ ssize_t btrfs_do_encoded_write(struct kiocb *iocb, struct iov_iter *from,
 	btrfs_delalloc_release_metadata(inode, disk_num_bytes, ret < 0);
 out_qgroup_free_data:
 	if (ret < 0)
-		btrfs_qgroup_free_data(inode, data_reserved, start, num_bytes);
+		btrfs_qgroup_free_data(inode, data_reserved, start, num_bytes, NULL);
 out_free_data_space:
 	/*
 	 * If btrfs_reserve_extent() succeeded, then we already decremented

fs/btrfs/ordered-data.c

@@ -152,11 +152,12 @@ static struct btrfs_ordered_extent *alloc_ordered_extent(
 {
 	struct btrfs_ordered_extent *entry;
 	int ret;
+	u64 qgroup_rsv = 0;
 
 	if (flags &
 	    ((1 << BTRFS_ORDERED_NOCOW) | (1 << BTRFS_ORDERED_PREALLOC))) {
 		/* For nocow write, we can release the qgroup rsv right now */
-		ret = btrfs_qgroup_free_data(inode, NULL, file_offset, num_bytes);
+		ret = btrfs_qgroup_free_data(inode, NULL, file_offset, num_bytes, &qgroup_rsv);
 		if (ret < 0)
 			return ERR_PTR(ret);
 	} else {
@@ -164,7 +165,7 @@ static struct btrfs_ordered_extent *alloc_ordered_extent(
 		 * The ordered extent has reserved qgroup space, release now
 		 * and pass the reserved number for qgroup_record to free.
 		 */
-		ret = btrfs_qgroup_release_data(inode, file_offset, num_bytes);
+		ret = btrfs_qgroup_release_data(inode, file_offset, num_bytes, &qgroup_rsv);
 		if (ret < 0)
 			return ERR_PTR(ret);
 	}
@@ -182,7 +183,7 @@ static struct btrfs_ordered_extent *alloc_ordered_extent(
 	entry->inode = igrab(&inode->vfs_inode);
 	entry->compress_type = compress_type;
 	entry->truncated_len = (u64)-1;
-	entry->qgroup_rsv = ret;
+	entry->qgroup_rsv = qgroup_rsv;
 	entry->flags = flags;
 	refcount_set(&entry->refs, 1);
 	init_waitqueue_head(&entry->wait);
@@ -599,7 +600,9 @@ void btrfs_remove_ordered_extent(struct btrfs_inode *btrfs_inode,
 			release = entry->disk_num_bytes;
 		else
 			release = entry->num_bytes;
-		btrfs_delalloc_release_metadata(btrfs_inode, release, false);
+		btrfs_delalloc_release_metadata(btrfs_inode, release,
+						test_bit(BTRFS_ORDERED_IOERR,
+							 &entry->flags));
 	}
 
 	percpu_counter_add_batch(&fs_info->ordered_bytes, -entry->num_bytes,

fs/btrfs/qgroup.c

@@ -4057,13 +4057,14 @@ int btrfs_qgroup_reserve_data(struct btrfs_inode *inode,
 
 /* Free ranges specified by @reserved, normally in error path */
 static int qgroup_free_reserved_data(struct btrfs_inode *inode,
-			struct extent_changeset *reserved, u64 start, u64 len)
+				     struct extent_changeset *reserved,
+				     u64 start, u64 len, u64 *freed_ret)
 {
 	struct btrfs_root *root = inode->root;
 	struct ulist_node *unode;
 	struct ulist_iterator uiter;
 	struct extent_changeset changeset;
-	int freed = 0;
+	u64 freed = 0;
 	int ret;
 
 	extent_changeset_init(&changeset);
@@ -4104,7 +4105,9 @@ static int qgroup_free_reserved_data(struct btrfs_inode *inode,
 	}
 	btrfs_qgroup_free_refroot(root->fs_info, root->root_key.objectid, freed,
 				  BTRFS_QGROUP_RSV_DATA);
-	ret = freed;
+	if (freed_ret)
+		*freed_ret = freed;
+	ret = 0;
 out:
 	extent_changeset_release(&changeset);
 	return ret;
@@ -4112,7 +4115,7 @@ static int qgroup_free_reserved_data(struct btrfs_inode *inode,
 
 static int __btrfs_qgroup_release_data(struct btrfs_inode *inode,
 			struct extent_changeset *reserved, u64 start, u64 len,
-			int free)
+			u64 *released, int free)
 {
 	struct extent_changeset changeset;
 	int trace_op = QGROUP_RELEASE;
@@ -4128,7 +4131,7 @@ static int __btrfs_qgroup_release_data(struct btrfs_inode *inode,
 	/* In release case, we shouldn't have @reserved */
 	WARN_ON(!free && reserved);
 	if (free && reserved)
-		return qgroup_free_reserved_data(inode, reserved, start, len);
+		return qgroup_free_reserved_data(inode, reserved, start, len, released);
 	extent_changeset_init(&changeset);
 	ret = clear_record_extent_bits(&inode->io_tree, start, start + len -1,
 				       EXTENT_QGROUP_RESERVED, &changeset);
@@ -4143,7 +4146,8 @@ static int __btrfs_qgroup_release_data(struct btrfs_inode *inode,
 		btrfs_qgroup_free_refroot(inode->root->fs_info,
 				inode->root->root_key.objectid,
 				changeset.bytes_changed, BTRFS_QGROUP_RSV_DATA);
-	ret = changeset.bytes_changed;
+	if (released)
+		*released = changeset.bytes_changed;
 out:
 	extent_changeset_release(&changeset);
 	return ret;
@@ -4162,9 +4166,10 @@ static int __btrfs_qgroup_release_data(struct btrfs_inode *inode,
  * NOTE: This function may sleep for memory allocation.
  */
 int btrfs_qgroup_free_data(struct btrfs_inode *inode,
-			struct extent_changeset *reserved, u64 start, u64 len)
+			   struct extent_changeset *reserved,
+			   u64 start, u64 len, u64 *freed)
 {
-	return __btrfs_qgroup_release_data(inode, reserved, start, len, 1);
+	return __btrfs_qgroup_release_data(inode, reserved, start, len, freed, 1);
 }
 
 /*
@@ -4182,9 +4187,9 @@ int btrfs_qgroup_free_data(struct btrfs_inode *inode,
  *
  * NOTE: This function may sleep for memory allocation.
  */
-int btrfs_qgroup_release_data(struct btrfs_inode *inode, u64 start, u64 len)
+int btrfs_qgroup_release_data(struct btrfs_inode *inode, u64 start, u64 len, u64 *released)
 {
-	return __btrfs_qgroup_release_data(inode, NULL, start, len, 0);
+	return __btrfs_qgroup_release_data(inode, NULL, start, len, released, 0);
 }
 
 static void add_root_meta_rsv(struct btrfs_root *root, int num_bytes,
@@ -4332,8 +4337,9 @@ static void qgroup_convert_meta(struct btrfs_fs_info *fs_info, u64 ref_root,
 
 		qgroup_rsv_release(fs_info, qgroup, num_bytes,
 				BTRFS_QGROUP_RSV_META_PREALLOC);
-		qgroup_rsv_add(fs_info, qgroup, num_bytes,
-				BTRFS_QGROUP_RSV_META_PERTRANS);
+		if (!sb_rdonly(fs_info->sb))
+			qgroup_rsv_add(fs_info, qgroup, num_bytes,
+				       BTRFS_QGROUP_RSV_META_PERTRANS);
 
 		list_for_each_entry(glist, &qgroup->groups, next_group)
 			qgroup_iterator_add(&qgroup_list, glist->group);
@@ -4655,6 +4661,17 @@ void btrfs_qgroup_destroy_extent_records(struct btrfs_transaction *trans)
 	*root = RB_ROOT;
 }
 
+void btrfs_free_squota_rsv(struct btrfs_fs_info *fs_info, u64 root, u64 rsv_bytes)
+{
+	if (btrfs_qgroup_mode(fs_info) != BTRFS_QGROUP_MODE_SIMPLE)
+		return;
+
+	if (!is_fstree(root))
+		return;
+
+	btrfs_qgroup_free_refroot(fs_info, root, rsv_bytes, BTRFS_QGROUP_RSV_DATA);
+}
+
 int btrfs_record_squota_delta(struct btrfs_fs_info *fs_info,
 			      struct btrfs_squota_delta *delta)
 {
@@ -4699,8 +4716,5 @@ int btrfs_record_squota_delta(struct btrfs_fs_info *fs_info,
 
 out:
 	spin_unlock(&fs_info->qgroup_lock);
-	if (!ret && delta->rsv_bytes)
-		btrfs_qgroup_free_refroot(fs_info, root, delta->rsv_bytes,
-					  BTRFS_QGROUP_RSV_DATA);
 	return ret;
 }

fs/btrfs/qgroup.h

@@ -274,8 +274,6 @@ struct btrfs_squota_delta {
 	u64 root;
 	/* The number of bytes in the extent being counted. */
 	u64 num_bytes;
-	/* The number of bytes reserved for this extent. */
-	u64 rsv_bytes;
 	/* The generation the extent was created in. */
 	u64 generation;
 	/* Whether we are using or freeing the extent. */
@@ -358,10 +356,10 @@ int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid,
 /* New io_tree based accurate qgroup reserve API */
 int btrfs_qgroup_reserve_data(struct btrfs_inode *inode,
 			struct extent_changeset **reserved, u64 start, u64 len);
-int btrfs_qgroup_release_data(struct btrfs_inode *inode, u64 start, u64 len);
+int btrfs_qgroup_release_data(struct btrfs_inode *inode, u64 start, u64 len, u64 *released);
 int btrfs_qgroup_free_data(struct btrfs_inode *inode,
 			   struct extent_changeset *reserved, u64 start,
-			   u64 len);
+			   u64 len, u64 *freed);
 int btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
 			      enum btrfs_qgroup_rsv_type type, bool enforce);
 int __btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
@@ -422,6 +420,7 @@ int btrfs_qgroup_trace_subtree_after_cow(struct btrfs_trans_handle *trans,
 		struct btrfs_root *root, struct extent_buffer *eb);
 void btrfs_qgroup_destroy_extent_records(struct btrfs_transaction *trans);
 bool btrfs_check_quota_leak(struct btrfs_fs_info *fs_info);
+void btrfs_free_squota_rsv(struct btrfs_fs_info *fs_info, u64 root, u64 rsv_bytes);
 int btrfs_record_squota_delta(struct btrfs_fs_info *fs_info,
 			      struct btrfs_squota_delta *delta);
 

fs/btrfs/transaction.c

@@ -37,8 +37,6 @@
 
 static struct kmem_cache *btrfs_trans_handle_cachep;
 
-#define BTRFS_ROOT_TRANS_TAG 0
-
 /*
  * Transaction states and transitions
  *

fs/btrfs/transaction.h

@@ -12,6 +12,9 @@
 #include "ctree.h"
 #include "misc.h"
 
+/* Radix-tree tag for roots that are part of the trasaction. */
+#define BTRFS_ROOT_TRANS_TAG			0
+
 enum btrfs_trans_state {
 	TRANS_STATE_RUNNING,
 	TRANS_STATE_COMMIT_PREP,