8 files changed, 112 insertions, 47 deletions
diff --git a/fs/btrfs/bio.c b/fs/btrfs/bio.c
index 477f350a8bd0..e3a57196b0ee 100644
--- a/fs/btrfs/bio.c
+++ b/fs/btrfs/bio.c
@@ -741,7 +741,9 @@ static bool btrfs_submit_chunk(struct btrfs_bio *bbio, int mirror_num)
 			ret = btrfs_bio_csum(bbio);
 			if (ret)
 				goto fail_put_bio;
-		} else if (use_append) {
+		} else if (use_append ||
+			   (btrfs_is_zoned(fs_info) && inode &&
+			    inode->flags & BTRFS_INODE_NODATASUM)) {
 			ret = btrfs_alloc_dummy_sum(bbio);
 			if (ret)
 				goto fail_put_bio;
diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 1e09aeea69c2..1a66be33bb04 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -1785,6 +1785,7 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		container_of(work, struct btrfs_fs_info, reclaim_bgs_work);
 	struct btrfs_block_group *bg;
 	struct btrfs_space_info *space_info;
+	LIST_HEAD(retry_list);
 
 	if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))
 		return;
@@ -1921,8 +1922,11 @@ void btrfs_reclaim_bgs_work(struct work_struct *work)
 		}
 
 next:
-		if (ret)
-			btrfs_mark_bg_to_reclaim(bg);
+		if (ret) {
+			/* Refcount held by the reclaim_bgs list after splice. */
+			btrfs_get_block_group(bg);
+			list_add_tail(&bg->bg_list, &retry_list);
+		}
 		btrfs_put_block_group(bg);
 
 		mutex_unlock(&fs_info->reclaim_bgs_lock);
@@ -1942,6 +1946,9 @@ next:
 	spin_unlock(&fs_info->unused_bgs_lock);
 	mutex_unlock(&fs_info->reclaim_bgs_lock);
 end:
+	spin_lock(&fs_info->unused_bgs_lock);
+	list_splice_tail(&retry_list, &fs_info->reclaim_bgs);
+	spin_unlock(&fs_info->unused_bgs_lock);
 	btrfs_exclop_finish(fs_info);
 	sb_end_write(fs_info->sb);
 }
diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index 91c994b569f3..6ed495ca7a31 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -89,6 +89,16 @@ enum {
 	BTRFS_INODE_FREE_SPACE_INODE,
 	/* Set when there are no capabilities in XATTs for the inode. */
 	BTRFS_INODE_NO_CAP_XATTR,
+	/*
+	 * Set if an error happened when doing a COW write before submitting a
+	 * bio or during writeback. Used for both buffered writes and direct IO
+	 * writes. This is to signal a fast fsync that it has to wait for
+	 * ordered extents to complete and therefore not log extent maps that
+	 * point to unwritten extents (when an ordered extent completes and it
+	 * has the BTRFS_ORDERED_IOERR flag set, it drops extent maps in its
+	 * range).
+	 */
+	BTRFS_INODE_COW_WRITE_ERROR,
 };
 
 /* in memory btrfs inode */
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 1b20b3e390df..38cdb8875e8e 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -4538,18 +4538,10 @@ static void btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 				       struct btrfs_fs_info *fs_info)
 {
 	struct rb_node *node;
-	struct btrfs_delayed_ref_root *delayed_refs;
+	struct btrfs_delayed_ref_root *delayed_refs = &trans->delayed_refs;
 	struct btrfs_delayed_ref_node *ref;
 
-	delayed_refs = &trans->delayed_refs;
-
 	spin_lock(&delayed_refs->lock);
-	if (atomic_read(&delayed_refs->num_entries) == 0) {
-		spin_unlock(&delayed_refs->lock);
-		btrfs_debug(fs_info, "delayed_refs has NO entry");
-		return;
-	}
-
 	while ((node = rb_first_cached(&delayed_refs->href_root)) != NULL) {
 		struct btrfs_delayed_ref_head *head;
 		struct rb_node *n;
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 597387e9f040..f688fab55251 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -3689,6 +3689,8 @@ static struct extent_buffer *grab_extent_buffer(
 	struct folio *folio = page_folio(page);
 	struct extent_buffer *exists;
 
+	lockdep_assert_held(&page->mapping->i_private_lock);
+
 	/*
 	 * For subpage case, we completely rely on radix tree to ensure we
 	 * don't try to insert two ebs for the same bytenr.  So here we always
@@ -3756,13 +3758,14 @@ static int check_eb_alignment(struct btrfs_fs_info *fs_info, u64 start)
  * The caller needs to free the existing folios and retry using the same order.
  */
 static int attach_eb_folio_to_filemap(struct extent_buffer *eb, int i,
+				      struct btrfs_subpage *prealloc,
 				      struct extent_buffer **found_eb_ret)
 {
 
 	struct btrfs_fs_info *fs_info = eb->fs_info;
 	struct address_space *mapping = fs_info->btree_inode->i_mapping;
 	const unsigned long index = eb->start >> PAGE_SHIFT;
-	struct folio *existing_folio;
+	struct folio *existing_folio = NULL;
 	int ret;
 
 	ASSERT(found_eb_ret);
@@ -3774,12 +3777,14 @@ retry:
 	ret = filemap_add_folio(mapping, eb->folios[i], index + i,
 				GFP_NOFS | __GFP_NOFAIL);
 	if (!ret)
-		return 0;
+		goto finish;
 
 	existing_folio = filemap_lock_folio(mapping, index + i);
 	/* The page cache only exists for a very short time, just retry. */
-	if (IS_ERR(existing_folio))
+	if (IS_ERR(existing_folio)) {
+		existing_folio = NULL;
 		goto retry;
+	}
 
 	/* For now, we should only have single-page folios for btree inode. */
 	ASSERT(folio_nr_pages(existing_folio) == 1);
@@ -3790,14 +3795,13 @@ retry:
 		return -EAGAIN;
 	}
 
-	if (fs_info->nodesize < PAGE_SIZE) {
-		/*
-		 * We're going to reuse the existing page, can drop our page
-		 * and subpage structure now.
-		 */
+finish:
+	spin_lock(&mapping->i_private_lock);
+	if (existing_folio && fs_info->nodesize < PAGE_SIZE) {
+		/* We're going to reuse the existing page, can drop our folio now. */
 		__free_page(folio_page(eb->folios[i], 0));
 		eb->folios[i] = existing_folio;
-	} else {
+	} else if (existing_folio) {
 		struct extent_buffer *existing_eb;
 
 		existing_eb = grab_extent_buffer(fs_info,
@@ -3805,6 +3809,7 @@ retry:
 		if (existing_eb) {
 			/* The extent buffer still exists, we can use it directly. */
 			*found_eb_ret = existing_eb;
+			spin_unlock(&mapping->i_private_lock);
 			folio_unlock(existing_folio);
 			folio_put(existing_folio);
 			return 1;
@@ -3813,6 +3818,22 @@ retry:
 		__free_page(folio_page(eb->folios[i], 0));
 		eb->folios[i] = existing_folio;
 	}
+	eb->folio_size = folio_size(eb->folios[i]);
+	eb->folio_shift = folio_shift(eb->folios[i]);
+	/* Should not fail, as we have preallocated the memory. */
+	ret = attach_extent_buffer_folio(eb, eb->folios[i], prealloc);
+	ASSERT(!ret);
+	/*
+	 * To inform we have an extra eb under allocation, so that
+	 * detach_extent_buffer_page() won't release the folio private when the
+	 * eb hasn't been inserted into radix tree yet.
+	 *
+	 * The ref will be decreased when the eb releases the page, in
+	 * detach_extent_buffer_page().  Thus needs no special handling in the
+	 * error path.
+	 */
+	btrfs_folio_inc_eb_refs(fs_info, eb->folios[i]);
+	spin_unlock(&mapping->i_private_lock);
 	return 0;
 }
 
@@ -3824,7 +3845,6 @@ struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
 	int attached = 0;
 	struct extent_buffer *eb;
 	struct extent_buffer *existing_eb = NULL;
-	struct address_space *mapping = fs_info->btree_inode->i_mapping;
 	struct btrfs_subpage *prealloc = NULL;
 	u64 lockdep_owner = owner_root;
 	bool page_contig = true;
@@ -3890,7 +3910,7 @@ reallocate:
 	for (int i = 0; i < num_folios; i++) {
 		struct folio *folio;
 
-		ret = attach_eb_folio_to_filemap(eb, i, &existing_eb);
+		ret = attach_eb_folio_to_filemap(eb, i, prealloc, &existing_eb);
 		if (ret > 0) {
 			ASSERT(existing_eb);
 			goto out;
@@ -3927,24 +3947,6 @@ reallocate:
 		 * and free the allocated page.
 		 */
 		folio = eb->folios[i];
-		eb->folio_size = folio_size(folio);
-		eb->folio_shift = folio_shift(folio);
-		spin_lock(&mapping->i_private_lock);
-		/* Should not fail, as we have preallocated the memory */
-		ret = attach_extent_buffer_folio(eb, folio, prealloc);
-		ASSERT(!ret);
-		/*
-		 * To inform we have extra eb under allocation, so that
-		 * detach_extent_buffer_page() won't release the folio private
-		 * when the eb hasn't yet been inserted into radix tree.
-		 *
-		 * The ref will be decreased when the eb released the page, in
-		 * detach_extent_buffer_page().
-		 * Thus needs no special handling in error path.
-		 */
-		btrfs_folio_inc_eb_refs(fs_info, folio);
-		spin_unlock(&mapping->i_private_lock);
-
 		WARN_ON(btrfs_folio_test_dirty(fs_info, folio, eb->start, eb->len));
 
 		/*
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index e764ac3f22e2..d90138683a0a 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1885,6 +1885,7 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	 */
 	if (full_sync || btrfs_is_zoned(fs_info)) {
 		ret = btrfs_wait_ordered_range(inode, start, len);
+		clear_bit(BTRFS_INODE_COW_WRITE_ERROR, &BTRFS_I(inode)->runtime_flags);
 	} else {
 		/*
 		 * Get our ordered extents as soon as possible to avoid doing
@@ -1894,6 +1895,21 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 		btrfs_get_ordered_extents_for_logging(BTRFS_I(inode),
 						      &ctx.ordered_extents);
 		ret = filemap_fdatawait_range(inode->i_mapping, start, end);
+		if (ret)
+			goto out_release_extents;
+
+		/*
+		 * Check and clear the BTRFS_INODE_COW_WRITE_ERROR now after
+		 * starting and waiting for writeback, because for buffered IO
+		 * it may have been set during the end IO callback
+		 * (end_bbio_data_write() -> btrfs_finish_ordered_extent()) in
+		 * case an error happened and we need to wait for ordered
+		 * extents to complete so that any extent maps that point to
+		 * unwritten locations are dropped and we don't log them.
+		 */
+		if (test_and_clear_bit(BTRFS_INODE_COW_WRITE_ERROR,
+				       &BTRFS_I(inode)->runtime_flags))
+			ret = btrfs_wait_ordered_range(inode, start, len);
 	}
 
 	if (ret)
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index c5bdd674f55c..35a413ce935d 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -388,6 +388,37 @@ bool btrfs_finish_ordered_extent(struct btrfs_ordered_extent *ordered,
 	ret = can_finish_ordered_extent(ordered, page, file_offset, len, uptodate);
 	spin_unlock_irqrestore(&inode->ordered_tree_lock, flags);
 
+	/*
+	 * If this is a COW write it means we created new extent maps for the
+	 * range and they point to unwritten locations if we got an error either
+	 * before submitting a bio or during IO.
+	 *
+	 * We have marked the ordered extent with BTRFS_ORDERED_IOERR, and we
+	 * are queuing its completion below. During completion, at
+	 * btrfs_finish_one_ordered(), we will drop the extent maps for the
+	 * unwritten extents.
+	 *
+	 * However because completion runs in a work queue we can end up having
+	 * a fast fsync running before that. In the case of direct IO, once we
+	 * unlock the inode the fsync might start, and we queue the completion
+	 * before unlocking the inode. In the case of buffered IO when writeback
+	 * finishes (end_bbio_data_write()) we queue the completion, so if the
+	 * writeback was triggered by a fast fsync, the fsync might start
+	 * logging before ordered extent completion runs in the work queue.
+	 *
+	 * The fast fsync will log file extent items based on the extent maps it
+	 * finds, so if by the time it collects extent maps the ordered extent
+	 * completion didn't happen yet, it will log file extent items that
+	 * point to unwritten extents, resulting in a corruption if a crash
+	 * happens and the log tree is replayed. Note that a fast fsync does not
+	 * wait for completion of ordered extents in order to reduce latency.
+	 *
+	 * Set a flag in the inode so that the next fast fsync will wait for
+	 * ordered extents to complete before starting to log.
+	 */
+	if (!uptodate && !test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags))
+		set_bit(BTRFS_INODE_COW_WRITE_ERROR, &inode->runtime_flags);
+
 	if (ret)
 		btrfs_queue_ordered_fn(ordered);
 	return ret;
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index 5146387b416b..26a2e5aa08e9 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -4860,18 +4860,23 @@ static int btrfs_log_prealloc_extents(struct btrfs_trans_handle *trans,
 			path->slots[0]++;
 			continue;
 		}
-		if (!dropped_extents) {
-			/*
-			 * Avoid logging extent items logged in past fsync calls
-			 * and leading to duplicate keys in the log tree.
-			 */
+		/*
+		 * Avoid overlapping items in the log tree. The first time we
+		 * get here, get rid of everything from a past fsync. After
+		 * that, if the current extent starts before the end of the last
+		 * extent we copied, truncate the last one. This can happen if
+		 * an ordered extent completion modifies the subvolume tree
+		 * while btrfs_next_leaf() has the tree unlocked.
+		 */
+		if (!dropped_extents || key.offset < truncate_offset) {
 			ret = truncate_inode_items(trans, root->log_root, inode,
-						   truncate_offset,
+						   min(key.offset, truncate_offset),
 						   BTRFS_EXTENT_DATA_KEY);
 			if (ret)
 				goto out;
 			dropped_extents = true;
 		}
+		truncate_offset = btrfs_file_extent_end(path);
 		if (ins_nr == 0)
 			start_slot = slot;
 		ins_nr++;