btrfs: zoned: write out partially allocated region

cow_file_range() works in an all-or-nothing way: if it fails to allocate an
extent for a part of the given region, it gives up all the region including
the successfully allocated parts. On cow_file_range(), run_delalloc_zoned()
writes data for the region only when it successfully allocate all the
region.

This all-or-nothing allocation and write-out are problematic when available
space in all the block groups are get tight with the active zone
restriction. btrfs_reserve_extent() try hard to utilize the left space in
the active block groups and gives up finally and fails with
-ENOSPC. However, if we send IOs for the successfully allocated region, we
can finish a zone and can continue on the rest of the allocation on a newly
allocated block group.

This patch implements the partial write-out for run_delalloc_zoned(). With
this patch applied, cow_file_range() returns -EAGAIN to tell the caller to
do something to progress the further allocation, and tells the successfully
allocated region with done_offset. Furthermore, the zoned extent allocator
returns -EAGAIN to tell cow_file_range() going back to the caller side.

Actually, we still need to wait for an IO to complete to continue the
allocation. The next patch implements that part.

CC: stable@vger.kernel.org # 5.16+
Fixes: afba2bc036 ("btrfs: zoned: implement active zone tracking")
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>

fs/btrfs/extent-tree.c

@@ -3996,6 +3996,16 @@ static int can_allocate_chunk_zoned(struct btrfs_fs_info *fs_info,
 	if (ffe_ctl->max_extent_size >= ffe_ctl->min_alloc_size)
 		return -ENOSPC;
 
+	/*
+	 * Even min_alloc_size is not left in any block groups. Since we cannot
+	 * activate a new block group, allocating it may not help. Let's tell a
+	 * caller to try again and hope it progress something by writing some
+	 * parts of the region. That is only possible for data block groups,
+	 * where a part of the region can be written.
+	 */
+	if (ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA)
+		return -EAGAIN;
+
 	/*
 	 * We cannot activate a new block group and no enough space left in any
 	 * block groups. So, allocating a new block group may not help. But,

fs/btrfs/inode.c

@@ -117,7 +117,8 @@ static int btrfs_truncate(struct inode *inode, bool skip_writeback);
 static noinline int cow_file_range(struct btrfs_inode *inode,
 				   struct page *locked_page,
 				   u64 start, u64 end, int *page_started,
-				   unsigned long *nr_written, int unlock);
+				   unsigned long *nr_written, int unlock,
+				   u64 *done_offset);
 static struct extent_map *create_io_em(struct btrfs_inode *inode, u64 start,
 				       u64 len, u64 orig_start, u64 block_start,
 				       u64 block_len, u64 orig_block_len,
@@ -921,7 +922,7 @@ static int submit_uncompressed_range(struct btrfs_inode *inode,
 	 * can directly submit them without interruption.
 	 */
 	ret = cow_file_range(inode, locked_page, start, end, &page_started,
-			     &nr_written, 0);
+			     &nr_written, 0, NULL);
 	/* Inline extent inserted, page gets unlocked and everything is done */
 	if (page_started) {
 		ret = 0;
@@ -1170,7 +1171,8 @@ static u64 get_extent_allocation_hint(struct btrfs_inode *inode, u64 start,
 static noinline int cow_file_range(struct btrfs_inode *inode,
 				   struct page *locked_page,
 				   u64 start, u64 end, int *page_started,
-				   unsigned long *nr_written, int unlock)
+				   unsigned long *nr_written, int unlock,
+				   u64 *done_offset)
 {
 	struct btrfs_root *root = inode->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -1363,6 +1365,21 @@ out_reserve:
 	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
 	btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 1);
 out_unlock:
+	/*
+	 * If done_offset is non-NULL and ret == -EAGAIN, we expect the
+	 * caller to write out the successfully allocated region and retry.
+	 */
+	if (done_offset && ret == -EAGAIN) {
+		if (orig_start < start)
+			*done_offset = start - 1;
+		else
+			*done_offset = start;
+		return ret;
+	} else if (ret == -EAGAIN) {
+		/* Convert to -ENOSPC since the caller cannot retry. */
+		ret = -ENOSPC;
+	}
+
 	/*
 	 * Now, we have three regions to clean up:
 	 *
@@ -1608,19 +1625,37 @@ static noinline int run_delalloc_zoned(struct btrfs_inode *inode,
 				       u64 end, int *page_started,
 				       unsigned long *nr_written)
 {
+	u64 done_offset = end;
 	int ret;
+	bool locked_page_done = false;
 
+	while (start <= end) {
 		ret = cow_file_range(inode, locked_page, start, end, page_started,
-			     nr_written, 0);
+				     nr_written, 0, &done_offset);
-	if (ret)
+		if (ret && ret != -EAGAIN)
 			return ret;
 
-	if (*page_started)
+		if (*page_started) {
+			ASSERT(ret == 0);
 			return 0;
+		}
 
+		if (ret == 0)
+			done_offset = end;
+
+		if (done_offset == start)
+			return -ENOSPC;
+
+		if (!locked_page_done) {
 			__set_page_dirty_nobuffers(locked_page);
 			account_page_redirty(locked_page);
-	extent_write_locked_range(&inode->vfs_inode, start, end);
+		}
+		locked_page_done = true;
+		extent_write_locked_range(&inode->vfs_inode, start, done_offset);
+
+		start = done_offset + 1;
+	}
+
 	*page_started = 1;
 
 	return 0;
@@ -1712,7 +1747,7 @@ static int fallback_to_cow(struct btrfs_inode *inode, struct page *locked_page,
 	}
 
 	return cow_file_range(inode, locked_page, start, end, page_started,
-			      nr_written, 1);
+			      nr_written, 1, NULL);
 }
 
 struct can_nocow_file_extent_args {
@@ -2185,7 +2220,7 @@ int btrfs_run_delalloc_range(struct btrfs_inode *inode, struct page *locked_page
 						 page_started, nr_written);
 		else
 			ret = cow_file_range(inode, locked_page, start, end,
-					     page_started, nr_written, 1);
+					     page_started, nr_written, 1, NULL);
 	} else {
 		set_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, &inode->runtime_flags);
 		ret = cow_file_range_async(inode, wbc, locked_page, start, end,