btrfs: zoned: wait until zone is finished when allocation didn't progress

When the allocated position doesn't progress, we cannot submit IOs to
finish a block group, but there should be ongoing IOs that will finish a
block group. So, in that case, we wait for a zone to be finished and retry
the allocation after that.

Introduce a new flag BTRFS_FS_NEED_ZONE_FINISH for fs_info->flags to
indicate we need a zone finish to have proceeded. The flag is set when the
allocator detected it cannot activate a new block group. And, it is cleared
once a zone is finished.

CC: stable@vger.kernel.org # 5.16+
Fixes: afba2bc036b0 ("btrfs: zoned: implement active zone tracking")
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>

4 files changed, 19 insertions, 2 deletions

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 7859635d876e..202496172059 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -638,6 +638,9 @@ enum {
 	/* Indicate we have half completed snapshot deletions pending. */
 	BTRFS_FS_UNFINISHED_DROPS,
 
+	/* Indicate we have to finish a zone to do next allocation. */
+	BTRFS_FS_NEED_ZONE_FINISH,
+
 #if BITS_PER_LONG == 32
 	/* Indicate if we have error/warn message printed on 32bit systems */
 	BTRFS_FS_32BIT_ERROR,
@@ -1086,6 +1089,8 @@ struct btrfs_fs_info {
 
 	spinlock_t zone_active_bgs_lock;
 	struct list_head zone_active_bgs;
+	/* Waiters when BTRFS_FS_NEED_ZONE_FINISH is set */
+	wait_queue_head_t zone_finish_wait;
 
 	/* Updates are not protected by any lock */
 	struct btrfs_commit_stats commit_stats;
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 90e513e54b48..3fac429cf8a4 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -3152,6 +3152,7 @@ void btrfs_init_fs_info(struct btrfs_fs_info *fs_info)
 	init_waitqueue_head(&fs_info->transaction_blocked_wait);
 	init_waitqueue_head(&fs_info->async_submit_wait);
 	init_waitqueue_head(&fs_info->delayed_iputs_wait);
+	init_waitqueue_head(&fs_info->zone_finish_wait);
 
 	/* Usable values until the real ones are cached from the superblock */
 	fs_info->nodesize = 4096;
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index a8f972835544..16789bbacf75 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -1643,8 +1643,13 @@ static noinline int run_delalloc_zoned(struct btrfs_inode *inode,
 		if (ret == 0)
 			done_offset = end;
 
-		if (done_offset == start)
-			return -ENOSPC;
+		if (done_offset == start) {
+			struct btrfs_fs_info *info = inode->root->fs_info;
+
+			wait_var_event(&info->zone_finish_wait,
+				       !test_bit(BTRFS_FS_NEED_ZONE_FINISH, &info->flags));
+			continue;
+		}
 
 		if (!locked_page_done) {
 			__set_page_dirty_nobuffers(locked_page);
diff --git a/fs/btrfs/zoned.c b/fs/btrfs/zoned.c
index 6c391b5b4172..b150b07ba1a7 100644
--- a/fs/btrfs/zoned.c
+++ b/fs/btrfs/zoned.c
@@ -2007,6 +2007,9 @@ static int do_zone_finish(struct btrfs_block_group *block_group, bool fully_writ
 	/* For active_bg_list */
 	btrfs_put_block_group(block_group);
 
+	clear_bit(BTRFS_FS_NEED_ZONE_FINISH, &fs_info->flags);
+	wake_up_all(&fs_info->zone_finish_wait);
+
 	return 0;
 }
 
@@ -2043,6 +2046,9 @@ bool btrfs_can_activate_zone(struct btrfs_fs_devices *fs_devices, u64 flags)
 	}
 	mutex_unlock(&fs_info->chunk_mutex);
 
+	if (!ret)
+		set_bit(BTRFS_FS_NEED_ZONE_FINISH, &fs_info->flags);
+
 	return ret;
 }