zonefs: Fix management of open zones

The mount option "explicit_open" manages the device open zone
resources to ensure that if an application opens a sequential file for
writing, the file zone can always be written by explicitly opening
the zone and accounting for that state with the s_open_zones counter.

However, if some zones are already open when mounting, the device open
zone resource usage status will be larger than the initial s_open_zones
value of 0. Ensure that this inconsistency does not happen by closing
any sequential zone that is open when mounting.

Furthermore, with ZNS drives, closing an explicitly open zone that has
not been written will change the zone state to "closed", that is, the
zone will remain in an active state. Since this can then cause failures
of explicit open operations on other zones if the drive active zone
resources are exceeded, we need to make sure that the zone is not
active anymore by resetting it instead of closing it. To address this,
zonefs_zone_mgmt() is modified to change a REQ_OP_ZONE_CLOSE request
into a REQ_OP_ZONE_RESET for sequential zones that have not been
written.

Fixes: b5c00e9 ("zonefs: open/close zone on file open/close")
Cc: <[email protected]>
Signed-off-by: Damien Le Moal <[email protected]>
Reviewed-by: Johannes Thumshirn <[email protected]>
Reviewed-by: Hans Holmberg <[email protected]>

Author: Damien Le Moal

fs/zonefs/super.c

@@ -35,6 +35,17 @@ static inline int zonefs_zone_mgmt(struct inode *inode,
 
 	lockdep_assert_held(&zi->i_truncate_mutex);
 
+	/*
+	 * With ZNS drives, closing an explicitly open zone that has not been
+	 * written will change the zone state to "closed", that is, the zone
+	 * will remain active. Since this can then cause failure of explicit
+	 * open operation on other zones if the drive active zone resources
+	 * are exceeded, make sure that the zone does not remain active by
+	 * resetting it.
+	 */
+	if (op == REQ_OP_ZONE_CLOSE && !zi->i_wpoffset)
+		op = REQ_OP_ZONE_RESET;
+
 	trace_zonefs_zone_mgmt(inode, op);
 	ret = blkdev_zone_mgmt(inode->i_sb->s_bdev, op, zi->i_zsector,
 			       zi->i_zone_size >> SECTOR_SHIFT, GFP_NOFS);
@@ -1294,12 +1305,13 @@ static void zonefs_init_dir_inode(struct inode *parent, struct inode *inode,
 	inc_nlink(parent);
 }
 
-static void zonefs_init_file_inode(struct inode *inode, struct blk_zone *zone,
-				   enum zonefs_ztype type)
+static int zonefs_init_file_inode(struct inode *inode, struct blk_zone *zone,
+				  enum zonefs_ztype type)
 {
 	struct super_block *sb = inode->i_sb;
 	struct zonefs_sb_info *sbi = ZONEFS_SB(sb);
 	struct zonefs_inode_info *zi = ZONEFS_I(inode);
+	int ret = 0;
 
 	inode->i_ino = zone->start >> sbi->s_zone_sectors_shift;
 	inode->i_mode = S_IFREG | sbi->s_perm;
@@ -1324,6 +1336,22 @@ static void zonefs_init_file_inode(struct inode *inode, struct blk_zone *zone,
 	sb->s_maxbytes = max(zi->i_max_size, sb->s_maxbytes);
 	sbi->s_blocks += zi->i_max_size >> sb->s_blocksize_bits;
 	sbi->s_used_blocks += zi->i_wpoffset >> sb->s_blocksize_bits;
+
+	/*
+	 * For sequential zones, make sure that any open zone is closed first
+	 * to ensure that the initial number of open zones is 0, in sync with
+	 * the open zone accounting done when the mount option
+	 * ZONEFS_MNTOPT_EXPLICIT_OPEN is used.
+	 */
+	if (type == ZONEFS_ZTYPE_SEQ &&
+	    (zone->cond == BLK_ZONE_COND_IMP_OPEN ||
+	     zone->cond == BLK_ZONE_COND_EXP_OPEN)) {
+		mutex_lock(&zi->i_truncate_mutex);
+		ret = zonefs_zone_mgmt(inode, REQ_OP_ZONE_CLOSE);
+		mutex_unlock(&zi->i_truncate_mutex);
+	}
+
+	return ret;
 }
 
 static struct dentry *zonefs_create_inode(struct dentry *parent,
@@ -1333,6 +1361,7 @@ static struct dentry *zonefs_create_inode(struct dentry *parent,
 	struct inode *dir = d_inode(parent);
 	struct dentry *dentry;
 	struct inode *inode;
+	int ret;
 
 	dentry = d_alloc_name(parent, name);
 	if (!dentry)
@@ -1343,10 +1372,16 @@ static struct dentry *zonefs_create_inode(struct dentry *parent,
 		goto dput;
 
 	inode->i_ctime = inode->i_mtime = inode->i_atime = dir->i_ctime;
-	if (zone)
-		zonefs_init_file_inode(inode, zone, type);
-	else
+	if (zone) {
+		ret = zonefs_init_file_inode(inode, zone, type);
+		if (ret) {
+			iput(inode);
+			goto dput;
+		}
+	} else {
 		zonefs_init_dir_inode(dir, inode, type);
+	}
+
 	d_add(dentry, inode);
 	dir->i_size++;