f2fs-tools: zns zone-capacity support

NVM Express Zoned Namespace (ZNS) devices can have zone-capacity(zc) less
than the zone-size. ZNS defines a per zone capacity which can be equal
or less than the zone-size. Zone-capacity is the number of usable blocks
in the zone. If zone-capacity is less than zone-size, then the segments
which start at/after zone-capacity are considered unusable. Only those
segments which start before the zone-capacity are considered as usable
and added to the free_segment_count and free_segment_bitmap of the kernel.
In such cases, the filesystem should not write/read beyond the
zone-capacity.

Update the super block with the usable number of blocks and free segment
count in the ZNS device zones, if zone-capacity is less than zone-size.
Set reserved segment count and overprovision ratio based on the usable
segments in the zone.

Allow fsck to find the free_segment_count based on the zone-capacity and
compare with checkpoint values.

Signed-off-by: Aravind Ramesh <aravind.ramesh@wdc.com>
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
[Jaegeuk Kim: add UNUSED to is_usable_seg()]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

8 files changed, 190 insertions, 15 deletions

diff --git a/configure.ac b/configure.ac
index 9ac0c24..e9acd1a 100644
--- a/configure.ac
+++ b/configure.ac
@@ -213,6 +213,10 @@ AC_CONFIG_FILES([
 	tools/f2fs_io/Makefile
 ])
 
+AC_CHECK_MEMBER([struct blk_zone.capacity],
+		[AC_DEFINE(HAVE_BLK_ZONE_REP_V2, [1], [report zones includes zone capacity])],
+		[], [[#include <linux/blkzoned.h>]])
+
 # export library version info for mkfs/libf2fs_format_la
 AC_SUBST(FMT_CURRENT, 6)
 AC_SUBST(FMT_REVISION, 0)
diff --git a/fsck/fsck.c b/fsck/fsck.c
index 0f627eb..f97e9fb 100644
--- a/fsck/fsck.c
+++ b/fsck/fsck.c
@@ -1905,11 +1905,12 @@ int fsck_chk_meta(struct f2fs_sb_info *sbi)
 		if (IS_NODESEG(se->type))
 			sit_node_blks += se->valid_blocks;
 	}
-	if (fsck->chk.sit_free_segs + sit_valid_segs != TOTAL_SEGS(sbi)) {
+	if (fsck->chk.sit_free_segs + sit_valid_segs !=
+				get_usable_seg_count(sbi)) {
 		ASSERT_MSG("SIT usage does not match: sit_free_segs %u, "
 				"sit_valid_segs %u, total_segs %u",
 			fsck->chk.sit_free_segs, sit_valid_segs,
-			TOTAL_SEGS(sbi));
+			get_usable_seg_count(sbi));
 		return -EINVAL;
 	}
 
diff --git a/fsck/fsck.h b/fsck/fsck.h
index d8bab97..c5e85fe 100644
--- a/fsck/fsck.h
+++ b/fsck/fsck.h
@@ -235,6 +235,8 @@ extern u32 update_nat_bits_flags(struct f2fs_super_block *,
 				struct f2fs_checkpoint *, u32);
 extern void write_nat_bits(struct f2fs_sb_info *, struct f2fs_super_block *,
 			struct f2fs_checkpoint *, int);
+extern unsigned int get_usable_seg_count(struct f2fs_sb_info *);
+extern bool is_usable_seg(struct f2fs_sb_info *, unsigned int);
 
 /* dump.c */
 struct dump_option {
diff --git a/fsck/mount.c b/fsck/mount.c
index 2f55768..8ebc5b0 100644
--- a/fsck/mount.c
+++ b/fsck/mount.c
@@ -30,6 +30,76 @@
 #define ACL_OTHER		(0x20)
 #endif
 
+static int get_device_idx(struct f2fs_sb_info *sbi, u_int32_t segno)
+{
+	block_t seg_start_blkaddr;
+	int i;
+
+	seg_start_blkaddr = SM_I(sbi)->main_blkaddr +
+				segno * DEFAULT_BLOCKS_PER_SEGMENT;
+	for (i = 0; i < c.ndevs; i++)
+		if (c.devices[i].start_blkaddr <= seg_start_blkaddr &&
+			c.devices[i].end_blkaddr > seg_start_blkaddr)
+			return i;
+	return 0;
+}
+
+#ifdef HAVE_LINUX_BLKZONED_H
+
+static int get_zone_idx_from_dev(struct f2fs_sb_info *sbi,
+					u_int32_t segno, u_int32_t dev_idx)
+{
+	block_t seg_start_blkaddr = START_BLOCK(sbi, segno);
+
+	return (seg_start_blkaddr - c.devices[dev_idx].start_blkaddr) >>
+			log_base_2(sbi->segs_per_sec * sbi->blocks_per_seg);
+}
+
+bool is_usable_seg(struct f2fs_sb_info *sbi, unsigned int segno)
+{
+	unsigned int secno = segno / sbi->segs_per_sec;
+	block_t seg_start = START_BLOCK(sbi, segno);
+	block_t blocks_per_sec = sbi->blocks_per_seg * sbi->segs_per_sec;
+	unsigned int dev_idx = get_device_idx(sbi, segno);
+	unsigned int zone_idx = get_zone_idx_from_dev(sbi, segno, dev_idx);
+	unsigned int sec_off = SM_I(sbi)->main_blkaddr >>
+						log_base_2(blocks_per_sec);
+
+	if (zone_idx < c.devices[dev_idx].nr_rnd_zones)
+		return true;
+
+	if (c.devices[dev_idx].zoned_model != F2FS_ZONED_HM)
+		return true;
+
+	return seg_start < ((sec_off + secno) * blocks_per_sec) +
+				c.devices[dev_idx].zone_cap_blocks[zone_idx];
+}
+
+unsigned int get_usable_seg_count(struct f2fs_sb_info *sbi)
+{
+	unsigned int i, usable_seg_count = 0;
+
+	for (i = 0; i < TOTAL_SEGS(sbi); i++)
+		if (is_usable_seg(sbi, i))
+			usable_seg_count++;
+
+	return usable_seg_count;
+}
+
+#else
+
+bool is_usable_seg(struct f2fs_sb_info *UNUSED(sbi), unsigned int UNUSED(segno))
+{
+	return true;
+}
+
+unsigned int get_usable_seg_count(struct f2fs_sb_info *sbi)
+{
+	return TOTAL_SEGS(sbi);
+}
+
+#endif
+
 u32 get_free_segments(struct f2fs_sb_info *sbi)
 {
 	u32 i, free_segs = 0;
@@ -37,7 +107,8 @@ u32 get_free_segments(struct f2fs_sb_info *sbi)
 	for (i = 0; i < TOTAL_SEGS(sbi); i++) {
 		struct seg_entry *se = get_seg_entry(sbi, i);
 
-		if (se->valid_blocks == 0x0 && !IS_CUR_SEGNO(sbi, i))
+		if (se->valid_blocks == 0x0 && !IS_CUR_SEGNO(sbi, i) &&
+							is_usable_seg(sbi, i))
 			free_segs++;
 	}
 	return free_segs;
@@ -2351,7 +2422,7 @@ void build_sit_area_bitmap(struct f2fs_sb_info *sbi)
 		memcpy(ptr, se->cur_valid_map, SIT_VBLOCK_MAP_SIZE);
 		ptr += SIT_VBLOCK_MAP_SIZE;
 
-		if (se->valid_blocks == 0x0) {
+		if (se->valid_blocks == 0x0 && is_usable_seg(sbi, segno)) {
 			if (le32_to_cpu(sbi->ckpt->cur_node_segno[0]) == segno ||
 				le32_to_cpu(sbi->ckpt->cur_data_segno[0]) == segno ||
 				le32_to_cpu(sbi->ckpt->cur_node_segno[1]) == segno ||
diff --git a/include/f2fs_fs.h b/include/f2fs_fs.h
index 3323be1..b6bd4a7 100644
--- a/include/f2fs_fs.h
+++ b/include/f2fs_fs.h
@@ -332,6 +332,7 @@ struct device_info {
 	u_int32_t nr_zones;
 	u_int32_t nr_rnd_zones;
 	size_t zone_blocks;
+	size_t *zone_cap_blocks;
 };
 
 typedef struct {
@@ -537,6 +538,7 @@ struct f2fs_configuration {
 	(void) (&_max1 == &_max2);		\
 	_max1 > _max2 ? _max1 : _max2; })
 
+#define round_up(x, y)		(((x) + (y) - 1) / (y))
 /*
  * Copied from fs/f2fs/f2fs.h
  */
@@ -1325,13 +1327,42 @@ blk_zone_cond_str(struct blk_zone *blkz)
 	return "Unknown-cond";
 }
 
-#define blk_zone_empty(z)	(blk_zone_cond(z) == BLK_ZONE_COND_EMPTY)
+/*
+ * Handle kernel zone capacity support
+ */
+#ifndef HAVE_BLK_ZONE_REP_V2
+#define BLK_ZONE_REP_CAPACITY   (1 << 0)
+struct blk_zone_v2 {
+	__u64   start;          /* Zone start sector */
+	__u64   len;            /* Zone length in number of sectors */
+	__u64   wp;             /* Zone write pointer position */
+	__u8    type;           /* Zone type */
+	__u8    cond;           /* Zone condition */
+	__u8    non_seq;        /* Non-sequential write resources active */
+	__u8    reset;          /* Reset write pointer recommended */
+	__u8    resv[4];
+	__u64   capacity;       /* Zone capacity in number of sectors */
+	__u8    reserved[24];
+};
+#define blk_zone blk_zone_v2
 
+struct blk_zone_report_v2 {
+	__u64   sector;
+	__u32   nr_zones;
+	__u32   flags;
+struct blk_zone zones[0];
+};
+#define blk_zone_report blk_zone_report_v2
+#endif /* HAVE_BLK_ZONE_REP_V2 */
+
+#define blk_zone_empty(z)	(blk_zone_cond(z) == BLK_ZONE_COND_EMPTY)
 #define blk_zone_sector(z)	(z)->start
 #define blk_zone_length(z)	(z)->len
 #define blk_zone_wp_sector(z)	(z)->wp
 #define blk_zone_need_reset(z)	(int)(z)->reset
 #define blk_zone_non_seq(z)	(int)(z)->non_seq
+#define blk_zone_capacity(z, f) ((f & BLK_ZONE_REP_CAPACITY) ? \
+					(z)->capacity : (z)->len)
 
 #endif
 
@@ -1343,6 +1374,7 @@ extern int f2fs_report_zones(int, report_zones_cb_t *, void *);
 extern int f2fs_check_zones(int);
 int f2fs_reset_zone(int, void *);
 extern int f2fs_reset_zones(int);
+extern uint32_t f2fs_get_usable_segments(struct f2fs_super_block *sb);
 
 #define SIZE_ALIGN(val, size)	((val) + (size) - 1) / (size)
 #define SEG_ALIGN(blks)		SIZE_ALIGN(blks, c.blks_per_seg)
@@ -1353,6 +1385,7 @@ static inline double get_best_overprovision(struct f2fs_super_block *sb)
 {
 	double reserved, ovp, candidate, end, diff, space;
 	double max_ovp = 0, max_space = 0;
+	u_int32_t usable_main_segs = f2fs_get_usable_segments(sb);
 
 	if (get_sb(segment_count_main) < 256) {
 		candidate = 10;
@@ -1366,9 +1399,9 @@ static inline double get_best_overprovision(struct f2fs_super_block *sb)
 
 	for (; candidate <= end; candidate += diff) {
 		reserved = (2 * (100 / candidate + 1) + 6) *
-						get_sb(segs_per_sec);
-		ovp = (get_sb(segment_count_main) - reserved) * candidate / 100;
-		space = get_sb(segment_count_main) - reserved - ovp;
+				round_up(usable_main_segs, get_sb(section_count));
+		ovp = (usable_main_segs - reserved) * candidate / 100;
+		space = usable_main_segs - reserved - ovp;
 		if (max_space < space) {
 			max_space = space;
 			max_ovp = candidate;
diff --git a/lib/libf2fs_io.c b/lib/libf2fs_io.c
index 1f597a9..138285d 100644
--- a/lib/libf2fs_io.c
+++ b/lib/libf2fs_io.c
@@ -784,6 +784,7 @@ int f2fs_finalize_device(void)
 			break;
 		}
 		free(c.devices[i].path);
+		free(c.devices[i].zone_cap_blocks);
 	}
 	close(c.kd);
 
diff --git a/lib/libf2fs_zoned.c b/lib/libf2fs_zoned.c
index efc687c..56c97d1 100644
--- a/lib/libf2fs_zoned.c
+++ b/lib/libf2fs_zoned.c
@@ -291,6 +291,13 @@ int f2fs_check_zones(int j)
 		return -ENOMEM;
 	}
 
+	dev->zone_cap_blocks = malloc(dev->nr_zones * sizeof(size_t));
+	if (!dev->zone_cap_blocks) {
+		ERR_MSG("No memory for zone capacity list.\n");
+		return -ENOMEM;
+	}
+	memset(dev->zone_cap_blocks, 0, (dev->nr_zones * sizeof(size_t)));
+
 	dev->nr_rnd_zones = 0;
 	sector = 0;
 	total_sectors = (dev->total_sectors * c.sector_size) >> 9;
@@ -335,10 +342,15 @@ int f2fs_check_zones(int j)
 				    blk_zone_cond_str(blkz),
 				    blk_zone_sector(blkz),
 				    blk_zone_length(blkz));
+				dev->zone_cap_blocks[n] =
+					blk_zone_length(blkz) >>
+					(F2FS_BLKSIZE_BITS - SECTOR_SHIFT);
 			} else {
 				DBG(2,
-				    "Zone %05u: type 0x%x (%s), cond 0x%x (%s), need_reset %d, "
-				    "non_seq %d, sector %llu, %llu sectors, wp sector %llu\n",
+				    "Zone %05u: type 0x%x (%s), cond 0x%x (%s),"
+				    " need_reset %d, non_seq %d, sector %llu,"
+				    " %llu sectors, capacity %llu,"
+				    " wp sector %llu\n",
 				    n,
 				    blk_zone_type(blkz),
 				    blk_zone_type_str(blkz),
@@ -348,7 +360,11 @@ int f2fs_check_zones(int j)
 				    blk_zone_non_seq(blkz),
 				    blk_zone_sector(blkz),
 				    blk_zone_length(blkz),
+				    blk_zone_capacity(blkz, rep->flags),
 				    blk_zone_wp_sector(blkz));
+				dev->zone_cap_blocks[n] =
+					blk_zone_capacity(blkz, rep->flags) >>
+					(F2FS_BLKSIZE_BITS - SECTOR_SHIFT);
 			}
 
 			sector = blk_zone_sector(blkz) + blk_zone_length(blkz);
@@ -473,6 +489,34 @@ out:
 	return ret;
 }
 
+uint32_t f2fs_get_usable_segments(struct f2fs_super_block *sb)
+{
+#ifdef HAVE_BLK_ZONE_REP_V2
+	int i, j;
+	uint32_t usable_segs = 0, zone_segs;
+
+	for (i = 0; i < c.ndevs; i++) {
+		if (c.devices[i].zoned_model != F2FS_ZONED_HM) {
+			usable_segs += c.devices[i].total_segments;
+			continue;
+		}
+		for (j = 0; j < c.devices[i].nr_zones; j++) {
+			zone_segs = c.devices[i].zone_cap_blocks[j] >>
+					get_sb(log_blocks_per_seg);
+			if (c.devices[i].zone_cap_blocks[j] %
+						DEFAULT_BLOCKS_PER_SEGMENT)
+				usable_segs += zone_segs + 1;
+			else
+				usable_segs += zone_segs;
+		}
+	}
+	usable_segs -= (get_sb(main_blkaddr) - get_sb(segment0_blkaddr)) >>
+						get_sb(log_blocks_per_seg);
+	return usable_segs;
+#endif
+	return get_sb(segment_count_main);
+}
+
 #else
 
 int f2fs_report_zone(int i, u_int64_t UNUSED(sector), void *UNUSED(blkzone))
@@ -527,5 +571,9 @@ int f2fs_reset_zones(int i)
 	return -1;
 }
 
+uint32_t f2fs_get_usable_segments(struct f2fs_super_block *sb)
+{
+	return get_sb(segment_count_main);
+}
 #endif
 
diff --git a/mkfs/f2fs_format.c b/mkfs/f2fs_format.c
index be19c36..7c5d1eb 100644
--- a/mkfs/f2fs_format.c
+++ b/mkfs/f2fs_format.c
@@ -425,13 +425,19 @@ static int f2fs_prepare_super_block(void)
 
 	set_sb(segment_count_main, get_sb(section_count) * c.segs_per_sec);
 
-	/* Let's determine the best reserved and overprovisioned space */
+	/*
+	 * Let's determine the best reserved and overprovisioned space.
+	 * For Zoned device, if zone capacity less than zone size, the segments
+	 * starting after the zone capacity are unusable in each zone. So get
+	 * overprovision ratio and reserved seg count based on avg usable
+	 * segs_per_sec.
+	 */
 	if (c.overprovision == 0)
 		c.overprovision = get_best_overprovision(sb);
 
 	c.reserved_segments =
-			(2 * (100 / c.overprovision + 1) + NR_CURSEG_TYPE)
-			* c.segs_per_sec;
+			(2 * (100 / c.overprovision + 1) + NR_CURSEG_TYPE) *
+			round_up(f2fs_get_usable_segments(sb), get_sb(section_count));
 
 	if (c.overprovision == 0 || c.total_segments < F2FS_MIN_SEGMENTS ||
 		(c.devices[0].total_sectors *
@@ -679,19 +685,28 @@ static int f2fs_write_check_point_pack(void)
 	set_cp(valid_block_count, 2 + c.quota_inum + c.quota_dnum +
 			c.lpf_inum + c.lpf_dnum);
 	set_cp(rsvd_segment_count, c.reserved_segments);
-	set_cp(overprov_segment_count, (get_sb(segment_count_main) -
+
+	/*
+	 * For zoned devices, if zone capacity less than zone size, get
+	 * overprovision segment count based on usable segments in the device.
+	 */
+	set_cp(overprov_segment_count, (f2fs_get_usable_segments(sb) -
 			get_cp(rsvd_segment_count)) *
 			c.overprovision / 100);
 	set_cp(overprov_segment_count, get_cp(overprov_segment_count) +
 			get_cp(rsvd_segment_count));
 
+	if (f2fs_get_usable_segments(sb) <= get_cp(overprov_segment_count)) {
+		MSG(0, "\tError: Not enough segments to create F2FS Volume\n");
+		goto free_nat_bits;
+	}
 	MSG(0, "Info: Overprovision ratio = %.3lf%%\n", c.overprovision);
 	MSG(0, "Info: Overprovision segments = %u (GC reserved = %u)\n",
 					get_cp(overprov_segment_count),
 					c.reserved_segments);
 
 	/* main segments - reserved segments - (node + data segments) */
-	set_cp(free_segment_count, get_sb(segment_count_main) - 6);
+	set_cp(free_segment_count, f2fs_get_usable_segments(sb) - 6);
 	set_cp(user_block_count, ((get_cp(free_segment_count) + 6 -
 			get_cp(overprov_segment_count)) * c.blks_per_seg));
 	/* cp page (2), data summaries (1), node summaries (3) */