Linux-f2fs-devel Archive on lore.kernel.org
 help / color / Atom feed
* [f2fs-dev] [PATCH 0/2] f2fs-tools: add zone-capacity support
@ 2020-07-02 15:54 Aravind Ramesh
  2020-07-02 15:54 ` [f2fs-dev] [PATCH 1/2] mkfs.f2fs: zns " Aravind Ramesh
  2020-07-02 15:54 ` [f2fs-dev] [PATCH 2/2] fsck.f2fs: validate free seg count on zns device Aravind Ramesh
  0 siblings, 2 replies; 9+ messages in thread
From: Aravind Ramesh @ 2020-07-02 15:54 UTC (permalink / raw)
  To: jaegeuk, yuchao0, linux-f2fs-devel; +Cc: Aravind Ramesh, matias.bjorling

NVM Express Zoned Namespace command set specification allows host software
to communicate with a NVM subsystem using zones. ZNS defines a host-managed
zoned block device model for NVMe devices. It divides the logical address
space of a namespace into zones. Each zone provides a LBA range that shall
be written sequentially. An explicit reset of zone is needed to write to
the zone again.

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.
This patchset implements support for ZNS devices with a zone-capacity
that is less that the device zone-size.

Add checks in mkfs and fsck to consider zone-capacity and calculate the
overprovision ratio and free_user_blocks based on the usable segments
in a zone.

This series is based on the git tree
git://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs-tools.git branch dev

Aravind Ramesh (2):
  mkfs.f2fs: zns zone-capacity support.
  fsck.f2fs: validate free seg count on zns device

 configure.ac        |  4 +++
 fsck/fsck.c         |  5 +--
 fsck/fsck.h         |  2 ++
 fsck/mount.c        | 75 +++++++++++++++++++++++++++++++++++++++++++--
 include/f2fs_fs.h   | 40 +++++++++++++++++++++---
 lib/libf2fs_io.c    |  1 +
 lib/libf2fs_zoned.c | 51 ++++++++++++++++++++++++++++--
 mkfs/f2fs_format.c  | 26 +++++++++++++---
 8 files changed, 189 insertions(+), 15 deletions(-)

-- 
2.19.1



_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

^ permalink raw reply	[flat|nested] 9+ messages in thread

* [f2fs-dev] [PATCH 1/2] mkfs.f2fs: zns zone-capacity support.
  2020-07-02 15:54 [f2fs-dev] [PATCH 0/2] f2fs-tools: add zone-capacity support Aravind Ramesh
@ 2020-07-02 15:54 ` Aravind Ramesh
  2020-07-17  7:28   ` Chao Yu
  2020-07-02 15:54 ` [f2fs-dev] [PATCH 2/2] fsck.f2fs: validate free seg count on zns device Aravind Ramesh
  1 sibling, 1 reply; 9+ messages in thread
From: Aravind Ramesh @ 2020-07-02 15:54 UTC (permalink / raw)
  To: jaegeuk, yuchao0, linux-f2fs-devel; +Cc: Aravind Ramesh, matias.bjorling

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. 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.

Signed-off-by: Aravind Ramesh <aravind.ramesh@wdc.com>
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
---
 configure.ac        |  4 ++++
 include/f2fs_fs.h   | 40 +++++++++++++++++++++++++++++++----
 lib/libf2fs_io.c    |  1 +
 lib/libf2fs_zoned.c | 51 +++++++++++++++++++++++++++++++++++++++++++--
 mkfs/f2fs_format.c  | 26 ++++++++++++++++++-----
 5 files changed, 111 insertions(+), 11 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/include/f2fs_fs.h b/include/f2fs_fs.h
index 709bfd8..40165ed 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 {
@@ -1324,13 +1325,42 @@ 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
 
@@ -1342,6 +1372,7 @@ 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)
@@ -1352,6 +1383,7 @@ 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;
@@ -1365,9 +1397,9 @@ 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;
+				(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..f98fcdb 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 +489,33 @@ 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 +570,9 @@ 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 4999cac..74a81c8 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) *
+			(f2fs_get_usable_segments(sb) / get_sb(section_count));
 
 	if (c.overprovision == 0 || c.total_segments < F2FS_MIN_SEGMENTS ||
 		(c.devices[0].total_sectors *
@@ -672,19 +678,29 @@ 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(rsvd_segment_count) +
+					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) */
-- 
2.19.1



_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

^ permalink raw reply	[flat|nested] 9+ messages in thread

* [f2fs-dev] [PATCH 2/2] fsck.f2fs: validate free seg count on zns device
  2020-07-02 15:54 [f2fs-dev] [PATCH 0/2] f2fs-tools: add zone-capacity support Aravind Ramesh
  2020-07-02 15:54 ` [f2fs-dev] [PATCH 1/2] mkfs.f2fs: zns " Aravind Ramesh
@ 2020-07-02 15:54 ` Aravind Ramesh
  2020-07-13 19:07   ` Jaegeuk Kim
  2020-07-17  7:37   ` Chao Yu
  1 sibling, 2 replies; 9+ messages in thread
From: Aravind Ramesh @ 2020-07-02 15:54 UTC (permalink / raw)
  To: jaegeuk, yuchao0, linux-f2fs-devel; +Cc: Aravind Ramesh, matias.bjorling

NVMe Zoned Namespace devices can have zone-capacity less than zone-size.
Zone-capacity indicates the number of usable blocks in a 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.

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>
---
 fsck/fsck.c  |  5 ++--
 fsck/fsck.h  |  2 ++
 fsck/mount.c | 75 ++++++++++++++++++++++++++++++++++++++++++++++++++--
 3 files changed, 78 insertions(+), 4 deletions(-)

diff --git a/fsck/fsck.c b/fsck/fsck.c
index e110f3d..ba2340d 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 bc6a435..e86730c 100644
--- a/fsck/fsck.h
+++ b/fsck/fsck.h
@@ -224,6 +224,8 @@ 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 d0f2eab..72ca0cb 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 *sbi, unsigned int 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;
@@ -2337,7 +2408,7 @@ 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 ||
-- 
2.19.1



_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [f2fs-dev] [PATCH 2/2] fsck.f2fs: validate free seg count on zns device
  2020-07-02 15:54 ` [f2fs-dev] [PATCH 2/2] fsck.f2fs: validate free seg count on zns device Aravind Ramesh
@ 2020-07-13 19:07   ` Jaegeuk Kim
  2020-07-15 19:25     ` Aravind Ramesh
  2020-07-17  7:37   ` Chao Yu
  1 sibling, 1 reply; 9+ messages in thread
From: Jaegeuk Kim @ 2020-07-13 19:07 UTC (permalink / raw)
  To: Aravind Ramesh; +Cc: matias.bjorling, linux-f2fs-devel

Hi Aravind,

On 07/02, Aravind Ramesh wrote:
> NVMe Zoned Namespace devices can have zone-capacity less than zone-size.
> Zone-capacity indicates the number of usable blocks in a 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.
> 
> 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>
> ---
>  fsck/fsck.c  |  5 ++--
>  fsck/fsck.h  |  2 ++
>  fsck/mount.c | 75 ++++++++++++++++++++++++++++++++++++++++++++++++++--
>  3 files changed, 78 insertions(+), 4 deletions(-)
> 
> diff --git a/fsck/fsck.c b/fsck/fsck.c
> index e110f3d..ba2340d 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 bc6a435..e86730c 100644
> --- a/fsck/fsck.h
> +++ b/fsck/fsck.h
> @@ -224,6 +224,8 @@ 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 d0f2eab..72ca0cb 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 *sbi, unsigned int segno)

I've applied the below small change to avoid Android build error.

--- a/fsck/mount.c
+++ b/fsck/mount.c
@@ -88,7 +88,7 @@ unsigned int get_usable_seg_count(struct f2fs_sb_info *sbi)

 #else

-bool is_usable_seg(struct f2fs_sb_info *sbi, unsigned int segno)
+bool is_usable_seg(struct f2fs_sb_info *UNUSED(sbi), unsigned int UNUSED(segno))
 {
        return true;
 }

Thanks,

> +{
> +	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;
> @@ -2337,7 +2408,7 @@ 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 ||
> -- 
> 2.19.1


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [f2fs-dev] [PATCH 2/2] fsck.f2fs: validate free seg count on zns device
  2020-07-13 19:07   ` Jaegeuk Kim
@ 2020-07-15 19:25     ` Aravind Ramesh
  0 siblings, 0 replies; 9+ messages in thread
From: Aravind Ramesh @ 2020-07-15 19:25 UTC (permalink / raw)
  To: Jaegeuk Kim; +Cc: Matias Bjorling, linux-f2fs-devel



> -----Original Message-----
> From: Jaegeuk Kim <jaegeuk@kernel.org>
> Sent: Tuesday, July 14, 2020 12:38 AM
> To: Aravind Ramesh <Aravind.Ramesh@wdc.com>
> Cc: yuchao0@huawei.com; linux-f2fs-devel@lists.sourceforge.net; Shinichiro
> Kawasaki <shinichiro.kawasaki@wdc.com>; Matias Bjorling
> <Matias.Bjorling@wdc.com>
> Subject: Re: [PATCH 2/2] fsck.f2fs: validate free seg count on zns device
> 
> Hi Aravind,
> 
> On 07/02, Aravind Ramesh wrote:
> > NVMe Zoned Namespace devices can have zone-capacity less than zone-size.
> > Zone-capacity indicates the number of usable blocks in a 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.
> >
> > 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>
> > ---
> >  fsck/fsck.c  |  5 ++--
> >  fsck/fsck.h  |  2 ++
> >  fsck/mount.c | 75
> > ++++++++++++++++++++++++++++++++++++++++++++++++++--
> >  3 files changed, 78 insertions(+), 4 deletions(-)
> >
> > diff --git a/fsck/fsck.c b/fsck/fsck.c index e110f3d..ba2340d 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 bc6a435..e86730c 100644
> > --- a/fsck/fsck.h
> > +++ b/fsck/fsck.h
> > @@ -224,6 +224,8 @@ 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 d0f2eab..72ca0cb 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 *sbi, unsigned int segno)
> 
> I've applied the below small change to avoid Android build error.
> 
> --- a/fsck/mount.c
> +++ b/fsck/mount.c
> @@ -88,7 +88,7 @@ unsigned int get_usable_seg_count(struct f2fs_sb_info *sbi)
> 
>  #else
> 
> -bool is_usable_seg(struct f2fs_sb_info *sbi, unsigned int segno)
> +bool is_usable_seg(struct f2fs_sb_info *UNUSED(sbi), unsigned int
> +UNUSED(segno))
>  {
>         return true;
>  }
> 
> Thanks,

Ok. 

Thank you,
Aravind

> 
> > +{
> > +	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;
> > @@ -2337,7 +2408,7 @@ 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 ||
> > --
> > 2.19.1


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [f2fs-dev] [PATCH 1/2] mkfs.f2fs: zns zone-capacity support.
  2020-07-02 15:54 ` [f2fs-dev] [PATCH 1/2] mkfs.f2fs: zns " Aravind Ramesh
@ 2020-07-17  7:28   ` Chao Yu
  2020-07-17  8:36     ` Aravind Ramesh
  0 siblings, 1 reply; 9+ messages in thread
From: Chao Yu @ 2020-07-17  7:28 UTC (permalink / raw)
  To: Aravind Ramesh, jaegeuk, linux-f2fs-devel; +Cc: matias.bjorling

On 2020/7/2 23:54, Aravind Ramesh wrote:
> 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. 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.
> 
> Signed-off-by: Aravind Ramesh <aravind.ramesh@wdc.com>
> Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
> ---
>  configure.ac        |  4 ++++
>  include/f2fs_fs.h   | 40 +++++++++++++++++++++++++++++++----
>  lib/libf2fs_io.c    |  1 +
>  lib/libf2fs_zoned.c | 51 +++++++++++++++++++++++++++++++++++++++++++--
>  mkfs/f2fs_format.c  | 26 ++++++++++++++++++-----
>  5 files changed, 111 insertions(+), 11 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/include/f2fs_fs.h b/include/f2fs_fs.h
> index 709bfd8..40165ed 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 {
> @@ -1324,13 +1325,42 @@ 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
>  
> @@ -1342,6 +1372,7 @@ 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)
> @@ -1352,6 +1383,7 @@ 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;
> @@ -1365,9 +1397,9 @@ 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;
> +				(usable_main_segs / get_sb(section_count));

It looks segs_per_sec becomes an average value in whole zns device, I'm not sure
whether calculating with divide round down way is safe enough to reserve space,
as it may cut several segments, which may be needed during foreground GC, so I
suggest to use DIV_ROUND_UP() here to avoid boundary issue.

> +		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..f98fcdb 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 +489,33 @@ 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 +570,9 @@ 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 4999cac..74a81c8 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) *
> +			(f2fs_get_usable_segments(sb) / get_sb(section_count));

Ditto, DIV_ROUND_UP()

>  
>  	if (c.overprovision == 0 || c.total_segments < F2FS_MIN_SEGMENTS ||
>  		(c.devices[0].total_sectors *
> @@ -672,19 +678,29 @@ 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(rsvd_segment_count) +

equal is not allowed as well?

> +					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) */
> 


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [f2fs-dev] [PATCH 2/2] fsck.f2fs: validate free seg count on zns device
  2020-07-02 15:54 ` [f2fs-dev] [PATCH 2/2] fsck.f2fs: validate free seg count on zns device Aravind Ramesh
  2020-07-13 19:07   ` Jaegeuk Kim
@ 2020-07-17  7:37   ` Chao Yu
  2020-07-17  8:36     ` Aravind Ramesh
  1 sibling, 1 reply; 9+ messages in thread
From: Chao Yu @ 2020-07-17  7:37 UTC (permalink / raw)
  To: Aravind Ramesh, jaegeuk, linux-f2fs-devel; +Cc: matias.bjorling

On 2020/7/2 23:54, Aravind Ramesh wrote:
> NVMe Zoned Namespace devices can have zone-capacity less than zone-size.
> Zone-capacity indicates the number of usable blocks in a 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.
> 
> 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>

It's better to merge these two related patch into one, otherwise, it
looks good to me.

Thanks,


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [f2fs-dev] [PATCH 1/2] mkfs.f2fs: zns zone-capacity support.
  2020-07-17  7:28   ` Chao Yu
@ 2020-07-17  8:36     ` Aravind Ramesh
  0 siblings, 0 replies; 9+ messages in thread
From: Aravind Ramesh @ 2020-07-17  8:36 UTC (permalink / raw)
  To: Chao Yu, jaegeuk, linux-f2fs-devel; +Cc: Matias Bjorling


> -----Original Message-----
> From: Chao Yu <yuchao0@huawei.com>
> Sent: Friday, July 17, 2020 12:58 PM
> To: Aravind Ramesh <Aravind.Ramesh@wdc.com>; jaegeuk@kernel.org; linux-f2fs-
> devel@lists.sourceforge.net
> Cc: Shinichiro Kawasaki <shinichiro.kawasaki@wdc.com>; Matias Bjorling
> <Matias.Bjorling@wdc.com>
> Subject: Re: [PATCH 1/2] mkfs.f2fs: zns zone-capacity support.
> 
> On 2020/7/2 23:54, Aravind Ramesh wrote:
> > 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. 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.
> >
> > Signed-off-by: Aravind Ramesh <aravind.ramesh@wdc.com>
> > Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
> > ---
> >  configure.ac        |  4 ++++
> >  include/f2fs_fs.h   | 40 +++++++++++++++++++++++++++++++----
> >  lib/libf2fs_io.c    |  1 +
> >  lib/libf2fs_zoned.c | 51
> > +++++++++++++++++++++++++++++++++++++++++++--
> >  mkfs/f2fs_format.c  | 26 ++++++++++++++++++-----
> >  5 files changed, 111 insertions(+), 11 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/include/f2fs_fs.h b/include/f2fs_fs.h index 709bfd8..40165ed 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 {
> > @@ -1324,13 +1325,42 @@ 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];
> > +};

[snip...]

> > @@ -1352,6 +1383,7 @@ 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;
> > @@ -1365,9 +1397,9 @@ 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;
> > +				(usable_main_segs / get_sb(section_count));
> 
> It looks segs_per_sec becomes an average value in whole zns device, I'm not sure
> whether calculating with divide round down way is safe enough to reserve space, as
> it may cut several segments, which may be needed during foreground GC, so I
> suggest to use DIV_ROUND_UP() here to avoid boundary issue.

Yes, it tries to reflect the usable segs_per_sec rather than using a value based on zone-size,
which could, reserve more than the intended amount of segments.

I do see a round_down(x, y) in the f2fs-tools code. But no, round_up().

Does this look ok?
DIV_ROUND_UP(n, d) (((n) + (d) - 1) / (d))
(from kernel.h)

> 
> > +		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..f98fcdb 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
> > +489,33 @@ 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 +570,9 @@ 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
> > 4999cac..74a81c8 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) *
> > +			(f2fs_get_usable_segments(sb) / get_sb(section_count));
> 
> Ditto, DIV_ROUND_UP()

Ok

> 
> >
> >  	if (c.overprovision == 0 || c.total_segments < F2FS_MIN_SEGMENTS ||
> >  		(c.devices[0].total_sectors *
> > @@ -672,19 +678,29 @@ 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(rsvd_segment_count) +
> 
> equal is not allowed as well?

You are right, will make it " <= "

Thanks for the feedback,
Aravind
> 
> > +					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) */
> >


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [f2fs-dev] [PATCH 2/2] fsck.f2fs: validate free seg count on zns device
  2020-07-17  7:37   ` Chao Yu
@ 2020-07-17  8:36     ` Aravind Ramesh
  0 siblings, 0 replies; 9+ messages in thread
From: Aravind Ramesh @ 2020-07-17  8:36 UTC (permalink / raw)
  To: Chao Yu, jaegeuk, linux-f2fs-devel; +Cc: Matias Bjorling



> -----Original Message-----
> From: Chao Yu <yuchao0@huawei.com>
> Sent: Friday, July 17, 2020 1:07 PM
> To: Aravind Ramesh <Aravind.Ramesh@wdc.com>; jaegeuk@kernel.org; linux-f2fs-
> devel@lists.sourceforge.net
> Cc: Shinichiro Kawasaki <shinichiro.kawasaki@wdc.com>; Matias Bjorling
> <Matias.Bjorling@wdc.com>
> Subject: Re: [PATCH 2/2] fsck.f2fs: validate free seg count on zns device
> 
> On 2020/7/2 23:54, Aravind Ramesh wrote:
> > NVMe Zoned Namespace devices can have zone-capacity less than zone-size.
> > Zone-capacity indicates the number of usable blocks in a 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.
> >
> > 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>
> 
> It's better to merge these two related patch into one, otherwise, it looks good to me.

Ok, will merge it to one patch and send a V2.
Thanks for the review and feedback.

Aravind
> 
> Thanks,


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

^ permalink raw reply	[flat|nested] 9+ messages in thread

end of thread, back to index

Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-07-02 15:54 [f2fs-dev] [PATCH 0/2] f2fs-tools: add zone-capacity support Aravind Ramesh
2020-07-02 15:54 ` [f2fs-dev] [PATCH 1/2] mkfs.f2fs: zns " Aravind Ramesh
2020-07-17  7:28   ` Chao Yu
2020-07-17  8:36     ` Aravind Ramesh
2020-07-02 15:54 ` [f2fs-dev] [PATCH 2/2] fsck.f2fs: validate free seg count on zns device Aravind Ramesh
2020-07-13 19:07   ` Jaegeuk Kim
2020-07-15 19:25     ` Aravind Ramesh
2020-07-17  7:37   ` Chao Yu
2020-07-17  8:36     ` Aravind Ramesh

Linux-f2fs-devel Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-f2fs-devel/0 linux-f2fs-devel/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-f2fs-devel linux-f2fs-devel/ https://lore.kernel.org/linux-f2fs-devel \
		linux-f2fs-devel@lists.sourceforge.net
	public-inbox-index linux-f2fs-devel

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/net.sourceforge.lists.linux-f2fs-devel


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git