f2fs: Use allocate_section_policy to control write priority in multi-devices setups

Introduces two new sys nodes: allocate_section_hint and
allocate_section_policy. The allocate_section_hint identifies the boundary
between devices, measured in sections; it defaults to the end of the device
for single storage setups, and the end of the first device for multiple
storage setups. The allocate_section_policy determines the write strategy,
with a default value of 0 for normal sequential write strategy. A value of
1 prioritizes writes before the allocate_section_hint, while a value of 2
prioritizes writes after it.

This strategy addresses the issue where, despite F2FS supporting multiple
devices, SOC vendors lack multi-devices support (currently only supporting
zoned devices). As a workaround, multiple storage devices are mapped to a
single dm device. Both this workaround and the F2FS multi-devices solution
may require prioritizing writing to certain devices, such as a device with
better performance or when switching is needed due to performance
degradation near a device's end. For scenarios with more than two devices,
sort them at mount time to utilize this feature.

When using this feature with a single storage device, it has almost no
impact. However, for configurations where multiple storage devices are
mapped to the same dm device using F2FS, utilizing this feature can provide
some optimization benefits. Therefore, I believe it should not be limited
to just multi-devices usage.

Signed-off-by: Liao Yuanhong <liaoyuanhong@vivo.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

6 files changed, 74 insertions, 0 deletions

diff --git a/Documentation/ABI/testing/sysfs-fs-f2fs b/Documentation/ABI/testing/sysfs-fs-f2fs
index ee3acc8c2cb8..b590809869ca 100644
--- a/Documentation/ABI/testing/sysfs-fs-f2fs
+++ b/Documentation/ABI/testing/sysfs-fs-f2fs
@@ -911,3 +911,25 @@ Description:	Used to adjust the BG_GC priority when pending IO, with a default v
 		bggc_io_aware = 1   skip background GC if there is pending read IO
 		bggc_io_aware = 2   don't aware IO for background GC
 		==================  ======================================================
+
+What:		/sys/fs/f2fs/<disk>/allocate_section_hint
+Date:		August 2025
+Contact:	"Liao Yuanhong" <liaoyuanhong@vivo.com>
+Description:	Indicates the hint section between the first device and others in multi-devices
+		setup. It defaults to the end of the first device in sections. For a single storage
+		device, it defaults to the total number of sections. It can be manually set to match
+		scenarios where multi-devices are mapped to the same dm device.
+
+What:		/sys/fs/f2fs/<disk>/allocate_section_policy
+Date:		August 2025
+Contact:	"Liao Yuanhong" <liaoyuanhong@vivo.com>
+Description:	Controls write priority in multi-devices setups. A value of 0 means normal writing.
+		A value of 1 prioritizes writing to devices before the allocate_section_hint. A value of 2
+		prioritizes writing to devices after the allocate_section_hint. The default is 0.
+
+		===========================  ==========================================================
+		value					     description
+		allocate_section_policy = 0  Normal writing
+		allocate_section_policy = 1  Prioritize writing to section before allocate_section_hint
+		allocate_section_policy = 2  Prioritize writing to section after allocate_section_hint
+		===========================  ==========================================================
diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 69ff43f81c88..9d3bc9633c1d 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -162,6 +162,12 @@ enum bggc_io_aware_policy {
 	AWARE_NONE,			/* don't aware IO for background GC */
 };
 
+enum device_allocation_policy {
+	ALLOCATE_FORWARD_NOHINT,
+	ALLOCATE_FORWARD_WITHIN_HINT,
+	ALLOCATE_FORWARD_FROM_HINT,
+};
+
 /*
  * An implementation of an rwsem that is explicitly unfair to readers. This
  * prevents priority inversion when a low-priority reader acquires the read lock
@@ -1850,6 +1856,8 @@ struct f2fs_sb_info {
 	bool aligned_blksize;			/* all devices has the same logical blksize */
 	unsigned int first_seq_zone_segno;	/* first segno in sequential zone */
 	unsigned int bggc_io_aware;		/* For adjust the BG_GC priority when pending IO */
+	unsigned int allocate_section_hint;	/* the boundary position between devices */
+	unsigned int allocate_section_policy;	/* determine the section writing priority */
 
 	/* For write statistics */
 	u64 sectors_written_start;
diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
index c0f209f74688..ed3acbfc83ca 100644
--- a/fs/f2fs/gc.c
+++ b/fs/f2fs/gc.c
@@ -2182,6 +2182,8 @@ static void update_fs_metadata(struct f2fs_sb_info *sbi, int secs)
 	SM_I(sbi)->segment_count = (int)SM_I(sbi)->segment_count + segs;
 	MAIN_SEGS(sbi) = (int)MAIN_SEGS(sbi) + segs;
 	MAIN_SECS(sbi) += secs;
+	if (sbi->allocate_section_hint > MAIN_SECS(sbi))
+		sbi->allocate_section_hint = MAIN_SECS(sbi);
 	FREE_I(sbi)->free_sections = (int)FREE_I(sbi)->free_sections + secs;
 	FREE_I(sbi)->free_segments = (int)FREE_I(sbi)->free_segments + segs;
 	F2FS_CKPT(sbi)->user_block_count = cpu_to_le64(user_block_count + blks);
@@ -2189,6 +2191,9 @@ static void update_fs_metadata(struct f2fs_sb_info *sbi, int secs)
 	if (f2fs_is_multi_device(sbi)) {
 		int last_dev = sbi->s_ndevs - 1;
 
+		sbi->allocate_section_hint = FDEV(0).total_segments /
+					SEGS_PER_SEC(sbi);
+
 		FDEV(last_dev).total_segments =
 				(int)FDEV(last_dev).total_segments + segs;
 		FDEV(last_dev).end_blk =
diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 04b0a3c1804d..e4d71755a60f 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -2774,6 +2774,8 @@ static int get_new_segment(struct f2fs_sb_info *sbi,
 	unsigned int total_zones = MAIN_SECS(sbi) / sbi->secs_per_zone;
 	unsigned int hint = GET_SEC_FROM_SEG(sbi, *newseg);
 	unsigned int old_zoneno = GET_ZONE_FROM_SEG(sbi, *newseg);
+	unsigned int alloc_policy = sbi->allocate_section_policy;
+	unsigned int alloc_hint = sbi->allocate_section_hint;
 	bool init = true;
 	int i;
 	int ret = 0;
@@ -2807,6 +2809,21 @@ static int get_new_segment(struct f2fs_sb_info *sbi,
 	}
 #endif
 
+	/*
+	 * Prevent allocate_section_hint from exceeding MAIN_SECS()
+	 * due to desynchronization.
+	 */
+	if (alloc_policy != ALLOCATE_FORWARD_NOHINT &&
+		alloc_hint > MAIN_SECS(sbi))
+		alloc_hint = MAIN_SECS(sbi);
+
+	if (alloc_policy == ALLOCATE_FORWARD_FROM_HINT &&
+		hint < alloc_hint)
+		hint = alloc_hint;
+	else if (alloc_policy == ALLOCATE_FORWARD_WITHIN_HINT &&
+			hint >= alloc_hint)
+		hint = 0;
+
 find_other_zone:
 	secno = find_next_zero_bit(free_i->free_secmap, MAIN_SECS(sbi), hint);
 
diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c
index 07f6c8cac07a..1e0678e37a30 100644
--- a/fs/f2fs/super.c
+++ b/fs/f2fs/super.c
@@ -4234,6 +4234,8 @@ static void init_sb_info(struct f2fs_sb_info *sbi)
 	sbi->total_node_count = SEGS_TO_BLKS(sbi,
 			((le32_to_cpu(raw_super->segment_count_nat) / 2) *
 			NAT_ENTRY_PER_BLOCK));
+	sbi->allocate_section_hint = le32_to_cpu(raw_super->section_count);
+	sbi->allocate_section_policy = ALLOCATE_FORWARD_NOHINT;
 	F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
 	F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
 	F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
@@ -4748,6 +4750,8 @@ static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
 					SEGS_TO_BLKS(sbi,
 					FDEV(i).total_segments) - 1 +
 					le32_to_cpu(raw_super->segment0_blkaddr);
+				sbi->allocate_section_hint = FDEV(i).total_segments /
+							SEGS_PER_SEC(sbi);
 			} else {
 				FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
 				FDEV(i).end_blk = FDEV(i).start_blk +
diff --git a/fs/f2fs/sysfs.c b/fs/f2fs/sysfs.c
index 86ebe97c256d..081b5df0e664 100644
--- a/fs/f2fs/sysfs.c
+++ b/fs/f2fs/sysfs.c
@@ -889,6 +889,20 @@ out:
 		return count;
 	}
 
+	if (!strcmp(a->attr.name, "allocate_section_hint")) {
+		if (t < 0 || t > MAIN_SECS(sbi))
+			return -EINVAL;
+		sbi->allocate_section_hint = t;
+		return count;
+	}
+
+	if (!strcmp(a->attr.name, "allocate_section_policy")) {
+		if (t < ALLOCATE_FORWARD_NOHINT || t > ALLOCATE_FORWARD_FROM_HINT)
+			return -EINVAL;
+		sbi->allocate_section_policy = t;
+		return count;
+	}
+
 	*ui = (unsigned int)t;
 
 	return count;
@@ -1161,6 +1175,8 @@ F2FS_SBI_GENERAL_RW_ATTR(max_victim_search);
 F2FS_SBI_GENERAL_RW_ATTR(migration_granularity);
 F2FS_SBI_GENERAL_RW_ATTR(migration_window_granularity);
 F2FS_SBI_GENERAL_RW_ATTR(dir_level);
+F2FS_SBI_GENERAL_RW_ATTR(allocate_section_hint);
+F2FS_SBI_GENERAL_RW_ATTR(allocate_section_policy);
 #ifdef CONFIG_F2FS_IOSTAT
 F2FS_SBI_GENERAL_RW_ATTR(iostat_enable);
 F2FS_SBI_GENERAL_RW_ATTR(iostat_period_ms);
@@ -1398,6 +1414,8 @@ static struct attribute *f2fs_attrs[] = {
 	ATTR_LIST(max_read_extent_count),
 	ATTR_LIST(carve_out),
 	ATTR_LIST(reserved_pin_section),
+	ATTR_LIST(allocate_section_hint),
+	ATTR_LIST(allocate_section_policy),
 	NULL,
 };
 ATTRIBUTE_GROUPS(f2fs);