I think this is a simple hack, so NAK.
It seems the ask is to support storage tiering, and we need to design a better
approach. Something like allocating a section per defined device priority.
On 07/23, Liao Yuanhong wrote:
> During the development process, we encounter the following two issues:
>
> 1.In a multi-device scenario, it's likely that two devices exhibit
> inconsistent performance, causing fluctuations in performance and making
> usage and testing inconvenient. Under normal circumstances, we hope to
> prioritize the use of the device with better performance and consider the
> device with poorer performance when space is insufficient. Using reuse
> mode can solve some of the issues, but tests reveal that the fragmentation
> degree in reuse mode is significantly higher than in default mode, which
> poses performance risks.
>
> 2.We need to examine the differences in data placement for different sizes
> of storage devices under specific write patterns. Currently, this
> comparison can only be made by switching storage devices.
>
> To address the above issues, I am considering adding a last_secno node. By
> adjusting this node, we can change the end of the addressing in
> get_new_segment so that it readdresses from 0 once the set value is
> reached. The default value of the node is the maximum number of sections
> for the current storage device, so making no modifications will not affect
> the current logic. If the space before the set value is already filled with
> valid data, it will normally write into the free area after the set value.
>
> Signed-off-by: Liao Yuanhong <liaoyuanh...@vivo.com>
> ---
> Documentation/ABI/testing/sysfs-fs-f2fs | 8 ++++++++
> fs/f2fs/f2fs.h | 1 +
> fs/f2fs/segment.c | 4 ++--
> fs/f2fs/super.c | 1 +
> fs/f2fs/sysfs.c | 9 +++++++++
> 5 files changed, 21 insertions(+), 2 deletions(-)
>
> diff --git a/Documentation/ABI/testing/sysfs-fs-f2fs
> b/Documentation/ABI/testing/sysfs-fs-f2fs
> index c2a233f2a085..e5ad8dc70cb6 100644
> --- a/Documentation/ABI/testing/sysfs-fs-f2fs
> +++ b/Documentation/ABI/testing/sysfs-fs-f2fs
> @@ -870,3 +870,11 @@ Description: This threshold is used to control
> triggering garbage collection whi
> reserved section before preallocating on pinned file.
> By default, the value is ovp_sections, especially, for zoned
> ufs, the
> value is 1.
> +
> +What: /sys/fs/f2fs/<disk>/last_secno
> +Date: July 2025
> +Contact: "Yuanhong Liao" <liaoyuanh...@vivo.com>
> +Description: This node is used to adjust the addressing end of f2fs when
> writing.
> + The default value is the maximum number of sections in the
> storage. If the
> + frontend space is full after adjusting the address, it will
> still address to
> + the back of the address until reaching the end of the memory.
> \ No newline at end of file
> diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> index e6dcd7e6f47c..7d93d8671033 100644
> --- a/fs/f2fs/f2fs.h
> +++ b/fs/f2fs/f2fs.h
> @@ -1802,6 +1802,7 @@ struct f2fs_sb_info {
> spinlock_t dev_lock; /* protect dirty_device */
> bool aligned_blksize; /* all devices has the same
> logical blksize */
> unsigned int first_seq_zone_segno; /* first segno in sequential
> zone */
> + unsigned int last_secno; /* for adjust the end of target
> device */
>
> /* For write statistics */
> u64 sectors_written_start;
> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> index cc82d42ef14c..192619bb2034 100644
> --- a/fs/f2fs/segment.c
> +++ b/fs/f2fs/segment.c
> @@ -2811,7 +2811,7 @@ static int get_new_segment(struct f2fs_sb_info *sbi,
> secno = find_next_zero_bit(free_i->free_secmap, MAIN_SECS(sbi), hint);
>
> #ifdef CONFIG_BLK_DEV_ZONED
> - if (secno >= MAIN_SECS(sbi) && f2fs_sb_has_blkzoned(sbi)) {
> + if (secno >= sbi->last_secno && f2fs_sb_has_blkzoned(sbi)) {
> /* Write only to sequential zones */
> if (sbi->blkzone_alloc_policy == BLKZONE_ALLOC_ONLY_SEQ) {
> hint = GET_SEC_FROM_SEG(sbi, sbi->first_seq_zone_segno);
> @@ -2827,7 +2827,7 @@ static int get_new_segment(struct f2fs_sb_info *sbi,
> }
> #endif
>
> - if (secno >= MAIN_SECS(sbi)) {
> + if (secno >= sbi->last_secno) {
> secno = find_first_zero_bit(free_i->free_secmap,
> MAIN_SECS(sbi));
> if (secno >= MAIN_SECS(sbi)) {
> diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c
> index 30c038413040..5b7de0734da7 100644
> --- a/fs/f2fs/super.c
> +++ b/fs/f2fs/super.c
> @@ -3859,6 +3859,7 @@ static int sanity_check_raw_super(struct f2fs_sb_info
> *sbi,
> segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
> secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
> total_sections = le32_to_cpu(raw_super->section_count);
> + sbi->last_secno = total_sections;
>
> /* blocks_per_seg should be 512, given the above check */
> blocks_per_seg = BIT(le32_to_cpu(raw_super->log_blocks_per_seg));
> diff --git a/fs/f2fs/sysfs.c b/fs/f2fs/sysfs.c
> index bdef926b3377..a4ab5c38e883 100644
> --- a/fs/f2fs/sysfs.c
> +++ b/fs/f2fs/sysfs.c
> @@ -649,6 +649,13 @@ static ssize_t __sbi_store(struct f2fs_attr *a,
> return count;
> }
>
> + if (!strcmp(a->attr.name, "last_secno")) {
> + if (t < 0 || t > MAIN_SECS(sbi))
> + return -EINVAL;
> + sbi->last_secno = t;
> + return count;
> + }
> +
> #ifdef CONFIG_F2FS_IOSTAT
> if (!strcmp(a->attr.name, "iostat_enable")) {
> sbi->iostat_enable = !!t;
> @@ -1122,6 +1129,7 @@ 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(last_secno);
> #ifdef CONFIG_F2FS_IOSTAT
> F2FS_SBI_GENERAL_RW_ATTR(iostat_enable);
> F2FS_SBI_GENERAL_RW_ATTR(iostat_period_ms);
> @@ -1285,6 +1293,7 @@ static struct attribute *f2fs_attrs[] = {
> ATTR_LIST(discard_idle_interval),
> ATTR_LIST(gc_idle_interval),
> ATTR_LIST(umount_discard_timeout),
> + ATTR_LIST(last_secno),
> #ifdef CONFIG_F2FS_IOSTAT
> ATTR_LIST(iostat_enable),
> ATTR_LIST(iostat_period_ms),
> --
> 2.34.1
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
