From: Jaegeuk Kim via Linux-f2fs-devel <linux-f2fs-devel@lists.sourceforge.net>
To: Chao Yu <chao@kernel.org>
Cc: linux-kernel@vger.kernel.org, linux-f2fs-devel@lists.sourceforge.net
Subject: Re: [f2fs-dev] [PATCH v2 1/2] f2fs: use per-log target_bitmap to improve lookup performace of ssr allocation
Date: Mon, 7 Oct 2024 16:52:14 +0000 [thread overview]
Message-ID: <ZwQRvmVoDr_O6vLH@google.com> (raw)
In-Reply-To: <20240411082354.1691820-1-chao@kernel.org>
Hi Chao,
This introduces another bitmap increasing memory footprint. Do we know how
much performance benefit with this?
On 04/11, Chao Yu wrote:
> After commit 899fee36fac0 ("f2fs: fix to avoid data corruption by
> forbidding SSR overwrite"), valid block bitmap of current openned
> segment is fixed, let's introduce a per-log bitmap instead of temp
> bitmap to avoid unnecessary calculation overhead whenever allocating
> free slot w/ SSR allocator.
>
> Signed-off-by: Chao Yu <chao@kernel.org>
> ---
> v2:
> - rebase to last dev-test branch.
> fs/f2fs/segment.c | 30 ++++++++++++++++++++++--------
> fs/f2fs/segment.h | 1 +
> 2 files changed, 23 insertions(+), 8 deletions(-)
>
> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> index 6474b7338e81..af716925db19 100644
> --- a/fs/f2fs/segment.c
> +++ b/fs/f2fs/segment.c
> @@ -2840,31 +2840,39 @@ static int new_curseg(struct f2fs_sb_info *sbi, int type, bool new_sec)
> return 0;
> }
>
> -static int __next_free_blkoff(struct f2fs_sb_info *sbi,
> - int segno, block_t start)
> +static void __get_segment_bitmap(struct f2fs_sb_info *sbi,
> + unsigned long *target_map,
> + int segno)
> {
> struct seg_entry *se = get_seg_entry(sbi, segno);
> int entries = SIT_VBLOCK_MAP_SIZE / sizeof(unsigned long);
> - unsigned long *target_map = SIT_I(sbi)->tmp_map;
> unsigned long *ckpt_map = (unsigned long *)se->ckpt_valid_map;
> unsigned long *cur_map = (unsigned long *)se->cur_valid_map;
> int i;
>
> for (i = 0; i < entries; i++)
> target_map[i] = ckpt_map[i] | cur_map[i];
> +}
> +
> +static int __next_free_blkoff(struct f2fs_sb_info *sbi, unsigned long *bitmap,
> + int segno, block_t start)
> +{
> + __get_segment_bitmap(sbi, bitmap, segno);
>
> - return __find_rev_next_zero_bit(target_map, BLKS_PER_SEG(sbi), start);
> + return __find_rev_next_zero_bit(bitmap, BLKS_PER_SEG(sbi), start);
> }
>
> static int f2fs_find_next_ssr_block(struct f2fs_sb_info *sbi,
> - struct curseg_info *seg)
> + struct curseg_info *seg)
> {
> - return __next_free_blkoff(sbi, seg->segno, seg->next_blkoff + 1);
> + return __find_rev_next_zero_bit(seg->target_map,
> + BLKS_PER_SEG(sbi), seg->next_blkoff + 1);
> }
>
> bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno)
> {
> - return __next_free_blkoff(sbi, segno, 0) < BLKS_PER_SEG(sbi);
> + return __next_free_blkoff(sbi, SIT_I(sbi)->tmp_map, segno, 0) <
> + BLKS_PER_SEG(sbi);
> }
>
> /*
> @@ -2890,7 +2898,8 @@ static int change_curseg(struct f2fs_sb_info *sbi, int type)
>
> reset_curseg(sbi, type, 1);
> curseg->alloc_type = SSR;
> - curseg->next_blkoff = __next_free_blkoff(sbi, curseg->segno, 0);
> + curseg->next_blkoff = __next_free_blkoff(sbi, curseg->target_map,
> + curseg->segno, 0);
>
> sum_page = f2fs_get_sum_page(sbi, new_segno);
> if (IS_ERR(sum_page)) {
> @@ -4635,6 +4644,10 @@ static int build_curseg(struct f2fs_sb_info *sbi)
> sizeof(struct f2fs_journal), GFP_KERNEL);
> if (!array[i].journal)
> return -ENOMEM;
> + array[i].target_map = f2fs_kzalloc(sbi, SIT_VBLOCK_MAP_SIZE,
> + GFP_KERNEL);
> + if (!array[i].target_map)
> + return -ENOMEM;
> if (i < NR_PERSISTENT_LOG)
> array[i].seg_type = CURSEG_HOT_DATA + i;
> else if (i == CURSEG_COLD_DATA_PINNED)
> @@ -5453,6 +5466,7 @@ static void destroy_curseg(struct f2fs_sb_info *sbi)
> for (i = 0; i < NR_CURSEG_TYPE; i++) {
> kfree(array[i].sum_blk);
> kfree(array[i].journal);
> + kfree(array[i].target_map);
> }
> kfree(array);
> }
> diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
> index e1c0f418aa11..10f3e44f036f 100644
> --- a/fs/f2fs/segment.h
> +++ b/fs/f2fs/segment.h
> @@ -292,6 +292,7 @@ struct curseg_info {
> struct f2fs_summary_block *sum_blk; /* cached summary block */
> struct rw_semaphore journal_rwsem; /* protect journal area */
> struct f2fs_journal *journal; /* cached journal info */
> + unsigned long *target_map; /* bitmap for SSR allocator */
> unsigned char alloc_type; /* current allocation type */
> unsigned short seg_type; /* segment type like CURSEG_XXX_TYPE */
> unsigned int segno; /* current segment number */
> --
> 2.40.1
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
WARNING: multiple messages have this Message-ID (diff)
From: Jaegeuk Kim <jaegeuk@kernel.org>
To: Chao Yu <chao@kernel.org>
Cc: linux-f2fs-devel@lists.sourceforge.net, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 1/2] f2fs: use per-log target_bitmap to improve lookup performace of ssr allocation
Date: Mon, 7 Oct 2024 16:52:14 +0000 [thread overview]
Message-ID: <ZwQRvmVoDr_O6vLH@google.com> (raw)
In-Reply-To: <20240411082354.1691820-1-chao@kernel.org>
Hi Chao,
This introduces another bitmap increasing memory footprint. Do we know how
much performance benefit with this?
On 04/11, Chao Yu wrote:
> After commit 899fee36fac0 ("f2fs: fix to avoid data corruption by
> forbidding SSR overwrite"), valid block bitmap of current openned
> segment is fixed, let's introduce a per-log bitmap instead of temp
> bitmap to avoid unnecessary calculation overhead whenever allocating
> free slot w/ SSR allocator.
>
> Signed-off-by: Chao Yu <chao@kernel.org>
> ---
> v2:
> - rebase to last dev-test branch.
> fs/f2fs/segment.c | 30 ++++++++++++++++++++++--------
> fs/f2fs/segment.h | 1 +
> 2 files changed, 23 insertions(+), 8 deletions(-)
>
> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> index 6474b7338e81..af716925db19 100644
> --- a/fs/f2fs/segment.c
> +++ b/fs/f2fs/segment.c
> @@ -2840,31 +2840,39 @@ static int new_curseg(struct f2fs_sb_info *sbi, int type, bool new_sec)
> return 0;
> }
>
> -static int __next_free_blkoff(struct f2fs_sb_info *sbi,
> - int segno, block_t start)
> +static void __get_segment_bitmap(struct f2fs_sb_info *sbi,
> + unsigned long *target_map,
> + int segno)
> {
> struct seg_entry *se = get_seg_entry(sbi, segno);
> int entries = SIT_VBLOCK_MAP_SIZE / sizeof(unsigned long);
> - unsigned long *target_map = SIT_I(sbi)->tmp_map;
> unsigned long *ckpt_map = (unsigned long *)se->ckpt_valid_map;
> unsigned long *cur_map = (unsigned long *)se->cur_valid_map;
> int i;
>
> for (i = 0; i < entries; i++)
> target_map[i] = ckpt_map[i] | cur_map[i];
> +}
> +
> +static int __next_free_blkoff(struct f2fs_sb_info *sbi, unsigned long *bitmap,
> + int segno, block_t start)
> +{
> + __get_segment_bitmap(sbi, bitmap, segno);
>
> - return __find_rev_next_zero_bit(target_map, BLKS_PER_SEG(sbi), start);
> + return __find_rev_next_zero_bit(bitmap, BLKS_PER_SEG(sbi), start);
> }
>
> static int f2fs_find_next_ssr_block(struct f2fs_sb_info *sbi,
> - struct curseg_info *seg)
> + struct curseg_info *seg)
> {
> - return __next_free_blkoff(sbi, seg->segno, seg->next_blkoff + 1);
> + return __find_rev_next_zero_bit(seg->target_map,
> + BLKS_PER_SEG(sbi), seg->next_blkoff + 1);
> }
>
> bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno)
> {
> - return __next_free_blkoff(sbi, segno, 0) < BLKS_PER_SEG(sbi);
> + return __next_free_blkoff(sbi, SIT_I(sbi)->tmp_map, segno, 0) <
> + BLKS_PER_SEG(sbi);
> }
>
> /*
> @@ -2890,7 +2898,8 @@ static int change_curseg(struct f2fs_sb_info *sbi, int type)
>
> reset_curseg(sbi, type, 1);
> curseg->alloc_type = SSR;
> - curseg->next_blkoff = __next_free_blkoff(sbi, curseg->segno, 0);
> + curseg->next_blkoff = __next_free_blkoff(sbi, curseg->target_map,
> + curseg->segno, 0);
>
> sum_page = f2fs_get_sum_page(sbi, new_segno);
> if (IS_ERR(sum_page)) {
> @@ -4635,6 +4644,10 @@ static int build_curseg(struct f2fs_sb_info *sbi)
> sizeof(struct f2fs_journal), GFP_KERNEL);
> if (!array[i].journal)
> return -ENOMEM;
> + array[i].target_map = f2fs_kzalloc(sbi, SIT_VBLOCK_MAP_SIZE,
> + GFP_KERNEL);
> + if (!array[i].target_map)
> + return -ENOMEM;
> if (i < NR_PERSISTENT_LOG)
> array[i].seg_type = CURSEG_HOT_DATA + i;
> else if (i == CURSEG_COLD_DATA_PINNED)
> @@ -5453,6 +5466,7 @@ static void destroy_curseg(struct f2fs_sb_info *sbi)
> for (i = 0; i < NR_CURSEG_TYPE; i++) {
> kfree(array[i].sum_blk);
> kfree(array[i].journal);
> + kfree(array[i].target_map);
> }
> kfree(array);
> }
> diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
> index e1c0f418aa11..10f3e44f036f 100644
> --- a/fs/f2fs/segment.h
> +++ b/fs/f2fs/segment.h
> @@ -292,6 +292,7 @@ struct curseg_info {
> struct f2fs_summary_block *sum_blk; /* cached summary block */
> struct rw_semaphore journal_rwsem; /* protect journal area */
> struct f2fs_journal *journal; /* cached journal info */
> + unsigned long *target_map; /* bitmap for SSR allocator */
> unsigned char alloc_type; /* current allocation type */
> unsigned short seg_type; /* segment type like CURSEG_XXX_TYPE */
> unsigned int segno; /* current segment number */
> --
> 2.40.1
next prev parent reply other threads:[~2024-10-07 16:52 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-04-11 8:23 [f2fs-dev] [PATCH v2 1/2] f2fs: use per-log target_bitmap to improve lookup performace of ssr allocation Chao Yu
2024-04-11 8:23 ` Chao Yu
2024-04-11 8:23 ` [f2fs-dev] [PATCH v2 2/2] f2fs: introduce written_map to indicate written datas Chao Yu
2024-04-11 8:23 ` Chao Yu
2024-04-23 2:07 ` [f2fs-dev] [PATCH v2 1/2] f2fs: use per-log target_bitmap to improve lookup performace of ssr allocation Chao Yu
2024-04-23 2:07 ` Chao Yu
2024-05-29 11:13 ` [f2fs-dev] " Chao Yu
2024-05-29 11:13 ` Chao Yu
2024-05-30 23:39 ` Jaegeuk Kim
2024-05-30 23:39 ` Jaegeuk Kim
2024-05-31 1:10 ` Chao Yu
2024-05-31 1:10 ` Chao Yu
2024-09-18 2:17 ` Chao Yu via Linux-f2fs-devel
2024-09-18 2:17 ` Chao Yu
2024-10-07 16:52 ` Jaegeuk Kim via Linux-f2fs-devel [this message]
2024-10-07 16:52 ` Jaegeuk Kim
-- strict thread matches above, loose matches on Subject: below --
2024-04-07 13:49 [f2fs-dev] " Chao Yu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=ZwQRvmVoDr_O6vLH@google.com \
--to=linux-f2fs-devel@lists.sourceforge.net \
--cc=chao@kernel.org \
--cc=jaegeuk@kernel.org \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.