lkml.org 
[lkml]   [2013]   [Oct]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
SubjectRE: [f2fs-dev] [PATCH V2] f2fs: check all ones or zeros bitmap with bitops for better mount performance
Date
Hi Lee,

> -----Original Message-----
> From: 이창만 [mailto:cm224.lee@samsung.com]
> Sent: Monday, October 28, 2013 10:20 AM
> To: 'Chao Yu'; jaegeuk.kim@samsung.com
> Cc: linux-fsdevel@vger.kernel.org; '谭姝'; linux-kernel@vger.kernel.org;
> linux-f2fs-devel@lists.sourceforge.net
> Subject: RE: [f2fs-dev] [PATCH V2] f2fs: check all ones or zeros bitmap with
> bitops for better mount performance
>
> To check whether bitmap are all zeros or all ones, I think memcmp is more
> neat.
> But I don't know exactly performance gap between memcmp and
> find_next_bit.

With the result of my test, find_next_bit takes less time than memcmp.
If we could use {test, set, clear}_bit_le intead of f2fs_{test, set, clear}_bit, the following patch
could be used for better performance.
This one is better than the V2 patch for performance with mixed bitmap in my test.

diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
index 7f94d78..32153eb
--- a/fs/f2fs/segment.h
+++ b/fs/f2fs/segment.h
@@ -544,7 +544,8 @@ static inline void check_block_count(struct f2fs_sb_info *sbi,
struct f2fs_sm_info *sm_info = SM_I(sbi);
unsigned int end_segno = sm_info->segment_count - 1;
int valid_blocks = 0;
- int i;
+ int i, cur_pos = 0, next_pos;
+ bool is_valid;

/* check segment usage */
BUG_ON(GET_SIT_VBLOCKS(raw_sit) > sbi->blocks_per_seg);
@@ -553,9 +554,21 @@ static inline void check_block_count(struct f2fs_sb_info *sbi,
BUG_ON(segno > end_segno);

/* check bitmap with valid block count */
- for (i = 0; i < sbi->blocks_per_seg; i++)
- if (f2fs_test_bit(i, raw_sit->valid_map))
- valid_blocks++;
+ cur_pos = 0;
+ is_valid = test_bit_le(0, raw_sit->valid_map) ? true : false;
+ do {
+ if (is_valid) {
+ next_pos = find_next_zero_bit_le(&raw_sit->valid_map,
+ sbi->blocks_per_seg,
+ cur_pos);
+ valid_blocks += next_pos - cur_pos;
+ } else
+ next_pos = find_next_bit_le(&raw_sit->valid_map,
+ sbi->blocks_per_seg,
+ cur_pos);
+ cur_pos = next_pos;
+ is_valid = !is_valid;
+ } while (cur_pos < sbi->blocks_per_seg);
BUG_ON(GET_SIT_VBLOCKS(raw_sit) != valid_blocks);
}
>
>
> -----Original Message-----
> From: Chao Yu [mailto:chao2.yu@samsung.com]
> Sent: Thursday, October 24, 2013 5:21 PM
> To: jaegeuk.kim@samsung.com
> Cc: linux-fsdevel@vger.kernel.org; '谭姝'; linux-kernel@vger.kernel.org;
> linux-f2fs-devel@lists.sourceforge.net
> Subject: [f2fs-dev] [PATCH V2] f2fs: check all ones or zeros bitmap with bitops
> for better mount performance
>
> Previously, check_block_count check valid_map with bit data type in common
> scenario that sit has all ones or zeros bitmap, it makes low mount
> performance.
> So let's check the special bitmap with integer data type instead of the bit one.
>
> v2:
> use find_next_bit_le/find_next_zero_bit_le for better performance and
> readable as Jaegeuk suggested.
>
> Suggested-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
> Signed-off-by: Tan Shu <shu.tan@samsung.com>
> Signed-off-by: Yu Chao <chao2.yu@samsung.com>
> ---
> fs/f2fs/segment.h | 17 +++++++++++++++++
> 1 file changed, 17 insertions(+)
>
> diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h index 7f94d78..d25b6af
> 100644
> --- a/fs/f2fs/segment.h
> +++ b/fs/f2fs/segment.h
> @@ -552,6 +552,23 @@ static inline void check_block_count(struct
> f2fs_sb_info *sbi,
> /* check boundary of a given segment number */
> BUG_ON(segno > end_segno);
>
> + /* check all ones or zeros valid_map */
> + if (GET_SIT_VBLOCKS(raw_sit) == 0) {
> + int pos = find_next_bit_le(&raw_sit->valid_map,
> + sbi->blocks_per_seg,
> + 0);
> + if (pos != sbi->blocks_per_seg)
> + BUG();
> + return;
> + } else if (GET_SIT_VBLOCKS(raw_sit) == sbi->blocks_per_seg) {
> + int pos = find_next_zero_bit_le(&raw_sit->valid_map,
> + sbi->blocks_per_seg,
> + 0);
> + if (pos != sbi->blocks_per_seg)
> + BUG();
> + return;
> + }
> +
> /* check bitmap with valid block count */
> for (i = 0; i < sbi->blocks_per_seg; i++)
> if (f2fs_test_bit(i, raw_sit->valid_map))
> --
> 1.7.9.5
>
>
> ------------------------------------------------------------------------------
> October Webinars: Code for Performance
> Free Intel webinars can help you accelerate application performance.
> Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
> the latest Intel processors and coprocessors. See abstracts and register >
> http://pubads.g.doubleclick.net/gampad/clk?id=60135991&iu=/4140/ostg.clktr
> k
> _______________________________________________
> Linux-f2fs-devel mailing list
> Linux-f2fs-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2013-10-28 08:41    [W:0.081 / U:2.356 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site