[f2fs-dev] [PATCH] f2fs: check all ones or zeros bitmap with integer data type for better mount performance

[f2fs-dev] [PATCH] f2fs: check all ones or zeros bitmap with integer data type for better mount performance

[Chao Yu]

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.

Signed-off-by: Tan Shu <shu.tan@samsung.com>
Signed-off-by: Yu Chao <chao2.yu@samsung.com>
---
 fs/f2fs/segment.h |   13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
index 7f94d78..d43ab9f 100644
--- a/fs/f2fs/segment.h
+++ b/fs/f2fs/segment.h
@@ -543,6 +543,7 @@ 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_map = (int *)raw_sit->valid_map;
 	int valid_blocks = 0;
 	int i;
 
@@ -552,6 +553,19 @@ 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) {
+		for (i = 0; i < SIT_VBLOCK_MAP_SIZE / sizeof(int); i++)
+			if (unlikely(valid_map[i] != 0))
+				BUG();
+		return;
+	} else if (GET_SIT_VBLOCKS(raw_sit) == sbi->blocks_per_seg) {
+		for (i = 0; i < SIT_VBLOCK_MAP_SIZE / sizeof(int); i++)
+			if (unlikely(valid_map[i] != -1))
+				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))
---

Re: [f2fs-dev] [PATCH] f2fs: check all ones or zeros bitmap with integer data type for better mount performance

[Jaegeuk Kim]

Hi,

2013-10-22 (화), 17:28 +0800, Chao Yu:
> 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.
> 
> Signed-off-by: Tan Shu <shu.tan@samsung.com>
> Signed-off-by: Yu Chao <chao2.yu@samsung.com>
> ---
>  fs/f2fs/segment.h |   13 +++++++++++++
>  1 file changed, 13 insertions(+)
> 
> diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
> index 7f94d78..d43ab9f 100644
> --- a/fs/f2fs/segment.h
> +++ b/fs/f2fs/segment.h
> @@ -543,6 +543,7 @@ 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_map = (int *)raw_sit->valid_map;
>  	int valid_blocks = 0;
>  	int i;
>  
> @@ -552,6 +553,19 @@ 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) {
> +		for (i = 0; i < SIT_VBLOCK_MAP_SIZE / sizeof(int); i++)

We cannot guarantee all the time that SIT_VBLOCK_MAP_SIZE is multiple of
sizeof(int).
How about using memcmp() with __u8?

> +			if (unlikely(valid_map[i] != 0))
> +				BUG();
> +		return;
> +	} else if (GET_SIT_VBLOCKS(raw_sit) == sbi->blocks_per_seg) {
> +		for (i = 0; i < SIT_VBLOCK_MAP_SIZE / sizeof(int); i++)
> +			if (unlikely(valid_map[i] != -1))
> +				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))
> ---
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

-- 
Jaegeuk Kim
Samsung

RE: [f2fs-dev] [PATCH] f2fs: check all ones or zeros bitmap with integer data type for better mount performance

[Chao Yu]

Hi, Kim:

> -----Original Message-----
> From: Jaegeuk Kim [mailto:jaegeuk.kim@samsung.com]
> Sent: Tuesday, October 22, 2013 8:24 PM
> To: Chao Yu
> Cc: linux-f2fs-devel@lists.sourceforge.net; linux-fsdevel@vger.kernel.org;
> linux-kernel@vger.kernel.org; 谭姝
> Subject: Re: [f2fs-dev] [PATCH] f2fs: check all ones or zeros bitmap with integer
> data type for better mount performance
> 
> Hi,
> 
> 2013-10-22 (화), 17:28 +0800, Chao Yu:
> > 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.
> >
> > Signed-off-by: Tan Shu <shu.tan@samsung.com>
> > Signed-off-by: Yu Chao <chao2.yu@samsung.com>
> > ---
> >  fs/f2fs/segment.h |   13 +++++++++++++
> >  1 file changed, 13 insertions(+)
> >
> > diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h index
> > 7f94d78..d43ab9f 100644
> > --- a/fs/f2fs/segment.h
> > +++ b/fs/f2fs/segment.h
> > @@ -543,6 +543,7 @@ 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_map = (int *)raw_sit->valid_map;
> >  	int valid_blocks = 0;
> >  	int i;
> >
> > @@ -552,6 +553,19 @@ 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) {
> > +		for (i = 0; i < SIT_VBLOCK_MAP_SIZE / sizeof(int); i++)
> 
> We cannot guarantee all the time that SIT_VBLOCK_MAP_SIZE is multiple of
> sizeof(int).
Well, It's really large changes for f2fs if SIT_VBLOCK_MAP_SIZE value is being modified.

> How about using memcmp() with __u8?
Do you mean that we can alloc all zeros or ones memory in SIT_VBLOCK_MAP_SIZE size,
then memcmp() it with sit bitmap by __u8?

> 
> > +			if (unlikely(valid_map[i] != 0))
> > +				BUG();
> > +		return;
> > +	} else if (GET_SIT_VBLOCKS(raw_sit) == sbi->blocks_per_seg) {
> > +		for (i = 0; i < SIT_VBLOCK_MAP_SIZE / sizeof(int); i++)
> > +			if (unlikely(valid_map[i] != -1))
> > +				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))
> > ---
> >
> > --
> > To unsubscribe from this list: send the line "unsubscribe
> > linux-fsdevel" in the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> --
> Jaegeuk Kim
> Samsung

RE: [f2fs-dev] [PATCH] f2fs: check all ones or zeros bitmap with integer data type for better mount performance

[Jaegeuk Kim]

Hi,

2013-10-23 (수), 11:23 +0800, Chao Yu:
> Hi, Kim:
> 
> > -----Original Message-----
> > From: Jaegeuk Kim [mailto:jaegeuk.kim@samsung.com]
> > Sent: Tuesday, October 22, 2013 8:24 PM
> > To: Chao Yu
> > Cc: linux-f2fs-devel@lists.sourceforge.net; linux-fsdevel@vger.kernel.org;
> > linux-kernel@vger.kernel.org; 谭姝
> > Subject: Re: [f2fs-dev] [PATCH] f2fs: check all ones or zeros bitmap with integer
> > data type for better mount performance
> > 
> > Hi,
> > 
> > 2013-10-22 (화), 17:28 +0800, Chao Yu:
> > > 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.
> > >
> > > Signed-off-by: Tan Shu <shu.tan@samsung.com>
> > > Signed-off-by: Yu Chao <chao2.yu@samsung.com>
> > > ---
> > >  fs/f2fs/segment.h |   13 +++++++++++++
> > >  1 file changed, 13 insertions(+)
> > >
> > > diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h index
> > > 7f94d78..d43ab9f 100644
> > > --- a/fs/f2fs/segment.h
> > > +++ b/fs/f2fs/segment.h
> > > @@ -543,6 +543,7 @@ 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_map = (int *)raw_sit->valid_map;
> > >  	int valid_blocks = 0;
> > >  	int i;
> > >
> > > @@ -552,6 +553,19 @@ 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) {
> > > +		for (i = 0; i < SIT_VBLOCK_MAP_SIZE / sizeof(int); i++)
> > 
> > We cannot guarantee all the time that SIT_VBLOCK_MAP_SIZE is multiple of
> > sizeof(int).
> Well, It's really large changes for f2fs if SIT_VBLOCK_MAP_SIZE value is being modified.

But, it can be changed.
Please do not add any unnecessary assumption.

> 
> > How about using memcmp() with __u8?
> Do you mean that we can alloc all zeros or ones memory in SIT_VBLOCK_MAP_SIZE size,
> then memcmp() it with sit bitmap by __u8?

Yap.
Ah, but there is another one.
It would be better to use find_next_bit_le() and
find_next_zero_bit_le().
Any idea?

> 
> > 
> > > +			if (unlikely(valid_map[i] != 0))
> > > +				BUG();
> > > +		return;
> > > +	} else if (GET_SIT_VBLOCKS(raw_sit) == sbi->blocks_per_seg) {
> > > +		for (i = 0; i < SIT_VBLOCK_MAP_SIZE / sizeof(int); i++)
> > > +			if (unlikely(valid_map[i] != -1))
> > > +				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))
> > > ---
> > >
> > > --
> > > To unsubscribe from this list: send the line "unsubscribe
> > > linux-fsdevel" in the body of a message to majordomo@vger.kernel.org
> > > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > 
> > --
> > Jaegeuk Kim
> > Samsung
> 

-- 
Jaegeuk Kim
Samsung

RE: [f2fs-dev] [PATCH] f2fs: check all ones or zeros bitmap with integer data type for better mount performance

[Chao Yu]

Hi Kim,

> -----Original Message-----
> From: Jaegeuk Kim [mailto:jaegeuk.kim@samsung.com]
> Sent: Wednesday, October 23, 2013 5:32 PM
> To: Chao Yu
> Cc: linux-f2fs-devel@lists.sourceforge.net; linux-fsdevel@vger.kernel.org;
> linux-kernel@vger.kernel.org; '谭姝'
> Subject: RE: [f2fs-dev] [PATCH] f2fs: check all ones or zeros bitmap with integer
> data type for better mount performance
> 
> Hi,
> 
> 2013-10-23 (수), 11:23 +0800, Chao Yu:
> > Hi, Kim:
> >
> > > -----Original Message-----
> > > From: Jaegeuk Kim [mailto:jaegeuk.kim@samsung.com]
> > > Sent: Tuesday, October 22, 2013 8:24 PM
> > > To: Chao Yu
> > > Cc: linux-f2fs-devel@lists.sourceforge.net;
> > > linux-fsdevel@vger.kernel.org; linux-kernel@vger.kernel.org; 谭姝
> > > Subject: Re: [f2fs-dev] [PATCH] f2fs: check all ones or zeros bitmap
> > > with integer data type for better mount performance
> > >
> > > Hi,
> > >
> > > 2013-10-22 (화), 17:28 +0800, Chao Yu:
> > > > 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.
> > > >
> > > > Signed-off-by: Tan Shu <shu.tan@samsung.com>
> > > > Signed-off-by: Yu Chao <chao2.yu@samsung.com>
> > > > ---
> > > >  fs/f2fs/segment.h |   13 +++++++++++++
> > > >  1 file changed, 13 insertions(+)
> > > >
> > > > diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h index
> > > > 7f94d78..d43ab9f 100644
> > > > --- a/fs/f2fs/segment.h
> > > > +++ b/fs/f2fs/segment.h
> > > > @@ -543,6 +543,7 @@ 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_map = (int *)raw_sit->valid_map;
> > > >  	int valid_blocks = 0;
> > > >  	int i;
> > > >
> > > > @@ -552,6 +553,19 @@ 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) {
> > > > +		for (i = 0; i < SIT_VBLOCK_MAP_SIZE / sizeof(int); i++)
> > >
> > > We cannot guarantee all the time that SIT_VBLOCK_MAP_SIZE is
> > > multiple of sizeof(int).
> > Well, It's really large changes for f2fs if SIT_VBLOCK_MAP_SIZE value is being
> modified.
> 
> But, it can be changed.
> Please do not add any unnecessary assumption.
Got it, sorry for the unmeaning assumption.

> 
> >
> > > How about using memcmp() with __u8?
> > Do you mean that we can alloc all zeros or ones memory in
> > SIT_VBLOCK_MAP_SIZE size, then memcmp() it with sit bitmap by __u8?
> 
> Yap.
> Ah, but there is another one.
> It would be better to use find_next_bit_le() and find_next_zero_bit_le().
> Any idea?
Good point.
I try to use memcmp(bitmap, bitmap+1, size-1) and bitmap[0], 
But yours got better performance and readable.
Thanks.

> 
> >
> > >
> > > > +			if (unlikely(valid_map[i] != 0))
> > > > +				BUG();
> > > > +		return;
> > > > +	} else if (GET_SIT_VBLOCKS(raw_sit) == sbi->blocks_per_seg) {
> > > > +		for (i = 0; i < SIT_VBLOCK_MAP_SIZE / sizeof(int); i++)
> > > > +			if (unlikely(valid_map[i] != -1))
> > > > +				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))
> > > > ---
> > > >
> > > > --
> > > > To unsubscribe from this list: send the line "unsubscribe
> > > > linux-fsdevel" in the body of a message to
> > > > majordomo@vger.kernel.org More majordomo info at
> > > > http://vger.kernel.org/majordomo-info.html
> > >
> > > --
> > > Jaegeuk Kim
> > > Samsung
> >
> 
> --
> Jaegeuk Kim
> Samsung