Message-ID: <20180124022253.GA82000@jaegeuk-macbookpro.roam.corp.google.com>
Date:   Tue, 23 Jan 2018 18:22:53 -0800
From:   Jaegeuk Kim <jaegeuk@...nel.org>
To:     Chao Yu <yuchao0@...wei.com>
Cc:     linux-f2fs-devel@...ts.sourceforge.net,
        linux-kernel@...r.kernel.org, chao@...nel.org
Subject: Re: [PATCH v2] mkfs.f2fs: expand scalability of nat bitmap

On 01/24, Chao Yu wrote:
> On 2018/1/24 6:19, Jaegeuk Kim wrote:
> > On 01/23, Jaegeuk Kim wrote:
> >> On 01/23, Chao Yu wrote:
> >>> On 2018/1/23 7:00, Jaegeuk Kim wrote:
> >>>> On 01/17, Chao Yu wrote:
> >>>>> Hi Jaegeuk,
> >>>>>
> >>>>> On 2018/1/17 8:47, Jaegeuk Kim wrote:
> >>>>>> Hi Chao,
> >>>>>>
> >>>>>> On 01/15, Chao Yu wrote:
> >>>>>>> Previously, our total node number (nat_bitmap) and total nat segment count
> >>>>>>> will not monotonously increase along with image size, and max nat_bitmap size
> >>>>>>> is limited by "CHECKSUM_OFFSET - sizeof(struct f2fs_checkpoint) + 1", it is
> >>>>>>> with bad scalability when user wants to create more inode/node in larger image.
> >>>>>>>
> >>>>>>> So this patch tries to relieve the limitation, by default, limitting total nat
> >>>>>>> entry number with 20% of total block number.
> >>>>>>>
> >>>>>>> Before:
> >>>>>>> image_size(GB)	nat_bitmap	sit_bitmap	nat_segment	sit_segment
> >>>>>>> 16		3836		64		36		2
> >>>>>>> 32		3836		64		72		2
> >>>>>>> 64		3772		128		116		4
> >>>>>>> 128		3708		192		114		6
> >>>>>>> 256		3580		320		110		10
> >>>>>
> >>>>> As you see, nat_segment count will reduce when image size increases
> >>>>> starting from 64GB, that means nat segment count will not monotonously
> >>>>> increase when image size is increasing, so it would be better to active
> >>>>> this when image size is larger than 32GB?
> >>>>>
> >>>>> IMO, configuring basic nid ratio to fixed value like ext4 ("free inode" :
> >>>>> "free block" is about 1 : 4) would be better:
> >>>>> a. It will be easy for user to predict nid count or nat segment count with
> >>>>> fix-sized image;
> >>>>> b. If user wants to reserve more nid count, we can support -N option in
> >>>>> mkfs.f2fs to specify total nid count as user wish.
> >>>>
> >>>> My concern is about a CTS failure in terms of # of free inodes.
> >>>
> >>> You mean testSaneInodes()?
> >>>
> >>>         final long maxsize = stat.f_blocks * stat.f_frsize;
> >>>         final long maxInodes = maxsize / 4096;
> >>>         final long minsize = stat.f_bavail * stat.f_frsize;
> >>>         final long minInodes = minsize / 32768;
> >>>
> >>> The range is about [1/8, 1], so our 20% threshold can just let it passed,
> >>> right?
> >>
> >> Yes, thanks for checking the codes. Let me play with this for some time.
> > 
> > It simply triggers a panic, if kernel does not have the patch to detect the
> > feature. Hmm...
> 
> Yes, because we have changed disk layout of nat/sit_version_bitmap in
> mkfs.f2fs, if kernel can not detect that, we will encounter panic simply.

, which means we can't do this by default at least.

> 
> Thanks,
> 
> > 
> > Thanks,
> > 
> >>
> >> Thanks,
> >>
> >>>
> >>> Thanks,
> >>>
> >>>>
> >>>> Thanks,
> >>>>
> >>>>>
> >>>>> How do you think?
> >>>>>
> >>>>> Thanks,
> >>>>>
> >>>>>>> 512		3260		640		100		20
> >>>>>>> 1024		2684		1216		82		38
> >>>>>>> 2048		1468		2432		44		76
> >>>>>>> 4096		3900		4800		120		150
> >>>>>>>
> >>>>>>> After:
> >>>>>>> image_size(GB)	nat_bitmap	sit_bitmap	nat_segment	sit_segment
> >>>>>>> 16		256		64		8		2
> >>>>>>> 32		512		64		16		2
> >>>>>>> 64		960		128		30		4
> >>>>>>> 128		1856		192		58		6
> >>>>>>> 256		3712		320		116		10
> >>>>>>
> >>>>>> Can we activate this, if size is larger than 256GB or something around that?
> >>>>>>
> >>>>>> Thanks,
> >>>>>>
> >>>>>>> 512		7424		640		232		20
> >>>>>>> 1024		14787		1216		462		38
> >>>>>>> 2048		29504		2432		922		76
> >>>>>>> 4096		59008		4800		1844		150
> >>>>>>>
> >>>>>>> Signed-off-by: Chao Yu <yuchao0@...wei.com>
> >>>>>>> ---
> >>>>>>> v2:
> >>>>>>> - add CP_LARGE_NAT_BITMAP_FLAG flag to indicate new layout of nat/sit bitmap.
> >>>>>>>  fsck/f2fs.h        | 19 +++++++++++++------
> >>>>>>>  fsck/resize.c      | 35 +++++++++++++++++------------------
> >>>>>>>  include/f2fs_fs.h  |  8 ++++++--
> >>>>>>>  lib/libf2fs.c      |  1 +
> >>>>>>>  mkfs/f2fs_format.c | 45 +++++++++++++++++++++++----------------------
> >>>>>>>  5 files changed, 60 insertions(+), 48 deletions(-)
> >>>>>>>
> >>>>>>> diff --git a/fsck/f2fs.h b/fsck/f2fs.h
> >>>>>>> index f5970d9dafc0..8a5ce365282d 100644
> >>>>>>> --- a/fsck/f2fs.h
> >>>>>>> +++ b/fsck/f2fs.h
> >>>>>>> @@ -239,6 +239,12 @@ static inline unsigned int ofs_of_node(struct f2fs_node *node_blk)
> >>>>>>>  	return flag >> OFFSET_BIT_SHIFT;
> >>>>>>>  }
> >>>>>>>  
> >>>>>>> +static inline bool is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
> >>>>>>> +{
> >>>>>>> +	unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
> >>>>>>> +	return ckpt_flags & f ? 1 : 0;
> >>>>>>> +}
> >>>>>>> +
> >>>>>>>  static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag)
> >>>>>>>  {
> >>>>>>>  	struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
> >>>>>>> @@ -256,6 +262,13 @@ static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
> >>>>>>>  {
> >>>>>>>  	struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
> >>>>>>>  	int offset;
> >>>>>>> +
> >>>>>>> +	if (is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG)) {
> >>>>>>> +		offset = (flag == SIT_BITMAP) ?
> >>>>>>> +			le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0;
> >>>>>>> +		return &ckpt->sit_nat_version_bitmap + offset;
> >>>>>>> +	}
> >>>>>>> +
> >>>>>>>  	if (le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload) > 0) {
> >>>>>>>  		if (flag == NAT_BITMAP)
> >>>>>>>  			return &ckpt->sit_nat_version_bitmap;
> >>>>>>> @@ -268,12 +281,6 @@ static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag)
> >>>>>>>  	}
> >>>>>>>  }
> >>>>>>>  
> >>>>>>> -static inline bool is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f)
> >>>>>>> -{
> >>>>>>> -	unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags);
> >>>>>>> -	return ckpt_flags & f ? 1 : 0;
> >>>>>>> -}
> >>>>>>> -
> >>>>>>>  static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi)
> >>>>>>>  {
> >>>>>>>  	block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr);
> >>>>>>> diff --git a/fsck/resize.c b/fsck/resize.c
> >>>>>>> index 143ad5d3c0a1..f3547c86f351 100644
> >>>>>>> --- a/fsck/resize.c
> >>>>>>> +++ b/fsck/resize.c
> >>>>>>> @@ -13,10 +13,10 @@ static int get_new_sb(struct f2fs_super_block *sb)
> >>>>>>>  {
> >>>>>>>  	u_int32_t zone_size_bytes, zone_align_start_offset;
> >>>>>>>  	u_int32_t blocks_for_sit, blocks_for_nat, blocks_for_ssa;
> >>>>>>> -	u_int32_t sit_segments, diff, total_meta_segments;
> >>>>>>> +	u_int32_t sit_segments, nat_segments, diff, total_meta_segments;
> >>>>>>>  	u_int32_t total_valid_blks_available;
> >>>>>>>  	u_int32_t sit_bitmap_size, max_sit_bitmap_size;
> >>>>>>> -	u_int32_t max_nat_bitmap_size, max_nat_segments;
> >>>>>>> +	u_int32_t max_nat_bitmap_size;
> >>>>>>>  	u_int32_t segment_size_bytes = 1 << (get_sb(log_blocksize) +
> >>>>>>>  					get_sb(log_blocks_per_seg));
> >>>>>>>  	u_int32_t blks_per_seg = 1 << get_sb(log_blocks_per_seg);
> >>>>>>> @@ -47,7 +47,15 @@ static int get_new_sb(struct f2fs_super_block *sb)
> >>>>>>>  			get_sb(segment_count_sit))) * blks_per_seg;
> >>>>>>>  	blocks_for_nat = SIZE_ALIGN(total_valid_blks_available,
> >>>>>>>  					NAT_ENTRY_PER_BLOCK);
> >>>>>>> -	set_sb(segment_count_nat, SEG_ALIGN(blocks_for_nat));
> >>>>>>> +	nat_segments = SEG_ALIGN(blocks_for_nat) *
> >>>>>>> +					DEFAULT_NAT_ENTRY_RATIO / 100;
> >>>>>>> +	set_sb(segment_count_nat, nat_segments ? nat_segments : 1);
> >>>>>>> +
> >>>>>>> +	max_nat_bitmap_size = (get_sb(segment_count_nat) <<
> >>>>>>> +					get_sb(log_blocks_per_seg)) / 8;
> >>>>>>> +	set_sb(segment_count_nat, get_sb(segment_count_nat) * 2);
> >>>>>>> +
> >>>>>>> +	c.large_nat_bitmap = 1;
> >>>>>>>  
> >>>>>>>  	sit_bitmap_size = ((get_sb(segment_count_sit) / 2) <<
> >>>>>>>  				get_sb(log_blocks_per_seg)) / 8;
> >>>>>>> @@ -56,25 +64,16 @@ static int get_new_sb(struct f2fs_super_block *sb)
> >>>>>>>  	else
> >>>>>>>  		max_sit_bitmap_size = sit_bitmap_size;
> >>>>>>>  
> >>>>>>> -	/*
> >>>>>>> -	 * It should be reserved minimum 1 segment for nat.
> >>>>>>> -	 * When sit is too large, we should expand cp area. It requires more pages for cp.
> >>>>>>> -	 */
> >>>>>>> -	if (max_sit_bitmap_size > MAX_SIT_BITMAP_SIZE_IN_CKPT) {
> >>>>>>> -		max_nat_bitmap_size = CHECKSUM_OFFSET - sizeof(struct f2fs_checkpoint) + 1;
> >>>>>>> -		set_sb(cp_payload, F2FS_BLK_ALIGN(max_sit_bitmap_size));
> >>>>>>> +	/* use cp_payload if free space of f2fs_checkpoint is not enough */
> >>>>>>> +	if (max_sit_bitmap_size + max_nat_bitmap_size >
> >>>>>>> +					MAX_BITMAP_SIZE_IN_CKPT) {
> >>>>>>> +		u_int32_t diff =  max_sit_bitmap_size + max_nat_bitmap_size -
> >>>>>>> +							MAX_BITMAP_SIZE_IN_CKPT;
> >>>>>>> +		set_sb(cp_payload, F2FS_BLK_ALIGN(diff));
> >>>>>>>  	} else {
> >>>>>>> -		max_nat_bitmap_size = CHECKSUM_OFFSET - sizeof(struct f2fs_checkpoint) + 1
> >>>>>>> -			- max_sit_bitmap_size;
> >>>>>>>  		set_sb(cp_payload, 0);
> >>>>>>>  	}
> >>>>>>>  
> >>>>>>> -	max_nat_segments = (max_nat_bitmap_size * 8) >>
> >>>>>>> -					get_sb(log_blocks_per_seg);
> >>>>>>> -
> >>>>>>> -	if (get_sb(segment_count_nat) > max_nat_segments)
> >>>>>>> -		set_sb(segment_count_nat, max_nat_segments);
> >>>>>>> -
> >>>>>>>  	set_sb(segment_count_nat, get_sb(segment_count_nat) * 2);
> >>>>>>>  
> >>>>>>>  	set_sb(ssa_blkaddr, get_sb(nat_blkaddr) +
> >>>>>>> diff --git a/include/f2fs_fs.h b/include/f2fs_fs.h
> >>>>>>> index 4739085ed98f..edf351412702 100644
> >>>>>>> --- a/include/f2fs_fs.h
> >>>>>>> +++ b/include/f2fs_fs.h
> >>>>>>> @@ -362,6 +362,7 @@ struct f2fs_configuration {
> >>>>>>>  	int preen_mode;
> >>>>>>>  	int ro;
> >>>>>>>  	int preserve_limits;		/* preserve quota limits */
> >>>>>>> +	int large_nat_bitmap;
> >>>>>>>  	__le32 feature;			/* defined features */
> >>>>>>>  
> >>>>>>>  	/* defragmentation parameters */
> >>>>>>> @@ -613,6 +614,7 @@ struct f2fs_super_block {
> >>>>>>>  /*
> >>>>>>>   * For checkpoint
> >>>>>>>   */
> >>>>>>> +#define CP_LARGE_NAT_BITMAP_FLAG	0x00000200
> >>>>>>>  #define CP_TRIMMED_FLAG		0x00000100
> >>>>>>>  #define CP_NAT_BITS_FLAG	0x00000080
> >>>>>>>  #define CP_CRC_RECOVERY_FLAG	0x00000040
> >>>>>>> @@ -657,8 +659,8 @@ struct f2fs_checkpoint {
> >>>>>>>  	unsigned char sit_nat_version_bitmap[1];
> >>>>>>>  } __attribute__((packed));
> >>>>>>>  
> >>>>>>> -#define MAX_SIT_BITMAP_SIZE_IN_CKPT	\
> >>>>>>> -	(CHECKSUM_OFFSET - sizeof(struct f2fs_checkpoint) + 1 - 64)
> >>>>>>> +#define MAX_BITMAP_SIZE_IN_CKPT	\
> >>>>>>> +	(CHECKSUM_OFFSET - sizeof(struct f2fs_checkpoint) + 1)
> >>>>>>>  
> >>>>>>>  /*
> >>>>>>>   * For orphan inode management
> >>>>>>> @@ -846,6 +848,8 @@ struct f2fs_node {
> >>>>>>>  #define NAT_ENTRY_PER_BLOCK (PAGE_CACHE_SIZE / sizeof(struct f2fs_nat_entry))
> >>>>>>>  #define NAT_BLOCK_OFFSET(start_nid) (start_nid / NAT_ENTRY_PER_BLOCK)
> >>>>>>>  
> >>>>>>> +#define DEFAULT_NAT_ENTRY_RATIO		20
> >>>>>>> +
> >>>>>>>  #ifdef ANDROID_WINDOWS_HOST
> >>>>>>>  #pragma pack(1)
> >>>>>>>  #endif
> >>>>>>> diff --git a/lib/libf2fs.c b/lib/libf2fs.c
> >>>>>>> index ffdbccb34627..e8b1842b7391 100644
> >>>>>>> --- a/lib/libf2fs.c
> >>>>>>> +++ b/lib/libf2fs.c
> >>>>>>> @@ -623,6 +623,7 @@ void f2fs_init_configuration(void)
> >>>>>>>  	c.ro = 0;
> >>>>>>>  	c.kd = -1;
> >>>>>>>  	c.dry_run = 0;
> >>>>>>> +	c.large_nat_bitmap = 0;
> >>>>>>>  	c.fixed_time = -1;
> >>>>>>>  }
> >>>>>>>  
> >>>>>>> diff --git a/mkfs/f2fs_format.c b/mkfs/f2fs_format.c
> >>>>>>> index a13000184300..23eaf40c5962 100644
> >>>>>>> --- a/mkfs/f2fs_format.c
> >>>>>>> +++ b/mkfs/f2fs_format.c
> >>>>>>> @@ -151,13 +151,13 @@ static int f2fs_prepare_super_block(void)
> >>>>>>>  	u_int32_t log_sectorsize, log_sectors_per_block;
> >>>>>>>  	u_int32_t log_blocksize, log_blks_per_seg;
> >>>>>>>  	u_int32_t segment_size_bytes, zone_size_bytes;
> >>>>>>> -	u_int32_t sit_segments;
> >>>>>>> +	u_int32_t sit_segments, nat_segments;
> >>>>>>>  	u_int32_t blocks_for_sit, blocks_for_nat, blocks_for_ssa;
> >>>>>>>  	u_int32_t total_valid_blks_available;
> >>>>>>>  	u_int64_t zone_align_start_offset, diff;
> >>>>>>>  	u_int64_t total_meta_zones, total_meta_segments;
> >>>>>>>  	u_int32_t sit_bitmap_size, max_sit_bitmap_size;
> >>>>>>> -	u_int32_t max_nat_bitmap_size, max_nat_segments;
> >>>>>>> +	u_int32_t max_nat_bitmap_size;
> >>>>>>>  	u_int32_t total_zones;
> >>>>>>>  	u_int32_t next_ino;
> >>>>>>>  	enum quota_type qtype;
> >>>>>>> @@ -272,7 +272,18 @@ static int f2fs_prepare_super_block(void)
> >>>>>>>  	blocks_for_nat = SIZE_ALIGN(total_valid_blks_available,
> >>>>>>>  			NAT_ENTRY_PER_BLOCK);
> >>>>>>>  
> >>>>>>> -	set_sb(segment_count_nat, SEG_ALIGN(blocks_for_nat));
> >>>>>>> +	nat_segments = SEG_ALIGN(blocks_for_nat) *
> >>>>>>> +					DEFAULT_NAT_ENTRY_RATIO / 100;
> >>>>>>> +
> >>>>>>> +	set_sb(segment_count_nat, nat_segments ? nat_segments : 1);
> >>>>>>> +
> >>>>>>> +	max_nat_bitmap_size = (get_sb(segment_count_nat) <<
> >>>>>>> +					log_blks_per_seg) / 8;
> >>>>>>> +
> >>>>>>> +	set_sb(segment_count_nat, get_sb(segment_count_nat) * 2);
> >>>>>>> +
> >>>>>>> +	c.large_nat_bitmap = 1;
> >>>>>>> +
> >>>>>>>  	/*
> >>>>>>>  	 * The number of node segments should not be exceeded a "Threshold".
> >>>>>>>  	 * This number resizes NAT bitmap area in a CP page.
> >>>>>>> @@ -286,29 +297,16 @@ static int f2fs_prepare_super_block(void)
> >>>>>>>  	else
> >>>>>>>  		max_sit_bitmap_size = sit_bitmap_size;
> >>>>>>>  
> >>>>>>> -	/*
> >>>>>>> -	 * It should be reserved minimum 1 segment for nat.
> >>>>>>> -	 * When sit is too large, we should expand cp area. It requires more
> >>>>>>> -	 * pages for cp.
> >>>>>>> -	 */
> >>>>>>> -	if (max_sit_bitmap_size > MAX_SIT_BITMAP_SIZE_IN_CKPT) {
> >>>>>>> -		max_nat_bitmap_size = CHECKSUM_OFFSET -
> >>>>>>> -				sizeof(struct f2fs_checkpoint) + 1;
> >>>>>>> -		set_sb(cp_payload, F2FS_BLK_ALIGN(max_sit_bitmap_size));
> >>>>>>> +	/* use cp_payload if free space of f2fs_checkpoint is not enough */
> >>>>>>> +	if (max_sit_bitmap_size + max_nat_bitmap_size >
> >>>>>>> +					MAX_BITMAP_SIZE_IN_CKPT) {
> >>>>>>> +		u_int32_t diff =  max_sit_bitmap_size + max_nat_bitmap_size -
> >>>>>>> +							MAX_BITMAP_SIZE_IN_CKPT;
> >>>>>>> +		set_sb(cp_payload, F2FS_BLK_ALIGN(diff));
> >>>>>>>  	} else {
> >>>>>>> -		max_nat_bitmap_size =
> >>>>>>> -			CHECKSUM_OFFSET - sizeof(struct f2fs_checkpoint) + 1
> >>>>>>> -			- max_sit_bitmap_size;
> >>>>>>>  		set_sb(cp_payload, 0);
> >>>>>>>  	}
> >>>>>>>  
> >>>>>>> -	max_nat_segments = (max_nat_bitmap_size * 8) >> log_blks_per_seg;
> >>>>>>> -
> >>>>>>> -	if (get_sb(segment_count_nat) > max_nat_segments)
> >>>>>>> -		set_sb(segment_count_nat, max_nat_segments);
> >>>>>>> -
> >>>>>>> -	set_sb(segment_count_nat, get_sb(segment_count_nat) * 2);
> >>>>>>> -
> >>>>>>>  	set_sb(ssa_blkaddr, get_sb(nat_blkaddr) + get_sb(segment_count_nat) *
> >>>>>>>  			c.blks_per_seg);
> >>>>>>>  
> >>>>>>> @@ -622,6 +620,9 @@ static int f2fs_write_check_point_pack(void)
> >>>>>>>  	if (c.trimmed)
> >>>>>>>  		flags |= CP_TRIMMED_FLAG;
> >>>>>>>  
> >>>>>>> +	if (c.large_nat_bitmap)
> >>>>>>> +		flags |= CP_LARGE_NAT_BITMAP_FLAG;
> >>>>>>> +
> >>>>>>>  	set_cp(ckpt_flags, flags);
> >>>>>>>  	set_cp(cp_pack_start_sum, 1 + get_sb(cp_payload));
> >>>>>>>  	set_cp(valid_node_count, 1 + quota_inum);
> >>>>>>> -- 
> >>>>>>> 2.15.0.55.gc2ece9dc4de6
> >>>>>>
> >>>>>> .
> >>>>>>
> >>>>
> >>>> .
> >>>>
> > 
> > .
> > 

Powered by blists - more mailing lists