2020-08-31 02:19:19

by Daeho Jeong

[permalink] [raw]
Subject: [PATCH] f2fs: change compr_blocks of superblock info to 64bit

From: Daeho Jeong <[email protected]>

Current compr_blocks of superblock info is not 64bit value. We are
accumulating each i_compr_blocks count of inodes to this value and
those are 64bit values. So, need to change this to 64bit value.

Signed-off-by: Daeho Jeong <[email protected]>
---
fs/f2fs/debug.c | 6 +++---
fs/f2fs/f2fs.h | 9 +++++----
2 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/fs/f2fs/debug.c b/fs/f2fs/debug.c
index 4276c0f79beb..257e658b3a5e 100644
--- a/fs/f2fs/debug.c
+++ b/fs/f2fs/debug.c
@@ -131,7 +131,7 @@ static void update_general_status(struct f2fs_sb_info *sbi)
si->inline_inode = atomic_read(&sbi->inline_inode);
si->inline_dir = atomic_read(&sbi->inline_dir);
si->compr_inode = atomic_read(&sbi->compr_inode);
- si->compr_blocks = atomic_read(&sbi->compr_blocks);
+ si->compr_blocks = atomic64_read(&sbi->compr_blocks);
si->append = sbi->im[APPEND_INO].ino_num;
si->update = sbi->im[UPDATE_INO].ino_num;
si->orphans = sbi->im[ORPHAN_INO].ino_num;
@@ -342,7 +342,7 @@ static int stat_show(struct seq_file *s, void *v)
si->inline_inode);
seq_printf(s, " - Inline_dentry Inode: %u\n",
si->inline_dir);
- seq_printf(s, " - Compressed Inode: %u, Blocks: %u\n",
+ seq_printf(s, " - Compressed Inode: %u, Blocks: %llu\n",
si->compr_inode, si->compr_blocks);
seq_printf(s, " - Orphan/Append/Update Inode: %u, %u, %u\n",
si->orphans, si->append, si->update);
@@ -542,7 +542,7 @@ int f2fs_build_stats(struct f2fs_sb_info *sbi)
atomic_set(&sbi->inline_inode, 0);
atomic_set(&sbi->inline_dir, 0);
atomic_set(&sbi->compr_inode, 0);
- atomic_set(&sbi->compr_blocks, 0);
+ atomic64_set(&sbi->compr_blocks, 0);
atomic_set(&sbi->inplace_count, 0);
for (i = META_CP; i < META_MAX; i++)
atomic_set(&sbi->meta_count[i], 0);
diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 02811ce15059..f60414805e05 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -1552,7 +1552,7 @@ struct f2fs_sb_info {
atomic_t inline_inode; /* # of inline_data inodes */
atomic_t inline_dir; /* # of inline_dentry inodes */
atomic_t compr_inode; /* # of compressed inodes */
- atomic_t compr_blocks; /* # of compressed blocks */
+ atomic64_t compr_blocks; /* # of compressed blocks */
atomic_t vw_cnt; /* # of volatile writes */
atomic_t max_aw_cnt; /* max # of atomic writes */
atomic_t max_vw_cnt; /* max # of volatile writes */
@@ -3533,7 +3533,8 @@ struct f2fs_stat_info {
int nr_discard_cmd;
unsigned int undiscard_blks;
int inline_xattr, inline_inode, inline_dir, append, update, orphans;
- int compr_inode, compr_blocks;
+ int compr_inode;
+ unsigned long long compr_blocks;
int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt;
unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks;
unsigned int bimodal, avg_vblocks;
@@ -3618,9 +3619,9 @@ static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi)
(atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \
} while (0)
#define stat_add_compr_blocks(inode, blocks) \
- (atomic_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
+ (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks))
#define stat_sub_compr_blocks(inode, blocks) \
- (atomic_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
+ (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks))
#define stat_inc_meta_count(sbi, blkaddr) \
do { \
if (blkaddr < SIT_I(sbi)->sit_base_addr) \
--
2.28.0.402.g5ffc5be6b7-goog


2020-08-31 02:49:59

by Chao Yu

[permalink] [raw]
Subject: Re: [f2fs-dev] [PATCH] f2fs: change compr_blocks of superblock info to 64bit

On 2020/8/31 10:09, Daeho Jeong wrote:
> From: Daeho Jeong <[email protected]>
>
> Current compr_blocks of superblock info is not 64bit value. We are
> accumulating each i_compr_blocks count of inodes to this value and
> those are 64bit values. So, need to change this to 64bit value.

It's not a big deal, as f2fs use 32bits to index block device's address space,
so there is at most 1 << 32 blocks in f2fs, we can not save more blocks than
the number...

If it overflows 32bits variable's space, it must be a bug.

>
> Signed-off-by: Daeho Jeong <[email protected]>

Reviewed-by: Chao Yu <[email protected]>

Thanks,