Commit 952fc18e authored by Theodore Ts'o's avatar Theodore Ts'o

ext4: fix overhead calculation used by ext4_statfs()

Commit f975d6bc introduced bug which caused ext4_statfs() to
miscalculate the number of file system overhead blocks.  This causes
the f_blocks field in the statfs structure to be larger than it should
be.  This would in turn cause the "df" output to show the number of
data blocks in the file system and the number of data blocks used to
be larger than they should be.
Signed-off-by: default avatar"Theodore Ts'o" <tytso@mit.edu>
Cc: stable@kernel.org
parent f6fb99ca
...@@ -11,8 +11,6 @@ ...@@ -11,8 +11,6 @@
#include <linux/jbd2.h> #include <linux/jbd2.h>
#include "ext4.h" #include "ext4.h"
#ifdef EXT4FS_DEBUG
static const int nibblemap[] = {4, 3, 3, 2, 3, 2, 2, 1, 3, 2, 2, 1, 2, 1, 1, 0}; static const int nibblemap[] = {4, 3, 3, 2, 3, 2, 2, 1, 3, 2, 2, 1, 2, 1, 1, 0};
unsigned int ext4_count_free(char *bitmap, unsigned int numchars) unsigned int ext4_count_free(char *bitmap, unsigned int numchars)
...@@ -25,8 +23,6 @@ unsigned int ext4_count_free(char *bitmap, unsigned int numchars) ...@@ -25,8 +23,6 @@ unsigned int ext4_count_free(char *bitmap, unsigned int numchars)
return sum; return sum;
} }
#endif /* EXT4FS_DEBUG */
int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group, int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
struct ext4_group_desc *gdp, struct ext4_group_desc *gdp,
struct buffer_head *bh, int sz) struct buffer_head *bh, int sz)
......
...@@ -1161,8 +1161,7 @@ struct ext4_sb_info { ...@@ -1161,8 +1161,7 @@ struct ext4_sb_info {
unsigned long s_desc_per_block; /* Number of group descriptors per block */ unsigned long s_desc_per_block; /* Number of group descriptors per block */
ext4_group_t s_groups_count; /* Number of groups in the fs */ ext4_group_t s_groups_count; /* Number of groups in the fs */
ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */ ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
unsigned long s_overhead_last; /* Last calculated overhead */ unsigned long s_overhead; /* # of fs overhead clusters */
unsigned long s_blocks_last; /* Last seen block count */
unsigned int s_cluster_ratio; /* Number of blocks per cluster */ unsigned int s_cluster_ratio; /* Number of blocks per cluster */
unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */ unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */ loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
...@@ -2037,6 +2036,7 @@ extern int ext4_group_extend(struct super_block *sb, ...@@ -2037,6 +2036,7 @@ extern int ext4_group_extend(struct super_block *sb,
extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count); extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
/* super.c */ /* super.c */
extern int ext4_calculate_overhead(struct super_block *sb);
extern int ext4_superblock_csum_verify(struct super_block *sb, extern int ext4_superblock_csum_verify(struct super_block *sb,
struct ext4_super_block *es); struct ext4_super_block *es);
extern void ext4_superblock_csum_set(struct super_block *sb, extern void ext4_superblock_csum_set(struct super_block *sb,
......
...@@ -1197,7 +1197,7 @@ static void ext4_update_super(struct super_block *sb, ...@@ -1197,7 +1197,7 @@ static void ext4_update_super(struct super_block *sb,
struct ext4_new_group_data *group_data = flex_gd->groups; struct ext4_new_group_data *group_data = flex_gd->groups;
struct ext4_sb_info *sbi = EXT4_SB(sb); struct ext4_sb_info *sbi = EXT4_SB(sb);
struct ext4_super_block *es = sbi->s_es; struct ext4_super_block *es = sbi->s_es;
int i; int i, ret;
BUG_ON(flex_gd->count == 0 || group_data == NULL); BUG_ON(flex_gd->count == 0 || group_data == NULL);
/* /*
...@@ -1272,6 +1272,11 @@ static void ext4_update_super(struct super_block *sb, ...@@ -1272,6 +1272,11 @@ static void ext4_update_super(struct super_block *sb,
&sbi->s_flex_groups[flex_group].free_inodes); &sbi->s_flex_groups[flex_group].free_inodes);
} }
/*
* Update the fs overhead information
*/
ext4_calculate_overhead(sb);
if (test_opt(sb, DEBUG)) if (test_opt(sb, DEBUG))
printk(KERN_DEBUG "EXT4-fs: added group %u:" printk(KERN_DEBUG "EXT4-fs: added group %u:"
"%llu blocks(%llu free %llu reserved)\n", flex_gd->count, "%llu blocks(%llu free %llu reserved)\n", flex_gd->count,
......
...@@ -3085,6 +3085,114 @@ static int set_journal_csum_feature_set(struct super_block *sb) ...@@ -3085,6 +3085,114 @@ static int set_journal_csum_feature_set(struct super_block *sb)
return ret; return ret;
} }
/*
* Note: calculating the overhead so we can be compatible with
* historical BSD practice is quite difficult in the face of
* clusters/bigalloc. This is because multiple metadata blocks from
* different block group can end up in the same allocation cluster.
* Calculating the exact overhead in the face of clustered allocation
* requires either O(all block bitmaps) in memory or O(number of block
* groups**2) in time. We will still calculate the superblock for
* older file systems --- and if we come across with a bigalloc file
* system with zero in s_overhead_clusters the estimate will be close to
* correct especially for very large cluster sizes --- but for newer
* file systems, it's better to calculate this figure once at mkfs
* time, and store it in the superblock. If the superblock value is
* present (even for non-bigalloc file systems), we will use it.
*/
static int count_overhead(struct super_block *sb, ext4_group_t grp,
char *buf)
{
struct ext4_sb_info *sbi = EXT4_SB(sb);
struct ext4_group_desc *gdp;
ext4_fsblk_t first_block, last_block, b;
ext4_group_t i, ngroups = ext4_get_groups_count(sb);
int s, j, count = 0;
first_block = le32_to_cpu(sbi->s_es->s_first_data_block) +
(grp * EXT4_BLOCKS_PER_GROUP(sb));
last_block = first_block + EXT4_BLOCKS_PER_GROUP(sb) - 1;
for (i = 0; i < ngroups; i++) {
gdp = ext4_get_group_desc(sb, i, NULL);
b = ext4_block_bitmap(sb, gdp);
if (b >= first_block && b <= last_block) {
ext4_set_bit(EXT4_B2C(sbi, b - first_block), buf);
count++;
}
b = ext4_inode_bitmap(sb, gdp);
if (b >= first_block && b <= last_block) {
ext4_set_bit(EXT4_B2C(sbi, b - first_block), buf);
count++;
}
b = ext4_inode_table(sb, gdp);
if (b >= first_block && b + sbi->s_itb_per_group <= last_block)
for (j = 0; j < sbi->s_itb_per_group; j++, b++) {
int c = EXT4_B2C(sbi, b - first_block);
ext4_set_bit(c, buf);
count++;
}
if (i != grp)
continue;
s = 0;
if (ext4_bg_has_super(sb, grp)) {
ext4_set_bit(s++, buf);
count++;
}
for (j = ext4_bg_num_gdb(sb, grp); j > 0; j--) {
ext4_set_bit(EXT4_B2C(sbi, s++), buf);
count++;
}
}
if (!count)
return 0;
return EXT4_CLUSTERS_PER_GROUP(sb) -
ext4_count_free(buf, EXT4_CLUSTERS_PER_GROUP(sb) / 8);
}
/*
* Compute the overhead and stash it in sbi->s_overhead
*/
int ext4_calculate_overhead(struct super_block *sb)
{
struct ext4_sb_info *sbi = EXT4_SB(sb);
struct ext4_super_block *es = sbi->s_es;
ext4_group_t i, ngroups = ext4_get_groups_count(sb);
ext4_fsblk_t overhead = 0;
char *buf = (char *) get_zeroed_page(GFP_KERNEL);
memset(buf, 0, PAGE_SIZE);
if (!buf)
return -ENOMEM;
/*
* Compute the overhead (FS structures). This is constant
* for a given filesystem unless the number of block groups
* changes so we cache the previous value until it does.
*/
/*
* All of the blocks before first_data_block are overhead
*/
overhead = EXT4_B2C(sbi, le32_to_cpu(es->s_first_data_block));
/*
* Add the overhead found in each block group
*/
for (i = 0; i < ngroups; i++) {
int blks;
blks = count_overhead(sb, i, buf);
overhead += blks;
if (blks)
memset(buf, 0, PAGE_SIZE);
cond_resched();
}
sbi->s_overhead = overhead;
smp_wmb();
free_page((unsigned long) buf);
return 0;
}
static int ext4_fill_super(struct super_block *sb, void *data, int silent) static int ext4_fill_super(struct super_block *sb, void *data, int silent)
{ {
char *orig_data = kstrdup(data, GFP_KERNEL); char *orig_data = kstrdup(data, GFP_KERNEL);
...@@ -3734,6 +3842,18 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) ...@@ -3734,6 +3842,18 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
percpu_counter_set(&sbi->s_dirtyclusters_counter, 0); percpu_counter_set(&sbi->s_dirtyclusters_counter, 0);
no_journal: no_journal:
/*
* Get the # of file system overhead blocks from the
* superblock if present.
*/
if (es->s_overhead_clusters)
sbi->s_overhead = le32_to_cpu(es->s_overhead_clusters);
else {
ret = ext4_calculate_overhead(sb);
if (ret)
goto failed_mount_wq;
}
/* /*
* The maximum number of concurrent works can be high and * The maximum number of concurrent works can be high and
* concurrency isn't really necessary. Limit it to 1. * concurrency isn't really necessary. Limit it to 1.
...@@ -4600,67 +4720,21 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data) ...@@ -4600,67 +4720,21 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
return err; return err;
} }
/*
* Note: calculating the overhead so we can be compatible with
* historical BSD practice is quite difficult in the face of
* clusters/bigalloc. This is because multiple metadata blocks from
* different block group can end up in the same allocation cluster.
* Calculating the exact overhead in the face of clustered allocation
* requires either O(all block bitmaps) in memory or O(number of block
* groups**2) in time. We will still calculate the superblock for
* older file systems --- and if we come across with a bigalloc file
* system with zero in s_overhead_clusters the estimate will be close to
* correct especially for very large cluster sizes --- but for newer
* file systems, it's better to calculate this figure once at mkfs
* time, and store it in the superblock. If the superblock value is
* present (even for non-bigalloc file systems), we will use it.
*/
static int ext4_statfs(struct dentry *dentry, struct kstatfs *buf) static int ext4_statfs(struct dentry *dentry, struct kstatfs *buf)
{ {
struct super_block *sb = dentry->d_sb; struct super_block *sb = dentry->d_sb;
struct ext4_sb_info *sbi = EXT4_SB(sb); struct ext4_sb_info *sbi = EXT4_SB(sb);
struct ext4_super_block *es = sbi->s_es; struct ext4_super_block *es = sbi->s_es;
struct ext4_group_desc *gdp; ext4_fsblk_t overhead = 0;
u64 fsid; u64 fsid;
s64 bfree; s64 bfree;
if (test_opt(sb, MINIX_DF)) { if (!test_opt(sb, MINIX_DF))
sbi->s_overhead_last = 0; overhead = sbi->s_overhead;
} else if (es->s_overhead_clusters) {
sbi->s_overhead_last = le32_to_cpu(es->s_overhead_clusters);
} else if (sbi->s_blocks_last != ext4_blocks_count(es)) {
ext4_group_t i, ngroups = ext4_get_groups_count(sb);
ext4_fsblk_t overhead = 0;
/*
* Compute the overhead (FS structures). This is constant
* for a given filesystem unless the number of block groups
* changes so we cache the previous value until it does.
*/
/*
* All of the blocks before first_data_block are
* overhead
*/
overhead = EXT4_B2C(sbi, le32_to_cpu(es->s_first_data_block));
/*
* Add the overhead found in each block group
*/
for (i = 0; i < ngroups; i++) {
gdp = ext4_get_group_desc(sb, i, NULL);
overhead += ext4_num_overhead_clusters(sb, i, gdp);
cond_resched();
}
sbi->s_overhead_last = overhead;
smp_wmb();
sbi->s_blocks_last = ext4_blocks_count(es);
}
buf->f_type = EXT4_SUPER_MAGIC; buf->f_type = EXT4_SUPER_MAGIC;
buf->f_bsize = sb->s_blocksize; buf->f_bsize = sb->s_blocksize;
buf->f_blocks = (ext4_blocks_count(es) - buf->f_blocks = ext4_blocks_count(es) - EXT4_C2B(sbi, sbi->s_overhead);
EXT4_C2B(sbi, sbi->s_overhead_last));
bfree = percpu_counter_sum_positive(&sbi->s_freeclusters_counter) - bfree = percpu_counter_sum_positive(&sbi->s_freeclusters_counter) -
percpu_counter_sum_positive(&sbi->s_dirtyclusters_counter); percpu_counter_sum_positive(&sbi->s_dirtyclusters_counter);
/* prevent underflow in case that few free space is available */ /* prevent underflow in case that few free space is available */
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment