ext4: Avoid races caused by on-line resizing and SMP memory reordering
Ext4's on-line resizing adds a new block group and then, only at the
last step adjusts s_groups_count. However, it's possible on SMP
systems that another CPU could see the updated the s_group_count and
not see the newly initialized data structures for the just-added block
group. For this reason, it's important to insert a SMP read barrier
after reading s_groups_count and before reading any (for example) the
new block group descriptors allowed by the increased value of
s_groups_count.
Unfortunately, we rather blatently violate this locking protocol
documented in fs/ext4/resize.c. Fortunately, (1) on-line resizes
happen relatively rarely, and (2) it seems rare that the filesystem
code will immediately try to use just-added block group before any
memory ordering issues resolve themselves. So apparently problems
here are relatively hard to hit, since ext3 has been vulnerable to the
same issue for years with no one apparently complaining.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c
index 53c72ad..a5ba039 100644
--- a/fs/ext4/balloc.c
+++ b/fs/ext4/balloc.c
@@ -88,6 +88,7 @@
ext4_group_t block_group, struct ext4_group_desc *gdp)
{
int bit, bit_max;
+ ext4_group_t ngroups = ext4_get_groups_count(sb);
unsigned free_blocks, group_blocks;
struct ext4_sb_info *sbi = EXT4_SB(sb);
@@ -123,7 +124,7 @@
bit_max += ext4_bg_num_gdb(sb, block_group);
}
- if (block_group == sbi->s_groups_count - 1) {
+ if (block_group == ngroups - 1) {
/*
* Even though mke2fs always initialize first and last group
* if some other tool enabled the EXT4_BG_BLOCK_UNINIT we need
@@ -131,7 +132,7 @@
*/
group_blocks = ext4_blocks_count(sbi->s_es) -
le32_to_cpu(sbi->s_es->s_first_data_block) -
- (EXT4_BLOCKS_PER_GROUP(sb) * (sbi->s_groups_count - 1));
+ (EXT4_BLOCKS_PER_GROUP(sb) * (ngroups - 1));
} else {
group_blocks = EXT4_BLOCKS_PER_GROUP(sb);
}
@@ -205,18 +206,18 @@
{
unsigned int group_desc;
unsigned int offset;
+ ext4_group_t ngroups = ext4_get_groups_count(sb);
struct ext4_group_desc *desc;
struct ext4_sb_info *sbi = EXT4_SB(sb);
- if (block_group >= sbi->s_groups_count) {
+ if (block_group >= ngroups) {
ext4_error(sb, "ext4_get_group_desc",
"block_group >= groups_count - "
"block_group = %u, groups_count = %u",
- block_group, sbi->s_groups_count);
+ block_group, ngroups);
return NULL;
}
- smp_rmb();
group_desc = block_group >> EXT4_DESC_PER_BLOCK_BITS(sb);
offset = block_group & (EXT4_DESC_PER_BLOCK(sb) - 1);
@@ -665,7 +666,7 @@
ext4_fsblk_t desc_count;
struct ext4_group_desc *gdp;
ext4_group_t i;
- ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
+ ext4_group_t ngroups = ext4_get_groups_count(sb);
#ifdef EXT4FS_DEBUG
struct ext4_super_block *es;
ext4_fsblk_t bitmap_count;
@@ -677,7 +678,6 @@
bitmap_count = 0;
gdp = NULL;
- smp_rmb();
for (i = 0; i < ngroups; i++) {
gdp = ext4_get_group_desc(sb, i, NULL);
if (!gdp)
@@ -700,7 +700,6 @@
return bitmap_count;
#else
desc_count = 0;
- smp_rmb();
for (i = 0; i < ngroups; i++) {
gdp = ext4_get_group_desc(sb, i, NULL);
if (!gdp)