ext4: fix potential race between online resizing and write operations
commit 1d0c3924a92e69bfa91163bda83c12a994b4d106 upstream.
During an online resize an array of pointers to buffer heads gets
replaced so it can get enlarged. If there is a racing block
allocation or deallocation which uses the old array, and the old array
has gotten reused this can lead to a GPF or some other random kernel
memory getting modified.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=206443
Link: https://lore.kernel.org/r/20200221053458.730016-2-tytso@mit.edu
Reported-by: Suraj Jitindar Singh <surajjs@amazon.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org # 4.4.x
Cc: stable@kernel.org # 4.9.x
Signed-off-by: Sasha Levin <sashal@kernel.org>
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index b69a78c..75840ee 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -826,6 +826,7 @@
{
struct ext4_sb_info *sbi = EXT4_SB(sb);
struct ext4_super_block *es = sbi->s_es;
+ struct buffer_head **group_desc;
int aborted = 0;
int i, err;
@@ -857,9 +858,12 @@
if (!(sb->s_flags & MS_RDONLY))
ext4_commit_super(sb, 1);
+ rcu_read_lock();
+ group_desc = rcu_dereference(sbi->s_group_desc);
for (i = 0; i < sbi->s_gdb_count; i++)
- brelse(sbi->s_group_desc[i]);
- kvfree(sbi->s_group_desc);
+ brelse(group_desc[i]);
+ kvfree(group_desc);
+ rcu_read_unlock();
kvfree(sbi->s_flex_groups);
percpu_counter_destroy(&sbi->s_freeclusters_counter);
percpu_counter_destroy(&sbi->s_freeinodes_counter);
@@ -3403,7 +3407,7 @@
static int ext4_fill_super(struct super_block *sb, void *data, int silent)
{
char *orig_data = kstrdup(data, GFP_KERNEL);
- struct buffer_head *bh;
+ struct buffer_head *bh, **group_desc;
struct ext4_super_block *es = NULL;
struct ext4_sb_info *sbi = kzalloc(sizeof(*sbi), GFP_KERNEL);
ext4_fsblk_t block;
@@ -3955,9 +3959,10 @@
goto failed_mount;
}
}
- sbi->s_group_desc = ext4_kvmalloc(db_count *
+ rcu_assign_pointer(sbi->s_group_desc,
+ ext4_kvmalloc(db_count *
sizeof(struct buffer_head *),
- GFP_KERNEL);
+ GFP_KERNEL));
if (sbi->s_group_desc == NULL) {
ext4_msg(sb, KERN_ERR, "not enough memory");
ret = -ENOMEM;
@@ -3967,14 +3972,19 @@
bgl_lock_init(sbi->s_blockgroup_lock);
for (i = 0; i < db_count; i++) {
+ struct buffer_head *bh;
+
block = descriptor_loc(sb, logical_sb_block, i);
- sbi->s_group_desc[i] = sb_bread_unmovable(sb, block);
- if (!sbi->s_group_desc[i]) {
+ bh = sb_bread_unmovable(sb, block);
+ if (!bh) {
ext4_msg(sb, KERN_ERR,
"can't read group descriptor %d", i);
db_count = i;
goto failed_mount2;
}
+ rcu_read_lock();
+ rcu_dereference(sbi->s_group_desc)[i] = bh;
+ rcu_read_unlock();
}
sbi->s_gdb_count = db_count;
if (!ext4_check_descriptors(sb, logical_sb_block, &first_not_zeroed)) {
@@ -4349,9 +4359,12 @@
if (sbi->s_mmp_tsk)
kthread_stop(sbi->s_mmp_tsk);
failed_mount2:
+ rcu_read_lock();
+ group_desc = rcu_dereference(sbi->s_group_desc);
for (i = 0; i < db_count; i++)
- brelse(sbi->s_group_desc[i]);
- kvfree(sbi->s_group_desc);
+ brelse(group_desc[i]);
+ kvfree(group_desc);
+ rcu_read_unlock();
failed_mount:
if (sbi->s_chksum_driver)
crypto_free_shash(sbi->s_chksum_driver);