[v2] ocfs2: fix use-after-free when unmounting read-only filesystem

Message ID 20230523093310.2628-1-lhenriques@suse.de
State New
Headers
Series [v2] ocfs2: fix use-after-free when unmounting read-only filesystem |

Commit Message

Luis Henriques May 23, 2023, 9:33 a.m. UTC
  It's trivial to trigger a use-after-free bug in the ocfs2 quotas code using
fstest generic/452.  After a read-only remount, quotas are suspended and
ocfs2_mem_dqinfo is freed through ->ocfs2_local_free_info().  When unmounting
the filesystem, an UAF access to the oinfo will eventually cause a crash.

BUG: KASAN: slab-use-after-free in timer_delete+0x54/0xc0
Read of size 8 at addr ffff8880389a8208 by task umount/669
...
Call Trace:
 <TASK>
 ...
 timer_delete+0x54/0xc0
 try_to_grab_pending+0x31/0x230
 __cancel_work_timer+0x6c/0x270
 ocfs2_disable_quotas.isra.0+0x3e/0xf0 [ocfs2]
 ocfs2_dismount_volume+0xdd/0x450 [ocfs2]
 generic_shutdown_super+0xaa/0x280
 kill_block_super+0x46/0x70
 deactivate_locked_super+0x4d/0xb0
 cleanup_mnt+0x135/0x1f0
 ...
 </TASK>

Allocated by task 632:
 kasan_save_stack+0x1c/0x40
 kasan_set_track+0x21/0x30
 __kasan_kmalloc+0x8b/0x90
 ocfs2_local_read_info+0xe3/0x9a0 [ocfs2]
 dquot_load_quota_sb+0x34b/0x680
 dquot_load_quota_inode+0xfe/0x1a0
 ocfs2_enable_quotas+0x190/0x2f0 [ocfs2]
 ocfs2_fill_super+0x14ef/0x2120 [ocfs2]
 mount_bdev+0x1be/0x200
 legacy_get_tree+0x6c/0xb0
 vfs_get_tree+0x3e/0x110
 path_mount+0xa90/0xe10
 __x64_sys_mount+0x16f/0x1a0
 do_syscall_64+0x43/0x90
 entry_SYSCALL_64_after_hwframe+0x72/0xdc

Freed by task 650:
 kasan_save_stack+0x1c/0x40
 kasan_set_track+0x21/0x30
 kasan_save_free_info+0x2a/0x50
 __kasan_slab_free+0xf9/0x150
 __kmem_cache_free+0x89/0x180
 ocfs2_local_free_info+0x2ba/0x3f0 [ocfs2]
 dquot_disable+0x35f/0xa70
 ocfs2_susp_quotas.isra.0+0x159/0x1a0 [ocfs2]
 ocfs2_remount+0x150/0x580 [ocfs2]
 reconfigure_super+0x1a5/0x3a0
 path_mount+0xc8a/0xe10
 __x64_sys_mount+0x16f/0x1a0
 do_syscall_64+0x43/0x90
 entry_SYSCALL_64_after_hwframe+0x72/0xdc

Cc: <stable@vger.kernel.org>
Signed-off-by: Luís Henriques <lhenriques@suse.de>
---
Changes since v1:
 - Fixed changelog to explicitly state 'read-only remount'
 - Added KASAN splat to changelog

 fs/ocfs2/super.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
  

Comments

Joseph Qi May 24, 2023, 3:38 a.m. UTC | #1
On 5/23/23 5:33 PM, Luís Henriques wrote:
> It's trivial to trigger a use-after-free bug in the ocfs2 quotas code using
> fstest generic/452.  After a read-only remount, quotas are suspended and
> ocfs2_mem_dqinfo is freed through ->ocfs2_local_free_info().  When unmounting
> the filesystem, an UAF access to the oinfo will eventually cause a crash.
> 
> BUG: KASAN: slab-use-after-free in timer_delete+0x54/0xc0
> Read of size 8 at addr ffff8880389a8208 by task umount/669
> ...
> Call Trace:
>  <TASK>
>  ...
>  timer_delete+0x54/0xc0
>  try_to_grab_pending+0x31/0x230
>  __cancel_work_timer+0x6c/0x270
>  ocfs2_disable_quotas.isra.0+0x3e/0xf0 [ocfs2]
>  ocfs2_dismount_volume+0xdd/0x450 [ocfs2]
>  generic_shutdown_super+0xaa/0x280
>  kill_block_super+0x46/0x70
>  deactivate_locked_super+0x4d/0xb0
>  cleanup_mnt+0x135/0x1f0
>  ...
>  </TASK>
> 
> Allocated by task 632:
>  kasan_save_stack+0x1c/0x40
>  kasan_set_track+0x21/0x30
>  __kasan_kmalloc+0x8b/0x90
>  ocfs2_local_read_info+0xe3/0x9a0 [ocfs2]
>  dquot_load_quota_sb+0x34b/0x680
>  dquot_load_quota_inode+0xfe/0x1a0
>  ocfs2_enable_quotas+0x190/0x2f0 [ocfs2]
>  ocfs2_fill_super+0x14ef/0x2120 [ocfs2]
>  mount_bdev+0x1be/0x200
>  legacy_get_tree+0x6c/0xb0
>  vfs_get_tree+0x3e/0x110
>  path_mount+0xa90/0xe10
>  __x64_sys_mount+0x16f/0x1a0
>  do_syscall_64+0x43/0x90
>  entry_SYSCALL_64_after_hwframe+0x72/0xdc
> 
> Freed by task 650:
>  kasan_save_stack+0x1c/0x40
>  kasan_set_track+0x21/0x30
>  kasan_save_free_info+0x2a/0x50
>  __kasan_slab_free+0xf9/0x150
>  __kmem_cache_free+0x89/0x180
>  ocfs2_local_free_info+0x2ba/0x3f0 [ocfs2]
>  dquot_disable+0x35f/0xa70
>  ocfs2_susp_quotas.isra.0+0x159/0x1a0 [ocfs2]
>  ocfs2_remount+0x150/0x580 [ocfs2]
>  reconfigure_super+0x1a5/0x3a0
>  path_mount+0xc8a/0xe10
>  __x64_sys_mount+0x16f/0x1a0
>  do_syscall_64+0x43/0x90
>  entry_SYSCALL_64_after_hwframe+0x72/0xdc
> 
> Cc: <stable@vger.kernel.org>
> Signed-off-by: Luís Henriques <lhenriques@suse.de>
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Tested-by: Joseph Qi <joseph.qi@linux.alibaba.com>

Tested with following procedures:
mkfs.ocfs2 -b 4k -C 1M -N 2 --fs-features=usrquota,grpquota /dev/vdc
/etc/init.d/o2cb.init online ocfs2cluster
mount.ocfs2 -o usrquota,grpquota /dev/vdc /mnt/ocfs2
mount.ocfs2 -o remount,ro,heartbeat=local /dev/vdc /mnt/ocfs2
umount /mnt/ocfs2

> ---
> Changes since v1:
>  - Fixed changelog to explicitly state 'read-only remount'
>  - Added KASAN splat to changelog
> 
>  fs/ocfs2/super.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
> index 0b0e6a132101..988d1c076861 100644
> --- a/fs/ocfs2/super.c
> +++ b/fs/ocfs2/super.c
> @@ -952,8 +952,10 @@ static void ocfs2_disable_quotas(struct ocfs2_super *osb)
>  	for (type = 0; type < OCFS2_MAXQUOTAS; type++) {
>  		if (!sb_has_quota_loaded(sb, type))
>  			continue;
> -		oinfo = sb_dqinfo(sb, type)->dqi_priv;
> -		cancel_delayed_work_sync(&oinfo->dqi_sync_work);
> +		if (!sb_has_quota_suspended(sb, type)) {
> +			oinfo = sb_dqinfo(sb, type)->dqi_priv;
> +			cancel_delayed_work_sync(&oinfo->dqi_sync_work);
> +		}
>  		inode = igrab(sb->s_dquot.files[type]);
>  		/* Turn off quotas. This will remove all dquot structures from
>  		 * memory and so they will be automatically synced to global
  

Patch

diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index 0b0e6a132101..988d1c076861 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -952,8 +952,10 @@  static void ocfs2_disable_quotas(struct ocfs2_super *osb)
 	for (type = 0; type < OCFS2_MAXQUOTAS; type++) {
 		if (!sb_has_quota_loaded(sb, type))
 			continue;
-		oinfo = sb_dqinfo(sb, type)->dqi_priv;
-		cancel_delayed_work_sync(&oinfo->dqi_sync_work);
+		if (!sb_has_quota_suspended(sb, type)) {
+			oinfo = sb_dqinfo(sb, type)->dqi_priv;
+			cancel_delayed_work_sync(&oinfo->dqi_sync_work);
+		}
 		inode = igrab(sb->s_dquot.files[type]);
 		/* Turn off quotas. This will remove all dquot structures from
 		 * memory and so they will be automatically synced to global