dm thin metadata: Fix use-after-free in dm_bm_set_read_only

From: Ye Bin <[email protected]>

Hi Greg,

Ye Bin had ever fixed a use-after-free of dm-thin in v5.9, and the complete patchset contains three patches:

[1/3] d16ff19e69ab dm cache metadata: Avoid returning cmd->bm wild pointer on error [2/3] 219403d7e56f dm thin metadata: Avoid returning cmd->bm wild pointer on error [3/3] 3a653b205f29 dm thin metadata: Fix use-after-free in dm_bm_set_read_only

However, 4.19.y stable only picks the former two patches [1]:
[1/3] 67f03c3d6829 dm cache metadata: Avoid returning cmd->bm wild pointer on error [2/3] 2c00ee626ed4 dm thin metadata: Avoid returning cmd->bm wild pointer on error

We encountered a NULL crash and xiejingfeng found that the omitted patch 3 can fix that. I'm not sure why patch 3 is not picked then, and we need this patch to fix this issue.

[32402.449200] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008
[32402.459553] Oops: 0002 [#1] SMP NOPTI
[32402.483982] RIP: 0010:dm_bm_set_read_only+0x5/0x10 [dm_persistent_data]
[32402.588073] Call Trace:
[32402.590522] dm_pool_metadata_read_only+0x22/0x30 [dm_thin_pool]
[32402.596526] set_pool_mode+0x209/0x2e0 [dm_thin_pool]
[32402.601579] metadata_operation_failed+0xd5/0xf0 [dm_thin_pool]
[32402.607499] commit+0x91/0xf0 [dm_thin_pool]
[32402.611771] pool_status+0x28a/0x700 [dm_thin_pool]
[32402.616652] retrieve_status+0xa1/0x1c0 [dm_mod]
[32402.627794] table_status+0x61/0xa0 [dm_mod]
[32402.632068] ctl_ioctl+0x1b3/0x480 [dm_mod]
[32402.636253] dm_ctl_ioctl+0xa/0x10 [dm_mod]
[32402.640440] do_vfs_ioctl+0x9f/0x610
[32402.653081] ksys_ioctl+0x70/0x80
[32402.656393] __x64_sys_ioctl+0x16/0x20
[32402.660145] do_syscall_64+0x7b/0x200
[32402.663813] entry_SYSCALL_64_after_hwframe+0x44/0xa9

[1] https://lore.kernel.org/lkml/[email protected]/

---
commit 3a653b205f29b3f9827a01a0c88bfbcb0d169494 upstream.

The following error ocurred when testing disk online/offline:

[  301.798344] device-mapper: thin: 253:5: aborting current metadata transaction
[  301.848441] device-mapper: thin: 253:5: failed to abort metadata transaction
[  301.849206] Aborting journal on device dm-26-8.
[  301.850489] EXT4-fs error (device dm-26) in __ext4_new_inode:943: Journal has aborted
[  301.851095] EXT4-fs (dm-26): Delayed block allocation failed for inode 398742 at logical offset 181 with max blocks 19 with error 30
[  301.854476] BUG: KASAN: use-after-free in dm_bm_set_read_only+0x3a/0x40 [dm_persistent_data]

Reason is:

metadata_operation_failed

    abort_transaction
        dm_pool_abort_metadata
	    __create_persistent_data_objects
	        r = __open_or_format_metadata
	        if (r) --> If failed will free pmd->bm but pmd->bm not set NULL
		    dm_block_manager_destroy(pmd->bm);
    set_pool_mode
	dm_pool_metadata_read_only(pool->pmd);
	dm_bm_set_read_only(pmd->bm);  --> use-after-free

Add checks to see if pmd->bm is NULL in dm_bm_set_read_only and dm_bm_set_read_write functions. If bm is NULL it means creating the bm failed and so dm_bm_is_read_only must return true.

Signed-off-by Ye Bin <[email protected]>

Cc [email protected]

Signed-off-by Mike Snitzer <[email protected]>

Signed-off-by xiejingfeng <[email protected]>

Signed-off-by Jeffle Xu <[email protected]>

---
 drivers/md/dm-thin-metadata.c                 |  2 +-
 drivers/md/persistent-data/dm-block-manager.c | 14 ++++++++------
 2 files changed, 9 insertions(+), 7 deletions(-)

drivers/md/dm-thin-metadata.c +1 -1

index 85077f4d257a..a6a5cee6b943 100644
@@ -901,7 +901,7 @@ int dm_pool_metadata_close(struct dm_pool_metadata *pmd)
 		return -EBUSY;
 	}
 
-	if (!dm_bm_is_read_only(pmd->bm) && !pmd->fail_io) {
+	if (!pmd->fail_io && !dm_bm_is_read_only(pmd->bm)) {
 		r = __commit_transaction(pmd);
 		if (r < 0)
 			DMWARN("%s: __commit_transaction() failed, error = %d",

drivers/md/persistent-data/dm-block-manager.c +8 -6

index 492a3f8ac119..0401daa0f7fb 100644
@@ -494,7 +494,7 @@ int dm_bm_write_lock(struct dm_block_manager *bm,
 	void *p;
 	int r;
 
-	if (bm->read_only)
+	if (dm_bm_is_read_only(bm))
 		return -EPERM;
 
 	p = dm_bufio_read(bm->bufio, b, (struct dm_buffer **) result);

@@ -563,7 +563,7 @@ int dm_bm_write_lock_zero(struct dm_block_manager *bm,
 	struct buffer_aux *aux;
 	void *p;
 
-	if (bm->read_only)
+	if (dm_bm_is_read_only(bm))
 		return -EPERM;
 
 	p = dm_bufio_new(bm->bufio, b, (struct dm_buffer **) result);

@@ -603,7 +603,7 @@ EXPORT_SYMBOL_GPL(dm_bm_unlock);
 
 int dm_bm_flush(struct dm_block_manager *bm)
 {
-	if (bm->read_only)
+	if (dm_bm_is_read_only(bm))
 		return -EPERM;
 
 	return dm_bufio_write_dirty_buffers(bm->bufio);

@@ -617,19 +617,21 @@ void dm_bm_prefetch(struct dm_block_manager *bm, dm_block_t b)
 
 bool dm_bm_is_read_only(struct dm_block_manager *bm)
 {
-	return bm->read_only;
+	return (bm ? bm->read_only : true);
 }
 EXPORT_SYMBOL_GPL(dm_bm_is_read_only);
 
 void dm_bm_set_read_only(struct dm_block_manager *bm)
 {
-	bm->read_only = true;
+	if (bm)
+		bm->read_only = true;
 }
 EXPORT_SYMBOL_GPL(dm_bm_set_read_only);
 
 void dm_bm_set_read_write(struct dm_block_manager *bm)
 {
-	bm->read_only = false;
+	if (bm)
+		bm->read_only = false;
 }
 EXPORT_SYMBOL_GPL(dm_bm_set_read_write);