mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH 1/1] dm-vdo: Refactor VDO_ASSERT usage in start_restoring_volume functions
@ 2025-09-08 22:24 Ivan Abramov
  2025-09-09 18:58 ` Matthew Sakai
  0 siblings, 1 reply; 3+ messages in thread
From: Ivan Abramov @ 2025-09-08 22:24 UTC (permalink / raw)
  To: Matthew Sakai; +Cc: Ivan Abramov, dm-devel, linux-kernel, lvc-project

There's an incorrect VDO_ASSERT macro usage in 
start_restoring volume_index() and start_restoring_volume_sub_index(),
since assert's return value is not used anywhere.

Thus, use VDO_ASSERT_LOG_ONLY macro in such cases.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Signed-off-by: Ivan Abramov <i.abramov@mt-integration.ru>
---
 drivers/md/dm-vdo/indexer/volume-index.c | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)

diff --git a/drivers/md/dm-vdo/indexer/volume-index.c b/drivers/md/dm-vdo/indexer/volume-index.c
index 12f954a0c532..99608f14400a 100644
--- a/drivers/md/dm-vdo/indexer/volume-index.c
+++ b/drivers/md/dm-vdo/indexer/volume-index.c
@@ -832,11 +832,9 @@ static int start_restoring_volume_sub_index(struct volume_sub_index *sub_index,
 		decode_u32_le(buffer, &offset, &header.first_list);
 		decode_u32_le(buffer, &offset, &header.list_count);
 
-		result = VDO_ASSERT(offset == sizeof(buffer),
+		VDO_ASSERT_LOG_ONLY(offset == sizeof(buffer),
 				    "%zu bytes decoded of %zu expected", offset,
 				    sizeof(buffer));
-		if (result != VDO_SUCCESS)
-			result = UDS_CORRUPT_DATA;
 
 		if (memcmp(header.magic, MAGIC_START_5, MAGIC_SIZE) != 0) {
 			return vdo_log_warning_strerror(UDS_CORRUPT_DATA,
@@ -924,11 +922,9 @@ static int start_restoring_volume_index(struct volume_index *volume_index,
 		offset += MAGIC_SIZE;
 		decode_u32_le(buffer, &offset, &header.sparse_sample_rate);
 
-		result = VDO_ASSERT(offset == sizeof(buffer),
+		VDO_ASSERT_LOG_ONLY(offset == sizeof(buffer),
 				    "%zu bytes decoded of %zu expected", offset,
 				    sizeof(buffer));
-		if (result != VDO_SUCCESS)
-			result = UDS_CORRUPT_DATA;
 
 		if (memcmp(header.magic, MAGIC_START_6, MAGIC_SIZE) != 0)
 			return vdo_log_warning_strerror(UDS_CORRUPT_DATA,
-- 
2.39.5


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH 1/1] dm-vdo: Refactor VDO_ASSERT usage in start_restoring_volume functions
  2025-09-08 22:24 [PATCH 1/1] dm-vdo: Refactor VDO_ASSERT usage in start_restoring_volume functions Ivan Abramov
@ 2025-09-09 18:58 ` Matthew Sakai
  2025-09-09 19:23   ` Matthew Sakai
  0 siblings, 1 reply; 3+ messages in thread
From: Matthew Sakai @ 2025-09-09 18:58 UTC (permalink / raw)
  To: Ivan Abramov; +Cc: dm-devel, linux-kernel, lvc-project

On 9/8/25 6:24 PM, Ivan Abramov wrote:
> There's an incorrect VDO_ASSERT macro usage in
> start_restoring volume_index() and start_restoring_volume_sub_index(),
> since assert's return value is not used anywhere.
> 
> Thus, use VDO_ASSERT_LOG_ONLY macro in such cases.
> 
> Found by Linux Verification Center (linuxtesting.org) with SVACE.
> 
> Signed-off-by: Ivan Abramov <i.abramov@mt-integration.ru>
> ---
>   drivers/md/dm-vdo/indexer/volume-index.c | 8 ++------
>   1 file changed, 2 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/md/dm-vdo/indexer/volume-index.c b/drivers/md/dm-vdo/indexer/volume-index.c
> index 12f954a0c532..99608f14400a 100644
> --- a/drivers/md/dm-vdo/indexer/volume-index.c
> +++ b/drivers/md/dm-vdo/indexer/volume-index.c
> @@ -832,11 +832,9 @@ static int start_restoring_volume_sub_index(struct volume_sub_index *sub_index,
>   		decode_u32_le(buffer, &offset, &header.first_list);
>   		decode_u32_le(buffer, &offset, &header.list_count);
>   
> -		result = VDO_ASSERT(offset == sizeof(buffer),
> +		VDO_ASSERT_LOG_ONLY(offset == sizeof(buffer),
>   				    "%zu bytes decoded of %zu expected", offset,
>   				    sizeof(buffer));
> -		if (result != VDO_SUCCESS)
> -			result = UDS_CORRUPT_DATA;

This is not a correct fix so we should not take this.

This assertion (and the one below) should only fail if the stored 
metadata has been corrupted, but in that case we definitely need to 
return the error and fail the load.

You are correct that we don't seem to use this result, but the correct 
fix is to return UDS_CORRUPT_DATA when this assertion fails.

Matt


>   		if (memcmp(header.magic, MAGIC_START_5, MAGIC_SIZE) != 0) {
>   			return vdo_log_warning_strerror(UDS_CORRUPT_DATA,
> @@ -924,11 +922,9 @@ static int start_restoring_volume_index(struct volume_index *volume_index,
>   		offset += MAGIC_SIZE;
>   		decode_u32_le(buffer, &offset, &header.sparse_sample_rate);
>   
> -		result = VDO_ASSERT(offset == sizeof(buffer),
> +		VDO_ASSERT_LOG_ONLY(offset == sizeof(buffer),
>   				    "%zu bytes decoded of %zu expected", offset,
>   				    sizeof(buffer));
> -		if (result != VDO_SUCCESS)
> -			result = UDS_CORRUPT_DATA;
>   
>   		if (memcmp(header.magic, MAGIC_START_6, MAGIC_SIZE) != 0)
>   			return vdo_log_warning_strerror(UDS_CORRUPT_DATA,


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH 1/1] dm-vdo: Refactor VDO_ASSERT usage in start_restoring_volume functions
  2025-09-09 18:58 ` Matthew Sakai
@ 2025-09-09 19:23   ` Matthew Sakai
  0 siblings, 0 replies; 3+ messages in thread
From: Matthew Sakai @ 2025-09-09 19:23 UTC (permalink / raw)
  To: Ivan Abramov; +Cc: dm-devel, linux-kernel, lvc-project



On 9/9/25 2:58 PM, Matthew Sakai wrote:
> On 9/8/25 6:24 PM, Ivan Abramov wrote:
>> There's an incorrect VDO_ASSERT macro usage in
>> start_restoring volume_index() and start_restoring_volume_sub_index(),
>> since assert's return value is not used anywhere.
>>
>> Thus, use VDO_ASSERT_LOG_ONLY macro in such cases.
>>
>> Found by Linux Verification Center (linuxtesting.org) with SVACE.
>>
>> Signed-off-by: Ivan Abramov <i.abramov@mt-integration.ru>
>> ---
>>   drivers/md/dm-vdo/indexer/volume-index.c | 8 ++------
>>   1 file changed, 2 insertions(+), 6 deletions(-)
>>
>> diff --git a/drivers/md/dm-vdo/indexer/volume-index.c b/drivers/md/dm- 
>> vdo/indexer/volume-index.c
>> index 12f954a0c532..99608f14400a 100644
>> --- a/drivers/md/dm-vdo/indexer/volume-index.c
>> +++ b/drivers/md/dm-vdo/indexer/volume-index.c
>> @@ -832,11 +832,9 @@ static int 
>> start_restoring_volume_sub_index(struct volume_sub_index *sub_index,
>>           decode_u32_le(buffer, &offset, &header.first_list);
>>           decode_u32_le(buffer, &offset, &header.list_count);
>> -        result = VDO_ASSERT(offset == sizeof(buffer),
>> +        VDO_ASSERT_LOG_ONLY(offset == sizeof(buffer),
>>                       "%zu bytes decoded of %zu expected", offset,
>>                       sizeof(buffer));
>> -        if (result != VDO_SUCCESS)
>> -            result = UDS_CORRUPT_DATA;
> 
> This is not a correct fix so we should not take this.
> 
> This assertion (and the one below) should only fail if the stored 
> metadata has been corrupted, but in that case we definitely need to 
> return the error and fail the load.
> 
> You are correct that we don't seem to use this result, but the correct 
> fix is to return UDS_CORRUPT_DATA when this assertion fails.

Replying to myself, feel free to correct this patch, or I can probably 
get to posting a proper fix later in the week.

> Matt
> 
> 
>>           if (memcmp(header.magic, MAGIC_START_5, MAGIC_SIZE) != 0) {
>>               return vdo_log_warning_strerror(UDS_CORRUPT_DATA,
>> @@ -924,11 +922,9 @@ static int start_restoring_volume_index(struct 
>> volume_index *volume_index,
>>           offset += MAGIC_SIZE;
>>           decode_u32_le(buffer, &offset, &header.sparse_sample_rate);
>> -        result = VDO_ASSERT(offset == sizeof(buffer),
>> +        VDO_ASSERT_LOG_ONLY(offset == sizeof(buffer),
>>                       "%zu bytes decoded of %zu expected", offset,
>>                       sizeof(buffer));
>> -        if (result != VDO_SUCCESS)
>> -            result = UDS_CORRUPT_DATA;
>>           if (memcmp(header.magic, MAGIC_START_6, MAGIC_SIZE) != 0)
>>               return vdo_log_warning_strerror(UDS_CORRUPT_DATA,
> 


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2025-09-09 19:23 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2025-09-08 22:24 [PATCH 1/1] dm-vdo: Refactor VDO_ASSERT usage in start_restoring_volume functions Ivan Abramov
2025-09-09 18:58 ` Matthew Sakai
2025-09-09 19:23   ` Matthew Sakai

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®