f2fs: data: fix possible overflow in check_swap_activate()

[PATCH] f2fs: data: fix possible overflow in check_swap_activate()

Posted by Sergey Shtylyov 2 years, 1 month ago

In check_swap_activate(), if the *while* loop exits early (0- or 1-page
long swap file), an overflow happens while calculating the value of the
span parameter as the lowest_pblock variable ends up being greater than
the highest_pblock variable. Let's set *span to 0 in this case...

Found by Linux Verification Center (linuxtesting.org) with the SVACE static
analysis tool.

Signed-off-by: Sergey Shtylyov <s.shtylyov@omp.ru>
---
This patch is against the 'master' branch of Jaegeuk Kim's F2FS repo...

 fs/f2fs/data.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index 916e317ac925..342cb0d5056d 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -4047,7 +4047,10 @@ static int check_swap_activate(struct swap_info_struct *sis,
 		cur_lblock += nr_pblocks;
 	}
 	ret = nr_extents;
-	*span = 1 + highest_pblock - lowest_pblock;
+	if (lowest_pblock <= highest_pblock)
+		*span = 1 + highest_pblock - lowest_pblock;
+	else
+		*span = 0;
 	if (cur_lblock == 0)
 		cur_lblock = 1;	/* force Empty message */
 	sis->max = cur_lblock;
-- 
2.26.3

Re: [PATCH] f2fs: data: fix possible overflow in check_swap_activate()

Posted by Chao Yu 2 years, 1 month ago

Hi Sergey,

Thanks for the patch.

On 2023/10/26 4:20, Sergey Shtylyov wrote:
> In check_swap_activate(), if the *while* loop exits early (0- or 1-page
> long swap file), an overflow happens while calculating the value of the
> span parameter as the lowest_pblock variable ends up being greater than
> the highest_pblock variable. Let's set *span to 0 in this case...

What do you think of returning -EINVAL for such case? I assume this is a
corner case.

> 
> Found by Linux Verification Center (linuxtesting.org) with the SVACE static
> analysis tool.
> 
> Signed-off-by: Sergey Shtylyov <s.shtylyov@omp.ru>
> ---
> This patch is against the 'master' branch of Jaegeuk Kim's F2FS repo...
> 
>   fs/f2fs/data.c | 5 ++++-
>   1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
> index 916e317ac925..342cb0d5056d 100644
> --- a/fs/f2fs/data.c
> +++ b/fs/f2fs/data.c
> @@ -4047,7 +4047,10 @@ static int check_swap_activate(struct swap_info_struct *sis,
>   		cur_lblock += nr_pblocks;
>   	}
>   	ret = nr_extents;
> -	*span = 1 + highest_pblock - lowest_pblock;
> +	if (lowest_pblock <= highest_pblock)

if (unlikely(higest_pblock < lowest_pblock))
	return -EINVAL;

*span = 1 + highest_pblock - lowest_pblock;

Thanks,

> +		*span = 1 + highest_pblock - lowest_pblock;
> +	else
> +		*span = 0;
>   	if (cur_lblock == 0)
>   		cur_lblock = 1;	/* force Empty message */
>   	sis->max = cur_lblock;

Re: [PATCH] f2fs: data: fix possible overflow in check_swap_activate()

Posted by Sergey Shtylyov 2 years ago

Hello!

   Sorry for replying a month later than I should -- I got distracted by the
other Svace reports... It took a significant part of the weekend to swap this
stuff back in... :-/

On 11/7/23 6:29 PM, Chao Yu wrote:
[...]
>> In check_swap_activate(), if the *while* loop exits early (0- or 1-page
>> long swap file), an overflow happens while calculating the value of the
>> span parameter as the lowest_pblock variable ends up being greater than
>> the highest_pblock variable. Let's set *span to 0 in this case...
> 
> What do you think of returning -EINVAL for such case? I assume this is a
> corner case.

   I don't know the code well enough but I got the impression that iff
we have a file containing a single page, we'd have one successful call
of add_swap_extent(). Am I missing something?

>> Found by Linux Verification Center (linuxtesting.org) with the SVACE static
>> analysis tool.
>>
>> Signed-off-by: Sergey Shtylyov <s.shtylyov@omp.ru>
>> ---
>> This patch is against the 'master' branch of Jaegeuk Kim's F2FS repo...
>>
>>   fs/f2fs/data.c | 5 ++++-
>>   1 file changed, 4 insertions(+), 1 deletion(-)
>>
>> diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
>> index 916e317ac925..342cb0d5056d 100644
>> --- a/fs/f2fs/data.c
>> +++ b/fs/f2fs/data.c
>> @@ -4047,7 +4047,10 @@ static int check_swap_activate(struct swap_info_struct *sis,
>>           cur_lblock += nr_pblocks;
>>       }
>>       ret = nr_extents;
>> -    *span = 1 + highest_pblock - lowest_pblock;
>> +    if (lowest_pblock <= highest_pblock)
> 
> if (unlikely(higest_pblock < lowest_pblock))

   Well, Greg KH says we shouldn't use unlikely() unless we can prove
that it indeed improves things...

>     return -EINVAL;
> 
> *span = 1 + highest_pblock - lowest_pblock;
> 
> Thanks,
> 
>> +        *span = 1 + highest_pblock - lowest_pblock;
>> +    else
>> +        *span = 0;
>>       if (cur_lblock == 0)
>>           cur_lblock = 1;    /* force Empty message */
>>       sis->max = cur_lblock;

MBR, Sergey