[PATCH 2/3] memcg: check more strictly to avoid PAGE_ALIGN wrapped to 0

[PATCH 2/3] memcg: check more strictly to avoid PAGE_ALIGN wrapped to 0

[Sha Zhengju]

Since PAGE_ALIGN is aligning up(the next page boundary), this can
prevent input values wrapped to 0 and cause strange result to user.

This patch also rename the second arg of
res_counter_memparse_write_strategy() to 'resp' and add a local
variable 'res' to save the too often dereferences. Thanks Andrew
for pointing it out!


Signed-off-by: Sha Zhengju <handai.szj-3b8fjiQLQpfQT0dZR+AlfA@public.gmane.org>
Reported-by: Li Wenpeng <xingke.lwp-3b8fjiQLQpfQT0dZR+AlfA@public.gmane.org>
---
 kernel/res_counter.c |   18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)

diff --git a/kernel/res_counter.c b/kernel/res_counter.c
index 3f0417f..be8ddda 100644
--- a/kernel/res_counter.c
+++ b/kernel/res_counter.c
@@ -178,23 +178,29 @@ u64 res_counter_read_u64(struct res_counter *counter, int member)
 #endif
 
 int res_counter_memparse_write_strategy(const char *buf,
-					unsigned long long *res)
+					unsigned long long *resp)
 {
 	char *end;
+	unsigned long long res;
 
 	/* return RES_COUNTER_MAX(unlimited) if "-1" is specified */
 	if (*buf == '-') {
-		*res = simple_strtoull(buf + 1, &end, 10);
-		if (*res != 1 || *end != '\0')
+		res = simple_strtoull(buf + 1, &end, 10);
+		if (res != 1 || *end != '\0')
 			return -EINVAL;
-		*res = RES_COUNTER_MAX;
+		*resp = RES_COUNTER_MAX;
 		return 0;
 	}
 
-	*res = memparse(buf, &end);
+	res = memparse(buf, &end);
 	if (*end != '\0')
 		return -EINVAL;
 
-	*res = PAGE_ALIGN(*res);
+	if (PAGE_ALIGN(res) >= res)
+		res = PAGE_ALIGN(res);
+	else
+		res = RES_COUNTER_MAX; /* avoid PAGE_ALIGN wrapping to zero */
+
+	*resp = res;
 	return 0;
 }
-- 
1.7.9.5

Re: [PATCH 2/3] memcg: check more strictly to avoid PAGE_ALIGN wrapped to 0

[Michal Hocko]

On Sun 05-05-13 23:43:10, Sha Zhengju wrote:
> Since PAGE_ALIGN is aligning up(the next page boundary), this can
> prevent input values wrapped to 0 and cause strange result to user.

I guess you wanted to say that it can cause an overflow, right?
"
Since PAGE_ALIGN is aligning up (to the next page boundary), this can
cause an overflow to 0 if >= ULLONG_MAX-4094 value is given in the
buffer.
"
> 
> This patch also rename the second arg of
> res_counter_memparse_write_strategy() to 'resp' and add a local
> variable 'res' to save the too often dereferences. Thanks Andrew
> for pointing it out!

Again, it would be nicer to have this cleanup in a separate patch.

> Signed-off-by: Sha Zhengju <handai.szj@taobao.com>
> Reported-by: Li Wenpeng <xingke.lwp@taobao.com>

Acked-by: Michal Hocko <mhocko@suse.cz>

We have this bug since ever and nobody has noticed so nobody seems to
use 

> ---
>  kernel/res_counter.c |   18 ++++++++++++------
>  1 file changed, 12 insertions(+), 6 deletions(-)
> 
> diff --git a/kernel/res_counter.c b/kernel/res_counter.c
> index 3f0417f..be8ddda 100644
> --- a/kernel/res_counter.c
> +++ b/kernel/res_counter.c
> @@ -178,23 +178,29 @@ u64 res_counter_read_u64(struct res_counter *counter, int member)
>  #endif
>  
>  int res_counter_memparse_write_strategy(const char *buf,
> -					unsigned long long *res)
> +					unsigned long long *resp)
>  {
>  	char *end;
> +	unsigned long long res;
>  
>  	/* return RES_COUNTER_MAX(unlimited) if "-1" is specified */
>  	if (*buf == '-') {
> -		*res = simple_strtoull(buf + 1, &end, 10);
> -		if (*res != 1 || *end != '\0')
> +		res = simple_strtoull(buf + 1, &end, 10);
> +		if (res != 1 || *end != '\0')
>  			return -EINVAL;
> -		*res = RES_COUNTER_MAX;
> +		*resp = RES_COUNTER_MAX;
>  		return 0;
>  	}
>  
> -	*res = memparse(buf, &end);
> +	res = memparse(buf, &end);
>  	if (*end != '\0')
>  		return -EINVAL;
>  
> -	*res = PAGE_ALIGN(*res);
> +	if (PAGE_ALIGN(res) >= res)
> +		res = PAGE_ALIGN(res);
> +	else
> +		res = RES_COUNTER_MAX; /* avoid PAGE_ALIGN wrapping to zero */
> +
> +	*resp = res;
>  	return 0;
>  }
> -- 
> 1.7.9.5
> 

-- 
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH 2/3] memcg: check more strictly to avoid PAGE_ALIGN wrapped to 0

[Sha Zhengju]

On Tue, May 7, 2013 at 10:04 PM, Michal Hocko <mhocko@suse.cz> wrote:
> On Sun 05-05-13 23:43:10, Sha Zhengju wrote:
>> Since PAGE_ALIGN is aligning up(the next page boundary), this can
>> prevent input values wrapped to 0 and cause strange result to user.
>
> I guess you wanted to say that it can cause an overflow, right?
> "
> Since PAGE_ALIGN is aligning up (to the next page boundary), this can
> cause an overflow to 0 if >= ULLONG_MAX-4094 value is given in the
> buffer.
> "

Yes!

>>
>> This patch also rename the second arg of
>> res_counter_memparse_write_strategy() to 'resp' and add a local
>> variable 'res' to save the too often dereferences. Thanks Andrew
>> for pointing it out!
>
> Again, it would be nicer to have this cleanup in a separate patch.

Okay.

>
>> Signed-off-by: Sha Zhengju <handai.szj@taobao.com>
>> Reported-by: Li Wenpeng <xingke.lwp@taobao.com>
>
> Acked-by: Michal Hocko <mhocko@suse.cz>
>
> We have this bug since ever and nobody has noticed so nobody seems to
> use

Yeah, that's rarely occur, but we happened to run into it.
Thank you for the comments!

>
>> ---
>>  kernel/res_counter.c |   18 ++++++++++++------
>>  1 file changed, 12 insertions(+), 6 deletions(-)
>>
>> diff --git a/kernel/res_counter.c b/kernel/res_counter.c
>> index 3f0417f..be8ddda 100644
>> --- a/kernel/res_counter.c
>> +++ b/kernel/res_counter.c
>> @@ -178,23 +178,29 @@ u64 res_counter_read_u64(struct res_counter *counter, int member)
>>  #endif
>>
>>  int res_counter_memparse_write_strategy(const char *buf,
>> -                                     unsigned long long *res)
>> +                                     unsigned long long *resp)
>>  {
>>       char *end;
>> +     unsigned long long res;
>>
>>       /* return RES_COUNTER_MAX(unlimited) if "-1" is specified */
>>       if (*buf == '-') {
>> -             *res = simple_strtoull(buf + 1, &end, 10);
>> -             if (*res != 1 || *end != '\0')
>> +             res = simple_strtoull(buf + 1, &end, 10);
>> +             if (res != 1 || *end != '\0')
>>                       return -EINVAL;
>> -             *res = RES_COUNTER_MAX;
>> +             *resp = RES_COUNTER_MAX;
>>               return 0;
>>       }
>>
>> -     *res = memparse(buf, &end);
>> +     res = memparse(buf, &end);
>>       if (*end != '\0')
>>               return -EINVAL;
>>
>> -     *res = PAGE_ALIGN(*res);
>> +     if (PAGE_ALIGN(res) >= res)
>> +             res = PAGE_ALIGN(res);
>> +     else
>> +             res = RES_COUNTER_MAX; /* avoid PAGE_ALIGN wrapping to zero */
>> +
>> +     *resp = res;
>>       return 0;
>>  }
>> --
>> 1.7.9.5
>>
>
> --
> Michal Hocko
> SUSE Labs



--
Thanks,
Sha

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>