* Re: [PATCH] kasan: use IS_ALIGNED in memory_is_poisoned_8()
@ 2015-09-11 22:47 ` Andrew Morton
0 siblings, 0 replies; 8+ messages in thread
From: Andrew Morton @ 2015-09-11 22:47 UTC (permalink / raw)
To: Xishi Qiu
Cc: Andrey Konovalov, Rusty Russell, Michal Marek, long.wanglong,
Linux MM, LKML, Andrey Ryabinin
On Fri, 11 Sep 2015 10:02:29 +0800 Xishi Qiu <qiuxishi@huawei.com> wrote:
> Use IS_ALIGNED() to determine whether the shadow span two bytes.
> It generates less code and more readable.
>
Please cc Andrey Ryabinin on kasan patches.
> --- a/mm/kasan/kasan.c
> +++ b/mm/kasan/kasan.c
> @@ -120,7 +120,7 @@ static __always_inline bool memory_is_poisoned_8(unsigned long addr)
> if (memory_is_poisoned_1(addr + 7))
> return true;
>
> - if (likely(((addr + 7) & KASAN_SHADOW_MASK) >= 7))
> + if (likely(IS_ALIGNED(addr, 8)))
> return false;
Wouldn't IS_ALIGNED(addr, KASAN_SHADOW_SCALE_SIZE) be more appropriate?
But I'm not really sure what the original code is trying to do.
if ((addr + 7) & 7) >= 7)
can only evaluate true if ((addr + 7) & 7) equals 7, so the ">=" could
be "==".
I think. The code looks a bit weird. A code comment would help.
And how come memory_is_poisoned_16() does IS_ALIGNED(addr, 8)? Should
it be 16?
--
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>
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH] kasan: use IS_ALIGNED in memory_is_poisoned_8()
2015-09-11 22:47 ` Andrew Morton
@ 2015-09-14 1:17 ` Xishi Qiu
-1 siblings, 0 replies; 8+ messages in thread
From: Xishi Qiu @ 2015-09-14 1:17 UTC (permalink / raw)
To: Andrew Morton
Cc: Andrey Konovalov, Rusty Russell, Michal Marek, long.wanglong,
Linux MM, LKML, Andrey Ryabinin
On 2015/9/12 6:47, Andrew Morton wrote:
> On Fri, 11 Sep 2015 10:02:29 +0800 Xishi Qiu <qiuxishi@huawei.com> wrote:
>
>> Use IS_ALIGNED() to determine whether the shadow span two bytes.
>> It generates less code and more readable.
>>
>
> Please cc Andrey Ryabinin on kasan patches.
Sorry, my mistake.
>
>> --- a/mm/kasan/kasan.c
>> +++ b/mm/kasan/kasan.c
>> @@ -120,7 +120,7 @@ static __always_inline bool memory_is_poisoned_8(unsigned long addr)
>> if (memory_is_poisoned_1(addr + 7))
>> return true;
>>
>> - if (likely(((addr + 7) & KASAN_SHADOW_MASK) >= 7))
>> + if (likely(IS_ALIGNED(addr, 8)))
>> return false;
>
> Wouldn't IS_ALIGNED(addr, KASAN_SHADOW_SCALE_SIZE) be more appropriate?
>
OK, I'll send V2.
> But I'm not really sure what the original code is trying to do.
>
> if ((addr + 7) & 7) >= 7)
>
> can only evaluate true if ((addr + 7) & 7) equals 7, so the ">=" could
> be "==".
>
I think it should be "==", the value will not "> 7"
> I think. The code looks a bit weird. A code comment would help.
>
> And how come memory_is_poisoned_16() does IS_ALIGNED(addr, 8)? Should
> it be 16?
No, it is to determine whether the shadow span two bytes(8 bytes, not 16).
Thanks,
Xishi Qiu
>
>
> .
>
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH] kasan: use IS_ALIGNED in memory_is_poisoned_8()
@ 2015-09-14 1:17 ` Xishi Qiu
0 siblings, 0 replies; 8+ messages in thread
From: Xishi Qiu @ 2015-09-14 1:17 UTC (permalink / raw)
To: Andrew Morton
Cc: Andrey Konovalov, Rusty Russell, Michal Marek, long.wanglong,
Linux MM, LKML, Andrey Ryabinin
On 2015/9/12 6:47, Andrew Morton wrote:
> On Fri, 11 Sep 2015 10:02:29 +0800 Xishi Qiu <qiuxishi@huawei.com> wrote:
>
>> Use IS_ALIGNED() to determine whether the shadow span two bytes.
>> It generates less code and more readable.
>>
>
> Please cc Andrey Ryabinin on kasan patches.
Sorry, my mistake.
>
>> --- a/mm/kasan/kasan.c
>> +++ b/mm/kasan/kasan.c
>> @@ -120,7 +120,7 @@ static __always_inline bool memory_is_poisoned_8(unsigned long addr)
>> if (memory_is_poisoned_1(addr + 7))
>> return true;
>>
>> - if (likely(((addr + 7) & KASAN_SHADOW_MASK) >= 7))
>> + if (likely(IS_ALIGNED(addr, 8)))
>> return false;
>
> Wouldn't IS_ALIGNED(addr, KASAN_SHADOW_SCALE_SIZE) be more appropriate?
>
OK, I'll send V2.
> But I'm not really sure what the original code is trying to do.
>
> if ((addr + 7) & 7) >= 7)
>
> can only evaluate true if ((addr + 7) & 7) equals 7, so the ">=" could
> be "==".
>
I think it should be "==", the value will not "> 7"
> I think. The code looks a bit weird. A code comment would help.
>
> And how come memory_is_poisoned_16() does IS_ALIGNED(addr, 8)? Should
> it be 16?
No, it is to determine whether the shadow span two bytes(8 bytes, not 16).
Thanks,
Xishi Qiu
>
>
> .
>
--
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>
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH] kasan: use IS_ALIGNED in memory_is_poisoned_8()
2015-09-11 22:47 ` Andrew Morton
@ 2015-09-14 13:19 ` Andrey Ryabinin
-1 siblings, 0 replies; 8+ messages in thread
From: Andrey Ryabinin @ 2015-09-14 13:19 UTC (permalink / raw)
To: Andrew Morton
Cc: Xishi Qiu, Andrey Konovalov, Rusty Russell, Michal Marek,
long.wanglong, Linux MM, LKML
2015-09-12 1:47 GMT+03:00 Andrew Morton <akpm@linux-foundation.org>:
> On Fri, 11 Sep 2015 10:02:29 +0800 Xishi Qiu <qiuxishi@huawei.com> wrote:
>> - if (likely(((addr + 7) & KASAN_SHADOW_MASK) >= 7))
>> + if (likely(IS_ALIGNED(addr, 8)))
>> return false;
>
> Wouldn't IS_ALIGNED(addr, KASAN_SHADOW_SCALE_SIZE) be more appropriate?
>
> But I'm not really sure what the original code is trying to do.
>
Original code is trying to estimate whether we should check 2 shadow
bytes or just 1 should be enough.
> if ((addr + 7) & 7) >= 7)
>
> can only evaluate true if ((addr + 7) & 7) equals 7, so the ">=" could
> be "==".
>
Yes, it could be "==".
">=" is just for consistency with similar code in memory_is_poisoned_2/4.
If I'm not mistaken generic formula for such check looks like this:
((addr + size - 1) & KASAN_SHADOW_MASK) >= ((size - 1) &
KASAN_SHADOW_MASK)
But when size >= KASAN_SHADOW_SCALE_SIZE we could just check for alignment.
> I think. The code looks a bit weird. A code comment would help.
>
> And how come memory_is_poisoned_16() does IS_ALIGNED(addr, 8)? Should
> it be 16?
>
No, If 16 bytes are 8-byte aligned, then shadow is 2-bytes.
^ permalink raw reply [flat|nested] 8+ messages in thread* Re: [PATCH] kasan: use IS_ALIGNED in memory_is_poisoned_8()
@ 2015-09-14 13:19 ` Andrey Ryabinin
0 siblings, 0 replies; 8+ messages in thread
From: Andrey Ryabinin @ 2015-09-14 13:19 UTC (permalink / raw)
To: Andrew Morton
Cc: Xishi Qiu, Andrey Konovalov, Rusty Russell, Michal Marek,
long.wanglong, Linux MM, LKML
2015-09-12 1:47 GMT+03:00 Andrew Morton <akpm@linux-foundation.org>:
> On Fri, 11 Sep 2015 10:02:29 +0800 Xishi Qiu <qiuxishi@huawei.com> wrote:
>> - if (likely(((addr + 7) & KASAN_SHADOW_MASK) >= 7))
>> + if (likely(IS_ALIGNED(addr, 8)))
>> return false;
>
> Wouldn't IS_ALIGNED(addr, KASAN_SHADOW_SCALE_SIZE) be more appropriate?
>
> But I'm not really sure what the original code is trying to do.
>
Original code is trying to estimate whether we should check 2 shadow
bytes or just 1 should be enough.
> if ((addr + 7) & 7) >= 7)
>
> can only evaluate true if ((addr + 7) & 7) equals 7, so the ">=" could
> be "==".
>
Yes, it could be "==".
">=" is just for consistency with similar code in memory_is_poisoned_2/4.
If I'm not mistaken generic formula for such check looks like this:
((addr + size - 1) & KASAN_SHADOW_MASK) >= ((size - 1) &
KASAN_SHADOW_MASK)
But when size >= KASAN_SHADOW_SCALE_SIZE we could just check for alignment.
> I think. The code looks a bit weird. A code comment would help.
>
> And how come memory_is_poisoned_16() does IS_ALIGNED(addr, 8)? Should
> it be 16?
>
No, If 16 bytes are 8-byte aligned, then shadow is 2-bytes.
--
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>
^ permalink raw reply [flat|nested] 8+ messages in thread