2014-11-04 10:04:57

by Maxime Coquelin

Subject: [PATCH v2] bitops: Fix shift overflow in GENMASK macros

On some 32 bits architectures, including x86, GENMASK(31, 0) returns 0
instead of the expected ~0UL.

This is the same on some 64 bits architectures with GENMASK_ULL(63, 0).

This is due to an overflow in the shift operand, 1 << 32 for GENMASK,
1 << 64 for GENMASK_ULL.

Fixes: 10ef6b0dffe404bcc54e94cb2ca1a5b18445a66b
Cc: <[email protected]> #v3.13+
Reported-by: Eric Paire <[email protected]>
Suggested-by: Peter Zijlstra <[email protected]>
Signed-off-by: Maxime Coquelin <[email protected]>
---
include/linux/bitops.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/include/linux/bitops.h b/include/linux/bitops.h
index be5fd38..a49d7b7 100644
--- a/include/linux/bitops.h
+++ b/include/linux/bitops.h
@@ -18,8 +18,8 @@
* position @h. For example
* GENMASK_ULL(39, 21) gives us the 64bit vector 0x000000ffffe00000.
*/
-#define GENMASK(h, l) (((U32_C(1) << ((h) - (l) + 1)) - 1) << (l))
-#define GENMASK_ULL(h, l) (((U64_C(1) << ((h) - (l) + 1)) - 1) << (l))
+#define GENMASK(h, l) ((~0UL >> (BITS_PER_LONG - (h - l + 1))) << l)
+#define GENMASK_ULL(h, l) ((~0ULL >> (BITS_PER_LONG_LONG - (h - l + 1))) << l)

extern unsigned int __sw_hweight8(unsigned int w);
extern unsigned int __sw_hweight16(unsigned int w);
--
1.9.1

2014-11-04 10:44:29

by Peter Zijlstra

Subject: Re: [PATCH v2] bitops: Fix shift overflow in GENMASK macros

On Tue, Nov 04, 2014 at 11:03:57AM +0100, Maxime COQUELIN wrote:

> -#define GENMASK(h, l) (((U32_C(1) << ((h) - (l) + 1)) - 1) << (l))
> -#define GENMASK_ULL(h, l) (((U64_C(1) << ((h) - (l) + 1)) - 1) << (l))
> +#define GENMASK(h, l) ((~0UL >> (BITS_PER_LONG - (h - l + 1))) << l)
> +#define GENMASK_ULL(h, l) ((~0ULL >> (BITS_PER_LONG_LONG - (h - l + 1))) << l)

OK, so you need to keep the (h) and (l) bits, macro arguments should be
wrapped in seemingly superfluous braces in order to preserve precedence
on expansion.

My bad for not explicitly doing that when suggesting the alternative.

2014-11-04 10:48:26

by Maxime Coquelin

Subject: Re: [PATCH v2] bitops: Fix shift overflow in GENMASK macros

On 11/04/2014 11:44 AM, Peter Zijlstra wrote:
> On Tue, Nov 04, 2014 at 11:03:57AM +0100, Maxime COQUELIN wrote:
>
>> -#define GENMASK(h, l) (((U32_C(1) << ((h) - (l) + 1)) - 1) << (l))
>> -#define GENMASK_ULL(h, l) (((U64_C(1) << ((h) - (l) + 1)) - 1) << (l))
>> +#define GENMASK(h, l) ((~0UL >> (BITS_PER_LONG - (h - l + 1))) << l)
>> +#define GENMASK_ULL(h, l) ((~0ULL >> (BITS_PER_LONG_LONG - (h - l + 1))) << l)
> OK, so you need to keep the (h) and (l) bits, macro arguments should be
> wrapped in seemingly superfluous braces in order to preserve precedence
> on expansion.
You're right, I should have seen this..
>
> My bad for not explicitly doing that when suggesting the alternative.
Not a problem, v3 is coming.

Maxime