2014-11-04 10:59:09

by Maxime Coquelin

Subject: [PATCH v3] 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 | 7 +++++--
1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/include/linux/bitops.h b/include/linux/bitops.h
index be5fd38..c2ce66c 100644
--- a/include/linux/bitops.h
+++ b/include/linux/bitops.h
@@ -18,8 +18,11 @@
* 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 13:27:42

by Peter Zijlstra

Subject: Re: [PATCH v3] bitops: Fix shift overflow in GENMASK macros

On Tue, Nov 04, 2014 at 11:58:17AM +0100, Maxime COQUELIN wrote:
> 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]>


There doesn't appear to be a clear maintainer for this file, so I'll try
and stick in in tip.

Thanks!

2014-11-05 11:10:50

by Rasmus Villemoes

Subject: Re: [PATCH v3] bitops: Fix shift overflow in GENMASK macros

On Tue, Nov 04 2014, Maxime COQUELIN <[email protected]> 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))

Slightly bikeshedding here, but may I suggest spelling it

(((~0UL) << (l)) & (~0UL >> (BITS_PER_LONG - 1 - (h))))

? This will ensure the arguments are expanded once each, and will, IMHO,
DTRT if/when l > h (namely, yield 0).

Rasmus

2014-11-05 15:48:11

by Maxime Coquelin

Subject: Re: [PATCH v3] bitops: Fix shift overflow in GENMASK macros

On 11/05/2014 12:10 PM, Rasmus Villemoes wrote:
> On Tue, Nov 04 2014, Maxime COQUELIN <[email protected]> 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))
> Slightly bikeshedding here, but may I suggest spelling it
>
> (((~0UL) << (l)) & (~0UL >> (BITS_PER_LONG - 1 - (h))))
>
> ? This will ensure the arguments are expanded once each, and will, IMHO,
> DTRT if/when l > h (namely, yield 0).
That's a good point.

Peter, what is your view?

Kind regards,
Maxime
>
> Rasmus

2014-11-05 16:45:08

by Peter Zijlstra

Subject: Re: [PATCH v3] bitops: Fix shift overflow in GENMASK macros

On Wed, Nov 05, 2014 at 04:47:14PM +0100, Maxime Coquelin wrote:
>
> On 11/05/2014 12:10 PM, Rasmus Villemoes wrote:
> >On Tue, Nov 04 2014, Maxime COQUELIN <[email protected]> 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))
> >Slightly bikeshedding here, but may I suggest spelling it
> >
> > (((~0UL) << (l)) & (~0UL >> (BITS_PER_LONG - 1 - (h))))
> >
> >? This will ensure the arguments are expanded once each, and will, IMHO,
> >DTRT if/when l > h (namely, yield 0).
> That's a good point.
>
> Peter, what is your view?

I'm fine with that, feel free to post another version and I'll replace
v3.

2014-11-06 09:57:42

by Maxime Coquelin

Subject: Re: [PATCH v3] bitops: Fix shift overflow in GENMASK macros

On 11/05/2014 05:44 PM, Peter Zijlstra wrote:
> On Wed, Nov 05, 2014 at 04:47:14PM +0100, Maxime Coquelin wrote:
>> On 11/05/2014 12:10 PM, Rasmus Villemoes wrote:
>>> On Tue, Nov 04 2014, Maxime COQUELIN <[email protected]> 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))
>>> Slightly bikeshedding here, but may I suggest spelling it
>>>
>>> (((~0UL) << (l)) & (~0UL >> (BITS_PER_LONG - 1 - (h))))
>>>
>>> ? This will ensure the arguments are expanded once each, and will, IMHO,
>>> DTRT if/when l > h (namely, yield 0).
>> That's a good point.
>>
>> Peter, what is your view?
> I'm fine with that, feel free to post another version and I'll replace
> v3.

Thanks Peter, v4 just sent.

Regards,
Maxime