Re: [PATCH] x86, AMD: Correct F15h IC aliasing issue

["H. Peter Anvin" <hpa@zytor.com>]

On 07/27/2011 10:10 AM, Borislav Petkov wrote:
> 
> How about something like the following instead - it should take care
> of all your bitmask generating needs. There are also a couple of
> GENMASK/BITMASK identical definitions around the tree which can be
> unified while I'm at it too.
> 
> diff --git a/include/linux/bitops.h b/include/linux/bitops.h
> index a3ef66a..b1970e3 100644
> --- a/include/linux/bitops.h
> +++ b/include/linux/bitops.h
> @@ -6,6 +6,19 @@
>  #define BIT(nr)                        (1UL << (nr))
>  #define BIT_MASK(nr)           (1UL << ((nr) % BITS_PER_LONG))
>  #define BIT_WORD(nr)           ((nr) / BITS_PER_LONG)
> +
> +/*
> + * Create a contiguous bitmask starting at bit position @lo and ending at
> + * position @hi. For example
> + *
> + * GENMASK_ULL(21, 39) gives us the 64bit vector 0x000000ffffe00000.
> + */
> +#define _GENMASK_T(cast, type, lo, hi) \
> +       (((cast)(1##type << ((hi) - (lo) + 1)) - 1) << (lo))
> +#define GENMASK(lo, hi)                _GENMASK_T(unsigned, U, lo, hi)
> +#define GENMASK_UL(lo, hi)     _GENMASK_T(unsigned long, UL, lo, hi)
> +#define GENMASK_ULL(lo, hi)    _GENMASK_T(unsigned long long, ULL, lo, hi)
> +

These really need to be usable from assembly language, too (in which
case you of course need to not have the cast and suffix), so it probably
should be defined in <linux/const.h> with the other constant macros.

	-hpa

-- 
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.