On Sat, Mar 01, 2025 at 10:23:54PM +0800, Kuan-Wei Chiu wrote:
> Introduce parity16(), parity32(), and parity64() functions for
> computing parity on 16-bit, 32-bit, and 64-bit integers, respectively.
> These functions use __builtin_parity() or __builtin_parityll() when
> available, ensuring efficient computation. If the input is a
> compile-time constant, they expand using the _parity_const() macro to
> allow constant folding.
>
> These additions provide parity computation helpers for larger integer
> types, ensuring consistency and performance across different
> bit-widths.
>
> Co-developed-by: Yu-Chun Lin <eleanor15x@xxxxxxxxx>
> Signed-off-by: Yu-Chun Lin <eleanor15x@xxxxxxxxx>
> Signed-off-by: Kuan-Wei Chiu <visitorckw@xxxxxxxxx>
> ---
> include/linux/bitops.h | 63 ++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 63 insertions(+)
>
> diff --git a/include/linux/bitops.h b/include/linux/bitops.h
> index 4c307d9c1545..41e9e7fb894b 100644
> --- a/include/linux/bitops.h
> +++ b/include/linux/bitops.h
> @@ -276,6 +276,69 @@ static inline __attribute_const__ int parity8(u8 val)
> return __builtin_constant_p(val) ? _parity_const(val) : _parity8(val);
> }
>
> +#ifndef _parity16
> +static inline __attribute_const__ int _parity16(u16 val)
> +{
> + return __builtin_parity(val);
> +}
> +#endif
> +
> +/**
> + * parity16 - get the parity of an u16 value
> + * @value: the value to be examined
nit: This really ought to be @val to match the function signature.
Likewise for parity8, parity32, and parity64.
> + *
> + * Determine the parity of the u16 argument.
> + *
> + * Returns:
> + * 0 for even parity, 1 for odd parity
> + */
> +static inline __attribute_const__ int parity16(u16 val)
> +{
> + return __builtin_constant_p(val) ? _parity_const(val) : _parity16(val);
> +}
...
