On 04/04/17 23:08, Brian Brooks wrote:
> On 04/04 22:17:48, Maxim Uvarov wrote:
>> On 04/04/17 21:48, Brian Brooks wrote:
>>> Signed-off-by: Ola Liljedahl <ola.liljed...@arm.com>
>>> Reviewed-by: Brian Brooks <brian.bro...@arm.com>
>>> Reviewed-by: Honnappa Nagarahalli <honnappa.nagaraha...@arm.com>
>>> ---
>>> platform/linux-generic/include/odp_bitset.h | 155
>>> ++++++++++++++++++++++++++++
>>> 1 file changed, 155 insertions(+)
>>> create mode 100644 platform/linux-generic/include/odp_bitset.h
>>>
>>> diff --git a/platform/linux-generic/include/odp_bitset.h
>>> b/platform/linux-generic/include/odp_bitset.h
>>> new file mode 100644
>>> index 00000000..db004267
>>> --- /dev/null
>>> +++ b/platform/linux-generic/include/odp_bitset.h
>>> @@ -0,0 +1,155 @@
>>> +/* Copyright (c) 2017, ARM Limited
>>> + * All rights reserved.
>>> + *
>>> + * SPDX-License-Identifier: BSD-3-Clause
>>> + */
>>
>>
>> Should be Linaro copyright and patches needs to be resend from Linaro
>> address.
>
> Can you raise this concern through the appropriate channels?
>
>>> +
>>> +#ifndef _ODP_BITSET_H_
>>> +#define _ODP_BITSET_H_
>>> +
>>> +/******************************************************************************
>>> + * bitset abstract data type
>>> +
>>> *****************************************************************************/
>>> +/* This could be a struct of scalars to support larger bit sets */
>>> +
>>> +#if ATOM_BITSET_SIZE <= 32
>>> +
>>> +typedef uint32_t bitset_t;
>>> +
>>> +static inline bitset_t bitset_mask(uint32_t bit)
>>> +{
>>> + return 1UL << bit;
>>> +}
>>> +
>>> +/* Return first-bit-set with StdC ffs() semantics */
>>> +static inline uint32_t bitset_ffs(bitset_t b)
>>> +{
>>> + return __builtin_ffsl(b);
>>> +}
>>> +
>>> +/* Load-exclusive with memory ordering */
>>> +static inline bitset_t bitset_ldex(bitset_t *bs, int mo)
>>> +{
>>> + return LDXR32(bs, mo);
>>
>> is it arm specific?
>
> Can you please explain?
>
is macros LDXR32 arm specific? Or that can be also compiled on other
arches (x86, mips, ppc)?
Maxim.
>>> +}
>>> +
>>> +#elif ATOM_BITSET_SIZE <= 64
>>> +
>>> +typedef uint64_t bitset_t;
>>> +
>>> +static inline bitset_t bitset_mask(uint32_t bit)
>>> +{
>>> + return 1ULL << bit;
>>> +}
>>> +
>>> +/* Return first-bit-set with StdC ffs() semantics */
>>> +static inline uint32_t bitset_ffs(bitset_t b)
>>> +{
>>> + return __builtin_ffsll(b);
>>> +}
>>> +
>>> +/* Load-exclusive with memory ordering */
>>> +static inline bitset_t bitset_ldex(bitset_t *bs, int mo)
>>> +{
>>> + return LDXR64(bs, mo);
>>> +}
>>> +
>>> +#elif ATOM_BITSET_SIZE <= 128
>>> +
>>> +#if __SIZEOF_INT128__ == 16
>>> +typedef unsigned __int128 bitset_t;
>>> +
>>
>> so bitset_t can be 32 or 64 or 128. That might be difficult. And sort of
>> this needs to be done at ./configure stage, I think.
>
> Can you please explain?
>
>>> +static inline bitset_t bitset_mask(uint32_t bit)
>>> +{
>>> + if (bit < 64)
>>> + return 1ULL << bit;
>>> + else
>>> + return (unsigned __int128)(1ULL << (bit - 64)) << 64;
>>> +}
>>
>> you set 32 bit mast to 128 bit value, other 96 bits unreachable. Why is
>> that needed?
>
> 'bit' represents a bit position in the scalar type. Unfortunately GCC
> doesn't support 128-bit numeric literals, so you have to generate the
> mask using 'ull' and then cast that to the scalar type and then shift
> that even further. Please take another look.
>
>>> +
>>> +/* Return first-bit-set with StdC ffs() semantics */
>>> +static inline uint32_t bitset_ffs(bitset_t b)
>>> +{
>>> + if ((uint64_t)b != 0)
>>> + return __builtin_ffsll((uint64_t)b);
>>> + else if ((b >> 64) != 0)
>>> + return __builtin_ffsll((uint64_t)(b >> 64)) + 64;
>>> + else
>>> + return 0;
>>> +}
>>> +
>>> +/* Load-exclusive with memory ordering */
>>> +static inline bitset_t bitset_ldex(bitset_t *bs, int mo)
>>> +{
>>> + return LDXR128(bs, mo);
>>> +}
>>> +
>>> +#else
>>> +#error __int128 not supported by compiler
>>
>>
>> this should be captured at configure stage, not while compiling project
>> code.
>
> When you use the macros provided by the compiler, e.g. system-specific
> macros, you can eliminate the dependency on the build system.. for achieving
> the same thing. This is an advantage.
>
>> Maxim.
>>
>>> +#endif
>>> +
>>> +#else
>>> +#error Unsupported size of bit sets (ATOM_BITSET_SIZE)
>>> +#endif
>>> +
>>> +/* Atomic load with memory ordering */
>>> +static inline bitset_t atom_bitset_load(bitset_t *bs, int mo)
>>> +{
>>> + return __atomic_load_n(bs, mo);
>>> +}
>>> +
>>> +/* Atomic bit set with memory ordering */
>>> +static inline void atom_bitset_set(bitset_t *bs, uint32_t bit, int mo)
>>> +{
>>> + (void)__atomic_fetch_or(bs, bitset_mask(bit), mo);
>>> +}
>>> +
>>> +/* Atomic bit clear with memory ordering */
>>> +static inline void atom_bitset_clr(bitset_t *bs, uint32_t bit, int mo)
>>> +{
>>> + (void)__atomic_fetch_and(bs, ~bitset_mask(bit), mo);
>>> +}
>>> +
>>> +/* Atomic exchange with memory ordering */
>>> +static inline bitset_t atom_bitset_xchg(bitset_t *bs, bitset_t neu, int mo)
>>> +{
>>> + return __atomic_exchange_n(bs, neu, mo);
>>> +}
>>> +
>>> +/* Return a & ~b */
>>> +static inline bitset_t bitset_andn(bitset_t a, bitset_t b)
>>> +{
>>> + return a & ~b;
>>> +}
>>> +
>>> +static inline bool bitset_is_eql(bitset_t a, bitset_t b)
>>> +{
>>> + return a == b;
>>> +}
>>> +
>>> +static inline bitset_t bitset_clr(bitset_t bs, uint32_t bit)
>>> +{
>>> + return bs & ~bitset_mask(bit);
>>> +}
>>> +
>>> +static inline bitset_t bitset_set(bitset_t bs, uint32_t bit)
>>> +{
>>> + return bs | bitset_mask(bit);
>>> +}
>>> +
>>> +static inline bitset_t bitset_null(void)
>>> +{
>>> + return 0U;
>>> +}
>>> +
>>> +static inline bool bitset_is_null(bitset_t a)
>>> +{
>>> + return a == 0U;
>>> +}
>>> +
>>> +static inline bool bitset_is_set(bitset_t a, uint32_t bit)
>>> +{
>>> + return (a & bitset_mask(bit)) != 0;
>>> +}
>>> +
>>> +#endif
>>>
>>
