122 lines
4.5 KiB
C
122 lines
4.5 KiB
C
|
/*
|
||
|
* Copyright 2011-present Facebook, Inc.
|
||
|
*
|
||
|
* Licensed under the Apache License, Version 2.0 (the "License");
|
||
|
* you may not use this file except in compliance with the License.
|
||
|
* You may obtain a copy of the License at
|
||
|
*
|
||
|
* http://www.apache.org/licenses/LICENSE-2.0
|
||
|
*
|
||
|
* Unless required by applicable law or agreed to in writing, software
|
||
|
* distributed under the License is distributed on an "AS IS" BASIS,
|
||
|
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||
|
* See the License for the specific language governing permissions and
|
||
|
* limitations under the License.
|
||
|
*/
|
||
|
|
||
|
#pragma once
|
||
|
|
||
|
#include <cstddef>
|
||
|
|
||
|
#include <folly/Portability.h>
|
||
|
|
||
|
namespace folly {
|
||
|
|
||
|
namespace detail {
|
||
|
|
||
|
// Implemented this way because of a bug in Clang for ARMv7, which gives the
|
||
|
// wrong result for `alignof` a `union` with a field of each scalar type.
|
||
|
constexpr size_t max_align_(std::size_t a) {
|
||
|
return a;
|
||
|
}
|
||
|
template <typename... Es>
|
||
|
constexpr std::size_t max_align_(std::size_t a, std::size_t e, Es... es) {
|
||
|
return !(a < e) ? a : max_align_(e, es...);
|
||
|
}
|
||
|
template <typename... Ts>
|
||
|
struct max_align_t_ {
|
||
|
static constexpr std::size_t value = max_align_(0u, alignof(Ts)...);
|
||
|
};
|
||
|
using max_align_v_ = max_align_t_<
|
||
|
long double,
|
||
|
double,
|
||
|
float,
|
||
|
long long int,
|
||
|
long int,
|
||
|
int,
|
||
|
short int,
|
||
|
bool,
|
||
|
char,
|
||
|
char16_t,
|
||
|
char32_t,
|
||
|
wchar_t,
|
||
|
void*,
|
||
|
std::max_align_t>;
|
||
|
|
||
|
} // namespace detail
|
||
|
|
||
|
// max_align_v is the alignment of max_align_t.
|
||
|
//
|
||
|
// max_align_t is a type which is aligned at least as strictly as the
|
||
|
// most-aligned basic type (see the specification of std::max_align_t). This
|
||
|
// implementation exists because 32-bit iOS platforms have a broken
|
||
|
// std::max_align_t (see below).
|
||
|
//
|
||
|
// You should refer to this as `::folly::max_align_t` in portable code, even if
|
||
|
// you have `using namespace folly;` because C11 defines a global namespace
|
||
|
// `max_align_t` type.
|
||
|
//
|
||
|
// To be certain, we consider every non-void fundamental type specified by the
|
||
|
// standard. On most platforms `long double` would be enough, but iOS 32-bit
|
||
|
// has an 8-byte aligned `double` and `long long int` and a 4-byte aligned
|
||
|
// `long double`.
|
||
|
//
|
||
|
// So far we've covered locals and other non-allocated storage, but we also need
|
||
|
// confidence that allocated storage from `malloc`, `new`, etc will also be
|
||
|
// suitable for objects with this alignment requirement.
|
||
|
//
|
||
|
// Apple document that their implementation of malloc will issue 16-byte
|
||
|
// granularity chunks for small allocations (large allocations are page-size
|
||
|
// granularity and page-aligned). We think that allocated storage will be
|
||
|
// suitable for these objects based on the following assumptions:
|
||
|
//
|
||
|
// 1. 16-byte granularity also means 16-byte aligned.
|
||
|
// 2. `new` and other allocators follow the `malloc` rules.
|
||
|
//
|
||
|
// We also have some anecdotal evidence: we don't see lots of misaligned-storage
|
||
|
// crashes on 32-bit iOS apps that use `double`.
|
||
|
//
|
||
|
// Apple's allocation reference: http://bit.ly/malloc-small
|
||
|
constexpr std::size_t max_align_v = detail::max_align_v_::value;
|
||
|
struct alignas(max_align_v) max_align_t {};
|
||
|
|
||
|
// Memory locations within the same cache line are subject to destructive
|
||
|
// interference, also known as false sharing, which is when concurrent
|
||
|
// accesses to these different memory locations from different cores, where at
|
||
|
// least one of the concurrent accesses is or involves a store operation,
|
||
|
// induce contention and harm performance.
|
||
|
//
|
||
|
// Microbenchmarks indicate that pairs of cache lines also see destructive
|
||
|
// interference under heavy use of atomic operations, as observed for atomic
|
||
|
// increment on Sandy Bridge.
|
||
|
//
|
||
|
// We assume a cache line size of 64, so we use a cache line pair size of 128
|
||
|
// to avoid destructive interference.
|
||
|
//
|
||
|
// mimic: std::hardware_destructive_interference_size, C++17
|
||
|
constexpr std::size_t hardware_destructive_interference_size =
|
||
|
kIsArchArm ? 64 : 128;
|
||
|
static_assert(hardware_destructive_interference_size >= max_align_v, "math?");
|
||
|
|
||
|
// Memory locations within the same cache line are subject to constructive
|
||
|
// interference, also known as true sharing, which is when accesses to some
|
||
|
// memory locations induce all memory locations within the same cache line to
|
||
|
// be cached, benefiting subsequent accesses to different memory locations
|
||
|
// within the same cache line and heping performance.
|
||
|
//
|
||
|
// mimic: std::hardware_constructive_interference_size, C++17
|
||
|
constexpr std::size_t hardware_constructive_interference_size = 64;
|
||
|
static_assert(hardware_constructive_interference_size >= max_align_v, "math?");
|
||
|
|
||
|
} // namespace folly
|