302 lines
8.6 KiB
Groff
302 lines
8.6 KiB
Groff
.\" Copyright (c) 2011 Ed Schouten <ed@FreeBSD.org>
|
|
.\" All rights reserved.
|
|
.\"
|
|
.\" Redistribution and use in source and binary forms, with or without
|
|
.\" modification, are permitted provided that the following conditions
|
|
.\" are met:
|
|
.\" 1. Redistributions of source code must retain the above copyright
|
|
.\" notice, this list of conditions and the following disclaimer.
|
|
.\" 2. Redistributions in binary form must reproduce the above copyright
|
|
.\" notice, this list of conditions and the following disclaimer in the
|
|
.\" documentation and/or other materials provided with the distribution.
|
|
.\"
|
|
.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
|
.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
|
.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|
.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|
.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|
.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
.\" SUCH DAMAGE.
|
|
.\"
|
|
.\" $FreeBSD$
|
|
.\"
|
|
.Dd December 27, 2011
|
|
.Dt ATOMIC_VAR_INIT 3
|
|
.Os
|
|
.Sh NAME
|
|
.Nm ATOMIC_VAR_INIT ,
|
|
.Nm atomic_init ,
|
|
.Nm atomic_load ,
|
|
.Nm atomic_store ,
|
|
.Nm atomic_exchange ,
|
|
.Nm atomic_compare_exchange_strong ,
|
|
.Nm atomic_compare_exchange_weak ,
|
|
.Nm atomic_fetch_add ,
|
|
.Nm atomic_fetch_and ,
|
|
.Nm atomic_fetch_or ,
|
|
.Nm atomic_fetch_sub ,
|
|
.Nm atomic_fetch_xor ,
|
|
.Nm atomic_is_lock_free
|
|
.Nd type-generic atomic operations
|
|
.Sh SYNOPSIS
|
|
.In stdatomic.h
|
|
.Pp
|
|
_Atomic(T)
|
|
.Fa v
|
|
= ATOMIC_VAR_INIT(c);
|
|
.Ft void
|
|
.Fn atomic_init "_Atomic(T) *object" "T value"
|
|
.Ft T
|
|
.Fn atomic_load "_Atomic(T) *object"
|
|
.Ft T
|
|
.Fn atomic_load_explicit "_Atomic(T) *object" "memory_order order"
|
|
.Ft void
|
|
.Fn atomic_store "_Atomic(T) *object" "T desired"
|
|
.Ft void
|
|
.Fn atomic_store_explicit "_Atomic(T) *object" "T desired" "memory_order order"
|
|
.Ft T
|
|
.Fn atomic_exchange "_Atomic(T) *object" "T desired"
|
|
.Ft T
|
|
.Fn atomic_exchange_explicit "_Atomic(T) *object" "T desired" "memory_order order"
|
|
.Ft _Bool
|
|
.Fn atomic_compare_exchange_strong "_Atomic(T) *object" "T *expected" "T desired"
|
|
.Ft _Bool
|
|
.Fn atomic_compare_exchange_strong_explicit "_Atomic(T) *object" "T *expected" "T desired" "memory_order success" "memory_order failure"
|
|
.Ft _Bool
|
|
.Fn atomic_compare_exchange_weak "_Atomic(T) *object" "T *expected" "T desired"
|
|
.Ft _Bool
|
|
.Fn atomic_compare_exchange_weak_explicit "_Atomic(T) *object" "T *expected" "T desired" "memory_order success" "memory_order failure"
|
|
.Ft T
|
|
.Fn atomic_fetch_add "_Atomic(T) *object" "T operand"
|
|
.Ft T
|
|
.Fn atomic_fetch_add_explicit "_Atomic(T) *object" "T operand" "memory_order order"
|
|
.Ft T
|
|
.Fn atomic_fetch_and "_Atomic(T) *object" "T operand"
|
|
.Ft T
|
|
.Fn atomic_fetch_and_explicit "_Atomic(T) *object" "T operand" "memory_order order"
|
|
.Ft T
|
|
.Fn atomic_fetch_or "_Atomic(T) *object" "T operand"
|
|
.Ft T
|
|
.Fn atomic_fetch_or_explicit "_Atomic(T) *object" "T operand" "memory_order order"
|
|
.Ft T
|
|
.Fn atomic_fetch_sub "_Atomic(T) *object" "T operand"
|
|
.Ft T
|
|
.Fn atomic_fetch_sub_explicit "_Atomic(T) *object" "T operand" "memory_order order"
|
|
.Ft T
|
|
.Fn atomic_fetch_xor "_Atomic(T) *object" "T operand"
|
|
.Ft T
|
|
.Fn atomic_fetch_xor_explicit "_Atomic(T) *object" "T operand" "memory_order order"
|
|
.Ft _Bool
|
|
.Fn atomic_is_lock_free "const _Atomic(T) *object"
|
|
.Sh DESCRIPTION
|
|
The header
|
|
.In stdatomic.h
|
|
provides type-generic macros for atomic operations.
|
|
Atomic operations can be used by multithreaded programs to provide
|
|
shared variables between threads that in most cases may be modified
|
|
without acquiring locks.
|
|
.Pp
|
|
Atomic variables are declared using the
|
|
.Fn _Atomic
|
|
type specifier.
|
|
These variables are not type-compatible with their non-atomic
|
|
counterparts.
|
|
Depending on the compiler used, atomic variables may be opaque and can
|
|
therefore only be influenced using the macros described.
|
|
.Pp
|
|
The
|
|
.Fn atomic_init
|
|
macro initializes the atomic variable
|
|
.Fa object
|
|
with a
|
|
.Fa value .
|
|
Atomic variables can be initialized while being declared using
|
|
.Fn ATOMIC_VAR_INIT .
|
|
.Pp
|
|
The
|
|
.Fn atomic_load
|
|
macro returns the value of atomic variable
|
|
.Fa object .
|
|
The
|
|
.Fn atomic_store
|
|
macro sets the atomic variable
|
|
.Fa object
|
|
to its
|
|
.Fa desired
|
|
value.
|
|
.Pp
|
|
The
|
|
.Fn atomic_exchange
|
|
macro combines the behaviour of
|
|
.Fn atomic_load
|
|
and
|
|
.Fn atomic_store .
|
|
It sets the atomic variable
|
|
.Fa object
|
|
to its desired
|
|
.Fa value
|
|
and returns the original contents of the atomic variable.
|
|
.Pp
|
|
The
|
|
.Fn atomic_compare_exchange_strong
|
|
macro stores a
|
|
.Fa desired
|
|
value into atomic variable
|
|
.Fa object ,
|
|
only if the atomic variable is equal to its
|
|
.Fa expected
|
|
value.
|
|
Upon success, the macro returns
|
|
.Dv true .
|
|
Upon failure, the
|
|
.Fa desired
|
|
value is overwritten with the value of the atomic variable and
|
|
.Dv false
|
|
is returned.
|
|
The
|
|
.Fn atomic_compare_exchange_weak
|
|
macro is identical to
|
|
.Fn atomic_compare_exchange_strong ,
|
|
but is allowed to fail even if atomic variable
|
|
.Fa object
|
|
is equal to its
|
|
.Fa expected
|
|
value.
|
|
.Pp
|
|
The
|
|
.Fn atomic_fetch_add
|
|
macro adds the value
|
|
.Fa operand
|
|
to atomic variable
|
|
.Fa object
|
|
and returns the original contents of the atomic variable.
|
|
.Pp
|
|
The
|
|
.Fn atomic_fetch_and
|
|
macro applies the
|
|
.Em and
|
|
operator to atomic variable
|
|
.Fa object
|
|
and
|
|
.Fa operand
|
|
and stores the value into
|
|
.Fa object ,
|
|
while returning the original contents of the atomic variable.
|
|
.Pp
|
|
The
|
|
.Fn atomic_fetch_or
|
|
macro applies the
|
|
.Em or
|
|
operator to atomic variable
|
|
.Fa object
|
|
and
|
|
.Fa operand
|
|
and stores the value into
|
|
.Fa object ,
|
|
while returning the original contents of the atomic variable.
|
|
.Pp
|
|
The
|
|
.Fn atomic_fetch_sub
|
|
macro subtracts the value
|
|
.Fa operand
|
|
from atomic variable
|
|
.Fa object
|
|
and returns the original contents of the atomic variable.
|
|
.Pp
|
|
The
|
|
.Fn atomic_fetch_xor
|
|
macro applies the
|
|
.Em xor
|
|
operator to atomic variable
|
|
.Fa object
|
|
and
|
|
.Fa operand
|
|
and stores the value into
|
|
.Fa object ,
|
|
while returning the original contents of the atomic variable.
|
|
.Pp
|
|
The
|
|
.Fn atomic_is_lock_free
|
|
macro returns whether atomic variable
|
|
.Fa object
|
|
uses locks when using atomic operations.
|
|
.Sh BARRIERS
|
|
The atomic operations described previously are implemented in such a way
|
|
that they disallow both the compiler and the executing processor to
|
|
re-order any nearby memory operations across the atomic operation.
|
|
In certain cases this behaviour may cause suboptimal performance.
|
|
To mitigate this, every atomic operation has an
|
|
.Fn _explicit
|
|
version that allows the re-ordering to be configured.
|
|
.Pp
|
|
The
|
|
.Fa order
|
|
parameter of these
|
|
.Fn _explicit
|
|
macros can have one of the following values.
|
|
.Bl -tag -width memory_order_relaxed
|
|
.It Dv memory_order_relaxed
|
|
No operation orders memory.
|
|
.It Dv memory_order_consume
|
|
Perform consume operation.
|
|
.It Dv memory_order_acquire
|
|
Acquire fence.
|
|
.It Dv memory_order_release
|
|
Release fence.
|
|
.It Dv memory_order_acq_rel
|
|
Acquire and release fence.
|
|
.It Dv memory_order_seq_cst
|
|
Sequentially consistent acquire and release fence.
|
|
.El
|
|
.Pp
|
|
The previously described macros are identical to the
|
|
.Fn _explicit
|
|
macros, when
|
|
.Fa order
|
|
is
|
|
.Dv memory_order_seq_cst .
|
|
.Sh COMPILER SUPPORT
|
|
These atomic operations are typically implemented by the compiler, as
|
|
they must be implemented type-generically and must often use special
|
|
hardware instructions.
|
|
As this interface has not been adopted by most compilers yet, the
|
|
.In stdatomic.h
|
|
header implements these macros on top of existing compiler intrinsics to
|
|
provide forward compatibility.
|
|
.Pp
|
|
This means that certain aspects of the interface, such as support for
|
|
different barrier types may simply be ignored.
|
|
When using GCC, all atomic operations are executed as if they are using
|
|
.Dv memory_order_seq_cst .
|
|
.Pp
|
|
Instead of using the atomic operations provided by this interface,
|
|
.St -isoC-2011
|
|
allows the atomic variables to be modified directly using built-in
|
|
language operators.
|
|
This behaviour cannot be emulated for older compilers.
|
|
To prevent unintended non-atomic access to these variables, this header
|
|
file places the atomic variable in a structure when using an older
|
|
compiler.
|
|
.Pp
|
|
When using GCC on architectures on which it lacks support for built-in
|
|
atomic intrinsics, these macros may emit function calls to fallback
|
|
routines.
|
|
These fallback routines are only implemented for 32-bits and 64-bits
|
|
datatypes, if supported by the CPU.
|
|
.Sh SEE ALSO
|
|
.Xr pthread 3 ,
|
|
.Xr atomic 9
|
|
.Sh STANDARDS
|
|
These macros attempt to conform to
|
|
.St -isoC-2011 .
|
|
.Sh HISTORY
|
|
These macros appeared in
|
|
.Fx 10.0 .
|
|
.Sh AUTHORS
|
|
.An Ed Schouten Aq Mt ed@FreeBSD.org
|
|
.An David Chisnall Aq Mt theraven@FreeBSD.org
|