2 * Copyright © 2015 Intel
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
27 #include "util/futex.h"
28 #include "util/macros.h"
30 #include "c11/threads.h"
32 #if defined(__GNUC__) && defined(HAVE_LINUX_FUTEX_H)
34 /* mtx_t - Fast, simple mutex
36 * While modern pthread mutexes are very fast (implemented using futex), they
37 * still incur a call to an external DSO and overhead of the generality and
38 * features of pthread mutexes. Most mutexes in mesa only needs lock/unlock,
39 * and the idea here is that we can inline the atomic operation and make the
40 * fast case just two intructions. Mutexes are subtle and finicky to
41 * implement, so we carefully copy the implementation from Ulrich Dreppers
42 * well-written and well-reviewed paper:
44 * "Futexes Are Tricky"
45 * http://www.akkadia.org/drepper/futex.pdf
47 * We implement "mutex3", which gives us a mutex that has no syscalls on
48 * uncontended lock or unlock. Further, the uncontended case boils down to a
49 * locked cmpxchg and an untaken branch, the uncontended unlock is just a
50 * locked decr and an untaken branch. We use __builtin_expect() to indicate
51 * that contention is unlikely so that gcc will put the contention code out of
54 * A fast mutex only supports lock/unlock, can't be recursive or used with
55 * condition variables.
62 #define _SIMPLE_MTX_INITIALIZER_NP { 0 }
64 #define _SIMPLE_MTX_INVALID_VALUE 0xd0d0d0d0
67 simple_mtx_init(simple_mtx_t
*mtx
, ASSERTED
int type
)
69 assert(type
== mtx_plain
);
75 simple_mtx_destroy(ASSERTED simple_mtx_t
*mtx
)
78 mtx
->val
= _SIMPLE_MTX_INVALID_VALUE
;
83 simple_mtx_lock(simple_mtx_t
*mtx
)
87 c
= __sync_val_compare_and_swap(&mtx
->val
, 0, 1);
89 assert(c
!= _SIMPLE_MTX_INVALID_VALUE
);
91 if (__builtin_expect(c
!= 0, 0)) {
93 c
= __sync_lock_test_and_set(&mtx
->val
, 2);
95 futex_wait(&mtx
->val
, 2, NULL
);
96 c
= __sync_lock_test_and_set(&mtx
->val
, 2);
102 simple_mtx_unlock(simple_mtx_t
*mtx
)
106 c
= __sync_fetch_and_sub(&mtx
->val
, 1);
108 assert(c
!= _SIMPLE_MTX_INVALID_VALUE
);
110 if (__builtin_expect(c
!= 1, 0)) {
112 futex_wake(&mtx
->val
, 1);
118 typedef mtx_t simple_mtx_t
;
120 #define _SIMPLE_MTX_INITIALIZER_NP _MTX_INITIALIZER_NP
123 simple_mtx_init(simple_mtx_t
*mtx
, int type
)
129 simple_mtx_destroy(simple_mtx_t
*mtx
)
135 simple_mtx_lock(simple_mtx_t
*mtx
)
141 simple_mtx_unlock(simple_mtx_t
*mtx
)