|
145
|
1 /* Copyright (C) 2012-2020 Free Software Foundation, Inc.
|
|
111
|
2 Contributed by Richard Henderson <rth@redhat.com>.
|
|
|
3
|
|
|
4 This file is part of the GNU Atomic Library (libatomic).
|
|
|
5
|
|
|
6 Libatomic is free software; you can redistribute it and/or modify it
|
|
|
7 under the terms of the GNU General Public License as published by
|
|
|
8 the Free Software Foundation; either version 3 of the License, or
|
|
|
9 (at your option) any later version.
|
|
|
10
|
|
|
11 Libatomic is distributed in the hope that it will be useful, but WITHOUT ANY
|
|
|
12 WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
|
|
|
13 FOR A PARTICULAR PURPOSE. See the GNU General Public License for
|
|
|
14 more details.
|
|
|
15
|
|
|
16 Under Section 7 of GPL version 3, you are granted additional
|
|
|
17 permissions described in the GCC Runtime Library Exception, version
|
|
|
18 3.1, as published by the Free Software Foundation.
|
|
|
19
|
|
|
20 You should have received a copy of the GNU General Public License and
|
|
|
21 a copy of the GCC Runtime Library Exception along with this program;
|
|
|
22 see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
|
|
|
23 <http://www.gnu.org/licenses/>. */
|
|
|
24
|
|
|
25 #include "libatomic_i.h"
|
|
|
26
|
|
|
27
|
|
|
28 /* If we natively support the exchange, and if we're unconcerned with extra
|
|
|
29 barriers (e.g. fully in-order cpu for which barriers are a nop), then
|
|
|
30 go ahead and expand the operation inline. */
|
|
|
31 #if !defined(WANT_SPECIALCASE_RELAXED) && !defined(__OPTIMIZE_SIZE__)
|
|
|
32 # define EXACT_INLINE(N) \
|
|
|
33 if (C2(HAVE_ATOMIC_EXCHANGE_,N)) \
|
|
|
34 { \
|
|
|
35 *PTR(N,rptr) = __atomic_exchange_n \
|
|
|
36 (PTR(N,mptr), *PTR(N,vptr), __ATOMIC_SEQ_CST); \
|
|
|
37 return; \
|
|
|
38 }
|
|
|
39 #else
|
|
|
40 # define EXACT_INLINE(N)
|
|
|
41 #endif
|
|
|
42
|
|
|
43
|
|
|
44 #define EXACT(N) \
|
|
|
45 do { \
|
|
|
46 if (!C2(HAVE_INT,N)) break; \
|
|
|
47 if ((uintptr_t)mptr & (N - 1)) break; \
|
|
|
48 EXACT_INLINE (N); \
|
|
|
49 *PTR(N,rptr) = C3(local_,exchange_,N) \
|
|
|
50 (PTR(N,mptr), *PTR(N,vptr), smodel); \
|
|
|
51 return; \
|
|
|
52 } while (0)
|
|
|
53
|
|
|
54
|
|
|
55 #define LARGER(N) \
|
|
|
56 do { \
|
|
|
57 if (!C2(HAVE_INT,N)) break; \
|
|
|
58 if (!C2(MAYBE_HAVE_ATOMIC_CAS_,N)) break; \
|
|
|
59 r = (uintptr_t)mptr & (N - 1); \
|
|
|
60 a = (uintptr_t)mptr & -N; \
|
|
|
61 if (r + n <= N) \
|
|
|
62 { \
|
|
|
63 pre_barrier (smodel); \
|
|
|
64 u.C2(i,N) = *PTR(N,a); \
|
|
|
65 do { \
|
|
|
66 v = u; \
|
|
|
67 memcpy (v.b + r, vptr, n); \
|
|
|
68 } while (!(C2(HAVE_ATOMIC_CAS_,N) \
|
|
|
69 ? __atomic_compare_exchange_n (PTR(N,a), \
|
|
|
70 &u.C2(i,N), v.C2(i,N), true, \
|
|
|
71 __ATOMIC_RELAXED, __ATOMIC_RELAXED) \
|
|
|
72 : C3(local_,compare_exchange_,N) (PTR(N,a), \
|
|
|
73 &u.C2(i,N), v.C2(i,N), \
|
|
|
74 __ATOMIC_RELAXED, __ATOMIC_RELAXED))); \
|
|
|
75 goto Lfinish; \
|
|
|
76 } \
|
|
|
77 } while (0)
|
|
|
78
|
|
|
79
|
|
|
80 static void __attribute__((noinline))
|
|
|
81 libat_exchange_large_inplace (size_t n, void *mptr, void *vptr)
|
|
|
82 {
|
|
|
83 #define BUF 1024
|
|
|
84
|
|
|
85 char temp[BUF];
|
|
|
86 size_t i = 0;
|
|
|
87
|
|
|
88 for (i = 0; n >= BUF; i += BUF, n -= BUF)
|
|
|
89 {
|
|
|
90 memcpy (temp, mptr + i, BUF);
|
|
|
91 memcpy (mptr + i, vptr + i, BUF);
|
|
|
92 memcpy (vptr + i, temp, BUF);
|
|
|
93 }
|
|
|
94 if (n > 0)
|
|
|
95 {
|
|
|
96 memcpy (temp, mptr + i, n);
|
|
|
97 memcpy (mptr + i, vptr + i, n);
|
|
|
98 memcpy (vptr + i, temp, n);
|
|
|
99 }
|
|
|
100
|
|
|
101 #undef BUF
|
|
|
102 }
|
|
|
103
|
|
|
104 void
|
|
|
105 libat_exchange (size_t n, void *mptr, void *vptr, void *rptr, int smodel)
|
|
|
106 {
|
|
|
107 union max_size_u u, v;
|
|
|
108 uintptr_t r, a;
|
|
|
109
|
|
|
110 switch (n)
|
|
|
111 {
|
|
|
112 case 0: return;
|
|
|
113 case 1: EXACT(1); goto L4;
|
|
|
114 case 2: EXACT(2); goto L4;
|
|
|
115 case 4: EXACT(4); goto L8;
|
|
|
116 case 8: EXACT(8); goto L16;
|
|
|
117 case 16: EXACT(16); break;
|
|
|
118
|
|
|
119 case 3: L4: LARGER(4); /* FALLTHRU */
|
|
|
120 case 5 ... 7: L8: LARGER(8); /* FALLTHRU */
|
|
|
121 case 9 ... 15: L16: LARGER(16); break;
|
|
|
122
|
|
|
123 Lfinish:
|
|
|
124 post_barrier (smodel);
|
|
|
125 memcpy (rptr, u.b + r, n);
|
|
|
126 return;
|
|
|
127 }
|
|
|
128
|
|
|
129 pre_seq_barrier (smodel);
|
|
|
130 libat_lock_n (mptr, n);
|
|
|
131
|
|
|
132 if (vptr != rptr)
|
|
|
133 {
|
|
|
134 memcpy (rptr, mptr, n);
|
|
|
135 memcpy (mptr, vptr, n);
|
|
|
136 }
|
|
|
137 else
|
|
|
138 libat_exchange_large_inplace (n, mptr, vptr);
|
|
|
139
|
|
|
140 libat_unlock_n (mptr, n);
|
|
|
141 post_seq_barrier (smodel);
|
|
|
142 }
|
|
|
143
|
|
|
144 EXPORT_ALIAS (exchange);
|
