1 /* Copyright (C) 2012-2023 Free Software Foundation, Inc.
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 #if HAVE_IFUNC
26 #include <cpuid.h>
27
28 #ifdef __x86_64__
29 # define FEAT1_REGISTER ecx
30 #else
31 # define FEAT1_REGISTER edx
32 #endif
33
34 /* Value of the CPUID feature register FEAT1_REGISTER for the cmpxchg
35 bit for IFUNC_COND1 below. */
36 extern unsigned int __libat_feat1 HIDDEN;
37
38 /* Initialize libat_feat1 and return its value. */
39 unsigned int __libat_feat1_init (void) HIDDEN;
40
41 /* Return the value of the relevant feature register for the relevant
42 cmpxchg bit, or 0 if there is no CPUID support. */
43 static inline unsigned int
44 __attribute__ ((const))
45 load_feat1 (void)
46 {
47 /* See the store in __libat_feat1_init. */
48 unsigned int feat1 = __atomic_load_n (&__libat_feat1, __ATOMIC_RELAXED);
49 if (feat1 == 0)
50 /* Assume that initialization has not happened yet. This may get
51 called repeatedly if the CPU does not have any feature bits at
52 all. */
53 feat1 = __libat_feat1_init ();
54 return feat1;
55 }
56
57 #ifdef __x86_64__
58 # define IFUNC_COND_1 ((load_feat1 () & (bit_AVX | bit_CMPXCHG16B)) \
59 == (bit_AVX | bit_CMPXCHG16B))
60 # define IFUNC_COND_2 (load_feat1 () & bit_CMPXCHG16B)
61 #else
62 # define IFUNC_COND_1 (load_feat1 () & bit_CMPXCHG8B)
63 #endif
64
65 #ifdef __x86_64__
66 # define IFUNC_NCOND(N) (2 * (N == 16))
67 #else
68 # define IFUNC_NCOND(N) (N == 8)
69 #endif
70
71 #ifdef __x86_64__
72 # undef MAYBE_HAVE_ATOMIC_CAS_16
73 # define MAYBE_HAVE_ATOMIC_CAS_16 IFUNC_COND_2
74 # undef MAYBE_HAVE_ATOMIC_EXCHANGE_16
75 # define MAYBE_HAVE_ATOMIC_EXCHANGE_16 IFUNC_COND_2
76 # undef MAYBE_HAVE_ATOMIC_LDST_16
77 # define MAYBE_HAVE_ATOMIC_LDST_16 IFUNC_COND_2
78 /* Since load and store are implemented with CAS, they are not fast. */
79 # undef FAST_ATOMIC_LDST_16
80 # define FAST_ATOMIC_LDST_16 0
81 # if IFUNC_ALT != 0
82 # undef HAVE_ATOMIC_CAS_16
83 # define HAVE_ATOMIC_CAS_16 1
84 # endif
85 # if IFUNC_ALT == 1
86 # undef HAVE_ATOMIC_LDST_16
87 # define HAVE_ATOMIC_LDST_16 1
88 # endif
89 #else
90 # undef MAYBE_HAVE_ATOMIC_CAS_8
91 # define MAYBE_HAVE_ATOMIC_CAS_8 IFUNC_COND_1
92 # undef MAYBE_HAVE_ATOMIC_EXCHANGE_8
93 # define MAYBE_HAVE_ATOMIC_EXCHANGE_8 IFUNC_COND_1
94 # undef MAYBE_HAVE_ATOMIC_LDST_8
95 # define MAYBE_HAVE_ATOMIC_LDST_8 IFUNC_COND_1
96 # if IFUNC_ALT == 1
97 # undef HAVE_ATOMIC_CAS_8
98 # define HAVE_ATOMIC_CAS_8 1
99 # endif
100 #endif
101
102 #if defined(__x86_64__) && N == 16 && IFUNC_ALT != 0
103 static inline bool
104 atomic_compare_exchange_n (UTYPE *mptr, UTYPE *eptr, UTYPE newval,
105 bool weak_p UNUSED, int sm UNUSED, int fm UNUSED)
106 {
107 UTYPE cmpval = *eptr;
108 UTYPE oldval = __sync_val_compare_and_swap_16 (mptr, cmpval, newval);
109 if (oldval == cmpval)
110 return true;
111 *eptr = oldval;
112 return false;
113 }
114 # define atomic_compare_exchange_n atomic_compare_exchange_n
115 #endif /* Have CAS 16 */
116
117 #if defined(__x86_64__) && N == 16 && IFUNC_ALT == 1
118 #define __atomic_load_n(ptr, model) \
119 (sizeof (*ptr) == 16 ? atomic_load_n (ptr, model) \
120 : (__atomic_load_n) (ptr, model))
121 #define __atomic_store_n(ptr, val, model) \
122 (sizeof (*ptr) == 16 ? atomic_store_n (ptr, val, model) \
123 : (__atomic_store_n) (ptr, val, model))
124
125 static inline UTYPE
126 atomic_load_n (UTYPE *ptr, int model UNUSED)
127 {
128 UTYPE ret;
129 __asm__ ("vmovdqa\t{%1, %0|%0, %1}" : "=x" (ret) : "m" (*ptr));
130 return ret;
131 }
132
133 static inline void
134 atomic_store_n (UTYPE *ptr, UTYPE val, int model UNUSED)
135 {
136 __asm__ ("vmovdqa\t{%1, %0|%0, %1}\n\tmfence" : "=m" (*ptr) : "x" (val));
137 }
138 #endif
139
140 #endif /* HAVE_IFUNC */
141
142 #include_next <host-config.h>