1 | /* Multiple versions of strcmp. |
2 | All versions must be listed in ifunc-impl-list.c. |
3 | Copyright (C) 2017-2022 Free Software Foundation, Inc. |
4 | This file is part of the GNU C Library. |
5 | |
6 | The GNU C Library is free software; you can redistribute it and/or |
7 | modify it under the terms of the GNU Lesser General Public |
8 | License as published by the Free Software Foundation; either |
9 | version 2.1 of the License, or (at your option) any later version. |
10 | |
11 | The GNU C Library is distributed in the hope that it will be useful, |
12 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
14 | Lesser General Public License for more details. |
15 | |
16 | You should have received a copy of the GNU Lesser General Public |
17 | License along with the GNU C Library; if not, see |
18 | <https://www.gnu.org/licenses/>. */ |
19 | |
20 | /* Define multiple versions only for the definition in libc. */ |
21 | #if IS_IN (libc) |
22 | # define strcmp __redirect_strcmp |
23 | # include <string.h> |
24 | # undef strcmp |
25 | |
26 | # define SYMBOL_NAME strcmp |
27 | # include <init-arch.h> |
28 | |
29 | extern __typeof (REDIRECT_NAME) OPTIMIZE (evex) attribute_hidden; |
30 | |
31 | extern __typeof (REDIRECT_NAME) OPTIMIZE (avx2) attribute_hidden; |
32 | extern __typeof (REDIRECT_NAME) OPTIMIZE (avx2_rtm) attribute_hidden; |
33 | |
34 | extern __typeof (REDIRECT_NAME) OPTIMIZE (sse42) attribute_hidden; |
35 | |
36 | extern __typeof (REDIRECT_NAME) |
37 | OPTIMIZE (sse2_unaligned) attribute_hidden; |
38 | extern __typeof (REDIRECT_NAME) OPTIMIZE (sse2) attribute_hidden; |
39 | |
40 | |
41 | |
42 | static inline void * |
43 | IFUNC_SELECTOR (void) |
44 | { |
45 | const struct cpu_features *cpu_features = __get_cpu_features (); |
46 | |
47 | if (X86_ISA_CPU_FEATURE_USABLE_P (cpu_features, AVX2) |
48 | && X86_ISA_CPU_FEATURES_ARCH_P (cpu_features, |
49 | AVX_Fast_Unaligned_Load, )) |
50 | { |
51 | if (X86_ISA_CPU_FEATURE_USABLE_P (cpu_features, AVX512VL) |
52 | && X86_ISA_CPU_FEATURE_USABLE_P (cpu_features, AVX512BW) |
53 | && X86_ISA_CPU_FEATURE_USABLE_P (cpu_features, BMI2)) |
54 | return OPTIMIZE (evex); |
55 | |
56 | if (CPU_FEATURE_USABLE_P (cpu_features, RTM)) |
57 | return OPTIMIZE (avx2_rtm); |
58 | |
59 | if (X86_ISA_CPU_FEATURES_ARCH_P (cpu_features, |
60 | Prefer_No_VZEROUPPER, !)) |
61 | return OPTIMIZE (avx2); |
62 | } |
63 | |
64 | if (X86_ISA_CPU_FEATURE_USABLE_P (cpu_features, SSE4_2) |
65 | /* Keep this as runtime check. Some ISA level >= 2 CPUs such as |
66 | Tremont, Silvermont, and more check this. */ |
67 | && !CPU_FEATURES_ARCH_P (cpu_features, Slow_SSE4_2)) |
68 | return OPTIMIZE (sse42); |
69 | |
70 | /* Keep this as runtime check. The standard SSE2 version has |
71 | meaningful optimizations around keeping all loads aligned in the |
72 | main loop which can benefit some ISA level >= 2 CPUs. */ |
73 | if (CPU_FEATURES_ARCH_P (cpu_features, Fast_Unaligned_Load)) |
74 | return OPTIMIZE (sse2_unaligned); |
75 | |
76 | return OPTIMIZE (sse2); |
77 | } |
78 | |
79 | libc_ifunc_redirected (__redirect_strcmp, strcmp, IFUNC_SELECTOR ()); |
80 | |
81 | # ifdef SHARED |
82 | __hidden_ver1 (strcmp, __GI_strcmp, __redirect_strcmp) |
83 | __attribute__ ((visibility ("hidden" ))) __attribute_copy__ (strcmp); |
84 | # endif |
85 | #endif |
86 | |