| 1 | /* Common definition for memset/memset_chk ifunc selections. |
| 2 | All versions must be listed in ifunc-impl-list.c. |
| 3 | Copyright (C) 2017-2021 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 | #include <init-arch.h> |
| 21 | |
| 22 | extern __typeof (REDIRECT_NAME) OPTIMIZE (erms) attribute_hidden; |
| 23 | extern __typeof (REDIRECT_NAME) OPTIMIZE (sse2_unaligned) |
| 24 | attribute_hidden; |
| 25 | extern __typeof (REDIRECT_NAME) OPTIMIZE (sse2_unaligned_erms) |
| 26 | attribute_hidden; |
| 27 | extern __typeof (REDIRECT_NAME) OPTIMIZE (avx2_unaligned) attribute_hidden; |
| 28 | extern __typeof (REDIRECT_NAME) OPTIMIZE (avx2_unaligned_erms) |
| 29 | attribute_hidden; |
| 30 | extern __typeof (REDIRECT_NAME) OPTIMIZE (avx2_unaligned_rtm) |
| 31 | attribute_hidden; |
| 32 | extern __typeof (REDIRECT_NAME) OPTIMIZE (avx2_unaligned_erms_rtm) |
| 33 | attribute_hidden; |
| 34 | extern __typeof (REDIRECT_NAME) OPTIMIZE (evex_unaligned) |
| 35 | attribute_hidden; |
| 36 | extern __typeof (REDIRECT_NAME) OPTIMIZE (evex_unaligned_erms) |
| 37 | attribute_hidden; |
| 38 | extern __typeof (REDIRECT_NAME) OPTIMIZE (avx512_unaligned) |
| 39 | attribute_hidden; |
| 40 | extern __typeof (REDIRECT_NAME) OPTIMIZE (avx512_unaligned_erms) |
| 41 | attribute_hidden; |
| 42 | extern __typeof (REDIRECT_NAME) OPTIMIZE (avx512_no_vzeroupper) |
| 43 | attribute_hidden; |
| 44 | |
| 45 | static inline void * |
| 46 | IFUNC_SELECTOR (void) |
| 47 | { |
| 48 | const struct cpu_features* cpu_features = __get_cpu_features (); |
| 49 | |
| 50 | if (CPU_FEATURES_ARCH_P (cpu_features, Prefer_ERMS)) |
| 51 | return OPTIMIZE (erms); |
| 52 | |
| 53 | if (CPU_FEATURE_USABLE_P (cpu_features, AVX512F) |
| 54 | && !CPU_FEATURES_ARCH_P (cpu_features, Prefer_No_AVX512)) |
| 55 | { |
| 56 | if (CPU_FEATURE_USABLE_P (cpu_features, AVX512VL) |
| 57 | && CPU_FEATURE_USABLE_P (cpu_features, AVX512BW) |
| 58 | && CPU_FEATURE_USABLE_P (cpu_features, BMI2)) |
| 59 | { |
| 60 | if (CPU_FEATURE_USABLE_P (cpu_features, ERMS)) |
| 61 | return OPTIMIZE (avx512_unaligned_erms); |
| 62 | |
| 63 | return OPTIMIZE (avx512_unaligned); |
| 64 | } |
| 65 | |
| 66 | return OPTIMIZE (avx512_no_vzeroupper); |
| 67 | } |
| 68 | |
| 69 | if (CPU_FEATURE_USABLE_P (cpu_features, AVX2)) |
| 70 | { |
| 71 | if (CPU_FEATURE_USABLE_P (cpu_features, AVX512VL) |
| 72 | && CPU_FEATURE_USABLE_P (cpu_features, AVX512BW) |
| 73 | && CPU_FEATURE_USABLE_P (cpu_features, BMI2)) |
| 74 | { |
| 75 | if (CPU_FEATURE_USABLE_P (cpu_features, ERMS)) |
| 76 | return OPTIMIZE (evex_unaligned_erms); |
| 77 | |
| 78 | return OPTIMIZE (evex_unaligned); |
| 79 | } |
| 80 | |
| 81 | if (CPU_FEATURE_USABLE_P (cpu_features, RTM)) |
| 82 | { |
| 83 | if (CPU_FEATURE_USABLE_P (cpu_features, ERMS)) |
| 84 | return OPTIMIZE (avx2_unaligned_erms_rtm); |
| 85 | |
| 86 | return OPTIMIZE (avx2_unaligned_rtm); |
| 87 | } |
| 88 | |
| 89 | if (!CPU_FEATURES_ARCH_P (cpu_features, Prefer_No_VZEROUPPER)) |
| 90 | { |
| 91 | if (CPU_FEATURE_USABLE_P (cpu_features, ERMS)) |
| 92 | return OPTIMIZE (avx2_unaligned_erms); |
| 93 | |
| 94 | return OPTIMIZE (avx2_unaligned); |
| 95 | } |
| 96 | } |
| 97 | |
| 98 | if (CPU_FEATURE_USABLE_P (cpu_features, ERMS)) |
| 99 | return OPTIMIZE (sse2_unaligned_erms); |
| 100 | |
| 101 | return OPTIMIZE (sse2_unaligned); |
| 102 | } |
| 103 | |