| 1 | /* |
| 2 | * Copyright(c) 2019-2023 Qualcomm Innovation Center, Inc. All Rights Reserved. |
| 3 | * |
| 4 | * This program is free software; you can redistribute it and/or modify |
| 5 | * it under the terms of the GNU General Public License as published by |
| 6 | * the Free Software Foundation; either version 2 of the License, or |
| 7 | * (at your option) any later version. |
| 8 | * |
| 9 | * This program is distributed in the hope that it will be useful, |
| 10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 12 | * GNU General Public License for more details. |
| 13 | * |
| 14 | * You should have received a copy of the GNU General Public License |
| 15 | * along with this program; if not, see <http://www.gnu.org/licenses/>. |
| 16 | */ |
| 17 | |
| 18 | #include <stdio.h> |
| 19 | #include <stdint.h> |
| 20 | #include <stdbool.h> |
| 21 | |
| 22 | int err; |
| 23 | |
| 24 | #include "hex_test.h" |
| 25 | |
| 26 | static int32_t sfrecipa(int32_t Rs, int32_t Rt, bool *pred_result) |
| 27 | { |
| 28 | int32_t result; |
| 29 | bool predval; |
| 30 | |
| 31 | asm volatile("%0,p0 = sfrecipa(%2, %3)\n\t" |
| 32 | "%1 = p0\n\t" |
| 33 | : "+r"(result), "=r"(predval) |
| 34 | : "r"(Rs), "r"(Rt) |
| 35 | : "p0"); |
| 36 | *pred_result = predval; |
| 37 | return result; |
| 38 | } |
| 39 | |
| 40 | static int32_t sfinvsqrta(int32_t Rs, int32_t *pred_result) |
| 41 | { |
| 42 | int32_t result; |
| 43 | int32_t predval; |
| 44 | |
| 45 | asm volatile("%0,p0 = sfinvsqrta(%2)\n\t" |
| 46 | "%1 = p0\n\t" |
| 47 | : "+r"(result), "=r"(predval) |
| 48 | : "r"(Rs) |
| 49 | : "p0"); |
| 50 | *pred_result = predval; |
| 51 | return result; |
| 52 | } |
| 53 | |
| 54 | static int64_t vacsh(int64_t Rxx, int64_t Rss, int64_t Rtt, |
| 55 | int *pred_result, bool *ovf_result) |
| 56 | { |
| 57 | int64_t result = Rxx; |
| 58 | int predval; |
| 59 | uint32_t usr; |
| 60 | |
| 61 | /* |
| 62 | * This instruction can set bit 0 (OVF/overflow) in usr |
| 63 | * Clear the bit first, then return that bit to the caller |
| 64 | */ |
| 65 | asm volatile("r2 = usr\n\t" |
| 66 | "r2 = clrbit(r2, #0)\n\t" /* clear overflow bit */ |
| 67 | "usr = r2\n\t" |
| 68 | "%0,p0 = vacsh(%3, %4)\n\t" |
| 69 | "%1 = p0\n\t" |
| 70 | "%2 = usr\n\t" |
| 71 | : "+r"(result), "=r"(predval), "=r"(usr) |
| 72 | : "r"(Rss), "r"(Rtt) |
| 73 | : "r2", "p0", "usr"); |
| 74 | *pred_result = predval; |
| 75 | *ovf_result = (usr & 1); |
| 76 | return result; |
| 77 | } |
| 78 | |
| 79 | static int64_t vminub(int64_t Rtt, int64_t Rss, int32_t *pred_result) |
| 80 | { |
| 81 | int64_t result; |
| 82 | int32_t predval; |
| 83 | |
| 84 | asm volatile("%0,p0 = vminub(%2, %3)\n\t" |
| 85 | "%1 = p0\n\t" |
| 86 | : "=r"(result), "=r"(predval) |
| 87 | : "r"(Rtt), "r"(Rss) |
| 88 | : "p0"); |
| 89 | *pred_result = predval; |
| 90 | return result; |
| 91 | } |
| 92 | |
| 93 | static int64_t add_carry(int64_t Rss, int64_t Rtt, |
| 94 | int32_t pred_in, int32_t *pred_result) |
| 95 | { |
| 96 | int64_t result; |
| 97 | int32_t predval = pred_in; |
| 98 | |
| 99 | asm volatile("p0 = %1\n\t" |
| 100 | "%0 = add(%2, %3, p0):carry\n\t" |
| 101 | "%1 = p0\n\t" |
| 102 | : "=r"(result), "+r"(predval) |
| 103 | : "r"(Rss), "r"(Rtt) |
| 104 | : "p0"); |
| 105 | *pred_result = predval; |
| 106 | return result; |
| 107 | } |
| 108 | |
| 109 | static int64_t sub_carry(int64_t Rss, int64_t Rtt, |
| 110 | int32_t pred_in, int32_t *pred_result) |
| 111 | { |
| 112 | int64_t result; |
| 113 | int32_t predval = pred_in; |
| 114 | |
| 115 | asm volatile("p0 = !cmp.eq(%1, #0)\n\t" |
| 116 | "%0 = sub(%2, %3, p0):carry\n\t" |
| 117 | "%1 = p0\n\t" |
| 118 | : "=r"(result), "+r"(predval) |
| 119 | : "r"(Rss), "r"(Rtt) |
| 120 | : "p0"); |
| 121 | *pred_result = predval; |
| 122 | return result; |
| 123 | } |
| 124 | |
| 125 | static void test_sfrecipa() |
| 126 | { |
| 127 | int32_t res; |
| 128 | bool pred_result; |
| 129 | |
| 130 | res = sfrecipa(0x04030201, 0x05060708, &pred_result); |
| 131 | check32(res, 0x59f38001); |
| 132 | check32(pred_result, false); |
| 133 | } |
| 134 | |
| 135 | static void test_sfinvsqrta() |
| 136 | { |
| 137 | int32_t res; |
| 138 | int32_t pred_result; |
| 139 | |
| 140 | res = sfinvsqrta(0x04030201, &pred_result); |
| 141 | check32(res, 0x4d330000); |
| 142 | check32(pred_result, 0xe0); |
| 143 | |
| 144 | res = sfinvsqrta(0x0, &pred_result); |
| 145 | check32(res, 0x3f800000); |
| 146 | check32(pred_result, 0x0); |
| 147 | } |
| 148 | |
| 149 | static void test_vacsh() |
| 150 | { |
| 151 | int64_t res64; |
| 152 | int32_t pred_result; |
| 153 | bool ovf_result; |
| 154 | |
| 155 | res64 = vacsh(0x0004000300020001LL, |
| 156 | 0x0001000200030004LL, |
| 157 | 0x0000000000000000LL, &pred_result, &ovf_result); |
| 158 | check64(res64, 0x0004000300030004LL); |
| 159 | check32(pred_result, 0xf0); |
| 160 | check32(ovf_result, false); |
| 161 | |
| 162 | res64 = vacsh(0x0004000300020001LL, |
| 163 | 0x0001000200030004LL, |
| 164 | 0x000affff000d0000LL, &pred_result, &ovf_result); |
| 165 | check64(res64, 0x000e0003000f0004LL); |
| 166 | check32(pred_result, 0xcc); |
| 167 | check32(ovf_result, false); |
| 168 | |
| 169 | res64 = vacsh(0x00047fff00020001LL, |
| 170 | 0x00017fff00030004LL, |
| 171 | 0x000a0fff000d0000LL, &pred_result, &ovf_result); |
| 172 | check64(res64, 0x000e7fff000f0004LL); |
| 173 | check32(pred_result, 0xfc); |
| 174 | check32(ovf_result, true); |
| 175 | |
| 176 | res64 = vacsh(0x0004000300020001LL, |
| 177 | 0x0001000200030009LL, |
| 178 | 0x000affff000d0001LL, &pred_result, &ovf_result); |
| 179 | check64(res64, 0x000e0003000f0008LL); |
| 180 | check32(pred_result, 0xcc); |
| 181 | check32(ovf_result, false); |
| 182 | } |
| 183 | |
| 184 | static void test_vminub() |
| 185 | { |
| 186 | int64_t res64; |
| 187 | int32_t pred_result; |
| 188 | |
| 189 | res64 = vminub(0x0807060504030201LL, |
| 190 | 0x0102030405060708LL, |
| 191 | &pred_result); |
| 192 | check64(res64, 0x0102030404030201LL); |
| 193 | check32(pred_result, 0xf0); |
| 194 | |
| 195 | res64 = vminub(0x0802060405030701LL, |
| 196 | 0x0107030504060208LL, |
| 197 | &pred_result); |
| 198 | check64(res64, 0x0102030404030201LL); |
| 199 | check32(pred_result, 0xaa); |
| 200 | } |
| 201 | |
| 202 | static void test_add_carry() |
| 203 | { |
| 204 | int64_t res64; |
| 205 | int32_t pred_result; |
| 206 | |
| 207 | res64 = add_carry(0x0000000000000000LL, |
| 208 | 0xffffffffffffffffLL, |
| 209 | 1, &pred_result); |
| 210 | check64(res64, 0x0000000000000000LL); |
| 211 | check32(pred_result, 0xff); |
| 212 | |
| 213 | res64 = add_carry(0x0000000100000000LL, |
| 214 | 0xffffffffffffffffLL, |
| 215 | 0, &pred_result); |
| 216 | check64(res64, 0x00000000ffffffffLL); |
| 217 | check32(pred_result, 0xff); |
| 218 | |
| 219 | res64 = add_carry(0x0000000100000000LL, |
| 220 | 0xffffffffffffffffLL, |
| 221 | 0, &pred_result); |
| 222 | check64(res64, 0x00000000ffffffffLL); |
| 223 | check32(pred_result, 0xff); |
| 224 | } |
| 225 | |
| 226 | static void test_sub_carry() |
| 227 | { |
| 228 | int64_t res64; |
| 229 | int32_t pred_result; |
| 230 | |
| 231 | res64 = sub_carry(0x0000000000000000LL, |
| 232 | 0x0000000000000000LL, |
| 233 | 1, &pred_result); |
| 234 | check64(res64, 0x0000000000000000LL); |
| 235 | check32(pred_result, 0xff); |
| 236 | |
| 237 | res64 = sub_carry(0x0000000100000000LL, |
| 238 | 0x0000000000000000LL, |
| 239 | 0, &pred_result); |
| 240 | check64(res64, 0x00000000ffffffffLL); |
| 241 | check32(pred_result, 0xff); |
| 242 | |
| 243 | res64 = sub_carry(0x0000000100000000LL, |
| 244 | 0x0000000000000000LL, |
| 245 | 0, &pred_result); |
| 246 | check64(res64, 0x00000000ffffffffLL); |
| 247 | check32(pred_result, 0xff); |
| 248 | } |
| 249 | |
| 250 | int main() |
| 251 | { |
| 252 | test_sfrecipa(); |
| 253 | test_sfinvsqrta(); |
| 254 | test_vacsh(); |
| 255 | test_vminub(); |
| 256 | test_add_carry(); |
| 257 | test_sub_carry(); |
| 258 | |
| 259 | puts(err ? "FAIL" : "PASS"); |
| 260 | return err; |
| 261 | } |