| 1 | /* gameplaySP |
| 2 | * |
| 3 | * Copyright (C) 2006 Exophase <exophase@gmail.com> |
| 4 | * |
| 5 | * This program is free software; you can redistribute it and/or |
| 6 | * modify it under the terms of the GNU General Public License as |
| 7 | * published by the Free Software Foundation; either version 2 of |
| 8 | * the License, or (at your option) any later version. |
| 9 | * |
| 10 | * This program is distributed in the hope that it will be useful, |
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 13 | * General Public License for more details. |
| 14 | * |
| 15 | * You should have received a copy of the GNU General Public License |
| 16 | * along with this program; if not, write to the Free Software |
| 17 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 18 | */ |
| 19 | |
| 20 | #ifndef ARM_EMIT_H |
| 21 | #define ARM_EMIT_H |
| 22 | |
| 23 | #include "arm_codegen.h" |
| 24 | |
| 25 | u32 arm_update_gba_arm(u32 pc); |
| 26 | u32 arm_update_gba_thumb(u32 pc); |
| 27 | u32 arm_update_gba_idle_arm(u32 pc); |
| 28 | u32 arm_update_gba_idle_thumb(u32 pc); |
| 29 | |
| 30 | // Although these are defined as a function, don't call them as |
| 31 | // such (jump to it instead) |
| 32 | void arm_indirect_branch_arm(u32 address); |
| 33 | void arm_indirect_branch_thumb(u32 address); |
| 34 | void arm_indirect_branch_dual_arm(u32 address); |
| 35 | void arm_indirect_branch_dual_thumb(u32 address); |
| 36 | |
| 37 | void execute_store_cpsr(u32 new_cpsr, u32 store_mask, u32 address); |
| 38 | u32 execute_store_cpsr_body(u32 _cpsr, u32 store_mask, u32 address); |
| 39 | void execute_store_spsr(u32 new_cpsr, u32 store_mask); |
| 40 | u32 execute_read_spsr(); |
| 41 | u32 execute_spsr_restore(u32 address); |
| 42 | |
| 43 | void execute_swi_arm(u32 pc); |
| 44 | void execute_swi_thumb(u32 pc); |
| 45 | |
| 46 | void function_cc execute_store_u32_safe(u32 address, u32 source); |
| 47 | |
| 48 | void step_debug_arm(u32 pc); |
| 49 | |
| 50 | |
| 51 | #define write32(value) \ |
| 52 | *((u32 *)translation_ptr) = value; \ |
| 53 | translation_ptr += 4 \ |
| 54 | |
| 55 | #define arm_relative_offset(source, offset) \ |
| 56 | (((((u32)offset - (u32)source) - 8) >> 2) & 0xFFFFFF) \ |
| 57 | |
| 58 | |
| 59 | // reg_base_offset is the amount of bytes after reg_base where the registers |
| 60 | // actually begin. |
| 61 | |
| 62 | #define reg_base_offset 1024 |
| 63 | |
| 64 | |
| 65 | #define reg_a0 ARMREG_R0 |
| 66 | #define reg_a1 ARMREG_R1 |
| 67 | #define reg_a2 ARMREG_R2 |
| 68 | |
| 69 | #define reg_s0 ARMREG_R9 |
| 70 | #define reg_base ARMREG_SP |
| 71 | #define reg_flags ARMREG_R11 |
| 72 | |
| 73 | #define reg_cycles ARMREG_R12 |
| 74 | |
| 75 | #define reg_rv ARMREG_R0 |
| 76 | |
| 77 | #define reg_rm ARMREG_R0 |
| 78 | #define reg_rn ARMREG_R1 |
| 79 | #define reg_rs ARMREG_R14 |
| 80 | #define reg_rd ARMREG_R0 |
| 81 | |
| 82 | |
| 83 | // Register allocation layout for ARM and Thumb: |
| 84 | // Map from a GBA register to a host ARM register. -1 means load it |
| 85 | // from memory into one of the temp registers. |
| 86 | |
| 87 | // The following registers are chosen based on statistical analysis |
| 88 | // of a few games (see below), but might not be the best ones. Results |
| 89 | // vary tremendously between ARM and Thumb (for obvious reasons), so |
| 90 | // two sets are used. Take care to not call any function which can |
| 91 | // overwrite any of these registers from the dynarec - only call |
| 92 | // trusted functions in arm_stub.S which know how to save/restore |
| 93 | // them and know how to transfer them to the C functions it calls |
| 94 | // if necessary. |
| 95 | |
| 96 | // The following define the actual registers available for allocation. |
| 97 | // As registers are freed up add them to this list. |
| 98 | |
| 99 | // Note that r15 is linked to the a0 temp reg - this register will |
| 100 | // be preloaded with a constant upon read, and used to link to |
| 101 | // indirect branch functions upon write. |
| 102 | |
| 103 | #define reg_x0 ARMREG_R3 |
| 104 | #define reg_x1 ARMREG_R4 |
| 105 | #define reg_x2 ARMREG_R5 |
| 106 | #define reg_x3 ARMREG_R6 |
| 107 | #define reg_x4 ARMREG_R7 |
| 108 | #define reg_x5 ARMREG_R8 |
| 109 | |
| 110 | #define mem_reg -1 |
| 111 | |
| 112 | /* |
| 113 | |
| 114 | ARM register usage (38.775138% ARM instructions): |
| 115 | r00: 18.263814% (-- 18.263814%) |
| 116 | r12: 11.531477% (-- 29.795291%) |
| 117 | r09: 11.500162% (-- 41.295453%) |
| 118 | r14: 9.063440% (-- 50.358893%) |
| 119 | r06: 7.837682% (-- 58.196574%) |
| 120 | r01: 7.401049% (-- 65.597623%) |
| 121 | r07: 6.778340% (-- 72.375963%) |
| 122 | r05: 5.445009% (-- 77.820973%) |
| 123 | r02: 5.427288% (-- 83.248260%) |
| 124 | r03: 5.293743% (-- 88.542003%) |
| 125 | r04: 3.601103% (-- 92.143106%) |
| 126 | r11: 3.207311% (-- 95.350417%) |
| 127 | r10: 2.334864% (-- 97.685281%) |
| 128 | r08: 1.708207% (-- 99.393488%) |
| 129 | r15: 0.311270% (-- 99.704757%) |
| 130 | r13: 0.295243% (-- 100.000000%) |
| 131 | |
| 132 | Thumb register usage (61.224862% Thumb instructions): |
| 133 | r00: 34.788858% (-- 34.788858%) |
| 134 | r01: 26.564083% (-- 61.352941%) |
| 135 | r03: 10.983500% (-- 72.336441%) |
| 136 | r02: 8.303127% (-- 80.639567%) |
| 137 | r04: 4.900381% (-- 85.539948%) |
| 138 | r05: 3.941292% (-- 89.481240%) |
| 139 | r06: 3.257582% (-- 92.738822%) |
| 140 | r07: 2.644851% (-- 95.383673%) |
| 141 | r13: 1.408824% (-- 96.792497%) |
| 142 | r08: 0.906433% (-- 97.698930%) |
| 143 | r09: 0.679693% (-- 98.378623%) |
| 144 | r10: 0.656446% (-- 99.035069%) |
| 145 | r12: 0.453668% (-- 99.488737%) |
| 146 | r14: 0.248909% (-- 99.737646%) |
| 147 | r11: 0.171066% (-- 99.908713%) |
| 148 | r15: 0.091287% (-- 100.000000%) |
| 149 | |
| 150 | */ |
| 151 | |
| 152 | s32 arm_register_allocation[] = |
| 153 | { |
| 154 | reg_x0, // GBA r0 |
| 155 | reg_x1, // GBA r1 |
| 156 | mem_reg, // GBA r2 |
| 157 | mem_reg, // GBA r3 |
| 158 | mem_reg, // GBA r4 |
| 159 | mem_reg, // GBA r5 |
| 160 | reg_x2, // GBA r6 |
| 161 | mem_reg, // GBA r7 |
| 162 | mem_reg, // GBA r8 |
| 163 | reg_x3, // GBA r9 |
| 164 | mem_reg, // GBA r10 |
| 165 | mem_reg, // GBA r11 |
| 166 | reg_x4, // GBA r12 |
| 167 | mem_reg, // GBA r13 |
| 168 | reg_x5, // GBA r14 |
| 169 | reg_a0 // GBA r15 |
| 170 | |
| 171 | mem_reg, |
| 172 | mem_reg, |
| 173 | mem_reg, |
| 174 | mem_reg, |
| 175 | mem_reg, |
| 176 | mem_reg, |
| 177 | mem_reg, |
| 178 | mem_reg, |
| 179 | mem_reg, |
| 180 | mem_reg, |
| 181 | mem_reg, |
| 182 | mem_reg, |
| 183 | mem_reg, |
| 184 | mem_reg, |
| 185 | mem_reg, |
| 186 | mem_reg, |
| 187 | }; |
| 188 | |
| 189 | s32 thumb_register_allocation[] = |
| 190 | { |
| 191 | reg_x0, // GBA r0 |
| 192 | reg_x1, // GBA r1 |
| 193 | reg_x2, // GBA r2 |
| 194 | reg_x3, // GBA r3 |
| 195 | reg_x4, // GBA r4 |
| 196 | reg_x5, // GBA r5 |
| 197 | mem_reg, // GBA r6 |
| 198 | mem_reg, // GBA r7 |
| 199 | mem_reg, // GBA r8 |
| 200 | mem_reg, // GBA r9 |
| 201 | mem_reg, // GBA r10 |
| 202 | mem_reg, // GBA r11 |
| 203 | mem_reg, // GBA r12 |
| 204 | mem_reg, // GBA r13 |
| 205 | mem_reg, // GBA r14 |
| 206 | reg_a0 // GBA r15 |
| 207 | |
| 208 | mem_reg, |
| 209 | mem_reg, |
| 210 | mem_reg, |
| 211 | mem_reg, |
| 212 | mem_reg, |
| 213 | mem_reg, |
| 214 | mem_reg, |
| 215 | mem_reg, |
| 216 | mem_reg, |
| 217 | mem_reg, |
| 218 | mem_reg, |
| 219 | mem_reg, |
| 220 | mem_reg, |
| 221 | mem_reg, |
| 222 | mem_reg, |
| 223 | mem_reg, |
| 224 | }; |
| 225 | |
| 226 | |
| 227 | |
| 228 | #define arm_imm_lsl_to_rot(value) \ |
| 229 | (32 - value) \ |
| 230 | |
| 231 | |
| 232 | u32 arm_disect_imm_32bit(u32 imm, u32 *stores, u32 *rotations) |
| 233 | { |
| 234 | u32 store_count = 0; |
| 235 | u32 left_shift = 0; |
| 236 | u32 i; |
| 237 | |
| 238 | // Otherwise it'll return 0 things to store because it'll never |
| 239 | // find anything. |
| 240 | if(imm == 0) |
| 241 | { |
| 242 | rotations[0] = 0; |
| 243 | stores[0] = 0; |
| 244 | return 1; |
| 245 | } |
| 246 | |
| 247 | // Find chunks of non-zero data at 2 bit alignments. |
| 248 | while(1) |
| 249 | { |
| 250 | for(; left_shift < 32; left_shift += 2) |
| 251 | { |
| 252 | if((imm >> left_shift) & 0x03) |
| 253 | break; |
| 254 | } |
| 255 | |
| 256 | if(left_shift == 32) |
| 257 | { |
| 258 | // We've hit the end of the useful data. |
| 259 | return store_count; |
| 260 | } |
| 261 | |
| 262 | // Hit the end, it might wrap back around to the beginning. |
| 263 | if(left_shift >= 24) |
| 264 | { |
| 265 | // Make a mask for the residual bits. IE, if we have |
| 266 | // 5 bits of data at the end we can wrap around to 3 |
| 267 | // bits of data in the beginning. Thus the first |
| 268 | // thing, after being shifted left, has to be less |
| 269 | // than 111b, 0x7, or (1 << 3) - 1. |
| 270 | u32 top_bits = 32 - left_shift; |
| 271 | u32 residual_bits = 8 - top_bits; |
| 272 | u32 residual_mask = (1 << residual_bits) - 1; |
| 273 | |
| 274 | if((store_count > 1) && (left_shift > 24) && |
| 275 | ((stores[0] << ((32 - rotations[0]) & 0x1F)) < residual_mask)) |
| 276 | { |
| 277 | // Then we can throw out the last bit and tack it on |
| 278 | // to the first bit. |
| 279 | u32 initial_bits = rotations[0]; |
| 280 | stores[0] = |
| 281 | (stores[0] << ((top_bits + (32 - rotations[0])) & 0x1F)) | |
| 282 | ((imm >> left_shift) & 0xFF); |
| 283 | rotations[0] = top_bits; |
| 284 | |
| 285 | return store_count; |
| 286 | } |
| 287 | else |
| 288 | { |
| 289 | // There's nothing to wrap over to in the beginning |
| 290 | stores[store_count] = (imm >> left_shift) & 0xFF; |
| 291 | rotations[store_count] = (32 - left_shift) & 0x1F; |
| 292 | return store_count + 1; |
| 293 | } |
| 294 | break; |
| 295 | } |
| 296 | |
| 297 | stores[store_count] = (imm >> left_shift) & 0xFF; |
| 298 | rotations[store_count] = (32 - left_shift) & 0x1F; |
| 299 | |
| 300 | store_count++; |
| 301 | left_shift += 8; |
| 302 | } |
| 303 | } |
| 304 | |
| 305 | #define arm_load_imm_32bit(ireg, imm) \ |
| 306 | { \ |
| 307 | u32 stores[4]; \ |
| 308 | u32 rotations[4]; \ |
| 309 | u32 store_count = arm_disect_imm_32bit(imm, stores, rotations); \ |
| 310 | u32 i; \ |
| 311 | \ |
| 312 | ARM_MOV_REG_IMM(0, ireg, stores[0], rotations[0]); \ |
| 313 | \ |
| 314 | for(i = 1; i < store_count; i++) \ |
| 315 | { \ |
| 316 | ARM_ORR_REG_IMM(0, ireg, ireg, stores[i], rotations[i]); \ |
| 317 | } \ |
| 318 | } \ |
| 319 | |
| 320 | |
| 321 | #define generate_load_pc(ireg, new_pc) \ |
| 322 | arm_load_imm_32bit(ireg, new_pc) \ |
| 323 | |
| 324 | #define generate_load_imm(ireg, imm, imm_ror) \ |
| 325 | ARM_MOV_REG_IMM(0, ireg, imm, imm_ror) \ |
| 326 | |
| 327 | |
| 328 | |
| 329 | #define generate_shift_left(ireg, imm) \ |
| 330 | ARM_MOV_REG_IMMSHIFT(0, ireg, ireg, ARMSHIFT_LSL, imm) \ |
| 331 | |
| 332 | #define generate_shift_right(ireg, imm) \ |
| 333 | ARM_MOV_REG_IMMSHIFT(0, ireg, ireg, ARMSHIFT_LSR, imm) \ |
| 334 | |
| 335 | #define generate_shift_right_arithmetic(ireg, imm) \ |
| 336 | ARM_MOV_REG_IMMSHIFT(0, ireg, ireg, ARMSHIFT_ASR, imm) \ |
| 337 | |
| 338 | #define generate_rotate_right(ireg, imm) \ |
| 339 | ARM_MOV_REG_IMMSHIFT(0, ireg, ireg, ARMSHIFT_ROR, imm) \ |
| 340 | |
| 341 | #define generate_add(ireg_dest, ireg_src) \ |
| 342 | ARM_ADD_REG_REG(0, ireg_dest, ireg_dest, ireg_src) \ |
| 343 | |
| 344 | #define generate_sub(ireg_dest, ireg_src) \ |
| 345 | ARM_SUB_REG_REG(0, ireg_dest, ireg_dest, ireg_src) \ |
| 346 | |
| 347 | #define generate_or(ireg_dest, ireg_src) \ |
| 348 | ARM_ORR_REG_REG(0, ireg_dest, ireg_dest, ireg_src) \ |
| 349 | |
| 350 | #define generate_xor(ireg_dest, ireg_src) \ |
| 351 | ARM_EOR_REG_REG(0, ireg_dest, ireg_dest, ireg_src) \ |
| 352 | |
| 353 | #define generate_add_imm(ireg, imm, imm_ror) \ |
| 354 | ARM_ADD_REG_IMM(0, ireg, ireg, imm, imm_ror) \ |
| 355 | |
| 356 | #define generate_sub_imm(ireg, imm, imm_ror) \ |
| 357 | ARM_SUB_REG_IMM(0, ireg, ireg, imm, imm_ror) \ |
| 358 | |
| 359 | #define generate_xor_imm(ireg, imm, imm_ror) \ |
| 360 | ARM_EOR_REG_IMM(0, ireg, ireg, imm, imm_ror) \ |
| 361 | |
| 362 | #define generate_add_reg_reg_imm(ireg_dest, ireg_src, imm, imm_ror) \ |
| 363 | ARM_ADD_REG_IMM(0, ireg_dest, ireg_src, imm, imm_ror) \ |
| 364 | |
| 365 | #define generate_and_imm(ireg, imm, imm_ror) \ |
| 366 | ARM_AND_REG_IMM(0, ireg, ireg, imm, imm_ror) \ |
| 367 | |
| 368 | #define generate_mov(ireg_dest, ireg_src) \ |
| 369 | if(ireg_dest != ireg_src) \ |
| 370 | { \ |
| 371 | ARM_MOV_REG_REG(0, ireg_dest, ireg_src); \ |
| 372 | } \ |
| 373 | |
| 374 | #define generate_function_call(function_location) \ |
| 375 | ARM_BL(0, arm_relative_offset(translation_ptr, function_location)) \ |
| 376 | |
| 377 | #define generate_exit_block() \ |
| 378 | ARM_BX(0, ARMREG_LR) \ |
| 379 | |
| 380 | // The branch target is to be filled in later (thus a 0 for now) |
| 381 | |
| 382 | #define generate_branch_filler(condition_code, writeback_location) \ |
| 383 | (writeback_location) = translation_ptr; \ |
| 384 | ARM_B_COND(0, condition_code, 0) \ |
| 385 | |
| 386 | #define generate_update_pc(new_pc) \ |
| 387 | generate_load_pc(reg_a0, new_pc) \ |
| 388 | |
| 389 | #define generate_cycle_update() \ |
| 390 | if(cycle_count) \ |
| 391 | { \ |
| 392 | if(cycle_count >> 8) \ |
| 393 | { \ |
| 394 | ARM_ADD_REG_IMM(0, reg_cycles, reg_cycles, (cycle_count >> 8) & 0xFF, \ |
| 395 | arm_imm_lsl_to_rot(8)); \ |
| 396 | } \ |
| 397 | ARM_ADD_REG_IMM(0, reg_cycles, reg_cycles, (cycle_count & 0xFF), 0); \ |
| 398 | cycle_count = 0; \ |
| 399 | } \ |
| 400 | |
| 401 | #define generate_cycle_update_flag_set() \ |
| 402 | if(cycle_count >> 8) \ |
| 403 | { \ |
| 404 | ARM_ADD_REG_IMM(0, reg_cycles, reg_cycles, (cycle_count >> 8) & 0xFF, \ |
| 405 | arm_imm_lsl_to_rot(8)); \ |
| 406 | } \ |
| 407 | generate_save_flags(); \ |
| 408 | ARM_ADDS_REG_IMM(0, reg_cycles, reg_cycles, (cycle_count & 0xFF), 0); \ |
| 409 | cycle_count = 0 \ |
| 410 | |
| 411 | #define generate_branch_patch_conditional(dest, offset) \ |
| 412 | *((u32 *)(dest)) = (*((u32 *)dest) & 0xFF000000) | \ |
| 413 | arm_relative_offset(dest, offset) \ |
| 414 | |
| 415 | #define generate_branch_patch_unconditional(dest, offset) \ |
| 416 | *((u32 *)(dest)) = (*((u32 *)dest) & 0xFF000000) | \ |
| 417 | arm_relative_offset(dest, offset) \ |
| 418 | |
| 419 | // A different function is called for idle updates because of the relative |
| 420 | // location of the embedded PC. The idle version could be optimized to put |
| 421 | // the CPU into halt mode too, however. |
| 422 | |
| 423 | #define generate_branch_idle_eliminate(writeback_location, new_pc, mode) \ |
| 424 | generate_function_call(arm_update_gba_idle_##mode); \ |
| 425 | write32(new_pc); \ |
| 426 | generate_branch_filler(ARMCOND_AL, writeback_location) \ |
| 427 | |
| 428 | #define generate_branch_update(writeback_location, new_pc, mode) \ |
| 429 | ARM_MOV_REG_IMMSHIFT(0, reg_a0, reg_cycles, ARMSHIFT_LSR, 31); \ |
| 430 | ARM_ADD_REG_IMMSHIFT(0, ARMREG_PC, ARMREG_PC, reg_a0, ARMSHIFT_LSL, 2); \ |
| 431 | write32(new_pc); \ |
| 432 | generate_function_call(arm_update_gba_##mode); \ |
| 433 | generate_branch_filler(ARMCOND_AL, writeback_location) \ |
| 434 | |
| 435 | |
| 436 | #define generate_branch_no_cycle_update(writeback_location, new_pc, mode) \ |
| 437 | if(pc == idle_loop_target_pc) \ |
| 438 | { \ |
| 439 | generate_branch_idle_eliminate(writeback_location, new_pc, mode); \ |
| 440 | } \ |
| 441 | else \ |
| 442 | { \ |
| 443 | generate_branch_update(writeback_location, new_pc, mode); \ |
| 444 | } \ |
| 445 | |
| 446 | #define generate_branch_cycle_update(writeback_location, new_pc, mode) \ |
| 447 | generate_cycle_update(); \ |
| 448 | generate_branch_no_cycle_update(writeback_location, new_pc, mode) \ |
| 449 | |
| 450 | // a0 holds the destination |
| 451 | |
| 452 | #define generate_indirect_branch_no_cycle_update(type) \ |
| 453 | ARM_B(0, arm_relative_offset(translation_ptr, arm_indirect_branch_##type)) \ |
| 454 | |
| 455 | #define generate_indirect_branch_cycle_update(type) \ |
| 456 | generate_cycle_update(); \ |
| 457 | generate_indirect_branch_no_cycle_update(type) \ |
| 458 | |
| 459 | #define generate_block_prologue() \ |
| 460 | |
| 461 | #define generate_block_extra_vars_arm() \ |
| 462 | void generate_indirect_branch_arm() \ |
| 463 | { \ |
| 464 | if(condition == 0x0E) \ |
| 465 | { \ |
| 466 | generate_cycle_update(); \ |
| 467 | } \ |
| 468 | generate_indirect_branch_no_cycle_update(arm); \ |
| 469 | } \ |
| 470 | \ |
| 471 | void generate_indirect_branch_dual() \ |
| 472 | { \ |
| 473 | if(condition == 0x0E) \ |
| 474 | { \ |
| 475 | generate_cycle_update(); \ |
| 476 | } \ |
| 477 | generate_indirect_branch_no_cycle_update(dual_arm); \ |
| 478 | } \ |
| 479 | \ |
| 480 | u32 prepare_load_reg(u32 scratch_reg, u32 reg_index) \ |
| 481 | { \ |
| 482 | u32 reg_use = arm_register_allocation[reg_index]; \ |
| 483 | if(reg_use == mem_reg) \ |
| 484 | { \ |
| 485 | ARM_LDR_IMM(0, scratch_reg, reg_base, \ |
| 486 | (reg_base_offset + (reg_index * 4))); \ |
| 487 | return scratch_reg; \ |
| 488 | } \ |
| 489 | \ |
| 490 | return reg_use; \ |
| 491 | } \ |
| 492 | \ |
| 493 | u32 prepare_load_reg_pc(u32 scratch_reg, u32 reg_index, u32 pc_offset) \ |
| 494 | { \ |
| 495 | if(reg_index == 15) \ |
| 496 | { \ |
| 497 | generate_load_pc(scratch_reg, pc + pc_offset); \ |
| 498 | return scratch_reg; \ |
| 499 | } \ |
| 500 | return prepare_load_reg(scratch_reg, reg_index); \ |
| 501 | } \ |
| 502 | \ |
| 503 | u32 prepare_store_reg(u32 scratch_reg, u32 reg_index) \ |
| 504 | { \ |
| 505 | u32 reg_use = arm_register_allocation[reg_index]; \ |
| 506 | if(reg_use == mem_reg) \ |
| 507 | return scratch_reg; \ |
| 508 | \ |
| 509 | return reg_use; \ |
| 510 | } \ |
| 511 | \ |
| 512 | void complete_store_reg(u32 scratch_reg, u32 reg_index) \ |
| 513 | { \ |
| 514 | if(arm_register_allocation[reg_index] == mem_reg) \ |
| 515 | { \ |
| 516 | ARM_STR_IMM(0, scratch_reg, reg_base, \ |
| 517 | (reg_base_offset + (reg_index * 4))); \ |
| 518 | } \ |
| 519 | } \ |
| 520 | \ |
| 521 | void complete_store_reg_pc_no_flags(u32 scratch_reg, u32 reg_index) \ |
| 522 | { \ |
| 523 | if(reg_index == 15) \ |
| 524 | { \ |
| 525 | generate_indirect_branch_arm(); \ |
| 526 | } \ |
| 527 | else \ |
| 528 | { \ |
| 529 | complete_store_reg(scratch_reg, reg_index); \ |
| 530 | } \ |
| 531 | } \ |
| 532 | \ |
| 533 | void complete_store_reg_pc_flags(u32 scratch_reg, u32 reg_index) \ |
| 534 | { \ |
| 535 | if(reg_index == 15) \ |
| 536 | { \ |
| 537 | if(condition == 0x0E) \ |
| 538 | { \ |
| 539 | generate_cycle_update(); \ |
| 540 | } \ |
| 541 | generate_function_call(execute_spsr_restore); \ |
| 542 | } \ |
| 543 | else \ |
| 544 | { \ |
| 545 | complete_store_reg(scratch_reg, reg_index); \ |
| 546 | } \ |
| 547 | } \ |
| 548 | \ |
| 549 | void generate_load_reg(u32 ireg, u32 reg_index) \ |
| 550 | { \ |
| 551 | s32 load_src = arm_register_allocation[reg_index]; \ |
| 552 | if(load_src != mem_reg) \ |
| 553 | { \ |
| 554 | ARM_MOV_REG_REG(0, ireg, load_src); \ |
| 555 | } \ |
| 556 | else \ |
| 557 | { \ |
| 558 | ARM_LDR_IMM(0, ireg, reg_base, (reg_base_offset + (reg_index * 4))); \ |
| 559 | } \ |
| 560 | } \ |
| 561 | \ |
| 562 | void generate_store_reg(u32 ireg, u32 reg_index) \ |
| 563 | { \ |
| 564 | s32 store_dest = arm_register_allocation[reg_index]; \ |
| 565 | if(store_dest != mem_reg) \ |
| 566 | { \ |
| 567 | ARM_MOV_REG_REG(0, store_dest, ireg); \ |
| 568 | } \ |
| 569 | else \ |
| 570 | { \ |
| 571 | ARM_STR_IMM(0, ireg, reg_base, (reg_base_offset + (reg_index * 4))); \ |
| 572 | } \ |
| 573 | } \ |
| 574 | |
| 575 | |
| 576 | #define generate_block_extra_vars_thumb() \ |
| 577 | u32 prepare_load_reg(u32 scratch_reg, u32 reg_index) \ |
| 578 | { \ |
| 579 | u32 reg_use = thumb_register_allocation[reg_index]; \ |
| 580 | if(reg_use == mem_reg) \ |
| 581 | { \ |
| 582 | ARM_LDR_IMM(0, scratch_reg, reg_base, \ |
| 583 | (reg_base_offset + (reg_index * 4))); \ |
| 584 | return scratch_reg; \ |
| 585 | } \ |
| 586 | \ |
| 587 | return reg_use; \ |
| 588 | } \ |
| 589 | \ |
| 590 | u32 prepare_load_reg_pc(u32 scratch_reg, u32 reg_index, u32 pc_offset) \ |
| 591 | { \ |
| 592 | if(reg_index == 15) \ |
| 593 | { \ |
| 594 | generate_load_pc(scratch_reg, pc + pc_offset); \ |
| 595 | return scratch_reg; \ |
| 596 | } \ |
| 597 | return prepare_load_reg(scratch_reg, reg_index); \ |
| 598 | } \ |
| 599 | \ |
| 600 | u32 prepare_store_reg(u32 scratch_reg, u32 reg_index) \ |
| 601 | { \ |
| 602 | u32 reg_use = thumb_register_allocation[reg_index]; \ |
| 603 | if(reg_use == mem_reg) \ |
| 604 | return scratch_reg; \ |
| 605 | \ |
| 606 | return reg_use; \ |
| 607 | } \ |
| 608 | \ |
| 609 | void complete_store_reg(u32 scratch_reg, u32 reg_index) \ |
| 610 | { \ |
| 611 | if(thumb_register_allocation[reg_index] == mem_reg) \ |
| 612 | { \ |
| 613 | ARM_STR_IMM(0, scratch_reg, reg_base, \ |
| 614 | (reg_base_offset + (reg_index * 4))); \ |
| 615 | } \ |
| 616 | } \ |
| 617 | \ |
| 618 | void generate_load_reg(u32 ireg, u32 reg_index) \ |
| 619 | { \ |
| 620 | s32 load_src = thumb_register_allocation[reg_index]; \ |
| 621 | if(load_src != mem_reg) \ |
| 622 | { \ |
| 623 | ARM_MOV_REG_REG(0, ireg, load_src); \ |
| 624 | } \ |
| 625 | else \ |
| 626 | { \ |
| 627 | ARM_LDR_IMM(0, ireg, reg_base, (reg_base_offset + (reg_index * 4))); \ |
| 628 | } \ |
| 629 | } \ |
| 630 | \ |
| 631 | void generate_store_reg(u32 ireg, u32 reg_index) \ |
| 632 | { \ |
| 633 | s32 store_dest = thumb_register_allocation[reg_index]; \ |
| 634 | if(store_dest != mem_reg) \ |
| 635 | { \ |
| 636 | ARM_MOV_REG_REG(0, store_dest, ireg); \ |
| 637 | } \ |
| 638 | else \ |
| 639 | { \ |
| 640 | ARM_STR_IMM(0, ireg, reg_base, (reg_base_offset + (reg_index * 4))); \ |
| 641 | } \ |
| 642 | } \ |
| 643 | |
| 644 | u8 *last_rom_translation_ptr = rom_translation_cache; |
| 645 | u8 *last_ram_translation_ptr = ram_translation_cache; |
| 646 | u8 *last_bios_translation_ptr = bios_translation_cache; |
| 647 | |
| 648 | #define translate_invalidate_dcache_one(which) \ |
| 649 | if (which##_translation_ptr > last_##which##_translation_ptr) \ |
| 650 | { \ |
| 651 | warm_cache_op_range(WOP_D_CLEAN, last_##which##_translation_ptr, \ |
| 652 | which##_translation_ptr - last_##which##_translation_ptr); \ |
| 653 | warm_cache_op_range(WOP_I_INVALIDATE, last_##which##_translation_ptr, 32);\ |
| 654 | last_##which##_translation_ptr = which##_translation_ptr; \ |
| 655 | } |
| 656 | |
| 657 | #define translate_invalidate_dcache() \ |
| 658 | { \ |
| 659 | translate_invalidate_dcache_one(rom) \ |
| 660 | translate_invalidate_dcache_one(ram) \ |
| 661 | translate_invalidate_dcache_one(bios) \ |
| 662 | } |
| 663 | |
| 664 | #define invalidate_icache_region(addr, size) \ |
| 665 | warm_cache_op_range(WOP_I_INVALIDATE, addr, size) |
| 666 | |
| 667 | |
| 668 | #define block_prologue_size 0 |
| 669 | |
| 670 | |
| 671 | // It should be okay to still generate result flags, spsr will overwrite them. |
| 672 | // This is pretty infrequent (returning from interrupt handlers, et al) so |
| 673 | // probably not worth optimizing for. |
| 674 | |
| 675 | #define check_for_interrupts() \ |
| 676 | if((io_registers[REG_IE] & io_registers[REG_IF]) && \ |
| 677 | io_registers[REG_IME] && ((reg[REG_CPSR] & 0x80) == 0)) \ |
| 678 | { \ |
| 679 | reg_mode[MODE_IRQ][6] = pc + 4; \ |
| 680 | spsr[MODE_IRQ] = reg[REG_CPSR]; \ |
| 681 | reg[REG_CPSR] = 0xD2; \ |
| 682 | pc = 0x00000018; \ |
| 683 | set_cpu_mode(MODE_IRQ); \ |
| 684 | } \ |
| 685 | |
| 686 | #define generate_load_reg_pc(ireg, reg_index, pc_offset) \ |
| 687 | if(reg_index == 15) \ |
| 688 | { \ |
| 689 | generate_load_pc(ireg, pc + pc_offset); \ |
| 690 | } \ |
| 691 | else \ |
| 692 | { \ |
| 693 | generate_load_reg(ireg, reg_index); \ |
| 694 | } \ |
| 695 | |
| 696 | #define generate_store_reg_pc_no_flags(ireg, reg_index) \ |
| 697 | generate_store_reg(ireg, reg_index); \ |
| 698 | if(reg_index == 15) \ |
| 699 | { \ |
| 700 | generate_indirect_branch_arm(); \ |
| 701 | } \ |
| 702 | |
| 703 | |
| 704 | u32 function_cc execute_spsr_restore_body(u32 pc) |
| 705 | { |
| 706 | set_cpu_mode(cpu_modes[reg[REG_CPSR] & 0x1F]); |
| 707 | check_for_interrupts(); |
| 708 | |
| 709 | return pc; |
| 710 | } |
| 711 | |
| 712 | |
| 713 | #define generate_store_reg_pc_flags(ireg, reg_index) \ |
| 714 | generate_store_reg(ireg, reg_index); \ |
| 715 | if(reg_index == 15) \ |
| 716 | { \ |
| 717 | if(condition == 0x0E) \ |
| 718 | { \ |
| 719 | generate_cycle_update(); \ |
| 720 | } \ |
| 721 | generate_function_call(execute_spsr_restore); \ |
| 722 | } \ |
| 723 | |
| 724 | |
| 725 | #define generate_load_flags() \ |
| 726 | /* ARM_MSR_REG(0, ARM_PSR_F, reg_flags, ARM_CPSR) */ \ |
| 727 | |
| 728 | #define generate_store_flags() \ |
| 729 | /* ARM_MRS_CPSR(0, reg_flags) */ \ |
| 730 | |
| 731 | #define generate_save_flags() \ |
| 732 | ARM_MRS_CPSR(0, reg_flags) \ |
| 733 | |
| 734 | #define generate_restore_flags() \ |
| 735 | ARM_MSR_REG(0, ARM_PSR_F, reg_flags, ARM_CPSR) \ |
| 736 | |
| 737 | |
| 738 | #define condition_opposite_eq ARMCOND_NE |
| 739 | #define condition_opposite_ne ARMCOND_EQ |
| 740 | #define condition_opposite_cs ARMCOND_CC |
| 741 | #define condition_opposite_cc ARMCOND_CS |
| 742 | #define condition_opposite_mi ARMCOND_PL |
| 743 | #define condition_opposite_pl ARMCOND_MI |
| 744 | #define condition_opposite_vs ARMCOND_VC |
| 745 | #define condition_opposite_vc ARMCOND_VS |
| 746 | #define condition_opposite_hi ARMCOND_LS |
| 747 | #define condition_opposite_ls ARMCOND_HI |
| 748 | #define condition_opposite_ge ARMCOND_LT |
| 749 | #define condition_opposite_lt ARMCOND_GE |
| 750 | #define condition_opposite_gt ARMCOND_LE |
| 751 | #define condition_opposite_le ARMCOND_GT |
| 752 | #define condition_opposite_al ARMCOND_NV |
| 753 | #define condition_opposite_nv ARMCOND_AL |
| 754 | |
| 755 | #define generate_branch(mode) \ |
| 756 | { \ |
| 757 | generate_branch_cycle_update( \ |
| 758 | block_exits[block_exit_position].branch_source, \ |
| 759 | block_exits[block_exit_position].branch_target, mode); \ |
| 760 | block_exit_position++; \ |
| 761 | } \ |
| 762 | |
| 763 | |
| 764 | #define generate_op_and_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 765 | ARM_AND_REG_IMMSHIFT(0, _rd, _rn, _rm, shift_type, shift) \ |
| 766 | |
| 767 | #define generate_op_orr_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 768 | ARM_ORR_REG_IMMSHIFT(0, _rd, _rn, _rm, shift_type, shift) \ |
| 769 | |
| 770 | #define generate_op_eor_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 771 | ARM_EOR_REG_IMMSHIFT(0, _rd, _rn, _rm, shift_type, shift) \ |
| 772 | |
| 773 | #define generate_op_bic_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 774 | ARM_BIC_REG_IMMSHIFT(0, _rd, _rn, _rm, shift_type, shift) \ |
| 775 | |
| 776 | #define generate_op_sub_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 777 | ARM_SUB_REG_IMMSHIFT(0, _rd, _rn, _rm, shift_type, shift) \ |
| 778 | |
| 779 | #define generate_op_rsb_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 780 | ARM_RSB_REG_IMMSHIFT(0, _rd, _rn, _rm, shift_type, shift) \ |
| 781 | |
| 782 | #define generate_op_sbc_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 783 | ARM_SBC_REG_IMMSHIFT(0, _rd, _rn, _rm, shift_type, shift) \ |
| 784 | |
| 785 | #define generate_op_rsc_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 786 | ARM_RSC_REG_IMMSHIFT(0, _rd, _rn, _rm, shift_type, shift) \ |
| 787 | |
| 788 | #define generate_op_add_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 789 | ARM_ADD_REG_IMMSHIFT(0, _rd, _rn, _rm, shift_type, shift) \ |
| 790 | |
| 791 | #define generate_op_adc_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 792 | ARM_ADC_REG_IMMSHIFT(0, _rd, _rn, _rm, shift_type, shift) \ |
| 793 | |
| 794 | #define generate_op_mov_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 795 | ARM_MOV_REG_IMMSHIFT(0, _rd, _rm, shift_type, shift) \ |
| 796 | |
| 797 | #define generate_op_mvn_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 798 | ARM_MVN_REG_IMMSHIFT(0, _rd, _rm, shift_type, shift) \ |
| 799 | |
| 800 | |
| 801 | #define generate_op_and_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 802 | ARM_AND_REG_REGSHIFT(0, _rd, _rn, _rm, shift_type, _rs) \ |
| 803 | |
| 804 | #define generate_op_orr_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 805 | ARM_ORR_REG_REGSHIFT(0, _rd, _rn, _rm, shift_type, _rs) \ |
| 806 | |
| 807 | #define generate_op_eor_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 808 | ARM_EOR_REG_REGSHIFT(0, _rd, _rn, _rm, shift_type, _rs) \ |
| 809 | |
| 810 | #define generate_op_bic_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 811 | ARM_BIC_REG_REGSHIFT(0, _rd, _rn, _rm, shift_type, _rs) \ |
| 812 | |
| 813 | #define generate_op_sub_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 814 | ARM_SUB_REG_REGSHIFT(0, _rd, _rn, _rm, shift_type, _rs) \ |
| 815 | |
| 816 | #define generate_op_rsb_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 817 | ARM_RSB_REG_REGSHIFT(0, _rd, _rn, _rm, shift_type, _rs) \ |
| 818 | |
| 819 | #define generate_op_sbc_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 820 | ARM_SBC_REG_REGSHIFT(0, _rd, _rn, _rm, shift_type, _rs) \ |
| 821 | |
| 822 | #define generate_op_rsc_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 823 | ARM_RSC_REG_REGSHIFT(0, _rd, _rn, _rm, shift_type, _rs) \ |
| 824 | |
| 825 | #define generate_op_add_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 826 | ARM_ADD_REG_REGSHIFT(0, _rd, _rn, _rm, shift_type, _rs) \ |
| 827 | |
| 828 | #define generate_op_adc_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 829 | ARM_ADC_REG_REGSHIFT(0, _rd, _rn, _rm, shift_type, _rs) \ |
| 830 | |
| 831 | #define generate_op_mov_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 832 | ARM_MOV_REG_REGSHIFT(0, _rd, _rm, shift_type, _rs) \ |
| 833 | |
| 834 | #define generate_op_mvn_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 835 | ARM_MVN_REG_REGSHIFT(0, _rd, _rm, shift_type, _rs) \ |
| 836 | |
| 837 | |
| 838 | #define generate_op_and_imm(_rd, _rn) \ |
| 839 | ARM_AND_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 840 | |
| 841 | #define generate_op_orr_imm(_rd, _rn) \ |
| 842 | ARM_ORR_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 843 | |
| 844 | #define generate_op_eor_imm(_rd, _rn) \ |
| 845 | ARM_EOR_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 846 | |
| 847 | #define generate_op_bic_imm(_rd, _rn) \ |
| 848 | ARM_BIC_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 849 | |
| 850 | #define generate_op_sub_imm(_rd, _rn) \ |
| 851 | ARM_SUB_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 852 | |
| 853 | #define generate_op_rsb_imm(_rd, _rn) \ |
| 854 | ARM_RSB_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 855 | |
| 856 | #define generate_op_sbc_imm(_rd, _rn) \ |
| 857 | ARM_SBC_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 858 | |
| 859 | #define generate_op_rsc_imm(_rd, _rn) \ |
| 860 | ARM_RSC_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 861 | |
| 862 | #define generate_op_add_imm(_rd, _rn) \ |
| 863 | ARM_ADD_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 864 | |
| 865 | #define generate_op_adc_imm(_rd, _rn) \ |
| 866 | ARM_ADC_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 867 | |
| 868 | #define generate_op_mov_imm(_rd, _rn) \ |
| 869 | ARM_MOV_REG_IMM(0, _rd, imm, imm_ror) \ |
| 870 | |
| 871 | #define generate_op_mvn_imm(_rd, _rn) \ |
| 872 | ARM_MVN_REG_IMM(0, _rd, imm, imm_ror) \ |
| 873 | |
| 874 | |
| 875 | #define generate_op_reg_immshift_lflags(name, _rd, _rn, _rm, st, shift) \ |
| 876 | ARM_##name##_REG_IMMSHIFT(0, _rd, _rn, _rm, st, shift) \ |
| 877 | |
| 878 | #define generate_op_reg_immshift_aflags(name, _rd, _rn, _rm, st, shift) \ |
| 879 | ARM_##name##_REG_IMMSHIFT(0, _rd, _rn, _rm, st, shift) \ |
| 880 | |
| 881 | #define generate_op_reg_immshift_aflags_load_c(name, _rd, _rn, _rm, st, sh) \ |
| 882 | ARM_##name##_REG_IMMSHIFT(0, _rd, _rn, _rm, st, sh) \ |
| 883 | |
| 884 | #define generate_op_reg_immshift_uflags(name, _rd, _rm, shift_type, shift) \ |
| 885 | ARM_##name##_REG_IMMSHIFT(0, _rd, _rm, shift_type, shift) \ |
| 886 | |
| 887 | #define generate_op_reg_immshift_tflags(name, _rn, _rm, shift_type, shift) \ |
| 888 | ARM_##name##_REG_IMMSHIFT(0, _rn, _rm, shift_type, shift) \ |
| 889 | |
| 890 | |
| 891 | #define generate_op_reg_regshift_lflags(name, _rd, _rn, _rm, shift_type, _rs) \ |
| 892 | ARM_##name##_REG_REGSHIFT(0, _rd, _rn, _rm, shift_type, _rs) \ |
| 893 | |
| 894 | #define generate_op_reg_regshift_aflags(name, _rd, _rn, _rm, st, _rs) \ |
| 895 | ARM_##name##_REG_REGSHIFT(0, _rd, _rn, _rm, st, _rs) \ |
| 896 | |
| 897 | #define generate_op_reg_regshift_aflags_load_c(name, _rd, _rn, _rm, st, _rs) \ |
| 898 | ARM_##name##_REG_REGSHIFT(0, _rd, _rn, _rm, st, _rs) \ |
| 899 | |
| 900 | #define generate_op_reg_regshift_uflags(name, _rd, _rm, shift_type, _rs) \ |
| 901 | ARM_##name##_REG_REGSHIFT(0, _rd, _rm, shift_type, _rs) \ |
| 902 | |
| 903 | #define generate_op_reg_regshift_tflags(name, _rn, _rm, shift_type, _rs) \ |
| 904 | ARM_##name##_REG_REGSHIFT(0, _rn, _rm, shift_type, _rs) \ |
| 905 | |
| 906 | |
| 907 | #define generate_op_imm_lflags(name, _rd, _rn) \ |
| 908 | ARM_##name##_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 909 | |
| 910 | #define generate_op_imm_aflags(name, _rd, _rn) \ |
| 911 | ARM_##name##_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 912 | |
| 913 | #define generate_op_imm_aflags_load_c(name, _rd, _rn) \ |
| 914 | ARM_##name##_REG_IMM(0, _rd, _rn, imm, imm_ror) \ |
| 915 | |
| 916 | #define generate_op_imm_uflags(name, _rd) \ |
| 917 | ARM_##name##_REG_IMM(0, _rd, imm, imm_ror) \ |
| 918 | |
| 919 | #define generate_op_imm_tflags(name, _rn) \ |
| 920 | ARM_##name##_REG_IMM(0, _rn, imm, imm_ror) \ |
| 921 | |
| 922 | |
| 923 | #define generate_op_ands_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 924 | generate_op_reg_immshift_lflags(ANDS, _rd, _rn, _rm, shift_type, shift) \ |
| 925 | |
| 926 | #define generate_op_orrs_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 927 | generate_op_reg_immshift_lflags(ORRS, _rd, _rn, _rm, shift_type, shift) \ |
| 928 | |
| 929 | #define generate_op_eors_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 930 | generate_op_reg_immshift_lflags(EORS, _rd, _rn, _rm, shift_type, shift) \ |
| 931 | |
| 932 | #define generate_op_bics_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 933 | generate_op_reg_immshift_lflags(BICS, _rd, _rn, _rm, shift_type, shift) \ |
| 934 | |
| 935 | #define generate_op_subs_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 936 | generate_op_reg_immshift_aflags(SUBS, _rd, _rn, _rm, shift_type, shift) \ |
| 937 | |
| 938 | #define generate_op_rsbs_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 939 | generate_op_reg_immshift_aflags(RSBS, _rd, _rn, _rm, shift_type, shift) \ |
| 940 | |
| 941 | #define generate_op_sbcs_reg_immshift(_rd, _rn, _rm, st, shift) \ |
| 942 | generate_op_reg_immshift_aflags_load_c(SBCS, _rd, _rn, _rm, st, shift) \ |
| 943 | |
| 944 | #define generate_op_rscs_reg_immshift(_rd, _rn, _rm, st, shift) \ |
| 945 | generate_op_reg_immshift_aflags_load_c(RSCS, _rd, _rn, _rm, st, shift) \ |
| 946 | |
| 947 | #define generate_op_adds_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 948 | generate_op_reg_immshift_aflags(ADDS, _rd, _rn, _rm, shift_type, shift) \ |
| 949 | |
| 950 | #define generate_op_adcs_reg_immshift(_rd, _rn, _rm, st, shift) \ |
| 951 | generate_op_reg_immshift_aflags_load_c(ADCS, _rd, _rn, _rm, st, shift) \ |
| 952 | |
| 953 | #define generate_op_movs_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 954 | generate_op_reg_immshift_uflags(MOVS, _rd, _rm, shift_type, shift) \ |
| 955 | |
| 956 | #define generate_op_mvns_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 957 | generate_op_reg_immshift_uflags(MVNS, _rd, _rm, shift_type, shift) \ |
| 958 | |
| 959 | // The reg operand is in reg_rm, not reg_rn like expected, so rsbs isn't |
| 960 | // being used here. When rsbs is fully inlined it can be used with the |
| 961 | // apropriate operands. |
| 962 | |
| 963 | #define generate_op_neg_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 964 | { \ |
| 965 | generate_load_imm(reg_rn, 0, 0); \ |
| 966 | generate_op_subs_reg_immshift(_rd, reg_rn, _rm, ARMSHIFT_LSL, 0); \ |
| 967 | } \ |
| 968 | |
| 969 | #define generate_op_muls_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 970 | generate_load_flags(); \ |
| 971 | ARM_MULS(0, _rd, _rn, _rm); \ |
| 972 | generate_store_flags() \ |
| 973 | |
| 974 | #define generate_op_cmp_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 975 | generate_op_reg_immshift_tflags(CMP, _rn, _rm, shift_type, shift) \ |
| 976 | |
| 977 | #define generate_op_cmn_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 978 | generate_op_reg_immshift_tflags(CMN, _rn, _rm, shift_type, shift) \ |
| 979 | |
| 980 | #define generate_op_tst_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 981 | generate_op_reg_immshift_tflags(TST, _rn, _rm, shift_type, shift) \ |
| 982 | |
| 983 | #define generate_op_teq_reg_immshift(_rd, _rn, _rm, shift_type, shift) \ |
| 984 | generate_op_reg_immshift_tflags(TEQ, _rn, _rm, shift_type, shift) \ |
| 985 | |
| 986 | |
| 987 | #define generate_op_ands_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 988 | generate_op_reg_regshift_lflags(ANDS, _rd, _rn, _rm, shift_type, _rs) \ |
| 989 | |
| 990 | #define generate_op_orrs_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 991 | generate_op_reg_regshift_lflags(ORRS, _rd, _rn, _rm, shift_type, _rs) \ |
| 992 | |
| 993 | #define generate_op_eors_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 994 | generate_op_reg_regshift_lflags(EORS, _rd, _rn, _rm, shift_type, _rs) \ |
| 995 | |
| 996 | #define generate_op_bics_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 997 | generate_op_reg_regshift_lflags(BICS, _rd, _rn, _rm, shift_type, _rs) \ |
| 998 | |
| 999 | #define generate_op_subs_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 1000 | generate_op_reg_regshift_aflags(SUBS, _rd, _rn, _rm, shift_type, _rs) \ |
| 1001 | |
| 1002 | #define generate_op_rsbs_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 1003 | generate_op_reg_regshift_aflags(RSBS, _rd, _rn, _rm, shift_type, _rs) \ |
| 1004 | |
| 1005 | #define generate_op_sbcs_reg_regshift(_rd, _rn, _rm, st, _rs) \ |
| 1006 | generate_op_reg_regshift_aflags_load_c(SBCS, _rd, _rn, _rm, st, _rs) \ |
| 1007 | |
| 1008 | #define generate_op_rscs_reg_regshift(_rd, _rn, _rm, st, _rs) \ |
| 1009 | generate_op_reg_regshift_aflags_load_c(RSCS, _rd, _rn, _rm, st, _rs) \ |
| 1010 | |
| 1011 | #define generate_op_adds_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 1012 | generate_op_reg_regshift_aflags(ADDS, _rd, _rn, _rm, shift_type, _rs) \ |
| 1013 | |
| 1014 | #define generate_op_adcs_reg_regshift(_rd, _rn, _rm, st, _rs) \ |
| 1015 | generate_op_reg_regshift_aflags_load_c(ADCS, _rd, _rn, _rm, st, _rs) \ |
| 1016 | |
| 1017 | #define generate_op_movs_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 1018 | generate_op_reg_regshift_uflags(MOVS, _rd, _rm, shift_type, _rs) \ |
| 1019 | |
| 1020 | #define generate_op_mvns_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 1021 | generate_op_reg_regshift_uflags(MVNS, _rd, _rm, shift_type, _rs) \ |
| 1022 | |
| 1023 | #define generate_op_cmp_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 1024 | generate_op_reg_regshift_tflags(CMP, _rn, _rm, shift_type, _rs) \ |
| 1025 | |
| 1026 | #define generate_op_cmn_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 1027 | generate_op_reg_regshift_tflags(CMN, _rn, _rm, shift_type, _rs) \ |
| 1028 | |
| 1029 | #define generate_op_tst_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 1030 | generate_op_reg_regshift_tflags(TST, _rn, _rm, shift_type, _rs) \ |
| 1031 | |
| 1032 | #define generate_op_teq_reg_regshift(_rd, _rn, _rm, shift_type, _rs) \ |
| 1033 | generate_op_reg_regshift_tflags(TEQ, _rn, _rm, shift_type, _rs) \ |
| 1034 | |
| 1035 | |
| 1036 | #define generate_op_ands_imm(_rd, _rn) \ |
| 1037 | generate_op_imm_lflags(ANDS, _rd, _rn) \ |
| 1038 | |
| 1039 | #define generate_op_orrs_imm(_rd, _rn) \ |
| 1040 | generate_op_imm_lflags(ORRS, _rd, _rn) \ |
| 1041 | |
| 1042 | #define generate_op_eors_imm(_rd, _rn) \ |
| 1043 | generate_op_imm_lflags(EORS, _rd, _rn) \ |
| 1044 | |
| 1045 | #define generate_op_bics_imm(_rd, _rn) \ |
| 1046 | generate_op_imm_lflags(BICS, _rd, _rn) \ |
| 1047 | |
| 1048 | #define generate_op_subs_imm(_rd, _rn) \ |
| 1049 | generate_op_imm_aflags(SUBS, _rd, _rn) \ |
| 1050 | |
| 1051 | #define generate_op_rsbs_imm(_rd, _rn) \ |
| 1052 | generate_op_imm_aflags(RSBS, _rd, _rn) \ |
| 1053 | |
| 1054 | #define generate_op_sbcs_imm(_rd, _rn) \ |
| 1055 | generate_op_imm_aflags_load_c(SBCS, _rd, _rn) \ |
| 1056 | |
| 1057 | #define generate_op_rscs_imm(_rd, _rn) \ |
| 1058 | generate_op_imm_aflags_load_c(RSCS, _rd, _rn) \ |
| 1059 | |
| 1060 | #define generate_op_adds_imm(_rd, _rn) \ |
| 1061 | generate_op_imm_aflags(ADDS, _rd, _rn) \ |
| 1062 | |
| 1063 | #define generate_op_adcs_imm(_rd, _rn) \ |
| 1064 | generate_op_imm_aflags_load_c(ADCS, _rd, _rn) \ |
| 1065 | |
| 1066 | #define generate_op_movs_imm(_rd, _rn) \ |
| 1067 | generate_op_imm_uflags(MOVS, _rd) \ |
| 1068 | |
| 1069 | #define generate_op_mvns_imm(_rd, _rn) \ |
| 1070 | generate_op_imm_uflags(MVNS, _rd) \ |
| 1071 | |
| 1072 | #define generate_op_cmp_imm(_rd, _rn) \ |
| 1073 | generate_op_imm_tflags(CMP, _rn) \ |
| 1074 | |
| 1075 | #define generate_op_cmn_imm(_rd, _rn) \ |
| 1076 | generate_op_imm_tflags(CMN, _rn) \ |
| 1077 | |
| 1078 | #define generate_op_tst_imm(_rd, _rn) \ |
| 1079 | generate_op_imm_tflags(TST, _rn) \ |
| 1080 | |
| 1081 | #define generate_op_teq_imm(_rd, _rn) \ |
| 1082 | generate_op_imm_tflags(TEQ, _rn) \ |
| 1083 | |
| 1084 | |
| 1085 | #define prepare_load_rn_yes() \ |
| 1086 | u32 _rn = prepare_load_reg_pc(reg_rn, rn, 8) \ |
| 1087 | |
| 1088 | #define prepare_load_rn_no() \ |
| 1089 | |
| 1090 | #define prepare_store_rd_yes() \ |
| 1091 | u32 _rd = prepare_store_reg(reg_rd, rd) \ |
| 1092 | |
| 1093 | #define prepare_store_rd_no() \ |
| 1094 | |
| 1095 | #define complete_store_rd_yes(flags_op) \ |
| 1096 | complete_store_reg_pc_##flags_op(_rd, rd) \ |
| 1097 | |
| 1098 | #define complete_store_rd_no(flags_op) \ |
| 1099 | |
| 1100 | #define arm_generate_op_reg(name, load_op, store_op, flags_op) \ |
| 1101 | u32 shift_type = (opcode >> 5) & 0x03; \ |
| 1102 | arm_decode_data_proc_reg(); \ |
| 1103 | prepare_load_rn_##load_op(); \ |
| 1104 | prepare_store_rd_##store_op(); \ |
| 1105 | \ |
| 1106 | if((opcode >> 4) & 0x01) \ |
| 1107 | { \ |
| 1108 | u32 rs = ((opcode >> 8) & 0x0F); \ |
| 1109 | u32 _rs = prepare_load_reg(reg_rs, rs); \ |
| 1110 | u32 _rm = prepare_load_reg_pc(reg_rm, rm, 12); \ |
| 1111 | generate_op_##name##_reg_regshift(_rd, _rn, _rm, shift_type, _rs); \ |
| 1112 | } \ |
| 1113 | else \ |
| 1114 | { \ |
| 1115 | u32 shift_imm = ((opcode >> 7) & 0x1F); \ |
| 1116 | u32 _rm = prepare_load_reg_pc(reg_rm, rm, 8); \ |
| 1117 | generate_op_##name##_reg_immshift(_rd, _rn, _rm, shift_type, shift_imm); \ |
| 1118 | } \ |
| 1119 | complete_store_rd_##store_op(flags_op) \ |
| 1120 | |
| 1121 | #define arm_generate_op_reg_flags(name, load_op, store_op, flags_op) \ |
| 1122 | arm_generate_op_reg(name, load_op, store_op, flags_op) \ |
| 1123 | |
| 1124 | // imm will be loaded by the called function if necessary. |
| 1125 | |
| 1126 | #define arm_generate_op_imm(name, load_op, store_op, flags_op) \ |
| 1127 | arm_decode_data_proc_imm(); \ |
| 1128 | prepare_load_rn_##load_op(); \ |
| 1129 | prepare_store_rd_##store_op(); \ |
| 1130 | generate_op_##name##_imm(_rd, _rn); \ |
| 1131 | complete_store_rd_##store_op(flags_op) \ |
| 1132 | |
| 1133 | #define arm_generate_op_imm_flags(name, load_op, store_op, flags_op) \ |
| 1134 | arm_generate_op_imm(name, load_op, store_op, flags_op) \ |
| 1135 | |
| 1136 | #define arm_data_proc(name, type, flags_op) \ |
| 1137 | { \ |
| 1138 | arm_generate_op_##type(name, yes, yes, flags_op); \ |
| 1139 | } \ |
| 1140 | |
| 1141 | #define arm_data_proc_test(name, type) \ |
| 1142 | { \ |
| 1143 | arm_generate_op_##type(name, yes, no, no); \ |
| 1144 | } \ |
| 1145 | |
| 1146 | #define arm_data_proc_unary(name, type, flags_op) \ |
| 1147 | { \ |
| 1148 | arm_generate_op_##type(name, no, yes, flags_op); \ |
| 1149 | } \ |
| 1150 | |
| 1151 | |
| 1152 | #define arm_multiply_add_no_flags_no() \ |
| 1153 | ARM_MUL(0, _rd, _rm, _rs) \ |
| 1154 | |
| 1155 | #define arm_multiply_add_yes_flags_no() \ |
| 1156 | u32 _rn = prepare_load_reg(reg_a2, rn); \ |
| 1157 | ARM_MLA(0, _rd, _rm, _rs, _rn) \ |
| 1158 | |
| 1159 | #define arm_multiply_add_no_flags_yes() \ |
| 1160 | generate_load_flags(); \ |
| 1161 | ARM_MULS(0, reg_a0, reg_a0, reg_a1) \ |
| 1162 | generate_store_flags() \ |
| 1163 | |
| 1164 | #define arm_multiply_add_yes_flags_yes() \ |
| 1165 | u32 _rn = prepare_load_reg(reg_a2, rn); \ |
| 1166 | generate_load_flags(); \ |
| 1167 | ARM_MLAS(0, _rd, _rm, _rs, _rn); \ |
| 1168 | generate_store_flags() |
| 1169 | |
| 1170 | |
| 1171 | #define arm_multiply(add_op, flags) \ |
| 1172 | { \ |
| 1173 | arm_decode_multiply(); \ |
| 1174 | u32 _rm = prepare_load_reg(reg_a0, rm); \ |
| 1175 | u32 _rs = prepare_load_reg(reg_a1, rs); \ |
| 1176 | u32 _rd = prepare_store_reg(reg_a0, rd); \ |
| 1177 | arm_multiply_add_##add_op##_flags_##flags(); \ |
| 1178 | complete_store_reg(_rd, rd); \ |
| 1179 | } \ |
| 1180 | |
| 1181 | |
| 1182 | #define arm_multiply_long_name_s64 SMULL |
| 1183 | #define arm_multiply_long_name_u64 UMULL |
| 1184 | #define arm_multiply_long_name_s64_add SMLAL |
| 1185 | #define arm_multiply_long_name_u64_add UMLAL |
| 1186 | |
| 1187 | |
| 1188 | #define arm_multiply_long_flags_no(name) \ |
| 1189 | ARM_##name(0, _rdlo, _rdhi, _rm, _rs) \ |
| 1190 | |
| 1191 | #define arm_multiply_long_flags_yes(name) \ |
| 1192 | generate_load_flags(); \ |
| 1193 | ARM_##name##S(0, _rdlo, _rdhi, _rm, _rs); \ |
| 1194 | generate_store_flags() \ |
| 1195 | |
| 1196 | |
| 1197 | #define arm_multiply_long_add_no(name) \ |
| 1198 | |
| 1199 | #define arm_multiply_long_add_yes(name) \ |
| 1200 | prepare_load_reg(reg_a0, rdlo); \ |
| 1201 | prepare_load_reg(reg_a1, rdhi) \ |
| 1202 | |
| 1203 | |
| 1204 | #define arm_multiply_long_op(flags, name) \ |
| 1205 | arm_multiply_long_flags_##flags(name) \ |
| 1206 | |
| 1207 | #define arm_multiply_long(name, add_op, flags) \ |
| 1208 | { \ |
| 1209 | arm_decode_multiply_long(); \ |
| 1210 | u32 _rm = prepare_load_reg(reg_a2, rm); \ |
| 1211 | u32 _rs = prepare_load_reg(reg_rs, rs); \ |
| 1212 | u32 _rdlo = prepare_store_reg(reg_a0, rdlo); \ |
| 1213 | u32 _rdhi = prepare_store_reg(reg_a1, rdhi); \ |
| 1214 | arm_multiply_long_add_##add_op(name); \ |
| 1215 | arm_multiply_long_op(flags, arm_multiply_long_name_##name); \ |
| 1216 | complete_store_reg(_rdlo, rdlo); \ |
| 1217 | complete_store_reg(_rdhi, rdhi); \ |
| 1218 | } \ |
| 1219 | |
| 1220 | #define arm_psr_read_cpsr() \ |
| 1221 | u32 _rd = prepare_store_reg(reg_a0, rd); \ |
| 1222 | generate_load_reg(_rd, REG_CPSR); \ |
| 1223 | ARM_BIC_REG_IMM(0, _rd, _rd, 0xF0, arm_imm_lsl_to_rot(24)); \ |
| 1224 | ARM_AND_REG_IMM(0, reg_flags, reg_flags, 0xF0, arm_imm_lsl_to_rot(24)); \ |
| 1225 | ARM_ORR_REG_REG(0, _rd, _rd, reg_flags); \ |
| 1226 | complete_store_reg(_rd, rd) \ |
| 1227 | |
| 1228 | #define arm_psr_read_spsr() \ |
| 1229 | generate_function_call(execute_read_spsr) \ |
| 1230 | generate_store_reg(reg_a0, rd) \ |
| 1231 | |
| 1232 | #define arm_psr_read(op_type, psr_reg) \ |
| 1233 | arm_psr_read_##psr_reg() \ |
| 1234 | |
| 1235 | // This function's okay because it's called from an ASM function that can |
| 1236 | // wrap it correctly. |
| 1237 | |
| 1238 | u32 execute_store_cpsr_body(u32 _cpsr, u32 store_mask, u32 address) |
| 1239 | { |
| 1240 | reg[REG_CPSR] = _cpsr; |
| 1241 | if(store_mask & 0xFF) |
| 1242 | { |
| 1243 | set_cpu_mode(cpu_modes[_cpsr & 0x1F]); |
| 1244 | if((io_registers[REG_IE] & io_registers[REG_IF]) && |
| 1245 | io_registers[REG_IME] && ((_cpsr & 0x80) == 0)) |
| 1246 | { |
| 1247 | reg_mode[MODE_IRQ][6] = address + 4; |
| 1248 | spsr[MODE_IRQ] = _cpsr; |
| 1249 | reg[REG_CPSR] = 0xD2; |
| 1250 | set_cpu_mode(MODE_IRQ); |
| 1251 | return 0x00000018; |
| 1252 | } |
| 1253 | } |
| 1254 | |
| 1255 | return 0; |
| 1256 | } |
| 1257 | |
| 1258 | #define arm_psr_load_new_reg() \ |
| 1259 | generate_load_reg(reg_a0, rm) \ |
| 1260 | |
| 1261 | #define arm_psr_load_new_imm() \ |
| 1262 | generate_load_imm(reg_a0, imm, imm_ror) \ |
| 1263 | |
| 1264 | #define arm_psr_store_cpsr() \ |
| 1265 | arm_load_imm_32bit(reg_a1, psr_masks[psr_field]); \ |
| 1266 | generate_function_call(execute_store_cpsr); \ |
| 1267 | write32(pc) \ |
| 1268 | |
| 1269 | #define arm_psr_store_spsr() \ |
| 1270 | generate_function_call(execute_store_spsr) \ |
| 1271 | |
| 1272 | #define arm_psr_store(op_type, psr_reg) \ |
| 1273 | arm_psr_load_new_##op_type(); \ |
| 1274 | arm_psr_store_##psr_reg() \ |
| 1275 | |
| 1276 | |
| 1277 | #define arm_psr(op_type, transfer_type, psr_reg) \ |
| 1278 | { \ |
| 1279 | arm_decode_psr_##op_type(); \ |
| 1280 | arm_psr_##transfer_type(op_type, psr_reg); \ |
| 1281 | } \ |
| 1282 | |
| 1283 | // TODO: loads will need the PC passed as well for open address, however can |
| 1284 | // eventually be rectified with a hash table on the memory accesses |
| 1285 | // (same with the stores) |
| 1286 | |
| 1287 | #define arm_access_memory_load(mem_type) \ |
| 1288 | cycle_count += 2; \ |
| 1289 | generate_function_call(execute_load_##mem_type); \ |
| 1290 | write32((pc + 8)); \ |
| 1291 | generate_store_reg_pc_no_flags(reg_rv, rd) \ |
| 1292 | |
| 1293 | #define arm_access_memory_store(mem_type) \ |
| 1294 | cycle_count++; \ |
| 1295 | generate_load_reg_pc(reg_a1, rd, 12); \ |
| 1296 | generate_function_call(execute_store_##mem_type); \ |
| 1297 | write32((pc + 4)) \ |
| 1298 | |
| 1299 | // Calculate the address into a0 from _rn, _rm |
| 1300 | |
| 1301 | #define arm_access_memory_adjust_reg_sh_up(ireg) \ |
| 1302 | ARM_ADD_REG_IMMSHIFT(0, ireg, _rn, _rm, ((opcode >> 5) & 0x03), \ |
| 1303 | ((opcode >> 7) & 0x1F)) \ |
| 1304 | |
| 1305 | #define arm_access_memory_adjust_reg_sh_down(ireg) \ |
| 1306 | ARM_SUB_REG_IMMSHIFT(0, ireg, _rn, _rm, ((opcode >> 5) & 0x03), \ |
| 1307 | ((opcode >> 7) & 0x1F)) \ |
| 1308 | |
| 1309 | #define arm_access_memory_adjust_reg_up(ireg) \ |
| 1310 | ARM_ADD_REG_REG(0, ireg, _rn, _rm) \ |
| 1311 | |
| 1312 | #define arm_access_memory_adjust_reg_down(ireg) \ |
| 1313 | ARM_SUB_REG_REG(0, ireg, _rn, _rm) \ |
| 1314 | |
| 1315 | #define arm_access_memory_adjust_imm(op, ireg) \ |
| 1316 | { \ |
| 1317 | u32 stores[4]; \ |
| 1318 | u32 rotations[4]; \ |
| 1319 | u32 store_count = arm_disect_imm_32bit(offset, stores, rotations); \ |
| 1320 | \ |
| 1321 | if(store_count > 1) \ |
| 1322 | { \ |
| 1323 | ARM_##op##_REG_IMM(0, ireg, _rn, stores[0], rotations[0]); \ |
| 1324 | ARM_##op##_REG_IMM(0, ireg, ireg, stores[1], rotations[1]); \ |
| 1325 | } \ |
| 1326 | else \ |
| 1327 | { \ |
| 1328 | ARM_##op##_REG_IMM(0, ireg, _rn, stores[0], rotations[0]); \ |
| 1329 | } \ |
| 1330 | } \ |
| 1331 | |
| 1332 | #define arm_access_memory_adjust_imm_up(ireg) \ |
| 1333 | arm_access_memory_adjust_imm(ADD, ireg) \ |
| 1334 | |
| 1335 | #define arm_access_memory_adjust_imm_down(ireg) \ |
| 1336 | arm_access_memory_adjust_imm(SUB, ireg) \ |
| 1337 | |
| 1338 | |
| 1339 | #define arm_access_memory_pre(type, direction) \ |
| 1340 | arm_access_memory_adjust_##type##_##direction(reg_a0) \ |
| 1341 | |
| 1342 | #define arm_access_memory_pre_wb(type, direction) \ |
| 1343 | arm_access_memory_adjust_##type##_##direction(reg_a0); \ |
| 1344 | generate_store_reg(reg_a0, rn) \ |
| 1345 | |
| 1346 | #define arm_access_memory_post(type, direction) \ |
| 1347 | u32 _rn_dest = prepare_store_reg(reg_a1, rn); \ |
| 1348 | if(_rn != reg_a0) \ |
| 1349 | { \ |
| 1350 | generate_load_reg(reg_a0, rn); \ |
| 1351 | } \ |
| 1352 | arm_access_memory_adjust_##type##_##direction(_rn_dest); \ |
| 1353 | complete_store_reg(_rn_dest, rn) \ |
| 1354 | |
| 1355 | |
| 1356 | #define arm_data_trans_reg(adjust_op, direction) \ |
| 1357 | arm_decode_data_trans_reg(); \ |
| 1358 | u32 _rn = prepare_load_reg_pc(reg_a0, rn, 8); \ |
| 1359 | u32 _rm = prepare_load_reg(reg_a1, rm); \ |
| 1360 | arm_access_memory_##adjust_op(reg_sh, direction) \ |
| 1361 | |
| 1362 | #define arm_data_trans_imm(adjust_op, direction) \ |
| 1363 | arm_decode_data_trans_imm(); \ |
| 1364 | u32 _rn = prepare_load_reg_pc(reg_a0, rn, 8); \ |
| 1365 | arm_access_memory_##adjust_op(imm, direction) \ |
| 1366 | |
| 1367 | |
| 1368 | #define arm_data_trans_half_reg(adjust_op, direction) \ |
| 1369 | arm_decode_half_trans_r(); \ |
| 1370 | u32 _rn = prepare_load_reg_pc(reg_a0, rn, 8); \ |
| 1371 | u32 _rm = prepare_load_reg(reg_a1, rm); \ |
| 1372 | arm_access_memory_##adjust_op(reg, direction) \ |
| 1373 | |
| 1374 | #define arm_data_trans_half_imm(adjust_op, direction) \ |
| 1375 | arm_decode_half_trans_of(); \ |
| 1376 | u32 _rn = prepare_load_reg_pc(reg_a0, rn, 8); \ |
| 1377 | arm_access_memory_##adjust_op(imm, direction) \ |
| 1378 | |
| 1379 | |
| 1380 | #define arm_access_memory(access_type, direction, adjust_op, mem_type, \ |
| 1381 | offset_type) \ |
| 1382 | { \ |
| 1383 | arm_data_trans_##offset_type(adjust_op, direction); \ |
| 1384 | arm_access_memory_##access_type(mem_type); \ |
| 1385 | } \ |
| 1386 | |
| 1387 | |
| 1388 | #define word_bit_count(word) \ |
| 1389 | (bit_count[word >> 8] + bit_count[word & 0xFF]) \ |
| 1390 | |
| 1391 | #define sprint_no(access_type, pre_op, post_op, wb) \ |
| 1392 | |
| 1393 | #define sprint_yes(access_type, pre_op, post_op, wb) \ |
| 1394 | printf("sbit on %s %s %s %s\n", #access_type, #pre_op, #post_op, #wb) \ |
| 1395 | |
| 1396 | |
| 1397 | // TODO: Make these use cached registers. Implement iwram_stack_optimize. |
| 1398 | |
| 1399 | #define arm_block_memory_load() \ |
| 1400 | generate_function_call(execute_load_u32); \ |
| 1401 | write32((pc + 8)); \ |
| 1402 | generate_store_reg(reg_rv, i) \ |
| 1403 | |
| 1404 | #define arm_block_memory_store() \ |
| 1405 | generate_load_reg_pc(reg_a1, i, 8); \ |
| 1406 | generate_function_call(execute_store_u32_safe) \ |
| 1407 | |
| 1408 | #define arm_block_memory_final_load() \ |
| 1409 | arm_block_memory_load() \ |
| 1410 | |
| 1411 | #define arm_block_memory_final_store() \ |
| 1412 | generate_load_reg_pc(reg_a1, i, 12); \ |
| 1413 | generate_function_call(execute_store_u32); \ |
| 1414 | write32((pc + 4)) \ |
| 1415 | |
| 1416 | #define arm_block_memory_adjust_pc_store() \ |
| 1417 | |
| 1418 | #define arm_block_memory_adjust_pc_load() \ |
| 1419 | if(reg_list & 0x8000) \ |
| 1420 | { \ |
| 1421 | generate_mov(reg_a0, reg_rv); \ |
| 1422 | generate_indirect_branch_arm(); \ |
| 1423 | } \ |
| 1424 | |
| 1425 | #define arm_block_memory_offset_down_a() \ |
| 1426 | generate_sub_imm(reg_s0, ((word_bit_count(reg_list) * 4) - 4), 0) \ |
| 1427 | |
| 1428 | #define arm_block_memory_offset_down_b() \ |
| 1429 | generate_sub_imm(reg_s0, (word_bit_count(reg_list) * 4), 0) \ |
| 1430 | |
| 1431 | #define arm_block_memory_offset_no() \ |
| 1432 | |
| 1433 | #define arm_block_memory_offset_up() \ |
| 1434 | generate_add_imm(reg_s0, 4, 0) \ |
| 1435 | |
| 1436 | #define arm_block_memory_writeback_down() \ |
| 1437 | generate_load_reg(reg_a0, rn); \ |
| 1438 | generate_sub_imm(reg_a0, (word_bit_count(reg_list) * 4), 0); \ |
| 1439 | generate_store_reg(reg_a0, rn) \ |
| 1440 | |
| 1441 | #define arm_block_memory_writeback_up() \ |
| 1442 | generate_load_reg(reg_a0, rn); \ |
| 1443 | generate_add_imm(reg_a0, (word_bit_count(reg_list) * 4), 0); \ |
| 1444 | generate_store_reg(reg_a0, rn) \ |
| 1445 | |
| 1446 | #define arm_block_memory_writeback_no() |
| 1447 | |
| 1448 | // Only emit writeback if the register is not in the list |
| 1449 | |
| 1450 | #define arm_block_memory_writeback_load(writeback_type) \ |
| 1451 | if(!((reg_list >> rn) & 0x01)) \ |
| 1452 | { \ |
| 1453 | arm_block_memory_writeback_##writeback_type(); \ |
| 1454 | } \ |
| 1455 | |
| 1456 | #define arm_block_memory_writeback_store(writeback_type) \ |
| 1457 | arm_block_memory_writeback_##writeback_type() \ |
| 1458 | |
| 1459 | #define arm_block_memory(access_type, offset_type, writeback_type, s_bit) \ |
| 1460 | { \ |
| 1461 | arm_decode_block_trans(); \ |
| 1462 | u32 offset = 0; \ |
| 1463 | u32 i; \ |
| 1464 | \ |
| 1465 | generate_load_reg(reg_s0, rn); \ |
| 1466 | arm_block_memory_offset_##offset_type(); \ |
| 1467 | arm_block_memory_writeback_##access_type(writeback_type); \ |
| 1468 | ARM_BIC_REG_IMM(0, reg_s0, reg_s0, 0x03, 0); \ |
| 1469 | \ |
| 1470 | for(i = 0; i < 16; i++) \ |
| 1471 | { \ |
| 1472 | if((reg_list >> i) & 0x01) \ |
| 1473 | { \ |
| 1474 | cycle_count++; \ |
| 1475 | generate_add_reg_reg_imm(reg_a0, reg_s0, offset, 0); \ |
| 1476 | if(reg_list & ~((2 << i) - 1)) \ |
| 1477 | { \ |
| 1478 | arm_block_memory_##access_type(); \ |
| 1479 | offset += 4; \ |
| 1480 | } \ |
| 1481 | else \ |
| 1482 | { \ |
| 1483 | arm_block_memory_final_##access_type(); \ |
| 1484 | break; \ |
| 1485 | } \ |
| 1486 | } \ |
| 1487 | } \ |
| 1488 | \ |
| 1489 | arm_block_memory_adjust_pc_##access_type(); \ |
| 1490 | } \ |
| 1491 | |
| 1492 | #define arm_swap(type) \ |
| 1493 | { \ |
| 1494 | arm_decode_swap(); \ |
| 1495 | cycle_count += 3; \ |
| 1496 | generate_load_reg(reg_a0, rn); \ |
| 1497 | generate_function_call(execute_load_##type); \ |
| 1498 | write32((pc + 8)); \ |
| 1499 | generate_mov(reg_s0, reg_rv); \ |
| 1500 | generate_load_reg(reg_a0, rn); \ |
| 1501 | generate_load_reg(reg_a1, rm); \ |
| 1502 | generate_function_call(execute_store_##type); \ |
| 1503 | write32((pc + 4)); \ |
| 1504 | generate_store_reg(reg_s0, rd); \ |
| 1505 | } \ |
| 1506 | |
| 1507 | |
| 1508 | #define thumb_generate_op_reg(name, _rd, _rs, _rn) \ |
| 1509 | u32 __rm = prepare_load_reg(reg_rm, _rn); \ |
| 1510 | generate_op_##name##_reg_immshift(__rd, __rn, __rm, ARMSHIFT_LSL, 0) \ |
| 1511 | |
| 1512 | #define thumb_generate_op_imm(name, _rd, _rs, imm_) \ |
| 1513 | { \ |
| 1514 | u32 imm_ror = 0; \ |
| 1515 | generate_op_##name##_imm(__rd, __rn); \ |
| 1516 | } \ |
| 1517 | |
| 1518 | |
| 1519 | #define thumb_data_proc(type, name, op_type, _rd, _rs, _rn) \ |
| 1520 | { \ |
| 1521 | thumb_decode_##type(); \ |
| 1522 | u32 __rn = prepare_load_reg(reg_rn, _rs); \ |
| 1523 | u32 __rd = prepare_store_reg(reg_rd, _rd); \ |
| 1524 | generate_load_reg(reg_rn, _rs); \ |
| 1525 | thumb_generate_op_##op_type(name, _rd, _rs, _rn); \ |
| 1526 | complete_store_reg(__rd, _rd); \ |
| 1527 | } \ |
| 1528 | |
| 1529 | #define thumb_data_proc_test(type, name, op_type, _rd, _rs) \ |
| 1530 | { \ |
| 1531 | thumb_decode_##type(); \ |
| 1532 | u32 __rn = prepare_load_reg(reg_rn, _rd); \ |
| 1533 | thumb_generate_op_##op_type(name, 0, _rd, _rs); \ |
| 1534 | } \ |
| 1535 | |
| 1536 | #define thumb_data_proc_unary(type, name, op_type, _rd, _rs) \ |
| 1537 | { \ |
| 1538 | thumb_decode_##type(); \ |
| 1539 | u32 __rd = prepare_store_reg(reg_rd, _rd); \ |
| 1540 | thumb_generate_op_##op_type(name, _rd, 0, _rs); \ |
| 1541 | complete_store_reg(__rd, _rd); \ |
| 1542 | } \ |
| 1543 | |
| 1544 | |
| 1545 | #define complete_store_reg_pc_thumb() \ |
| 1546 | if(rd == 15) \ |
| 1547 | { \ |
| 1548 | generate_indirect_branch_cycle_update(thumb); \ |
| 1549 | } \ |
| 1550 | else \ |
| 1551 | { \ |
| 1552 | complete_store_reg(_rd, rd); \ |
| 1553 | } \ |
| 1554 | |
| 1555 | #define thumb_data_proc_hi(name) \ |
| 1556 | { \ |
| 1557 | thumb_decode_hireg_op(); \ |
| 1558 | u32 _rd = prepare_load_reg_pc(reg_rd, rd, 4); \ |
| 1559 | u32 _rs = prepare_load_reg_pc(reg_rn, rs, 4); \ |
| 1560 | generate_op_##name##_reg_immshift(_rd, _rd, _rs, ARMSHIFT_LSL, 0); \ |
| 1561 | complete_store_reg_pc_thumb(); \ |
| 1562 | } \ |
| 1563 | |
| 1564 | #define thumb_data_proc_test_hi(name) \ |
| 1565 | { \ |
| 1566 | thumb_decode_hireg_op(); \ |
| 1567 | u32 _rd = prepare_load_reg_pc(reg_rd, rd, 4); \ |
| 1568 | u32 _rs = prepare_load_reg_pc(reg_rn, rs, 4); \ |
| 1569 | generate_op_##name##_reg_immshift(0, _rd, _rs, ARMSHIFT_LSL, 0); \ |
| 1570 | } \ |
| 1571 | |
| 1572 | #define thumb_data_proc_mov_hi() \ |
| 1573 | { \ |
| 1574 | thumb_decode_hireg_op(); \ |
| 1575 | u32 _rs = prepare_load_reg_pc(reg_rn, rs, 4); \ |
| 1576 | u32 _rd = prepare_store_reg(reg_rd, rd); \ |
| 1577 | ARM_MOV_REG_REG(0, _rd, _rs); \ |
| 1578 | complete_store_reg_pc_thumb(); \ |
| 1579 | } \ |
| 1580 | |
| 1581 | |
| 1582 | |
| 1583 | #define thumb_load_pc(_rd) \ |
| 1584 | { \ |
| 1585 | thumb_decode_imm(); \ |
| 1586 | u32 __rd = prepare_store_reg(reg_rd, _rd); \ |
| 1587 | generate_load_pc(__rd, (((pc & ~2) + 4) + (imm * 4))); \ |
| 1588 | complete_store_reg(__rd, _rd); \ |
| 1589 | } \ |
| 1590 | |
| 1591 | #define thumb_load_sp(_rd) \ |
| 1592 | { \ |
| 1593 | thumb_decode_imm(); \ |
| 1594 | u32 __sp = prepare_load_reg(reg_a0, REG_SP); \ |
| 1595 | u32 __rd = prepare_store_reg(reg_a0, _rd); \ |
| 1596 | ARM_ADD_REG_IMM(0, __rd, __sp, imm, arm_imm_lsl_to_rot(2)); \ |
| 1597 | complete_store_reg(__rd, _rd); \ |
| 1598 | } \ |
| 1599 | |
| 1600 | #define thumb_adjust_sp_up() \ |
| 1601 | ARM_ADD_REG_IMM(0, _sp, _sp, imm, arm_imm_lsl_to_rot(2)) \ |
| 1602 | |
| 1603 | #define thumb_adjust_sp_down() \ |
| 1604 | ARM_SUB_REG_IMM(0, _sp, _sp, imm, arm_imm_lsl_to_rot(2)) \ |
| 1605 | |
| 1606 | #define thumb_adjust_sp(direction) \ |
| 1607 | { \ |
| 1608 | thumb_decode_add_sp(); \ |
| 1609 | u32 _sp = prepare_load_reg(reg_a0, REG_SP); \ |
| 1610 | thumb_adjust_sp_##direction(); \ |
| 1611 | complete_store_reg(_sp, REG_SP); \ |
| 1612 | } \ |
| 1613 | |
| 1614 | #define generate_op_lsl_reg(_rd, _rm, _rs) \ |
| 1615 | generate_op_movs_reg_regshift(_rd, 0, _rm, ARMSHIFT_LSL, _rs) \ |
| 1616 | |
| 1617 | #define generate_op_lsr_reg(_rd, _rm, _rs) \ |
| 1618 | generate_op_movs_reg_regshift(_rd, 0, _rm, ARMSHIFT_LSR, _rs) \ |
| 1619 | |
| 1620 | #define generate_op_asr_reg(_rd, _rm, _rs) \ |
| 1621 | generate_op_movs_reg_regshift(_rd, 0, _rm, ARMSHIFT_ASR, _rs) \ |
| 1622 | |
| 1623 | #define generate_op_ror_reg(_rd, _rm, _rs) \ |
| 1624 | generate_op_movs_reg_regshift(_rd, 0, _rm, ARMSHIFT_ROR, _rs) \ |
| 1625 | |
| 1626 | |
| 1627 | #define generate_op_lsl_imm(_rd, _rm) \ |
| 1628 | generate_op_movs_reg_immshift(_rd, 0, _rm, ARMSHIFT_LSL, imm) \ |
| 1629 | |
| 1630 | #define generate_op_lsr_imm(_rd, _rm) \ |
| 1631 | generate_op_movs_reg_immshift(_rd, 0, _rm, ARMSHIFT_LSR, imm) \ |
| 1632 | |
| 1633 | #define generate_op_asr_imm(_rd, _rm) \ |
| 1634 | generate_op_movs_reg_immshift(_rd, 0, _rm, ARMSHIFT_ASR, imm) \ |
| 1635 | |
| 1636 | #define generate_op_ror_imm(_rd, _rm) \ |
| 1637 | generate_op_movs_reg_immshift(_rd, 0, _rm, ARMSHIFT_ROR, imm) \ |
| 1638 | |
| 1639 | |
| 1640 | #define generate_shift_reg(op_type) \ |
| 1641 | u32 __rm = prepare_load_reg(reg_rd, rd); \ |
| 1642 | u32 __rs = prepare_load_reg(reg_rs, rs); \ |
| 1643 | generate_op_##op_type##_reg(__rd, __rm, __rs) \ |
| 1644 | |
| 1645 | #define generate_shift_imm(op_type) \ |
| 1646 | u32 __rs = prepare_load_reg(reg_rs, rs); \ |
| 1647 | generate_op_##op_type##_imm(__rd, __rs) \ |
| 1648 | |
| 1649 | |
| 1650 | #define thumb_shift(decode_type, op_type, value_type) \ |
| 1651 | { \ |
| 1652 | thumb_decode_##decode_type(); \ |
| 1653 | u32 __rd = prepare_store_reg(reg_rd, rd); \ |
| 1654 | generate_shift_##value_type(op_type); \ |
| 1655 | complete_store_reg(__rd, rd); \ |
| 1656 | } \ |
| 1657 | |
| 1658 | // Operation types: imm, mem_reg, mem_imm |
| 1659 | |
| 1660 | #define thumb_access_memory_load(mem_type, _rd) \ |
| 1661 | cycle_count += 2; \ |
| 1662 | generate_function_call(execute_load_##mem_type); \ |
| 1663 | write32((pc + 4)); \ |
| 1664 | generate_store_reg(reg_rv, _rd) \ |
| 1665 | |
| 1666 | #define thumb_access_memory_store(mem_type, _rd) \ |
| 1667 | cycle_count++; \ |
| 1668 | generate_load_reg(reg_a1, _rd); \ |
| 1669 | generate_function_call(execute_store_##mem_type); \ |
| 1670 | write32((pc + 2)) \ |
| 1671 | |
| 1672 | #define thumb_access_memory_generate_address_pc_relative(offset, _rb, _ro) \ |
| 1673 | generate_load_pc(reg_a0, (offset)) \ |
| 1674 | |
| 1675 | #define thumb_access_memory_generate_address_reg_imm(offset, _rb, _ro) \ |
| 1676 | u32 __rb = prepare_load_reg(reg_a0, _rb); \ |
| 1677 | ARM_ADD_REG_IMM(0, reg_a0, __rb, offset, 0) \ |
| 1678 | |
| 1679 | #define thumb_access_memory_generate_address_reg_imm_sp(offset, _rb, _ro) \ |
| 1680 | u32 __rb = prepare_load_reg(reg_a0, _rb); \ |
| 1681 | ARM_ADD_REG_IMM(0, reg_a0, __rb, offset, arm_imm_lsl_to_rot(2)) \ |
| 1682 | |
| 1683 | #define thumb_access_memory_generate_address_reg_reg(offset, _rb, _ro) \ |
| 1684 | u32 __rb = prepare_load_reg(reg_a0, _rb); \ |
| 1685 | u32 __ro = prepare_load_reg(reg_a1, _ro); \ |
| 1686 | ARM_ADD_REG_REG(0, reg_a0, __rb, __ro) \ |
| 1687 | |
| 1688 | #define thumb_access_memory(access_type, op_type, _rd, _rb, _ro, \ |
| 1689 | address_type, offset, mem_type) \ |
| 1690 | { \ |
| 1691 | thumb_decode_##op_type(); \ |
| 1692 | thumb_access_memory_generate_address_##address_type(offset, _rb, _ro); \ |
| 1693 | thumb_access_memory_##access_type(mem_type, _rd); \ |
| 1694 | } \ |
| 1695 | |
| 1696 | // TODO: Make these use cached registers. Implement iwram_stack_optimize. |
| 1697 | |
| 1698 | #define thumb_block_address_preadjust_up() \ |
| 1699 | generate_add_imm(reg_s0, (bit_count[reg_list] * 4), 0) \ |
| 1700 | |
| 1701 | #define thumb_block_address_preadjust_down() \ |
| 1702 | generate_sub_imm(reg_s0, (bit_count[reg_list] * 4), 0) \ |
| 1703 | |
| 1704 | #define thumb_block_address_preadjust_push_lr() \ |
| 1705 | generate_sub_imm(reg_s0, ((bit_count[reg_list] + 1) * 4), 0) \ |
| 1706 | |
| 1707 | #define thumb_block_address_preadjust_no() \ |
| 1708 | |
| 1709 | #define thumb_block_address_postadjust_no(base_reg) \ |
| 1710 | generate_store_reg(reg_s0, base_reg) \ |
| 1711 | |
| 1712 | #define thumb_block_address_postadjust_up(base_reg) \ |
| 1713 | generate_add_reg_reg_imm(reg_a0, reg_s0, (bit_count[reg_list] * 4), 0); \ |
| 1714 | generate_store_reg(reg_a0, base_reg) \ |
| 1715 | |
| 1716 | #define thumb_block_address_postadjust_down(base_reg) \ |
| 1717 | generate_mov(reg_a0, reg_s0); \ |
| 1718 | generate_sub_imm(reg_a0, (bit_count[reg_list] * 4), 0); \ |
| 1719 | generate_store_reg(reg_a0, base_reg) \ |
| 1720 | |
| 1721 | #define thumb_block_address_postadjust_pop_pc(base_reg) \ |
| 1722 | generate_add_reg_reg_imm(reg_a0, reg_s0, \ |
| 1723 | ((bit_count[reg_list] + 1) * 4), 0); \ |
| 1724 | generate_store_reg(reg_a0, base_reg) \ |
| 1725 | |
| 1726 | #define thumb_block_address_postadjust_push_lr(base_reg) \ |
| 1727 | generate_store_reg(reg_s0, base_reg) \ |
| 1728 | |
| 1729 | #define thumb_block_memory_extra_no() \ |
| 1730 | |
| 1731 | #define thumb_block_memory_extra_up() \ |
| 1732 | |
| 1733 | #define thumb_block_memory_extra_down() \ |
| 1734 | |
| 1735 | #define thumb_block_memory_extra_pop_pc() \ |
| 1736 | generate_add_reg_reg_imm(reg_a0, reg_s0, (bit_count[reg_list] * 4), 0); \ |
| 1737 | generate_function_call(execute_load_u32); \ |
| 1738 | write32((pc + 4)); \ |
| 1739 | generate_mov(reg_a0, reg_rv); \ |
| 1740 | generate_indirect_branch_cycle_update(thumb) \ |
| 1741 | |
| 1742 | #define thumb_block_memory_extra_push_lr(base_reg) \ |
| 1743 | generate_add_reg_reg_imm(reg_a0, reg_s0, (bit_count[reg_list] * 4), 0); \ |
| 1744 | generate_load_reg(reg_a1, REG_LR); \ |
| 1745 | generate_function_call(execute_store_u32_safe) \ |
| 1746 | |
| 1747 | #define thumb_block_memory_load() \ |
| 1748 | generate_function_call(execute_load_u32); \ |
| 1749 | write32((pc + 4)); \ |
| 1750 | generate_store_reg(reg_rv, i) \ |
| 1751 | |
| 1752 | #define thumb_block_memory_store() \ |
| 1753 | generate_load_reg(reg_a1, i); \ |
| 1754 | generate_function_call(execute_store_u32_safe) \ |
| 1755 | |
| 1756 | #define thumb_block_memory_final_load() \ |
| 1757 | thumb_block_memory_load() \ |
| 1758 | |
| 1759 | #define thumb_block_memory_final_store() \ |
| 1760 | generate_load_reg(reg_a1, i); \ |
| 1761 | generate_function_call(execute_store_u32); \ |
| 1762 | write32((pc + 2)) \ |
| 1763 | |
| 1764 | #define thumb_block_memory_final_no(access_type) \ |
| 1765 | thumb_block_memory_final_##access_type() \ |
| 1766 | |
| 1767 | #define thumb_block_memory_final_up(access_type) \ |
| 1768 | thumb_block_memory_final_##access_type() \ |
| 1769 | |
| 1770 | #define thumb_block_memory_final_down(access_type) \ |
| 1771 | thumb_block_memory_final_##access_type() \ |
| 1772 | |
| 1773 | #define thumb_block_memory_final_push_lr(access_type) \ |
| 1774 | thumb_block_memory_##access_type() \ |
| 1775 | |
| 1776 | #define thumb_block_memory_final_pop_pc(access_type) \ |
| 1777 | thumb_block_memory_##access_type() \ |
| 1778 | |
| 1779 | #define thumb_block_memory(access_type, pre_op, post_op, base_reg) \ |
| 1780 | { \ |
| 1781 | thumb_decode_rlist(); \ |
| 1782 | u32 i; \ |
| 1783 | u32 offset = 0; \ |
| 1784 | \ |
| 1785 | generate_load_reg(reg_s0, base_reg); \ |
| 1786 | ARM_BIC_REG_IMM(0, reg_s0, reg_s0, 0x03, 0); \ |
| 1787 | thumb_block_address_preadjust_##pre_op(); \ |
| 1788 | thumb_block_address_postadjust_##post_op(base_reg); \ |
| 1789 | \ |
| 1790 | for(i = 0; i < 8; i++) \ |
| 1791 | { \ |
| 1792 | if((reg_list >> i) & 0x01) \ |
| 1793 | { \ |
| 1794 | cycle_count++; \ |
| 1795 | generate_add_reg_reg_imm(reg_a0, reg_s0, offset, 0); \ |
| 1796 | if(reg_list & ~((2 << i) - 1)) \ |
| 1797 | { \ |
| 1798 | thumb_block_memory_##access_type(); \ |
| 1799 | offset += 4; \ |
| 1800 | } \ |
| 1801 | else \ |
| 1802 | { \ |
| 1803 | thumb_block_memory_final_##post_op(access_type); \ |
| 1804 | break; \ |
| 1805 | } \ |
| 1806 | } \ |
| 1807 | } \ |
| 1808 | \ |
| 1809 | thumb_block_memory_extra_##post_op(); \ |
| 1810 | } \ |
| 1811 | |
| 1812 | #define thumb_conditional_branch(condition) \ |
| 1813 | { \ |
| 1814 | generate_cycle_update(); \ |
| 1815 | generate_load_flags(); \ |
| 1816 | generate_branch_filler(condition_opposite_##condition, backpatch_address); \ |
| 1817 | generate_branch_no_cycle_update( \ |
| 1818 | block_exits[block_exit_position].branch_source, \ |
| 1819 | block_exits[block_exit_position].branch_target, thumb); \ |
| 1820 | generate_branch_patch_conditional(backpatch_address, translation_ptr); \ |
| 1821 | block_exit_position++; \ |
| 1822 | } \ |
| 1823 | |
| 1824 | |
| 1825 | #define arm_conditional_block_header() \ |
| 1826 | generate_cycle_update(); \ |
| 1827 | generate_load_flags(); \ |
| 1828 | /* This will choose the opposite condition */ \ |
| 1829 | condition ^= 0x01; \ |
| 1830 | generate_branch_filler(condition, backpatch_address) \ |
| 1831 | |
| 1832 | #define arm_b() \ |
| 1833 | generate_branch(arm) \ |
| 1834 | |
| 1835 | #define arm_bl() \ |
| 1836 | generate_update_pc((pc + 4)); \ |
| 1837 | generate_store_reg(reg_a0, REG_LR); \ |
| 1838 | generate_branch(arm) \ |
| 1839 | |
| 1840 | #define arm_bx() \ |
| 1841 | arm_decode_branchx(); \ |
| 1842 | generate_load_reg(reg_a0, rn); \ |
| 1843 | generate_indirect_branch_dual(); \ |
| 1844 | |
| 1845 | #define arm_swi() \ |
| 1846 | generate_swi_hle_handler((opcode >> 16) & 0xFF, arm); \ |
| 1847 | generate_function_call(execute_swi_arm); \ |
| 1848 | write32((pc + 4)); \ |
| 1849 | generate_branch(arm) \ |
| 1850 | |
| 1851 | #define thumb_b() \ |
| 1852 | generate_branch(thumb) \ |
| 1853 | |
| 1854 | #define thumb_bl() \ |
| 1855 | generate_update_pc(((pc + 2) | 0x01)); \ |
| 1856 | generate_store_reg(reg_a0, REG_LR); \ |
| 1857 | generate_branch(thumb) \ |
| 1858 | |
| 1859 | #define thumb_blh() \ |
| 1860 | { \ |
| 1861 | thumb_decode_branch(); \ |
| 1862 | generate_update_pc(((pc + 2) | 0x01)); \ |
| 1863 | generate_load_reg(reg_a1, REG_LR); \ |
| 1864 | generate_store_reg(reg_a0, REG_LR); \ |
| 1865 | generate_mov(reg_a0, reg_a1); \ |
| 1866 | generate_add_imm(reg_a0, (offset * 2), 0); \ |
| 1867 | generate_indirect_branch_cycle_update(thumb); \ |
| 1868 | } \ |
| 1869 | |
| 1870 | #define thumb_bx() \ |
| 1871 | { \ |
| 1872 | thumb_decode_hireg_op(); \ |
| 1873 | generate_load_reg_pc(reg_a0, rs, 4); \ |
| 1874 | generate_indirect_branch_cycle_update(dual_thumb); \ |
| 1875 | } \ |
| 1876 | |
| 1877 | #define thumb_swi() \ |
| 1878 | generate_swi_hle_handler(opcode & 0xFF, thumb); \ |
| 1879 | generate_function_call(execute_swi_thumb); \ |
| 1880 | write32((pc + 2)); \ |
| 1881 | /* We're in ARM mode now */ \ |
| 1882 | generate_branch(arm) \ |
| 1883 | |
| 1884 | u8 swi_hle_handle[256] = |
| 1885 | { |
| 1886 | 0x0, // SWI 0: SoftReset |
| 1887 | 0x0, // SWI 1: RegisterRAMReset |
| 1888 | 0x0, // SWI 2: Halt |
| 1889 | 0x0, // SWI 3: Stop/Sleep |
| 1890 | 0x0, // SWI 4: IntrWait |
| 1891 | 0x0, // SWI 5: VBlankIntrWait |
| 1892 | 0x1, // SWI 6: Div |
| 1893 | 0x0, // SWI 7: DivArm |
| 1894 | 0x0, // SWI 8: Sqrt |
| 1895 | 0x0, // SWI 9: ArcTan |
| 1896 | 0x0, // SWI A: ArcTan2 |
| 1897 | 0x0, // SWI B: CpuSet |
| 1898 | 0x0, // SWI C: CpuFastSet |
| 1899 | 0x0, // SWI D: GetBIOSCheckSum |
| 1900 | 0x0, // SWI E: BgAffineSet |
| 1901 | 0x0, // SWI F: ObjAffineSet |
| 1902 | 0x0, // SWI 10: BitUnpack |
| 1903 | 0x0, // SWI 11: LZ77UnCompWram |
| 1904 | 0x0, // SWI 12: LZ77UnCompVram |
| 1905 | 0x0, // SWI 13: HuffUnComp |
| 1906 | 0x0, // SWI 14: RLUnCompWram |
| 1907 | 0x0, // SWI 15: RLUnCompVram |
| 1908 | 0x0, // SWI 16: Diff8bitUnFilterWram |
| 1909 | 0x0, // SWI 17: Diff8bitUnFilterVram |
| 1910 | 0x0, // SWI 18: Diff16bitUnFilter |
| 1911 | 0x0, // SWI 19: SoundBias |
| 1912 | 0x0, // SWI 1A: SoundDriverInit |
| 1913 | 0x0, // SWI 1B: SoundDriverMode |
| 1914 | 0x0, // SWI 1C: SoundDriverMain |
| 1915 | 0x0, // SWI 1D: SoundDriverVSync |
| 1916 | 0x0, // SWI 1E: SoundChannelClear |
| 1917 | 0x0, // SWI 1F: MidiKey2Freq |
| 1918 | 0x0, // SWI 20: SoundWhatever0 |
| 1919 | 0x0, // SWI 21: SoundWhatever1 |
| 1920 | 0x0, // SWI 22: SoundWhatever2 |
| 1921 | 0x0, // SWI 23: SoundWhatever3 |
| 1922 | 0x0, // SWI 24: SoundWhatever4 |
| 1923 | 0x0, // SWI 25: MultiBoot |
| 1924 | 0x0, // SWI 26: HardReset |
| 1925 | 0x0, // SWI 27: CustomHalt |
| 1926 | 0x0, // SWI 28: SoundDriverVSyncOff |
| 1927 | 0x0, // SWI 29: SoundDriverVSyncOn |
| 1928 | 0x0 // SWI 2A: SoundGetJumpList |
| 1929 | }; |
| 1930 | |
| 1931 | void execute_swi_hle_div_arm(); |
| 1932 | void execute_swi_hle_div_thumb(); |
| 1933 | |
| 1934 | void execute_swi_hle_div_c() |
| 1935 | { |
| 1936 | s32 result = (s32)reg[0] / (s32)reg[1]; |
| 1937 | reg[1] = (s32)reg[0] % (s32)reg[1]; |
| 1938 | reg[0] = result; |
| 1939 | |
| 1940 | reg[3] = (result ^ (result >> 31)) - (result >> 31); |
| 1941 | } |
| 1942 | |
| 1943 | #define generate_swi_hle_handler(_swi_number, mode) \ |
| 1944 | { \ |
| 1945 | u32 swi_number = _swi_number; \ |
| 1946 | if(swi_hle_handle[swi_number]) \ |
| 1947 | { \ |
| 1948 | /* Div */ \ |
| 1949 | if(swi_number == 0x06) \ |
| 1950 | { \ |
| 1951 | generate_function_call(execute_swi_hle_div_##mode); \ |
| 1952 | } \ |
| 1953 | break; \ |
| 1954 | } \ |
| 1955 | } \ |
| 1956 | |
| 1957 | #define generate_translation_gate(type) \ |
| 1958 | generate_update_pc(pc); \ |
| 1959 | generate_indirect_branch_no_cycle_update(type) \ |
| 1960 | |
| 1961 | #define generate_step_debug() \ |
| 1962 | generate_function_call(step_debug_arm); \ |
| 1963 | write32(pc) \ |
| 1964 | |
| 1965 | #endif |
| 1966 | |