4 * SH4 emulation core, and parent module for all the SH4 peripheral
7 * Copyright (c) 2005 Nathan Keynes.
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2 of the License, or
12 * (at your option) any later version.
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
20 #define MODULE sh4_module
24 #include "dreamcast.h"
29 #include "sh4/sh4core.h"
30 #include "sh4/sh4mmio.h"
31 #include "sh4/sh4stat.h"
34 #define SH4_CALLTRACE 1
36 #define MAX_INT 0x7FFFFFFF
37 #define MIN_INT 0x80000000
38 #define MAX_INTF 2147483647.0
39 #define MIN_INTF -2147483648.0
41 /********************** SH4 Module Definition ****************************/
43 uint32_t sh4_emulate_run_slice( uint32_t nanosecs )
47 if( sh4_breakpoint_count == 0 ) {
48 for( ; sh4r.slice_cycle < nanosecs; sh4r.slice_cycle += sh4_cpu_period ) {
49 if( SH4_EVENT_PENDING() ) {
50 if( sh4r.event_types & PENDING_EVENT ) {
53 /* Eventq execute may (quite likely) deliver an immediate IRQ */
54 if( sh4r.event_types & PENDING_IRQ ) {
55 sh4_accept_interrupt();
58 if( !sh4_execute_instruction() ) {
63 for( ;sh4r.slice_cycle < nanosecs; sh4r.slice_cycle += sh4_cpu_period ) {
64 if( SH4_EVENT_PENDING() ) {
65 if( sh4r.event_types & PENDING_EVENT ) {
68 /* Eventq execute may (quite likely) deliver an immediate IRQ */
69 if( sh4r.event_types & PENDING_IRQ ) {
70 sh4_accept_interrupt();
74 if( !sh4_execute_instruction() )
76 #ifdef ENABLE_DEBUG_MODE
77 for( i=0; i<sh4_breakpoint_count; i++ ) {
78 if( sh4_breakpoints[i].address == sh4r.pc ) {
82 if( i != sh4_breakpoint_count ) {
83 sh4_core_exit( CORE_EXIT_BREAKPOINT );
89 /* If we aborted early, but the cpu is still technically running,
90 * we're doing a hard abort - cut the timeslice back to what we
93 if( sh4r.slice_cycle != nanosecs && sh4r.sh4_state == SH4_STATE_RUNNING ) {
94 nanosecs = sh4r.slice_cycle;
96 if( sh4r.sh4_state != SH4_STATE_STANDBY ) {
97 TMU_run_slice( nanosecs );
98 SCIF_run_slice( nanosecs );
103 /********************** SH4 emulation core ****************************/
105 #if(SH4_CALLTRACE == 1)
106 #define MAX_CALLSTACK 32
107 static struct call_stack {
109 sh4addr_t target_addr;
110 sh4addr_t stack_pointer;
111 } call_stack[MAX_CALLSTACK];
113 static int call_stack_depth = 0;
114 int sh4_call_trace_on = 0;
116 static inline void trace_call( sh4addr_t source, sh4addr_t dest )
118 if( call_stack_depth < MAX_CALLSTACK ) {
119 call_stack[call_stack_depth].call_addr = source;
120 call_stack[call_stack_depth].target_addr = dest;
121 call_stack[call_stack_depth].stack_pointer = sh4r.r[15];
126 static inline void trace_return( sh4addr_t source, sh4addr_t dest )
128 if( call_stack_depth > 0 ) {
133 void fprint_stack_trace( FILE *f )
135 int i = call_stack_depth -1;
136 if( i >= MAX_CALLSTACK )
137 i = MAX_CALLSTACK - 1;
138 for( ; i >= 0; i-- ) {
139 fprintf( f, "%d. Call from %08X => %08X, SP=%08X\n",
140 (call_stack_depth - i), call_stack[i].call_addr,
141 call_stack[i].target_addr, call_stack[i].stack_pointer );
145 #define TRACE_CALL( source, dest ) trace_call(source, dest)
146 #define TRACE_RETURN( source, dest ) trace_return(source, dest)
148 #define TRACE_CALL( dest, rts )
149 #define TRACE_RETURN( source, dest )
152 static gboolean FASTCALL sh4_raise_slot_exception( int normal_code, int slot_code ) {
153 if( sh4r.in_delay_slot ) {
154 sh4_raise_exception(slot_code);
156 sh4_raise_exception(normal_code);
162 #define CHECKPRIV() if( !IS_SH4_PRIVMODE() ) { return sh4_raise_slot_exception( EXC_ILLEGAL, EXC_SLOT_ILLEGAL ); }
163 #define CHECKRALIGN16(addr) if( (addr)&0x01 ) { sh4_raise_exception( EXC_DATA_ADDR_READ ); return TRUE; }
164 #define CHECKRALIGN32(addr) if( (addr)&0x03 ) { sh4_raise_exception( EXC_DATA_ADDR_READ ); return TRUE; }
165 #define CHECKRALIGN64(addr) if( (addr)&0x07 ) { sh4_raise_exception( EXC_DATA_ADDR_READ ); return TRUE; }
166 #define CHECKWALIGN16(addr) if( (addr)&0x01 ) { sh4_raise_exception( EXC_DATA_ADDR_WRITE ); return TRUE; }
167 #define CHECKWALIGN32(addr) if( (addr)&0x03 ) { sh4_raise_exception( EXC_DATA_ADDR_WRITE ); return TRUE; }
168 #define CHECKWALIGN64(addr) if( (addr)&0x07 ) { sh4_raise_exception( EXC_DATA_ADDR_WRITE ); return TRUE; }
170 #define CHECKFPUEN() if( !IS_FPU_ENABLED() ) { if( ir == 0xFFFD ) { UNDEF(ir); } else { return sh4_raise_slot_exception( EXC_FPU_DISABLED, EXC_SLOT_FPU_DISABLED ); } }
171 #define CHECKDEST(p) if( (p) == 0 ) { ERROR( "%08X: Branch/jump to NULL, CPU halted", sh4r.pc ); sh4_core_exit(CORE_EXIT_HALT); return FALSE; }
172 #define CHECKSLOTILLEGAL() if(sh4r.in_delay_slot) { sh4_raise_exception(EXC_SLOT_ILLEGAL); return TRUE; }
174 #define ADDRSPACE (IS_SH4_PRIVMODE() ? sh4_address_space : sh4_user_address_space)
175 #define SQADDRSPACE (IS_SH4_PRIVMODE() ? storequeue_address_space : storequeue_user_address_space)
177 #ifdef HAVE_FRAME_ADDRESS
178 static FASTCALL __attribute__((noinline)) void *__first_arg(void *a, void *b) { return a; }
179 #define INIT_EXCEPTIONS(label) goto *__first_arg(&&fnstart,&&label); fnstart:
180 #define MEM_READ_BYTE( addr, val ) val = ((mem_read_exc_fn_t)ADDRSPACE[(addr)>>12]->read_byte)((addr), &&except)
181 #define MEM_READ_WORD( addr, val ) val = ((mem_read_exc_fn_t)ADDRSPACE[(addr)>>12]->read_word)((addr), &&except)
182 #define MEM_READ_LONG( addr, val ) val = ((mem_read_exc_fn_t)ADDRSPACE[(addr)>>12]->read_long)((addr), &&except)
183 #define MEM_WRITE_BYTE( addr, val ) ((mem_write_exc_fn_t)ADDRSPACE[(addr)>>12]->write_byte)((addr), (val), &&except)
184 #define MEM_WRITE_WORD( addr, val ) ((mem_write_exc_fn_t)ADDRSPACE[(addr)>>12]->write_word)((addr), (val), &&except)
185 #define MEM_WRITE_LONG( addr, val ) ((mem_write_exc_fn_t)ADDRSPACE[(addr)>>12]->write_long)((addr), (val), &&except)
186 #define MEM_PREFETCH( addr ) ((mem_prefetch_exc_fn_t)ADDRSPACE[(addr)>>12]->prefetch)((addr), &&except)
188 #define INIT_EXCEPTIONS(label)
189 #define MEM_READ_BYTE( addr, val ) val = ADDRSPACE[(addr)>>12]->read_byte(addr)
190 #define MEM_READ_WORD( addr, val ) val = ADDRSPACE[(addr)>>12]->read_word(addr)
191 #define MEM_READ_LONG( addr, val ) val = ADDRSPACE[(addr)>>12]->read_long(addr)
192 #define MEM_WRITE_BYTE( addr, val ) ADDRSPACE[(addr)>>12]->write_byte(addr, val)
193 #define MEM_WRITE_WORD( addr, val ) ADDRSPACE[(addr)>>12]->write_word(addr, val)
194 #define MEM_WRITE_LONG( addr, val ) ADDRSPACE[(addr)>>12]->write_long(addr, val)
195 #define MEM_PREFETCH( addr ) ADDRSPACE[(addr)>>12]->prefetch(addr)
198 #define FP_WIDTH (IS_FPU_DOUBLESIZE() ? 8 : 4)
200 #define MEM_FP_READ( addr, reg ) \
201 if( IS_FPU_DOUBLESIZE() ) { \
202 CHECKRALIGN64(addr); \
204 MEM_READ_LONG( addr, *((uint32_t *)&XF((reg) & 0x0E)) ); \
205 MEM_READ_LONG( addr+4, *((uint32_t *)&XF(reg)) ); \
207 MEM_READ_LONG( addr, *((uint32_t *)&FR(reg)) ); \
208 MEM_READ_LONG( addr+4, *((uint32_t *)&FR((reg)|0x01)) ); \
211 CHECKRALIGN32(addr); \
212 MEM_READ_LONG( addr, *((uint32_t *)&FR(reg)) ); \
214 #define MEM_FP_WRITE( addr, reg ) \
215 if( IS_FPU_DOUBLESIZE() ) { \
216 CHECKWALIGN64(addr); \
218 MEM_WRITE_LONG( addr, *((uint32_t *)&XF((reg)&0x0E)) ); \
219 MEM_WRITE_LONG( addr+4, *((uint32_t *)&XF(reg)) ); \
221 MEM_WRITE_LONG( addr, *((uint32_t *)&FR(reg)) ); \
222 MEM_WRITE_LONG( addr+4, *((uint32_t *)&FR((reg)|0x01)) ); \
225 CHECKWALIGN32(addr); \
226 MEM_WRITE_LONG(addr, *((uint32_t *)&FR((reg))) ); \
233 * Perform instruction-completion following core exit of a partially completed
234 * instruction. NOTE: This is only allowed on memory writes, operation is not
235 * guaranteed in any other case.
237 void sh4_finalize_instruction( void )
242 assert( IS_IN_ICACHE(sh4r.pc) );
243 ir = *(uint16_t *)GET_ICACHE_PTR(sh4r.pc);
246 * Note - we can't take an exit on a control transfer instruction itself,
247 * which means the exit must have happened in the delay slot. So for these
248 * cases, finalize the delay slot instruction, and re-execute the control transfer.
250 * For delay slots which modify the argument used in the branch instruction,
251 * we pretty much just assume that that can't have already happened in an exit case.
257 sh4_finalize_instruction();
259 sh4r.slice_cycle += sh4_cpu_period;
264 sh4_finalize_instruction();
266 sh4r.slice_cycle += sh4_cpu_period;
269 /* Note: PR is already set */
271 sh4_finalize_instruction();
273 sh4r.slice_cycle += sh4_cpu_period;
276 /* Note: PR is already set */
279 sh4_finalize_instruction();
281 sh4r.slice_cycle += sh4_cpu_period;
285 sh4_finalize_instruction();
289 sh4r.slice_cycle += sh4_cpu_period;
293 sh4_finalize_instruction();
297 sh4r.slice_cycle += sh4_cpu_period;
302 sh4_finalize_instruction();
304 sh4r.new_pc = tmp + 2;
305 sh4r.slice_cycle += 2*sh4_cpu_period;
309 /* Note: PR is already set */
312 sh4_finalize_instruction();
314 sh4r.new_pc = tmp + 2;
315 sh4r.slice_cycle += 2*sh4_cpu_period;
320 sh4_finalize_instruction();
322 sh4r.new_pc = sh4r.pr + 2;
323 sh4r.slice_cycle += 2*sh4_cpu_period;
327 /* SR is already set */
329 sh4_finalize_instruction();
331 sh4r.new_pc = sh4r.pr + 2;
332 sh4r.slice_cycle += 2*sh4_cpu_period;
335 MOV.B Rm, @-Rn {: sh4r.r[Rn]--; :}
336 MOV.W Rm, @-Rn {: sh4r.r[Rn] -= 2; :}
337 MOV.L Rm, @-Rn {: sh4r.r[Rn] -= 4; :}
338 MOV.B @Rm+, Rn {: if( Rm != Rn ) { sh4r.r[Rm] ++; } :}
339 MOV.W @Rm+, Rn {: if( Rm != Rn ) { sh4r.r[Rm] += 2; } :}
340 MOV.L @Rm+, Rn {: if( Rm != Rn ) { sh4r.r[Rm] += 4; } :}
343 sh4r.new_pc = sh4r.pc+2;
344 sh4r.slice_cycle += sh4_cpu_period;
350 #define UNDEF(ir) return sh4_raise_slot_exception(EXC_ILLEGAL, EXC_SLOT_ILLEGAL)
351 #define UNIMP(ir) do{ ERROR( "Halted on unimplemented instruction at %08x, opcode = %04x", sh4r.pc, ir ); sh4_core_exit(CORE_EXIT_HALT); return FALSE; }while(0)
354 gboolean sh4_execute_instruction( void )
361 int64_t memtmp; // temporary holder for memory reads
363 INIT_EXCEPTIONS(except)
367 if( pc > 0xFFFFFF00 ) {
369 syscall_invoke( pc );
370 sh4r.in_delay_slot = 0;
371 pc = sh4r.pc = sh4r.pr;
372 sh4r.new_pc = sh4r.pc + 2;
377 #ifdef ENABLE_SH4STATS
378 sh4_stats_add_by_pc(sh4r.pc);
381 /* Read instruction */
382 if( !IS_IN_ICACHE(pc) ) {
383 if( !mmu_update_icache(pc) ) {
384 // Fault - look for the fault handler
385 if( !mmu_update_icache(sh4r.pc) ) {
386 // double fault - halt
387 ERROR( "Double fault - halting" );
388 sh4_core_exit(CORE_EXIT_HALT);
394 assert( IS_IN_ICACHE(pc) );
395 ir = *(uint16_t *)GET_ICACHE_PTR(sh4r.pc);
397 /* FIXME: This is a bit of a hack, but the PC of the delay slot should not
398 * be visible until after the instruction has executed (for exception
401 if( sh4r.in_delay_slot ) {
405 AND Rm, Rn {: sh4r.r[Rn] &= sh4r.r[Rm]; :}
406 AND #imm, R0 {: R0 &= imm; :}
407 AND.B #imm, @(R0, GBR) {: MEM_READ_BYTE(R0+sh4r.gbr, tmp); MEM_WRITE_BYTE( R0 + sh4r.gbr, imm & tmp ); :}
408 NOT Rm, Rn {: sh4r.r[Rn] = ~sh4r.r[Rm]; :}
409 OR Rm, Rn {: sh4r.r[Rn] |= sh4r.r[Rm]; :}
410 OR #imm, R0 {: R0 |= imm; :}
411 OR.B #imm, @(R0, GBR) {: MEM_READ_BYTE(R0+sh4r.gbr, tmp); MEM_WRITE_BYTE( R0 + sh4r.gbr, imm | tmp ); :}
413 MEM_READ_BYTE( sh4r.r[Rn], tmp );
414 sh4r.t = ( tmp == 0 ? 1 : 0 );
415 MEM_WRITE_BYTE( sh4r.r[Rn], tmp | 0x80 );
417 TST Rm, Rn {: sh4r.t = (sh4r.r[Rn]&sh4r.r[Rm] ? 0 : 1); :}
418 TST #imm, R0 {: sh4r.t = (R0 & imm ? 0 : 1); :}
419 TST.B #imm, @(R0, GBR) {: MEM_READ_BYTE(R0+sh4r.gbr, tmp); sh4r.t = ( tmp & imm ? 0 : 1 ); :}
420 XOR Rm, Rn {: sh4r.r[Rn] ^= sh4r.r[Rm]; :}
421 XOR #imm, R0 {: R0 ^= imm; :}
422 XOR.B #imm, @(R0, GBR) {: MEM_READ_BYTE(R0+sh4r.gbr, tmp); MEM_WRITE_BYTE( R0 + sh4r.gbr, imm ^ tmp ); :}
423 XTRCT Rm, Rn {: sh4r.r[Rn] = (sh4r.r[Rn]>>16) | (sh4r.r[Rm]<<16); :}
426 sh4r.t = sh4r.r[Rn] >> 31;
428 sh4r.r[Rn] |= sh4r.t;
431 sh4r.t = sh4r.r[Rn] & 0x00000001;
433 sh4r.r[Rn] |= (sh4r.t << 31);
436 tmp = sh4r.r[Rn] >> 31;
438 sh4r.r[Rn] |= sh4r.t;
442 tmp = sh4r.r[Rn] & 0x00000001;
444 sh4r.r[Rn] |= (sh4r.t << 31 );
449 if( (tmp & 0x80000000) == 0 ) sh4r.r[Rn] <<= (tmp&0x1f);
450 else if( (tmp & 0x1F) == 0 )
451 sh4r.r[Rn] = ((int32_t)sh4r.r[Rn]) >> 31;
453 sh4r.r[Rn] = ((int32_t)sh4r.r[Rn]) >> (((~sh4r.r[Rm]) & 0x1F)+1);
457 if( (tmp & 0x80000000) == 0 ) sh4r.r[Rn] <<= (tmp&0x1f);
458 else if( (tmp & 0x1F) == 0 ) sh4r.r[Rn] = 0;
459 else sh4r.r[Rn] >>= (((~tmp) & 0x1F)+1);
462 sh4r.t = sh4r.r[Rn] >> 31;
466 sh4r.t = sh4r.r[Rn] & 0x00000001;
467 sh4r.r[Rn] = ((int32_t)sh4r.r[Rn]) >> 1;
469 SHLL Rn {: sh4r.t = sh4r.r[Rn] >> 31; sh4r.r[Rn] <<= 1; :}
470 SHLR Rn {: sh4r.t = sh4r.r[Rn] & 0x00000001; sh4r.r[Rn] >>= 1; :}
471 SHLL2 Rn {: sh4r.r[Rn] <<= 2; :}
472 SHLR2 Rn {: sh4r.r[Rn] >>= 2; :}
473 SHLL8 Rn {: sh4r.r[Rn] <<= 8; :}
474 SHLR8 Rn {: sh4r.r[Rn] >>= 8; :}
475 SHLL16 Rn {: sh4r.r[Rn] <<= 16; :}
476 SHLR16 Rn {: sh4r.r[Rn] >>= 16; :}
478 EXTU.B Rm, Rn {: sh4r.r[Rn] = sh4r.r[Rm]&0x000000FF; :}
479 EXTU.W Rm, Rn {: sh4r.r[Rn] = sh4r.r[Rm]&0x0000FFFF; :}
480 EXTS.B Rm, Rn {: sh4r.r[Rn] = SIGNEXT8( sh4r.r[Rm]&0x000000FF ); :}
481 EXTS.W Rm, Rn {: sh4r.r[Rn] = SIGNEXT16( sh4r.r[Rm]&0x0000FFFF ); :}
482 SWAP.B Rm, Rn {: sh4r.r[Rn] = (sh4r.r[Rm]&0xFFFF0000) | ((sh4r.r[Rm]&0x0000FF00)>>8) | ((sh4r.r[Rm]&0x000000FF)<<8); :}
483 SWAP.W Rm, Rn {: sh4r.r[Rn] = (sh4r.r[Rm]>>16) | (sh4r.r[Rm]<<16); :}
485 CLRT {: sh4r.t = 0; :}
486 SETT {: sh4r.t = 1; :}
487 CLRMAC {: sh4r.mac = 0; :}
488 LDTLB {: MMU_ldtlb(); :}
489 CLRS {: sh4r.s = 0; :}
490 SETS {: sh4r.s = 1; :}
491 MOVT Rn {: sh4r.r[Rn] = sh4r.t; :}
495 MEM_PREFETCH(sh4r.r[Rn]);
503 MEM_WRITE_LONG( tmp, R0 );
505 MOV.B Rm, @(R0, Rn) {: MEM_WRITE_BYTE( R0 + sh4r.r[Rn], sh4r.r[Rm] ); :}
506 MOV.W Rm, @(R0, Rn) {:
507 CHECKWALIGN16( R0 + sh4r.r[Rn] );
508 MEM_WRITE_WORD( R0 + sh4r.r[Rn], sh4r.r[Rm] );
510 MOV.L Rm, @(R0, Rn) {:
511 CHECKWALIGN32( R0 + sh4r.r[Rn] );
512 MEM_WRITE_LONG( R0 + sh4r.r[Rn], sh4r.r[Rm] );
514 MOV.B @(R0, Rm), Rn {: MEM_READ_BYTE( R0 + sh4r.r[Rm], sh4r.r[Rn] ); :}
515 MOV.W @(R0, Rm), Rn {: CHECKRALIGN16( R0 + sh4r.r[Rm] );
516 MEM_READ_WORD( R0 + sh4r.r[Rm], sh4r.r[Rn] );
518 MOV.L @(R0, Rm), Rn {: CHECKRALIGN32( R0 + sh4r.r[Rm] );
519 MEM_READ_LONG( R0 + sh4r.r[Rm], sh4r.r[Rn] );
521 MOV.L Rm, @(disp, Rn) {:
522 tmp = sh4r.r[Rn] + disp;
523 CHECKWALIGN32( tmp );
524 MEM_WRITE_LONG( tmp, sh4r.r[Rm] );
526 MOV.B Rm, @Rn {: MEM_WRITE_BYTE( sh4r.r[Rn], sh4r.r[Rm] ); :}
527 MOV.W Rm, @Rn {: CHECKWALIGN16( sh4r.r[Rn] ); MEM_WRITE_WORD( sh4r.r[Rn], sh4r.r[Rm] ); :}
528 MOV.L Rm, @Rn {: CHECKWALIGN32( sh4r.r[Rn] ); MEM_WRITE_LONG( sh4r.r[Rn], sh4r.r[Rm] ); :}
529 MOV.B Rm, @-Rn {: MEM_WRITE_BYTE( sh4r.r[Rn]-1, sh4r.r[Rm] ); sh4r.r[Rn]--; :}
530 MOV.W Rm, @-Rn {: CHECKWALIGN16( sh4r.r[Rn] ); MEM_WRITE_WORD( sh4r.r[Rn]-2, sh4r.r[Rm] ); sh4r.r[Rn] -= 2; :}
531 MOV.L Rm, @-Rn {: CHECKWALIGN32( sh4r.r[Rn] ); MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.r[Rm] ); sh4r.r[Rn] -= 4; :}
532 MOV.L @(disp, Rm), Rn {:
533 tmp = sh4r.r[Rm] + disp;
534 CHECKRALIGN32( tmp );
535 MEM_READ_LONG( tmp, sh4r.r[Rn] );
537 MOV.B @Rm, Rn {: MEM_READ_BYTE( sh4r.r[Rm], sh4r.r[Rn] ); :}
538 MOV.W @Rm, Rn {: CHECKRALIGN16( sh4r.r[Rm] ); MEM_READ_WORD( sh4r.r[Rm], sh4r.r[Rn] ); :}
539 MOV.L @Rm, Rn {: CHECKRALIGN32( sh4r.r[Rm] ); MEM_READ_LONG( sh4r.r[Rm], sh4r.r[Rn] ); :}
540 MOV Rm, Rn {: sh4r.r[Rn] = sh4r.r[Rm]; :}
541 MOV.B @Rm+, Rn {: MEM_READ_BYTE( sh4r.r[Rm], sh4r.r[Rn] ); if( Rm != Rn ) { sh4r.r[Rm] ++; } :}
542 MOV.W @Rm+, Rn {: CHECKRALIGN16( sh4r.r[Rm] ); MEM_READ_WORD( sh4r.r[Rm], sh4r.r[Rn] ); if( Rm != Rn ) { sh4r.r[Rm] += 2; } :}
543 MOV.L @Rm+, Rn {: CHECKRALIGN32( sh4r.r[Rm] ); MEM_READ_LONG( sh4r.r[Rm], sh4r.r[Rn] ); if( Rm != Rn ) { sh4r.r[Rm] += 4; } :}
544 MOV.L @(disp, PC), Rn {:
546 tmp = (pc&0xFFFFFFFC) + disp + 4;
547 MEM_READ_LONG( tmp, sh4r.r[Rn] );
549 MOV.B R0, @(disp, GBR) {: MEM_WRITE_BYTE( sh4r.gbr + disp, R0 ); :}
550 MOV.W R0, @(disp, GBR) {:
551 tmp = sh4r.gbr + disp;
552 CHECKWALIGN16( tmp );
553 MEM_WRITE_WORD( tmp, R0 );
555 MOV.L R0, @(disp, GBR) {:
556 tmp = sh4r.gbr + disp;
557 CHECKWALIGN32( tmp );
558 MEM_WRITE_LONG( tmp, R0 );
560 MOV.B @(disp, GBR), R0 {: MEM_READ_BYTE( sh4r.gbr + disp, R0 ); :}
561 MOV.W @(disp, GBR), R0 {:
562 tmp = sh4r.gbr + disp;
563 CHECKRALIGN16( tmp );
564 MEM_READ_WORD( tmp, R0 );
566 MOV.L @(disp, GBR), R0 {:
567 tmp = sh4r.gbr + disp;
568 CHECKRALIGN32( tmp );
569 MEM_READ_LONG( tmp, R0 );
571 MOV.B R0, @(disp, Rn) {: MEM_WRITE_BYTE( sh4r.r[Rn] + disp, R0 ); :}
572 MOV.W R0, @(disp, Rn) {:
573 tmp = sh4r.r[Rn] + disp;
574 CHECKWALIGN16( tmp );
575 MEM_WRITE_WORD( tmp, R0 );
577 MOV.B @(disp, Rm), R0 {: MEM_READ_BYTE( sh4r.r[Rm] + disp, R0 ); :}
578 MOV.W @(disp, Rm), R0 {:
579 tmp = sh4r.r[Rm] + disp;
580 CHECKRALIGN16( tmp );
581 MEM_READ_WORD( tmp, R0 );
583 MOV.W @(disp, PC), Rn {:
586 MEM_READ_WORD( tmp, sh4r.r[Rn] );
588 MOVA @(disp, PC), R0 {:
590 R0 = (pc&0xFFFFFFFC) + disp + 4;
592 MOV #imm, Rn {: sh4r.r[Rn] = imm; :}
594 FMOV @(R0, Rm), FRn {: MEM_FP_READ( sh4r.r[Rm] + R0, FRn ); :}
595 FMOV FRm, @(R0, Rn) {: MEM_FP_WRITE( sh4r.r[Rn] + R0, FRm ); :}
596 FMOV @Rm, FRn {: MEM_FP_READ( sh4r.r[Rm], FRn ); :}
597 FMOV @Rm+, FRn {: MEM_FP_READ( sh4r.r[Rm], FRn ); sh4r.r[Rm] += FP_WIDTH; :}
598 FMOV FRm, @Rn {: MEM_FP_WRITE( sh4r.r[Rn], FRm ); :}
599 FMOV FRm, @-Rn {: MEM_FP_WRITE( sh4r.r[Rn] - FP_WIDTH, FRm ); sh4r.r[Rn] -= FP_WIDTH; :}
601 if( IS_FPU_DOUBLESIZE() )
607 CMP/EQ #imm, R0 {: sh4r.t = ( R0 == imm ? 1 : 0 ); :}
608 CMP/EQ Rm, Rn {: sh4r.t = ( sh4r.r[Rm] == sh4r.r[Rn] ? 1 : 0 ); :}
609 CMP/GE Rm, Rn {: sh4r.t = ( ((int32_t)sh4r.r[Rn]) >= ((int32_t)sh4r.r[Rm]) ? 1 : 0 ); :}
610 CMP/GT Rm, Rn {: sh4r.t = ( ((int32_t)sh4r.r[Rn]) > ((int32_t)sh4r.r[Rm]) ? 1 : 0 ); :}
611 CMP/HI Rm, Rn {: sh4r.t = ( sh4r.r[Rn] > sh4r.r[Rm] ? 1 : 0 ); :}
612 CMP/HS Rm, Rn {: sh4r.t = ( sh4r.r[Rn] >= sh4r.r[Rm] ? 1 : 0 ); :}
613 CMP/PL Rn {: sh4r.t = ( ((int32_t)sh4r.r[Rn]) > 0 ? 1 : 0 ); :}
614 CMP/PZ Rn {: sh4r.t = ( ((int32_t)sh4r.r[Rn]) >= 0 ? 1 : 0 ); :}
616 /* set T = 1 if any byte in RM & RN is the same */
617 tmp = sh4r.r[Rm] ^ sh4r.r[Rn];
618 sh4r.t = ((tmp&0x000000FF)==0 || (tmp&0x0000FF00)==0 ||
619 (tmp&0x00FF0000)==0 || (tmp&0xFF000000)==0)?1:0;
622 ADD Rm, Rn {: sh4r.r[Rn] += sh4r.r[Rm]; :}
623 ADD #imm, Rn {: sh4r.r[Rn] += imm; :}
626 sh4r.r[Rn] += sh4r.r[Rm] + sh4r.t;
627 sh4r.t = ( sh4r.r[Rn] < tmp || (sh4r.r[Rn] == tmp && sh4r.t != 0) ? 1 : 0 );
630 tmp = sh4r.r[Rn] + sh4r.r[Rm];
631 sh4r.t = ( (sh4r.r[Rn]>>31) == (sh4r.r[Rm]>>31) && ((sh4r.r[Rn]>>31) != (tmp>>31)) );
634 DIV0U {: sh4r.m = sh4r.q = sh4r.t = 0; :}
636 sh4r.q = sh4r.r[Rn]>>31;
637 sh4r.m = sh4r.r[Rm]>>31;
638 sh4r.t = sh4r.q ^ sh4r.m;
641 /* This is derived from the sh4 manual with some simplifications */
642 uint32_t tmp0, tmp1, tmp2, dir;
644 dir = sh4r.q ^ sh4r.m;
645 sh4r.q = (sh4r.r[Rn] >> 31);
647 sh4r.r[Rn] = (sh4r.r[Rn] << 1) | sh4r.t;
651 tmp1 = (sh4r.r[Rn]<tmp0 ? 1 : 0 );
654 tmp1 = (sh4r.r[Rn]>tmp0 ? 1 : 0 );
656 sh4r.q ^= sh4r.m ^ tmp1;
657 sh4r.t = ( sh4r.q == sh4r.m ? 1 : 0 );
659 DMULS.L Rm, Rn {: sh4r.mac = SIGNEXT32(sh4r.r[Rm]) * SIGNEXT32(sh4r.r[Rn]); :}
660 DMULU.L Rm, Rn {: sh4r.mac = ((uint64_t)sh4r.r[Rm]) * ((uint64_t)sh4r.r[Rn]); :}
663 sh4r.t = ( sh4r.r[Rn] == 0 ? 1 : 0 );
668 CHECKRALIGN16(sh4r.r[Rn]);
669 MEM_READ_WORD( sh4r.r[Rn], tmp );
670 stmp = SIGNEXT16(tmp);
671 MEM_READ_WORD( sh4r.r[Rn]+2, tmp );
672 stmp *= SIGNEXT16(tmp);
675 CHECKRALIGN16( sh4r.r[Rn] );
676 CHECKRALIGN16( sh4r.r[Rm] );
677 MEM_READ_WORD(sh4r.r[Rn], tmp);
678 stmp = SIGNEXT16(tmp);
679 MEM_READ_WORD(sh4r.r[Rm], tmp);
680 stmp = stmp * SIGNEXT16(tmp);
685 int64_t tmpl = (int64_t)((int32_t)sh4r.mac) + (int64_t)stmp;
686 if( tmpl > (int64_t)0x000000007FFFFFFFLL ) {
687 sh4r.mac = 0x000000017FFFFFFFLL;
688 } else if( tmpl < (int64_t)0xFFFFFFFF80000000LL ) {
689 sh4r.mac = 0x0000000180000000LL;
691 sh4r.mac = (sh4r.mac & 0xFFFFFFFF00000000LL) |
692 ((uint32_t)(sh4r.mac + stmp));
695 sh4r.mac += SIGNEXT32(stmp);
701 CHECKRALIGN32( sh4r.r[Rn] );
702 MEM_READ_LONG(sh4r.r[Rn], tmp);
703 tmpl = SIGNEXT32(tmp);
704 MEM_READ_LONG(sh4r.r[Rn]+4, tmp);
705 tmpl = tmpl * SIGNEXT32(tmp) + sh4r.mac;
708 CHECKRALIGN32( sh4r.r[Rm] );
709 CHECKRALIGN32( sh4r.r[Rn] );
710 MEM_READ_LONG(sh4r.r[Rn], tmp);
711 tmpl = SIGNEXT32(tmp);
712 MEM_READ_LONG(sh4r.r[Rm], tmp);
713 tmpl = tmpl * SIGNEXT32(tmp) + sh4r.mac;
718 /* 48-bit Saturation. Yuch */
719 if( tmpl < (int64_t)0xFFFF800000000000LL )
720 tmpl = 0xFFFF800000000000LL;
721 else if( tmpl > (int64_t)0x00007FFFFFFFFFFFLL )
722 tmpl = 0x00007FFFFFFFFFFFLL;
726 MUL.L Rm, Rn {: sh4r.mac = (sh4r.mac&0xFFFFFFFF00000000LL) |
727 (sh4r.r[Rm] * sh4r.r[Rn]); :}
729 sh4r.mac = (sh4r.mac&0xFFFFFFFF00000000LL) |
730 (uint32_t)((sh4r.r[Rm]&0xFFFF) * (sh4r.r[Rn]&0xFFFF));
733 sh4r.mac = (sh4r.mac&0xFFFFFFFF00000000LL) |
734 (uint32_t)(SIGNEXT32(sh4r.r[Rm]&0xFFFF) * SIGNEXT32(sh4r.r[Rn]&0xFFFF));
737 tmp = 0 - sh4r.r[Rm];
738 sh4r.r[Rn] = tmp - sh4r.t;
739 sh4r.t = ( 0<tmp || tmp<sh4r.r[Rn] ? 1 : 0 );
741 NEG Rm, Rn {: sh4r.r[Rn] = 0 - sh4r.r[Rm]; :}
742 SUB Rm, Rn {: sh4r.r[Rn] -= sh4r.r[Rm]; :}
745 sh4r.r[Rn] = sh4r.r[Rn] - sh4r.r[Rm] - sh4r.t;
746 sh4r.t = (sh4r.r[Rn] > tmp || (sh4r.r[Rn] == tmp && sh4r.t == 1));
751 CHECKDEST( pc + 4 + sh4r.r[Rn] );
752 sh4r.in_delay_slot = 1;
753 sh4r.pc = sh4r.new_pc;
754 sh4r.new_pc = pc + 4 + sh4r.r[Rn];
759 CHECKDEST( pc + 4 + sh4r.r[Rn] );
760 sh4r.in_delay_slot = 1;
761 sh4r.pr = sh4r.pc + 4;
762 sh4r.pc = sh4r.new_pc;
763 sh4r.new_pc = pc + 4 + sh4r.r[Rn];
764 TRACE_CALL( pc, sh4r.new_pc );
770 CHECKDEST( sh4r.pc + disp + 4 )
772 sh4r.new_pc = sh4r.pc + 2;
779 CHECKDEST( sh4r.pc + disp + 4 )
781 sh4r.new_pc = sh4r.pc + 2;
788 CHECKDEST( sh4r.pc + disp + 4 )
789 sh4r.in_delay_slot = 1;
790 sh4r.pc = sh4r.new_pc;
791 sh4r.new_pc = pc + disp + 4;
792 sh4r.in_delay_slot = 1;
799 CHECKDEST( sh4r.pc + disp + 4 )
800 sh4r.in_delay_slot = 1;
801 sh4r.pc = sh4r.new_pc;
802 sh4r.new_pc = pc + disp + 4;
808 CHECKDEST( sh4r.pc + disp + 4 );
809 sh4r.in_delay_slot = 1;
810 sh4r.pc = sh4r.new_pc;
811 sh4r.new_pc = pc + 4 + disp;
815 CHECKDEST( sh4r.pc + disp + 4 );
817 sh4r.in_delay_slot = 1;
819 sh4r.pc = sh4r.new_pc;
820 sh4r.new_pc = pc + 4 + disp;
821 TRACE_CALL( pc, sh4r.new_pc );
827 sh4_raise_trap( imm );
832 CHECKDEST( sh4r.pr );
833 sh4r.in_delay_slot = 1;
834 sh4r.pc = sh4r.new_pc;
835 sh4r.new_pc = sh4r.pr;
836 TRACE_RETURN( pc, sh4r.new_pc );
840 if( MMIO_READ( CPG, STBCR ) & 0x80 ) {
841 sh4r.sh4_state = SH4_STATE_STANDBY;
843 sh4r.sh4_state = SH4_STATE_SLEEP;
845 return FALSE; /* Halt CPU */
849 CHECKDEST( sh4r.spc );
851 sh4r.in_delay_slot = 1;
852 sh4r.pc = sh4r.new_pc;
853 sh4r.new_pc = sh4r.spc;
854 sh4_write_sr( sh4r.ssr );
858 CHECKDEST( sh4r.r[Rn] );
860 sh4r.in_delay_slot = 1;
861 sh4r.pc = sh4r.new_pc;
862 sh4r.new_pc = sh4r.r[Rn];
866 CHECKDEST( sh4r.r[Rn] );
868 sh4r.in_delay_slot = 1;
869 sh4r.pc = sh4r.new_pc;
870 sh4r.new_pc = sh4r.r[Rn];
872 TRACE_CALL( pc, sh4r.new_pc );
875 STS MACH, Rn {: sh4r.r[Rn] = (sh4r.mac>>32); :}
877 CHECKWALIGN32( sh4r.r[Rn] );
878 MEM_WRITE_LONG( sh4r.r[Rn]-4, (sh4r.mac>>32) );
883 CHECKWALIGN32( sh4r.r[Rn] );
884 MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4_read_sr() );
888 CHECKRALIGN32( sh4r.r[Rm] );
889 MEM_READ_LONG(sh4r.r[Rm], tmp);
890 sh4r.mac = (sh4r.mac & 0x00000000FFFFFFFF) |
891 (((uint64_t)tmp)<<32);
897 CHECKWALIGN32( sh4r.r[Rm] );
898 MEM_READ_LONG(sh4r.r[Rm], tmp);
903 sh4r.mac = (sh4r.mac & 0x00000000FFFFFFFF) |
904 (((uint64_t)sh4r.r[Rm])<<32);
909 sh4_write_sr( sh4r.r[Rm] );
913 sh4r.sgr = sh4r.r[Rm];
917 CHECKRALIGN32( sh4r.r[Rm] );
918 MEM_READ_LONG(sh4r.r[Rm], sh4r.sgr);
921 STS MACL, Rn {: sh4r.r[Rn] = (uint32_t)sh4r.mac; :}
923 CHECKWALIGN32( sh4r.r[Rn] );
924 MEM_WRITE_LONG( sh4r.r[Rn]-4, (uint32_t)sh4r.mac );
928 CHECKWALIGN32( sh4r.r[Rn] );
929 MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.gbr );
933 CHECKRALIGN32( sh4r.r[Rm] );
934 MEM_READ_LONG(sh4r.r[Rm], tmp);
935 sh4r.mac = (sh4r.mac & 0xFFFFFFFF00000000LL) |
936 (uint64_t)((uint32_t)tmp);
940 CHECKRALIGN32( sh4r.r[Rm] );
941 MEM_READ_LONG(sh4r.r[Rm], sh4r.gbr);
945 sh4r.mac = (sh4r.mac & 0xFFFFFFFF00000000LL) |
946 (uint64_t)((uint32_t)(sh4r.r[Rm]));
948 LDC Rm, GBR {: sh4r.gbr = sh4r.r[Rm]; :}
949 STS PR, Rn {: sh4r.r[Rn] = sh4r.pr; :}
951 CHECKWALIGN32( sh4r.r[Rn] );
952 MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.pr );
957 CHECKWALIGN32( sh4r.r[Rn] );
958 MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.vbr );
962 CHECKRALIGN32( sh4r.r[Rm] );
963 MEM_READ_LONG( sh4r.r[Rm], sh4r.pr );
968 CHECKRALIGN32( sh4r.r[Rm] );
969 MEM_READ_LONG(sh4r.r[Rm], sh4r.vbr);
972 LDS Rm, PR {: sh4r.pr = sh4r.r[Rm]; :}
975 sh4r.vbr = sh4r.r[Rm];
979 sh4r.r[Rn] = sh4r.sgr;
983 CHECKWALIGN32( sh4r.r[Rn] );
984 MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.sgr );
989 CHECKWALIGN32( sh4r.r[Rn] );
990 MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.ssr );
995 CHECKRALIGN32( sh4r.r[Rm] );
996 MEM_READ_LONG(sh4r.r[Rm], sh4r.ssr);
1001 sh4r.ssr = sh4r.r[Rm];
1005 CHECKWALIGN32( sh4r.r[Rn] );
1006 MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.spc );
1011 CHECKRALIGN32( sh4r.r[Rm] );
1012 MEM_READ_LONG(sh4r.r[Rm], sh4r.spc);
1017 sh4r.spc = sh4r.r[Rm];
1025 CHECKWALIGN32( sh4r.r[Rn] );
1026 MEM_WRITE_LONG( sh4r.r[Rn]-4, FPULi );
1031 CHECKRALIGN32( sh4r.r[Rm] );
1032 MEM_READ_LONG(sh4r.r[Rm], FPULi);
1041 sh4r.r[Rn] = sh4r.fpscr;
1043 STS.L FPSCR, @-Rn {:
1045 CHECKWALIGN32( sh4r.r[Rn] );
1046 MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.fpscr );
1049 LDS.L @Rm+, FPSCR {:
1051 CHECKRALIGN32( sh4r.r[Rm] );
1052 MEM_READ_LONG(sh4r.r[Rm], tmp);
1054 sh4_write_fpscr( tmp );
1058 sh4_write_fpscr( sh4r.r[Rm] );
1060 STC DBR, Rn {: CHECKPRIV(); sh4r.r[Rn] = sh4r.dbr; :}
1063 CHECKWALIGN32( sh4r.r[Rn] );
1064 MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.dbr );
1069 CHECKRALIGN32( sh4r.r[Rm] );
1070 MEM_READ_LONG(sh4r.r[Rm], sh4r.dbr);
1075 sh4r.dbr = sh4r.r[Rm];
1077 STC.L Rm_BANK, @-Rn {:
1079 CHECKWALIGN32( sh4r.r[Rn] );
1080 MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.r_bank[Rm_BANK] );
1083 LDC.L @Rm+, Rn_BANK {:
1085 CHECKRALIGN32( sh4r.r[Rm] );
1086 MEM_READ_LONG( sh4r.r[Rm], sh4r.r_bank[Rn_BANK] );
1091 sh4r.r_bank[Rn_BANK] = sh4r.r[Rm];
1095 sh4r.r[Rn] = sh4_read_sr();
1098 sh4r.r[Rn] = sh4r.gbr;
1102 sh4r.r[Rn] = sh4r.vbr;
1106 sh4r.r[Rn] = sh4r.ssr;
1110 sh4r.r[Rn] = sh4r.spc;
1114 sh4r.r[Rn] = sh4r.r_bank[Rm_BANK];
1119 if( IS_FPU_DOUBLEPREC() ) {
1127 if( IS_FPU_DOUBLEPREC() ) {
1136 if( IS_FPU_DOUBLEPREC() ) {
1145 if( IS_FPU_DOUBLEPREC() ) {
1154 if( IS_FPU_DOUBLEPREC() ) {
1155 sh4r.t = ( DR(FRn) == DR(FRm) ? 1 : 0 );
1157 sh4r.t = ( FR(FRn) == FR(FRm) ? 1 : 0 );
1163 if( IS_FPU_DOUBLEPREC() ) {
1164 sh4r.t = ( DR(FRn) > DR(FRm) ? 1 : 0 );
1166 sh4r.t = ( FR(FRn) > FR(FRm) ? 1 : 0 );
1170 FSTS FPUL, FRn {: CHECKFPUEN(); FR(FRn) = FPULf; :}
1171 FLDS FRm, FPUL {: CHECKFPUEN(); FPULf = FR(FRm); :}
1174 if( IS_FPU_DOUBLEPREC() ) {
1175 if( FRn&1 ) { // No, really...
1176 dtmp = (double)FPULi;
1177 FR(FRn) = *(((float *)&dtmp)+1);
1179 DRF(FRn>>1) = (double)FPULi;
1182 FR(FRn) = (float)FPULi;
1187 if( IS_FPU_DOUBLEPREC() ) {
1190 *(((float *)&dtmp)+1) = FR(FRm);
1194 if( dtmp >= MAX_INTF )
1196 else if( dtmp <= MIN_INTF )
1199 FPULi = (int32_t)dtmp;
1202 if( ftmp >= MAX_INTF )
1204 else if( ftmp <= MIN_INTF )
1207 FPULi = (int32_t)ftmp;
1212 if( IS_FPU_DOUBLEPREC() ) {
1220 if( IS_FPU_DOUBLEPREC() ) {
1221 DR(FRn) = fabs(DR(FRn));
1223 FR(FRn) = fabsf(FR(FRn));
1228 if( IS_FPU_DOUBLEPREC() ) {
1229 DR(FRn) = sqrt(DR(FRn));
1231 FR(FRn) = sqrtf(FR(FRn));
1236 if( IS_FPU_DOUBLEPREC() ) {
1244 if( IS_FPU_DOUBLEPREC() ) {
1250 FMAC FR0, FRm, FRn {:
1252 if( IS_FPU_DOUBLEPREC() ) {
1253 DR(FRn) += DR(FRm)*DR(0);
1255 FR(FRn) += FR(FRm)*FR(0);
1260 sh4r.fpscr ^= FPSCR_FR;
1261 sh4_switch_fr_banks();
1263 FSCHG {: CHECKFPUEN(); sh4r.fpscr ^= FPSCR_SZ; :}
1266 if( IS_FPU_DOUBLEPREC() && !IS_FPU_DOUBLESIZE() ) {
1267 DR(FRn) = (double)FPULf;
1272 if( IS_FPU_DOUBLEPREC() && !IS_FPU_DOUBLESIZE() ) {
1273 FPULf = (float)DR(FRm);
1279 if( !IS_FPU_DOUBLEPREC() ) {
1280 FR(FRn) = 1.0/sqrtf(FR(FRn));
1285 if( !IS_FPU_DOUBLEPREC() ) {
1288 FR(tmp2+3) = FR(tmp)*FR(tmp2) +
1289 FR(tmp+1)*FR(tmp2+1) +
1290 FR(tmp+2)*FR(tmp2+2) +
1291 FR(tmp+3)*FR(tmp2+3);
1296 if( !IS_FPU_DOUBLEPREC() ) {
1297 sh4_fsca( FPULi, (float *)&(DRF(FRn>>1)) );
1299 float angle = (((float)(FPULi&0xFFFF))/65536.0) * 2 * M_PI;
1300 FR(FRn) = sinf(angle);
1301 FR((FRn)+1) = cosf(angle);
1307 if( !IS_FPU_DOUBLEPREC() ) {
1308 sh4_ftrv((float *)&(DRF(FVn<<1)) );
1315 sh4r.pc = sh4r.new_pc;
1319 sh4r.in_delay_slot = 0;
.