9 void sh4_save( FILE *f );
10 int sh4_load( FILE *f );
12 struct dreamcast_module sh4_module = { "SH4", sh4_init, sh4_reset,
16 struct sh4_registers sh4r;
17 static int running = 0;
21 register_io_regions( mmio_list_sh4mmio );
28 sh4r.new_pc= 0xA0000002;
29 sh4r.vbr = 0x00000000;
30 sh4r.fpscr = 0x00040001;
33 /* Everything else is undefined anyway, so don't bother setting it */
37 void sh4_set_pc( int pc )
48 void sh4_save( FILE *f )
50 fwrite( &sh4r, sizeof(sh4r), 1, f );
51 /* Save all additional on-board MMIO state */
54 int sh4_load( FILE * f )
56 fread( &sh4r, sizeof(sh4r), 1, f );
64 sh4_execute_instruction();
68 void sh4_runfor(uint32_t count)
71 while( running && count--) {
73 sh4_execute_instruction();
75 if( sh4r.pc == 0x8C0C1636 ||
76 sh4r.pc == 0x8C0C1634 ) {
77 WARN( "Branching to %08X from %08X", sh4r.pc, pc );
83 int sh4_isrunning(void)
88 void sh4_runto( uint32_t target_pc, uint32_t count )
91 while( running && count--) {
92 sh4_execute_instruction();
93 if( sh4r.pc == target_pc ) {
100 #define UNDEF(ir) do{ ERROR( "Raising exception on undefined instruction at %08x, opcode = %04x", sh4r.pc, ir ); sh4_stop(); RAISE( EXC_ILLEGAL, EXV_ILLEGAL ); }while(0)
101 #define UNIMP(ir) do{ ERROR( "Halted on unimplemented instruction at %08x, opcode = %04x", sh4r.pc, ir ); sh4_stop(); return; }while(0)
103 #define RAISE( x, v ) do{ \
104 if( sh4r.vbr == 0 ) { \
105 ERROR( "%08X: VBR not initialized while raising exception %03X, halting", sh4r.pc, x ); \
108 sh4r.spc = sh4r.pc + 2; \
109 sh4r.ssr = sh4_read_sr(); \
110 sh4r.sgr = sh4r.r[15]; \
111 MMIO_WRITE(MMU,EXPEVT,x); \
112 sh4r.pc = sh4r.vbr + v; \
113 sh4r.new_pc = sh4r.pc + 2; \
114 sh4_load_sr( sh4r.ssr |SR_MD|SR_BL|SR_RB ); \
118 #define MEM_READ_BYTE( addr ) sh4_read_byte(addr)
119 #define MEM_READ_WORD( addr ) sh4_read_word(addr)
120 #define MEM_READ_LONG( addr ) sh4_read_long(addr)
121 #define MEM_WRITE_BYTE( addr, val ) sh4_write_byte(addr, val)
122 #define MEM_WRITE_WORD( addr, val ) sh4_write_word(addr, val)
123 #define MEM_WRITE_LONG( addr, val ) sh4_write_long(addr, val)
125 #define MEM_FP_READ( addr, reg ) if( IS_FPU_DOUBLESIZE() ) { \
126 ((uint32_t *)FR)[(reg)&0xE0] = sh4_read_long(addr); \
127 ((uint32_t *)FR)[(reg)|1] = sh4_read_long(addr+4); \
128 } else ((uint32_t *)FR)[reg] = sh4_read_long(addr)
130 #define MEM_FP_WRITE( addr, reg ) if( IS_FPU_DOUBLESIZE() ) { \
131 sh4_write_long( addr, ((uint32_t *)FR)[(reg)&0xE0] ); \
132 sh4_write_long( addr+4, ((uint32_t *)FR)[(reg)|1] ); \
133 } else sh4_write_long( addr, ((uint32_t *)FR)[reg] )
135 #define FP_WIDTH (IS_FPU_DOUBLESIZE() ? 8 : 4)
137 #define EXC_POWER_RESET 0x000 /* vector special */
138 #define EXC_MANUAL_RESET 0x020
139 #define EXC_SLOT_ILLEGAL 0x1A0
140 #define EXC_ILLEGAL 0x180
141 #define EXV_ILLEGAL 0x100
142 #define EXC_TRAP 0x160
143 #define EXV_TRAP 0x100
144 #define EXC_FPDISABLE 0x800
145 #define EXV_FPDISABLE 0x100
147 #define CHECK( x, c, v ) if( !x ) RAISE( c, v )
148 #define CHECKPRIV() CHECK( IS_SH4_PRIVMODE(), EXC_ILLEGAL, EXV_ILLEGAL )
149 #define CHECKFPUEN() CHECK( IS_FPU_ENABLED(), EXC_FPDISABLE, EXV_FPDISABLE )
150 #define CHECKDEST(p) if( (p) == 0 ) { ERROR( "%08X: Branch/jump to NULL, CPU halted", sh4r.pc ); sh4_stop(); return; }
151 #define CHECKSLOTILLEGAL() if(sh4r.in_delay_slot) { RAISE(EXC_SLOT_ILLEGAL,EXV_ILLEGAL); }
153 static void sh4_switch_banks( )
157 memcpy( tmp, sh4r.r, sizeof(uint32_t)*8 );
158 memcpy( sh4r.r, sh4r.r_bank, sizeof(uint32_t)*8 );
159 memcpy( sh4r.r_bank, tmp, sizeof(uint32_t)*8 );
162 static void sh4_load_sr( uint32_t newval )
164 if( (newval ^ sh4r.sr) & SR_RB )
167 sh4r.t = (newval&SR_T) ? 1 : 0;
168 sh4r.s = (newval&SR_S) ? 1 : 0;
169 sh4r.m = (newval&SR_M) ? 1 : 0;
170 sh4r.q = (newval&SR_Q) ? 1 : 0;
174 static uint32_t sh4_read_sr( void )
176 /* synchronize sh4r.sr with the various bitflags */
177 sh4r.sr &= SR_MQSTMASK;
178 if( sh4r.t ) sh4r.sr |= SR_T;
179 if( sh4r.s ) sh4r.sr |= SR_S;
180 if( sh4r.m ) sh4r.sr |= SR_M;
181 if( sh4r.q ) sh4r.sr |= SR_Q;
184 /* function for external use */
185 void sh4_raise_exception( int code, int vector )
190 static void sh4_accept_interrupt( void )
192 uint32_t code = intc_accept_interrupt();
193 sh4r.ssr = sh4_read_sr();
195 sh4r.sgr = sh4r.r[15];
196 sh4_load_sr( sh4r.ssr|SR_BL|SR_MD|SR_RB );
197 MMIO_WRITE( MMU, INTEVT, code );
198 sh4r.pc = sh4r.vbr + 0x600;
199 sh4r.new_pc = sh4r.pc + 2;
200 WARN( "Accepting interrupt %03X, from %08X => %08X", code, sh4r.spc, sh4r.pc );
203 void sh4_execute_instruction( void )
212 #define RN(ir) sh4r.r[(ir&0x0F00)>>8]
213 #define RN_BANK(ir) sh4r.r_bank[(ir&0x0070)>>4]
214 #define RM(ir) sh4r.r[(ir&0x00F0)>>4]
215 #define DISP4(ir) (ir&0x000F) /* 4-bit displacements are *NOT* sign-extended */
216 #define DISP8(ir) (ir&0x00FF)
217 #define PCDISP8(ir) SIGNEXT8(ir&0x00FF)
218 #define IMM8(ir) SIGNEXT8(ir&0x00FF)
219 #define UIMM8(ir) (ir&0x00FF) /* Unsigned immmediate */
220 #define DISP12(ir) SIGNEXT12(ir&0x0FFF)
221 #define FVN(ir) ((ir&0x0C00)>>8)
222 #define FVM(ir) ((ir&0x0300)>>6)
223 #define FRN(ir) (FR[(ir&0x0F00)>>8])
224 #define FRM(ir) (FR[(ir&0x00F0)>>4])
225 #define FRNi(ir) (((uint32_t *)FR)[(ir&0x0F00)>>8])
226 #define FRMi(ir) (((uint32_t *)FR)[(ir&0x00F0)>>4])
227 #define DRN(ir) (((double *)FR)[(ir&0x0E00)>>9])
228 #define DRM(ir) (((double *)FR)[(ir&0x00E0)>>5])
229 #define DRNi(ir) (((uint64_t *)FR)[(ir&0x0E00)>>9])
230 #define DRMi(ir) (((uint64_t *)FR)[(ir&0x00E0)>>5])
231 #define FRNn(ir) ((ir&0x0F00)>>8)
232 #define FRMn(ir) ((ir&0x00F0)>>4)
233 #define FPULf *((float *)&sh4r.fpul)
234 #define FPULi (sh4r.fpul)
236 if( SH4_INT_PENDING() )
237 sh4_accept_interrupt();
240 ir = MEM_READ_WORD(pc);
243 switch( (ir&0xF000)>>12 ) {
244 case 0: /* 0000nnnnmmmmxxxx */
245 switch( ir&0x000F ) {
247 switch( (ir&0x00F0)>>4 ) {
248 case 0: /* STC SR, Rn */
250 RN(ir) = sh4_read_sr();
252 case 1: /* STC GBR, Rn */
255 case 2: /* STC VBR, Rn */
259 case 3: /* STC SSR, Rn */
263 case 4: /* STC SPC, Rn */
267 case 8: case 9: case 10: case 11: case 12: case 13:
268 case 14: case 15:/* STC Rm_bank, Rn */
270 RN(ir) = RN_BANK(ir);
276 switch( (ir&0x00F0)>>4 ) {
277 case 0: /* BSRF Rn */
278 CHECKDEST( pc + 4 + RN(ir) );
280 sh4r.in_delay_slot = 1;
281 sh4r.pr = sh4r.pc + 4;
282 sh4r.pc = sh4r.new_pc;
283 sh4r.new_pc = pc + 4 + RN(ir);
285 case 2: /* BRAF Rn */
286 CHECKDEST( pc + 4 + RN(ir) );
288 sh4r.in_delay_slot = 1;
289 sh4r.pc = sh4r.new_pc;
290 sh4r.new_pc = pc + 4 + RN(ir);
292 case 8: /* PREF [Rn] */
294 if( (tmp & 0xFC000000) == 0xE0000000 ) {
295 /* Store queue operation */
296 int queue = (tmp&0x20)>>2;
297 int32_t *src = &sh4r.store_queue[queue];
298 uint32_t hi = (MMIO_READ( MMU, (queue == 0 ? QACR0 : QACR1) ) & 0x1C) << 24;
299 uint32_t target = tmp&0x03FFFFE0 | hi;
300 mem_copy_to_sh4( target, src, 32 );
301 WARN( "Executed SQ%c => %08X",
302 (queue == 0 ? '0' : '1'), target );
305 case 9: /* OCBI [Rn] */
306 case 10:/* OCBP [Rn] */
307 case 11:/* OCBWB [Rn] */
310 case 12:/* MOVCA.L R0, [Rn] */
315 case 4: /* MOV.B Rm, [R0 + Rn] */
316 MEM_WRITE_BYTE( R0 + RN(ir), RM(ir) );
318 case 5: /* MOV.W Rm, [R0 + Rn] */
319 MEM_WRITE_WORD( R0 + RN(ir), RM(ir) );
321 case 6: /* MOV.L Rm, [R0 + Rn] */
322 MEM_WRITE_LONG( R0 + RN(ir), RM(ir) );
324 case 7: /* MUL.L Rm, Rn */
325 sh4r.mac = (sh4r.mac&0xFFFFFFFF00000000LL) |
329 switch( (ir&0x0FF0)>>4 ) {
351 if( (ir&0x00F0) == 0x20 ) /* MOVT Rn */
353 else if( ir == 0x0019 ) /* DIV0U */
354 sh4r.m = sh4r.q = sh4r.t = 0;
355 else if( ir == 0x0009 )
360 switch( (ir&0x00F0) >> 4 ) {
361 case 0: /* STS MACH, Rn */
362 RN(ir) = sh4r.mac >> 32;
364 case 1: /* STS MACL, Rn */
365 RN(ir) = (uint32_t)sh4r.mac;
367 case 2: /* STS PR, Rn */
370 case 3: /* STC SGR, Rn */
374 case 5:/* STS FPUL, Rn */
377 case 6: /* STS FPSCR, Rn */
380 case 15:/* STC DBR, Rn */
388 switch( (ir&0x0FF0)>>4 ) {
390 CHECKDEST( sh4r.pr );
392 sh4r.in_delay_slot = 1;
393 sh4r.pc = sh4r.new_pc;
394 sh4r.new_pc = sh4r.pr;
401 CHECKDEST( sh4r.spc );
403 sh4r.in_delay_slot = 1;
404 sh4r.pc = sh4r.new_pc;
405 sh4r.new_pc = sh4r.spc;
406 sh4_load_sr( sh4r.ssr );
407 WARN( "RTE => %08X", sh4r.new_pc );
412 case 12:/* MOV.B [R0+R%d], R%d */
413 RN(ir) = MEM_READ_BYTE( R0 + RM(ir) );
415 case 13:/* MOV.W [R0+R%d], R%d */
416 RN(ir) = MEM_READ_WORD( R0 + RM(ir) );
418 case 14:/* MOV.L [R0+R%d], R%d */
419 RN(ir) = MEM_READ_LONG( R0 + RM(ir) );
421 case 15:/* MAC.L [Rm++], [Rn++] */
422 tmpl = ( SIGNEXT32(MEM_READ_LONG(RM(ir))) *
423 SIGNEXT32(MEM_READ_LONG(RN(ir))) );
425 /* 48-bit Saturation. Yuch */
426 tmpl += SIGNEXT48(sh4r.mac);
427 if( tmpl < 0xFFFF800000000000LL )
428 tmpl = 0xFFFF800000000000LL;
429 else if( tmpl > 0x00007FFFFFFFFFFFLL )
430 tmpl = 0x00007FFFFFFFFFFFLL;
431 sh4r.mac = (sh4r.mac&0xFFFF000000000000LL) |
432 (tmpl&0x0000FFFFFFFFFFFFLL);
433 } else sh4r.mac = tmpl;
442 case 1: /* 0001nnnnmmmmdddd */
443 /* MOV.L Rm, [Rn + disp4*4] */
444 MEM_WRITE_LONG( RN(ir) + (DISP4(ir)<<2), RM(ir) );
446 case 2: /* 0010nnnnmmmmxxxx */
447 switch( ir&0x000F ) {
448 case 0: /* MOV.B Rm, [Rn] */
449 MEM_WRITE_BYTE( RN(ir), RM(ir) );
451 case 1: /* MOV.W Rm, [Rn] */
452 MEM_WRITE_WORD( RN(ir), RM(ir) );
454 case 2: /* MOV.L Rm, [Rn] */
455 MEM_WRITE_LONG( RN(ir), RM(ir) );
459 case 4: /* MOV.B Rm, [--Rn] */
461 MEM_WRITE_BYTE( RN(ir), RM(ir) );
463 case 5: /* MOV.W Rm, [--Rn] */
465 MEM_WRITE_WORD( RN(ir), RM(ir) );
467 case 6: /* MOV.L Rm, [--Rn] */
469 MEM_WRITE_LONG( RN(ir), RM(ir) );
471 case 7: /* DIV0S Rm, Rn */
474 sh4r.t = sh4r.q ^ sh4r.m;
476 case 8: /* TST Rm, Rn */
477 sh4r.t = (RN(ir)&RM(ir) ? 0 : 1);
479 case 9: /* AND Rm, Rn */
482 case 10:/* XOR Rm, Rn */
485 case 11:/* OR Rm, Rn */
488 case 12:/* CMP/STR Rm, Rn */
489 /* set T = 1 if any byte in RM & RN is the same */
490 tmp = RM(ir) ^ RN(ir);
491 sh4r.t = ((tmp&0x000000FF)==0 || (tmp&0x0000FF00)==0 ||
492 (tmp&0x00FF0000)==0 || (tmp&0xFF000000)==0)?1:0;
494 case 13:/* XTRCT Rm, Rn */
495 RN(ir) = (RN(ir)>>16) | (RM(ir)<<16);
497 case 14:/* MULU.W Rm, Rn */
498 sh4r.mac = (sh4r.mac&0xFFFFFFFF00000000LL) |
499 (uint32_t)((RM(ir)&0xFFFF) * (RN(ir)&0xFFFF));
501 case 15:/* MULS.W Rm, Rn */
502 sh4r.mac = (sh4r.mac&0xFFFFFFFF00000000LL) |
503 (uint32_t)(SIGNEXT32(RM(ir)&0xFFFF) * SIGNEXT32(RN(ir)&0xFFFF));
507 case 3: /* 0011nnnnmmmmxxxx */
508 switch( ir&0x000F ) {
509 case 0: /* CMP/EQ Rm, Rn */
510 sh4r.t = ( RM(ir) == RN(ir) ? 1 : 0 );
512 case 2: /* CMP/HS Rm, Rn */
513 sh4r.t = ( RN(ir) >= RM(ir) ? 1 : 0 );
515 case 3: /* CMP/GE Rm, Rn */
516 sh4r.t = ( ((int32_t)RN(ir)) >= ((int32_t)RM(ir)) ? 1 : 0 );
518 case 4: { /* DIV1 Rm, Rn */
519 /* This is just from the sh4p manual with some
520 * simplifications (someone want to check it's correct? :)
521 * Why they couldn't just provide a real DIV instruction...
522 * Please oh please let the translator batch these things
523 * up into a single DIV... */
524 uint32_t tmp0, tmp1, tmp2, dir;
526 dir = sh4r.q ^ sh4r.m;
527 sh4r.q = (RN(ir) >> 31);
529 RN(ir) = (RN(ir) << 1) | sh4r.t;
533 tmp1 = (RN(ir)<tmp0 ? 1 : 0 );
536 tmp1 = (RN(ir)>tmp0 ? 1 : 0 );
538 sh4r.q ^= sh4r.m ^ tmp1;
539 sh4r.t = ( sh4r.q == sh4r.m ? 1 : 0 );
541 case 5: /* DMULU.L Rm, Rn */
542 sh4r.mac = ((uint64_t)RM(ir)) * ((uint64_t)RN(ir));
544 case 6: /* CMP/HI Rm, Rn */
545 sh4r.t = ( RN(ir) > RM(ir) ? 1 : 0 );
547 case 7: /* CMP/GT Rm, Rn */
548 sh4r.t = ( ((int32_t)RN(ir)) > ((int32_t)RM(ir)) ? 1 : 0 );
550 case 8: /* SUB Rm, Rn */
553 case 10:/* SUBC Rm, Rn */
555 RN(ir) = RN(ir) - RM(ir) - sh4r.t;
556 sh4r.t = (RN(ir) > tmp || (RN(ir) == tmp && sh4r.t == 1));
558 case 11:/* SUBV Rm, Rn */
561 case 12:/* ADD Rm, Rn */
564 case 13:/* DMULS.L Rm, Rn */
565 sh4r.mac = SIGNEXT32(RM(ir)) * SIGNEXT32(RN(ir));
567 case 14:/* ADDC Rm, Rn */
569 RN(ir) += RM(ir) + sh4r.t;
570 sh4r.t = ( RN(ir) < tmp || (RN(ir) == tmp && sh4r.t != 0) ? 1 : 0 );
572 case 15:/* ADDV Rm, Rn */
578 case 4: /* 0100nnnnxxxxxxxx */
579 switch( ir&0x00FF ) {
580 case 0x00: /* SHLL Rn */
581 sh4r.t = RN(ir) >> 31;
584 case 0x01: /* SHLR Rn */
585 sh4r.t = RN(ir) & 0x00000001;
588 case 0x02: /* STS.L MACH, [--Rn] */
590 MEM_WRITE_LONG( RN(ir), (sh4r.mac>>32) );
592 case 0x03: /* STC.L SR, [--Rn] */
595 MEM_WRITE_LONG( RN(ir), sh4_read_sr() );
597 case 0x04: /* ROTL Rn */
598 sh4r.t = RN(ir) >> 31;
602 case 0x05: /* ROTR Rn */
603 sh4r.t = RN(ir) & 0x00000001;
605 RN(ir) |= (sh4r.t << 31);
607 case 0x06: /* LDS.L [Rn++], MACH */
608 sh4r.mac = (sh4r.mac & 0x00000000FFFFFFFF) |
609 (((uint64_t)MEM_READ_LONG(RN(ir)))<<32);
612 case 0x07: /* LDC.L [Rn++], SR */
614 sh4_load_sr( MEM_READ_LONG(RN(ir)) );
617 case 0x08: /* SHLL2 Rn */
620 case 0x09: /* SHLR2 Rn */
623 case 0x0A: /* LDS Rn, MACH */
624 sh4r.mac = (sh4r.mac & 0x00000000FFFFFFFF) |
625 (((uint64_t)RN(ir))<<32);
627 case 0x0B: /* JSR [Rn] */
630 sh4r.in_delay_slot = 1;
631 sh4r.pc = sh4r.new_pc;
632 sh4r.new_pc = RN(ir);
635 case 0x0E: /* LDC Rn, SR */
637 sh4_load_sr( RN(ir) );
639 case 0x10: /* DT Rn */
641 sh4r.t = ( RN(ir) == 0 ? 1 : 0 );
643 case 0x11: /* CMP/PZ Rn */
644 sh4r.t = ( ((int32_t)RN(ir)) >= 0 ? 1 : 0 );
646 case 0x12: /* STS.L MACL, [--Rn] */
648 MEM_WRITE_LONG( RN(ir), (uint32_t)sh4r.mac );
650 case 0x13: /* STC.L GBR, [--Rn] */
652 MEM_WRITE_LONG( RN(ir), sh4r.gbr );
654 case 0x15: /* CMP/PL Rn */
655 sh4r.t = ( ((int32_t)RN(ir)) > 0 ? 1 : 0 );
657 case 0x16: /* LDS.L [Rn++], MACL */
658 sh4r.mac = (sh4r.mac & 0xFFFFFFFF00000000LL) |
659 (uint64_t)((uint32_t)MEM_READ_LONG(RN(ir)));
662 case 0x17: /* LDC.L [Rn++], GBR */
663 sh4r.gbr = MEM_READ_LONG(RN(ir));
666 case 0x18: /* SHLL8 Rn */
669 case 0x19: /* SHLR8 Rn */
672 case 0x1A: /* LDS Rn, MACL */
673 sh4r.mac = (sh4r.mac & 0xFFFFFFFF00000000LL) |
674 (uint64_t)((uint32_t)(RN(ir)));
676 case 0x1B: /* TAS.B [Rn] */
677 tmp = MEM_READ_BYTE( RN(ir) );
678 sh4r.t = ( tmp == 0 ? 1 : 0 );
679 MEM_WRITE_BYTE( RN(ir), tmp | 0x80 );
681 case 0x1E: /* LDC Rn, GBR */
684 case 0x20: /* SHAL Rn */
685 sh4r.t = RN(ir) >> 31;
688 case 0x21: /* SHAR Rn */
689 sh4r.t = RN(ir) & 0x00000001;
690 RN(ir) = ((int32_t)RN(ir)) >> 1;
692 case 0x22: /* STS.L PR, [--Rn] */
694 MEM_WRITE_LONG( RN(ir), sh4r.pr );
696 case 0x23: /* STC.L VBR, [--Rn] */
699 MEM_WRITE_LONG( RN(ir), sh4r.vbr );
701 case 0x24: /* ROTCL Rn */
707 case 0x25: /* ROTCR Rn */
708 tmp = RN(ir) & 0x00000001;
710 RN(ir) |= (sh4r.t << 31 );
713 case 0x26: /* LDS.L [Rn++], PR */
714 sh4r.pr = MEM_READ_LONG( RN(ir) );
717 case 0x27: /* LDC.L [Rn++], VBR */
719 sh4r.vbr = MEM_READ_LONG(RN(ir));
722 case 0x28: /* SHLL16 Rn */
725 case 0x29: /* SHLR16 Rn */
728 case 0x2A: /* LDS Rn, PR */
731 case 0x2B: /* JMP [Rn] */
734 sh4r.in_delay_slot = 1;
735 sh4r.pc = sh4r.new_pc;
736 sh4r.new_pc = RN(ir);
738 case 0x2E: /* LDC Rn, VBR */
742 case 0x32: /* STC.L SGR, [--Rn] */
745 MEM_WRITE_LONG( RN(ir), sh4r.sgr );
747 case 0x33: /* STC.L SSR, [--Rn] */
750 MEM_WRITE_LONG( RN(ir), sh4r.ssr );
752 case 0x37: /* LDC.L [Rn++], SSR */
754 sh4r.ssr = MEM_READ_LONG(RN(ir));
757 case 0x3E: /* LDC Rn, SSR */
761 case 0x43: /* STC.L SPC, [--Rn] */
764 MEM_WRITE_LONG( RN(ir), sh4r.spc );
766 case 0x47: /* LDC.L [Rn++], SPC */
768 sh4r.spc = MEM_READ_LONG(RN(ir));
771 case 0x4E: /* LDC Rn, SPC */
775 case 0x52: /* STS.L FPUL, [--Rn] */
777 MEM_WRITE_LONG( RN(ir), sh4r.fpul );
779 case 0x56: /* LDS.L [Rn++], FPUL */
780 sh4r.fpul = MEM_READ_LONG(RN(ir));
783 case 0x5A: /* LDS Rn, FPUL */
786 case 0x62: /* STS.L FPSCR, [--Rn] */
788 MEM_WRITE_LONG( RN(ir), sh4r.fpscr );
790 case 0x66: /* LDS.L [Rn++], FPSCR */
791 sh4r.fpscr = MEM_READ_LONG(RN(ir));
794 case 0x6A: /* LDS Rn, FPSCR */
797 case 0xF2: /* STC.L DBR, [--Rn] */
800 MEM_WRITE_LONG( RN(ir), sh4r.dbr );
802 case 0xF6: /* LDC.L [Rn++], DBR */
804 sh4r.dbr = MEM_READ_LONG(RN(ir));
807 case 0xFA: /* LDC Rn, DBR */
811 case 0x83: case 0x93: case 0xA3: case 0xB3: case 0xC3:
812 case 0xD3: case 0xE3: case 0xF3: /* STC.L Rn_BANK, [--Rn] */
815 MEM_WRITE_LONG( RN(ir), RN_BANK(ir) );
817 case 0x87: case 0x97: case 0xA7: case 0xB7: case 0xC7:
818 case 0xD7: case 0xE7: case 0xF7: /* LDC.L [Rn++], Rn_BANK */
820 RN_BANK(ir) = MEM_READ_LONG( RN(ir) );
823 case 0x8E: case 0x9E: case 0xAE: case 0xBE: case 0xCE:
824 case 0xDE: case 0xEE: case 0xFE: /* LDC Rm, Rn_BANK */
826 RN_BANK(ir) = RM(ir);
829 if( (ir&0x000F) == 0x0F ) {
830 /* MAC.W [Rm++], [Rn++] */
831 tmp = SIGNEXT16(MEM_READ_WORD(RM(ir))) *
832 SIGNEXT16(MEM_READ_WORD(RN(ir)));
836 } else sh4r.mac += SIGNEXT32(tmp);
839 } else if( (ir&0x000F) == 0x0C ) {
842 if( (tmp & 0x80000000) == 0 ) RN(ir) <<= (tmp&0x1f);
843 else if( (tmp & 0x1F) == 0 )
844 RN(ir) = ((int32_t)RN(ir)) >> 31;
846 RN(ir) = ((int32_t)RN(ir)) >> (((~RM(ir)) & 0x1F)+1);
847 } else if( (ir&0x000F) == 0x0D ) {
850 if( (tmp & 0x80000000) == 0 ) RN(ir) <<= (tmp&0x1f);
851 else if( (tmp & 0x1F) == 0 ) RN(ir) = 0;
852 else RN(ir) >>= (((~tmp) & 0x1F)+1);
856 case 5: /* 0101nnnnmmmmdddd */
857 /* MOV.L [Rm + disp4*4], Rn */
858 RN(ir) = MEM_READ_LONG( RM(ir) + (DISP4(ir)<<2) );
860 case 6: /* 0110xxxxxxxxxxxx */
861 switch( ir&0x000f ) {
862 case 0: /* MOV.B [Rm], Rn */
863 RN(ir) = MEM_READ_BYTE( RM(ir) );
865 case 1: /* MOV.W [Rm], Rn */
866 RN(ir) = MEM_READ_WORD( RM(ir) );
868 case 2: /* MOV.L [Rm], Rn */
869 RN(ir) = MEM_READ_LONG( RM(ir) );
871 case 3: /* MOV Rm, Rn */
874 case 4: /* MOV.B [Rm++], Rn */
875 RN(ir) = MEM_READ_BYTE( RM(ir) );
878 case 5: /* MOV.W [Rm++], Rn */
879 RN(ir) = MEM_READ_WORD( RM(ir) );
882 case 6: /* MOV.L [Rm++], Rn */
883 RN(ir) = MEM_READ_LONG( RM(ir) );
886 case 7: /* NOT Rm, Rn */
889 case 8: /* SWAP.B Rm, Rn */
890 RN(ir) = (RM(ir)&0xFFFF0000) | ((RM(ir)&0x0000FF00)>>8) |
891 ((RM(ir)&0x000000FF)<<8);
893 case 9: /* SWAP.W Rm, Rn */
894 RN(ir) = (RM(ir)>>16) | (RM(ir)<<16);
896 case 10:/* NEGC Rm, Rn */
898 RN(ir) = tmp - sh4r.t;
899 sh4r.t = ( 0<tmp || tmp<RN(ir) ? 1 : 0 );
901 case 11:/* NEG Rm, Rn */
904 case 12:/* EXTU.B Rm, Rn */
905 RN(ir) = RM(ir)&0x000000FF;
907 case 13:/* EXTU.W Rm, Rn */
908 RN(ir) = RM(ir)&0x0000FFFF;
910 case 14:/* EXTS.B Rm, Rn */
911 RN(ir) = SIGNEXT8( RM(ir)&0x000000FF );
913 case 15:/* EXTS.W Rm, Rn */
914 RN(ir) = SIGNEXT16( RM(ir)&0x0000FFFF );
918 case 7: /* 0111nnnniiiiiiii */
922 case 8: /* 1000xxxxxxxxxxxx */
923 switch( (ir&0x0F00) >> 8 ) {
924 case 0: /* MOV.B R0, [Rm + disp4] */
925 MEM_WRITE_BYTE( RM(ir) + DISP4(ir), R0 );
927 case 1: /* MOV.W R0, [Rm + disp4*2] */
928 MEM_WRITE_WORD( RM(ir) + (DISP4(ir)<<1), R0 );
930 case 4: /* MOV.B [Rm + disp4], R0 */
931 R0 = MEM_READ_BYTE( RM(ir) + DISP4(ir) );
933 case 5: /* MOV.W [Rm + disp4*2], R0 */
934 R0 = MEM_READ_WORD( RM(ir) + (DISP4(ir)<<1) );
936 case 8: /* CMP/EQ imm, R0 */
937 sh4r.t = ( R0 == IMM8(ir) ? 1 : 0 );
939 case 9: /* BT disp8 */
942 CHECKDEST( sh4r.pc + (PCDISP8(ir)<<1) + 4 )
943 sh4r.pc += (PCDISP8(ir)<<1) + 4;
944 sh4r.new_pc = sh4r.pc + 2;
948 case 11:/* BF disp8 */
951 CHECKDEST( sh4r.pc + (PCDISP8(ir)<<1) + 4 )
952 sh4r.pc += (PCDISP8(ir)<<1) + 4;
953 sh4r.new_pc = sh4r.pc + 2;
957 case 13:/* BT/S disp8 */
960 CHECKDEST( sh4r.pc + (PCDISP8(ir)<<1) + 4 )
961 sh4r.in_delay_slot = 1;
962 sh4r.pc = sh4r.new_pc;
963 sh4r.new_pc = pc + (PCDISP8(ir)<<1) + 4;
964 sh4r.in_delay_slot = 1;
968 case 15:/* BF/S disp8 */
971 CHECKDEST( sh4r.pc + (PCDISP8(ir)<<1) + 4 )
972 sh4r.in_delay_slot = 1;
973 sh4r.pc = sh4r.new_pc;
974 sh4r.new_pc = pc + (PCDISP8(ir)<<1) + 4;
981 case 9: /* 1001xxxxxxxxxxxx */
982 /* MOV.W [disp8*2 + pc + 4], Rn */
983 RN(ir) = MEM_READ_WORD( pc + 4 + (DISP8(ir)<<1) );
985 case 10:/* 1010dddddddddddd */
987 CHECKDEST( sh4r.pc + (DISP12(ir)<<1) + 4 )
989 sh4r.in_delay_slot = 1;
990 sh4r.pc = sh4r.new_pc;
991 sh4r.new_pc = pc + 4 + (DISP12(ir)<<1);
993 case 11:/* 1011dddddddddddd */
995 CHECKDEST( sh4r.pc + (DISP12(ir)<<1) + 4 )
997 sh4r.in_delay_slot = 1;
999 sh4r.pc = sh4r.new_pc;
1000 sh4r.new_pc = pc + 4 + (DISP12(ir)<<1);
1002 case 12:/* 1100xxxxdddddddd */
1003 switch( (ir&0x0F00)>>8 ) {
1004 case 0: /* MOV.B R0, [GBR + disp8] */
1005 MEM_WRITE_BYTE( sh4r.gbr + DISP8(ir), R0 );
1007 case 1: /* MOV.W R0, [GBR + disp8*2] */
1008 MEM_WRITE_WORD( sh4r.gbr + (DISP8(ir)<<1), R0 );
1010 case 2: /*MOV.L R0, [GBR + disp8*4] */
1011 MEM_WRITE_LONG( sh4r.gbr + (DISP8(ir)<<2), R0 );
1013 case 3: /* TRAPA imm8 */
1015 sh4r.in_delay_slot = 1;
1016 MMIO_WRITE( MMU, TRA, UIMM8(ir) );
1017 sh4r.pc = sh4r.new_pc; /* RAISE ends the instruction */
1019 RAISE( EXC_TRAP, EXV_TRAP );
1021 case 4: /* MOV.B [GBR + disp8], R0 */
1022 R0 = MEM_READ_BYTE( sh4r.gbr + DISP8(ir) );
1024 case 5: /* MOV.W [GBR + disp8*2], R0 */
1025 R0 = MEM_READ_WORD( sh4r.gbr + (DISP8(ir)<<1) );
1027 case 6: /* MOV.L [GBR + disp8*4], R0 */
1028 R0 = MEM_READ_LONG( sh4r.gbr + (DISP8(ir)<<2) );
1030 case 7: /* MOVA disp8 + pc&~3 + 4, R0 */
1031 R0 = (pc&0xFFFFFFFC) + (DISP8(ir)<<2) + 4;
1033 case 8: /* TST imm8, R0 */
1034 sh4r.t = (R0 & UIMM8(ir) ? 0 : 1);
1036 case 9: /* AND imm8, R0 */
1039 case 10:/* XOR imm8, R0 */
1042 case 11:/* OR imm8, R0 */
1045 case 12:/* TST.B imm8, [R0+GBR] */
1046 sh4r.t = ( MEM_READ_BYTE(R0 + sh4r.gbr) & UIMM8(ir) ? 0 : 1 );
1048 case 13:/* AND.B imm8, [R0+GBR] */
1049 MEM_WRITE_BYTE( R0 + sh4r.gbr,
1050 UIMM8(ir) & MEM_READ_BYTE(R0 + sh4r.gbr) );
1052 case 14:/* XOR.B imm8, [R0+GBR] */
1053 MEM_WRITE_BYTE( R0 + sh4r.gbr,
1054 UIMM8(ir) ^ MEM_READ_BYTE(R0 + sh4r.gbr) );
1056 case 15:/* OR.B imm8, [R0+GBR] */
1057 MEM_WRITE_BYTE( R0 + sh4r.gbr,
1058 UIMM8(ir) | MEM_READ_BYTE(R0 + sh4r.gbr) );
1062 case 13:/* 1101nnnndddddddd */
1063 /* MOV.L [disp8*4 + pc&~3 + 4], Rn */
1064 RN(ir) = MEM_READ_LONG( (pc&0xFFFFFFFC) + (DISP8(ir)<<2) + 4 );
1066 case 14:/* 1110nnnniiiiiiii */
1070 case 15:/* 1111xxxxxxxxxxxx */
1072 switch( ir&0x000F ) {
1073 case 0: /* FADD FRm, FRn */
1076 case 1: /* FSUB FRm, FRn */
1079 case 2: /* FMUL FRm, FRn */
1080 FRN(ir) = FRN(ir) * FRM(ir);
1082 case 3: /* FDIV FRm, FRn */
1083 FRN(ir) = FRN(ir) / FRM(ir);
1085 case 4: /* FCMP/EQ FRm, FRn */
1086 sh4r.t = ( FRN(ir) == FRM(ir) ? 1 : 0 );
1088 case 5: /* FCMP/GT FRm, FRn */
1089 sh4r.t = ( FRN(ir) > FRM(ir) ? 1 : 0 );
1091 case 6: /* FMOV.S [Rm+R0], FRn */
1092 MEM_FP_READ( RM(ir) + R0, FRNn(ir) );
1094 case 7: /* FMOV.S FRm, [Rn+R0] */
1095 MEM_FP_WRITE( RN(ir) + R0, FRMn(ir) );
1097 case 8: /* FMOV.S [Rm], FRn */
1098 MEM_FP_READ( RM(ir), FRNn(ir) );
1100 case 9: /* FMOV.S [Rm++], FRn */
1101 MEM_FP_READ( RM(ir), FRNn(ir) );
1104 case 10:/* FMOV.S FRm, [Rn] */
1105 MEM_FP_WRITE( RN(ir), FRMn(ir) );
1107 case 11:/* FMOV.S FRm, [--Rn] */
1109 MEM_FP_WRITE( RN(ir), FRMn(ir) );
1111 case 12:/* FMOV FRm, FRn */
1112 if( IS_FPU_DOUBLESIZE() ) {
1119 switch( (ir&0x00F0) >> 4 ) {
1120 case 0: /* FSTS FPUL, FRn */
1123 case 1: /* FLDS FRn, FPUL */
1126 case 2: /* FLOAT FPUL, FRn */
1127 FRN(ir) = (float)FPULi;
1129 case 3: /* FTRC FRn, FPUL */
1130 FPULi = (uint32_t)FRN(ir);
1131 /* FIXME: is this sufficient? */
1133 case 4: /* FNEG FRn */
1136 case 5: /* FABS FRn */
1137 FRN(ir) = fabsf(FRN(ir));
1139 case 6: /* FSQRT FRn */
1140 FRN(ir) = sqrtf(FRN(ir));
1142 case 7: /* FSRRA FRn */
1143 FRN(ir) = 1.0/sqrtf(FRN(ir));
1145 case 8: /* FLDI0 FRn */
1148 case 9: /* FLDI1 FRn */
1151 case 10: /* FCNVSD FPUL, DRn */
1152 if( IS_FPU_DOUBLEPREC() )
1153 DRN(ir) = (double)FPULf;
1156 case 11: /* FCNVDS DRn, FPUL */
1157 if( IS_FPU_DOUBLEPREC() )
1158 FPULf = (float)DRN(ir);
1161 case 14:/* FIPR FVm, FVn */
1162 /* FIXME: This is not going to be entirely accurate
1163 * as the SH4 instruction is less precise. Also
1164 * need to check for 0s and infinities.
1167 float *fr_bank = FR;
1170 fr_bank[tmp2+3] = fr_bank[tmp]*fr_bank[tmp2] +
1171 fr_bank[tmp+1]*fr_bank[tmp2+1] +
1172 fr_bank[tmp+2]*fr_bank[tmp2+2] +
1173 fr_bank[tmp+3]*fr_bank[tmp2+3];
1177 if( (ir&0x0300) == 0x0100 ) { /* FTRV XMTRX,FVn */
1178 float *fvout = FR+FVN(ir);
1180 float fv[4] = { fvout[0], fvout[1], fvout[2], fvout[3] };
1181 fvout[0] = xm[0] * fv[0] + xm[4]*fv[1] +
1182 xm[8]*fv[2] + xm[12]*fv[3];
1183 fvout[1] = xm[1] * fv[0] + xm[5]*fv[1] +
1184 xm[9]*fv[2] + xm[13]*fv[3];
1185 fvout[2] = xm[2] * fv[0] + xm[6]*fv[1] +
1186 xm[10]*fv[2] + xm[14]*fv[3];
1187 fvout[3] = xm[3] * fv[0] + xm[7]*fv[1] +
1188 xm[11]*fv[2] + xm[15]*fv[3];
1191 else if( (ir&0x0100) == 0 ) { /* FSCA FPUL, DRn */
1192 float angle = (((float)(short)(FPULi>>16)) +
1193 ((float)(FPULi&16)/65536.0)) *
1196 FR[reg] = sinf(angle);
1197 FR[reg+1] = cosf(angle);
1200 else if( ir == 0xFBFD ) {
1202 sh4r.fpscr ^= FPSCR_FR;
1205 else if( ir == 0xF3FD ) {
1207 sh4r.fpscr ^= FPSCR_SZ;
1213 case 14:/* FMAC FR0, FRm, FRn */
1214 FRN(ir) += FRM(ir)*FR0;
1220 sh4r.pc = sh4r.new_pc;
1222 sh4r.in_delay_slot = 0;
.