/**

 * $Id$

 * 

 * SH4 emulation core, and parent module for all the SH4 peripheral

 * modules.

 *

 * Copyright (c) 2005 Nathan Keynes.

 *

 * This program is free software; you can redistribute it and/or modify

 * it under the terms of the GNU General Public License as published by

 * the Free Software Foundation; either version 2 of the License, or

 * (at your option) any later version.

 *

 * This program is distributed in the hope that it will be useful,

 * but WITHOUT ANY WARRANTY; without even the implied warranty of

 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

 * GNU General Public License for more details.

 */

#define MODULE sh4_module

#include <assert.h>

#include <math.h>

#include "dream.h"

#include "dreamcast.h"

#include "eventq.h"

#include "mem.h"

#include "clock.h"

#include "syscall.h"

#include "sh4/sh4core.h"

#include "sh4/sh4mmio.h"

#include "sh4/sh4stat.h"

#include "sh4/mmu.h"

#define SH4_CALLTRACE 1

#define MAX_INT 0x7FFFFFFF

#define MIN_INT 0x80000000

#define MAX_INTF 2147483647.0

#define MIN_INTF -2147483648.0

/********************** SH4 Module Definition ****************************/

uint32_t sh4_emulate_run_slice( uint32_t nanosecs ) 

{

    int i;

    if( sh4_breakpoint_count == 0 ) {

	for( ; sh4r.slice_cycle < nanosecs; sh4r.slice_cycle += sh4_cpu_period ) {

	    if( SH4_EVENT_PENDING() ) {

		if( sh4r.event_types & PENDING_EVENT ) {

		    event_execute();

		}

		/* Eventq execute may (quite likely) deliver an immediate IRQ */

		if( sh4r.event_types & PENDING_IRQ ) {

		    sh4_accept_interrupt();

		}

	    }

	    if( !sh4_execute_instruction() ) {

		break;

	    }

	}

    } else {

	for( ;sh4r.slice_cycle < nanosecs; sh4r.slice_cycle += sh4_cpu_period ) {

	    if( SH4_EVENT_PENDING() ) {

		if( sh4r.event_types & PENDING_EVENT ) {

		    event_execute();

		}

		/* Eventq execute may (quite likely) deliver an immediate IRQ */

		if( sh4r.event_types & PENDING_IRQ ) {

		    sh4_accept_interrupt();

		}

	    }

	    if( !sh4_execute_instruction() )

		break;

#ifdef ENABLE_DEBUG_MODE

	    for( i=0; i<sh4_breakpoint_count; i++ ) {

		if( sh4_breakpoints[i].address == sh4r.pc ) {

		    break;

		}

	    }

	    if( i != sh4_breakpoint_count ) {

	    	sh4_core_exit( CORE_EXIT_BREAKPOINT );

	    }

#endif	

	}

    }

    /* If we aborted early, but the cpu is still technically running,

     * we're doing a hard abort - cut the timeslice back to what we

     * actually executed

     */

    if( sh4r.slice_cycle != nanosecs && sh4r.sh4_state == SH4_STATE_RUNNING ) {

	nanosecs = sh4r.slice_cycle;

    }

    if( sh4r.sh4_state != SH4_STATE_STANDBY ) {

	TMU_run_slice( nanosecs );

	SCIF_run_slice( nanosecs );

    }

    return nanosecs;

}

/********************** SH4 emulation core  ****************************/

#if(SH4_CALLTRACE == 1)

#define MAX_CALLSTACK 32

static struct call_stack {

    sh4addr_t call_addr;

    sh4addr_t target_addr;

    sh4addr_t stack_pointer;

} call_stack[MAX_CALLSTACK];

static int call_stack_depth = 0;

int sh4_call_trace_on = 0;

static inline void trace_call( sh4addr_t source, sh4addr_t dest ) 

{

    if( call_stack_depth < MAX_CALLSTACK ) {

	call_stack[call_stack_depth].call_addr = source;

	call_stack[call_stack_depth].target_addr = dest;

	call_stack[call_stack_depth].stack_pointer = sh4r.r[15];

    }

    call_stack_depth++;

}

static inline void trace_return( sh4addr_t source, sh4addr_t dest )

{

    if( call_stack_depth > 0 ) {

	call_stack_depth--;

    }

}

void fprint_stack_trace( FILE *f )

{

    int i = call_stack_depth -1;

    if( i >= MAX_CALLSTACK )

	i = MAX_CALLSTACK - 1;

    for( ; i >= 0; i-- ) {

	fprintf( f, "%d. Call from %08X => %08X, SP=%08X\n", 

		 (call_stack_depth - i), call_stack[i].call_addr,

		 call_stack[i].target_addr, call_stack[i].stack_pointer );

    }

}

#define TRACE_CALL( source, dest ) trace_call(source, dest)

#define TRACE_RETURN( source, dest ) trace_return(source, dest)

#else

#define TRACE_CALL( dest, rts ) 

#define TRACE_RETURN( source, dest )

#endif

static gboolean FASTCALL sh4_raise_slot_exception( int normal_code, int slot_code ) {

    if( sh4r.in_delay_slot ) {

        sh4_raise_exception(slot_code);

    } else {

        sh4_raise_exception(normal_code);

    }

    return TRUE;

}

#define CHECKPRIV() if( !IS_SH4_PRIVMODE() ) { return sh4_raise_slot_exception( EXC_ILLEGAL, EXC_SLOT_ILLEGAL ); }

#define CHECKRALIGN16(addr) if( (addr)&0x01 ) { sh4_raise_exception( EXC_DATA_ADDR_READ ); return TRUE; }

#define CHECKRALIGN32(addr) if( (addr)&0x03 ) { sh4_raise_exception( EXC_DATA_ADDR_READ ); return TRUE; }

#define CHECKRALIGN64(addr) if( (addr)&0x07 ) { sh4_raise_exception( EXC_DATA_ADDR_READ ); return TRUE; }

#define CHECKWALIGN16(addr) if( (addr)&0x01 ) { sh4_raise_exception( EXC_DATA_ADDR_WRITE ); return TRUE; }

#define CHECKWALIGN32(addr) if( (addr)&0x03 ) { sh4_raise_exception( EXC_DATA_ADDR_WRITE ); return TRUE; }

#define CHECKWALIGN64(addr) if( (addr)&0x07 ) { sh4_raise_exception( EXC_DATA_ADDR_WRITE ); return TRUE; }

#define CHECKFPUEN() if( !IS_FPU_ENABLED() ) { if( ir == 0xFFFD ) { UNDEF(ir); } else { return sh4_raise_slot_exception( EXC_FPU_DISABLED, EXC_SLOT_FPU_DISABLED ); } }

#define CHECKDEST(p) if( (p) == 0 ) { ERROR( "%08X: Branch/jump to NULL, CPU halted", sh4r.pc ); sh4_core_exit(CORE_EXIT_HALT); return FALSE; }

#define CHECKSLOTILLEGAL() if(sh4r.in_delay_slot) { sh4_raise_exception(EXC_SLOT_ILLEGAL); return TRUE; }

#define ADDRSPACE (IS_SH4_PRIVMODE() ? sh4_address_space : sh4_user_address_space)

#define SQADDRSPACE (IS_SH4_PRIVMODE() ? storequeue_address_space : storequeue_user_address_space)

#ifdef HAVE_FRAME_ADDRESS

static FASTCALL __attribute__((noinline)) void *__first_arg(void *a, void *b) { return a; }

#define INIT_EXCEPTIONS(label) goto *__first_arg(&&fnstart,&&label); fnstart:

#define MEM_READ_BYTE( addr, val ) val = ((mem_read_exc_fn_t)ADDRSPACE[(addr)>>12]->read_byte)((addr), &&except)

#define MEM_READ_WORD( addr, val ) val = ((mem_read_exc_fn_t)ADDRSPACE[(addr)>>12]->read_word)((addr), &&except)

#define MEM_READ_LONG( addr, val ) val = ((mem_read_exc_fn_t)ADDRSPACE[(addr)>>12]->read_long)((addr), &&except)

#define MEM_WRITE_BYTE( addr, val ) ((mem_write_exc_fn_t)ADDRSPACE[(addr)>>12]->write_byte)((addr), (val), &&except)

#define MEM_WRITE_WORD( addr, val ) ((mem_write_exc_fn_t)ADDRSPACE[(addr)>>12]->write_word)((addr), (val), &&except)

#define MEM_WRITE_LONG( addr, val ) ((mem_write_exc_fn_t)ADDRSPACE[(addr)>>12]->write_long)((addr), (val), &&except)

#define MEM_PREFETCH( addr ) ((mem_prefetch_exc_fn_t)ADDRSPACE[(addr)>>12]->prefetch)((addr), &&except)

#else

#define INIT_EXCEPTIONS(label)

#define MEM_READ_BYTE( addr, val ) val = ADDRSPACE[(addr)>>12]->read_byte(addr)

#define MEM_READ_WORD( addr, val ) val = ADDRSPACE[(addr)>>12]->read_word(addr)

#define MEM_READ_LONG( addr, val ) val = ADDRSPACE[(addr)>>12]->read_long(addr)

#define MEM_WRITE_BYTE( addr, val ) ADDRSPACE[(addr)>>12]->write_byte(addr, val)

#define MEM_WRITE_WORD( addr, val ) ADDRSPACE[(addr)>>12]->write_word(addr, val)

#define MEM_WRITE_LONG( addr, val ) ADDRSPACE[(addr)>>12]->write_long(addr, val)

#define MEM_PREFETCH( addr ) ADDRSPACE[(addr)>>12]->prefetch(addr)

#endif

#define FP_WIDTH (IS_FPU_DOUBLESIZE() ? 8 : 4)

#define MEM_FP_READ( addr, reg ) \

    if( IS_FPU_DOUBLESIZE() ) { \

	CHECKRALIGN64(addr); \

        if( reg & 1 ) { \

            MEM_READ_LONG( addr, *((uint32_t *)&XF((reg) & 0x0E)) ); \

            MEM_READ_LONG( addr+4, *((uint32_t *)&XF(reg)) ); \

        } else { \

            MEM_READ_LONG( addr, *((uint32_t *)&FR(reg)) ); \

            MEM_READ_LONG( addr+4, *((uint32_t *)&FR((reg)|0x01)) ); \

	} \

    } else { \

        CHECKRALIGN32(addr); \

        MEM_READ_LONG( addr, *((uint32_t *)&FR(reg)) ); \

    }

#define MEM_FP_WRITE( addr, reg ) \

    if( IS_FPU_DOUBLESIZE() ) { \

        CHECKWALIGN64(addr); \

        if( reg & 1 ) { \

	    MEM_WRITE_LONG( addr, *((uint32_t *)&XF((reg)&0x0E)) ); \

	    MEM_WRITE_LONG( addr+4, *((uint32_t *)&XF(reg)) ); \

        } else { \

	    MEM_WRITE_LONG( addr, *((uint32_t *)&FR(reg)) ); \

	    MEM_WRITE_LONG( addr+4, *((uint32_t *)&FR((reg)|0x01)) ); \

	} \

    } else { \

    	CHECKWALIGN32(addr); \

        MEM_WRITE_LONG(addr, *((uint32_t *)&FR((reg))) ); \

    }

#define UNDEF(ir)

#define UNIMP(ir)

/**

 * Perform instruction-completion following core exit of a partially completed

 * instruction. NOTE: This is only allowed on memory writes, operation is not

 * guaranteed in any other case.

 */

void sh4_finalize_instruction( void )

{

    unsigned short ir;

    uint32_t tmp;

    if( IS_SYSCALL(sh4r.pc) ) {

        return;

    }

    assert( IS_IN_ICACHE(sh4r.pc) );

    ir = *(uint16_t *)GET_ICACHE_PTR(sh4r.pc);

    /**

     * Note - we can't take an exit on a control transfer instruction itself,

     * which means the exit must have happened in the delay slot. So for these

     * cases, finalize the delay slot instruction, and re-execute the control transfer.

     *

     * For delay slots which modify the argument used in the branch instruction,

     * we pretty much just assume that that can't have already happened in an exit case.

     */

%%

BRA disp {: 

    sh4r.pc += 2; 

    sh4_finalize_instruction(); 

    sh4r.pc += disp;

:}

BRAF Rn {: 

    sh4r.pc += 2; 

    tmp = sh4r.r[Rn];

    sh4_finalize_instruction(); 

    sh4r.pc += tmp;

:}

BSR disp {: 

    /* Note: PR is already set */ 

    sh4r.pc += 2;

    sh4_finalize_instruction();

    sh4r.pc += disp;

:}

BSRF Rn {:

    /* Note: PR is already set */ 

    sh4r.pc += 2;

    tmp = sh4r.r[Rn];

    sh4_finalize_instruction();

    sh4r.pc += tmp;

:}

BF/S disp {: 

    sh4r.pc += 2;

    sh4_finalize_instruction();

    if( !sh4r.t ) {

        sh4r.pc += disp;

    }

:}

BT/S disp {: 

    sh4r.pc += 2;

    sh4_finalize_instruction();

    if( sh4r.t ) {

        sh4r.pc += disp;

    }

:}

JMP @Rn {:

    sh4r.pc += 2;

    tmp = sh4r.r[Rn];

    sh4_finalize_instruction();

    sh4r.pc = tmp;

    sh4r.new_pc = tmp + 2;

    sh4r.slice_cycle += sh4_cpu_period;

    return;

:}

JSR @Rn {: 

    /* Note: PR is already set */ 

    sh4r.pc += 2;

    tmp = sh4r.r[Rn];

    sh4_finalize_instruction();

    sh4r.pc = tmp;

    sh4r.new_pc = tmp + 2;

    sh4r.slice_cycle += sh4_cpu_period;

    return;

:}

RTS {: 

    sh4r.pc += 2;

    sh4_finalize_instruction();

    sh4r.pc = sh4r.pr;

    sh4r.new_pc = sh4r.pr + 2;

    sh4r.slice_cycle += sh4_cpu_period;

    return;

:}

RTE {: 

    /* SR is already set */

    sh4r.pc += 2;

    sh4_finalize_instruction();

    sh4r.pc = sh4r.spc;

    sh4r.new_pc = sh4r.pr + 2;

    sh4r.slice_cycle += sh4_cpu_period;

    return;

:}

MOV.B Rm, @-Rn {: sh4r.r[Rn]--; :}

MOV.W Rm, @-Rn {: sh4r.r[Rn] -= 2; :}

MOV.L Rm, @-Rn {: sh4r.r[Rn] -= 4; :}

MOV.B @Rm+, Rn {: if( Rm != Rn ) { sh4r.r[Rm] ++;  } :}

MOV.W @Rm+, Rn {: if( Rm != Rn ) { sh4r.r[Rm] += 2; } :}

MOV.L @Rm+, Rn {: if( Rm != Rn ) { sh4r.r[Rm] += 4; } :}

%%

    sh4r.in_delay_slot = 0;

    sh4r.pc += 2;

    sh4r.new_pc = sh4r.pc+2;

    sh4r.slice_cycle += sh4_cpu_period;

}

#undef UNDEF

#undef UNIMP

#define UNDEF(ir) return sh4_raise_slot_exception(EXC_ILLEGAL, EXC_SLOT_ILLEGAL)

#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)

gboolean sh4_execute_instruction( void )

{

    uint32_t pc;

    unsigned short ir;

    uint32_t tmp;

    float ftmp;

    double dtmp;

    int64_t memtmp; // temporary holder for memory reads

    INIT_EXCEPTIONS(except)

#define R0 sh4r.r[0]

    pc = sh4r.pc;

    if( pc > 0xFFFFFF00 ) {

	/* SYSCALL Magic */

	syscall_invoke( pc );

	sh4r.in_delay_slot = 0;

	pc = sh4r.pc = sh4r.pr;

	sh4r.new_pc = sh4r.pc + 2;

        return TRUE;

    }

    CHECKRALIGN16(pc);

#ifdef ENABLE_SH4STATS

    sh4_stats_add_by_pc(sh4r.pc);

#endif

    /* Read instruction */

    if( !IS_IN_ICACHE(pc) ) {

        gboolean delay_slot = sh4r.in_delay_slot;

	if( !mmu_update_icache(pc) ) {

	    if( delay_slot ) {

	        sh4r.spc -= 2;

	    }

	    // Fault - look for the fault handler

	    if( !mmu_update_icache(sh4r.pc) ) {

		// double fault - halt

		ERROR( "Double fault - halting" );

		sh4_core_exit(CORE_EXIT_HALT);

		return FALSE;

	    }

	}

	pc = sh4r.pc;

    }

    assert( IS_IN_ICACHE(pc) );

    ir = *(uint16_t *)GET_ICACHE_PTR(sh4r.pc);

    /* FIXME: This is a bit of a hack, but the PC of the delay slot should not

     * be visible until after the instruction has executed (for exception 

     * correctness)

     */

    if( sh4r.in_delay_slot ) {

    	sh4r.pc -= 2;

    }

%%

AND Rm, Rn {: sh4r.r[Rn] &= sh4r.r[Rm]; :}

AND #imm, R0 {: R0 &= imm; :}

 AND.B #imm, @(R0, GBR) {: MEM_READ_BYTE(R0+sh4r.gbr, tmp); MEM_WRITE_BYTE( R0 + sh4r.gbr, imm & tmp ); :}

NOT Rm, Rn {: sh4r.r[Rn] = ~sh4r.r[Rm]; :}

OR Rm, Rn {: sh4r.r[Rn] |= sh4r.r[Rm]; :}

OR #imm, R0  {: R0 |= imm; :}

 OR.B #imm, @(R0, GBR) {: MEM_READ_BYTE(R0+sh4r.gbr, tmp); MEM_WRITE_BYTE( R0 + sh4r.gbr, imm | tmp ); :}

TAS.B @Rn {:

    MEM_READ_BYTE( sh4r.r[Rn], tmp );

    sh4r.t = ( tmp == 0 ? 1 : 0 );

    MEM_WRITE_BYTE( sh4r.r[Rn], tmp | 0x80 );

:}

TST Rm, Rn {: sh4r.t = (sh4r.r[Rn]&sh4r.r[Rm] ? 0 : 1); :}

TST #imm, R0 {: sh4r.t = (R0 & imm ? 0 : 1); :}

 TST.B #imm, @(R0, GBR) {: MEM_READ_BYTE(R0+sh4r.gbr, tmp); sh4r.t = ( tmp & imm ? 0 : 1 ); :}

XOR Rm, Rn {: sh4r.r[Rn] ^= sh4r.r[Rm]; :}

XOR #imm, R0 {: R0 ^= imm; :}

 XOR.B #imm, @(R0, GBR) {: MEM_READ_BYTE(R0+sh4r.gbr, tmp); MEM_WRITE_BYTE( R0 + sh4r.gbr, imm ^ tmp ); :}

XTRCT Rm, Rn {: sh4r.r[Rn] = (sh4r.r[Rn]>>16) | (sh4r.r[Rm]<<16); :}

ROTL Rn {:

    sh4r.t = sh4r.r[Rn] >> 31;

    sh4r.r[Rn] <<= 1;

    sh4r.r[Rn] |= sh4r.t;

:}

ROTR Rn {:

    sh4r.t = sh4r.r[Rn] & 0x00000001;

    sh4r.r[Rn] >>= 1;

    sh4r.r[Rn] |= (sh4r.t << 31);

:}

ROTCL Rn {:

    tmp = sh4r.r[Rn] >> 31;

    sh4r.r[Rn] <<= 1;

    sh4r.r[Rn] |= sh4r.t;

    sh4r.t = tmp;

:}

ROTCR Rn {:

    tmp = sh4r.r[Rn] & 0x00000001;

    sh4r.r[Rn] >>= 1;

    sh4r.r[Rn] |= (sh4r.t << 31 );

    sh4r.t = tmp;

:}

SHAD Rm, Rn {:

    tmp = sh4r.r[Rm];

    if( (tmp & 0x80000000) == 0 ) sh4r.r[Rn] <<= (tmp&0x1f);

    else if( (tmp & 0x1F) == 0 )  

        sh4r.r[Rn] = ((int32_t)sh4r.r[Rn]) >> 31;

    else 

	sh4r.r[Rn] = ((int32_t)sh4r.r[Rn]) >> (((~sh4r.r[Rm]) & 0x1F)+1);

:}

SHLD Rm, Rn {:

    tmp = sh4r.r[Rm];

    if( (tmp & 0x80000000) == 0 ) sh4r.r[Rn] <<= (tmp&0x1f);

    else if( (tmp & 0x1F) == 0 ) sh4r.r[Rn] = 0;

    else sh4r.r[Rn] >>= (((~tmp) & 0x1F)+1);

:}

SHAL Rn {:

    sh4r.t = sh4r.r[Rn] >> 31;

    sh4r.r[Rn] <<= 1;

:}

SHAR Rn {:

    sh4r.t = sh4r.r[Rn] & 0x00000001;

    sh4r.r[Rn] = ((int32_t)sh4r.r[Rn]) >> 1;

:}

SHLL Rn {: sh4r.t = sh4r.r[Rn] >> 31; sh4r.r[Rn] <<= 1; :}

SHLR Rn {: sh4r.t = sh4r.r[Rn] & 0x00000001; sh4r.r[Rn] >>= 1; :}

SHLL2 Rn {: sh4r.r[Rn] <<= 2; :}

SHLR2 Rn {: sh4r.r[Rn] >>= 2; :}

SHLL8 Rn {: sh4r.r[Rn] <<= 8; :}

SHLR8 Rn {: sh4r.r[Rn] >>= 8; :}

SHLL16 Rn {: sh4r.r[Rn] <<= 16; :}

SHLR16 Rn {: sh4r.r[Rn] >>= 16; :}

EXTU.B Rm, Rn {: sh4r.r[Rn] = sh4r.r[Rm]&0x000000FF; :}

EXTU.W Rm, Rn {: sh4r.r[Rn] = sh4r.r[Rm]&0x0000FFFF; :}

EXTS.B Rm, Rn {: sh4r.r[Rn] = SIGNEXT8( sh4r.r[Rm]&0x000000FF ); :}

EXTS.W Rm, Rn {: sh4r.r[Rn] = SIGNEXT16( sh4r.r[Rm]&0x0000FFFF ); :}

SWAP.B Rm, Rn {: sh4r.r[Rn] = (sh4r.r[Rm]&0xFFFF0000) | ((sh4r.r[Rm]&0x0000FF00)>>8) | ((sh4r.r[Rm]&0x000000FF)<<8); :}

SWAP.W Rm, Rn {: sh4r.r[Rn] = (sh4r.r[Rm]>>16) | (sh4r.r[Rm]<<16); :}

CLRT {: sh4r.t = 0; :}

SETT {: sh4r.t = 1; :}

CLRMAC {: sh4r.mac = 0; :}

LDTLB {: MMU_ldtlb(); :}

CLRS {: sh4r.s = 0; :}

SETS {: sh4r.s = 1; :}

MOVT Rn {: sh4r.r[Rn] = sh4r.t; :}

NOP {: /* NOP */ :}

PREF @Rn {:

    MEM_PREFETCH(sh4r.r[Rn]);

:}

OCBI @Rn {: :}

OCBP @Rn {: :}

OCBWB @Rn {: :}

MOVCA.L R0, @Rn {:

    tmp = sh4r.r[Rn];

    CHECKWALIGN32(tmp);

    MEM_WRITE_LONG( tmp, R0 );

:}

MOV.B Rm, @(R0, Rn) {: MEM_WRITE_BYTE( R0 + sh4r.r[Rn], sh4r.r[Rm] ); :}

MOV.W Rm, @(R0, Rn) {: 

    CHECKWALIGN16( R0 + sh4r.r[Rn] );

    MEM_WRITE_WORD( R0 + sh4r.r[Rn], sh4r.r[Rm] );

:}

MOV.L Rm, @(R0, Rn) {:

    CHECKWALIGN32( R0 + sh4r.r[Rn] );

    MEM_WRITE_LONG( R0 + sh4r.r[Rn], sh4r.r[Rm] );

:}

MOV.B @(R0, Rm), Rn {: MEM_READ_BYTE( R0 + sh4r.r[Rm], sh4r.r[Rn] ); :}

MOV.W @(R0, Rm), Rn {: CHECKRALIGN16( R0 + sh4r.r[Rm] );

    MEM_READ_WORD( R0 + sh4r.r[Rm], sh4r.r[Rn] );

:}

MOV.L @(R0, Rm), Rn {: CHECKRALIGN32( R0 + sh4r.r[Rm] );

    MEM_READ_LONG( R0 + sh4r.r[Rm], sh4r.r[Rn] );

:}

MOV.L Rm, @(disp, Rn) {:

    tmp = sh4r.r[Rn] + disp;

    CHECKWALIGN32( tmp );

    MEM_WRITE_LONG( tmp, sh4r.r[Rm] );

:}

MOV.B Rm, @Rn {: MEM_WRITE_BYTE( sh4r.r[Rn], sh4r.r[Rm] ); :}

MOV.W Rm, @Rn {: CHECKWALIGN16( sh4r.r[Rn] ); MEM_WRITE_WORD( sh4r.r[Rn], sh4r.r[Rm] ); :}

MOV.L Rm, @Rn {: CHECKWALIGN32( sh4r.r[Rn] ); MEM_WRITE_LONG( sh4r.r[Rn], sh4r.r[Rm] ); :}

 MOV.B Rm, @-Rn {: MEM_WRITE_BYTE( sh4r.r[Rn]-1, sh4r.r[Rm] ); sh4r.r[Rn]--; :}

 MOV.W Rm, @-Rn {: CHECKWALIGN16( sh4r.r[Rn] ); MEM_WRITE_WORD( sh4r.r[Rn]-2, sh4r.r[Rm] ); sh4r.r[Rn] -= 2; :}

 MOV.L Rm, @-Rn {: CHECKWALIGN32( sh4r.r[Rn] ); MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.r[Rm] ); sh4r.r[Rn] -= 4; :}

MOV.L @(disp, Rm), Rn {:

    tmp = sh4r.r[Rm] + disp;

    CHECKRALIGN32( tmp );

    MEM_READ_LONG( tmp, sh4r.r[Rn] );

:}

MOV.B @Rm, Rn {: MEM_READ_BYTE( sh4r.r[Rm], sh4r.r[Rn] ); :}

 MOV.W @Rm, Rn {: CHECKRALIGN16( sh4r.r[Rm] ); MEM_READ_WORD( sh4r.r[Rm], sh4r.r[Rn] ); :}

 MOV.L @Rm, Rn {: CHECKRALIGN32( sh4r.r[Rm] ); MEM_READ_LONG( sh4r.r[Rm], sh4r.r[Rn] ); :}

MOV Rm, Rn {: sh4r.r[Rn] = sh4r.r[Rm]; :}

 MOV.B @Rm+, Rn {: MEM_READ_BYTE( sh4r.r[Rm], sh4r.r[Rn] ); if( Rm != Rn ) { sh4r.r[Rm] ++; } :}

 MOV.W @Rm+, Rn {: CHECKRALIGN16( sh4r.r[Rm] ); MEM_READ_WORD( sh4r.r[Rm], sh4r.r[Rn] ); if( Rm != Rn ) { sh4r.r[Rm] += 2; } :}

 MOV.L @Rm+, Rn {: CHECKRALIGN32( sh4r.r[Rm] ); MEM_READ_LONG( sh4r.r[Rm], sh4r.r[Rn] ); if( Rm != Rn ) { sh4r.r[Rm] += 4; } :}

MOV.L @(disp, PC), Rn {:

    CHECKSLOTILLEGAL();

    tmp = (pc&0xFFFFFFFC) + disp + 4;

    MEM_READ_LONG( tmp, sh4r.r[Rn] );

:}

MOV.B R0, @(disp, GBR) {: MEM_WRITE_BYTE( sh4r.gbr + disp, R0 ); :}

MOV.W R0, @(disp, GBR) {:

    tmp = sh4r.gbr + disp;

    CHECKWALIGN16( tmp );

    MEM_WRITE_WORD( tmp, R0 );

:}

MOV.L R0, @(disp, GBR) {:

    tmp = sh4r.gbr + disp;

    CHECKWALIGN32( tmp );

    MEM_WRITE_LONG( tmp, R0 );

:}

 MOV.B @(disp, GBR), R0 {: MEM_READ_BYTE( sh4r.gbr + disp, R0 ); :}

MOV.W @(disp, GBR), R0 {: 

    tmp = sh4r.gbr + disp;

    CHECKRALIGN16( tmp );

    MEM_READ_WORD( tmp, R0 );

:}

MOV.L @(disp, GBR), R0 {:

    tmp = sh4r.gbr + disp;

    CHECKRALIGN32( tmp );

    MEM_READ_LONG( tmp, R0 );

:}

MOV.B R0, @(disp, Rn) {: MEM_WRITE_BYTE( sh4r.r[Rn] + disp, R0 ); :}

MOV.W R0, @(disp, Rn) {: 

    tmp = sh4r.r[Rn] + disp;

    CHECKWALIGN16( tmp );

    MEM_WRITE_WORD( tmp, R0 );

:}

 MOV.B @(disp, Rm), R0 {: MEM_READ_BYTE( sh4r.r[Rm] + disp, R0 ); :}

MOV.W @(disp, Rm), R0 {: 

    tmp = sh4r.r[Rm] + disp;

    CHECKRALIGN16( tmp );

    MEM_READ_WORD( tmp, R0 );

:}

MOV.W @(disp, PC), Rn {:

    CHECKSLOTILLEGAL();

    tmp = pc + 4 + disp;

    MEM_READ_WORD( tmp, sh4r.r[Rn] );

:}

MOVA @(disp, PC), R0 {:

    CHECKSLOTILLEGAL();

    R0 = (pc&0xFFFFFFFC) + disp + 4;

:}

MOV #imm, Rn {:  sh4r.r[Rn] = imm; :}

FMOV @(R0, Rm), FRn {: MEM_FP_READ( sh4r.r[Rm] + R0, FRn ); :}

FMOV FRm, @(R0, Rn) {: MEM_FP_WRITE( sh4r.r[Rn] + R0, FRm ); :}

FMOV @Rm, FRn {: MEM_FP_READ( sh4r.r[Rm], FRn ); :}

FMOV @Rm+, FRn {: MEM_FP_READ( sh4r.r[Rm], FRn ); sh4r.r[Rm] += FP_WIDTH; :}

FMOV FRm, @Rn {: MEM_FP_WRITE( sh4r.r[Rn], FRm ); :}

 FMOV FRm, @-Rn {: MEM_FP_WRITE( sh4r.r[Rn] - FP_WIDTH, FRm ); sh4r.r[Rn] -= FP_WIDTH; :}

FMOV FRm, FRn {: 

    if( IS_FPU_DOUBLESIZE() )

	DR(FRn) = DR(FRm);

    else

	FR(FRn) = FR(FRm);

:}

CMP/EQ #imm, R0 {: sh4r.t = ( R0 == imm ? 1 : 0 ); :}

CMP/EQ Rm, Rn {: sh4r.t = ( sh4r.r[Rm] == sh4r.r[Rn] ? 1 : 0 ); :}

CMP/GE Rm, Rn {: sh4r.t = ( ((int32_t)sh4r.r[Rn]) >= ((int32_t)sh4r.r[Rm]) ? 1 : 0 ); :}

CMP/GT Rm, Rn {: sh4r.t = ( ((int32_t)sh4r.r[Rn]) > ((int32_t)sh4r.r[Rm]) ? 1 : 0 ); :}

CMP/HI Rm, Rn {: sh4r.t = ( sh4r.r[Rn] > sh4r.r[Rm] ? 1 : 0 ); :}

CMP/HS Rm, Rn {: sh4r.t = ( sh4r.r[Rn] >= sh4r.r[Rm] ? 1 : 0 ); :}

CMP/PL Rn {: sh4r.t = ( ((int32_t)sh4r.r[Rn]) > 0 ? 1 : 0 ); :}

CMP/PZ Rn {: sh4r.t = ( ((int32_t)sh4r.r[Rn]) >= 0 ? 1 : 0 ); :}

CMP/STR Rm, Rn {: 

    /* set T = 1 if any byte in RM & RN is the same */

    tmp = sh4r.r[Rm] ^ sh4r.r[Rn];

    sh4r.t = ((tmp&0x000000FF)==0 || (tmp&0x0000FF00)==0 ||

             (tmp&0x00FF0000)==0 || (tmp&0xFF000000)==0)?1:0;

:}

ADD Rm, Rn {: sh4r.r[Rn] += sh4r.r[Rm]; :}

ADD #imm, Rn {: sh4r.r[Rn] += imm; :}

ADDC Rm, Rn {:

    tmp = sh4r.r[Rn];

    sh4r.r[Rn] += sh4r.r[Rm] + sh4r.t;

    sh4r.t = ( sh4r.r[Rn] < tmp || (sh4r.r[Rn] == tmp && sh4r.t != 0) ? 1 : 0 );

:}

ADDV Rm, Rn {:

    tmp = sh4r.r[Rn] + sh4r.r[Rm];

    sh4r.t = ( (sh4r.r[Rn]>>31) == (sh4r.r[Rm]>>31) && ((sh4r.r[Rn]>>31) != (tmp>>31)) );

    sh4r.r[Rn] = tmp;

:}

DIV0U {: sh4r.m = sh4r.q = sh4r.t = 0; :}

DIV0S Rm, Rn {: 

    sh4r.q = sh4r.r[Rn]>>31;

    sh4r.m = sh4r.r[Rm]>>31;

    sh4r.t = sh4r.q ^ sh4r.m;

:}

DIV1 Rm, Rn {:

    /* This is derived from the sh4 manual with some simplifications */

    uint32_t tmp0, tmp1, tmp2, dir;

    dir = sh4r.q ^ sh4r.m;

    sh4r.q = (sh4r.r[Rn] >> 31);

    tmp2 = sh4r.r[Rm];

    sh4r.r[Rn] = (sh4r.r[Rn] << 1) | sh4r.t;

    tmp0 = sh4r.r[Rn];

    if( dir ) {

         sh4r.r[Rn] += tmp2;

         tmp1 = (sh4r.r[Rn]<tmp0 ? 1 : 0 );

    } else {

         sh4r.r[Rn] -= tmp2;

         tmp1 = (sh4r.r[Rn]>tmp0 ? 1 : 0 );

    }

    sh4r.q ^= sh4r.m ^ tmp1;

    sh4r.t = ( sh4r.q == sh4r.m ? 1 : 0 );

:}

DMULS.L Rm, Rn {: sh4r.mac = SIGNEXT32(sh4r.r[Rm]) * SIGNEXT32(sh4r.r[Rn]); :}

DMULU.L Rm, Rn {: sh4r.mac = ((uint64_t)sh4r.r[Rm]) * ((uint64_t)sh4r.r[Rn]); :}

DT Rn {:

    sh4r.r[Rn] --;

    sh4r.t = ( sh4r.r[Rn] == 0 ? 1 : 0 );

:}

MAC.W @Rm+, @Rn+ {:

    int32_t stmp;

    if( Rm == Rn ) {

	CHECKRALIGN16(sh4r.r[Rn]);

	MEM_READ_WORD( sh4r.r[Rn], tmp );

	stmp = SIGNEXT16(tmp);

	MEM_READ_WORD( sh4r.r[Rn]+2, tmp );

	stmp *= SIGNEXT16(tmp);

	sh4r.r[Rn] += 4;

    } else {

	CHECKRALIGN16( sh4r.r[Rn] );

	CHECKRALIGN16( sh4r.r[Rm] );

	MEM_READ_WORD(sh4r.r[Rn], tmp);

	stmp = SIGNEXT16(tmp);

	MEM_READ_WORD(sh4r.r[Rm], tmp);

	stmp = stmp * SIGNEXT16(tmp);

	sh4r.r[Rn] += 2;

	sh4r.r[Rm] += 2;

    }

    if( sh4r.s ) {

	int64_t tmpl = (int64_t)((int32_t)sh4r.mac) + (int64_t)stmp;

	if( tmpl > (int64_t)0x000000007FFFFFFFLL ) {

	    sh4r.mac = 0x000000017FFFFFFFLL;

	} else if( tmpl < (int64_t)0xFFFFFFFF80000000LL ) {

	    sh4r.mac = 0x0000000180000000LL;

	} else {

	    sh4r.mac = (sh4r.mac & 0xFFFFFFFF00000000LL) |

		((uint32_t)(sh4r.mac + stmp));

	}

    } else {

	sh4r.mac += SIGNEXT32(stmp);

    }

:}

MAC.L @Rm+, @Rn+ {:

    int64_t tmpl;

    if( Rm == Rn ) {

	CHECKRALIGN32( sh4r.r[Rn] );

	MEM_READ_LONG(sh4r.r[Rn], tmp);

	tmpl = SIGNEXT32(tmp);

	MEM_READ_LONG(sh4r.r[Rn]+4, tmp);

	tmpl = tmpl * SIGNEXT32(tmp) + sh4r.mac;

	sh4r.r[Rn] += 8;

    } else {

	CHECKRALIGN32( sh4r.r[Rm] );

	CHECKRALIGN32( sh4r.r[Rn] );

	MEM_READ_LONG(sh4r.r[Rn], tmp);

	tmpl = SIGNEXT32(tmp);

	MEM_READ_LONG(sh4r.r[Rm], tmp);

	tmpl = tmpl * SIGNEXT32(tmp) + sh4r.mac;

	sh4r.r[Rn] += 4;

	sh4r.r[Rm] += 4;

    }

    if( sh4r.s ) {

        /* 48-bit Saturation. Yuch */

        if( tmpl < (int64_t)0xFFFF800000000000LL )

            tmpl = 0xFFFF800000000000LL;

        else if( tmpl > (int64_t)0x00007FFFFFFFFFFFLL )

            tmpl = 0x00007FFFFFFFFFFFLL;

    }

    sh4r.mac = tmpl;

:}

MUL.L Rm, Rn {: sh4r.mac = (sh4r.mac&0xFFFFFFFF00000000LL) |

                        (sh4r.r[Rm] * sh4r.r[Rn]); :}

MULU.W Rm, Rn {:

    sh4r.mac = (sh4r.mac&0xFFFFFFFF00000000LL) |

               (uint32_t)((sh4r.r[Rm]&0xFFFF) * (sh4r.r[Rn]&0xFFFF));

:}

MULS.W Rm, Rn {:

    sh4r.mac = (sh4r.mac&0xFFFFFFFF00000000LL) |

               (uint32_t)(SIGNEXT32(sh4r.r[Rm]&0xFFFF) * SIGNEXT32(sh4r.r[Rn]&0xFFFF));

:}

NEGC Rm, Rn {:

    tmp = 0 - sh4r.r[Rm];

    sh4r.r[Rn] = tmp - sh4r.t;

    sh4r.t = ( 0<tmp || tmp<sh4r.r[Rn] ? 1 : 0 );

:}

NEG Rm, Rn {: sh4r.r[Rn] = 0 - sh4r.r[Rm]; :}

SUB Rm, Rn {: sh4r.r[Rn] -= sh4r.r[Rm]; :}

SUBC Rm, Rn {: 

    tmp = sh4r.r[Rn];

    sh4r.r[Rn] = sh4r.r[Rn] - sh4r.r[Rm] - sh4r.t;

    sh4r.t = (sh4r.r[Rn] > tmp || (sh4r.r[Rn] == tmp && sh4r.t == 1));

:}

SUBV Rm, Rn {:

    tmp = sh4r.r[Rn] - sh4r.r[Rm];

    sh4r.t = ( (sh4r.r[Rn]>>31) != (sh4r.r[Rm]>>31) && ((sh4r.r[Rn]>>31) != (tmp>>31)) );

    sh4r.r[Rn] = tmp;

:}

BRAF Rn {:

     CHECKSLOTILLEGAL();

     CHECKDEST( pc + 4 + sh4r.r[Rn] );

     sh4r.in_delay_slot = 1;

     sh4r.pc = sh4r.new_pc;

     sh4r.new_pc = pc + 4 + sh4r.r[Rn];

     return TRUE;

:}

BSRF Rn {:

     CHECKSLOTILLEGAL();

     CHECKDEST( pc + 4 + sh4r.r[Rn] );

     sh4r.in_delay_slot = 1;

     sh4r.pr = sh4r.pc + 4;

     sh4r.pc = sh4r.new_pc;

     sh4r.new_pc = pc + 4 + sh4r.r[Rn];

     TRACE_CALL( pc, sh4r.new_pc );

     return TRUE;

:}

BT disp {:

    CHECKSLOTILLEGAL();

    if( sh4r.t ) {

        CHECKDEST( sh4r.pc + disp + 4 )

        sh4r.pc += disp + 4;

        sh4r.new_pc = sh4r.pc + 2;

        return TRUE;

    }

:}

BF disp {:

    CHECKSLOTILLEGAL();

    if( !sh4r.t ) {

        CHECKDEST( sh4r.pc + disp + 4 )

        sh4r.pc += disp + 4;

        sh4r.new_pc = sh4r.pc + 2;

        return TRUE;

    }

:}

BT/S disp {:

    CHECKSLOTILLEGAL();

    if( sh4r.t ) {

        CHECKDEST( sh4r.pc + disp + 4 )

        sh4r.in_delay_slot = 1;

        sh4r.pc = sh4r.new_pc;

        sh4r.new_pc = pc + disp + 4;

        sh4r.in_delay_slot = 1;

        return TRUE;

    }

:}

BF/S disp {:

    CHECKSLOTILLEGAL();

    if( !sh4r.t ) {

        CHECKDEST( sh4r.pc + disp + 4 )

        sh4r.in_delay_slot = 1;

        sh4r.pc = sh4r.new_pc;

        sh4r.new_pc = pc + disp + 4;

        return TRUE;

    }

:}

BRA disp {:

    CHECKSLOTILLEGAL();

    CHECKDEST( sh4r.pc + disp + 4 );

    sh4r.in_delay_slot = 1;

    sh4r.pc = sh4r.new_pc;

    sh4r.new_pc = pc + 4 + disp;

    return TRUE;

:}

BSR disp {:

    CHECKDEST( sh4r.pc + disp + 4 );

    CHECKSLOTILLEGAL();

    sh4r.in_delay_slot = 1;

    sh4r.pr = pc + 4;

    sh4r.pc = sh4r.new_pc;

    sh4r.new_pc = pc + 4 + disp;

    TRACE_CALL( pc, sh4r.new_pc );

    return TRUE;

:}

TRAPA #imm {:

    CHECKSLOTILLEGAL();

    sh4r.pc += 2;

    sh4_raise_trap( imm );

    return TRUE;

:}

RTS {: 

    CHECKSLOTILLEGAL();

    CHECKDEST( sh4r.pr );

    sh4r.in_delay_slot = 1;

    sh4r.pc = sh4r.new_pc;

    sh4r.new_pc = sh4r.pr;

    TRACE_RETURN( pc, sh4r.new_pc );

    return TRUE;

:}

SLEEP {:

    if( MMIO_READ( CPG, STBCR ) & 0x80 ) {

	sh4r.sh4_state = SH4_STATE_STANDBY;

    } else {

	sh4r.sh4_state = SH4_STATE_SLEEP;

    }

    return FALSE; /* Halt CPU */

:}

RTE {:

    CHECKPRIV();

    CHECKDEST( sh4r.spc );

    CHECKSLOTILLEGAL();

    sh4r.in_delay_slot = 1;

    sh4r.pc = sh4r.new_pc;

    sh4r.new_pc = sh4r.spc;

    sh4_write_sr( sh4r.ssr );

    return TRUE;

:}

JMP @Rn {:

    CHECKDEST( sh4r.r[Rn] );

    CHECKSLOTILLEGAL();

    sh4r.in_delay_slot = 1;

    sh4r.pc = sh4r.new_pc;

    sh4r.new_pc = sh4r.r[Rn];

    return TRUE;

:}

JSR @Rn {:

    CHECKDEST( sh4r.r[Rn] );

    CHECKSLOTILLEGAL();

    sh4r.in_delay_slot = 1;

    sh4r.pc = sh4r.new_pc;

    sh4r.new_pc = sh4r.r[Rn];

    sh4r.pr = pc + 4;

    TRACE_CALL( pc, sh4r.new_pc );

    return TRUE;

:}

STS MACH, Rn {: sh4r.r[Rn] = (sh4r.mac>>32); :}

STS.L MACH, @-Rn {:

    CHECKWALIGN32( sh4r.r[Rn] );

    MEM_WRITE_LONG( sh4r.r[Rn]-4, (sh4r.mac>>32) );

    sh4r.r[Rn] -= 4;

:}

STC.L SR, @-Rn {:

    CHECKPRIV();

    CHECKWALIGN32( sh4r.r[Rn] );

    MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4_read_sr() );

    sh4r.r[Rn] -= 4;

:}

LDS.L @Rm+, MACH {:

    CHECKRALIGN32( sh4r.r[Rm] );

    MEM_READ_LONG(sh4r.r[Rm], tmp);

    sh4r.mac = (sh4r.mac & 0x00000000FFFFFFFF) |

	(((uint64_t)tmp)<<32);

    sh4r.r[Rm] += 4;

:}

LDC.L @Rm+, SR {:

    CHECKSLOTILLEGAL();

    CHECKPRIV();

    CHECKWALIGN32( sh4r.r[Rm] );

    MEM_READ_LONG(sh4r.r[Rm], tmp);

    sh4_write_sr( tmp );

    sh4r.r[Rm] +=4;

:}

LDS Rm, MACH {:

    sh4r.mac = (sh4r.mac & 0x00000000FFFFFFFF) |

               (((uint64_t)sh4r.r[Rm])<<32);

:}

LDC Rm, SR {:

    CHECKSLOTILLEGAL();

    CHECKPRIV();

    sh4_write_sr( sh4r.r[Rm] );

:}

LDC Rm, SGR {:

    CHECKPRIV();

    sh4r.sgr = sh4r.r[Rm];

:}

LDC.L @Rm+, SGR {:

    CHECKPRIV();

    CHECKRALIGN32( sh4r.r[Rm] );

    MEM_READ_LONG(sh4r.r[Rm], sh4r.sgr);

    sh4r.r[Rm] +=4;

:}

STS MACL, Rn {: sh4r.r[Rn] = (uint32_t)sh4r.mac; :}

STS.L MACL, @-Rn {:

    CHECKWALIGN32( sh4r.r[Rn] );

    MEM_WRITE_LONG( sh4r.r[Rn]-4, (uint32_t)sh4r.mac );

    sh4r.r[Rn] -= 4;

:}

STC.L GBR, @-Rn {:

    CHECKWALIGN32( sh4r.r[Rn] );

    MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.gbr );

    sh4r.r[Rn] -= 4;

:}

LDS.L @Rm+, MACL {:

    CHECKRALIGN32( sh4r.r[Rm] );

    MEM_READ_LONG(sh4r.r[Rm], tmp);

    sh4r.mac = (sh4r.mac & 0xFFFFFFFF00000000LL) |

               (uint64_t)((uint32_t)tmp);

    sh4r.r[Rm] += 4;

:}

LDC.L @Rm+, GBR {:

    CHECKRALIGN32( sh4r.r[Rm] );

    MEM_READ_LONG(sh4r.r[Rm], sh4r.gbr);

    sh4r.r[Rm] +=4;

:}

LDS Rm, MACL {:

    sh4r.mac = (sh4r.mac & 0xFFFFFFFF00000000LL) |

               (uint64_t)((uint32_t)(sh4r.r[Rm]));

:}

LDC Rm, GBR {: sh4r.gbr = sh4r.r[Rm]; :}

STS PR, Rn {: sh4r.r[Rn] = sh4r.pr; :}

STS.L PR, @-Rn {:

    CHECKWALIGN32( sh4r.r[Rn] );

    MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.pr );

    sh4r.r[Rn] -= 4;

:}

STC.L VBR, @-Rn {:

    CHECKPRIV();

    CHECKWALIGN32( sh4r.r[Rn] );

    MEM_WRITE_LONG( sh4r.r[Rn]-4, sh4r.vbr );

    sh4r.r[Rn] -= 4;

:}

LDS.L @Rm+, PR {:

    CHECKRALIGN32( sh4r.r[Rm] );

    MEM_READ_LONG( sh4r.r[Rm], sh4r.pr );

    sh4r.r[Rm] += 4;

:}

LDC.L @Rm+, VBR {:

    CHECKPRIV();

    CHECKRALIGN32( sh4r.r[Rm] );

    MEM_READ_LONG(sh4r.r[Rm], sh4r.vbr);

    sh4r.r[Rm] +=4;

:}

LDS Rm, PR {: sh4r.pr = sh4r.r[Rm]; :}

LDC Rm, VBR {:

    CHECKPRIV();

    sh4r.vbr = sh4r.r[Rm];

:}

STC SGR, Rn {:

    CHECKPRIV();

    sh4r.r[Rn] = sh4r.sgr;

:}