/**

  * $Id$

  * 

  * SH4 translation core module. This part handles the non-target-specific

  * section of the translation.

  *

  * 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.

  */

 #include <assert.h>

 #include <setjmp.h>

 #include "eventq.h"

 #include "syscall.h"

 #include "clock.h"

 #include "dreamcast.h"

 #include "sh4/sh4core.h"

 #include "sh4/sh4trans.h"

 #include "sh4/xltcache.h"

 static jmp_buf xlat_jmp_buf;

 static gboolean xlat_running = FALSE;

 gboolean sh4_xlat_is_running()

 {

     return xlat_running;

 }

 /**

  * Execute a timeslice using translated code only (ie translate/execute loop)

  */

 uint32_t sh4_xlat_run_slice( uint32_t nanosecs ) 

 {

     sh4r.slice_cycle = 0;

     if( sh4r.sh4_state != SH4_STATE_RUNNING ) {

         sh4_sleep_run_slice(nanosecs);

     }

     switch( setjmp(xlat_jmp_buf) ) {

     case XLAT_EXIT_BREAKPOINT:

         sh4_clear_breakpoint( sh4r.pc, BREAK_ONESHOT );

         /* fallthrough */

     case XLAT_EXIT_HALT:

         if( sh4r.sh4_state != SH4_STATE_STANDBY ) {

             TMU_run_slice( sh4r.slice_cycle );

             SCIF_run_slice( sh4r.slice_cycle );

             dreamcast_stop();

             return sh4r.slice_cycle;

         }

     case XLAT_EXIT_SYSRESET:

         dreamcast_reset();

         break;

     case XLAT_EXIT_SLEEP:

         sh4_sleep_run_slice(nanosecs);

         break;	

     }

     xlat_running = TRUE;

     void * (*code)() = NULL;

     while( sh4r.slice_cycle < nanosecs ) {

         if( sh4r.event_pending <= sh4r.slice_cycle ) {

             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();

                 code = NULL;

             }

         }

         if( code == NULL ) {

             if( sh4r.pc > 0xFFFFFF00 ) {

                 syscall_invoke( sh4r.pc );

                 sh4r.in_delay_slot = 0;

                 sh4r.pc = sh4r.pr;

             }

             code = xlat_get_code_by_vma( sh4r.pc );

             if( code == NULL ) {

                 code = sh4_translate_basic_block( sh4r.pc );

             }

         }

         code = code();

     }

     xlat_running = FALSE;

     sh4_starting = FALSE;

     sh4r.slice_cycle = nanosecs;

     if( sh4r.sh4_state != SH4_STATE_STANDBY ) {

         TMU_run_slice( nanosecs );

         SCIF_run_slice( nanosecs );

     }

     return nanosecs;

 }

 uint8_t *xlat_output;

 xlat_cache_block_t xlat_current_block;

 struct xlat_recovery_record xlat_recovery[MAX_RECOVERY_SIZE];

 uint32_t xlat_recovery_posn;

 void sh4_translate_add_recovery( uint32_t icount )

 {

     xlat_recovery[xlat_recovery_posn].xlat_offset = 

         ((uintptr_t)xlat_output) - ((uintptr_t)xlat_current_block->code);

     xlat_recovery[xlat_recovery_posn].sh4_icount = icount;

     xlat_recovery_posn++;

 }

 /**

  * Translate a linear basic block, ie all instructions from the start address

  * (inclusive) until the next branch/jump instruction or the end of the page

  * is reached.

  * @return the address of the translated block

  * eg due to lack of buffer space.

  */

 void * sh4_translate_basic_block( sh4addr_t start )

 {

     sh4addr_t pc = start;

     sh4addr_t lastpc = (pc&0xFFFFF000)+0x1000;

     int done, i;

     xlat_current_block = xlat_start_block( start );

     xlat_output = (uint8_t *)xlat_current_block->code;

     xlat_recovery_posn = 0;

     uint8_t *eob = xlat_output + xlat_current_block->size;

     if( GET_ICACHE_END() < lastpc ) {

         lastpc = GET_ICACHE_END();

     }

     sh4_translate_begin_block(pc);

     do {

         /* check for breakpoints at this pc */

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

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

                 sh4_translate_emit_breakpoint(pc);

                 break;

             }

         }

         if( eob - xlat_output < MAX_INSTRUCTION_SIZE ) {

             uint8_t *oldstart = xlat_current_block->code;

             xlat_current_block = xlat_extend_block( xlat_output - oldstart + MAX_INSTRUCTION_SIZE );

             xlat_output = xlat_current_block->code + (xlat_output - oldstart);

             eob = xlat_current_block->code + xlat_current_block->size;

         }

         done = sh4_translate_instruction( pc ); 

         assert( xlat_output <= eob );

         pc += 2;

         if ( pc >= lastpc ) {

             done = 2;

         }

     } while( !done );

     pc += (done - 2);

     // Add end-of-block recovery for post-instruction checks

     sh4_translate_add_recovery( (pc - start)>>1 ); 

     int epilogue_size = sh4_translate_end_block_size();

     uint32_t recovery_size = sizeof(struct xlat_recovery_record)*xlat_recovery_posn;

     uint32_t finalsize = (xlat_output - xlat_current_block->code) + epilogue_size + recovery_size;

     if( xlat_current_block->size < finalsize ) {

         uint8_t *oldstart = xlat_current_block->code;

         xlat_current_block = xlat_extend_block( finalsize );

         xlat_output = xlat_current_block->code + (xlat_output - oldstart);

     }	

     sh4_translate_end_block(pc);

     assert( xlat_output <= (xlat_current_block->code + xlat_current_block->size - recovery_size) );

     /* Write the recovery records onto the end of the code block */

     memcpy( xlat_output, xlat_recovery, recovery_size);

     xlat_current_block->recover_table_offset = xlat_output - (uint8_t *)xlat_current_block->code;

     xlat_current_block->recover_table_size = xlat_recovery_posn;

     xlat_commit_block( finalsize, pc-start );

     return xlat_current_block->code;

 }

 /**

  * "Execute" the supplied recovery record. Currently this only updates

  * sh4r.pc and sh4r.slice_cycle according to the currently executing

  * instruction. In future this may be more sophisticated (ie will

  * call into generated code).

  */

 void sh4_translate_run_recovery( xlat_recovery_record_t recovery )

 {

     sh4r.slice_cycle += (recovery->sh4_icount * sh4_cpu_period);

     sh4r.pc += (recovery->sh4_icount<<1);

 }

 void sh4_translate_unwind_stack( gboolean abort_after, unwind_thunk_t thunk )

 {

     void *pc = xlat_get_native_pc();

     assert( pc != NULL );

     void *code = xlat_get_code( sh4r.pc );

     xlat_recovery_record_t recover = xlat_get_recovery(code, pc, TRUE);

     if( recover != NULL ) {

         // Can be null if there is no recovery necessary

         sh4_translate_run_recovery(recover);

     }

     if( thunk != NULL ) {

         thunk();

     }

     // finally longjmp back into sh4_xlat_run_slice

     xlat_running = FALSE;

     longjmp(xlat_jmp_buf, XLAT_EXIT_CONTINUE);

 } 

 void sh4_translate_exit( int exit_code )

 {

     void *pc = xlat_get_native_pc();

     if( pc != NULL ) {

         // could be null if we're not actually running inside the translator

         void *code = xlat_get_code( sh4r.pc );

         xlat_recovery_record_t recover = xlat_get_recovery(code, pc, TRUE);

         if( recover != NULL ) {

             // Can be null if there is no recovery necessary

             sh4_translate_run_recovery(recover);

         }

     }

     // finally longjmp back into sh4_xlat_run_slice

     xlat_running = FALSE;

     longjmp(xlat_jmp_buf, exit_code);

 }

 void sh4_translate_breakpoint_hit(uint32_t pc)

 {

     if( sh4_starting && sh4r.slice_cycle == 0 && pc == sh4r.pc ) {

         return;

     }

     sh4_translate_exit( XLAT_EXIT_BREAKPOINT );

 }

 /**

  * Exit the current block at the end of the current instruction, flush the

  * translation cache (completely) and return control to sh4_xlat_run_slice.

  *

  * As a special case, if the current instruction is actually the last 

  * instruction in the block (ie it's in a delay slot), this function 

  * returns to allow normal completion of the translation block. Otherwise

  * this function never returns.

  *

  * Must only be invoked (indirectly) from within translated code.

  */

 void sh4_translate_flush_cache()

 {

     void *pc = xlat_get_native_pc();

     assert( pc != NULL );

     void *code = xlat_get_code( sh4r.pc );

     xlat_recovery_record_t recover = xlat_get_recovery(code, pc, TRUE);

     if( recover != NULL ) {

         // Can be null if there is no recovery necessary

         sh4_translate_run_recovery(recover);

         xlat_flush_cache();

         xlat_running = FALSE;

         longjmp(xlat_jmp_buf, XLAT_EXIT_CONTINUE);

     } else {

         xlat_flush_cache();

         return;

     }

 }

 void *xlat_get_code_by_vma( sh4vma_t vma )

 {

     void *result = NULL;

     if( IS_IN_ICACHE(vma) ) {

         return xlat_get_code( GET_ICACHE_PHYS(vma) );

     }

     if( vma > 0xFFFFFF00 ) {

         // lxdream hook

         return NULL;

     }

     if( !mmu_update_icache(vma) ) {

         // fault - off to the fault handler

         if( !mmu_update_icache(sh4r.pc) ) {

             // double fault - halt

             ERROR( "Double fault - halting" );

             dreamcast_stop();

             return NULL;

         }

     }

     assert( IS_IN_ICACHE(sh4r.pc) );

     result = xlat_get_code( GET_ICACHE_PHYS(sh4r.pc) );

     return result;

 }