/**

  * $Id$

  * 

  * x86-specific MMU code - this emits simple TLB stubs for TLB indirection.

   *

  * Copyright (c) 2008 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 "lxdream.h"

 #include "mem.h"

 #include "sh4/sh4core.h"

 #include "sh4/sh4mmio.h"

 #include "sh4/sh4trans.h"

 #include "sh4/mmu.h"

 #include "sh4/x86op.h"

 #if SIZEOF_VOID_P == 8

 #define ARG1 R_EDI

 #define ARG2 R_ESI

 #define DECODE() \

     MOV_imm64_r32((uintptr_t)addr_space, R_EAX);     /* movq ptr, %rax */ \

     REXW(); OP(0x8B); OP(0x0C); OP(0xC8)                    /* movq [%rax + %rcx*8], %rcx */

 #else

 #define ARG1 R_EAX

 #define ARG2 R_EDX

 #define DECODE() \

     MOV_r32disp32x4_r32( R_ECX, (uintptr_t)addr_space, R_ECX );

 #endif

 void mmu_utlb_init_vtable( struct utlb_entry *ent, struct utlb_page_entry *page, gboolean writable )

 {

     uint32_t mask = ent->mask;

     uint32_t vpn = ent->vpn & mask;

     uint32_t ppn = ent->ppn & mask;

     struct mem_region_fn **addr_space;

     uint8_t **out = (uint8_t **)&page->fn;

     uint8_t **fn;

     int inc = writable ? 1 : 2; 

     int i;

     xlat_output = page->code;

     if( (ppn & 0x1FFFFFFF) >= 0x1C000000 ) {

         /* SH4 control region */

         ppn |= 0xE0000000;

         addr_space = sh4_address_space;

     } else {

         addr_space = ext_address_space;

     }

     fn = (uint8_t **)addr_space[ppn>>12];

     for( i=0; i<10; i+= inc, fn += inc, out += inc ) {

         *out = xlat_output;

         if( i != 9 ) { /* read_byte_for_write doesn't increment mmu_urc, everything else does */

 #if SIZEOF_VOID_P == 8

             MOV_imm64_r32((uintptr_t)&mmu_urc, R_EAX );

             OP(0x83); OP(0x00); OP(0x01); // ADD #1, [RAX]

 #else 

             OP(0x83); MODRM_r32_disp32(0, (uintptr_t)&mmu_urc); OP(0x01); // ADD #1, mmu_urc

 #endif

         }

         ADD_imm32_r32( ppn-vpn, ARG1 ); // 6

         if( ent->mask >= 0xFFFFF000 ) {

             // Maps to a single page, so jump directly there

             int rel = (*fn - xlat_output);

             JMP_rel( rel ); // 5

         } else {

             MOV_r32_r32( ARG1, R_ECX ); // 2

             SHR_imm8_r32( 12, R_ECX );  // 3

             DECODE();                   // 14

             JMP_r32disp8(R_ECX, (((uintptr_t)out) - ((uintptr_t)&page->fn)) );    // 3

         }

     }

     page->fn.prefetch = unmapped_prefetch; // FIXME

 }

 void mmu_utlb_init_storequeue_vtable( struct utlb_entry *ent, struct utlb_page_entry *page )

 {

     uint32_t mask = ent->mask;

     uint32_t vpn = ent->vpn & mask;

     uint32_t ppn = ent->ppn & mask;

     xlat_output = page->code;

     memcpy( page, &p4_region_storequeue, sizeof(struct mem_region_fn) );

     page->fn.prefetch = (mem_prefetch_fn_t)xlat_output;

 #if SIZEOF_VOID_P == 8

     MOV_imm64_r32((uintptr_t)&mmu_urc, R_EAX );

     OP(0x83); OP(0x00); OP(0x01); // ADD #1, [RAX]

 #else 

     OP(0x83); MODRM_r32_disp32(0, (uintptr_t)&mmu_urc); OP(0x01); // ADD #1, mmu_urc

 #endif

     ADD_imm32_r32( ppn-vpn, ARG1 );

     int rel = ((uint8_t *)ccn_storequeue_prefetch_tlb) - xlat_output;

     JMP_rel( rel );

 }

 void mmu_utlb_1k_init_vtable( struct utlb_1k_entry *entry )

 {

     xlat_output = entry->code;

     int i;

     uint8_t **out = (uint8_t **)&entry->fn;

     for( i=0; i<9; i++, out++ ) {

         *out = xlat_output;

         MOV_r32_r32( ARG1, R_ECX );

         SHR_imm8_r32( 10, R_ECX );

         AND_imm8s_r32( 0x3, R_ECX );

 #if SIZEOF_VOID_P == 8

         MOV_imm64_r32( (uintptr_t)&entry->subpages[0], R_EAX );

         REXW(); OP(0x8B); OP(0x0C); OP(0xC8);                   /* movq [%rax + %rcx*8], %rcx */

 #else

         MOV_r32disp32x4_r32( R_ECX, ((uintptr_t)&entry->subpages[0]), R_ECX );

 #endif                

         JMP_r32disp8(R_ECX, (((uintptr_t)out) - ((uintptr_t)&entry->fn)) );    // 3

     }

     out = (uint8_t **)&entry->user_fn;

     for( i=0; i<9; i++, out++ ) {

         *out = xlat_output;

         MOV_r32_r32( ARG1, R_ECX );

         SHR_imm8_r32( 10, R_ECX );

         AND_imm8s_r32( 0x3, R_ECX );

 #if SIZEOF_VOID_P == 8

         MOV_imm64_r32( (uintptr_t)&entry->user_subpages[0], R_EAX );

         REXW(); OP(0x8B); OP(0x0C); OP(0xC8);                   /* movq [%rax + %rcx*8], %rcx */

 #else

         MOV_r32disp32x4_r32( R_ECX, ((uintptr_t)&entry->user_subpages[0]), R_ECX );

 #endif                

         JMP_r32disp8(R_ECX, (((uintptr_t)out) - ((uintptr_t)&entry->user_fn)) );    // 3

     }

 }