lifning
/
gpsp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
gpsp/cpu.c

3733 lines
160 KiB
C
Raw Normal View History

original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
/* gameplaySP
*
* Copyright (C) 2006 Exophase <exophase@gmail.com>
*
* 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.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
// Important todo:
// - stm reglist writeback when base is in the list needs adjustment
// - block memory needs psr swapping and user mode reg swapping
#include "common.h"
Improve tracing a bit, also in interpreter mode
2023-04-15 02:02:17 +02:00
#include "cpu_instrument.h"
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
u32 memory_region_access_read_u8[16];
u32 memory_region_access_read_s8[16];
u32 memory_region_access_read_u16[16];
u32 memory_region_access_read_s16[16];
u32 memory_region_access_read_u32[16];
u32 memory_region_access_write_u8[16];
u32 memory_region_access_write_u16[16];
u32 memory_region_access_write_u32[16];
u32 memory_reads_u8;
u32 memory_reads_s8;
u32 memory_reads_u16;
u32 memory_reads_s16;
u32 memory_reads_u32;
u32 memory_writes_u8;
u32 memory_writes_u16;
u32 memory_writes_u32;
const u8 bit_count[256] =
{
0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4, 1, 2, 2, 3, 2, 3, 3,
4, 2, 3, 3, 4, 3, 4, 4, 5, 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4,
4, 5, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 1, 2, 2, 3, 2,
3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5,
4, 5, 5, 6, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 3, 4, 4,
5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, 1, 2, 2, 3, 2, 3, 3, 4, 2, 3,
3, 4, 3, 4, 4, 5, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 2,
3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 3, 4, 4, 5, 4, 5, 5, 6,
4, 5, 5, 6, 5, 6, 6, 7, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5,
6, 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, 3, 4, 4, 5, 4, 5,
5, 6, 4, 5, 5, 6, 5, 6, 6, 7, 4, 5, 5, 6, 5, 6, 6, 7, 5, 6, 6, 7, 6,
7, 7, 8
};
Add macro parameter 'opcode' to some macros
2014-12-10 01:17:37 +01:00
#define arm_decode_data_proc_reg(opcode) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
u32 rn = (opcode >> 16) & 0x0F; \
u32 rd = (opcode >> 12) & 0x0F; \
u32 rm = opcode & 0x0F; \
Prevent bunch of warnings
2015-07-26 04:44:17 +02:00
(void)rd; \
(void)rn; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
using_register(arm, rd, op_dest); \
using_register(arm, rn, op_src); \
using_register(arm, rm, op_src) \
Add macro parameter 'opcode' to some macros
2014-12-10 01:17:37 +01:00
#define arm_decode_data_proc_imm(opcode) \
Prevent bunch of warnings
2015-07-26 04:44:17 +02:00
u32 imm; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
u32 rn = (opcode >> 16) & 0x0F; \
u32 rd = (opcode >> 12) & 0x0F; \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
u32 imm_ror = ((opcode >> 8) & 0xF) << 1; \
Prevent bunch of warnings
2015-07-26 04:44:17 +02:00
(void)rd; \
(void)rn; \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
ror(imm, opcode & 0xFF, imm_ror); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
using_register(arm, rd, op_dest); \
using_register(arm, rn, op_src) \
Add macro parameter 'opcode' to some macros
2014-12-10 01:17:37 +01:00
#define arm_decode_psr_reg(opcode) \
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
u32 psr_pfield = ((opcode >> 16) & 1) | ((opcode >> 18) & 2); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
u32 rd = (opcode >> 12) & 0x0F; \
u32 rm = opcode & 0x0F; \
Prevent bunch of warnings
2015-07-26 04:44:17 +02:00
(void)rd; \
(void)rm; \
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
(void)psr_pfield; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
using_register(arm, rd, op_dest); \
using_register(arm, rm, op_src) \
Add macro parameter 'opcode' to some macros
2014-12-10 01:17:37 +01:00
#define arm_decode_psr_imm(opcode) \
Prevent bunch of warnings
2015-07-26 04:44:17 +02:00
u32 imm; \
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
u32 psr_pfield = ((opcode >> 16) & 1) | ((opcode >> 18) & 2); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
u32 rd = (opcode >> 12) & 0x0F; \
Prevent bunch of warnings
2015-07-26 04:44:17 +02:00
(void)rd; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
ror(imm, opcode & 0xFF, ((opcode >> 8) & 0x0F) * 2); \
using_register(arm, rd, op_dest) \
Add macro parameter 'opcode' to some macros
2014-12-10 01:17:37 +01:00
#define arm_decode_branchx(opcode) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
u32 rn = opcode & 0x0F; \
using_register(arm, rn, branch_target) \
#define arm_decode_multiply() \
u32 rd = (opcode >> 16) & 0x0F; \
u32 rn = (opcode >> 12) & 0x0F; \
u32 rs = (opcode >> 8) & 0x0F; \
u32 rm = opcode & 0x0F; \
Prevent bunch of warnings
2015-07-26 04:44:17 +02:00
(void)rn; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
using_register(arm, rd, op_dest); \
using_register(arm, rn, op_src); \
using_register(arm, rm, op_src) \
#define arm_decode_multiply_long() \
u32 rdhi = (opcode >> 16) & 0x0F; \
u32 rdlo = (opcode >> 12) & 0x0F; \
u32 rn = (opcode >> 8) & 0x0F; \
u32 rm = opcode & 0x0F; \
using_register(arm, rdhi, op_dest); \
using_register(arm, rdlo, op_dest); \
using_register(arm, rn, op_src); \
using_register(arm, rm, op_src) \
#define arm_decode_swap() \
u32 rn = (opcode >> 16) & 0x0F; \
u32 rd = (opcode >> 12) & 0x0F; \
u32 rm = opcode & 0x0F; \
using_register(arm, rd, memory_target); \
using_register(arm, rn, memory_base); \
using_register(arm, rm, memory_target) \
#define arm_decode_half_trans_r() \
u32 rn = (opcode >> 16) & 0x0F; \
u32 rd = (opcode >> 12) & 0x0F; \
u32 rm = opcode & 0x0F; \
using_register(arm, rd, memory_target); \
using_register(arm, rn, memory_base); \
using_register(arm, rm, memory_offset) \
#define arm_decode_half_trans_of() \
u32 rn = (opcode >> 16) & 0x0F; \
u32 rd = (opcode >> 12) & 0x0F; \
u32 offset = ((opcode >> 4) & 0xF0) | (opcode & 0x0F); \
using_register(arm, rd, memory_target); \
using_register(arm, rn, memory_base) \
#define arm_decode_data_trans_imm() \
u32 rn = (opcode >> 16) & 0x0F; \
u32 rd = (opcode >> 12) & 0x0F; \
u32 offset = opcode & 0x0FFF; \
using_register(arm, rd, memory_target); \
using_register(arm, rn, memory_base) \
#define arm_decode_data_trans_reg() \
u32 rn = (opcode >> 16) & 0x0F; \
u32 rd = (opcode >> 12) & 0x0F; \
u32 rm = opcode & 0x0F; \
using_register(arm, rd, memory_target); \
using_register(arm, rn, memory_base); \
using_register(arm, rm, memory_offset) \
#define arm_decode_block_trans() \
u32 rn = (opcode >> 16) & 0x0F; \
u32 reg_list = opcode & 0xFFFF; \
using_register(arm, rn, memory_base); \
using_register_list(arm, reg_list, 16) \
#define arm_decode_branch() \
Fix some UB behaviour
2021-06-30 00:29:21 +02:00
s32 offset = ((s32)((u32)(opcode << 8))) >> 6 \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define thumb_decode_shift() \
u32 imm = (opcode >> 6) & 0x1F; \
u32 rs = (opcode >> 3) & 0x07; \
u32 rd = opcode & 0x07; \
using_register(thumb, rd, op_dest); \
using_register(thumb, rs, op_shift) \
#define thumb_decode_add_sub() \
u32 rn = (opcode >> 6) & 0x07; \
u32 rs = (opcode >> 3) & 0x07; \
u32 rd = opcode & 0x07; \
using_register(thumb, rd, op_dest); \
using_register(thumb, rn, op_src); \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
using_register(thumb, rs, op_src) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define thumb_decode_add_sub_imm() \
u32 imm = (opcode >> 6) & 0x07; \
u32 rs = (opcode >> 3) & 0x07; \
u32 rd = opcode & 0x07; \
using_register(thumb, rd, op_src_dest); \
using_register(thumb, rs, op_src) \
#define thumb_decode_imm() \
u32 imm = opcode & 0xFF; \
using_register(thumb, ((opcode >> 8) & 0x07), op_dest) \
#define thumb_decode_alu_op() \
u32 rs = (opcode >> 3) & 0x07; \
u32 rd = opcode & 0x07; \
using_register(thumb, rd, op_src_dest); \
using_register(thumb, rs, op_src) \
#define thumb_decode_hireg_op() \
u32 rs = (opcode >> 3) & 0x0F; \
u32 rd = ((opcode >> 4) & 0x08) | (opcode & 0x07); \
Prevent bunch of warnings
2015-07-26 04:44:17 +02:00
(void)rd; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
using_register(thumb, rd, op_src_dest); \
using_register(thumb, rs, op_src) \
#define thumb_decode_mem_reg() \
u32 ro = (opcode >> 6) & 0x07; \
u32 rb = (opcode >> 3) & 0x07; \
u32 rd = opcode & 0x07; \
using_register(thumb, rd, memory_target); \
using_register(thumb, rb, memory_base); \
using_register(thumb, ro, memory_offset) \
#define thumb_decode_mem_imm() \
u32 imm = (opcode >> 6) & 0x1F; \
u32 rb = (opcode >> 3) & 0x07; \
u32 rd = opcode & 0x07; \
using_register(thumb, rd, memory_target); \
using_register(thumb, rb, memory_base) \
#define thumb_decode_add_sp() \
u32 imm = opcode & 0x7F; \
using_register(thumb, REG_SP, op_dest) \
#define thumb_decode_rlist() \
u32 reg_list = opcode & 0xFF; \
using_register_list(thumb, rlist, 8) \
#define thumb_decode_branch_cond() \
s32 offset = (s8)(opcode & 0xFF) \
#define thumb_decode_swi() \
u32 comment = opcode & 0xFF \
#define thumb_decode_branch() \
u32 offset = opcode & 0x07FF \
#define get_shift_register(dest) \
Fix shift by register in interpreter Only the LSB byte is actually used.
2021-09-17 20:17:51 +02:00
u32 shift = reg[(opcode >> 8) & 0x0F] & 0xFF; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
using_register(arm, ((opcode >> 8) & 0x0F), op_shift); \
dest = reg[rm]; \
if(rm == 15) \
dest += 4 \
#define calculate_z_flag(dest) \
z_flag = (dest == 0) \
#define calculate_n_flag(dest) \
n_flag = ((signed)dest < 0) \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
#define calculate_c_flag_sub(dest, src_a, src_b, carry) \
c_flag = (carry) ? ((unsigned)src_b <= (unsigned)src_a) : \
((unsigned)src_b < (unsigned)src_a); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define calculate_v_flag_sub(dest, src_a, src_b) \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
v_flag = (((src_a ^ src_b) & (~src_b ^ dest)) >> 31)
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define calculate_v_flag_add(dest, src_a, src_b) \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
v_flag = ((~((src_a) ^ (src_b)) & ((src_a) ^ (dest))) >> 31)
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define calculate_reg_sh() \
enable -Wall and fix warnings reported by it
2011-09-05 18:31:58 +02:00
u32 reg_sh = 0; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
switch((opcode >> 4) & 0x07) \
{ \
/* LSL imm */ \
case 0x0: \
{ \
reg_sh = reg[rm] << ((opcode >> 7) & 0x1F); \
break; \
} \
\
/* LSL reg */ \
case 0x1: \
{ \
get_shift_register(reg_sh); \
if(shift <= 31) \
reg_sh = reg_sh << shift; \
else \
reg_sh = 0; \
break; \
} \
\
/* LSR imm */ \
case 0x2: \
{ \
u32 imm = (opcode >> 7) & 0x1F; \
if(imm == 0) \
reg_sh = 0; \
else \
reg_sh = reg[rm] >> imm; \
break; \
} \
\
/* LSR reg */ \
case 0x3: \
{ \
get_shift_register(reg_sh); \
if(shift <= 31) \
reg_sh = reg_sh >> shift; \
else \
reg_sh = 0; \
break; \
} \
\
/* ASR imm */ \
case 0x4: \
{ \
u32 imm = (opcode >> 7) & 0x1F; \
reg_sh = reg[rm]; \
\
if(imm == 0) \
reg_sh = (s32)reg_sh >> 31; \
else \
reg_sh = (s32)reg_sh >> imm; \
break; \
} \
\
/* ASR reg */ \
case 0x5: \
{ \
get_shift_register(reg_sh); \
if(shift <= 31) \
reg_sh = (s32)reg_sh >> shift; \
else \
reg_sh = (s32)reg_sh >> 31; \
break; \
} \
\
/* ROR imm */ \
case 0x6: \
{ \
u32 imm = (opcode >> 7) & 0x1F; \
\
if(imm == 0) \
reg_sh = (reg[rm] >> 1) | (c_flag << 31); \
else \
ror(reg_sh, reg[rm], imm); \
break; \
} \
\
/* ROR reg */ \
case 0x7: \
{ \
get_shift_register(reg_sh); \
ror(reg_sh, reg_sh, shift); \
break; \
} \
} \
#define calculate_reg_sh_flags() \
enable -Wall and fix warnings reported by it
2011-09-05 18:31:58 +02:00
u32 reg_sh = 0; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
switch((opcode >> 4) & 0x07) \
{ \
/* LSL imm */ \
case 0x0: \
{ \
u32 imm = (opcode >> 7) & 0x1F; \
reg_sh = reg[rm]; \
\
if(imm != 0) \
{ \
c_flag = (reg_sh >> (32 - imm)) & 0x01; \
reg_sh <<= imm; \
} \
\
break; \
} \
\
/* LSL reg */ \
case 0x1: \
{ \
get_shift_register(reg_sh); \
if(shift != 0) \
{ \
if(shift > 31) \
{ \
if(shift == 32) \
c_flag = reg_sh & 0x01; \
else \
c_flag = 0; \
reg_sh = 0; \
} \
else \
{ \
c_flag = (reg_sh >> (32 - shift)) & 0x01; \
reg_sh <<= shift; \
} \
} \
break; \
} \
\
/* LSR imm */ \
case 0x2: \
{ \
u32 imm = (opcode >> 7) & 0x1F; \
reg_sh = reg[rm]; \
if(imm == 0) \
{ \
c_flag = reg_sh >> 31; \
reg_sh = 0; \
} \
else \
{ \
c_flag = (reg_sh >> (imm - 1)) & 0x01; \
reg_sh >>= imm; \
} \
break; \
} \
\
/* LSR reg */ \
case 0x3: \
{ \
get_shift_register(reg_sh); \
if(shift != 0) \
{ \
if(shift > 31) \
{ \
if(shift == 32) \
c_flag = (reg_sh >> 31) & 0x01; \
else \
c_flag = 0; \
reg_sh = 0; \
} \
else \
{ \
c_flag = (reg_sh >> (shift - 1)) & 0x01; \
reg_sh >>= shift; \
} \
} \
break; \
} \
\
/* ASR imm */ \
case 0x4: \
{ \
u32 imm = (opcode >> 7) & 0x1F; \
reg_sh = reg[rm]; \
if(imm == 0) \
{ \
reg_sh = (s32)reg_sh >> 31; \
c_flag = reg_sh & 0x01; \
} \
else \
{ \
c_flag = (reg_sh >> (imm - 1)) & 0x01; \
reg_sh = (s32)reg_sh >> imm; \
} \
break; \
} \
\
/* ASR reg */ \
case 0x5: \
{ \
get_shift_register(reg_sh); \
if(shift != 0) \
{ \
if(shift > 31) \
{ \
reg_sh = (s32)reg_sh >> 31; \
c_flag = reg_sh & 0x01; \
} \
else \
{ \
c_flag = (reg_sh >> (shift - 1)) & 0x01; \
reg_sh = (s32)reg_sh >> shift; \
} \
} \
break; \
} \
\
/* ROR imm */ \
case 0x6: \
{ \
u32 imm = (opcode >> 7) & 0x1F; \
reg_sh = reg[rm]; \
if(imm == 0) \
{ \
u32 old_c_flag = c_flag; \
c_flag = reg_sh & 0x01; \
reg_sh = (reg_sh >> 1) | (old_c_flag << 31); \
} \
else \
{ \
c_flag = (reg_sh >> (imm - 1)) & 0x01; \
ror(reg_sh, reg_sh, imm); \
} \
break; \
} \
\
/* ROR reg */ \
case 0x7: \
{ \
get_shift_register(reg_sh); \
if(shift != 0) \
{ \
c_flag = (reg_sh >> (shift - 1)) & 0x01; \
ror(reg_sh, reg_sh, shift); \
} \
break; \
} \
} \
#define calculate_reg_offset() \
enable -Wall and fix warnings reported by it
2011-09-05 18:31:58 +02:00
u32 reg_offset = 0; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
switch((opcode >> 5) & 0x03) \
{ \
/* LSL imm */ \
case 0x0: \
{ \
reg_offset = reg[rm] << ((opcode >> 7) & 0x1F); \
break; \
} \
\
/* LSR imm */ \
case 0x1: \
{ \
u32 imm = (opcode >> 7) & 0x1F; \
if(imm == 0) \
reg_offset = 0; \
else \
reg_offset = reg[rm] >> imm; \
break; \
} \
\
/* ASR imm */ \
case 0x2: \
{ \
u32 imm = (opcode >> 7) & 0x1F; \
if(imm == 0) \
reg_offset = (s32)reg[rm] >> 31; \
else \
reg_offset = (s32)reg[rm] >> imm; \
break; \
} \
\
/* ROR imm */ \
case 0x3: \
{ \
u32 imm = (opcode >> 7) & 0x1F; \
if(imm == 0) \
reg_offset = (reg[rm] >> 1) | (c_flag << 31); \
else \
ror(reg_offset, reg[rm], imm); \
break; \
} \
} \
#define calculate_flags_add(dest, src_a, src_b) \
calculate_z_flag(dest); \
calculate_n_flag(dest); \
calculate_v_flag_add(dest, src_a, src_b) \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
#define calculate_flags_sub(dest, src_a, src_b, carry) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
calculate_z_flag(dest); \
calculate_n_flag(dest); \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
calculate_c_flag_sub(dest, src_a, src_b, carry); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
calculate_v_flag_sub(dest, src_a, src_b) \
#define calculate_flags_logic(dest) \
calculate_z_flag(dest); \
calculate_n_flag(dest) \
#define extract_flags() \
n_flag = reg[REG_CPSR] >> 31; \
z_flag = (reg[REG_CPSR] >> 30) & 0x01; \
c_flag = (reg[REG_CPSR] >> 29) & 0x01; \
v_flag = (reg[REG_CPSR] >> 28) & 0x01; \
#define collapse_flags() \
reg[REG_CPSR] = (n_flag << 31) | (z_flag << 30) | (c_flag << 29) | \
(v_flag << 28) | (reg[REG_CPSR] & 0xFF) \
#define check_pc_region() \
new_pc_region = (pc >> 15); \
if(new_pc_region != pc_region) \
{ \
pc_region = new_pc_region; \
pc_address_block = memory_map_read[new_pc_region]; \
Rewrite ROM buffer allocation This improves the existing on-demand ROM paging and also breaks down ROM buffers into 1MB blocks for platforms with memory fragmentation issues. Fixed some potential RTC reg issue in said platforms too. Fixed page pinning on interpreter (would crash due to LRU evictions).
2021-07-14 01:17:55 +02:00
touch_gamepak_page(pc_region); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
\
Code nits - get rid of '== NULL' / '!= NULL'
2014-12-10 16:40:43 +01:00
if(!pc_address_block) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
pc_address_block = load_gamepak_page(pc_region & 0x3FF); \
} \
#define arm_update_pc() \
pc = reg[REG_PC] \
#define arm_pc_offset(val) \
pc += val; \
reg[REG_PC] = pc \
#define arm_pc_offset_update(val) \
pc += val; \
reg[REG_PC] = pc \
#define arm_pc_offset_update_direct(val) \
pc = val; \
reg[REG_PC] = pc \
// It should be okay to still generate result flags, spsr will overwrite them.
// This is pretty infrequent (returning from interrupt handlers, et al) so
// probably not worth optimizing for.
#define check_for_interrupts() \
Enable big-endian devices: gc/wii This patch adds big-endian compatibility in gpsp (in general but only for the interpreter). There's no performance hit for little-endian platforms (should be a no-op) and only add a small overhead in memory accesses for big-endian platforms. Most memory accesses are wrapped with a byteswap instruction and I/O reg accesses are also rewired for proper access (using macros). Video rendering has been fixed to also do byteswaps but there's a couple of games and rendering modes that still seem broken (but they amount to less than 20 games in my tests with 1K ROMs). This also adds build rules and CI for NGC/WII/WIIU (untested)
2021-06-27 01:16:28 +02:00
if((read_ioreg(REG_IE) & read_ioreg(REG_IF)) && \
read_ioreg(REG_IME) && ((reg[REG_CPSR] & 0x80) == 0)) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{ \
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
REG_MODE(MODE_IRQ)[6] = reg[REG_PC] + 4; \
REG_SPSR(MODE_IRQ) = reg[REG_CPSR]; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
reg[REG_CPSR] = 0xD2; \
reg[REG_PC] = 0x00000018; \
arm_update_pc(); \
set_cpu_mode(MODE_IRQ); \
goto arm_loop; \
} \
#define arm_spsr_restore() \
if(rd == 15) \
{ \
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
if(reg[CPU_MODE] != MODE_USER && reg[CPU_MODE] != MODE_SYSTEM) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{ \
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
reg[REG_CPSR] = REG_SPSR(reg[CPU_MODE]); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
extract_flags(); \
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
set_cpu_mode(cpu_modes[reg[REG_CPSR] & 0xF]); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
check_for_interrupts(); \
} \
arm_update_pc(); \
\
if(reg[REG_CPSR] & 0x20) \
goto thumb_loop; \
} \
#define arm_data_proc_flags_reg() \
Add macro parameter 'opcode' to some macros
2014-12-10 01:17:37 +01:00
arm_decode_data_proc_reg(opcode); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
calculate_reg_sh_flags() \
#define arm_data_proc_reg() \
Add macro parameter 'opcode' to some macros
2014-12-10 01:17:37 +01:00
arm_decode_data_proc_reg(opcode); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
calculate_reg_sh() \
#define arm_data_proc_flags_imm() \
Add macro parameter 'opcode' to some macros
2014-12-10 01:17:37 +01:00
arm_decode_data_proc_imm(opcode) \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
if(imm_ror) \
c_flag = (imm >> 31); /* imm is rotated already! */ \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define arm_data_proc_imm() \
Add macro parameter 'opcode' to some macros
2014-12-10 01:17:37 +01:00
arm_decode_data_proc_imm(opcode) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define arm_data_proc(expr, type) \
{ \
u32 dest; \
arm_pc_offset(8); \
arm_data_proc_##type(); \
dest = expr; \
arm_pc_offset(-4); \
reg[rd] = dest; \
\
if(rd == 15) \
{ \
arm_update_pc(); \
} \
} \
#define flags_vars(src_a, src_b) \
u32 dest; \
const u32 _sa = src_a; \
const u32 _sb = src_b \
#define arm_data_proc_logic_flags(expr, type) \
{ \
arm_pc_offset(8); \
arm_data_proc_flags_##type(); \
u32 dest = expr; \
calculate_flags_logic(dest); \
arm_pc_offset(-4); \
reg[rd] = dest; \
arm_spsr_restore(); \
} \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
#define arm_data_proc_add_flags(src_a, src_b, src_c, type) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{ \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
u32 _sc = src_c; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
arm_pc_offset(8); \
arm_data_proc_##type(); \
flags_vars(src_a, src_b); \
dest = _sa + _sb; \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
c_flag = (dest < _sb); \
dest += _sc; \
c_flag |= (dest < _sc); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
calculate_flags_add(dest, _sa, _sb); \
arm_pc_offset(-4); \
reg[rd] = dest; \
arm_spsr_restore(); \
}
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
#define arm_data_proc_sub_flags(src_a, src_b, src_c, type) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{ \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
u32 _sc = src_c; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
arm_pc_offset(8); \
arm_data_proc_##type(); \
flags_vars(src_a, src_b); \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
dest = _sa + (~(_sb)) + _sc; \
calculate_flags_sub(dest, _sa, _sb, _sc); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
arm_pc_offset(-4); \
reg[rd] = dest; \
arm_spsr_restore(); \
} \
#define arm_data_proc_test_logic(expr, type) \
{ \
arm_pc_offset(8); \
arm_data_proc_flags_##type(); \
u32 dest = expr; \
calculate_flags_logic(dest); \
arm_pc_offset(-4); \
} \
#define arm_data_proc_test_add(src_a, src_b, type) \
{ \
arm_pc_offset(8); \
arm_data_proc_##type(); \
flags_vars(src_a, src_b); \
dest = _sa + _sb; \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
c_flag = (dest < _sb); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
calculate_flags_add(dest, _sa, _sb); \
arm_pc_offset(-4); \
} \
#define arm_data_proc_test_sub(src_a, src_b, type) \
{ \
arm_pc_offset(8); \
arm_data_proc_##type(); \
flags_vars(src_a, src_b); \
dest = _sa - _sb; \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
calculate_flags_sub(dest, _sa, _sb, 1); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
arm_pc_offset(-4); \
} \
#define arm_multiply_flags_yes(_dest) \
calculate_z_flag(_dest); \
calculate_n_flag(_dest); \
#define arm_multiply_flags_no(_dest) \
#define arm_multiply_long_flags_yes(_dest_lo, _dest_hi) \
z_flag = (_dest_lo == 0) & (_dest_hi == 0); \
calculate_n_flag(_dest_hi) \
#define arm_multiply_long_flags_no(_dest_lo, _dest_hi) \
#define arm_multiply(add_op, flags) \
{ \
u32 dest; \
arm_decode_multiply(); \
dest = (reg[rm] * reg[rs]) add_op; \
arm_multiply_flags_##flags(dest); \
reg[rd] = dest; \
arm_pc_offset(4); \
} \
#define arm_multiply_long_addop(type) \
+ ((type##64)((((type##64)reg[rdhi]) << 32) | reg[rdlo])); \
#define arm_multiply_long(add_op, flags, type) \
{ \
type##64 dest; \
u32 dest_lo; \
u32 dest_hi; \
arm_decode_multiply_long(); \
dest = ((type##64)((type##32)reg[rm]) * \
(type##64)((type##32)reg[rn])) add_op; \
dest_lo = (u32)dest; \
dest_hi = (u32)(dest >> 32); \
arm_multiply_long_flags_##flags(dest_lo, dest_hi); \
reg[rdlo] = dest_lo; \
reg[rdhi] = dest_hi; \
arm_pc_offset(4); \
} \
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
// Index by PRS fields (1 and 4 only!) and User-Privileged mode
// In user mode some bits are read only
// Bit #4 is always set to one (so all modes are 1XXXX)
// Reserved bits are always zero and cannot be modified
const u32 cpsr_masks[4][2] =
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
// User, Privileged
{0x00000000, 0x00000000},
{0x00000020, 0x000000EF},
{0xF0000000, 0xF0000000},
{0xF0000020, 0xF00000EF}
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
};
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
// SPSR is always a privileged instruction
const u32 spsr_masks[4] = { 0x00000000, 0x000000EF, 0xF0000000, 0xF00000EF };
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define arm_psr_read(dummy, psr_reg) \
collapse_flags(); \
reg[rd] = psr_reg \
#define arm_psr_store_cpsr(source) \
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
const u32 store_mask = cpsr_masks[psr_pfield][PRIVMODE(reg[CPU_MODE])]; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
reg[REG_CPSR] = (source & store_mask) | (reg[REG_CPSR] & (~store_mask)); \
extract_flags(); \
if(store_mask & 0xFF) \
{ \
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
set_cpu_mode(cpu_modes[reg[REG_CPSR] & 0xF]); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
check_for_interrupts(); \
} \
#define arm_psr_store_spsr(source) \
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
const u32 store_mask = spsr_masks[psr_pfield]; \
u32 _psr = REG_SPSR(reg[CPU_MODE]); \
REG_SPSR(reg[CPU_MODE]) = (source & store_mask) | (_psr & (~store_mask)) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define arm_psr_store(source, psr_reg) \
arm_psr_store_##psr_reg(source) \
#define arm_psr_src_reg reg[rm]
#define arm_psr_src_imm imm
#define arm_psr(op_type, transfer_type, psr_reg) \
{ \
Add macro parameter 'opcode' to some macros
2014-12-10 01:17:37 +01:00
arm_decode_psr_##op_type(opcode); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
arm_pc_offset(4); \
arm_psr_##transfer_type(arm_psr_src_##op_type, psr_reg); \
} \
#define arm_data_trans_reg() \
arm_decode_data_trans_reg(); \
calculate_reg_offset() \
#define arm_data_trans_imm() \
arm_decode_data_trans_imm() \
#define arm_data_trans_half_reg() \
arm_decode_half_trans_r() \
#define arm_data_trans_half_imm() \
arm_decode_half_trans_of() \
#define aligned_address_mask8 0xF0000000
#define aligned_address_mask16 0xF0000001
#define aligned_address_mask32 0xF0000003
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
#define fast_read_memory(size, type, addr, dest, readfn) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{ \
u8 *map; \
Enable big-endian devices: gc/wii This patch adds big-endian compatibility in gpsp (in general but only for the interpreter). There's no performance hit for little-endian platforms (should be a no-op) and only add a small overhead in memory accesses for big-endian platforms. Most memory accesses are wrapped with a byteswap instruction and I/O reg accesses are also rewired for proper access (using macros). Video rendering has been fixed to also do byteswaps but there's a couple of games and rendering modes that still seem broken (but they amount to less than 20 games in my tests with 1K ROMs). This also adds build rules and CI for NGC/WII/WIIU (untested)
2021-06-27 01:16:28 +02:00
u32 _address = addr; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
\
if(_address < 0x10000000) \
{ \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
/* Account for cycles and other stats */ \
u8 region = _address >> 24; \
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_nseq[region][(size - 8) / 16]; \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
memory_region_access_read_##type[region]++; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
memory_reads_##type++; \
} \
\
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
if ( \
(((_address >> 24) == 0) && (pc >= 0x4000)) || /* BIOS reading */ \
(_address & aligned_address_mask##size) || /* Unaligned access */ \
!(map = memory_map_read[_address >> 15]) /* Unmapped memory */ \
) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{ \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
dest = (type)(readfn)(_address); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
} \
else \
{ \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
/* Aligned and mapped read */ \
dest = (type)readaddress##size(map, (_address & 0x7FFF)); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
} \
} \
#define fast_write_memory(size, type, address, value) \
{ \
u32 _address = (address) & ~(aligned_address_mask##size & 0x03); \
if(_address < 0x10000000) \
{ \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
u8 region = _address >> 24; \
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_nseq[region][(size - 8) / 16]; \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
memory_region_access_write_##type[region]++; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
memory_writes_##type++; \
} \
\
Make ewram memory lineal This saves a few cycles in MIPS and simplifies a bit the core. Removed the write map, only affects interpreter performance very minimally. Rewired ARM and x86 handlers to support direct access to I/EWRAM (and VRAM on ARM) to compensate. Overall performance is slightly better but code is cleaner and allows for further improvements in the dynarecs.
2021-03-23 19:05:35 +01:00
cpu_alert = write_memory##size(_address, value); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
} \
#define load_aligned32(address, dest) \
{ \
enable -Wall and fix warnings reported by it
2011-09-05 18:31:58 +02:00
u32 _address = address; \
u8 *map = memory_map_read[_address >> 15]; \
if(_address < 0x10000000) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{ \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
/* Account for cycles and other stats */ \
u8 region = _address >> 24; \
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_seq[region][1]; \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
memory_region_access_read_u32[region]++; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
memory_reads_u32++; \
} \
Fix aligned32 reads in interpreter mode An address check was missing to read aligned 32 (stm/ldm) data from high mem areas (0xX0000000). This fixes SM4 EU that for some reason has some weird memory access (perhaps a bug?)
2021-05-05 21:31:24 +02:00
if(_address < 0x10000000 && map) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{ \
Enable big-endian devices: gc/wii This patch adds big-endian compatibility in gpsp (in general but only for the interpreter). There's no performance hit for little-endian platforms (should be a no-op) and only add a small overhead in memory accesses for big-endian platforms. Most memory accesses are wrapped with a byteswap instruction and I/O reg accesses are also rewired for proper access (using macros). Video rendering has been fixed to also do byteswaps but there's a couple of games and rendering modes that still seem broken (but they amount to less than 20 games in my tests with 1K ROMs). This also adds build rules and CI for NGC/WII/WIIU (untested)
2021-06-27 01:16:28 +02:00
dest = readaddress32(map, _address & 0x7FFF); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
} \
else \
{ \
enable -Wall and fix warnings reported by it
2011-09-05 18:31:58 +02:00
dest = read_memory32(_address); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
} \
} \
#define store_aligned32(address, value) \
{ \
enable -Wall and fix warnings reported by it
2011-09-05 18:31:58 +02:00
u32 _address = address; \
if(_address < 0x10000000) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{ \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
/* Account for cycles and other stats */ \
u8 region = _address >> 24; \
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_seq[region][1]; \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
memory_region_access_write_u32[region]++; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
memory_writes_u32++; \
} \
Make ewram memory lineal This saves a few cycles in MIPS and simplifies a bit the core. Removed the write map, only affects interpreter performance very minimally. Rewired ARM and x86 handlers to support direct access to I/EWRAM (and VRAM on ARM) to compensate. Overall performance is slightly better but code is cleaner and allows for further improvements in the dynarecs.
2021-03-23 19:05:35 +01:00
cpu_alert = write_memory32(_address, value); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
} \
#define load_memory_u8(address, dest) \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
fast_read_memory(8, u8, address, dest, read_memory8) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define load_memory_u16(address, dest) \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
fast_read_memory(16, u16, address, dest, read_memory16) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define load_memory_u32(address, dest) \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
fast_read_memory(32, u32, address, dest, read_memory32) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define load_memory_s8(address, dest) \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
fast_read_memory(8, s8, address, dest, read_memory8) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define load_memory_s16(address, dest) \
[interp] Simplify memory bindings a bit
2023-04-26 21:08:33 +02:00
fast_read_memory(16, s16, address, dest, read_memory16_signed) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define store_memory_u8(address, value) \
fast_write_memory(8, u8, address, value) \
#define store_memory_u16(address, value) \
fast_write_memory(16, u16, address, value) \
#define store_memory_u32(address, value) \
fast_write_memory(32, u32, address, value) \
#define no_op \
#define arm_access_memory_writeback_yes(off_op) \
reg[rn] = address off_op \
#define arm_access_memory_writeback_no(off_op) \
#define arm_access_memory_pc_preadjust_load() \
#define arm_access_memory_pc_preadjust_store() \
u32 reg_op = reg[rd]; \
if(rd == 15) \
reg_op += 4 \
#define arm_access_memory_pc_postadjust_load() \
arm_update_pc() \
#define arm_access_memory_pc_postadjust_store() \
#define load_reg_op reg[rd] \
#define store_reg_op reg_op \
#define arm_access_memory(access_type, off_op, off_type, mem_type, \
wb, wb_off_op) \
{ \
arm_pc_offset(8); \
arm_data_trans_##off_type(); \
u32 address = reg[rn] off_op; \
arm_access_memory_pc_preadjust_##access_type(); \
\
arm_pc_offset(-4); \
arm_access_memory_writeback_##wb(wb_off_op); \
access_type##_memory_##mem_type(address, access_type##_reg_op); \
arm_access_memory_pc_postadjust_##access_type(); \
} \
#define word_bit_count(word) \
(bit_count[word >> 8] + bit_count[word & 0xFF]) \
#define arm_block_writeback_load() \
if(!((reg_list >> rn) & 0x01)) \
{ \
reg[rn] = address; \
} \
#define arm_block_writeback_store() \
reg[rn] = address \
#define arm_block_writeback_yes(access_type) \
arm_block_writeback_##access_type() \
#define arm_block_writeback_no(access_type) \
#define load_block_memory(address, dest) \
Enable big-endian devices: gc/wii This patch adds big-endian compatibility in gpsp (in general but only for the interpreter). There's no performance hit for little-endian platforms (should be a no-op) and only add a small overhead in memory accesses for big-endian platforms. Most memory accesses are wrapped with a byteswap instruction and I/O reg accesses are also rewired for proper access (using macros). Video rendering has been fixed to also do byteswaps but there's a couple of games and rendering modes that still seem broken (but they amount to less than 20 games in my tests with 1K ROMs). This also adds build rules and CI for NGC/WII/WIIU (untested)
2021-06-27 01:16:28 +02:00
dest = readaddress32(address_region, (address + offset) & 0x7FFF) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define store_block_memory(address, dest) \
Enable big-endian devices: gc/wii This patch adds big-endian compatibility in gpsp (in general but only for the interpreter). There's no performance hit for little-endian platforms (should be a no-op) and only add a small overhead in memory accesses for big-endian platforms. Most memory accesses are wrapped with a byteswap instruction and I/O reg accesses are also rewired for proper access (using macros). Video rendering has been fixed to also do byteswaps but there's a couple of games and rendering modes that still seem broken (but they amount to less than 20 games in my tests with 1K ROMs). This also adds build rules and CI for NGC/WII/WIIU (untested)
2021-06-27 01:16:28 +02:00
address32(address_region, (address + offset) & 0x7FFF) = eswap32(dest) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
#define arm_block_memory_offset_down_a() \
(base - (word_bit_count(reg_list) * 4) + 4) \
#define arm_block_memory_offset_down_b() \
(base - (word_bit_count(reg_list) * 4)) \
#define arm_block_memory_offset_no() \
(base) \
#define arm_block_memory_offset_up() \
(base + 4) \
#define arm_block_memory_writeback_down() \
reg[rn] = base - (word_bit_count(reg_list) * 4) \
#define arm_block_memory_writeback_up() \
reg[rn] = base + (word_bit_count(reg_list) * 4) \
#define arm_block_memory_writeback_no() \
#define arm_block_memory_load_pc() \
load_aligned32(address, pc); \
reg[REG_PC] = pc \
#define arm_block_memory_store_pc() \
store_aligned32(address, pc + 4) \
#define arm_block_memory(access_type, offset_type, writeback_type, s_bit) \
{ \
arm_decode_block_trans(); \
u32 base = reg[rn]; \
u32 address = arm_block_memory_offset_##offset_type() & 0xFFFFFFFC; \
u32 i; \
\
arm_block_memory_writeback_##writeback_type(); \
\
for(i = 0; i < 15; i++) \
{ \
if((reg_list >> i) & 0x01) \
{ \
access_type##_aligned32(address, reg[i]); \
address += 4; \
} \
} \
\
arm_pc_offset(4); \
if(reg_list & 0x8000) \
{ \
arm_block_memory_##access_type##_pc(); \
} \
} \
#define arm_swap(type) \
{ \
arm_decode_swap(); \
u32 temp; \
load_memory_##type(reg[rn], temp); \
store_memory_##type(reg[rn], reg[rm]); \
reg[rd] = temp; \
arm_pc_offset(4); \
} \
#define arm_next_instruction() \
{ \
arm_pc_offset(4); \
goto skip_instruction; \
} \
#define thumb_update_pc() \
pc = reg[REG_PC] \
#define thumb_pc_offset(val) \
pc += val; \
reg[REG_PC] = pc \
#define thumb_pc_offset_update(val) \
pc += val; \
reg[REG_PC] = pc \
#define thumb_pc_offset_update_direct(val) \
pc = val; \
reg[REG_PC] = pc \
// Types: add_sub, add_sub_imm, alu_op, imm
// Affects N/Z/C/V flags
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
#define thumb_add(type, dest_reg, src_a, src_b, src_c) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{ \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
const u32 _sc = src_c; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
thumb_decode_##type(); \
const u32 _sa = src_a; \
const u32 _sb = src_b; \
u32 dest = _sa + _sb; \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
c_flag = (dest < _sb); \
dest += _sc; \
c_flag |= (dest < _sc); \
enable -Wall and fix warnings reported by it
2011-09-05 18:31:58 +02:00
calculate_flags_add(dest, _sa, _sb); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
reg[dest_reg] = dest; \
thumb_pc_offset(2); \
} \
#define thumb_add_noflags(type, dest_reg, src_a, src_b) \
{ \
thumb_decode_##type(); \
enable -Wall and fix warnings reported by it
2011-09-05 18:31:58 +02:00
u32 dest = (src_a) + (src_b); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
reg[dest_reg] = dest; \
thumb_pc_offset(2); \
} \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
#define thumb_sub(type, dest_reg, src_a, src_b, src_c) \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{ \
thumb_decode_##type(); \
const u32 _sa = src_a; \
const u32 _sb = src_b; \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
const u32 _sc = src_c; \
u32 dest = _sa + (~_sb) + _sc; \
calculate_flags_sub(dest, _sa, _sb, _sc); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
reg[dest_reg] = dest; \
thumb_pc_offset(2); \
} \
// Affects N/Z flags
#define thumb_logic(type, dest_reg, expr) \
{ \
thumb_decode_##type(); \
u32 dest = expr; \
calculate_flags_logic(dest); \
reg[dest_reg] = dest; \
thumb_pc_offset(2); \
} \
// Decode types: shift, alu_op
// Operation types: lsl, lsr, asr, ror
// Affects N/Z/C flags
#define thumb_shift_lsl_reg() \
u32 shift = reg[rs]; \
u32 dest = reg[rd]; \
if(shift != 0) \
{ \
if(shift > 31) \
{ \
if(shift == 32) \
c_flag = dest & 0x01; \
else \
c_flag = 0; \
dest = 0; \
} \
else \
{ \
c_flag = (dest >> (32 - shift)) & 0x01; \
dest <<= shift; \
} \
} \
#define thumb_shift_lsr_reg() \
u32 shift = reg[rs]; \
u32 dest = reg[rd]; \
if(shift != 0) \
{ \
if(shift > 31) \
{ \
if(shift == 32) \
c_flag = dest >> 31; \
else \
c_flag = 0; \
dest = 0; \
} \
else \
{ \
c_flag = (dest >> (shift - 1)) & 0x01; \
dest >>= shift; \
} \
} \
#define thumb_shift_asr_reg() \
u32 shift = reg[rs]; \
u32 dest = reg[rd]; \
if(shift != 0) \
{ \
if(shift > 31) \
{ \
dest = (s32)dest >> 31; \
c_flag = dest & 0x01; \
} \
else \
{ \
c_flag = (dest >> (shift - 1)) & 0x01; \
dest = (s32)dest >> shift; \
} \
} \
#define thumb_shift_ror_reg() \
u32 shift = reg[rs]; \
u32 dest = reg[rd]; \
if(shift != 0) \
{ \
c_flag = (dest >> (shift - 1)) & 0x01; \
ror(dest, dest, shift); \
} \
#define thumb_shift_lsl_imm() \
u32 dest = reg[rs]; \
if(imm != 0) \
{ \
c_flag = (dest >> (32 - imm)) & 0x01; \
dest <<= imm; \
} \
#define thumb_shift_lsr_imm() \
u32 dest; \
if(imm == 0) \
{ \
dest = 0; \
c_flag = reg[rs] >> 31; \
} \
else \
{ \
dest = reg[rs]; \
c_flag = (dest >> (imm - 1)) & 0x01; \
dest >>= imm; \
} \
#define thumb_shift_asr_imm() \
u32 dest; \
if(imm == 0) \
{ \
dest = (s32)reg[rs] >> 31; \
c_flag = dest & 0x01; \
} \
else \
{ \
dest = reg[rs]; \
c_flag = (dest >> (imm - 1)) & 0x01; \
dest = (s32)dest >> imm; \
} \
#define thumb_shift_ror_imm() \
u32 dest = reg[rs]; \
if(imm == 0) \
{ \
u32 old_c_flag = c_flag; \
c_flag = dest & 0x01; \
dest = (dest >> 1) | (old_c_flag << 31); \
} \
else \
{ \
c_flag = (dest >> (imm - 1)) & 0x01; \
ror(dest, dest, imm); \
} \
#define thumb_shift(decode_type, op_type, value_type) \
{ \
thumb_decode_##decode_type(); \
thumb_shift_##op_type##_##value_type(); \
calculate_flags_logic(dest); \
reg[rd] = dest; \
thumb_pc_offset(2); \
} \
#define thumb_test_add(type, src_a, src_b) \
{ \
thumb_decode_##type(); \
const u32 _sa = src_a; \
const u32 _sb = src_b; \
u32 dest = _sa + _sb; \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
c_flag = (dest < _sb); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
calculate_flags_add(dest, src_a, src_b); \
thumb_pc_offset(2); \
} \
#define thumb_test_sub(type, src_a, src_b) \
{ \
thumb_decode_##type(); \
const u32 _sa = src_a; \
const u32 _sb = src_b; \
u32 dest = _sa - _sb; \
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
calculate_flags_sub(dest, src_a, src_b, 1); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
thumb_pc_offset(2); \
} \
#define thumb_test_logic(type, expr) \
{ \
thumb_decode_##type(); \
u32 dest = expr; \
calculate_flags_logic(dest); \
thumb_pc_offset(2); \
}
#define thumb_hireg_op(expr) \
{ \
thumb_pc_offset(4); \
thumb_decode_hireg_op(); \
u32 dest = expr; \
thumb_pc_offset(-2); \
if(rd == 15) \
{ \
reg[REG_PC] = dest & ~0x01; \
thumb_update_pc(); \
} \
else \
{ \
reg[rd] = dest; \
} \
} \
// Operation types: imm, mem_reg, mem_imm
#define thumb_access_memory(access_type, op_type, address, reg_op, \
mem_type) \
{ \
Fix interpreter bug on store-triggered DMAs In thumb mode, a store that triggers a DMA and its correspondoing DMA IRQ (since emulate DMAs being instantanious, which is another can of worms tho...) overrwrites the PC with the PC-next value (disregarding the IRQ handler address). This causes quite a few bugs. I'd expect it causing way more bugs, but it doesnt...? Anyway this fixes a few games like Buffy, Woody, Medabots, Super Duper Sumos and Power Rangers
2023-04-03 19:54:07 +02:00
thumb_pc_offset(2); \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
thumb_decode_##op_type(); \
access_type##_memory_##mem_type(address, reg_op); \
} \
#define thumb_block_address_preadjust_no_op() \
#define thumb_block_address_preadjust_up() \
address += bit_count[reg_list] * 4 \
#define thumb_block_address_preadjust_down() \
address -= bit_count[reg_list] * 4 \
#define thumb_block_address_preadjust_push_lr() \
address -= (bit_count[reg_list] + 1) * 4 \
#define thumb_block_address_postadjust_no_op() \
#define thumb_block_address_postadjust_up() \
address += offset \
#define thumb_block_address_postadjust_down() \
address -= offset \
#define thumb_block_address_postadjust_pop_pc() \
load_memory_u32(address + offset, pc); \
pc &= ~0x01; \
reg[REG_PC] = pc; \
address += offset + 4 \
#define thumb_block_address_postadjust_push_lr() \
store_memory_u32(address + offset, reg[REG_LR]); \
#define thumb_block_memory_wb_load(base_reg) \
if(!((reg_list >> base_reg) & 0x01)) \
{ \
reg[base_reg] = address; \
} \
#define thumb_block_memory_wb_store(base_reg) \
reg[base_reg] = address \
#define thumb_block_memory(access_type, pre_op, post_op, base_reg) \
{ \
u32 i; \
u32 offset = 0; \
thumb_decode_rlist(); \
using_register(thumb, base_reg, memory_base); \
u32 address = reg[base_reg] & ~0x03; \
thumb_block_address_preadjust_##pre_op(); \
\
for(i = 0; i < 8; i++) \
{ \
if((reg_list >> i) & 1) \
{ \
access_type##_aligned32(address + offset, reg[i]); \
offset += 4; \
} \
} \
\
thumb_pc_offset(2); \
\
thumb_block_address_postadjust_##post_op(); \
thumb_block_memory_wb_##access_type(base_reg); \
} \
#define thumb_conditional_branch(condition) \
{ \
thumb_decode_branch_cond(); \
if(condition) \
{ \
thumb_pc_offset((offset * 2) + 4); \
} \
else \
{ \
thumb_pc_offset(2); \
} \
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_nseq[pc >> 24][0]; \
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
} \
// When a mode change occurs from non-FIQ to non-FIQ retire the current
// reg[13] and reg[14] into reg_mode[cpu_mode][5] and reg_mode[cpu_mode][6]
// respectively and load into reg[13] and reg[14] reg_mode[new_mode][5] and
// reg_mode[new_mode][6]. When swapping to/from FIQ retire/load reg[8]
// through reg[14] to/from reg_mode[MODE_FIQ][0] through reg_mode[MODE_FIQ][6].
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
const u32 cpu_modes[16] =
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
MODE_USER, MODE_FIQ, MODE_IRQ, MODE_SUPERVISOR,
MODE_INVALID, MODE_INVALID, MODE_INVALID, MODE_ABORT,
MODE_UNDEFINED is 0x1B (not 0x1C) (#184) In the cpu_modes array, MODE_UNDEFINED is at position 0x1C, but this mode value is actually 0x1B.
2023-01-21 21:44:23 +01:00
MODE_INVALID, MODE_INVALID, MODE_INVALID, MODE_UNDEFINED,
MODE_INVALID, MODE_INVALID, MODE_INVALID, MODE_SYSTEM
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
};
// ARM/Thumb mode is stored in the flags directly, this is simpler than
// shadowing it since it has a constant 1bit represenation.
u32 instruction_count = 0;
void set_cpu_mode(cpu_mode_type new_mode)
{
u32 i;
cpu_mode_type cpu_mode = reg[CPU_MODE];
Style nits
2014-12-10 15:47:19 +01:00
if(cpu_mode == new_mode)
return;
if(new_mode == MODE_FIQ)
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{
Style nits
2014-12-10 15:47:19 +01:00
for(i = 8; i < 15; i++)
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
REG_MODE(cpu_mode)[i - 8] = reg[i];
Style nits
2014-12-10 15:47:19 +01:00
}
else
{
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
REG_MODE(cpu_mode)[5] = reg[REG_SP];
REG_MODE(cpu_mode)[6] = reg[REG_LR];
Style nits
2014-12-10 15:47:19 +01:00
}
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
Style nits
2014-12-10 15:47:19 +01:00
if(cpu_mode == MODE_FIQ)
{
for(i = 8; i < 15; i++)
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
reg[i] = REG_MODE(new_mode)[i - 8];
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
}
Style nits
2014-12-10 15:47:19 +01:00
else
{
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
reg[REG_SP] = REG_MODE(new_mode)[5];
reg[REG_LR] = REG_MODE(new_mode)[6];
Style nits
2014-12-10 15:47:19 +01:00
}
reg[CPU_MODE] = new_mode;
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
}
Rewrite I/O side effects write and IRQ triggers This rewrites the way that CPU alerts work, making them a bitmap (since multiple alerts can happen simultaneously, like SMC and IRQ). This doesn't really fix many games but improves accuracy overall and improves performance on some I/O writes (the ones without side effects). The IRQ raising is now decoupled and explicitely called via a new function (check_and_raise_interrupts) to avoid issues such as invalid CPSR values (doesn't seem to bother most games!). There's more side effects missing, so this just lays the ground for more fixes.
2023-04-14 01:41:55 +02:00
#define cpu_has_interrupt() \
(!(reg[REG_CPSR] & 0x80) && read_ioreg(REG_IME) && \
Wrap I/O register reads This can cause issues in big-endian platforms otherwise
2023-04-19 20:03:11 +02:00
(read_ioreg(REG_IE) & read_ioreg(REG_IF)))
Rewrite I/O side effects write and IRQ triggers This rewrites the way that CPU alerts work, making them a bitmap (since multiple alerts can happen simultaneously, like SMC and IRQ). This doesn't really fix many games but improves accuracy overall and improves performance on some I/O writes (the ones without side effects). The IRQ raising is now decoupled and explicitely called via a new function (check_and_raise_interrupts) to avoid issues such as invalid CPSR values (doesn't seem to bother most games!). There's more side effects missing, so this just lays the ground for more fixes.
2023-04-14 01:41:55 +02:00
// Returns whether the CPU has a pending interrupt.
cpu_alert_type check_interrupt() {
return (cpu_has_interrupt()) ? CPU_ALERT_IRQ : CPU_ALERT_NONE;
}
// Checks for pending IRQs and raises them. This changes the CPU mode
// which means that it must be called with a valid CPU state.
Remove CHANGED_PC_STATUS, simplify update flow
2023-04-24 20:24:03 +02:00
u32 check_and_raise_interrupts()
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{
Fix pending interrupts not raised on IME/IE writes Whenever an interrupt is pending and interrupts are disabled (via IME/IE), an IE/IME write that re-enables IRQs will fail to raise an IRQ. This makes some games hang. Most games seem to use CPSR.IRQ to enable and disable interruts, so they are not affected. However some others use IME/IE (or all of them), causing these deadlocks and some race conditions. This fixes a bunch of games that did not crash but would "hang" in some interesting ways.
2023-04-10 12:58:03 +02:00
// Check any IRQ flag pending, IME and CPSR-IRQ enabled
Rewrite I/O side effects write and IRQ triggers This rewrites the way that CPU alerts work, making them a bitmap (since multiple alerts can happen simultaneously, like SMC and IRQ). This doesn't really fix many games but improves accuracy overall and improves performance on some I/O writes (the ones without side effects). The IRQ raising is now decoupled and explicitely called via a new function (check_and_raise_interrupts) to avoid issues such as invalid CPSR values (doesn't seem to bother most games!). There's more side effects missing, so this just lays the ground for more fixes.
2023-04-14 01:41:55 +02:00
if (cpu_has_interrupt())
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{
Minor mem fixes around mirrors and u8/u16 accesses Improve the open bus access a bit too.
2021-09-17 21:30:33 +02:00
// Value after the FIQ returns, should be improved
Improve open bus reads on ARM/MIPS
2022-01-26 19:03:14 +01:00
reg[REG_BUS_VALUE] = 0xe55ec002;
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
// Interrupt handler in BIOS
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
REG_MODE(MODE_IRQ)[6] = reg[REG_PC] + 4;
REG_SPSR(MODE_IRQ) = reg[REG_CPSR];
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
reg[REG_CPSR] = 0xD2;
reg[REG_PC] = 0x00000018;
set_cpu_mode(MODE_IRQ);
reg[CPU_HALT_STATE] = CPU_ACTIVE;
Remove CHANGED_PC_STATUS, simplify update flow
2023-04-24 20:24:03 +02:00
return 1;
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
}
Remove CHANGED_PC_STATUS, simplify update flow
2023-04-24 20:24:03 +02:00
return 0;
Fix pending interrupts not raised on IME/IE writes Whenever an interrupt is pending and interrupts are disabled (via IME/IE), an IE/IME write that re-enables IRQs will fail to raise an IRQ. This makes some games hang. Most games seem to use CPSR.IRQ to enable and disable interruts, so they are not affected. However some others use IME/IE (or all of them), causing these deadlocks and some race conditions. This fixes a bunch of games that did not crash but would "hang" in some interesting ways.
2023-04-10 12:58:03 +02:00
}
Rewrite I/O side effects write and IRQ triggers This rewrites the way that CPU alerts work, making them a bitmap (since multiple alerts can happen simultaneously, like SMC and IRQ). This doesn't really fix many games but improves accuracy overall and improves performance on some I/O writes (the ones without side effects). The IRQ raising is now decoupled and explicitely called via a new function (check_and_raise_interrupts) to avoid issues such as invalid CPSR values (doesn't seem to bother most games!). There's more side effects missing, so this just lays the ground for more fixes.
2023-04-14 01:41:55 +02:00
// This function marks a pending interrupt but does not raise it.
// It simply updates IF register and returns whether the IRQ needs
// to be raised (that is, IE/IME/CPSR enable the IRQ).
// Safe to call via dynarec without proper registers saved.
cpu_alert_type flag_interrupt(irq_type irq_raised)
Fix pending interrupts not raised on IME/IE writes Whenever an interrupt is pending and interrupts are disabled (via IME/IE), an IE/IME write that re-enables IRQs will fail to raise an IRQ. This makes some games hang. Most games seem to use CPSR.IRQ to enable and disable interruts, so they are not affected. However some others use IME/IE (or all of them), causing these deadlocks and some race conditions. This fixes a bunch of games that did not crash but would "hang" in some interesting ways.
2023-04-10 12:58:03 +02:00
{
Rewrite I/O side effects write and IRQ triggers This rewrites the way that CPU alerts work, making them a bitmap (since multiple alerts can happen simultaneously, like SMC and IRQ). This doesn't really fix many games but improves accuracy overall and improves performance on some I/O writes (the ones without side effects). The IRQ raising is now decoupled and explicitely called via a new function (check_and_raise_interrupts) to avoid issues such as invalid CPSR values (doesn't seem to bother most games!). There's more side effects missing, so this just lays the ground for more fixes.
2023-04-14 01:41:55 +02:00
// Flag interrupt
Fix pending interrupts not raised on IME/IE writes Whenever an interrupt is pending and interrupts are disabled (via IME/IE), an IE/IME write that re-enables IRQs will fail to raise an IRQ. This makes some games hang. Most games seem to use CPSR.IRQ to enable and disable interruts, so they are not affected. However some others use IME/IE (or all of them), causing these deadlocks and some race conditions. This fixes a bunch of games that did not crash but would "hang" in some interesting ways.
2023-04-10 12:58:03 +02:00
write_ioreg(REG_IF, read_ioreg(REG_IF) | irq_raised);
Rewrite I/O side effects write and IRQ triggers This rewrites the way that CPU alerts work, making them a bitmap (since multiple alerts can happen simultaneously, like SMC and IRQ). This doesn't really fix many games but improves accuracy overall and improves performance on some I/O writes (the ones without side effects). The IRQ raising is now decoupled and explicitely called via a new function (check_and_raise_interrupts) to avoid issues such as invalid CPSR values (doesn't seem to bother most games!). There's more side effects missing, so this just lays the ground for more fixes.
2023-04-14 01:41:55 +02:00
return check_interrupt();
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
}
Add HAVE_DYNAREC ifdef
2014-12-10 17:06:09 +01:00
#ifndef HAVE_DYNAREC
Move membuffers close to dynarec area to fix x86 relocs This essentially makes it easier to get a relocation-free text area for x86 so that Android loaders are happy.
2021-07-28 19:12:43 +02:00
// When switching modes set spsr[new_mode] to cpsr. Modifying PC as the
// target of a data proc instruction will set cpsr to spsr[cpu_mode].
u32 reg[64];
u32 spsr[6];
u32 reg_mode[7][7];
Add HAVE_DYNAREC ifdef
2014-12-10 17:06:09 +01:00
u8 *memory_map_read [8 * 1024];
Move OAM RAM to stubs also Makes accesses more efficient for MIPS. Make accesses also fast for palette reads.
2021-03-25 23:01:20 +01:00
u16 oam_ram[512];
Move palettes around to simplify MIPS dynarec Will move also OAM structures to gain a few cycles per load/store. Loads can also be optimized for an extra instruction per access.
2021-03-15 02:25:02 +01:00
u16 palette_ram[512];
u16 palette_ram_converted[512];
Move membuffers close to dynarec area to fix x86 relocs This essentially makes it easier to get a relocation-free text area for x86 so that Android loaders are happy.
2021-07-28 19:12:43 +02:00
u8 ewram[1024 * 256 * 2];
u8 iwram[1024 * 32 * 2];
u8 vram[1024 * 96];
Fix and reenable Android arm 32 bit builds Removed the last bits of text relocations by moving all relevant RAMs to the stub reachable area. This should be as fast or even faster than previous code.
2021-07-31 17:45:46 +02:00
u16 io_registers[512];
Add HAVE_DYNAREC ifdef
2014-12-10 17:06:09 +01:00
#endif
enable -Wall and fix warnings reported by it
2011-09-05 18:31:58 +02:00
void execute_arm(u32 cycles)
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{
u32 pc = reg[REG_PC];
u32 opcode;
u32 condition;
u32 n_flag, z_flag, c_flag, v_flag;
u32 pc_region = (pc >> 15);
u8 *pc_address_block = memory_map_read[pc_region];
u32 new_pc_region;
s32 cycles_remaining;
cpu_alert_type cpu_alert;
u32 old_pc;
Prevent bunch of warnings
2015-07-26 04:44:17 +02:00
(void)old_pc;
Code nits - get rid of '== NULL' / '!= NULL'
2014-12-10 16:40:43 +01:00
if(!pc_address_block)
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
pc_address_block = load_gamepak_page(pc_region & 0x3FF);
Rewrite ROM buffer allocation This improves the existing on-demand ROM paging and also breaks down ROM buffers into 1MB blocks for platforms with memory fragmentation issues. Fixed some potential RTC reg issue in said platforms too. Fixed page pinning on interpreter (would crash due to LRU evictions).
2021-07-14 01:17:55 +02:00
touch_gamepak_page(pc_region);
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
Remove libco This removes libco and all the usages of it (+pthreads). Rewired all dynarecs and interpreter to return after every frame so that libretro can process events. This required to make dynarec re-entrant. Dynarecs were updated to check for new frame on every update (IRQ, cycle exhaustion, I/O write, etc). The performance impact of doing so should be minimal (and definitely outweight the libco gains). While at it, fixed small issues to get a bit more perf: arm dynarec was not idling correctly, mips was using stack when not needed, etc. Tested on PSP (mips), OGA (armv7), Linux (x86 and interpreter). Not tested on Android though.
2021-03-08 18:44:03 +01:00
cycles_remaining = cycles;
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
while(1)
{
Remove libco This removes libco and all the usages of it (+pthreads). Rewired all dynarecs and interpreter to return after every frame so that libretro can process events. This required to make dynarec re-entrant. Dynarecs were updated to check for new frame on every update (IRQ, cycle exhaustion, I/O write, etc). The performance impact of doing so should be minimal (and definitely outweight the libco gains). While at it, fixed small issues to get a bit more perf: arm dynarec was not idling correctly, mips was using stack when not needed, etc. Tested on PSP (mips), OGA (armv7), Linux (x86 and interpreter). Not tested on Android though.
2021-03-08 18:44:03 +01:00
/* Do not execute until CPU is active */
while(reg[CPU_HALT_STATE] != CPU_ACTIVE) {
Remove CHANGED_PC_STATUS, simplify update flow
2023-04-24 20:24:03 +02:00
cycles_remaining = update_gba_cycles(cycles_remaining);
Remove libco This removes libco and all the usages of it (+pthreads). Rewired all dynarecs and interpreter to return after every frame so that libretro can process events. This required to make dynarec re-entrant. Dynarecs were updated to check for new frame on every update (IRQ, cycle exhaustion, I/O write, etc). The performance impact of doing so should be minimal (and definitely outweight the libco gains). While at it, fixed small issues to get a bit more perf: arm dynarec was not idling correctly, mips was using stack when not needed, etc. Tested on PSP (mips), OGA (armv7), Linux (x86 and interpreter). Not tested on Android though.
2021-03-08 18:44:03 +01:00
if (reg[COMPLETED_FRAME])
return;
}
Fix inifinite loop in the interpreter, fixes ~6 games Seems that, on IRQ, it is assumed that the PC will change (which happens most of the time). However if the IRQ is masked it resumes execution. On masked DMA IRQ the interpreter jumps to alert without incrementing the instruction pointer (right after).
2021-08-15 23:02:12 +02:00
cpu_alert = CPU_ALERT_NONE;
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
pc = reg[REG_PC];
extract_flags();
if(reg[REG_CPSR] & 0x20)
goto thumb_loop;
do
{
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
arm_loop:
collapse_flags();
Adding Code Breaker cheat support This works on both interpreter and dynarec. Tested in MIPS, ARM and x86, still needs some more testing, some edge cases can be buggy.
2021-05-05 02:20:00 +02:00
/* Process cheats if we are about to execute the cheat hook */
if (pc == cheat_master_hook)
process_cheats();
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
old_pc = pc;
/* Execute ARM instruction */
using_instruction(arm);
check_pc_region();
pc &= ~0x03;
Enable big-endian devices: gc/wii This patch adds big-endian compatibility in gpsp (in general but only for the interpreter). There's no performance hit for little-endian platforms (should be a no-op) and only add a small overhead in memory accesses for big-endian platforms. Most memory accesses are wrapped with a byteswap instruction and I/O reg accesses are also rewired for proper access (using macros). Video rendering has been fixed to also do byteswaps but there's a couple of games and rendering modes that still seem broken (but they amount to less than 20 games in my tests with 1K ROMs). This also adds build rules and CI for NGC/WII/WIIU (untested)
2021-06-27 01:16:28 +02:00
opcode = readaddress32(pc_address_block, (pc & 0x7FFF));
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
condition = opcode >> 28;
switch(condition)
{
case 0x0:
/* EQ */
if(!z_flag)
arm_next_instruction();
break;
case 0x1:
/* NE */
if(z_flag)
arm_next_instruction();
break;
case 0x2:
/* CS */
if(!c_flag)
arm_next_instruction();
break;
case 0x3:
/* CC */
if(c_flag)
arm_next_instruction();
break;
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0x4:
/* MI */
if(!n_flag)
arm_next_instruction();
break;
Rename cpu_orig.c to cpu.c
2014-12-14 02:51:43 +01:00
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0x5:
/* PL */
if(n_flag)
arm_next_instruction();
break;
Rename cpu_orig.c to cpu.c
2014-12-14 02:51:43 +01:00
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0x6:
/* VS */
if(!v_flag)
arm_next_instruction();
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0x7:
/* VC */
if(v_flag)
arm_next_instruction();
break;
Rename cpu_orig.c to cpu.c
2014-12-14 02:51:43 +01:00
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0x8:
/* HI */
if((c_flag == 0) | z_flag)
arm_next_instruction();
break;
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0x9:
/* LS */
if(c_flag & (z_flag ^ 1))
arm_next_instruction();
break;
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0xA:
/* GE */
if(n_flag != v_flag)
arm_next_instruction();
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0xB:
/* LT */
if(n_flag == v_flag)
arm_next_instruction();
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0xC:
/* GT */
if(z_flag | (n_flag != v_flag))
arm_next_instruction();
break;
Adding Code Breaker cheat support This works on both interpreter and dynarec. Tested in MIPS, ARM and x86, still needs some more testing, some edge cases can be buggy.
2021-05-05 02:20:00 +02:00
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0xD:
/* LE */
if((z_flag == 0) & (n_flag == v_flag))
arm_next_instruction();
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0xE:
/* AL */
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0xF:
/* Reserved - treat as "never" */
arm_next_instruction();
break;
}
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Improve tracing a bit, also in interpreter mode
2023-04-15 02:02:17 +02:00
#ifdef TRACE_INSTRUCTIONS
interp_trace_instruction(pc, 1);
#endif
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
switch((opcode >> 20) & 0xFF)
{
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x00:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
if(opcode & 0x20)
{
/* STRH rd, [rn], -rm */
arm_access_memory(store, no_op, half_reg, u16, yes, - reg[rm]);
}
else
{
/* MUL rd, rm, rs */
arm_multiply(no_op, no);
}
}
else
{
/* AND rd, rn, reg_op */
arm_data_proc(reg[rn] & reg_sh, reg);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x01:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 0:
/* MULS rd, rm, rs */
arm_multiply(no_op, yes);
break;
case 1:
/* LDRH rd, [rn], -rm */
arm_access_memory(load, no_op, half_reg, u16, yes, - reg[rm]);
break;
case 2:
/* LDRSB rd, [rn], -rm */
arm_access_memory(load, no_op, half_reg, s8, yes, - reg[rm]);
break;
case 3:
/* LDRSH rd, [rn], -rm */
arm_access_memory(load, no_op, half_reg, s16, yes, - reg[rm]);
break;
}
}
else
{
/* ANDS rd, rn, reg_op */
arm_data_proc_logic_flags(reg[rn] & reg_sh, reg);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x02:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
if(opcode & 0x20)
{
/* STRH rd, [rn], -rm */
arm_access_memory(store, no_op, half_reg, u16, yes, - reg[rm]);
}
else
{
/* MLA rd, rm, rs, rn */
arm_multiply(+ reg[rn], no);
}
}
else
{
/* EOR rd, rn, reg_op */
arm_data_proc(reg[rn] ^ reg_sh, reg);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x03:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 0:
/* MLAS rd, rm, rs, rn */
arm_multiply(+ reg[rn], yes);
break;
case 1:
/* LDRH rd, [rn], -rm */
arm_access_memory(load, no_op, half_reg, u16, yes, - reg[rm]);
break;
case 2:
/* LDRSB rd, [rn], -rm */
arm_access_memory(load, no_op, half_reg, s8, yes, - reg[rm]);
break;
case 3:
/* LDRSH rd, [rn], -rm */
arm_access_memory(load, no_op, half_reg, s16, yes, - reg[rm]);
break;
}
}
else
{
/* EORS rd, rn, reg_op */
arm_data_proc_logic_flags(reg[rn] ^ reg_sh, reg);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x04:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
/* STRH rd, [rn], -imm */
arm_access_memory(store, no_op, half_imm, u16, yes, - offset);
}
else
{
/* SUB rd, rn, reg_op */
arm_data_proc(reg[rn] - reg_sh, reg);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x05:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 1:
/* LDRH rd, [rn], -imm */
arm_access_memory(load, no_op, half_imm, u16, yes, - offset);
break;
case 2:
/* LDRSB rd, [rn], -imm */
arm_access_memory(load, no_op, half_imm, s8, yes, - offset);
break;
case 3:
/* LDRSH rd, [rn], -imm */
arm_access_memory(load, no_op, half_imm, s16, yes, - offset);
break;
}
}
else
{
/* SUBS rd, rn, reg_op */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_sub_flags(reg[rn], reg_sh, 1, reg);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x06:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
/* STRH rd, [rn], -imm */
arm_access_memory(store, no_op, half_imm, u16, yes, - offset);
}
else
{
/* RSB rd, rn, reg_op */
arm_data_proc(reg_sh - reg[rn], reg);
}
break;
case 0x07:
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 1:
/* LDRH rd, [rn], -imm */
arm_access_memory(load, no_op, half_imm, u16, yes, - offset);
break;
case 2:
/* LDRSB rd, [rn], -imm */
arm_access_memory(load, no_op, half_imm, s8, yes, - offset);
break;
case 3:
/* LDRSH rd, [rn], -imm */
arm_access_memory(load, no_op, half_imm, s16, yes, - offset);
break;
}
}
else
{
/* RSBS rd, rn, reg_op */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_sub_flags(reg_sh, reg[rn], 1, reg);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
}
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x08:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
if(opcode & 0x20)
{
/* STRH rd, [rn], +rm */
arm_access_memory(store, no_op, half_reg, u16, yes, + reg[rm]);
}
else
{
/* UMULL rd, rm, rs */
arm_multiply_long(no_op, no, u);
}
}
else
{
/* ADD rd, rn, reg_op */
arm_data_proc(reg[rn] + reg_sh, reg);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x09:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 0:
/* UMULLS rdlo, rdhi, rm, rs */
arm_multiply_long(no_op, yes, u);
break;
case 1:
/* LDRH rd, [rn], +rm */
arm_access_memory(load, no_op, half_reg, u16, yes, + reg[rm]);
break;
case 2:
/* LDRSB rd, [rn], +rm */
arm_access_memory(load, no_op, half_reg, s8, yes, + reg[rm]);
break;
case 3:
/* LDRSH rd, [rn], +rm */
arm_access_memory(load, no_op, half_reg, s16, yes, + reg[rm]);
break;
}
}
else
{
/* ADDS rd, rn, reg_op */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_add_flags(reg[rn], reg_sh, 0, reg);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x0A:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
if(opcode & 0x20)
{
/* STRH rd, [rn], +rm */
arm_access_memory(store, no_op, half_reg, u16, yes, + reg[rm]);
}
else
{
/* UMLAL rd, rm, rs */
arm_multiply_long(arm_multiply_long_addop(u), no, u);
}
}
else
{
/* ADC rd, rn, reg_op */
arm_data_proc(reg[rn] + reg_sh + c_flag, reg);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x0B:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 0:
/* UMLALS rdlo, rdhi, rm, rs */
arm_multiply_long(arm_multiply_long_addop(u), yes, u);
break;
case 1:
/* LDRH rd, [rn], +rm */
arm_access_memory(load, no_op, half_reg, u16, yes, + reg[rm]);
break;
case 2:
/* LDRSB rd, [rn], +rm */
arm_access_memory(load, no_op, half_reg, s8, yes, + reg[rm]);
break;
case 3:
/* LDRSH rd, [rn], +rm */
arm_access_memory(load, no_op, half_reg, s16, yes, + reg[rm]);
break;
}
}
else
{
/* ADCS rd, rn, reg_op */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_add_flags(reg[rn], reg_sh, c_flag, reg);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x0C:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
if(opcode & 0x20)
{
/* STRH rd, [rn], +imm */
arm_access_memory(store, no_op, half_imm, u16, yes, + offset);
}
else
{
/* SMULL rd, rm, rs */
arm_multiply_long(no_op, no, s);
}
}
else
{
/* SBC rd, rn, reg_op */
arm_data_proc(reg[rn] - (reg_sh + (c_flag ^ 1)), reg);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x0D:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 0:
/* SMULLS rdlo, rdhi, rm, rs */
arm_multiply_long(no_op, yes, s);
break;
case 1:
/* LDRH rd, [rn], +imm */
arm_access_memory(load, no_op, half_imm, u16, yes, + offset);
break;
case 2:
/* LDRSB rd, [rn], +imm */
arm_access_memory(load, no_op, half_imm, s8, yes, + offset);
break;
case 3:
/* LDRSH rd, [rn], +imm */
arm_access_memory(load, no_op, half_imm, s16, yes, + offset);
break;
}
}
else
{
/* SBCS rd, rn, reg_op */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_sub_flags(reg[rn], reg_sh, c_flag, reg);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x0E:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
if(opcode & 0x20)
{
/* STRH rd, [rn], +imm */
arm_access_memory(store, no_op, half_imm, u16, yes, + offset);
}
else
{
/* SMLAL rd, rm, rs */
arm_multiply_long(arm_multiply_long_addop(s), no, s);
}
}
else
{
/* RSC rd, rn, reg_op */
arm_data_proc(reg_sh - reg[rn] + c_flag - 1, reg);
}
break;
case 0x0F:
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 0:
/* SMLALS rdlo, rdhi, rm, rs */
arm_multiply_long(arm_multiply_long_addop(s), yes, s);
break;
case 1:
/* LDRH rd, [rn], +imm */
arm_access_memory(load, no_op, half_imm, u16, yes, + offset);
break;
case 2:
/* LDRSB rd, [rn], +imm */
arm_access_memory(load, no_op, half_imm, s8, yes, + offset);
break;
case 3:
/* LDRSH rd, [rn], +imm */
arm_access_memory(load, no_op, half_imm, s16, yes, + offset);
break;
}
}
else
{
/* RSCS rd, rn, reg_op */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_sub_flags(reg_sh, reg[rn], c_flag, reg);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
}
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x10:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
if(opcode & 0x20)
{
/* STRH rd, [rn - rm] */
arm_access_memory(store, - reg[rm], half_reg, u16, no, no_op);
}
else
{
/* SWP rd, rm, [rn] */
arm_swap(u32);
}
}
else
{
/* MRS rd, cpsr */
arm_psr(reg, read, reg[REG_CPSR]);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x11:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 1:
/* LDRH rd, [rn - rm] */
arm_access_memory(load, - reg[rm], half_reg, u16, no, no_op);
break;
case 2:
/* LDRSB rd, [rn - rm] */
arm_access_memory(load, - reg[rm], half_reg, s8, no, no_op);
break;
case 3:
/* LDRSH rd, [rn - rm] */
arm_access_memory(load, - reg[rm], half_reg, s16, no, no_op);
break;
}
}
else
{
/* TST rd, rn, reg_op */
arm_data_proc_test_logic(reg[rn] & reg_sh, reg);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x12:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
/* STRH rd, [rn - rm]! */
arm_access_memory(store, - reg[rm], half_reg, u16, yes, no_op);
}
else
{
if(opcode & 0x10)
{
/* BX rn */
arm_decode_branchx(opcode);
u32 src = reg[rn];
if(src & 0x01)
{
src -= 1;
arm_pc_offset_update_direct(src);
reg[REG_CPSR] |= 0x20;
goto thumb_loop;
}
else
{
arm_pc_offset_update_direct(src);
}
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_nseq[pc >> 24][1];
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
}
else
{
/* MSR cpsr, rm */
arm_psr(reg, store, cpsr);
}
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x13:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 1:
/* LDRH rd, [rn - rm]! */
arm_access_memory(load, - reg[rm], half_reg, u16, yes, no_op);
break;
case 2:
/* LDRSB rd, [rn - rm]! */
arm_access_memory(load, - reg[rm], half_reg, s8, yes, no_op);
break;
case 3:
/* LDRSH rd, [rn - rm]! */
arm_access_memory(load, - reg[rm], half_reg, s16, yes, no_op);
break;
}
}
else
{
/* TEQ rd, rn, reg_op */
arm_data_proc_test_logic(reg[rn] ^ reg_sh, reg);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x14:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
if(opcode & 0x20)
{
/* STRH rd, [rn - imm] */
arm_access_memory(store, - offset, half_imm, u16, no, no_op);
}
else
{
/* SWPB rd, rm, [rn] */
arm_swap(u8);
}
}
else
{
/* MRS rd, spsr */
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
arm_psr(reg, read, REG_SPSR(reg[CPU_MODE]));
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x15:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 1:
/* LDRH rd, [rn - imm] */
arm_access_memory(load, - offset, half_imm, u16, no, no_op);
break;
case 2:
/* LDRSB rd, [rn - imm] */
arm_access_memory(load, - offset, half_imm, s8, no, no_op);
break;
case 3:
/* LDRSH rd, [rn - imm] */
arm_access_memory(load, - offset, half_imm, s16, no, no_op);
break;
}
}
else
{
/* CMP rn, reg_op */
arm_data_proc_test_sub(reg[rn], reg_sh, reg);
}
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x16:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
/* STRH rd, [rn - imm]! */
arm_access_memory(store, - offset, half_imm, u16, yes, no_op);
}
else
{
/* MSR spsr, rm */
arm_psr(reg, store, spsr);
}
break;
case 0x17:
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 1:
/* LDRH rd, [rn - imm]! */
arm_access_memory(load, - offset, half_imm, u16, yes, no_op);
break;
case 2:
/* LDRSB rd, [rn - imm]! */
arm_access_memory(load, - offset, half_imm, s8, yes, no_op);
break;
case 3:
/* LDRSH rd, [rn - imm]! */
arm_access_memory(load, - offset, half_imm, s16, yes, no_op);
break;
}
}
else
{
/* CMN rd, rn, reg_op */
arm_data_proc_test_add(reg[rn], reg_sh, reg);
}
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x18:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
/* STRH rd, [rn + rm] */
arm_access_memory(store, + reg[rm], half_reg, u16, no, no_op);
}
else
{
/* ORR rd, rn, reg_op */
arm_data_proc(reg[rn] | reg_sh, reg);
}
break;
case 0x19:
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 1:
/* LDRH rd, [rn + rm] */
arm_access_memory(load, + reg[rm], half_reg, u16, no, no_op);
break;
case 2:
/* LDRSB rd, [rn + rm] */
arm_access_memory(load, + reg[rm], half_reg, s8, no, no_op);
break;
case 3:
/* LDRSH rd, [rn + rm] */
arm_access_memory(load, + reg[rm], half_reg, s16, no, no_op);
break;
}
}
else
{
/* ORRS rd, rn, reg_op */
arm_data_proc_logic_flags(reg[rn] | reg_sh, reg);
}
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x1A:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
/* STRH rd, [rn + rm]! */
arm_access_memory(store, + reg[rm], half_reg, u16, yes, no_op);
}
else
{
/* MOV rd, reg_op */
arm_data_proc(reg_sh, reg);
}
break;
case 0x1B:
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 1:
/* LDRH rd, [rn + rm]! */
arm_access_memory(load, + reg[rm], half_reg, u16, yes, no_op);
break;
case 2:
/* LDRSB rd, [rn + rm]! */
arm_access_memory(load, + reg[rm], half_reg, s8, yes, no_op);
break;
case 3:
/* LDRSH rd, [rn + rm]! */
arm_access_memory(load, + reg[rm], half_reg, s16, yes, no_op);
break;
}
}
else
{
/* MOVS rd, reg_op */
arm_data_proc_logic_flags(reg_sh, reg);
}
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x1C:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
/* STRH rd, [rn + imm] */
arm_access_memory(store, + offset, half_imm, u16, no, no_op);
}
else
{
/* BIC rd, rn, reg_op */
arm_data_proc(reg[rn] & (~reg_sh), reg);
}
break;
case 0x1D:
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 1:
/* LDRH rd, [rn + imm] */
arm_access_memory(load, + offset, half_imm, u16, no, no_op);
break;
case 2:
/* LDRSB rd, [rn + imm] */
arm_access_memory(load, + offset, half_imm, s8, no, no_op);
break;
case 3:
/* LDRSH rd, [rn + imm] */
arm_access_memory(load, + offset, half_imm, s16, no, no_op);
break;
}
}
else
{
/* BICS rd, rn, reg_op */
arm_data_proc_logic_flags(reg[rn] & (~reg_sh), reg);
}
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x1E:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if((opcode & 0x90) == 0x90)
{
/* STRH rd, [rn + imm]! */
arm_access_memory(store, + offset, half_imm, u16, yes, no_op);
}
else
{
/* MVN rd, reg_op */
arm_data_proc(~reg_sh, reg);
}
break;
case 0x1F:
if((opcode & 0x90) == 0x90)
{
switch((opcode >> 5) & 0x03)
{
case 1:
/* LDRH rd, [rn + imm]! */
arm_access_memory(load, + offset, half_imm, u16, yes, no_op);
break;
case 2:
/* LDRSB rd, [rn + imm]! */
arm_access_memory(load, + offset, half_imm, s8, yes, no_op);
break;
case 3:
/* LDRSH rd, [rn + imm]! */
arm_access_memory(load, + offset, half_imm, s16, yes, no_op);
break;
}
}
else
{
/* MVNS rd, rn, reg_op */
arm_data_proc_logic_flags(~reg_sh, reg);
}
break;
case 0x20:
/* AND rd, rn, imm */
arm_data_proc(reg[rn] & imm, imm);
break;
case 0x21:
/* ANDS rd, rn, imm */
arm_data_proc_logic_flags(reg[rn] & imm, imm);
break;
case 0x22:
/* EOR rd, rn, imm */
arm_data_proc(reg[rn] ^ imm, imm);
break;
case 0x23:
/* EORS rd, rn, imm */
arm_data_proc_logic_flags(reg[rn] ^ imm, imm);
break;
case 0x24:
/* SUB rd, rn, imm */
arm_data_proc(reg[rn] - imm, imm);
break;
case 0x25:
/* SUBS rd, rn, imm */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_sub_flags(reg[rn], imm, 1, imm);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x26:
/* RSB rd, rn, imm */
arm_data_proc(imm - reg[rn], imm);
break;
case 0x27:
/* RSBS rd, rn, imm */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_sub_flags(imm, reg[rn], 1, imm);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x28:
/* ADD rd, rn, imm */
arm_data_proc(reg[rn] + imm, imm);
break;
case 0x29:
/* ADDS rd, rn, imm */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_add_flags(reg[rn], imm, 0, imm);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x2A:
/* ADC rd, rn, imm */
arm_data_proc(reg[rn] + imm + c_flag, imm);
break;
case 0x2B:
/* ADCS rd, rn, imm */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_add_flags(reg[rn], imm, c_flag, imm);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x2C:
/* SBC rd, rn, imm */
arm_data_proc(reg[rn] - imm + c_flag - 1, imm);
break;
case 0x2D:
/* SBCS rd, rn, imm */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_sub_flags(reg[rn], imm, c_flag, imm);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x2E:
/* RSC rd, rn, imm */
arm_data_proc(imm - reg[rn] + c_flag - 1, imm);
break;
case 0x2F:
/* RSCS rd, rn, imm */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
arm_data_proc_sub_flags(imm, reg[rn], c_flag, imm);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x30:
case 0x31:
/* TST rn, imm */
arm_data_proc_test_logic(reg[rn] & imm, imm);
break;
case 0x32:
/* MSR cpsr, imm */
arm_psr(imm, store, cpsr);
break;
case 0x33:
/* TEQ rn, imm */
arm_data_proc_test_logic(reg[rn] ^ imm, imm);
break;
case 0x34:
case 0x35:
/* CMP rn, imm */
arm_data_proc_test_sub(reg[rn], imm, imm);
break;
case 0x36:
/* MSR spsr, imm */
arm_psr(imm, store, spsr);
break;
case 0x37:
/* CMN rn, imm */
arm_data_proc_test_add(reg[rn], imm, imm);
break;
case 0x38:
/* ORR rd, rn, imm */
arm_data_proc(reg[rn] | imm, imm);
break;
case 0x39:
/* ORRS rd, rn, imm */
arm_data_proc_logic_flags(reg[rn] | imm, imm);
break;
case 0x3A:
/* MOV rd, imm */
arm_data_proc(imm, imm);
break;
case 0x3B:
/* MOVS rd, imm */
arm_data_proc_logic_flags(imm, imm);
break;
case 0x3C:
/* BIC rd, rn, imm */
arm_data_proc(reg[rn] & (~imm), imm);
break;
case 0x3D:
/* BICS rd, rn, imm */
arm_data_proc_logic_flags(reg[rn] & (~imm), imm);
break;
case 0x3E:
/* MVN rd, imm */
arm_data_proc(~imm, imm);
break;
case 0x3F:
/* MVNS rd, imm */
arm_data_proc_logic_flags(~imm, imm);
break;
case 0x40:
/* STR rd, [rn], -imm */
arm_access_memory(store, no_op, imm, u32, yes, - offset);
break;
case 0x41:
/* LDR rd, [rn], -imm */
arm_access_memory(load, no_op, imm, u32, yes, - offset);
break;
case 0x42:
/* STRT rd, [rn], -imm */
arm_access_memory(store, no_op, imm, u32, yes, - offset);
break;
case 0x43:
/* LDRT rd, [rn], -imm */
arm_access_memory(load, no_op, imm, u32, yes, - offset);
break;
case 0x44:
/* STRB rd, [rn], -imm */
arm_access_memory(store, no_op, imm, u8, yes, - offset);
break;
case 0x45:
/* LDRB rd, [rn], -imm */
arm_access_memory(load, no_op, imm, u8, yes, - offset);
break;
case 0x46:
/* STRBT rd, [rn], -imm */
arm_access_memory(store, no_op, imm, u8, yes, - offset);
break;
case 0x47:
/* LDRBT rd, [rn], -imm */
arm_access_memory(load, no_op, imm, u8, yes, - offset);
break;
case 0x48:
/* STR rd, [rn], +imm */
arm_access_memory(store, no_op, imm, u32, yes, + offset);
break;
case 0x49:
/* LDR rd, [rn], +imm */
arm_access_memory(load, no_op, imm, u32, yes, + offset);
break;
case 0x4A:
/* STRT rd, [rn], +imm */
arm_access_memory(store, no_op, imm, u32, yes, + offset);
break;
case 0x4B:
/* LDRT rd, [rn], +imm */
arm_access_memory(load, no_op, imm, u32, yes, + offset);
break;
case 0x4C:
/* STRB rd, [rn], +imm */
arm_access_memory(store, no_op, imm, u8, yes, + offset);
break;
case 0x4D:
/* LDRB rd, [rn], +imm */
arm_access_memory(load, no_op, imm, u8, yes, + offset);
break;
case 0x4E:
/* STRBT rd, [rn], +imm */
arm_access_memory(store, no_op, imm, u8, yes, + offset);
break;
case 0x4F:
/* LDRBT rd, [rn], +imm */
arm_access_memory(load, no_op, imm, u8, yes, + offset);
break;
Rename cpu_orig.c to cpu.c
2014-12-14 02:51:43 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x50:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STR rd, [rn - imm] */
arm_access_memory(store, - offset, imm, u32, no, no_op);
break;
case 0x51:
/* LDR rd, [rn - imm] */
arm_access_memory(load, - offset, imm, u32, no, no_op);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x52:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STR rd, [rn - imm]! */
arm_access_memory(store, - offset, imm, u32, yes, no_op);
break;
case 0x53:
/* LDR rd, [rn - imm]! */
arm_access_memory(load, - offset, imm, u32, yes, no_op);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x54:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRB rd, [rn - imm] */
arm_access_memory(store, - offset, imm, u8, no, no_op);
break;
Rename cpu_orig.c to cpu.c
2014-12-14 02:51:43 +01:00
case 0x55:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDRB rd, [rn - imm] */
arm_access_memory(load, - offset, imm, u8, no, no_op);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x56:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRB rd, [rn - imm]! */
arm_access_memory(store, - offset, imm, u8, yes, no_op);
break;
Rename cpu_orig.c to cpu.c
2014-12-14 02:51:43 +01:00
case 0x57:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDRB rd, [rn - imm]! */
arm_access_memory(load, - offset, imm, u8, yes, no_op);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x58:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STR rd, [rn + imm] */
arm_access_memory(store, + offset, imm, u32, no, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x59:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDR rd, [rn + imm] */
arm_access_memory(load, + offset, imm, u32, no, no_op);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x5A:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STR rd, [rn + imm]! */
arm_access_memory(store, + offset, imm, u32, yes, no_op);
break;
Rename cpu_orig.c to cpu.c
2014-12-14 02:51:43 +01:00
case 0x5B:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDR rd, [rn + imm]! */
arm_access_memory(load, + offset, imm, u32, yes, no_op);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x5C:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRB rd, [rn + imm] */
arm_access_memory(store, + offset, imm, u8, no, no_op);
break;
Rename cpu_orig.c to cpu.c
2014-12-14 02:51:43 +01:00
case 0x5D:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDRB rd, [rn + imm] */
arm_access_memory(load, + offset, imm, u8, no, no_op);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x5E:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRB rd, [rn + imm]! */
arm_access_memory(store, + offset, imm, u8, yes, no_op);
break;
Rename cpu_orig.c to cpu.c
2014-12-14 02:51:43 +01:00
case 0x5F:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDRBT rd, [rn + imm]! */
arm_access_memory(load, + offset, imm, u8, yes, no_op);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x60:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STR rd, [rn], -reg_op */
arm_access_memory(store, no_op, reg, u32, yes, - reg_offset);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x61:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDR rd, [rn], -reg_op */
arm_access_memory(load, no_op, reg, u32, yes, - reg_offset);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x62:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRT rd, [rn], -reg_op */
arm_access_memory(store, no_op, reg, u32, yes, - reg_offset);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x63:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDRT rd, [rn], -reg_op */
arm_access_memory(load, no_op, reg, u32, yes, - reg_offset);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x64:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRB rd, [rn], -reg_op */
arm_access_memory(store, no_op, reg, u8, yes, - reg_offset);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x65:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDRB rd, [rn], -reg_op */
arm_access_memory(load, no_op, reg, u8, yes, - reg_offset);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x66:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRBT rd, [rn], -reg_op */
arm_access_memory(store, no_op, reg, u8, yes, - reg_offset);
break;
case 0x67:
/* LDRBT rd, [rn], -reg_op */
arm_access_memory(load, no_op, reg, u8, yes, - reg_offset);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x68:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STR rd, [rn], +reg_op */
arm_access_memory(store, no_op, reg, u32, yes, + reg_offset);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x69:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDR rd, [rn], +reg_op */
arm_access_memory(load, no_op, reg, u32, yes, + reg_offset);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x6A:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRT rd, [rn], +reg_op */
arm_access_memory(store, no_op, reg, u32, yes, + reg_offset);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x6B:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDRT rd, [rn], +reg_op */
arm_access_memory(load, no_op, reg, u32, yes, + reg_offset);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x6C:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRB rd, [rn], +reg_op */
arm_access_memory(store, no_op, reg, u8, yes, + reg_offset);
break;
case 0x6D:
/* LDRB rd, [rn], +reg_op */
arm_access_memory(load, no_op, reg, u8, yes, + reg_offset);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x6E:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRBT rd, [rn], +reg_op */
arm_access_memory(store, no_op, reg, u8, yes, + reg_offset);
break;
case 0x6F:
/* LDRBT rd, [rn], +reg_op */
arm_access_memory(load, no_op, reg, u8, yes, + reg_offset);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x70:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STR rd, [rn - reg_op] */
arm_access_memory(store, - reg_offset, reg, u32, no, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x71:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDR rd, [rn - reg_op] */
arm_access_memory(load, - reg_offset, reg, u32, no, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x72:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STR rd, [rn - reg_op]! */
arm_access_memory(store, - reg_offset, reg, u32, yes, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x73:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDR rd, [rn - reg_op]! */
arm_access_memory(load, - reg_offset, reg, u32, yes, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x74:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRB rd, [rn - reg_op] */
arm_access_memory(store, - reg_offset, reg, u8, no, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x75:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDRB rd, [rn - reg_op] */
arm_access_memory(load, - reg_offset, reg, u8, no, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x76:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRB rd, [rn - reg_op]! */
arm_access_memory(store, - reg_offset, reg, u8, yes, no_op);
break;
case 0x77:
/* LDRB rd, [rn - reg_op]! */
arm_access_memory(load, - reg_offset, reg, u8, yes, no_op);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x78:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STR rd, [rn + reg_op] */
arm_access_memory(store, + reg_offset, reg, u32, no, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x79:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDR rd, [rn + reg_op] */
arm_access_memory(load, + reg_offset, reg, u32, no, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x7A:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STR rd, [rn + reg_op]! */
arm_access_memory(store, + reg_offset, reg, u32, yes, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x7B:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDR rd, [rn + reg_op]! */
arm_access_memory(load, + reg_offset, reg, u32, yes, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x7C:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRB rd, [rn + reg_op] */
arm_access_memory(store, + reg_offset, reg, u8, no, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x7D:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDRB rd, [rn + reg_op] */
arm_access_memory(load, + reg_offset, reg, u8, no, no_op);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x7E:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STRB rd, [rn + reg_op]! */
arm_access_memory(store, + reg_offset, reg, u8, yes, no_op);
break;
case 0x7F:
/* LDRBT rd, [rn + reg_op]! */
arm_access_memory(load, + reg_offset, reg, u8, yes, no_op);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x80:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STMDA rn, rlist */
arm_block_memory(store, down_a, no, no);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x81:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDMDA rn, rlist */
arm_block_memory(load, down_a, no, no);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x82:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STMDA rn!, rlist */
arm_block_memory(store, down_a, down, no);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x83:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDMDA rn!, rlist */
arm_block_memory(load, down_a, down, no);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x84:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STMDA rn, rlist^ */
arm_block_memory(store, down_a, no, yes);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x85:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDMDA rn, rlist^ */
arm_block_memory(load, down_a, no, yes);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x86:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STMDA rn!, rlist^ */
arm_block_memory(store, down_a, down, yes);
break;
case 0x87:
/* LDMDA rn!, rlist^ */
arm_block_memory(load, down_a, down, yes);
break;
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x88:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STMIA rn, rlist */
arm_block_memory(store, no, no, no);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x89:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDMIA rn, rlist */
arm_block_memory(load, no, no, no);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x8A:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STMIA rn!, rlist */
arm_block_memory(store, no, up, no);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x8B:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDMIA rn!, rlist */
arm_block_memory(load, no, up, no);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x8C:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STMIA rn, rlist^ */
arm_block_memory(store, no, no, yes);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x8D:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* LDMIA rn, rlist^ */
arm_block_memory(load, no, no, yes);
break;
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0x8E:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
/* STMIA rn!, rlist^ */
arm_block_memory(store, no, up, yes);
break;
case 0x8F:
/* LDMIA rn!, rlist^ */
arm_block_memory(load, no, up, yes);
break;
case 0x90:
/* STMDB rn, rlist */
arm_block_memory(store, down_b, no, no);
break;
case 0x91:
/* LDMDB rn, rlist */
arm_block_memory(load, down_b, no, no);
break;
case 0x92:
/* STMDB rn!, rlist */
arm_block_memory(store, down_b, down, no);
break;
case 0x93:
/* LDMDB rn!, rlist */
arm_block_memory(load, down_b, down, no);
break;
case 0x94:
/* STMDB rn, rlist^ */
arm_block_memory(store, down_b, no, yes);
break;
case 0x95:
/* LDMDB rn, rlist^ */
arm_block_memory(load, down_b, no, yes);
break;
case 0x96:
/* STMDB rn!, rlist^ */
arm_block_memory(store, down_b, down, yes);
break;
case 0x97:
/* LDMDB rn!, rlist^ */
arm_block_memory(load, down_b, down, yes);
break;
case 0x98:
/* STMIB rn, rlist */
arm_block_memory(store, up, no, no);
break;
case 0x99:
/* LDMIB rn, rlist */
arm_block_memory(load, up, no, no);
break;
case 0x9A:
/* STMIB rn!, rlist */
arm_block_memory(store, up, up, no);
break;
case 0x9B:
/* LDMIB rn!, rlist */
arm_block_memory(load, up, up, no);
break;
case 0x9C:
/* STMIB rn, rlist^ */
arm_block_memory(store, up, no, yes);
break;
case 0x9D:
/* LDMIB rn, rlist^ */
arm_block_memory(load, up, no, yes);
break;
case 0x9E:
/* STMIB rn!, rlist^ */
arm_block_memory(store, up, up, yes);
break;
case 0x9F:
/* LDMIB rn!, rlist^ */
arm_block_memory(load, up, up, yes);
break;
case 0xA0 ... 0xAF:
{
/* B offset */
arm_decode_branch();
arm_pc_offset_update(offset + 8);
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_nseq[pc >> 24][1];
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
}
case 0xB0 ... 0xBF:
{
/* BL offset */
arm_decode_branch();
reg[REG_LR] = pc + 4;
arm_pc_offset_update(offset + 8);
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_nseq[pc >> 24][1];
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
}
#ifdef HAVE_UNUSED
case 0xC0 ... 0xEF:
/* coprocessor instructions, reserved on GBA */
break;
#endif
case 0xF0 ... 0xFF:
{
/* SWI comment */
u32 swi_comment = opcode & 0x00FFFFFF;
switch(swi_comment >> 16)
{
/* Jump to BIOS SWI handler */
default:
Minor mem fixes around mirrors and u8/u16 accesses Improve the open bus access a bit too.
2021-09-17 21:30:33 +02:00
// After SWI, we read bios[0xE4]
Improve open bus reads on ARM/MIPS
2022-01-26 19:03:14 +01:00
reg[REG_BUS_VALUE] = 0xe3a02004;
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
REG_MODE(MODE_SUPERVISOR)[6] = pc + 4;
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
collapse_flags();
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
REG_SPSR(MODE_SUPERVISOR) = reg[REG_CPSR];
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
reg[REG_PC] = 0x00000008;
arm_update_pc();
Fix SWI handling (disable IRQs) This introduced a potential race condition between the start of a SWI and the BIOS handling the exception by returning to system mode. During this ~10 instruction window, having an IRQ that issues a SWI causes bad behaviour that results in crashes or other weirdness. Fixes a couple of games and potentially many weird and obscure bugs here and there (hard to reproduce sometimes).
2021-09-17 22:22:01 +02:00
// Move to ARM mode, Supervisor mode and disable IRQs
reg[REG_CPSR] = (reg[REG_CPSR] & ~0x3F) | 0x13 | 0x80;
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
set_cpu_mode(MODE_SUPERVISOR);
break;
}
break;
}
}
skip_instruction:
/* End of Execute ARM instruction */
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_seq[(pc >> 24) & 0xF][1];
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if (pc == idle_loop_target_pc && cycles_remaining > 0) cycles_remaining = 0;
Fix inifinite loop in the interpreter, fixes ~6 games Seems that, on IRQ, it is assumed that the PC will change (which happens most of the time). However if the IRQ is masked it resumes execution. On masked DMA IRQ the interpreter jumps to alert without incrementing the instruction pointer (right after).
2021-08-15 23:02:12 +02:00
Rewrite I/O side effects write and IRQ triggers This rewrites the way that CPU alerts work, making them a bitmap (since multiple alerts can happen simultaneously, like SMC and IRQ). This doesn't really fix many games but improves accuracy overall and improves performance on some I/O writes (the ones without side effects). The IRQ raising is now decoupled and explicitely called via a new function (check_and_raise_interrupts) to avoid issues such as invalid CPSR values (doesn't seem to bother most games!). There's more side effects missing, so this just lays the ground for more fixes.
2023-04-14 01:41:55 +02:00
if (cpu_alert & (CPU_ALERT_HALT | CPU_ALERT_IRQ))
Fix inifinite loop in the interpreter, fixes ~6 games Seems that, on IRQ, it is assumed that the PC will change (which happens most of the time). However if the IRQ is masked it resumes execution. On masked DMA IRQ the interpreter jumps to alert without incrementing the instruction pointer (right after).
2021-08-15 23:02:12 +02:00
goto alert;
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
} while(cycles_remaining > 0);
collapse_flags();
Remove CHANGED_PC_STATUS, simplify update flow
2023-04-24 20:24:03 +02:00
cycles_remaining = update_gba_cycles(cycles_remaining);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
if (reg[COMPLETED_FRAME])
return;
continue;
do
{
thumb_loop:
collapse_flags();
/* Process cheats if we are about to execute the cheat hook */
if (pc == cheat_master_hook)
process_cheats();
old_pc = pc;
/* Execute THUMB instruction */
using_instruction(thumb);
check_pc_region();
pc &= ~0x01;
Enable big-endian devices: gc/wii This patch adds big-endian compatibility in gpsp (in general but only for the interpreter). There's no performance hit for little-endian platforms (should be a no-op) and only add a small overhead in memory accesses for big-endian platforms. Most memory accesses are wrapped with a byteswap instruction and I/O reg accesses are also rewired for proper access (using macros). Video rendering has been fixed to also do byteswaps but there's a couple of games and rendering modes that still seem broken (but they amount to less than 20 games in my tests with 1K ROMs). This also adds build rules and CI for NGC/WII/WIIU (untested)
2021-06-27 01:16:28 +02:00
opcode = readaddress16(pc_address_block, (pc & 0x7FFF));
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
Improve tracing a bit, also in interpreter mode
2023-04-15 02:02:17 +02:00
#ifdef TRACE_INSTRUCTIONS
interp_trace_instruction(pc, 0);
#endif
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
switch((opcode >> 8) & 0xFF)
{
case 0x00 ... 0x07:
/* LSL rd, rs, offset */
thumb_shift(shift, lsl, imm);
break;
case 0x08 ... 0x0F:
/* LSR rd, rs, offset */
thumb_shift(shift, lsr, imm);
break;
case 0x10 ... 0x17:
/* ASR rd, rs, offset */
thumb_shift(shift, asr, imm);
break;
case 0x18:
case 0x19:
/* ADD rd, rs, rn */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
thumb_add(add_sub, rd, reg[rs], reg[rn], 0);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x1A:
case 0x1B:
/* SUB rd, rs, rn */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
thumb_sub(add_sub, rd, reg[rs], reg[rn], 1);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x1C:
case 0x1D:
/* ADD rd, rs, imm */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
thumb_add(add_sub_imm, rd, reg[rs], imm, 0);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x1E:
case 0x1F:
/* SUB rd, rs, imm */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
thumb_sub(add_sub_imm, rd, reg[rs], imm, 1);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x20 ... 0x27:
/* MOV r0..7, imm */
thumb_logic(imm, ((opcode >> 8) & 7), imm);
break;
case 0x28 ... 0x2F:
/* CMP r0..7, imm */
thumb_test_sub(imm, reg[(opcode >> 8) & 7], imm);
break;
case 0x30 ... 0x37:
/* ADD r0..7, imm */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
thumb_add(imm, ((opcode >> 8) & 7), reg[(opcode >> 8) & 7], imm, 0);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x38 ... 0x3F:
/* SUB r0..7, imm */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
thumb_sub(imm, ((opcode >> 8) & 7), reg[(opcode >> 8) & 7], imm, 1);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x40:
switch((opcode >> 6) & 0x03)
{
case 0x00:
/* AND rd, rs */
thumb_logic(alu_op, rd, reg[rd] & reg[rs]);
break;
case 0x01:
/* EOR rd, rs */
thumb_logic(alu_op, rd, reg[rd] ^ reg[rs]);
break;
case 0x02:
/* LSL rd, rs */
thumb_shift(alu_op, lsl, reg);
break;
case 0x03:
/* LSR rd, rs */
thumb_shift(alu_op, lsr, reg);
break;
}
break;
case 0x41:
switch((opcode >> 6) & 0x03)
{
case 0x00:
/* ASR rd, rs */
thumb_shift(alu_op, asr, reg);
break;
case 0x01:
/* ADC rd, rs */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
thumb_add(alu_op, rd, reg[rd], reg[rs], c_flag);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x02:
/* SBC rd, rs */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
thumb_sub(alu_op, rd, reg[rd], reg[rs], c_flag);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x03:
/* ROR rd, rs */
thumb_shift(alu_op, ror, reg);
break;
}
break;
case 0x42:
switch((opcode >> 6) & 0x03)
{
case 0x00:
/* TST rd, rs */
thumb_test_logic(alu_op, reg[rd] & reg[rs]);
break;
case 0x01:
/* NEG rd, rs */
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
thumb_sub(alu_op, rd, 0, reg[rs], 1);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x02:
/* CMP rd, rs */
thumb_test_sub(alu_op, reg[rd], reg[rs]);
break;
case 0x03:
/* CMN rd, rs */
thumb_test_add(alu_op, reg[rd], reg[rs]);
break;
}
break;
case 0x43:
switch((opcode >> 6) & 0x03)
{
case 0x00:
/* ORR rd, rs */
thumb_logic(alu_op, rd, reg[rd] | reg[rs]);
break;
case 0x01:
/* MUL rd, rs */
thumb_logic(alu_op, rd, reg[rd] * reg[rs]);
break;
case 0x02:
/* BIC rd, rs */
thumb_logic(alu_op, rd, reg[rd] & (~reg[rs]));
break;
case 0x03:
/* MVN rd, rs */
thumb_logic(alu_op, rd, ~reg[rs]);
break;
}
break;
case 0x44:
/* ADD rd, rs */
thumb_hireg_op(reg[rd] + reg[rs]);
break;
case 0x45:
/* CMP rd, rs */
{
thumb_pc_offset(4);
thumb_decode_hireg_op();
u32 _sa = reg[rd];
u32 _sb = reg[rs];
u32 dest = _sa - _sb;
thumb_pc_offset(-2);
[interpreter] Fix many flag related bugs including Op2 C flag This fixes a few games and GBA-V too. C flag was neglected on Op2 immediate. Many add/adc related ops were broken due to carry in.
2021-12-14 00:46:37 +01:00
calculate_flags_sub(dest, _sa, _sb, 1);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
}
break;
case 0x46:
/* MOV rd, rs */
thumb_hireg_op(reg[rs]);
break;
case 0x47:
/* BX rs */
{
thumb_decode_hireg_op();
u32 src;
thumb_pc_offset(4);
src = reg[rs];
if(src & 0x01)
{
src -= 1;
thumb_pc_offset_update_direct(src);
}
else
{
/* Switch to ARM mode */
thumb_pc_offset_update_direct(src);
reg[REG_CPSR] &= ~0x20;
collapse_flags();
goto arm_loop;
}
}
break;
case 0x48 ... 0x4F:
/* LDR r0..7, [pc + imm] */
Fix interpreter bug on store-triggered DMAs In thumb mode, a store that triggers a DMA and its correspondoing DMA IRQ (since emulate DMAs being instantanious, which is another can of worms tho...) overrwrites the PC with the PC-next value (disregarding the IRQ handler address). This causes quite a few bugs. I'd expect it causing way more bugs, but it doesnt...? Anyway this fixes a few games like Buffy, Woody, Medabots, Super Duper Sumos and Power Rangers
2023-04-03 19:54:07 +02:00
thumb_access_memory(load, imm, ((pc-2) & ~2) + (imm * 4) + 4, reg[(opcode >> 8) & 7], u32);
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
case 0x50:
case 0x51:
/* STR rd, [rb + ro] */
thumb_access_memory(store, mem_reg, reg[rb] + reg[ro], reg[rd], u32);
break;
case 0x52:
case 0x53:
/* STRH rd, [rb + ro] */
thumb_access_memory(store, mem_reg, reg[rb] + reg[ro], reg[rd], u16);
break;
case 0x54:
case 0x55:
/* STRB rd, [rb + ro] */
thumb_access_memory(store, mem_reg, reg[rb] + reg[ro], reg[rd], u8);
break;
case 0x56:
case 0x57:
/* LDSB rd, [rb + ro] */
thumb_access_memory(load, mem_reg, reg[rb] + reg[ro], reg[rd], s8);
break;
case 0x58:
case 0x59:
/* LDR rd, [rb + ro] */
thumb_access_memory(load, mem_reg, reg[rb] + reg[ro], reg[rd], u32);
break;
case 0x5A:
case 0x5B:
/* LDRH rd, [rb + ro] */
thumb_access_memory(load, mem_reg, reg[rb] + reg[ro], reg[rd], u16);
break;
case 0x5C:
case 0x5D:
/* LDRB rd, [rb + ro] */
thumb_access_memory(load, mem_reg, reg[rb] + reg[ro], reg[rd], u8);
break;
case 0x5E:
case 0x5F:
/* LDSH rd, [rb + ro] */
thumb_access_memory(load, mem_reg, reg[rb] + reg[ro], reg[rd], s16);
break;
case 0x60 ... 0x67:
/* STR rd, [rb + imm] */
thumb_access_memory(store, mem_imm, reg[rb] + (imm * 4), reg[rd], u32);
break;
case 0x68 ... 0x6F:
/* LDR rd, [rb + imm] */
thumb_access_memory(load, mem_imm, reg[rb] + (imm * 4), reg[rd], u32);
break;
case 0x70 ... 0x77:
/* STRB rd, [rb + imm] */
thumb_access_memory(store, mem_imm, reg[rb] + imm, reg[rd], u8);
break;
case 0x78 ... 0x7F:
/* LDRB rd, [rb + imm] */
thumb_access_memory(load, mem_imm, reg[rb] + imm, reg[rd], u8);
break;
case 0x80 ... 0x87:
/* STRH rd, [rb + imm] */
thumb_access_memory(store, mem_imm, reg[rb] + (imm * 2), reg[rd], u16);
break;
case 0x88 ... 0x8F:
/* LDRH rd, [rb + imm] */
thumb_access_memory(load, mem_imm, reg[rb] + (imm * 2), reg[rd], u16);
break;
case 0x90 ... 0x97:
/* STR r0..7, [sp + imm] */
thumb_access_memory(store, imm, reg[REG_SP] + (imm * 4), reg[(opcode >> 8) & 7], u32);
break;
case 0x98 ... 0x9F:
/* LDR r0..7, [sp + imm] */
thumb_access_memory(load, imm, reg[REG_SP] + (imm * 4), reg[(opcode >> 8) & 7], u32);
break;
case 0xA0 ... 0xA7:
/* ADD r0..7, pc, +imm */
thumb_add_noflags(imm, ((opcode >> 8) & 7), (pc & ~2) + 4, (imm * 4));
break;
case 0xA8 ... 0xAF:
/* ADD r0..7, sp, +imm */
thumb_add_noflags(imm, ((opcode >> 8) & 7), reg[REG_SP], (imm * 4));
break;
Rename cpu_orig.c to cpu.c
2014-12-14 02:51:43 +01:00
Remove a lot of the GNU case range extensions instances
2014-12-13 04:50:48 +01:00
case 0xB0:
case 0xB1:
case 0xB2:
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
case 0xB3:
if((opcode >> 7) & 0x01)
{
/* ADD sp, -imm */
thumb_add_noflags(add_sp, 13, reg[REG_SP], -(imm * 4));
}
else
{
/* ADD sp, +imm */
thumb_add_noflags(add_sp, 13, reg[REG_SP], (imm * 4));
}
break;
case 0xB4:
/* PUSH rlist */
thumb_block_memory(store, down, no_op, 13);
break;
case 0xB5:
/* PUSH rlist, lr */
thumb_block_memory(store, push_lr, push_lr, 13);
break;
case 0xBC:
/* POP rlist */
thumb_block_memory(load, no_op, up, 13);
break;
case 0xBD:
/* POP rlist, pc */
thumb_block_memory(load, no_op, pop_pc, 13);
break;
case 0xC0 ... 0xC7:
/* STMIA r0..7!, rlist */
thumb_block_memory(store, no_op, up, ((opcode >> 8) & 7));
break;
case 0xC8 ... 0xCF:
/* LDMIA r0..7!, rlist */
thumb_block_memory(load, no_op, up, ((opcode >> 8) & 7));
break;
case 0xD0:
/* BEQ label */
thumb_conditional_branch(z_flag == 1);
break;
case 0xD1:
/* BNE label */
thumb_conditional_branch(z_flag == 0);
break;
case 0xD2:
/* BCS label */
thumb_conditional_branch(c_flag == 1);
break;
case 0xD3:
/* BCC label */
thumb_conditional_branch(c_flag == 0);
break;
case 0xD4:
/* BMI label */
thumb_conditional_branch(n_flag == 1);
break;
case 0xD5:
/* BPL label */
thumb_conditional_branch(n_flag == 0);
break;
case 0xD6:
/* BVS label */
thumb_conditional_branch(v_flag == 1);
break;
case 0xD7:
/* BVC label */
thumb_conditional_branch(v_flag == 0);
break;
case 0xD8:
/* BHI label */
thumb_conditional_branch(c_flag & (z_flag ^ 1));
break;
case 0xD9:
/* BLS label */
thumb_conditional_branch((c_flag == 0) | z_flag);
break;
case 0xDA:
/* BGE label */
thumb_conditional_branch(n_flag == v_flag);
break;
case 0xDB:
/* BLT label */
thumb_conditional_branch(n_flag != v_flag);
break;
case 0xDC:
/* BGT label */
thumb_conditional_branch((z_flag == 0) & (n_flag == v_flag));
break;
case 0xDD:
/* BLE label */
thumb_conditional_branch(z_flag | (n_flag != v_flag));
break;
case 0xDF:
{
/* SWI comment */
u32 swi_comment = opcode & 0xFF;
switch(swi_comment)
{
default:
Minor mem fixes around mirrors and u8/u16 accesses Improve the open bus access a bit too.
2021-09-17 21:30:33 +02:00
// After SWI, we read bios[0xE4]
Improve open bus reads on ARM/MIPS
2022-01-26 19:03:14 +01:00
reg[REG_BUS_VALUE] = 0xe3a02004;
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
REG_MODE(MODE_SUPERVISOR)[6] = pc + 2;
REG_SPSR(MODE_SUPERVISOR) = reg[REG_CPSR];
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
reg[REG_PC] = 0x00000008;
thumb_update_pc();
Fix SWI handling (disable IRQs) This introduced a potential race condition between the start of a SWI and the BIOS handling the exception by returning to system mode. During this ~10 instruction window, having an IRQ that issues a SWI causes bad behaviour that results in crashes or other weirdness. Fixes a couple of games and potentially many weird and obscure bugs here and there (hard to reproduce sometimes).
2021-09-17 22:22:01 +02:00
// Move to ARM mode, Supervisor mode and disable IRQs
reg[REG_CPSR] = (reg[REG_CPSR] & ~0x3F) | 0x13 | 0x80;
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
set_cpu_mode(MODE_SUPERVISOR);
collapse_flags();
goto arm_loop;
}
break;
}
case 0xE0 ... 0xE7:
{
/* B label */
thumb_decode_branch();
thumb_pc_offset_update(((s32)(offset << 21) >> 20) + 4);
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_nseq[pc >> 24][0];
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
}
case 0xF0 ... 0xF7:
{
/* (low word) BL label */
thumb_decode_branch();
reg[REG_LR] = pc + 4 + ((s32)(offset << 21) >> 9);
thumb_pc_offset(2);
break;
}
case 0xF8 ... 0xFF:
{
/* (high word) BL label */
thumb_decode_branch();
u32 lr = (pc + 2) | 0x01;
pc = reg[REG_LR] + (offset * 2);
reg[REG_LR] = lr;
reg[REG_PC] = pc;
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_nseq[pc >> 24][0];
Small optimization (~2-4%) and whitespace cleanup! Cleans up a ton of whitespace in cpu.c (like 100KB!) and improves readability of some massive decode statements. Added an optimization for PC-relative loads (pool load) in ROM (since it's read only and cannot possibily change) that directly emits an immediate load. This is way faster, specially in MIPS/x86, ARM can be even faster if we rewrite the immediate load macros to also use a pool.
2021-05-07 20:41:54 +02:00
break;
}
}
execute_arm_instruction and execute_thumb_instruction - throw them straight into execute_arm/execute_thumb functions
2014-12-12 06:12:00 +01:00
/* End of Execute THUMB instruction */
[interp] Improve interpreter timings and honor WAITCNT This fixes a few games and makes the interpreter faster (since it doesn't run an overclocked CPU anymore).
2023-06-07 19:40:27 +02:00
cycles_remaining -= ws_cyc_seq[(pc >> 24) & 0xF][0];
Rename cpu_orig.c to cpu.c
2014-12-14 02:51:43 +01:00
if (pc == idle_loop_target_pc && cycles_remaining > 0) cycles_remaining = 0;
Fix inifinite loop in the interpreter, fixes ~6 games Seems that, on IRQ, it is assumed that the PC will change (which happens most of the time). However if the IRQ is masked it resumes execution. On masked DMA IRQ the interpreter jumps to alert without incrementing the instruction pointer (right after).
2021-08-15 23:02:12 +02:00
Rewrite I/O side effects write and IRQ triggers This rewrites the way that CPU alerts work, making them a bitmap (since multiple alerts can happen simultaneously, like SMC and IRQ). This doesn't really fix many games but improves accuracy overall and improves performance on some I/O writes (the ones without side effects). The IRQ raising is now decoupled and explicitely called via a new function (check_and_raise_interrupts) to avoid issues such as invalid CPSR values (doesn't seem to bother most games!). There's more side effects missing, so this just lays the ground for more fixes.
2023-04-14 01:41:55 +02:00
if (cpu_alert & (CPU_ALERT_HALT | CPU_ALERT_IRQ))
Fix inifinite loop in the interpreter, fixes ~6 games Seems that, on IRQ, it is assumed that the PC will change (which happens most of the time). However if the IRQ is masked it resumes execution. On masked DMA IRQ the interpreter jumps to alert without incrementing the instruction pointer (right after).
2021-08-15 23:02:12 +02:00
goto alert;
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
} while(cycles_remaining > 0);
collapse_flags();
Remove CHANGED_PC_STATUS, simplify update flow
2023-04-24 20:24:03 +02:00
cycles_remaining = update_gba_cycles(cycles_remaining);
Remove libco This removes libco and all the usages of it (+pthreads). Rewired all dynarecs and interpreter to return after every frame so that libretro can process events. This required to make dynarec re-entrant. Dynarecs were updated to check for new frame on every update (IRQ, cycle exhaustion, I/O write, etc). The performance impact of doing so should be minimal (and definitely outweight the libco gains). While at it, fixed small issues to get a bit more perf: arm dynarec was not idling correctly, mips was using stack when not needed, etc. Tested on PSP (mips), OGA (armv7), Linux (x86 and interpreter). Not tested on Android though.
2021-03-08 18:44:03 +01:00
if (reg[COMPLETED_FRAME])
return;
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
continue;
alert:
Rewrite I/O side effects write and IRQ triggers This rewrites the way that CPU alerts work, making them a bitmap (since multiple alerts can happen simultaneously, like SMC and IRQ). This doesn't really fix many games but improves accuracy overall and improves performance on some I/O writes (the ones without side effects). The IRQ raising is now decoupled and explicitely called via a new function (check_and_raise_interrupts) to avoid issues such as invalid CPSR values (doesn't seem to bother most games!). There's more side effects missing, so this just lays the ground for more fixes.
2023-04-14 01:41:55 +02:00
/* CPU stopped or switch to IRQ handler */
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
collapse_flags();
}
}
Style nits
2014-12-10 15:47:19 +01:00
void init_cpu(void)
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
{
Move membuffers close to dynarec area to fix x86 relocs This essentially makes it easier to get a relocation-free text area for x86 so that Android loaders are happy.
2021-07-28 19:12:43 +02:00
// Initialize CPU registers
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
int i;
[ARM] Remove cross calls between cache and text (far calls) This is to allow cache to be mapped far from the regular .text section
2021-10-29 22:09:22 +02:00
memset(reg, 0, REG_USERDEF * sizeof(u32));
Move membuffers close to dynarec area to fix x86 relocs This essentially makes it easier to get a relocation-free text area for x86 so that Android loaders are happy.
2021-07-28 19:12:43 +02:00
memset(reg_mode, 0, sizeof(reg_mode));
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
for (i = 0; i < sizeof(spsr)/sizeof(spsr[0]); i++)
spsr[i] = 0x00000010;
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
reg[CPU_HALT_STATE] = CPU_ACTIVE;
Adding Normmatt's BIOS as a built-in BIOS Add options to select whether to boot from BIOS (default is no, as it is now) and whether to use the original bios or the builtin one (default is auto, which tries to use the official but falls back to the builtin if not found).
2021-03-22 21:45:52 +01:00
if (selected_boot_mode == boot_game) {
reg[REG_SP] = 0x03007F00;
reg[REG_PC] = 0x08000000;
reg[REG_CPSR] = 0x0000001F; // system mode
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
reg[CPU_MODE] = MODE_SYSTEM;
Adding Normmatt's BIOS as a built-in BIOS Add options to select whether to boot from BIOS (default is no, as it is now) and whether to use the original bios or the builtin one (default is auto, which tries to use the official but falls back to the builtin if not found).
2021-03-22 21:45:52 +01:00
} else {
reg[REG_SP] = 0x03007F00;
reg[REG_PC] = 0x00000000;
reg[REG_CPSR] = 0x00000013 | 0xC0; // supervisor
reg[CPU_MODE] = MODE_SUPERVISOR;
}
// Stack pointers are set by BIOS, we set them
// nevertheless, should we not boot from BIOS
[all] Fix CPSR and CPU modes gpsp doesn't differentiate between USER and SYSTEM mode, most likely cause it is not that important for most games. This implements the modes correctly and adds checks for privileged operations. Still some bugs/hacks but it mostly fixes CPSR/SPSR reads/writes. To implement PSR writes we are using a more refined masks and force mode bit num. 4 to always be one. Reserved bits are forced to zero (this needs to be validated on a real device).
2023-01-11 20:44:56 +01:00
REG_MODE(MODE_USER)[5] = 0x03007F00;
REG_MODE(MODE_IRQ)[5] = 0x03007FA0;
REG_MODE(MODE_FIQ)[5] = 0x03007FA0;
REG_MODE(MODE_SUPERVISOR)[5] = 0x03007FE0;
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
}
New savestate implementation This uses BSON as savestate format, to allow external tools to parse it (so that we can add proper test of the states). The BSON is not 100% correct according to spec (no ordered keys) but can be parsed by most libraries. This fixes also a bug in the savestate palette color recalculation that was wrongly overwritting the original palette (which could cause some problems on some games). Also fixes some potential issues by serializing some more stuff and cleans up unused stuff. Testing shows that states look good and there's only minor differences in audio ticks, related to buffer sizes (since buffer flushes are de-synced from video frames due to different frequency).
2021-08-24 17:15:27 +02:00
bool cpu_read_savestate(const u8 *src)
{
const u8 *cpudoc = bson_find_key(src, "cpu");
Improve open bus reads on ARM/MIPS
2022-01-26 19:03:14 +01:00
return bson_read_int32(cpudoc, "bus-value", &reg[REG_BUS_VALUE]) &&
bson_read_int32_array(cpudoc, "regs", reg, REG_ARCH_COUNT) &&
New savestate implementation This uses BSON as savestate format, to allow external tools to parse it (so that we can add proper test of the states). The BSON is not 100% correct according to spec (no ordered keys) but can be parsed by most libraries. This fixes also a bug in the savestate palette color recalculation that was wrongly overwritting the original palette (which could cause some problems on some games). Also fixes some potential issues by serializing some more stuff and cleans up unused stuff. Testing shows that states look good and there's only minor differences in audio ticks, related to buffer sizes (since buffer flushes are de-synced from video frames due to different frequency).
2021-08-24 17:15:27 +02:00
bson_read_int32_array(cpudoc, "spsr", spsr, 6) &&
bson_read_int32_array(cpudoc, "regmod", (u32*)reg_mode, 7*7);
implement retro_serialize/unserialize
2014-12-10 11:06:17 +01:00
}
original source from gpsp09-2xb_src.tar.bz2
2009-05-21 17:48:31 +02:00
New savestate implementation This uses BSON as savestate format, to allow external tools to parse it (so that we can add proper test of the states). The BSON is not 100% correct according to spec (no ordered keys) but can be parsed by most libraries. This fixes also a bug in the savestate palette color recalculation that was wrongly overwritting the original palette (which could cause some problems on some games). Also fixes some potential issues by serializing some more stuff and cleans up unused stuff. Testing shows that states look good and there's only minor differences in audio ticks, related to buffer sizes (since buffer flushes are de-synced from video frames due to different frequency).
2021-08-24 17:15:27 +02:00
unsigned cpu_write_savestate(u8 *dst)
{
u8 *wbptr, *startp = dst;
bson_start_document(dst, "cpu", wbptr);
Improve open bus reads on ARM/MIPS
2022-01-26 19:03:14 +01:00
bson_write_int32array(dst, "regs", reg, REG_ARCH_COUNT);
New savestate implementation This uses BSON as savestate format, to allow external tools to parse it (so that we can add proper test of the states). The BSON is not 100% correct according to spec (no ordered keys) but can be parsed by most libraries. This fixes also a bug in the savestate palette color recalculation that was wrongly overwritting the original palette (which could cause some problems on some games). Also fixes some potential issues by serializing some more stuff and cleans up unused stuff. Testing shows that states look good and there's only minor differences in audio ticks, related to buffer sizes (since buffer flushes are de-synced from video frames due to different frequency).
2021-08-24 17:15:27 +02:00
bson_write_int32array(dst, "spsr", spsr, 6);
bson_write_int32array(dst, "regmod", reg_mode, 7*7);
Improve open bus reads on ARM/MIPS
2022-01-26 19:03:14 +01:00
bson_write_int32(dst, "bus-value", reg[REG_BUS_VALUE]);
New savestate implementation This uses BSON as savestate format, to allow external tools to parse it (so that we can add proper test of the states). The BSON is not 100% correct according to spec (no ordered keys) but can be parsed by most libraries. This fixes also a bug in the savestate palette color recalculation that was wrongly overwritting the original palette (which could cause some problems on some games). Also fixes some potential issues by serializing some more stuff and cleans up unused stuff. Testing shows that states look good and there's only minor differences in audio ticks, related to buffer sizes (since buffer flushes are de-synced from video frames due to different frequency).
2021-08-24 17:15:27 +02:00
bson_finish_document(dst, wbptr);
return (unsigned int)(dst - startp);
}