/* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
- * Mupen64plus - linkage_arm.s *
- * Copyright (C) 2009-2010 Ari64 *
+ * linkage_arm.s for PCSX *
+ * Copyright (C) 2009-2011 Ari64 *
+ * Copyright (C) 2010-2011 GraÅžvydas "notaz" Ignotas *
* *
* 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 *
* Free Software Foundation, Inc., *
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
- .cpu arm9tdmi
- .fpu softvfp
- .eabi_attribute 20, 1
- .eabi_attribute 21, 1
- .eabi_attribute 23, 3
- .eabi_attribute 24, 1
- .eabi_attribute 25, 1
- .eabi_attribute 26, 2
- .eabi_attribute 30, 6
- .eabi_attribute 18, 4
- .file "linkage_arm.s"
+
+/* .equiv HAVE_ARMV7, 1 */
+
.global rdram
rdram = 0x80000000
.global dynarec_local
.global reg
.global hi
.global lo
- .global reg_cop1_simple
- .global reg_cop1_double
.global reg_cop0
+ .global reg_cop2d
+ .global reg_cop2c
.global FCR0
.global FCR31
- .global rounding_modes
.global next_interupt
.global cycle_count
.global last_count
.global stop
.global invc_ptr
.global address
- .global readmem_dword
- .global dword
- .global word
- .global hword
- .global byte
.global branch_target
.global PC
- .global fake_pc
.global mini_ht
.global restore_candidate
- .global memory_map
+ /* psx */
+ .global psxRegs
+ .global mem_rtab
+ .global mem_wtab
+ .global psxH_ptr
+ .global inv_code_start
+ .global inv_code_end
+ .global rcnts
+
.bss
.align 4
.type dynarec_local, %object
- .size dynarec_local, 64
+ .size dynarec_local, dynarec_local_end-dynarec_local
dynarec_local:
- .space 64+16+16+8+8+8+8+256+8+8+128+128+128+16+8+132+4+256+512+4194304
+ .space dynarec_local_end-dynarec_local
next_interupt = dynarec_local + 64
.type next_interupt, %object
.size next_interupt, 4
pending_exception = last_count + 4
.type pending_exception, %object
.size pending_exception, 4
-pcaddr = pending_exception + 4
- .type pcaddr, %object
- .size pcaddr, 4
-stop = pcaddr + 4
+stop = pending_exception + 4
.type stop, %object
.size stop, 4
invc_ptr = stop + 4
address = invc_ptr + 4
.type address, %object
.size address, 4
-readmem_dword = address + 4
- .type readmem_dword, %object
- .size readmem_dword, 8
-dword = readmem_dword + 8
- .type dword, %object
- .size dword, 8
-word = dword + 8
- .type word, %object
- .size word, 4
-hword = word + 4
- .type hword, %object
- .size hword, 2
-byte = hword + 2
- .type byte, %object
- .size byte, 1 /* 1 byte free */
-FCR0 = hword + 4
- .type FCR0, %object
- .size FCR0, 4
-FCR31 = FCR0 + 4
- .type FCR31, %object
- .size FCR31, 4
-reg = FCR31 + 4
+psxRegs = address + 4
+
+/* psxRegs */
+ .type psxRegs, %object
+ .size psxRegs, psxRegs_end-psxRegs
+reg = psxRegs
.type reg, %object
- .size reg, 256
-hi = reg + 256
- .type hi, %object
- .size hi, 8
-lo = hi + 8
+ .size reg, 128
+lo = reg + 128
.type lo, %object
- .size lo, 8
-reg_cop0 = lo + 8
+ .size lo, 4
+hi = lo + 4
+ .type hi, %object
+ .size hi, 4
+reg_cop0 = hi + 4
.type reg_cop0, %object
.size reg_cop0, 128
-reg_cop1_simple = reg_cop0 + 128
- .type reg_cop1_simple, %object
- .size reg_cop1_simple, 128
-reg_cop1_double = reg_cop1_simple + 128
- .type reg_cop1_double, %object
- .size reg_cop1_double, 128
-rounding_modes = reg_cop1_double + 128
- .type rounding_modes, %object
- .size rounding_modes, 16
-branch_target = rounding_modes + 16
- .type branch_target, %object
- .size branch_target, 4
-PC = branch_target + 4
+reg_cop2d = reg_cop0 + 128
+ .type reg_cop2d, %object
+ .size reg_cop2d, 128
+reg_cop2c = reg_cop2d + 128
+ .type reg_cop2c, %object
+ .size reg_cop2c, 128
+PC = reg_cop2c + 128
+pcaddr = PC
.type PC, %object
.size PC, 4
-fake_pc = PC + 4
- .type fake_pc, %object
- .size fake_pc, 132
-/* 4 bytes free */
-mini_ht = fake_pc + 136
+code = PC + 4
+ .type code, %object
+ .size code, 4
+cycle = code + 4
+ .type cycle, %object
+ .size cycle, 4
+interrupt = cycle + 4
+ .type interrupt, %object
+ .size interrupt, 4
+intCycle = interrupt + 4
+ .type intCycle, %object
+ .size intCycle, 256
+psxRegs_end = intCycle + 256
+
+rcnts = psxRegs_end
+ .type rcnts, %object
+ .size rcnts, 7*4*4
+rcnts_end = rcnts + 7*4*4
+
+mem_rtab = rcnts_end
+ .type mem_rtab, %object
+ .size mem_rtab, 4
+mem_wtab = mem_rtab + 4
+ .type mem_wtab, %object
+ .size mem_wtab, 4
+psxH_ptr = mem_wtab + 4
+ .type psxH_ptr, %object
+ .size psxH_ptr, 4
+inv_code_start = psxH_ptr + 4
+ .type inv_code_start, %object
+ .size inv_code_start, 4
+inv_code_end = inv_code_start + 4
+ .type inv_code_end, %object
+ .size inv_code_end, 4
+branch_target = inv_code_end + 4
+ .type branch_target, %object
+ .size branch_target, 4
+align0 = branch_target + 4 /* unused/alignment */
+ .type align0, %object
+ .size align0, 4
+mini_ht = align0 + 4
.type mini_ht, %object
.size mini_ht, 256
restore_candidate = mini_ht + 256
.type restore_candidate, %object
.size restore_candidate, 512
-memory_map = restore_candidate + 512
- .type memory_map, %object
- .size memory_map, 4194304
+dynarec_local_end = restore_candidate + 512
- .text
- .align 2
- .global dyna_linker
- .type dyna_linker, %function
-dyna_linker:
+/* unused */
+FCR0 = align0
+ .type FCR0, %object
+ .size FCR0, 4
+FCR31 = align0
+ .type FCR31, %object
+ .size FCR31, 4
+
+.macro load_var_adr reg var
+.if HAVE_ARMV7
+ movw \reg, #:lower16:\var
+ movt \reg, #:upper16:\var
+.else
+ ldr \reg, =\var
+.endif
+.endm
+
+.macro mov_16 reg imm
+.if HAVE_ARMV7
+ movw \reg, #\imm
+.else
+ mov \reg, #(\imm & 0x00ff)
+ orr \reg, #(\imm & 0xff00)
+.endif
+.endm
+
+.macro mov_24 reg imm
+.if HAVE_ARMV7
+ movw \reg, #(\imm & 0xffff)
+ movt \reg, #(\imm >> 16)
+.else
+ mov \reg, #(\imm & 0x0000ff)
+ orr \reg, #(\imm & 0x00ff00)
+ orr \reg, #(\imm & 0xff0000)
+.endif
+.endm
+
+.macro dyna_linker_main
/* r0 = virtual target address */
/* r1 = instruction to patch */
- ldr r4, .tlbptr
- lsr r5, r0, #12
- mov r12, r0
- cmp r0, #0xC0000000
- mov r6, #4096
- ldrge r12, [r4, r5, lsl #2]
- mov r2, #0x80000
ldr r3, .jiptr
- tst r12, r12
+ /* get_page */
+ lsr r2, r0, #12
+ mov r6, #4096
+ bic r2, r2, #0xe0000
sub r6, r6, #1
- moveq r12, r0
+ cmp r2, #0x1000
ldr r7, [r1]
- eor r2, r2, r12, lsr #12
- and r6, r6, r12, lsr #12
+ biclt r2, #0x0e00
+ and r6, r6, r2
cmp r2, #2048
add r12, r7, #2
orrcs r2, r6, #2048
ldr r5, [r3, r2, lsl #2]
lsl r12, r12, #8
+ add r6, r1, r12, asr #6
+ mov r8, #0
/* jump_in lookup */
-.A1:
+1:
movs r4, r5
- beq .A3
+ beq 2f
ldr r3, [r5]
ldr r5, [r4, #12]
teq r3, r0
- bne .A1
+ bne 1b
ldr r3, [r4, #4]
ldr r4, [r4, #8]
tst r3, r3
- bne .A1
-.A2:
- mov r5, r1
- add r1, r1, r12, asr #6
- teq r1, r4
+ bne 1b
+ teq r4, r6
moveq pc, r4 /* Stale i-cache */
+ mov r8, r4
+ b 1b /* jump_in may have dupes, continue search */
+2:
+ tst r8, r8
+ beq 3f /* r0 not in jump_in */
+
+ mov r5, r1
+ mov r1, r6
bl add_link
- sub r2, r4, r5
+ sub r2, r8, r5
and r1, r7, #0xff000000
lsl r2, r2, #6
sub r1, r1, #2
add r1, r1, r2, lsr #8
str r1, [r5]
- mov pc, r4
-.A3:
+ mov pc, r8
+3:
/* hash_table lookup */
cmp r2, #2048
ldr r3, .jdptr
teq r7, r0
ldreq pc, [r6, #12]
/* jump_dirty lookup */
-.A6:
+6:
movs r4, r5
- beq .A8
+ beq 8f
ldr r3, [r5]
ldr r5, [r4, #12]
teq r3, r0
- bne .A6
-.A7:
+ bne 6b
+7:
ldr r1, [r4, #8]
/* hash_table insert */
ldr r2, [r6]
str r2, [r6, #8]
str r3, [r6, #12]
mov pc, r1
-.A8:
+8:
+.endm
+
+ .text
+ .align 2
+ .global dyna_linker
+ .type dyna_linker, %function
+dyna_linker:
+ /* r0 = virtual target address */
+ /* r1 = instruction to patch */
+ dyna_linker_main
+
mov r4, r0
mov r5, r1
bl new_recompile_block
bl get_addr_ht
mov pc, r0
.size exec_pagefault, .-exec_pagefault
+
/* Special dynamic linker for the case where a page fault
may occur in a branch delay slot */
.global dyna_linker_ds
dyna_linker_ds:
/* r0 = virtual target address */
/* r1 = instruction to patch */
- ldr r4, .tlbptr
- lsr r5, r0, #12
- mov r12, r0
- cmp r0, #0xC0000000
- mov r6, #4096
- ldrge r12, [r4, r5, lsl #2]
- mov r2, #0x80000
- ldr r3, .jiptr
- tst r12, r12
- sub r6, r6, #1
- moveq r12, r0
- ldr r7, [r1]
- eor r2, r2, r12, lsr #12
- and r6, r6, r12, lsr #12
- cmp r2, #2048
- add r12, r7, #2
- orrcs r2, r6, #2048
- ldr r5, [r3, r2, lsl #2]
- lsl r12, r12, #8
- /* jump_in lookup */
-.B1:
- movs r4, r5
- beq .B3
- ldr r3, [r5]
- ldr r5, [r4, #12]
- teq r3, r0
- bne .B1
- ldr r3, [r4, #4]
- ldr r4, [r4, #8]
- tst r3, r3
- bne .B1
-.B2:
- mov r5, r1
- add r1, r1, r12, asr #6
- teq r1, r4
- moveq pc, r4 /* Stale i-cache */
- bl add_link
- sub r2, r4, r5
- and r1, r7, #0xff000000
- lsl r2, r2, #6
- sub r1, r1, #2
- add r1, r1, r2, lsr #8
- str r1, [r5]
- mov pc, r4
-.B3:
- /* hash_table lookup */
- cmp r2, #2048
- ldr r3, .jdptr
- eor r4, r0, r0, lsl #16
- lslcc r2, r0, #9
- ldr r6, .htptr
- lsr r4, r4, #12
- lsrcc r2, r2, #21
- bic r4, r4, #15
- ldr r5, [r3, r2, lsl #2]
- ldr r7, [r6, r4]!
- teq r7, r0
- ldreq pc, [r6, #4]
- ldr r7, [r6, #8]
- teq r7, r0
- ldreq pc, [r6, #12]
- /* jump_dirty lookup */
-.B6:
- movs r4, r5
- beq .B8
- ldr r3, [r5]
- ldr r5, [r4, #12]
- teq r3, r0
- bne .B6
-.B7:
- ldr r1, [r4, #8]
- /* hash_table insert */
- ldr r2, [r6]
- ldr r3, [r6, #4]
- str r0, [r6]
- str r1, [r6, #4]
- str r2, [r6, #8]
- str r3, [r6, #12]
- mov pc, r1
-.B8:
+ dyna_linker_main
+
mov r4, r0
bic r0, r0, #7
mov r5, r1
.word jump_in
.jdptr:
.word jump_dirty
-.tlbptr:
- .word tlb_LUT_r
.htptr:
.word hash_table
+
.align 2
.global jump_vaddr_r0
.type jump_vaddr_r0, %function
ldr r10, [fp, #cycle_count-dynarec_local]
mov pc, r0
.size jump_vaddr, .-jump_vaddr
+
.align 2
.global verify_code_ds
.type verify_code_ds, %function
.global verify_code_vm
.type verify_code_vm, %function
verify_code_vm:
- /* r0 = instruction pointer (virtual address) */
- /* r1 = source (virtual address) */
- /* r2 = target */
- /* r3 = length */
- cmp r1, #0xC0000000
- blt verify_code
- add r12, fp, #memory_map-dynarec_local
- lsr r4, r1, #12
- add r5, r1, r3
- sub r5, #1
- ldr r6, [r12, r4, lsl #2]
- lsr r5, r5, #12
- movs r7, r6
- bmi .D5
- add r1, r1, r6, lsl #2
- lsl r6, r6, #2
-.D1:
- add r4, r4, #1
- teq r6, r7, lsl #2
- bne .D5
- ldr r7, [r12, r4, lsl #2]
- cmp r4, r5
- bls .D1
- .size verify_code_vm, .-verify_code_vm
.global verify_code
.type verify_code, %function
verify_code:
bl get_addr
mov pc, r0
.size verify_code, .-verify_code
+ .size verify_code_vm, .-verify_code_vm
+
.align 2
.global cc_interrupt
.type cc_interrupt, %function
str r1, [fp, #pending_exception-dynarec_local]
and r2, r2, r10, lsr #17
add r3, fp, #restore_candidate-dynarec_local
- str r10, [fp, #reg_cop0+36-dynarec_local] /* Count */
+ str r10, [fp, #cycle-dynarec_local] /* PCSX cycles */
+@@ str r10, [fp, #reg_cop0+36-dynarec_local] /* Count */
ldr r4, [r2, r3]
mov r10, lr
tst r4, r4
.E1:
bl gen_interupt
mov lr, r10
- ldr r10, [fp, #reg_cop0+36-dynarec_local] /* Count */
+ ldr r10, [fp, #cycle-dynarec_local]
ldr r0, [fp, #next_interupt-dynarec_local]
ldr r1, [fp, #pending_exception-dynarec_local]
ldr r2, [fp, #stop-dynarec_local]
str r0, [fp, #last_count-dynarec_local]
sub r10, r10, r0
tst r2, r2
- bne .E3
+ ldmnefd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, ip, pc}
tst r1, r1
moveq pc, lr
.E2:
ldr r0, [fp, #pcaddr-dynarec_local]
bl get_addr_ht
mov pc, r0
-.E3:
- add r12, fp, #28
- ldmia r12, {r4, r5, r6, r7, r8, r9, sl, fp, pc}
.E4:
/* Move 'dirty' blocks to the 'clean' list */
lsl r5, r2, #3
tst r5, #31
bne .E5
b .E1
-
.size cc_interrupt, .-cc_interrupt
+
.align 2
.global do_interrupt
.type do_interrupt, %function
do_interrupt:
ldr r0, [fp, #pcaddr-dynarec_local]
bl get_addr_ht
- ldr r1, [fp, #next_interupt-dynarec_local]
- ldr r10, [fp, #reg_cop0+36-dynarec_local] /* Count */
- str r1, [fp, #last_count-dynarec_local]
- sub r10, r10, r1
add r10, r10, #2
mov pc, r0
.size do_interrupt, .-do_interrupt
+
.align 2
.global fp_exception
.type fp_exception, %function
add r2, r2, #0x2c
str r1, [fp, #reg_cop0+48-dynarec_local] /* Status */
str r2, [fp, #reg_cop0+52-dynarec_local] /* Cause */
- add r0, r3, #0x180
+ add r0, r3, #0x80
bl get_addr_ht
mov pc, r0
.size fp_exception, .-fp_exception
mov r2, #0x90000000 /* Set high bit if delay slot */
b .E7
.size fp_exception_ds, .-fp_exception_ds
+
.align 2
.global jump_syscall
.type jump_syscall, %function
mov r2, #0x20
str r1, [fp, #reg_cop0+48-dynarec_local] /* Status */
str r2, [fp, #reg_cop0+52-dynarec_local] /* Cause */
- add r0, r3, #0x180
+ add r0, r3, #0x80
bl get_addr_ht
mov pc, r0
.size jump_syscall, .-jump_syscall
.align 2
- .global indirect_jump_indexed
- .type indirect_jump_indexed, %function
-indirect_jump_indexed:
- ldr r0, [r0, r1, lsl #2]
- .size indirect_jump_indexed, .-indirect_jump_indexed
- .align 2
- .global indirect_jump
- .type indirect_jump, %function
-indirect_jump:
- ldr r12, [fp, #last_count-dynarec_local]
- add r2, r2, r12
- str r2, [fp, #reg_cop0+36-dynarec_local] /* Count */
- mov pc, r0
- .size indirect_jump, .-indirect_jump
+
.align 2
- .global jump_eret
- .type jump_eret, %function
-jump_eret:
- ldr r1, [fp, #reg_cop0+48-dynarec_local] /* Status */
- ldr r0, [fp, #last_count-dynarec_local]
- bic r1, r1, #2
- add r10, r0, r10
- str r1, [fp, #reg_cop0+48-dynarec_local] /* Status */
- str r10, [fp, #reg_cop0+36-dynarec_local] /* Count */
- bl check_interupt
+ .global jump_syscall_hle
+ .type jump_syscall_hle, %function
+jump_syscall_hle:
+ str r0, [fp, #pcaddr-dynarec_local] /* PC must be set to EPC for psxException */
+ ldr r2, [fp, #last_count-dynarec_local]
+ mov r1, #0 /* in delay slot */
+ add r2, r2, r10
+ mov r0, #0x20 /* cause */
+ str r2, [fp, #cycle-dynarec_local] /* PCSX cycle counter */
+ bl psxException
+
+ /* note: psxException might do recorsive recompiler call from it's HLE code,
+ * so be ready for this */
+pcsx_return:
ldr r1, [fp, #next_interupt-dynarec_local]
- ldr r0, [fp, #reg_cop0+56-dynarec_local] /* EPC */
+ ldr r10, [fp, #cycle-dynarec_local]
+ ldr r0, [fp, #pcaddr-dynarec_local]
+ sub r10, r10, r1
str r1, [fp, #last_count-dynarec_local]
- subs r10, r10, r1
- bpl .E11
-.E8:
- add r6, fp, #reg+256-dynarec_local
- mov r5, #248
- mov r1, #0
-.E9:
- ldr r2, [r6, #-8]!
- ldr r3, [r6, #4]
- eor r3, r3, r2, asr #31
- subs r3, r3, #1
- adc r1, r1, r1
- subs r5, r5, #8
- bne .E9
- ldr r2, [fp, #hi-dynarec_local]
- ldr r3, [fp, #hi+4-dynarec_local]
- eors r3, r3, r2, asr #31
- ldr r2, [fp, #lo-dynarec_local]
- ldreq r3, [fp, #lo+4-dynarec_local]
- eoreq r3, r3, r2, asr #31
- subs r3, r3, #1
- adc r1, r1, r1
- bl get_addr_32
+ bl get_addr_ht
mov pc, r0
-.E11:
+ .size jump_syscall_hle, .-jump_syscall_hle
+
+ .align 2
+ .global jump_hlecall
+ .type jump_hlecall, %function
+jump_hlecall:
+ ldr r2, [fp, #last_count-dynarec_local]
str r0, [fp, #pcaddr-dynarec_local]
- bl cc_interrupt
- ldr r0, [fp, #pcaddr-dynarec_local]
- b .E8
- .size jump_eret, .-jump_eret
+ add r2, r2, r10
+ adr lr, pcsx_return
+ str r2, [fp, #cycle-dynarec_local] /* PCSX cycle counter */
+ bx r1
+ .size jump_hlecall, .-jump_hlecall
+
.align 2
- .global new_dyna_start
- .type new_dyna_start, %function
-new_dyna_start:
- ldr r12, .dlptr
- mov r0, #0xa4000000
- stmia r12, {r4, r5, r6, r7, r8, r9, sl, fp, lr}
- sub fp, r12, #28
- add r0, r0, #0x40
- bl new_recompile_block
- ldr r0, [fp, #next_interupt-dynarec_local]
- ldr r10, [fp, #reg_cop0+36-dynarec_local] /* Count */
- str r0, [fp, #last_count-dynarec_local]
- sub r10, r10, r0
- mov pc, #0x7000000
-.dlptr:
- .word dynarec_local+28
- .size new_dyna_start, .-new_dyna_start
+ .global jump_intcall
+ .type jump_intcall, %function
+jump_intcall:
+ ldr r2, [fp, #last_count-dynarec_local]
+ str r0, [fp, #pcaddr-dynarec_local]
+ add r2, r2, r10
+ adr lr, pcsx_return
+ str r2, [fp, #cycle-dynarec_local] /* PCSX cycle counter */
+ b execI
+ .size jump_hlecall, .-jump_hlecall
+
+new_dyna_leave:
.align 2
- .global write_rdram_new
- .type write_rdram_new, %function
-write_rdram_new:
- ldr r2, [fp, #address-dynarec_local]
- ldr r0, [fp, #word-dynarec_local]
- str r0, [r2]
- b .E12
- .size write_rdram_new, .-write_rdram_new
+ .global new_dyna_leave
+ .type new_dyna_leave, %function
+ ldr r0, [fp, #last_count-dynarec_local]
+ add r12, fp, #28
+ add r10, r0, r10
+ str r10, [fp, #cycle-dynarec_local]
+ ldmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, ip, pc}
+ .size new_dyna_leave, .-new_dyna_leave
+
.align 2
- .global write_rdramb_new
- .type write_rdramb_new, %function
-write_rdramb_new:
- ldr r2, [fp, #address-dynarec_local]
- ldrb r0, [fp, #byte-dynarec_local]
- eor r2, r2, #3
- strb r0, [r2]
- b .E12
- .size write_rdramb_new, .-write_rdramb_new
+ .global invalidate_addr_r0
+ .type invalidate_addr_r0, %function
+invalidate_addr_r0:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ b invalidate_addr_call
+ .size invalidate_addr_r0, .-invalidate_addr_r0
.align 2
- .global write_rdramh_new
- .type write_rdramh_new, %function
-write_rdramh_new:
- ldr r2, [fp, #address-dynarec_local]
- ldrh r0, [fp, #hword-dynarec_local]
- eor r2, r2, #2
- strh r0, [r2]
- b .E12
- .size write_rdramh_new, .-write_rdramh_new
+ .global invalidate_addr_r1
+ .type invalidate_addr_r1, %function
+invalidate_addr_r1:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ mov r0, r1
+ b invalidate_addr_call
+ .size invalidate_addr_r1, .-invalidate_addr_r1
.align 2
- .global write_rdramd_new
- .type write_rdramd_new, %function
-write_rdramd_new:
- ldr r2, [fp, #address-dynarec_local]
-/* ldrd r0, [fp, #dword-dynarec_local]*/
- ldr r0, [fp, #dword-dynarec_local]
- ldr r1, [fp, #dword+4-dynarec_local]
- str r0, [r2, #4]
- str r1, [r2]
- b .E12
- .size write_rdramd_new, .-write_rdramd_new
+ .global invalidate_addr_r2
+ .type invalidate_addr_r2, %function
+invalidate_addr_r2:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ mov r0, r2
+ b invalidate_addr_call
+ .size invalidate_addr_r2, .-invalidate_addr_r2
.align 2
- .global do_invalidate
- .type do_invalidate, %function
-do_invalidate:
- ldr r2, [fp, #address-dynarec_local]
-.E12:
- ldr r1, [fp, #invc_ptr-dynarec_local]
- lsr r0, r2, #12
- ldrb r2, [r1, r0]
- tst r2, r2
- beq invalidate_block
- mov pc, lr
- .size do_invalidate, .-do_invalidate
+ .global invalidate_addr_r3
+ .type invalidate_addr_r3, %function
+invalidate_addr_r3:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ mov r0, r3
+ b invalidate_addr_call
+ .size invalidate_addr_r3, .-invalidate_addr_r3
+ .align 2
+ .global invalidate_addr_r4
+ .type invalidate_addr_r4, %function
+invalidate_addr_r4:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ mov r0, r4
+ b invalidate_addr_call
+ .size invalidate_addr_r4, .-invalidate_addr_r4
.align 2
- .global read_nomem_new
- .type read_nomem_new, %function
-read_nomem_new:
- ldr r2, [fp, #address-dynarec_local]
- add r12, fp, #memory_map-dynarec_local
- lsr r0, r2, #12
- ldr r12, [r12, r0, lsl #2]
- mov r1, #8
- tst r12, r12
- bmi tlb_exception
- ldr r0, [r2, r12, lsl #2]
- str r0, [fp, #readmem_dword-dynarec_local]
- mov pc, lr
- .size read_nomem_new, .-read_nomem_new
+ .global invalidate_addr_r5
+ .type invalidate_addr_r5, %function
+invalidate_addr_r5:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ mov r0, r5
+ b invalidate_addr_call
+ .size invalidate_addr_r5, .-invalidate_addr_r5
.align 2
- .global read_nomemb_new
- .type read_nomemb_new, %function
-read_nomemb_new:
- ldr r2, [fp, #address-dynarec_local]
- add r12, fp, #memory_map-dynarec_local
- lsr r0, r2, #12
- ldr r12, [r12, r0, lsl #2]
- mov r1, #8
- tst r12, r12
- bmi tlb_exception
- eor r2, r2, #3
- ldrb r0, [r2, r12, lsl #2]
- str r0, [fp, #readmem_dword-dynarec_local]
- mov pc, lr
- .size read_nomemb_new, .-read_nomemb_new
+ .global invalidate_addr_r6
+ .type invalidate_addr_r6, %function
+invalidate_addr_r6:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ mov r0, r6
+ b invalidate_addr_call
+ .size invalidate_addr_r6, .-invalidate_addr_r6
.align 2
- .global read_nomemh_new
- .type read_nomemh_new, %function
-read_nomemh_new:
- ldr r2, [fp, #address-dynarec_local]
- add r12, fp, #memory_map-dynarec_local
- lsr r0, r2, #12
- ldr r12, [r12, r0, lsl #2]
- mov r1, #8
- tst r12, r12
- bmi tlb_exception
- lsl r12, r12, #2
- eor r2, r2, #2
- ldrh r0, [r2, r12]
- str r0, [fp, #readmem_dword-dynarec_local]
- mov pc, lr
- .size read_nomemh_new, .-read_nomemh_new
+ .global invalidate_addr_r7
+ .type invalidate_addr_r7, %function
+invalidate_addr_r7:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ mov r0, r7
+ b invalidate_addr_call
+ .size invalidate_addr_r7, .-invalidate_addr_r7
.align 2
- .global read_nomemd_new
- .type read_nomemd_new, %function
-read_nomemd_new:
- ldr r2, [fp, #address-dynarec_local]
- add r12, fp, #memory_map-dynarec_local
- lsr r0, r2, #12
- ldr r12, [r12, r0, lsl #2]
- mov r1, #8
- tst r12, r12
- bmi tlb_exception
- lsl r12, r12, #2
-/* ldrd r0, [r2, r12]*/
- add r3, r2, #4
- ldr r0, [r2, r12]
- ldr r1, [r3, r12]
- str r0, [fp, #readmem_dword+4-dynarec_local]
- str r1, [fp, #readmem_dword-dynarec_local]
- mov pc, lr
- .size read_nomemd_new, .-read_nomemd_new
+ .global invalidate_addr_r8
+ .type invalidate_addr_r8, %function
+invalidate_addr_r8:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ mov r0, r8
+ b invalidate_addr_call
+ .size invalidate_addr_r8, .-invalidate_addr_r8
.align 2
- .global write_nomem_new
- .type write_nomem_new, %function
-write_nomem_new:
- str r3, [fp, #24]
- str lr, [fp, #28]
- bl do_invalidate
- ldr r2, [fp, #address-dynarec_local]
- add r12, fp, #memory_map-dynarec_local
- ldr lr, [fp, #28]
- lsr r0, r2, #12
- ldr r3, [fp, #24]
- ldr r12, [r12, r0, lsl #2]
- mov r1, #0xc
- tst r12, #0x40000000
- bne tlb_exception
- ldr r0, [fp, #word-dynarec_local]
- str r0, [r2, r12, lsl #2]
- mov pc, lr
- .size write_nomem_new, .-write_nomem_new
+ .global invalidate_addr_r9
+ .type invalidate_addr_r9, %function
+invalidate_addr_r9:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ mov r0, r9
+ b invalidate_addr_call
+ .size invalidate_addr_r9, .-invalidate_addr_r9
.align 2
- .global write_nomemb_new
- .type write_nomemb_new, %function
-write_nomemb_new:
- str r3, [fp, #24]
- str lr, [fp, #28]
- bl do_invalidate
- ldr r2, [fp, #address-dynarec_local]
- add r12, fp, #memory_map-dynarec_local
- ldr lr, [fp, #28]
- lsr r0, r2, #12
- ldr r3, [fp, #24]
- ldr r12, [r12, r0, lsl #2]
- mov r1, #0xc
- tst r12, #0x40000000
- bne tlb_exception
- eor r2, r2, #3
- ldrb r0, [fp, #byte-dynarec_local]
- strb r0, [r2, r12, lsl #2]
- mov pc, lr
- .size write_nomemb_new, .-write_nomemb_new
+ .global invalidate_addr_r10
+ .type invalidate_addr_r10, %function
+invalidate_addr_r10:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ mov r0, r10
+ b invalidate_addr_call
+ .size invalidate_addr_r10, .-invalidate_addr_r10
.align 2
- .global write_nomemh_new
- .type write_nomemh_new, %function
-write_nomemh_new:
- str r3, [fp, #24]
- str lr, [fp, #28]
- bl do_invalidate
- ldr r2, [fp, #address-dynarec_local]
- add r12, fp, #memory_map-dynarec_local
- ldr lr, [fp, #28]
- lsr r0, r2, #12
- ldr r3, [fp, #24]
- ldr r12, [r12, r0, lsl #2]
- mov r1, #0xc
- lsls r12, #2
- bcs tlb_exception
- eor r2, r2, #2
- ldrh r0, [fp, #hword-dynarec_local]
- strh r0, [r2, r12]
- mov pc, lr
- .size write_nomemh_new, .-write_nomemh_new
+ .global invalidate_addr_r12
+ .type invalidate_addr_r12, %function
+invalidate_addr_r12:
+ stmia fp, {r0, r1, r2, r3, r12, lr}
+ mov r0, r12
+ .size invalidate_addr_r12, .-invalidate_addr_r12
.align 2
- .global write_nomemd_new
- .type write_nomemd_new, %function
-write_nomemd_new:
- str r3, [fp, #24]
- str lr, [fp, #28]
- bl do_invalidate
- ldr r2, [fp, #address-dynarec_local]
- add r12, fp, #memory_map-dynarec_local
- ldr lr, [fp, #28]
- lsr r0, r2, #12
- ldr r3, [fp, #24]
- ldr r12, [r12, r0, lsl #2]
- mov r1, #0xc
- lsls r12, #2
- bcs tlb_exception
- add r3, r2, #4
- ldr r0, [fp, #dword+4-dynarec_local]
- ldr r1, [fp, #dword-dynarec_local]
-/* strd r0, [r2, r12]*/
- str r0, [r2, r12]
- str r1, [r3, r12]
- mov pc, lr
- .size write_nomemd_new, .-write_nomemd_new
+ .global invalidate_addr_call
+ .type invalidate_addr_call, %function
+invalidate_addr_call:
+ ldr r12, [fp, #inv_code_start-dynarec_local]
+ ldr lr, [fp, #inv_code_end-dynarec_local]
+ cmp r0, r12
+ cmpcs lr, r0
+ blcc invalidate_addr
+ ldmia fp, {r0, r1, r2, r3, r12, pc}
+ .size invalidate_addr_call, .-invalidate_addr_call
+
.align 2
- .global tlb_exception
- .type tlb_exception, %function
-tlb_exception:
- /* r1 = cause */
- /* r2 = address */
- /* r3 = instr addr/flags */
- ldr r4, [fp, #reg_cop0+48-dynarec_local] /* Status */
- add r5, fp, #memory_map-dynarec_local
- lsr r6, r3, #12
- orr r1, r1, r3, lsl #31
- orr r4, r4, #2
- ldr r7, [r5, r6, lsl #2]
- bic r8, r3, #3
- str r4, [fp, #reg_cop0+48-dynarec_local] /* Status */
- mov r6, #0x6000000
- str r1, [fp, #reg_cop0+52-dynarec_local] /* Cause */
- orr r6, r6, #0x22
- ldr r0, [r8, r7, lsl #2]
- add r4, r8, r1, asr #29
- add r5, fp, #reg-dynarec_local
- str r4, [fp, #reg_cop0+56-dynarec_local] /* EPC */
- mov r7, #0xf8
- ldr r8, [fp, #reg_cop0+16-dynarec_local] /* Context */
- lsl r1, r0, #16
- lsr r4, r0, #26
- and r7, r7, r0, lsr #18
- mvn r9, #0xF000000F
- sub r2, r2, r1, asr #16
- bic r9, r9, #0x0F800000
- rors r6, r6, r4
- mov r0, #0x80000000
- ldrcs r2, [r5, r7]
- bic r8, r8, r9
- tst r3, #2
- str r2, [r5, r7]
- add r4, r2, r1, asr #16
- add r6, fp, #reg+4-dynarec_local
- asr r3, r2, #31
- str r4, [fp, #reg_cop0+32-dynarec_local] /* BadVAddr */
- add r0, r0, #0x180
- and r4, r9, r4, lsr #9
- strne r3, [r6, r7]
- orr r8, r8, r4
- str r8, [fp, #reg_cop0+16-dynarec_local] /* Context */
+ .global new_dyna_start
+ .type new_dyna_start, %function
+new_dyna_start:
+ /* ip is stored to conform EABI alignment */
+ stmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, ip, lr}
+ load_var_adr fp, dynarec_local
+ ldr r0, [fp, #pcaddr-dynarec_local]
bl get_addr_ht
ldr r1, [fp, #next_interupt-dynarec_local]
- ldr r10, [fp, #reg_cop0+36-dynarec_local] /* Count */
+ ldr r10, [fp, #cycle-dynarec_local]
str r1, [fp, #last_count-dynarec_local]
sub r10, r10, r1
- mov pc, r0
- .size tlb_exception, .-tlb_exception
- .align 2
- .global breakpoint
- .type breakpoint, %function
-breakpoint:
- /* Set breakpoint here for debugging */
- mov pc, lr
- .size breakpoint, .-breakpoint
- .section .note.GNU-stack,"",%progbits
+ mov pc, r0
+ .size new_dyna_start, .-new_dyna_start
+
+/* --------------------------------------- */
+
+.align 2
+.global jump_handler_read8
+.global jump_handler_read16
+.global jump_handler_read32
+.global jump_handler_write8
+.global jump_handler_write16
+.global jump_handler_write32
+.global jump_handler_write_h
+.global jump_handle_swl
+.global jump_handle_swr
+.global rcnt0_read_count_m0
+.global rcnt0_read_count_m1
+.global rcnt1_read_count_m0
+.global rcnt1_read_count_m1
+.global rcnt2_read_count_m0
+.global rcnt2_read_count_m1
+
+
+.macro pcsx_read_mem readop tab_shift
+ /* r0 = address, r1 = handler_tab, r2 = cycles */
+ lsl r3, r0, #20
+ lsr r3, #(20+\tab_shift)
+ ldr r12, [fp, #last_count-dynarec_local]
+ ldr r1, [r1, r3, lsl #2]
+ add r2, r2, r12
+ lsls r1, #1
+.if \tab_shift == 1
+ lsl r3, #1
+ \readop r0, [r1, r3]
+.else
+ \readop r0, [r1, r3, lsl #\tab_shift]
+.endif
+ movcc pc, lr
+ str r2, [fp, #cycle-dynarec_local]
+ bx r1
+.endm
+
+jump_handler_read8:
+ add r1, #0x1000/4*4 + 0x1000/2*4 @ shift to r8 part
+ pcsx_read_mem ldrccb, 0
+
+jump_handler_read16:
+ add r1, #0x1000/4*4 @ shift to r16 part
+ pcsx_read_mem ldrcch, 1
+
+jump_handler_read32:
+ pcsx_read_mem ldrcc, 2
+
+
+.macro pcsx_write_mem wrtop tab_shift
+ /* r0 = address, r1 = data, r2 = cycles, r3 = handler_tab */
+ lsl r12,r0, #20
+ lsr r12, #(20+\tab_shift)
+ ldr r3, [r3, r12, lsl #2]
+ str r0, [fp, #address-dynarec_local] @ some handlers still need it..
+ lsls r3, #1
+ mov r0, r2 @ cycle return in case of direct store
+.if \tab_shift == 1
+ lsl r12, #1
+ \wrtop r1, [r3, r12]
+.else
+ \wrtop r1, [r3, r12, lsl #\tab_shift]
+.endif
+ movcc pc, lr
+ ldr r12, [fp, #last_count-dynarec_local]
+ mov r0, r1
+ add r2, r2, r12
+ push {r2, lr}
+ str r2, [fp, #cycle-dynarec_local]
+ blx r3
+
+ ldr r0, [fp, #next_interupt-dynarec_local]
+ pop {r2, r3}
+ str r0, [fp, #last_count-dynarec_local]
+ sub r0, r2, r0
+ bx r3
+.endm
+
+jump_handler_write8:
+ add r3, #0x1000/4*4 + 0x1000/2*4 @ shift to r8 part
+ pcsx_write_mem strccb, 0
+
+jump_handler_write16:
+ add r3, #0x1000/4*4 @ shift to r16 part
+ pcsx_write_mem strcch, 1
+
+jump_handler_write32:
+ pcsx_write_mem strcc, 2
+
+jump_handler_write_h:
+ /* r0 = address, r1 = data, r2 = cycles, r3 = handler */
+ ldr r12, [fp, #last_count-dynarec_local]
+ str r0, [fp, #address-dynarec_local] @ some handlers still need it..
+ add r2, r2, r12
+ mov r0, r1
+ push {r2, lr}
+ str r2, [fp, #cycle-dynarec_local]
+ blx r3
+
+ ldr r0, [fp, #next_interupt-dynarec_local]
+ pop {r2, r3}
+ str r0, [fp, #last_count-dynarec_local]
+ sub r0, r2, r0
+ bx r3
+
+jump_handle_swl:
+ /* r0 = address, r1 = data, r2 = cycles */
+ ldr r3, [fp, #mem_wtab-dynarec_local]
+ mov r12,r0,lsr #12
+ ldr r3, [r3, r12, lsl #2]
+ lsls r3, #1
+ bcs 4f
+ add r3, r0, r3
+ mov r0, r2
+ tst r3, #2
+ beq 101f
+ tst r3, #1
+ beq 2f
+3:
+ str r1, [r3, #-3]
+ bx lr
+2:
+ lsr r2, r1, #8
+ lsr r1, #24
+ strh r2, [r3, #-2]
+ strb r1, [r3]
+ bx lr
+101:
+ tst r3, #1
+ lsrne r1, #16 @ 1
+ lsreq r12, r1, #24 @ 0
+ strneh r1, [r3, #-1]
+ streqb r12, [r3]
+ bx lr
+4:
+ mov r0, r2
+@ b abort
+ bx lr @ TODO?
+
+
+jump_handle_swr:
+ /* r0 = address, r1 = data, r2 = cycles */
+ ldr r3, [fp, #mem_wtab-dynarec_local]
+ mov r12,r0,lsr #12
+ ldr r3, [r3, r12, lsl #2]
+ lsls r3, #1
+ bcs 4f
+ add r3, r0, r3
+ and r12,r3, #3
+ mov r0, r2
+ cmp r12,#2
+ strgtb r1, [r3] @ 3
+ streqh r1, [r3] @ 2
+ cmp r12,#1
+ strlt r1, [r3] @ 0
+ bxne lr
+ lsr r2, r1, #8 @ 1
+ strb r1, [r3]
+ strh r2, [r3, #1]
+ bx lr
+4:
+ mov r0, r2
+@ b abort
+ bx lr @ TODO?
+
+
+.macro rcntx_read_mode0 num
+ /* r0 = address, r2 = cycles */
+ ldr r3, [fp, #rcnts-dynarec_local+6*4+7*4*\num] @ cycleStart
+ mov r0, r2, lsl #16
+ sub r0, r3, lsl #16
+ lsr r0, #16
+ bx lr
+.endm
+
+rcnt0_read_count_m0:
+ rcntx_read_mode0 0
+
+rcnt1_read_count_m0:
+ rcntx_read_mode0 1
+
+rcnt2_read_count_m0:
+ rcntx_read_mode0 2
+
+rcnt0_read_count_m1:
+ /* r0 = address, r2 = cycles */
+ ldr r3, [fp, #rcnts-dynarec_local+6*4+7*4*0] @ cycleStart
+ mov_16 r1, 0x3334
+ sub r2, r2, r3
+ mul r0, r1, r2 @ /= 5
+ lsr r0, #16
+ bx lr
+
+rcnt1_read_count_m1:
+ /* r0 = address, r2 = cycles */
+ ldr r3, [fp, #rcnts-dynarec_local+6*4+7*4*1]
+ mov_24 r1, 0x1e6cde
+ sub r2, r2, r3
+ umull r3, r0, r1, r2 @ ~ /= hsync_cycles, max ~0x1e6cdd
+ bx lr
+
+rcnt2_read_count_m1:
+ /* r0 = address, r2 = cycles */
+ ldr r3, [fp, #rcnts-dynarec_local+6*4+7*4*2]
+ mov r0, r2, lsl #16-3
+ sub r0, r3, lsl #16-3
+ lsr r0, #16 @ /= 8
+ bx lr
+
+@ vim:filetype=armasm
notaz.gp2x.de / pcsx_rearmed.git / blobdiff