* Free Software Foundation, Inc., *
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
+
+.equiv HAVE_ARMV7, 1
+
+.if HAVE_ARMV7
+ .cpu cortex-a8
+ .fpu vfp
+.else
.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"
+.endif
.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 invc_ptr
.global address
.global readmem_dword
+ .global readmem_word
.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
+
.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 /*0x400630*/
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
.type address, %object
.size address, 4
readmem_dword = address + 4
+readmem_word = readmem_dword
.type readmem_dword, %object
.size readmem_dword, 8
dword = readmem_dword + 8
.type FCR31, %object
.size FCR31, 4
reg = FCR31 + 4
+
+/* psxRegs */
+psxRegs = reg
.type reg, %object
- .size reg, 256
-hi = reg + 256
- .type hi, %object
- .size hi, 8
-lo = hi + 8
+ .size reg, 128
+ .size psxRegs, psxRegs_end-psxRegs
+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, 128
+psxRegs_end = intCycle + 128
+
+align0 = psxRegs_end /* just for alignment */
+ .type align0, %object
+ .size align0, 8
+branch_target = align0 + 8
+ .type branch_target, %object
+ .size branch_target, 4
+mini_ht = branch_target + 4
.type mini_ht, %object
.size mini_ht, 256
restore_candidate = mini_ht + 256
memory_map = restore_candidate + 512
.type memory_map, %object
.size memory_map, 4194304
+dynarec_local_end = memory_map + 4194304
.text
.align 2
dyna_linker:
/* 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
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
.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:
+ /* FIXME: cycles already calculated, not needed? */
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
add r10, r10, #2
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
+
+ .align 2
+ .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 r10, [fp, #cycle-dynarec_local]
+ ldr r0, [fp, #pcaddr-dynarec_local]
+ sub r10, r10, r1
+ str r1, [fp, #last_count-dynarec_local]
+ bl get_addr_ht
+ mov pc, r0
+ .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]
+ add r2, r2, r10
+ str r2, [fp, #cycle-dynarec_local] /* PCSX cycle counter */
+ adr lr, pcsx_return
+ bx r1
+ .size jump_hlecall, .-jump_hlecall
+
+new_dyna_leave:
+ .align 2
+ .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
+
+ /* these are used to call memhandlers */
+ .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 */
+ str r2, [fp, #cycle-dynarec_local]
mov pc, r0
.size indirect_jump, .-indirect_jump
+ .size indirect_jump_indexed, .-indirect_jump_indexed
+
.align 2
.global jump_eret
.type jump_eret, %function
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 */
+ str r10, [fp, #cycle-dynarec_local]
bl check_interupt
ldr r1, [fp, #next_interupt-dynarec_local]
ldr r0, [fp, #reg_cop0+56-dynarec_local] /* EPC */
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
mov pc, r0
.E11:
str r0, [fp, #pcaddr-dynarec_local]
ldr r0, [fp, #pcaddr-dynarec_local]
b .E8
.size jump_eret, .-jump_eret
+
.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
+ /* ip is stored to conform EABI alignment */
+ stmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, ip, lr}
+.if HAVE_ARMV7
+ movw fp, #:lower16:dynarec_local
+ movt fp, #:upper16:dynarec_local
+.else
+ ldr fp, .dlptr
+.endif
+ ldr r0, [fp, #pcaddr-dynarec_local]
+ bl get_addr_ht
+ ldr r1, [fp, #next_interupt-dynarec_local]
+ ldr r10, [fp, #cycle-dynarec_local]
+ str r1, [fp, #last_count-dynarec_local]
+ sub r10, r10, r1
+ mov pc, r0
.dlptr:
- .word dynarec_local+28
+ .word dynarec_local
.size new_dyna_start, .-new_dyna_start
+
.align 2
.global write_rdram_new
.type write_rdram_new, %function
strh r0, [r2]
b .E12
.size write_rdramh_new, .-write_rdramh_new
- .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
+
.align 2
.global do_invalidate
.type do_invalidate, %function
beq invalidate_block
mov pc, lr
.size do_invalidate, .-do_invalidate
+
.align 2
.global read_nomem_new
.type read_nomem_new, %function
-read_nomem_new:
+/*read_nomem_new:*/
+read_nomemb_new:
+read_nomemh_new:
+read_nomemd_new:
+ /* should never happen */
+ b read_nomem_new
+/*
ldr r2, [fp, #address-dynarec_local]
add r12, fp, #memory_map-dynarec_local
lsr r0, r2, #12
ldr r0, [r2, r12, lsl #2]
str r0, [fp, #readmem_dword-dynarec_local]
mov pc, lr
+*/
.size read_nomem_new, .-read_nomem_new
+/*
.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
- .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
- .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
- .align 2
- .global write_nomem_new
- .type write_nomem_new, %function
write_nomem_new:
str r3, [fp, #24]
str lr, [fp, #28]
str r0, [r2, r12, lsl #2]
mov pc, lr
.size write_nomem_new, .-write_nomem_new
+
.align 2
.global write_nomemb_new
.type write_nomemb_new, %function
strb r0, [r2, r12, lsl #2]
mov pc, lr
.size write_nomemb_new, .-write_nomemb_new
+
.align 2
.global write_nomemh_new
.type write_nomemh_new, %function
strh r0, [r2, r12]
mov pc, lr
.size write_nomemh_new, .-write_nomemh_new
- .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
- .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 */
- 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
- mov pc, r0
- .size tlb_exception, .-tlb_exception
+*/
.align 2
.global breakpoint
.type breakpoint, %function