X-Git-Url: https://notaz.gp2x.de/cgi-bin/gitweb.cgi?p=pcsx_rearmed.git;a=blobdiff_plain;f=libpcsxcore%2Fnew_dynarec%2Flinkage_arm.s;h=c1bc8c97f538315b12561cb75c881392d1ee56ed;hp=aeb8baa3b08da87de70f9cf865ee583af85ad8a7;hb=ffb0b9e0f051789f97f5efdcfab0b261e62688f9;hpb=b9b61529b6c9bf30a3146178e2dda31b15ff3614

diff --git a/libpcsxcore/new_dynarec/linkage_arm.s b/libpcsxcore/new_dynarec/linkage_arm.s
index aeb8baa3..c1bc8c97 100644
--- a/libpcsxcore/new_dynarec/linkage_arm.s
+++ b/libpcsxcore/new_dynarec/linkage_arm.s
@@ -1,6 +1,7 @@
 /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
- *   Mupen64plus - linkage_arm.s                                           *
- *   Copyright (C) 2009-2010 Ari64                                         *
+ *   linkage_arm.s for PCSX                                                *
+ *   Copyright (C) 2009-2011 Ari64                                         *
+ *   Copyright (C) 2010-2011 Gražvydas "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  *
@@ -17,17 +18,9 @@
  *   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
@@ -60,7 +53,12 @@ rdram = 0x80000000
 	.global	memory_map
 	/* psx */
 	.global psxRegs
-	.global psxHLEt_addr
+	.global mem_rtab
+	.global mem_wtab
+	.global nd_pcsx_io
+	.global psxH_ptr
+	.global inv_code_start
+	.global inv_code_end
 
 	.bss
 	.align	4
@@ -111,13 +109,14 @@ FCR0 = hword + 4
 FCR31 = FCR0 + 4
 	.type	FCR31, %object
 	.size	FCR31, 4
-reg = FCR31 + 4
+psxRegs = FCR31 + 4
 
 /* psxRegs */
-psxRegs = reg
+	.type	psxRegs, %object
+	.size	psxRegs, psxRegs_end-psxRegs
+reg = psxRegs
 	.type	reg, %object
 	.size	reg, 128
-	.size	psxRegs, psxRegs_end-psxRegs
 lo = reg + 128
 	.type	lo, %object
 	.size	lo, 4
@@ -148,13 +147,56 @@ interrupt = cycle + 4
 	.size	interrupt, 4
 intCycle = interrupt + 4
 	.type	intCycle, %object
-	.size	intCycle, 128
-psxRegs_end = intCycle + 128
+	.size	intCycle, 256
+psxRegs_end = intCycle + 256
 
-psxHLEt_addr = psxRegs_end
-	.type	psxHLEt_addr, %object
-	.size	psxHLEt_addr, 4
-align0 = psxHLEt_addr + 4 /* just for alignment */
+mem_rtab = psxRegs_end
+	.type	mem_rtab, %object
+	.size	mem_rtab, 4
+mem_wtab = mem_rtab + 4
+	.type	mem_wtab, %object
+	.size	mem_wtab, 4
+
+/* nd_pcsx_io */
+nd_pcsx_io = mem_wtab + 4
+	.type	nd_pcsx_io, %object
+	.size	nd_pcsx_io, nd_pcsx_io_end-nd_pcsx_io
+tab_read8 = nd_pcsx_io
+	.type	tab_read8, %object
+	.size	tab_read8, 4
+tab_read16 = tab_read8 + 4
+	.type	tab_read16, %object
+	.size	tab_read16, 4
+tab_read32 = tab_read16 + 4
+	.type	tab_read32, %object
+	.size	tab_read32, 4
+tab_write8 = tab_read32 + 4
+	.type	tab_write8, %object
+	.size	tab_write8, 4
+tab_write16 = tab_write8 + 4
+	.type	tab_write16, %object
+	.size	tab_write16, 4
+tab_write32 = tab_write16 + 4
+	.type	tab_write32, %object
+	.size	tab_write32, 4
+spu_readf = tab_write32 + 4
+	.type	spu_readf, %object
+	.size	spu_readf, 4
+spu_writef = spu_readf + 4
+	.type	spu_writef, %object
+	.size	spu_writef, 4
+nd_pcsx_io_end = spu_writef + 4
+
+psxH_ptr = nd_pcsx_io_end
+	.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
+align0 = inv_code_end + 4 /* just for alignment */
 	.type	align0, %object
 	.size	align0, 4
 branch_target = align0 + 4
@@ -171,52 +213,66 @@ memory_map = restore_candidate + 512
 	.size	memory_map, 4194304
 dynarec_local_end = memory_map + 4194304
 
-	.text
-	.align	2
-	.global	dyna_linker
-	.type	dyna_linker, %function
-dyna_linker:
+.macro load_var_adr reg var
+.if HAVE_ARMV7
+	movw	\reg, #:lower16:\var
+	movt	\reg, #:upper16:\var
+.else
+	ldr	\reg, =\var
+.endif
+.endm
+
+.macro dyna_linker_main
 	/* r0 = virtual target address */
 	/* r1 = instruction to patch */
-	mov	r12, r0
-	mov	r6, #4096
-	mov	r2, #0x80000
 	ldr	r3, .jiptr
+	/* get_page */
+	lsr     r2, r0, #12
+	mov	r6, #4096
+	bic	r2, r2, #0xe0000
 	sub	r6, r6, #1
+	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
@@ -234,14 +290,14 @@ dyna_linker:
 	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]
@@ -251,7 +307,18 @@ dyna_linker:
 	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
@@ -296,80 +363,8 @@ exec_pagefault:
 dyna_linker_ds:
 	/* r0 = virtual target address */
 	/* r1 = instruction to patch */
-	mov	r12, r0
-	mov	r6, #4096
-	mov	r2, #0x80000
-	ldr	r3, .jiptr
-	sub	r6, r6, #1
-	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
@@ -567,16 +562,13 @@ cc_interrupt:
 	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
@@ -597,13 +589,10 @@ cc_interrupt:
 do_interrupt:
 	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
 	add	r10, r10, #2
 	mov	pc, r0
 	.size	do_interrupt, .-do_interrupt
+
 	.align	2
 	.global	fp_exception
 	.type	fp_exception, %function
@@ -676,12 +665,22 @@ pcsx_return:
 jump_hlecall:
 	ldr	r2, [fp, #last_count-dynarec_local]
 	str	r0, [fp, #pcaddr-dynarec_local]
-	and	r1, r1, #7
 	add	r2, r2, r10
-	ldr	r3, [fp, #psxHLEt_addr-dynarec_local] /* psxHLEt */
+	adr	lr, pcsx_return
 	str	r2, [fp, #cycle-dynarec_local] /* PCSX cycle counter */
+	bx	r1
+	.size	jump_hlecall, .-jump_hlecall
+
+	.align	2
+	.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
-	ldr	pc, [r3, r1, lsl #2] @ FIXME
+	str	r2, [fp, #cycle-dynarec_local] /* PCSX cycle counter */
+	b	execI
 	.size	jump_hlecall, .-jump_hlecall
 
 new_dyna_leave:
@@ -692,7 +691,7 @@ new_dyna_leave:
 	add	r12, fp, #28
 	add	r10, r0, r10
 	str	r10, [fp, #cycle-dynarec_local]
-	ldmia	r12, {r4, r5, r6, r7, r8, r9, sl, fp, pc}
+	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 */
@@ -712,187 +711,573 @@ indirect_jump:
 	.size	indirect_jump_indexed, .-indirect_jump_indexed
 
 	.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, #cycle-dynarec_local]
-	bl	check_interupt
-	ldr	r1, [fp, #next_interupt-dynarec_local]
-	ldr	r0, [fp, #reg_cop0+56-dynarec_local] /* EPC */
-	str	r1, [fp, #last_count-dynarec_local]
-	subs	r10, r10, r1
-	bpl	.E11
-.E8:
-	bl	get_addr
-	mov	pc, r0
-.E11:
-	str	r0, [fp, #pcaddr-dynarec_local]
-	bl	cc_interrupt
-	ldr	r0, [fp, #pcaddr-dynarec_local]
-	b	.E8
-	.size	jump_eret, .-jump_eret
+	.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	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	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	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	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	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	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	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	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	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	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	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	new_dyna_start
 	.type	new_dyna_start, %function
 new_dyna_start:
-	ldr	r12, .dlptr
-	stmia	r12, {r4, r5, r6, r7, r8, r9, sl, fp, lr}
-	sub	fp, r12, #28
+	/* 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	new_recompile_block*/
 	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
 	.size	new_dyna_start, .-new_dyna_start
 
-	.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
-	.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
-	.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
+/* --------------------------------------- */
 
-	.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
+.align	2
+.global	ari_read_ram8
+.global	ari_read_ram16
+.global	ari_read_ram32
+.global	ari_read_ram_mirror8
+.global	ari_read_ram_mirror16
+.global	ari_read_ram_mirror32
+.global	ari_write_ram8
+.global	ari_write_ram16
+.global	ari_write_ram32
+.global	ari_write_ram_mirror8
+.global	ari_write_ram_mirror16
+.global	ari_write_ram_mirror32
+.global	ari_write_ram_mirror_ro32
+.global	ari_read_bios8
+.global	ari_read_bios16
+.global	ari_read_bios32
+.global	ari_read_io8
+.global	ari_read_io16
+.global	ari_read_io32
+.global	ari_write_io8
+.global	ari_write_io16
+.global	ari_write_io32
+.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
+
+.macro ari_read_ram bic_const op
+	ldr	r0, [fp, #address-dynarec_local]
+.if \bic_const
+	bic	r0, r0, #\bic_const
+.endif
+	\op	r0, [r0]
+	str	r0, [fp, #readmem_dword-dynarec_local]
 	mov	pc, lr
-	.size	do_invalidate, .-do_invalidate
+.endm
 
-	.align	2
-	.global	read_nomem_new
-	.type	read_nomem_new, %function
-/*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	r12, [r12, r0, lsl #2]
-	mov	r1, #8
-	tst	r12, r12
-	bmi	tlb_exception
-	ldr	r0, [r2, r12, lsl #2]
+ari_read_ram8:
+	ari_read_ram 0, ldrb
+
+ari_read_ram16:
+	ari_read_ram 1, ldrh
+
+ari_read_ram32:
+	ari_read_ram 3, ldr
+
+.macro ari_read_ram_mirror mvn_const, op
+	ldr	r0, [fp, #address-dynarec_local]
+	mvn	r1, #\mvn_const
+	and	r0, r1, lsr #11
+	orr	r0, r0, #1<<31
+	\op	r0, [r0]
 	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
-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]
+.endm
+
+ari_read_ram_mirror8:
+	ari_read_ram_mirror 0, ldrb
+
+ari_read_ram_mirror16:
+	ari_read_ram_mirror (1<<11), ldrh
+
+ari_read_ram_mirror32:
+	ari_read_ram_mirror (3<<11), ldr
+
+/* invalidation is already taken care of by the caller */
+.macro ari_write_ram bic_const var pf
+	ldr	r0, [fp, #address-dynarec_local]
+	ldr\pf	r1, [fp, #\var-dynarec_local]
+.if \bic_const
+	bic	r0, r0, #\bic_const
+.endif
+	str\pf	r1, [r0]
 	mov	pc, lr
-	.size	write_nomem_new, .-write_nomem_new
+.endm
 
-	.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]
+ari_write_ram8:
+	ari_write_ram 0, byte, b
+
+ari_write_ram16:
+	ari_write_ram 1, hword, h
+
+ari_write_ram32:
+	ari_write_ram 3, word,
+
+.macro ari_write_ram_mirror mvn_const var pf
+	ldr	r0, [fp, #address-dynarec_local]
+	mvn	r3, #\mvn_const
+	ldr\pf	r1, [fp, #\var-dynarec_local]
+	and	r0, r3, lsr #11
+	ldr	r2, [fp, #invc_ptr-dynarec_local]
+	orr	r0, r0, #1<<31
+	ldrb	r2, [r2, r0, lsr #12]
+	str\pf	r1, [r0]
+	tst	r2, r2
+	movne	pc, lr
+	ldr	r1, [fp, #inv_code_start-dynarec_local]
+	ldr	r2, [fp, #inv_code_end-dynarec_local]
+	cmp	r0, r1
+	cmpcs   r2, r0
+	movcs	pc, lr
+	nop
+	b	invalidate_addr
+.endm
+
+ari_write_ram_mirror8:
+	ari_write_ram_mirror 0, byte, b
+
+ari_write_ram_mirror16:
+	ari_write_ram_mirror (1<<11), hword, h
+
+ari_write_ram_mirror32:
+	ari_write_ram_mirror (3<<11), word,
+
+ari_write_ram_mirror_ro32:
+	load_var_adr r0, pcsx_ram_is_ro
+	ldr	r0, [r0]
+	tst	r0, r0
+	movne	pc, lr
+	nop
+	b	ari_write_ram_mirror32
+
+
+.macro ari_read_bios_mirror bic_const op
+	ldr	r0, [fp, #address-dynarec_local]
+	orr	r0, r0, #0x80000000
+	bic	r0, r0, #(0x20000000|\bic_const)	@ map to 0x9fc...
+	\op	r0, [r0]
+	str	r0, [fp, #readmem_dword-dynarec_local]
 	mov	pc, lr
-	.size	write_nomemb_new, .-write_nomemb_new
+.endm
 
-	.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]
+ari_read_bios8:
+	ari_read_bios_mirror 0, ldrb
+
+ari_read_bios16:
+	ari_read_bios_mirror 1, ldrh
+
+ari_read_bios32:
+	ari_read_bios_mirror 3, ldr
+
+
+@ for testing
+.macro ari_read_io_old tab_shift
+	str     lr, [sp, #-8]! @ EABI alignment..
+.if \tab_shift == 0
+	bl	psxHwRead32
+.endif
+.if \tab_shift == 1
+	bl	psxHwRead16
+.endif
+.if \tab_shift == 2
+	bl	psxHwRead8
+.endif
+	str	r0, [fp, #readmem_dword-dynarec_local]
+	ldr     pc, [sp], #8
+.endm
+
+.macro ari_read_io readop mem_tab tab_shift
+	ldr	r0, [fp, #address-dynarec_local]
+	ldr	r1, [fp, #psxH_ptr-dynarec_local]
+.if \tab_shift == 0
+	bic	r0, r0, #3
+.endif
+.if \tab_shift == 1
+	bic	r0, r0, #1
+.endif
+	bic	r2, r0, #0x1f800000
+	ldr	r12,[fp, #\mem_tab-dynarec_local]
+	subs	r3, r2, #0x1000
+	blo	2f
+@	ari_read_io_old \tab_shift
+	cmp	r3, #0x880
+	bhs	1f
+	ldr	r12,[r12, r3, lsl #\tab_shift]
+	tst	r12,r12
+	beq	2f
+0:
+	str     lr, [sp, #-8]! @ EABI alignment..
+	blx	r12
+	str	r0, [fp, #readmem_dword-dynarec_local]
+	ldr     pc, [sp], #8
+
+1:
+.if \tab_shift == 1 @ read16
+	cmp	r2, #0x1c00
+	blo	2f
+	cmp	r2, #0x1e00
+	bhs	2f
+	ldr	r12,[fp, #spu_readf-dynarec_local]
+	b	0b
+.endif
+2:
+	@ no handler, just read psxH
+	\readop	r0, [r1, r2]
+	str	r0, [fp, #readmem_dword-dynarec_local]
 	mov	pc, lr
-	.size	write_nomemh_new, .-write_nomemh_new
-*/
-	.align	2
-	.global	breakpoint
-	.type	breakpoint, %function
-breakpoint:
-	/* Set breakpoint here for debugging */
+.endm
+
+ari_read_io8:
+	ari_read_io ldrb, tab_read8, 2
+
+ari_read_io16:
+	ari_read_io ldrh, tab_read16, 1
+
+ari_read_io32:
+	ari_read_io ldr, tab_read32, 0
+
+.macro ari_write_io_old tab_shift
+.if \tab_shift == 0
+	b	psxHwWrite32
+.endif
+.if \tab_shift == 1
+	b	psxHwWrite16
+.endif
+.if \tab_shift == 2
+	b	psxHwWrite8
+.endif
+.endm
+
+.macro ari_write_io pf var mem_tab tab_shift
+	ldr	r0, [fp, #address-dynarec_local]
+	ldr\pf	r1, [fp, #\var-dynarec_local]
+.if \tab_shift == 0
+	bic	r0, r0, #3
+.endif
+.if \tab_shift == 1
+	bic	r0, r0, #1
+.endif
+	bic	r2, r0, #0x1f800000
+	ldr	r12,[fp, #\mem_tab-dynarec_local]
+	subs	r3, r2, #0x1000
+	blo	0f
+@	ari_write_io_old \tab_shift
+	cmp	r3, #0x880
+	bhs	1f
+	ldr	r12,[r12, r3, lsl #\tab_shift]
+	mov	r0, r1
+	tst	r12,r12
+	bxne	r12
+0:
+	ldr	r3, [fp, #psxH_ptr-dynarec_local]
+	str\pf	r1, [r2, r3]
 	mov	pc, lr
-	.size	breakpoint, .-breakpoint
-	.section	.note.GNU-stack,"",%progbits
+1:
+	cmp	r2, #0x1c00
+	blo	0b
+	cmp	r2, #0x1e00
+.if \tab_shift != 0
+	ldrlo	pc, [fp, #spu_writef-dynarec_local]
+.else
+	@ write32 to SPU - very rare case (is this correct?)
+	bhs	0b
+	add	r2, r0, #2
+	mov	r3, r1, lsr #16
+	push	{r2,r3,lr}
+	mov	lr, pc
+	ldr	pc, [fp, #spu_writef-dynarec_local]
+	pop	{r0,r1,lr}
+	ldr	pc, [fp, #spu_writef-dynarec_local]
+.endif
+	nop
+	b	0b
+.endm
+
+ari_write_io8:
+	@ PCSX always writes to psxH, so do we for consistency
+	ldr	r0, [fp, #address-dynarec_local]
+	ldr	r3, [fp, #psxH_ptr-dynarec_local]
+	ldrb	r1, [fp, #byte-dynarec_local]
+	bic	r2, r0, #0x1f800000
+	ldr	r12,[fp, #tab_write8-dynarec_local]
+	strb	r1, [r2, r3]
+	subs	r3, r2, #0x1000
+	movlo	pc, lr
+@	ari_write_io_old 2
+	cmp	r3, #0x880
+	movhs	pc, lr
+	ldr	r12,[r12, r3, lsl #2]
+	mov	r0, r1
+	tst	r12,r12
+	bxne	r12
+	mov	pc, lr
+
+ari_write_io16:
+	ari_write_io h, hword, tab_write16, 1
+
+ari_write_io32:
+	ari_write_io , word, tab_write32, 0
+
+/* */
+
+.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?
+
+
+@ vim:filetype=armasm