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=d32dc0bffcfff1efc7d789eefe16e99dd3811373;hp=b48ec30677a8f53def64e3e1b1a5c343c11ba722;hb=0e4ad31902f206e2c6945632bb1f558eae941ff1;hpb=c6d5790c709c4efcc01718fbe99572af550d3dff

diff --git a/libpcsxcore/new_dynarec/linkage_arm.S b/libpcsxcore/new_dynarec/linkage_arm.S
index b48ec306..d32dc0bf 100644
--- a/libpcsxcore/new_dynarec/linkage_arm.S
+++ b/libpcsxcore/new_dynarec/linkage_arm.S
@@ -20,6 +20,7 @@
  * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
 
 #include "arm_features.h"
+#include "new_dynarec_config.h"
 #include "linkage_offsets.h"
 
 
@@ -92,6 +93,22 @@ DRC_VAR(restore_candidate, 512)
 DRC_VAR(FCR0, 4)
 DRC_VAR(FCR31, 4)
 
+#ifdef TEXRELS_FORBIDDEN
+	.data
+	.align 2
+ptr_jump_in:
+	.word	ESYM(jump_in)
+ptr_jump_dirty:
+	.word	ESYM(jump_dirty)
+ptr_hash_table:
+	.word	ESYM(hash_table)
+#endif
+
+
+	.syntax unified
+	.text
+	.align	2
+
 #ifndef HAVE_ARMV5
 .macro blx rd
 	mov	lr, pc
@@ -100,7 +117,12 @@ DRC_VAR(FCR31, 4)
 #endif
 
 .macro load_varadr reg var
-#if defined(__ARM_ARCH_7A__) && !defined(__PIC__)
+#if defined(HAVE_ARMV7) && defined(TEXRELS_FORBIDDEN)
+	movw	\reg, #:lower16:(\var-(1678f+8))
+	movt	\reg, #:upper16:(\var-(1678f+8))
+1678:
+	add	\reg, pc
+#elif defined(HAVE_ARMV7) && !defined(__PIC__)
 	movw	\reg, #:lower16:\var
 	movt	\reg, #:upper16:\var
 #else
@@ -108,8 +130,19 @@ DRC_VAR(FCR31, 4)
 #endif
 .endm
 
+.macro load_varadr_ext reg var
+#if defined(HAVE_ARMV7) && defined(TEXRELS_FORBIDDEN)
+	movw	\reg, #:lower16:(ptr_\var-(1678f+8))
+	movt	\reg, #:upper16:(ptr_\var-(1678f+8))
+1678:
+	ldr	\reg, [pc, \reg]
+#else
+	load_varadr \reg \var
+#endif
+.endm
+
 .macro mov_16 reg imm
-#ifdef __ARM_ARCH_7A__
+#ifdef HAVE_ARMV7
 	movw	\reg, #\imm
 #else
 	mov	\reg, #(\imm & 0x00ff)
@@ -118,7 +151,7 @@ DRC_VAR(FCR31, 4)
 .endm
 
 .macro mov_24 reg imm
-#ifdef __ARM_ARCH_7A__
+#ifdef HAVE_ARMV7
 	movw	\reg, #(\imm & 0xffff)
 	movt	\reg, #(\imm >> 16)
 #else
@@ -128,10 +161,11 @@ DRC_VAR(FCR31, 4)
 #endif
 .endm
 
+/* r0 = virtual target address */
+/* r1 = instruction to patch */
 .macro dyna_linker_main
-	/* r0 = virtual target address */
-	/* r1 = instruction to patch */
-	ldr	r3, .jiptr
+#ifndef NO_WRITE_EXEC
+	load_varadr_ext r3, jump_in
 	/* get_page */
 	lsr     r2, r0, #12
 	mov	r6, #4096
@@ -152,14 +186,10 @@ DRC_VAR(FCR31, 4)
 1:
 	movs	r4, r5
 	beq	2f
-	ldr	r3, [r5]
-	ldr	r5, [r4, #12]
+	ldr	r3, [r5]	 /* ll_entry .vaddr */
+	ldrd	r4, r5, [r4, #8] /* ll_entry .next, .addr */
 	teq	r3, r0
 	bne	1b
-	ldr	r3, [r4, #4]
-	ldr	r4, [r4, #8]
-	tst	r3, r3
-	bne	1b
 	teq	r4, r6
 	moveq	pc, r4 /* Stale i-cache */
 	mov	r8, r4
@@ -181,10 +211,10 @@ DRC_VAR(FCR31, 4)
 3:
 	/* hash_table lookup */
 	cmp	r2, #2048
-	ldr	r3, .jdptr
+	load_varadr_ext r3, jump_dirty
 	eor	r4, r0, r0, lsl #16
 	lslcc	r2, r0, #9
-	ldr	r6, .htptr
+	load_varadr_ext r6, hash_table
 	lsr	r4, r4, #12
 	lsrcc	r2, r2, #21
 	bic	r4, r4, #15
@@ -214,10 +244,13 @@ DRC_VAR(FCR31, 4)
 	str	r3, [r6, #12]
 	mov	pc, r1
 8:
+#else
+	/* XXX: should be able to do better than this... */
+	bl	get_addr_ht
+	mov	pc, r0
+#endif
 .endm
 
-	.text
-	.align	2
 
 FUNCTION(dyna_linker):
 	/* r0 = virtual target address */
@@ -282,12 +315,6 @@ FUNCTION(dyna_linker_ds):
 	sub	r0, r1, #4
 	b	exec_pagefault
 	.size	dyna_linker_ds, .-dyna_linker_ds
-.jiptr:
-	.word	jump_in
-.jdptr:
-	.word	jump_dirty
-.htptr:
-	.word	hash_table
 
 	.align	2
 
@@ -350,7 +377,7 @@ FUNCTION(jump_vaddr_r7):
 	add	r0, r7, #0
 	.size	jump_vaddr_r7, .-jump_vaddr_r7
 FUNCTION(jump_vaddr):
-	ldr	r1, .htptr
+	load_varadr_ext r1, hash_table
 	mvn	r3, #15
 	and	r2, r3, r2, lsr #12
 	ldr	r2, [r1, r2]!
@@ -431,7 +458,7 @@ FUNCTION(cc_interrupt):
 	str	r0, [fp, #LO_last_count]
 	sub	r10, r10, r0
 	tst	r2, r2
-	ldmnefd	sp!, {r4, r5, r6, r7, r8, r9, sl, fp, ip, pc}
+	ldmfdne	sp!, {r4, r5, r6, r7, r8, r9, sl, fp, ip, pc}
 	tst	r1, r1
 	moveq	pc, lr
 .E2:
@@ -549,72 +576,72 @@ FUNCTION(new_dyna_leave):
 
 	.align	2
 FUNCTION(invalidate_addr_r0):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	b	invalidate_addr_call
 	.size	invalidate_addr_r0, .-invalidate_addr_r0
 	.align	2
 FUNCTION(invalidate_addr_r1):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	mov	r0, r1
 	b	invalidate_addr_call
 	.size	invalidate_addr_r1, .-invalidate_addr_r1
 	.align	2
 FUNCTION(invalidate_addr_r2):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	mov	r0, r2
 	b	invalidate_addr_call
 	.size	invalidate_addr_r2, .-invalidate_addr_r2
 	.align	2
 FUNCTION(invalidate_addr_r3):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	mov	r0, r3
 	b	invalidate_addr_call
 	.size	invalidate_addr_r3, .-invalidate_addr_r3
 	.align	2
 FUNCTION(invalidate_addr_r4):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	mov	r0, r4
 	b	invalidate_addr_call
 	.size	invalidate_addr_r4, .-invalidate_addr_r4
 	.align	2
 FUNCTION(invalidate_addr_r5):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	mov	r0, r5
 	b	invalidate_addr_call
 	.size	invalidate_addr_r5, .-invalidate_addr_r5
 	.align	2
 FUNCTION(invalidate_addr_r6):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	mov	r0, r6
 	b	invalidate_addr_call
 	.size	invalidate_addr_r6, .-invalidate_addr_r6
 	.align	2
 FUNCTION(invalidate_addr_r7):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	mov	r0, r7
 	b	invalidate_addr_call
 	.size	invalidate_addr_r7, .-invalidate_addr_r7
 	.align	2
 FUNCTION(invalidate_addr_r8):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	mov	r0, r8
 	b	invalidate_addr_call
 	.size	invalidate_addr_r8, .-invalidate_addr_r8
 	.align	2
 FUNCTION(invalidate_addr_r9):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	mov	r0, r9
 	b	invalidate_addr_call
 	.size	invalidate_addr_r9, .-invalidate_addr_r9
 	.align	2
 FUNCTION(invalidate_addr_r10):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	mov	r0, r10
 	b	invalidate_addr_call
 	.size	invalidate_addr_r10, .-invalidate_addr_r10
 	.align	2
 FUNCTION(invalidate_addr_r12):
-	stmia	fp, {r0, r1, r2, r3, r12, lr}
+	stmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, lr}
 	mov	r0, r12
 	.size	invalidate_addr_r12, .-invalidate_addr_r12
 	.align	2
@@ -624,7 +651,7 @@ invalidate_addr_call:
 	cmp	r0, r12
 	cmpcs   lr, r0
 	blcc	invalidate_addr
-	ldmia	fp, {r0, r1, r2, r3, r12, pc}
+	ldmia	fp, {r0, r1, r2, r3, EXTRA_UNSAVED_REGS r12, pc}
 	.size	invalidate_addr_call, .-invalidate_addr_call
 
 	.align	2
@@ -666,11 +693,11 @@ FUNCTION(new_dyna_start):
 
 FUNCTION(jump_handler_read8):
 	add     r1, #0x1000/4*4 + 0x1000/2*4 @ shift to r8 part
-	pcsx_read_mem ldrccb, 0
+	pcsx_read_mem ldrbcc, 0
 
 FUNCTION(jump_handler_read16):
 	add     r1, #0x1000/4*4              @ shift to r16 part
-	pcsx_read_mem ldrcch, 1
+	pcsx_read_mem ldrhcc, 1
 
 FUNCTION(jump_handler_read32):
 	pcsx_read_mem ldrcc, 2
@@ -707,11 +734,11 @@ FUNCTION(jump_handler_read32):
 
 FUNCTION(jump_handler_write8):
 	add     r3, #0x1000/4*4 + 0x1000/2*4 @ shift to r8 part
-	pcsx_write_mem strccb, 0
+	pcsx_write_mem strbcc, 0
 
 FUNCTION(jump_handler_write16):
 	add     r3, #0x1000/4*4              @ shift to r16 part
-	pcsx_write_mem strcch, 1
+	pcsx_write_mem strhcc, 1
 
 FUNCTION(jump_handler_write32):
 	pcsx_write_mem strcc, 2
@@ -758,8 +785,8 @@ FUNCTION(jump_handle_swl):
 	tst	r3, #1
 	lsrne	r1, #16		@ 1
 	lsreq	r12, r1, #24	@ 0
-	strneh	r1, [r3, #-1]
-	streqb	r12, [r3]
+	strhne	r1, [r3, #-1]
+	strbeq	r12, [r3]
 	bx	lr
 4:
 	mov	r0, r2
@@ -778,8 +805,8 @@ FUNCTION(jump_handle_swr):
 	and	r12,r3, #3
 	mov	r0, r2
 	cmp	r12,#2
-	strgtb	r1, [r3]	@ 3
-	streqh	r1, [r3]	@ 2
+	strbgt	r1, [r3]	@ 3
+	strheq	r1, [r3]	@ 2
 	cmp	r12,#1
 	strlt	r1, [r3]	@ 0
 	bxne	lr
@@ -797,7 +824,7 @@ FUNCTION(jump_handle_swr):
 	/* r0 = address, r2 = cycles */
 	ldr	r3, [fp, #LO_rcnts+6*4+7*4*\num] @ cycleStart
 	mov	r0, r2, lsl #16
-	sub	r0, r3, lsl #16
+	sub	r0, r0, r3, lsl #16
 	lsr	r0, #16
 	bx	lr
 .endm
@@ -832,7 +859,7 @@ FUNCTION(rcnt2_read_count_m1):
 	/* r0 = address, r2 = cycles */
 	ldr	r3, [fp, #LO_rcnts+6*4+7*4*2]
 	mov	r0, r2, lsl #16-3
-	sub	r0, r3, lsl #16-3
+	sub	r0, r0, r3, lsl #16-3
 	lsr	r0, #16			@ /= 8
 	bx	lr