diff options
-rw-r--r-- | libpcsxcore/gte_neon.S (renamed from libpcsxcore/gte_neon.s) | 15 | ||||
-rw-r--r-- | libpcsxcore/new_dynarec/assem_arm.c | 2 | ||||
-rw-r--r-- | libpcsxcore/new_dynarec/assem_arm.h | 2 | ||||
-rw-r--r-- | libpcsxcore/new_dynarec/linkage_arm.S (renamed from libpcsxcore/new_dynarec/linkage_arm.s) | 24 | ||||
-rw-r--r-- | plugins/dfsound/arm_utils.S (renamed from plugins/dfsound/arm_utils.s) | 41 |
5 files changed, 46 insertions, 38 deletions
diff --git a/libpcsxcore/gte_neon.s b/libpcsxcore/gte_neon.S index 470c3e3..9fafb27 100644 --- a/libpcsxcore/gte_neon.s +++ b/libpcsxcore/gte_neon.S @@ -17,6 +17,15 @@ scratch: .text .align 2 +.macro ldr_scratch rd +#ifndef __PIC__ + movw \rd, #:lower16:scratch + movt \rd, #:upper16:scratch +#else + ldr \rd, =scratch +#endif +.endm + @ XXX: gteMAC calc shouldn't be saturating, but it is here @ approximate gteMAC|123 flags @@ -139,8 +148,7 @@ gteRTPS_neon: push {r4-r6,lr} @ fmrx r4, fpscr @ vmrs? at least 40 cycle hit - movw r1, #:lower16:scratch - movt r1, #:upper16:scratch + ldr_scratch r1 mov r12, #0 vldmia r0, {d8} @ VXYZ(0) @@ -293,8 +301,7 @@ gteRTPS_neon: gteRTPT_neon: push {r4-r11,lr} - movw r1, #:lower16:scratch - movt r1, #:upper16:scratch + ldr_scratch r1 mov r12, #0 rtpx_preload diff --git a/libpcsxcore/new_dynarec/assem_arm.c b/libpcsxcore/new_dynarec/assem_arm.c index f2c2efa..77cfafa 100644 --- a/libpcsxcore/new_dynarec/assem_arm.c +++ b/libpcsxcore/new_dynarec/assem_arm.c @@ -2702,7 +2702,7 @@ void literal_pool_jumpover(int n) set_jump_target(jaddr,(int)out); } -emit_extjump2(int addr, int target, int linker) +emit_extjump2(u_int addr, int target, int linker) { u_char *ptr=(u_char *)addr; assert((ptr[3]&0x0e)==0xa); diff --git a/libpcsxcore/new_dynarec/assem_arm.h b/libpcsxcore/new_dynarec/assem_arm.h index 2d9efe1..f4e36a9 100644 --- a/libpcsxcore/new_dynarec/assem_arm.h +++ b/libpcsxcore/new_dynarec/assem_arm.h @@ -66,5 +66,5 @@ extern char *invc_ptr; #define BASE_ADDR 0x1000000 #else extern char translation_cache[1 << TARGET_SIZE_2]; -#define BASE_ADDR translation_cache +#define BASE_ADDR (u_int)translation_cache #endif diff --git a/libpcsxcore/new_dynarec/linkage_arm.s b/libpcsxcore/new_dynarec/linkage_arm.S index bd5a03d..4748078 100644 --- a/libpcsxcore/new_dynarec/linkage_arm.s +++ b/libpcsxcore/new_dynarec/linkage_arm.S @@ -19,8 +19,6 @@ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */ -/* .equiv HAVE_ARMV7, 1 */ - .global dynarec_local .global reg .global hi @@ -165,33 +163,33 @@ FCR31 = align0 .type FCR31, %object .size FCR31, 4 -.macro load_var_adr reg var -.if HAVE_ARMV7 +.macro load_varadr reg var +#if defined(__ARM_ARCH_7A__) && !defined(__PIC__) movw \reg, #:lower16:\var movt \reg, #:upper16:\var -.else +#else ldr \reg, =\var -.endif +#endif .endm .macro mov_16 reg imm -.if HAVE_ARMV7 +#ifdef __ARM_ARCH_7A__ movw \reg, #\imm -.else +#else mov \reg, #(\imm & 0x00ff) orr \reg, #(\imm & 0xff00) -.endif +#endif .endm .macro mov_24 reg imm -.if HAVE_ARMV7 +#ifdef __ARM_ARCH_7A__ movw \reg, #(\imm & 0xffff) movt \reg, #(\imm >> 16) -.else +#else mov \reg, #(\imm & 0x0000ff) orr \reg, #(\imm & 0x00ff00) orr \reg, #(\imm & 0xff0000) -.endif +#endif .endm .macro dyna_linker_main @@ -778,7 +776,7 @@ invalidate_addr_call: 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 + load_varadr fp, dynarec_local ldr r0, [fp, #pcaddr-dynarec_local] bl get_addr_ht ldr r1, [fp, #next_interupt-dynarec_local] diff --git a/plugins/dfsound/arm_utils.s b/plugins/dfsound/arm_utils.S index 70ff24d..22e5844 100644 --- a/plugins/dfsound/arm_utils.s +++ b/plugins/dfsound/arm_utils.S @@ -12,18 +12,24 @@ .text .align 2 -@ XXX: should be HAVE_NEON -.if HAVE_ARMV7 +.macro load_varadr reg var +#if defined(__ARM_ARCH_7A__) && !defined(__PIC__) + movw \reg, #:lower16:\var + movt \reg, #:upper16:\var +#else + ldr \reg, =\var +#endif +.endm + +#ifdef __ARM_NEON__ .global mix_chan @ (int start, int count, int lv, int rv) mix_chan: vmov.32 d14[0], r2 vmov.32 d14[1], r3 @ multipliers mov r12, r0 - movw r0, #:lower16:ChanBuf - movw r2, #:lower16:SSumLR - movt r0, #:upper16:ChanBuf - movt r2, #:upper16:SSumLR + load_varadr r0, ChanBuf + load_varadr r2, SSumLR add r0, r12, lsl #2 add r2, r12, lsl #3 0: @@ -56,12 +62,9 @@ mix_chan_rvb: vmov.32 d14[0], r2 vmov.32 d14[1], r3 @ multipliers mov r12, r0 - movw r0, #:lower16:ChanBuf - movw r3, #:lower16:sRVBStart - movw r2, #:lower16:SSumLR - movt r0, #:upper16:ChanBuf - movt r3, #:upper16:sRVBStart - movt r2, #:upper16:SSumLR + load_varadr r0, ChanBuf + load_varadr r3, sRVBStart + load_varadr r2, SSumLR ldr r3, [r3] add r0, r12, lsl #2 add r2, r12, lsl #3 @@ -97,7 +100,7 @@ mcr_finish: vstmiage r3!, {d8} bx lr -.else +#else .global mix_chan @ (int start, int count, int lv, int rv) mix_chan: @@ -105,8 +108,8 @@ mix_chan: orr r3, r2, r3, lsl #16 lsl r3, #1 @ packed multipliers << 1 mov r12, r0 - ldr r0, =ChanBuf - ldr r2, =SSumLR + load_varadr r0, ChanBuf + load_varadr r2, SSumLR add r0, r12, lsl #2 add r2, r12, lsl #3 0: @@ -134,9 +137,9 @@ mix_chan_rvb: stmfd sp!, {r4-r8,lr} orr lr, r2, r3, lsl #16 lsl lr, #1 - ldr r3, =sRVBStart - ldr r2, =SSumLR - ldr r4, =ChanBuf + load_varadr r3, sRVBStart + load_varadr r2, SSumLR + load_varadr r4, ChanBuf ldr r3, [r3] add r2, r2, r0, lsl #3 add r3, r3, r0, lsl #3 @@ -156,6 +159,6 @@ mix_chan_rvb: bgt 0b ldmfd sp!, {r4-r8,pc} -.endif +#endif @ vim:filetype=armasm |