ref: 0afe5e405dfc09e46cff10895aabb18950b85708
parent: b2d690187e0386546ba482a877770e67761f2f80
parent: f51f67602e19cc086e29aa8e503d05b919f0a7b0
author: James Zern <[email protected]>
date: Tue Jun 28 15:22:10 EDT 2016
Merge "*.asm: normalize label format"
--- a/vp8/common/x86/mfqe_sse2.asm
+++ b/vp8/common/x86/mfqe_sse2.asm
@@ -45,7 +45,7 @@
mov rcx, 16 ; loop count
pxor xmm6, xmm6
-.combine
+.combine:
movdqa xmm2, [rax]
movdqa xmm4, [rdx]
add rax, rsi
@@ -122,7 +122,7 @@
mov rcx, 8 ; loop count
pxor xmm4, xmm4
-.combine
+.combine:
movq xmm2, [rax]
movq xmm3, [rdx]
add rax, rsi
@@ -189,7 +189,7 @@
; Because we're working with the actual output frames
; we can't depend on any kind of data alignment.
-.accumulate
+.accumulate:
movdqa xmm0, [rax] ; src1
movdqa xmm1, [rdx] ; src2
add rax, rcx ; src1 + stride1
--- a/vp8/common/x86/postproc_mmx.asm
+++ b/vp8/common/x86/postproc_mmx.asm
@@ -59,7 +59,7 @@
lea rdi, [rdi+rdx]
movq mm1, QWORD ptr[rdi] ; first row
mov rcx, 8
-.init_borderd ; initialize borders
+.init_borderd: ; initialize borders
lea rdi, [rdi + rax]
movq [rdi], mm1
@@ -72,7 +72,7 @@
mov rdi, rsi
movq mm1, QWORD ptr[rdi] ; first row
mov rcx, 8
-.init_border ; initialize borders
+.init_border: ; initialize borders
lea rdi, [rdi + rax]
movq [rdi], mm1
@@ -213,7 +213,7 @@
movd mm1, DWORD PTR [rsp+rcx*4] ;d[rcx*4]
movd [rsi], mm1
-.skip_assignment
+.skip_assignment:
lea rsi, [rsi+rax]
lea rdi, [rdi+rax]
--- a/vp8/common/x86/postproc_sse2.asm
+++ b/vp8/common/x86/postproc_sse2.asm
@@ -198,7 +198,7 @@
UPDATE_FLIMIT
jmp .acrossnextcol
-.acrossdone
+.acrossdone:
; last 16 pixels
movq QWORD PTR [rdi+rdx-16], mm0
@@ -278,7 +278,7 @@
lea rdi, [rdi+rdx]
movq xmm1, QWORD ptr[rdi] ; first row
mov rcx, 8
-.init_borderd ; initialize borders
+.init_borderd: ; initialize borders
lea rdi, [rdi + rax]
movq [rdi], xmm1
@@ -291,7 +291,7 @@
mov rdi, rsi
movq xmm1, QWORD ptr[rdi] ; first row
mov rcx, 8
-.init_border ; initialize borders
+.init_border: ; initialize borders
lea rdi, [rdi + rax]
movq [rdi], xmm1
@@ -434,7 +434,7 @@
movq mm0, [rsp + rcx*8] ;d[rcx*8]
movq [rsi], mm0
-.skip_assignment
+.skip_assignment:
lea rsi, [rsi+rax]
lea rdi, [rdi+rax]
--- a/vp9/common/x86/vp9_mfqe_sse2.asm
+++ b/vp9/common/x86/vp9_mfqe_sse2.asm
@@ -46,7 +46,7 @@
mov rcx, 16 ; loop count
pxor xmm6, xmm6
-.combine
+.combine:
movdqa xmm2, [rax]
movdqa xmm4, [rdx]
add rax, rsi
@@ -123,7 +123,7 @@
mov rcx, 8 ; loop count
pxor xmm4, xmm4
-.combine
+.combine:
movq xmm2, [rax]
movq xmm3, [rdx]
add rax, rsi
@@ -190,7 +190,7 @@
; Because we're working with the actual output frames
; we can't depend on any kind of data alignment.
-.accumulate
+.accumulate:
movdqa xmm0, [rax] ; src1
movdqa xmm1, [rdx] ; src2
add rax, rcx ; src1 + stride1
--- a/vpx_dsp/x86/intrapred_sse2.asm
+++ b/vpx_dsp/x86/intrapred_sse2.asm
@@ -756,7 +756,7 @@
psubw m0, m2 ; t1-tl t2-tl ... t8-tl [word]
movq m2, [leftq]
punpcklbw m2, m1 ; l1 l2 l3 l4 l5 l6 l7 l8 [word]
-.loop
+.loop:
pshuflw m4, m2, 0x0 ; [63:0] l1 l1 l1 l1 [word]
pshuflw m3, m2, 0x55 ; [63:0] l2 l2 l2 l2 [word]
punpcklqdq m4, m4 ; l1 l1 l1 l1 l1 l1 l1 l1 [word]
--- a/vpx_dsp/x86/vpx_convolve_copy_sse2.asm
+++ b/vpx_dsp/x86/vpx_convolve_copy_sse2.asm
@@ -87,7 +87,7 @@
RET
%endif
-.w64
+.w64:
mov r4d, dword hm
.loop64:
movu m0, [srcq]
--- a/vpx_dsp/x86/vpx_subpixel_8t_ssse3.asm
+++ b/vpx_dsp/x86/vpx_subpixel_8t_ssse3.asm
@@ -211,7 +211,7 @@
pavgb m1, m0
%endif
movd [dstq], m1
-.done
+.done:
RET
%endm