aboutsummaryrefslogtreecommitdiffhomepage
path: root/src
diff options
context:
space:
mode:
authorGravatar Mike Klein <mtklein@chromium.org>2017-06-05 13:14:02 -0400
committerGravatar Skia Commit-Bot <skia-commit-bot@chromium.org>2017-06-05 19:20:48 +0000
commitf36031b68aa5b92204187c154fc5bc717db20a3a (patch)
tree973fe8af3aef7ea5ad817bb9c259ea46c76bf2c8 /src
parent12a4dc985b894083a9130d8f5770e67656418b52 (diff)
lowp: add some big easy stages
srcover_rgba_8888, lerp_u8, lerp_1_float, scale_u8, scale_1_float... this is enough for _lots_ of drawing. Change-Id: Ibe42adb8b1da6c66db3085851561dc9070556ee3 Reviewed-on: https://skia-review.googlesource.com/18622 Reviewed-by: Herb Derby <herb@google.com> Commit-Queue: Mike Klein <mtklein@chromium.org>
Diffstat (limited to 'src')
-rw-r--r--src/jumper/SkJumper.cpp21
-rw-r--r--src/jumper/SkJumper_generated.S408
-rw-r--r--src/jumper/SkJumper_generated_win.S407
-rw-r--r--src/jumper/SkJumper_stages_lowp.cpp60
4 files changed, 880 insertions, 16 deletions
diff --git a/src/jumper/SkJumper.cpp b/src/jumper/SkJumper.cpp
index 4244bfda9f..8912499d4f 100644
--- a/src/jumper/SkJumper.cpp
+++ b/src/jumper/SkJumper.cpp
@@ -70,14 +70,19 @@ using StartPipelineFn = void(size_t,size_t,size_t,void**,K*);
#endif
// Some stages have low-precision (~15 bit) versions from SkJumper_stages_lowp.cpp.
-#define LOWP_STAGES(M) \
- M(constant_color) \
- M(load_8888) \
- M(store_8888) \
- M(swap_rb) \
- M(swap) \
- M(move_src_dst) \
- M(move_dst_src) \
+#define LOWP_STAGES(M) \
+ M(constant_color) \
+ M(load_8888) \
+ M(store_8888) \
+ M(srcover_rgba_8888) \
+ M(lerp_1_float) \
+ M(lerp_u8) \
+ M(scale_1_float) \
+ M(scale_u8) \
+ M(swap_rb) \
+ M(swap) \
+ M(move_src_dst) \
+ M(move_dst_src) \
M(srcover)
extern "C" {
diff --git a/src/jumper/SkJumper_generated.S b/src/jumper/SkJumper_generated.S
index eba1759a3f..c9f498b330 100644
--- a/src/jumper/SkJumper_generated.S
+++ b/src/jumper/SkJumper_generated.S
@@ -36948,7 +36948,7 @@ HIDDEN _sk_constant_color_ssse3_lowp
FUNCTION(_sk_constant_color_ssse3_lowp)
_sk_constant_color_ssse3_lowp:
.byte 72,173 // lods %ds:(%rsi),%rax
- .byte 243,15,16,29,114,3,0,0 // movss 0x372(%rip),%xmm3 # 428 <_sk_srcover_ssse3_lowp+0x65>
+ .byte 243,15,16,29,122,9,0,0 // movss 0x97a(%rip),%xmm3 # a30 <_sk_srcover_ssse3_lowp+0x65>
.byte 243,15,16,0 // movss (%rax),%xmm0
.byte 243,15,89,195 // mulss %xmm3,%xmm0
.byte 243,68,15,44,200 // cvttss2si %xmm0,%r9d
@@ -36985,7 +36985,7 @@ _sk_load_8888_ssse3_lowp:
.byte 117,113 // jne 1a2 <_sk_load_8888_ssse3_lowp+0x7b>
.byte 69,15,16,76,147,16 // movups 0x10(%r11,%rdx,4),%xmm9
.byte 69,15,16,4,147 // movups (%r11,%rdx,4),%xmm8
- .byte 102,15,111,5,236,2,0,0 // movdqa 0x2ec(%rip),%xmm0 # 430 <_sk_srcover_ssse3_lowp+0x6d>
+ .byte 102,15,111,5,252,8,0,0 // movdqa 0x8fc(%rip),%xmm0 # a40 <_sk_srcover_ssse3_lowp+0x75>
.byte 102,68,15,56,0,192 // pshufb %xmm0,%xmm8
.byte 102,68,15,56,0,200 // pshufb %xmm0,%xmm9
.byte 102,65,15,111,208 // movdqa %xmm8,%xmm2
@@ -36999,7 +36999,7 @@ _sk_load_8888_ssse3_lowp:
.byte 102,15,239,210 // pxor %xmm2,%xmm2
.byte 102,65,15,96,208 // punpcklbw %xmm8,%xmm2
.byte 102,65,15,104,216 // punpckhbw %xmm8,%xmm3
- .byte 102,68,15,111,5,182,2,0,0 // movdqa 0x2b6(%rip),%xmm8 # 440 <_sk_srcover_ssse3_lowp+0x7d>
+ .byte 102,68,15,111,5,198,8,0,0 // movdqa 0x8c6(%rip),%xmm8 # a50 <_sk_srcover_ssse3_lowp+0x85>
.byte 102,65,15,228,192 // pmulhuw %xmm8,%xmm0
.byte 102,65,15,228,200 // pmulhuw %xmm8,%xmm1
.byte 102,65,15,228,208 // pmulhuw %xmm8,%xmm2
@@ -37130,6 +37130,385 @@ _sk_store_8888_ssse3_lowp:
.byte 255 // (bad)
.byte 255 // .byte 0xff
+HIDDEN _sk_srcover_rgba_8888_ssse3_lowp
+.globl _sk_srcover_rgba_8888_ssse3_lowp
+FUNCTION(_sk_srcover_rgba_8888_ssse3_lowp)
+_sk_srcover_rgba_8888_ssse3_lowp:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 76,139,24 // mov (%rax),%r11
+ .byte 77,133,192 // test %r8,%r8
+ .byte 15,133,66,1,0,0 // jne 4b4 <_sk_srcover_rgba_8888_ssse3_lowp+0x150>
+ .byte 69,15,16,76,147,16 // movups 0x10(%r11,%rdx,4),%xmm9
+ .byte 69,15,16,4,147 // movups (%r11,%rdx,4),%xmm8
+ .byte 77,133,192 // test %r8,%r8
+ .byte 102,15,111,37,216,6,0,0 // movdqa 0x6d8(%rip),%xmm4 # a60 <_sk_srcover_ssse3_lowp+0x95>
+ .byte 102,68,15,56,0,196 // pshufb %xmm4,%xmm8
+ .byte 102,68,15,56,0,204 // pshufb %xmm4,%xmm9
+ .byte 102,65,15,111,240 // movdqa %xmm8,%xmm6
+ .byte 102,65,15,98,241 // punpckldq %xmm9,%xmm6
+ .byte 102,15,239,255 // pxor %xmm7,%xmm7
+ .byte 102,15,239,228 // pxor %xmm4,%xmm4
+ .byte 102,15,96,230 // punpcklbw %xmm6,%xmm4
+ .byte 102,15,239,237 // pxor %xmm5,%xmm5
+ .byte 102,15,104,238 // punpckhbw %xmm6,%xmm5
+ .byte 102,69,15,106,193 // punpckhdq %xmm9,%xmm8
+ .byte 102,15,239,246 // pxor %xmm6,%xmm6
+ .byte 102,65,15,96,240 // punpcklbw %xmm8,%xmm6
+ .byte 102,65,15,104,248 // punpckhbw %xmm8,%xmm7
+ .byte 102,68,15,111,5,162,6,0,0 // movdqa 0x6a2(%rip),%xmm8 # a70 <_sk_srcover_ssse3_lowp+0xa5>
+ .byte 102,65,15,228,224 // pmulhuw %xmm8,%xmm4
+ .byte 102,65,15,228,232 // pmulhuw %xmm8,%xmm5
+ .byte 102,65,15,228,240 // pmulhuw %xmm8,%xmm6
+ .byte 102,65,15,228,248 // pmulhuw %xmm8,%xmm7
+ .byte 102,68,15,111,29,149,6,0,0 // movdqa 0x695(%rip),%xmm11 # a80 <_sk_srcover_ssse3_lowp+0xb5>
+ .byte 102,68,15,249,219 // psubw %xmm3,%xmm11
+ .byte 102,68,15,111,196 // movdqa %xmm4,%xmm8
+ .byte 102,69,15,56,11,195 // pmulhrsw %xmm11,%xmm8
+ .byte 102,69,15,56,29,192 // pabsw %xmm8,%xmm8
+ .byte 102,68,15,253,192 // paddw %xmm0,%xmm8
+ .byte 102,15,111,197 // movdqa %xmm5,%xmm0
+ .byte 102,65,15,56,11,195 // pmulhrsw %xmm11,%xmm0
+ .byte 102,68,15,56,29,200 // pabsw %xmm0,%xmm9
+ .byte 102,68,15,253,201 // paddw %xmm1,%xmm9
+ .byte 102,15,111,198 // movdqa %xmm6,%xmm0
+ .byte 102,65,15,56,11,195 // pmulhrsw %xmm11,%xmm0
+ .byte 102,68,15,56,29,208 // pabsw %xmm0,%xmm10
+ .byte 102,68,15,253,210 // paddw %xmm2,%xmm10
+ .byte 102,68,15,56,11,223 // pmulhrsw %xmm7,%xmm11
+ .byte 102,69,15,56,29,219 // pabsw %xmm11,%xmm11
+ .byte 102,68,15,253,219 // paddw %xmm3,%xmm11
+ .byte 102,65,15,111,208 // movdqa %xmm8,%xmm2
+ .byte 102,15,113,210,7 // psrlw $0x7,%xmm2
+ .byte 102,65,15,111,194 // movdqa %xmm10,%xmm0
+ .byte 102,15,113,208,7 // psrlw $0x7,%xmm0
+ .byte 102,15,103,208 // packuswb %xmm0,%xmm2
+ .byte 102,65,15,111,193 // movdqa %xmm9,%xmm0
+ .byte 102,15,113,208,7 // psrlw $0x7,%xmm0
+ .byte 102,65,15,111,203 // movdqa %xmm11,%xmm1
+ .byte 102,15,113,209,7 // psrlw $0x7,%xmm1
+ .byte 102,15,103,193 // packuswb %xmm1,%xmm0
+ .byte 102,15,111,202 // movdqa %xmm2,%xmm1
+ .byte 102,15,96,200 // punpcklbw %xmm0,%xmm1
+ .byte 102,15,104,208 // punpckhbw %xmm0,%xmm2
+ .byte 102,15,111,193 // movdqa %xmm1,%xmm0
+ .byte 102,15,97,194 // punpcklwd %xmm2,%xmm0
+ .byte 102,15,105,202 // punpckhwd %xmm2,%xmm1
+ .byte 15,133,207,0,0,0 // jne 55e <_sk_srcover_rgba_8888_ssse3_lowp+0x1fa>
+ .byte 243,65,15,127,76,147,16 // movdqu %xmm1,0x10(%r11,%rdx,4)
+ .byte 243,65,15,127,4,147 // movdqu %xmm0,(%r11,%rdx,4)
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 102,65,15,111,192 // movdqa %xmm8,%xmm0
+ .byte 102,65,15,111,201 // movdqa %xmm9,%xmm1
+ .byte 102,65,15,111,210 // movdqa %xmm10,%xmm2
+ .byte 102,65,15,111,219 // movdqa %xmm11,%xmm3
+ .byte 255,224 // jmpq *%rax
+ .byte 69,137,193 // mov %r8d,%r9d
+ .byte 65,128,225,7 // and $0x7,%r9b
+ .byte 102,69,15,239,201 // pxor %xmm9,%xmm9
+ .byte 69,15,87,192 // xorps %xmm8,%xmm8
+ .byte 65,254,201 // dec %r9b
+ .byte 65,128,249,6 // cmp $0x6,%r9b
+ .byte 15,135,172,254,255,255 // ja 37d <_sk_srcover_rgba_8888_ssse3_lowp+0x19>
+ .byte 69,15,182,201 // movzbl %r9b,%r9d
+ .byte 76,141,21,248,0,0,0 // lea 0xf8(%rip),%r10 # 5d4 <_sk_srcover_rgba_8888_ssse3_lowp+0x270>
+ .byte 75,99,4,138 // movslq (%r10,%r9,4),%rax
+ .byte 76,1,208 // add %r10,%rax
+ .byte 255,224 // jmpq *%rax
+ .byte 102,65,15,110,100,147,24 // movd 0x18(%r11,%rdx,4),%xmm4
+ .byte 102,68,15,112,204,69 // pshufd $0x45,%xmm4,%xmm9
+ .byte 69,15,87,192 // xorps %xmm8,%xmm8
+ .byte 243,65,15,16,100,147,20 // movss 0x14(%r11,%rdx,4),%xmm4
+ .byte 65,15,198,225,0 // shufps $0x0,%xmm9,%xmm4
+ .byte 65,15,198,225,226 // shufps $0xe2,%xmm9,%xmm4
+ .byte 68,15,40,204 // movaps %xmm4,%xmm9
+ .byte 243,65,15,16,100,147,16 // movss 0x10(%r11,%rdx,4),%xmm4
+ .byte 243,68,15,16,204 // movss %xmm4,%xmm9
+ .byte 243,65,15,16,100,147,12 // movss 0xc(%r11,%rdx,4),%xmm4
+ .byte 65,15,198,224,32 // shufps $0x20,%xmm8,%xmm4
+ .byte 68,15,198,196,36 // shufps $0x24,%xmm4,%xmm8
+ .byte 243,65,15,16,100,147,8 // movss 0x8(%r11,%rdx,4),%xmm4
+ .byte 65,15,198,224,48 // shufps $0x30,%xmm8,%xmm4
+ .byte 68,15,198,196,132 // shufps $0x84,%xmm4,%xmm8
+ .byte 243,65,15,16,100,147,4 // movss 0x4(%r11,%rdx,4),%xmm4
+ .byte 65,15,198,224,0 // shufps $0x0,%xmm8,%xmm4
+ .byte 65,15,198,224,226 // shufps $0xe2,%xmm8,%xmm4
+ .byte 68,15,40,196 // movaps %xmm4,%xmm8
+ .byte 243,65,15,16,36,147 // movss (%r11,%rdx,4),%xmm4
+ .byte 243,68,15,16,196 // movss %xmm4,%xmm8
+ .byte 233,31,254,255,255 // jmpq 37d <_sk_srcover_rgba_8888_ssse3_lowp+0x19>
+ .byte 69,137,193 // mov %r8d,%r9d
+ .byte 65,128,225,7 // and $0x7,%r9b
+ .byte 65,254,201 // dec %r9b
+ .byte 65,128,249,6 // cmp $0x6,%r9b
+ .byte 15,135,42,255,255,255 // ja 49c <_sk_srcover_rgba_8888_ssse3_lowp+0x138>
+ .byte 65,15,182,193 // movzbl %r9b,%eax
+ .byte 76,141,13,115,0,0,0 // lea 0x73(%rip),%r9 # 5f0 <_sk_srcover_rgba_8888_ssse3_lowp+0x28c>
+ .byte 73,99,4,129 // movslq (%r9,%rax,4),%rax
+ .byte 76,1,200 // add %r9,%rax
+ .byte 255,224 // jmpq *%rax
+ .byte 102,15,112,209,78 // pshufd $0x4e,%xmm1,%xmm2
+ .byte 102,65,15,126,84,147,24 // movd %xmm2,0x18(%r11,%rdx,4)
+ .byte 102,15,112,209,229 // pshufd $0xe5,%xmm1,%xmm2
+ .byte 102,65,15,126,84,147,20 // movd %xmm2,0x14(%r11,%rdx,4)
+ .byte 102,65,15,126,76,147,16 // movd %xmm1,0x10(%r11,%rdx,4)
+ .byte 102,15,112,200,231 // pshufd $0xe7,%xmm0,%xmm1
+ .byte 102,65,15,126,76,147,12 // movd %xmm1,0xc(%r11,%rdx,4)
+ .byte 102,15,112,200,78 // pshufd $0x4e,%xmm0,%xmm1
+ .byte 102,65,15,126,76,147,8 // movd %xmm1,0x8(%r11,%rdx,4)
+ .byte 102,15,112,200,229 // pshufd $0xe5,%xmm0,%xmm1
+ .byte 102,65,15,126,76,147,4 // movd %xmm1,0x4(%r11,%rdx,4)
+ .byte 102,65,15,126,4,147 // movd %xmm0,(%r11,%rdx,4)
+ .byte 233,200,254,255,255 // jmpq 49c <_sk_srcover_rgba_8888_ssse3_lowp+0x138>
+ .byte 122,255 // jp 5d5 <_sk_srcover_rgba_8888_ssse3_lowp+0x271>
+ .byte 255 // (bad)
+ .byte 255,101,255 // jmpq *-0x1(%rbp)
+ .byte 255 // (bad)
+ .byte 255,84,255,255 // callq *-0x1(%rdi,%rdi,8)
+ .byte 255,67,255 // incl -0x1(%rbx)
+ .byte 255 // (bad)
+ .byte 255,55 // pushq (%rdi)
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255,34 // jmpq *(%rdx)
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255,17 // callq *(%rcx)
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 217,255 // fcos
+ .byte 255 // (bad)
+ .byte 255,205 // dec %ebp
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255,193 // inc %ecx
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255,181,255,255,255,174 // pushq -0x51000001(%rbp)
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255,162,255,255,255,150 // jmpq *-0x69000001(%rdx)
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255 // .byte 0xff
+
+HIDDEN _sk_scale_1_float_ssse3_lowp
+.globl _sk_scale_1_float_ssse3_lowp
+FUNCTION(_sk_scale_1_float_ssse3_lowp)
+_sk_scale_1_float_ssse3_lowp:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 243,68,15,16,0 // movss (%rax),%xmm8
+ .byte 243,68,15,89,5,24,4,0,0 // mulss 0x418(%rip),%xmm8 # a34 <_sk_srcover_ssse3_lowp+0x69>
+ .byte 243,65,15,44,192 // cvttss2si %xmm8,%eax
+ .byte 102,68,15,110,192 // movd %eax,%xmm8
+ .byte 242,69,15,112,192,0 // pshuflw $0x0,%xmm8,%xmm8
+ .byte 102,69,15,112,192,80 // pshufd $0x50,%xmm8,%xmm8
+ .byte 102,65,15,56,11,192 // pmulhrsw %xmm8,%xmm0
+ .byte 102,15,56,29,192 // pabsw %xmm0,%xmm0
+ .byte 102,65,15,56,11,200 // pmulhrsw %xmm8,%xmm1
+ .byte 102,15,56,29,201 // pabsw %xmm1,%xmm1
+ .byte 102,65,15,56,11,208 // pmulhrsw %xmm8,%xmm2
+ .byte 102,15,56,29,210 // pabsw %xmm2,%xmm2
+ .byte 102,65,15,56,11,216 // pmulhrsw %xmm8,%xmm3
+ .byte 102,15,56,29,219 // pabsw %xmm3,%xmm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
+HIDDEN _sk_scale_u8_ssse3_lowp
+.globl _sk_scale_u8_ssse3_lowp
+FUNCTION(_sk_scale_u8_ssse3_lowp)
+_sk_scale_u8_ssse3_lowp:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 76,139,24 // mov (%rax),%r11
+ .byte 77,133,192 // test %r8,%r8
+ .byte 117,74 // jne 6b6 <_sk_scale_u8_ssse3_lowp+0x54>
+ .byte 243,69,15,126,4,19 // movq (%r11,%rdx,1),%xmm8
+ .byte 102,68,15,96,192 // punpcklbw %xmm0,%xmm8
+ .byte 102,65,15,113,240,8 // psllw $0x8,%xmm8
+ .byte 102,68,15,228,5,10,4,0,0 // pmulhuw 0x40a(%rip),%xmm8 # a90 <_sk_srcover_ssse3_lowp+0xc5>
+ .byte 102,65,15,56,11,192 // pmulhrsw %xmm8,%xmm0
+ .byte 102,15,56,29,192 // pabsw %xmm0,%xmm0
+ .byte 102,65,15,56,11,200 // pmulhrsw %xmm8,%xmm1
+ .byte 102,15,56,29,201 // pabsw %xmm1,%xmm1
+ .byte 102,65,15,56,11,208 // pmulhrsw %xmm8,%xmm2
+ .byte 102,15,56,29,210 // pabsw %xmm2,%xmm2
+ .byte 102,65,15,56,11,216 // pmulhrsw %xmm8,%xmm3
+ .byte 102,15,56,29,219 // pabsw %xmm3,%xmm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+ .byte 69,137,193 // mov %r8d,%r9d
+ .byte 65,128,225,7 // and $0x7,%r9b
+ .byte 102,69,15,239,192 // pxor %xmm8,%xmm8
+ .byte 65,254,201 // dec %r9b
+ .byte 65,128,249,6 // cmp $0x6,%r9b
+ .byte 119,172 // ja 677 <_sk_scale_u8_ssse3_lowp+0x15>
+ .byte 69,15,182,201 // movzbl %r9b,%r9d
+ .byte 76,141,21,102,0,0,0 // lea 0x66(%rip),%r10 # 73c <_sk_scale_u8_ssse3_lowp+0xda>
+ .byte 75,99,4,138 // movslq (%r10,%r9,4),%rax
+ .byte 76,1,208 // add %r10,%rax
+ .byte 255,224 // jmpq *%rax
+ .byte 65,15,182,68,19,6 // movzbl 0x6(%r11,%rdx,1),%eax
+ .byte 102,69,15,239,192 // pxor %xmm8,%xmm8
+ .byte 102,68,15,196,192,6 // pinsrw $0x6,%eax,%xmm8
+ .byte 65,15,182,68,19,5 // movzbl 0x5(%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,5 // pinsrw $0x5,%eax,%xmm8
+ .byte 65,15,182,68,19,4 // movzbl 0x4(%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,4 // pinsrw $0x4,%eax,%xmm8
+ .byte 65,15,182,68,19,3 // movzbl 0x3(%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,3 // pinsrw $0x3,%eax,%xmm8
+ .byte 65,15,182,68,19,2 // movzbl 0x2(%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,2 // pinsrw $0x2,%eax,%xmm8
+ .byte 65,15,182,68,19,1 // movzbl 0x1(%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,1 // pinsrw $0x1,%eax,%xmm8
+ .byte 65,15,182,4,19 // movzbl (%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,0 // pinsrw $0x0,%eax,%xmm8
+ .byte 233,59,255,255,255 // jmpq 677 <_sk_scale_u8_ssse3_lowp+0x15>
+ .byte 240,255 // lock (bad)
+ .byte 255 // (bad)
+ .byte 255,228 // jmpq *%rsp
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 216,255 // fdivr %st(7),%st
+ .byte 255 // (bad)
+ .byte 255,204 // dec %esp
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255,192 // inc %eax
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255,180,255,255,255,163,255 // pushq -0x5c0001(%rdi,%rdi,8)
+ .byte 255 // (bad)
+ .byte 255 // .byte 0xff
+
+HIDDEN _sk_lerp_1_float_ssse3_lowp
+.globl _sk_lerp_1_float_ssse3_lowp
+FUNCTION(_sk_lerp_1_float_ssse3_lowp)
+_sk_lerp_1_float_ssse3_lowp:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 243,68,15,16,0 // movss (%rax),%xmm8
+ .byte 243,68,15,89,5,208,2,0,0 // mulss 0x2d0(%rip),%xmm8 # a38 <_sk_srcover_ssse3_lowp+0x6d>
+ .byte 243,65,15,44,192 // cvttss2si %xmm8,%eax
+ .byte 102,68,15,110,192 // movd %eax,%xmm8
+ .byte 242,69,15,112,192,0 // pshuflw $0x0,%xmm8,%xmm8
+ .byte 102,69,15,112,192,80 // pshufd $0x50,%xmm8,%xmm8
+ .byte 102,65,15,56,11,192 // pmulhrsw %xmm8,%xmm0
+ .byte 102,68,15,56,29,200 // pabsw %xmm0,%xmm9
+ .byte 102,68,15,111,21,13,3,0,0 // movdqa 0x30d(%rip),%xmm10 # aa0 <_sk_srcover_ssse3_lowp+0xd5>
+ .byte 102,69,15,249,208 // psubw %xmm8,%xmm10
+ .byte 102,15,111,196 // movdqa %xmm4,%xmm0
+ .byte 102,65,15,56,11,194 // pmulhrsw %xmm10,%xmm0
+ .byte 102,15,56,29,192 // pabsw %xmm0,%xmm0
+ .byte 102,65,15,253,193 // paddw %xmm9,%xmm0
+ .byte 102,65,15,56,11,200 // pmulhrsw %xmm8,%xmm1
+ .byte 102,68,15,56,29,201 // pabsw %xmm1,%xmm9
+ .byte 102,15,111,205 // movdqa %xmm5,%xmm1
+ .byte 102,65,15,56,11,202 // pmulhrsw %xmm10,%xmm1
+ .byte 102,15,56,29,201 // pabsw %xmm1,%xmm1
+ .byte 102,65,15,253,201 // paddw %xmm9,%xmm1
+ .byte 102,65,15,56,11,208 // pmulhrsw %xmm8,%xmm2
+ .byte 102,68,15,56,29,202 // pabsw %xmm2,%xmm9
+ .byte 102,15,111,214 // movdqa %xmm6,%xmm2
+ .byte 102,65,15,56,11,210 // pmulhrsw %xmm10,%xmm2
+ .byte 102,15,56,29,210 // pabsw %xmm2,%xmm2
+ .byte 102,65,15,253,209 // paddw %xmm9,%xmm2
+ .byte 102,65,15,56,11,216 // pmulhrsw %xmm8,%xmm3
+ .byte 102,68,15,56,29,195 // pabsw %xmm3,%xmm8
+ .byte 102,68,15,56,11,215 // pmulhrsw %xmm7,%xmm10
+ .byte 102,65,15,56,29,218 // pabsw %xmm10,%xmm3
+ .byte 102,65,15,253,216 // paddw %xmm8,%xmm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+
+HIDDEN _sk_lerp_u8_ssse3_lowp
+.globl _sk_lerp_u8_ssse3_lowp
+FUNCTION(_sk_lerp_u8_ssse3_lowp)
+_sk_lerp_u8_ssse3_lowp:
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 76,139,24 // mov (%rax),%r11
+ .byte 77,133,192 // test %r8,%r8
+ .byte 15,133,169,0,0,0 // jne 8c4 <_sk_lerp_u8_ssse3_lowp+0xb7>
+ .byte 243,69,15,126,4,19 // movq (%r11,%rdx,1),%xmm8
+ .byte 102,68,15,96,192 // punpcklbw %xmm0,%xmm8
+ .byte 102,65,15,113,240,8 // psllw $0x8,%xmm8
+ .byte 102,68,15,228,5,123,2,0,0 // pmulhuw 0x27b(%rip),%xmm8 # ab0 <_sk_srcover_ssse3_lowp+0xe5>
+ .byte 102,65,15,56,11,192 // pmulhrsw %xmm8,%xmm0
+ .byte 102,68,15,56,29,200 // pabsw %xmm0,%xmm9
+ .byte 102,68,15,111,21,118,2,0,0 // movdqa 0x276(%rip),%xmm10 # ac0 <_sk_srcover_ssse3_lowp+0xf5>
+ .byte 102,69,15,249,208 // psubw %xmm8,%xmm10
+ .byte 102,15,111,196 // movdqa %xmm4,%xmm0
+ .byte 102,65,15,56,11,194 // pmulhrsw %xmm10,%xmm0
+ .byte 102,15,56,29,192 // pabsw %xmm0,%xmm0
+ .byte 102,65,15,253,193 // paddw %xmm9,%xmm0
+ .byte 102,65,15,56,11,200 // pmulhrsw %xmm8,%xmm1
+ .byte 102,68,15,56,29,201 // pabsw %xmm1,%xmm9
+ .byte 102,15,111,205 // movdqa %xmm5,%xmm1
+ .byte 102,65,15,56,11,202 // pmulhrsw %xmm10,%xmm1
+ .byte 102,15,56,29,201 // pabsw %xmm1,%xmm1
+ .byte 102,65,15,253,201 // paddw %xmm9,%xmm1
+ .byte 102,65,15,56,11,208 // pmulhrsw %xmm8,%xmm2
+ .byte 102,68,15,56,29,202 // pabsw %xmm2,%xmm9
+ .byte 102,15,111,214 // movdqa %xmm6,%xmm2
+ .byte 102,65,15,56,11,210 // pmulhrsw %xmm10,%xmm2
+ .byte 102,15,56,29,210 // pabsw %xmm2,%xmm2
+ .byte 102,65,15,253,209 // paddw %xmm9,%xmm2
+ .byte 102,65,15,56,11,216 // pmulhrsw %xmm8,%xmm3
+ .byte 102,68,15,56,29,195 // pabsw %xmm3,%xmm8
+ .byte 102,68,15,56,11,215 // pmulhrsw %xmm7,%xmm10
+ .byte 102,65,15,56,29,218 // pabsw %xmm10,%xmm3
+ .byte 102,65,15,253,216 // paddw %xmm8,%xmm3
+ .byte 72,173 // lods %ds:(%rsi),%rax
+ .byte 255,224 // jmpq *%rax
+ .byte 69,137,193 // mov %r8d,%r9d
+ .byte 65,128,225,7 // and $0x7,%r9b
+ .byte 102,69,15,239,192 // pxor %xmm8,%xmm8
+ .byte 65,254,201 // dec %r9b
+ .byte 65,128,249,6 // cmp $0x6,%r9b
+ .byte 15,135,73,255,255,255 // ja 826 <_sk_lerp_u8_ssse3_lowp+0x19>
+ .byte 69,15,182,201 // movzbl %r9b,%r9d
+ .byte 76,141,21,104,0,0,0 // lea 0x68(%rip),%r10 # 950 <_sk_lerp_u8_ssse3_lowp+0x143>
+ .byte 75,99,4,138 // movslq (%r10,%r9,4),%rax
+ .byte 76,1,208 // add %r10,%rax
+ .byte 255,224 // jmpq *%rax
+ .byte 65,15,182,68,19,6 // movzbl 0x6(%r11,%rdx,1),%eax
+ .byte 102,69,15,239,192 // pxor %xmm8,%xmm8
+ .byte 102,68,15,196,192,6 // pinsrw $0x6,%eax,%xmm8
+ .byte 65,15,182,68,19,5 // movzbl 0x5(%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,5 // pinsrw $0x5,%eax,%xmm8
+ .byte 65,15,182,68,19,4 // movzbl 0x4(%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,4 // pinsrw $0x4,%eax,%xmm8
+ .byte 65,15,182,68,19,3 // movzbl 0x3(%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,3 // pinsrw $0x3,%eax,%xmm8
+ .byte 65,15,182,68,19,2 // movzbl 0x2(%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,2 // pinsrw $0x2,%eax,%xmm8
+ .byte 65,15,182,68,19,1 // movzbl 0x1(%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,1 // pinsrw $0x1,%eax,%xmm8
+ .byte 65,15,182,4,19 // movzbl (%r11,%rdx,1),%eax
+ .byte 102,68,15,196,192,0 // pinsrw $0x0,%eax,%xmm8
+ .byte 233,216,254,255,255 // jmpq 826 <_sk_lerp_u8_ssse3_lowp+0x19>
+ .byte 102,144 // xchg %ax,%ax
+ .byte 238 // out %al,(%dx)
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255,226 // jmpq *%rdx
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255,214 // callq *%rsi
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255,202 // dec %edx
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 190,255,255,255,178 // mov $0xb2ffffff,%esi
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255 // .byte 0xff
+ .byte 161 // .byte 0xa1
+ .byte 255 // (bad)
+ .byte 255 // (bad)
+ .byte 255 // .byte 0xff
+
HIDDEN _sk_swap_rb_ssse3_lowp
.globl _sk_swap_rb_ssse3_lowp
FUNCTION(_sk_swap_rb_ssse3_lowp)
@@ -37185,7 +37564,7 @@ HIDDEN _sk_srcover_ssse3_lowp
.globl _sk_srcover_ssse3_lowp
FUNCTION(_sk_srcover_ssse3_lowp)
_sk_srcover_ssse3_lowp:
- .byte 102,68,15,111,5,132,0,0,0 // movdqa 0x84(%rip),%xmm8 # 450 <_sk_srcover_ssse3_lowp+0x8d>
+ .byte 102,68,15,111,5,252,0,0,0 // movdqa 0xfc(%rip),%xmm8 # ad0 <_sk_srcover_ssse3_lowp+0x105>
.byte 102,68,15,249,195 // psubw %xmm3,%xmm8
.byte 102,68,15,111,204 // movdqa %xmm4,%xmm9
.byte 102,69,15,56,11,200 // pmulhrsw %xmm8,%xmm9
@@ -37207,6 +37586,9 @@ _sk_srcover_ssse3_lowp:
BALIGN4
.byte 0,0 // add %al,(%rax)
+ .byte 0,71,0 // add %al,0x0(%rdi)
+ .byte 0,0 // add %al,(%rax)
+ .byte 71,0,0 // rex.RXB add %r8b,(%r8)
.byte 0 // .byte 0x0
.byte 71 // rex.RXB
@@ -37218,7 +37600,25 @@ BALIGN16
.byte 3,7 // add (%rdi),%eax
.byte 11,15 // or (%rdi),%ecx
.byte 129,128,129,128,129,128,129,128,129,128// addl $0x80818081,-0x7f7e7f7f(%rax)
+ .byte 129,128,129,128,129,128,0,4,8,12 // addl $0xc080400,-0x7f7e7f7f(%rax)
+ .byte 1,5,9,13,2,6 // add %eax,0x6020d09(%rip) # 6021773 <_sk_srcover_ssse3_lowp+0x6020da8>
+ .byte 10,14 // or (%rsi),%cl
+ .byte 3,7 // add (%rdi),%eax
+ .byte 11,15 // or (%rdi),%ecx
+ .byte 129,128,129,128,129,128,129,128,129,128// addl $0x80818081,-0x7f7e7f7f(%rax)
.byte 129,128,129,128,129,128,0,128,0,128 // addl $0x80008000,-0x7f7e7f7f(%rax)
.byte 0,128,0,128,0,128 // add %al,-0x7fff8000(%rax)
.byte 0,128,0,128,0,128 // add %al,-0x7fff8000(%rax)
+ .byte 129,128,129,128,129,128,129,128,129,128// addl $0x80818081,-0x7f7e7f7f(%rax)
+ .byte 129,128,129,128,129,128,0,128,0,128 // addl $0x80008000,-0x7f7e7f7f(%rax)
+ .byte 0,128,0,128,0,128 // add %al,-0x7fff8000(%rax)
+ .byte 0,128,0,128,0,128 // add %al,-0x7fff8000(%rax)
+ .byte 129,128,129,128,129,128,129,128,129,128// addl $0x80818081,-0x7f7e7f7f(%rax)
+ .byte 129,128,129,128,129,128,0,128,0,128 // addl $0x80008000,-0x7f7e7f7f(%rax)
+ .byte 0,128,0,128,0,128 // add %al,-0x7fff8000(%rax)
+ .byte 0,128,0,128,0,128 // add %al,-0x7fff8000(%rax)
+ .byte 0,128,0,128,0,128 // add %al,-0x7fff8000(%rax)
+ .byte 0,128,0,128,0,128 // add %al,-0x7fff8000(%rax)
+ .byte 0 // .byte 0x0
+ .byte 128,0,128 // addb $0x80,(%rax)
#endif
diff --git a/src/jumper/SkJumper_generated_win.S b/src/jumper/SkJumper_generated_win.S
index a7848d3706..e75d982b53 100644
--- a/src/jumper/SkJumper_generated_win.S
+++ b/src/jumper/SkJumper_generated_win.S
@@ -26361,7 +26361,7 @@ _sk_just_return_ssse3_lowp LABEL PROC
PUBLIC _sk_constant_color_ssse3_lowp
_sk_constant_color_ssse3_lowp LABEL PROC
DB 72,173 ; lods %ds:(%rsi),%rax
- DB 243,15,16,29,114,3,0,0 ; movss 0x372(%rip),%xmm3 # 4b8 <_sk_srcover_ssse3_lowp+0x65>
+ DB 243,15,16,29,130,9,0,0 ; movss 0x982(%rip),%xmm3 # ac8 <_sk_srcover_ssse3_lowp+0x65>
DB 243,15,16,0 ; movss (%rax),%xmm0
DB 243,15,89,195 ; mulss %xmm3,%xmm0
DB 243,68,15,44,200 ; cvttss2si %xmm0,%r9d
@@ -26396,7 +26396,7 @@ _sk_load_8888_ssse3_lowp LABEL PROC
DB 117,113 ; jne 232 <_sk_load_8888_ssse3_lowp+0x7b>
DB 69,15,16,76,147,16 ; movups 0x10(%r11,%rdx,4),%xmm9
DB 69,15,16,4,147 ; movups (%r11,%rdx,4),%xmm8
- DB 102,15,111,5,236,2,0,0 ; movdqa 0x2ec(%rip),%xmm0 # 4c0 <_sk_srcover_ssse3_lowp+0x6d>
+ DB 102,15,111,5,12,9,0,0 ; movdqa 0x90c(%rip),%xmm0 # ae0 <_sk_srcover_ssse3_lowp+0x7d>
DB 102,68,15,56,0,192 ; pshufb %xmm0,%xmm8
DB 102,68,15,56,0,200 ; pshufb %xmm0,%xmm9
DB 102,65,15,111,208 ; movdqa %xmm8,%xmm2
@@ -26410,7 +26410,7 @@ _sk_load_8888_ssse3_lowp LABEL PROC
DB 102,15,239,210 ; pxor %xmm2,%xmm2
DB 102,65,15,96,208 ; punpcklbw %xmm8,%xmm2
DB 102,65,15,104,216 ; punpckhbw %xmm8,%xmm3
- DB 102,68,15,111,5,182,2,0,0 ; movdqa 0x2b6(%rip),%xmm8 # 4d0 <_sk_srcover_ssse3_lowp+0x7d>
+ DB 102,68,15,111,5,214,8,0,0 ; movdqa 0x8d6(%rip),%xmm8 # af0 <_sk_srcover_ssse3_lowp+0x8d>
DB 102,65,15,228,192 ; pmulhuw %xmm8,%xmm0
DB 102,65,15,228,200 ; pmulhuw %xmm8,%xmm1
DB 102,65,15,228,208 ; pmulhuw %xmm8,%xmm2
@@ -26539,6 +26539,384 @@ _sk_store_8888_ssse3_lowp LABEL PROC
DB 255 ; (bad)
DB 255 ; .byte 0xff
+PUBLIC _sk_srcover_rgba_8888_ssse3_lowp
+_sk_srcover_rgba_8888_ssse3_lowp LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 76,139,24 ; mov (%rax),%r11
+ DB 77,133,192 ; test %r8,%r8
+ DB 15,133,66,1,0,0 ; jne 544 <_sk_srcover_rgba_8888_ssse3_lowp+0x150>
+ DB 69,15,16,76,147,16 ; movups 0x10(%r11,%rdx,4),%xmm9
+ DB 69,15,16,4,147 ; movups (%r11,%rdx,4),%xmm8
+ DB 77,133,192 ; test %r8,%r8
+ DB 102,15,111,37,232,6,0,0 ; movdqa 0x6e8(%rip),%xmm4 # b00 <_sk_srcover_ssse3_lowp+0x9d>
+ DB 102,68,15,56,0,196 ; pshufb %xmm4,%xmm8
+ DB 102,68,15,56,0,204 ; pshufb %xmm4,%xmm9
+ DB 102,65,15,111,240 ; movdqa %xmm8,%xmm6
+ DB 102,65,15,98,241 ; punpckldq %xmm9,%xmm6
+ DB 102,15,239,255 ; pxor %xmm7,%xmm7
+ DB 102,15,239,228 ; pxor %xmm4,%xmm4
+ DB 102,15,96,230 ; punpcklbw %xmm6,%xmm4
+ DB 102,15,239,237 ; pxor %xmm5,%xmm5
+ DB 102,15,104,238 ; punpckhbw %xmm6,%xmm5
+ DB 102,69,15,106,193 ; punpckhdq %xmm9,%xmm8
+ DB 102,15,239,246 ; pxor %xmm6,%xmm6
+ DB 102,65,15,96,240 ; punpcklbw %xmm8,%xmm6
+ DB 102,65,15,104,248 ; punpckhbw %xmm8,%xmm7
+ DB 102,68,15,111,5,178,6,0,0 ; movdqa 0x6b2(%rip),%xmm8 # b10 <_sk_srcover_ssse3_lowp+0xad>
+ DB 102,65,15,228,224 ; pmulhuw %xmm8,%xmm4
+ DB 102,65,15,228,232 ; pmulhuw %xmm8,%xmm5
+ DB 102,65,15,228,240 ; pmulhuw %xmm8,%xmm6
+ DB 102,65,15,228,248 ; pmulhuw %xmm8,%xmm7
+ DB 102,68,15,111,29,165,6,0,0 ; movdqa 0x6a5(%rip),%xmm11 # b20 <_sk_srcover_ssse3_lowp+0xbd>
+ DB 102,68,15,249,219 ; psubw %xmm3,%xmm11
+ DB 102,68,15,111,196 ; movdqa %xmm4,%xmm8
+ DB 102,69,15,56,11,195 ; pmulhrsw %xmm11,%xmm8
+ DB 102,69,15,56,29,192 ; pabsw %xmm8,%xmm8
+ DB 102,68,15,253,192 ; paddw %xmm0,%xmm8
+ DB 102,15,111,197 ; movdqa %xmm5,%xmm0
+ DB 102,65,15,56,11,195 ; pmulhrsw %xmm11,%xmm0
+ DB 102,68,15,56,29,200 ; pabsw %xmm0,%xmm9
+ DB 102,68,15,253,201 ; paddw %xmm1,%xmm9
+ DB 102,15,111,198 ; movdqa %xmm6,%xmm0
+ DB 102,65,15,56,11,195 ; pmulhrsw %xmm11,%xmm0
+ DB 102,68,15,56,29,208 ; pabsw %xmm0,%xmm10
+ DB 102,68,15,253,210 ; paddw %xmm2,%xmm10
+ DB 102,68,15,56,11,223 ; pmulhrsw %xmm7,%xmm11
+ DB 102,69,15,56,29,219 ; pabsw %xmm11,%xmm11
+ DB 102,68,15,253,219 ; paddw %xmm3,%xmm11
+ DB 102,65,15,111,208 ; movdqa %xmm8,%xmm2
+ DB 102,15,113,210,7 ; psrlw $0x7,%xmm2
+ DB 102,65,15,111,194 ; movdqa %xmm10,%xmm0
+ DB 102,15,113,208,7 ; psrlw $0x7,%xmm0
+ DB 102,15,103,208 ; packuswb %xmm0,%xmm2
+ DB 102,65,15,111,193 ; movdqa %xmm9,%xmm0
+ DB 102,15,113,208,7 ; psrlw $0x7,%xmm0
+ DB 102,65,15,111,203 ; movdqa %xmm11,%xmm1
+ DB 102,15,113,209,7 ; psrlw $0x7,%xmm1
+ DB 102,15,103,193 ; packuswb %xmm1,%xmm0
+ DB 102,15,111,202 ; movdqa %xmm2,%xmm1
+ DB 102,15,96,200 ; punpcklbw %xmm0,%xmm1
+ DB 102,15,104,208 ; punpckhbw %xmm0,%xmm2
+ DB 102,15,111,193 ; movdqa %xmm1,%xmm0
+ DB 102,15,97,194 ; punpcklwd %xmm2,%xmm0
+ DB 102,15,105,202 ; punpckhwd %xmm2,%xmm1
+ DB 15,133,207,0,0,0 ; jne 5ee <_sk_srcover_rgba_8888_ssse3_lowp+0x1fa>
+ DB 243,65,15,127,76,147,16 ; movdqu %xmm1,0x10(%r11,%rdx,4)
+ DB 243,65,15,127,4,147 ; movdqu %xmm0,(%r11,%rdx,4)
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 102,65,15,111,192 ; movdqa %xmm8,%xmm0
+ DB 102,65,15,111,201 ; movdqa %xmm9,%xmm1
+ DB 102,65,15,111,210 ; movdqa %xmm10,%xmm2
+ DB 102,65,15,111,219 ; movdqa %xmm11,%xmm3
+ DB 255,224 ; jmpq *%rax
+ DB 69,137,193 ; mov %r8d,%r9d
+ DB 65,128,225,7 ; and $0x7,%r9b
+ DB 102,69,15,239,201 ; pxor %xmm9,%xmm9
+ DB 69,15,87,192 ; xorps %xmm8,%xmm8
+ DB 65,254,201 ; dec %r9b
+ DB 65,128,249,6 ; cmp $0x6,%r9b
+ DB 15,135,172,254,255,255 ; ja 40d <_sk_srcover_rgba_8888_ssse3_lowp+0x19>
+ DB 69,15,182,201 ; movzbl %r9b,%r9d
+ DB 76,141,21,248,0,0,0 ; lea 0xf8(%rip),%r10 # 664 <_sk_srcover_rgba_8888_ssse3_lowp+0x270>
+ DB 75,99,4,138 ; movslq (%r10,%r9,4),%rax
+ DB 76,1,208 ; add %r10,%rax
+ DB 255,224 ; jmpq *%rax
+ DB 102,65,15,110,100,147,24 ; movd 0x18(%r11,%rdx,4),%xmm4
+ DB 102,68,15,112,204,69 ; pshufd $0x45,%xmm4,%xmm9
+ DB 69,15,87,192 ; xorps %xmm8,%xmm8
+ DB 243,65,15,16,100,147,20 ; movss 0x14(%r11,%rdx,4),%xmm4
+ DB 65,15,198,225,0 ; shufps $0x0,%xmm9,%xmm4
+ DB 65,15,198,225,226 ; shufps $0xe2,%xmm9,%xmm4
+ DB 68,15,40,204 ; movaps %xmm4,%xmm9
+ DB 243,65,15,16,100,147,16 ; movss 0x10(%r11,%rdx,4),%xmm4
+ DB 243,68,15,16,204 ; movss %xmm4,%xmm9
+ DB 243,65,15,16,100,147,12 ; movss 0xc(%r11,%rdx,4),%xmm4
+ DB 65,15,198,224,32 ; shufps $0x20,%xmm8,%xmm4
+ DB 68,15,198,196,36 ; shufps $0x24,%xmm4,%xmm8
+ DB 243,65,15,16,100,147,8 ; movss 0x8(%r11,%rdx,4),%xmm4
+ DB 65,15,198,224,48 ; shufps $0x30,%xmm8,%xmm4
+ DB 68,15,198,196,132 ; shufps $0x84,%xmm4,%xmm8
+ DB 243,65,15,16,100,147,4 ; movss 0x4(%r11,%rdx,4),%xmm4
+ DB 65,15,198,224,0 ; shufps $0x0,%xmm8,%xmm4
+ DB 65,15,198,224,226 ; shufps $0xe2,%xmm8,%xmm4
+ DB 68,15,40,196 ; movaps %xmm4,%xmm8
+ DB 243,65,15,16,36,147 ; movss (%r11,%rdx,4),%xmm4
+ DB 243,68,15,16,196 ; movss %xmm4,%xmm8
+ DB 233,31,254,255,255 ; jmpq 40d <_sk_srcover_rgba_8888_ssse3_lowp+0x19>
+ DB 69,137,193 ; mov %r8d,%r9d
+ DB 65,128,225,7 ; and $0x7,%r9b
+ DB 65,254,201 ; dec %r9b
+ DB 65,128,249,6 ; cmp $0x6,%r9b
+ DB 15,135,42,255,255,255 ; ja 52c <_sk_srcover_rgba_8888_ssse3_lowp+0x138>
+ DB 65,15,182,193 ; movzbl %r9b,%eax
+ DB 76,141,13,115,0,0,0 ; lea 0x73(%rip),%r9 # 680 <_sk_srcover_rgba_8888_ssse3_lowp+0x28c>
+ DB 73,99,4,129 ; movslq (%r9,%rax,4),%rax
+ DB 76,1,200 ; add %r9,%rax
+ DB 255,224 ; jmpq *%rax
+ DB 102,15,112,209,78 ; pshufd $0x4e,%xmm1,%xmm2
+ DB 102,65,15,126,84,147,24 ; movd %xmm2,0x18(%r11,%rdx,4)
+ DB 102,15,112,209,229 ; pshufd $0xe5,%xmm1,%xmm2
+ DB 102,65,15,126,84,147,20 ; movd %xmm2,0x14(%r11,%rdx,4)
+ DB 102,65,15,126,76,147,16 ; movd %xmm1,0x10(%r11,%rdx,4)
+ DB 102,15,112,200,231 ; pshufd $0xe7,%xmm0,%xmm1
+ DB 102,65,15,126,76,147,12 ; movd %xmm1,0xc(%r11,%rdx,4)
+ DB 102,15,112,200,78 ; pshufd $0x4e,%xmm0,%xmm1
+ DB 102,65,15,126,76,147,8 ; movd %xmm1,0x8(%r11,%rdx,4)
+ DB 102,15,112,200,229 ; pshufd $0xe5,%xmm0,%xmm1
+ DB 102,65,15,126,76,147,4 ; movd %xmm1,0x4(%r11,%rdx,4)
+ DB 102,65,15,126,4,147 ; movd %xmm0,(%r11,%rdx,4)
+ DB 233,200,254,255,255 ; jmpq 52c <_sk_srcover_rgba_8888_ssse3_lowp+0x138>
+ DB 122,255 ; jp 665 <_sk_srcover_rgba_8888_ssse3_lowp+0x271>
+ DB 255 ; (bad)
+ DB 255,101,255 ; jmpq *-0x1(%rbp)
+ DB 255 ; (bad)
+ DB 255,84,255,255 ; callq *-0x1(%rdi,%rdi,8)
+ DB 255,67,255 ; incl -0x1(%rbx)
+ DB 255 ; (bad)
+ DB 255,55 ; pushq (%rdi)
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255,34 ; jmpq *(%rdx)
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255,17 ; callq *(%rcx)
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 217,255 ; fcos
+ DB 255 ; (bad)
+ DB 255,205 ; dec %ebp
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255,193 ; inc %ecx
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255,181,255,255,255,174 ; pushq -0x51000001(%rbp)
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255,162,255,255,255,150 ; jmpq *-0x69000001(%rdx)
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255 ; .byte 0xff
+
+PUBLIC _sk_scale_1_float_ssse3_lowp
+_sk_scale_1_float_ssse3_lowp LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 243,68,15,16,0 ; movss (%rax),%xmm8
+ DB 243,68,15,89,5,32,4,0,0 ; mulss 0x420(%rip),%xmm8 # acc <_sk_srcover_ssse3_lowp+0x69>
+ DB 243,65,15,44,192 ; cvttss2si %xmm8,%eax
+ DB 102,68,15,110,192 ; movd %eax,%xmm8
+ DB 242,69,15,112,192,0 ; pshuflw $0x0,%xmm8,%xmm8
+ DB 102,69,15,112,192,80 ; pshufd $0x50,%xmm8,%xmm8
+ DB 102,65,15,56,11,192 ; pmulhrsw %xmm8,%xmm0
+ DB 102,15,56,29,192 ; pabsw %xmm0,%xmm0
+ DB 102,65,15,56,11,200 ; pmulhrsw %xmm8,%xmm1
+ DB 102,15,56,29,201 ; pabsw %xmm1,%xmm1
+ DB 102,65,15,56,11,208 ; pmulhrsw %xmm8,%xmm2
+ DB 102,15,56,29,210 ; pabsw %xmm2,%xmm2
+ DB 102,65,15,56,11,216 ; pmulhrsw %xmm8,%xmm3
+ DB 102,15,56,29,219 ; pabsw %xmm3,%xmm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
+PUBLIC _sk_scale_u8_ssse3_lowp
+_sk_scale_u8_ssse3_lowp LABEL PROC
+ DB 80 ; push %rax
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 76,139,24 ; mov (%rax),%r11
+ DB 77,133,192 ; test %r8,%r8
+ DB 117,76 ; jne 749 <_sk_scale_u8_ssse3_lowp+0x57>
+ DB 243,69,15,126,4,19 ; movq (%r11,%rdx,1),%xmm8
+ DB 102,68,15,96,192 ; punpcklbw %xmm0,%xmm8
+ DB 102,65,15,113,240,8 ; psllw $0x8,%xmm8
+ DB 102,68,15,228,5,25,4,0,0 ; pmulhuw 0x419(%rip),%xmm8 # b30 <_sk_srcover_ssse3_lowp+0xcd>
+ DB 102,65,15,56,11,192 ; pmulhrsw %xmm8,%xmm0
+ DB 102,15,56,29,192 ; pabsw %xmm0,%xmm0
+ DB 102,65,15,56,11,200 ; pmulhrsw %xmm8,%xmm1
+ DB 102,15,56,29,201 ; pabsw %xmm1,%xmm1
+ DB 102,65,15,56,11,208 ; pmulhrsw %xmm8,%xmm2
+ DB 102,15,56,29,210 ; pabsw %xmm2,%xmm2
+ DB 102,65,15,56,11,216 ; pmulhrsw %xmm8,%xmm3
+ DB 102,15,56,29,219 ; pabsw %xmm3,%xmm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 65,89 ; pop %r9
+ DB 255,224 ; jmpq *%rax
+ DB 69,137,193 ; mov %r8d,%r9d
+ DB 65,128,225,7 ; and $0x7,%r9b
+ DB 102,69,15,239,192 ; pxor %xmm8,%xmm8
+ DB 65,254,201 ; dec %r9b
+ DB 65,128,249,6 ; cmp $0x6,%r9b
+ DB 119,170 ; ja 708 <_sk_scale_u8_ssse3_lowp+0x16>
+ DB 69,15,182,201 ; movzbl %r9b,%r9d
+ DB 76,141,21,103,0,0,0 ; lea 0x67(%rip),%r10 # 7d0 <_sk_scale_u8_ssse3_lowp+0xde>
+ DB 75,99,4,138 ; movslq (%r10,%r9,4),%rax
+ DB 76,1,208 ; add %r10,%rax
+ DB 255,224 ; jmpq *%rax
+ DB 65,15,182,68,19,6 ; movzbl 0x6(%r11,%rdx,1),%eax
+ DB 102,69,15,239,192 ; pxor %xmm8,%xmm8
+ DB 102,68,15,196,192,6 ; pinsrw $0x6,%eax,%xmm8
+ DB 65,15,182,68,19,5 ; movzbl 0x5(%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,5 ; pinsrw $0x5,%eax,%xmm8
+ DB 65,15,182,68,19,4 ; movzbl 0x4(%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,4 ; pinsrw $0x4,%eax,%xmm8
+ DB 65,15,182,68,19,3 ; movzbl 0x3(%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,3 ; pinsrw $0x3,%eax,%xmm8
+ DB 65,15,182,68,19,2 ; movzbl 0x2(%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,2 ; pinsrw $0x2,%eax,%xmm8
+ DB 65,15,182,68,19,1 ; movzbl 0x1(%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,1 ; pinsrw $0x1,%eax,%xmm8
+ DB 65,15,182,4,19 ; movzbl (%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,0 ; pinsrw $0x0,%eax,%xmm8
+ DB 233,57,255,255,255 ; jmpq 708 <_sk_scale_u8_ssse3_lowp+0x16>
+ DB 144 ; nop
+ DB 239 ; out %eax,(%dx)
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255,227 ; jmpq *%rbx
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255,215 ; callq *%rdi
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255,203 ; dec %ebx
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 191,255,255,255,179 ; mov $0xb3ffffff,%edi
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255 ; .byte 0xff
+ DB 162 ; .byte 0xa2
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255 ; .byte 0xff
+
+PUBLIC _sk_lerp_1_float_ssse3_lowp
+_sk_lerp_1_float_ssse3_lowp LABEL PROC
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 243,68,15,16,0 ; movss (%rax),%xmm8
+ DB 243,68,15,89,5,212,2,0,0 ; mulss 0x2d4(%rip),%xmm8 # ad0 <_sk_srcover_ssse3_lowp+0x6d>
+ DB 243,65,15,44,192 ; cvttss2si %xmm8,%eax
+ DB 102,68,15,110,192 ; movd %eax,%xmm8
+ DB 242,69,15,112,192,0 ; pshuflw $0x0,%xmm8,%xmm8
+ DB 102,69,15,112,192,80 ; pshufd $0x50,%xmm8,%xmm8
+ DB 102,65,15,56,11,192 ; pmulhrsw %xmm8,%xmm0
+ DB 102,68,15,56,29,200 ; pabsw %xmm0,%xmm9
+ DB 102,68,15,111,21,25,3,0,0 ; movdqa 0x319(%rip),%xmm10 # b40 <_sk_srcover_ssse3_lowp+0xdd>
+ DB 102,69,15,249,208 ; psubw %xmm8,%xmm10
+ DB 102,15,111,196 ; movdqa %xmm4,%xmm0
+ DB 102,65,15,56,11,194 ; pmulhrsw %xmm10,%xmm0
+ DB 102,15,56,29,192 ; pabsw %xmm0,%xmm0
+ DB 102,65,15,253,193 ; paddw %xmm9,%xmm0
+ DB 102,65,15,56,11,200 ; pmulhrsw %xmm8,%xmm1
+ DB 102,68,15,56,29,201 ; pabsw %xmm1,%xmm9
+ DB 102,15,111,205 ; movdqa %xmm5,%xmm1
+ DB 102,65,15,56,11,202 ; pmulhrsw %xmm10,%xmm1
+ DB 102,15,56,29,201 ; pabsw %xmm1,%xmm1
+ DB 102,65,15,253,201 ; paddw %xmm9,%xmm1
+ DB 102,65,15,56,11,208 ; pmulhrsw %xmm8,%xmm2
+ DB 102,68,15,56,29,202 ; pabsw %xmm2,%xmm9
+ DB 102,15,111,214 ; movdqa %xmm6,%xmm2
+ DB 102,65,15,56,11,210 ; pmulhrsw %xmm10,%xmm2
+ DB 102,15,56,29,210 ; pabsw %xmm2,%xmm2
+ DB 102,65,15,253,209 ; paddw %xmm9,%xmm2
+ DB 102,65,15,56,11,216 ; pmulhrsw %xmm8,%xmm3
+ DB 102,68,15,56,29,195 ; pabsw %xmm3,%xmm8
+ DB 102,68,15,56,11,215 ; pmulhrsw %xmm7,%xmm10
+ DB 102,65,15,56,29,218 ; pabsw %xmm10,%xmm3
+ DB 102,65,15,253,216 ; paddw %xmm8,%xmm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 255,224 ; jmpq *%rax
+
+PUBLIC _sk_lerp_u8_ssse3_lowp
+_sk_lerp_u8_ssse3_lowp LABEL PROC
+ DB 80 ; push %rax
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 76,139,24 ; mov (%rax),%r11
+ DB 77,133,192 ; test %r8,%r8
+ DB 15,133,171,0,0,0 ; jne 95b <_sk_lerp_u8_ssse3_lowp+0xba>
+ DB 243,69,15,126,4,19 ; movq (%r11,%rdx,1),%xmm8
+ DB 102,68,15,96,192 ; punpcklbw %xmm0,%xmm8
+ DB 102,65,15,113,240,8 ; psllw $0x8,%xmm8
+ DB 102,68,15,228,5,134,2,0,0 ; pmulhuw 0x286(%rip),%xmm8 # b50 <_sk_srcover_ssse3_lowp+0xed>
+ DB 102,65,15,56,11,192 ; pmulhrsw %xmm8,%xmm0
+ DB 102,68,15,56,29,200 ; pabsw %xmm0,%xmm9
+ DB 102,68,15,111,21,129,2,0,0 ; movdqa 0x281(%rip),%xmm10 # b60 <_sk_srcover_ssse3_lowp+0xfd>
+ DB 102,69,15,249,208 ; psubw %xmm8,%xmm10
+ DB 102,15,111,196 ; movdqa %xmm4,%xmm0
+ DB 102,65,15,56,11,194 ; pmulhrsw %xmm10,%xmm0
+ DB 102,15,56,29,192 ; pabsw %xmm0,%xmm0
+ DB 102,65,15,253,193 ; paddw %xmm9,%xmm0
+ DB 102,65,15,56,11,200 ; pmulhrsw %xmm8,%xmm1
+ DB 102,68,15,56,29,201 ; pabsw %xmm1,%xmm9
+ DB 102,15,111,205 ; movdqa %xmm5,%xmm1
+ DB 102,65,15,56,11,202 ; pmulhrsw %xmm10,%xmm1
+ DB 102,15,56,29,201 ; pabsw %xmm1,%xmm1
+ DB 102,65,15,253,201 ; paddw %xmm9,%xmm1
+ DB 102,65,15,56,11,208 ; pmulhrsw %xmm8,%xmm2
+ DB 102,68,15,56,29,202 ; pabsw %xmm2,%xmm9
+ DB 102,15,111,214 ; movdqa %xmm6,%xmm2
+ DB 102,65,15,56,11,210 ; pmulhrsw %xmm10,%xmm2
+ DB 102,15,56,29,210 ; pabsw %xmm2,%xmm2
+ DB 102,65,15,253,209 ; paddw %xmm9,%xmm2
+ DB 102,65,15,56,11,216 ; pmulhrsw %xmm8,%xmm3
+ DB 102,68,15,56,29,195 ; pabsw %xmm3,%xmm8
+ DB 102,68,15,56,11,215 ; pmulhrsw %xmm7,%xmm10
+ DB 102,65,15,56,29,218 ; pabsw %xmm10,%xmm3
+ DB 102,65,15,253,216 ; paddw %xmm8,%xmm3
+ DB 72,173 ; lods %ds:(%rsi),%rax
+ DB 65,89 ; pop %r9
+ DB 255,224 ; jmpq *%rax
+ DB 69,137,193 ; mov %r8d,%r9d
+ DB 65,128,225,7 ; and $0x7,%r9b
+ DB 102,69,15,239,192 ; pxor %xmm8,%xmm8
+ DB 65,254,201 ; dec %r9b
+ DB 65,128,249,6 ; cmp $0x6,%r9b
+ DB 15,135,71,255,255,255 ; ja 8bb <_sk_lerp_u8_ssse3_lowp+0x1a>
+ DB 69,15,182,201 ; movzbl %r9b,%r9d
+ DB 76,141,21,105,0,0,0 ; lea 0x69(%rip),%r10 # 9e8 <_sk_lerp_u8_ssse3_lowp+0x147>
+ DB 75,99,4,138 ; movslq (%r10,%r9,4),%rax
+ DB 76,1,208 ; add %r10,%rax
+ DB 255,224 ; jmpq *%rax
+ DB 65,15,182,68,19,6 ; movzbl 0x6(%r11,%rdx,1),%eax
+ DB 102,69,15,239,192 ; pxor %xmm8,%xmm8
+ DB 102,68,15,196,192,6 ; pinsrw $0x6,%eax,%xmm8
+ DB 65,15,182,68,19,5 ; movzbl 0x5(%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,5 ; pinsrw $0x5,%eax,%xmm8
+ DB 65,15,182,68,19,4 ; movzbl 0x4(%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,4 ; pinsrw $0x4,%eax,%xmm8
+ DB 65,15,182,68,19,3 ; movzbl 0x3(%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,3 ; pinsrw $0x3,%eax,%xmm8
+ DB 65,15,182,68,19,2 ; movzbl 0x2(%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,2 ; pinsrw $0x2,%eax,%xmm8
+ DB 65,15,182,68,19,1 ; movzbl 0x1(%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,1 ; pinsrw $0x1,%eax,%xmm8
+ DB 65,15,182,4,19 ; movzbl (%r11,%rdx,1),%eax
+ DB 102,68,15,196,192,0 ; pinsrw $0x0,%eax,%xmm8
+ DB 233,214,254,255,255 ; jmpq 8bb <_sk_lerp_u8_ssse3_lowp+0x1a>
+ DB 15,31,0 ; nopl (%rax)
+ DB 237 ; in (%dx),%eax
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255,225 ; jmpq *%rcx
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255,213 ; callq *%rbp
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255,201 ; dec %ecx
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 189,255,255,255,177 ; mov $0xb1ffffff,%ebp
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255 ; .byte 0xff
+ DB 160 ; .byte 0xa0
+ DB 255 ; (bad)
+ DB 255 ; (bad)
+ DB 255 ; .byte 0xff
+
PUBLIC _sk_swap_rb_ssse3_lowp
_sk_swap_rb_ssse3_lowp LABEL PROC
DB 68,15,40,192 ; movaps %xmm0,%xmm8
@@ -26584,7 +26962,7 @@ _sk_move_dst_src_ssse3_lowp LABEL PROC
PUBLIC _sk_srcover_ssse3_lowp
_sk_srcover_ssse3_lowp LABEL PROC
- DB 102,68,15,111,5,132,0,0,0 ; movdqa 0x84(%rip),%xmm8 # 4e0 <_sk_srcover_ssse3_lowp+0x8d>
+ DB 102,68,15,111,5,4,1,0,0 ; movdqa 0x104(%rip),%xmm8 # b70 <_sk_srcover_ssse3_lowp+0x10d>
DB 102,68,15,249,195 ; psubw %xmm3,%xmm8
DB 102,68,15,111,204 ; movdqa %xmm4,%xmm9
DB 102,69,15,56,11,200 ; pmulhrsw %xmm8,%xmm9
@@ -26606,6 +26984,9 @@ _sk_srcover_ssse3_lowp LABEL PROC
ALIGN 4
DB 0,0 ; add %al,(%rax)
+ DB 0,71,0 ; add %al,0x0(%rdi)
+ DB 0,0 ; add %al,(%rax)
+ DB 71,0,0 ; rex.RXB add %r8b,(%r8)
DB 0 ; .byte 0x0
DB 71 ; rex.RXB
@@ -26617,8 +26998,26 @@ ALIGN 16
DB 3,7 ; add (%rdi),%eax
DB 11,15 ; or (%rdi),%ecx
DB 129,128,129,128,129,128,129,128,129,128; addl $0x80818081,-0x7f7e7f7f(%rax)
+ DB 129,128,129,128,129,128,0,4,8,12 ; addl $0xc080400,-0x7f7e7f7f(%rax)
+ DB 1,5,9,13,2,6 ; add %eax,0x6020d09(%rip) # 6021813 <_sk_srcover_ssse3_lowp+0x6020db0>
+ DB 10,14 ; or (%rsi),%cl
+ DB 3,7 ; add (%rdi),%eax
+ DB 11,15 ; or (%rdi),%ecx
+ DB 129,128,129,128,129,128,129,128,129,128; addl $0x80818081,-0x7f7e7f7f(%rax)
+ DB 129,128,129,128,129,128,0,128,0,128 ; addl $0x80008000,-0x7f7e7f7f(%rax)
+ DB 0,128,0,128,0,128 ; add %al,-0x7fff8000(%rax)
+ DB 0,128,0,128,0,128 ; add %al,-0x7fff8000(%rax)
+ DB 129,128,129,128,129,128,129,128,129,128; addl $0x80818081,-0x7f7e7f7f(%rax)
DB 129,128,129,128,129,128,0,128,0,128 ; addl $0x80008000,-0x7f7e7f7f(%rax)
DB 0,128,0,128,0,128 ; add %al,-0x7fff8000(%rax)
DB 0,128,0,128,0,128 ; add %al,-0x7fff8000(%rax)
+ DB 129,128,129,128,129,128,129,128,129,128; addl $0x80818081,-0x7f7e7f7f(%rax)
+ DB 129,128,129,128,129,128,0,128,0,128 ; addl $0x80008000,-0x7f7e7f7f(%rax)
+ DB 0,128,0,128,0,128 ; add %al,-0x7fff8000(%rax)
+ DB 0,128,0,128,0,128 ; add %al,-0x7fff8000(%rax)
+ DB 0,128,0,128,0,128 ; add %al,-0x7fff8000(%rax)
+ DB 0,128,0,128,0,128 ; add %al,-0x7fff8000(%rax)
+ DB 0 ; .byte 0x0
+ DB 128,0,128 ; addb $0x80,(%rax)
ENDIF
END
diff --git a/src/jumper/SkJumper_stages_lowp.cpp b/src/jumper/SkJumper_stages_lowp.cpp
index 987bfa6518..d78de209fc 100644
--- a/src/jumper/SkJumper_stages_lowp.cpp
+++ b/src/jumper/SkJumper_stages_lowp.cpp
@@ -39,6 +39,7 @@ SI F operator-(F x, F y) { return x.vec - y.vec; }
SI F operator*(F x, F y) { return _mm_abs_epi16(_mm_mulhrs_epi16(x.vec, y.vec)); }
SI F mad(F f, F m, F a) { return f*m+a; }
SI F inv(F v) { return 1.0f - v; }
+SI F lerp(F from, F to, F t) { return to*t + from*inv(t); }
SI F operator<<(F x, int bits) { return x.vec << bits; }
SI F operator>>(F x, int bits) { return x.vec >> bits; }
@@ -144,6 +145,12 @@ SI void from_8888(U32 rgba, F* r, F* g, F* b, F* a) {
*b = _mm_mulhi_epu16(B, U16(32897));
*a = _mm_mulhi_epu16(A, U16(32897));
}
+SI F from_byte(U8 bytes) {
+ // See from_8888() just above.
+ U16 hi = _mm_unpacklo_epi8(U16(0), widen_cast<__m128i>(bytes));
+ return (F)_mm_mulhi_epu16(hi, U16(32897));
+}
+
SI U32 to_8888(F r, F g, F b, F a) {
// We want to interlace and pack these values from [0,32768] to [0,255].
// Luckily the simplest possible thing works great: >>7, then saturate.
@@ -182,6 +189,59 @@ STAGE(store_8888) {
store(ptr, to_8888(r,g,b,a), tail);
}
+STAGE(srcover_rgba_8888) {
+ auto ptr = *(uint32_t**)ctx + x;
+
+ from_8888(load<U32>(ptr, tail), &dr,&dg,&db,&da);
+
+ r = mad(dr, inv(a), r);
+ g = mad(dg, inv(a), g);
+ b = mad(db, inv(a), b);
+ a = mad(da, inv(a), a);
+
+ store(ptr, to_8888(r,g,b,a), tail);
+}
+
+STAGE(scale_1_float) {
+ float c = *(const float*)ctx;
+
+ r = r * c;
+ g = g * c;
+ b = b * c;
+ a = a * c;
+}
+STAGE(scale_u8) {
+ auto ptr = *(const uint8_t**)ctx + x;
+
+ U8 scales = load<U8>(ptr, tail);
+ F c = from_byte(scales);
+
+ r = r * c;
+ g = g * c;
+ b = b * c;
+ a = a * c;
+}
+
+STAGE(lerp_1_float) {
+ float c = *(const float*)ctx;
+
+ r = lerp(dr, r, c);
+ g = lerp(dg, g, c);
+ b = lerp(db, b, c);
+ a = lerp(da, a, c);
+}
+STAGE(lerp_u8) {
+ auto ptr = *(const uint8_t**)ctx + x;
+
+ U8 scales = load<U8>(ptr, tail);
+ F c = from_byte(scales);
+
+ r = lerp(dr, r, c);
+ g = lerp(dg, g, c);
+ b = lerp(db, b, c);
+ a = lerp(da, a, c);
+}
+
STAGE(swap_rb) {
auto tmp = r;
r = b;
generated by cgit v1.2.3 (git 2.39.1) at 2024-06-23 07:09:10 +0000