diff options
| -rw-r--r-- | src/jumper/SkJumper.cpp | 2 | ||||
| -rw-r--r-- | src/jumper/SkJumper_generated.S | 1763 | ||||
| -rw-r--r-- | src/jumper/SkJumper_stages.cpp | 2 | ||||
| -rw-r--r-- | src/jumper/SkJumper_vectors.h | 44 | ||||
| -rw-r--r-- | tests/SkRasterPipelineTest.cpp | 13 |
5 files changed, 1127 insertions, 697 deletions
diff --git a/src/jumper/SkJumper.cpp b/src/jumper/SkJumper.cpp index 428a7249b6..3ded4e47d1 100644 --- a/src/jumper/SkJumper.cpp +++ b/src/jumper/SkJumper.cpp @@ -122,7 +122,7 @@ static SkJumper_Engine choose_engine() { return { #define M(stage) ASM(stage, aarch64), { SK_RASTER_PIPELINE_STAGES(M) }, - 4, M(start_pipeline) M(just_return) + 1, M(start_pipeline) M(just_return) #undef M }; diff --git a/src/jumper/SkJumper_generated.S b/src/jumper/SkJumper_generated.S index 35f8a340e6..a760686440 100644 --- a/src/jumper/SkJumper_generated.S +++ b/src/jumper/SkJumper_generated.S @@ -33,14 +33,14 @@ _sk_start_pipeline_aarch64: .long 0xa9027bf3 // stp x19, x30, [sp, #32] .long 0xaa0103f4 // mov x20, x1 .long 0xf8408697 // ldr x23, [x20], #8 - .long 0xaa0003f5 // mov x21, x0 + .long 0xaa0003f6 // mov x22, x0 .long 0xaa0303f3 // mov x19, x3 - .long 0x910012a8 // add x8, x21, #0x4 + .long 0x910012c8 // add x8, x22, #0x4 .long 0xeb13011f // cmp x8, x19 - .long 0xaa0203f6 // mov x22, x2 + .long 0xaa0203f5 // mov x21, x2 .long 0x54000069 // b.ls 34 <sk_start_pipeline_aarch64+0x34> // b.plast - .long 0xaa1503e0 // mov x0, x21 - .long 0x14000012 // b 78 <sk_start_pipeline_aarch64+0x78> + .long 0xaa1603e0 // mov x0, x22 + .long 0x14000013 // b 7c <sk_start_pipeline_aarch64+0x7c> .long 0x6f00e400 // movi v0.2d, #0x0 .long 0x6f00e401 // movi v1.2d, #0x0 .long 0x6f00e402 // movi v2.2d, #0x0 @@ -49,15 +49,30 @@ _sk_start_pipeline_aarch64: .long 0x6f00e405 // movi v5.2d, #0x0 .long 0x6f00e406 // movi v6.2d, #0x0 .long 0x6f00e407 // movi v7.2d, #0x0 - .long 0xaa1503e0 // mov x0, x21 + .long 0xaa1603e0 // mov x0, x22 .long 0xaa1403e1 // mov x1, x20 - .long 0xaa1603e2 // mov x2, x22 + .long 0xaa1503e2 // mov x2, x21 + .long 0xaa1f03e3 // mov x3, xzr .long 0xd63f02e0 // blr x23 - .long 0x910012a0 // add x0, x21, #0x4 - .long 0x910022a8 // add x8, x21, #0x8 + .long 0x910012c0 // add x0, x22, #0x4 + .long 0x910022c8 // add x8, x22, #0x8 .long 0xeb13011f // cmp x8, x19 - .long 0xaa0003f5 // mov x21, x0 - .long 0x54fffe09 // b.ls 34 <sk_start_pipeline_aarch64+0x34> // b.plast + .long 0xaa0003f6 // mov x22, x0 + .long 0x54fffde9 // b.ls 34 <sk_start_pipeline_aarch64+0x34> // b.plast + .long 0xcb000263 // sub x3, x19, x0 + .long 0xb4000183 // cbz x3, b0 <sk_start_pipeline_aarch64+0xb0> + .long 0x6f00e400 // movi v0.2d, #0x0 + .long 0x6f00e401 // movi v1.2d, #0x0 + .long 0x6f00e402 // movi v2.2d, #0x0 + .long 0x6f00e403 // movi v3.2d, #0x0 + .long 0x6f00e404 // movi v4.2d, #0x0 + .long 0x6f00e405 // movi v5.2d, #0x0 + .long 0x6f00e406 // movi v6.2d, #0x0 + .long 0x6f00e407 // movi v7.2d, #0x0 + .long 0xaa1403e1 // mov x1, x20 + .long 0xaa1503e2 // mov x2, x21 + .long 0xd63f02e0 // blr x23 + .long 0xaa1303e0 // mov x0, x19 .long 0xa9427bf3 // ldp x19, x30, [sp, #32] .long 0xa94153f5 // ldp x21, x20, [sp, #16] .long 0xa8c35bf7 // ldp x23, x22, [sp], #48 @@ -73,7 +88,7 @@ HIDDEN _sk_seed_shader_aarch64 .globl _sk_seed_shader_aarch64 FUNCTION(_sk_seed_shader_aarch64) _sk_seed_shader_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x3dc00046 // ldr q6, [x2] .long 0x4e040c00 // dup v0.4s, w0 .long 0x4f0167e7 // movi v7.4s, #0x3f, lsl #24 @@ -89,13 +104,13 @@ _sk_seed_shader_aarch64: .long 0x6f00e406 // movi v6.2d, #0x0 .long 0x4e27d421 // fadd v1.4s, v1.4s, v7.4s .long 0x6f00e407 // movi v7.2d, #0x0 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_dither_aarch64 .globl _sk_dither_aarch64 FUNCTION(_sk_dither_aarch64) _sk_dither_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x3dc00851 // ldr q17, [x2, #32] .long 0x4e040c10 // dup v16.4s, w0 .long 0x4f000432 // movi v18.4s, #0x1 @@ -140,13 +155,13 @@ _sk_dither_aarch64: .long 0x4e20f660 // fmax v0.4s, v19.4s, v0.4s .long 0x4e21f661 // fmax v1.4s, v19.4s, v1.4s .long 0x4e22f662 // fmax v2.4s, v19.4s, v2.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_constant_color_aarch64 .globl _sk_constant_color_aarch64 FUNCTION(_sk_constant_color_aarch64) _sk_constant_color_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xaa0803ea // mov x10, x8 .long 0x4ddfc940 // ld1r {v0.4s}, [x10], #4 .long 0x91002109 // add x9, x8, #0x8 @@ -154,16 +169,16 @@ _sk_constant_color_aarch64: .long 0x4d40c922 // ld1r {v2.4s}, [x9] .long 0x4d40c903 // ld1r {v3.4s}, [x8] .long 0x4d40c941 // ld1r {v1.4s}, [x10] - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_load_rgba_aarch64 .globl _sk_load_rgba_aarch64 FUNCTION(_sk_load_rgba_aarch64) _sk_load_rgba_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xad400500 // ldp q0, q1, [x8] .long 0xad410d02 // ldp q2, q3, [x8, #32] - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_store_rgba_aarch64 .globl _sk_store_rgba_aarch64 @@ -172,20 +187,20 @@ _sk_store_rgba_aarch64: .long 0xf9400028 // ldr x8, [x1] .long 0xad000500 // stp q0, q1, [x8] .long 0xad010d02 // stp q2, q3, [x8, #32] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_clear_aarch64 .globl _sk_clear_aarch64 FUNCTION(_sk_clear_aarch64) _sk_clear_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6f00e400 // movi v0.2d, #0x0 .long 0x6f00e401 // movi v1.2d, #0x0 .long 0x6f00e402 // movi v2.2d, #0x0 .long 0x6f00e403 // movi v3.2d, #0x0 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_srcatop_aarch64 .globl _sk_srcatop_aarch64 @@ -196,14 +211,14 @@ _sk_srcatop_aarch64: .long 0x6e27dc21 // fmul v1.4s, v1.4s, v7.4s .long 0x6e27dc42 // fmul v2.4s, v2.4s, v7.4s .long 0x4ea3d610 // fsub v16.4s, v16.4s, v3.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4e30cc80 // fmla v0.4s, v4.4s, v16.4s .long 0x4e30cca1 // fmla v1.4s, v5.4s, v16.4s .long 0x4e30ccc2 // fmla v2.4s, v6.4s, v16.4s .long 0x6e27de10 // fmul v16.4s, v16.4s, v7.4s .long 0x4e23ccf0 // fmla v16.4s, v7.4s, v3.4s .long 0x4eb01e03 // mov v3.16b, v16.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_dstatop_aarch64 .globl _sk_dstatop_aarch64 @@ -211,7 +226,7 @@ FUNCTION(_sk_dstatop_aarch64) _sk_dstatop_aarch64: .long 0x4f03f610 // fmov v16.4s, #1.000000000000000000e+00 .long 0x4ea7d610 // fsub v16.4s, v16.4s, v7.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e20de00 // fmul v0.4s, v16.4s, v0.4s .long 0x6e21de01 // fmul v1.4s, v16.4s, v1.4s .long 0x6e22de02 // fmul v2.4s, v16.4s, v2.4s @@ -221,75 +236,75 @@ _sk_dstatop_aarch64: .long 0x4e23cca1 // fmla v1.4s, v5.4s, v3.4s .long 0x4e23ccc2 // fmla v2.4s, v6.4s, v3.4s .long 0x4eb01e03 // mov v3.16b, v16.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_srcin_aarch64 .globl _sk_srcin_aarch64 FUNCTION(_sk_srcin_aarch64) _sk_srcin_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e27dc00 // fmul v0.4s, v0.4s, v7.4s .long 0x6e27dc21 // fmul v1.4s, v1.4s, v7.4s .long 0x6e27dc42 // fmul v2.4s, v2.4s, v7.4s .long 0x6e27dc63 // fmul v3.4s, v3.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_dstin_aarch64 .globl _sk_dstin_aarch64 FUNCTION(_sk_dstin_aarch64) _sk_dstin_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e24dc60 // fmul v0.4s, v3.4s, v4.4s .long 0x6e25dc61 // fmul v1.4s, v3.4s, v5.4s .long 0x6e26dc62 // fmul v2.4s, v3.4s, v6.4s .long 0x6e27dc63 // fmul v3.4s, v3.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_srcout_aarch64 .globl _sk_srcout_aarch64 FUNCTION(_sk_srcout_aarch64) _sk_srcout_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4f03f610 // fmov v16.4s, #1.000000000000000000e+00 .long 0x4ea7d610 // fsub v16.4s, v16.4s, v7.4s .long 0x6e20de00 // fmul v0.4s, v16.4s, v0.4s .long 0x6e21de01 // fmul v1.4s, v16.4s, v1.4s .long 0x6e22de02 // fmul v2.4s, v16.4s, v2.4s .long 0x6e23de03 // fmul v3.4s, v16.4s, v3.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_dstout_aarch64 .globl _sk_dstout_aarch64 FUNCTION(_sk_dstout_aarch64) _sk_dstout_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4f03f600 // fmov v0.4s, #1.000000000000000000e+00 .long 0x4ea3d403 // fsub v3.4s, v0.4s, v3.4s .long 0x6e24dc60 // fmul v0.4s, v3.4s, v4.4s .long 0x6e25dc61 // fmul v1.4s, v3.4s, v5.4s .long 0x6e26dc62 // fmul v2.4s, v3.4s, v6.4s .long 0x6e27dc63 // fmul v3.4s, v3.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_srcover_aarch64 .globl _sk_srcover_aarch64 FUNCTION(_sk_srcover_aarch64) _sk_srcover_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4f03f610 // fmov v16.4s, #1.000000000000000000e+00 .long 0x4ea3d610 // fsub v16.4s, v16.4s, v3.4s .long 0x4e24ce00 // fmla v0.4s, v16.4s, v4.4s .long 0x4e25ce01 // fmla v1.4s, v16.4s, v5.4s .long 0x4e26ce02 // fmla v2.4s, v16.4s, v6.4s .long 0x4e27ce03 // fmla v3.4s, v16.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_dstover_aarch64 .globl _sk_dstover_aarch64 FUNCTION(_sk_dstover_aarch64) _sk_dstover_aarch64: .long 0x4f03f611 // fmov v17.4s, #1.000000000000000000e+00 - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4ea41c90 // mov v16.16b, v4.16b .long 0x4ea7d634 // fsub v20.4s, v17.4s, v7.4s .long 0x4ea51cb1 // mov v17.16b, v5.16b @@ -303,18 +318,18 @@ _sk_dstover_aarch64: .long 0x4eb11e21 // mov v1.16b, v17.16b .long 0x4eb21e42 // mov v2.16b, v18.16b .long 0x4eb31e63 // mov v3.16b, v19.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_modulate_aarch64 .globl _sk_modulate_aarch64 FUNCTION(_sk_modulate_aarch64) _sk_modulate_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e24dc00 // fmul v0.4s, v0.4s, v4.4s .long 0x6e25dc21 // fmul v1.4s, v1.4s, v5.4s .long 0x6e26dc42 // fmul v2.4s, v2.4s, v6.4s .long 0x6e27dc63 // fmul v3.4s, v3.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_multiply_aarch64 .globl _sk_multiply_aarch64 @@ -327,7 +342,7 @@ _sk_multiply_aarch64: .long 0x6e21de71 // fmul v17.4s, v19.4s, v1.4s .long 0x6e22de72 // fmul v18.4s, v19.4s, v2.4s .long 0x6e23de73 // fmul v19.4s, v19.4s, v3.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4e34cc90 // fmla v16.4s, v4.4s, v20.4s .long 0x4e34ccb1 // fmla v17.4s, v5.4s, v20.4s .long 0x4e34ccd2 // fmla v18.4s, v6.4s, v20.4s @@ -340,24 +355,24 @@ _sk_multiply_aarch64: .long 0x4eb11e21 // mov v1.16b, v17.16b .long 0x4eb21e42 // mov v2.16b, v18.16b .long 0x4eb31e63 // mov v3.16b, v19.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_plus__aarch64 .globl _sk_plus__aarch64 FUNCTION(_sk_plus__aarch64) _sk_plus__aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4e24d400 // fadd v0.4s, v0.4s, v4.4s .long 0x4e25d421 // fadd v1.4s, v1.4s, v5.4s .long 0x4e26d442 // fadd v2.4s, v2.4s, v6.4s .long 0x4e27d463 // fadd v3.4s, v3.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_screen_aarch64 .globl _sk_screen_aarch64 FUNCTION(_sk_screen_aarch64) _sk_screen_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4e24d410 // fadd v16.4s, v0.4s, v4.4s .long 0x4e25d431 // fadd v17.4s, v1.4s, v5.4s .long 0x4e26d452 // fadd v18.4s, v2.4s, v6.4s @@ -370,14 +385,14 @@ _sk_screen_aarch64: .long 0x4eb11e21 // mov v1.16b, v17.16b .long 0x4eb21e42 // mov v2.16b, v18.16b .long 0x4eb31e63 // mov v3.16b, v19.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_xor__aarch64 .globl _sk_xor__aarch64 FUNCTION(_sk_xor__aarch64) _sk_xor__aarch64: .long 0x4f03f610 // fmov v16.4s, #1.000000000000000000e+00 - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4ea7d611 // fsub v17.4s, v16.4s, v7.4s .long 0x4ea3d610 // fsub v16.4s, v16.4s, v3.4s .long 0x6e20de20 // fmul v0.4s, v17.4s, v0.4s @@ -388,7 +403,7 @@ _sk_xor__aarch64: .long 0x4e30cca1 // fmla v1.4s, v5.4s, v16.4s .long 0x4e30ccc2 // fmla v2.4s, v6.4s, v16.4s .long 0x4e30cce3 // fmla v3.4s, v7.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_darken_aarch64 .globl _sk_darken_aarch64 @@ -400,7 +415,7 @@ _sk_darken_aarch64: .long 0x6e25dc73 // fmul v19.4s, v3.4s, v5.4s .long 0x4e31f610 // fmax v16.4s, v16.4s, v17.4s .long 0x4e24d400 // fadd v0.4s, v0.4s, v4.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e27dc51 // fmul v17.4s, v2.4s, v7.4s .long 0x4e33f652 // fmax v18.4s, v18.4s, v19.4s .long 0x6e26dc73 // fmul v19.4s, v3.4s, v6.4s @@ -413,7 +428,7 @@ _sk_darken_aarch64: .long 0x4eb2d421 // fsub v1.4s, v1.4s, v18.4s .long 0x4eb1d442 // fsub v2.4s, v2.4s, v17.4s .long 0x4e27ce03 // fmla v3.4s, v16.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_lighten_aarch64 .globl _sk_lighten_aarch64 @@ -425,7 +440,7 @@ _sk_lighten_aarch64: .long 0x6e25dc73 // fmul v19.4s, v3.4s, v5.4s .long 0x4eb1f610 // fmin v16.4s, v16.4s, v17.4s .long 0x4e24d400 // fadd v0.4s, v0.4s, v4.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e27dc51 // fmul v17.4s, v2.4s, v7.4s .long 0x4eb3f652 // fmin v18.4s, v18.4s, v19.4s .long 0x6e26dc73 // fmul v19.4s, v3.4s, v6.4s @@ -438,7 +453,7 @@ _sk_lighten_aarch64: .long 0x4eb2d421 // fsub v1.4s, v1.4s, v18.4s .long 0x4eb1d442 // fsub v2.4s, v2.4s, v17.4s .long 0x4e27ce03 // fmla v3.4s, v16.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_difference_aarch64 .globl _sk_difference_aarch64 @@ -457,7 +472,7 @@ _sk_difference_aarch64: .long 0x4eb0d400 // fsub v0.4s, v0.4s, v16.4s .long 0x4e25d421 // fadd v1.4s, v1.4s, v5.4s .long 0x4e32d650 // fadd v16.4s, v18.4s, v18.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4eb3f631 // fmin v17.4s, v17.4s, v19.4s .long 0x4eb0d421 // fsub v1.4s, v1.4s, v16.4s .long 0x4f03f610 // fmov v16.4s, #1.000000000000000000e+00 @@ -466,7 +481,7 @@ _sk_difference_aarch64: .long 0x4ea3d610 // fsub v16.4s, v16.4s, v3.4s .long 0x4eb1d442 // fsub v2.4s, v2.4s, v17.4s .long 0x4e27ce03 // fmla v3.4s, v16.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_exclusion_aarch64 .globl _sk_exclusion_aarch64 @@ -483,12 +498,12 @@ _sk_exclusion_aarch64: .long 0x4e26d450 // fadd v16.4s, v2.4s, v6.4s .long 0x6e26dc42 // fmul v2.4s, v2.4s, v6.4s .long 0x4e22d442 // fadd v2.4s, v2.4s, v2.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4ea2d602 // fsub v2.4s, v16.4s, v2.4s .long 0x4f03f610 // fmov v16.4s, #1.000000000000000000e+00 .long 0x4ea3d610 // fsub v16.4s, v16.4s, v3.4s .long 0x4e27ce03 // fmla v3.4s, v16.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_colorburn_aarch64 .globl _sk_colorburn_aarch64 @@ -526,7 +541,7 @@ _sk_colorburn_aarch64: .long 0x4e20cc61 // fmla v1.4s, v3.4s, v0.4s .long 0x4ea0d840 // fcmeq v0.4s, v2.4s, #0.0 .long 0x4ea3d631 // fsub v17.4s, v17.4s, v3.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4e27e490 // fcmeq v16.4s, v4.4s, v7.4s .long 0x6e611c40 // bsl v0.16b, v2.16b, v1.16b .long 0x4e31cc92 // fmla v18.4s, v4.4s, v17.4s @@ -539,7 +554,7 @@ _sk_colorburn_aarch64: .long 0x6e601e82 // bsl v2.16b, v20.16b, v0.16b .long 0x4e27ce23 // fmla v3.4s, v17.4s, v7.4s .long 0x4eb01e00 // mov v0.16b, v16.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_colordodge_aarch64 .globl _sk_colordodge_aarch64 @@ -575,7 +590,7 @@ _sk_colordodge_aarch64: .long 0x4e23e440 // fcmeq v0.4s, v2.4s, v3.4s .long 0x6e611c40 // bsl v0.16b, v2.16b, v1.16b .long 0x4ea3d641 // fsub v1.4s, v18.4s, v3.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4ea0d890 // fcmeq v16.4s, v4.4s, #0.0 .long 0x4ea0d8b1 // fcmeq v17.4s, v5.4s, #0.0 .long 0x4e21cc93 // fmla v19.4s, v4.4s, v1.4s @@ -588,7 +603,7 @@ _sk_colordodge_aarch64: .long 0x4e27cc23 // fmla v3.4s, v1.4s, v7.4s .long 0x4eb01e00 // mov v0.16b, v16.16b .long 0x4eb11e21 // mov v1.16b, v17.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_hardlight_aarch64 .globl _sk_hardlight_aarch64 @@ -625,7 +640,7 @@ _sk_hardlight_aarch64: .long 0x6e21dea1 // fmul v1.4s, v21.4s, v1.4s .long 0x6e22deb5 // fmul v21.4s, v21.4s, v2.4s .long 0x6e26dc42 // fmul v2.4s, v2.4s, v6.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4e32d652 // fadd v18.4s, v18.4s, v18.4s .long 0x4ea3d610 // fsub v16.4s, v16.4s, v3.4s .long 0x6e34e474 // fcmge v20.4s, v3.4s, v20.4s @@ -639,7 +654,7 @@ _sk_hardlight_aarch64: .long 0x4e36d421 // fadd v1.4s, v1.4s, v22.4s .long 0x4e34d6a2 // fadd v2.4s, v21.4s, v20.4s .long 0x4e27ce03 // fmla v3.4s, v16.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_overlay_aarch64 .globl _sk_overlay_aarch64 @@ -676,7 +691,7 @@ _sk_overlay_aarch64: .long 0x6e21dea1 // fmul v1.4s, v21.4s, v1.4s .long 0x6e22deb5 // fmul v21.4s, v21.4s, v2.4s .long 0x6e26dc42 // fmul v2.4s, v2.4s, v6.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4e32d652 // fadd v18.4s, v18.4s, v18.4s .long 0x4ea3d610 // fsub v16.4s, v16.4s, v3.4s .long 0x6e34e4f4 // fcmge v20.4s, v7.4s, v20.4s @@ -690,7 +705,7 @@ _sk_overlay_aarch64: .long 0x4e36d421 // fadd v1.4s, v1.4s, v22.4s .long 0x4e34d6a2 // fadd v2.4s, v21.4s, v20.4s .long 0x4e27ce03 // fmla v3.4s, v16.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_softlight_aarch64 .globl _sk_softlight_aarch64 @@ -782,7 +797,7 @@ _sk_softlight_aarch64: .long 0x6e27df7b // fmul v27.4s, v27.4s, v7.4s .long 0x6e27ded6 // fmul v22.4s, v22.4s, v7.4s .long 0x6e27df18 // fmul v24.4s, v24.4s, v7.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e37df77 // fmul v23.4s, v27.4s, v23.4s .long 0x6e3aded6 // fmul v22.4s, v22.4s, v26.4s .long 0x6e34df14 // fmul v20.4s, v24.4s, v20.4s @@ -806,7 +821,7 @@ _sk_softlight_aarch64: .long 0x4e35d421 // fadd v1.4s, v1.4s, v21.4s .long 0x4e3cd442 // fadd v2.4s, v2.4s, v28.4s .long 0x4e27ce43 // fmla v3.4s, v18.4s, v7.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_hue_aarch64 .globl _sk_hue_aarch64 @@ -908,7 +923,7 @@ _sk_hue_aarch64: .long 0x6e35fe31 // fdiv v17.4s, v17.4s, v21.4s .long 0x6e35fe52 // fdiv v18.4s, v18.4s, v21.4s .long 0x6e35fe73 // fdiv v19.4s, v19.4s, v21.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4eb71ee3 // mov v3.16b, v23.16b .long 0x4eb71efc // mov v28.16b, v23.16b .long 0x4e31d711 // fadd v17.4s, v24.4s, v17.4s @@ -925,7 +940,7 @@ _sk_hue_aarch64: .long 0x4e31d421 // fadd v1.4s, v1.4s, v17.4s .long 0x4e30d442 // fadd v2.4s, v2.4s, v16.4s .long 0x4eb4d743 // fsub v3.4s, v26.4s, v20.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_saturation_aarch64 .globl _sk_saturation_aarch64 @@ -1027,7 +1042,7 @@ _sk_saturation_aarch64: .long 0x6e35fe31 // fdiv v17.4s, v17.4s, v21.4s .long 0x6e35fe52 // fdiv v18.4s, v18.4s, v21.4s .long 0x6e35fe73 // fdiv v19.4s, v19.4s, v21.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4eb71ee3 // mov v3.16b, v23.16b .long 0x4eb71efc // mov v28.16b, v23.16b .long 0x4e31d711 // fadd v17.4s, v24.4s, v17.4s @@ -1044,7 +1059,7 @@ _sk_saturation_aarch64: .long 0x4e31d421 // fadd v1.4s, v1.4s, v17.4s .long 0x4e30d442 // fadd v2.4s, v2.4s, v16.4s .long 0x4eb4d743 // fsub v3.4s, v26.4s, v20.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_color_aarch64 .globl _sk_color_aarch64 @@ -1122,7 +1137,7 @@ _sk_color_aarch64: .long 0x6e37fe31 // fdiv v17.4s, v17.4s, v23.4s .long 0x6e37fe73 // fdiv v19.4s, v19.4s, v23.4s .long 0x6e37fe94 // fdiv v20.4s, v20.4s, v23.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4eb91f23 // mov v3.16b, v25.16b .long 0x4eb91f3c // mov v28.16b, v25.16b .long 0x4e31d6b1 // fadd v17.4s, v21.4s, v17.4s @@ -1139,7 +1154,7 @@ _sk_color_aarch64: .long 0x4e31d421 // fadd v1.4s, v1.4s, v17.4s .long 0x4e32d442 // fadd v2.4s, v2.4s, v18.4s .long 0x4eb0d703 // fsub v3.4s, v24.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_luminosity_aarch64 .globl _sk_luminosity_aarch64 @@ -1217,7 +1232,7 @@ _sk_luminosity_aarch64: .long 0x6e33fc00 // fdiv v0.4s, v0.4s, v19.4s .long 0x6e33fc63 // fdiv v3.4s, v3.4s, v19.4s .long 0x6e33fe31 // fdiv v17.4s, v17.4s, v19.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4eb51ea2 // mov v2.16b, v21.16b .long 0x4eb51ebb // mov v27.16b, v21.16b .long 0x4e20d680 // fadd v0.4s, v20.4s, v0.4s @@ -1234,40 +1249,41 @@ _sk_luminosity_aarch64: .long 0x4e22d421 // fadd v1.4s, v1.4s, v2.4s .long 0x4e23d742 // fadd v2.4s, v26.4s, v3.4s .long 0x4eb0d703 // fsub v3.4s, v24.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_srcover_rgba_8888_aarch64 .globl _sk_srcover_rgba_8888_aarch64 FUNCTION(_sk_srcover_rgba_8888_aarch64) _sk_srcover_rgba_8888_aarch64: .long 0xf9400028 // ldr x8, [x1] - .long 0xd37ef409 // lsl x9, x0, #2 - .long 0x6f00e625 // movi v5.2d, #0xff000000ff - .long 0x4f03f604 // fmov v4.4s, #1.000000000000000000e+00 .long 0xf9400108 // ldr x8, [x8] - .long 0x52a86fea // mov w10, #0x437f0000 - .long 0x4ea3d490 // fsub v16.4s, v4.4s, v3.4s - .long 0x4e040d44 // dup v4.4s, w10 - .long 0x3ce96906 // ldr q6, [x8, x9] - .long 0x6e24dc00 // fmul v0.4s, v0.4s, v4.4s - .long 0x6e24dc21 // fmul v1.4s, v1.4s, v4.4s - .long 0x6e24dc42 // fmul v2.4s, v2.4s, v4.4s - .long 0x6f3804c7 // ushr v7.4s, v6.4s, #8 - .long 0x6f3004d1 // ushr v17.4s, v6.4s, #16 + .long 0x8b000908 // add x8, x8, x0, lsl #2 + .long 0xb5000503 // cbnz x3, 1130 <sk_srcover_rgba_8888_aarch64+0xac> + .long 0x3dc00104 // ldr q4, [x8] + .long 0x6f00e625 // movi v5.2d, #0xff000000ff + .long 0x6f380486 // ushr v6.4s, v4.4s, #8 + .long 0x6f300487 // ushr v7.4s, v4.4s, #16 + .long 0x6f280490 // ushr v16.4s, v4.4s, #24 + .long 0x4f03f611 // fmov v17.4s, #1.000000000000000000e+00 + .long 0x52a86fe9 // mov w9, #0x437f0000 + .long 0x4e251cc6 // and v6.16b, v6.16b, v5.16b .long 0x4e251cf2 // and v18.16b, v7.16b, v5.16b - .long 0x6e24dc63 // fmul v3.4s, v3.4s, v4.4s - .long 0x4e251cc4 // and v4.16b, v6.16b, v5.16b - .long 0x4e251e31 // and v17.16b, v17.16b, v5.16b - .long 0x4e21da45 // scvtf v5.4s, v18.4s - .long 0x6f2804c6 // ushr v6.4s, v6.4s, #24 + .long 0x4e21da07 // scvtf v7.4s, v16.4s + .long 0x4ea3d630 // fsub v16.4s, v17.4s, v3.4s + .long 0x4e040d31 // dup v17.4s, w9 + .long 0x4e251c84 // and v4.16b, v4.16b, v5.16b + .long 0x4e21d8c5 // scvtf v5.4s, v6.4s + .long 0x6e31dc21 // fmul v1.4s, v1.4s, v17.4s .long 0x4e21d884 // scvtf v4.4s, v4.4s + .long 0x6e31dc00 // fmul v0.4s, v0.4s, v17.4s .long 0x4e25ce01 // fmla v1.4s, v16.4s, v5.4s - .long 0x4e21d8c7 // scvtf v7.4s, v6.4s - .long 0x4e21da26 // scvtf v6.4s, v17.4s + .long 0x4e21da46 // scvtf v6.4s, v18.4s + .long 0x6e31dc42 // fmul v2.4s, v2.4s, v17.4s + .long 0x6e31dc63 // fmul v3.4s, v3.4s, v17.4s .long 0x4e24ce00 // fmla v0.4s, v16.4s, v4.4s .long 0x6e21a831 // fcvtnu v17.4s, v1.4s - .long 0x4e27ce03 // fmla v3.4s, v16.4s, v7.4s .long 0x4e26ce02 // fmla v2.4s, v16.4s, v6.4s + .long 0x4e27ce03 // fmla v3.4s, v16.4s, v7.4s .long 0x6e21a810 // fcvtnu v16.4s, v0.4s .long 0x4f285631 // shl v17.4s, v17.4s, #8 .long 0x4eb01e30 // orr v16.16b, v17.16b, v16.16b @@ -1277,74 +1293,103 @@ _sk_srcover_rgba_8888_aarch64: .long 0x6e21a871 // fcvtnu v17.4s, v3.4s .long 0x4f385631 // shl v17.4s, v17.4s, #24 .long 0x4eb11e10 // orr v16.16b, v16.16b, v17.16b - .long 0x3ca96910 // str q16, [x8, x9] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xb5000283 // cbnz x3, 116c <sk_srcover_rgba_8888_aarch64+0xe8> + .long 0x3d800110 // str q16, [x8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x6f00e404 // movi v4.2d, #0x0 + .long 0x54000140 // b.eq 1164 <sk_srcover_rgba_8888_aarch64+0xe0> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000c0 // b.eq 115c <sk_srcover_rgba_8888_aarch64+0xd8> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffa61 // b.ne 1098 <sk_srcover_rgba_8888_aarch64+0x14> // b.any + .long 0x91002109 // add x9, x8, #0x8 + .long 0x4e040fe4 // dup v4.4s, wzr + .long 0x4d408124 // ld1 {v4.s}[2], [x9] + .long 0x91001109 // add x9, x8, #0x4 + .long 0x0d409124 // ld1 {v4.s}[1], [x9] + .long 0x0d408104 // ld1 {v4.s}[0], [x8] + .long 0x17ffffcc // b 1098 <sk_srcover_rgba_8888_aarch64+0x14> + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x54000120 // b.eq 1198 <sk_srcover_rgba_8888_aarch64+0x114> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000a0 // b.eq 1190 <sk_srcover_rgba_8888_aarch64+0x10c> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffd01 // b.ne 1124 <sk_srcover_rgba_8888_aarch64+0xa0> // b.any + .long 0x91002109 // add x9, x8, #0x8 + .long 0x4d008130 // st1 {v16.s}[2], [x9] + .long 0x91001109 // add x9, x8, #0x4 + .long 0x0d009130 // st1 {v16.s}[1], [x9] + .long 0x0d008110 // st1 {v16.s}[0], [x8] + .long 0x17ffffe2 // b 1124 <sk_srcover_rgba_8888_aarch64+0xa0> HIDDEN _sk_clamp_0_aarch64 .globl _sk_clamp_0_aarch64 FUNCTION(_sk_clamp_0_aarch64) _sk_clamp_0_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6f00e410 // movi v16.2d, #0x0 .long 0x4e30f400 // fmax v0.4s, v0.4s, v16.4s .long 0x4e30f421 // fmax v1.4s, v1.4s, v16.4s .long 0x4e30f442 // fmax v2.4s, v2.4s, v16.4s .long 0x4e30f463 // fmax v3.4s, v3.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_clamp_1_aarch64 .globl _sk_clamp_1_aarch64 FUNCTION(_sk_clamp_1_aarch64) _sk_clamp_1_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4f03f610 // fmov v16.4s, #1.000000000000000000e+00 .long 0x4eb0f400 // fmin v0.4s, v0.4s, v16.4s .long 0x4eb0f421 // fmin v1.4s, v1.4s, v16.4s .long 0x4eb0f442 // fmin v2.4s, v2.4s, v16.4s .long 0x4eb0f463 // fmin v3.4s, v3.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_clamp_a_aarch64 .globl _sk_clamp_a_aarch64 FUNCTION(_sk_clamp_a_aarch64) _sk_clamp_a_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4f03f610 // fmov v16.4s, #1.000000000000000000e+00 .long 0x4eb0f463 // fmin v3.4s, v3.4s, v16.4s .long 0x4ea3f400 // fmin v0.4s, v0.4s, v3.4s .long 0x4ea3f421 // fmin v1.4s, v1.4s, v3.4s .long 0x4ea3f442 // fmin v2.4s, v2.4s, v3.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_set_rgb_aarch64 .globl _sk_set_rgb_aarch64 FUNCTION(_sk_set_rgb_aarch64) _sk_set_rgb_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xaa0803e9 // mov x9, x8 .long 0x4ddfc920 // ld1r {v0.4s}, [x9], #4 .long 0x91002108 // add x8, x8, #0x8 .long 0x4d40c902 // ld1r {v2.4s}, [x8] .long 0x4d40c921 // ld1r {v1.4s}, [x9] - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_swap_rb_aarch64 .globl _sk_swap_rb_aarch64 FUNCTION(_sk_swap_rb_aarch64) _sk_swap_rb_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4ea01c10 // mov v16.16b, v0.16b .long 0x4ea21c40 // mov v0.16b, v2.16b .long 0x4eb01e02 // mov v2.16b, v16.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_swap_aarch64 .globl _sk_swap_aarch64 FUNCTION(_sk_swap_aarch64) _sk_swap_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4ea31c70 // mov v16.16b, v3.16b .long 0x4ea21c51 // mov v17.16b, v2.16b .long 0x4ea11c32 // mov v18.16b, v1.16b @@ -1357,53 +1402,53 @@ _sk_swap_aarch64: .long 0x4eb21e45 // mov v5.16b, v18.16b .long 0x4eb11e26 // mov v6.16b, v17.16b .long 0x4eb01e07 // mov v7.16b, v16.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_move_src_dst_aarch64 .globl _sk_move_src_dst_aarch64 FUNCTION(_sk_move_src_dst_aarch64) _sk_move_src_dst_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4ea01c04 // mov v4.16b, v0.16b .long 0x4ea11c25 // mov v5.16b, v1.16b .long 0x4ea21c46 // mov v6.16b, v2.16b .long 0x4ea31c67 // mov v7.16b, v3.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_move_dst_src_aarch64 .globl _sk_move_dst_src_aarch64 FUNCTION(_sk_move_dst_src_aarch64) _sk_move_dst_src_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4ea41c80 // mov v0.16b, v4.16b .long 0x4ea51ca1 // mov v1.16b, v5.16b .long 0x4ea61cc2 // mov v2.16b, v6.16b .long 0x4ea71ce3 // mov v3.16b, v7.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_premul_aarch64 .globl _sk_premul_aarch64 FUNCTION(_sk_premul_aarch64) _sk_premul_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e23dc00 // fmul v0.4s, v0.4s, v3.4s .long 0x6e23dc21 // fmul v1.4s, v1.4s, v3.4s .long 0x6e23dc42 // fmul v2.4s, v2.4s, v3.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_unpremul_aarch64 .globl _sk_unpremul_aarch64 FUNCTION(_sk_unpremul_aarch64) _sk_unpremul_aarch64: .long 0x4f03f611 // fmov v17.4s, #1.000000000000000000e+00 - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4ea0d870 // fcmeq v16.4s, v3.4s, #0.0 .long 0x6e23fe31 // fdiv v17.4s, v17.4s, v3.4s .long 0x4e701e30 // bic v16.16b, v17.16b, v16.16b .long 0x6e20de00 // fmul v0.4s, v16.4s, v0.4s .long 0x6e21de01 // fmul v1.4s, v16.4s, v1.4s .long 0x6e22de02 // fmul v2.4s, v16.4s, v2.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_from_srgb_aarch64 .globl _sk_from_srgb_aarch64 @@ -1432,7 +1477,7 @@ _sk_from_srgb_aarch64: .long 0x6e20dc13 // fmul v19.4s, v0.4s, v0.4s .long 0x7288f5c8 // movk w8, #0x47ae .long 0x4e20ce15 // fmla v21.4s, v16.4s, v0.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e21dc34 // fmul v20.4s, v1.4s, v1.4s .long 0x4e33ceb6 // fmla v22.4s, v21.4s, v19.4s .long 0x4e040d13 // dup v19.4s, w8 @@ -1447,7 +1492,7 @@ _sk_from_srgb_aarch64: .long 0x6e761ea0 // bsl v0.16b, v21.16b, v22.16b .long 0x6e781e01 // bsl v1.16b, v16.16b, v24.16b .long 0x6e721e22 // bsl v2.16b, v17.16b, v18.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_to_srgb_aarch64 .globl _sk_to_srgb_aarch64 @@ -1509,7 +1554,7 @@ _sk_to_srgb_aarch64: .long 0x4e35fe94 // frecps v20.4s, v20.4s, v21.4s .long 0x6e34deb4 // fmul v20.4s, v21.4s, v20.4s .long 0x4ea1da75 // frecpe v21.4s, v19.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4e35fe73 // frecps v19.4s, v19.4s, v21.4s .long 0x6e33deb3 // fmul v19.4s, v21.4s, v19.4s .long 0x6e30de50 // fmul v16.4s, v18.4s, v16.4s @@ -1518,7 +1563,7 @@ _sk_to_srgb_aarch64: .long 0x6e701f20 // bsl v0.16b, v25.16b, v16.16b .long 0x6e721f41 // bsl v1.16b, v26.16b, v18.16b .long 0x6e731e22 // bsl v2.16b, v17.16b, v19.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_rgb_to_hsl_aarch64 .globl _sk_rgb_to_hsl_aarch64 @@ -1551,7 +1596,7 @@ _sk_rgb_to_hsl_aarch64: .long 0x4e38cc59 // fmla v25.4s, v2.4s, v24.4s .long 0x6e3ade02 // fmul v2.4s, v16.4s, v26.4s .long 0x72955568 // movk w8, #0xaaab - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4eb1d651 // fsub v17.4s, v18.4s, v17.4s .long 0x6ebae452 // fcmgt v18.4s, v2.4s, v26.4s .long 0x6e791ee1 // bsl v1.16b, v23.16b, v25.16b @@ -1562,7 +1607,7 @@ _sk_rgb_to_hsl_aarch64: .long 0x6e33dc00 // fmul v0.4s, v0.4s, v19.4s .long 0x4e761c00 // bic v0.16b, v0.16b, v22.16b .long 0x4e761c21 // bic v1.16b, v1.16b, v22.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_hsl_to_rgb_aarch64 .globl _sk_hsl_to_rgb_aarch64 @@ -1628,7 +1673,7 @@ _sk_hsl_to_rgb_aarch64: .long 0x6e31e435 // fcmge v21.4s, v1.4s, v17.4s .long 0x6e31e711 // fcmge v17.4s, v24.4s, v17.4s .long 0x6e3ee718 // fcmge v24.4s, v24.4s, v30.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e741ef1 // bsl v17.16b, v23.16b, v20.16b .long 0x6e3ee733 // fcmge v19.4s, v25.4s, v30.4s .long 0x6e3ee439 // fcmge v25.4s, v1.4s, v30.4s @@ -1643,52 +1688,72 @@ _sk_hsl_to_rgb_aarch64: .long 0x6e791c40 // bsl v0.16b, v2.16b, v25.16b .long 0x6e731c41 // bsl v1.16b, v2.16b, v19.16b .long 0x4eb01e02 // mov v2.16b, v16.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_scale_1_float_aarch64 .globl _sk_scale_1_float_aarch64 FUNCTION(_sk_scale_1_float_aarch64) _sk_scale_1_float_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xbd400110 // ldr s16, [x8] .long 0x4f909000 // fmul v0.4s, v0.4s, v16.s[0] .long 0x4f909021 // fmul v1.4s, v1.4s, v16.s[0] .long 0x4f909042 // fmul v2.4s, v2.4s, v16.s[0] .long 0x4f909063 // fmul v3.4s, v3.4s, v16.s[0] - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_scale_u8_aarch64 .globl _sk_scale_u8_aarch64 FUNCTION(_sk_scale_u8_aarch64) _sk_scale_u8_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 - .long 0x52a77009 // mov w9, #0x3b800000 - .long 0x72901029 // movk w9, #0x8081 - .long 0x4e040d30 // dup v16.4s, w9 + .long 0xf9400028 // ldr x8, [x1] .long 0xf9400108 // ldr x8, [x8] .long 0x8b000108 // add x8, x8, x0 + .long 0xb50002e3 // cbnz x3, 16b8 <sk_scale_u8_aarch64+0x68> .long 0x39400109 // ldrb w9, [x8] .long 0x3940050a // ldrb w10, [x8, #1] .long 0x3940090b // ldrb w11, [x8, #2] .long 0x39400d08 // ldrb w8, [x8, #3] - .long 0x4e021d31 // mov v17.h[0], w9 - .long 0x4e061d51 // mov v17.h[1], w10 - .long 0x4e0a1d71 // mov v17.h[2], w11 - .long 0x4e0e1d11 // mov v17.h[3], w8 - .long 0x2f10a631 // uxtl v17.4s, v17.4h - .long 0x6e21da31 // ucvtf v17.4s, v17.4s - .long 0x6e30de30 // fmul v16.4s, v17.4s, v16.4s + .long 0x4e021d30 // mov v16.h[0], w9 + .long 0x4e061d50 // mov v16.h[1], w10 + .long 0x4e0a1d70 // mov v16.h[2], w11 + .long 0x4e0e1d10 // mov v16.h[3], w8 + .long 0x2f07b7f0 // bic v16.4h, #0xff, lsl #8 + .long 0x52a77008 // mov w8, #0x3b800000 + .long 0x72901028 // movk w8, #0x8081 + .long 0xf9400424 // ldr x4, [x1, #8] + .long 0x2f10a610 // uxtl v16.4s, v16.4h + .long 0x4e040d11 // dup v17.4s, w8 + .long 0x6e21da10 // ucvtf v16.4s, v16.4s + .long 0x6e31de10 // fmul v16.4s, v16.4s, v17.4s .long 0x6e20de00 // fmul v0.4s, v16.4s, v0.4s .long 0x6e21de01 // fmul v1.4s, v16.4s, v1.4s .long 0x6e22de02 // fmul v2.4s, v16.4s, v2.4s .long 0x6e23de03 // fmul v3.4s, v16.4s, v3.4s - .long 0xd61f0060 // br x3 + .long 0x91004021 // add x1, x1, #0x10 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x2f00e410 // movi d16, #0x0 + .long 0x54000140 // b.eq 16ec <sk_scale_u8_aarch64+0x9c> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000c0 // b.eq 16e4 <sk_scale_u8_aarch64+0x94> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffd61 // b.ne 1680 <sk_scale_u8_aarch64+0x30> // b.any + .long 0x39400909 // ldrb w9, [x8, #2] + .long 0x0e020ff0 // dup v16.4h, wzr + .long 0x4e0a1d30 // mov v16.h[2], w9 + .long 0x39400509 // ldrb w9, [x8, #1] + .long 0x4e061d30 // mov v16.h[1], w9 + .long 0x39400108 // ldrb w8, [x8] + .long 0x4e021d10 // mov v16.h[0], w8 + .long 0x17ffffe3 // b 1680 <sk_scale_u8_aarch64+0x30> HIDDEN _sk_lerp_1_float_aarch64 .globl _sk_lerp_1_float_aarch64 FUNCTION(_sk_lerp_1_float_aarch64) _sk_lerp_1_float_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x4ea4d411 // fsub v17.4s, v0.4s, v4.4s .long 0x4ea41c80 // mov v0.16b, v4.16b .long 0x4ea5d432 // fsub v18.4s, v1.4s, v5.4s @@ -1702,275 +1767,361 @@ _sk_lerp_1_float_aarch64: .long 0x4ea71ce3 // mov v3.16b, v7.16b .long 0x4f901222 // fmla v2.4s, v17.4s, v16.s[0] .long 0x4f901243 // fmla v3.4s, v18.4s, v16.s[0] - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_lerp_u8_aarch64 .globl _sk_lerp_u8_aarch64 FUNCTION(_sk_lerp_u8_aarch64) _sk_lerp_u8_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 - .long 0x52a77009 // mov w9, #0x3b800000 - .long 0x72901029 // movk w9, #0x8081 - .long 0x4e040d30 // dup v16.4s, w9 + .long 0xf9400028 // ldr x8, [x1] .long 0xf9400108 // ldr x8, [x8] - .long 0x4ea4d412 // fsub v18.4s, v0.4s, v4.4s .long 0x8b000108 // add x8, x8, x0 - .long 0x3940010a // ldrb w10, [x8] - .long 0x39400509 // ldrb w9, [x8, #1] + .long 0xb50003e3 // cbnz x3, 17bc <sk_lerp_u8_aarch64+0x88> + .long 0x39400109 // ldrb w9, [x8] + .long 0x3940050a // ldrb w10, [x8, #1] .long 0x3940090b // ldrb w11, [x8, #2] .long 0x39400d08 // ldrb w8, [x8, #3] - .long 0x4e021d51 // mov v17.h[0], w10 - .long 0x4e061d31 // mov v17.h[1], w9 - .long 0x4e0a1d71 // mov v17.h[2], w11 - .long 0x4e0e1d11 // mov v17.h[3], w8 - .long 0x2f10a620 // uxtl v0.4s, v17.4h + .long 0x4e021d30 // mov v16.h[0], w9 + .long 0x4e061d50 // mov v16.h[1], w10 + .long 0x4e0a1d70 // mov v16.h[2], w11 + .long 0x4e0e1d10 // mov v16.h[3], w8 + .long 0x2f07b7f0 // bic v16.4h, #0xff, lsl #8 + .long 0x52a77008 // mov w8, #0x3b800000 + .long 0x72901028 // movk w8, #0x8081 + .long 0x4ea4d411 // fsub v17.4s, v0.4s, v4.4s + .long 0x2f10a600 // uxtl v0.4s, v16.4h + .long 0x4e040d10 // dup v16.4s, w8 .long 0x6e21d800 // ucvtf v0.4s, v0.4s + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x6e30dc10 // fmul v16.4s, v0.4s, v16.4s .long 0x4ea41c80 // mov v0.16b, v4.16b - .long 0x4ea5d431 // fsub v17.4s, v1.4s, v5.4s + .long 0x4ea5d432 // fsub v18.4s, v1.4s, v5.4s .long 0x4ea51ca1 // mov v1.16b, v5.16b - .long 0x4e32ce00 // fmla v0.4s, v16.4s, v18.4s - .long 0x4ea6d452 // fsub v18.4s, v2.4s, v6.4s - .long 0x4e31ce01 // fmla v1.4s, v16.4s, v17.4s + .long 0x4e31ce00 // fmla v0.4s, v16.4s, v17.4s + .long 0x4ea6d451 // fsub v17.4s, v2.4s, v6.4s + .long 0x4e32ce01 // fmla v1.4s, v16.4s, v18.4s .long 0x4ea61cc2 // mov v2.16b, v6.16b - .long 0x4ea7d471 // fsub v17.4s, v3.4s, v7.4s + .long 0x4ea7d472 // fsub v18.4s, v3.4s, v7.4s .long 0x4ea71ce3 // mov v3.16b, v7.16b - .long 0x4e32ce02 // fmla v2.4s, v16.4s, v18.4s - .long 0x4e31ce03 // fmla v3.4s, v16.4s, v17.4s - .long 0xd61f0060 // br x3 + .long 0x4e31ce02 // fmla v2.4s, v16.4s, v17.4s + .long 0x4e32ce03 // fmla v3.4s, v16.4s, v18.4s + .long 0x91004021 // add x1, x1, #0x10 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x2f00e410 // movi d16, #0x0 + .long 0x54000140 // b.eq 17f0 <sk_lerp_u8_aarch64+0xbc> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000c0 // b.eq 17e8 <sk_lerp_u8_aarch64+0xb4> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffc61 // b.ne 1764 <sk_lerp_u8_aarch64+0x30> // b.any + .long 0x39400909 // ldrb w9, [x8, #2] + .long 0x0e020ff0 // dup v16.4h, wzr + .long 0x4e0a1d30 // mov v16.h[2], w9 + .long 0x39400509 // ldrb w9, [x8, #1] + .long 0x4e061d30 // mov v16.h[1], w9 + .long 0x39400108 // ldrb w8, [x8] + .long 0x4e021d10 // mov v16.h[0], w8 + .long 0x17ffffdb // b 1764 <sk_lerp_u8_aarch64+0x30> HIDDEN _sk_lerp_565_aarch64 .globl _sk_lerp_565_aarch64 FUNCTION(_sk_lerp_565_aarch64) _sk_lerp_565_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 - .long 0xd37ff809 // lsl x9, x0, #1 - .long 0x4f072711 // movi v17.4s, #0xf8, lsl #8 - .long 0x4ea4d416 // fsub v22.4s, v0.4s, v4.4s + .long 0xf9400028 // ldr x8, [x1] .long 0xf9400108 // ldr x8, [x8] - .long 0x4f0007f4 // movi v20.4s, #0x1f - .long 0x4ea7d463 // fsub v3.4s, v3.4s, v7.4s - .long 0xfc696910 // ldr d16, [x8, x9] + .long 0x8b000508 // add x8, x8, x0, lsl #1 + .long 0xb50005c3 // cbnz x3, 18c0 <sk_lerp_565_aarch64+0xc4> + .long 0xfd400110 // ldr d16, [x8] + .long 0x321b17e8 // orr w8, wzr, #0x7e0 + .long 0x4e040d12 // dup v18.4s, w8 .long 0x52a6f088 // mov w8, #0x37840000 + .long 0x2f10a610 // uxtl v16.4s, v16.4h + .long 0x4f072711 // movi v17.4s, #0xf8, lsl #8 .long 0x72842108 // movk w8, #0x2108 + .long 0x4f0007f3 // movi v19.4s, #0x1f + .long 0x4e311e11 // and v17.16b, v16.16b, v17.16b + .long 0x4e331e13 // and v19.16b, v16.16b, v19.16b + .long 0x4e321e10 // and v16.16b, v16.16b, v18.16b .long 0x4e040d12 // dup v18.4s, w8 - .long 0x321b17e8 // orr w8, wzr, #0x7e0 - .long 0x4e040d13 // dup v19.4s, w8 - .long 0x52a74048 // mov w8, #0x3a020000 - .long 0x2f10a600 // uxtl v0.4s, v16.4h - .long 0x72810428 // movk w8, #0x821 - .long 0x4e311c10 // and v16.16b, v0.16b, v17.16b - .long 0x4e040d15 // dup v21.4s, w8 .long 0x52a7a088 // mov w8, #0x3d040000 - .long 0x4e331c11 // and v17.16b, v0.16b, v19.16b - .long 0x4e341c13 // and v19.16b, v0.16b, v20.16b - .long 0x4ea5d434 // fsub v20.4s, v1.4s, v5.4s - .long 0x4e21da01 // scvtf v1.4s, v16.4s .long 0x72842108 // movk w8, #0x2108 - .long 0x6e32dc30 // fmul v16.4s, v1.4s, v18.4s - .long 0x4ea6d452 // fsub v18.4s, v2.4s, v6.4s - .long 0x4e21da22 // scvtf v2.4s, v17.4s - .long 0x4ea51ca1 // mov v1.16b, v5.16b - .long 0x6e35dc51 // fmul v17.4s, v2.4s, v21.4s - .long 0x4e040d02 // dup v2.4s, w8 - .long 0x4e21da73 // scvtf v19.4s, v19.4s - .long 0x6e22de73 // fmul v19.4s, v19.4s, v2.4s - .long 0x4ea61cc2 // mov v2.16b, v6.16b - .long 0x4ea71cf5 // mov v21.16b, v7.16b - .long 0x4e34ce21 // fmla v1.4s, v17.4s, v20.4s + .long 0x4e21da31 // scvtf v17.4s, v17.4s + .long 0x6e32de31 // fmul v17.4s, v17.4s, v18.4s + .long 0x4e040d12 // dup v18.4s, w8 + .long 0x52a74048 // mov w8, #0x3a020000 + .long 0x72810428 // movk w8, #0x821 + .long 0x4ea4d414 // fsub v20.4s, v0.4s, v4.4s + .long 0x4e21da60 // scvtf v0.4s, v19.4s + .long 0x6e32dc12 // fmul v18.4s, v0.4s, v18.4s + .long 0x4e040d00 // dup v0.4s, w8 + .long 0x4e21da10 // scvtf v16.4s, v16.4s + .long 0x6e20de10 // fmul v16.4s, v16.4s, v0.4s + .long 0x4ea41c80 // mov v0.16b, v4.16b + .long 0x4e34ce20 // fmla v0.4s, v17.4s, v20.4s + .long 0x4ea7d463 // fsub v3.4s, v3.4s, v7.4s .long 0x4ea71cf4 // mov v20.16b, v7.16b - .long 0x4e32ce62 // fmla v2.4s, v19.4s, v18.4s + .long 0x4ea5d433 // fsub v19.4s, v1.4s, v5.4s + .long 0x4ea6d441 // fsub v1.4s, v2.4s, v6.4s + .long 0x4ea61cc2 // mov v2.16b, v6.16b + .long 0x4e23ce34 // fmla v20.4s, v17.4s, v3.4s + .long 0x4ea71cf1 // mov v17.16b, v7.16b + .long 0x4e21ce42 // fmla v2.4s, v18.4s, v1.4s + .long 0x4e23ce51 // fmla v17.4s, v18.4s, v3.4s .long 0x4ea71cf2 // mov v18.16b, v7.16b - .long 0x4e23ce35 // fmla v21.4s, v17.4s, v3.4s - .long 0x4e23ce74 // fmla v20.4s, v19.4s, v3.4s - .long 0x4ea41c80 // mov v0.16b, v4.16b + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4e23ce12 // fmla v18.4s, v16.4s, v3.4s - .long 0x4e34f6a3 // fmax v3.4s, v21.4s, v20.4s - .long 0x4e36ce00 // fmla v0.4s, v16.4s, v22.4s - .long 0x4e23f643 // fmax v3.4s, v18.4s, v3.4s - .long 0xd61f0060 // br x3 + .long 0x4ea51ca1 // mov v1.16b, v5.16b + .long 0x4e31f643 // fmax v3.4s, v18.4s, v17.4s + .long 0x4e33ce01 // fmla v1.4s, v16.4s, v19.4s + .long 0x4e23f683 // fmax v3.4s, v20.4s, v3.4s + .long 0x91004021 // add x1, x1, #0x10 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x2f00e410 // movi d16, #0x0 + .long 0x54000140 // b.eq 18f4 <sk_lerp_565_aarch64+0xf8> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000c0 // b.eq 18ec <sk_lerp_565_aarch64+0xf0> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fff9a1 // b.ne 1810 <sk_lerp_565_aarch64+0x14> // b.any + .long 0x91001109 // add x9, x8, #0x4 + .long 0x0e020ff0 // dup v16.4h, wzr + .long 0x0d405130 // ld1 {v16.h}[2], [x9] + .long 0x91000909 // add x9, x8, #0x2 + .long 0x0d404930 // ld1 {v16.h}[1], [x9] + .long 0x0d404110 // ld1 {v16.h}[0], [x8] + .long 0x17ffffc6 // b 1810 <sk_lerp_565_aarch64+0x14> HIDDEN _sk_load_tables_aarch64 .globl _sk_load_tables_aarch64 FUNCTION(_sk_load_tables_aarch64) _sk_load_tables_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 - .long 0xd37ef409 // lsl x9, x0, #2 + .long 0xf9400028 // ldr x8, [x1] + .long 0xf9400109 // ldr x9, [x8] + .long 0x8b000929 // add x9, x9, x0, lsl #2 + .long 0xb50006e3 // cbnz x3, 19e4 <sk_load_tables_aarch64+0xe8> + .long 0x3dc00122 // ldr q2, [x9] + .long 0xa940a909 // ldp x9, x10, [x8, #8] .long 0x6f00e620 // movi v0.2d, #0xff000000ff + .long 0x6f380441 // ushr v1.4s, v2.4s, #8 + .long 0x4e201c50 // and v16.16b, v2.16b, v0.16b + .long 0x4e201c21 // and v1.16b, v1.16b, v0.16b + .long 0x1e26020e // fmov w14, s16 + .long 0xf9400d08 // ldr x8, [x8, #24] + .long 0x6f300443 // ushr v3.4s, v2.4s, #16 + .long 0x1e260032 // fmov w18, s1 + .long 0x8b2e492e // add x14, x9, w14, uxtw #2 .long 0x52a7700b // mov w11, #0x3b800000 - .long 0xa940310a // ldp x10, x12, [x8] - .long 0x7290102b // movk w11, #0x8081 - .long 0x4e040d63 // dup v3.4s, w11 - .long 0x3ce96942 // ldr q2, [x10, x9] - .long 0xa9412109 // ldp x9, x8, [x8, #16] - .long 0x4e201c41 // and v1.16b, v2.16b, v0.16b - .long 0x1e26002e // fmov w14, s1 - .long 0x6f380450 // ushr v16.4s, v2.4s, #8 - .long 0x6f300451 // ushr v17.4s, v2.4s, #16 - .long 0x8b2e498e // add x14, x12, w14, uxtw #2 - .long 0x0e0c3c2a // mov w10, v1.s[1] - .long 0x0e143c2b // mov w11, v1.s[2] - .long 0x0e1c3c2d // mov w13, v1.s[3] - .long 0x4e201e01 // and v1.16b, v16.16b, v0.16b - .long 0x4e201e30 // and v16.16b, v17.16b, v0.16b + .long 0x4e201c63 // and v3.16b, v3.16b, v0.16b .long 0x0d4081c0 // ld1 {v0.s}[0], [x14] - .long 0x8b2a498a // add x10, x12, w10, uxtw #2 - .long 0xbc6b5991 // ldr s17, [x12, w11, uxtw #2] - .long 0xbc6d5992 // ldr s18, [x12, w13, uxtw #2] - .long 0x0e0c3c2b // mov w11, v1.s[1] - .long 0x0e143c2c // mov w12, v1.s[2] - .long 0x0e1c3c2d // mov w13, v1.s[3] - .long 0x1e26002e // fmov w14, s1 - .long 0x8b2e492e // add x14, x9, w14, uxtw #2 - .long 0xbc6c5933 // ldr s19, [x9, w12, uxtw #2] - .long 0xbc6d5934 // ldr s20, [x9, w13, uxtw #2] - .long 0x8b2b4929 // add x9, x9, w11, uxtw #2 - .long 0x1e26020b // fmov w11, s16 + .long 0x8b324952 // add x18, x10, w18, uxtw #2 + .long 0x7290102b // movk w11, #0x8081 + .long 0x0e0c3c2f // mov w15, v1.s[1] + .long 0x0e143c30 // mov w16, v1.s[2] + .long 0x0e1c3c31 // mov w17, v1.s[3] + .long 0x0d408241 // ld1 {v1.s}[0], [x18] + .long 0x4e040d71 // dup v17.4s, w11 + .long 0x0e0c3e0b // mov w11, v16.s[1] + .long 0x1e26006e // fmov w14, s3 .long 0x6f280442 // ushr v2.4s, v2.4s, #24 - .long 0x0d409140 // ld1 {v0.s}[1], [x10] - .long 0x4e21d842 // scvtf v2.4s, v2.4s - .long 0x8b2b490a // add x10, x8, w11, uxtw #2 - .long 0x0d4081c1 // ld1 {v1.s}[0], [x14] - .long 0x6e23dc43 // fmul v3.4s, v2.4s, v3.4s - .long 0x0d408142 // ld1 {v2.s}[0], [x10] - .long 0x0e0c3e0f // mov w15, v16.s[1] .long 0x0e143e0c // mov w12, v16.s[2] - .long 0x8b2f490a // add x10, x8, w15, uxtw #2 + .long 0xbc705952 // ldr s18, [x10, w16, uxtw #2] + .long 0x0e143c70 // mov w16, v3.s[2] + .long 0x4e21d842 // scvtf v2.4s, v2.4s + .long 0x8b2e490e // add x14, x8, w14, uxtw #2 + .long 0x8b2b492b // add x11, x9, w11, uxtw #2 .long 0x0e1c3e0d // mov w13, v16.s[3] - .long 0xbc6c5910 // ldr s16, [x8, w12, uxtw #2] - .long 0x0d409121 // ld1 {v1.s}[1], [x9] + .long 0xbc6c5930 // ldr s16, [x9, w12, uxtw #2] + .long 0x0e0c3c6c // mov w12, v3.s[1] + .long 0xbc705913 // ldr s19, [x8, w16, uxtw #2] + .long 0x0e1c3c70 // mov w16, v3.s[3] + .long 0x6e31dc43 // fmul v3.4s, v2.4s, v17.4s + .long 0x0d4081c2 // ld1 {v2.s}[0], [x14] + .long 0x0d409160 // ld1 {v0.s}[1], [x11] + .long 0x8b2f494b // add x11, x10, w15, uxtw #2 + .long 0x0d409161 // ld1 {v1.s}[1], [x11] + .long 0x8b2c490b // add x11, x8, w12, uxtw #2 + .long 0x0d409162 // ld1 {v2.s}[1], [x11] + .long 0x6e140600 // mov v0.s[2], v16.s[0] + .long 0xbc6d5930 // ldr s16, [x9, w13, uxtw #2] + .long 0x6e140641 // mov v1.s[2], v18.s[0] + .long 0xbc715951 // ldr s17, [x10, w17, uxtw #2] + .long 0xbc705912 // ldr s18, [x8, w16, uxtw #2] + .long 0xf9400424 // ldr x4, [x1, #8] + .long 0x6e140662 // mov v2.s[2], v19.s[0] + .long 0x6e1c0600 // mov v0.s[3], v16.s[0] + .long 0x6e1c0621 // mov v1.s[3], v17.s[0] + .long 0x6e1c0642 // mov v2.s[3], v18.s[0] + .long 0x91004021 // add x1, x1, #0x10 + .long 0xd61f0080 // br x4 + .long 0x1200046a // and w10, w3, #0x3 + .long 0x7100055f // cmp w10, #0x1 + .long 0x6f00e402 // movi v2.2d, #0x0 + .long 0x54000140 // b.eq 1a18 <sk_load_tables_aarch64+0x11c> // b.none + .long 0x7100095f // cmp w10, #0x2 + .long 0x540000c0 // b.eq 1a10 <sk_load_tables_aarch64+0x114> // b.none + .long 0x71000d5f // cmp w10, #0x3 + .long 0x54fff881 // b.ne 1910 <sk_load_tables_aarch64+0x14> // b.any + .long 0x9100212a // add x10, x9, #0x8 + .long 0x4e040fe2 // dup v2.4s, wzr + .long 0x4d408142 // ld1 {v2.s}[2], [x10] + .long 0x9100112a // add x10, x9, #0x4 .long 0x0d409142 // ld1 {v2.s}[1], [x10] - .long 0x6e140620 // mov v0.s[2], v17.s[0] - .long 0xbc6d5911 // ldr s17, [x8, w13, uxtw #2] - .long 0x6e140661 // mov v1.s[2], v19.s[0] - .long 0x6e140602 // mov v2.s[2], v16.s[0] - .long 0x6e1c0640 // mov v0.s[3], v18.s[0] - .long 0x6e1c0681 // mov v1.s[3], v20.s[0] - .long 0x6e1c0622 // mov v2.s[3], v17.s[0] - .long 0xd61f0060 // br x3 + .long 0x0d408122 // ld1 {v2.s}[0], [x9] + .long 0x17ffffbd // b 1910 <sk_load_tables_aarch64+0x14> HIDDEN _sk_load_tables_u16_be_aarch64 .globl _sk_load_tables_u16_be_aarch64 FUNCTION(_sk_load_tables_u16_be_aarch64) _sk_load_tables_u16_be_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 - .long 0x52a6f00a // mov w10, #0x37800000 - .long 0x7280100a // movk w10, #0x80 - .long 0x4e040d50 // dup v16.4s, w10 - .long 0xa9402d09 // ldp x9, x11, [x8] + .long 0xf9400028 // ldr x8, [x1] + .long 0xf9400109 // ldr x9, [x8] .long 0x8b000d29 // add x9, x9, x0, lsl #3 + .long 0xb5000743 // cbnz x3, 1b14 <sk_load_tables_u16_be_aarch64+0xf4> .long 0x0c400520 // ld4 {v0.4h-v3.4h}, [x9] - .long 0xa9412109 // ldp x9, x8, [x8, #16] + .long 0xa940a909 // ldp x9, x10, [x8, #8] .long 0x2f07b7e0 // bic v0.4h, #0xff, lsl #8 - .long 0x2f10a411 // uxtl v17.4s, v0.4h + .long 0x52a6f00b // mov w11, #0x37800000 + .long 0x7280100b // movk w11, #0x80 + .long 0x2f10a410 // uxtl v16.4s, v0.4h .long 0x2f07b7e1 // bic v1.4h, #0xff, lsl #8 - .long 0x1e26022e // fmov w14, s17 - .long 0x0e0c3e2a // mov w10, v17.s[1] - .long 0x0e143e2c // mov w12, v17.s[2] - .long 0x0e1c3e2d // mov w13, v17.s[3] - .long 0x8b2e496e // add x14, x11, w14, uxtw #2 - .long 0x2f10a431 // uxtl v17.4s, v1.4h + .long 0xf9400d08 // ldr x8, [x8, #24] + .long 0x4e040d71 // dup v17.4s, w11 + .long 0x0e0c3e0b // mov w11, v16.s[1] + .long 0x0e143e0c // mov w12, v16.s[2] + .long 0x0e1c3e0d // mov w13, v16.s[3] + .long 0x1e26020e // fmov w14, s16 + .long 0x2f10a430 // uxtl v16.4s, v1.4h .long 0x2f07b7e2 // bic v2.4h, #0xff, lsl #8 - .long 0x2f10a453 // uxtl v19.4s, v2.4h - .long 0x2f180462 // ushr v2.4h, v3.4h, #8 - .long 0x0d4081c0 // ld1 {v0.s}[0], [x14] + .long 0x0e0c3e0f // mov w15, v16.s[1] + .long 0xbc6c5932 // ldr s18, [x9, w12, uxtw #2] + .long 0x0e143e0c // mov w12, v16.s[2] + .long 0xbc6d5933 // ldr s19, [x9, w13, uxtw #2] + .long 0x0e1c3e0d // mov w13, v16.s[3] + .long 0x8b2e492e // add x14, x9, w14, uxtw #2 + .long 0x8b2b4929 // add x9, x9, w11, uxtw #2 + .long 0x1e26020b // fmov w11, s16 + .long 0x2f10a450 // uxtl v16.4s, v2.4h .long 0x0f185474 // shl v20.4h, v3.4h, #8 + .long 0x2f180462 // ushr v2.4h, v3.4h, #8 .long 0x0ea21e82 // orr v2.8b, v20.8b, v2.8b - .long 0x8b2a496a // add x10, x11, w10, uxtw #2 - .long 0x1e26022e // fmov w14, s17 - .long 0x1e26026f // fmov w15, s19 + .long 0x8b2b494b // add x11, x10, w11, uxtw #2 + .long 0x0d4081c0 // ld1 {v0.s}[0], [x14] + .long 0x1e26020e // fmov w14, s16 .long 0x2f10a442 // uxtl v2.4s, v2.4h - .long 0x8b2e492e // add x14, x9, w14, uxtw #2 - .long 0x0d409140 // ld1 {v0.s}[1], [x10] - .long 0x8b2f490a // add x10, x8, w15, uxtw #2 + .long 0x0d408161 // ld1 {v1.s}[0], [x11] + .long 0x8b2e490b // add x11, x8, w14, uxtw #2 .long 0x6e21d842 // ucvtf v2.4s, v2.4s - .long 0x0d4081c1 // ld1 {v1.s}[0], [x14] - .long 0x6e30dc43 // fmul v3.4s, v2.4s, v16.4s - .long 0x0d408142 // ld1 {v2.s}[0], [x10] - .long 0xbc6c5972 // ldr s18, [x11, w12, uxtw #2] - .long 0xbc6d5975 // ldr s21, [x11, w13, uxtw #2] - .long 0x0e0c3e2b // mov w11, v17.s[1] - .long 0x0e143e2c // mov w12, v17.s[2] - .long 0x0e1c3e2d // mov w13, v17.s[3] - .long 0x8b2b492b // add x11, x9, w11, uxtw #2 - .long 0xbc6c5931 // ldr s17, [x9, w12, uxtw #2] - .long 0xbc6d5936 // ldr s22, [x9, w13, uxtw #2] - .long 0x0e0c3e69 // mov w9, v19.s[1] - .long 0x0e143e6c // mov w12, v19.s[2] - .long 0x8b294909 // add x9, x8, w9, uxtw #2 - .long 0x0e1c3e6d // mov w13, v19.s[3] - .long 0xbc6c5913 // ldr s19, [x8, w12, uxtw #2] - .long 0x0d409161 // ld1 {v1.s}[1], [x11] + .long 0x6e31dc43 // fmul v3.4s, v2.4s, v17.4s + .long 0x0d408162 // ld1 {v2.s}[0], [x11] + .long 0x8b2f494f // add x15, x10, w15, uxtw #2 + .long 0xbc6c5955 // ldr s21, [x10, w12, uxtw #2] + .long 0xbc6d5956 // ldr s22, [x10, w13, uxtw #2] + .long 0x0e0c3e0a // mov w10, v16.s[1] + .long 0x0e143e0c // mov w12, v16.s[2] + .long 0x0d409120 // ld1 {v0.s}[1], [x9] + .long 0x8b2a4909 // add x9, x8, w10, uxtw #2 + .long 0x0e1c3e0d // mov w13, v16.s[3] + .long 0xbc6c5910 // ldr s16, [x8, w12, uxtw #2] + .long 0x0d4091e1 // ld1 {v1.s}[1], [x15] .long 0x0d409122 // ld1 {v2.s}[1], [x9] - .long 0xbc6d5910 // ldr s16, [x8, w13, uxtw #2] + .long 0xbc6d5911 // ldr s17, [x8, w13, uxtw #2] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x6e140640 // mov v0.s[2], v18.s[0] - .long 0x6e140621 // mov v1.s[2], v17.s[0] - .long 0x6e140662 // mov v2.s[2], v19.s[0] - .long 0x6e1c06a0 // mov v0.s[3], v21.s[0] + .long 0x6e1406a1 // mov v1.s[2], v21.s[0] + .long 0x6e140602 // mov v2.s[2], v16.s[0] + .long 0x6e1c0660 // mov v0.s[3], v19.s[0] .long 0x6e1c06c1 // mov v1.s[3], v22.s[0] - .long 0x6e1c0602 // mov v2.s[3], v16.s[0] - .long 0xd61f0060 // br x3 + .long 0x6e1c0622 // mov v2.s[3], v17.s[0] + .long 0x91004021 // add x1, x1, #0x10 + .long 0xd61f0080 // br x4 + .long 0x0d606120 // ld4 {v0.h-v3.h}[0], [x9] + .long 0xf100047f // cmp x3, #0x1 + .long 0x54fff8c0 // b.eq 1a34 <sk_load_tables_u16_be_aarch64+0x14> // b.none + .long 0x9100212a // add x10, x9, #0x8 + .long 0x0d606940 // ld4 {v0.h-v3.h}[1], [x10] + .long 0xf1000c7f // cmp x3, #0x3 + .long 0x54fff843 // b.cc 1a34 <sk_load_tables_u16_be_aarch64+0x14> // b.lo, b.ul, b.last + .long 0x91004129 // add x9, x9, #0x10 + .long 0x0d607120 // ld4 {v0.h-v3.h}[2], [x9] + .long 0x17ffffbf // b 1a34 <sk_load_tables_u16_be_aarch64+0x14> HIDDEN _sk_load_tables_rgb_u16_be_aarch64 .globl _sk_load_tables_rgb_u16_be_aarch64 FUNCTION(_sk_load_tables_rgb_u16_be_aarch64) _sk_load_tables_rgb_u16_be_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xf9400028 // ldr x8, [x1] .long 0x321f07ea // orr w10, wzr, #0x6 - .long 0xa9402d09 // ldp x9, x11, [x8] + .long 0xf9400109 // ldr x9, [x8] .long 0x9b0a2409 // madd x9, x0, x10, x9 - .long 0x0c404521 // ld3 {v1.4h-v3.4h}, [x9] - .long 0xa9412109 // ldp x9, x8, [x8, #16] + .long 0xb5000663 // cbnz x3, 1c18 <sk_load_tables_rgb_u16_be_aarch64+0xdc> + .long 0x0c404520 // ld3 {v0.4h-v2.4h}, [x9] + .long 0xa940a909 // ldp x9, x10, [x8, #8] + .long 0x2f07b7e0 // bic v0.4h, #0xff, lsl #8 + .long 0x2f10a403 // uxtl v3.4s, v0.4h .long 0x2f07b7e1 // bic v1.4h, #0xff, lsl #8 - .long 0x2f10a420 // uxtl v0.4s, v1.4h + .long 0xf9400d0b // ldr x11, [x8, #24] + .long 0x0e0c3c68 // mov w8, v3.s[1] + .long 0x0e143c6c // mov w12, v3.s[2] + .long 0x0e1c3c6d // mov w13, v3.s[3] + .long 0x1e26006e // fmov w14, s3 + .long 0x2f10a423 // uxtl v3.4s, v1.4h .long 0x2f07b7e2 // bic v2.4h, #0xff, lsl #8 - .long 0x1e26000e // fmov w14, s0 - .long 0x0e143c0c // mov w12, v0.s[2] - .long 0x8b2e496e // add x14, x11, w14, uxtw #2 - .long 0x2f10a450 // uxtl v16.4s, v2.4h - .long 0x0e0c3c0a // mov w10, v0.s[1] - .long 0x0e1c3c0d // mov w13, v0.s[3] - .long 0xbc6c5971 // ldr s17, [x11, w12, uxtw #2] - .long 0x0d4081c0 // ld1 {v0.s}[0], [x14] - .long 0x0e143e0c // mov w12, v16.s[2] - .long 0x1e26020e // fmov w14, s16 - .long 0x2f07b7e3 // bic v3.4h, #0xff, lsl #8 - .long 0x8b2a496a // add x10, x11, w10, uxtw #2 - .long 0xbc6d5972 // ldr s18, [x11, w13, uxtw #2] - .long 0x0e0c3e0b // mov w11, v16.s[1] - .long 0x0e1c3e0d // mov w13, v16.s[3] .long 0xbc6c5930 // ldr s16, [x9, w12, uxtw #2] - .long 0x8b2e492c // add x12, x9, w14, uxtw #2 - .long 0x2f10a462 // uxtl v2.4s, v3.4h - .long 0xbc6d5923 // ldr s3, [x9, w13, uxtw #2] - .long 0x0d408181 // ld1 {v1.s}[0], [x12] - .long 0x0e143c4c // mov w12, v2.s[2] - .long 0x1e26004d // fmov w13, s2 - .long 0xbc6c5913 // ldr s19, [x8, w12, uxtw #2] - .long 0x8b2d490c // add x12, x8, w13, uxtw #2 - .long 0x8b2b492b // add x11, x9, w11, uxtw #2 - .long 0x0e0c3c49 // mov w9, v2.s[1] - .long 0x0d409140 // ld1 {v0.s}[1], [x10] - .long 0x0e1c3c4a // mov w10, v2.s[3] - .long 0x0d408182 // ld1 {v2.s}[0], [x12] - .long 0x8b294909 // add x9, x8, w9, uxtw #2 - .long 0x0d409161 // ld1 {v1.s}[1], [x11] - .long 0x6e140620 // mov v0.s[2], v17.s[0] - .long 0x0d409122 // ld1 {v2.s}[1], [x9] - .long 0xbc6a5911 // ldr s17, [x8, w10, uxtw #2] - .long 0x6e140601 // mov v1.s[2], v16.s[0] - .long 0x6e1c0640 // mov v0.s[3], v18.s[0] - .long 0x6e140662 // mov v2.s[2], v19.s[0] - .long 0x6e1c0461 // mov v1.s[3], v3.s[0] - .long 0x6e1c0622 // mov v2.s[3], v17.s[0] + .long 0xbc6d5931 // ldr s17, [x9, w13, uxtw #2] + .long 0x8b2e492e // add x14, x9, w14, uxtw #2 + .long 0x8b284928 // add x8, x9, w8, uxtw #2 + .long 0x1e260069 // fmov w9, s3 + .long 0x2f10a442 // uxtl v2.4s, v2.4h + .long 0x8b294949 // add x9, x10, w9, uxtw #2 + .long 0x0d4081c0 // ld1 {v0.s}[0], [x14] + .long 0x0e143c4e // mov w14, v2.s[2] + .long 0x0d408121 // ld1 {v1.s}[0], [x9] + .long 0xbc6e5972 // ldr s18, [x11, w14, uxtw #2] + .long 0x1e26004e // fmov w14, s2 + .long 0x0e0c3c6f // mov w15, v3.s[1] + .long 0x0e143c6c // mov w12, v3.s[2] + .long 0x8b2e496e // add x14, x11, w14, uxtw #2 + .long 0x0e1c3c6d // mov w13, v3.s[3] + .long 0xbc6c5943 // ldr s3, [x10, w12, uxtw #2] + .long 0x0e0c3c4c // mov w12, v2.s[1] + .long 0x0e1c3c49 // mov w9, v2.s[3] + .long 0x0d4081c2 // ld1 {v2.s}[0], [x14] + .long 0x0d409100 // ld1 {v0.s}[1], [x8] + .long 0x8b2f4948 // add x8, x10, w15, uxtw #2 + .long 0x0d409101 // ld1 {v1.s}[1], [x8] + .long 0x8b2c4968 // add x8, x11, w12, uxtw #2 + .long 0x0d409102 // ld1 {v2.s}[1], [x8] + .long 0x6e140600 // mov v0.s[2], v16.s[0] + .long 0xbc6d5950 // ldr s16, [x10, w13, uxtw #2] + .long 0x6e140461 // mov v1.s[2], v3.s[0] + .long 0xbc695963 // ldr s3, [x11, w9, uxtw #2] + .long 0xf9400424 // ldr x4, [x1, #8] + .long 0x91004028 // add x8, x1, #0x10 + .long 0x6e140642 // mov v2.s[2], v18.s[0] + .long 0x6e1c0620 // mov v0.s[3], v17.s[0] + .long 0x6e1c0601 // mov v1.s[3], v16.s[0] + .long 0x6e1c0462 // mov v2.s[3], v3.s[0] .long 0x4f03f603 // fmov v3.4s, #1.000000000000000000e+00 - .long 0xd61f0060 // br x3 + .long 0xaa0803e1 // mov x1, x8 + .long 0xd61f0080 // br x4 + .long 0x0d406120 // ld3 {v0.h-v2.h}[0], [x9] + .long 0xf100047f // cmp x3, #0x1 + .long 0x54fff9a0 // b.eq 1b54 <sk_load_tables_rgb_u16_be_aarch64+0x18> // b.none + .long 0x9100192a // add x10, x9, #0x6 + .long 0x0d406940 // ld3 {v0.h-v2.h}[1], [x10] + .long 0xf1000c7f // cmp x3, #0x3 + .long 0x54fff923 // b.cc 1b54 <sk_load_tables_rgb_u16_be_aarch64+0x18> // b.lo, b.ul, b.last + .long 0x91003129 // add x9, x9, #0xc + .long 0x0d407120 // ld3 {v0.h-v2.h}[2], [x9] + .long 0x17ffffc6 // b 1b54 <sk_load_tables_rgb_u16_be_aarch64+0x18> HIDDEN _sk_byte_tables_aarch64 .globl _sk_byte_tables_aarch64 FUNCTION(_sk_byte_tables_aarch64) _sk_byte_tables_aarch64: - .long 0xd10083ff // sub sp, sp, #0x20 + .long 0xf81e0ff5 // str x21, [sp, #-32]! .long 0xaa0103e8 // mov x8, x1 .long 0x91002109 // add x9, x8, #0x8 .long 0xa9014ff4 // stp x20, x19, [sp, #16] @@ -1979,7 +2130,7 @@ _sk_byte_tables_aarch64: .long 0x52a86fea // mov w10, #0x437f0000 .long 0x4e040d51 // dup v17.4s, w10 .long 0x52a7700b // mov w11, #0x3b800000 - .long 0xa9405126 // ldp x6, x20, [x9] + .long 0xa9405527 // ldp x7, x21, [x9] .long 0x6e31dc00 // fmul v0.4s, v0.4s, v17.4s .long 0x7290102b // movk w11, #0x8081 .long 0x6e21a800 // fcvtnu v0.4s, v0.4s @@ -1988,11 +2139,11 @@ _sk_byte_tables_aarch64: .long 0x0e143c0b // mov w11, v0.s[2] .long 0x0e1c3c0c // mov w12, v0.s[3] .long 0x1e26000d // fmov w13, s0 - .long 0x386d48cd // ldrb w13, [x6, w13, uxtw] - .long 0x386a48ca // ldrb w10, [x6, w10, uxtw] - .long 0x386b48cb // ldrb w11, [x6, w11, uxtw] - .long 0x386c48cc // ldrb w12, [x6, w12, uxtw] - .long 0xa9412526 // ldp x6, x9, [x9, #16] + .long 0x386d48ed // ldrb w13, [x7, w13, uxtw] + .long 0x386a48ea // ldrb w10, [x7, w10, uxtw] + .long 0x386b48eb // ldrb w11, [x7, w11, uxtw] + .long 0x386c48ec // ldrb w12, [x7, w12, uxtw] + .long 0xa9412527 // ldp x7, x9, [x9, #16] .long 0x6e31dc42 // fmul v2.4s, v2.4s, v17.4s .long 0x6e31dc21 // fmul v1.4s, v1.4s, v17.4s .long 0x6e31dc63 // fmul v3.4s, v3.4s, v17.4s @@ -2000,46 +2151,46 @@ _sk_byte_tables_aarch64: .long 0x6e21a821 // fcvtnu v1.4s, v1.4s .long 0x6e21a863 // fcvtnu v3.4s, v3.4s .long 0x0e0c3c52 // mov w18, v2.s[1] - .long 0x0e143c43 // mov w3, v2.s[2] - .long 0x0e1c3c44 // mov w4, v2.s[3] - .long 0x1e260045 // fmov w5, s2 + .long 0x0e143c44 // mov w4, v2.s[2] + .long 0x0e1c3c45 // mov w5, v2.s[3] + .long 0x1e260046 // fmov w6, s2 .long 0x1e260031 // fmov w17, s1 - .long 0x386548c5 // ldrb w5, [x6, w5, uxtw] - .long 0x387248d2 // ldrb w18, [x6, w18, uxtw] - .long 0x386348c3 // ldrb w3, [x6, w3, uxtw] - .long 0x386448c4 // ldrb w4, [x6, w4, uxtw] - .long 0x1e260066 // fmov w6, s3 + .long 0x386648e6 // ldrb w6, [x7, w6, uxtw] + .long 0x387248f2 // ldrb w18, [x7, w18, uxtw] + .long 0x386448e4 // ldrb w4, [x7, w4, uxtw] + .long 0x386548e5 // ldrb w5, [x7, w5, uxtw] + .long 0x1e260067 // fmov w7, s3 .long 0x0e0c3c2e // mov w14, v1.s[1] - .long 0x0e0c3c67 // mov w7, v3.s[1] - .long 0x38714a91 // ldrb w17, [x20, w17, uxtw] - .long 0x38664926 // ldrb w6, [x9, w6, uxtw] + .long 0x0e0c3c73 // mov w19, v3.s[1] + .long 0x38714ab1 // ldrb w17, [x21, w17, uxtw] + .long 0x38674927 // ldrb w7, [x9, w7, uxtw] .long 0x0e143c2f // mov w15, v1.s[2] .long 0x0e1c3c30 // mov w16, v1.s[3] - .long 0x0e143c73 // mov w19, v3.s[2] - .long 0x386e4a8e // ldrb w14, [x20, w14, uxtw] - .long 0x38674927 // ldrb w7, [x9, w7, uxtw] - .long 0x386f4a8f // ldrb w15, [x20, w15, uxtw] - .long 0x38704a90 // ldrb w16, [x20, w16, uxtw] - .long 0x0e1c3c74 // mov w20, v3.s[3] + .long 0x0e143c74 // mov w20, v3.s[2] + .long 0x386e4aae // ldrb w14, [x21, w14, uxtw] .long 0x38734933 // ldrb w19, [x9, w19, uxtw] - .long 0x38744929 // ldrb w9, [x9, w20, uxtw] + .long 0x386f4aaf // ldrb w15, [x21, w15, uxtw] + .long 0x38704ab0 // ldrb w16, [x21, w16, uxtw] + .long 0x0e1c3c75 // mov w21, v3.s[3] + .long 0x38744934 // ldrb w20, [x9, w20, uxtw] + .long 0x38754929 // ldrb w9, [x9, w21, uxtw] .long 0x4e021da0 // mov v0.h[0], w13 .long 0x4e021e21 // mov v1.h[0], w17 - .long 0x4e021ca2 // mov v2.h[0], w5 - .long 0x4e021cc3 // mov v3.h[0], w6 + .long 0x4e021cc2 // mov v2.h[0], w6 + .long 0x4e021ce3 // mov v3.h[0], w7 .long 0x4e061d40 // mov v0.h[1], w10 .long 0x4e061dc1 // mov v1.h[1], w14 .long 0x4e061e42 // mov v2.h[1], w18 - .long 0x4e061ce3 // mov v3.h[1], w7 + .long 0x4e061e63 // mov v3.h[1], w19 .long 0x4e0a1d60 // mov v0.h[2], w11 .long 0x4e0a1de1 // mov v1.h[2], w15 - .long 0x4e0a1c62 // mov v2.h[2], w3 - .long 0x4e0a1e63 // mov v3.h[2], w19 + .long 0x4e0a1c82 // mov v2.h[2], w4 + .long 0x4e0a1e83 // mov v3.h[2], w20 .long 0x4e0e1d80 // mov v0.h[3], w12 .long 0x4e0e1e01 // mov v1.h[3], w16 - .long 0x4e0e1c82 // mov v2.h[3], w4 + .long 0x4e0e1ca2 // mov v2.h[3], w5 .long 0x4e0e1d23 // mov v3.h[3], w9 - .long 0xf9400505 // ldr x5, [x8, #8] + .long 0xf9400506 // ldr x6, [x8, #8] .long 0x2f07b7e0 // bic v0.4h, #0xff, lsl #8 .long 0x2f07b7e1 // bic v1.4h, #0xff, lsl #8 .long 0x2f07b7e2 // bic v2.4h, #0xff, lsl #8 @@ -2057,14 +2208,14 @@ _sk_byte_tables_aarch64: .long 0x6e30dc21 // fmul v1.4s, v1.4s, v16.4s .long 0x6e30dc42 // fmul v2.4s, v2.4s, v16.4s .long 0x6e30dc63 // fmul v3.4s, v3.4s, v16.4s - .long 0x910083ff // add sp, sp, #0x20 - .long 0xd61f00a0 // br x5 + .long 0xf84207f5 // ldr x21, [sp], #32 + .long 0xd61f00c0 // br x6 HIDDEN _sk_byte_tables_rgb_aarch64 .globl _sk_byte_tables_rgb_aarch64 FUNCTION(_sk_byte_tables_rgb_aarch64) _sk_byte_tables_rgb_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x52a77009 // mov w9, #0x3b800000 .long 0x72901029 // movk w9, #0x8081 .long 0x4e040d30 // dup v16.4s, w9 @@ -2092,27 +2243,27 @@ _sk_byte_tables_rgb_aarch64: .long 0x1e260051 // fmov w17, s2 .long 0x0e0c3c09 // mov w9, v0.s[1] .long 0x386e494e // ldrb w14, [x10, w14, uxtw] - .long 0x0e0c3c44 // mov w4, v2.s[1] + .long 0x0e0c3c45 // mov w5, v2.s[1] .long 0x38714911 // ldrb w17, [x8, w17, uxtw] .long 0x0e143c0c // mov w12, v0.s[2] .long 0x0e1c3c0d // mov w13, v0.s[3] - .long 0x0e143c45 // mov w5, v2.s[2] + .long 0x0e143c46 // mov w6, v2.s[2] .long 0x38694949 // ldrb w9, [x10, w9, uxtw] - .long 0x38644904 // ldrb w4, [x8, w4, uxtw] + .long 0x38654905 // ldrb w5, [x8, w5, uxtw] .long 0x386c494c // ldrb w12, [x10, w12, uxtw] .long 0x386d494a // ldrb w10, [x10, w13, uxtw] .long 0x0e1c3c4d // mov w13, v2.s[3] - .long 0x38654905 // ldrb w5, [x8, w5, uxtw] + .long 0x38664906 // ldrb w6, [x8, w6, uxtw] .long 0x386d4908 // ldrb w8, [x8, w13, uxtw] .long 0x4e021dc0 // mov v0.h[0], w14 .long 0x4e021e41 // mov v1.h[0], w18 .long 0x4e021e22 // mov v2.h[0], w17 .long 0x4e061d20 // mov v0.h[1], w9 .long 0x4e061de1 // mov v1.h[1], w15 - .long 0x4e061c82 // mov v2.h[1], w4 + .long 0x4e061ca2 // mov v2.h[1], w5 .long 0x4e0a1d80 // mov v0.h[2], w12 .long 0x4e0a1e01 // mov v1.h[2], w16 - .long 0x4e0a1ca2 // mov v2.h[2], w5 + .long 0x4e0a1cc2 // mov v2.h[2], w6 .long 0x4e0e1d40 // mov v0.h[3], w10 .long 0x4e0e1d61 // mov v1.h[3], w11 .long 0x4e0e1d02 // mov v2.h[3], w8 @@ -2128,13 +2279,13 @@ _sk_byte_tables_rgb_aarch64: .long 0x6e30dc00 // fmul v0.4s, v0.4s, v16.4s .long 0x6e30dc21 // fmul v1.4s, v1.4s, v16.4s .long 0x6e30dc42 // fmul v2.4s, v2.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_table_r_aarch64 .globl _sk_table_r_aarch64 FUNCTION(_sk_table_r_aarch64) _sk_table_r_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xb9400909 // ldr w9, [x8, #8] .long 0xf9400108 // ldr x8, [x8] .long 0x51000529 // sub w9, w9, #0x1 @@ -2154,13 +2305,13 @@ _sk_table_r_aarch64: .long 0xbc6b5911 // ldr s17, [x8, w11, uxtw #2] .long 0x6e140600 // mov v0.s[2], v16.s[0] .long 0x6e1c0620 // mov v0.s[3], v17.s[0] - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_table_g_aarch64 .globl _sk_table_g_aarch64 FUNCTION(_sk_table_g_aarch64) _sk_table_g_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xb9400909 // ldr w9, [x8, #8] .long 0xf9400108 // ldr x8, [x8] .long 0x51000529 // sub w9, w9, #0x1 @@ -2180,13 +2331,13 @@ _sk_table_g_aarch64: .long 0xbc6b5911 // ldr s17, [x8, w11, uxtw #2] .long 0x6e140601 // mov v1.s[2], v16.s[0] .long 0x6e1c0621 // mov v1.s[3], v17.s[0] - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_table_b_aarch64 .globl _sk_table_b_aarch64 FUNCTION(_sk_table_b_aarch64) _sk_table_b_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xb9400909 // ldr w9, [x8, #8] .long 0xf9400108 // ldr x8, [x8] .long 0x51000529 // sub w9, w9, #0x1 @@ -2206,13 +2357,13 @@ _sk_table_b_aarch64: .long 0xbc6b5911 // ldr s17, [x8, w11, uxtw #2] .long 0x6e140602 // mov v2.s[2], v16.s[0] .long 0x6e1c0622 // mov v2.s[3], v17.s[0] - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_table_a_aarch64 .globl _sk_table_a_aarch64 FUNCTION(_sk_table_a_aarch64) _sk_table_a_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xb9400909 // ldr w9, [x8, #8] .long 0xf9400108 // ldr x8, [x8] .long 0x51000529 // sub w9, w9, #0x1 @@ -2232,7 +2383,7 @@ _sk_table_a_aarch64: .long 0xbc6b5911 // ldr s17, [x8, w11, uxtw #2] .long 0x6e140603 // mov v3.s[2], v16.s[0] .long 0x6e1c0623 // mov v3.s[3], v17.s[0] - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_parametric_r_aarch64 .globl _sk_parametric_r_aarch64 @@ -2299,7 +2450,7 @@ _sk_parametric_r_aarch64: .long 0x4f026573 // movi v19.4s, #0x4b, lsl #24 .long 0x6e33de31 // fmul v17.4s, v17.4s, v19.4s .long 0x6e21aa31 // fcvtnu v17.4s, v17.4s - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4e34d631 // fadd v17.4s, v17.4s, v20.4s .long 0x6f00e412 // movi v18.2d, #0x0 .long 0x6e711e00 // bsl v0.16b, v16.16b, v17.16b @@ -2307,7 +2458,7 @@ _sk_parametric_r_aarch64: .long 0x4e32f400 // fmax v0.4s, v0.4s, v18.4s .long 0x4eb5f400 // fmin v0.4s, v0.4s, v21.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_parametric_g_aarch64 .globl _sk_parametric_g_aarch64 @@ -2374,7 +2525,7 @@ _sk_parametric_g_aarch64: .long 0x4f026573 // movi v19.4s, #0x4b, lsl #24 .long 0x6e33de31 // fmul v17.4s, v17.4s, v19.4s .long 0x6e21aa31 // fcvtnu v17.4s, v17.4s - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4e34d631 // fadd v17.4s, v17.4s, v20.4s .long 0x6f00e412 // movi v18.2d, #0x0 .long 0x6e711e01 // bsl v1.16b, v16.16b, v17.16b @@ -2382,7 +2533,7 @@ _sk_parametric_g_aarch64: .long 0x4e32f421 // fmax v1.4s, v1.4s, v18.4s .long 0x4eb5f421 // fmin v1.4s, v1.4s, v21.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_parametric_b_aarch64 .globl _sk_parametric_b_aarch64 @@ -2449,7 +2600,7 @@ _sk_parametric_b_aarch64: .long 0x4f026573 // movi v19.4s, #0x4b, lsl #24 .long 0x6e33de31 // fmul v17.4s, v17.4s, v19.4s .long 0x6e21aa31 // fcvtnu v17.4s, v17.4s - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4e34d631 // fadd v17.4s, v17.4s, v20.4s .long 0x6f00e412 // movi v18.2d, #0x0 .long 0x6e711e02 // bsl v2.16b, v16.16b, v17.16b @@ -2457,7 +2608,7 @@ _sk_parametric_b_aarch64: .long 0x4e32f442 // fmax v2.4s, v2.4s, v18.4s .long 0x4eb5f442 // fmin v2.4s, v2.4s, v21.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_parametric_a_aarch64 .globl _sk_parametric_a_aarch64 @@ -2524,7 +2675,7 @@ _sk_parametric_a_aarch64: .long 0x4f026573 // movi v19.4s, #0x4b, lsl #24 .long 0x6e33de31 // fmul v17.4s, v17.4s, v19.4s .long 0x6e21aa31 // fcvtnu v17.4s, v17.4s - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4e34d631 // fadd v17.4s, v17.4s, v20.4s .long 0x6f00e412 // movi v18.2d, #0x0 .long 0x6e711e03 // bsl v3.16b, v16.16b, v17.16b @@ -2532,7 +2683,7 @@ _sk_parametric_a_aarch64: .long 0x4e32f463 // fmax v3.4s, v3.4s, v18.4s .long 0x4eb5f463 // fmin v3.4s, v3.4s, v21.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_lab_to_xyz_aarch64 .globl _sk_lab_to_xyz_aarch64 @@ -2590,7 +2741,7 @@ _sk_lab_to_xyz_aarch64: .long 0x52a7ea68 // mov w8, #0x3f530000 .long 0x4e30d6b0 // fadd v16.4s, v21.4s, v16.4s .long 0x6e32deb2 // fmul v18.4s, v21.4s, v18.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x72881ec8 // movk w8, #0x40f6 .long 0x6e22de02 // fmul v2.4s, v16.4s, v2.4s .long 0x6ea0e640 // fcmgt v0.4s, v18.4s, v0.4s @@ -2598,39 +2749,60 @@ _sk_lab_to_xyz_aarch64: .long 0x6e621e40 // bsl v0.16b, v18.16b, v2.16b .long 0x6e31dc00 // fmul v0.4s, v0.4s, v17.4s .long 0x6e34de62 // fmul v2.4s, v19.4s, v20.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_load_a8_aarch64 .globl _sk_load_a8_aarch64 FUNCTION(_sk_load_a8_aarch64) _sk_load_a8_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 - .long 0x52a77009 // mov w9, #0x3b800000 - .long 0x72901029 // movk w9, #0x8081 - .long 0x4e040d22 // dup v2.4s, w9 + .long 0xf9400028 // ldr x8, [x1] .long 0xf9400108 // ldr x8, [x8] - .long 0x6f00e400 // movi v0.2d, #0x0 - .long 0x6f00e401 // movi v1.2d, #0x0 .long 0x8b000108 // add x8, x8, x0 - .long 0x3940010a // ldrb w10, [x8] - .long 0x3940050b // ldrb w11, [x8, #1] - .long 0x3940090c // ldrb w12, [x8, #2] + .long 0xb50002e3 // cbnz x3, 25b4 <sk_load_a8_aarch64+0x68> + .long 0x39400109 // ldrb w9, [x8] + .long 0x3940050a // ldrb w10, [x8, #1] + .long 0x3940090b // ldrb w11, [x8, #2] .long 0x39400d08 // ldrb w8, [x8, #3] - .long 0x4e021d43 // mov v3.h[0], w10 - .long 0x4e061d63 // mov v3.h[1], w11 - .long 0x4e0a1d83 // mov v3.h[2], w12 - .long 0x4e0e1d03 // mov v3.h[3], w8 - .long 0x2f10a463 // uxtl v3.4s, v3.4h - .long 0x6e21d863 // ucvtf v3.4s, v3.4s - .long 0x6e22dc63 // fmul v3.4s, v3.4s, v2.4s + .long 0x4e021d22 // mov v2.h[0], w9 + .long 0x4e061d42 // mov v2.h[1], w10 + .long 0x4e0a1d62 // mov v2.h[2], w11 + .long 0x4e0e1d02 // mov v2.h[3], w8 + .long 0x2f07b7e2 // bic v2.4h, #0xff, lsl #8 + .long 0x52a77008 // mov w8, #0x3b800000 + .long 0xf9400424 // ldr x4, [x1, #8] + .long 0x72901028 // movk w8, #0x8081 + .long 0x2f10a442 // uxtl v2.4s, v2.4h + .long 0x4e040d03 // dup v3.4s, w8 + .long 0x91004028 // add x8, x1, #0x10 + .long 0x6e21d842 // ucvtf v2.4s, v2.4s + .long 0x6f00e400 // movi v0.2d, #0x0 + .long 0x6f00e401 // movi v1.2d, #0x0 + .long 0x6e23dc43 // fmul v3.4s, v2.4s, v3.4s .long 0x6f00e402 // movi v2.2d, #0x0 - .long 0xd61f0060 // br x3 + .long 0xaa0803e1 // mov x1, x8 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x2f00e402 // movi d2, #0x0 + .long 0x54000140 // b.eq 25e8 <sk_load_a8_aarch64+0x9c> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000c0 // b.eq 25e0 <sk_load_a8_aarch64+0x94> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffd61 // b.ne 257c <sk_load_a8_aarch64+0x30> // b.any + .long 0x39400909 // ldrb w9, [x8, #2] + .long 0x0e020fe2 // dup v2.4h, wzr + .long 0x4e0a1d22 // mov v2.h[2], w9 + .long 0x39400509 // ldrb w9, [x8, #1] + .long 0x4e061d22 // mov v2.h[1], w9 + .long 0x39400108 // ldrb w8, [x8] + .long 0x4e021d02 // mov v2.h[0], w8 + .long 0x17ffffe3 // b 257c <sk_load_a8_aarch64+0x30> HIDDEN _sk_gather_a8_aarch64 .globl _sk_gather_a8_aarch64 FUNCTION(_sk_gather_a8_aarch64) _sk_gather_a8_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x4ea1b821 // fcvtzs v1.4s, v1.4s .long 0x4ea1b800 // fcvtzs v0.4s, v0.4s .long 0x91004109 // add x9, x8, #0x10 @@ -2659,7 +2831,7 @@ _sk_gather_a8_aarch64: .long 0x6f00e401 // movi v1.2d, #0x0 .long 0x6e23dc43 // fmul v3.4s, v2.4s, v3.4s .long 0x6f00e402 // movi v2.2d, #0x0 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_store_a8_aarch64 .globl _sk_store_a8_aarch64 @@ -2672,50 +2844,86 @@ _sk_store_a8_aarch64: .long 0xf9400108 // ldr x8, [x8] .long 0x6e21aa10 // fcvtnu v16.4s, v16.4s .long 0x0e612a10 // xtn v16.4h, v16.4s - .long 0x0e0e3e09 // umov w9, v16.h[3] .long 0x8b000108 // add x8, x8, x0 + .long 0xb5000183 // cbnz x3, 26bc <sk_store_a8_aarch64+0x50> + .long 0x0e0e3e09 // umov w9, v16.h[3] + .long 0x0e0a3e0a // umov w10, v16.h[2] + .long 0x0e063e0b // umov w11, v16.h[1] + .long 0x0e023e0c // umov w12, v16.h[0] .long 0x39000d09 // strb w9, [x8, #3] + .long 0x3900090a // strb w10, [x8, #2] + .long 0x3900050b // strb w11, [x8, #1] + .long 0x3900010c // strb w12, [x8] + .long 0xf9400424 // ldr x4, [x1, #8] + .long 0x91004021 // add x1, x1, #0x10 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x54000120 // b.eq 26e8 <sk_store_a8_aarch64+0x7c> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000a0 // b.eq 26e0 <sk_store_a8_aarch64+0x74> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffee1 // b.ne 26b0 <sk_store_a8_aarch64+0x44> // b.any .long 0x0e0a3e09 // umov w9, v16.h[2] .long 0x39000909 // strb w9, [x8, #2] .long 0x0e063e09 // umov w9, v16.h[1] .long 0x39000509 // strb w9, [x8, #1] .long 0x0e023e09 // umov w9, v16.h[0] .long 0x39000109 // strb w9, [x8] - .long 0xf9400423 // ldr x3, [x1, #8] - .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0x17fffff0 // b 26b0 <sk_store_a8_aarch64+0x44> HIDDEN _sk_load_g8_aarch64 .globl _sk_load_g8_aarch64 FUNCTION(_sk_load_g8_aarch64) _sk_load_g8_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 - .long 0x52a77009 // mov w9, #0x3b800000 - .long 0x72901029 // movk w9, #0x8081 - .long 0x4e040d20 // dup v0.4s, w9 + .long 0xf9400028 // ldr x8, [x1] .long 0xf9400108 // ldr x8, [x8] - .long 0x4f03f603 // fmov v3.4s, #1.000000000000000000e+00 .long 0x8b000108 // add x8, x8, x0 - .long 0x3940010a // ldrb w10, [x8] - .long 0x39400509 // ldrb w9, [x8, #1] + .long 0xb50002e3 // cbnz x3, 275c <sk_load_g8_aarch64+0x68> + .long 0x39400109 // ldrb w9, [x8] + .long 0x3940050a // ldrb w10, [x8, #1] .long 0x3940090b // ldrb w11, [x8, #2] .long 0x39400d08 // ldrb w8, [x8, #3] - .long 0x4e021d41 // mov v1.h[0], w10 - .long 0x4e061d21 // mov v1.h[1], w9 - .long 0x4e0a1d61 // mov v1.h[2], w11 - .long 0x4e0e1d01 // mov v1.h[3], w8 - .long 0x2f10a421 // uxtl v1.4s, v1.4h - .long 0x6e21d821 // ucvtf v1.4s, v1.4s - .long 0x6e20dc20 // fmul v0.4s, v1.4s, v0.4s + .long 0x4e021d20 // mov v0.h[0], w9 + .long 0x4e061d40 // mov v0.h[1], w10 + .long 0x4e0a1d60 // mov v0.h[2], w11 + .long 0x4e0e1d00 // mov v0.h[3], w8 + .long 0x2f07b7e0 // bic v0.4h, #0xff, lsl #8 + .long 0x52a77008 // mov w8, #0x3b800000 + .long 0x72901028 // movk w8, #0x8081 + .long 0xf9400424 // ldr x4, [x1, #8] + .long 0x2f10a400 // uxtl v0.4s, v0.4h + .long 0x4e040d01 // dup v1.4s, w8 + .long 0x6e21d800 // ucvtf v0.4s, v0.4s + .long 0x91004028 // add x8, x1, #0x10 + .long 0x6e21dc00 // fmul v0.4s, v0.4s, v1.4s + .long 0x4f03f603 // fmov v3.4s, #1.000000000000000000e+00 + .long 0xaa0803e1 // mov x1, x8 .long 0x4ea01c01 // mov v1.16b, v0.16b .long 0x4ea01c02 // mov v2.16b, v0.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x2f00e400 // movi d0, #0x0 + .long 0x54000140 // b.eq 2790 <sk_load_g8_aarch64+0x9c> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000c0 // b.eq 2788 <sk_load_g8_aarch64+0x94> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffd61 // b.ne 2724 <sk_load_g8_aarch64+0x30> // b.any + .long 0x39400909 // ldrb w9, [x8, #2] + .long 0x0e020fe0 // dup v0.4h, wzr + .long 0x4e0a1d20 // mov v0.h[2], w9 + .long 0x39400509 // ldrb w9, [x8, #1] + .long 0x4e061d20 // mov v0.h[1], w9 + .long 0x39400108 // ldrb w8, [x8] + .long 0x4e021d00 // mov v0.h[0], w8 + .long 0x17ffffe3 // b 2724 <sk_load_g8_aarch64+0x30> HIDDEN _sk_gather_g8_aarch64 .globl _sk_gather_g8_aarch64 FUNCTION(_sk_gather_g8_aarch64) _sk_gather_g8_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x4ea1b821 // fcvtzs v1.4s, v1.4s .long 0x4ea1b800 // fcvtzs v0.4s, v0.4s .long 0x91004109 // add x9, x8, #0x10 @@ -2744,7 +2952,7 @@ _sk_gather_g8_aarch64: .long 0x4f03f603 // fmov v3.4s, #1.000000000000000000e+00 .long 0x4ea01c01 // mov v1.16b, v0.16b .long 0x4ea01c02 // mov v2.16b, v0.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_gather_i8_aarch64 .globl _sk_gather_i8_aarch64 @@ -2752,9 +2960,9 @@ FUNCTION(_sk_gather_i8_aarch64) _sk_gather_i8_aarch64: .long 0xaa0103e8 // mov x8, x1 .long 0xf8408429 // ldr x9, [x1], #8 - .long 0xb4000069 // cbz x9, 24e8 <sk_gather_i8_aarch64+0x14> + .long 0xb4000069 // cbz x9, 2828 <sk_gather_i8_aarch64+0x14> .long 0xaa0903ea // mov x10, x9 - .long 0x14000003 // b 24f0 <sk_gather_i8_aarch64+0x1c> + .long 0x14000003 // b 2830 <sk_gather_i8_aarch64+0x1c> .long 0xf940050a // ldr x10, [x8, #8] .long 0x91004101 // add x1, x8, #0x10 .long 0xf8410548 // ldr x8, [x10], #16 @@ -2790,7 +2998,7 @@ _sk_gather_i8_aarch64: .long 0x0d409100 // ld1 {v0.s}[1], [x8] .long 0xb86b5928 // ldr w8, [x9, w11, uxtw #2] .long 0x7290102c // movk w12, #0x8081 - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4e141d40 // mov v0.s[2], w10 .long 0x4e1c1d00 // mov v0.s[3], w8 .long 0x4e231c01 // and v1.16b, v0.16b, v3.16b @@ -2808,47 +3016,66 @@ _sk_gather_i8_aarch64: .long 0x6e30dc21 // fmul v1.4s, v1.4s, v16.4s .long 0x6e30dc42 // fmul v2.4s, v2.4s, v16.4s .long 0x6e30de23 // fmul v3.4s, v17.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_load_565_aarch64 .globl _sk_load_565_aarch64 FUNCTION(_sk_load_565_aarch64) _sk_load_565_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 - .long 0xd37ff809 // lsl x9, x0, #1 - .long 0x4f072701 // movi v1.4s, #0xf8, lsl #8 - .long 0x4f0007e3 // movi v3.4s, #0x1f + .long 0xf9400028 // ldr x8, [x1] .long 0xf9400108 // ldr x8, [x8] - .long 0xfc696900 // ldr d0, [x8, x9] + .long 0x8b000508 // add x8, x8, x0, lsl #1 + .long 0xb50003c3 // cbnz x3, 2984 <sk_load_565_aarch64+0x84> + .long 0xfd400100 // ldr d0, [x8] .long 0x321b17e8 // orr w8, wzr, #0x7e0 .long 0x4e040d02 // dup v2.4s, w8 .long 0x52a6f088 // mov w8, #0x37840000 - .long 0x72842108 // movk w8, #0x2108 .long 0x2f10a400 // uxtl v0.4s, v0.4h + .long 0x4f072701 // movi v1.4s, #0xf8, lsl #8 + .long 0x72842108 // movk w8, #0x2108 + .long 0x4f0007e3 // movi v3.4s, #0x1f .long 0x4e211c01 // and v1.16b, v0.16b, v1.16b - .long 0x4e221c02 // and v2.16b, v0.16b, v2.16b .long 0x4e231c03 // and v3.16b, v0.16b, v3.16b + .long 0x4e221c10 // and v16.16b, v0.16b, v2.16b .long 0x4e040d00 // dup v0.4s, w8 - .long 0x52a74048 // mov w8, #0x3a020000 - .long 0x72810428 // movk w8, #0x821 + .long 0x52a7a088 // mov w8, #0x3d040000 + .long 0x72842108 // movk w8, #0x2108 .long 0x4e21d821 // scvtf v1.4s, v1.4s .long 0x6e20dc20 // fmul v0.4s, v1.4s, v0.4s .long 0x4e040d01 // dup v1.4s, w8 - .long 0x52a7a088 // mov w8, #0x3d040000 - .long 0x72842108 // movk w8, #0x2108 - .long 0x4e21d842 // scvtf v2.4s, v2.4s - .long 0x6e21dc41 // fmul v1.4s, v2.4s, v1.4s - .long 0x4e040d02 // dup v2.4s, w8 - .long 0x4e21d863 // scvtf v3.4s, v3.4s - .long 0x6e22dc62 // fmul v2.4s, v3.4s, v2.4s + .long 0x52a74048 // mov w8, #0x3a020000 + .long 0xf9400424 // ldr x4, [x1, #8] + .long 0x72810428 // movk w8, #0x821 + .long 0x4e21d862 // scvtf v2.4s, v3.4s + .long 0x6e21dc42 // fmul v2.4s, v2.4s, v1.4s + .long 0x4e040d01 // dup v1.4s, w8 + .long 0x91004028 // add x8, x1, #0x10 + .long 0x4e21da03 // scvtf v3.4s, v16.4s + .long 0x6e21dc61 // fmul v1.4s, v3.4s, v1.4s .long 0x4f03f603 // fmov v3.4s, #1.000000000000000000e+00 - .long 0xd61f0060 // br x3 + .long 0xaa0803e1 // mov x1, x8 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x2f00e400 // movi d0, #0x0 + .long 0x54000140 // b.eq 29b8 <sk_load_565_aarch64+0xb8> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000c0 // b.eq 29b0 <sk_load_565_aarch64+0xb0> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffba1 // b.ne 2914 <sk_load_565_aarch64+0x14> // b.any + .long 0x91001109 // add x9, x8, #0x4 + .long 0x0e020fe0 // dup v0.4h, wzr + .long 0x0d405120 // ld1 {v0.h}[2], [x9] + .long 0x91000909 // add x9, x8, #0x2 + .long 0x0d404920 // ld1 {v0.h}[1], [x9] + .long 0x0d404100 // ld1 {v0.h}[0], [x8] + .long 0x17ffffd6 // b 2914 <sk_load_565_aarch64+0x14> HIDDEN _sk_gather_565_aarch64 .globl _sk_gather_565_aarch64 FUNCTION(_sk_gather_565_aarch64) _sk_gather_565_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x4ea1b821 // fcvtzs v1.4s, v1.4s .long 0x4ea1b800 // fcvtzs v0.4s, v0.4s .long 0x91004109 // add x9, x8, #0x10 @@ -2891,49 +3118,64 @@ _sk_gather_565_aarch64: .long 0x4e21d842 // scvtf v2.4s, v2.4s .long 0x6e23dc42 // fmul v2.4s, v2.4s, v3.4s .long 0x4f03f603 // fmov v3.4s, #1.000000000000000000e+00 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_store_565_aarch64 .globl _sk_store_565_aarch64 FUNCTION(_sk_store_565_aarch64) _sk_store_565_aarch64: .long 0xf9400028 // ldr x8, [x1] - .long 0x52a84f8a // mov w10, #0x427c0000 + .long 0x52a84f89 // mov w9, #0x427c0000 .long 0x4f01f7f0 // fmov v16.4s, #3.100000000000000000e+01 - .long 0x4e040d52 // dup v18.4s, w10 + .long 0x4e040d32 // dup v18.4s, w9 .long 0x6e30dc11 // fmul v17.4s, v0.4s, v16.4s .long 0x6e32dc32 // fmul v18.4s, v1.4s, v18.4s .long 0x6e21aa31 // fcvtnu v17.4s, v17.4s .long 0x6e21aa52 // fcvtnu v18.4s, v18.4s + .long 0xf9400108 // ldr x8, [x8] .long 0x6e30dc50 // fmul v16.4s, v2.4s, v16.4s .long 0x4f2b5631 // shl v17.4s, v17.4s, #11 - .long 0xf9400108 // ldr x8, [x8] .long 0x4f255652 // shl v18.4s, v18.4s, #5 - .long 0x4eb11e51 // orr v17.16b, v18.16b, v17.16b .long 0x6e21aa10 // fcvtnu v16.4s, v16.4s + .long 0x4eb11e51 // orr v17.16b, v18.16b, v17.16b .long 0x4eb01e30 // orr v16.16b, v17.16b, v16.16b - .long 0xd37ff809 // lsl x9, x0, #1 + .long 0x8b000508 // add x8, x8, x0, lsl #1 .long 0x0e612a10 // xtn v16.4h, v16.4s - .long 0xfc296910 // str d16, [x8, x9] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xb50000a3 // cbnz x3, 2ac8 <sk_store_565_aarch64+0x58> + .long 0xfd000110 // str d16, [x8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x54000120 // b.eq 2af4 <sk_store_565_aarch64+0x84> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000a0 // b.eq 2aec <sk_store_565_aarch64+0x7c> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffee1 // b.ne 2abc <sk_store_565_aarch64+0x4c> // b.any + .long 0x91001109 // add x9, x8, #0x4 + .long 0x0d005130 // st1 {v16.h}[2], [x9] + .long 0x91000909 // add x9, x8, #0x2 + .long 0x0d004930 // st1 {v16.h}[1], [x9] + .long 0x0d004110 // st1 {v16.h}[0], [x8] + .long 0x17fffff1 // b 2abc <sk_store_565_aarch64+0x4c> HIDDEN _sk_load_4444_aarch64 .globl _sk_load_4444_aarch64 FUNCTION(_sk_load_4444_aarch64) _sk_load_4444_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 - .long 0xd37ff809 // lsl x9, x0, #1 + .long 0xf9400028 // ldr x8, [x1] + .long 0xf9400108 // ldr x8, [x8] + .long 0x8b000508 // add x8, x8, x0, lsl #1 + .long 0xb5000443 // cbnz x3, 2b90 <sk_load_4444_aarch64+0x94> + .long 0xfd400100 // ldr d0, [x8] + .long 0x52a6f108 // mov w8, #0x37880000 + .long 0x2f10a400 // uxtl v0.4s, v0.4h .long 0x4f072601 // movi v1.4s, #0xf0, lsl #8 + .long 0x72911128 // movk w8, #0x8889 .long 0x4f0025e2 // movi v2.4s, #0xf, lsl #8 - .long 0xf9400108 // ldr x8, [x8] .long 0x4f070603 // movi v3.4s, #0xf0 .long 0x4f0005f0 // movi v16.4s, #0xf - .long 0xfc696900 // ldr d0, [x8, x9] - .long 0x52a6f108 // mov w8, #0x37880000 - .long 0x72911128 // movk w8, #0x8889 - .long 0x2f10a400 // uxtl v0.4s, v0.4h .long 0x4e211c01 // and v1.16b, v0.16b, v1.16b .long 0x4e221c02 // and v2.16b, v0.16b, v2.16b .long 0x4e231c03 // and v3.16b, v0.16b, v3.16b @@ -2950,19 +3192,36 @@ _sk_load_4444_aarch64: .long 0x6e21dc41 // fmul v1.4s, v2.4s, v1.4s .long 0x4e040d02 // dup v2.4s, w8 .long 0x52a7b108 // mov w8, #0x3d880000 + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x72911128 // movk w8, #0x8889 .long 0x4e21d863 // scvtf v3.4s, v3.4s .long 0x6e22dc62 // fmul v2.4s, v3.4s, v2.4s .long 0x4e040d03 // dup v3.4s, w8 .long 0x4e21da10 // scvtf v16.4s, v16.4s .long 0x6e23de03 // fmul v3.4s, v16.4s, v3.4s - .long 0xd61f0060 // br x3 + .long 0x91004021 // add x1, x1, #0x10 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x2f00e400 // movi d0, #0x0 + .long 0x54000140 // b.eq 2bc4 <sk_load_4444_aarch64+0xc8> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000c0 // b.eq 2bbc <sk_load_4444_aarch64+0xc0> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffb21 // b.ne 2b10 <sk_load_4444_aarch64+0x14> // b.any + .long 0x91001109 // add x9, x8, #0x4 + .long 0x0e020fe0 // dup v0.4h, wzr + .long 0x0d405120 // ld1 {v0.h}[2], [x9] + .long 0x91000909 // add x9, x8, #0x2 + .long 0x0d404920 // ld1 {v0.h}[1], [x9] + .long 0x0d404100 // ld1 {v0.h}[0], [x8] + .long 0x17ffffd2 // b 2b10 <sk_load_4444_aarch64+0x14> HIDDEN _sk_gather_4444_aarch64 .globl _sk_gather_4444_aarch64 FUNCTION(_sk_gather_4444_aarch64) _sk_gather_4444_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x4ea1b821 // fcvtzs v1.4s, v1.4s .long 0x4ea1b800 // fcvtzs v0.4s, v0.4s .long 0x4f070603 // movi v3.4s, #0xf0 @@ -3010,16 +3269,16 @@ _sk_gather_4444_aarch64: .long 0x4e040d03 // dup v3.4s, w8 .long 0x4e21da10 // scvtf v16.4s, v16.4s .long 0x6e23de03 // fmul v3.4s, v16.4s, v3.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_store_4444_aarch64 .globl _sk_store_4444_aarch64 FUNCTION(_sk_store_4444_aarch64) _sk_store_4444_aarch64: .long 0x4f01f5d0 // fmov v16.4s, #1.500000000000000000e+01 + .long 0xf9400028 // ldr x8, [x1] .long 0x6e30dc11 // fmul v17.4s, v0.4s, v16.4s .long 0x6e30dc32 // fmul v18.4s, v1.4s, v16.4s - .long 0xf9400028 // ldr x8, [x1] .long 0x6e21aa31 // fcvtnu v17.4s, v17.4s .long 0x6e21aa52 // fcvtnu v18.4s, v18.4s .long 0x4f2c5631 // shl v17.4s, v17.4s, #12 @@ -3027,52 +3286,84 @@ _sk_store_4444_aarch64: .long 0x4eb11e51 // orr v17.16b, v18.16b, v17.16b .long 0x6e30dc52 // fmul v18.4s, v2.4s, v16.4s .long 0x6e21aa52 // fcvtnu v18.4s, v18.4s + .long 0xf9400108 // ldr x8, [x8] .long 0x6e30dc70 // fmul v16.4s, v3.4s, v16.4s .long 0x4f245652 // shl v18.4s, v18.4s, #4 - .long 0xf9400108 // ldr x8, [x8] - .long 0x4eb21e31 // orr v17.16b, v17.16b, v18.16b .long 0x6e21aa10 // fcvtnu v16.4s, v16.4s + .long 0x4eb21e31 // orr v17.16b, v17.16b, v18.16b .long 0x4eb01e30 // orr v16.16b, v17.16b, v16.16b - .long 0xd37ff809 // lsl x9, x0, #1 + .long 0x8b000508 // add x8, x8, x0, lsl #1 .long 0x0e612a10 // xtn v16.4h, v16.4s - .long 0xfc296910 // str d16, [x8, x9] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xb50000a3 // cbnz x3, 2cf0 <sk_store_4444_aarch64+0x60> + .long 0xfd000110 // str d16, [x8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x54000120 // b.eq 2d1c <sk_store_4444_aarch64+0x8c> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000a0 // b.eq 2d14 <sk_store_4444_aarch64+0x84> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffee1 // b.ne 2ce4 <sk_store_4444_aarch64+0x54> // b.any + .long 0x91001109 // add x9, x8, #0x4 + .long 0x0d005130 // st1 {v16.h}[2], [x9] + .long 0x91000909 // add x9, x8, #0x2 + .long 0x0d004930 // st1 {v16.h}[1], [x9] + .long 0x0d004110 // st1 {v16.h}[0], [x8] + .long 0x17fffff1 // b 2ce4 <sk_store_4444_aarch64+0x54> HIDDEN _sk_load_8888_aarch64 .globl _sk_load_8888_aarch64 FUNCTION(_sk_load_8888_aarch64) _sk_load_8888_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 - .long 0xd37ef409 // lsl x9, x0, #2 - .long 0x6f00e621 // movi v1.2d, #0xff000000ff + .long 0xf9400028 // ldr x8, [x1] .long 0xf9400108 // ldr x8, [x8] - .long 0x3ce96900 // ldr q0, [x8, x9] + .long 0x8b000908 // add x8, x8, x0, lsl #2 + .long 0xb50002e3 // cbnz x3, 2d8c <sk_load_8888_aarch64+0x68> + .long 0x3dc00100 // ldr q0, [x8] + .long 0x6f00e621 // movi v1.2d, #0xff000000ff .long 0x52a77008 // mov w8, #0x3b800000 + .long 0x6f380402 // ushr v2.4s, v0.4s, #8 + .long 0x6f300403 // ushr v3.4s, v0.4s, #16 + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x72901028 // movk w8, #0x8081 - .long 0x4e040d02 // dup v2.4s, w8 - .long 0x6f380410 // ushr v16.4s, v0.4s, #8 - .long 0x6f300411 // ushr v17.4s, v0.4s, #16 - .long 0x4e211c03 // and v3.16b, v0.16b, v1.16b - .long 0x6f280400 // ushr v0.4s, v0.4s, #24 - .long 0x4e211e10 // and v16.16b, v16.16b, v1.16b - .long 0x4e211e21 // and v1.16b, v17.16b, v1.16b - .long 0x4e21d863 // scvtf v3.4s, v3.4s - .long 0x4e21d811 // scvtf v17.4s, v0.4s - .long 0x4e21da10 // scvtf v16.4s, v16.4s - .long 0x4e21d832 // scvtf v18.4s, v1.4s - .long 0x6e22dc60 // fmul v0.4s, v3.4s, v2.4s - .long 0x6e22de23 // fmul v3.4s, v17.4s, v2.4s - .long 0x6e22de01 // fmul v1.4s, v16.4s, v2.4s - .long 0x6e22de42 // fmul v2.4s, v18.4s, v2.4s - .long 0xd61f0060 // br x3 + .long 0x6f280410 // ushr v16.4s, v0.4s, #24 + .long 0x4e211c00 // and v0.16b, v0.16b, v1.16b + .long 0x4e211c42 // and v2.16b, v2.16b, v1.16b + .long 0x4e211c61 // and v1.16b, v3.16b, v1.16b + .long 0x4e040d11 // dup v17.4s, w8 + .long 0x4e21da03 // scvtf v3.4s, v16.4s + .long 0x4e21d800 // scvtf v0.4s, v0.4s + .long 0x4e21d842 // scvtf v2.4s, v2.4s + .long 0x4e21d830 // scvtf v16.4s, v1.4s + .long 0x6e31dc63 // fmul v3.4s, v3.4s, v17.4s + .long 0x6e31dc00 // fmul v0.4s, v0.4s, v17.4s + .long 0x6e31dc41 // fmul v1.4s, v2.4s, v17.4s + .long 0x6e31de02 // fmul v2.4s, v16.4s, v17.4s + .long 0x91004021 // add x1, x1, #0x10 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x6f00e400 // movi v0.2d, #0x0 + .long 0x54000140 // b.eq 2dc0 <sk_load_8888_aarch64+0x9c> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000c0 // b.eq 2db8 <sk_load_8888_aarch64+0x94> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffc81 // b.ne 2d38 <sk_load_8888_aarch64+0x14> // b.any + .long 0x91002109 // add x9, x8, #0x8 + .long 0x4e040fe0 // dup v0.4s, wzr + .long 0x4d408120 // ld1 {v0.s}[2], [x9] + .long 0x91001109 // add x9, x8, #0x4 + .long 0x0d409120 // ld1 {v0.s}[1], [x9] + .long 0x0d408100 // ld1 {v0.s}[0], [x8] + .long 0x17ffffdd // b 2d38 <sk_load_8888_aarch64+0x14> HIDDEN _sk_gather_8888_aarch64 .globl _sk_gather_8888_aarch64 FUNCTION(_sk_gather_8888_aarch64) _sk_gather_8888_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x4ea1b821 // fcvtzs v1.4s, v1.4s .long 0x4ea1b800 // fcvtzs v0.4s, v0.4s .long 0x91004109 // add x9, x8, #0x10 @@ -3109,15 +3400,15 @@ _sk_gather_8888_aarch64: .long 0x6e22de23 // fmul v3.4s, v17.4s, v2.4s .long 0x6e22de01 // fmul v1.4s, v16.4s, v2.4s .long 0x6e22de42 // fmul v2.4s, v18.4s, v2.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_store_8888_aarch64 .globl _sk_store_8888_aarch64 FUNCTION(_sk_store_8888_aarch64) _sk_store_8888_aarch64: - .long 0x52a86fea // mov w10, #0x437f0000 - .long 0x4e040d50 // dup v16.4s, w10 + .long 0x52a86fe9 // mov w9, #0x437f0000 .long 0xf9400028 // ldr x8, [x1] + .long 0x4e040d30 // dup v16.4s, w9 .long 0x6e30dc32 // fmul v18.4s, v1.4s, v16.4s .long 0x6e30dc11 // fmul v17.4s, v0.4s, v16.4s .long 0x6e21aa52 // fcvtnu v18.4s, v18.4s @@ -3125,33 +3416,60 @@ _sk_store_8888_aarch64: .long 0x4f285652 // shl v18.4s, v18.4s, #8 .long 0x4eb11e51 // orr v17.16b, v18.16b, v17.16b .long 0x6e30dc52 // fmul v18.4s, v2.4s, v16.4s + .long 0xf9400108 // ldr x8, [x8] .long 0x6e30dc70 // fmul v16.4s, v3.4s, v16.4s .long 0x6e21aa52 // fcvtnu v18.4s, v18.4s - .long 0xf9400108 // ldr x8, [x8] .long 0x6e21aa10 // fcvtnu v16.4s, v16.4s .long 0x4f305652 // shl v18.4s, v18.4s, #16 - .long 0x4eb21e31 // orr v17.16b, v17.16b, v18.16b .long 0x4f385610 // shl v16.4s, v16.4s, #24 - .long 0xd37ef409 // lsl x9, x0, #2 + .long 0x4eb21e31 // orr v17.16b, v17.16b, v18.16b + .long 0x8b000908 // add x8, x8, x0, lsl #2 .long 0x4eb01e30 // orr v16.16b, v17.16b, v16.16b - .long 0x3ca96910 // str q16, [x8, x9] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xb50000a3 // cbnz x3, 2ec0 <sk_store_8888_aarch64+0x60> + .long 0x3d800110 // str q16, [x8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 + .long 0x12000469 // and w9, w3, #0x3 + .long 0x7100053f // cmp w9, #0x1 + .long 0x54000120 // b.eq 2eec <sk_store_8888_aarch64+0x8c> // b.none + .long 0x7100093f // cmp w9, #0x2 + .long 0x540000a0 // b.eq 2ee4 <sk_store_8888_aarch64+0x84> // b.none + .long 0x71000d3f // cmp w9, #0x3 + .long 0x54fffee1 // b.ne 2eb4 <sk_store_8888_aarch64+0x54> // b.any + .long 0x91002109 // add x9, x8, #0x8 + .long 0x4d008130 // st1 {v16.s}[2], [x9] + .long 0x91001109 // add x9, x8, #0x4 + .long 0x0d009130 // st1 {v16.s}[1], [x9] + .long 0x0d008110 // st1 {v16.s}[0], [x8] + .long 0x17fffff1 // b 2eb4 <sk_store_8888_aarch64+0x54> HIDDEN _sk_load_f16_aarch64 .globl _sk_load_f16_aarch64 FUNCTION(_sk_load_f16_aarch64) _sk_load_f16_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xf9400028 // ldr x8, [x1] .long 0xf9400108 // ldr x8, [x8] .long 0x8b000d08 // add x8, x8, x0, lsl #3 + .long 0xb5000123 // cbnz x3, 2f24 <sk_load_f16_aarch64+0x30> .long 0x0c400510 // ld4 {v16.4h-v19.4h}, [x8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x0e217a00 // fcvtl v0.4s, v16.4h .long 0x0e217a21 // fcvtl v1.4s, v17.4h .long 0x0e217a42 // fcvtl v2.4s, v18.4h .long 0x0e217a63 // fcvtl v3.4s, v19.4h - .long 0xd61f0060 // br x3 + .long 0x91004021 // add x1, x1, #0x10 + .long 0xd61f0080 // br x4 + .long 0x0d606110 // ld4 {v16.h-v19.h}[0], [x8] + .long 0xf100047f // cmp x3, #0x1 + .long 0x54fffee0 // b.eq 2f08 <sk_load_f16_aarch64+0x14> // b.none + .long 0x91002109 // add x9, x8, #0x8 + .long 0x0d606930 // ld4 {v16.h-v19.h}[1], [x9] + .long 0xf1000c7f // cmp x3, #0x3 + .long 0x54fffe63 // b.cc 2f08 <sk_load_f16_aarch64+0x14> // b.lo, b.ul, b.last + .long 0x91004108 // add x8, x8, #0x10 + .long 0x0d607110 // ld4 {v16.h-v19.h}[2], [x8] + .long 0x17fffff0 // b 2f08 <sk_load_f16_aarch64+0x14> HIDDEN _sk_gather_f16_aarch64 .globl _sk_gather_f16_aarch64 @@ -3205,21 +3523,31 @@ _sk_store_f16_aarch64: .long 0xf9400108 // ldr x8, [x8] .long 0x0e216873 // fcvtn v19.4h, v3.4s .long 0x8b000d08 // add x8, x8, x0, lsl #3 + .long 0xb50000a3 // cbnz x3, 300c <sk_store_f16_aarch64+0x30> .long 0x0c000510 // st4 {v16.4h-v19.4h}, [x8] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 + .long 0xf100047f // cmp x3, #0x1 + .long 0x0d206110 // st4 {v16.h-v19.h}[0], [x8] + .long 0x54ffff60 // b.eq 3000 <sk_store_f16_aarch64+0x24> // b.none + .long 0x91002109 // add x9, x8, #0x8 + .long 0xf1000c7f // cmp x3, #0x3 + .long 0x0d206930 // st4 {v16.h-v19.h}[1], [x9] + .long 0x54fffee3 // b.cc 3000 <sk_store_f16_aarch64+0x24> // b.lo, b.ul, b.last + .long 0x91004108 // add x8, x8, #0x10 + .long 0x0d207110 // st4 {v16.h-v19.h}[2], [x8] + .long 0x17fffff4 // b 3000 <sk_store_f16_aarch64+0x24> HIDDEN _sk_load_u16_be_aarch64 .globl _sk_load_u16_be_aarch64 FUNCTION(_sk_load_u16_be_aarch64) _sk_load_u16_be_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xf9400028 // ldr x8, [x1] .long 0xf9400108 // ldr x8, [x8] .long 0x8b000d08 // add x8, x8, x0, lsl #3 + .long 0xb5000403 // cbnz x3, 30c0 <sk_load_u16_be_aarch64+0x8c> .long 0x0c400500 // ld4 {v0.4h-v3.4h}, [x8] - .long 0x52a6f008 // mov w8, #0x37800000 - .long 0x72801008 // movk w8, #0x80 .long 0x0f185410 // shl v16.4h, v0.4h, #8 .long 0x2f180411 // ushr v17.4h, v0.4h, #8 .long 0x0f185432 // shl v18.4h, v1.4h, #8 @@ -3228,10 +3556,13 @@ _sk_load_u16_be_aarch64: .long 0x2f180455 // ushr v21.4h, v2.4h, #8 .long 0x0f185476 // shl v22.4h, v3.4h, #8 .long 0x2f180460 // ushr v0.4h, v3.4h, #8 + .long 0x52a6f008 // mov w8, #0x37800000 + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x0eb11e01 // orr v1.8b, v16.8b, v17.8b .long 0x0eb31e42 // orr v2.8b, v18.8b, v19.8b .long 0x0eb51e90 // orr v16.8b, v20.8b, v21.8b .long 0x0ea01ec0 // orr v0.8b, v22.8b, v0.8b + .long 0x72801008 // movk w8, #0x80 .long 0x2f10a421 // uxtl v1.4s, v1.4h .long 0x2f10a442 // uxtl v2.4s, v2.4h .long 0x2f10a610 // uxtl v16.4s, v16.4h @@ -3245,32 +3576,46 @@ _sk_load_u16_be_aarch64: .long 0x6e23dc41 // fmul v1.4s, v2.4s, v3.4s .long 0x6e23de02 // fmul v2.4s, v16.4s, v3.4s .long 0x6e23de23 // fmul v3.4s, v17.4s, v3.4s - .long 0xd61f0060 // br x3 + .long 0x91004021 // add x1, x1, #0x10 + .long 0xd61f0080 // br x4 + .long 0x0d606100 // ld4 {v0.h-v3.h}[0], [x8] + .long 0xf100047f // cmp x3, #0x1 + .long 0x54fffc00 // b.eq 3048 <sk_load_u16_be_aarch64+0x14> // b.none + .long 0x91002109 // add x9, x8, #0x8 + .long 0x0d606920 // ld4 {v0.h-v3.h}[1], [x9] + .long 0xf1000c7f // cmp x3, #0x3 + .long 0x54fffb83 // b.cc 3048 <sk_load_u16_be_aarch64+0x14> // b.lo, b.ul, b.last + .long 0x91004108 // add x8, x8, #0x10 + .long 0x0d607100 // ld4 {v0.h-v3.h}[2], [x8] + .long 0x17ffffd9 // b 3048 <sk_load_u16_be_aarch64+0x14> HIDDEN _sk_load_rgb_u16_be_aarch64 .globl _sk_load_rgb_u16_be_aarch64 FUNCTION(_sk_load_rgb_u16_be_aarch64) _sk_load_rgb_u16_be_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xf9400028 // ldr x8, [x1] .long 0x321f07e9 // orr w9, wzr, #0x6 .long 0xf9400108 // ldr x8, [x8] .long 0x9b092008 // madd x8, x0, x9, x8 + .long 0xb5000383 // cbnz x3, 3168 <sk_load_rgb_u16_be_aarch64+0x80> .long 0x0c404500 // ld3 {v0.4h-v2.4h}, [x8] - .long 0x52a6f008 // mov w8, #0x37800000 - .long 0x72801008 // movk w8, #0x80 .long 0x0f185403 // shl v3.4h, v0.4h, #8 .long 0x2f180410 // ushr v16.4h, v0.4h, #8 .long 0x0f185431 // shl v17.4h, v1.4h, #8 .long 0x2f180432 // ushr v18.4h, v1.4h, #8 .long 0x0f185453 // shl v19.4h, v2.4h, #8 .long 0x2f180440 // ushr v0.4h, v2.4h, #8 + .long 0x52a6f008 // mov w8, #0x37800000 + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x0eb01c61 // orr v1.8b, v3.8b, v16.8b .long 0x0eb21e23 // orr v3.8b, v17.8b, v18.8b .long 0x0ea01e60 // orr v0.8b, v19.8b, v0.8b + .long 0x72801008 // movk w8, #0x80 .long 0x2f10a421 // uxtl v1.4s, v1.4h .long 0x2f10a463 // uxtl v3.4s, v3.4h .long 0x2f10a400 // uxtl v0.4s, v0.4h .long 0x4e040d02 // dup v2.4s, w8 + .long 0x91004028 // add x8, x1, #0x10 .long 0x6e21d821 // ucvtf v1.4s, v1.4s .long 0x6e21d863 // ucvtf v3.4s, v3.4s .long 0x6e21d810 // ucvtf v16.4s, v0.4s @@ -3278,7 +3623,18 @@ _sk_load_rgb_u16_be_aarch64: .long 0x6e22dc61 // fmul v1.4s, v3.4s, v2.4s .long 0x6e22de02 // fmul v2.4s, v16.4s, v2.4s .long 0x4f03f603 // fmov v3.4s, #1.000000000000000000e+00 - .long 0xd61f0060 // br x3 + .long 0xaa0803e1 // mov x1, x8 + .long 0xd61f0080 // br x4 + .long 0x0d406100 // ld3 {v0.h-v2.h}[0], [x8] + .long 0xf100047f // cmp x3, #0x1 + .long 0x54fffc80 // b.eq 3100 <sk_load_rgb_u16_be_aarch64+0x18> // b.none + .long 0x91001909 // add x9, x8, #0x6 + .long 0x0d406920 // ld3 {v0.h-v2.h}[1], [x9] + .long 0xf1000c7f // cmp x3, #0x3 + .long 0x54fffc03 // b.cc 3100 <sk_load_rgb_u16_be_aarch64+0x18> // b.lo, b.ul, b.last + .long 0x91003108 // add x8, x8, #0xc + .long 0x0d407100 // ld3 {v0.h-v2.h}[2], [x8] + .long 0x17ffffdd // b 3100 <sk_load_rgb_u16_be_aarch64+0x18> HIDDEN _sk_store_u16_be_aarch64 .globl _sk_store_u16_be_aarch64 @@ -3286,48 +3642,72 @@ FUNCTION(_sk_store_u16_be_aarch64) _sk_store_u16_be_aarch64: .long 0x52a8efe9 // mov w9, #0x477f0000 .long 0x729fe009 // movk w9, #0xff00 - .long 0x4e040d30 // dup v16.4s, w9 - .long 0x6e30dc11 // fmul v17.4s, v0.4s, v16.4s + .long 0x4e040d34 // dup v20.4s, w9 + .long 0x6e34dc10 // fmul v16.4s, v0.4s, v20.4s + .long 0x6e34dc31 // fmul v17.4s, v1.4s, v20.4s + .long 0x6e21aa10 // fcvtnu v16.4s, v16.4s .long 0xf9400028 // ldr x8, [x1] .long 0x6e21aa31 // fcvtnu v17.4s, v17.4s - .long 0x0e612a31 // xtn v17.4h, v17.4s - .long 0x6e30dc32 // fmul v18.4s, v1.4s, v16.4s - .long 0x0f185633 // shl v19.4h, v17.4h, #8 - .long 0x2f180631 // ushr v17.4h, v17.4h, #8 - .long 0x6e21aa52 // fcvtnu v18.4s, v18.4s - .long 0x0eb11e75 // orr v21.8b, v19.8b, v17.8b - .long 0x6e30dc51 // fmul v17.4s, v2.4s, v16.4s - .long 0x0e612a52 // xtn v18.4h, v18.4s - .long 0x6e30dc70 // fmul v16.4s, v3.4s, v16.4s - .long 0x6e21aa31 // fcvtnu v17.4s, v17.4s - .long 0xf9400108 // ldr x8, [x8] - .long 0x0f185654 // shl v20.4h, v18.4h, #8 - .long 0x2f180652 // ushr v18.4h, v18.4h, #8 - .long 0x6e21aa10 // fcvtnu v16.4s, v16.4s - .long 0x0e612a31 // xtn v17.4h, v17.4s - .long 0x0eb21e96 // orr v22.8b, v20.8b, v18.8b .long 0x0e612a10 // xtn v16.4h, v16.4s - .long 0x0f185632 // shl v18.4h, v17.4h, #8 - .long 0x2f180631 // ushr v17.4h, v17.4h, #8 - .long 0x0eb11e57 // orr v23.8b, v18.8b, v17.8b - .long 0x0f185611 // shl v17.4h, v16.4h, #8 + .long 0x0e612a31 // xtn v17.4h, v17.4s + .long 0x0f185612 // shl v18.4h, v16.4h, #8 .long 0x2f180610 // ushr v16.4h, v16.4h, #8 + .long 0x0f185635 // shl v21.4h, v17.4h, #8 + .long 0x2f180636 // ushr v22.4h, v17.4h, #8 + .long 0x0eb01e50 // orr v16.8b, v18.8b, v16.8b + .long 0x0eb61eb1 // orr v17.8b, v21.8b, v22.8b + .long 0x6e34dc55 // fmul v21.4s, v2.4s, v20.4s + .long 0x6e34dc74 // fmul v20.4s, v3.4s, v20.4s + .long 0x6e21aab5 // fcvtnu v21.4s, v21.4s + .long 0xf9400108 // ldr x8, [x8] + .long 0x6e21aa94 // fcvtnu v20.4s, v20.4s + .long 0x0e612ab5 // xtn v21.4h, v21.4s + .long 0x0e612a94 // xtn v20.4h, v20.4s + .long 0x0f1856b6 // shl v22.4h, v21.4h, #8 + .long 0x2f1806b5 // ushr v21.4h, v21.4h, #8 + .long 0x0eb51ed2 // orr v18.8b, v22.8b, v21.8b + .long 0x0f185695 // shl v21.4h, v20.4h, #8 + .long 0x2f180694 // ushr v20.4h, v20.4h, #8 .long 0x8b000d08 // add x8, x8, x0, lsl #3 - .long 0x0eb01e38 // orr v24.8b, v17.8b, v16.8b - .long 0x0c000515 // st4 {v21.4h-v24.4h}, [x8] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0x0eb41eb3 // orr v19.8b, v21.8b, v20.8b + .long 0xb50000a3 // cbnz x3, 321c <sk_store_u16_be_aarch64+0x8c> + .long 0x0c000510 // st4 {v16.4h-v19.4h}, [x8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 + .long 0xf100047f // cmp x3, #0x1 + .long 0x0d206110 // st4 {v16.h-v19.h}[0], [x8] + .long 0x54ffff60 // b.eq 3210 <sk_store_u16_be_aarch64+0x80> // b.none + .long 0x91002109 // add x9, x8, #0x8 + .long 0xf1000c7f // cmp x3, #0x3 + .long 0x0d206930 // st4 {v16.h-v19.h}[1], [x9] + .long 0x54fffee3 // b.cc 3210 <sk_store_u16_be_aarch64+0x80> // b.lo, b.ul, b.last + .long 0x91004108 // add x8, x8, #0x10 + .long 0x0d207110 // st4 {v16.h-v19.h}[2], [x8] + .long 0x17fffff4 // b 3210 <sk_store_u16_be_aarch64+0x80> HIDDEN _sk_load_f32_aarch64 .globl _sk_load_f32_aarch64 FUNCTION(_sk_load_f32_aarch64) _sk_load_f32_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xf9400028 // ldr x8, [x1] .long 0xf9400108 // ldr x8, [x8] .long 0x8b001108 // add x8, x8, x0, lsl #4 + .long 0xb50000a3 // cbnz x3, 3264 <sk_load_f32_aarch64+0x20> .long 0x4c400900 // ld4 {v0.4s-v3.4s}, [x8] - .long 0xd61f0060 // br x3 + .long 0xf9400424 // ldr x4, [x1, #8] + .long 0x91004021 // add x1, x1, #0x10 + .long 0xd61f0080 // br x4 + .long 0x0d60a100 // ld4 {v0.s-v3.s}[0], [x8] + .long 0xf100047f // cmp x3, #0x1 + .long 0x54ffff60 // b.eq 3258 <sk_load_f32_aarch64+0x14> // b.none + .long 0x91004109 // add x9, x8, #0x10 + .long 0x0d60b120 // ld4 {v0.s-v3.s}[1], [x9] + .long 0xf1000c7f // cmp x3, #0x3 + .long 0x54fffee3 // b.cc 3258 <sk_load_f32_aarch64+0x14> // b.lo, b.ul, b.last + .long 0x91008108 // add x8, x8, #0x20 + .long 0x4d60a100 // ld4 {v0.s-v3.s}[2], [x8] + .long 0x17fffff4 // b 3258 <sk_load_f32_aarch64+0x14> HIDDEN _sk_store_f32_aarch64 .globl _sk_store_f32_aarch64 @@ -3336,62 +3716,73 @@ _sk_store_f32_aarch64: .long 0xf9400028 // ldr x8, [x1] .long 0xf9400108 // ldr x8, [x8] .long 0x8b001108 // add x8, x8, x0, lsl #4 + .long 0xb50000a3 // cbnz x3, 32ac <sk_store_f32_aarch64+0x20> .long 0x4c000900 // st4 {v0.4s-v3.4s}, [x8] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 + .long 0xf100047f // cmp x3, #0x1 + .long 0x0d20a100 // st4 {v0.s-v3.s}[0], [x8] + .long 0x54ffff60 // b.eq 32a0 <sk_store_f32_aarch64+0x14> // b.none + .long 0x91004109 // add x9, x8, #0x10 + .long 0xf1000c7f // cmp x3, #0x3 + .long 0x0d20b120 // st4 {v0.s-v3.s}[1], [x9] + .long 0x54fffee3 // b.cc 32a0 <sk_store_f32_aarch64+0x14> // b.lo, b.ul, b.last + .long 0x91008108 // add x8, x8, #0x20 + .long 0x4d20a100 // st4 {v0.s-v3.s}[2], [x8] + .long 0x17fffff4 // b 32a0 <sk_store_f32_aarch64+0x14> HIDDEN _sk_clamp_x_aarch64 .globl _sk_clamp_x_aarch64 FUNCTION(_sk_clamp_x_aarch64) _sk_clamp_x_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x6f00e411 // movi v17.2d, #0x0 .long 0x4e20f620 // fmax v0.4s, v17.4s, v0.4s .long 0x4d40c910 // ld1r {v16.4s}, [x8] .long 0x4eb0f400 // fmin v0.4s, v0.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_clamp_y_aarch64 .globl _sk_clamp_y_aarch64 FUNCTION(_sk_clamp_y_aarch64) _sk_clamp_y_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x6f00e411 // movi v17.2d, #0x0 .long 0x4e21f621 // fmax v1.4s, v17.4s, v1.4s .long 0x4d40c910 // ld1r {v16.4s}, [x8] .long 0x4eb0f421 // fmin v1.4s, v1.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_repeat_x_aarch64 .globl _sk_repeat_x_aarch64 FUNCTION(_sk_repeat_x_aarch64) _sk_repeat_x_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xbd400110 // ldr s16, [x8] .long 0x4e040611 // dup v17.4s, v16.s[0] .long 0x6e31fc11 // fdiv v17.4s, v0.4s, v17.4s .long 0x4e219a31 // frintm v17.4s, v17.4s .long 0x4f905220 // fmls v0.4s, v17.4s, v16.s[0] - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_repeat_y_aarch64 .globl _sk_repeat_y_aarch64 FUNCTION(_sk_repeat_y_aarch64) _sk_repeat_y_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xbd400110 // ldr s16, [x8] .long 0x4e040611 // dup v17.4s, v16.s[0] .long 0x6e31fc31 // fdiv v17.4s, v1.4s, v17.4s .long 0x4e219a31 // frintm v17.4s, v17.4s .long 0x4f905221 // fmls v1.4s, v17.4s, v16.s[0] - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_mirror_x_aarch64 .globl _sk_mirror_x_aarch64 FUNCTION(_sk_mirror_x_aarch64) _sk_mirror_x_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xbd400110 // ldr s16, [x8] .long 0x4e040611 // dup v17.4s, v16.s[0] .long 0x1e302a10 // fadd s16, s16, s16 @@ -3402,13 +3793,13 @@ _sk_mirror_x_aarch64: .long 0x4f905240 // fmls v0.4s, v18.4s, v16.s[0] .long 0x4eb1d400 // fsub v0.4s, v0.4s, v17.4s .long 0x4ea0f800 // fabs v0.4s, v0.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_mirror_y_aarch64 .globl _sk_mirror_y_aarch64 FUNCTION(_sk_mirror_y_aarch64) _sk_mirror_y_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xbd400110 // ldr s16, [x8] .long 0x4e040611 // dup v17.4s, v16.s[0] .long 0x1e302a10 // fadd s16, s16, s16 @@ -3419,27 +3810,27 @@ _sk_mirror_y_aarch64: .long 0x4f905241 // fmls v1.4s, v18.4s, v16.s[0] .long 0x4eb1d421 // fsub v1.4s, v1.4s, v17.4s .long 0x4ea0f821 // fabs v1.4s, v1.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_clamp_x_1_aarch64 .globl _sk_clamp_x_1_aarch64 FUNCTION(_sk_clamp_x_1_aarch64) _sk_clamp_x_1_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6f00e410 // movi v16.2d, #0x0 .long 0x4e20f600 // fmax v0.4s, v16.4s, v0.4s .long 0x4f03f610 // fmov v16.4s, #1.000000000000000000e+00 .long 0x4eb0f400 // fmin v0.4s, v0.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_repeat_x_1_aarch64 .globl _sk_repeat_x_1_aarch64 FUNCTION(_sk_repeat_x_1_aarch64) _sk_repeat_x_1_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4e219810 // frintm v16.4s, v0.4s .long 0x4eb0d400 // fsub v0.4s, v0.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_mirror_x_1_aarch64 .globl _sk_mirror_x_1_aarch64 @@ -3451,11 +3842,11 @@ _sk_mirror_x_1_aarch64: .long 0x6e31dc11 // fmul v17.4s, v0.4s, v17.4s .long 0x4e219a31 // frintm v17.4s, v17.4s .long 0x4e31d631 // fadd v17.4s, v17.4s, v17.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4eb1d400 // fsub v0.4s, v0.4s, v17.4s .long 0x4e30d400 // fadd v0.4s, v0.4s, v16.4s .long 0x4ea0f800 // fabs v0.4s, v0.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_luminance_to_alpha_aarch64 .globl _sk_luminance_to_alpha_aarch64 @@ -3469,7 +3860,7 @@ _sk_luminance_to_alpha_aarch64: .long 0x4ea01c10 // mov v16.16b, v0.16b .long 0x4e040d00 // dup v0.4s, w8 .long 0x52a7b268 // mov w8, #0x3d930000 - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x729bb308 // movk w8, #0xdd98 .long 0x6e20dc23 // fmul v3.4s, v1.4s, v0.4s .long 0x4e30ce23 // fmla v3.4s, v17.4s, v16.4s @@ -3478,13 +3869,13 @@ _sk_luminance_to_alpha_aarch64: .long 0x6f00e401 // movi v1.2d, #0x0 .long 0x4e22ce03 // fmla v3.4s, v16.4s, v2.4s .long 0x6f00e402 // movi v2.2d, #0x0 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_matrix_2x3_aarch64 .globl _sk_matrix_2x3_aarch64 FUNCTION(_sk_matrix_2x3_aarch64) _sk_matrix_2x3_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xaa0803e9 // mov x9, x8 .long 0x9100410a // add x10, x8, #0x10 .long 0x4ddfc932 // ld1r {v18.4s}, [x9], #4 @@ -3499,13 +3890,13 @@ _sk_matrix_2x3_aarch64: .long 0x4f931011 // fmla v17.4s, v0.4s, v19.s[0] .long 0x4eb01e00 // mov v0.16b, v16.16b .long 0x4eb11e21 // mov v1.16b, v17.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_matrix_3x4_aarch64 .globl _sk_matrix_3x4_aarch64 FUNCTION(_sk_matrix_3x4_aarch64) _sk_matrix_3x4_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xaa0803e9 // mov x9, x8 .long 0x9100910a // add x10, x8, #0x24 .long 0x4ddfc933 // ld1r {v19.4s}, [x9], #4 @@ -3531,7 +3922,7 @@ _sk_matrix_3x4_aarch64: .long 0x4eb01e00 // mov v0.16b, v16.16b .long 0x4eb11e21 // mov v1.16b, v17.16b .long 0x4eb21e42 // mov v2.16b, v18.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_matrix_4x5_aarch64 .globl _sk_matrix_4x5_aarch64 @@ -3549,7 +3940,7 @@ _sk_matrix_4x5_aarch64: .long 0x2d465533 // ldp s19, s21, [x9, #48] .long 0x2d475d36 // ldp s22, s23, [x9, #56] .long 0x9101312a // add x10, x9, #0x4c - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f931070 // fmla v16.4s, v3.4s, v19.s[0] .long 0x4d40c953 // ld1r {v19.4s}, [x10] .long 0x4f951071 // fmla v17.4s, v3.4s, v21.s[0] @@ -3578,7 +3969,7 @@ _sk_matrix_4x5_aarch64: .long 0x4eb11e21 // mov v1.16b, v17.16b .long 0x4eb21e42 // mov v2.16b, v18.16b .long 0x4eb31e63 // mov v3.16b, v19.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_matrix_4x3_aarch64 .globl _sk_matrix_4x3_aarch64 @@ -3604,20 +3995,20 @@ _sk_matrix_4x3_aarch64: .long 0x4f951022 // fmla v2.4s, v1.4s, v21.s[0] .long 0x4f961023 // fmla v3.4s, v1.4s, v22.s[0] .long 0x2d414d01 // ldp s1, s19, [x8, #8] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f921011 // fmla v17.4s, v0.4s, v18.s[0] .long 0x91004021 // add x1, x1, #0x10 .long 0x4f811002 // fmla v2.4s, v0.4s, v1.s[0] .long 0x4f931003 // fmla v3.4s, v0.4s, v19.s[0] .long 0x4eb01e00 // mov v0.16b, v16.16b .long 0x4eb11e21 // mov v1.16b, v17.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_matrix_perspective_aarch64 .globl _sk_matrix_perspective_aarch64 FUNCTION(_sk_matrix_perspective_aarch64) _sk_matrix_perspective_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xaa0803e9 // mov x9, x8 .long 0x9100510a // add x10, x8, #0x14 .long 0x4ddfc930 // ld1r {v16.4s}, [x9], #4 @@ -3640,7 +4031,7 @@ _sk_matrix_perspective_aarch64: .long 0x4e20ce14 // fmla v20.4s, v16.4s, v0.4s .long 0x6e32de21 // fmul v1.4s, v17.4s, v18.4s .long 0x6e32de80 // fmul v0.4s, v20.4s, v18.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_evenly_spaced_gradient_aarch64 .globl _sk_evenly_spaced_gradient_aarch64 @@ -3663,70 +4054,70 @@ _sk_evenly_spaced_gradient_aarch64: .long 0x6f20a422 // uxtl2 v2.2d, v1.4s .long 0x2f20a421 // uxtl v1.2d, v1.2s .long 0x9e660032 // fmov x18, d1 - .long 0x9e660044 // fmov x4, d2 + .long 0x9e660045 // fmov x5, d2 .long 0x4e183c2b // mov x11, v1.d[1] - .long 0x4e183c43 // mov x3, v2.d[1] - .long 0xbc647921 // ldr s1, [x9, x4, lsl #2] - .long 0xbc6479a2 // ldr s2, [x13, x4, lsl #2] - .long 0xbc6479c3 // ldr s3, [x14, x4, lsl #2] - .long 0xbc647a11 // ldr s17, [x16, x4, lsl #2] - .long 0xbc6479f2 // ldr s18, [x15, x4, lsl #2] - .long 0xbc647a33 // ldr s19, [x17, x4, lsl #2] - .long 0xbc647994 // ldr s20, [x12, x4, lsl #2] - .long 0xbc647955 // ldr s21, [x10, x4, lsl #2] - .long 0x8b120924 // add x4, x9, x18, lsl #2 - .long 0x0d408096 // ld1 {v22.s}[0], [x4] - .long 0x8b1209a4 // add x4, x13, x18, lsl #2 - .long 0x0d408090 // ld1 {v16.s}[0], [x4] - .long 0x8b0b0924 // add x4, x9, x11, lsl #2 - .long 0x0d409096 // ld1 {v22.s}[1], [x4] - .long 0x8b1209c4 // add x4, x14, x18, lsl #2 - .long 0x0d408097 // ld1 {v23.s}[0], [x4] - .long 0x8b120a04 // add x4, x16, x18, lsl #2 + .long 0x4e183c44 // mov x4, v2.d[1] + .long 0xbc657921 // ldr s1, [x9, x5, lsl #2] + .long 0xbc6579a2 // ldr s2, [x13, x5, lsl #2] + .long 0xbc6579c3 // ldr s3, [x14, x5, lsl #2] + .long 0xbc657a11 // ldr s17, [x16, x5, lsl #2] + .long 0xbc6579f2 // ldr s18, [x15, x5, lsl #2] + .long 0xbc657a33 // ldr s19, [x17, x5, lsl #2] + .long 0xbc657994 // ldr s20, [x12, x5, lsl #2] + .long 0xbc657955 // ldr s21, [x10, x5, lsl #2] + .long 0x8b120925 // add x5, x9, x18, lsl #2 + .long 0x0d4080b6 // ld1 {v22.s}[0], [x5] + .long 0x8b1209a5 // add x5, x13, x18, lsl #2 + .long 0x0d4080b0 // ld1 {v16.s}[0], [x5] + .long 0x8b0b0925 // add x5, x9, x11, lsl #2 + .long 0x0d4090b6 // ld1 {v22.s}[1], [x5] + .long 0x8b1209c5 // add x5, x14, x18, lsl #2 + .long 0x0d4080b7 // ld1 {v23.s}[0], [x5] + .long 0x8b120a05 // add x5, x16, x18, lsl #2 .long 0x6e140436 // mov v22.s[2], v1.s[0] - .long 0x0d408081 // ld1 {v1.s}[0], [x4] - .long 0x8b0b09a4 // add x4, x13, x11, lsl #2 - .long 0x0d409090 // ld1 {v16.s}[1], [x4] - .long 0x8b0b09c4 // add x4, x14, x11, lsl #2 - .long 0x0d409097 // ld1 {v23.s}[1], [x4] - .long 0x8b1209e4 // add x4, x15, x18, lsl #2 - .long 0x0d408098 // ld1 {v24.s}[0], [x4] - .long 0x8b120a24 // add x4, x17, x18, lsl #2 + .long 0x0d4080a1 // ld1 {v1.s}[0], [x5] + .long 0x8b0b09a5 // add x5, x13, x11, lsl #2 + .long 0x0d4090b0 // ld1 {v16.s}[1], [x5] + .long 0x8b0b09c5 // add x5, x14, x11, lsl #2 + .long 0x0d4090b7 // ld1 {v23.s}[1], [x5] + .long 0x8b1209e5 // add x5, x15, x18, lsl #2 + .long 0x0d4080b8 // ld1 {v24.s}[0], [x5] + .long 0x8b120a25 // add x5, x17, x18, lsl #2 .long 0x6e140450 // mov v16.s[2], v2.s[0] - .long 0x0d408082 // ld1 {v2.s}[0], [x4] - .long 0x8b0b0a04 // add x4, x16, x11, lsl #2 - .long 0x0d409081 // ld1 {v1.s}[1], [x4] - .long 0x8b0b09e4 // add x4, x15, x11, lsl #2 - .long 0x0d409098 // ld1 {v24.s}[1], [x4] - .long 0x8b120984 // add x4, x12, x18, lsl #2 + .long 0x0d4080a2 // ld1 {v2.s}[0], [x5] + .long 0x8b0b0a05 // add x5, x16, x11, lsl #2 + .long 0x0d4090a1 // ld1 {v1.s}[1], [x5] + .long 0x8b0b09e5 // add x5, x15, x11, lsl #2 + .long 0x0d4090b8 // ld1 {v24.s}[1], [x5] + .long 0x8b120985 // add x5, x12, x18, lsl #2 .long 0x8b120952 // add x18, x10, x18, lsl #2 .long 0x6e140477 // mov v23.s[2], v3.s[0] .long 0x0d408243 // ld1 {v3.s}[0], [x18] .long 0x8b0b0a32 // add x18, x17, x11, lsl #2 .long 0x6e140621 // mov v1.s[2], v17.s[0] - .long 0x0d408091 // ld1 {v17.s}[0], [x4] + .long 0x0d4080b1 // ld1 {v17.s}[0], [x5] .long 0x0d409242 // ld1 {v2.s}[1], [x18] .long 0x8b0b0992 // add x18, x12, x11, lsl #2 .long 0x6e140658 // mov v24.s[2], v18.s[0] .long 0x0d409251 // ld1 {v17.s}[1], [x18] .long 0x6e140662 // mov v2.s[2], v19.s[0] - .long 0xbc637932 // ldr s18, [x9, x3, lsl #2] - .long 0xbc6379b3 // ldr s19, [x13, x3, lsl #2] + .long 0xbc647932 // ldr s18, [x9, x4, lsl #2] + .long 0xbc6479b3 // ldr s19, [x13, x4, lsl #2] .long 0x6e140691 // mov v17.s[2], v20.s[0] - .long 0xbc6379d4 // ldr s20, [x14, x3, lsl #2] + .long 0xbc6479d4 // ldr s20, [x14, x4, lsl #2] .long 0x6e1c0656 // mov v22.s[3], v18.s[0] - .long 0xbc637a12 // ldr s18, [x16, x3, lsl #2] + .long 0xbc647a12 // ldr s18, [x16, x4, lsl #2] .long 0x6e1c0670 // mov v16.s[3], v19.s[0] - .long 0xbc6379f3 // ldr s19, [x15, x3, lsl #2] + .long 0xbc6479f3 // ldr s19, [x15, x4, lsl #2] .long 0x8b0b094b // add x11, x10, x11, lsl #2 .long 0x0d409163 // ld1 {v3.s}[1], [x11] .long 0x6e1c0697 // mov v23.s[3], v20.s[0] - .long 0xbc637a34 // ldr s20, [x17, x3, lsl #2] + .long 0xbc647a34 // ldr s20, [x17, x4, lsl #2] .long 0x6e1c0641 // mov v1.s[3], v18.s[0] - .long 0xbc637992 // ldr s18, [x12, x3, lsl #2] + .long 0xbc647992 // ldr s18, [x12, x4, lsl #2] .long 0x6e1c0678 // mov v24.s[3], v19.s[0] - .long 0xbc637953 // ldr s19, [x10, x3, lsl #2] - .long 0xf9400503 // ldr x3, [x8, #8] + .long 0xbc647953 // ldr s19, [x10, x4, lsl #2] + .long 0xf9400504 // ldr x4, [x8, #8] .long 0x6e1406a3 // mov v3.s[2], v21.s[0] .long 0x6e1c0682 // mov v2.s[3], v20.s[0] .long 0x6e1c0651 // mov v17.s[3], v18.s[0] @@ -3737,7 +4128,7 @@ _sk_evenly_spaced_gradient_aarch64: .long 0x4e20ce23 // fmla v3.4s, v17.4s, v0.4s .long 0x4eb01e00 // mov v0.16b, v16.16b .long 0x910043ff // add sp, sp, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_gauss_a_to_rgba_aarch64 .globl _sk_gauss_a_to_rgba_aarch64 @@ -3758,7 +4149,7 @@ _sk_gauss_a_to_rgba_aarch64: .long 0x4e23cc01 // fmla v1.4s, v0.4s, v3.4s .long 0x72830008 // movk w8, #0x1800 .long 0x4e040d30 // dup v16.4s, w9 - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x4e23cc22 // fmla v2.4s, v1.4s, v3.4s .long 0x4e040d00 // dup v0.4s, w8 .long 0x4e23cc50 // fmla v16.4s, v2.4s, v3.4s @@ -3766,7 +4157,7 @@ _sk_gauss_a_to_rgba_aarch64: .long 0x4ea01c01 // mov v1.16b, v0.16b .long 0x4ea01c02 // mov v2.16b, v0.16b .long 0x4ea01c03 // mov v3.16b, v0.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_gradient_aarch64 .globl _sk_gradient_aarch64 @@ -3780,7 +4171,7 @@ _sk_gradient_aarch64: .long 0x6f00e411 // movi v17.2d, #0x0 .long 0xf9400109 // ldr x9, [x8] .long 0xf100093f // cmp x9, #0x2 - .long 0x540001c3 // b.cc 3268 <sk_gradient_aarch64+0x58> // b.lo, b.ul, b.last + .long 0x540001c3 // b.cc 3884 <sk_gradient_aarch64+0x58> // b.lo, b.ul, b.last .long 0xf940250a // ldr x10, [x8, #72] .long 0xd1000529 // sub x9, x9, #0x1 .long 0x6f00e401 // movi v1.2d, #0x0 @@ -3791,7 +4182,7 @@ _sk_gradient_aarch64: .long 0x6e23e403 // fcmge v3.4s, v0.4s, v3.4s .long 0x4e221c63 // and v3.16b, v3.16b, v2.16b .long 0x4ea18461 // add v1.4s, v3.4s, v1.4s - .long 0xb5ffff69 // cbnz x9, 3248 <sk_gradient_aarch64+0x38> + .long 0xb5ffff69 // cbnz x9, 3864 <sk_gradient_aarch64+0x38> .long 0x6f20a431 // uxtl2 v17.2d, v1.4s .long 0x2f20a421 // uxtl v1.2d, v1.2s .long 0xa940b10a // ldp x10, x12, [x8, #8] @@ -3864,7 +4255,7 @@ _sk_gradient_aarch64: .long 0x6e1c06d4 // mov v20.s[3], v22.s[0] .long 0xbc6b7916 // ldr s22, [x8, x11, lsl #2] .long 0x4d4081e3 // ld1 {v3.s}[2], [x15] - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e1c06e2 // mov v2.s[3], v23.s[0] .long 0x6e1c0635 // mov v21.s[3], v17.s[0] .long 0x6e1c06c3 // mov v3.s[3], v22.s[0] @@ -3874,13 +4265,13 @@ _sk_gradient_aarch64: .long 0x4e20cea3 // fmla v3.4s, v21.4s, v0.4s .long 0x4eb01e00 // mov v0.16b, v16.16b .long 0x910043ff // add sp, sp, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_evenly_spaced_2_stop_gradient_aarch64 .globl _sk_evenly_spaced_2_stop_gradient_aarch64 FUNCTION(_sk_evenly_spaced_2_stop_gradient_aarch64) _sk_evenly_spaced_2_stop_gradient_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0xaa0803e9 // mov x9, x8 .long 0x9100410a // add x10, x8, #0x10 .long 0x4ddfc931 // ld1r {v17.4s}, [x9], #4 @@ -3898,7 +4289,7 @@ _sk_evenly_spaced_2_stop_gradient_aarch64: .long 0x4f931003 // fmla v3.4s, v0.4s, v19.s[0] .long 0x4f911001 // fmla v1.4s, v0.4s, v17.s[0] .long 0x4eb01e00 // mov v0.16b, v16.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_xy_to_unit_angle_aarch64 .globl _sk_xy_to_unit_angle_aarch64 @@ -3937,23 +4328,23 @@ _sk_xy_to_unit_angle_aarch64: .long 0x6e701e40 // bsl v0.16b, v18.16b, v16.16b .long 0x4ea0e831 // fcmlt v17.4s, v1.4s, #0.0 .long 0x4ea0d690 // fsub v16.4s, v20.4s, v0.4s - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e601e11 // bsl v17.16b, v16.16b, v0.16b .long 0x6ea0ca20 // fcmge v0.4s, v17.4s, #0.0 .long 0x4ea0ea30 // fcmlt v16.4s, v17.4s, #0.0 .long 0x4ea01e00 // orr v0.16b, v16.16b, v0.16b .long 0x4e201e20 // and v0.16b, v17.16b, v0.16b - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_xy_to_radius_aarch64 .globl _sk_xy_to_radius_aarch64 FUNCTION(_sk_xy_to_radius_aarch64) _sk_xy_to_radius_aarch64: - .long 0xf8408423 // ldr x3, [x1], #8 + .long 0xf8408424 // ldr x4, [x1], #8 .long 0x6e21dc30 // fmul v16.4s, v1.4s, v1.4s .long 0x4e20cc10 // fmla v16.4s, v0.4s, v0.4s .long 0x6ea1fa00 // fsqrt v0.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_save_xy_aarch64 .globl _sk_save_xy_aarch64 @@ -3971,15 +4362,15 @@ _sk_save_xy_aarch64: .long 0x3d800901 // str q1, [x8, #32] .long 0x3d801111 // str q17, [x8, #64] .long 0x3d801910 // str q16, [x8, #96] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_accumulate_aarch64 .globl _sk_accumulate_aarch64 FUNCTION(_sk_accumulate_aarch64) _sk_accumulate_aarch64: - .long 0xa8c10c28 // ldp x8, x3, [x1], #16 + .long 0xa8c11028 // ldp x8, x4, [x1], #16 .long 0x3dc02110 // ldr q16, [x8, #128] .long 0x3dc02911 // ldr q17, [x8, #160] .long 0x6e31de10 // fmul v16.4s, v16.4s, v17.4s @@ -3987,7 +4378,7 @@ _sk_accumulate_aarch64: .long 0x4e30cc25 // fmla v5.4s, v1.4s, v16.4s .long 0x4e30cc46 // fmla v6.4s, v2.4s, v16.4s .long 0x4e30cc67 // fmla v7.4s, v3.4s, v16.4s - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bilinear_nx_aarch64 .globl _sk_bilinear_nx_aarch64 @@ -3999,11 +4390,11 @@ _sk_bilinear_nx_aarch64: .long 0x3dc00110 // ldr q16, [x8] .long 0x4ea0d620 // fsub v0.4s, v17.4s, v0.4s .long 0x3d802100 // str q0, [x8, #128] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f0567e0 // movi v0.4s, #0xbf, lsl #24 .long 0x4e20d600 // fadd v0.4s, v16.4s, v0.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bilinear_px_aarch64 .globl _sk_bilinear_px_aarch64 @@ -4013,11 +4404,11 @@ _sk_bilinear_px_aarch64: .long 0x3dc01100 // ldr q0, [x8, #64] .long 0x3dc00110 // ldr q16, [x8] .long 0x3d802100 // str q0, [x8, #128] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f0167e0 // movi v0.4s, #0x3f, lsl #24 .long 0x4e20d600 // fadd v0.4s, v16.4s, v0.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bilinear_ny_aarch64 .globl _sk_bilinear_ny_aarch64 @@ -4029,11 +4420,11 @@ _sk_bilinear_ny_aarch64: .long 0x3dc00910 // ldr q16, [x8, #32] .long 0x4ea1d621 // fsub v1.4s, v17.4s, v1.4s .long 0x3d802901 // str q1, [x8, #160] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f0567e1 // movi v1.4s, #0xbf, lsl #24 .long 0x4e21d601 // fadd v1.4s, v16.4s, v1.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bilinear_py_aarch64 .globl _sk_bilinear_py_aarch64 @@ -4043,11 +4434,11 @@ _sk_bilinear_py_aarch64: .long 0x3dc01901 // ldr q1, [x8, #96] .long 0x3dc00910 // ldr q16, [x8, #32] .long 0x3d802901 // str q1, [x8, #160] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f0167e1 // movi v1.4s, #0x3f, lsl #24 .long 0x4e21d601 // fadd v1.4s, v16.4s, v1.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bicubic_n3x_aarch64 .globl _sk_bicubic_n3x_aarch64 @@ -4068,11 +4459,11 @@ _sk_bicubic_n3x_aarch64: .long 0x6e32de20 // fmul v0.4s, v17.4s, v18.4s .long 0x3dc00113 // ldr q19, [x8] .long 0x3d802100 // str q0, [x8, #128] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f07f700 // fmov v0.4s, #-1.500000000000000000e+00 .long 0x4e20d660 // fadd v0.4s, v19.4s, v0.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bicubic_n1x_aarch64 .globl _sk_bicubic_n1x_aarch64 @@ -4095,11 +4486,11 @@ _sk_bicubic_n1x_aarch64: .long 0x4e20ce51 // fmla v17.4s, v18.4s, v0.4s .long 0x3dc00110 // ldr q16, [x8] .long 0x3d802111 // str q17, [x8, #128] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f0567e0 // movi v0.4s, #0xbf, lsl #24 .long 0x4e20d600 // fadd v0.4s, v16.4s, v0.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bicubic_p1x_aarch64 .globl _sk_bicubic_p1x_aarch64 @@ -4121,9 +4512,9 @@ _sk_bicubic_p1x_aarch64: .long 0x4e040d31 // dup v17.4s, w9 .long 0x4e32ce11 // fmla v17.4s, v16.4s, v18.4s .long 0x3d802111 // str q17, [x8, #128] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bicubic_p3x_aarch64 .globl _sk_bicubic_p3x_aarch64 @@ -4142,11 +4533,11 @@ _sk_bicubic_p3x_aarch64: .long 0x6e31de60 // fmul v0.4s, v19.4s, v17.4s .long 0x3dc00112 // ldr q18, [x8] .long 0x3d802100 // str q0, [x8, #128] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f03f700 // fmov v0.4s, #1.500000000000000000e+00 .long 0x4e20d640 // fadd v0.4s, v18.4s, v0.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bicubic_n3y_aarch64 .globl _sk_bicubic_n3y_aarch64 @@ -4167,11 +4558,11 @@ _sk_bicubic_n3y_aarch64: .long 0x6e32de21 // fmul v1.4s, v17.4s, v18.4s .long 0x3dc00913 // ldr q19, [x8, #32] .long 0x3d802901 // str q1, [x8, #160] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f07f701 // fmov v1.4s, #-1.500000000000000000e+00 .long 0x4e21d661 // fadd v1.4s, v19.4s, v1.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bicubic_n1y_aarch64 .globl _sk_bicubic_n1y_aarch64 @@ -4194,11 +4585,11 @@ _sk_bicubic_n1y_aarch64: .long 0x4e21ce51 // fmla v17.4s, v18.4s, v1.4s .long 0x3dc00910 // ldr q16, [x8, #32] .long 0x3d802911 // str q17, [x8, #160] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f0567e1 // movi v1.4s, #0xbf, lsl #24 .long 0x4e21d601 // fadd v1.4s, v16.4s, v1.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bicubic_p1y_aarch64 .globl _sk_bicubic_p1y_aarch64 @@ -4220,9 +4611,9 @@ _sk_bicubic_p1y_aarch64: .long 0x4e040d31 // dup v17.4s, w9 .long 0x4e32ce11 // fmla v17.4s, v16.4s, v18.4s .long 0x3d802911 // str q17, [x8, #160] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_bicubic_p3y_aarch64 .globl _sk_bicubic_p3y_aarch64 @@ -4241,45 +4632,49 @@ _sk_bicubic_p3y_aarch64: .long 0x6e31de61 // fmul v1.4s, v19.4s, v17.4s .long 0x3dc00912 // ldr q18, [x8, #32] .long 0x3d802901 // str q1, [x8, #160] - .long 0xf9400423 // ldr x3, [x1, #8] + .long 0xf9400424 // ldr x4, [x1, #8] .long 0x4f03f701 // fmov v1.4s, #1.500000000000000000e+00 .long 0x4e21d641 // fadd v1.4s, v18.4s, v1.4s .long 0x91004021 // add x1, x1, #0x10 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 HIDDEN _sk_callback_aarch64 .globl _sk_callback_aarch64 FUNCTION(_sk_callback_aarch64) _sk_callback_aarch64: .long 0xd101c3ff // sub sp, sp, #0x70 - .long 0xf90023f6 // str x22, [sp, #64] + .long 0xa9045bf7 // stp x23, x22, [sp, #64] .long 0xa90553f5 // stp x21, x20, [sp, #80] .long 0xa9067bf3 // stp x19, x30, [sp, #96] .long 0xad011fe6 // stp q6, q7, [sp, #32] .long 0xad0017e4 // stp q4, q5, [sp] - .long 0xaa0103f4 // mov x20, x1 - .long 0xf9400295 // ldr x21, [x20] - .long 0xaa0003f6 // mov x22, x0 - .long 0x321e03e1 // orr w1, wzr, #0x4 - .long 0xaa0203f3 // mov x19, x2 - .long 0x910022a8 // add x8, x21, #0x8 + .long 0xaa0103f5 // mov x21, x1 + .long 0xf94002b7 // ldr x23, [x21] + .long 0xaa0303f3 // mov x19, x3 + .long 0xf100027f // cmp x19, #0x0 + .long 0x321e03e9 // orr w9, wzr, #0x4 + .long 0x910022e8 // add x8, x23, #0x8 .long 0x4c000900 // st4 {v0.4s-v3.4s}, [x8] - .long 0xf94002a8 // ldr x8, [x21] - .long 0xaa1503e0 // mov x0, x21 + .long 0xf94002e8 // ldr x8, [x23] + .long 0xaa0003f6 // mov x22, x0 + .long 0x1a891261 // csel w1, w19, w9, ne // ne = any + .long 0xaa1703e0 // mov x0, x23 + .long 0xaa0203f4 // mov x20, x2 .long 0xd63f0100 // blr x8 - .long 0xf94046a8 // ldr x8, [x21, #136] - .long 0xf9400683 // ldr x3, [x20, #8] - .long 0x91004281 // add x1, x20, #0x10 + .long 0xf94046e8 // ldr x8, [x23, #136] + .long 0xf94006a4 // ldr x4, [x21, #8] + .long 0x910042a1 // add x1, x21, #0x10 .long 0xaa1603e0 // mov x0, x22 .long 0x4c400900 // ld4 {v0.4s-v3.4s}, [x8] - .long 0xaa1303e2 // mov x2, x19 + .long 0xaa1403e2 // mov x2, x20 + .long 0xaa1303e3 // mov x3, x19 .long 0xad4017e4 // ldp q4, q5, [sp] .long 0xad411fe6 // ldp q6, q7, [sp, #32] .long 0xa9467bf3 // ldp x19, x30, [sp, #96] .long 0xa94553f5 // ldp x21, x20, [sp, #80] - .long 0xf94023f6 // ldr x22, [sp, #64] + .long 0xa9445bf7 // ldp x23, x22, [sp, #64] .long 0x9101c3ff // add sp, sp, #0x70 - .long 0xd61f0060 // br x3 + .long 0xd61f0080 // br x4 #elif defined(__arm__) BALIGN4 diff --git a/src/jumper/SkJumper_stages.cpp b/src/jumper/SkJumper_stages.cpp index 7cf537d214..46a18d4390 100644 --- a/src/jumper/SkJumper_stages.cpp +++ b/src/jumper/SkJumper_stages.cpp @@ -76,7 +76,7 @@ struct LazyCtx { // We're finally going to get to what a Stage function looks like! // It's best to jump down to the #else case first, then to come back up here for AVX. -#if defined(JUMPER) && (defined(__SSE2__) || defined(__arm__)) +#if defined(JUMPER) && (defined(__SSE2__) || defined(__arm__) || defined(__aarch64__)) // Process the tail on all x86 processors with SSE2 or better instructions. // tail == 0 ~~> work on a full kStride pixels // tail != 0 ~~> work on only the first tail pixels diff --git a/src/jumper/SkJumper_vectors.h b/src/jumper/SkJumper_vectors.h index 4deec83a28..a7898398a6 100644 --- a/src/jumper/SkJumper_vectors.h +++ b/src/jumper/SkJumper_vectors.h @@ -108,31 +108,63 @@ } SI void load3(const uint16_t* ptr, size_t tail, U16* r, U16* g, U16* b) { - uint16x4x3_t rgb = vld3_u16(ptr); + uint16x4x3_t rgb; + if (__builtin_expect(tail,0)) { + if ( true ) { rgb = vld3_lane_u16(ptr + 0, rgb, 0); } + if (tail > 1) { rgb = vld3_lane_u16(ptr + 3, rgb, 1); } + if (tail > 2) { rgb = vld3_lane_u16(ptr + 6, rgb, 2); } + } else { + rgb = vld3_u16(ptr); + } *r = rgb.val[0]; *g = rgb.val[1]; *b = rgb.val[2]; } SI void load4(const uint16_t* ptr, size_t tail, U16* r, U16* g, U16* b, U16* a) { - uint16x4x4_t rgba = vld4_u16(ptr); + uint16x4x4_t rgba; + if (__builtin_expect(tail,0)) { + if ( true ) { rgba = vld4_lane_u16(ptr + 0, rgba, 0); } + if (tail > 1) { rgba = vld4_lane_u16(ptr + 4, rgba, 1); } + if (tail > 2) { rgba = vld4_lane_u16(ptr + 8, rgba, 2); } + } else { + rgba = vld4_u16(ptr); + } *r = rgba.val[0]; *g = rgba.val[1]; *b = rgba.val[2]; *a = rgba.val[3]; } SI void store4(uint16_t* ptr, size_t tail, U16 r, U16 g, U16 b, U16 a) { - vst4_u16(ptr, (uint16x4x4_t{{r,g,b,a}})); + if (__builtin_expect(tail,0)) { + if ( true ) { vst4_lane_u16(ptr + 0, (uint16x4x4_t{{r,g,b,a}}), 0); } + if (tail > 1) { vst4_lane_u16(ptr + 4, (uint16x4x4_t{{r,g,b,a}}), 1); } + if (tail > 2) { vst4_lane_u16(ptr + 8, (uint16x4x4_t{{r,g,b,a}}), 2); } + } else { + vst4_u16(ptr, (uint16x4x4_t{{r,g,b,a}})); + } } - SI void load4(const float* ptr, size_t tail, F* r, F* g, F* b, F* a) { - float32x4x4_t rgba = vld4q_f32(ptr); + float32x4x4_t rgba; + if (__builtin_expect(tail,0)) { + if ( true ) { rgba = vld4q_lane_f32(ptr + 0, rgba, 0); } + if (tail > 1) { rgba = vld4q_lane_f32(ptr + 4, rgba, 1); } + if (tail > 2) { rgba = vld4q_lane_f32(ptr + 8, rgba, 2); } + } else { + rgba = vld4q_f32(ptr); + } *r = rgba.val[0]; *g = rgba.val[1]; *b = rgba.val[2]; *a = rgba.val[3]; } SI void store4(float* ptr, size_t tail, F r, F g, F b, F a) { - vst4q_f32(ptr, (float32x4x4_t{{r,g,b,a}})); + if (__builtin_expect(tail,0)) { + if ( true ) { vst4q_lane_f32(ptr + 0, (float32x4x4_t{{r,g,b,a}}), 0); } + if (tail > 1) { vst4q_lane_f32(ptr + 4, (float32x4x4_t{{r,g,b,a}}), 1); } + if (tail > 2) { vst4q_lane_f32(ptr + 8, (float32x4x4_t{{r,g,b,a}}), 2); } + } else { + vst4q_f32(ptr, (float32x4x4_t{{r,g,b,a}})); + } } #elif defined(__arm__) diff --git a/tests/SkRasterPipelineTest.cpp b/tests/SkRasterPipelineTest.cpp index 5f6a0b3ff0..1af40fd5f6 100644 --- a/tests/SkRasterPipelineTest.cpp +++ b/tests/SkRasterPipelineTest.cpp @@ -115,15 +115,18 @@ DEF_TEST(SkRasterPipeline_tail, r) { float* src = &data[0][0]; float* dst = &buffer[0][0]; - for (unsigned i = 0; i < 4; i++) { + for (unsigned i = 1; i <= 4; i++) { memset(buffer, 0xff, sizeof(buffer)); SkRasterPipeline_<256> p; p.append(SkRasterPipeline::load_f32, &src); p.append(SkRasterPipeline::store_f32, &dst); p.run(0, i); for (unsigned j = 0; j < i; j++) { - REPORTER_ASSERT(r, - !memcmp(&data[j][0], &buffer[j][0], sizeof(buffer[j]))); + for (unsigned k = 0; k < 4; k++) { + if (buffer[j][k] != data[j][k]) { + ERRORF(r, "(%u, %u) - a: %g r: %g\n", j, k, data[j][k], buffer[j][k]); + } + } } for (int j = i; j < 4; j++) { for (auto f : buffer[j]) { @@ -144,7 +147,7 @@ DEF_TEST(SkRasterPipeline_tail, r) { uint16_t* src = &data[0][0]; uint16_t* dst = &buffer[0][0]; - for (unsigned i = 0; i < 4; i++) { + for (unsigned i = 1; i <= 4; i++) { memset(buffer, 0xff, sizeof(buffer)); SkRasterPipeline_<256> p; p.append(SkRasterPipeline::load_f16, &src); @@ -181,7 +184,7 @@ DEF_TEST(SkRasterPipeline_tail, r) { uint16_t* src = &data[0][0]; float* dst = &buffer[0][0]; - for (unsigned i = 0; i < 4; i++) { + for (unsigned i = 1; i <= 4; i++) { memset(buffer, 0xff, sizeof(buffer)); SkRasterPipeline_<256> p; p.append(SkRasterPipeline::load_rgb_u16_be, &src); |