blob: c36b06138a3b6ae132e45f3aaf0665025f04e009 [file] [log] [blame]
/*
* Copyright 2016 Google Inc.
*
* Use of this source code is governed by a BSD-style license that can be
* found in the LICENSE file.
*/
#include "include/private/SkImageInfoPriv.h"
#include "include/private/SkTemplates.h"
#include "include/private/SkVx.h"
#include "include/third_party/skcms/skcms.h"
#include "src/core/SkColorSpacePriv.h"
#include "src/core/SkOpts.h"
#include "src/core/SkRasterPipeline.h"
#include <algorithm>
bool gForceHighPrecisionRasterPipeline;
SkRasterPipeline::SkRasterPipeline(SkArenaAlloc* alloc) : fAlloc(alloc) {
this->reset();
}
void SkRasterPipeline::reset() {
fStages = nullptr;
fNumStages = 0;
fSlotsNeeded = 1; // We always need one extra slot for just_return().
}
void SkRasterPipeline::append(StockStage stage, void* ctx) {
SkASSERT(stage != uniform_color); // Please use append_constant_color().
SkASSERT(stage != unbounded_uniform_color); // Please use append_constant_color().
SkASSERT(stage != set_rgb); // Please use append_set_rgb().
SkASSERT(stage != unbounded_set_rgb); // Please use append_set_rgb().
SkASSERT(stage != clamp_gamut); // Please use append_gamut_clamp_if_normalized().
SkASSERT(stage != parametric); // Please use append_transfer_function().
SkASSERT(stage != gamma_); // Please use append_transfer_function().
SkASSERT(stage != PQish); // Please use append_transfer_function().
SkASSERT(stage != HLGish); // Please use append_transfer_function().
SkASSERT(stage != HLGinvish); // Please use append_transfer_function().
this->unchecked_append(stage, ctx);
}
void SkRasterPipeline::unchecked_append(StockStage stage, void* ctx) {
fStages = fAlloc->make<StageList>( StageList{fStages, stage, ctx} );
fNumStages += 1;
fSlotsNeeded += ctx ? 2 : 1;
}
void SkRasterPipeline::append(StockStage stage, uintptr_t ctx) {
void* ptrCtx;
memcpy(&ptrCtx, &ctx, sizeof(ctx));
this->append(stage, ptrCtx);
}
void SkRasterPipeline::extend(const SkRasterPipeline& src) {
if (src.empty()) {
return;
}
auto stages = fAlloc->makeArrayDefault<StageList>(src.fNumStages);
int n = src.fNumStages;
const StageList* st = src.fStages;
while (n --> 1) {
stages[n] = *st;
stages[n].prev = &stages[n-1];
st = st->prev;
}
stages[0] = *st;
stages[0].prev = fStages;
fStages = &stages[src.fNumStages - 1];
fNumStages += src.fNumStages;
fSlotsNeeded += src.fSlotsNeeded - 1; // Don't double count just_returns().
}
void SkRasterPipeline::dump() const {
SkDebugf("SkRasterPipeline, %d stages\n", fNumStages);
std::vector<const char*> stages;
for (auto st = fStages; st; st = st->prev) {
const char* name = "";
switch (st->stage) {
#define M(x) case x: name = #x; break;
SK_RASTER_PIPELINE_STAGES(M)
#undef M
}
stages.push_back(name);
}
std::reverse(stages.begin(), stages.end());
for (const char* name : stages) {
SkDebugf("\t%s\n", name);
}
SkDebugf("\n");
}
void SkRasterPipeline::append_set_rgb(SkArenaAlloc* alloc, const float rgb[3]) {
auto arg = alloc->makeArrayDefault<float>(3);
arg[0] = rgb[0];
arg[1] = rgb[1];
arg[2] = rgb[2];
auto stage = unbounded_set_rgb;
if (0 <= rgb[0] && rgb[0] <= 1 &&
0 <= rgb[1] && rgb[1] <= 1 &&
0 <= rgb[2] && rgb[2] <= 1)
{
stage = set_rgb;
}
this->unchecked_append(stage, arg);
}
void SkRasterPipeline::append_constant_color(SkArenaAlloc* alloc, const float rgba[4]) {
// r,g,b might be outside [0,1], but alpha should probably always be in [0,1].
SkASSERT(0 <= rgba[3] && rgba[3] <= 1);
if (rgba[0] == 0 && rgba[1] == 0 && rgba[2] == 0 && rgba[3] == 1) {
this->append(black_color);
} else if (rgba[0] == 1 && rgba[1] == 1 && rgba[2] == 1 && rgba[3] == 1) {
this->append(white_color);
} else {
auto ctx = alloc->make<SkRasterPipeline_UniformColorCtx>();
skvx::float4 color = skvx::float4::Load(rgba);
color.store(&ctx->r);
// uniform_color requires colors in range and can go lowp,
// while unbounded_uniform_color supports out-of-range colors too but not lowp.
if (0 <= rgba[0] && rgba[0] <= rgba[3] &&
0 <= rgba[1] && rgba[1] <= rgba[3] &&
0 <= rgba[2] && rgba[2] <= rgba[3]) {
// To make loads more direct, we store 8-bit values in 16-bit slots.
color = color * 255.0f + 0.5f;
ctx->rgba[0] = (uint16_t)color[0];
ctx->rgba[1] = (uint16_t)color[1];
ctx->rgba[2] = (uint16_t)color[2];
ctx->rgba[3] = (uint16_t)color[3];
this->unchecked_append(uniform_color, ctx);
} else {
this->unchecked_append(unbounded_uniform_color, ctx);
}
}
}
void SkRasterPipeline::append_matrix(SkArenaAlloc* alloc, const SkMatrix& matrix) {
SkMatrix::TypeMask mt = matrix.getType();
if (mt == SkMatrix::kIdentity_Mask) {
return;
}
if (mt == SkMatrix::kTranslate_Mask) {
float* trans = alloc->makeArrayDefault<float>(2);
trans[0] = matrix.getTranslateX();
trans[1] = matrix.getTranslateY();
this->append(SkRasterPipeline::matrix_translate, trans);
} else if ((mt | (SkMatrix::kScale_Mask | SkMatrix::kTranslate_Mask)) ==
(SkMatrix::kScale_Mask | SkMatrix::kTranslate_Mask)) {
float* scaleTrans = alloc->makeArrayDefault<float>(4);
scaleTrans[0] = matrix.getScaleX();
scaleTrans[1] = matrix.getScaleY();
scaleTrans[2] = matrix.getTranslateX();
scaleTrans[3] = matrix.getTranslateY();
this->append(SkRasterPipeline::matrix_scale_translate, scaleTrans);
} else {
float* storage = alloc->makeArrayDefault<float>(9);
matrix.get9(storage);
if (!matrix.hasPerspective()) {
// note: asAffine and the 2x3 stage really only need 6 entries
this->append(SkRasterPipeline::matrix_2x3, storage);
} else {
this->append(SkRasterPipeline::matrix_perspective, storage);
}
}
}
void SkRasterPipeline::append_load(SkColorType ct, const SkRasterPipeline_MemoryCtx* ctx) {
switch (ct) {
case kUnknown_SkColorType: SkASSERT(false); break;
case kAlpha_8_SkColorType: this->append(load_a8, ctx); break;
case kA16_unorm_SkColorType: this->append(load_a16, ctx); break;
case kA16_float_SkColorType: this->append(load_af16, ctx); break;
case kRGB_565_SkColorType: this->append(load_565, ctx); break;
case kARGB_4444_SkColorType: this->append(load_4444, ctx); break;
case kR8G8_unorm_SkColorType: this->append(load_rg88, ctx); break;
case kR16G16_unorm_SkColorType: this->append(load_rg1616, ctx); break;
case kR16G16_float_SkColorType: this->append(load_rgf16, ctx); break;
case kRGBA_8888_SkColorType: this->append(load_8888, ctx); break;
case kRGBA_1010102_SkColorType: this->append(load_1010102, ctx); break;
case kR16G16B16A16_unorm_SkColorType:this->append(load_16161616,ctx); break;
case kRGBA_F16Norm_SkColorType:
case kRGBA_F16_SkColorType: this->append(load_f16, ctx); break;
case kRGBA_F32_SkColorType: this->append(load_f32, ctx); break;
case kGray_8_SkColorType: this->append(load_a8, ctx);
this->append(alpha_to_gray);
break;
case kR8_unorm_SkColorType: this->append(load_a8, ctx);
this->append(alpha_to_red);
break;
case kRGB_888x_SkColorType: this->append(load_8888, ctx);
this->append(force_opaque);
break;
case kBGRA_1010102_SkColorType: this->append(load_1010102, ctx);
this->append(swap_rb);
break;
case kRGB_101010x_SkColorType: this->append(load_1010102, ctx);
this->append(force_opaque);
break;
case kBGR_101010x_SkColorType: this->append(load_1010102, ctx);
this->append(force_opaque);
this->append(swap_rb);
break;
case kBGRA_8888_SkColorType: this->append(load_8888, ctx);
this->append(swap_rb);
break;
case kSRGBA_8888_SkColorType:
this->append(load_8888, ctx);
this->append_transfer_function(*skcms_sRGB_TransferFunction());
break;
}
}
void SkRasterPipeline::append_load_dst(SkColorType ct, const SkRasterPipeline_MemoryCtx* ctx) {
switch (ct) {
case kUnknown_SkColorType: SkASSERT(false); break;
case kAlpha_8_SkColorType: this->append(load_a8_dst, ctx); break;
case kA16_unorm_SkColorType: this->append(load_a16_dst, ctx); break;
case kA16_float_SkColorType: this->append(load_af16_dst, ctx); break;
case kRGB_565_SkColorType: this->append(load_565_dst, ctx); break;
case kARGB_4444_SkColorType: this->append(load_4444_dst, ctx); break;
case kR8G8_unorm_SkColorType: this->append(load_rg88_dst, ctx); break;
case kR16G16_unorm_SkColorType: this->append(load_rg1616_dst, ctx); break;
case kR16G16_float_SkColorType: this->append(load_rgf16_dst, ctx); break;
case kRGBA_8888_SkColorType: this->append(load_8888_dst, ctx); break;
case kRGBA_1010102_SkColorType: this->append(load_1010102_dst, ctx); break;
case kR16G16B16A16_unorm_SkColorType: this->append(load_16161616_dst,ctx); break;
case kRGBA_F16Norm_SkColorType:
case kRGBA_F16_SkColorType: this->append(load_f16_dst, ctx); break;
case kRGBA_F32_SkColorType: this->append(load_f32_dst, ctx); break;
case kGray_8_SkColorType: this->append(load_a8_dst, ctx);
this->append(alpha_to_gray_dst);
break;
case kR8_unorm_SkColorType: this->append(load_a8_dst, ctx);
this->append(alpha_to_red_dst);
break;
case kRGB_888x_SkColorType: this->append(load_8888_dst, ctx);
this->append(force_opaque_dst);
break;
case kBGRA_1010102_SkColorType: this->append(load_1010102_dst, ctx);
this->append(swap_rb_dst);
break;
case kRGB_101010x_SkColorType: this->append(load_1010102_dst, ctx);
this->append(force_opaque_dst);
break;
case kBGR_101010x_SkColorType: this->append(load_1010102_dst, ctx);
this->append(force_opaque_dst);
this->append(swap_rb_dst);
break;
case kBGRA_8888_SkColorType: this->append(load_8888_dst, ctx);
this->append(swap_rb_dst);
break;
case kSRGBA_8888_SkColorType:
// TODO: We could remove the double-swap if we had _dst versions of all the TF stages
this->append(load_8888_dst, ctx);
this->append(swap_src_dst);
this->append_transfer_function(*skcms_sRGB_TransferFunction());
this->append(swap_src_dst);
break;
}
}
void SkRasterPipeline::append_store(SkColorType ct, const SkRasterPipeline_MemoryCtx* ctx) {
switch (ct) {
case kUnknown_SkColorType: SkASSERT(false); break;
case kAlpha_8_SkColorType: this->append(store_a8, ctx); break;
case kR8_unorm_SkColorType: this->append(store_r8, ctx); break;
case kA16_unorm_SkColorType: this->append(store_a16, ctx); break;
case kA16_float_SkColorType: this->append(store_af16, ctx); break;
case kRGB_565_SkColorType: this->append(store_565, ctx); break;
case kARGB_4444_SkColorType: this->append(store_4444, ctx); break;
case kR8G8_unorm_SkColorType: this->append(store_rg88, ctx); break;
case kR16G16_unorm_SkColorType: this->append(store_rg1616, ctx); break;
case kR16G16_float_SkColorType: this->append(store_rgf16, ctx); break;
case kRGBA_8888_SkColorType: this->append(store_8888, ctx); break;
case kRGBA_1010102_SkColorType: this->append(store_1010102, ctx); break;
case kR16G16B16A16_unorm_SkColorType: this->append(store_16161616,ctx); break;
case kRGBA_F16Norm_SkColorType:
case kRGBA_F16_SkColorType: this->append(store_f16, ctx); break;
case kRGBA_F32_SkColorType: this->append(store_f32, ctx); break;
case kRGB_888x_SkColorType: this->append(force_opaque);
this->append(store_8888, ctx);
break;
case kBGRA_1010102_SkColorType: this->append(swap_rb);
this->append(store_1010102, ctx);
break;
case kRGB_101010x_SkColorType: this->append(force_opaque);
this->append(store_1010102, ctx);
break;
case kBGR_101010x_SkColorType: this->append(force_opaque);
this->append(swap_rb);
this->append(store_1010102, ctx);
break;
case kGray_8_SkColorType: this->append(bt709_luminance_or_luma_to_alpha);
this->append(store_a8, ctx);
break;
case kBGRA_8888_SkColorType: this->append(swap_rb);
this->append(store_8888, ctx);
break;
case kSRGBA_8888_SkColorType:
this->append_transfer_function(*skcms_sRGB_Inverse_TransferFunction());
this->append(store_8888, ctx);
break;
}
}
void SkRasterPipeline::append_transfer_function(const skcms_TransferFunction& tf) {
void* ctx = const_cast<void*>(static_cast<const void*>(&tf));
switch (classify_transfer_fn(tf)) {
case Bad_TF: SkASSERT(false); break;
case TFKind::sRGBish_TF:
if (tf.a == 1 && tf.b == 0 && tf.c == 0 && tf.d == 0 && tf.e == 0 && tf.f == 0) {
this->unchecked_append(gamma_, ctx);
} else {
this->unchecked_append(parametric, ctx);
}
break;
case PQish_TF: this->unchecked_append(PQish, ctx); break;
case HLGish_TF: this->unchecked_append(HLGish, ctx); break;
case HLGinvish_TF: this->unchecked_append(HLGinvish, ctx); break;
}
}
// Clamp premul values to [0,alpha] (logical [0,1]) to avoid the confusing
// scenario of being able to store a logical color channel > 1.0 when alpha < 1.0.
// Most software that works with normalized premul values expect r,g,b channels all <= a.
//
// In addition, GL clamps all its color channels to limits of the format just
// before the blend step (~here). To match that auto-clamp, we clamp alpha to
// [0,1] too, just in case someone gave us a crazy alpha.
void SkRasterPipeline::append_gamut_clamp_if_normalized(const SkImageInfo& info) {
if (info.alphaType() == kPremul_SkAlphaType && SkColorTypeIsNormalized(info.colorType())) {
this->unchecked_append(SkRasterPipeline::clamp_gamut, nullptr);
}
}
SkRasterPipeline::StartPipelineFn SkRasterPipeline::build_pipeline(void** ip) const {
if (!gForceHighPrecisionRasterPipeline) {
// We'll try to build a lowp pipeline, but if that fails fallback to a highp float pipeline.
void** reset_point = ip;
// Stages are stored backwards in fStages, so we reverse here, back to front.
*--ip = (void*)SkOpts::just_return_lowp;
for (const StageList* st = fStages; st; st = st->prev) {
if (auto fn = SkOpts::stages_lowp[st->stage]) {
if (st->ctx) {
*--ip = st->ctx;
}
*--ip = (void*)fn;
} else {
ip = reset_point;
break;
}
}
if (ip != reset_point) {
return SkOpts::start_pipeline_lowp;
}
}
*--ip = (void*)SkOpts::just_return_highp;
for (const StageList* st = fStages; st; st = st->prev) {
if (st->ctx) {
*--ip = st->ctx;
}
*--ip = (void*)SkOpts::stages_highp[st->stage];
}
return SkOpts::start_pipeline_highp;
}
void SkRasterPipeline::run(size_t x, size_t y, size_t w, size_t h) const {
if (this->empty()) {
return;
}
// Best to not use fAlloc here... we can't bound how often run() will be called.
SkAutoSTMalloc<64, void*> program(fSlotsNeeded);
auto start_pipeline = this->build_pipeline(program.get() + fSlotsNeeded);
start_pipeline(x,y,x+w,y+h, program.get());
}
std::function<void(size_t, size_t, size_t, size_t)> SkRasterPipeline::compile() const {
if (this->empty()) {
return [](size_t, size_t, size_t, size_t) {};
}
void** program = fAlloc->makeArray<void*>(fSlotsNeeded);
auto start_pipeline = this->build_pipeline(program + fSlotsNeeded);
return [=](size_t x, size_t y, size_t w, size_t h) {
start_pipeline(x,y,x+w,y+h, program);
};
}