1605 lines
56 KiB
C
Raw Normal View History

/*
* Copyright (C) 2013 The Android Open Source Project
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#ifndef ANDROID_HWUI_DISPLAY_OPERATION_H
#define ANDROID_HWUI_DISPLAY_OPERATION_H
#ifndef LOG_TAG
#define LOG_TAG "OpenGLRenderer"
#endif
#include <SkPath.h>
#include <SkPathOps.h>
#include <SkXfermode.h>
#include <private/hwui/DrawGlInfo.h>
#include "OpenGLRenderer.h"
Pack preloaded framework assets in a texture atlas When the Android runtime starts, the system preloads a series of assets in the Zygote process. These assets are shared across all processes. Unfortunately, each one of these assets is later uploaded in its own OpenGL texture, once per process. This wastes memory and generates unnecessary OpenGL state changes. This CL introduces an asset server that provides an atlas to all processes. Note: bitmaps used by skia shaders are *not* sampled from the atlas. It's an uncommon use case and would require extra texture transforms in the GL shaders. WHAT IS THE ASSETS ATLAS The "assets atlas" is a single, shareable graphic buffer that contains all the system's preloaded bitmap drawables (this includes 9-patches.) The atlas is made of two distinct objects: the graphic buffer that contains the actual pixels and the map which indicates where each preloaded bitmap can be found in the atlas (essentially a pair of x and y coordinates.) HOW IS THE ASSETS ATLAS GENERATED Because we need to support a wide variety of devices and because it is easy to change the list of preloaded drawables, the atlas is generated at runtime, during the startup phase of the system process. There are several steps that lead to the atlas generation: 1. If the device is booting for the first time, or if the device was updated, we need to find the best atlas configuration. To do so, the atlas service tries a number of width, height and algorithm variations that allows us to pack as many assets as possible while using as little memory as possible. Once a best configuration is found, it gets written to disk in /data/system/framework_atlas 2. Given a best configuration (algorithm variant, dimensions and number of bitmaps that can be packed in the atlas), the atlas service packs all the preloaded bitmaps into a single graphic buffer object. 3. The packing is done using Skia in a temporary native bitmap. The Skia bitmap is then copied into the graphic buffer using OpenGL ES to benefit from texture swizzling. HOW PROCESSES USE THE ATLAS Whenever a process' hardware renderer initializes its EGL context, it queries the atlas service for the graphic buffer and the map. It is important to remember that both the context and the map will be valid for the lifetime of the hardware renderer (if the system process goes down, all apps get killed as well.) Every time the hardware renderer needs to render a bitmap, it first checks whether the bitmap can be found in the assets atlas. When the bitmap is part of the atlas, texture coordinates are remapped appropriately before rendering. Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
2013-04-17 18:54:38 -07:00
#include "AssetAtlas.h"
#include "DeferredDisplayList.h"
#include "DisplayListRenderer.h"
Pack preloaded framework assets in a texture atlas When the Android runtime starts, the system preloads a series of assets in the Zygote process. These assets are shared across all processes. Unfortunately, each one of these assets is later uploaded in its own OpenGL texture, once per process. This wastes memory and generates unnecessary OpenGL state changes. This CL introduces an asset server that provides an atlas to all processes. Note: bitmaps used by skia shaders are *not* sampled from the atlas. It's an uncommon use case and would require extra texture transforms in the GL shaders. WHAT IS THE ASSETS ATLAS The "assets atlas" is a single, shareable graphic buffer that contains all the system's preloaded bitmap drawables (this includes 9-patches.) The atlas is made of two distinct objects: the graphic buffer that contains the actual pixels and the map which indicates where each preloaded bitmap can be found in the atlas (essentially a pair of x and y coordinates.) HOW IS THE ASSETS ATLAS GENERATED Because we need to support a wide variety of devices and because it is easy to change the list of preloaded drawables, the atlas is generated at runtime, during the startup phase of the system process. There are several steps that lead to the atlas generation: 1. If the device is booting for the first time, or if the device was updated, we need to find the best atlas configuration. To do so, the atlas service tries a number of width, height and algorithm variations that allows us to pack as many assets as possible while using as little memory as possible. Once a best configuration is found, it gets written to disk in /data/system/framework_atlas 2. Given a best configuration (algorithm variant, dimensions and number of bitmaps that can be packed in the atlas), the atlas service packs all the preloaded bitmaps into a single graphic buffer object. 3. The packing is done using Skia in a temporary native bitmap. The Skia bitmap is then copied into the graphic buffer using OpenGL ES to benefit from texture swizzling. HOW PROCESSES USE THE ATLAS Whenever a process' hardware renderer initializes its EGL context, it queries the atlas service for the graphic buffer and the map. It is important to remember that both the context and the map will be valid for the lifetime of the hardware renderer (if the system process goes down, all apps get killed as well.) Every time the hardware renderer needs to render a bitmap, it first checks whether the bitmap can be found in the assets atlas. When the bitmap is part of the atlas, texture coordinates are remapped appropriately before rendering. Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
2013-04-17 18:54:38 -07:00
#include "UvMapper.h"
#include "utils/LinearAllocator.h"
#define CRASH() do { \
*(int *)(uintptr_t) 0xbbadbeef = 0; \
((void(*)())0)(); /* More reliable, but doesn't say BBADBEEF */ \
} while(false)
// Use OP_LOG for logging with arglist, OP_LOGS if just printing char*
#define OP_LOGS(s) OP_LOG("%s", (s))
#define OP_LOG(s, ...) ALOGD( "%*s" s, level * 2, "", __VA_ARGS__ )
namespace android {
namespace uirenderer {
/**
* Structure for storing canvas operations when they are recorded into a DisplayList, so that they
* may be replayed to an OpenGLRenderer.
*
* To avoid individual memory allocations, DisplayListOps may only be allocated into a
* LinearAllocator's managed memory buffers. Each pointer held by a DisplayListOp is either a
* pointer into memory also allocated in the LinearAllocator (mostly for text and float buffers) or
* references a externally refcounted object (Sk... and Skia... objects). ~DisplayListOp() is
* never called as LinearAllocators are simply discarded, so no memory management should be done in
* this class.
*/
class DisplayListOp {
public:
// These objects should always be allocated with a LinearAllocator, and never destroyed/deleted.
// standard new() intentionally not implemented, and delete/deconstructor should never be used.
virtual ~DisplayListOp() { CRASH(); }
static void operator delete(void* ptr) { CRASH(); }
/** static void* operator new(size_t size); PURPOSELY OMITTED **/
static void* operator new(size_t size, LinearAllocator& allocator) {
return allocator.alloc(size);
}
enum OpLogFlag {
kOpLogFlag_Recurse = 0x1,
kOpLogFlag_JSON = 0x2 // TODO: add?
};
virtual void defer(DeferStateStruct& deferStruct, int saveCount, int level,
bool useQuickReject) = 0;
virtual void replay(ReplayStateStruct& replayStruct, int saveCount, int level,
bool useQuickReject) = 0;
virtual void output(int level, uint32_t logFlags = 0) const = 0;
// NOTE: it would be nice to declare constants and overriding the implementation in each op to
// point at the constants, but that seems to require a .cpp file
virtual const char* name() = 0;
};
class StateOp : public DisplayListOp {
public:
StateOp() {};
virtual ~StateOp() {}
virtual void defer(DeferStateStruct& deferStruct, int saveCount, int level,
bool useQuickReject) {
// default behavior only affects immediate, deferrable state, issue directly to renderer
applyState(deferStruct.mRenderer, saveCount);
}
/**
* State operations are applied directly to the renderer, but can cause the deferred drawing op
* list to flush
*/
virtual void replay(ReplayStateStruct& replayStruct, int saveCount, int level,
bool useQuickReject) {
applyState(replayStruct.mRenderer, saveCount);
}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const = 0;
};
class DrawOp : public DisplayListOp {
friend class MergingDrawBatch;
public:
DrawOp(const SkPaint* paint)
: mPaint(paint), mQuickRejected(false) {}
virtual void defer(DeferStateStruct& deferStruct, int saveCount, int level,
bool useQuickReject) {
if (mQuickRejected && CC_LIKELY(useQuickReject)) {
return;
}
deferStruct.mDeferredList.addDrawOp(deferStruct.mRenderer, this);
}
virtual void replay(ReplayStateStruct& replayStruct, int saveCount, int level,
bool useQuickReject) {
if (mQuickRejected && CC_LIKELY(useQuickReject)) {
return;
}
replayStruct.mDrawGlStatus |= applyDraw(replayStruct.mRenderer, replayStruct.mDirty);
}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) = 0;
/**
* Draw multiple instances of an operation, must be overidden for operations that merge
*
* Currently guarantees certain similarities between ops (see MergingDrawBatch::canMergeWith),
* and pure translation transformations. Other guarantees of similarity should be enforced by
* reducing which operations are tagged as mergeable.
*/
virtual status_t multiDraw(OpenGLRenderer& renderer, Rect& dirty,
const Vector<OpStatePair>& ops, const Rect& bounds) {
status_t status = DrawGlInfo::kStatusDone;
for (unsigned int i = 0; i < ops.size(); i++) {
renderer.restoreDisplayState(*(ops[i].state), true);
status |= ops[i].op->applyDraw(renderer, dirty);
}
return status;
}
/**
* When this method is invoked the state field is initialized to have the
* final rendering state. We can thus use it to process data as it will be
* used at draw time.
*
* Additionally, this method allows subclasses to provide defer-time preferences for batching
* and merging.
*
* if a subclass can set deferInfo.mergeable to true, it should implement multiDraw()
*/
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {}
/**
* Query the conservative, local bounds (unmapped) bounds of the op.
*
* returns true if bounds exist
*/
virtual bool getLocalBounds(const DrawModifiers& drawModifiers, Rect& localBounds) {
return false;
}
// TODO: better refine localbounds usage
void setQuickRejected(bool quickRejected) { mQuickRejected = quickRejected; }
bool getQuickRejected() { return mQuickRejected; }
inline int getPaintAlpha() const {
return OpenGLRenderer::getAlphaDirect(mPaint);
}
virtual bool hasTextShadow() const {
return false;
}
inline float strokeWidthOutset() {
// since anything AA stroke with less than 1.0 pixel width is drawn with an alpha-reduced
// 1.0 stroke, treat 1.0 as minimum.
// TODO: it would be nice if this could take scale into account, but scale isn't stable
// since higher levels of the view hierarchy can change scale out from underneath it.
return fmaxf(mPaint->getStrokeWidth(), 1) * 0.5f;
}
protected:
const SkPaint* getPaint(OpenGLRenderer& renderer) {
return renderer.filterPaint(mPaint);
}
// Helper method for determining op opaqueness. Assumes op fills its bounds in local
// coordinates, and that paint's alpha is used
inline bool isOpaqueOverBounds(const DeferredDisplayState& state) {
// ensure that local bounds cover mapped bounds
if (!state.mMatrix.isSimple()) return false;
// check state/paint for transparency
Inspect SkShader to determine hw shader. Instead of duplicating internal info about SkShader, inspect the SkShader installed on the SkPaint. core/java/android/view/GLES20Canvas.java: Remove setupModifiers, nResetModifiers, and nSetupShader. core/jni/android/graphics/Shader.cpp: Remove calls to create/destroy the (previously) attached SkiaShader. core/jni/android_view_GLES20Canvas.cpp: Remove native code for setupShader and resetModifiers. graphics/java/android/graphics/BitmapShader.java: graphics/java/android/graphics/ComposeShader.java: graphics/java/android/graphics/LinearGradient.java: graphics/java/android/graphics/RadialGradient.java: graphics/java/android/graphics/Shader.java: graphics/java/android/graphics/SweepGradient.java: Remove code keeping track of native SkiaShader. libs/hwui/Caches.h: Include Extensions.h. libs/hwui/DeferredDisplayList.cpp: Compare shaders on the paint, instead of on DrawModifiers. libs/hwui/DisplayList.cpp: libs/hwui/DisplayList.h: Remove vector of SkiaShaders. libs/hwui/DisplayListOp.h: Access the SkShader on mPaint. Remove SetupShaderOp and ResetShaderOp. libs/hwui/DisplayListRenderer.cpp: libs/hwui/DisplayListRenderer.h: Remove resetShader, setupShader, refShader, and mShaderMap. libs/hwui/FontRenderer.cpp: Pass SkShader to setupDrawShader and setupDrawShaderUniforms. libs/hwui/OpenGLRenderer.cpp: libs/hwui/OpenGLRenderer.h: Add LayerShader, a class inheriting from SkShader, to mimic the behavior of SkiaLayerShader. Unlike SkiaLayerShader, it can be set on the SkPaint so it can be inspected later. Set a LayerShader instead of a SkiaLayerShader. setupDrawShader and setupDrawShaderUniforms now inspect an SkShader passed in. Inspect SkShader instead of mDrawModifiers.mShader. Remove resetShader and setupShader. setupDrawColorUniforms now takes a boolean indicating whether there is a shader. Add an inline function for accessing the SkShader on an SkPaint. In setupDrawBlending(Layer*, bool), do not check the shader (which will never be set), but do check whether the color filter may change the alpha (newly fixed behavior). In setupDrawBlending(SkPaint, ...), check the SkShader and whether the color filter affects alpha (the latter is new behavior). libs/hwui/Renderer.h: Remove pure virtual functions setupShader and resetShader. libs/hwui/ResourceCache.cpp: libs/hwui/ResourceCache.h: Remove functions for refing/unrefing shaders. libs/hwui/SkiaShader.cpp: libs/hwui/SkiaShader.h: Much of this code was redundant and has been removed. Convert structs into class with nothing but static functions for calling describe/setupProgram. libs/hwui/TextureCache.cpp: libs/hwui/TextureCache.h: Use the SkPixelRef as the key to the bitmap Lru cache, since shader inspection will provide a different SkBitmap pointer (though it will hold the correct SkPixelRef with the correct generation ID). tests/CanvasCompare/src/com/android/test/hwuicompare/DisplayModifier.java: tests/CanvasCompare/src/com/android/test/hwuicompare/ResourceModifiers.java: Update manual test to have more shaders: radial, sweep, compose, invalid compose. BUG:10650594 Change-Id: Iaa7189178bda1c55f96da044d2a9fa602ba36034
2014-05-05 12:50:38 -04:00
if (mPaint) {
if (mPaint->getShader() && !mPaint->getShader()->isOpaque()) {
return false;
}
if (mPaint->getAlpha() != 0xFF) {
return false;
}
}
if (state.mAlpha != 1.0f) return false;
SkXfermode::Mode mode = OpenGLRenderer::getXfermodeDirect(mPaint);
return (mode == SkXfermode::kSrcOver_Mode ||
mode == SkXfermode::kSrc_Mode);
}
const SkPaint* mPaint; // should be accessed via getPaint() when applying
bool mQuickRejected;
};
class DrawBoundedOp : public DrawOp {
public:
DrawBoundedOp(float left, float top, float right, float bottom, const SkPaint* paint)
: DrawOp(paint), mLocalBounds(left, top, right, bottom) {}
DrawBoundedOp(const Rect& localBounds, const SkPaint* paint)
: DrawOp(paint), mLocalBounds(localBounds) {}
// Calculates bounds as smallest rect encompassing all points
// NOTE: requires at least 1 vertex, and doesn't account for stroke size (should be handled in
// subclass' constructor)
DrawBoundedOp(const float* points, int count, const SkPaint* paint)
: DrawOp(paint), mLocalBounds(points[0], points[1], points[0], points[1]) {
for (int i = 2; i < count; i += 2) {
mLocalBounds.left = fminf(mLocalBounds.left, points[i]);
mLocalBounds.right = fmaxf(mLocalBounds.right, points[i]);
mLocalBounds.top = fminf(mLocalBounds.top, points[i + 1]);
mLocalBounds.bottom = fmaxf(mLocalBounds.bottom, points[i + 1]);
}
}
// default empty constructor for bounds, to be overridden in child constructor body
DrawBoundedOp(const SkPaint* paint): DrawOp(paint) { }
bool getLocalBounds(const DrawModifiers& drawModifiers, Rect& localBounds) {
localBounds.set(mLocalBounds);
OpenGLRenderer::TextShadow textShadow;
if (OpenGLRenderer::getTextShadow(mPaint, &textShadow)) {
Rect shadow(mLocalBounds);
shadow.translate(textShadow.dx, textShadow.dx);
shadow.outset(textShadow.radius);
localBounds.unionWith(shadow);
}
return true;
}
protected:
Rect mLocalBounds; // displayed area in LOCAL coord. doesn't incorporate stroke, so check paint
};
///////////////////////////////////////////////////////////////////////////////
// STATE OPERATIONS - these may affect the state of the canvas/renderer, but do
// not directly draw or alter output
///////////////////////////////////////////////////////////////////////////////
class SaveOp : public StateOp {
public:
SaveOp(int flags)
: mFlags(flags) {}
virtual void defer(DeferStateStruct& deferStruct, int saveCount, int level,
bool useQuickReject) {
int newSaveCount = deferStruct.mRenderer.save(mFlags);
deferStruct.mDeferredList.addSave(deferStruct.mRenderer, this, newSaveCount);
}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.save(mFlags);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Save flags %x", mFlags);
}
virtual const char* name() { return "Save"; }
int getFlags() const { return mFlags; }
private:
int mFlags;
};
class RestoreToCountOp : public StateOp {
public:
RestoreToCountOp(int count)
: mCount(count) {}
virtual void defer(DeferStateStruct& deferStruct, int saveCount, int level,
bool useQuickReject) {
deferStruct.mDeferredList.addRestoreToCount(deferStruct.mRenderer,
this, saveCount + mCount);
deferStruct.mRenderer.restoreToCount(saveCount + mCount);
}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.restoreToCount(saveCount + mCount);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Restore to count %d", mCount);
}
virtual const char* name() { return "RestoreToCount"; }
private:
int mCount;
};
class SaveLayerOp : public StateOp {
public:
SaveLayerOp(float left, float top, float right, float bottom, int alpha, int flags)
: mArea(left, top, right, bottom)
, mPaint(&mCachedPaint)
, mFlags(flags)
, mConvexMask(NULL) {
mCachedPaint.setAlpha(alpha);
}
SaveLayerOp(float left, float top, float right, float bottom, const SkPaint* paint, int flags)
: mArea(left, top, right, bottom)
, mPaint(paint)
, mFlags(flags)
, mConvexMask(NULL)
{}
virtual void defer(DeferStateStruct& deferStruct, int saveCount, int level,
bool useQuickReject) {
// NOTE: don't bother with actual saveLayer, instead issuing it at flush time
int newSaveCount = deferStruct.mRenderer.getSaveCount();
deferStruct.mDeferredList.addSaveLayer(deferStruct.mRenderer, this, newSaveCount);
// NOTE: don't issue full saveLayer, since that has side effects/is costly. instead just
// setup the snapshot for deferral, and re-issue the op at flush time
deferStruct.mRenderer.saveLayerDeferred(mArea.left, mArea.top, mArea.right, mArea.bottom,
mPaint, mFlags);
}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.saveLayer(mArea.left, mArea.top, mArea.right, mArea.bottom,
mPaint, mFlags, mConvexMask);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("SaveLayer%s of area " RECT_STRING,
(isSaveLayerAlpha() ? "Alpha" : ""),RECT_ARGS(mArea));
}
virtual const char* name() { return isSaveLayerAlpha() ? "SaveLayerAlpha" : "SaveLayer"; }
int getFlags() { return mFlags; }
// Called to make SaveLayerOp clip to the provided mask when drawing back/restored
void setMask(const SkPath* convexMask) {
mConvexMask = convexMask;
}
private:
bool isSaveLayerAlpha() const {
SkXfermode::Mode mode = OpenGLRenderer::getXfermodeDirect(mPaint);
int alpha = OpenGLRenderer::getAlphaDirect(mPaint);
return alpha < 255 && mode == SkXfermode::kSrcOver_Mode;
}
Rect mArea;
const SkPaint* mPaint;
SkPaint mCachedPaint;
int mFlags;
// Convex path, points at data in RenderNode, valid for the duration of the frame only
// Only used for masking the SaveLayer which wraps projected RenderNodes
const SkPath* mConvexMask;
};
class TranslateOp : public StateOp {
public:
TranslateOp(float dx, float dy)
: mDx(dx), mDy(dy) {}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.translate(mDx, mDy);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Translate by %f %f", mDx, mDy);
}
virtual const char* name() { return "Translate"; }
private:
float mDx;
float mDy;
};
class RotateOp : public StateOp {
public:
RotateOp(float degrees)
: mDegrees(degrees) {}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.rotate(mDegrees);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Rotate by %f degrees", mDegrees);
}
virtual const char* name() { return "Rotate"; }
private:
float mDegrees;
};
class ScaleOp : public StateOp {
public:
ScaleOp(float sx, float sy)
: mSx(sx), mSy(sy) {}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.scale(mSx, mSy);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Scale by %f %f", mSx, mSy);
}
virtual const char* name() { return "Scale"; }
private:
float mSx;
float mSy;
};
class SkewOp : public StateOp {
public:
SkewOp(float sx, float sy)
: mSx(sx), mSy(sy) {}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.skew(mSx, mSy);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Skew by %f %f", mSx, mSy);
}
virtual const char* name() { return "Skew"; }
private:
float mSx;
float mSy;
};
class SetMatrixOp : public StateOp {
public:
SetMatrixOp(const SkMatrix& matrix)
: mMatrix(matrix) {}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.setMatrix(mMatrix);
}
virtual void output(int level, uint32_t logFlags) const {
if (mMatrix.isIdentity()) {
OP_LOGS("SetMatrix (reset)");
} else {
OP_LOG("SetMatrix " SK_MATRIX_STRING, SK_MATRIX_ARGS(&mMatrix));
}
}
virtual const char* name() { return "SetMatrix"; }
private:
const SkMatrix mMatrix;
};
class ConcatMatrixOp : public StateOp {
public:
ConcatMatrixOp(const SkMatrix& matrix)
: mMatrix(matrix) {}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.concatMatrix(mMatrix);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("ConcatMatrix " SK_MATRIX_STRING, SK_MATRIX_ARGS(&mMatrix));
}
virtual const char* name() { return "ConcatMatrix"; }
private:
const SkMatrix mMatrix;
};
class ClipOp : public StateOp {
public:
ClipOp(SkRegion::Op op) : mOp(op) {}
virtual void defer(DeferStateStruct& deferStruct, int saveCount, int level,
bool useQuickReject) {
// NOTE: must defer op BEFORE applying state, since it may read clip
deferStruct.mDeferredList.addClip(deferStruct.mRenderer, this);
// TODO: Can we avoid applying complex clips at defer time?
applyState(deferStruct.mRenderer, saveCount);
}
bool canCauseComplexClip() {
return ((mOp != SkRegion::kIntersect_Op) && (mOp != SkRegion::kReplace_Op)) || !isRect();
}
protected:
virtual bool isRect() { return false; }
SkRegion::Op mOp;
};
class ClipRectOp : public ClipOp {
public:
ClipRectOp(float left, float top, float right, float bottom, SkRegion::Op op)
: ClipOp(op), mArea(left, top, right, bottom) {}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.clipRect(mArea.left, mArea.top, mArea.right, mArea.bottom, mOp);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("ClipRect " RECT_STRING, RECT_ARGS(mArea));
}
virtual const char* name() { return "ClipRect"; }
protected:
virtual bool isRect() { return true; }
private:
Rect mArea;
};
class ClipPathOp : public ClipOp {
public:
ClipPathOp(const SkPath* path, SkRegion::Op op)
: ClipOp(op), mPath(path) {}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.clipPath(mPath, mOp);
}
virtual void output(int level, uint32_t logFlags) const {
SkRect bounds = mPath->getBounds();
OP_LOG("ClipPath bounds " RECT_STRING,
bounds.left(), bounds.top(), bounds.right(), bounds.bottom());
}
virtual const char* name() { return "ClipPath"; }
private:
const SkPath* mPath;
};
class ClipRegionOp : public ClipOp {
public:
ClipRegionOp(const SkRegion* region, SkRegion::Op op)
: ClipOp(op), mRegion(region) {}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.clipRegion(mRegion, mOp);
}
virtual void output(int level, uint32_t logFlags) const {
SkIRect bounds = mRegion->getBounds();
OP_LOG("ClipRegion bounds %d %d %d %d",
bounds.left(), bounds.top(), bounds.right(), bounds.bottom());
}
virtual const char* name() { return "ClipRegion"; }
private:
const SkRegion* mRegion;
};
class ResetPaintFilterOp : public StateOp {
public:
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.resetPaintFilter();
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOGS("ResetPaintFilter");
}
virtual const char* name() { return "ResetPaintFilter"; }
};
class SetupPaintFilterOp : public StateOp {
public:
SetupPaintFilterOp(int clearBits, int setBits)
: mClearBits(clearBits), mSetBits(setBits) {}
virtual void applyState(OpenGLRenderer& renderer, int saveCount) const {
renderer.setupPaintFilter(mClearBits, mSetBits);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("SetupPaintFilter, clear %#x, set %#x", mClearBits, mSetBits);
}
virtual const char* name() { return "SetupPaintFilter"; }
private:
int mClearBits;
int mSetBits;
};
///////////////////////////////////////////////////////////////////////////////
// DRAW OPERATIONS - these are operations that can draw to the canvas's device
///////////////////////////////////////////////////////////////////////////////
class DrawBitmapOp : public DrawBoundedOp {
public:
DrawBitmapOp(const SkBitmap* bitmap, float left, float top, const SkPaint* paint)
Pack preloaded framework assets in a texture atlas When the Android runtime starts, the system preloads a series of assets in the Zygote process. These assets are shared across all processes. Unfortunately, each one of these assets is later uploaded in its own OpenGL texture, once per process. This wastes memory and generates unnecessary OpenGL state changes. This CL introduces an asset server that provides an atlas to all processes. Note: bitmaps used by skia shaders are *not* sampled from the atlas. It's an uncommon use case and would require extra texture transforms in the GL shaders. WHAT IS THE ASSETS ATLAS The "assets atlas" is a single, shareable graphic buffer that contains all the system's preloaded bitmap drawables (this includes 9-patches.) The atlas is made of two distinct objects: the graphic buffer that contains the actual pixels and the map which indicates where each preloaded bitmap can be found in the atlas (essentially a pair of x and y coordinates.) HOW IS THE ASSETS ATLAS GENERATED Because we need to support a wide variety of devices and because it is easy to change the list of preloaded drawables, the atlas is generated at runtime, during the startup phase of the system process. There are several steps that lead to the atlas generation: 1. If the device is booting for the first time, or if the device was updated, we need to find the best atlas configuration. To do so, the atlas service tries a number of width, height and algorithm variations that allows us to pack as many assets as possible while using as little memory as possible. Once a best configuration is found, it gets written to disk in /data/system/framework_atlas 2. Given a best configuration (algorithm variant, dimensions and number of bitmaps that can be packed in the atlas), the atlas service packs all the preloaded bitmaps into a single graphic buffer object. 3. The packing is done using Skia in a temporary native bitmap. The Skia bitmap is then copied into the graphic buffer using OpenGL ES to benefit from texture swizzling. HOW PROCESSES USE THE ATLAS Whenever a process' hardware renderer initializes its EGL context, it queries the atlas service for the graphic buffer and the map. It is important to remember that both the context and the map will be valid for the lifetime of the hardware renderer (if the system process goes down, all apps get killed as well.) Every time the hardware renderer needs to render a bitmap, it first checks whether the bitmap can be found in the assets atlas. When the bitmap is part of the atlas, texture coordinates are remapped appropriately before rendering. Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
2013-04-17 18:54:38 -07:00
: DrawBoundedOp(left, top, left + bitmap->width(), top + bitmap->height(), paint),
mBitmap(bitmap), mAtlas(Caches::getInstance().assetAtlas) {
mEntry = mAtlas.getEntry(bitmap);
if (mEntry) {
mEntryGenerationId = mAtlas.getGenerationId();
mUvMapper = mEntry->uvMapper;
}
Pack preloaded framework assets in a texture atlas When the Android runtime starts, the system preloads a series of assets in the Zygote process. These assets are shared across all processes. Unfortunately, each one of these assets is later uploaded in its own OpenGL texture, once per process. This wastes memory and generates unnecessary OpenGL state changes. This CL introduces an asset server that provides an atlas to all processes. Note: bitmaps used by skia shaders are *not* sampled from the atlas. It's an uncommon use case and would require extra texture transforms in the GL shaders. WHAT IS THE ASSETS ATLAS The "assets atlas" is a single, shareable graphic buffer that contains all the system's preloaded bitmap drawables (this includes 9-patches.) The atlas is made of two distinct objects: the graphic buffer that contains the actual pixels and the map which indicates where each preloaded bitmap can be found in the atlas (essentially a pair of x and y coordinates.) HOW IS THE ASSETS ATLAS GENERATED Because we need to support a wide variety of devices and because it is easy to change the list of preloaded drawables, the atlas is generated at runtime, during the startup phase of the system process. There are several steps that lead to the atlas generation: 1. If the device is booting for the first time, or if the device was updated, we need to find the best atlas configuration. To do so, the atlas service tries a number of width, height and algorithm variations that allows us to pack as many assets as possible while using as little memory as possible. Once a best configuration is found, it gets written to disk in /data/system/framework_atlas 2. Given a best configuration (algorithm variant, dimensions and number of bitmaps that can be packed in the atlas), the atlas service packs all the preloaded bitmaps into a single graphic buffer object. 3. The packing is done using Skia in a temporary native bitmap. The Skia bitmap is then copied into the graphic buffer using OpenGL ES to benefit from texture swizzling. HOW PROCESSES USE THE ATLAS Whenever a process' hardware renderer initializes its EGL context, it queries the atlas service for the graphic buffer and the map. It is important to remember that both the context and the map will be valid for the lifetime of the hardware renderer (if the system process goes down, all apps get killed as well.) Every time the hardware renderer needs to render a bitmap, it first checks whether the bitmap can be found in the assets atlas. When the bitmap is part of the atlas, texture coordinates are remapped appropriately before rendering. Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
2013-04-17 18:54:38 -07:00
}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawBitmap(mBitmap, mLocalBounds.left, mLocalBounds.top,
getPaint(renderer));
}
AssetAtlas::Entry* getAtlasEntry() {
// The atlas entry is stale, let's get a new one
if (mEntry && mEntryGenerationId != mAtlas.getGenerationId()) {
mEntryGenerationId = mAtlas.getGenerationId();
mEntry = mAtlas.getEntry(mBitmap);
mUvMapper = mEntry->uvMapper;
}
return mEntry;
}
#define SET_TEXTURE(ptr, posRect, offsetRect, texCoordsRect, xDim, yDim) \
TextureVertex::set(ptr++, posRect.xDim - offsetRect.left, posRect.yDim - offsetRect.top, \
texCoordsRect.xDim, texCoordsRect.yDim)
/**
* This multi-draw operation builds a mesh on the stack by generating a quad
* for each bitmap in the batch. This method is also responsible for dirtying
* the current layer, if any.
*/
virtual status_t multiDraw(OpenGLRenderer& renderer, Rect& dirty,
const Vector<OpStatePair>& ops, const Rect& bounds) {
const DeferredDisplayState& firstState = *(ops[0].state);
renderer.restoreDisplayState(firstState, true); // restore all but the clip
TextureVertex vertices[6 * ops.size()];
TextureVertex* vertex = &vertices[0];
const bool hasLayer = renderer.hasLayer();
bool pureTranslate = true;
Pack preloaded framework assets in a texture atlas When the Android runtime starts, the system preloads a series of assets in the Zygote process. These assets are shared across all processes. Unfortunately, each one of these assets is later uploaded in its own OpenGL texture, once per process. This wastes memory and generates unnecessary OpenGL state changes. This CL introduces an asset server that provides an atlas to all processes. Note: bitmaps used by skia shaders are *not* sampled from the atlas. It's an uncommon use case and would require extra texture transforms in the GL shaders. WHAT IS THE ASSETS ATLAS The "assets atlas" is a single, shareable graphic buffer that contains all the system's preloaded bitmap drawables (this includes 9-patches.) The atlas is made of two distinct objects: the graphic buffer that contains the actual pixels and the map which indicates where each preloaded bitmap can be found in the atlas (essentially a pair of x and y coordinates.) HOW IS THE ASSETS ATLAS GENERATED Because we need to support a wide variety of devices and because it is easy to change the list of preloaded drawables, the atlas is generated at runtime, during the startup phase of the system process. There are several steps that lead to the atlas generation: 1. If the device is booting for the first time, or if the device was updated, we need to find the best atlas configuration. To do so, the atlas service tries a number of width, height and algorithm variations that allows us to pack as many assets as possible while using as little memory as possible. Once a best configuration is found, it gets written to disk in /data/system/framework_atlas 2. Given a best configuration (algorithm variant, dimensions and number of bitmaps that can be packed in the atlas), the atlas service packs all the preloaded bitmaps into a single graphic buffer object. 3. The packing is done using Skia in a temporary native bitmap. The Skia bitmap is then copied into the graphic buffer using OpenGL ES to benefit from texture swizzling. HOW PROCESSES USE THE ATLAS Whenever a process' hardware renderer initializes its EGL context, it queries the atlas service for the graphic buffer and the map. It is important to remember that both the context and the map will be valid for the lifetime of the hardware renderer (if the system process goes down, all apps get killed as well.) Every time the hardware renderer needs to render a bitmap, it first checks whether the bitmap can be found in the assets atlas. When the bitmap is part of the atlas, texture coordinates are remapped appropriately before rendering. Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
2013-04-17 18:54:38 -07:00
// TODO: manually handle rect clip for bitmaps by adjusting texCoords per op,
// and allowing them to be merged in getBatchId()
for (unsigned int i = 0; i < ops.size(); i++) {
const DeferredDisplayState& state = *(ops[i].state);
const Rect& opBounds = state.mBounds;
// When we reach multiDraw(), the matrix can be either
// pureTranslate or simple (translate and/or scale).
// If the matrix is not pureTranslate, then we have a scale
pureTranslate &= state.mMatrix.isPureTranslate();
Pack preloaded framework assets in a texture atlas When the Android runtime starts, the system preloads a series of assets in the Zygote process. These assets are shared across all processes. Unfortunately, each one of these assets is later uploaded in its own OpenGL texture, once per process. This wastes memory and generates unnecessary OpenGL state changes. This CL introduces an asset server that provides an atlas to all processes. Note: bitmaps used by skia shaders are *not* sampled from the atlas. It's an uncommon use case and would require extra texture transforms in the GL shaders. WHAT IS THE ASSETS ATLAS The "assets atlas" is a single, shareable graphic buffer that contains all the system's preloaded bitmap drawables (this includes 9-patches.) The atlas is made of two distinct objects: the graphic buffer that contains the actual pixels and the map which indicates where each preloaded bitmap can be found in the atlas (essentially a pair of x and y coordinates.) HOW IS THE ASSETS ATLAS GENERATED Because we need to support a wide variety of devices and because it is easy to change the list of preloaded drawables, the atlas is generated at runtime, during the startup phase of the system process. There are several steps that lead to the atlas generation: 1. If the device is booting for the first time, or if the device was updated, we need to find the best atlas configuration. To do so, the atlas service tries a number of width, height and algorithm variations that allows us to pack as many assets as possible while using as little memory as possible. Once a best configuration is found, it gets written to disk in /data/system/framework_atlas 2. Given a best configuration (algorithm variant, dimensions and number of bitmaps that can be packed in the atlas), the atlas service packs all the preloaded bitmaps into a single graphic buffer object. 3. The packing is done using Skia in a temporary native bitmap. The Skia bitmap is then copied into the graphic buffer using OpenGL ES to benefit from texture swizzling. HOW PROCESSES USE THE ATLAS Whenever a process' hardware renderer initializes its EGL context, it queries the atlas service for the graphic buffer and the map. It is important to remember that both the context and the map will be valid for the lifetime of the hardware renderer (if the system process goes down, all apps get killed as well.) Every time the hardware renderer needs to render a bitmap, it first checks whether the bitmap can be found in the assets atlas. When the bitmap is part of the atlas, texture coordinates are remapped appropriately before rendering. Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
2013-04-17 18:54:38 -07:00
Rect texCoords(0, 0, 1, 1);
((DrawBitmapOp*) ops[i].op)->mUvMapper.map(texCoords);
Pack preloaded framework assets in a texture atlas When the Android runtime starts, the system preloads a series of assets in the Zygote process. These assets are shared across all processes. Unfortunately, each one of these assets is later uploaded in its own OpenGL texture, once per process. This wastes memory and generates unnecessary OpenGL state changes. This CL introduces an asset server that provides an atlas to all processes. Note: bitmaps used by skia shaders are *not* sampled from the atlas. It's an uncommon use case and would require extra texture transforms in the GL shaders. WHAT IS THE ASSETS ATLAS The "assets atlas" is a single, shareable graphic buffer that contains all the system's preloaded bitmap drawables (this includes 9-patches.) The atlas is made of two distinct objects: the graphic buffer that contains the actual pixels and the map which indicates where each preloaded bitmap can be found in the atlas (essentially a pair of x and y coordinates.) HOW IS THE ASSETS ATLAS GENERATED Because we need to support a wide variety of devices and because it is easy to change the list of preloaded drawables, the atlas is generated at runtime, during the startup phase of the system process. There are several steps that lead to the atlas generation: 1. If the device is booting for the first time, or if the device was updated, we need to find the best atlas configuration. To do so, the atlas service tries a number of width, height and algorithm variations that allows us to pack as many assets as possible while using as little memory as possible. Once a best configuration is found, it gets written to disk in /data/system/framework_atlas 2. Given a best configuration (algorithm variant, dimensions and number of bitmaps that can be packed in the atlas), the atlas service packs all the preloaded bitmaps into a single graphic buffer object. 3. The packing is done using Skia in a temporary native bitmap. The Skia bitmap is then copied into the graphic buffer using OpenGL ES to benefit from texture swizzling. HOW PROCESSES USE THE ATLAS Whenever a process' hardware renderer initializes its EGL context, it queries the atlas service for the graphic buffer and the map. It is important to remember that both the context and the map will be valid for the lifetime of the hardware renderer (if the system process goes down, all apps get killed as well.) Every time the hardware renderer needs to render a bitmap, it first checks whether the bitmap can be found in the assets atlas. When the bitmap is part of the atlas, texture coordinates are remapped appropriately before rendering. Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
2013-04-17 18:54:38 -07:00
SET_TEXTURE(vertex, opBounds, bounds, texCoords, left, top);
SET_TEXTURE(vertex, opBounds, bounds, texCoords, right, top);
SET_TEXTURE(vertex, opBounds, bounds, texCoords, left, bottom);
SET_TEXTURE(vertex, opBounds, bounds, texCoords, left, bottom);
SET_TEXTURE(vertex, opBounds, bounds, texCoords, right, top);
SET_TEXTURE(vertex, opBounds, bounds, texCoords, right, bottom);
if (hasLayer) {
renderer.dirtyLayer(opBounds.left, opBounds.top, opBounds.right, opBounds.bottom);
}
}
return renderer.drawBitmaps(mBitmap, mEntry, ops.size(), &vertices[0],
pureTranslate, bounds, mPaint);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw bitmap %p at %f %f", mBitmap, mLocalBounds.left, mLocalBounds.top);
}
virtual const char* name() { return "DrawBitmap"; }
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
deferInfo.batchId = DeferredDisplayList::kOpBatch_Bitmap;
deferInfo.mergeId = getAtlasEntry() ?
(mergeid_t) mEntry->getMergeId() : (mergeid_t) mBitmap;
// Don't merge non-simply transformed or neg scale ops, SET_TEXTURE doesn't handle rotation
// Don't merge A8 bitmaps - the paint's color isn't compared by mergeId, or in
// MergingDrawBatch::canMergeWith()
// TODO: support clipped bitmaps by handling them in SET_TEXTURE
deferInfo.mergeable = state.mMatrix.isSimple() && state.mMatrix.positiveScale() &&
!state.mClipSideFlags &&
OpenGLRenderer::getXfermodeDirect(mPaint) == SkXfermode::kSrcOver_Mode &&
(mBitmap->config() != SkBitmap::kA8_Config);
}
const SkBitmap* bitmap() { return mBitmap; }
protected:
const SkBitmap* mBitmap;
const AssetAtlas& mAtlas;
uint32_t mEntryGenerationId;
AssetAtlas::Entry* mEntry;
Pack preloaded framework assets in a texture atlas When the Android runtime starts, the system preloads a series of assets in the Zygote process. These assets are shared across all processes. Unfortunately, each one of these assets is later uploaded in its own OpenGL texture, once per process. This wastes memory and generates unnecessary OpenGL state changes. This CL introduces an asset server that provides an atlas to all processes. Note: bitmaps used by skia shaders are *not* sampled from the atlas. It's an uncommon use case and would require extra texture transforms in the GL shaders. WHAT IS THE ASSETS ATLAS The "assets atlas" is a single, shareable graphic buffer that contains all the system's preloaded bitmap drawables (this includes 9-patches.) The atlas is made of two distinct objects: the graphic buffer that contains the actual pixels and the map which indicates where each preloaded bitmap can be found in the atlas (essentially a pair of x and y coordinates.) HOW IS THE ASSETS ATLAS GENERATED Because we need to support a wide variety of devices and because it is easy to change the list of preloaded drawables, the atlas is generated at runtime, during the startup phase of the system process. There are several steps that lead to the atlas generation: 1. If the device is booting for the first time, or if the device was updated, we need to find the best atlas configuration. To do so, the atlas service tries a number of width, height and algorithm variations that allows us to pack as many assets as possible while using as little memory as possible. Once a best configuration is found, it gets written to disk in /data/system/framework_atlas 2. Given a best configuration (algorithm variant, dimensions and number of bitmaps that can be packed in the atlas), the atlas service packs all the preloaded bitmaps into a single graphic buffer object. 3. The packing is done using Skia in a temporary native bitmap. The Skia bitmap is then copied into the graphic buffer using OpenGL ES to benefit from texture swizzling. HOW PROCESSES USE THE ATLAS Whenever a process' hardware renderer initializes its EGL context, it queries the atlas service for the graphic buffer and the map. It is important to remember that both the context and the map will be valid for the lifetime of the hardware renderer (if the system process goes down, all apps get killed as well.) Every time the hardware renderer needs to render a bitmap, it first checks whether the bitmap can be found in the assets atlas. When the bitmap is part of the atlas, texture coordinates are remapped appropriately before rendering. Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
2013-04-17 18:54:38 -07:00
UvMapper mUvMapper;
};
class DrawBitmapMatrixOp : public DrawBoundedOp {
public:
DrawBitmapMatrixOp(const SkBitmap* bitmap, const SkMatrix& matrix, const SkPaint* paint)
: DrawBoundedOp(paint), mBitmap(bitmap), mMatrix(matrix) {
mLocalBounds.set(0, 0, bitmap->width(), bitmap->height());
const mat4 transform(matrix);
transform.mapRect(mLocalBounds);
}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawBitmap(mBitmap, mMatrix, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw bitmap %p matrix " SK_MATRIX_STRING, mBitmap, SK_MATRIX_ARGS(&mMatrix));
}
virtual const char* name() { return "DrawBitmapMatrix"; }
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
deferInfo.batchId = DeferredDisplayList::kOpBatch_Bitmap;
}
private:
const SkBitmap* mBitmap;
const SkMatrix mMatrix;
};
class DrawBitmapRectOp : public DrawBoundedOp {
public:
DrawBitmapRectOp(const SkBitmap* bitmap,
float srcLeft, float srcTop, float srcRight, float srcBottom,
float dstLeft, float dstTop, float dstRight, float dstBottom, const SkPaint* paint)
: DrawBoundedOp(dstLeft, dstTop, dstRight, dstBottom, paint),
mBitmap(bitmap), mSrc(srcLeft, srcTop, srcRight, srcBottom) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawBitmap(mBitmap, mSrc.left, mSrc.top, mSrc.right, mSrc.bottom,
mLocalBounds.left, mLocalBounds.top, mLocalBounds.right, mLocalBounds.bottom,
getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw bitmap %p src=" RECT_STRING ", dst=" RECT_STRING,
mBitmap, RECT_ARGS(mSrc), RECT_ARGS(mLocalBounds));
}
virtual const char* name() { return "DrawBitmapRect"; }
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
deferInfo.batchId = DeferredDisplayList::kOpBatch_Bitmap;
}
private:
const SkBitmap* mBitmap;
Rect mSrc;
};
class DrawBitmapDataOp : public DrawBitmapOp {
public:
DrawBitmapDataOp(const SkBitmap* bitmap, float left, float top, const SkPaint* paint)
: DrawBitmapOp(bitmap, left, top, paint) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawBitmapData(mBitmap, mLocalBounds.left,
mLocalBounds.top, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw bitmap %p", mBitmap);
}
virtual const char* name() { return "DrawBitmapData"; }
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
deferInfo.batchId = DeferredDisplayList::kOpBatch_Bitmap;
}
};
class DrawBitmapMeshOp : public DrawBoundedOp {
public:
DrawBitmapMeshOp(const SkBitmap* bitmap, int meshWidth, int meshHeight,
const float* vertices, const int* colors, const SkPaint* paint)
: DrawBoundedOp(vertices, 2 * (meshWidth + 1) * (meshHeight + 1), paint),
mBitmap(bitmap), mMeshWidth(meshWidth), mMeshHeight(meshHeight),
mVertices(vertices), mColors(colors) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawBitmapMesh(mBitmap, mMeshWidth, mMeshHeight,
mVertices, mColors, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw bitmap %p mesh %d x %d", mBitmap, mMeshWidth, mMeshHeight);
}
virtual const char* name() { return "DrawBitmapMesh"; }
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
deferInfo.batchId = DeferredDisplayList::kOpBatch_Bitmap;
}
private:
const SkBitmap* mBitmap;
int mMeshWidth;
int mMeshHeight;
const float* mVertices;
const int* mColors;
};
class DrawPatchOp : public DrawBoundedOp {
public:
DrawPatchOp(const SkBitmap* bitmap, const Res_png_9patch* patch,
float left, float top, float right, float bottom, const SkPaint* paint)
: DrawBoundedOp(left, top, right, bottom, paint),
mBitmap(bitmap), mPatch(patch), mGenerationId(0), mMesh(NULL),
mAtlas(Caches::getInstance().assetAtlas) {
mEntry = mAtlas.getEntry(bitmap);
if (mEntry) {
mEntryGenerationId = mAtlas.getGenerationId();
}
Pack preloaded framework assets in a texture atlas When the Android runtime starts, the system preloads a series of assets in the Zygote process. These assets are shared across all processes. Unfortunately, each one of these assets is later uploaded in its own OpenGL texture, once per process. This wastes memory and generates unnecessary OpenGL state changes. This CL introduces an asset server that provides an atlas to all processes. Note: bitmaps used by skia shaders are *not* sampled from the atlas. It's an uncommon use case and would require extra texture transforms in the GL shaders. WHAT IS THE ASSETS ATLAS The "assets atlas" is a single, shareable graphic buffer that contains all the system's preloaded bitmap drawables (this includes 9-patches.) The atlas is made of two distinct objects: the graphic buffer that contains the actual pixels and the map which indicates where each preloaded bitmap can be found in the atlas (essentially a pair of x and y coordinates.) HOW IS THE ASSETS ATLAS GENERATED Because we need to support a wide variety of devices and because it is easy to change the list of preloaded drawables, the atlas is generated at runtime, during the startup phase of the system process. There are several steps that lead to the atlas generation: 1. If the device is booting for the first time, or if the device was updated, we need to find the best atlas configuration. To do so, the atlas service tries a number of width, height and algorithm variations that allows us to pack as many assets as possible while using as little memory as possible. Once a best configuration is found, it gets written to disk in /data/system/framework_atlas 2. Given a best configuration (algorithm variant, dimensions and number of bitmaps that can be packed in the atlas), the atlas service packs all the preloaded bitmaps into a single graphic buffer object. 3. The packing is done using Skia in a temporary native bitmap. The Skia bitmap is then copied into the graphic buffer using OpenGL ES to benefit from texture swizzling. HOW PROCESSES USE THE ATLAS Whenever a process' hardware renderer initializes its EGL context, it queries the atlas service for the graphic buffer and the map. It is important to remember that both the context and the map will be valid for the lifetime of the hardware renderer (if the system process goes down, all apps get killed as well.) Every time the hardware renderer needs to render a bitmap, it first checks whether the bitmap can be found in the assets atlas. When the bitmap is part of the atlas, texture coordinates are remapped appropriately before rendering. Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
2013-04-17 18:54:38 -07:00
};
AssetAtlas::Entry* getAtlasEntry() {
// The atlas entry is stale, let's get a new one
if (mEntry && mEntryGenerationId != mAtlas.getGenerationId()) {
mEntryGenerationId = mAtlas.getGenerationId();
mEntry = mAtlas.getEntry(mBitmap);
}
return mEntry;
}
const Patch* getMesh(OpenGLRenderer& renderer) {
if (!mMesh || renderer.getCaches().patchCache.getGenerationId() != mGenerationId) {
PatchCache& cache = renderer.getCaches().patchCache;
mMesh = cache.get(getAtlasEntry(), mBitmap->width(), mBitmap->height(),
mLocalBounds.getWidth(), mLocalBounds.getHeight(), mPatch);
mGenerationId = cache.getGenerationId();
}
return mMesh;
}
/**
* This multi-draw operation builds an indexed mesh on the stack by copying
* and transforming the vertices of each 9-patch in the batch. This method
* is also responsible for dirtying the current layer, if any.
*/
virtual status_t multiDraw(OpenGLRenderer& renderer, Rect& dirty,
const Vector<OpStatePair>& ops, const Rect& bounds) {
const DeferredDisplayState& firstState = *(ops[0].state);
renderer.restoreDisplayState(firstState, true); // restore all but the clip
// Batches will usually contain a small number of items so it's
// worth performing a first iteration to count the exact number
// of vertices we need in the new mesh
uint32_t totalVertices = 0;
for (unsigned int i = 0; i < ops.size(); i++) {
totalVertices += ((DrawPatchOp*) ops[i].op)->getMesh(renderer)->verticesCount;
}
const bool hasLayer = renderer.hasLayer();
uint32_t indexCount = 0;
TextureVertex vertices[totalVertices];
TextureVertex* vertex = &vertices[0];
// Create a mesh that contains the transformed vertices for all the
// 9-patch objects that are part of the batch. Note that onDefer()
// enforces ops drawn by this function to have a pure translate or
// identity matrix
for (unsigned int i = 0; i < ops.size(); i++) {
DrawPatchOp* patchOp = (DrawPatchOp*) ops[i].op;
const DeferredDisplayState* state = ops[i].state;
const Patch* opMesh = patchOp->getMesh(renderer);
uint32_t vertexCount = opMesh->verticesCount;
if (vertexCount == 0) continue;
// We use the bounds to know where to translate our vertices
// Using patchOp->state.mBounds wouldn't work because these
// bounds are clipped
const float tx = (int) floorf(state->mMatrix.getTranslateX() +
patchOp->mLocalBounds.left + 0.5f);
const float ty = (int) floorf(state->mMatrix.getTranslateY() +
patchOp->mLocalBounds.top + 0.5f);
// Copy & transform all the vertices for the current operation
TextureVertex* opVertices = opMesh->vertices;
for (uint32_t j = 0; j < vertexCount; j++, opVertices++) {
TextureVertex::set(vertex++,
opVertices->x + tx, opVertices->y + ty,
opVertices->u, opVertices->v);
}
// Dirty the current layer if possible. When the 9-patch does not
// contain empty quads we can take a shortcut and simply set the
// dirty rect to the object's bounds.
if (hasLayer) {
if (!opMesh->hasEmptyQuads) {
renderer.dirtyLayer(tx, ty,
tx + patchOp->mLocalBounds.getWidth(),
ty + patchOp->mLocalBounds.getHeight());
} else {
const size_t count = opMesh->quads.size();
for (size_t i = 0; i < count; i++) {
const Rect& quadBounds = opMesh->quads[i];
const float x = tx + quadBounds.left;
const float y = ty + quadBounds.top;
renderer.dirtyLayer(x, y,
x + quadBounds.getWidth(), y + quadBounds.getHeight());
}
}
}
indexCount += opMesh->indexCount;
}
return renderer.drawPatches(mBitmap, getAtlasEntry(),
&vertices[0], indexCount, getPaint(renderer));
}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
// We're not calling the public variant of drawPatch() here
// This method won't perform the quickReject() since we've already done it at this point
return renderer.drawPatch(mBitmap, getMesh(renderer), getAtlasEntry(),
mLocalBounds.left, mLocalBounds.top, mLocalBounds.right, mLocalBounds.bottom,
getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw patch " RECT_STRING, RECT_ARGS(mLocalBounds));
}
virtual const char* name() { return "DrawPatch"; }
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
deferInfo.batchId = DeferredDisplayList::kOpBatch_Patch;
deferInfo.mergeId = getAtlasEntry() ? (mergeid_t) mEntry->getMergeId() : (mergeid_t) mBitmap;
deferInfo.mergeable = state.mMatrix.isPureTranslate() &&
OpenGLRenderer::getXfermodeDirect(mPaint) == SkXfermode::kSrcOver_Mode;
deferInfo.opaqueOverBounds = isOpaqueOverBounds(state) && mBitmap->isOpaque();
}
private:
const SkBitmap* mBitmap;
const Res_png_9patch* mPatch;
uint32_t mGenerationId;
const Patch* mMesh;
const AssetAtlas& mAtlas;
uint32_t mEntryGenerationId;
Pack preloaded framework assets in a texture atlas When the Android runtime starts, the system preloads a series of assets in the Zygote process. These assets are shared across all processes. Unfortunately, each one of these assets is later uploaded in its own OpenGL texture, once per process. This wastes memory and generates unnecessary OpenGL state changes. This CL introduces an asset server that provides an atlas to all processes. Note: bitmaps used by skia shaders are *not* sampled from the atlas. It's an uncommon use case and would require extra texture transforms in the GL shaders. WHAT IS THE ASSETS ATLAS The "assets atlas" is a single, shareable graphic buffer that contains all the system's preloaded bitmap drawables (this includes 9-patches.) The atlas is made of two distinct objects: the graphic buffer that contains the actual pixels and the map which indicates where each preloaded bitmap can be found in the atlas (essentially a pair of x and y coordinates.) HOW IS THE ASSETS ATLAS GENERATED Because we need to support a wide variety of devices and because it is easy to change the list of preloaded drawables, the atlas is generated at runtime, during the startup phase of the system process. There are several steps that lead to the atlas generation: 1. If the device is booting for the first time, or if the device was updated, we need to find the best atlas configuration. To do so, the atlas service tries a number of width, height and algorithm variations that allows us to pack as many assets as possible while using as little memory as possible. Once a best configuration is found, it gets written to disk in /data/system/framework_atlas 2. Given a best configuration (algorithm variant, dimensions and number of bitmaps that can be packed in the atlas), the atlas service packs all the preloaded bitmaps into a single graphic buffer object. 3. The packing is done using Skia in a temporary native bitmap. The Skia bitmap is then copied into the graphic buffer using OpenGL ES to benefit from texture swizzling. HOW PROCESSES USE THE ATLAS Whenever a process' hardware renderer initializes its EGL context, it queries the atlas service for the graphic buffer and the map. It is important to remember that both the context and the map will be valid for the lifetime of the hardware renderer (if the system process goes down, all apps get killed as well.) Every time the hardware renderer needs to render a bitmap, it first checks whether the bitmap can be found in the assets atlas. When the bitmap is part of the atlas, texture coordinates are remapped appropriately before rendering. Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
2013-04-17 18:54:38 -07:00
AssetAtlas::Entry* mEntry;
};
class DrawColorOp : public DrawOp {
public:
DrawColorOp(int color, SkXfermode::Mode mode)
: DrawOp(NULL), mColor(color), mMode(mode) {};
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawColor(mColor, mMode);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw color %#x, mode %d", mColor, mMode);
}
virtual const char* name() { return "DrawColor"; }
private:
int mColor;
SkXfermode::Mode mMode;
};
class DrawStrokableOp : public DrawBoundedOp {
public:
DrawStrokableOp(float left, float top, float right, float bottom, const SkPaint* paint)
: DrawBoundedOp(left, top, right, bottom, paint) {};
bool getLocalBounds(const DrawModifiers& drawModifiers, Rect& localBounds) {
localBounds.set(mLocalBounds);
if (mPaint && mPaint->getStyle() != SkPaint::kFill_Style) {
localBounds.outset(strokeWidthOutset());
}
return true;
}
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
if (mPaint->getPathEffect()) {
deferInfo.batchId = DeferredDisplayList::kOpBatch_AlphaMaskTexture;
} else {
deferInfo.batchId = mPaint->isAntiAlias() ?
DeferredDisplayList::kOpBatch_AlphaVertices :
DeferredDisplayList::kOpBatch_Vertices;
}
}
};
class DrawRectOp : public DrawStrokableOp {
public:
DrawRectOp(float left, float top, float right, float bottom, const SkPaint* paint)
: DrawStrokableOp(left, top, right, bottom, paint) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawRect(mLocalBounds.left, mLocalBounds.top,
mLocalBounds.right, mLocalBounds.bottom, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Rect " RECT_STRING, RECT_ARGS(mLocalBounds));
}
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
DrawStrokableOp::onDefer(renderer, deferInfo, state);
deferInfo.opaqueOverBounds = isOpaqueOverBounds(state) &&
mPaint->getStyle() == SkPaint::kFill_Style;
}
virtual const char* name() { return "DrawRect"; }
};
class DrawRectsOp : public DrawBoundedOp {
public:
DrawRectsOp(const float* rects, int count, const SkPaint* paint)
: DrawBoundedOp(rects, count, paint),
mRects(rects), mCount(count) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawRects(mRects, mCount, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Rects count %d", mCount);
}
virtual const char* name() { return "DrawRects"; }
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
deferInfo.batchId = DeferredDisplayList::kOpBatch_Vertices;
}
private:
const float* mRects;
int mCount;
};
class DrawRoundRectOp : public DrawStrokableOp {
public:
DrawRoundRectOp(float left, float top, float right, float bottom,
float rx, float ry, const SkPaint* paint)
: DrawStrokableOp(left, top, right, bottom, paint), mRx(rx), mRy(ry) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawRoundRect(mLocalBounds.left, mLocalBounds.top,
mLocalBounds.right, mLocalBounds.bottom, mRx, mRy, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw RoundRect " RECT_STRING ", rx %f, ry %f", RECT_ARGS(mLocalBounds), mRx, mRy);
}
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
DrawStrokableOp::onDefer(renderer, deferInfo, state);
if (!mPaint->getPathEffect()) {
renderer.getCaches().tessellationCache.precacheRoundRect(state.mMatrix, *mPaint,
mLocalBounds.getWidth(), mLocalBounds.getHeight(), mRx, mRy);
}
}
virtual const char* name() { return "DrawRoundRect"; }
private:
float mRx;
float mRy;
};
class DrawCircleOp : public DrawStrokableOp {
public:
DrawCircleOp(float x, float y, float radius, const SkPaint* paint)
: DrawStrokableOp(x - radius, y - radius, x + radius, y + radius, paint),
mX(x), mY(y), mRadius(radius) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawCircle(mX, mY, mRadius, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Circle x %f, y %f, r %f", mX, mY, mRadius);
}
virtual const char* name() { return "DrawCircle"; }
private:
float mX;
float mY;
float mRadius;
};
class DrawCirclePropsOp : public DrawOp {
public:
DrawCirclePropsOp(float* x, float* y, float* radius, const SkPaint* paint)
: DrawOp(paint), mX(x), mY(y), mRadius(radius) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawCircle(*mX, *mY, *mRadius, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Circle Props x %p, y %p, r %p", mX, mY, mRadius);
}
virtual const char* name() { return "DrawCircleProps"; }
private:
float* mX;
float* mY;
float* mRadius;
};
class DrawOvalOp : public DrawStrokableOp {
public:
DrawOvalOp(float left, float top, float right, float bottom, const SkPaint* paint)
: DrawStrokableOp(left, top, right, bottom, paint) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawOval(mLocalBounds.left, mLocalBounds.top,
mLocalBounds.right, mLocalBounds.bottom, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Oval " RECT_STRING, RECT_ARGS(mLocalBounds));
}
virtual const char* name() { return "DrawOval"; }
};
class DrawArcOp : public DrawStrokableOp {
public:
DrawArcOp(float left, float top, float right, float bottom,
float startAngle, float sweepAngle, bool useCenter, const SkPaint* paint)
: DrawStrokableOp(left, top, right, bottom, paint),
mStartAngle(startAngle), mSweepAngle(sweepAngle), mUseCenter(useCenter) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawArc(mLocalBounds.left, mLocalBounds.top,
mLocalBounds.right, mLocalBounds.bottom,
mStartAngle, mSweepAngle, mUseCenter, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Arc " RECT_STRING ", start %f, sweep %f, useCenter %d",
RECT_ARGS(mLocalBounds), mStartAngle, mSweepAngle, mUseCenter);
}
virtual const char* name() { return "DrawArc"; }
private:
float mStartAngle;
float mSweepAngle;
bool mUseCenter;
};
class DrawPathOp : public DrawBoundedOp {
public:
DrawPathOp(const SkPath* path, const SkPaint* paint)
: DrawBoundedOp(paint), mPath(path) {
float left, top, offset;
uint32_t width, height;
PathCache::computePathBounds(path, paint, left, top, offset, width, height);
left -= offset;
top -= offset;
mLocalBounds.set(left, top, left + width, top + height);
}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawPath(mPath, getPaint(renderer));
}
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
const SkPaint* paint = getPaint(renderer);
renderer.getCaches().pathCache.precache(mPath, paint);
deferInfo.batchId = DeferredDisplayList::kOpBatch_AlphaMaskTexture;
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Path %p in " RECT_STRING, mPath, RECT_ARGS(mLocalBounds));
}
virtual const char* name() { return "DrawPath"; }
private:
const SkPath* mPath;
};
class DrawLinesOp : public DrawBoundedOp {
public:
DrawLinesOp(const float* points, int count, const SkPaint* paint)
: DrawBoundedOp(points, count, paint),
mPoints(points), mCount(count) {
mLocalBounds.outset(strokeWidthOutset());
}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawLines(mPoints, mCount, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Lines count %d", mCount);
}
virtual const char* name() { return "DrawLines"; }
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
deferInfo.batchId = mPaint->isAntiAlias() ?
DeferredDisplayList::kOpBatch_AlphaVertices :
DeferredDisplayList::kOpBatch_Vertices;
}
protected:
const float* mPoints;
int mCount;
};
class DrawPointsOp : public DrawLinesOp {
public:
DrawPointsOp(const float* points, int count, const SkPaint* paint)
: DrawLinesOp(points, count, paint) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawPoints(mPoints, mCount, getPaint(renderer));
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Points count %d", mCount);
}
virtual const char* name() { return "DrawPoints"; }
};
class DrawSomeTextOp : public DrawOp {
public:
DrawSomeTextOp(const char* text, int bytesCount, int count, const SkPaint* paint)
: DrawOp(paint), mText(text), mBytesCount(bytesCount), mCount(count) {};
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw some text, %d bytes", mBytesCount);
}
virtual bool hasTextShadow() const {
return OpenGLRenderer::hasTextShadow(mPaint);
}
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
const SkPaint* paint = getPaint(renderer);
FontRenderer& fontRenderer = renderer.getCaches().fontRenderer->getFontRenderer(paint);
fontRenderer.precache(paint, mText, mCount, mat4::identity());
deferInfo.batchId = mPaint->getColor() == 0xff000000 ?
DeferredDisplayList::kOpBatch_Text :
DeferredDisplayList::kOpBatch_ColorText;
}
protected:
const char* mText;
int mBytesCount;
int mCount;
};
class DrawTextOnPathOp : public DrawSomeTextOp {
public:
DrawTextOnPathOp(const char* text, int bytesCount, int count,
const SkPath* path, float hOffset, float vOffset, const SkPaint* paint)
: DrawSomeTextOp(text, bytesCount, count, paint),
mPath(path), mHOffset(hOffset), mVOffset(vOffset) {
/* TODO: inherit from DrawBounded and init mLocalBounds */
}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawTextOnPath(mText, mBytesCount, mCount, mPath,
mHOffset, mVOffset, getPaint(renderer));
}
virtual const char* name() { return "DrawTextOnPath"; }
private:
const SkPath* mPath;
float mHOffset;
float mVOffset;
};
class DrawPosTextOp : public DrawSomeTextOp {
public:
DrawPosTextOp(const char* text, int bytesCount, int count,
const float* positions, const SkPaint* paint)
: DrawSomeTextOp(text, bytesCount, count, paint), mPositions(positions) {
/* TODO: inherit from DrawBounded and init mLocalBounds */
}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawPosText(mText, mBytesCount, mCount, mPositions, getPaint(renderer));
}
virtual const char* name() { return "DrawPosText"; }
private:
const float* mPositions;
};
class DrawTextOp : public DrawBoundedOp {
public:
DrawTextOp(const char* text, int bytesCount, int count, float x, float y,
const float* positions, const SkPaint* paint, float totalAdvance, const Rect& bounds)
: DrawBoundedOp(bounds, paint), mText(text), mBytesCount(bytesCount), mCount(count),
mX(x), mY(y), mPositions(positions), mTotalAdvance(totalAdvance) {
memset(&mPrecacheTransform.data[0], 0xff, 16 * sizeof(float));
}
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
const SkPaint* paint = getPaint(renderer);
FontRenderer& fontRenderer = renderer.getCaches().fontRenderer->getFontRenderer(paint);
const mat4& transform = renderer.findBestFontTransform(state.mMatrix);
if (mPrecacheTransform != transform) {
fontRenderer.precache(paint, mText, mCount, transform);
mPrecacheTransform = transform;
}
deferInfo.batchId = mPaint->getColor() == 0xff000000 ?
DeferredDisplayList::kOpBatch_Text :
DeferredDisplayList::kOpBatch_ColorText;
deferInfo.mergeId = reinterpret_cast<mergeid_t>(mPaint->getColor());
// don't merge decorated text - the decorations won't draw in order
bool noDecorations = !(mPaint->getFlags() & (SkPaint::kUnderlineText_Flag |
SkPaint::kStrikeThruText_Flag));
deferInfo.mergeable = state.mMatrix.isPureTranslate() && noDecorations &&
OpenGLRenderer::getXfermodeDirect(mPaint) == SkXfermode::kSrcOver_Mode;
}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
Rect bounds;
getLocalBounds(renderer.getDrawModifiers(), bounds);
return renderer.drawText(mText, mBytesCount, mCount, mX, mY,
mPositions, getPaint(renderer), mTotalAdvance, bounds);
}
virtual status_t multiDraw(OpenGLRenderer& renderer, Rect& dirty,
const Vector<OpStatePair>& ops, const Rect& bounds) {
status_t status = DrawGlInfo::kStatusDone;
for (unsigned int i = 0; i < ops.size(); i++) {
const DeferredDisplayState& state = *(ops[i].state);
DrawOpMode drawOpMode = (i == ops.size() - 1) ? kDrawOpMode_Flush : kDrawOpMode_Defer;
renderer.restoreDisplayState(state, true); // restore all but the clip
DrawTextOp& op = *((DrawTextOp*)ops[i].op);
// quickReject() will not occure in drawText() so we can use mLocalBounds
// directly, we do not need to account for shadow by calling getLocalBounds()
status |= renderer.drawText(op.mText, op.mBytesCount, op.mCount, op.mX, op.mY,
op.mPositions, op.getPaint(renderer), op.mTotalAdvance, op.mLocalBounds,
drawOpMode);
}
return status;
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Text of count %d, bytes %d", mCount, mBytesCount);
}
virtual const char* name() { return "DrawText"; }
private:
const char* mText;
int mBytesCount;
int mCount;
float mX;
float mY;
const float* mPositions;
float mTotalAdvance;
mat4 mPrecacheTransform;
};
///////////////////////////////////////////////////////////////////////////////
// SPECIAL DRAW OPERATIONS
///////////////////////////////////////////////////////////////////////////////
class DrawFunctorOp : public DrawOp {
public:
DrawFunctorOp(Functor* functor)
: DrawOp(NULL), mFunctor(functor) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
renderer.startMark("GL functor");
status_t ret = renderer.callDrawGLFunction(mFunctor, dirty);
renderer.endMark();
return ret;
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Functor %p", mFunctor);
}
virtual const char* name() { return "DrawFunctor"; }
private:
Functor* mFunctor;
};
class DrawRenderNodeOp : public DrawBoundedOp {
friend class RenderNode; // grant RenderNode access to info of child
public:
DrawRenderNodeOp(RenderNode* renderNode, int flags, const mat4& transformFromParent)
: DrawBoundedOp(0, 0, renderNode->getWidth(), renderNode->getHeight(), 0),
mRenderNode(renderNode), mFlags(flags), mTransformFromParent(transformFromParent) {}
virtual void defer(DeferStateStruct& deferStruct, int saveCount, int level,
bool useQuickReject) {
if (mRenderNode && mRenderNode->isRenderable() && !mSkipInOrderDraw) {
mRenderNode->defer(deferStruct, level + 1);
}
}
virtual void replay(ReplayStateStruct& replayStruct, int saveCount, int level,
bool useQuickReject) {
if (mRenderNode && mRenderNode->isRenderable() && !mSkipInOrderDraw) {
mRenderNode->replay(replayStruct, level + 1);
}
}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
LOG_ALWAYS_FATAL("should not be called, because replay() is overridden");
return 0;
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Display List %p, flags %#x", mRenderNode, mFlags);
if (mRenderNode && (logFlags & kOpLogFlag_Recurse)) {
mRenderNode->output(level + 1);
}
}
virtual const char* name() { return "DrawRenderNode"; }
RenderNode* renderNode() { return mRenderNode; }
private:
RenderNode* mRenderNode;
const int mFlags;
///////////////////////////
// Properties below are used by RenderNode::computeOrderingImpl() and issueOperations()
///////////////////////////
/**
* Records transform vs parent, used for computing total transform without rerunning DL contents
*/
const mat4 mTransformFromParent;
/**
* Holds the transformation between the projection surface ViewGroup and this RenderNode
* drawing instance. Represents any translations / transformations done within the drawing of
* the compositing ancestor ViewGroup's draw, before the draw of the View represented by this
* DisplayList draw instance.
*
* Note: doesn't include transformation within the RenderNode, or its properties.
*/
mat4 mTransformFromCompositingAncestor;
bool mSkipInOrderDraw;
};
/**
* Not a canvas operation, used only by 3d / z ordering logic in RenderNode::iterate()
*/
class DrawShadowOp : public DrawOp {
public:
DrawShadowOp(const mat4& transformXY, const mat4& transformZ,
float casterAlpha, bool casterUnclipped,
const SkPath* casterOutline, const SkPath* revealClip)
: DrawOp(NULL), mTransformXY(transformXY), mTransformZ(transformZ),
mCasterAlpha(casterAlpha), mCasterUnclipped(casterUnclipped) {
mOutline = *casterOutline;
if (revealClip) {
// intersect the outline with the convex reveal clip
Op(mOutline, *revealClip, kIntersect_PathOp, &mOutline);
}
}
virtual void onDefer(OpenGLRenderer& renderer, DeferInfo& deferInfo,
const DeferredDisplayState& state) {
renderer.getCaches().tessellationCache.precacheShadows(&state.mMatrix,
renderer.getLocalClipBounds(), isCasterOpaque(), &mOutline,
&mTransformXY, &mTransformZ, renderer.getLightCenter(), renderer.getLightRadius());
}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
TessellationCache::vertexBuffer_pair_t buffers;
Matrix4 drawTransform;
renderer.getMatrix(&drawTransform);
renderer.getCaches().tessellationCache.getShadowBuffers(&drawTransform,
renderer.getLocalClipBounds(), isCasterOpaque(), &mOutline,
&mTransformXY, &mTransformZ, renderer.getLightCenter(), renderer.getLightRadius(),
buffers);
return renderer.drawShadow(mCasterAlpha, buffers.first, buffers.second);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOGS("DrawShadow");
}
virtual const char* name() { return "DrawShadow"; }
private:
bool isCasterOpaque() { return mCasterAlpha >= 1.0f && mCasterUnclipped; }
const mat4 mTransformXY;
const mat4 mTransformZ;
const float mCasterAlpha;
const bool mCasterUnclipped;
SkPath mOutline;
};
class DrawLayerOp : public DrawOp {
public:
DrawLayerOp(Layer* layer, float x, float y)
: DrawOp(NULL), mLayer(layer), mX(x), mY(y) {}
virtual status_t applyDraw(OpenGLRenderer& renderer, Rect& dirty) {
return renderer.drawLayer(mLayer, mX, mY);
}
virtual void output(int level, uint32_t logFlags) const {
OP_LOG("Draw Layer %p at %f %f", mLayer, mX, mY);
}
virtual const char* name() { return "DrawLayer"; }
private:
Layer* mLayer;
float mX;
float mY;
};
}; // namespace uirenderer
}; // namespace android
#endif // ANDROID_HWUI_DISPLAY_OPERATION_H