blob: c11c32b38db334072e821b4007e0e4687581f24b [file] [log] [blame]
/*
* Copyright 2015 Google Inc.
*
* Use of this source code is governed by a BSD-style license that can be
* found in the LICENSE file.
*/
#include "GrBatchAtlas.h"
#include "GrBatchTarget.h"
#include "GrGpu.h"
#include "GrRectanizer.h"
#include "GrTracing.h"
#include "GrVertexBuffer.h"
static inline void adjust_for_offset(SkIPoint16* loc, const SkIPoint16& offset) {
loc->fX += offset.fX;
loc->fY += offset.fY;
}
static GrBatchAtlas::AtlasID create_id(int index, int generation) {
// Generation ID can roll over because we only check for equality
SkASSERT(index < (1 << 16));
return generation << 16 | index;
}
// The backing GrTexture for a GrBatchAtlas is broken into a spatial grid of GrBatchPlots.
// The GrBatchPlots keep track of subimage placement via their GrRectanizer. In turn, a GrBatchPlot
// manages the lifetime of its data using two tokens, a last ref toke and a last upload token.
// Once a GrBatchPlot is "full" (i.e. there is no room for the new subimage according to the
// GrRectanizer), it can no longer be used unless the last ref on the GrPlot has already been
// flushed through to the gpu.
class BatchPlot : public SkRefCnt {
public:
typedef GrBatchAtlas::BatchToken BatchToken;
SK_DECLARE_INTERNAL_LLIST_INTERFACE(BatchPlot);
// index() refers to the index of the plot in the owning GrAtlas's plot array. genID() is a
// monotonically incrementing number which is bumped every time the cpu backing store is
// wiped, or when the plot itself is evicted from the atlas(ie, there is continuity in genID()
// across atlas spills)
int index() const { return fIndex; }
int genID() const { return fGenID; }
GrBatchAtlas::AtlasID id() { return fID; }
GrTexture* texture() const { return fTexture; }
bool addSubImage(int width, int height, const void* image, SkIPoint16* loc, size_t rowBytes) {
if (!fRects->addRect(width, height, loc)) {
return false;
}
if (!fData) {
fData = reinterpret_cast<unsigned char*>(sk_calloc_throw(fBytesPerPixel * fWidth *
fHeight));
}
const unsigned char* imagePtr = (const unsigned char*)image;
// point ourselves at the right starting spot
unsigned char* dataPtr = fData;
dataPtr += fBytesPerPixel * fWidth * loc->fY;
dataPtr += fBytesPerPixel * loc->fX;
// copy into the data buffer
for (int i = 0; i < height; ++i) {
memcpy(dataPtr, imagePtr, rowBytes);
dataPtr += fBytesPerPixel * fWidth;
imagePtr += rowBytes;
}
fDirtyRect.join(loc->fX, loc->fY, loc->fX + width, loc->fY + height);
adjust_for_offset(loc, fOffset);
SkDEBUGCODE(fDirty = true;)
return true;
}
// to manage the lifetime of a plot, we use two tokens. We use last upload token to know when
// we can 'piggy back' uploads, ie if the last upload hasn't been flushed to gpu, we don't need
// to issue a new upload even if we update the cpu backing store. We use lastref to determine
// when we can evict a plot from the cache, ie if the last ref has already flushed through
// the gpu then we can reuse the plot
BatchToken lastUploadToken() const { return fLastUpload; }
BatchToken lastUseToken() const { return fLastUse; }
void setLastUploadToken(BatchToken batchToken) {
SkASSERT(batchToken >= fLastUpload);
fLastUpload = batchToken;
}
void setLastUseToken(BatchToken batchToken) {
SkASSERT(batchToken >= fLastUse);
fLastUse = batchToken;
}
void uploadToTexture(GrBatchTarget::TextureUploader uploader) {
// We should only be issuing uploads if we are in fact dirty
SkASSERT(fDirty && fData && fTexture);
TRACE_EVENT0(TRACE_DISABLED_BY_DEFAULT("skia.gpu"), "GrBatchPlot::uploadToTexture");
size_t rowBytes = fBytesPerPixel * fRects->width();
const unsigned char* dataPtr = fData;
dataPtr += rowBytes * fDirtyRect.fTop;
dataPtr += fBytesPerPixel * fDirtyRect.fLeft;
uploader.writeTexturePixels(fTexture,
fOffset.fX + fDirtyRect.fLeft, fOffset.fY + fDirtyRect.fTop,
fDirtyRect.width(), fDirtyRect.height(),
fTexture->config(), dataPtr, rowBytes);
fDirtyRect.setEmpty();
SkDEBUGCODE(fDirty = false;)
}
void resetRects() {
SkASSERT(fRects);
fRects->reset();
fGenID++;
fID = create_id(fIndex, fGenID);
// zero out the plot
if (fData) {
sk_bzero(fData, fBytesPerPixel * fWidth * fHeight);
}
fDirtyRect.setEmpty();
SkDEBUGCODE(fDirty = false;)
}
int x() const { return fX; }
int y() const { return fY; }
private:
BatchPlot()
: fLastUpload(0)
, fLastUse(0)
, fIndex(-1)
, fGenID(-1)
, fID(0)
, fData(NULL)
, fWidth(0)
, fHeight(0)
, fX(0)
, fY(0)
, fTexture(NULL)
, fRects(NULL)
, fAtlas(NULL)
, fBytesPerPixel(1)
#ifdef SK_DEBUG
, fDirty(false)
#endif
{
fOffset.set(0, 0);
}
~BatchPlot() {
sk_free(fData);
fData = NULL;
delete fRects;
}
void init(GrBatchAtlas* atlas, GrTexture* texture, int index, uint32_t generation,
int offX, int offY, int width, int height, size_t bpp) {
fIndex = index;
fGenID = generation;
fID = create_id(index, generation);
fWidth = width;
fHeight = height;
fX = offX;
fY = offY;
fRects = GrRectanizer::Factory(width, height);
fAtlas = atlas;
fOffset.set(offX * width, offY * height);
fBytesPerPixel = bpp;
fData = NULL;
fDirtyRect.setEmpty();
SkDEBUGCODE(fDirty = false;)
fTexture = texture;
}
BatchToken fLastUpload;
BatchToken fLastUse;
uint32_t fIndex;
uint32_t fGenID;
GrBatchAtlas::AtlasID fID;
unsigned char* fData;
int fWidth;
int fHeight;
int fX;
int fY;
GrTexture* fTexture;
GrRectanizer* fRects;
GrBatchAtlas* fAtlas;
SkIPoint16 fOffset; // the offset of the plot in the backing texture
size_t fBytesPerPixel;
SkIRect fDirtyRect;
SkDEBUGCODE(bool fDirty;)
friend class GrBatchAtlas;
typedef SkRefCnt INHERITED;
};
////////////////////////////////////////////////////////////////////////////////
class GrPlotUploader : public GrBatchTarget::Uploader {
public:
GrPlotUploader(BatchPlot* plot)
: INHERITED(plot->lastUploadToken())
, fPlot(SkRef(plot)) {
SkASSERT(plot);
}
void upload(GrBatchTarget::TextureUploader uploader) override {
fPlot->uploadToTexture(uploader);
}
private:
SkAutoTUnref<BatchPlot> fPlot;
typedef GrBatchTarget::Uploader INHERITED;
};
///////////////////////////////////////////////////////////////////////////////
GrBatchAtlas::GrBatchAtlas(GrTexture* texture, int numPlotsX, int numPlotsY)
: fTexture(texture)
, fNumPlotsX(numPlotsX)
, fNumPlotsY(numPlotsY)
, fPlotWidth(texture->width() / numPlotsX)
, fPlotHeight(texture->height() / numPlotsY)
, fAtlasGeneration(kInvalidAtlasGeneration + 1) {
SkASSERT(fNumPlotsX * fNumPlotsY <= BulkUseTokenUpdater::kMaxPlots);
SkASSERT(fPlotWidth * fNumPlotsX == texture->width());
SkASSERT(fPlotHeight * fNumPlotsY == texture->height());
// We currently do not support compressed atlases...
SkASSERT(!GrPixelConfigIsCompressed(texture->desc().fConfig));
// set up allocated plots
fBPP = GrBytesPerPixel(texture->desc().fConfig);
fPlotArray = SkNEW_ARRAY(SkAutoTUnref<BatchPlot>, (fNumPlotsX * fNumPlotsY));
SkAutoTUnref<BatchPlot>* currPlot = fPlotArray;
for (int y = fNumPlotsY - 1, r = 0; y >= 0; --y, ++r) {
for (int x = fNumPlotsX - 1, c = 0; x >= 0; --x, ++c) {
int id = r * fNumPlotsX + c;
currPlot->reset(SkNEW(BatchPlot));
(*currPlot)->init(this, texture, id, 1, x, y, fPlotWidth, fPlotHeight, fBPP);
// build LRU list
fPlotList.addToHead(currPlot->get());
++currPlot;
}
}
}
GrBatchAtlas::~GrBatchAtlas() {
SkSafeUnref(fTexture);
SkDELETE_ARRAY(fPlotArray);
}
void GrBatchAtlas::processEviction(AtlasID id) {
for (int i = 0; i < fEvictionCallbacks.count(); i++) {
(*fEvictionCallbacks[i].fFunc)(id, fEvictionCallbacks[i].fData);
}
}
void GrBatchAtlas::makeMRU(BatchPlot* plot) {
if (fPlotList.head() == plot) {
return;
}
fPlotList.remove(plot);
fPlotList.addToHead(plot);
}
inline void GrBatchAtlas::updatePlot(GrBatchTarget* batchTarget, AtlasID* id, BatchPlot* plot) {
this->makeMRU(plot);
// If our most recent upload has already occurred then we have to insert a new
// upload. Otherwise, we already have a scheduled upload that hasn't yet ocurred.
// This new update will piggy back on that previously scheduled update.
if (batchTarget->isIssued(plot->lastUploadToken())) {
plot->setLastUploadToken(batchTarget->asapToken());
SkAutoTUnref<GrPlotUploader> uploader(SkNEW_ARGS(GrPlotUploader, (plot)));
batchTarget->upload(uploader);
}
*id = plot->id();
}
bool GrBatchAtlas::addToAtlas(AtlasID* id, GrBatchTarget* batchTarget,
int width, int height, const void* image, SkIPoint16* loc) {
// We should already have a texture, TODO clean this up
SkASSERT(fTexture && width <= fPlotWidth && height <= fPlotHeight);
// now look through all allocated plots for one we can share, in Most Recently Refed order
GrBatchPlotList::Iter plotIter;
plotIter.init(fPlotList, GrBatchPlotList::Iter::kHead_IterStart);
BatchPlot* plot;
while ((plot = plotIter.get())) {
if (plot->addSubImage(width, height, image, loc, fBPP * width)) {
this->updatePlot(batchTarget, id, plot);
return true;
}
plotIter.next();
}
// If the above fails, then see if the least recently refed plot has already been flushed to the
// gpu
plotIter.init(fPlotList, GrBatchPlotList::Iter::kTail_IterStart);
plot = plotIter.get();
SkASSERT(plot);
if (batchTarget->isIssued(plot->lastUseToken())) {
this->processEviction(plot->id());
plot->resetRects();
SkDEBUGCODE(bool verify = )plot->addSubImage(width, height, image, loc, fBPP * width);
SkASSERT(verify);
this->updatePlot(batchTarget, id, plot);
fAtlasGeneration++;
return true;
}
// The least recently refed plot hasn't been flushed to the gpu yet, however, if we have flushed
// it to the batch target than we can reuse it. Our last ref token is guaranteed to be less
// than or equal to the current token. If its 'less than' the current token, than we can spin
// off the plot(ie let the batch target manage it) and create a new plot in its place in our
// array. If it is equal to the currentToken, then the caller has to flush draws to the batch
// target so we can spin off the plot
if (plot->lastUseToken() == batchTarget->currentToken()) {
return false;
}
// We take an extra ref here so our plot isn't deleted when we reset its index in the array.
plot->ref();
int index = plot->index();
int x = plot->x();
int y = plot->y();
int generation = plot->genID();
this->processEviction(plot->id());
fPlotList.remove(plot);
SkAutoTUnref<BatchPlot>& newPlot = fPlotArray[plot->index()];
newPlot.reset(SkNEW(BatchPlot));
newPlot->init(this, fTexture, index, ++generation, x, y, fPlotWidth, fPlotHeight, fBPP);
fPlotList.addToHead(newPlot.get());
SkDEBUGCODE(bool verify = )newPlot->addSubImage(width, height, image, loc, fBPP * width);
SkASSERT(verify);
newPlot->setLastUploadToken(batchTarget->currentToken());
SkAutoTUnref<GrPlotUploader> uploader(SkNEW_ARGS(GrPlotUploader, (newPlot)));
batchTarget->upload(uploader);
*id = newPlot->id();
plot->unref();
fAtlasGeneration++;
return true;
}
bool GrBatchAtlas::hasID(AtlasID id) {
int index = GetIndexFromID(id);
SkASSERT(index < fNumPlotsX * fNumPlotsY);
return fPlotArray[index]->genID() == GetGenerationFromID(id);
}
void GrBatchAtlas::setLastUseToken(AtlasID id, BatchToken batchToken) {
SkASSERT(this->hasID(id));
int index = GetIndexFromID(id);
SkASSERT(index < fNumPlotsX * fNumPlotsY);
this->makeMRU(fPlotArray[index]);
fPlotArray[index]->setLastUseToken(batchToken);
}
void GrBatchAtlas::setLastUseTokenBulk(const BulkUseTokenUpdater& updater, BatchToken batchToken) {
int count = updater.fPlotsToUpdate.count();
for (int i = 0; i < count; i++) {
BatchPlot* plot = fPlotArray[updater.fPlotsToUpdate[i]];
this->makeMRU(plot);
plot->setLastUseToken(batchToken);
}
}