mirror of
https://gitlab.winehq.org/wine/wine-gecko.git
synced 2024-09-13 09:24:08 -07:00
519 lines
18 KiB
C++
519 lines
18 KiB
C++
/* -*- Mode: C++; tab-width: 20; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
|
|
/* This Source Code Form is subject to the terms of the Mozilla Public
|
|
* License, v. 2.0. If a copy of the MPL was not distributed with this
|
|
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
|
|
|
|
#include "TiledContentHost.h"
|
|
#include "ThebesLayerComposite.h" // for ThebesLayerComposite
|
|
#include "mozilla/gfx/BaseSize.h" // for BaseSize
|
|
#include "mozilla/gfx/Matrix.h" // for Matrix4x4
|
|
#include "mozilla/layers/Compositor.h" // for Compositor
|
|
#include "mozilla/layers/Effects.h" // for TexturedEffect, Effect, etc
|
|
#include "mozilla/layers/TextureHostOGL.h" // for TextureHostOGL
|
|
#include "nsAString.h"
|
|
#include "nsDebug.h" // for NS_WARNING
|
|
#include "nsPoint.h" // for nsIntPoint
|
|
#include "nsPrintfCString.h" // for nsPrintfCString
|
|
#include "nsRect.h" // for nsIntRect
|
|
#include "nsSize.h" // for nsIntSize
|
|
#include "mozilla/layers/TiledContentClient.h"
|
|
|
|
class gfxReusableSurfaceWrapper;
|
|
|
|
namespace mozilla {
|
|
using namespace gfx;
|
|
namespace layers {
|
|
|
|
class Layer;
|
|
|
|
TiledLayerBufferComposite::TiledLayerBufferComposite()
|
|
: mFrameResolution(1.0)
|
|
, mHasDoubleBufferedTiles(false)
|
|
, mUninitialized(true)
|
|
{}
|
|
|
|
/* static */ void
|
|
TiledLayerBufferComposite::RecycleCallback(TextureHost* textureHost, void* aClosure)
|
|
{
|
|
textureHost->CompositorRecycle();
|
|
}
|
|
|
|
TiledLayerBufferComposite::TiledLayerBufferComposite(ISurfaceAllocator* aAllocator,
|
|
const SurfaceDescriptorTiles& aDescriptor,
|
|
const nsIntRegion& aOldPaintedRegion)
|
|
{
|
|
mUninitialized = false;
|
|
mHasDoubleBufferedTiles = false;
|
|
mValidRegion = aDescriptor.validRegion();
|
|
mPaintedRegion = aDescriptor.paintedRegion();
|
|
mRetainedWidth = aDescriptor.retainedWidth();
|
|
mRetainedHeight = aDescriptor.retainedHeight();
|
|
mResolution = aDescriptor.resolution();
|
|
mFrameResolution = CSSToParentLayerScale(aDescriptor.frameResolution());
|
|
|
|
// Combine any valid content that wasn't already uploaded
|
|
nsIntRegion oldPaintedRegion(aOldPaintedRegion);
|
|
oldPaintedRegion.And(oldPaintedRegion, mValidRegion);
|
|
mPaintedRegion.Or(mPaintedRegion, oldPaintedRegion);
|
|
|
|
const InfallibleTArray<TileDescriptor>& tiles = aDescriptor.tiles();
|
|
for(size_t i = 0; i < tiles.Length(); i++) {
|
|
RefPtr<TextureHost> texture;
|
|
const TileDescriptor& tileDesc = tiles[i];
|
|
switch (tileDesc.type()) {
|
|
case TileDescriptor::TTexturedTileDescriptor : {
|
|
texture = TextureHost::AsTextureHost(tileDesc.get_TexturedTileDescriptor().textureParent());
|
|
#if defined(MOZ_WIDGET_GONK) && ANDROID_VERSION >= 17
|
|
if (!gfxPrefs::LayersUseSimpleTiles()) {
|
|
texture->SetRecycleCallback(RecycleCallback, nullptr);
|
|
}
|
|
#endif
|
|
const TileLock& ipcLock = tileDesc.get_TexturedTileDescriptor().sharedLock();
|
|
nsRefPtr<gfxSharedReadLock> sharedLock;
|
|
if (ipcLock.type() == TileLock::TShmemSection) {
|
|
sharedLock = gfxShmSharedReadLock::Open(aAllocator, ipcLock.get_ShmemSection());
|
|
} else {
|
|
sharedLock = reinterpret_cast<gfxMemorySharedReadLock*>(ipcLock.get_uintptr_t());
|
|
if (sharedLock) {
|
|
// The corresponding AddRef is in TiledClient::GetTileDescriptor
|
|
sharedLock->Release();
|
|
}
|
|
}
|
|
|
|
mRetainedTiles.AppendElement(TileHost(sharedLock, texture));
|
|
break;
|
|
}
|
|
default:
|
|
NS_WARNING("Unrecognised tile descriptor type");
|
|
// Fall through
|
|
case TileDescriptor::TPlaceholderTileDescriptor :
|
|
mRetainedTiles.AppendElement(GetPlaceholderTile());
|
|
break;
|
|
}
|
|
if (texture && !texture->HasInternalBuffer()) {
|
|
mHasDoubleBufferedTiles = true;
|
|
}
|
|
}
|
|
}
|
|
|
|
void
|
|
TiledLayerBufferComposite::ReadUnlock()
|
|
{
|
|
if (!IsValid()) {
|
|
return;
|
|
}
|
|
for (size_t i = 0; i < mRetainedTiles.Length(); i++) {
|
|
mRetainedTiles[i].ReadUnlock();
|
|
}
|
|
}
|
|
|
|
void
|
|
TiledLayerBufferComposite::ReleaseTextureHosts()
|
|
{
|
|
if (!IsValid()) {
|
|
return;
|
|
}
|
|
for (size_t i = 0; i < mRetainedTiles.Length(); i++) {
|
|
mRetainedTiles[i].mTextureHost = nullptr;
|
|
}
|
|
}
|
|
|
|
void
|
|
TiledLayerBufferComposite::Upload()
|
|
{
|
|
if(!IsValid()) {
|
|
return;
|
|
}
|
|
// The TextureClients were created with the TextureFlags::IMMEDIATE_UPLOAD flag,
|
|
// so calling Update on all the texture hosts will perform the texture upload.
|
|
Update(mValidRegion, mPaintedRegion);
|
|
ClearPaintedRegion();
|
|
}
|
|
|
|
TileHost
|
|
TiledLayerBufferComposite::ValidateTile(TileHost aTile,
|
|
const nsIntPoint& aTileOrigin,
|
|
const nsIntRegion& aDirtyRect)
|
|
{
|
|
if (aTile.IsPlaceholderTile()) {
|
|
NS_WARNING("Placeholder tile encountered in painted region");
|
|
return aTile;
|
|
}
|
|
|
|
#ifdef GFX_TILEDLAYER_PREF_WARNINGS
|
|
printf_stderr("Upload tile %i, %i\n", aTileOrigin.x, aTileOrigin.y);
|
|
long start = PR_IntervalNow();
|
|
#endif
|
|
|
|
MOZ_ASSERT(aTile.mTextureHost->GetFlags() & TextureFlags::IMMEDIATE_UPLOAD);
|
|
// We possibly upload the entire texture contents here. This is a purposeful
|
|
// decision, as sub-image upload can often be slow and/or unreliable, but
|
|
// we may want to reevaluate this in the future.
|
|
// For !HasInternalBuffer() textures, this is likely a no-op.
|
|
aTile.mTextureHost->Updated(nullptr);
|
|
|
|
#ifdef GFX_TILEDLAYER_PREF_WARNINGS
|
|
if (PR_IntervalNow() - start > 1) {
|
|
printf_stderr("Tile Time to upload %i\n", PR_IntervalNow() - start);
|
|
}
|
|
#endif
|
|
return aTile;
|
|
}
|
|
|
|
void
|
|
TiledLayerBufferComposite::SetCompositor(Compositor* aCompositor)
|
|
{
|
|
if (!IsValid()) {
|
|
return;
|
|
}
|
|
for (size_t i = 0; i < mRetainedTiles.Length(); i++) {
|
|
if (mRetainedTiles[i].IsPlaceholderTile()) continue;
|
|
mRetainedTiles[i].mTextureHost->SetCompositor(aCompositor);
|
|
}
|
|
}
|
|
|
|
#if defined(MOZ_WIDGET_GONK) && ANDROID_VERSION >= 17
|
|
void
|
|
TiledLayerBufferComposite::SetReleaseFence(const android::sp<android::Fence>& aReleaseFence)
|
|
{
|
|
for (size_t i = 0; i < mRetainedTiles.Length(); i++) {
|
|
if (!mRetainedTiles[i].mTextureHost) {
|
|
continue;
|
|
}
|
|
TextureHostOGL* texture = mRetainedTiles[i].mTextureHost->AsHostOGL();
|
|
if (!texture) {
|
|
continue;
|
|
}
|
|
texture->SetReleaseFence(new android::Fence(aReleaseFence->dup()));
|
|
}
|
|
}
|
|
#endif
|
|
|
|
TiledContentHost::TiledContentHost(const TextureInfo& aTextureInfo)
|
|
: ContentHost(aTextureInfo)
|
|
, mTiledBuffer(TiledLayerBufferComposite())
|
|
, mLowPrecisionTiledBuffer(TiledLayerBufferComposite())
|
|
, mOldTiledBuffer(TiledLayerBufferComposite())
|
|
, mOldLowPrecisionTiledBuffer(TiledLayerBufferComposite())
|
|
, mPendingUpload(false)
|
|
, mPendingLowPrecisionUpload(false)
|
|
{
|
|
MOZ_COUNT_CTOR(TiledContentHost);
|
|
}
|
|
|
|
TiledContentHost::~TiledContentHost()
|
|
{
|
|
MOZ_COUNT_DTOR(TiledContentHost);
|
|
|
|
// Unlock any buffers that may still be locked. If we have a pending upload,
|
|
// we will need to unlock the buffer that was about to be uploaded.
|
|
// If a buffer that was being composited had double-buffered tiles, we will
|
|
// need to unlock that buffer too.
|
|
if (mPendingUpload) {
|
|
mTiledBuffer.ReadUnlock();
|
|
if (mOldTiledBuffer.HasDoubleBufferedTiles()) {
|
|
mOldTiledBuffer.ReadUnlock();
|
|
}
|
|
} else if (mTiledBuffer.HasDoubleBufferedTiles()) {
|
|
mTiledBuffer.ReadUnlock();
|
|
}
|
|
|
|
if (mPendingLowPrecisionUpload) {
|
|
mLowPrecisionTiledBuffer.ReadUnlock();
|
|
if (mOldLowPrecisionTiledBuffer.HasDoubleBufferedTiles()) {
|
|
mOldLowPrecisionTiledBuffer.ReadUnlock();
|
|
}
|
|
} else if (mLowPrecisionTiledBuffer.HasDoubleBufferedTiles()) {
|
|
mLowPrecisionTiledBuffer.ReadUnlock();
|
|
}
|
|
}
|
|
|
|
void
|
|
TiledContentHost::Attach(Layer* aLayer,
|
|
Compositor* aCompositor,
|
|
AttachFlags aFlags /* = NO_FLAGS */)
|
|
{
|
|
CompositableHost::Attach(aLayer, aCompositor, aFlags);
|
|
static_cast<ThebesLayerComposite*>(aLayer)->EnsureTiled();
|
|
}
|
|
|
|
void
|
|
TiledContentHost::UseTiledLayerBuffer(ISurfaceAllocator* aAllocator,
|
|
const SurfaceDescriptorTiles& aTiledDescriptor)
|
|
{
|
|
if (aTiledDescriptor.resolution() < 1) {
|
|
if (mPendingLowPrecisionUpload) {
|
|
mLowPrecisionTiledBuffer.ReadUnlock();
|
|
} else {
|
|
mPendingLowPrecisionUpload = true;
|
|
// If the old buffer has double-buffered tiles, hang onto it so we can
|
|
// unlock it after we've composited the new buffer.
|
|
// We only need to hang onto the locks, but not the textures.
|
|
// Releasing the textures here can help prevent a memory spike in the
|
|
// situation that the client starts rendering new content before we get
|
|
// to composite the new buffer.
|
|
if (mLowPrecisionTiledBuffer.HasDoubleBufferedTiles()) {
|
|
mOldLowPrecisionTiledBuffer = mLowPrecisionTiledBuffer;
|
|
mOldLowPrecisionTiledBuffer.ReleaseTextureHosts();
|
|
}
|
|
}
|
|
mLowPrecisionTiledBuffer =
|
|
TiledLayerBufferComposite(aAllocator, aTiledDescriptor,
|
|
mLowPrecisionTiledBuffer.GetPaintedRegion());
|
|
} else {
|
|
if (mPendingUpload) {
|
|
mTiledBuffer.ReadUnlock();
|
|
} else {
|
|
mPendingUpload = true;
|
|
if (mTiledBuffer.HasDoubleBufferedTiles()) {
|
|
mOldTiledBuffer = mTiledBuffer;
|
|
mOldTiledBuffer.ReleaseTextureHosts();
|
|
}
|
|
}
|
|
mTiledBuffer = TiledLayerBufferComposite(aAllocator, aTiledDescriptor,
|
|
mTiledBuffer.GetPaintedRegion());
|
|
}
|
|
}
|
|
|
|
void
|
|
TiledContentHost::Composite(EffectChain& aEffectChain,
|
|
float aOpacity,
|
|
const gfx::Matrix4x4& aTransform,
|
|
const gfx::Filter& aFilter,
|
|
const gfx::Rect& aClipRect,
|
|
const nsIntRegion* aVisibleRegion /* = nullptr */,
|
|
TiledLayerProperties* aLayerProperties /* = nullptr */)
|
|
{
|
|
MOZ_ASSERT(aLayerProperties, "aLayerProperties required for TiledContentHost");
|
|
|
|
if (mPendingUpload) {
|
|
mTiledBuffer.SetCompositor(mCompositor);
|
|
mTiledBuffer.Upload();
|
|
|
|
// For a single-buffered tiled buffer, Upload will upload the shared memory
|
|
// surface to texture memory and we no longer need to read from them.
|
|
if (!mTiledBuffer.HasDoubleBufferedTiles()) {
|
|
mTiledBuffer.ReadUnlock();
|
|
}
|
|
}
|
|
if (mPendingLowPrecisionUpload) {
|
|
mLowPrecisionTiledBuffer.SetCompositor(mCompositor);
|
|
mLowPrecisionTiledBuffer.Upload();
|
|
|
|
if (!mLowPrecisionTiledBuffer.HasDoubleBufferedTiles()) {
|
|
mLowPrecisionTiledBuffer.ReadUnlock();
|
|
}
|
|
}
|
|
|
|
RenderLayerBuffer(mLowPrecisionTiledBuffer, aEffectChain, aOpacity, aFilter,
|
|
aClipRect, aLayerProperties->mVisibleRegion, aTransform);
|
|
RenderLayerBuffer(mTiledBuffer, aEffectChain, aOpacity, aFilter,
|
|
aClipRect, aLayerProperties->mVisibleRegion, aTransform);
|
|
|
|
// Now release the old buffer if it had double-buffered tiles, as we can
|
|
// guarantee that they're no longer on the screen (and so any locks that may
|
|
// have been held have been released).
|
|
if (mPendingUpload && mOldTiledBuffer.HasDoubleBufferedTiles()) {
|
|
mOldTiledBuffer.ReadUnlock();
|
|
mOldTiledBuffer = TiledLayerBufferComposite();
|
|
}
|
|
if (mPendingLowPrecisionUpload && mOldLowPrecisionTiledBuffer.HasDoubleBufferedTiles()) {
|
|
mOldLowPrecisionTiledBuffer.ReadUnlock();
|
|
mOldLowPrecisionTiledBuffer = TiledLayerBufferComposite();
|
|
}
|
|
mPendingUpload = mPendingLowPrecisionUpload = false;
|
|
}
|
|
|
|
|
|
void
|
|
TiledContentHost::RenderTile(const TileHost& aTile,
|
|
EffectChain& aEffectChain,
|
|
float aOpacity,
|
|
const gfx::Matrix4x4& aTransform,
|
|
const gfx::Filter& aFilter,
|
|
const gfx::Rect& aClipRect,
|
|
const nsIntRegion& aScreenRegion,
|
|
const nsIntPoint& aTextureOffset,
|
|
const nsIntSize& aTextureBounds)
|
|
{
|
|
if (aTile.IsPlaceholderTile()) {
|
|
// This shouldn't ever happen, but let's fail semi-gracefully. No need
|
|
// to warn, the texture update would have already caught this.
|
|
return;
|
|
}
|
|
|
|
nsIntRect screenBounds = aScreenRegion.GetBounds();
|
|
Rect quad(screenBounds.x, screenBounds.y, screenBounds.width, screenBounds.height);
|
|
quad = aTransform.TransformBounds(quad);
|
|
|
|
if (!quad.Intersects(mCompositor->ClipRectInLayersCoordinates(aClipRect))) {
|
|
return;
|
|
}
|
|
|
|
AutoLockTextureHost autoLock(aTile.mTextureHost);
|
|
if (autoLock.Failed()) {
|
|
NS_WARNING("Failed to lock tile");
|
|
return;
|
|
}
|
|
RefPtr<NewTextureSource> source = aTile.mTextureHost->GetTextureSources();
|
|
if (!source) {
|
|
return;
|
|
}
|
|
|
|
RefPtr<TexturedEffect> effect =
|
|
CreateTexturedEffect(aTile.mTextureHost->GetFormat(), source, aFilter);
|
|
if (!effect) {
|
|
return;
|
|
}
|
|
|
|
aEffectChain.mPrimaryEffect = effect;
|
|
|
|
nsIntRegionRectIterator it(aScreenRegion);
|
|
for (const nsIntRect* rect = it.Next(); rect != nullptr; rect = it.Next()) {
|
|
Rect graphicsRect(rect->x, rect->y, rect->width, rect->height);
|
|
Rect textureRect(rect->x - aTextureOffset.x, rect->y - aTextureOffset.y,
|
|
rect->width, rect->height);
|
|
|
|
effect->mTextureCoords = Rect(textureRect.x / aTextureBounds.width,
|
|
textureRect.y / aTextureBounds.height,
|
|
textureRect.width / aTextureBounds.width,
|
|
textureRect.height / aTextureBounds.height);
|
|
mCompositor->DrawQuad(graphicsRect, aClipRect, aEffectChain, aOpacity, aTransform);
|
|
}
|
|
mCompositor->DrawDiagnostics(DiagnosticFlags::CONTENT | DiagnosticFlags::TILE,
|
|
aScreenRegion, aClipRect, aTransform, mFlashCounter);
|
|
}
|
|
|
|
void
|
|
TiledContentHost::RenderLayerBuffer(TiledLayerBufferComposite& aLayerBuffer,
|
|
EffectChain& aEffectChain,
|
|
float aOpacity,
|
|
const gfx::Filter& aFilter,
|
|
const gfx::Rect& aClipRect,
|
|
nsIntRegion aVisibleRegion,
|
|
gfx::Matrix4x4 aTransform)
|
|
{
|
|
if (!mCompositor) {
|
|
NS_WARNING("Can't render tiled content host - no compositor");
|
|
return;
|
|
}
|
|
float resolution = aLayerBuffer.GetResolution();
|
|
gfx::Size layerScale(1, 1);
|
|
|
|
// We assume that the current frame resolution is the one used in our high
|
|
// precision layer buffer. Compensate for a changing frame resolution when
|
|
// rendering the low precision buffer.
|
|
if (aLayerBuffer.GetFrameResolution() != mTiledBuffer.GetFrameResolution()) {
|
|
const CSSToParentLayerScale& layerResolution = aLayerBuffer.GetFrameResolution();
|
|
const CSSToParentLayerScale& localResolution = mTiledBuffer.GetFrameResolution();
|
|
layerScale.width = layerScale.height = layerResolution.scale / localResolution.scale;
|
|
aVisibleRegion.ScaleRoundOut(layerScale.width, layerScale.height);
|
|
}
|
|
|
|
// If we're drawing the low precision buffer, make sure the high precision
|
|
// buffer is masked out to avoid overdraw and rendering artifacts with
|
|
// non-opaque layers.
|
|
nsIntRegion maskRegion;
|
|
if (resolution != mTiledBuffer.GetResolution()) {
|
|
maskRegion = mTiledBuffer.GetValidRegion();
|
|
// XXX This should be ScaleRoundIn, but there is no such function on
|
|
// nsIntRegion.
|
|
maskRegion.ScaleRoundOut(layerScale.width, layerScale.height);
|
|
}
|
|
|
|
// Make sure the resolution and difference in frame resolution are accounted
|
|
// for in the layer transform.
|
|
aTransform.Scale(1/(resolution * layerScale.width),
|
|
1/(resolution * layerScale.height), 1);
|
|
|
|
uint32_t rowCount = 0;
|
|
uint32_t tileX = 0;
|
|
nsIntRect visibleRect = aVisibleRegion.GetBounds();
|
|
gfx::IntSize scaledTileSize = aLayerBuffer.GetScaledTileSize();
|
|
for (int32_t x = visibleRect.x; x < visibleRect.x + visibleRect.width;) {
|
|
rowCount++;
|
|
int32_t tileStartX = aLayerBuffer.GetTileStart(x, scaledTileSize.width);
|
|
int32_t w = scaledTileSize.width - tileStartX;
|
|
if (x + w > visibleRect.x + visibleRect.width) {
|
|
w = visibleRect.x + visibleRect.width - x;
|
|
}
|
|
int tileY = 0;
|
|
for (int32_t y = visibleRect.y; y < visibleRect.y + visibleRect.height;) {
|
|
int32_t tileStartY = aLayerBuffer.GetTileStart(y, scaledTileSize.height);
|
|
int32_t h = scaledTileSize.height - tileStartY;
|
|
if (y + h > visibleRect.y + visibleRect.height) {
|
|
h = visibleRect.y + visibleRect.height - y;
|
|
}
|
|
|
|
TileHost tileTexture = aLayerBuffer.
|
|
GetTile(nsIntPoint(aLayerBuffer.RoundDownToTileEdge(x, scaledTileSize.width),
|
|
aLayerBuffer.RoundDownToTileEdge(y, scaledTileSize.height)));
|
|
if (tileTexture != aLayerBuffer.GetPlaceholderTile()) {
|
|
nsIntRegion tileDrawRegion;
|
|
tileDrawRegion.And(nsIntRect(x, y, w, h), aLayerBuffer.GetValidRegion());
|
|
tileDrawRegion.And(tileDrawRegion, aVisibleRegion);
|
|
tileDrawRegion.Sub(tileDrawRegion, maskRegion);
|
|
|
|
if (!tileDrawRegion.IsEmpty()) {
|
|
tileDrawRegion.ScaleRoundOut(resolution, resolution);
|
|
nsIntPoint tileOffset((x - tileStartX) * resolution,
|
|
(y - tileStartY) * resolution);
|
|
gfx::IntSize tileSize = aLayerBuffer.GetTileSize();
|
|
RenderTile(tileTexture, aEffectChain, aOpacity, aTransform, aFilter, aClipRect, tileDrawRegion,
|
|
tileOffset, nsIntSize(tileSize.width, tileSize.height));
|
|
}
|
|
}
|
|
tileY++;
|
|
y += h;
|
|
}
|
|
tileX++;
|
|
x += w;
|
|
}
|
|
gfx::Rect rect(visibleRect.x, visibleRect.y,
|
|
visibleRect.width, visibleRect.height);
|
|
GetCompositor()->DrawDiagnostics(DiagnosticFlags::CONTENT,
|
|
rect, aClipRect, aTransform, mFlashCounter);
|
|
}
|
|
|
|
void
|
|
TiledContentHost::PrintInfo(nsACString& aTo, const char* aPrefix)
|
|
{
|
|
aTo += aPrefix;
|
|
aTo += nsPrintfCString("TiledContentHost (0x%p)", this);
|
|
|
|
}
|
|
|
|
#ifdef MOZ_DUMP_PAINTING
|
|
void
|
|
TiledContentHost::Dump(FILE* aFile,
|
|
const char* aPrefix,
|
|
bool aDumpHtml)
|
|
{
|
|
if (!aFile) {
|
|
aFile = stderr;
|
|
}
|
|
|
|
TiledLayerBufferComposite::Iterator it = mTiledBuffer.TilesBegin();
|
|
TiledLayerBufferComposite::Iterator stop = mTiledBuffer.TilesEnd();
|
|
if (aDumpHtml) {
|
|
fprintf_stderr(aFile, "<ul>");
|
|
}
|
|
for (;it != stop; ++it) {
|
|
fprintf_stderr(aFile, "%s", aPrefix);
|
|
fprintf_stderr(aFile, aDumpHtml ? "<li> <a href=" : "Tile ");
|
|
if (it->IsPlaceholderTile()) {
|
|
fprintf_stderr(aFile, "empty tile");
|
|
} else {
|
|
DumpTextureHost(aFile, it->mTextureHost);
|
|
}
|
|
fprintf_stderr(aFile, aDumpHtml ? " >Tile</a></li>" : " ");
|
|
}
|
|
if (aDumpHtml) {
|
|
fprintf_stderr(aFile, "</ul>");
|
|
}
|
|
}
|
|
#endif
|
|
|
|
} // namespace
|
|
} // namespace
|