зеркало из https://github.com/mozilla/gecko-dev.git
10781 строка
396 KiB
C++
10781 строка
396 KiB
C++
/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
|
|
/* vim: set ts=8 sts=2 et sw=2 tw=80: */
|
|
/* 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/.
|
|
*/
|
|
|
|
/*
|
|
* structures that represent things to be painted (ordered in z-order),
|
|
* used during painting and hit testing
|
|
*/
|
|
|
|
#include "nsDisplayList.h"
|
|
|
|
#include <stdint.h>
|
|
#include <algorithm>
|
|
#include <limits>
|
|
|
|
#include "gfxContext.h"
|
|
#include "gfxUtils.h"
|
|
#include "mozilla/dom/BrowserChild.h"
|
|
#include "mozilla/dom/HTMLCanvasElement.h"
|
|
#include "mozilla/dom/KeyframeEffect.h"
|
|
#include "mozilla/dom/Selection.h"
|
|
#include "mozilla/dom/ServiceWorkerRegistrar.h"
|
|
#include "mozilla/dom/ServiceWorkerRegistration.h"
|
|
#include "mozilla/dom/SVGElement.h"
|
|
#include "mozilla/dom/TouchEvent.h"
|
|
#include "mozilla/gfx/2D.h"
|
|
#include "mozilla/layers/PLayerTransaction.h"
|
|
#include "mozilla/PresShell.h"
|
|
#include "mozilla/ShapeUtils.h"
|
|
#include "mozilla/StaticPrefs_apz.h"
|
|
#include "mozilla/StaticPrefs_gfx.h"
|
|
#include "mozilla/StaticPrefs_layers.h"
|
|
#include "mozilla/StaticPrefs_layout.h"
|
|
#include "nsCSSRendering.h"
|
|
#include "nsCSSRenderingGradients.h"
|
|
#include "nsRegion.h"
|
|
#include "nsStyleStructInlines.h"
|
|
#include "nsStyleTransformMatrix.h"
|
|
#include "nsTransitionManager.h"
|
|
#include "gfxMatrix.h"
|
|
#include "nsSVGIntegrationUtils.h"
|
|
#include "nsSVGUtils.h"
|
|
#include "nsLayoutUtils.h"
|
|
#include "nsIScrollableFrame.h"
|
|
#include "nsIFrameInlines.h"
|
|
#include "nsStyleConsts.h"
|
|
#include "BorderConsts.h"
|
|
#include "LayerTreeInvalidation.h"
|
|
#include "mozilla/MathAlgorithms.h"
|
|
|
|
#include "imgIContainer.h"
|
|
#include "BasicLayers.h"
|
|
#include "nsBoxFrame.h"
|
|
#include "nsImageFrame.h"
|
|
#include "nsSubDocumentFrame.h"
|
|
#include "SVGObserverUtils.h"
|
|
#include "nsSVGClipPathFrame.h"
|
|
#include "GeckoProfiler.h"
|
|
#include "nsViewManager.h"
|
|
#include "ImageLayers.h"
|
|
#include "ImageContainer.h"
|
|
#include "nsCanvasFrame.h"
|
|
#include "nsSubDocumentFrame.h"
|
|
#include "StickyScrollContainer.h"
|
|
#include "mozilla/AnimationPerformanceWarning.h"
|
|
#include "mozilla/AnimationUtils.h"
|
|
#include "mozilla/AutoRestore.h"
|
|
#include "mozilla/EffectCompositor.h"
|
|
#include "mozilla/EffectSet.h"
|
|
#include "mozilla/EventStates.h"
|
|
#include "mozilla/HashTable.h"
|
|
#include "mozilla/LookAndFeel.h"
|
|
#include "mozilla/OperatorNewExtensions.h"
|
|
#include "mozilla/PendingAnimationTracker.h"
|
|
#include "mozilla/Preferences.h"
|
|
#include "mozilla/StyleAnimationValue.h"
|
|
#include "mozilla/ServoBindings.h"
|
|
#include "mozilla/Telemetry.h"
|
|
#include "mozilla/UniquePtr.h"
|
|
#include "mozilla/Unused.h"
|
|
#include "mozilla/ViewportFrame.h"
|
|
#include "mozilla/gfx/gfxVars.h"
|
|
#include "ActiveLayerTracker.h"
|
|
#include "nsPrintfCString.h"
|
|
#include "UnitTransforms.h"
|
|
#include "LayerAnimationInfo.h"
|
|
#include "LayersLogging.h"
|
|
#include "FrameLayerBuilder.h"
|
|
#include "mozilla/EventStateManager.h"
|
|
#include "nsCaret.h"
|
|
#include "nsDOMTokenList.h"
|
|
#include "nsCSSProps.h"
|
|
#include "nsSVGMaskFrame.h"
|
|
#include "nsTableCellFrame.h"
|
|
#include "nsTableColFrame.h"
|
|
#include "nsTextFrame.h"
|
|
#include "nsSliderFrame.h"
|
|
#include "nsFocusManager.h"
|
|
#include "ClientLayerManager.h"
|
|
#include "mozilla/layers/AnimationHelper.h"
|
|
#include "mozilla/layers/RenderRootStateManager.h"
|
|
#include "mozilla/layers/StackingContextHelper.h"
|
|
#include "mozilla/layers/TreeTraversal.h"
|
|
#include "mozilla/layers/WebRenderBridgeChild.h"
|
|
#include "mozilla/layers/WebRenderLayerManager.h"
|
|
#include "mozilla/layers/WebRenderMessages.h"
|
|
#include "mozilla/layers/WebRenderScrollData.h"
|
|
|
|
using namespace mozilla;
|
|
using namespace mozilla::layers;
|
|
using namespace mozilla::dom;
|
|
using namespace mozilla::layout;
|
|
using namespace mozilla::gfx;
|
|
|
|
typedef ScrollableLayerGuid::ViewID ViewID;
|
|
typedef nsStyleTransformMatrix::TransformReferenceBox TransformReferenceBox;
|
|
|
|
#ifdef DEBUG
|
|
static bool SpammyLayoutWarningsEnabled() {
|
|
static bool sValue = false;
|
|
static bool sValueInitialized = false;
|
|
|
|
if (!sValueInitialized) {
|
|
Preferences::GetBool("layout.spammy_warnings.enabled", &sValue);
|
|
sValueInitialized = true;
|
|
}
|
|
|
|
return sValue;
|
|
}
|
|
#endif
|
|
|
|
#ifdef MOZ_DIAGNOSTIC_ASSERT_ENABLED
|
|
void AssertUniqueItem(nsDisplayItem* aItem) {
|
|
nsIFrame::DisplayItemArray* items =
|
|
aItem->Frame()->GetProperty(nsIFrame::DisplayItems());
|
|
if (!items) {
|
|
return;
|
|
}
|
|
for (nsDisplayItemBase* i : *items) {
|
|
if (i != aItem && !i->HasDeletedFrame() && i->Frame() == aItem->Frame() &&
|
|
i->GetPerFrameKey() == aItem->GetPerFrameKey()) {
|
|
if (i->IsPreProcessedItem()) {
|
|
continue;
|
|
}
|
|
MOZ_DIAGNOSTIC_ASSERT(false, "Duplicate display item!");
|
|
}
|
|
}
|
|
}
|
|
#endif
|
|
|
|
bool ShouldBuildItemForEventsOrPlugins(const DisplayItemType aType) {
|
|
return aType == DisplayItemType::TYPE_COMPOSITOR_HITTEST_INFO ||
|
|
aType == DisplayItemType::TYPE_PLUGIN ||
|
|
(GetDisplayItemFlagsForType(aType) & TYPE_IS_CONTAINER);
|
|
}
|
|
|
|
void UpdateDisplayItemData(nsPaintedDisplayItem* aItem) {
|
|
for (mozilla::DisplayItemData* did : aItem->Frame()->DisplayItemData()) {
|
|
if (did->GetDisplayItemKey() == aItem->GetPerFrameKey() &&
|
|
did->GetLayer()->AsPaintedLayer()) {
|
|
if (!did->HasMergedFrames()) {
|
|
aItem->SetDisplayItemData(did, did->GetLayer()->Manager());
|
|
}
|
|
|
|
return;
|
|
}
|
|
}
|
|
}
|
|
|
|
/* static */
|
|
bool ActiveScrolledRoot::IsAncestor(const ActiveScrolledRoot* aAncestor,
|
|
const ActiveScrolledRoot* aDescendant) {
|
|
if (!aAncestor) {
|
|
// nullptr is the root
|
|
return true;
|
|
}
|
|
if (Depth(aAncestor) > Depth(aDescendant)) {
|
|
return false;
|
|
}
|
|
const ActiveScrolledRoot* asr = aDescendant;
|
|
while (asr) {
|
|
if (asr == aAncestor) {
|
|
return true;
|
|
}
|
|
asr = asr->mParent;
|
|
}
|
|
return false;
|
|
}
|
|
|
|
/* static */
|
|
nsCString ActiveScrolledRoot::ToString(
|
|
const ActiveScrolledRoot* aActiveScrolledRoot) {
|
|
nsAutoCString str;
|
|
for (auto* asr = aActiveScrolledRoot; asr; asr = asr->mParent) {
|
|
str.AppendPrintf("<0x%p>", asr->mScrollableFrame);
|
|
if (asr->mParent) {
|
|
str.AppendLiteral(", ");
|
|
}
|
|
}
|
|
return std::move(str);
|
|
}
|
|
|
|
static StyleTransformOperation ResolveTranslate(
|
|
TransformReferenceBox& aRefBox, const LengthPercentage& aX,
|
|
const LengthPercentage& aY = LengthPercentage::Zero(),
|
|
const Length& aZ = Length{0}) {
|
|
float x = nsStyleTransformMatrix::ProcessTranslatePart(
|
|
aX, &aRefBox, &TransformReferenceBox::Width);
|
|
float y = nsStyleTransformMatrix::ProcessTranslatePart(
|
|
aY, &aRefBox, &TransformReferenceBox::Height);
|
|
return StyleTransformOperation::Translate3D(
|
|
LengthPercentage::FromPixels(x), LengthPercentage::FromPixels(y), aZ);
|
|
}
|
|
|
|
static StyleTranslate ResolveTranslate(const StyleTranslate& aValue,
|
|
TransformReferenceBox& aRefBox) {
|
|
if (aValue.IsTranslate()) {
|
|
const auto& t = aValue.AsTranslate();
|
|
float x = nsStyleTransformMatrix::ProcessTranslatePart(
|
|
t._0, &aRefBox, &TransformReferenceBox::Width);
|
|
float y = nsStyleTransformMatrix::ProcessTranslatePart(
|
|
t._1, &aRefBox, &TransformReferenceBox::Height);
|
|
return StyleTranslate::Translate(LengthPercentage::FromPixels(x),
|
|
LengthPercentage::FromPixels(y), t._2);
|
|
}
|
|
|
|
MOZ_ASSERT(aValue.IsNone());
|
|
return StyleTranslate::None();
|
|
}
|
|
|
|
static StyleTransform ResolveTransformOperations(
|
|
const StyleTransform& aTransform, TransformReferenceBox& aRefBox) {
|
|
auto convertMatrix = [](const Matrix4x4& aM) {
|
|
return StyleTransformOperation::Matrix3D(StyleGenericMatrix3D<StyleNumber>{
|
|
aM._11, aM._12, aM._13, aM._14, aM._21, aM._22, aM._23, aM._24, aM._31,
|
|
aM._32, aM._33, aM._34, aM._41, aM._42, aM._43, aM._44});
|
|
};
|
|
|
|
Vector<StyleTransformOperation> result;
|
|
MOZ_RELEASE_ASSERT(
|
|
result.initCapacity(aTransform.Operations().Length()),
|
|
"Allocating vector of transform operations should be successful.");
|
|
|
|
for (const StyleTransformOperation& op : aTransform.Operations()) {
|
|
switch (op.tag) {
|
|
case StyleTransformOperation::Tag::TranslateX:
|
|
result.infallibleAppend(ResolveTranslate(aRefBox, op.AsTranslateX()));
|
|
break;
|
|
case StyleTransformOperation::Tag::TranslateY:
|
|
result.infallibleAppend(ResolveTranslate(
|
|
aRefBox, LengthPercentage::Zero(), op.AsTranslateY()));
|
|
break;
|
|
case StyleTransformOperation::Tag::TranslateZ:
|
|
result.infallibleAppend(
|
|
ResolveTranslate(aRefBox, LengthPercentage::Zero(),
|
|
LengthPercentage::Zero(), op.AsTranslateZ()));
|
|
break;
|
|
case StyleTransformOperation::Tag::Translate: {
|
|
const auto& translate = op.AsTranslate();
|
|
result.infallibleAppend(
|
|
ResolveTranslate(aRefBox, translate._0, translate._1));
|
|
break;
|
|
}
|
|
case StyleTransformOperation::Tag::Translate3D: {
|
|
const auto& translate = op.AsTranslate3D();
|
|
result.infallibleAppend(ResolveTranslate(aRefBox, translate._0,
|
|
translate._1, translate._2));
|
|
break;
|
|
}
|
|
case StyleTransformOperation::Tag::InterpolateMatrix: {
|
|
Matrix4x4 matrix;
|
|
nsStyleTransformMatrix::ProcessInterpolateMatrix(matrix, op, aRefBox);
|
|
result.infallibleAppend(convertMatrix(matrix));
|
|
break;
|
|
}
|
|
case StyleTransformOperation::Tag::AccumulateMatrix: {
|
|
Matrix4x4 matrix;
|
|
nsStyleTransformMatrix::ProcessAccumulateMatrix(matrix, op, aRefBox);
|
|
result.infallibleAppend(convertMatrix(matrix));
|
|
break;
|
|
}
|
|
case StyleTransformOperation::Tag::RotateX:
|
|
case StyleTransformOperation::Tag::RotateY:
|
|
case StyleTransformOperation::Tag::RotateZ:
|
|
case StyleTransformOperation::Tag::Rotate:
|
|
case StyleTransformOperation::Tag::Rotate3D:
|
|
case StyleTransformOperation::Tag::ScaleX:
|
|
case StyleTransformOperation::Tag::ScaleY:
|
|
case StyleTransformOperation::Tag::ScaleZ:
|
|
case StyleTransformOperation::Tag::Scale:
|
|
case StyleTransformOperation::Tag::Scale3D:
|
|
case StyleTransformOperation::Tag::SkewX:
|
|
case StyleTransformOperation::Tag::SkewY:
|
|
case StyleTransformOperation::Tag::Skew:
|
|
case StyleTransformOperation::Tag::Matrix:
|
|
case StyleTransformOperation::Tag::Matrix3D:
|
|
case StyleTransformOperation::Tag::Perspective:
|
|
result.infallibleAppend(op);
|
|
break;
|
|
default:
|
|
MOZ_ASSERT_UNREACHABLE("Function not handled yet!");
|
|
}
|
|
}
|
|
|
|
auto transform = StyleTransform{
|
|
StyleOwnedSlice<StyleTransformOperation>(std::move(result))};
|
|
MOZ_ASSERT(!transform.HasPercent());
|
|
MOZ_ASSERT(transform.Operations().Length() ==
|
|
aTransform.Operations().Length());
|
|
return transform;
|
|
}
|
|
|
|
static TimingFunction ToTimingFunction(
|
|
const Maybe<ComputedTimingFunction>& aCTF) {
|
|
if (aCTF.isNothing()) {
|
|
return TimingFunction(null_t());
|
|
}
|
|
|
|
if (aCTF->HasSpline()) {
|
|
const SMILKeySpline* spline = aCTF->GetFunction();
|
|
return TimingFunction(CubicBezierFunction(spline->X1(), spline->Y1(),
|
|
spline->X2(), spline->Y2()));
|
|
}
|
|
|
|
return TimingFunction(StepFunction(
|
|
aCTF->GetSteps().mSteps, static_cast<uint8_t>(aCTF->GetSteps().mPos)));
|
|
}
|
|
|
|
// FIXME: Bug 1489392: We don't have to normalize the path here if we accept
|
|
// the spec issue which would like to normalize svg paths at computed time.
|
|
static StyleOffsetPath NormalizeOffsetPath(const StyleOffsetPath& aOffsetPath) {
|
|
if (aOffsetPath.IsPath()) {
|
|
return StyleOffsetPath::Path(
|
|
MotionPathUtils::NormalizeSVGPathData(aOffsetPath.AsPath()));
|
|
}
|
|
return StyleOffsetPath(aOffsetPath);
|
|
}
|
|
|
|
static void SetAnimatable(nsCSSPropertyID aProperty,
|
|
const AnimationValue& aAnimationValue,
|
|
nsIFrame* aFrame, TransformReferenceBox& aRefBox,
|
|
layers::Animatable& aAnimatable) {
|
|
MOZ_ASSERT(aFrame);
|
|
|
|
if (aAnimationValue.IsNull()) {
|
|
aAnimatable = null_t();
|
|
return;
|
|
}
|
|
|
|
switch (aProperty) {
|
|
case eCSSProperty_background_color: {
|
|
// We don't support color animation on the compositor yet so that we can
|
|
// resolve currentColor at this moment.
|
|
nscolor foreground =
|
|
aFrame->Style()->GetVisitedDependentColor(&nsStyleText::mColor);
|
|
aAnimatable = aAnimationValue.GetColor(foreground);
|
|
break;
|
|
}
|
|
case eCSSProperty_opacity:
|
|
aAnimatable = aAnimationValue.GetOpacity();
|
|
break;
|
|
case eCSSProperty_rotate:
|
|
aAnimatable = aAnimationValue.GetRotateProperty();
|
|
break;
|
|
case eCSSProperty_scale:
|
|
aAnimatable = aAnimationValue.GetScaleProperty();
|
|
break;
|
|
case eCSSProperty_translate:
|
|
aAnimatable =
|
|
ResolveTranslate(aAnimationValue.GetTranslateProperty(), aRefBox);
|
|
break;
|
|
case eCSSProperty_transform:
|
|
aAnimatable = ResolveTransformOperations(
|
|
aAnimationValue.GetTransformProperty(), aRefBox);
|
|
break;
|
|
case eCSSProperty_offset_path:
|
|
aAnimatable =
|
|
NormalizeOffsetPath(aAnimationValue.GetOffsetPathProperty());
|
|
break;
|
|
case eCSSProperty_offset_distance:
|
|
aAnimatable = aAnimationValue.GetOffsetDistanceProperty();
|
|
break;
|
|
case eCSSProperty_offset_rotate:
|
|
aAnimatable = aAnimationValue.GetOffsetRotateProperty();
|
|
break;
|
|
case eCSSProperty_offset_anchor:
|
|
aAnimatable = aAnimationValue.GetOffsetAnchorProperty();
|
|
break;
|
|
default:
|
|
MOZ_ASSERT_UNREACHABLE("Unsupported property");
|
|
}
|
|
}
|
|
|
|
enum class Send {
|
|
NextTransaction,
|
|
Immediate,
|
|
};
|
|
static void AddAnimationForProperty(nsIFrame* aFrame,
|
|
const AnimationProperty& aProperty,
|
|
dom::Animation* aAnimation,
|
|
const CompositorAnimationData& aData,
|
|
Send aSendFlag,
|
|
AnimationInfo& aAnimationInfo) {
|
|
MOZ_ASSERT(aAnimation->GetEffect(),
|
|
"Should not be adding an animation without an effect");
|
|
MOZ_ASSERT(!aAnimation->GetCurrentOrPendingStartTime().IsNull() ||
|
|
!aAnimation->IsPlaying() ||
|
|
(aAnimation->GetTimeline() &&
|
|
aAnimation->GetTimeline()->TracksWallclockTime()),
|
|
"If the animation has an unresolved start time it should either"
|
|
" be static (so we don't need a start time) or else have a"
|
|
" timeline capable of converting TimeStamps (so we can calculate"
|
|
" one later");
|
|
|
|
layers::Animation* animation =
|
|
(aSendFlag == Send::NextTransaction)
|
|
? aAnimationInfo.AddAnimationForNextTransaction()
|
|
: aAnimationInfo.AddAnimation();
|
|
|
|
const TimingParams& timing = aAnimation->GetEffect()->SpecifiedTiming();
|
|
|
|
// If we are starting a new transition that replaces an existing transition
|
|
// running on the compositor, it is possible that the animation on the
|
|
// compositor will have advanced ahead of the main thread. If we use as
|
|
// the starting point of the new transition, the current value of the
|
|
// replaced transition as calculated on the main thread using the refresh
|
|
// driver time, the new transition will jump when it starts. Instead, we
|
|
// re-calculate the starting point of the new transition by applying the
|
|
// current TimeStamp to the parameters of the replaced transition.
|
|
//
|
|
// We need to do this here, rather than when we generate the new transition,
|
|
// since after generating the new transition other requestAnimationFrame
|
|
// callbacks may run that introduce further lag between the main thread and
|
|
// the compositor.
|
|
if (aAnimation->AsCSSTransition() && aAnimation->GetEffect() &&
|
|
aAnimation->GetEffect()->AsTransition()) {
|
|
// We update startValue from the replaced transition only if the effect is
|
|
// an ElementPropertyTransition.
|
|
aAnimation->GetEffect()
|
|
->AsTransition()
|
|
->UpdateStartValueFromReplacedTransition();
|
|
}
|
|
|
|
animation->originTime() =
|
|
!aAnimation->GetTimeline()
|
|
? TimeStamp()
|
|
: aAnimation->GetTimeline()->ToTimeStamp(TimeDuration());
|
|
|
|
Nullable<TimeDuration> startTime = aAnimation->GetCurrentOrPendingStartTime();
|
|
if (startTime.IsNull()) {
|
|
animation->startTime() = Nothing();
|
|
} else {
|
|
animation->startTime() = Some(startTime.Value());
|
|
}
|
|
|
|
animation->holdTime() = aAnimation->GetCurrentTimeAsDuration().Value();
|
|
|
|
const ComputedTiming computedTiming =
|
|
aAnimation->GetEffect()->GetComputedTiming();
|
|
animation->delay() = timing.Delay();
|
|
animation->endDelay() = timing.EndDelay();
|
|
animation->duration() = computedTiming.mDuration;
|
|
animation->iterations() = computedTiming.mIterations;
|
|
animation->iterationStart() = computedTiming.mIterationStart;
|
|
animation->direction() = static_cast<uint8_t>(timing.Direction());
|
|
animation->fillMode() = static_cast<uint8_t>(computedTiming.mFill);
|
|
animation->property() = aProperty.mProperty;
|
|
animation->playbackRate() = aAnimation->CurrentOrPendingPlaybackRate();
|
|
animation->previousPlaybackRate() =
|
|
aAnimation->HasPendingPlaybackRate()
|
|
? aAnimation->PlaybackRate()
|
|
: std::numeric_limits<float>::quiet_NaN();
|
|
animation->transformData() = aData.mTransform;
|
|
animation->motionPathData() = aData.mMotionPath;
|
|
animation->easingFunction() = ToTimingFunction(timing.TimingFunction());
|
|
animation->iterationComposite() = static_cast<uint8_t>(
|
|
aAnimation->GetEffect()->AsKeyframeEffect()->IterationComposite());
|
|
animation->isNotPlaying() = !aAnimation->IsPlaying();
|
|
animation->isNotAnimating() = false;
|
|
|
|
TransformReferenceBox refBox(aFrame);
|
|
|
|
// If the animation is additive or accumulates, we need to pass its base value
|
|
// to the compositor.
|
|
|
|
AnimationValue baseStyle =
|
|
aAnimation->GetEffect()->AsKeyframeEffect()->BaseStyle(
|
|
aProperty.mProperty);
|
|
if (!baseStyle.IsNull()) {
|
|
SetAnimatable(aProperty.mProperty, baseStyle, aFrame, refBox,
|
|
animation->baseStyle());
|
|
} else {
|
|
animation->baseStyle() = null_t();
|
|
}
|
|
|
|
for (uint32_t segIdx = 0; segIdx < aProperty.mSegments.Length(); segIdx++) {
|
|
const AnimationPropertySegment& segment = aProperty.mSegments[segIdx];
|
|
|
|
AnimationSegment* animSegment = animation->segments().AppendElement();
|
|
SetAnimatable(aProperty.mProperty, segment.mFromValue, aFrame, refBox,
|
|
animSegment->startState());
|
|
SetAnimatable(aProperty.mProperty, segment.mToValue, aFrame, refBox,
|
|
animSegment->endState());
|
|
|
|
animSegment->startPortion() = segment.mFromKey;
|
|
animSegment->endPortion() = segment.mToKey;
|
|
animSegment->startComposite() =
|
|
static_cast<uint8_t>(segment.mFromComposite);
|
|
animSegment->endComposite() = static_cast<uint8_t>(segment.mToComposite);
|
|
animSegment->sampleFn() = ToTimingFunction(segment.mTimingFunction);
|
|
}
|
|
}
|
|
|
|
// Let's use an example to explain this function:
|
|
//
|
|
// We have 4 playing animations (without any !important rule or transition):
|
|
// Animation A: [ transform, rotate ].
|
|
// Animation B: [ rotate, scale ].
|
|
// Animation C: [ transform, margin-left ].
|
|
// Animation D: [ opacity, margin-left ].
|
|
//
|
|
// Normally, GetAnimationsForCompositor(|transform-like properties|) returns:
|
|
// [ Animation A, Animation B, Animation C ], which is the first argument of
|
|
// this function.
|
|
//
|
|
// In this function, we want to re-organize the list as (Note: don't care
|
|
// the order of properties):
|
|
// [
|
|
// { rotate: [ Animation A, Animation B ] },
|
|
// { scale: [ Animation B ] },
|
|
// { transform: [ Animation A, Animation C ] },
|
|
// ]
|
|
//
|
|
// Therefore, AddAnimationsForProperty() will append each animation property
|
|
// into AnimationInfo, as a final list of layers::Animation:
|
|
// [
|
|
// { rotate: Animation A },
|
|
// { rotate: Animation B },
|
|
// { scale: Animation B },
|
|
// { transform: Animation A },
|
|
// { transform: Animation C },
|
|
// ]
|
|
//
|
|
// And then, for each transaction, we send this list to the compositor thread.
|
|
static HashMap<nsCSSPropertyID, nsTArray<RefPtr<dom::Animation>>>
|
|
GroupAnimationsByProperty(const nsTArray<RefPtr<dom::Animation>>& aAnimations,
|
|
const nsCSSPropertyIDSet& aPropertySet) {
|
|
HashMap<nsCSSPropertyID, nsTArray<RefPtr<dom::Animation>>> groupedAnims;
|
|
for (const RefPtr<dom::Animation>& anim : aAnimations) {
|
|
const KeyframeEffect* effect = anim->GetEffect()->AsKeyframeEffect();
|
|
MOZ_ASSERT(effect);
|
|
for (const AnimationProperty& property : effect->Properties()) {
|
|
if (!aPropertySet.HasProperty(property.mProperty)) {
|
|
continue;
|
|
}
|
|
|
|
auto animsForPropertyPtr = groupedAnims.lookupForAdd(property.mProperty);
|
|
if (!animsForPropertyPtr) {
|
|
DebugOnly<bool> rv =
|
|
groupedAnims.add(animsForPropertyPtr, property.mProperty,
|
|
nsTArray<RefPtr<dom::Animation>>());
|
|
MOZ_ASSERT(rv, "Should have enough memory");
|
|
}
|
|
animsForPropertyPtr->value().AppendElement(anim);
|
|
}
|
|
}
|
|
return groupedAnims;
|
|
}
|
|
|
|
static bool AddAnimationsForProperty(
|
|
nsIFrame* aFrame, const EffectSet* aEffects,
|
|
const nsTArray<RefPtr<dom::Animation>>& aCompositorAnimations,
|
|
const CompositorAnimationData& aData, nsCSSPropertyID aProperty,
|
|
Send aSendFlag, AnimationInfo& aAnimationInfo) {
|
|
bool addedAny = false;
|
|
// Add from first to last (since last overrides)
|
|
for (dom::Animation* anim : aCompositorAnimations) {
|
|
if (!anim->IsRelevant()) {
|
|
continue;
|
|
}
|
|
|
|
dom::KeyframeEffect* keyframeEffect =
|
|
anim->GetEffect() ? anim->GetEffect()->AsKeyframeEffect() : nullptr;
|
|
MOZ_ASSERT(keyframeEffect,
|
|
"A playing animation should have a keyframe effect");
|
|
const AnimationProperty* property =
|
|
keyframeEffect->GetEffectiveAnimationOfProperty(aProperty, *aEffects);
|
|
if (!property) {
|
|
continue;
|
|
}
|
|
|
|
// Note that if the property is overridden by !important rules,
|
|
// GetEffectiveAnimationOfProperty returns null instead.
|
|
// This is what we want, since if we have animations overridden by
|
|
// !important rules, we don't want to send them to the compositor.
|
|
MOZ_ASSERT(
|
|
anim->CascadeLevel() != EffectCompositor::CascadeLevel::Animations ||
|
|
!aEffects->PropertiesWithImportantRules().HasProperty(aProperty),
|
|
"GetEffectiveAnimationOfProperty already tested the property "
|
|
"is not overridden by !important rules");
|
|
|
|
// Don't add animations that are pending if their timeline does not
|
|
// track wallclock time. This is because any pending animations on layers
|
|
// will have their start time updated with the current wallclock time.
|
|
// If we can't convert that wallclock time back to an equivalent timeline
|
|
// time, we won't be able to update the content animation and it will end
|
|
// up being out of sync with the layer animation.
|
|
//
|
|
// Currently this only happens when the timeline is driven by a refresh
|
|
// driver under test control. In this case, the next time the refresh
|
|
// driver is advanced it will trigger any pending animations.
|
|
if (anim->Pending() &&
|
|
(anim->GetTimeline() && !anim->GetTimeline()->TracksWallclockTime())) {
|
|
continue;
|
|
}
|
|
|
|
AddAnimationForProperty(aFrame, *property, anim, aData, aSendFlag,
|
|
aAnimationInfo);
|
|
keyframeEffect->SetIsRunningOnCompositor(aProperty, true);
|
|
addedAny = true;
|
|
}
|
|
return addedAny;
|
|
}
|
|
|
|
enum class AnimationDataType {
|
|
WithMotionPath,
|
|
WithoutMotionPath,
|
|
};
|
|
static CompositorAnimationData CreateAnimationData(
|
|
nsIFrame* aFrame, nsDisplayItem* aItem, DisplayItemType aType,
|
|
layers::LayersBackend aLayersBackend, AnimationDataType aDataType) {
|
|
CompositorAnimationData result;
|
|
|
|
if (aType != DisplayItemType::TYPE_TRANSFORM) {
|
|
return result;
|
|
}
|
|
|
|
// XXX Performance here isn't ideal for SVG. We'd prefer to avoid resolving
|
|
// the dimensions of refBox. That said, we only get here if there are CSS
|
|
// animations or transitions on this element, and that is likely to be a
|
|
// lot rarer than transforms on SVG (the frequency of which drives the need
|
|
// for TransformReferenceBox).
|
|
TransformReferenceBox refBox(aFrame);
|
|
nsRect bounds(0, 0, refBox.Width(), refBox.Height());
|
|
|
|
// all data passed directly to the compositor should be in dev pixels
|
|
int32_t devPixelsToAppUnits = aFrame->PresContext()->AppUnitsPerDevPixel();
|
|
float scale = devPixelsToAppUnits;
|
|
Point3D offsetToTransformOrigin =
|
|
nsDisplayTransform::GetDeltaToTransformOrigin(aFrame, refBox, scale);
|
|
nsPoint origin;
|
|
float scaleX = 1.0f;
|
|
float scaleY = 1.0f;
|
|
bool hasPerspectiveParent = false;
|
|
if (aLayersBackend == layers::LayersBackend::LAYERS_WR) {
|
|
// leave origin empty, because we are sending it separately on the
|
|
// stacking context that we are pushing to WR, and WR will automatically
|
|
// include it when picking up the animated transform values
|
|
} else if (aItem) {
|
|
// This branch is for display items to leverage the cache of
|
|
// nsDisplayListBuilder.
|
|
origin = aItem->ToReferenceFrame();
|
|
} else {
|
|
// This branch is running for restyling.
|
|
// Animations are animated at the coordination of the reference
|
|
// frame outside, not the given frame itself. The given frame
|
|
// is also reference frame too, so the parent's reference frame
|
|
// are used.
|
|
nsIFrame* referenceFrame = nsLayoutUtils::GetReferenceFrame(
|
|
nsLayoutUtils::GetCrossDocParentFrame(aFrame));
|
|
origin = aFrame->GetOffsetToCrossDoc(referenceFrame);
|
|
}
|
|
|
|
result.mTransform = Some(TransformData(origin, offsetToTransformOrigin,
|
|
bounds, devPixelsToAppUnits, scaleX,
|
|
scaleY, hasPerspectiveParent));
|
|
|
|
if (aDataType == AnimationDataType::WithoutMotionPath) {
|
|
return result;
|
|
}
|
|
|
|
const StyleTransformOrigin& styleOrigin =
|
|
aFrame->StyleDisplay()->mTransformOrigin;
|
|
CSSPoint motionPathOrigin = nsStyleTransformMatrix::Convert2DPosition(
|
|
styleOrigin.horizontal, styleOrigin.vertical, refBox);
|
|
CSSPoint anchorAdjustment =
|
|
MotionPathUtils::ComputeAnchorPointAdjustment(*aFrame);
|
|
|
|
result.mMotionPath = Some(layers::MotionPathData(
|
|
motionPathOrigin, anchorAdjustment, RayReferenceData(aFrame)));
|
|
return result;
|
|
}
|
|
|
|
static void AddNonAnimatingTransformLikePropertiesStyles(
|
|
const nsCSSPropertyIDSet& aNonAnimatingProperties, nsIFrame* aFrame,
|
|
Send aSendFlag, AnimationInfo& aAnimationInfo) {
|
|
auto appendFakeAnimation = [&aAnimationInfo, aSendFlag](
|
|
nsCSSPropertyID aProperty,
|
|
Animatable&& aBaseStyle) {
|
|
layers::Animation* animation =
|
|
(aSendFlag == Send::NextTransaction)
|
|
? aAnimationInfo.AddAnimationForNextTransaction()
|
|
: aAnimationInfo.AddAnimation();
|
|
animation->property() = aProperty;
|
|
animation->baseStyle() = std::move(aBaseStyle);
|
|
animation->easingFunction() = null_t();
|
|
animation->isNotAnimating() = true;
|
|
};
|
|
|
|
const nsStyleDisplay* display = aFrame->StyleDisplay();
|
|
// A simple optimization. We don't need to send offset-* properties if we
|
|
// don't have offset-path and offset-position.
|
|
// FIXME: Bug 1559232: Add offset-position here.
|
|
bool hasMotion =
|
|
!display->mOffsetPath.IsNone() ||
|
|
!aNonAnimatingProperties.HasProperty(eCSSProperty_offset_path);
|
|
|
|
for (nsCSSPropertyID id : aNonAnimatingProperties) {
|
|
switch (id) {
|
|
case eCSSProperty_transform:
|
|
if (!display->mTransform.IsNone()) {
|
|
TransformReferenceBox refBox(aFrame);
|
|
appendFakeAnimation(
|
|
id, ResolveTransformOperations(display->mTransform, refBox));
|
|
}
|
|
break;
|
|
case eCSSProperty_translate:
|
|
if (!display->mTranslate.IsNone()) {
|
|
TransformReferenceBox refBox(aFrame);
|
|
appendFakeAnimation(id,
|
|
ResolveTranslate(display->mTranslate, refBox));
|
|
}
|
|
break;
|
|
case eCSSProperty_rotate:
|
|
if (!display->mRotate.IsNone()) {
|
|
appendFakeAnimation(id, display->mRotate);
|
|
}
|
|
break;
|
|
case eCSSProperty_scale:
|
|
if (!display->mScale.IsNone()) {
|
|
appendFakeAnimation(id, display->mScale);
|
|
}
|
|
break;
|
|
case eCSSProperty_offset_path:
|
|
if (!display->mOffsetPath.IsNone()) {
|
|
appendFakeAnimation(id, NormalizeOffsetPath(display->mOffsetPath));
|
|
}
|
|
break;
|
|
case eCSSProperty_offset_distance:
|
|
if (hasMotion && !display->mOffsetDistance.IsDefinitelyZero()) {
|
|
appendFakeAnimation(id, display->mOffsetDistance);
|
|
}
|
|
break;
|
|
case eCSSProperty_offset_rotate:
|
|
if (hasMotion && (!display->mOffsetRotate.auto_ ||
|
|
display->mOffsetRotate.angle.ToDegrees() != 0.0)) {
|
|
appendFakeAnimation(id, display->mOffsetRotate);
|
|
}
|
|
break;
|
|
case eCSSProperty_offset_anchor:
|
|
if (hasMotion && !display->mOffsetAnchor.IsAuto()) {
|
|
appendFakeAnimation(id, display->mOffsetAnchor);
|
|
}
|
|
break;
|
|
default:
|
|
MOZ_ASSERT_UNREACHABLE("Unsupported transform-like properties");
|
|
}
|
|
}
|
|
}
|
|
|
|
static void AddAnimationsForDisplayItem(nsIFrame* aFrame,
|
|
nsDisplayListBuilder* aBuilder,
|
|
nsDisplayItem* aItem,
|
|
DisplayItemType aType, Send aSendFlag,
|
|
layers::LayersBackend aLayersBackend,
|
|
AnimationInfo& aAnimationInfo) {
|
|
if (aSendFlag == Send::NextTransaction) {
|
|
aAnimationInfo.ClearAnimationsForNextTransaction();
|
|
} else {
|
|
aAnimationInfo.ClearAnimations();
|
|
}
|
|
|
|
// Update the animation generation on the layer. We need to do this before
|
|
// any early returns since even if we don't add any animations to the
|
|
// layer, we still need to mark it as up-to-date with regards to animations.
|
|
// Otherwise, in RestyleManager we'll notice the discrepancy between the
|
|
// animation generation numbers and update the layer indefinitely.
|
|
EffectSet* effects = EffectSet::GetEffectSetForFrame(aFrame, aType);
|
|
uint64_t animationGeneration =
|
|
effects ? effects->GetAnimationGeneration() : 0;
|
|
aAnimationInfo.SetAnimationGeneration(animationGeneration);
|
|
|
|
EffectCompositor::ClearIsRunningOnCompositor(aFrame, aType);
|
|
const nsCSSPropertyIDSet& propertySet =
|
|
LayerAnimationInfo::GetCSSPropertiesFor(aType);
|
|
const nsTArray<RefPtr<dom::Animation>> matchedAnimations =
|
|
EffectCompositor::GetAnimationsForCompositor(aFrame, propertySet);
|
|
if (matchedAnimations.IsEmpty()) {
|
|
return;
|
|
}
|
|
|
|
// If the frame is not prerendered, bail out.
|
|
// Do this check only during layer construction; during updating the
|
|
// caller is required to check it appropriately.
|
|
if (aItem && !aItem->CanUseAsyncAnimations(aBuilder)) {
|
|
// EffectCompositor needs to know that we refused to run this animation
|
|
// asynchronously so that it will not throttle the main thread
|
|
// animation.
|
|
aFrame->SetProperty(nsIFrame::RefusedAsyncAnimationProperty(), true);
|
|
return;
|
|
}
|
|
|
|
const HashMap<nsCSSPropertyID, nsTArray<RefPtr<dom::Animation>>>
|
|
compositorAnimations =
|
|
GroupAnimationsByProperty(matchedAnimations, propertySet);
|
|
CompositorAnimationData data =
|
|
CreateAnimationData(aFrame, aItem, aType, aLayersBackend,
|
|
compositorAnimations.has(eCSSProperty_offset_path) ||
|
|
!aFrame->StyleDisplay()->mOffsetPath.IsNone()
|
|
? AnimationDataType::WithMotionPath
|
|
: AnimationDataType::WithoutMotionPath);
|
|
// Bug 1424900: Drop this pref check after shipping individual transforms.
|
|
// Bug 1582554: Drop this pref check after shipping motion path.
|
|
const bool hasMultipleTransformLikeProperties =
|
|
(StaticPrefs::layout_css_individual_transform_enabled() ||
|
|
StaticPrefs::layout_css_motion_path_enabled()) &&
|
|
aType == DisplayItemType::TYPE_TRANSFORM;
|
|
nsCSSPropertyIDSet nonAnimatingProperties =
|
|
nsCSSPropertyIDSet::TransformLikeProperties();
|
|
for (auto iter = compositorAnimations.iter(); !iter.done(); iter.next()) {
|
|
// Note: We can skip offset-* if there is no offset-path/offset-position
|
|
// animations and styles. However, it should be fine and may be better to
|
|
// send these information to the compositor because 1) they are simple data
|
|
// structure, 2) AddAnimationsForProperty() marks these animations as
|
|
// running on the composiror, so CanThrottle() returns true for them, and
|
|
// we avoid running these animations on the main thread.
|
|
bool added =
|
|
AddAnimationsForProperty(aFrame, effects, iter.get().value(), data,
|
|
iter.get().key(), aSendFlag, aAnimationInfo);
|
|
if (added && data.HasData()) {
|
|
// Only copy TransformLikeMetaData in the first animation property.
|
|
data.Clear();
|
|
}
|
|
|
|
if (hasMultipleTransformLikeProperties && added) {
|
|
nonAnimatingProperties.RemoveProperty(iter.get().key());
|
|
}
|
|
}
|
|
|
|
// If some transform-like properties have animations, but others not, and
|
|
// those non-animating transform-like properties have non-none
|
|
// transform/translate/rotate/scale styles or non-initial value for motion
|
|
// path properties, we also pass their styles into the compositor, so the
|
|
// final transform matrix (on the compositor) could take them into account.
|
|
if (hasMultipleTransformLikeProperties &&
|
|
// For these cases we don't need to send the property style values to
|
|
// the compositor:
|
|
// 1. No property has running animations on the compositor. (i.e. All
|
|
// properties should be handled by main thread)
|
|
// 2. All properties have running animations on the compositor.
|
|
// (i.e. Those running animations should override the styles.)
|
|
!nonAnimatingProperties.Equals(
|
|
nsCSSPropertyIDSet::TransformLikeProperties()) &&
|
|
!nonAnimatingProperties.IsEmpty()) {
|
|
AddNonAnimatingTransformLikePropertiesStyles(nonAnimatingProperties, aFrame,
|
|
aSendFlag, aAnimationInfo);
|
|
}
|
|
}
|
|
|
|
static uint64_t AddAnimationsForWebRender(
|
|
nsDisplayItem* aItem, mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder, wr::RenderRoot aRenderRoot) {
|
|
RefPtr<WebRenderAnimationData> animationData =
|
|
aManager->CommandBuilder()
|
|
.CreateOrRecycleWebRenderUserData<WebRenderAnimationData>(
|
|
aItem, aRenderRoot);
|
|
AnimationInfo& animationInfo = animationData->GetAnimationInfo();
|
|
AddAnimationsForDisplayItem(aItem->Frame(), aDisplayListBuilder, aItem,
|
|
aItem->GetType(), Send::Immediate,
|
|
layers::LayersBackend::LAYERS_WR, animationInfo);
|
|
animationInfo.StartPendingAnimations(
|
|
aManager->LayerManager()->GetAnimationReadyTime());
|
|
|
|
// Note that animationsId can be 0 (uninitialized in AnimationInfo) if there
|
|
// are no active animations.
|
|
uint64_t animationsId = animationInfo.GetCompositorAnimationsId();
|
|
if (!animationInfo.GetAnimations().IsEmpty()) {
|
|
OpAddCompositorAnimations anim(
|
|
CompositorAnimations(animationInfo.GetAnimations(), animationsId));
|
|
aManager->WrBridge()->AddWebRenderParentCommand(anim, aRenderRoot);
|
|
aManager->AddActiveCompositorAnimationId(animationsId);
|
|
} else if (animationsId) {
|
|
aManager->AddCompositorAnimationsIdForDiscard(animationsId);
|
|
animationsId = 0;
|
|
}
|
|
|
|
return animationsId;
|
|
}
|
|
|
|
static bool GenerateAndPushTextMask(nsIFrame* aFrame, gfxContext* aContext,
|
|
const nsRect& aFillRect,
|
|
nsDisplayListBuilder* aBuilder) {
|
|
if (aBuilder->IsForGenerateGlyphMask()) {
|
|
return false;
|
|
}
|
|
|
|
SVGObserverUtils::GetAndObserveBackgroundClip(aFrame);
|
|
|
|
// The main function of enabling background-clip:text property value.
|
|
// When a nsDisplayBackgroundImage detects "text" bg-clip style, it will call
|
|
// this function to
|
|
// 1. Generate a mask by all descendant text frames
|
|
// 2. Push the generated mask into aContext.
|
|
|
|
gfxContext* sourceCtx = aContext;
|
|
LayoutDeviceRect bounds = LayoutDeviceRect::FromAppUnits(
|
|
aFillRect, aFrame->PresContext()->AppUnitsPerDevPixel());
|
|
|
|
// Create a mask surface.
|
|
RefPtr<DrawTarget> sourceTarget = sourceCtx->GetDrawTarget();
|
|
RefPtr<DrawTarget> maskDT = sourceTarget->CreateClippedDrawTarget(
|
|
bounds.ToUnknownRect(), SurfaceFormat::A8);
|
|
if (!maskDT || !maskDT->IsValid()) {
|
|
return false;
|
|
}
|
|
RefPtr<gfxContext> maskCtx =
|
|
gfxContext::CreatePreservingTransformOrNull(maskDT);
|
|
MOZ_ASSERT(maskCtx);
|
|
maskCtx->Multiply(Matrix::Translation(bounds.TopLeft().ToUnknownPoint()));
|
|
|
|
// Shade text shape into mask A8 surface.
|
|
nsLayoutUtils::PaintFrame(
|
|
maskCtx, aFrame, nsRect(nsPoint(0, 0), aFrame->GetSize()),
|
|
NS_RGB(255, 255, 255), nsDisplayListBuilderMode::GenerateGlyph);
|
|
|
|
// Push the generated mask into aContext, so that the caller can pop and
|
|
// blend with it.
|
|
|
|
Matrix currentMatrix = sourceCtx->CurrentMatrix();
|
|
Matrix invCurrentMatrix = currentMatrix;
|
|
invCurrentMatrix.Invert();
|
|
|
|
RefPtr<SourceSurface> maskSurface = maskDT->Snapshot();
|
|
sourceCtx->PushGroupForBlendBack(gfxContentType::COLOR_ALPHA, 1.0,
|
|
maskSurface, invCurrentMatrix);
|
|
|
|
return true;
|
|
}
|
|
|
|
/* static */
|
|
void nsDisplayListBuilder::AddAnimationsAndTransitionsToLayer(
|
|
Layer* aLayer, nsDisplayListBuilder* aBuilder, nsDisplayItem* aItem,
|
|
nsIFrame* aFrame, DisplayItemType aType) {
|
|
// This function can be called in two ways: from
|
|
// nsDisplay*::BuildLayer while constructing a layer (with all
|
|
// pointers non-null), or from RestyleManager's handling of
|
|
// UpdateOpacityLayer/UpdateTransformLayer hints.
|
|
MOZ_ASSERT(!aBuilder == !aItem,
|
|
"should only be called in two configurations, with both "
|
|
"aBuilder and aItem, or with neither");
|
|
MOZ_ASSERT(!aItem || aFrame == aItem->Frame(), "frame mismatch");
|
|
|
|
// Only send animations to a layer that is actually using
|
|
// off-main-thread compositing.
|
|
LayersBackend backend = aLayer->Manager()->GetBackendType();
|
|
if (!(backend == layers::LayersBackend::LAYERS_CLIENT ||
|
|
backend == layers::LayersBackend::LAYERS_WR)) {
|
|
return;
|
|
}
|
|
|
|
Send sendFlag = !aBuilder ? Send::NextTransaction : Send::Immediate;
|
|
AnimationInfo& animationInfo = aLayer->GetAnimationInfo();
|
|
AddAnimationsForDisplayItem(aFrame, aBuilder, aItem, aType, sendFlag,
|
|
layers::LayersBackend::LAYERS_CLIENT,
|
|
animationInfo);
|
|
animationInfo.TransferMutatedFlagToLayer(aLayer);
|
|
}
|
|
|
|
nsDisplayWrapList* nsDisplayListBuilder::MergeItems(
|
|
nsTArray<nsDisplayWrapList*>& aItems) {
|
|
// For merging, we create a temporary item by cloning the last item of the
|
|
// mergeable items list. This ensures that the temporary item will have the
|
|
// correct frame and bounds.
|
|
nsDisplayWrapList* merged = nullptr;
|
|
|
|
for (nsDisplayWrapList* item : Reversed(aItems)) {
|
|
MOZ_ASSERT(item);
|
|
|
|
if (!merged) {
|
|
// Create the temporary item.
|
|
merged = item->Clone(this);
|
|
MOZ_ASSERT(merged);
|
|
|
|
AddTemporaryItem(merged);
|
|
} else {
|
|
// Merge the item properties (frame/bounds/etc) with the previously
|
|
// created temporary item.
|
|
MOZ_ASSERT(merged->CanMerge(item));
|
|
merged->Merge(item);
|
|
}
|
|
|
|
// Create nsDisplayWrapList that points to the internal display list of the
|
|
// item we are merging. This nsDisplayWrapList is added to the display list
|
|
// of the temporary item.
|
|
merged->MergeDisplayListFromItem(this, item);
|
|
}
|
|
|
|
return merged;
|
|
}
|
|
|
|
void nsDisplayListBuilder::AutoCurrentActiveScrolledRootSetter::
|
|
SetCurrentActiveScrolledRoot(
|
|
const ActiveScrolledRoot* aActiveScrolledRoot) {
|
|
MOZ_ASSERT(!mUsed);
|
|
|
|
// Set the builder's mCurrentActiveScrolledRoot.
|
|
mBuilder->mCurrentActiveScrolledRoot = aActiveScrolledRoot;
|
|
|
|
// We also need to adjust the builder's mCurrentContainerASR.
|
|
// mCurrentContainerASR needs to be an ASR that all the container's
|
|
// contents have finite bounds with respect to. If aActiveScrolledRoot
|
|
// is an ancestor ASR of mCurrentContainerASR, that means we need to
|
|
// set mCurrentContainerASR to aActiveScrolledRoot, because otherwise
|
|
// the items that will be created with aActiveScrolledRoot wouldn't
|
|
// have finite bounds with respect to mCurrentContainerASR. There's one
|
|
// exception, in the case where there's a content clip on the builder
|
|
// that is scrolled by a descendant ASR of aActiveScrolledRoot. This
|
|
// content clip will clip all items that are created while this
|
|
// AutoCurrentActiveScrolledRootSetter exists. This means that the items
|
|
// created during our lifetime will have finite bounds with respect to
|
|
// the content clip's ASR, even if the items' actual ASR is an ancestor
|
|
// of that. And it also means that mCurrentContainerASR only needs to be
|
|
// set to the content clip's ASR and not all the way to aActiveScrolledRoot.
|
|
// This case is tested by fixed-pos-scrolled-clip-opacity-layerize.html
|
|
// and fixed-pos-scrolled-clip-opacity-inside-layerize.html.
|
|
|
|
// finiteBoundsASR is the leafmost ASR that all items created during
|
|
// object's lifetime have finite bounds with respect to.
|
|
const ActiveScrolledRoot* finiteBoundsASR =
|
|
ActiveScrolledRoot::PickDescendant(mContentClipASR, aActiveScrolledRoot);
|
|
|
|
// mCurrentContainerASR is adjusted so that it's still an ancestor of
|
|
// finiteBoundsASR.
|
|
mBuilder->mCurrentContainerASR = ActiveScrolledRoot::PickAncestor(
|
|
mBuilder->mCurrentContainerASR, finiteBoundsASR);
|
|
|
|
// If we are entering out-of-flow content inside a CSS filter, mark
|
|
// scroll frames wrt. which the content is fixed as containing such content.
|
|
if (mBuilder->mFilterASR && ActiveScrolledRoot::IsAncestor(
|
|
aActiveScrolledRoot, mBuilder->mFilterASR)) {
|
|
for (const ActiveScrolledRoot* asr = mBuilder->mFilterASR;
|
|
asr && asr != aActiveScrolledRoot; asr = asr->mParent) {
|
|
asr->mScrollableFrame->SetHasOutOfFlowContentInsideFilter();
|
|
}
|
|
}
|
|
|
|
mUsed = true;
|
|
}
|
|
|
|
void nsDisplayListBuilder::AutoCurrentActiveScrolledRootSetter::
|
|
InsertScrollFrame(nsIScrollableFrame* aScrollableFrame) {
|
|
MOZ_ASSERT(!mUsed);
|
|
size_t descendantsEndIndex = mBuilder->mActiveScrolledRoots.Length();
|
|
const ActiveScrolledRoot* parentASR = mBuilder->mCurrentActiveScrolledRoot;
|
|
const ActiveScrolledRoot* asr =
|
|
mBuilder->AllocateActiveScrolledRoot(parentASR, aScrollableFrame);
|
|
mBuilder->mCurrentActiveScrolledRoot = asr;
|
|
|
|
// All child ASRs of parentASR that were created while this
|
|
// AutoCurrentActiveScrolledRootSetter object was on the stack belong to us
|
|
// now. Reparent them to asr.
|
|
for (size_t i = mDescendantsStartIndex; i < descendantsEndIndex; i++) {
|
|
ActiveScrolledRoot* descendantASR = mBuilder->mActiveScrolledRoots[i];
|
|
if (ActiveScrolledRoot::IsAncestor(parentASR, descendantASR)) {
|
|
descendantASR->IncrementDepth();
|
|
if (descendantASR->mParent == parentASR) {
|
|
descendantASR->mParent = asr;
|
|
}
|
|
}
|
|
}
|
|
|
|
mUsed = true;
|
|
}
|
|
|
|
/* static */
|
|
nsRect nsDisplayListBuilder::OutOfFlowDisplayData::ComputeVisibleRectForFrame(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame,
|
|
const nsRect& aVisibleRect, const nsRect& aDirtyRect,
|
|
nsRect* aOutDirtyRect) {
|
|
nsRect visible = aVisibleRect;
|
|
nsRect dirtyRectRelativeToDirtyFrame = aDirtyRect;
|
|
|
|
if (StaticPrefs::apz_allow_zooming() &&
|
|
nsLayoutUtils::IsFixedPosFrameInDisplayPort(aFrame) &&
|
|
aBuilder->IsPaintingToWindow()) {
|
|
dirtyRectRelativeToDirtyFrame =
|
|
nsRect(nsPoint(0, 0), aFrame->GetParent()->GetSize());
|
|
|
|
// If there's a visual viewport size set, restrict the amount of the
|
|
// fixed-position element we paint to the visual viewport. (In general
|
|
// the fixed-position element can be as large as the layout viewport,
|
|
// which at a high zoom level can cause us to paint too large of an
|
|
// area.)
|
|
PresShell* presShell = aFrame->PresShell();
|
|
if (presShell->IsVisualViewportSizeSet()) {
|
|
dirtyRectRelativeToDirtyFrame =
|
|
nsRect(presShell->GetVisualViewportOffset(),
|
|
presShell->GetVisualViewportSize());
|
|
// But if we have a displayport, expand it to the displayport, so
|
|
// that async-scrolling the visual viewport within the layout viewport
|
|
// will not checkerboard.
|
|
if (nsIFrame* rootScrollFrame = presShell->GetRootScrollFrame()) {
|
|
nsRect displayport;
|
|
// Note that the displayport here is already in the right coordinate
|
|
// space: it's relative to the scroll port (= layout viewport), but
|
|
// covers the visual viewport with some margins around it, which is
|
|
// exactly what we want.
|
|
if (nsLayoutUtils::GetHighResolutionDisplayPort(
|
|
rootScrollFrame->GetContent(), &displayport)) {
|
|
dirtyRectRelativeToDirtyFrame = displayport;
|
|
}
|
|
}
|
|
}
|
|
visible = dirtyRectRelativeToDirtyFrame;
|
|
if (StaticPrefs::apz_test_logging_enabled() &&
|
|
presShell->GetDocument()->IsContentDocument()) {
|
|
nsLayoutUtils::LogAdditionalTestData(
|
|
aBuilder, "fixedPosDisplayport",
|
|
ToString(CSSSize::FromAppUnits(visible)));
|
|
}
|
|
}
|
|
|
|
*aOutDirtyRect = dirtyRectRelativeToDirtyFrame - aFrame->GetPosition();
|
|
visible -= aFrame->GetPosition();
|
|
|
|
nsRect overflowRect = aFrame->GetVisualOverflowRect();
|
|
|
|
if (aFrame->IsTransformed() &&
|
|
mozilla::EffectCompositor::HasAnimationsForCompositor(
|
|
aFrame, DisplayItemType::TYPE_TRANSFORM)) {
|
|
/**
|
|
* Add a fuzz factor to the overflow rectangle so that elements only
|
|
* just out of view are pulled into the display list, so they can be
|
|
* prerendered if necessary.
|
|
*/
|
|
overflowRect.Inflate(nsPresContext::CSSPixelsToAppUnits(32));
|
|
}
|
|
|
|
visible.IntersectRect(visible, overflowRect);
|
|
aOutDirtyRect->IntersectRect(*aOutDirtyRect, overflowRect);
|
|
|
|
return visible;
|
|
}
|
|
|
|
nsDisplayListBuilder::nsDisplayListBuilder(nsIFrame* aReferenceFrame,
|
|
nsDisplayListBuilderMode aMode,
|
|
bool aBuildCaret,
|
|
bool aRetainingDisplayList)
|
|
: mReferenceFrame(aReferenceFrame),
|
|
mIgnoreScrollFrame(nullptr),
|
|
mCurrentActiveScrolledRoot(nullptr),
|
|
mCurrentContainerASR(nullptr),
|
|
mCurrentFrame(aReferenceFrame),
|
|
mCurrentReferenceFrame(aReferenceFrame),
|
|
mNeedsDisplayListBuild{},
|
|
mRootAGR(AnimatedGeometryRoot::CreateAGRForFrame(
|
|
aReferenceFrame, nullptr, true, aRetainingDisplayList)),
|
|
mCurrentAGR(mRootAGR),
|
|
mBuildingExtraPagesForPageNum(0),
|
|
mUsedAGRBudget(0),
|
|
mDirtyRect(-1, -1, -1, -1),
|
|
mGlassDisplayItem(nullptr),
|
|
mCaretFrame(nullptr),
|
|
mScrollInfoItemsForHoisting(nullptr),
|
|
mFirstClipChainToDestroy(nullptr),
|
|
mMode(aMode),
|
|
mTableBackgroundSet(nullptr),
|
|
mCurrentScrollParentId(ScrollableLayerGuid::NULL_SCROLL_ID),
|
|
mCurrentScrollbarTarget(ScrollableLayerGuid::NULL_SCROLL_ID),
|
|
mFilterASR(nullptr),
|
|
mContainsBlendMode(false),
|
|
mIsBuildingScrollbar(false),
|
|
mCurrentScrollbarWillHaveLayer(false),
|
|
mBuildCaret(aBuildCaret),
|
|
mRetainingDisplayList(aRetainingDisplayList),
|
|
mPartialUpdate(false),
|
|
mIgnoreSuppression(false),
|
|
mIncludeAllOutOfFlows(false),
|
|
mDescendIntoSubdocuments(true),
|
|
mSelectedFramesOnly(false),
|
|
mAllowMergingAndFlattening(true),
|
|
mWillComputePluginGeometry(false),
|
|
mInTransform(false),
|
|
mInEventsAndPluginsOnly(false),
|
|
mInFilter(false),
|
|
mInPageSequence(false),
|
|
mIsInChromePresContext(false),
|
|
mSyncDecodeImages(false),
|
|
mIsPaintingToWindow(false),
|
|
mIsPaintingForWebRender(false),
|
|
mIsCompositingCheap(false),
|
|
mContainsPluginItem(false),
|
|
mAncestorHasApzAwareEventHandler(false),
|
|
mHaveScrollableDisplayPort(false),
|
|
mWindowDraggingAllowed(false),
|
|
mIsBuildingForPopup(nsLayoutUtils::IsPopup(aReferenceFrame)),
|
|
mForceLayerForScrollParent(false),
|
|
mAsyncPanZoomEnabled(nsLayoutUtils::AsyncPanZoomEnabled(aReferenceFrame)),
|
|
mBuildingInvisibleItems(false),
|
|
mHitTestIsForVisibility(false),
|
|
mIsBuilding(false),
|
|
mInInvalidSubtree(false),
|
|
mDisablePartialUpdates(false),
|
|
mPartialBuildFailed(false),
|
|
mIsInActiveDocShell(false),
|
|
mBuildAsyncZoomContainer(false),
|
|
mBuildBackdropRootContainer(false),
|
|
mContainsBackdropFilter(false),
|
|
mHitTestArea(),
|
|
mHitTestInfo(CompositorHitTestInvisibleToHit) {
|
|
MOZ_COUNT_CTOR(nsDisplayListBuilder);
|
|
|
|
mBuildCompositorHitTestInfo = mAsyncPanZoomEnabled && IsForPainting();
|
|
|
|
ShouldRebuildDisplayListDueToPrefChange();
|
|
|
|
static_assert(
|
|
static_cast<uint32_t>(DisplayItemType::TYPE_MAX) < (1 << TYPE_BITS),
|
|
"Check TYPE_MAX should not overflow");
|
|
}
|
|
|
|
static PresShell* GetFocusedPresShell() {
|
|
nsPIDOMWindowOuter* focusedWnd =
|
|
nsFocusManager::GetFocusManager()->GetFocusedWindow();
|
|
if (!focusedWnd) {
|
|
return nullptr;
|
|
}
|
|
|
|
nsCOMPtr<nsIDocShell> focusedDocShell = focusedWnd->GetDocShell();
|
|
if (!focusedDocShell) {
|
|
return nullptr;
|
|
}
|
|
|
|
return focusedDocShell->GetPresShell();
|
|
}
|
|
|
|
void nsDisplayListBuilder::BeginFrame() {
|
|
nsCSSRendering::BeginFrameTreesLocked();
|
|
mCurrentAGR = mRootAGR;
|
|
mFrameToAnimatedGeometryRootMap.Put(mReferenceFrame, mRootAGR);
|
|
|
|
mIsPaintingToWindow = false;
|
|
mIgnoreSuppression = false;
|
|
mInTransform = false;
|
|
mInFilter = false;
|
|
mSyncDecodeImages = false;
|
|
for (auto& needsDisplayListBuild : mNeedsDisplayListBuild) {
|
|
needsDisplayListBuild = false;
|
|
}
|
|
|
|
for (auto& renderRootRect : mRenderRootRects) {
|
|
renderRootRect = LayoutDeviceRect();
|
|
}
|
|
|
|
if (!mBuildCaret) {
|
|
return;
|
|
}
|
|
|
|
RefPtr<PresShell> presShell = GetFocusedPresShell();
|
|
if (presShell) {
|
|
RefPtr<nsCaret> caret = presShell->GetCaret();
|
|
mCaretFrame = caret->GetPaintGeometry(&mCaretRect);
|
|
|
|
// The focused pres shell may not be in the document that we're
|
|
// painting, or be in a popup. Check if the display root for
|
|
// the caret matches the display root that we're painting, and
|
|
// only use it if it matches.
|
|
if (mCaretFrame &&
|
|
nsLayoutUtils::GetDisplayRootFrame(mCaretFrame) !=
|
|
nsLayoutUtils::GetDisplayRootFrame(mReferenceFrame)) {
|
|
mCaretFrame = nullptr;
|
|
}
|
|
}
|
|
}
|
|
|
|
void nsDisplayListBuilder::EndFrame() {
|
|
NS_ASSERTION(!mInInvalidSubtree,
|
|
"Someone forgot to cleanup mInInvalidSubtree!");
|
|
mFrameToAnimatedGeometryRootMap.Clear();
|
|
mAGRBudgetSet.Clear();
|
|
mActiveScrolledRoots.Clear();
|
|
mEffectsUpdates.Clear();
|
|
FreeClipChains();
|
|
FreeTemporaryItems();
|
|
nsCSSRendering::EndFrameTreesLocked();
|
|
mCaretFrame = nullptr;
|
|
}
|
|
|
|
void nsDisplayListBuilder::MarkFrameForDisplay(nsIFrame* aFrame,
|
|
nsIFrame* aStopAtFrame) {
|
|
mFramesMarkedForDisplay.AppendElement(aFrame);
|
|
for (nsIFrame* f = aFrame; f;
|
|
f = nsLayoutUtils::GetParentOrPlaceholderForCrossDoc(f)) {
|
|
if (f->GetStateBits() & NS_FRAME_FORCE_DISPLAY_LIST_DESCEND_INTO) {
|
|
return;
|
|
}
|
|
f->AddStateBits(NS_FRAME_FORCE_DISPLAY_LIST_DESCEND_INTO);
|
|
if (f == aStopAtFrame) {
|
|
// we've reached a frame that we know will be painted, so we can stop.
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
void nsDisplayListBuilder::AddFrameMarkedForDisplayIfVisible(nsIFrame* aFrame) {
|
|
mFramesMarkedForDisplayIfVisible.AppendElement(aFrame);
|
|
}
|
|
|
|
void nsDisplayListBuilder::MarkFrameForDisplayIfVisible(
|
|
nsIFrame* aFrame, nsIFrame* aStopAtFrame) {
|
|
AddFrameMarkedForDisplayIfVisible(aFrame);
|
|
for (nsIFrame* f = aFrame; f; f = nsLayoutUtils::GetDisplayListParent(f)) {
|
|
if (f->ForceDescendIntoIfVisible()) {
|
|
return;
|
|
}
|
|
f->SetForceDescendIntoIfVisible(true);
|
|
if (f == aStopAtFrame) {
|
|
// we've reached a frame that we know will be painted, so we can stop.
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
void nsDisplayListBuilder::SetGlassDisplayItem(nsDisplayItem* aItem) {
|
|
// Web pages or extensions could trigger the "Multiple glass backgrounds
|
|
// found?" warning by using -moz-appearance:win-borderless-glass etc on their
|
|
// own elements (as long as they are DocElementBoxFrames, which is rare as
|
|
// each xul doc only gets one near the root). We only care about first one,
|
|
// since that will be the background of the root window.
|
|
|
|
if (IsPartialUpdate()) {
|
|
if (aItem->Frame()->IsDocElementBoxFrame()) {
|
|
#ifdef DEBUG
|
|
if (mHasGlassItemDuringPartial) {
|
|
NS_WARNING("Multiple glass backgrounds found?");
|
|
} else
|
|
#endif
|
|
if (!mHasGlassItemDuringPartial) {
|
|
mHasGlassItemDuringPartial = true;
|
|
aItem->SetIsGlassItem();
|
|
}
|
|
}
|
|
return;
|
|
}
|
|
|
|
if (aItem->Frame()->IsDocElementBoxFrame()) {
|
|
#ifdef DEBUG
|
|
if (mGlassDisplayItem) {
|
|
NS_WARNING("Multiple glass backgrounds found?");
|
|
} else
|
|
#endif
|
|
if (!mGlassDisplayItem) {
|
|
mGlassDisplayItem = aItem;
|
|
mGlassDisplayItem->SetIsGlassItem();
|
|
}
|
|
}
|
|
}
|
|
|
|
bool nsDisplayListBuilder::NeedToForceTransparentSurfaceForItem(
|
|
nsDisplayItem* aItem) {
|
|
return aItem == mGlassDisplayItem || aItem->ClearsBackground();
|
|
}
|
|
|
|
AnimatedGeometryRoot* nsDisplayListBuilder::WrapAGRForFrame(
|
|
nsIFrame* aAnimatedGeometryRoot, bool aIsAsync,
|
|
AnimatedGeometryRoot* aParent /* = nullptr */) {
|
|
DebugOnly<bool> dummy;
|
|
MOZ_ASSERT(IsAnimatedGeometryRoot(aAnimatedGeometryRoot, dummy) == AGR_YES);
|
|
|
|
RefPtr<AnimatedGeometryRoot> result;
|
|
if (!mFrameToAnimatedGeometryRootMap.Get(aAnimatedGeometryRoot, &result)) {
|
|
MOZ_ASSERT(nsLayoutUtils::IsAncestorFrameCrossDoc(RootReferenceFrame(),
|
|
aAnimatedGeometryRoot));
|
|
RefPtr<AnimatedGeometryRoot> parent = aParent;
|
|
if (!parent) {
|
|
nsIFrame* parentFrame =
|
|
nsLayoutUtils::GetCrossDocParentFrame(aAnimatedGeometryRoot);
|
|
if (parentFrame) {
|
|
bool isAsync;
|
|
nsIFrame* parentAGRFrame =
|
|
FindAnimatedGeometryRootFrameFor(parentFrame, isAsync);
|
|
parent = WrapAGRForFrame(parentAGRFrame, isAsync);
|
|
}
|
|
}
|
|
result = AnimatedGeometryRoot::CreateAGRForFrame(
|
|
aAnimatedGeometryRoot, parent, aIsAsync, IsRetainingDisplayList());
|
|
mFrameToAnimatedGeometryRootMap.Put(aAnimatedGeometryRoot, result);
|
|
}
|
|
MOZ_ASSERT(!aParent || result->mParentAGR == aParent);
|
|
return result;
|
|
}
|
|
|
|
AnimatedGeometryRoot* nsDisplayListBuilder::AnimatedGeometryRootForASR(
|
|
const ActiveScrolledRoot* aASR) {
|
|
if (!aASR) {
|
|
return GetRootAnimatedGeometryRoot();
|
|
}
|
|
nsIFrame* scrolledFrame = aASR->mScrollableFrame->GetScrolledFrame();
|
|
return FindAnimatedGeometryRootFor(scrolledFrame);
|
|
}
|
|
|
|
AnimatedGeometryRoot* nsDisplayListBuilder::FindAnimatedGeometryRootFor(
|
|
nsIFrame* aFrame) {
|
|
if (!IsPaintingToWindow()) {
|
|
return mRootAGR;
|
|
}
|
|
if (aFrame == mCurrentFrame) {
|
|
return mCurrentAGR;
|
|
}
|
|
RefPtr<AnimatedGeometryRoot> result;
|
|
if (mFrameToAnimatedGeometryRootMap.Get(aFrame, &result)) {
|
|
return result;
|
|
}
|
|
|
|
bool isAsync;
|
|
nsIFrame* agrFrame = FindAnimatedGeometryRootFrameFor(aFrame, isAsync);
|
|
result = WrapAGRForFrame(agrFrame, isAsync);
|
|
mFrameToAnimatedGeometryRootMap.Put(aFrame, result);
|
|
return result;
|
|
}
|
|
|
|
AnimatedGeometryRoot* nsDisplayListBuilder::FindAnimatedGeometryRootFor(
|
|
nsDisplayItem* aItem) {
|
|
if (aItem->ShouldFixToViewport(this)) {
|
|
// Make its active scrolled root be the active scrolled root of
|
|
// the enclosing viewport, since it shouldn't be scrolled by scrolled
|
|
// frames in its document. InvalidateFixedBackgroundFramesFromList in
|
|
// nsGfxScrollFrame will not repaint this item when scrolling occurs.
|
|
nsIFrame* viewportFrame = nsLayoutUtils::GetClosestFrameOfType(
|
|
aItem->Frame(), LayoutFrameType::Viewport, RootReferenceFrame());
|
|
if (viewportFrame) {
|
|
return FindAnimatedGeometryRootFor(viewportFrame);
|
|
}
|
|
}
|
|
return FindAnimatedGeometryRootFor(aItem->Frame());
|
|
}
|
|
|
|
void nsDisplayListBuilder::UpdateShouldBuildAsyncZoomContainer() {
|
|
Document* document = mReferenceFrame->PresContext()->Document();
|
|
mBuildAsyncZoomContainer = nsLayoutUtils::AllowZoomingForDocument(document);
|
|
}
|
|
|
|
void nsDisplayListBuilder::UpdateShouldBuildBackdropRootContainer() {
|
|
mBuildBackdropRootContainer =
|
|
StaticPrefs::layout_css_backdrop_filter_enabled();
|
|
}
|
|
|
|
// Certain prefs may cause display list items to be added or removed when they
|
|
// are toggled. In those cases, we need to fully rebuild the display list.
|
|
bool nsDisplayListBuilder::ShouldRebuildDisplayListDueToPrefChange() {
|
|
bool shouldRebuild = false;
|
|
|
|
// If we transition between wrapping the RCD-RSF contents into an async
|
|
// zoom container vs. not, we need to rebuild the display list. This only
|
|
// happens when the zooming or container scrolling prefs are toggled
|
|
// (manually by the user, or during test setup).
|
|
bool didBuildAsyncZoomContainer = mBuildAsyncZoomContainer;
|
|
UpdateShouldBuildAsyncZoomContainer();
|
|
if (didBuildAsyncZoomContainer != mBuildAsyncZoomContainer) {
|
|
shouldRebuild = true;
|
|
}
|
|
|
|
// If backdrop-filter is enabled, backdrop root containers are added to the
|
|
// display list. When the pref is toggled these containers may be added or
|
|
// removed, so the display list should be rebuilt.
|
|
bool didBuildBackdropRootContainer = mBuildBackdropRootContainer;
|
|
UpdateShouldBuildBackdropRootContainer();
|
|
if (didBuildBackdropRootContainer != mBuildBackdropRootContainer) {
|
|
shouldRebuild = true;
|
|
}
|
|
|
|
return shouldRebuild;
|
|
}
|
|
|
|
bool nsDisplayListBuilder::MarkOutOfFlowFrameForDisplay(nsIFrame* aDirtyFrame,
|
|
nsIFrame* aFrame) {
|
|
MOZ_ASSERT(aFrame->GetParent() == aDirtyFrame);
|
|
nsRect dirty;
|
|
nsRect visible = OutOfFlowDisplayData::ComputeVisibleRectForFrame(
|
|
this, aFrame, GetVisibleRect(), GetDirtyRect(), &dirty);
|
|
if (!(aFrame->GetStateBits() & NS_FRAME_FORCE_DISPLAY_LIST_DESCEND_INTO) &&
|
|
visible.IsEmpty()) {
|
|
return false;
|
|
}
|
|
|
|
// Only MarkFrameForDisplay if we're dirty. If this is a nested out-of-flow
|
|
// frame, then it will also mark any outer frames to ensure that building
|
|
// reaches the dirty feame.
|
|
if (!dirty.IsEmpty() || aFrame->ForceDescendIntoIfVisible()) {
|
|
MarkFrameForDisplay(aFrame, aDirtyFrame);
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
static void UnmarkFrameForDisplay(nsIFrame* aFrame, nsIFrame* aStopAtFrame) {
|
|
for (nsIFrame* f = aFrame; f;
|
|
f = nsLayoutUtils::GetParentOrPlaceholderForCrossDoc(f)) {
|
|
if (!(f->GetStateBits() & NS_FRAME_FORCE_DISPLAY_LIST_DESCEND_INTO)) {
|
|
return;
|
|
}
|
|
f->RemoveStateBits(NS_FRAME_FORCE_DISPLAY_LIST_DESCEND_INTO);
|
|
if (f == aStopAtFrame) {
|
|
// we've reached a frame that we know will be painted, so we can stop.
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
static void UnmarkFrameForDisplayIfVisible(nsIFrame* aFrame) {
|
|
for (nsIFrame* f = aFrame; f; f = nsLayoutUtils::GetDisplayListParent(f)) {
|
|
if (!f->ForceDescendIntoIfVisible()) {
|
|
return;
|
|
}
|
|
f->SetForceDescendIntoIfVisible(false);
|
|
}
|
|
}
|
|
|
|
nsDisplayListBuilder::~nsDisplayListBuilder() {
|
|
NS_ASSERTION(mFramesMarkedForDisplay.Length() == 0,
|
|
"All frames should have been unmarked");
|
|
NS_ASSERTION(mFramesWithOOFData.Length() == 0,
|
|
"All OOF data should have been removed");
|
|
NS_ASSERTION(mPresShellStates.Length() == 0,
|
|
"All presshells should have been exited");
|
|
|
|
DisplayItemClipChain* c = mFirstClipChainToDestroy;
|
|
while (c) {
|
|
DisplayItemClipChain* next = c->mNextClipChainToDestroy;
|
|
c->DisplayItemClipChain::~DisplayItemClipChain();
|
|
c = next;
|
|
}
|
|
|
|
MOZ_COUNT_DTOR(nsDisplayListBuilder);
|
|
}
|
|
|
|
uint32_t nsDisplayListBuilder::GetBackgroundPaintFlags() {
|
|
uint32_t flags = 0;
|
|
if (mSyncDecodeImages) {
|
|
flags |= nsCSSRendering::PAINTBG_SYNC_DECODE_IMAGES;
|
|
}
|
|
if (mIsPaintingToWindow) {
|
|
flags |= nsCSSRendering::PAINTBG_TO_WINDOW;
|
|
}
|
|
return flags;
|
|
}
|
|
|
|
uint32_t nsDisplayListBuilder::GetImageDecodeFlags() const {
|
|
uint32_t flags = imgIContainer::FLAG_ASYNC_NOTIFY;
|
|
if (mSyncDecodeImages) {
|
|
flags |= imgIContainer::FLAG_SYNC_DECODE;
|
|
} else {
|
|
flags |= imgIContainer::FLAG_SYNC_DECODE_IF_FAST;
|
|
}
|
|
if (mIsPaintingToWindow) {
|
|
flags |= imgIContainer::FLAG_HIGH_QUALITY_SCALING;
|
|
}
|
|
return flags;
|
|
}
|
|
|
|
void nsDisplayListBuilder::ComputeDefaultRenderRootRect(
|
|
LayoutDeviceIntSize aClientSize) {
|
|
LayoutDeviceIntRegion cutout;
|
|
LayoutDeviceIntRect clientRect(LayoutDeviceIntPoint(), aClientSize);
|
|
cutout.OrWith(clientRect);
|
|
cutout.SubOut(
|
|
RoundedToInt(mRenderRootRects[mozilla::wr::RenderRoot::Content]));
|
|
|
|
mRenderRootRects[mozilla::wr::RenderRoot::Default] =
|
|
LayoutDeviceRect(cutout.GetBounds());
|
|
}
|
|
|
|
void nsDisplayListBuilder::SubtractFromVisibleRegion(nsRegion* aVisibleRegion,
|
|
const nsRegion& aRegion) {
|
|
if (aRegion.IsEmpty()) {
|
|
return;
|
|
}
|
|
|
|
nsRegion tmp;
|
|
tmp.Sub(*aVisibleRegion, aRegion);
|
|
// Don't let *aVisibleRegion get too complex, but don't let it fluff out
|
|
// to its bounds either, which can be very bad (see bug 516740).
|
|
// Do let aVisibleRegion get more complex if by doing so we reduce its
|
|
// area by at least half.
|
|
if (GetAccurateVisibleRegions() || tmp.GetNumRects() <= 15 ||
|
|
tmp.Area() <= aVisibleRegion->Area() / 2) {
|
|
*aVisibleRegion = tmp;
|
|
}
|
|
}
|
|
|
|
nsCaret* nsDisplayListBuilder::GetCaret() {
|
|
RefPtr<nsCaret> caret = CurrentPresShellState()->mPresShell->GetCaret();
|
|
return caret;
|
|
}
|
|
|
|
void nsDisplayListBuilder::IncrementPresShellPaintCount(PresShell* aPresShell) {
|
|
if (mIsPaintingToWindow) {
|
|
mReferenceFrame->AddPaintedPresShell(aPresShell);
|
|
aPresShell->IncrementPaintCount();
|
|
}
|
|
}
|
|
|
|
void nsDisplayListBuilder::EnterPresShell(nsIFrame* aReferenceFrame,
|
|
bool aPointerEventsNoneDoc) {
|
|
PresShellState* state = mPresShellStates.AppendElement();
|
|
state->mPresShell = aReferenceFrame->PresShell();
|
|
state->mFirstFrameMarkedForDisplay = mFramesMarkedForDisplay.Length();
|
|
state->mFirstFrameWithOOFData = mFramesWithOOFData.Length();
|
|
|
|
nsIScrollableFrame* sf = state->mPresShell->GetRootScrollFrameAsScrollable();
|
|
if (sf && IsInSubdocument()) {
|
|
// We are forcing a rebuild of nsDisplayCanvasBackgroundColor to make sure
|
|
// that the canvas background color will be set correctly, and that only one
|
|
// unscrollable item will be created.
|
|
// This is done to avoid, for example, a case where only scrollbar frames
|
|
// are invalidated - we would skip creating nsDisplayCanvasBackgroundColor
|
|
// and possibly end up with an extra nsDisplaySolidColor item.
|
|
// We skip this for the root document, since we don't want to use
|
|
// MarkFrameForDisplayIfVisible before ComputeRebuildRegion. We'll
|
|
// do it manually there.
|
|
nsCanvasFrame* canvasFrame = do_QueryFrame(sf->GetScrolledFrame());
|
|
if (canvasFrame) {
|
|
MarkFrameForDisplayIfVisible(canvasFrame, aReferenceFrame);
|
|
}
|
|
}
|
|
|
|
#ifdef DEBUG
|
|
state->mAutoLayoutPhase.emplace(aReferenceFrame->PresContext(),
|
|
nsLayoutPhase::DisplayListBuilding);
|
|
#endif
|
|
|
|
state->mPresShell->UpdateCanvasBackground();
|
|
|
|
bool buildCaret = mBuildCaret;
|
|
if (mIgnoreSuppression || !state->mPresShell->IsPaintingSuppressed()) {
|
|
state->mIsBackgroundOnly = false;
|
|
} else {
|
|
state->mIsBackgroundOnly = true;
|
|
buildCaret = false;
|
|
}
|
|
|
|
bool pointerEventsNone = aPointerEventsNoneDoc;
|
|
if (IsInSubdocument()) {
|
|
pointerEventsNone |= mPresShellStates[mPresShellStates.Length() - 2]
|
|
.mInsidePointerEventsNoneDoc;
|
|
}
|
|
state->mInsidePointerEventsNoneDoc = pointerEventsNone;
|
|
|
|
state->mPresShellIgnoreScrollFrame =
|
|
state->mPresShell->IgnoringViewportScrolling()
|
|
? state->mPresShell->GetRootScrollFrame()
|
|
: nullptr;
|
|
|
|
nsPresContext* pc = aReferenceFrame->PresContext();
|
|
mIsInChromePresContext = pc->IsChrome();
|
|
nsIDocShell* docShell = pc->GetDocShell();
|
|
|
|
if (docShell) {
|
|
docShell->GetWindowDraggingAllowed(&mWindowDraggingAllowed);
|
|
}
|
|
|
|
state->mTouchEventPrefEnabledDoc = dom::TouchEvent::PrefEnabled(docShell);
|
|
|
|
if (!buildCaret) {
|
|
return;
|
|
}
|
|
|
|
// Caret frames add visual area to their frame, but we don't update the
|
|
// overflow area. Use flags to make sure we build display items for that frame
|
|
// instead.
|
|
if (mCaretFrame && mCaretFrame->PresShell() == state->mPresShell) {
|
|
MarkFrameForDisplay(mCaretFrame, aReferenceFrame);
|
|
}
|
|
}
|
|
|
|
// A non-blank paint is a paint that does not just contain the canvas
|
|
// background.
|
|
static bool DisplayListIsNonBlank(nsDisplayList* aList) {
|
|
for (nsDisplayItem* i : *aList) {
|
|
switch (i->GetType()) {
|
|
case DisplayItemType::TYPE_COMPOSITOR_HITTEST_INFO:
|
|
case DisplayItemType::TYPE_CANVAS_BACKGROUND_COLOR:
|
|
case DisplayItemType::TYPE_CANVAS_BACKGROUND_IMAGE:
|
|
continue;
|
|
case DisplayItemType::TYPE_SOLID_COLOR:
|
|
case DisplayItemType::TYPE_BACKGROUND:
|
|
case DisplayItemType::TYPE_BACKGROUND_COLOR:
|
|
if (i->Frame()->IsCanvasFrame()) {
|
|
continue;
|
|
}
|
|
return true;
|
|
default:
|
|
return true;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
|
|
// A contentful paint is a paint that does contains DOM content (text,
|
|
// images, non-blank canvases, SVG): "First Contentful Paint entry
|
|
// contains a DOMHighResTimeStamp reporting the time when the browser
|
|
// first rendered any text, image (including background images),
|
|
// non-white canvas or SVG. This excludes any content of iframes, but
|
|
// includes text with pending webfonts. This is the first time users
|
|
// could start consuming page content."
|
|
static bool DisplayListIsContentful(nsDisplayList* aList) {
|
|
for (nsDisplayItem* i : *aList) {
|
|
DisplayItemType type = i->GetType();
|
|
nsDisplayList* children = i->GetChildren();
|
|
|
|
switch (type) {
|
|
case DisplayItemType::TYPE_SUBDOCUMENT: // iframes are ignored
|
|
break;
|
|
// CANVASes check if they may have been modified (as a stand-in
|
|
// actually tracking all modifications)
|
|
default:
|
|
if (i->IsContentful()) {
|
|
return true;
|
|
}
|
|
if (children) {
|
|
if (DisplayListIsContentful(children)) {
|
|
return true;
|
|
}
|
|
}
|
|
break;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
|
|
void nsDisplayListBuilder::LeavePresShell(nsIFrame* aReferenceFrame,
|
|
nsDisplayList* aPaintedContents) {
|
|
NS_ASSERTION(
|
|
CurrentPresShellState()->mPresShell == aReferenceFrame->PresShell(),
|
|
"Presshell mismatch");
|
|
|
|
if (mIsPaintingToWindow && aPaintedContents) {
|
|
nsPresContext* pc = aReferenceFrame->PresContext();
|
|
if (!pc->HadNonBlankPaint()) {
|
|
if (!CurrentPresShellState()->mIsBackgroundOnly &&
|
|
DisplayListIsNonBlank(aPaintedContents)) {
|
|
pc->NotifyNonBlankPaint();
|
|
}
|
|
}
|
|
if (!pc->HadContentfulPaint()) {
|
|
if (!CurrentPresShellState()->mIsBackgroundOnly &&
|
|
DisplayListIsContentful(aPaintedContents)) {
|
|
pc->NotifyContentfulPaint();
|
|
}
|
|
}
|
|
}
|
|
|
|
ResetMarkedFramesForDisplayList(aReferenceFrame);
|
|
mPresShellStates.RemoveLastElement();
|
|
|
|
if (!mPresShellStates.IsEmpty()) {
|
|
nsPresContext* pc = CurrentPresContext();
|
|
nsIDocShell* docShell = pc->GetDocShell();
|
|
if (docShell) {
|
|
docShell->GetWindowDraggingAllowed(&mWindowDraggingAllowed);
|
|
}
|
|
mIsInChromePresContext = pc->IsChrome();
|
|
} else {
|
|
mCurrentAGR = mRootAGR;
|
|
|
|
for (uint32_t i = 0; i < mFramesMarkedForDisplayIfVisible.Length(); ++i) {
|
|
UnmarkFrameForDisplayIfVisible(mFramesMarkedForDisplayIfVisible[i]);
|
|
}
|
|
mFramesMarkedForDisplayIfVisible.SetLength(0);
|
|
}
|
|
}
|
|
|
|
void nsDisplayListBuilder::FreeClipChains() {
|
|
// Iterate the clip chains from newest to oldest (forward
|
|
// iteration), so that we destroy descendants first which
|
|
// will drop the ref count on their ancestors.
|
|
DisplayItemClipChain** indirect = &mFirstClipChainToDestroy;
|
|
|
|
while (*indirect) {
|
|
if (!(*indirect)->mRefCount) {
|
|
DisplayItemClipChain* next = (*indirect)->mNextClipChainToDestroy;
|
|
|
|
mClipDeduplicator.erase(*indirect);
|
|
(*indirect)->DisplayItemClipChain::~DisplayItemClipChain();
|
|
Destroy(DisplayListArenaObjectId::CLIPCHAIN, *indirect);
|
|
|
|
*indirect = next;
|
|
} else {
|
|
indirect = &(*indirect)->mNextClipChainToDestroy;
|
|
}
|
|
}
|
|
}
|
|
|
|
void nsDisplayListBuilder::FreeTemporaryItems() {
|
|
for (nsDisplayItem* i : mTemporaryItems) {
|
|
// Temporary display items are not added to the frames.
|
|
MOZ_ASSERT(i->Frame());
|
|
i->RemoveFrame(i->Frame());
|
|
i->Destroy(this);
|
|
}
|
|
|
|
mTemporaryItems.Clear();
|
|
}
|
|
|
|
void nsDisplayListBuilder::ResetMarkedFramesForDisplayList(
|
|
nsIFrame* aReferenceFrame) {
|
|
// Unmark and pop off the frames marked for display in this pres shell.
|
|
uint32_t firstFrameForShell =
|
|
CurrentPresShellState()->mFirstFrameMarkedForDisplay;
|
|
for (uint32_t i = firstFrameForShell; i < mFramesMarkedForDisplay.Length();
|
|
++i) {
|
|
UnmarkFrameForDisplay(mFramesMarkedForDisplay[i], aReferenceFrame);
|
|
}
|
|
mFramesMarkedForDisplay.SetLength(firstFrameForShell);
|
|
|
|
firstFrameForShell = CurrentPresShellState()->mFirstFrameWithOOFData;
|
|
for (uint32_t i = firstFrameForShell; i < mFramesWithOOFData.Length(); ++i) {
|
|
mFramesWithOOFData[i]->RemoveProperty(OutOfFlowDisplayDataProperty());
|
|
}
|
|
mFramesWithOOFData.SetLength(firstFrameForShell);
|
|
}
|
|
|
|
void nsDisplayListBuilder::ClearFixedBackgroundDisplayData() {
|
|
CurrentPresShellState()->mFixedBackgroundDisplayData = Nothing();
|
|
}
|
|
|
|
void nsDisplayListBuilder::MarkFramesForDisplayList(
|
|
nsIFrame* aDirtyFrame, const nsFrameList& aFrames) {
|
|
bool markedFrames = false;
|
|
for (nsIFrame* e : aFrames) {
|
|
// Skip the AccessibleCaret frame when building no caret.
|
|
if (!IsBuildingCaret()) {
|
|
nsIContent* content = e->GetContent();
|
|
if (content && content->IsInNativeAnonymousSubtree() &&
|
|
content->IsElement()) {
|
|
auto classList = content->AsElement()->ClassList();
|
|
if (classList->Contains(NS_LITERAL_STRING("moz-accessiblecaret"))) {
|
|
continue;
|
|
}
|
|
}
|
|
}
|
|
if (MarkOutOfFlowFrameForDisplay(aDirtyFrame, e)) {
|
|
markedFrames = true;
|
|
}
|
|
}
|
|
|
|
if (markedFrames) {
|
|
// mClipState.GetClipChainForContainingBlockDescendants can return pointers
|
|
// to objects on the stack, so we need to clone the chain.
|
|
const DisplayItemClipChain* clipChain =
|
|
CopyWholeChain(mClipState.GetClipChainForContainingBlockDescendants());
|
|
const DisplayItemClipChain* combinedClipChain =
|
|
mClipState.GetCurrentCombinedClipChain(this);
|
|
const ActiveScrolledRoot* asr = mCurrentActiveScrolledRoot;
|
|
OutOfFlowDisplayData* data = new OutOfFlowDisplayData(
|
|
clipChain, combinedClipChain, asr, GetVisibleRect(), GetDirtyRect());
|
|
aDirtyFrame->SetProperty(
|
|
nsDisplayListBuilder::OutOfFlowDisplayDataProperty(), data);
|
|
mFramesWithOOFData.AppendElement(aDirtyFrame);
|
|
}
|
|
|
|
if (!aDirtyFrame->GetParent()) {
|
|
// This is the viewport frame of aDirtyFrame's presshell.
|
|
// Store the current display data so that it can be used for fixed
|
|
// background images.
|
|
NS_ASSERTION(
|
|
CurrentPresShellState()->mPresShell == aDirtyFrame->PresShell(),
|
|
"Presshell mismatch");
|
|
MOZ_ASSERT(!CurrentPresShellState()->mFixedBackgroundDisplayData,
|
|
"already traversed this presshell's root frame?");
|
|
|
|
const DisplayItemClipChain* clipChain =
|
|
CopyWholeChain(mClipState.GetClipChainForContainingBlockDescendants());
|
|
const DisplayItemClipChain* combinedClipChain =
|
|
mClipState.GetCurrentCombinedClipChain(this);
|
|
const ActiveScrolledRoot* asr = mCurrentActiveScrolledRoot;
|
|
CurrentPresShellState()->mFixedBackgroundDisplayData.emplace(
|
|
clipChain, combinedClipChain, asr, GetVisibleRect(), GetDirtyRect());
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Mark all preserve-3d children with
|
|
* NS_FRAME_FORCE_DISPLAY_LIST_DESCEND_INTO to make sure
|
|
* nsFrame::BuildDisplayListForChild() would visit them. Also compute
|
|
* dirty rect for preserve-3d children.
|
|
*
|
|
* @param aDirtyFrame is the frame to mark children extending context.
|
|
*/
|
|
void nsDisplayListBuilder::MarkPreserve3DFramesForDisplayList(
|
|
nsIFrame* aDirtyFrame) {
|
|
AutoTArray<nsIFrame::ChildList, 4> childListArray;
|
|
aDirtyFrame->GetChildLists(&childListArray);
|
|
nsIFrame::ChildListArrayIterator lists(childListArray);
|
|
for (; !lists.IsDone(); lists.Next()) {
|
|
nsFrameList::Enumerator childFrames(lists.CurrentList());
|
|
for (; !childFrames.AtEnd(); childFrames.Next()) {
|
|
nsIFrame* child = childFrames.get();
|
|
if (child->Combines3DTransformWithAncestors()) {
|
|
MarkFrameForDisplay(child, aDirtyFrame);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
ActiveScrolledRoot* nsDisplayListBuilder::AllocateActiveScrolledRoot(
|
|
const ActiveScrolledRoot* aParent, nsIScrollableFrame* aScrollableFrame) {
|
|
RefPtr<ActiveScrolledRoot> asr = ActiveScrolledRoot::CreateASRForFrame(
|
|
aParent, aScrollableFrame, IsRetainingDisplayList());
|
|
mActiveScrolledRoots.AppendElement(asr);
|
|
return asr;
|
|
}
|
|
|
|
const DisplayItemClipChain* nsDisplayListBuilder::AllocateDisplayItemClipChain(
|
|
const DisplayItemClip& aClip, const ActiveScrolledRoot* aASR,
|
|
const DisplayItemClipChain* aParent) {
|
|
MOZ_ASSERT(!(aParent && aParent->mOnStack));
|
|
void* p = Allocate(sizeof(DisplayItemClipChain),
|
|
DisplayListArenaObjectId::CLIPCHAIN);
|
|
DisplayItemClipChain* c = new (KnownNotNull, p)
|
|
DisplayItemClipChain(aClip, aASR, aParent, mFirstClipChainToDestroy);
|
|
#ifdef DEBUG
|
|
c->mOnStack = false;
|
|
#endif
|
|
auto result = mClipDeduplicator.insert(c);
|
|
if (!result.second) {
|
|
// An equivalent clip chain item was already created, so let's return that
|
|
// instead. Destroy the one we just created.
|
|
// Note that this can cause clip chains from different coordinate systems to
|
|
// collapse into the same clip chain object, because clip chains do not keep
|
|
// track of the reference frame that they were created in.
|
|
c->DisplayItemClipChain::~DisplayItemClipChain();
|
|
Destroy(DisplayListArenaObjectId::CLIPCHAIN, c);
|
|
return *(result.first);
|
|
}
|
|
mFirstClipChainToDestroy = c;
|
|
return c;
|
|
}
|
|
|
|
struct ClipChainItem {
|
|
DisplayItemClip clip;
|
|
const ActiveScrolledRoot* asr;
|
|
};
|
|
|
|
const DisplayItemClipChain* nsDisplayListBuilder::CreateClipChainIntersection(
|
|
const DisplayItemClipChain* aAncestor,
|
|
const DisplayItemClipChain* aLeafClip1,
|
|
const DisplayItemClipChain* aLeafClip2) {
|
|
AutoTArray<ClipChainItem, 8> intersectedClips;
|
|
|
|
const DisplayItemClipChain* clip1 = aLeafClip1;
|
|
const DisplayItemClipChain* clip2 = aLeafClip2;
|
|
|
|
const ActiveScrolledRoot* asr = ActiveScrolledRoot::PickDescendant(
|
|
clip1 ? clip1->mASR : nullptr, clip2 ? clip2->mASR : nullptr);
|
|
|
|
// Build up the intersection from the leaf to the root and put it into
|
|
// intersectedClips. The loop below will convert intersectedClips into an
|
|
// actual DisplayItemClipChain.
|
|
// (We need to do this in two passes because we need the parent clip in order
|
|
// to create the DisplayItemClipChain object, but the parent clip has not
|
|
// been created at that point.)
|
|
while (!aAncestor || asr != aAncestor->mASR) {
|
|
if (clip1 && clip1->mASR == asr) {
|
|
if (clip2 && clip2->mASR == asr) {
|
|
DisplayItemClip intersection = clip1->mClip;
|
|
intersection.IntersectWith(clip2->mClip);
|
|
intersectedClips.AppendElement(ClipChainItem{intersection, asr});
|
|
clip2 = clip2->mParent;
|
|
} else {
|
|
intersectedClips.AppendElement(ClipChainItem{clip1->mClip, asr});
|
|
}
|
|
clip1 = clip1->mParent;
|
|
} else if (clip2 && clip2->mASR == asr) {
|
|
intersectedClips.AppendElement(ClipChainItem{clip2->mClip, asr});
|
|
clip2 = clip2->mParent;
|
|
}
|
|
if (!asr) {
|
|
MOZ_ASSERT(!aAncestor, "We should have exited this loop earlier");
|
|
break;
|
|
}
|
|
asr = asr->mParent;
|
|
}
|
|
|
|
// Convert intersectedClips into a DisplayItemClipChain.
|
|
const DisplayItemClipChain* parentSC = aAncestor;
|
|
for (auto& sc : Reversed(intersectedClips)) {
|
|
parentSC = AllocateDisplayItemClipChain(sc.clip, sc.asr, parentSC);
|
|
}
|
|
return parentSC;
|
|
}
|
|
|
|
const DisplayItemClipChain* nsDisplayListBuilder::CopyWholeChain(
|
|
const DisplayItemClipChain* aClipChain) {
|
|
return CreateClipChainIntersection(nullptr, aClipChain, nullptr);
|
|
}
|
|
|
|
const DisplayItemClipChain* nsDisplayListBuilder::FuseClipChainUpTo(
|
|
const DisplayItemClipChain* aClipChain, const ActiveScrolledRoot* aASR) {
|
|
if (!aClipChain) {
|
|
return nullptr;
|
|
}
|
|
|
|
const DisplayItemClipChain* sc = aClipChain;
|
|
DisplayItemClip mergedClip;
|
|
while (sc && ActiveScrolledRoot::PickDescendant(aASR, sc->mASR) == sc->mASR) {
|
|
mergedClip.IntersectWith(sc->mClip);
|
|
sc = sc->mParent;
|
|
}
|
|
|
|
if (!mergedClip.HasClip()) {
|
|
return nullptr;
|
|
}
|
|
|
|
return AllocateDisplayItemClipChain(mergedClip, aASR, sc);
|
|
}
|
|
|
|
const nsIFrame* nsDisplayListBuilder::FindReferenceFrameFor(
|
|
const nsIFrame* aFrame, nsPoint* aOffset) const {
|
|
if (aFrame == mCurrentFrame) {
|
|
if (aOffset) {
|
|
*aOffset = mCurrentOffsetToReferenceFrame;
|
|
}
|
|
return mCurrentReferenceFrame;
|
|
}
|
|
for (const nsIFrame* f = aFrame; f;
|
|
f = nsLayoutUtils::GetCrossDocParentFrame(f)) {
|
|
if (f == mReferenceFrame || f->IsTransformed()) {
|
|
if (aOffset) {
|
|
*aOffset = aFrame->GetOffsetToCrossDoc(f);
|
|
}
|
|
return f;
|
|
}
|
|
}
|
|
if (aOffset) {
|
|
*aOffset = aFrame->GetOffsetToCrossDoc(mReferenceFrame);
|
|
}
|
|
return mReferenceFrame;
|
|
}
|
|
|
|
// Sticky frames are active if their nearest scrollable frame is also active.
|
|
static bool IsStickyFrameActive(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame, nsIFrame* aParent) {
|
|
MOZ_ASSERT(aFrame->StyleDisplay()->mPosition ==
|
|
StylePositionProperty::Sticky);
|
|
|
|
// Find the nearest scrollframe.
|
|
nsIScrollableFrame* sf = nsLayoutUtils::GetNearestScrollableFrame(
|
|
aFrame->GetParent(), nsLayoutUtils::SCROLLABLE_SAME_DOC |
|
|
nsLayoutUtils::SCROLLABLE_INCLUDE_HIDDEN);
|
|
if (!sf) {
|
|
return false;
|
|
}
|
|
|
|
return sf->IsScrollingActive(aBuilder);
|
|
}
|
|
|
|
nsDisplayListBuilder::AGRState nsDisplayListBuilder::IsAnimatedGeometryRoot(
|
|
nsIFrame* aFrame, bool& aIsAsync, nsIFrame** aParent) {
|
|
// We can return once we know that this frame is an AGR, and we're either
|
|
// async, or sure that none of the later conditions might make us async.
|
|
// The exception to this is when IsPaintingToWindow() == false.
|
|
aIsAsync = false;
|
|
if (aFrame == mReferenceFrame) {
|
|
aIsAsync = true;
|
|
return AGR_YES;
|
|
}
|
|
|
|
if (!IsPaintingToWindow()) {
|
|
if (aParent) {
|
|
*aParent = nsLayoutUtils::GetCrossDocParentFrame(aFrame);
|
|
}
|
|
return AGR_NO;
|
|
}
|
|
|
|
nsIFrame* parent = nsLayoutUtils::GetCrossDocParentFrame(aFrame);
|
|
if (!parent) {
|
|
aIsAsync = true;
|
|
return AGR_YES;
|
|
}
|
|
|
|
if (aFrame->StyleDisplay()->mPosition == StylePositionProperty::Sticky &&
|
|
IsStickyFrameActive(this, aFrame, parent)) {
|
|
aIsAsync = true;
|
|
return AGR_YES;
|
|
}
|
|
|
|
if (aFrame->IsTransformed()) {
|
|
aIsAsync = EffectCompositor::HasAnimationsForCompositor(
|
|
aFrame, DisplayItemType::TYPE_TRANSFORM);
|
|
return AGR_YES;
|
|
}
|
|
|
|
LayoutFrameType parentType = parent->Type();
|
|
if (parentType == LayoutFrameType::Scroll ||
|
|
parentType == LayoutFrameType::ListControl) {
|
|
nsIScrollableFrame* sf = do_QueryFrame(parent);
|
|
if (sf->GetScrolledFrame() == aFrame && sf->IsScrollingActive(this)) {
|
|
MOZ_ASSERT(!aFrame->IsTransformed());
|
|
aIsAsync = sf->IsMaybeAsynchronouslyScrolled();
|
|
return AGR_YES;
|
|
}
|
|
}
|
|
|
|
// Treat the slider thumb as being as an active scrolled root when it wants
|
|
// its own layer so that it can move without repainting.
|
|
if (parentType == LayoutFrameType::Slider) {
|
|
auto* sf = static_cast<nsSliderFrame*>(parent)->GetScrollFrame();
|
|
// The word "Maybe" in IsMaybeScrollingActive might be confusing but we do
|
|
// indeed need to always consider scroll thumbs as AGRs if
|
|
// IsMaybeScrollingActive is true because that is the same condition we use
|
|
// in ScrollFrameHelper::AppendScrollPartsTo to layerize scroll thumbs.
|
|
if (sf && sf->IsMaybeScrollingActive()) {
|
|
return AGR_YES;
|
|
}
|
|
}
|
|
|
|
if (nsLayoutUtils::IsPopup(aFrame)) {
|
|
return AGR_YES;
|
|
}
|
|
|
|
if (ActiveLayerTracker::IsOffsetStyleAnimated(aFrame)) {
|
|
const bool inBudget = AddToAGRBudget(aFrame);
|
|
if (inBudget) {
|
|
return AGR_YES;
|
|
}
|
|
}
|
|
|
|
if (!aFrame->GetParent() &&
|
|
nsLayoutUtils::ViewportHasDisplayPort(aFrame->PresContext())) {
|
|
// Viewport frames in a display port need to be animated geometry roots
|
|
// for background-attachment:fixed elements.
|
|
return AGR_YES;
|
|
}
|
|
|
|
// Fixed-pos frames are parented by the viewport frame, which has no parent.
|
|
if (nsLayoutUtils::IsFixedPosFrameInDisplayPort(aFrame)) {
|
|
return AGR_YES;
|
|
}
|
|
|
|
if (aParent) {
|
|
*aParent = parent;
|
|
}
|
|
|
|
return AGR_NO;
|
|
}
|
|
|
|
nsIFrame* nsDisplayListBuilder::FindAnimatedGeometryRootFrameFor(
|
|
nsIFrame* aFrame, bool& aIsAsync) {
|
|
MOZ_ASSERT(
|
|
nsLayoutUtils::IsAncestorFrameCrossDoc(RootReferenceFrame(), aFrame));
|
|
nsIFrame* cursor = aFrame;
|
|
while (cursor != RootReferenceFrame()) {
|
|
nsIFrame* next;
|
|
if (IsAnimatedGeometryRoot(cursor, aIsAsync, &next) == AGR_YES) {
|
|
return cursor;
|
|
}
|
|
cursor = next;
|
|
}
|
|
// Root frame is always an async agr.
|
|
aIsAsync = true;
|
|
return cursor;
|
|
}
|
|
|
|
void nsDisplayListBuilder::RecomputeCurrentAnimatedGeometryRoot() {
|
|
bool isAsync;
|
|
if (*mCurrentAGR != mCurrentFrame &&
|
|
IsAnimatedGeometryRoot(const_cast<nsIFrame*>(mCurrentFrame), isAsync) ==
|
|
AGR_YES) {
|
|
AnimatedGeometryRoot* oldAGR = mCurrentAGR;
|
|
mCurrentAGR = WrapAGRForFrame(const_cast<nsIFrame*>(mCurrentFrame), isAsync,
|
|
mCurrentAGR);
|
|
|
|
// Iterate the AGR cache and look for any objects that reference the old AGR
|
|
// and check to see if they need to be updated. AGRs can be in the cache
|
|
// multiple times, so we may end up doing the work multiple times for AGRs
|
|
// that don't change.
|
|
for (auto iter = mFrameToAnimatedGeometryRootMap.Iter(); !iter.Done();
|
|
iter.Next()) {
|
|
RefPtr<AnimatedGeometryRoot> cached = iter.UserData();
|
|
if (cached->mParentAGR == oldAGR && cached != mCurrentAGR) {
|
|
// It's possible that this cached AGR struct that has the old AGR as a
|
|
// parent should instead have mCurrentFrame has a parent.
|
|
nsIFrame* parent = FindAnimatedGeometryRootFrameFor(*cached, isAsync);
|
|
MOZ_ASSERT(parent == mCurrentFrame || parent == *oldAGR);
|
|
if (parent == mCurrentFrame) {
|
|
cached->mParentAGR = mCurrentAGR;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
static nsRect ApplyAllClipNonRoundedIntersection(
|
|
const DisplayItemClipChain* aClipChain, const nsRect& aRect) {
|
|
nsRect result = aRect;
|
|
while (aClipChain) {
|
|
result = aClipChain->mClip.ApplyNonRoundedIntersection(result);
|
|
aClipChain = aClipChain->mParent;
|
|
}
|
|
return result;
|
|
}
|
|
|
|
void nsDisplayListBuilder::AdjustWindowDraggingRegion(nsIFrame* aFrame) {
|
|
if (!mWindowDraggingAllowed || !IsForPainting()) {
|
|
return;
|
|
}
|
|
|
|
const nsStyleUIReset* styleUI = aFrame->StyleUIReset();
|
|
if (styleUI->mWindowDragging == StyleWindowDragging::Default) {
|
|
// This frame has the default value and doesn't influence the window
|
|
// dragging region.
|
|
return;
|
|
}
|
|
|
|
LayoutDeviceToLayoutDeviceMatrix4x4 referenceFrameToRootReferenceFrame;
|
|
|
|
// The const_cast is for nsLayoutUtils::GetTransformToAncestor.
|
|
nsIFrame* referenceFrame =
|
|
const_cast<nsIFrame*>(FindReferenceFrameFor(aFrame));
|
|
|
|
if (IsInTransform()) {
|
|
// Only support 2d rectilinear transforms. Transform support is needed for
|
|
// the horizontal flip transform that's applied to the urlbar textbox in
|
|
// RTL mode - it should be able to exclude itself from the draggable region.
|
|
referenceFrameToRootReferenceFrame =
|
|
ViewAs<LayoutDeviceToLayoutDeviceMatrix4x4>(
|
|
nsLayoutUtils::GetTransformToAncestor(referenceFrame,
|
|
mReferenceFrame)
|
|
.GetMatrix());
|
|
Matrix referenceFrameToRootReferenceFrame2d;
|
|
if (!referenceFrameToRootReferenceFrame.Is2D(
|
|
&referenceFrameToRootReferenceFrame2d) ||
|
|
!referenceFrameToRootReferenceFrame2d.IsRectilinear()) {
|
|
return;
|
|
}
|
|
} else {
|
|
MOZ_ASSERT(referenceFrame == mReferenceFrame,
|
|
"referenceFrameToRootReferenceFrame needs to be adjusted");
|
|
}
|
|
|
|
// We do some basic visibility checking on the frame's border box here.
|
|
// We intersect it both with the current dirty rect and with the current
|
|
// clip. Either one is just a conservative approximation on its own, but
|
|
// their intersection luckily works well enough for our purposes, so that
|
|
// we don't have to do full-blown visibility computations.
|
|
// The most important case we need to handle is the scrolled-off tab:
|
|
// If the tab bar overflows, tab parts that are clipped by the scrollbox
|
|
// should not be allowed to interfere with the window dragging region. Using
|
|
// just the current DisplayItemClip is not enough to cover this case
|
|
// completely because clips are reset while building stacking context
|
|
// contents, so for example we'd fail to clip frames that have a clip path
|
|
// applied to them. But the current dirty rect doesn't get reset in that
|
|
// case, so we use it to make this case work.
|
|
nsRect borderBox = aFrame->GetRectRelativeToSelf().Intersect(mVisibleRect);
|
|
borderBox += ToReferenceFrame(aFrame);
|
|
const DisplayItemClipChain* clip =
|
|
ClipState().GetCurrentCombinedClipChain(this);
|
|
borderBox = ApplyAllClipNonRoundedIntersection(clip, borderBox);
|
|
if (borderBox.IsEmpty()) {
|
|
return;
|
|
}
|
|
|
|
LayoutDeviceRect devPixelBorderBox = LayoutDevicePixel::FromAppUnits(
|
|
borderBox, aFrame->PresContext()->AppUnitsPerDevPixel());
|
|
|
|
LayoutDeviceRect transformedDevPixelBorderBox =
|
|
TransformBy(referenceFrameToRootReferenceFrame, devPixelBorderBox);
|
|
transformedDevPixelBorderBox.Round();
|
|
LayoutDeviceIntRect transformedDevPixelBorderBoxInt;
|
|
|
|
if (!transformedDevPixelBorderBox.ToIntRect(
|
|
&transformedDevPixelBorderBoxInt)) {
|
|
return;
|
|
}
|
|
|
|
LayoutDeviceIntRegion& region =
|
|
styleUI->mWindowDragging == StyleWindowDragging::Drag
|
|
? mWindowDraggingRegion
|
|
: mWindowNoDraggingRegion;
|
|
|
|
if (!IsRetainingDisplayList()) {
|
|
region.OrWith(transformedDevPixelBorderBoxInt);
|
|
return;
|
|
}
|
|
|
|
mozilla::gfx::IntRect rect(transformedDevPixelBorderBoxInt.ToUnknownRect());
|
|
if (styleUI->mWindowDragging == StyleWindowDragging::Drag) {
|
|
mRetainedWindowDraggingRegion.Add(aFrame, rect);
|
|
} else {
|
|
mRetainedWindowNoDraggingRegion.Add(aFrame, rect);
|
|
}
|
|
}
|
|
|
|
LayoutDeviceIntRegion nsDisplayListBuilder::GetWindowDraggingRegion() const {
|
|
LayoutDeviceIntRegion result;
|
|
if (!IsRetainingDisplayList()) {
|
|
result.Sub(mWindowDraggingRegion, mWindowNoDraggingRegion);
|
|
return result;
|
|
}
|
|
|
|
LayoutDeviceIntRegion dragRegion =
|
|
mRetainedWindowDraggingRegion.ToLayoutDeviceIntRegion();
|
|
|
|
LayoutDeviceIntRegion noDragRegion =
|
|
mRetainedWindowNoDraggingRegion.ToLayoutDeviceIntRegion();
|
|
|
|
result.Sub(dragRegion, noDragRegion);
|
|
return result;
|
|
}
|
|
|
|
void nsDisplayHitTestInfoBase::AddSizeOfExcludingThis(
|
|
nsWindowSizes& aSizes) const {
|
|
nsPaintedDisplayItem::AddSizeOfExcludingThis(aSizes);
|
|
aSizes.mLayoutRetainedDisplayListSize +=
|
|
aSizes.mState.mMallocSizeOf(mHitTestInfo.get());
|
|
}
|
|
|
|
void nsDisplayTransform::AddSizeOfExcludingThis(nsWindowSizes& aSizes) const {
|
|
nsDisplayHitTestInfoBase::AddSizeOfExcludingThis(aSizes);
|
|
aSizes.mLayoutRetainedDisplayListSize +=
|
|
aSizes.mState.mMallocSizeOf(mTransformPreserves3D.get());
|
|
}
|
|
|
|
void nsDisplayListBuilder::AddSizeOfExcludingThis(nsWindowSizes& aSizes) const {
|
|
mPool.AddSizeOfExcludingThis(aSizes, Arena::ArenaKind::DisplayList);
|
|
|
|
size_t n = 0;
|
|
MallocSizeOf mallocSizeOf = aSizes.mState.mMallocSizeOf;
|
|
n += mDocumentWillChangeBudgets.ShallowSizeOfExcludingThis(mallocSizeOf);
|
|
n += mFrameWillChangeBudgets.ShallowSizeOfExcludingThis(mallocSizeOf);
|
|
n += mAGRBudgetSet.ShallowSizeOfExcludingThis(mallocSizeOf);
|
|
n += mEffectsUpdates.ShallowSizeOfExcludingThis(mallocSizeOf);
|
|
n += mWindowExcludeGlassRegion.SizeOfExcludingThis(mallocSizeOf);
|
|
n += mRetainedWindowDraggingRegion.SizeOfExcludingThis(mallocSizeOf);
|
|
n += mRetainedWindowNoDraggingRegion.SizeOfExcludingThis(mallocSizeOf);
|
|
n += mRetainedWindowOpaqueRegion.SizeOfExcludingThis(mallocSizeOf);
|
|
// XXX can't measure mClipDeduplicator since it uses std::unordered_set.
|
|
|
|
aSizes.mLayoutRetainedDisplayListSize += n;
|
|
}
|
|
|
|
void RetainedDisplayList::AddSizeOfExcludingThis(nsWindowSizes& aSizes) const {
|
|
for (nsDisplayItem* item : *this) {
|
|
item->AddSizeOfExcludingThis(aSizes);
|
|
if (RetainedDisplayList* children = item->GetChildren()) {
|
|
children->AddSizeOfExcludingThis(aSizes);
|
|
}
|
|
}
|
|
|
|
size_t n = 0;
|
|
|
|
n += mDAG.mDirectPredecessorList.ShallowSizeOfExcludingThis(
|
|
aSizes.mState.mMallocSizeOf);
|
|
n += mDAG.mNodesInfo.ShallowSizeOfExcludingThis(aSizes.mState.mMallocSizeOf);
|
|
n += mOldItems.ShallowSizeOfExcludingThis(aSizes.mState.mMallocSizeOf);
|
|
|
|
aSizes.mLayoutRetainedDisplayListSize += n;
|
|
}
|
|
|
|
size_t nsDisplayListBuilder::WeakFrameRegion::SizeOfExcludingThis(
|
|
MallocSizeOf aMallocSizeOf) const {
|
|
size_t n = 0;
|
|
n += mFrames.ShallowSizeOfExcludingThis(aMallocSizeOf);
|
|
for (auto& frame : mFrames) {
|
|
const UniquePtr<WeakFrame>& weakFrame = frame.mWeakFrame;
|
|
n += aMallocSizeOf(weakFrame.get());
|
|
}
|
|
n += mRects.ShallowSizeOfExcludingThis(aMallocSizeOf);
|
|
return n;
|
|
}
|
|
|
|
/**
|
|
* Removes modified frames and rects from this WeakFrameRegion.
|
|
*/
|
|
void nsDisplayListBuilder::WeakFrameRegion::RemoveModifiedFramesAndRects() {
|
|
MOZ_ASSERT(mFrames.Length() == mRects.Length());
|
|
|
|
uint32_t i = 0;
|
|
uint32_t length = mFrames.Length();
|
|
|
|
while (i < length) {
|
|
auto& wrapper = mFrames[i];
|
|
|
|
if (!wrapper.mWeakFrame->IsAlive() ||
|
|
AnyContentAncestorModified(wrapper.mWeakFrame->GetFrame())) {
|
|
// To avoid multiple O(n) shifts in the array, move the last element of
|
|
// the array to the current position and decrease the array length.
|
|
mFrameSet.RemoveEntry(wrapper.mFrame);
|
|
mFrames[i] = std::move(mFrames[length - 1]);
|
|
mRects[i] = std::move(mRects[length - 1]);
|
|
length--;
|
|
} else {
|
|
i++;
|
|
}
|
|
}
|
|
|
|
mFrames.TruncateLength(length);
|
|
mRects.TruncateLength(length);
|
|
}
|
|
|
|
void nsDisplayListBuilder::RemoveModifiedWindowRegions() {
|
|
mRetainedWindowDraggingRegion.RemoveModifiedFramesAndRects();
|
|
mRetainedWindowNoDraggingRegion.RemoveModifiedFramesAndRects();
|
|
mWindowExcludeGlassRegion.RemoveModifiedFramesAndRects();
|
|
mRetainedWindowOpaqueRegion.RemoveModifiedFramesAndRects();
|
|
|
|
mHasGlassItemDuringPartial = false;
|
|
}
|
|
|
|
void nsDisplayListBuilder::ClearRetainedWindowRegions() {
|
|
mRetainedWindowDraggingRegion.Clear();
|
|
mRetainedWindowNoDraggingRegion.Clear();
|
|
mWindowExcludeGlassRegion.Clear();
|
|
mRetainedWindowOpaqueRegion.Clear();
|
|
|
|
mGlassDisplayItem = nullptr;
|
|
}
|
|
|
|
const uint32_t gWillChangeAreaMultiplier = 3;
|
|
static uint32_t GetLayerizationCost(const nsSize& aSize) {
|
|
// There's significant overhead for each layer created from Gecko
|
|
// (IPC+Shared Objects) and from the backend (like an OpenGL texture).
|
|
// Therefore we set a minimum cost threshold of a 64x64 area.
|
|
const int minBudgetCost = 64 * 64;
|
|
|
|
const uint32_t budgetCost = std::max(
|
|
minBudgetCost, nsPresContext::AppUnitsToIntCSSPixels(aSize.width) *
|
|
nsPresContext::AppUnitsToIntCSSPixels(aSize.height));
|
|
|
|
return budgetCost;
|
|
}
|
|
|
|
bool nsDisplayListBuilder::AddToWillChangeBudget(nsIFrame* aFrame,
|
|
const nsSize& aSize) {
|
|
MOZ_ASSERT(IsForPainting());
|
|
|
|
if (aFrame->MayHaveWillChangeBudget()) {
|
|
// The frame is already in the will-change budget.
|
|
return true;
|
|
}
|
|
|
|
const nsPresContext* presContext = aFrame->PresContext();
|
|
const nsRect area = presContext->GetVisibleArea();
|
|
const uint32_t budgetLimit =
|
|
nsPresContext::AppUnitsToIntCSSPixels(area.width) *
|
|
nsPresContext::AppUnitsToIntCSSPixels(area.height);
|
|
const uint32_t cost = GetLayerizationCost(aSize);
|
|
|
|
DocumentWillChangeBudget& documentBudget =
|
|
mDocumentWillChangeBudgets.GetOrInsert(presContext);
|
|
|
|
const bool onBudget =
|
|
(documentBudget + cost) / gWillChangeAreaMultiplier < budgetLimit;
|
|
|
|
if (onBudget) {
|
|
documentBudget += cost;
|
|
mFrameWillChangeBudgets.Put(aFrame,
|
|
FrameWillChangeBudget(presContext, cost));
|
|
aFrame->SetMayHaveWillChangeBudget(true);
|
|
}
|
|
|
|
return onBudget;
|
|
}
|
|
|
|
bool nsDisplayListBuilder::IsInWillChangeBudget(nsIFrame* aFrame,
|
|
const nsSize& aSize) {
|
|
if (!IsForPainting()) {
|
|
// If this nsDisplayListBuilder is not for painting, the layerization should
|
|
// not matter. Do the simple thing and return false.
|
|
return false;
|
|
}
|
|
|
|
const bool onBudget = AddToWillChangeBudget(aFrame, aSize);
|
|
if (onBudget) {
|
|
return true;
|
|
}
|
|
|
|
auto* pc = aFrame->PresContext();
|
|
auto* doc = pc->Document();
|
|
if (!doc->HasWarnedAbout(Document::eIgnoringWillChangeOverBudget)) {
|
|
AutoTArray<nsString, 2> params;
|
|
params.AppendElement()->AppendInt(gWillChangeAreaMultiplier);
|
|
|
|
nsRect area = pc->GetVisibleArea();
|
|
uint32_t budgetLimit = nsPresContext::AppUnitsToIntCSSPixels(area.width) *
|
|
nsPresContext::AppUnitsToIntCSSPixels(area.height);
|
|
params.AppendElement()->AppendInt(budgetLimit);
|
|
|
|
doc->WarnOnceAbout(Document::eIgnoringWillChangeOverBudget, false, params);
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
void nsDisplayListBuilder::ClearWillChangeBudgetStatus(nsIFrame* aFrame) {
|
|
MOZ_ASSERT(IsForPainting());
|
|
|
|
if (!aFrame->MayHaveWillChangeBudget()) {
|
|
return;
|
|
}
|
|
|
|
aFrame->SetMayHaveWillChangeBudget(false);
|
|
RemoveFromWillChangeBudgets(aFrame);
|
|
}
|
|
|
|
void nsDisplayListBuilder::RemoveFromWillChangeBudgets(const nsIFrame* aFrame) {
|
|
if (auto entry = mFrameWillChangeBudgets.Lookup(aFrame)) {
|
|
const FrameWillChangeBudget& frameBudget = entry.Data();
|
|
|
|
DocumentWillChangeBudget* documentBudget =
|
|
mDocumentWillChangeBudgets.GetValue(frameBudget.mPresContext);
|
|
|
|
if (documentBudget) {
|
|
*documentBudget -= frameBudget.mUsage;
|
|
}
|
|
|
|
entry.Remove();
|
|
}
|
|
}
|
|
|
|
void nsDisplayListBuilder::ClearWillChangeBudgets() {
|
|
mFrameWillChangeBudgets.Clear();
|
|
mDocumentWillChangeBudgets.Clear();
|
|
}
|
|
|
|
#ifdef MOZ_GFX_OPTIMIZE_MOBILE
|
|
const float gAGRBudgetAreaMultiplier = 0.3;
|
|
#else
|
|
const float gAGRBudgetAreaMultiplier = 3.0;
|
|
#endif
|
|
|
|
bool nsDisplayListBuilder::AddToAGRBudget(nsIFrame* aFrame) {
|
|
if (mAGRBudgetSet.Contains(aFrame)) {
|
|
return true;
|
|
}
|
|
|
|
const nsPresContext* presContext =
|
|
aFrame->PresContext()->GetRootPresContext();
|
|
if (!presContext) {
|
|
return false;
|
|
}
|
|
|
|
const nsRect area = presContext->GetVisibleArea();
|
|
const uint32_t budgetLimit =
|
|
gAGRBudgetAreaMultiplier *
|
|
nsPresContext::AppUnitsToIntCSSPixels(area.width) *
|
|
nsPresContext::AppUnitsToIntCSSPixels(area.height);
|
|
|
|
const uint32_t cost = GetLayerizationCost(aFrame->GetSize());
|
|
const bool onBudget = mUsedAGRBudget + cost < budgetLimit;
|
|
|
|
if (onBudget) {
|
|
mUsedAGRBudget += cost;
|
|
mAGRBudgetSet.PutEntry(aFrame);
|
|
}
|
|
|
|
return onBudget;
|
|
}
|
|
|
|
void nsDisplayListBuilder::EnterSVGEffectsContents(
|
|
nsIFrame* aEffectsFrame, nsDisplayList* aHoistedItemsStorage) {
|
|
MOZ_ASSERT(aHoistedItemsStorage);
|
|
if (mSVGEffectsFrames.IsEmpty()) {
|
|
MOZ_ASSERT(!mScrollInfoItemsForHoisting);
|
|
mScrollInfoItemsForHoisting = aHoistedItemsStorage;
|
|
}
|
|
mSVGEffectsFrames.AppendElement(aEffectsFrame);
|
|
}
|
|
|
|
void nsDisplayListBuilder::ExitSVGEffectsContents() {
|
|
MOZ_ASSERT(!mSVGEffectsFrames.IsEmpty());
|
|
mSVGEffectsFrames.RemoveLastElement();
|
|
MOZ_ASSERT(mScrollInfoItemsForHoisting);
|
|
if (mSVGEffectsFrames.IsEmpty()) {
|
|
mScrollInfoItemsForHoisting = nullptr;
|
|
}
|
|
}
|
|
|
|
bool nsDisplayListBuilder::ShouldBuildScrollInfoItemsForHoisting() const {
|
|
/*
|
|
* Note: if changing the conditions under which scroll info layers
|
|
* are created, make a corresponding change to
|
|
* ScrollFrameWillBuildScrollInfoLayer() in nsSliderFrame.cpp.
|
|
*/
|
|
for (nsIFrame* frame : mSVGEffectsFrames) {
|
|
if (nsSVGIntegrationUtils::UsesSVGEffectsNotSupportedInCompositor(frame)) {
|
|
return true;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
|
|
void nsDisplayListBuilder::AppendNewScrollInfoItemForHoisting(
|
|
nsDisplayScrollInfoLayer* aScrollInfoItem) {
|
|
MOZ_ASSERT(ShouldBuildScrollInfoItemsForHoisting());
|
|
MOZ_ASSERT(mScrollInfoItemsForHoisting);
|
|
mScrollInfoItemsForHoisting->AppendToTop(aScrollInfoItem);
|
|
}
|
|
|
|
void nsDisplayListBuilder::BuildCompositorHitTestInfoIfNeeded(
|
|
nsIFrame* aFrame, nsDisplayList* aList, const bool aBuildNew) {
|
|
MOZ_ASSERT(aFrame);
|
|
MOZ_ASSERT(aList);
|
|
|
|
if (!BuildCompositorHitTestInfo()) {
|
|
return;
|
|
}
|
|
|
|
const CompositorHitTestInfo info = aFrame->GetCompositorHitTestInfo(this);
|
|
if (info == CompositorHitTestInvisibleToHit) {
|
|
return;
|
|
}
|
|
|
|
const nsRect area = aFrame->GetCompositorHitTestArea(this);
|
|
if (!aBuildNew && GetHitTestInfo() == info &&
|
|
GetHitTestArea().Contains(area)) {
|
|
return;
|
|
}
|
|
|
|
auto* item = MakeDisplayItem<nsDisplayCompositorHitTestInfo>(
|
|
this, aFrame, info, 0, Some(area));
|
|
MOZ_ASSERT(item);
|
|
|
|
SetCompositorHitTestInfo(area, info);
|
|
aList->AppendToTop(item);
|
|
}
|
|
|
|
void nsDisplayListSet::MoveTo(const nsDisplayListSet& aDestination) const {
|
|
aDestination.BorderBackground()->AppendToTop(BorderBackground());
|
|
aDestination.BlockBorderBackgrounds()->AppendToTop(BlockBorderBackgrounds());
|
|
aDestination.Floats()->AppendToTop(Floats());
|
|
aDestination.Content()->AppendToTop(Content());
|
|
aDestination.PositionedDescendants()->AppendToTop(PositionedDescendants());
|
|
aDestination.Outlines()->AppendToTop(Outlines());
|
|
}
|
|
|
|
static void MoveListTo(nsDisplayList* aList,
|
|
nsTArray<nsDisplayItem*>* aElements) {
|
|
nsDisplayItem* item;
|
|
while ((item = aList->RemoveBottom()) != nullptr) {
|
|
aElements->AppendElement(item);
|
|
}
|
|
}
|
|
|
|
nsRect nsDisplayList::GetClippedBounds(nsDisplayListBuilder* aBuilder) const {
|
|
nsRect bounds;
|
|
for (nsDisplayItem* i : *this) {
|
|
bounds.UnionRect(bounds, i->GetClippedBounds(aBuilder));
|
|
}
|
|
return bounds;
|
|
}
|
|
|
|
nsRect nsDisplayList::GetClippedBoundsWithRespectToASR(
|
|
nsDisplayListBuilder* aBuilder, const ActiveScrolledRoot* aASR,
|
|
nsRect* aBuildingRect) const {
|
|
nsRect bounds;
|
|
for (nsDisplayItem* i : *this) {
|
|
nsRect r = i->GetClippedBounds(aBuilder);
|
|
if (aASR != i->GetActiveScrolledRoot() && !r.IsEmpty()) {
|
|
if (Maybe<nsRect> clip = i->GetClipWithRespectToASR(aBuilder, aASR)) {
|
|
r = clip.ref();
|
|
}
|
|
}
|
|
if (aBuildingRect) {
|
|
aBuildingRect->UnionRect(*aBuildingRect, i->GetBuildingRect());
|
|
}
|
|
bounds.UnionRect(bounds, r);
|
|
}
|
|
return bounds;
|
|
}
|
|
|
|
nsRect nsDisplayList::GetBuildingRect() const {
|
|
nsRect result;
|
|
for (nsDisplayItem* i : *this) {
|
|
result.UnionRect(result, i->GetBuildingRect());
|
|
}
|
|
return result;
|
|
}
|
|
|
|
bool nsDisplayList::ComputeVisibilityForRoot(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
AUTO_PROFILER_LABEL("nsDisplayList::ComputeVisibilityForRoot", GRAPHICS);
|
|
|
|
nsRegion r;
|
|
const ActiveScrolledRoot* rootASR = nullptr;
|
|
r.And(*aVisibleRegion, GetClippedBoundsWithRespectToASR(aBuilder, rootASR));
|
|
return ComputeVisibilityForSublist(aBuilder, aVisibleRegion, r.GetBounds());
|
|
}
|
|
|
|
static nsRegion TreatAsOpaque(nsDisplayItem* aItem,
|
|
nsDisplayListBuilder* aBuilder) {
|
|
bool snap;
|
|
nsRegion opaque = aItem->GetOpaqueRegion(aBuilder, &snap);
|
|
MOZ_ASSERT(
|
|
(aBuilder->IsForEventDelivery() && aBuilder->HitTestIsForVisibility()) ||
|
|
!opaque.IsComplex());
|
|
if (aBuilder->IsForPluginGeometry() &&
|
|
aItem->GetType() != DisplayItemType::TYPE_COMPOSITOR_HITTEST_INFO) {
|
|
// Treat all leaf chrome items as opaque, unless their frames are opacity:0.
|
|
// Since opacity:0 frames generate an nsDisplayOpacity, that item will
|
|
// not be treated as opaque here, so opacity:0 chrome content will be
|
|
// effectively ignored, as it should be.
|
|
// We treat leaf chrome items as opaque to ensure that they cover
|
|
// content plugins, for security reasons.
|
|
// Non-leaf chrome items don't render contents of their own so shouldn't
|
|
// be treated as opaque (and their bounds is just the union of their
|
|
// children, which might be a large area their contents don't really cover).
|
|
nsIFrame* f = aItem->Frame();
|
|
if (f->PresContext()->IsChrome() && !aItem->GetChildren() &&
|
|
f->StyleEffects()->mOpacity != 0.0) {
|
|
opaque = aItem->GetBounds(aBuilder, &snap);
|
|
}
|
|
}
|
|
if (opaque.IsEmpty()) {
|
|
return opaque;
|
|
}
|
|
nsRegion opaqueClipped;
|
|
for (auto iter = opaque.RectIter(); !iter.Done(); iter.Next()) {
|
|
opaqueClipped.Or(opaqueClipped,
|
|
aItem->GetClip().ApproximateIntersectInward(iter.Get()));
|
|
}
|
|
return opaqueClipped;
|
|
}
|
|
|
|
bool nsDisplayList::ComputeVisibilityForSublist(
|
|
nsDisplayListBuilder* aBuilder, nsRegion* aVisibleRegion,
|
|
const nsRect& aListVisibleBounds) {
|
|
#ifdef DEBUG
|
|
nsRegion r;
|
|
r.And(*aVisibleRegion, GetClippedBounds(aBuilder));
|
|
// XXX this fails sometimes:
|
|
NS_WARNING_ASSERTION(r.GetBounds().IsEqualInterior(aListVisibleBounds),
|
|
"bad aListVisibleBounds");
|
|
#endif
|
|
|
|
bool anyVisible = false;
|
|
|
|
AutoTArray<nsDisplayItem*, 512> elements;
|
|
MoveListTo(this, &elements);
|
|
|
|
for (int32_t i = elements.Length() - 1; i >= 0; --i) {
|
|
nsDisplayItem* item = elements[i];
|
|
|
|
if (item->ForceNotVisible() && !item->GetSameCoordinateSystemChildren()) {
|
|
NS_ASSERTION(item->GetBuildingRect().IsEmpty(),
|
|
"invisible items should have empty vis rect");
|
|
item->SetPaintRect(nsRect());
|
|
} else {
|
|
nsRect bounds = item->GetClippedBounds(aBuilder);
|
|
|
|
nsRegion itemVisible;
|
|
itemVisible.And(*aVisibleRegion, bounds);
|
|
item->SetPaintRect(itemVisible.GetBounds());
|
|
}
|
|
|
|
if (item->ComputeVisibility(aBuilder, aVisibleRegion)) {
|
|
anyVisible = true;
|
|
|
|
nsRegion opaque = TreatAsOpaque(item, aBuilder);
|
|
// Subtract opaque item from the visible region
|
|
aBuilder->SubtractFromVisibleRegion(aVisibleRegion, opaque);
|
|
}
|
|
AppendToBottom(item);
|
|
}
|
|
|
|
mIsOpaque = !aVisibleRegion->Intersects(aListVisibleBounds);
|
|
return anyVisible;
|
|
}
|
|
|
|
static CallState TriggerPendingAnimationsOnSubDocuments(Document& aDoc,
|
|
void* aReadyTime) {
|
|
if (PendingAnimationTracker* tracker = aDoc.GetPendingAnimationTracker()) {
|
|
PresShell* presShell = aDoc.GetPresShell();
|
|
// If paint-suppression is in effect then we haven't finished painting
|
|
// this document yet so we shouldn't start animations
|
|
if (!presShell || !presShell->IsPaintingSuppressed()) {
|
|
const TimeStamp& readyTime = *static_cast<TimeStamp*>(aReadyTime);
|
|
tracker->TriggerPendingAnimationsOnNextTick(readyTime);
|
|
}
|
|
}
|
|
aDoc.EnumerateSubDocuments(TriggerPendingAnimationsOnSubDocuments,
|
|
aReadyTime);
|
|
return CallState::Continue;
|
|
}
|
|
|
|
static void TriggerPendingAnimations(Document& aDocument,
|
|
const TimeStamp& aReadyTime) {
|
|
MOZ_ASSERT(!aReadyTime.IsNull(),
|
|
"Animation ready time is not set. Perhaps we're using a layer"
|
|
" manager that doesn't update it");
|
|
TriggerPendingAnimationsOnSubDocuments(aDocument,
|
|
const_cast<TimeStamp*>(&aReadyTime));
|
|
}
|
|
|
|
LayerManager* nsDisplayListBuilder::GetWidgetLayerManager(nsView** aView) {
|
|
if (aView) {
|
|
*aView = RootReferenceFrame()->GetView();
|
|
}
|
|
if (RootReferenceFrame() !=
|
|
nsLayoutUtils::GetDisplayRootFrame(RootReferenceFrame())) {
|
|
return nullptr;
|
|
}
|
|
nsIWidget* window = RootReferenceFrame()->GetNearestWidget();
|
|
if (window) {
|
|
return window->GetLayerManager();
|
|
}
|
|
return nullptr;
|
|
}
|
|
|
|
// Find the layer which should house the root scroll metadata for a given
|
|
// layer tree. This is the async zoom container layer if there is one,
|
|
// otherwise it's the root layer.
|
|
Layer* GetLayerForRootMetadata(Layer* aRootLayer, ViewID aRootScrollId) {
|
|
Layer* asyncZoomContainer = DepthFirstSearch<ForwardIterator>(
|
|
aRootLayer, [aRootScrollId](Layer* aLayer) {
|
|
if (auto id = aLayer->IsAsyncZoomContainer()) {
|
|
return *id == aRootScrollId;
|
|
}
|
|
return false;
|
|
});
|
|
return asyncZoomContainer ? asyncZoomContainer : aRootLayer;
|
|
}
|
|
|
|
FrameLayerBuilder* nsDisplayList::BuildLayers(nsDisplayListBuilder* aBuilder,
|
|
LayerManager* aLayerManager,
|
|
uint32_t aFlags,
|
|
bool aIsWidgetTransaction) {
|
|
nsIFrame* frame = aBuilder->RootReferenceFrame();
|
|
nsPresContext* presContext = frame->PresContext();
|
|
PresShell* presShell = presContext->PresShell();
|
|
|
|
FrameLayerBuilder* layerBuilder = new FrameLayerBuilder();
|
|
layerBuilder->Init(aBuilder, aLayerManager);
|
|
|
|
if (aFlags & PAINT_COMPRESSED) {
|
|
layerBuilder->SetLayerTreeCompressionMode();
|
|
}
|
|
|
|
RefPtr<ContainerLayer> root;
|
|
{
|
|
AUTO_PROFILER_LABEL_CATEGORY_PAIR(GRAPHICS_LayerBuilding);
|
|
#ifdef MOZ_GECKO_PROFILER
|
|
nsCOMPtr<nsIDocShell> docShell = presContext->GetDocShell();
|
|
AUTO_PROFILER_TRACING_MARKER_DOCSHELL("Paint", "LayerBuilding", GRAPHICS,
|
|
docShell);
|
|
#endif
|
|
|
|
if (XRE_IsContentProcess() && StaticPrefs::gfx_content_always_paint()) {
|
|
FrameLayerBuilder::InvalidateAllLayers(aLayerManager);
|
|
}
|
|
|
|
if (aIsWidgetTransaction) {
|
|
layerBuilder->DidBeginRetainedLayerTransaction(aLayerManager);
|
|
}
|
|
|
|
// Clear any ScrollMetadata that may have been set on the root layer on a
|
|
// previous paint. This paint will set new metrics if necessary, and if we
|
|
// don't clear the old one here, we may be left with extra metrics.
|
|
if (Layer* rootLayer = aLayerManager->GetRoot()) {
|
|
rootLayer->SetScrollMetadata(nsTArray<ScrollMetadata>());
|
|
}
|
|
|
|
float resolutionUniform = 1.0f;
|
|
float resolutionX = resolutionUniform;
|
|
float resolutionY = resolutionUniform;
|
|
|
|
// If we are in a remote browser, then apply scaling from ancestor browsers
|
|
if (BrowserChild* browserChild = BrowserChild::GetFrom(presShell)) {
|
|
if (!browserChild->IsTopLevel()) {
|
|
resolutionX *= browserChild->GetEffectsInfo().mScaleX;
|
|
resolutionY *= browserChild->GetEffectsInfo().mScaleY;
|
|
}
|
|
}
|
|
|
|
ContainerLayerParameters containerParameters(resolutionX, resolutionY);
|
|
|
|
{
|
|
PaintTelemetry::AutoRecord record(PaintTelemetry::Metric::Layerization);
|
|
|
|
root = layerBuilder->BuildContainerLayerFor(aBuilder, aLayerManager,
|
|
frame, nullptr, this,
|
|
containerParameters, nullptr);
|
|
|
|
if (!record.GetStart().IsNull() &&
|
|
StaticPrefs::layers_acceleration_draw_fps()) {
|
|
if (PaintTiming* pt =
|
|
ClientLayerManager::MaybeGetPaintTiming(aLayerManager)) {
|
|
pt->flbMs() = (TimeStamp::Now() - record.GetStart()).ToMilliseconds();
|
|
}
|
|
}
|
|
}
|
|
|
|
if (!root) {
|
|
return nullptr;
|
|
}
|
|
// Root is being scaled up by the X/Y resolution. Scale it back down.
|
|
root->SetPostScale(1.0f / resolutionX, 1.0f / resolutionY);
|
|
|
|
auto callback = [root](ScrollableLayerGuid::ViewID aScrollId) -> bool {
|
|
return nsLayoutUtils::ContainsMetricsWithId(root, aScrollId);
|
|
};
|
|
if (Maybe<ScrollMetadata> rootMetadata = nsLayoutUtils::GetRootMetadata(
|
|
aBuilder, root->Manager(), containerParameters, callback)) {
|
|
GetLayerForRootMetadata(root, rootMetadata->GetMetrics().GetScrollId())
|
|
->SetScrollMetadata(rootMetadata.value());
|
|
}
|
|
|
|
// NS_WARNING is debug-only, so don't even bother checking the conditions
|
|
// in a release build.
|
|
#ifdef DEBUG
|
|
bool usingDisplayport = false;
|
|
if (nsIFrame* rootScrollFrame = presShell->GetRootScrollFrame()) {
|
|
nsIContent* content = rootScrollFrame->GetContent();
|
|
if (content) {
|
|
usingDisplayport = nsLayoutUtils::HasDisplayPort(content);
|
|
}
|
|
}
|
|
if (usingDisplayport &&
|
|
!(root->GetContentFlags() & Layer::CONTENT_OPAQUE) &&
|
|
SpammyLayoutWarningsEnabled()) {
|
|
// See bug 693938, attachment 567017
|
|
NS_WARNING("Transparent content with displayports can be expensive.");
|
|
}
|
|
#endif
|
|
|
|
aLayerManager->SetRoot(root);
|
|
layerBuilder->WillEndTransaction();
|
|
}
|
|
return layerBuilder;
|
|
}
|
|
|
|
/**
|
|
* We paint by executing a layer manager transaction, constructing a
|
|
* single layer representing the display list, and then making it the
|
|
* root of the layer manager, drawing into the PaintedLayers.
|
|
*/
|
|
already_AddRefed<LayerManager> nsDisplayList::PaintRoot(
|
|
nsDisplayListBuilder* aBuilder, gfxContext* aCtx, uint32_t aFlags) {
|
|
AUTO_PROFILER_LABEL("nsDisplayList::PaintRoot", GRAPHICS);
|
|
|
|
RefPtr<LayerManager> layerManager;
|
|
bool widgetTransaction = false;
|
|
bool doBeginTransaction = true;
|
|
nsView* view = nullptr;
|
|
if (aFlags & PAINT_USE_WIDGET_LAYERS) {
|
|
layerManager = aBuilder->GetWidgetLayerManager(&view);
|
|
if (layerManager) {
|
|
layerManager->SetContainsSVG(false);
|
|
|
|
doBeginTransaction = !(aFlags & PAINT_EXISTING_TRANSACTION);
|
|
widgetTransaction = true;
|
|
}
|
|
}
|
|
if (!layerManager) {
|
|
if (!aCtx) {
|
|
NS_WARNING("Nowhere to paint into");
|
|
return nullptr;
|
|
}
|
|
layerManager = new BasicLayerManager(BasicLayerManager::BLM_OFFSCREEN);
|
|
}
|
|
|
|
nsIFrame* frame = aBuilder->RootReferenceFrame();
|
|
nsPresContext* presContext = frame->PresContext();
|
|
PresShell* presShell = presContext->PresShell();
|
|
Document* document = presShell->GetDocument();
|
|
|
|
if (layerManager->GetBackendType() == layers::LayersBackend::LAYERS_WR) {
|
|
if (doBeginTransaction) {
|
|
if (aCtx) {
|
|
if (!layerManager->BeginTransactionWithTarget(aCtx)) {
|
|
return nullptr;
|
|
}
|
|
} else {
|
|
if (!layerManager->BeginTransaction()) {
|
|
return nullptr;
|
|
}
|
|
}
|
|
}
|
|
|
|
bool prevIsCompositingCheap =
|
|
aBuilder->SetIsCompositingCheap(layerManager->IsCompositingCheap());
|
|
MaybeSetupTransactionIdAllocator(layerManager, presContext);
|
|
|
|
bool sent = false;
|
|
if (aFlags & PAINT_IDENTICAL_DISPLAY_LIST) {
|
|
sent = layerManager->EndEmptyTransaction();
|
|
}
|
|
|
|
if (!sent) {
|
|
// Windowed plugins are not supported with WebRender enabled.
|
|
// But PluginGeometry needs to be updated to show plugin.
|
|
// Windowed plugins are going to be removed by Bug 1296400.
|
|
nsRootPresContext* rootPresContext = presContext->GetRootPresContext();
|
|
if (rootPresContext && XRE_IsContentProcess()) {
|
|
if (aBuilder->WillComputePluginGeometry()) {
|
|
rootPresContext->ComputePluginGeometryUpdates(
|
|
aBuilder->RootReferenceFrame(), aBuilder, this);
|
|
}
|
|
// This must be called even if PluginGeometryUpdates were not computed.
|
|
rootPresContext->CollectPluginGeometryUpdates(layerManager);
|
|
}
|
|
|
|
auto* wrManager = static_cast<WebRenderLayerManager*>(layerManager.get());
|
|
|
|
nsIDocShell* docShell = presContext->GetDocShell();
|
|
WrFiltersHolder wrFilters;
|
|
gfx::Matrix5x4* colorMatrix =
|
|
nsDocShell::Cast(docShell)->GetColorMatrix();
|
|
if (colorMatrix) {
|
|
wrFilters.filters.AppendElement(
|
|
wr::FilterOp::ColorMatrix(colorMatrix->components));
|
|
}
|
|
|
|
wrManager->EndTransactionWithoutLayer(this, aBuilder,
|
|
std::move(wrFilters));
|
|
}
|
|
|
|
// For layers-free mode, we check the invalidation state bits in the
|
|
// EndTransaction. So we clear the invalidation state bits after
|
|
// EndTransaction.
|
|
if (widgetTransaction ||
|
|
// SVG-as-an-image docs don't paint as part of the retained layer tree,
|
|
// but they still need the invalidation state bits cleared in order for
|
|
// invalidation for CSS/SMIL animation to work properly.
|
|
(document && document->IsBeingUsedAsImage())) {
|
|
frame->ClearInvalidationStateBits();
|
|
}
|
|
|
|
aBuilder->SetIsCompositingCheap(prevIsCompositingCheap);
|
|
if (document && widgetTransaction) {
|
|
TriggerPendingAnimations(*document,
|
|
layerManager->GetAnimationReadyTime());
|
|
}
|
|
|
|
if (presContext->RefreshDriver()->HasScheduleFlush()) {
|
|
presContext->NotifyInvalidation(layerManager->GetLastTransactionId(),
|
|
frame->GetRect());
|
|
}
|
|
|
|
return layerManager.forget();
|
|
}
|
|
|
|
NotifySubDocInvalidationFunc computeInvalidFunc =
|
|
presContext->MayHavePaintEventListenerInSubDocument()
|
|
? nsPresContext::NotifySubDocInvalidation
|
|
: nullptr;
|
|
|
|
UniquePtr<LayerProperties> props;
|
|
|
|
bool computeInvalidRect =
|
|
(computeInvalidFunc || (!layerManager->IsCompositingCheap() &&
|
|
layerManager->NeedsWidgetInvalidation())) &&
|
|
widgetTransaction;
|
|
|
|
if (computeInvalidRect) {
|
|
props = LayerProperties::CloneFrom(layerManager->GetRoot());
|
|
}
|
|
|
|
if (doBeginTransaction) {
|
|
if (aCtx) {
|
|
if (!layerManager->BeginTransactionWithTarget(aCtx)) {
|
|
return nullptr;
|
|
}
|
|
} else {
|
|
if (!layerManager->BeginTransaction()) {
|
|
return nullptr;
|
|
}
|
|
}
|
|
}
|
|
|
|
bool temp =
|
|
aBuilder->SetIsCompositingCheap(layerManager->IsCompositingCheap());
|
|
LayerManager::EndTransactionFlags flags = LayerManager::END_DEFAULT;
|
|
if (layerManager->NeedsWidgetInvalidation()) {
|
|
if (aFlags & PAINT_NO_COMPOSITE) {
|
|
flags = LayerManager::END_NO_COMPOSITE;
|
|
}
|
|
} else {
|
|
// Client layer managers never composite directly, so
|
|
// we don't need to worry about END_NO_COMPOSITE.
|
|
if (aBuilder->WillComputePluginGeometry()) {
|
|
flags = LayerManager::END_NO_REMOTE_COMPOSITE;
|
|
}
|
|
}
|
|
|
|
MaybeSetupTransactionIdAllocator(layerManager, presContext);
|
|
|
|
bool sent = false;
|
|
if (aFlags & PAINT_IDENTICAL_DISPLAY_LIST) {
|
|
sent = layerManager->EndEmptyTransaction(flags);
|
|
}
|
|
|
|
if (!sent) {
|
|
FrameLayerBuilder* layerBuilder =
|
|
BuildLayers(aBuilder, layerManager, aFlags, widgetTransaction);
|
|
|
|
if (!layerBuilder) {
|
|
layerManager->SetUserData(&gLayerManagerLayerBuilder, nullptr);
|
|
return nullptr;
|
|
}
|
|
|
|
// If this is the content process, we ship plugin geometry updates over with
|
|
// layer updates, so calculate that now before we call EndTransaction.
|
|
nsRootPresContext* rootPresContext = presContext->GetRootPresContext();
|
|
if (rootPresContext && XRE_IsContentProcess()) {
|
|
if (aBuilder->WillComputePluginGeometry()) {
|
|
rootPresContext->ComputePluginGeometryUpdates(
|
|
aBuilder->RootReferenceFrame(), aBuilder, this);
|
|
}
|
|
// The layer system caches plugin configuration information for forwarding
|
|
// with layer updates which needs to get set during reflow. This must be
|
|
// called even if there are no windowed plugins in the page.
|
|
rootPresContext->CollectPluginGeometryUpdates(layerManager);
|
|
}
|
|
|
|
layerManager->EndTransaction(FrameLayerBuilder::DrawPaintedLayer, aBuilder,
|
|
flags);
|
|
layerBuilder->DidEndTransaction();
|
|
}
|
|
|
|
if (widgetTransaction ||
|
|
// SVG-as-an-image docs don't paint as part of the retained layer tree,
|
|
// but they still need the invalidation state bits cleared in order for
|
|
// invalidation for CSS/SMIL animation to work properly.
|
|
(document && document->IsBeingUsedAsImage())) {
|
|
frame->ClearInvalidationStateBits();
|
|
}
|
|
|
|
aBuilder->SetIsCompositingCheap(temp);
|
|
|
|
if (document && widgetTransaction) {
|
|
TriggerPendingAnimations(*document, layerManager->GetAnimationReadyTime());
|
|
}
|
|
|
|
nsIntRegion invalid;
|
|
if (props) {
|
|
if (!props->ComputeDifferences(layerManager->GetRoot(), invalid,
|
|
computeInvalidFunc)) {
|
|
invalid = nsIntRect::MaxIntRect();
|
|
}
|
|
} else if (widgetTransaction) {
|
|
LayerProperties::ClearInvalidations(layerManager->GetRoot());
|
|
}
|
|
|
|
bool shouldInvalidate = layerManager->NeedsWidgetInvalidation();
|
|
|
|
if (view) {
|
|
if (props) {
|
|
if (!invalid.IsEmpty()) {
|
|
nsIntRect bounds = invalid.GetBounds();
|
|
nsRect rect(presContext->DevPixelsToAppUnits(bounds.x),
|
|
presContext->DevPixelsToAppUnits(bounds.y),
|
|
presContext->DevPixelsToAppUnits(bounds.width),
|
|
presContext->DevPixelsToAppUnits(bounds.height));
|
|
if (shouldInvalidate) {
|
|
view->GetViewManager()->InvalidateViewNoSuppression(view, rect);
|
|
}
|
|
presContext->NotifyInvalidation(layerManager->GetLastTransactionId(),
|
|
bounds);
|
|
}
|
|
} else if (shouldInvalidate) {
|
|
view->GetViewManager()->InvalidateView(view);
|
|
}
|
|
}
|
|
|
|
layerManager->SetUserData(&gLayerManagerLayerBuilder, nullptr);
|
|
return layerManager.forget();
|
|
}
|
|
|
|
nsDisplayItem* nsDisplayList::RemoveBottom() {
|
|
nsDisplayItem* item = mSentinel.mAbove;
|
|
if (!item) {
|
|
return nullptr;
|
|
}
|
|
mSentinel.mAbove = item->mAbove;
|
|
if (item == mTop) {
|
|
// must have been the only item
|
|
mTop = &mSentinel;
|
|
}
|
|
item->mAbove = nullptr;
|
|
mLength--;
|
|
return item;
|
|
}
|
|
|
|
void nsDisplayList::DeleteAll(nsDisplayListBuilder* aBuilder) {
|
|
nsDisplayItem* item;
|
|
while ((item = RemoveBottom()) != nullptr) {
|
|
item->Destroy(aBuilder);
|
|
}
|
|
}
|
|
|
|
static bool IsFrameReceivingPointerEvents(nsIFrame* aFrame) {
|
|
return StylePointerEvents::None !=
|
|
aFrame->StyleUI()->GetEffectivePointerEvents(aFrame);
|
|
}
|
|
|
|
// A list of frames, and their z depth. Used for sorting
|
|
// the results of hit testing.
|
|
struct FramesWithDepth {
|
|
explicit FramesWithDepth(float aDepth) : mDepth(aDepth) {}
|
|
|
|
bool operator<(const FramesWithDepth& aOther) const {
|
|
if (!FuzzyEqual(mDepth, aOther.mDepth, 0.1f)) {
|
|
// We want to sort so that the shallowest item (highest depth value) is
|
|
// first
|
|
return mDepth > aOther.mDepth;
|
|
}
|
|
return this < &aOther;
|
|
}
|
|
bool operator==(const FramesWithDepth& aOther) const {
|
|
return this == &aOther;
|
|
}
|
|
|
|
float mDepth;
|
|
nsTArray<nsIFrame*> mFrames;
|
|
};
|
|
|
|
// Sort the frames by depth and then moves all the contained frames to the
|
|
// destination
|
|
static void FlushFramesArray(nsTArray<FramesWithDepth>& aSource,
|
|
nsTArray<nsIFrame*>* aDest) {
|
|
if (aSource.IsEmpty()) {
|
|
return;
|
|
}
|
|
aSource.Sort();
|
|
uint32_t length = aSource.Length();
|
|
for (uint32_t i = 0; i < length; i++) {
|
|
aDest->AppendElements(std::move(aSource[i].mFrames));
|
|
}
|
|
aSource.Clear();
|
|
}
|
|
|
|
void nsDisplayList::HitTest(nsDisplayListBuilder* aBuilder, const nsRect& aRect,
|
|
nsDisplayItem::HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) const {
|
|
nsDisplayItem* item;
|
|
|
|
if (aState->mInPreserves3D) {
|
|
// Collect leaves of the current 3D rendering context.
|
|
for (nsDisplayItem* item : *this) {
|
|
auto itemType = item->GetType();
|
|
if (itemType != DisplayItemType::TYPE_TRANSFORM ||
|
|
!static_cast<nsDisplayTransform*>(item)->IsLeafOf3DContext()) {
|
|
item->HitTest(aBuilder, aRect, aState, aOutFrames);
|
|
} else {
|
|
// One of leaves in the current 3D rendering context.
|
|
aState->mItemBuffer.AppendElement(item);
|
|
}
|
|
}
|
|
return;
|
|
}
|
|
|
|
int32_t itemBufferStart = aState->mItemBuffer.Length();
|
|
for (nsDisplayItem* item : *this) {
|
|
aState->mItemBuffer.AppendElement(item);
|
|
}
|
|
|
|
AutoTArray<FramesWithDepth, 16> temp;
|
|
for (int32_t i = aState->mItemBuffer.Length() - 1; i >= itemBufferStart;
|
|
--i) {
|
|
// Pop element off the end of the buffer. We want to shorten the buffer
|
|
// so that recursive calls to HitTest have more buffer space.
|
|
item = aState->mItemBuffer[i];
|
|
aState->mItemBuffer.SetLength(i);
|
|
|
|
bool snap;
|
|
nsRect r = item->GetBounds(aBuilder, &snap).Intersect(aRect);
|
|
auto itemType = item->GetType();
|
|
bool same3DContext =
|
|
(itemType == DisplayItemType::TYPE_TRANSFORM &&
|
|
static_cast<nsDisplayTransform*>(item)->IsParticipating3DContext()) ||
|
|
(itemType == DisplayItemType::TYPE_PERSPECTIVE &&
|
|
item->Frame()->Extend3DContext());
|
|
if (same3DContext &&
|
|
(itemType != DisplayItemType::TYPE_TRANSFORM ||
|
|
!static_cast<nsDisplayTransform*>(item)->IsLeafOf3DContext())) {
|
|
if (!item->GetClip().MayIntersect(aRect)) {
|
|
continue;
|
|
}
|
|
AutoTArray<nsIFrame*, 1> neverUsed;
|
|
// Start gethering leaves of the 3D rendering context, and
|
|
// append leaves at the end of mItemBuffer. Leaves are
|
|
// processed at following iterations.
|
|
aState->mInPreserves3D = true;
|
|
item->HitTest(aBuilder, aRect, aState, &neverUsed);
|
|
aState->mInPreserves3D = false;
|
|
i = aState->mItemBuffer.Length();
|
|
continue;
|
|
}
|
|
if (same3DContext || item->GetClip().MayIntersect(r)) {
|
|
AutoTArray<nsIFrame*, 16> outFrames;
|
|
item->HitTest(aBuilder, aRect, aState, &outFrames);
|
|
|
|
// For 3d transforms with preserve-3d we add hit frames into the temp list
|
|
// so we can sort them later, otherwise we add them directly to the output
|
|
// list.
|
|
nsTArray<nsIFrame*>* writeFrames = aOutFrames;
|
|
if (item->GetType() == DisplayItemType::TYPE_TRANSFORM &&
|
|
static_cast<nsDisplayTransform*>(item)->IsLeafOf3DContext()) {
|
|
if (outFrames.Length()) {
|
|
nsDisplayTransform* transform =
|
|
static_cast<nsDisplayTransform*>(item);
|
|
nsPoint point = aRect.TopLeft();
|
|
// A 1x1 rect means a point, otherwise use the center of the rect
|
|
if (aRect.width != 1 || aRect.height != 1) {
|
|
point = aRect.Center();
|
|
}
|
|
temp.AppendElement(
|
|
FramesWithDepth(transform->GetHitDepthAtPoint(aBuilder, point)));
|
|
writeFrames = &temp[temp.Length() - 1].mFrames;
|
|
}
|
|
} else {
|
|
// We may have just finished a run of consecutive preserve-3d
|
|
// transforms, so flush these into the destination array before
|
|
// processing our frame list.
|
|
FlushFramesArray(temp, aOutFrames);
|
|
}
|
|
|
|
for (uint32_t j = 0; j < outFrames.Length(); j++) {
|
|
nsIFrame* f = outFrames.ElementAt(j);
|
|
// Filter out some frames depending on the type of hittest
|
|
// we are doing. For visibility tests, pass through all frames.
|
|
// For pointer tests, only pass through frames that are styled
|
|
// to receive pointer events.
|
|
if (aBuilder->HitTestIsForVisibility() ||
|
|
IsFrameReceivingPointerEvents(f)) {
|
|
writeFrames->AppendElement(f);
|
|
}
|
|
}
|
|
|
|
if (aBuilder->HitTestIsForVisibility() &&
|
|
item->GetOpaqueRegion(aBuilder, &snap).Contains(aRect)) {
|
|
// We're exiting early, so pop the remaining items off the buffer.
|
|
aState->mItemBuffer.SetLength(itemBufferStart);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
// Clear any remaining preserve-3d transforms.
|
|
FlushFramesArray(temp, aOutFrames);
|
|
NS_ASSERTION(aState->mItemBuffer.Length() == uint32_t(itemBufferStart),
|
|
"How did we forget to pop some elements?");
|
|
}
|
|
|
|
static nsIContent* FindContentInDocument(nsDisplayItem* aItem, Document* aDoc) {
|
|
nsIFrame* f = aItem->Frame();
|
|
while (f) {
|
|
nsPresContext* pc = f->PresContext();
|
|
if (pc->Document() == aDoc) {
|
|
return f->GetContent();
|
|
}
|
|
f = nsLayoutUtils::GetCrossDocParentFrame(pc->PresShell()->GetRootFrame());
|
|
}
|
|
return nullptr;
|
|
}
|
|
|
|
struct ZSortItem {
|
|
nsDisplayItem* item;
|
|
int32_t zIndex;
|
|
|
|
explicit ZSortItem(nsDisplayItem* aItem)
|
|
: item(aItem), zIndex(aItem->ZIndex()) {}
|
|
|
|
operator nsDisplayItem*() { return item; }
|
|
};
|
|
|
|
struct ZOrderComparator {
|
|
bool operator()(const ZSortItem& aLeft, const ZSortItem& aRight) const {
|
|
// Note that we can't just take the difference of the two
|
|
// z-indices here, because that might overflow a 32-bit int.
|
|
return aLeft.zIndex < aRight.zIndex;
|
|
}
|
|
};
|
|
|
|
void nsDisplayList::SortByZOrder() { Sort<ZSortItem>(ZOrderComparator()); }
|
|
|
|
struct ContentComparator {
|
|
nsIContent* mCommonAncestor;
|
|
|
|
explicit ContentComparator(nsIContent* aCommonAncestor)
|
|
: mCommonAncestor(aCommonAncestor) {}
|
|
|
|
bool operator()(nsDisplayItem* aLeft, nsDisplayItem* aRight) const {
|
|
// It's possible that the nsIContent for aItem1 or aItem2 is in a
|
|
// subdocument of commonAncestor, because display items for subdocuments
|
|
// have been mixed into the same list. Ensure that we're looking at content
|
|
// in commonAncestor's document.
|
|
Document* commonAncestorDoc = mCommonAncestor->OwnerDoc();
|
|
nsIContent* content1 = FindContentInDocument(aLeft, commonAncestorDoc);
|
|
nsIContent* content2 = FindContentInDocument(aRight, commonAncestorDoc);
|
|
if (!content1 || !content2) {
|
|
NS_ERROR("Document trees are mixed up!");
|
|
// Something weird going on
|
|
return true;
|
|
}
|
|
return nsLayoutUtils::CompareTreePosition(content1, content2,
|
|
mCommonAncestor) < 0;
|
|
}
|
|
};
|
|
|
|
void nsDisplayList::SortByContentOrder(nsIContent* aCommonAncestor) {
|
|
Sort<nsDisplayItem*>(ContentComparator(aCommonAncestor));
|
|
}
|
|
|
|
bool nsDisplayItemBase::HasModifiedFrame() const {
|
|
return mItemFlags.contains(ItemBaseFlag::ModifiedFrame);
|
|
}
|
|
|
|
void nsDisplayItemBase::SetModifiedFrame(bool aModified) {
|
|
if (aModified) {
|
|
mItemFlags += ItemBaseFlag::ModifiedFrame;
|
|
} else {
|
|
mItemFlags -= ItemBaseFlag::ModifiedFrame;
|
|
}
|
|
}
|
|
|
|
void nsDisplayItemBase::SetDeletedFrame() {
|
|
mItemFlags += ItemBaseFlag::DeletedFrame;
|
|
}
|
|
|
|
bool nsDisplayItemBase::HasDeletedFrame() const {
|
|
bool retval = mItemFlags.contains(ItemBaseFlag::DeletedFrame) ||
|
|
(GetType() == DisplayItemType::TYPE_REMOTE &&
|
|
!static_cast<const nsDisplayRemote*>(this)->GetFrameLoader());
|
|
MOZ_ASSERT(retval || mFrame);
|
|
return retval;
|
|
}
|
|
|
|
#if !defined(DEBUG) && !defined(MOZ_DIAGNOSTIC_ASSERT_ENABLED)
|
|
static_assert(sizeof(nsDisplayItem) <= 176, "nsDisplayItem has grown");
|
|
#endif
|
|
|
|
nsDisplayItem::nsDisplayItem(nsDisplayListBuilder* aBuilder, nsIFrame* aFrame)
|
|
: nsDisplayItem(aBuilder, aFrame, aBuilder->CurrentActiveScrolledRoot()) {}
|
|
|
|
nsDisplayItem::nsDisplayItem(nsDisplayListBuilder* aBuilder, nsIFrame* aFrame,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot)
|
|
: nsDisplayItemBase(aBuilder, aFrame),
|
|
mActiveScrolledRoot(aActiveScrolledRoot),
|
|
mAnimatedGeometryRoot(nullptr) {
|
|
MOZ_COUNT_CTOR(nsDisplayItem);
|
|
|
|
mReferenceFrame = aBuilder->FindReferenceFrameFor(aFrame, &mToReferenceFrame);
|
|
// This can return the wrong result if the item override
|
|
// ShouldFixToViewport(), the item needs to set it again in its constructor.
|
|
mAnimatedGeometryRoot = aBuilder->FindAnimatedGeometryRootFor(aFrame);
|
|
MOZ_ASSERT(nsLayoutUtils::IsAncestorFrameCrossDoc(
|
|
aBuilder->RootReferenceFrame(), *mAnimatedGeometryRoot),
|
|
"Bad");
|
|
NS_ASSERTION(
|
|
aBuilder->GetVisibleRect().width >= 0 || !aBuilder->IsForPainting(),
|
|
"visible rect not set");
|
|
|
|
nsDisplayItem::SetClipChain(
|
|
aBuilder->ClipState().GetCurrentCombinedClipChain(aBuilder), true);
|
|
|
|
// The visible rect is for mCurrentFrame, so we have to use
|
|
// mCurrentOffsetToReferenceFrame
|
|
nsRect visible = aBuilder->GetVisibleRect() +
|
|
aBuilder->GetCurrentFrameOffsetToReferenceFrame();
|
|
SetBuildingRect(visible);
|
|
|
|
const nsStyleDisplay* disp = mFrame->StyleDisplay();
|
|
if (mFrame->BackfaceIsHidden(disp)) {
|
|
mItemFlags += ItemFlag::BackfaceHidden;
|
|
}
|
|
if (mFrame->Combines3DTransformWithAncestors(disp)) {
|
|
mItemFlags += ItemFlag::Combines3DTransformWithAncestors;
|
|
}
|
|
}
|
|
|
|
/* static */
|
|
bool nsDisplayItem::ForceActiveLayers() {
|
|
return StaticPrefs::layers_force_active();
|
|
}
|
|
|
|
int32_t nsDisplayItem::ZIndex() const { return mFrame->ZIndex(); }
|
|
|
|
bool nsDisplayItem::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
return !GetPaintRect().IsEmpty() &&
|
|
!IsInvisibleInRect(aVisibleRegion->GetBounds());
|
|
}
|
|
|
|
bool nsDisplayItem::RecomputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
if (ForceNotVisible() && !GetSameCoordinateSystemChildren()) {
|
|
// mForceNotVisible wants to ensure that this display item doesn't render
|
|
// anything itself. If this item has contents, then we obviously want to
|
|
// render those, so we don't need this check in that case.
|
|
NS_ASSERTION(GetBuildingRect().IsEmpty(),
|
|
"invisible items without children should have empty vis rect");
|
|
SetPaintRect(nsRect());
|
|
} else {
|
|
bool snap;
|
|
nsRect bounds = GetBounds(aBuilder, &snap);
|
|
|
|
nsRegion itemVisible;
|
|
itemVisible.And(*aVisibleRegion, bounds);
|
|
SetPaintRect(itemVisible.GetBounds());
|
|
}
|
|
|
|
// When we recompute visibility within layers we don't need to
|
|
// expand the visible region for content behind plugins (the plugin
|
|
// is not in the layer).
|
|
if (!ComputeVisibility(aBuilder, aVisibleRegion)) {
|
|
SetPaintRect(nsRect());
|
|
return false;
|
|
}
|
|
|
|
nsRegion opaque = TreatAsOpaque(this, aBuilder);
|
|
aBuilder->SubtractFromVisibleRegion(aVisibleRegion, opaque);
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayItem::SetClipChain(const DisplayItemClipChain* aClipChain,
|
|
bool aStore) {
|
|
mClipChain = aClipChain;
|
|
mClip = DisplayItemClipChain::ClipForASR(aClipChain, mActiveScrolledRoot);
|
|
|
|
if (aStore) {
|
|
mState.mClipChain = mClipChain;
|
|
mState.mClip = mClip;
|
|
}
|
|
}
|
|
|
|
Maybe<nsRect> nsDisplayItem::GetClipWithRespectToASR(
|
|
nsDisplayListBuilder* aBuilder, const ActiveScrolledRoot* aASR) const {
|
|
if (const DisplayItemClip* clip =
|
|
DisplayItemClipChain::ClipForASR(GetClipChain(), aASR)) {
|
|
return Some(clip->GetClipRect());
|
|
}
|
|
#ifdef DEBUG
|
|
MOZ_ASSERT(false, "item should have finite clip with respect to aASR");
|
|
#endif
|
|
return Nothing();
|
|
}
|
|
|
|
void nsDisplayItem::FuseClipChainUpTo(nsDisplayListBuilder* aBuilder,
|
|
const ActiveScrolledRoot* aASR) {
|
|
mClipChain = aBuilder->FuseClipChainUpTo(mClipChain, aASR);
|
|
|
|
if (mClipChain) {
|
|
mClip = &mClipChain->mClip;
|
|
} else {
|
|
mClip = nullptr;
|
|
}
|
|
}
|
|
|
|
bool nsDisplayItem::ShouldUseAdvancedLayer(LayerManager* aManager,
|
|
PrefFunc aFunc) const {
|
|
return CanUseAdvancedLayer(aManager) ? aFunc() : false;
|
|
}
|
|
|
|
bool nsDisplayItem::CanUseAdvancedLayer(LayerManager* aManager) const {
|
|
return StaticPrefs::layers_advanced_basic_layer_enabled() || !aManager ||
|
|
aManager->GetBackendType() == layers::LayersBackend::LAYERS_WR;
|
|
}
|
|
|
|
static const DisplayItemClipChain* FindCommonAncestorClipForIntersection(
|
|
const DisplayItemClipChain* aOne, const DisplayItemClipChain* aTwo) {
|
|
for (const ActiveScrolledRoot* asr =
|
|
ActiveScrolledRoot::PickDescendant(aOne->mASR, aTwo->mASR);
|
|
asr; asr = asr->mParent) {
|
|
if (aOne == aTwo) {
|
|
return aOne;
|
|
}
|
|
if (aOne->mASR == asr) {
|
|
aOne = aOne->mParent;
|
|
}
|
|
if (aTwo->mASR == asr) {
|
|
aTwo = aTwo->mParent;
|
|
}
|
|
if (!aOne) {
|
|
return aTwo;
|
|
}
|
|
if (!aTwo) {
|
|
return aOne;
|
|
}
|
|
}
|
|
return nullptr;
|
|
}
|
|
|
|
void nsDisplayItem::IntersectClip(nsDisplayListBuilder* aBuilder,
|
|
const DisplayItemClipChain* aOther,
|
|
bool aStore) {
|
|
if (!aOther || mClipChain == aOther) {
|
|
return;
|
|
}
|
|
|
|
// aOther might be a reference to a clip on the stack. We need to make sure
|
|
// that CreateClipChainIntersection will allocate the actual intersected
|
|
// clip in the builder's arena, so for the mClipChain == nullptr case,
|
|
// we supply nullptr as the common ancestor so that
|
|
// CreateClipChainIntersection clones the whole chain.
|
|
const DisplayItemClipChain* ancestorClip =
|
|
mClipChain ? FindCommonAncestorClipForIntersection(mClipChain, aOther)
|
|
: nullptr;
|
|
|
|
SetClipChain(
|
|
aBuilder->CreateClipChainIntersection(ancestorClip, mClipChain, aOther),
|
|
aStore);
|
|
}
|
|
|
|
nsRect nsDisplayItem::GetClippedBounds(nsDisplayListBuilder* aBuilder) const {
|
|
bool snap;
|
|
nsRect r = GetBounds(aBuilder, &snap);
|
|
return GetClip().ApplyNonRoundedIntersection(r);
|
|
}
|
|
|
|
nsDisplayContainer::nsDisplayContainer(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot, nsDisplayList* aList)
|
|
: nsDisplayItem(aBuilder, aFrame, aActiveScrolledRoot) {
|
|
MOZ_COUNT_CTOR(nsDisplayContainer);
|
|
mChildren.AppendToTop(aList);
|
|
UpdateBounds(aBuilder);
|
|
|
|
// Clear and store the clip chain set by nsDisplayItem constructor.
|
|
nsDisplayItem::SetClipChain(nullptr, true);
|
|
}
|
|
|
|
bool nsDisplayContainer::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
aManager->CommandBuilder().CreateWebRenderCommandsFromDisplayList(
|
|
GetChildren(), this, aDisplayListBuilder, aSc, aBuilder, aResources);
|
|
return true;
|
|
}
|
|
|
|
/**
|
|
* Like |nsDisplayList::ComputeVisibilityForSublist()|, but restricts
|
|
* |aVisibleRegion| to given |aBounds| of the list.
|
|
*/
|
|
static bool ComputeClippedVisibilityForSubList(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion,
|
|
nsDisplayList* aList,
|
|
const nsRect& aBounds) {
|
|
nsRegion visibleRegion;
|
|
visibleRegion.And(*aVisibleRegion, aBounds);
|
|
nsRegion originalVisibleRegion = visibleRegion;
|
|
|
|
const bool anyItemVisible =
|
|
aList->ComputeVisibilityForSublist(aBuilder, &visibleRegion, aBounds);
|
|
nsRegion removed;
|
|
// removed = originalVisibleRegion - visibleRegion
|
|
removed.Sub(originalVisibleRegion, visibleRegion);
|
|
// aVisibleRegion = aVisibleRegion - removed (modulo any simplifications
|
|
// SubtractFromVisibleRegion does)
|
|
aBuilder->SubtractFromVisibleRegion(aVisibleRegion, removed);
|
|
|
|
return anyItemVisible;
|
|
}
|
|
|
|
bool nsDisplayContainer::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
return ::ComputeClippedVisibilityForSubList(aBuilder, aVisibleRegion,
|
|
GetChildren(), GetPaintRect());
|
|
}
|
|
|
|
nsRect nsDisplayContainer::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = false;
|
|
return mBounds;
|
|
}
|
|
|
|
nsRect nsDisplayContainer::GetComponentAlphaBounds(
|
|
nsDisplayListBuilder* aBuilder) const {
|
|
return mChildren.GetComponentAlphaBounds(aBuilder);
|
|
}
|
|
|
|
static nsRegion GetOpaqueRegion(nsDisplayListBuilder* aBuilder,
|
|
nsDisplayList* aList,
|
|
const nsRect& aListBounds) {
|
|
if (aList->IsOpaque()) {
|
|
// Everything within list bounds that's visible is opaque. This is an
|
|
// optimization to avoid calculating the opaque region.
|
|
return aListBounds;
|
|
}
|
|
|
|
if (aBuilder->HitTestIsForVisibility()) {
|
|
// If we care about an accurate opaque region, iterate the display list
|
|
// and build up a region of opaque bounds.
|
|
return aList->GetOpaqueRegion(aBuilder);
|
|
}
|
|
|
|
return nsRegion();
|
|
}
|
|
|
|
nsRegion nsDisplayContainer::GetOpaqueRegion(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
return ::GetOpaqueRegion(aBuilder, GetChildren(), GetBounds(aBuilder, aSnap));
|
|
}
|
|
|
|
Maybe<nsRect> nsDisplayContainer::GetClipWithRespectToASR(
|
|
nsDisplayListBuilder* aBuilder, const ActiveScrolledRoot* aASR) const {
|
|
// Our children should have finite bounds with respect to |aASR|.
|
|
if (aASR == mActiveScrolledRoot) {
|
|
return Some(mBounds);
|
|
}
|
|
|
|
return Some(mChildren.GetClippedBoundsWithRespectToASR(aBuilder, aASR));
|
|
}
|
|
|
|
void nsDisplayContainer::HitTest(nsDisplayListBuilder* aBuilder,
|
|
const nsRect& aRect, HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) {
|
|
mChildren.HitTest(aBuilder, aRect, aState, aOutFrames);
|
|
}
|
|
|
|
void nsDisplayContainer::UpdateBounds(nsDisplayListBuilder* aBuilder) {
|
|
// Container item bounds are expected to be clipped.
|
|
mBounds =
|
|
mChildren.GetClippedBoundsWithRespectToASR(aBuilder, mActiveScrolledRoot);
|
|
}
|
|
|
|
nsRect nsDisplaySolidColor::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = true;
|
|
return mBounds;
|
|
}
|
|
|
|
LayerState nsDisplaySolidColor::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
if (ForceActiveLayers()) {
|
|
return LayerState::LAYER_ACTIVE;
|
|
}
|
|
|
|
return LayerState::LAYER_NONE;
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplaySolidColor::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
RefPtr<ColorLayer> layer = static_cast<ColorLayer*>(
|
|
aManager->GetLayerBuilder()->GetLeafLayerFor(aBuilder, this));
|
|
if (!layer) {
|
|
layer = aManager->CreateColorLayer();
|
|
if (!layer) {
|
|
return nullptr;
|
|
}
|
|
}
|
|
layer->SetColor(gfx::Color::FromABGR(mColor));
|
|
|
|
const int32_t appUnitsPerDevPixel =
|
|
mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
layer->SetBounds(mBounds.ToNearestPixels(appUnitsPerDevPixel));
|
|
layer->SetBaseTransform(gfx::Matrix4x4::Translation(
|
|
aContainerParameters.mOffset.x, aContainerParameters.mOffset.y, 0));
|
|
|
|
return layer.forget();
|
|
}
|
|
|
|
void nsDisplaySolidColor::Paint(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx) {
|
|
int32_t appUnitsPerDevPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
DrawTarget* drawTarget = aCtx->GetDrawTarget();
|
|
Rect rect =
|
|
NSRectToSnappedRect(GetPaintRect(), appUnitsPerDevPixel, *drawTarget);
|
|
drawTarget->FillRect(rect, ColorPattern(ToDeviceColor(mColor)));
|
|
}
|
|
|
|
void nsDisplaySolidColor::WriteDebugInfo(std::stringstream& aStream) {
|
|
aStream << " (rgba " << (int)NS_GET_R(mColor) << "," << (int)NS_GET_G(mColor)
|
|
<< "," << (int)NS_GET_B(mColor) << "," << (int)NS_GET_A(mColor)
|
|
<< ")";
|
|
}
|
|
|
|
bool nsDisplaySolidColor::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
LayoutDeviceRect bounds = LayoutDeviceRect::FromAppUnits(
|
|
mBounds, mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
|
|
// There is one big solid color behind everything - just split it out if it
|
|
// intersects multiple render roots
|
|
if (aBuilder.GetRenderRoot() == wr::RenderRoot::Default) {
|
|
for (auto renderRoot : wr::kRenderRoots) {
|
|
// Skip the popover render root, as it's intended to overlay the others
|
|
// and be at least partially transparent.
|
|
if (renderRoot == wr::RenderRoot::Popover) {
|
|
continue;
|
|
}
|
|
if (aBuilder.HasSubBuilder(renderRoot)) {
|
|
LayoutDeviceRect renderRootRect =
|
|
aDisplayListBuilder->GetRenderRootRect(renderRoot);
|
|
wr::LayoutRect intersection =
|
|
wr::ToLayoutRect(bounds.Intersect(renderRootRect));
|
|
aBuilder.SubBuilder(renderRoot)
|
|
.PushRect(intersection, intersection, !BackfaceIsHidden(),
|
|
wr::ToColorF(ToDeviceColor(mColor)));
|
|
}
|
|
}
|
|
} else {
|
|
wr::LayoutRect r = wr::ToLayoutRect(bounds);
|
|
|
|
aBuilder.PushRect(r, r, !BackfaceIsHidden(),
|
|
wr::ToColorF(ToDeviceColor(mColor)));
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
nsRect nsDisplaySolidColorRegion::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = true;
|
|
return mRegion.GetBounds();
|
|
}
|
|
|
|
void nsDisplaySolidColorRegion::Paint(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx) {
|
|
int32_t appUnitsPerDevPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
DrawTarget* drawTarget = aCtx->GetDrawTarget();
|
|
ColorPattern color(mColor);
|
|
for (auto iter = mRegion.RectIter(); !iter.Done(); iter.Next()) {
|
|
Rect rect =
|
|
NSRectToSnappedRect(iter.Get(), appUnitsPerDevPixel, *drawTarget);
|
|
drawTarget->FillRect(rect, color);
|
|
}
|
|
}
|
|
|
|
void nsDisplaySolidColorRegion::WriteDebugInfo(std::stringstream& aStream) {
|
|
aStream << " (rgba " << int(mColor.r * 255) << "," << int(mColor.g * 255)
|
|
<< "," << int(mColor.b * 255) << "," << mColor.a << ")";
|
|
}
|
|
|
|
bool nsDisplaySolidColorRegion::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
for (auto iter = mRegion.RectIter(); !iter.Done(); iter.Next()) {
|
|
nsRect rect = iter.Get();
|
|
LayoutDeviceRect layerRects = LayoutDeviceRect::FromAppUnits(
|
|
rect, mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
wr::LayoutRect r = wr::ToLayoutRect(layerRects);
|
|
aBuilder.PushRect(r, r, !BackfaceIsHidden(),
|
|
wr::ToColorF(ToDeviceColor(mColor)));
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
static void RegisterThemeGeometry(nsDisplayListBuilder* aBuilder,
|
|
nsDisplayItem* aItem, nsIFrame* aFrame,
|
|
nsITheme::ThemeGeometryType aType) {
|
|
if (aBuilder->IsInChromeDocumentOrPopup() && !aBuilder->IsInTransform()) {
|
|
nsIFrame* displayRoot = nsLayoutUtils::GetDisplayRootFrame(aFrame);
|
|
nsPoint offset = aBuilder->IsInSubdocument()
|
|
? aBuilder->ToReferenceFrame(aFrame)
|
|
: aFrame->GetOffsetTo(displayRoot);
|
|
nsRect borderBox = nsRect(offset, aFrame->GetSize());
|
|
aBuilder->RegisterThemeGeometry(
|
|
aType, aItem,
|
|
LayoutDeviceIntRect::FromUnknownRect(borderBox.ToNearestPixels(
|
|
aFrame->PresContext()->AppUnitsPerDevPixel())));
|
|
}
|
|
}
|
|
|
|
// Return the bounds of the viewport relative to |aFrame|'s reference frame.
|
|
// Returns Nothing() if transforming into |aFrame|'s coordinate space fails.
|
|
static Maybe<nsRect> GetViewportRectRelativeToReferenceFrame(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame) {
|
|
nsIFrame* rootFrame = aFrame->PresShell()->GetRootFrame();
|
|
nsRect rootRect = rootFrame->GetRectRelativeToSelf();
|
|
if (nsLayoutUtils::TransformRect(rootFrame, aFrame, rootRect) ==
|
|
nsLayoutUtils::TRANSFORM_SUCCEEDED) {
|
|
return Some(rootRect + aBuilder->ToReferenceFrame(aFrame));
|
|
}
|
|
return Nothing();
|
|
}
|
|
|
|
/* static */ nsDisplayBackgroundImage::InitData
|
|
nsDisplayBackgroundImage::GetInitData(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame, uint16_t aLayer,
|
|
const nsRect& aBackgroundRect,
|
|
ComputedStyle* aBackgroundStyle) {
|
|
nsPresContext* presContext = aFrame->PresContext();
|
|
uint32_t flags = aBuilder->GetBackgroundPaintFlags();
|
|
const nsStyleImageLayers::Layer& layer =
|
|
aBackgroundStyle->StyleBackground()->mImage.mLayers[aLayer];
|
|
|
|
bool isTransformedFixed;
|
|
nsBackgroundLayerState state = nsCSSRendering::PrepareImageLayer(
|
|
presContext, aFrame, flags, aBackgroundRect, aBackgroundRect, layer,
|
|
&isTransformedFixed);
|
|
|
|
// background-attachment:fixed is treated as background-attachment:scroll
|
|
// if it's affected by a transform.
|
|
// See https://www.w3.org/Bugs/Public/show_bug.cgi?id=17521.
|
|
bool shouldTreatAsFixed =
|
|
layer.mAttachment == StyleImageLayerAttachment::Fixed &&
|
|
!isTransformedFixed;
|
|
|
|
bool shouldFixToViewport = shouldTreatAsFixed && !layer.mImage.IsNone();
|
|
bool isRasterImage = state.mImageRenderer.IsRasterImage();
|
|
nsCOMPtr<imgIContainer> image;
|
|
if (isRasterImage) {
|
|
image = state.mImageRenderer.GetImage();
|
|
}
|
|
return InitData{aBuilder, aBackgroundStyle, image,
|
|
aBackgroundRect, state.mFillArea, state.mDestArea,
|
|
aLayer, isRasterImage, shouldFixToViewport};
|
|
}
|
|
|
|
nsDisplayBackgroundImage::nsDisplayBackgroundImage(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, const InitData& aInitData,
|
|
nsIFrame* aFrameForBounds)
|
|
: nsDisplayImageContainer(aBuilder, aFrame),
|
|
mBackgroundStyle(aInitData.backgroundStyle),
|
|
mImage(aInitData.image),
|
|
mDependentFrame(nullptr),
|
|
mBackgroundRect(aInitData.backgroundRect),
|
|
mFillRect(aInitData.fillArea),
|
|
mDestRect(aInitData.destArea),
|
|
mLayer(aInitData.layer),
|
|
mIsRasterImage(aInitData.isRasterImage),
|
|
mShouldFixToViewport(aInitData.shouldFixToViewport),
|
|
mImageFlags(0),
|
|
mAssociatedImage(false) {
|
|
MOZ_COUNT_CTOR(nsDisplayBackgroundImage);
|
|
|
|
if (mBackgroundStyle && mBackgroundStyle != aFrame->Style()) {
|
|
MOZ_ASSERT(aFrame->IsCanvasFrame() || aFrame->IsTableCellFrame());
|
|
auto& layer = mBackgroundStyle->StyleBackground()->mImage.mLayers[mLayer];
|
|
if (aFrame->AssociateImage(layer.mImage)) {
|
|
mAssociatedImage = true;
|
|
}
|
|
}
|
|
|
|
mBounds = GetBoundsInternal(aInitData.builder, aFrameForBounds);
|
|
if (mShouldFixToViewport) {
|
|
mAnimatedGeometryRoot =
|
|
aInitData.builder->FindAnimatedGeometryRootFor(this);
|
|
|
|
// Expand the item's visible rect to cover the entire bounds, limited to the
|
|
// viewport rect. This is necessary because the background's clip can move
|
|
// asynchronously.
|
|
if (Maybe<nsRect> viewportRect = GetViewportRectRelativeToReferenceFrame(
|
|
aInitData.builder, mFrame)) {
|
|
SetBuildingRect(mBounds.Intersect(*viewportRect));
|
|
}
|
|
}
|
|
}
|
|
|
|
void nsDisplayBackgroundImage::DisassociateImage() {
|
|
MOZ_ASSERT(mAssociatedImage);
|
|
MOZ_ASSERT(mFrame);
|
|
|
|
if (mFrame->HasImageRequest()) {
|
|
// We need to check HasImageRequest because the frame may already have
|
|
// cleared all its requests in some other way before us.
|
|
auto& layer = mBackgroundStyle->StyleBackground()->mImage.mLayers[mLayer];
|
|
mFrame->DisassociateImage(layer.mImage);
|
|
}
|
|
|
|
mAssociatedImage = false;
|
|
}
|
|
|
|
nsDisplayBackgroundImage::~nsDisplayBackgroundImage() {
|
|
MOZ_COUNT_DTOR(nsDisplayBackgroundImage);
|
|
if (mAssociatedImage) {
|
|
DisassociateImage();
|
|
}
|
|
if (mDependentFrame) {
|
|
mDependentFrame->RemoveDisplayItem(this);
|
|
}
|
|
}
|
|
|
|
static nsIFrame* GetBackgroundComputedStyleFrame(nsIFrame* aFrame) {
|
|
nsIFrame* f;
|
|
if (!nsCSSRendering::FindBackgroundFrame(aFrame, &f)) {
|
|
// We don't want to bail out if moz-appearance is set on a root
|
|
// node. If it has a parent content node, bail because it's not
|
|
// a root, other wise keep going in order to let the theme stuff
|
|
// draw the background. The canvas really should be drawing the
|
|
// bg, but there's no way to hook that up via css.
|
|
if (!aFrame->StyleDisplay()->HasAppearance()) {
|
|
return nullptr;
|
|
}
|
|
|
|
nsIContent* content = aFrame->GetContent();
|
|
if (!content || content->GetParent()) {
|
|
return nullptr;
|
|
}
|
|
|
|
f = aFrame;
|
|
}
|
|
return f;
|
|
}
|
|
|
|
static void SetBackgroundClipRegion(
|
|
DisplayListClipState::AutoSaveRestore& aClipState, nsIFrame* aFrame,
|
|
const nsStyleImageLayers::Layer& aLayer, const nsRect& aBackgroundRect,
|
|
bool aWillPaintBorder) {
|
|
nsCSSRendering::ImageLayerClipState clip;
|
|
nsCSSRendering::GetImageLayerClip(
|
|
aLayer, aFrame, *aFrame->StyleBorder(), aBackgroundRect, aBackgroundRect,
|
|
aWillPaintBorder, aFrame->PresContext()->AppUnitsPerDevPixel(), &clip);
|
|
|
|
if (clip.mHasAdditionalBGClipArea) {
|
|
aClipState.ClipContentDescendants(
|
|
clip.mAdditionalBGClipArea, clip.mBGClipArea,
|
|
clip.mHasRoundedCorners ? clip.mRadii : nullptr);
|
|
} else {
|
|
aClipState.ClipContentDescendants(
|
|
clip.mBGClipArea, clip.mHasRoundedCorners ? clip.mRadii : nullptr);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* This is used for the find bar highlighter overlay. It's only accessible
|
|
* through the AnonymousContent API, so it's not exposed to general web pages.
|
|
*/
|
|
static bool SpecialCutoutRegionCase(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame,
|
|
const nsRect& aBackgroundRect,
|
|
nsDisplayList* aList, nscolor aColor) {
|
|
nsIContent* content = aFrame->GetContent();
|
|
if (!content) {
|
|
return false;
|
|
}
|
|
|
|
void* cutoutRegion = content->GetProperty(nsGkAtoms::cutoutregion);
|
|
if (!cutoutRegion) {
|
|
return false;
|
|
}
|
|
|
|
if (NS_GET_A(aColor) == 0) {
|
|
return true;
|
|
}
|
|
|
|
nsRegion region;
|
|
region.Sub(aBackgroundRect, *static_cast<nsRegion*>(cutoutRegion));
|
|
region.MoveBy(aBuilder->ToReferenceFrame(aFrame));
|
|
aList->AppendNewToTop<nsDisplaySolidColorRegion>(aBuilder, aFrame, region,
|
|
aColor);
|
|
|
|
return true;
|
|
}
|
|
|
|
/*static*/
|
|
bool nsDisplayBackgroundImage::AppendBackgroundItemsToTop(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame,
|
|
const nsRect& aBackgroundRect, nsDisplayList* aList,
|
|
bool aAllowWillPaintBorderOptimization, ComputedStyle* aComputedStyle,
|
|
const nsRect& aBackgroundOriginRect, nsIFrame* aSecondaryReferenceFrame,
|
|
Maybe<nsDisplayListBuilder::AutoBuildingDisplayList>*
|
|
aAutoBuildingDisplayList) {
|
|
ComputedStyle* bgSC = aComputedStyle;
|
|
const nsStyleBackground* bg = nullptr;
|
|
nsRect bgRect = aBackgroundRect;
|
|
nsRect bgOriginRect = bgRect;
|
|
if (!aBackgroundOriginRect.IsEmpty()) {
|
|
bgOriginRect = aBackgroundOriginRect;
|
|
}
|
|
nsPresContext* presContext = aFrame->PresContext();
|
|
bool isThemed = aFrame->IsThemed();
|
|
nsIFrame* dependentFrame = nullptr;
|
|
if (!isThemed) {
|
|
if (!bgSC) {
|
|
dependentFrame = GetBackgroundComputedStyleFrame(aFrame);
|
|
if (dependentFrame) {
|
|
bgSC = dependentFrame->Style();
|
|
if (dependentFrame == aFrame) {
|
|
dependentFrame = nullptr;
|
|
}
|
|
}
|
|
}
|
|
if (bgSC) {
|
|
bg = bgSC->StyleBackground();
|
|
}
|
|
}
|
|
|
|
bool drawBackgroundColor = false;
|
|
// Dummy initialisation to keep Valgrind/Memcheck happy.
|
|
// See bug 1122375 comment 1.
|
|
nscolor color = NS_RGBA(0, 0, 0, 0);
|
|
if (!nsCSSRendering::IsCanvasFrame(aFrame) && bg) {
|
|
bool drawBackgroundImage;
|
|
color = nsCSSRendering::DetermineBackgroundColor(
|
|
presContext, bgSC, aFrame, drawBackgroundImage, drawBackgroundColor);
|
|
}
|
|
|
|
if (SpecialCutoutRegionCase(aBuilder, aFrame, aBackgroundRect, aList,
|
|
color)) {
|
|
return false;
|
|
}
|
|
|
|
const nsStyleBorder* borderStyle = aFrame->StyleBorder();
|
|
const nsStyleEffects* effectsStyle = aFrame->StyleEffects();
|
|
bool hasInsetShadow = effectsStyle->HasBoxShadowWithInset(true);
|
|
bool willPaintBorder = aAllowWillPaintBorderOptimization && !isThemed &&
|
|
!hasInsetShadow && borderStyle->HasBorder();
|
|
|
|
// An auxiliary list is necessary in case we have background blending; if that
|
|
// is the case, background items need to be wrapped by a blend container to
|
|
// isolate blending to the background
|
|
nsDisplayList bgItemList;
|
|
// Even if we don't actually have a background color to paint, we may still
|
|
// need to create an item for hit testing.
|
|
if ((drawBackgroundColor && color != NS_RGBA(0, 0, 0, 0)) ||
|
|
aBuilder->IsForEventDelivery()) {
|
|
if (aAutoBuildingDisplayList && !*aAutoBuildingDisplayList) {
|
|
aAutoBuildingDisplayList->emplace(aBuilder, aFrame);
|
|
}
|
|
Maybe<DisplayListClipState::AutoSaveRestore> clipState;
|
|
nsRect bgColorRect = bgRect;
|
|
if (bg && !aBuilder->IsForEventDelivery()) {
|
|
// Disable the will-paint-border optimization for background
|
|
// colors with no border-radius. Enabling it for background colors
|
|
// doesn't help much (there are no tiling issues) and clipping the
|
|
// background breaks detection of the element's border-box being
|
|
// opaque. For nonzero border-radius we still need it because we
|
|
// want to inset the background if possible to avoid antialiasing
|
|
// artifacts along the rounded corners.
|
|
bool useWillPaintBorderOptimization =
|
|
willPaintBorder &&
|
|
nsLayoutUtils::HasNonZeroCorner(borderStyle->mBorderRadius);
|
|
|
|
nsCSSRendering::ImageLayerClipState clip;
|
|
nsCSSRendering::GetImageLayerClip(
|
|
bg->BottomLayer(), aFrame, *aFrame->StyleBorder(), bgRect, bgRect,
|
|
useWillPaintBorderOptimization,
|
|
aFrame->PresContext()->AppUnitsPerDevPixel(), &clip);
|
|
|
|
bgColorRect = bgColorRect.Intersect(clip.mBGClipArea);
|
|
if (clip.mHasAdditionalBGClipArea) {
|
|
bgColorRect = bgColorRect.Intersect(clip.mAdditionalBGClipArea);
|
|
}
|
|
if (clip.mHasRoundedCorners) {
|
|
clipState.emplace(aBuilder);
|
|
clipState->ClipContentDescendants(clip.mBGClipArea, clip.mRadii);
|
|
}
|
|
}
|
|
nsDisplayBackgroundColor* bgItem;
|
|
if (aSecondaryReferenceFrame) {
|
|
bgItem = MakeDisplayItem<nsDisplayTableBackgroundColor>(
|
|
aBuilder, aSecondaryReferenceFrame, bgColorRect, bgSC,
|
|
drawBackgroundColor ? color : NS_RGBA(0, 0, 0, 0), aFrame);
|
|
} else {
|
|
bgItem = MakeDisplayItem<nsDisplayBackgroundColor>(
|
|
aBuilder, aFrame, bgColorRect, bgSC,
|
|
drawBackgroundColor ? color : NS_RGBA(0, 0, 0, 0));
|
|
}
|
|
if (bgItem) {
|
|
bgItem->SetDependentFrame(aBuilder, dependentFrame);
|
|
bgItemList.AppendToTop(bgItem);
|
|
}
|
|
}
|
|
|
|
if (isThemed) {
|
|
nsITheme* theme = presContext->GetTheme();
|
|
if (theme->NeedToClearBackgroundBehindWidget(
|
|
aFrame, aFrame->StyleDisplay()->mAppearance) &&
|
|
aBuilder->IsInChromeDocumentOrPopup() && !aBuilder->IsInTransform()) {
|
|
bgItemList.AppendNewToTop<nsDisplayClearBackground>(aBuilder, aFrame);
|
|
}
|
|
if (aSecondaryReferenceFrame) {
|
|
nsDisplayTableThemedBackground* bgItem =
|
|
MakeDisplayItem<nsDisplayTableThemedBackground>(
|
|
aBuilder, aSecondaryReferenceFrame, bgRect, aFrame);
|
|
if (bgItem) {
|
|
bgItem->Init(aBuilder);
|
|
bgItemList.AppendToTop(bgItem);
|
|
}
|
|
} else {
|
|
nsDisplayThemedBackground* bgItem =
|
|
MakeDisplayItem<nsDisplayThemedBackground>(aBuilder, aFrame, bgRect);
|
|
if (bgItem) {
|
|
bgItem->Init(aBuilder);
|
|
bgItemList.AppendToTop(bgItem);
|
|
}
|
|
}
|
|
aList->AppendToTop(&bgItemList);
|
|
return true;
|
|
}
|
|
|
|
if (!bg) {
|
|
aList->AppendToTop(&bgItemList);
|
|
return false;
|
|
}
|
|
|
|
const ActiveScrolledRoot* asr = aBuilder->CurrentActiveScrolledRoot();
|
|
|
|
bool needBlendContainer = false;
|
|
|
|
// Passing bg == nullptr in this macro will result in one iteration with
|
|
// i = 0.
|
|
NS_FOR_VISIBLE_IMAGE_LAYERS_BACK_TO_FRONT(i, bg->mImage) {
|
|
if (bg->mImage.mLayers[i].mImage.IsNone()) {
|
|
continue;
|
|
}
|
|
|
|
if (aAutoBuildingDisplayList && !*aAutoBuildingDisplayList) {
|
|
aAutoBuildingDisplayList->emplace(aBuilder, aFrame);
|
|
}
|
|
|
|
if (bg->mImage.mLayers[i].mBlendMode != NS_STYLE_BLEND_NORMAL) {
|
|
needBlendContainer = true;
|
|
}
|
|
|
|
DisplayListClipState::AutoSaveRestore clipState(aBuilder);
|
|
if (!aBuilder->IsForEventDelivery()) {
|
|
const nsStyleImageLayers::Layer& layer = bg->mImage.mLayers[i];
|
|
SetBackgroundClipRegion(clipState, aFrame, layer, bgRect,
|
|
willPaintBorder);
|
|
}
|
|
|
|
nsDisplayList thisItemList;
|
|
nsDisplayBackgroundImage::InitData bgData =
|
|
nsDisplayBackgroundImage::GetInitData(aBuilder, aFrame, i, bgOriginRect,
|
|
bgSC);
|
|
|
|
if (bgData.shouldFixToViewport) {
|
|
auto* displayData = aBuilder->GetCurrentFixedBackgroundDisplayData();
|
|
nsDisplayListBuilder::AutoBuildingDisplayList buildingDisplayList(
|
|
aBuilder, aFrame, aBuilder->GetVisibleRect(),
|
|
aBuilder->GetDirtyRect());
|
|
|
|
nsDisplayListBuilder::AutoCurrentActiveScrolledRootSetter asrSetter(
|
|
aBuilder);
|
|
if (displayData) {
|
|
asrSetter.SetCurrentActiveScrolledRoot(
|
|
displayData->mContainingBlockActiveScrolledRoot);
|
|
if (nsLayoutUtils::UsesAsyncScrolling(aFrame)) {
|
|
// Override the dirty rect on the builder to be the dirty rect of
|
|
// the viewport.
|
|
// displayData->mDirtyRect is relative to the presshell's viewport
|
|
// frame (the root frame), and we need it to be relative to aFrame.
|
|
nsIFrame* rootFrame =
|
|
aBuilder->CurrentPresShellState()->mPresShell->GetRootFrame();
|
|
// There cannot be any transforms between aFrame and rootFrame
|
|
// because then bgData.shouldFixToViewport would have been false.
|
|
nsRect visibleRect =
|
|
displayData->mVisibleRect + aFrame->GetOffsetTo(rootFrame);
|
|
aBuilder->SetVisibleRect(visibleRect);
|
|
nsRect dirtyRect =
|
|
displayData->mDirtyRect + aFrame->GetOffsetTo(rootFrame);
|
|
aBuilder->SetDirtyRect(dirtyRect);
|
|
}
|
|
}
|
|
nsDisplayBackgroundImage* bgItem = nullptr;
|
|
{
|
|
// The clip is captured by the nsDisplayFixedPosition, so clear the
|
|
// clip for the nsDisplayBackgroundImage inside.
|
|
DisplayListClipState::AutoSaveRestore bgImageClip(aBuilder);
|
|
bgImageClip.Clear();
|
|
if (aSecondaryReferenceFrame) {
|
|
bgItem = MakeDisplayItem<nsDisplayTableBackgroundImage>(
|
|
aBuilder, aSecondaryReferenceFrame, bgData, aFrame);
|
|
} else {
|
|
bgItem = MakeDisplayItem<nsDisplayBackgroundImage>(aBuilder, aFrame,
|
|
bgData);
|
|
}
|
|
}
|
|
if (bgItem) {
|
|
bgItem->SetDependentFrame(aBuilder, dependentFrame);
|
|
if (aSecondaryReferenceFrame) {
|
|
thisItemList.AppendToTop(
|
|
nsDisplayTableFixedPosition::CreateForFixedBackground(
|
|
aBuilder, aSecondaryReferenceFrame, bgItem, i, aFrame));
|
|
} else {
|
|
thisItemList.AppendToTop(
|
|
nsDisplayFixedPosition::CreateForFixedBackground(aBuilder, aFrame,
|
|
bgItem, i));
|
|
}
|
|
}
|
|
|
|
} else {
|
|
nsDisplayBackgroundImage* bgItem;
|
|
if (aSecondaryReferenceFrame) {
|
|
bgItem = MakeDisplayItem<nsDisplayTableBackgroundImage>(
|
|
aBuilder, aSecondaryReferenceFrame, bgData, aFrame);
|
|
} else {
|
|
bgItem =
|
|
MakeDisplayItem<nsDisplayBackgroundImage>(aBuilder, aFrame, bgData);
|
|
}
|
|
if (bgItem) {
|
|
bgItem->SetDependentFrame(aBuilder, dependentFrame);
|
|
thisItemList.AppendToTop(bgItem);
|
|
}
|
|
}
|
|
|
|
if (bg->mImage.mLayers[i].mBlendMode != NS_STYLE_BLEND_NORMAL) {
|
|
DisplayListClipState::AutoSaveRestore blendClip(aBuilder);
|
|
// asr is scrolled. Even if we wrap a fixed background layer, that's
|
|
// fine, because the item will have a scrolled clip that limits the
|
|
// item with respect to asr.
|
|
if (aSecondaryReferenceFrame) {
|
|
thisItemList.AppendNewToTop<nsDisplayTableBlendMode>(
|
|
aBuilder, aSecondaryReferenceFrame, &thisItemList,
|
|
bg->mImage.mLayers[i].mBlendMode, asr, i + 1, aFrame);
|
|
} else {
|
|
thisItemList.AppendNewToTop<nsDisplayBlendMode>(
|
|
aBuilder, aFrame, &thisItemList, bg->mImage.mLayers[i].mBlendMode,
|
|
asr, i + 1);
|
|
}
|
|
}
|
|
bgItemList.AppendToTop(&thisItemList);
|
|
}
|
|
|
|
if (needBlendContainer) {
|
|
DisplayListClipState::AutoSaveRestore blendContainerClip(aBuilder);
|
|
if (aSecondaryReferenceFrame) {
|
|
bgItemList.AppendToTop(
|
|
nsDisplayTableBlendContainer::CreateForBackgroundBlendMode(
|
|
aBuilder, aSecondaryReferenceFrame, &bgItemList, asr, aFrame));
|
|
} else {
|
|
bgItemList.AppendToTop(
|
|
nsDisplayBlendContainer::CreateForBackgroundBlendMode(
|
|
aBuilder, aFrame, &bgItemList, asr));
|
|
}
|
|
}
|
|
|
|
aList->AppendToTop(&bgItemList);
|
|
return false;
|
|
}
|
|
|
|
// Check that the rounded border of aFrame, added to aToReferenceFrame,
|
|
// intersects aRect. Assumes that the unrounded border has already
|
|
// been checked for intersection.
|
|
static bool RoundedBorderIntersectsRect(nsIFrame* aFrame,
|
|
const nsPoint& aFrameToReferenceFrame,
|
|
const nsRect& aTestRect) {
|
|
if (!nsRect(aFrameToReferenceFrame, aFrame->GetSize()).Intersects(aTestRect))
|
|
return false;
|
|
|
|
nscoord radii[8];
|
|
return !aFrame->GetBorderRadii(radii) ||
|
|
nsLayoutUtils::RoundedRectIntersectsRect(
|
|
nsRect(aFrameToReferenceFrame, aFrame->GetSize()), radii,
|
|
aTestRect);
|
|
}
|
|
|
|
// Returns TRUE if aContainedRect is guaranteed to be contained in
|
|
// the rounded rect defined by aRoundedRect and aRadii. Complex cases are
|
|
// handled conservatively by returning FALSE in some situations where
|
|
// a more thorough analysis could return TRUE.
|
|
//
|
|
// See also RoundedRectIntersectsRect.
|
|
static bool RoundedRectContainsRect(const nsRect& aRoundedRect,
|
|
const nscoord aRadii[8],
|
|
const nsRect& aContainedRect) {
|
|
nsRegion rgn = nsLayoutUtils::RoundedRectIntersectRect(aRoundedRect, aRadii,
|
|
aContainedRect);
|
|
return rgn.Contains(aContainedRect);
|
|
}
|
|
|
|
bool nsDisplayBackgroundImage::CanOptimizeToImageLayer(
|
|
LayerManager* aManager, nsDisplayListBuilder* aBuilder) {
|
|
if (!mBackgroundStyle) {
|
|
return false;
|
|
}
|
|
|
|
// We currently can't handle tiled backgrounds.
|
|
if (!mDestRect.Contains(mFillRect)) {
|
|
return false;
|
|
}
|
|
|
|
// For 'contain' and 'cover', we allow any pixel of the image to be sampled
|
|
// because there isn't going to be any spriting/atlasing going on.
|
|
const nsStyleImageLayers::Layer& layer =
|
|
mBackgroundStyle->StyleBackground()->mImage.mLayers[mLayer];
|
|
bool allowPartialImages = layer.mSize.IsContain() || layer.mSize.IsCover();
|
|
if (!allowPartialImages && !mFillRect.Contains(mDestRect)) {
|
|
return false;
|
|
}
|
|
|
|
return nsDisplayImageContainer::CanOptimizeToImageLayer(aManager, aBuilder);
|
|
}
|
|
|
|
nsRect nsDisplayBackgroundImage::GetDestRect() const { return mDestRect; }
|
|
|
|
already_AddRefed<imgIContainer> nsDisplayBackgroundImage::GetImage() {
|
|
nsCOMPtr<imgIContainer> image = mImage;
|
|
return image.forget();
|
|
}
|
|
|
|
nsDisplayBackgroundImage::ImageLayerization
|
|
nsDisplayBackgroundImage::ShouldCreateOwnLayer(nsDisplayListBuilder* aBuilder,
|
|
LayerManager* aManager) {
|
|
if (ForceActiveLayers()) {
|
|
return WHENEVER_POSSIBLE;
|
|
}
|
|
|
|
nsIFrame* backgroundStyleFrame =
|
|
nsCSSRendering::FindBackgroundStyleFrame(StyleFrame());
|
|
if (ActiveLayerTracker::IsBackgroundPositionAnimated(aBuilder,
|
|
backgroundStyleFrame)) {
|
|
return WHENEVER_POSSIBLE;
|
|
}
|
|
|
|
if (StaticPrefs::layout_animated_image_layers_enabled() && mBackgroundStyle) {
|
|
const nsStyleImageLayers::Layer& layer =
|
|
mBackgroundStyle->StyleBackground()->mImage.mLayers[mLayer];
|
|
const auto* image = &layer.mImage;
|
|
if (auto* request = image->GetImageRequest()) {
|
|
nsCOMPtr<imgIContainer> image;
|
|
if (NS_SUCCEEDED(request->GetImage(getter_AddRefs(image))) && image) {
|
|
bool animated = false;
|
|
if (NS_SUCCEEDED(image->GetAnimated(&animated)) && animated) {
|
|
return WHENEVER_POSSIBLE;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
if (nsLayoutUtils::GPUImageScalingEnabled() &&
|
|
aManager->IsCompositingCheap()) {
|
|
return ONLY_FOR_SCALING;
|
|
}
|
|
|
|
return NO_LAYER_NEEDED;
|
|
}
|
|
|
|
static void CheckForBorderItem(nsDisplayItem* aItem, uint32_t& aFlags) {
|
|
nsDisplayItem* nextItem = aItem->GetAbove();
|
|
while (nextItem && nextItem->GetType() == DisplayItemType::TYPE_BACKGROUND) {
|
|
nextItem = nextItem->GetAbove();
|
|
}
|
|
if (nextItem && nextItem->Frame() == aItem->Frame() &&
|
|
nextItem->GetType() == DisplayItemType::TYPE_BORDER) {
|
|
aFlags |= nsCSSRendering::PAINTBG_WILL_PAINT_BORDER;
|
|
}
|
|
}
|
|
|
|
LayerState nsDisplayBackgroundImage::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
mImageFlags = aBuilder->GetBackgroundPaintFlags();
|
|
CheckForBorderItem(this, mImageFlags);
|
|
|
|
ImageLayerization shouldLayerize = ShouldCreateOwnLayer(aBuilder, aManager);
|
|
if (shouldLayerize == NO_LAYER_NEEDED) {
|
|
// We can skip the call to CanOptimizeToImageLayer if we don't want a
|
|
// layer anyway.
|
|
return LayerState::LAYER_NONE;
|
|
}
|
|
|
|
if (CanOptimizeToImageLayer(aManager, aBuilder)) {
|
|
if (shouldLayerize == WHENEVER_POSSIBLE) {
|
|
return LayerState::LAYER_ACTIVE;
|
|
}
|
|
|
|
MOZ_ASSERT(shouldLayerize == ONLY_FOR_SCALING,
|
|
"unhandled ImageLayerization value?");
|
|
|
|
MOZ_ASSERT(mImage);
|
|
int32_t imageWidth;
|
|
int32_t imageHeight;
|
|
mImage->GetWidth(&imageWidth);
|
|
mImage->GetHeight(&imageHeight);
|
|
NS_ASSERTION(imageWidth != 0 && imageHeight != 0, "Invalid image size!");
|
|
|
|
int32_t appUnitsPerDevPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
LayoutDeviceRect destRect =
|
|
LayoutDeviceRect::FromAppUnits(GetDestRect(), appUnitsPerDevPixel);
|
|
|
|
const LayerRect destLayerRect = destRect * aParameters.Scale();
|
|
|
|
// Calculate the scaling factor for the frame.
|
|
const gfxSize scale = gfxSize(destLayerRect.width / imageWidth,
|
|
destLayerRect.height / imageHeight);
|
|
|
|
if ((scale.width != 1.0f || scale.height != 1.0f) &&
|
|
(destLayerRect.width * destLayerRect.height >= 64 * 64)) {
|
|
// Separate this image into a layer.
|
|
// There's no point in doing this if we are not scaling at all or if the
|
|
// target size is pretty small.
|
|
return LayerState::LAYER_ACTIVE;
|
|
}
|
|
}
|
|
|
|
return LayerState::LAYER_NONE;
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayBackgroundImage::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
RefPtr<ImageLayer> layer = static_cast<ImageLayer*>(
|
|
aManager->GetLayerBuilder()->GetLeafLayerFor(aBuilder, this));
|
|
if (!layer) {
|
|
layer = aManager->CreateImageLayer();
|
|
if (!layer) {
|
|
return nullptr;
|
|
}
|
|
}
|
|
RefPtr<ImageContainer> imageContainer = GetContainer(aManager, aBuilder);
|
|
layer->SetContainer(imageContainer);
|
|
ConfigureLayer(layer, aParameters);
|
|
return layer.forget();
|
|
}
|
|
|
|
bool nsDisplayBackgroundImage::CanBuildWebRenderDisplayItems(
|
|
LayerManager* aManager, nsDisplayListBuilder* aDisplayListBuilder) {
|
|
if (aDisplayListBuilder) {
|
|
mImageFlags = aDisplayListBuilder->GetBackgroundPaintFlags();
|
|
}
|
|
|
|
return mBackgroundStyle->StyleBackground()->mImage.mLayers[mLayer].mClip !=
|
|
StyleGeometryBox::Text &&
|
|
nsCSSRendering::CanBuildWebRenderDisplayItemsForStyleImageLayer(
|
|
aManager, *StyleFrame()->PresContext(), StyleFrame(),
|
|
mBackgroundStyle->StyleBackground(), mLayer, mImageFlags);
|
|
}
|
|
|
|
bool nsDisplayBackgroundImage::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc, RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
if (!CanBuildWebRenderDisplayItems(aManager->LayerManager(),
|
|
aDisplayListBuilder)) {
|
|
return false;
|
|
}
|
|
|
|
CheckForBorderItem(this, mImageFlags);
|
|
nsCSSRendering::PaintBGParams params =
|
|
nsCSSRendering::PaintBGParams::ForSingleLayer(
|
|
*StyleFrame()->PresContext(), GetPaintRect(), mBackgroundRect,
|
|
StyleFrame(), mImageFlags, mLayer, CompositionOp::OP_OVER);
|
|
params.bgClipRect = &mBounds;
|
|
ImgDrawResult result =
|
|
nsCSSRendering::BuildWebRenderDisplayItemsForStyleImageLayer(
|
|
params, aBuilder, aResources, aSc, aManager, this);
|
|
if (result == ImgDrawResult::NOT_SUPPORTED) {
|
|
return false;
|
|
}
|
|
|
|
nsDisplayBackgroundGeometry::UpdateDrawResult(this, result);
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayBackgroundImage::HitTest(nsDisplayListBuilder* aBuilder,
|
|
const nsRect& aRect,
|
|
HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) {
|
|
if (RoundedBorderIntersectsRect(mFrame, ToReferenceFrame(), aRect)) {
|
|
aOutFrames->AppendElement(mFrame);
|
|
}
|
|
}
|
|
|
|
bool nsDisplayBackgroundImage::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
if (!nsDisplayImageContainer::ComputeVisibility(aBuilder, aVisibleRegion)) {
|
|
return false;
|
|
}
|
|
|
|
// Return false if the background was propagated away from this
|
|
// frame. We don't want this display item to show up and confuse
|
|
// anything.
|
|
return mBackgroundStyle;
|
|
}
|
|
|
|
/* static */
|
|
nsRegion nsDisplayBackgroundImage::GetInsideClipRegion(
|
|
const nsDisplayItem* aItem, StyleGeometryBox aClip, const nsRect& aRect,
|
|
const nsRect& aBackgroundRect) {
|
|
nsRegion result;
|
|
if (aRect.IsEmpty()) {
|
|
return result;
|
|
}
|
|
|
|
nsIFrame* frame = aItem->Frame();
|
|
|
|
nsRect clipRect = aBackgroundRect;
|
|
if (frame->IsCanvasFrame()) {
|
|
nsCanvasFrame* canvasFrame = static_cast<nsCanvasFrame*>(frame);
|
|
clipRect = canvasFrame->CanvasArea() + aItem->ToReferenceFrame();
|
|
} else if (aClip == StyleGeometryBox::PaddingBox ||
|
|
aClip == StyleGeometryBox::ContentBox) {
|
|
nsMargin border = frame->GetUsedBorder();
|
|
if (aClip == StyleGeometryBox::ContentBox) {
|
|
border += frame->GetUsedPadding();
|
|
}
|
|
border.ApplySkipSides(frame->GetSkipSides());
|
|
clipRect.Deflate(border);
|
|
}
|
|
|
|
return clipRect.Intersect(aRect);
|
|
}
|
|
|
|
nsRegion nsDisplayBackgroundImage::GetOpaqueRegion(
|
|
nsDisplayListBuilder* aBuilder, bool* aSnap) const {
|
|
nsRegion result;
|
|
*aSnap = false;
|
|
|
|
if (!mBackgroundStyle) {
|
|
return result;
|
|
}
|
|
|
|
*aSnap = true;
|
|
|
|
// For StyleBoxDecorationBreak::Slice, don't try to optimize here, since
|
|
// this could easily lead to O(N^2) behavior inside InlineBackgroundData,
|
|
// which expects frames to be sent to it in content order, not reverse
|
|
// content order which we'll produce here.
|
|
// Of course, if there's only one frame in the flow, it doesn't matter.
|
|
if (mFrame->StyleBorder()->mBoxDecorationBreak ==
|
|
StyleBoxDecorationBreak::Clone ||
|
|
(!mFrame->GetPrevContinuation() && !mFrame->GetNextContinuation())) {
|
|
const nsStyleImageLayers::Layer& layer =
|
|
mBackgroundStyle->StyleBackground()->mImage.mLayers[mLayer];
|
|
if (layer.mImage.IsOpaque() && layer.mBlendMode == NS_STYLE_BLEND_NORMAL &&
|
|
layer.mRepeat.mXRepeat != StyleImageLayerRepeat::Space &&
|
|
layer.mRepeat.mYRepeat != StyleImageLayerRepeat::Space &&
|
|
layer.mClip != StyleGeometryBox::Text) {
|
|
result = GetInsideClipRegion(this, layer.mClip, mBounds, mBackgroundRect);
|
|
}
|
|
}
|
|
|
|
return result;
|
|
}
|
|
|
|
Maybe<nscolor> nsDisplayBackgroundImage::IsUniform(
|
|
nsDisplayListBuilder* aBuilder) const {
|
|
if (!mBackgroundStyle) {
|
|
return Some(NS_RGBA(0, 0, 0, 0));
|
|
}
|
|
return Nothing();
|
|
}
|
|
|
|
nsRect nsDisplayBackgroundImage::GetPositioningArea() const {
|
|
if (!mBackgroundStyle) {
|
|
return nsRect();
|
|
}
|
|
nsIFrame* attachedToFrame;
|
|
bool transformedFixed;
|
|
return nsCSSRendering::ComputeImageLayerPositioningArea(
|
|
mFrame->PresContext(), mFrame, mBackgroundRect,
|
|
mBackgroundStyle->StyleBackground()->mImage.mLayers[mLayer],
|
|
&attachedToFrame, &transformedFixed) +
|
|
ToReferenceFrame();
|
|
}
|
|
|
|
bool nsDisplayBackgroundImage::RenderingMightDependOnPositioningAreaSizeChange()
|
|
const {
|
|
if (!mBackgroundStyle) {
|
|
return false;
|
|
}
|
|
|
|
nscoord radii[8];
|
|
if (mFrame->GetBorderRadii(radii)) {
|
|
// A change in the size of the positioning area might change the position
|
|
// of the rounded corners.
|
|
return true;
|
|
}
|
|
|
|
const nsStyleImageLayers::Layer& layer =
|
|
mBackgroundStyle->StyleBackground()->mImage.mLayers[mLayer];
|
|
if (layer.RenderingMightDependOnPositioningAreaSizeChange()) {
|
|
return true;
|
|
}
|
|
return false;
|
|
}
|
|
|
|
void nsDisplayBackgroundImage::Paint(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx) {
|
|
PaintInternal(aBuilder, aCtx, GetPaintRect(), &mBounds);
|
|
}
|
|
|
|
void nsDisplayBackgroundImage::PaintInternal(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx,
|
|
const nsRect& aBounds,
|
|
nsRect* aClipRect) {
|
|
gfxContext* ctx = aCtx;
|
|
StyleGeometryBox clip =
|
|
mBackgroundStyle->StyleBackground()->mImage.mLayers[mLayer].mClip;
|
|
|
|
if (clip == StyleGeometryBox::Text) {
|
|
if (!GenerateAndPushTextMask(StyleFrame(), aCtx, mBackgroundRect,
|
|
aBuilder)) {
|
|
return;
|
|
}
|
|
}
|
|
|
|
nsCSSRendering::PaintBGParams params =
|
|
nsCSSRendering::PaintBGParams::ForSingleLayer(
|
|
*StyleFrame()->PresContext(), aBounds, mBackgroundRect, StyleFrame(),
|
|
mImageFlags, mLayer, CompositionOp::OP_OVER);
|
|
params.bgClipRect = aClipRect;
|
|
ImgDrawResult result = nsCSSRendering::PaintStyleImageLayer(params, *aCtx);
|
|
|
|
if (clip == StyleGeometryBox::Text) {
|
|
ctx->PopGroupAndBlend();
|
|
}
|
|
|
|
nsDisplayBackgroundGeometry::UpdateDrawResult(this, result);
|
|
}
|
|
|
|
void nsDisplayBackgroundImage::ComputeInvalidationRegion(
|
|
nsDisplayListBuilder* aBuilder, const nsDisplayItemGeometry* aGeometry,
|
|
nsRegion* aInvalidRegion) const {
|
|
if (!mBackgroundStyle) {
|
|
return;
|
|
}
|
|
|
|
auto* geometry = static_cast<const nsDisplayBackgroundGeometry*>(aGeometry);
|
|
|
|
bool snap;
|
|
nsRect bounds = GetBounds(aBuilder, &snap);
|
|
nsRect positioningArea = GetPositioningArea();
|
|
if (positioningArea.TopLeft() != geometry->mPositioningArea.TopLeft() ||
|
|
(positioningArea.Size() != geometry->mPositioningArea.Size() &&
|
|
RenderingMightDependOnPositioningAreaSizeChange())) {
|
|
// Positioning area changed in a way that could cause everything to change,
|
|
// so invalidate everything (both old and new painting areas).
|
|
aInvalidRegion->Or(bounds, geometry->mBounds);
|
|
return;
|
|
}
|
|
if (!mDestRect.IsEqualInterior(geometry->mDestRect)) {
|
|
// Dest area changed in a way that could cause everything to change,
|
|
// so invalidate everything (both old and new painting areas).
|
|
aInvalidRegion->Or(bounds, geometry->mBounds);
|
|
return;
|
|
}
|
|
if (aBuilder->ShouldSyncDecodeImages()) {
|
|
const auto& image =
|
|
mBackgroundStyle->StyleBackground()->mImage.mLayers[mLayer].mImage;
|
|
if (image.IsImageRequestType() &&
|
|
geometry->ShouldInvalidateToSyncDecodeImages()) {
|
|
aInvalidRegion->Or(*aInvalidRegion, bounds);
|
|
}
|
|
}
|
|
if (!bounds.IsEqualInterior(geometry->mBounds)) {
|
|
// Positioning area is unchanged, so invalidate just the change in the
|
|
// painting area.
|
|
aInvalidRegion->Xor(bounds, geometry->mBounds);
|
|
}
|
|
}
|
|
|
|
nsRect nsDisplayBackgroundImage::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = true;
|
|
return mBounds;
|
|
}
|
|
|
|
nsRect nsDisplayBackgroundImage::GetBoundsInternal(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrameForBounds) {
|
|
// This allows nsDisplayTableBackgroundImage to change the frame used for
|
|
// bounds calculation.
|
|
nsIFrame* frame = aFrameForBounds ? aFrameForBounds : mFrame;
|
|
|
|
nsPresContext* presContext = frame->PresContext();
|
|
|
|
if (!mBackgroundStyle) {
|
|
return nsRect();
|
|
}
|
|
|
|
nsRect clipRect = mBackgroundRect;
|
|
if (frame->IsCanvasFrame()) {
|
|
nsCanvasFrame* canvasFrame = static_cast<nsCanvasFrame*>(frame);
|
|
clipRect = canvasFrame->CanvasArea() + ToReferenceFrame();
|
|
}
|
|
const nsStyleImageLayers::Layer& layer =
|
|
mBackgroundStyle->StyleBackground()->mImage.mLayers[mLayer];
|
|
return nsCSSRendering::GetBackgroundLayerRect(
|
|
presContext, frame, mBackgroundRect, clipRect, layer,
|
|
aBuilder->GetBackgroundPaintFlags());
|
|
}
|
|
|
|
nsDisplayTableBackgroundImage::nsDisplayTableBackgroundImage(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, const InitData& aData,
|
|
nsIFrame* aCellFrame)
|
|
: nsDisplayBackgroundImage(aBuilder, aFrame, aData, aCellFrame),
|
|
mStyleFrame(aCellFrame),
|
|
mTableType(GetTableTypeFromFrame(mStyleFrame)) {
|
|
if (aBuilder->IsRetainingDisplayList()) {
|
|
mStyleFrame->AddDisplayItem(this);
|
|
}
|
|
}
|
|
|
|
nsDisplayTableBackgroundImage::~nsDisplayTableBackgroundImage() {
|
|
if (mStyleFrame) {
|
|
mStyleFrame->RemoveDisplayItem(this);
|
|
}
|
|
}
|
|
|
|
bool nsDisplayTableBackgroundImage::IsInvalid(nsRect& aRect) const {
|
|
bool result = mStyleFrame ? mStyleFrame->IsInvalid(aRect) : false;
|
|
aRect += ToReferenceFrame();
|
|
return result;
|
|
}
|
|
|
|
nsDisplayThemedBackground::nsDisplayThemedBackground(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame,
|
|
const nsRect& aBackgroundRect)
|
|
: nsPaintedDisplayItem(aBuilder, aFrame), mBackgroundRect(aBackgroundRect) {
|
|
MOZ_COUNT_CTOR(nsDisplayThemedBackground);
|
|
}
|
|
|
|
void nsDisplayThemedBackground::Init(nsDisplayListBuilder* aBuilder) {
|
|
const nsStyleDisplay* disp = StyleFrame()->StyleDisplay();
|
|
mAppearance = disp->mAppearance;
|
|
StyleFrame()->IsThemed(disp, &mThemeTransparency);
|
|
|
|
// Perform necessary RegisterThemeGeometry
|
|
nsITheme* theme = StyleFrame()->PresContext()->GetTheme();
|
|
nsITheme::ThemeGeometryType type =
|
|
theme->ThemeGeometryTypeForWidget(StyleFrame(), disp->mAppearance);
|
|
if (type != nsITheme::eThemeGeometryTypeUnknown) {
|
|
RegisterThemeGeometry(aBuilder, this, StyleFrame(), type);
|
|
}
|
|
|
|
if (disp->mAppearance == StyleAppearance::MozWinBorderlessGlass ||
|
|
disp->mAppearance == StyleAppearance::MozWinGlass) {
|
|
aBuilder->SetGlassDisplayItem(this);
|
|
}
|
|
|
|
mBounds = GetBoundsInternal();
|
|
}
|
|
|
|
void nsDisplayThemedBackground::WriteDebugInfo(std::stringstream& aStream) {
|
|
aStream << " (themed, appearance:" << (int)mAppearance << ")";
|
|
}
|
|
|
|
void nsDisplayThemedBackground::HitTest(nsDisplayListBuilder* aBuilder,
|
|
const nsRect& aRect,
|
|
HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) {
|
|
// Assume that any point in our background rect is a hit.
|
|
if (mBackgroundRect.Intersects(aRect)) {
|
|
aOutFrames->AppendElement(mFrame);
|
|
}
|
|
}
|
|
|
|
nsRegion nsDisplayThemedBackground::GetOpaqueRegion(
|
|
nsDisplayListBuilder* aBuilder, bool* aSnap) const {
|
|
nsRegion result;
|
|
*aSnap = false;
|
|
|
|
if (mThemeTransparency == nsITheme::eOpaque) {
|
|
*aSnap = true;
|
|
result = mBackgroundRect;
|
|
}
|
|
return result;
|
|
}
|
|
|
|
Maybe<nscolor> nsDisplayThemedBackground::IsUniform(
|
|
nsDisplayListBuilder* aBuilder) const {
|
|
if (mAppearance == StyleAppearance::MozWinBorderlessGlass ||
|
|
mAppearance == StyleAppearance::MozWinGlass) {
|
|
return Some(NS_RGBA(0, 0, 0, 0));
|
|
}
|
|
return Nothing();
|
|
}
|
|
|
|
nsRect nsDisplayThemedBackground::GetPositioningArea() const {
|
|
return mBackgroundRect;
|
|
}
|
|
|
|
void nsDisplayThemedBackground::Paint(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx) {
|
|
PaintInternal(aBuilder, aCtx, GetPaintRect(), nullptr);
|
|
}
|
|
|
|
void nsDisplayThemedBackground::PaintInternal(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx,
|
|
const nsRect& aBounds,
|
|
nsRect* aClipRect) {
|
|
// XXXzw this ignores aClipRect.
|
|
nsPresContext* presContext = StyleFrame()->PresContext();
|
|
nsITheme* theme = presContext->GetTheme();
|
|
nsRect drawing(mBackgroundRect);
|
|
theme->GetWidgetOverflow(presContext->DeviceContext(), StyleFrame(),
|
|
mAppearance, &drawing);
|
|
drawing.IntersectRect(drawing, aBounds);
|
|
theme->DrawWidgetBackground(aCtx, StyleFrame(), mAppearance, mBackgroundRect,
|
|
drawing);
|
|
}
|
|
|
|
bool nsDisplayThemedBackground::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
nsITheme* theme = StyleFrame()->PresContext()->GetTheme();
|
|
return theme->CreateWebRenderCommandsForWidget(aBuilder, aResources, aSc,
|
|
aManager, StyleFrame(),
|
|
mAppearance, mBackgroundRect);
|
|
}
|
|
|
|
bool nsDisplayThemedBackground::IsWindowActive() const {
|
|
EventStates docState = mFrame->GetContent()->OwnerDoc()->GetDocumentState();
|
|
return !docState.HasState(NS_DOCUMENT_STATE_WINDOW_INACTIVE);
|
|
}
|
|
|
|
void nsDisplayThemedBackground::ComputeInvalidationRegion(
|
|
nsDisplayListBuilder* aBuilder, const nsDisplayItemGeometry* aGeometry,
|
|
nsRegion* aInvalidRegion) const {
|
|
auto* geometry =
|
|
static_cast<const nsDisplayThemedBackgroundGeometry*>(aGeometry);
|
|
|
|
bool snap;
|
|
nsRect bounds = GetBounds(aBuilder, &snap);
|
|
nsRect positioningArea = GetPositioningArea();
|
|
if (!positioningArea.IsEqualInterior(geometry->mPositioningArea)) {
|
|
// Invalidate everything (both old and new painting areas).
|
|
aInvalidRegion->Or(bounds, geometry->mBounds);
|
|
return;
|
|
}
|
|
if (!bounds.IsEqualInterior(geometry->mBounds)) {
|
|
// Positioning area is unchanged, so invalidate just the change in the
|
|
// painting area.
|
|
aInvalidRegion->Xor(bounds, geometry->mBounds);
|
|
}
|
|
nsITheme* theme = StyleFrame()->PresContext()->GetTheme();
|
|
if (theme->WidgetAppearanceDependsOnWindowFocus(mAppearance) &&
|
|
IsWindowActive() != geometry->mWindowIsActive) {
|
|
aInvalidRegion->Or(*aInvalidRegion, bounds);
|
|
}
|
|
}
|
|
|
|
nsRect nsDisplayThemedBackground::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = true;
|
|
return mBounds;
|
|
}
|
|
|
|
nsRect nsDisplayThemedBackground::GetBoundsInternal() {
|
|
nsPresContext* presContext = mFrame->PresContext();
|
|
|
|
nsRect r = mBackgroundRect - ToReferenceFrame();
|
|
presContext->GetTheme()->GetWidgetOverflow(
|
|
presContext->DeviceContext(), mFrame, mFrame->StyleDisplay()->mAppearance,
|
|
&r);
|
|
return r + ToReferenceFrame();
|
|
}
|
|
|
|
void nsDisplayImageContainer::ConfigureLayer(
|
|
ImageLayer* aLayer, const ContainerLayerParameters& aParameters) {
|
|
aLayer->SetSamplingFilter(nsLayoutUtils::GetSamplingFilterForFrame(mFrame));
|
|
|
|
nsCOMPtr<imgIContainer> image = GetImage();
|
|
MOZ_ASSERT(image);
|
|
int32_t imageWidth;
|
|
int32_t imageHeight;
|
|
image->GetWidth(&imageWidth);
|
|
image->GetHeight(&imageHeight);
|
|
NS_ASSERTION(imageWidth != 0 && imageHeight != 0, "Invalid image size!");
|
|
|
|
if (imageWidth > 0 && imageHeight > 0) {
|
|
// We're actually using the ImageContainer. Let our frame know that it
|
|
// should consider itself to have painted successfully.
|
|
UpdateDrawResult(ImgDrawResult::SUCCESS);
|
|
}
|
|
|
|
// It's possible (for example, due to downscale-during-decode) that the
|
|
// ImageContainer this ImageLayer is holding has a different size from the
|
|
// intrinsic size of the image. For this reason we compute the transform using
|
|
// the ImageContainer's size rather than the image's intrinsic size.
|
|
// XXX(seth): In reality, since the size of the ImageContainer may change
|
|
// asynchronously, this is not enough. Bug 1183378 will provide a more
|
|
// complete fix, but this solution is safe in more cases than simply relying
|
|
// on the intrinsic size.
|
|
IntSize containerSize = aLayer->GetContainer()
|
|
? aLayer->GetContainer()->GetCurrentSize()
|
|
: IntSize(imageWidth, imageHeight);
|
|
|
|
const int32_t factor = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
const LayoutDeviceRect destRect(
|
|
LayoutDeviceIntRect::FromAppUnitsToNearest(GetDestRect(), factor));
|
|
|
|
const LayoutDevicePoint p = destRect.TopLeft();
|
|
Matrix transform = Matrix::Translation(p.x + aParameters.mOffset.x,
|
|
p.y + aParameters.mOffset.y);
|
|
transform.PreScale(destRect.width / containerSize.width,
|
|
destRect.height / containerSize.height);
|
|
aLayer->SetBaseTransform(gfx::Matrix4x4::From2D(transform));
|
|
}
|
|
|
|
already_AddRefed<ImageContainer> nsDisplayImageContainer::GetContainer(
|
|
LayerManager* aManager, nsDisplayListBuilder* aBuilder) {
|
|
nsCOMPtr<imgIContainer> image = GetImage();
|
|
if (!image) {
|
|
MOZ_ASSERT_UNREACHABLE(
|
|
"Must call CanOptimizeToImage() and get true "
|
|
"before calling GetContainer()");
|
|
return nullptr;
|
|
}
|
|
|
|
uint32_t flags = imgIContainer::FLAG_ASYNC_NOTIFY;
|
|
if (aBuilder->ShouldSyncDecodeImages()) {
|
|
flags |= imgIContainer::FLAG_SYNC_DECODE;
|
|
}
|
|
|
|
RefPtr<ImageContainer> container = image->GetImageContainer(aManager, flags);
|
|
if (!container || !container->HasCurrentImage()) {
|
|
return nullptr;
|
|
}
|
|
|
|
return container.forget();
|
|
}
|
|
|
|
bool nsDisplayImageContainer::CanOptimizeToImageLayer(
|
|
LayerManager* aManager, nsDisplayListBuilder* aBuilder) {
|
|
uint32_t flags = aBuilder->ShouldSyncDecodeImages()
|
|
? imgIContainer::FLAG_SYNC_DECODE
|
|
: imgIContainer::FLAG_NONE;
|
|
|
|
nsCOMPtr<imgIContainer> image = GetImage();
|
|
if (!image) {
|
|
return false;
|
|
}
|
|
|
|
if (!image->IsImageContainerAvailable(aManager, flags)) {
|
|
return false;
|
|
}
|
|
|
|
int32_t imageWidth;
|
|
int32_t imageHeight;
|
|
image->GetWidth(&imageWidth);
|
|
image->GetHeight(&imageHeight);
|
|
|
|
if (imageWidth == 0 || imageHeight == 0) {
|
|
NS_ASSERTION(false, "invalid image size");
|
|
return false;
|
|
}
|
|
|
|
const int32_t factor = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
const LayoutDeviceRect destRect(
|
|
LayoutDeviceIntRect::FromAppUnitsToNearest(GetDestRect(), factor));
|
|
|
|
// Calculate the scaling factor for the frame.
|
|
const gfxSize scale =
|
|
gfxSize(destRect.width / imageWidth, destRect.height / imageHeight);
|
|
|
|
if (scale.width < 0.34 || scale.height < 0.34) {
|
|
// This would look awful as long as we can't use high-quality downscaling
|
|
// for image layers (bug 803703), so don't turn this into an image layer.
|
|
return false;
|
|
}
|
|
|
|
if (mFrame->IsImageFrame() || mFrame->IsImageControlFrame()) {
|
|
// Image layer doesn't support draw focus ring for image map.
|
|
nsImageFrame* f = static_cast<nsImageFrame*>(mFrame);
|
|
if (f->HasImageMap()) {
|
|
return false;
|
|
}
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayBackgroundColor::ApplyOpacity(nsDisplayListBuilder* aBuilder,
|
|
float aOpacity,
|
|
const DisplayItemClipChain* aClip) {
|
|
NS_ASSERTION(CanApplyOpacity(), "ApplyOpacity should be allowed");
|
|
mColor.a = mColor.a * aOpacity;
|
|
IntersectClip(aBuilder, aClip, false);
|
|
}
|
|
|
|
bool nsDisplayBackgroundColor::CanApplyOpacity() const {
|
|
// Don't apply opacity if the background color is animated since the color is
|
|
// going to be changed on the compositor.
|
|
return !EffectCompositor::HasAnimationsForCompositor(
|
|
mFrame, DisplayItemType::TYPE_BACKGROUND_COLOR);
|
|
}
|
|
|
|
LayerState nsDisplayBackgroundColor::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
if (ForceActiveLayers() && !HasBackgroundClipText()) {
|
|
return LayerState::LAYER_ACTIVE;
|
|
}
|
|
|
|
if (EffectCompositor::HasAnimationsForCompositor(
|
|
mFrame, DisplayItemType::TYPE_BACKGROUND_COLOR)) {
|
|
return LayerState::LAYER_ACTIVE_FORCE;
|
|
}
|
|
|
|
return LayerState::LAYER_NONE;
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayBackgroundColor::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
if (mColor == Color()) {
|
|
return nullptr;
|
|
}
|
|
|
|
RefPtr<ColorLayer> layer = static_cast<ColorLayer*>(
|
|
aManager->GetLayerBuilder()->GetLeafLayerFor(aBuilder, this));
|
|
if (!layer) {
|
|
layer = aManager->CreateColorLayer();
|
|
if (!layer) {
|
|
return nullptr;
|
|
}
|
|
}
|
|
layer->SetColor(mColor);
|
|
|
|
int32_t appUnitsPerDevPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
layer->SetBounds(mBackgroundRect.ToNearestPixels(appUnitsPerDevPixel));
|
|
layer->SetBaseTransform(gfx::Matrix4x4::Translation(
|
|
aContainerParameters.mOffset.x, aContainerParameters.mOffset.y, 0));
|
|
|
|
// Both nsDisplayBackgroundColor and nsDisplayTableBackgroundColor use this
|
|
// function, but only nsDisplayBackgroundColor supports compositor animations.
|
|
if (GetType() == DisplayItemType::TYPE_BACKGROUND_COLOR) {
|
|
nsDisplayListBuilder::AddAnimationsAndTransitionsToLayer(
|
|
layer, aBuilder, this, mFrame, GetType());
|
|
}
|
|
return layer.forget();
|
|
}
|
|
|
|
bool nsDisplayBackgroundColor::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
if (mColor == Color()) {
|
|
return true;
|
|
}
|
|
|
|
if (HasBackgroundClipText()) {
|
|
return false;
|
|
}
|
|
|
|
LayoutDeviceRect bounds = LayoutDeviceRect::FromAppUnits(
|
|
mBackgroundRect, mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
wr::LayoutRect r = wr::ToLayoutRect(bounds);
|
|
|
|
aBuilder.PushRect(r, r, !BackfaceIsHidden(),
|
|
wr::ToColorF(ToDeviceColor(mColor)));
|
|
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayBackgroundColor::PaintWithClip(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx,
|
|
const DisplayItemClip& aClip) {
|
|
MOZ_ASSERT(!HasBackgroundClipText());
|
|
|
|
if (mColor == Color()) {
|
|
return;
|
|
}
|
|
|
|
nsRect fillRect = mBackgroundRect;
|
|
if (aClip.HasClip()) {
|
|
fillRect.IntersectRect(fillRect, aClip.GetClipRect());
|
|
}
|
|
|
|
DrawTarget* dt = aCtx->GetDrawTarget();
|
|
int32_t A2D = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
Rect bounds = ToRect(nsLayoutUtils::RectToGfxRect(fillRect, A2D));
|
|
MaybeSnapToDevicePixels(bounds, *dt);
|
|
ColorPattern fill(ToDeviceColor(mColor));
|
|
|
|
if (aClip.GetRoundedRectCount()) {
|
|
MOZ_ASSERT(aClip.GetRoundedRectCount() == 1);
|
|
|
|
AutoTArray<DisplayItemClip::RoundedRect, 1> roundedRect;
|
|
aClip.AppendRoundedRects(&roundedRect);
|
|
|
|
bool pushedClip = false;
|
|
if (!fillRect.Contains(roundedRect[0].mRect)) {
|
|
dt->PushClipRect(bounds);
|
|
pushedClip = true;
|
|
}
|
|
|
|
RectCornerRadii pixelRadii;
|
|
nsCSSRendering::ComputePixelRadii(roundedRect[0].mRadii, A2D, &pixelRadii);
|
|
dt->FillRoundedRect(
|
|
RoundedRect(NSRectToSnappedRect(roundedRect[0].mRect, A2D, *dt),
|
|
pixelRadii),
|
|
fill);
|
|
if (pushedClip) {
|
|
dt->PopClip();
|
|
}
|
|
} else {
|
|
dt->FillRect(bounds, fill);
|
|
}
|
|
}
|
|
|
|
void nsDisplayBackgroundColor::Paint(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx) {
|
|
if (mColor == Color()) {
|
|
return;
|
|
}
|
|
|
|
#if 0
|
|
// See https://bugzilla.mozilla.org/show_bug.cgi?id=1148418#c21 for why this
|
|
// results in a precision induced rounding issue that makes the rect one
|
|
// pixel shorter in rare cases. Disabled in favor of the old code for now.
|
|
// Note that the pref layout.css.devPixelsPerPx needs to be set to 1 to
|
|
// reproduce the bug.
|
|
//
|
|
// TODO:
|
|
// This new path does not include support for background-clip:text; need to
|
|
// be fixed if/when we switch to this new code path.
|
|
|
|
DrawTarget& aDrawTarget = *aCtx->GetDrawTarget();
|
|
|
|
Rect rect = NSRectToSnappedRect(mBackgroundRect,
|
|
mFrame->PresContext()->AppUnitsPerDevPixel(),
|
|
aDrawTarget);
|
|
ColorPattern color(ToDeviceColor(mColor));
|
|
aDrawTarget.FillRect(rect, color);
|
|
#else
|
|
gfxContext* ctx = aCtx;
|
|
gfxRect bounds = nsLayoutUtils::RectToGfxRect(
|
|
mBackgroundRect, mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
|
|
if (HasBackgroundClipText()) {
|
|
if (!GenerateAndPushTextMask(mFrame, aCtx, mBackgroundRect, aBuilder)) {
|
|
return;
|
|
}
|
|
|
|
ctx->SetColor(mColor);
|
|
ctx->NewPath();
|
|
ctx->SnappedRectangle(bounds);
|
|
ctx->Fill();
|
|
ctx->PopGroupAndBlend();
|
|
return;
|
|
}
|
|
|
|
ctx->SetColor(mColor);
|
|
ctx->NewPath();
|
|
ctx->SnappedRectangle(bounds);
|
|
ctx->Fill();
|
|
#endif
|
|
}
|
|
|
|
nsRegion nsDisplayBackgroundColor::GetOpaqueRegion(
|
|
nsDisplayListBuilder* aBuilder, bool* aSnap) const {
|
|
*aSnap = false;
|
|
|
|
if (mColor.a != 1 ||
|
|
// Even if the current alpha channel is 1, we treat this item as if it's
|
|
// non-opaque if there is a background-color animation since the animation
|
|
// might change the alpha channel.
|
|
EffectCompositor::HasAnimationsForCompositor(
|
|
mFrame, DisplayItemType::TYPE_BACKGROUND_COLOR)) {
|
|
return nsRegion();
|
|
}
|
|
|
|
if (!mHasStyle || HasBackgroundClipText()) {
|
|
return nsRegion();
|
|
}
|
|
|
|
*aSnap = true;
|
|
return nsDisplayBackgroundImage::GetInsideClipRegion(
|
|
this, mBottomLayerClip, mBackgroundRect, mBackgroundRect);
|
|
}
|
|
|
|
Maybe<nscolor> nsDisplayBackgroundColor::IsUniform(
|
|
nsDisplayListBuilder* aBuilder) const {
|
|
return Some(mColor.ToABGR());
|
|
}
|
|
|
|
void nsDisplayBackgroundColor::HitTest(nsDisplayListBuilder* aBuilder,
|
|
const nsRect& aRect,
|
|
HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) {
|
|
if (!RoundedBorderIntersectsRect(mFrame, ToReferenceFrame(), aRect)) {
|
|
// aRect doesn't intersect our border-radius curve.
|
|
return;
|
|
}
|
|
|
|
aOutFrames->AppendElement(mFrame);
|
|
}
|
|
|
|
void nsDisplayBackgroundColor::WriteDebugInfo(std::stringstream& aStream) {
|
|
aStream << " (rgba " << mColor.r << "," << mColor.g << "," << mColor.b << ","
|
|
<< mColor.a << ")";
|
|
aStream << " backgroundRect" << mBackgroundRect;
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayClearBackground::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
RefPtr<ColorLayer> layer = static_cast<ColorLayer*>(
|
|
aManager->GetLayerBuilder()->GetLeafLayerFor(aBuilder, this));
|
|
if (!layer) {
|
|
layer = aManager->CreateColorLayer();
|
|
if (!layer) {
|
|
return nullptr;
|
|
}
|
|
}
|
|
layer->SetColor(Color());
|
|
layer->SetMixBlendMode(gfx::CompositionOp::OP_SOURCE);
|
|
|
|
bool snap;
|
|
nsRect bounds = GetBounds(aBuilder, &snap);
|
|
int32_t appUnitsPerDevPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
layer->SetBounds(bounds.ToNearestPixels(appUnitsPerDevPixel)); // XXX Do we
|
|
// need to
|
|
// respect the
|
|
// parent
|
|
// layer's
|
|
// scale here?
|
|
|
|
return layer.forget();
|
|
}
|
|
|
|
bool nsDisplayClearBackground::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
LayoutDeviceRect bounds = LayoutDeviceRect::FromAppUnits(
|
|
nsRect(ToReferenceFrame(), mFrame->GetSize()),
|
|
mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
|
|
aBuilder.PushClearRect(wr::ToLayoutRect(bounds));
|
|
|
|
return true;
|
|
}
|
|
|
|
nsRect nsDisplayOutline::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = false;
|
|
return mFrame->GetVisualOverflowRectRelativeToSelf() + ToReferenceFrame();
|
|
}
|
|
|
|
void nsDisplayOutline::Paint(nsDisplayListBuilder* aBuilder, gfxContext* aCtx) {
|
|
// TODO join outlines together
|
|
MOZ_ASSERT(mFrame->StyleOutline()->ShouldPaintOutline(),
|
|
"Should have not created a nsDisplayOutline!");
|
|
|
|
nsPoint offset = ToReferenceFrame();
|
|
nsCSSRendering::PaintOutline(
|
|
mFrame->PresContext(), *aCtx, mFrame, GetPaintRect(),
|
|
nsRect(offset, mFrame->GetSize()), mFrame->Style());
|
|
}
|
|
|
|
bool nsDisplayOutline::IsThemedOutline() const {
|
|
const auto& outlineStyle = mFrame->StyleOutline()->mOutlineStyle;
|
|
if (!outlineStyle.IsAuto() ||
|
|
!StaticPrefs::layout_css_outline_style_auto_enabled()) {
|
|
return false;
|
|
}
|
|
|
|
nsPresContext* pc = mFrame->PresContext();
|
|
nsITheme* theme = pc->GetTheme();
|
|
return theme &&
|
|
theme->ThemeSupportsWidget(pc, mFrame, StyleAppearance::FocusOutline);
|
|
}
|
|
|
|
bool nsDisplayOutline::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
if (IsThemedOutline()) {
|
|
return false;
|
|
}
|
|
|
|
nsPoint offset = ToReferenceFrame();
|
|
|
|
mozilla::Maybe<nsCSSBorderRenderer> borderRenderer =
|
|
nsCSSRendering::CreateBorderRendererForOutline(
|
|
mFrame->PresContext(), nullptr, mFrame, GetPaintRect(),
|
|
nsRect(offset, mFrame->GetSize()), mFrame->Style());
|
|
|
|
if (!borderRenderer) {
|
|
// No border renderer means "there is no outline".
|
|
// Paint nothing and return success.
|
|
return true;
|
|
}
|
|
|
|
borderRenderer->CreateWebRenderCommands(this, aBuilder, aResources, aSc);
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayOutline::IsInvisibleInRect(const nsRect& aRect) const {
|
|
const nsStyleOutline* outline = mFrame->StyleOutline();
|
|
nsRect borderBox(ToReferenceFrame(), mFrame->GetSize());
|
|
if (borderBox.Contains(aRect) &&
|
|
!nsLayoutUtils::HasNonZeroCorner(outline->mOutlineRadius)) {
|
|
if (outline->mOutlineOffset._0 >= 0.0f) {
|
|
// aRect is entirely inside the border-rect, and the outline isn't
|
|
// rendered inside the border-rect, so the outline is not visible.
|
|
return true;
|
|
}
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
void nsDisplayEventReceiver::HitTest(nsDisplayListBuilder* aBuilder,
|
|
const nsRect& aRect, HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) {
|
|
if (!RoundedBorderIntersectsRect(mFrame, ToReferenceFrame(), aRect)) {
|
|
// aRect doesn't intersect our border-radius curve.
|
|
return;
|
|
}
|
|
|
|
aOutFrames->AppendElement(mFrame);
|
|
}
|
|
|
|
nsDisplayCompositorHitTestInfo::nsDisplayCompositorHitTestInfo(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame,
|
|
const mozilla::gfx::CompositorHitTestInfo& aHitTestFlags, uint16_t aIndex,
|
|
const mozilla::Maybe<nsRect>& aArea)
|
|
: nsDisplayHitTestInfoBase(aBuilder, aFrame),
|
|
mIndex(aIndex),
|
|
mAppUnitsPerDevPixel(mFrame->PresContext()->AppUnitsPerDevPixel()) {
|
|
MOZ_COUNT_CTOR(nsDisplayCompositorHitTestInfo);
|
|
// We should never even create this display item if we're not building
|
|
// compositor hit-test info or if the computed hit info indicated the
|
|
// frame is invisible to hit-testing
|
|
MOZ_ASSERT(aBuilder->BuildCompositorHitTestInfo());
|
|
MOZ_ASSERT(aHitTestFlags != CompositorHitTestInvisibleToHit);
|
|
|
|
const nsRect& area =
|
|
aArea.isSome() ? *aArea : aFrame->GetCompositorHitTestArea(aBuilder);
|
|
|
|
SetHitTestInfo(area, aHitTestFlags);
|
|
InitializeScrollTarget(aBuilder);
|
|
}
|
|
|
|
nsDisplayCompositorHitTestInfo::nsDisplayCompositorHitTestInfo(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame,
|
|
mozilla::UniquePtr<HitTestInfo>&& aHitTestInfo)
|
|
: nsDisplayHitTestInfoBase(aBuilder, aFrame),
|
|
mIndex(0),
|
|
mAppUnitsPerDevPixel(mFrame->PresContext()->AppUnitsPerDevPixel()) {
|
|
MOZ_COUNT_CTOR(nsDisplayCompositorHitTestInfo);
|
|
SetHitTestInfo(std::move(aHitTestInfo));
|
|
InitializeScrollTarget(aBuilder);
|
|
}
|
|
|
|
void nsDisplayCompositorHitTestInfo::InitializeScrollTarget(
|
|
nsDisplayListBuilder* aBuilder) {
|
|
if (aBuilder->GetCurrentScrollbarDirection().isSome()) {
|
|
// In the case of scrollbar frames, we use the scrollbar's target
|
|
// scrollframe instead of the scrollframe with which the scrollbar actually
|
|
// moves.
|
|
MOZ_ASSERT(HitTestFlags().contains(CompositorHitTestFlags::eScrollbar));
|
|
mScrollTarget = mozilla::Some(aBuilder->GetCurrentScrollbarTarget());
|
|
}
|
|
}
|
|
|
|
bool nsDisplayCompositorHitTestInfo::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
if (HitTestArea().IsEmpty()) {
|
|
return true;
|
|
}
|
|
|
|
// XXX: eventually this scrollId computation and the SetHitTestInfo
|
|
// call will get moved out into the WR display item iteration code so that
|
|
// we don't need to do it as often, and so that we can do it for other
|
|
// display item types as well (reducing the need for as many instances of
|
|
// this display item).
|
|
ScrollableLayerGuid::ViewID scrollId =
|
|
mScrollTarget.valueOrFrom([&]() -> ScrollableLayerGuid::ViewID {
|
|
const ActiveScrolledRoot* asr = GetActiveScrolledRoot();
|
|
Maybe<ScrollableLayerGuid::ViewID> fixedTarget =
|
|
aBuilder.GetContainingFixedPosScrollTarget(asr);
|
|
if (fixedTarget) {
|
|
return *fixedTarget;
|
|
}
|
|
if (asr) {
|
|
return asr->GetViewId();
|
|
}
|
|
return ScrollableLayerGuid::NULL_SCROLL_ID;
|
|
});
|
|
|
|
Maybe<SideBits> sideBits =
|
|
aBuilder.GetContainingFixedPosSideBits(GetActiveScrolledRoot());
|
|
|
|
// Insert a transparent rectangle with the hit-test info
|
|
aBuilder.SetHitTestInfo(scrollId, HitTestFlags(),
|
|
sideBits.valueOr(SideBits::eNone));
|
|
|
|
const LayoutDeviceRect devRect =
|
|
LayoutDeviceRect::FromAppUnits(HitTestArea(), mAppUnitsPerDevPixel);
|
|
|
|
const wr::LayoutRect rect = wr::ToLayoutRect(devRect);
|
|
|
|
aBuilder.PushHitTest(rect, rect, !BackfaceIsHidden());
|
|
aBuilder.ClearHitTestInfo();
|
|
|
|
return true;
|
|
}
|
|
|
|
uint16_t nsDisplayCompositorHitTestInfo::CalculatePerFrameKey() const {
|
|
return mIndex;
|
|
}
|
|
|
|
int32_t nsDisplayCompositorHitTestInfo::ZIndex() const {
|
|
return mOverrideZIndex ? *mOverrideZIndex
|
|
: nsDisplayHitTestInfoBase::ZIndex();
|
|
}
|
|
|
|
void nsDisplayCompositorHitTestInfo::SetOverrideZIndex(int32_t aZIndex) {
|
|
mOverrideZIndex = Some(aZIndex);
|
|
}
|
|
|
|
nsDisplayCaret::nsDisplayCaret(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aCaretFrame)
|
|
: nsPaintedDisplayItem(aBuilder, aCaretFrame),
|
|
mCaret(aBuilder->GetCaret()),
|
|
mBounds(aBuilder->GetCaretRect() + ToReferenceFrame()) {
|
|
MOZ_COUNT_CTOR(nsDisplayCaret);
|
|
}
|
|
|
|
#ifdef NS_BUILD_REFCNT_LOGGING
|
|
nsDisplayCaret::~nsDisplayCaret() { MOZ_COUNT_DTOR(nsDisplayCaret); }
|
|
#endif
|
|
|
|
nsRect nsDisplayCaret::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = true;
|
|
// The caret returns a rect in the coordinates of mFrame.
|
|
return mBounds;
|
|
}
|
|
|
|
void nsDisplayCaret::Paint(nsDisplayListBuilder* aBuilder, gfxContext* aCtx) {
|
|
// Note: Because we exist, we know that the caret is visible, so we don't
|
|
// need to check for the caret's visibility.
|
|
mCaret->PaintCaret(*aCtx->GetDrawTarget(), mFrame, ToReferenceFrame());
|
|
}
|
|
|
|
bool nsDisplayCaret::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
using namespace mozilla::layers;
|
|
int32_t contentOffset;
|
|
nsIFrame* frame = mCaret->GetFrame(&contentOffset);
|
|
if (!frame) {
|
|
return true;
|
|
}
|
|
NS_ASSERTION(frame == mFrame, "We're referring different frame");
|
|
|
|
int32_t appUnitsPerDevPixel = frame->PresContext()->AppUnitsPerDevPixel();
|
|
|
|
nsRect caretRect;
|
|
nsRect hookRect;
|
|
mCaret->ComputeCaretRects(frame, contentOffset, &caretRect, &hookRect);
|
|
|
|
gfx::Color color = ToDeviceColor(frame->GetCaretColorAt(contentOffset));
|
|
LayoutDeviceRect devCaretRect = LayoutDeviceRect::FromAppUnits(
|
|
caretRect + ToReferenceFrame(), appUnitsPerDevPixel);
|
|
LayoutDeviceRect devHookRect = LayoutDeviceRect::FromAppUnits(
|
|
hookRect + ToReferenceFrame(), appUnitsPerDevPixel);
|
|
|
|
wr::LayoutRect caret = wr::ToLayoutRect(devCaretRect);
|
|
wr::LayoutRect hook = wr::ToLayoutRect(devHookRect);
|
|
|
|
// Note, WR will pixel snap anything that is layout aligned.
|
|
aBuilder.PushRect(caret, caret, !BackfaceIsHidden(), wr::ToColorF(color));
|
|
|
|
if (!devHookRect.IsEmpty()) {
|
|
aBuilder.PushRect(hook, hook, !BackfaceIsHidden(), wr::ToColorF(color));
|
|
}
|
|
return true;
|
|
}
|
|
|
|
nsDisplayBorder::nsDisplayBorder(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame)
|
|
: nsPaintedDisplayItem(aBuilder, aFrame) {
|
|
MOZ_COUNT_CTOR(nsDisplayBorder);
|
|
|
|
mBounds = CalculateBounds<nsRect>(*mFrame->StyleBorder());
|
|
}
|
|
|
|
bool nsDisplayBorder::IsInvisibleInRect(const nsRect& aRect) const {
|
|
nsRect paddingRect = GetPaddingRect();
|
|
const nsStyleBorder* styleBorder;
|
|
if (paddingRect.Contains(aRect) &&
|
|
!(styleBorder = mFrame->StyleBorder())->IsBorderImageSizeAvailable() &&
|
|
!nsLayoutUtils::HasNonZeroCorner(styleBorder->mBorderRadius)) {
|
|
// aRect is entirely inside the content rect, and no part
|
|
// of the border is rendered inside the content rect, so we are not
|
|
// visible
|
|
// Skip this if there's a border-image (which draws a background
|
|
// too) or if there is a border-radius (which makes the border draw
|
|
// further in).
|
|
return true;
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
nsDisplayItemGeometry* nsDisplayBorder::AllocateGeometry(
|
|
nsDisplayListBuilder* aBuilder) {
|
|
return new nsDisplayBorderGeometry(this, aBuilder);
|
|
}
|
|
|
|
void nsDisplayBorder::ComputeInvalidationRegion(
|
|
nsDisplayListBuilder* aBuilder, const nsDisplayItemGeometry* aGeometry,
|
|
nsRegion* aInvalidRegion) const {
|
|
auto* geometry = static_cast<const nsDisplayBorderGeometry*>(aGeometry);
|
|
bool snap;
|
|
|
|
if (!geometry->mBounds.IsEqualInterior(GetBounds(aBuilder, &snap))) {
|
|
// We can probably get away with only invalidating the difference
|
|
// between the border and padding rects, but the XUL ui at least
|
|
// is apparently painting a background with this?
|
|
aInvalidRegion->Or(GetBounds(aBuilder, &snap), geometry->mBounds);
|
|
}
|
|
|
|
if (aBuilder->ShouldSyncDecodeImages() &&
|
|
geometry->ShouldInvalidateToSyncDecodeImages()) {
|
|
aInvalidRegion->Or(*aInvalidRegion, GetBounds(aBuilder, &snap));
|
|
}
|
|
}
|
|
|
|
LayerState nsDisplayBorder::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
return LayerState::LAYER_NONE;
|
|
}
|
|
|
|
bool nsDisplayBorder::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
nsRect rect = nsRect(ToReferenceFrame(), mFrame->GetSize());
|
|
|
|
ImgDrawResult drawResult = nsCSSRendering::CreateWebRenderCommandsForBorder(
|
|
this, mFrame, rect, aBuilder, aResources, aSc, aManager,
|
|
aDisplayListBuilder);
|
|
|
|
if (drawResult == ImgDrawResult::NOT_SUPPORTED) {
|
|
return false;
|
|
}
|
|
|
|
nsDisplayBorderGeometry::UpdateDrawResult(this, drawResult);
|
|
return true;
|
|
};
|
|
|
|
void nsDisplayBorder::Paint(nsDisplayListBuilder* aBuilder, gfxContext* aCtx) {
|
|
nsPoint offset = ToReferenceFrame();
|
|
|
|
PaintBorderFlags flags = aBuilder->ShouldSyncDecodeImages()
|
|
? PaintBorderFlags::SyncDecodeImages
|
|
: PaintBorderFlags();
|
|
|
|
ImgDrawResult result = nsCSSRendering::PaintBorder(
|
|
mFrame->PresContext(), *aCtx, mFrame, GetPaintRect(),
|
|
nsRect(offset, mFrame->GetSize()), mFrame->Style(), flags,
|
|
mFrame->GetSkipSides());
|
|
|
|
nsDisplayBorderGeometry::UpdateDrawResult(this, result);
|
|
}
|
|
|
|
nsRect nsDisplayBorder::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = true;
|
|
return mBounds;
|
|
}
|
|
|
|
// Given a region, compute a conservative approximation to it as a list
|
|
// of rectangles that aren't vertically adjacent (i.e., vertically
|
|
// adjacent or overlapping rectangles are combined).
|
|
// Right now this is only approximate, some vertically overlapping rectangles
|
|
// aren't guaranteed to be combined.
|
|
static void ComputeDisjointRectangles(const nsRegion& aRegion,
|
|
nsTArray<nsRect>* aRects) {
|
|
nscoord accumulationMargin = nsPresContext::CSSPixelsToAppUnits(25);
|
|
nsRect accumulated;
|
|
|
|
for (auto iter = aRegion.RectIter(); !iter.Done(); iter.Next()) {
|
|
const nsRect& r = iter.Get();
|
|
if (accumulated.IsEmpty()) {
|
|
accumulated = r;
|
|
continue;
|
|
}
|
|
|
|
if (accumulated.YMost() >= r.y - accumulationMargin) {
|
|
accumulated.UnionRect(accumulated, r);
|
|
} else {
|
|
aRects->AppendElement(accumulated);
|
|
accumulated = r;
|
|
}
|
|
}
|
|
|
|
// Finish the in-flight rectangle, if there is one.
|
|
if (!accumulated.IsEmpty()) {
|
|
aRects->AppendElement(accumulated);
|
|
}
|
|
}
|
|
|
|
void nsDisplayBoxShadowOuter::Paint(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx) {
|
|
nsPoint offset = ToReferenceFrame();
|
|
nsRect borderRect = mFrame->VisualBorderRectRelativeToSelf() + offset;
|
|
nsPresContext* presContext = mFrame->PresContext();
|
|
AutoTArray<nsRect, 10> rects;
|
|
ComputeDisjointRectangles(mVisibleRegion, &rects);
|
|
|
|
AUTO_PROFILER_LABEL("nsDisplayBoxShadowOuter::Paint", GRAPHICS);
|
|
|
|
for (uint32_t i = 0; i < rects.Length(); ++i) {
|
|
nsCSSRendering::PaintBoxShadowOuter(presContext, *aCtx, mFrame, borderRect,
|
|
rects[i], mOpacity);
|
|
}
|
|
}
|
|
|
|
nsRect nsDisplayBoxShadowOuter::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = false;
|
|
return mBounds;
|
|
}
|
|
|
|
nsRect nsDisplayBoxShadowOuter::GetBoundsInternal() {
|
|
return nsLayoutUtils::GetBoxShadowRectForFrame(mFrame, mFrame->GetSize()) +
|
|
ToReferenceFrame();
|
|
}
|
|
|
|
bool nsDisplayBoxShadowOuter::IsInvisibleInRect(const nsRect& aRect) const {
|
|
nsPoint origin = ToReferenceFrame();
|
|
nsRect frameRect(origin, mFrame->GetSize());
|
|
if (!frameRect.Contains(aRect)) {
|
|
return false;
|
|
}
|
|
|
|
// the visible region is entirely inside the border-rect, and box shadows
|
|
// never render within the border-rect (unless there's a border radius).
|
|
nscoord twipsRadii[8];
|
|
bool hasBorderRadii = mFrame->GetBorderRadii(twipsRadii);
|
|
if (!hasBorderRadii) {
|
|
return true;
|
|
}
|
|
|
|
return RoundedRectContainsRect(frameRect, twipsRadii, aRect);
|
|
}
|
|
|
|
bool nsDisplayBoxShadowOuter::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
if (!nsPaintedDisplayItem::ComputeVisibility(aBuilder, aVisibleRegion)) {
|
|
return false;
|
|
}
|
|
|
|
mVisibleRegion.And(*aVisibleRegion, GetPaintRect());
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayBoxShadowOuter::CanBuildWebRenderDisplayItems() {
|
|
auto shadows = mFrame->StyleEffects()->mBoxShadow.AsSpan();
|
|
if (shadows.IsEmpty()) {
|
|
return false;
|
|
}
|
|
|
|
bool hasBorderRadius;
|
|
bool nativeTheme =
|
|
nsCSSRendering::HasBoxShadowNativeTheme(mFrame, hasBorderRadius);
|
|
|
|
// We don't support native themed things yet like box shadows around
|
|
// input buttons.
|
|
if (nativeTheme) {
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayBoxShadowOuter::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
if (!CanBuildWebRenderDisplayItems()) {
|
|
return false;
|
|
}
|
|
|
|
int32_t appUnitsPerDevPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
nsPoint offset = ToReferenceFrame();
|
|
nsRect borderRect = mFrame->VisualBorderRectRelativeToSelf() + offset;
|
|
AutoTArray<nsRect, 10> rects;
|
|
bool snap;
|
|
nsRect bounds = GetBounds(aDisplayListBuilder, &snap);
|
|
ComputeDisjointRectangles(bounds, &rects);
|
|
|
|
bool hasBorderRadius;
|
|
bool nativeTheme =
|
|
nsCSSRendering::HasBoxShadowNativeTheme(mFrame, hasBorderRadius);
|
|
|
|
// Don't need the full size of the shadow rect like we do in
|
|
// nsCSSRendering since WR takes care of calculations for blur
|
|
// and spread radius.
|
|
nsRect frameRect =
|
|
nsCSSRendering::GetShadowRect(borderRect, nativeTheme, mFrame);
|
|
|
|
RectCornerRadii borderRadii;
|
|
if (hasBorderRadius) {
|
|
hasBorderRadius = nsCSSRendering::GetBorderRadii(frameRect, borderRect,
|
|
mFrame, borderRadii);
|
|
}
|
|
|
|
// Everything here is in app units, change to device units.
|
|
for (uint32_t i = 0; i < rects.Length(); ++i) {
|
|
LayoutDeviceRect clipRect =
|
|
LayoutDeviceRect::FromAppUnits(rects[i], appUnitsPerDevPixel);
|
|
auto shadows = mFrame->StyleEffects()->mBoxShadow.AsSpan();
|
|
MOZ_ASSERT(!shadows.IsEmpty());
|
|
|
|
for (auto& shadow : Reversed(shadows)) {
|
|
if (shadow.inset) {
|
|
continue;
|
|
}
|
|
|
|
float blurRadius =
|
|
float(shadow.base.blur.ToAppUnits()) / float(appUnitsPerDevPixel);
|
|
gfx::Color shadowColor =
|
|
nsCSSRendering::GetShadowColor(shadow.base, mFrame, mOpacity);
|
|
|
|
// We don't move the shadow rect here since WR does it for us
|
|
// Now translate everything to device pixels.
|
|
const nsRect& shadowRect = frameRect;
|
|
LayoutDevicePoint shadowOffset = LayoutDevicePoint::FromAppUnits(
|
|
nsPoint(shadow.base.horizontal.ToAppUnits(),
|
|
shadow.base.vertical.ToAppUnits()),
|
|
appUnitsPerDevPixel);
|
|
|
|
LayoutDeviceRect deviceBox =
|
|
LayoutDeviceRect::FromAppUnits(shadowRect, appUnitsPerDevPixel);
|
|
wr::LayoutRect deviceBoxRect = wr::ToLayoutRect(deviceBox);
|
|
wr::LayoutRect deviceClipRect = wr::ToLayoutRect(clipRect);
|
|
|
|
LayoutDeviceSize zeroSize;
|
|
wr::BorderRadius borderRadius =
|
|
wr::ToBorderRadius(zeroSize, zeroSize, zeroSize, zeroSize);
|
|
if (hasBorderRadius) {
|
|
borderRadius = wr::ToBorderRadius(
|
|
LayoutDeviceSize::FromUnknownSize(borderRadii.TopLeft()),
|
|
LayoutDeviceSize::FromUnknownSize(borderRadii.TopRight()),
|
|
LayoutDeviceSize::FromUnknownSize(borderRadii.BottomLeft()),
|
|
LayoutDeviceSize::FromUnknownSize(borderRadii.BottomRight()));
|
|
}
|
|
|
|
float spreadRadius =
|
|
float(shadow.spread.ToAppUnits()) / float(appUnitsPerDevPixel);
|
|
|
|
aBuilder.PushBoxShadow(deviceBoxRect, deviceClipRect, !BackfaceIsHidden(),
|
|
deviceBoxRect, wr::ToLayoutVector2D(shadowOffset),
|
|
wr::ToColorF(shadowColor), blurRadius,
|
|
spreadRadius, borderRadius,
|
|
wr::BoxShadowClipMode::Outset);
|
|
}
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayBoxShadowOuter::ComputeInvalidationRegion(
|
|
nsDisplayListBuilder* aBuilder, const nsDisplayItemGeometry* aGeometry,
|
|
nsRegion* aInvalidRegion) const {
|
|
auto* geometry =
|
|
static_cast<const nsDisplayBoxShadowOuterGeometry*>(aGeometry);
|
|
bool snap;
|
|
if (!geometry->mBounds.IsEqualInterior(GetBounds(aBuilder, &snap)) ||
|
|
!geometry->mBorderRect.IsEqualInterior(GetBorderRect()) ||
|
|
mOpacity != geometry->mOpacity) {
|
|
nsRegion oldShadow, newShadow;
|
|
nscoord dontCare[8];
|
|
bool hasBorderRadius = mFrame->GetBorderRadii(dontCare);
|
|
if (hasBorderRadius) {
|
|
// If we have rounded corners then we need to invalidate the frame area
|
|
// too since we paint into it.
|
|
oldShadow = geometry->mBounds;
|
|
newShadow = GetBounds(aBuilder, &snap);
|
|
} else {
|
|
oldShadow.Sub(geometry->mBounds, geometry->mBorderRect);
|
|
newShadow.Sub(GetBounds(aBuilder, &snap), GetBorderRect());
|
|
}
|
|
aInvalidRegion->Or(oldShadow, newShadow);
|
|
}
|
|
}
|
|
|
|
void nsDisplayBoxShadowInner::Paint(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx) {
|
|
nsPoint offset = ToReferenceFrame();
|
|
nsRect borderRect = nsRect(offset, mFrame->GetSize());
|
|
nsPresContext* presContext = mFrame->PresContext();
|
|
AutoTArray<nsRect, 10> rects;
|
|
ComputeDisjointRectangles(mVisibleRegion, &rects);
|
|
|
|
AUTO_PROFILER_LABEL("nsDisplayBoxShadowInner::Paint", GRAPHICS);
|
|
|
|
DrawTarget* drawTarget = aCtx->GetDrawTarget();
|
|
gfxContext* gfx = aCtx;
|
|
int32_t appUnitsPerDevPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
|
|
for (uint32_t i = 0; i < rects.Length(); ++i) {
|
|
gfx->Save();
|
|
gfx->Clip(NSRectToSnappedRect(rects[i], appUnitsPerDevPixel, *drawTarget));
|
|
nsCSSRendering::PaintBoxShadowInner(presContext, *aCtx, mFrame, borderRect);
|
|
gfx->Restore();
|
|
}
|
|
}
|
|
|
|
bool nsDisplayBoxShadowInner::CanCreateWebRenderCommands(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame,
|
|
const nsPoint& aReferenceOffset) {
|
|
auto shadows = aFrame->StyleEffects()->mBoxShadow.AsSpan();
|
|
if (shadows.IsEmpty()) {
|
|
// Means we don't have to paint anything
|
|
return true;
|
|
}
|
|
|
|
bool hasBorderRadius;
|
|
bool nativeTheme =
|
|
nsCSSRendering::HasBoxShadowNativeTheme(aFrame, hasBorderRadius);
|
|
|
|
// We don't support native themed things yet like box shadows around
|
|
// input buttons.
|
|
if (nativeTheme) {
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
/* static */
|
|
void nsDisplayBoxShadowInner::CreateInsetBoxShadowWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder, const StackingContextHelper& aSc,
|
|
nsRegion& aVisibleRegion, nsIFrame* aFrame, const nsRect& aBorderRect) {
|
|
if (!nsCSSRendering::ShouldPaintBoxShadowInner(aFrame)) {
|
|
return;
|
|
}
|
|
|
|
int32_t appUnitsPerDevPixel = aFrame->PresContext()->AppUnitsPerDevPixel();
|
|
|
|
AutoTArray<nsRect, 10> rects;
|
|
ComputeDisjointRectangles(aVisibleRegion, &rects);
|
|
|
|
auto shadows = aFrame->StyleEffects()->mBoxShadow.AsSpan();
|
|
|
|
for (uint32_t i = 0; i < rects.Length(); ++i) {
|
|
LayoutDeviceRect clipRect =
|
|
LayoutDeviceRect::FromAppUnits(rects[i], appUnitsPerDevPixel);
|
|
|
|
for (auto& shadow : Reversed(shadows)) {
|
|
if (!shadow.inset) {
|
|
continue;
|
|
}
|
|
|
|
nsRect shadowRect =
|
|
nsCSSRendering::GetBoxShadowInnerPaddingRect(aFrame, aBorderRect);
|
|
RectCornerRadii innerRadii;
|
|
nsCSSRendering::GetShadowInnerRadii(aFrame, aBorderRect, innerRadii);
|
|
|
|
// Now translate everything to device pixels.
|
|
LayoutDeviceRect deviceBoxRect =
|
|
LayoutDeviceRect::FromAppUnits(shadowRect, appUnitsPerDevPixel);
|
|
wr::LayoutRect deviceClipRect = wr::ToLayoutRect(clipRect);
|
|
Color shadowColor =
|
|
nsCSSRendering::GetShadowColor(shadow.base, aFrame, 1.0);
|
|
|
|
LayoutDevicePoint shadowOffset = LayoutDevicePoint::FromAppUnits(
|
|
nsPoint(shadow.base.horizontal.ToAppUnits(),
|
|
shadow.base.vertical.ToAppUnits()),
|
|
appUnitsPerDevPixel);
|
|
|
|
float blurRadius =
|
|
float(shadow.base.blur.ToAppUnits()) / float(appUnitsPerDevPixel);
|
|
|
|
wr::BorderRadius borderRadius = wr::ToBorderRadius(
|
|
LayoutDeviceSize::FromUnknownSize(innerRadii.TopLeft()),
|
|
LayoutDeviceSize::FromUnknownSize(innerRadii.TopRight()),
|
|
LayoutDeviceSize::FromUnknownSize(innerRadii.BottomLeft()),
|
|
LayoutDeviceSize::FromUnknownSize(innerRadii.BottomRight()));
|
|
// NOTE: Any spread radius > 0 will render nothing. WR Bug.
|
|
float spreadRadius =
|
|
float(shadow.spread.ToAppUnits()) / float(appUnitsPerDevPixel);
|
|
|
|
aBuilder.PushBoxShadow(
|
|
wr::ToLayoutRect(deviceBoxRect), deviceClipRect,
|
|
!aFrame->BackfaceIsHidden(), wr::ToLayoutRect(deviceBoxRect),
|
|
wr::ToLayoutVector2D(shadowOffset), wr::ToColorF(shadowColor),
|
|
blurRadius, spreadRadius, borderRadius, wr::BoxShadowClipMode::Inset);
|
|
}
|
|
}
|
|
}
|
|
|
|
bool nsDisplayBoxShadowInner::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
if (!CanCreateWebRenderCommands(aDisplayListBuilder, mFrame,
|
|
ToReferenceFrame())) {
|
|
return false;
|
|
}
|
|
|
|
bool snap;
|
|
nsRegion visible = GetBounds(aDisplayListBuilder, &snap);
|
|
nsPoint offset = ToReferenceFrame();
|
|
nsRect borderRect = nsRect(offset, mFrame->GetSize());
|
|
nsDisplayBoxShadowInner::CreateInsetBoxShadowWebRenderCommands(
|
|
aBuilder, aSc, visible, mFrame, borderRect);
|
|
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayBoxShadowInner::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
if (!nsPaintedDisplayItem::ComputeVisibility(aBuilder, aVisibleRegion)) {
|
|
return false;
|
|
}
|
|
|
|
mVisibleRegion.And(*aVisibleRegion, GetPaintRect());
|
|
return true;
|
|
}
|
|
|
|
nsDisplayWrapList::nsDisplayWrapList(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame, nsDisplayList* aList)
|
|
: nsDisplayWrapList(aBuilder, aFrame, aList,
|
|
aBuilder->CurrentActiveScrolledRoot(), false, 0) {}
|
|
|
|
nsDisplayWrapList::nsDisplayWrapList(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot, bool aClearClipChain,
|
|
uint16_t aIndex)
|
|
: nsDisplayHitTestInfoBase(aBuilder, aFrame, aActiveScrolledRoot),
|
|
mFrameActiveScrolledRoot(aBuilder->CurrentActiveScrolledRoot()),
|
|
mOverrideZIndex(0),
|
|
mIndex(aIndex),
|
|
mHasZIndexOverride(false),
|
|
mClearingClipChain(aClearClipChain) {
|
|
MOZ_COUNT_CTOR(nsDisplayWrapList);
|
|
|
|
mBaseBuildingRect = GetBuildingRect();
|
|
|
|
mListPtr = &mList;
|
|
mListPtr->AppendToTop(aList);
|
|
nsDisplayWrapList::UpdateBounds(aBuilder);
|
|
|
|
if (!aFrame || !aFrame->IsTransformed()) {
|
|
return;
|
|
}
|
|
|
|
// If we're a transformed frame, then we need to find out if we're inside
|
|
// the nsDisplayTransform or outside of it. Frames inside the transform
|
|
// need mReferenceFrame == mFrame, outside needs the next ancestor
|
|
// reference frame.
|
|
// If we're inside the transform, then the nsDisplayItem constructor
|
|
// will have done the right thing.
|
|
// If we're outside the transform, then we should have only one child
|
|
// (since nsDisplayTransform wraps all actual content), and that child
|
|
// will have the correct reference frame set (since nsDisplayTransform
|
|
// handles this explictly).
|
|
nsDisplayItem* i = mListPtr->GetBottom();
|
|
if (i &&
|
|
(!i->GetAbove() || i->GetType() == DisplayItemType::TYPE_TRANSFORM) &&
|
|
i->Frame() == mFrame) {
|
|
mReferenceFrame = i->ReferenceFrame();
|
|
mToReferenceFrame = i->ToReferenceFrame();
|
|
}
|
|
|
|
nsRect visible = aBuilder->GetVisibleRect() +
|
|
aBuilder->GetCurrentFrameOffsetToReferenceFrame();
|
|
|
|
SetBuildingRect(visible);
|
|
}
|
|
|
|
nsDisplayWrapList::nsDisplayWrapList(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame, nsDisplayItem* aItem)
|
|
: nsDisplayHitTestInfoBase(aBuilder, aFrame,
|
|
aBuilder->CurrentActiveScrolledRoot()),
|
|
mOverrideZIndex(0),
|
|
mIndex(0),
|
|
mHasZIndexOverride(false) {
|
|
MOZ_COUNT_CTOR(nsDisplayWrapList);
|
|
|
|
mBaseBuildingRect = GetBuildingRect();
|
|
|
|
mListPtr = &mList;
|
|
mListPtr->AppendToTop(aItem);
|
|
nsDisplayWrapList::UpdateBounds(aBuilder);
|
|
|
|
if (!aFrame || !aFrame->IsTransformed()) {
|
|
return;
|
|
}
|
|
|
|
// See the previous nsDisplayWrapList constructor
|
|
if (aItem->Frame() == aFrame) {
|
|
mReferenceFrame = aItem->ReferenceFrame();
|
|
mToReferenceFrame = aItem->ToReferenceFrame();
|
|
}
|
|
|
|
nsRect visible = aBuilder->GetVisibleRect() +
|
|
aBuilder->GetCurrentFrameOffsetToReferenceFrame();
|
|
|
|
SetBuildingRect(visible);
|
|
}
|
|
|
|
nsDisplayWrapList::~nsDisplayWrapList() { MOZ_COUNT_DTOR(nsDisplayWrapList); }
|
|
|
|
void nsDisplayWrapList::MergeDisplayListFromItem(
|
|
nsDisplayListBuilder* aBuilder, const nsDisplayWrapList* aItem) {
|
|
const nsDisplayWrapList* wrappedItem = aItem->AsDisplayWrapList();
|
|
MOZ_ASSERT(wrappedItem);
|
|
|
|
// Create a new nsDisplayWrapList using a copy-constructor. This is done
|
|
// to preserve the information about bounds.
|
|
nsDisplayWrapList* wrapper =
|
|
MakeClone<nsDisplayWrapList>(aBuilder, wrappedItem);
|
|
MOZ_ASSERT(wrapper);
|
|
|
|
// Set the display list pointer of the new wrapper item to the display list
|
|
// of the wrapped item.
|
|
wrapper->mListPtr = wrappedItem->mListPtr;
|
|
|
|
mListPtr->AppendToBottom(wrapper);
|
|
}
|
|
|
|
void nsDisplayWrapList::HitTest(nsDisplayListBuilder* aBuilder,
|
|
const nsRect& aRect, HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) {
|
|
mListPtr->HitTest(aBuilder, aRect, aState, aOutFrames);
|
|
}
|
|
|
|
nsRect nsDisplayWrapList::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = false;
|
|
return mBounds;
|
|
}
|
|
|
|
bool nsDisplayWrapList::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
return ::ComputeClippedVisibilityForSubList(aBuilder, aVisibleRegion,
|
|
GetChildren(), GetPaintRect());
|
|
}
|
|
|
|
nsRegion nsDisplayWrapList::GetOpaqueRegion(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = false;
|
|
bool snap;
|
|
return ::GetOpaqueRegion(aBuilder, GetChildren(), GetBounds(aBuilder, &snap));
|
|
}
|
|
|
|
Maybe<nscolor> nsDisplayWrapList::IsUniform(
|
|
nsDisplayListBuilder* aBuilder) const {
|
|
// We could try to do something but let's conservatively just return Nothing.
|
|
return Nothing();
|
|
}
|
|
|
|
void nsDisplayWrapList::Paint(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx) {
|
|
NS_ERROR("nsDisplayWrapList should have been flattened away for painting");
|
|
}
|
|
|
|
/**
|
|
* Returns true if all descendant display items can be placed in the same
|
|
* PaintedLayer --- GetLayerState returns LayerState::LAYER_INACTIVE or
|
|
* LayerState::LAYER_NONE, and they all have the expected animated geometry
|
|
* root.
|
|
*/
|
|
static LayerState RequiredLayerStateForChildren(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters, const nsDisplayList& aList,
|
|
AnimatedGeometryRoot* aExpectedAnimatedGeometryRootForChildren) {
|
|
LayerState result = LayerState::LAYER_INACTIVE;
|
|
for (nsDisplayItem* i : aList) {
|
|
if (result == LayerState::LAYER_INACTIVE &&
|
|
i->GetAnimatedGeometryRoot() !=
|
|
aExpectedAnimatedGeometryRootForChildren) {
|
|
result = LayerState::LAYER_ACTIVE;
|
|
}
|
|
|
|
LayerState state = i->GetLayerState(aBuilder, aManager, aParameters);
|
|
if (state == LayerState::LAYER_ACTIVE &&
|
|
(i->GetType() == DisplayItemType::TYPE_BLEND_MODE ||
|
|
i->GetType() == DisplayItemType::TYPE_TABLE_BLEND_MODE)) {
|
|
// nsDisplayBlendMode always returns LayerState::LAYER_ACTIVE to ensure
|
|
// that the blending operation happens in the intermediate surface of its
|
|
// parent display item (usually an nsDisplayBlendContainer). But this does
|
|
// not mean that it needs all its ancestor display items to become active.
|
|
// So we ignore its layer state and look at its children instead.
|
|
state = RequiredLayerStateForChildren(
|
|
aBuilder, aManager, aParameters,
|
|
*i->GetSameCoordinateSystemChildren(), i->GetAnimatedGeometryRoot());
|
|
}
|
|
if ((state == LayerState::LAYER_ACTIVE ||
|
|
state == LayerState::LAYER_ACTIVE_FORCE) &&
|
|
state > result) {
|
|
result = state;
|
|
}
|
|
if (state == LayerState::LAYER_ACTIVE_EMPTY && state > result) {
|
|
result = LayerState::LAYER_ACTIVE_FORCE;
|
|
}
|
|
if (state == LayerState::LAYER_NONE) {
|
|
nsDisplayList* list = i->GetSameCoordinateSystemChildren();
|
|
if (list) {
|
|
LayerState childState = RequiredLayerStateForChildren(
|
|
aBuilder, aManager, aParameters, *list,
|
|
aExpectedAnimatedGeometryRootForChildren);
|
|
if (childState > result) {
|
|
result = childState;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
return result;
|
|
}
|
|
|
|
nsRect nsDisplayWrapList::GetComponentAlphaBounds(
|
|
nsDisplayListBuilder* aBuilder) const {
|
|
return mListPtr->GetComponentAlphaBounds(aBuilder);
|
|
}
|
|
|
|
void nsDisplayWrapList::SetReferenceFrame(const nsIFrame* aFrame) {
|
|
mReferenceFrame = aFrame;
|
|
mToReferenceFrame = mFrame->GetOffsetToCrossDoc(mReferenceFrame);
|
|
}
|
|
|
|
bool nsDisplayWrapList::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
aManager->CommandBuilder().CreateWebRenderCommandsFromDisplayList(
|
|
GetChildren(), this, aDisplayListBuilder, aSc, aBuilder, aResources);
|
|
return true;
|
|
}
|
|
|
|
static nsresult WrapDisplayList(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame, nsDisplayList* aList,
|
|
nsDisplayWrapper* aWrapper) {
|
|
if (!aList->GetTop()) {
|
|
return NS_OK;
|
|
}
|
|
nsDisplayItem* item = aWrapper->WrapList(aBuilder, aFrame, aList);
|
|
if (!item) {
|
|
return NS_ERROR_OUT_OF_MEMORY;
|
|
}
|
|
// aList was emptied
|
|
aList->AppendToTop(item);
|
|
return NS_OK;
|
|
}
|
|
|
|
static nsresult WrapEachDisplayItem(nsDisplayListBuilder* aBuilder,
|
|
nsDisplayList* aList,
|
|
nsDisplayWrapper* aWrapper) {
|
|
nsDisplayList newList;
|
|
nsDisplayItem* item;
|
|
while ((item = aList->RemoveBottom())) {
|
|
item = aWrapper->WrapItem(aBuilder, item);
|
|
if (!item) {
|
|
return NS_ERROR_OUT_OF_MEMORY;
|
|
}
|
|
newList.AppendToTop(item);
|
|
}
|
|
// aList was emptied
|
|
aList->AppendToTop(&newList);
|
|
return NS_OK;
|
|
}
|
|
|
|
nsresult nsDisplayWrapper::WrapLists(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame,
|
|
const nsDisplayListSet& aIn,
|
|
const nsDisplayListSet& aOut) {
|
|
nsresult rv = WrapListsInPlace(aBuilder, aFrame, aIn);
|
|
NS_ENSURE_SUCCESS(rv, rv);
|
|
|
|
if (&aOut == &aIn) {
|
|
return NS_OK;
|
|
}
|
|
aOut.BorderBackground()->AppendToTop(aIn.BorderBackground());
|
|
aOut.BlockBorderBackgrounds()->AppendToTop(aIn.BlockBorderBackgrounds());
|
|
aOut.Floats()->AppendToTop(aIn.Floats());
|
|
aOut.Content()->AppendToTop(aIn.Content());
|
|
aOut.PositionedDescendants()->AppendToTop(aIn.PositionedDescendants());
|
|
aOut.Outlines()->AppendToTop(aIn.Outlines());
|
|
return NS_OK;
|
|
}
|
|
|
|
nsresult nsDisplayWrapper::WrapListsInPlace(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame,
|
|
const nsDisplayListSet& aLists) {
|
|
nsresult rv;
|
|
if (WrapBorderBackground()) {
|
|
// Our border-backgrounds are in-flow
|
|
rv = WrapDisplayList(aBuilder, aFrame, aLists.BorderBackground(), this);
|
|
NS_ENSURE_SUCCESS(rv, rv);
|
|
}
|
|
// Our block border-backgrounds are in-flow
|
|
rv = WrapDisplayList(aBuilder, aFrame, aLists.BlockBorderBackgrounds(), this);
|
|
NS_ENSURE_SUCCESS(rv, rv);
|
|
// The floats are not in flow
|
|
rv = WrapEachDisplayItem(aBuilder, aLists.Floats(), this);
|
|
NS_ENSURE_SUCCESS(rv, rv);
|
|
// Our child content is in flow
|
|
rv = WrapDisplayList(aBuilder, aFrame, aLists.Content(), this);
|
|
NS_ENSURE_SUCCESS(rv, rv);
|
|
// The positioned descendants may not be in-flow
|
|
rv = WrapEachDisplayItem(aBuilder, aLists.PositionedDescendants(), this);
|
|
NS_ENSURE_SUCCESS(rv, rv);
|
|
// The outlines may not be in-flow
|
|
return WrapEachDisplayItem(aBuilder, aLists.Outlines(), this);
|
|
}
|
|
|
|
nsDisplayOpacity::nsDisplayOpacity(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot,
|
|
bool aForEventsAndPluginsOnly, bool aNeedsActiveLayer)
|
|
: nsDisplayWrapList(aBuilder, aFrame, aList, aActiveScrolledRoot, true),
|
|
mOpacity(aFrame->StyleEffects()->mOpacity),
|
|
mForEventsAndPluginsOnly(aForEventsAndPluginsOnly),
|
|
mNeedsActiveLayer(aNeedsActiveLayer),
|
|
mChildOpacityState(ChildOpacityState::Unknown) {
|
|
MOZ_COUNT_CTOR(nsDisplayOpacity);
|
|
mState.mOpacity = mOpacity;
|
|
}
|
|
|
|
void nsDisplayOpacity::HitTest(nsDisplayListBuilder* aBuilder,
|
|
const nsRect& aRect,
|
|
nsDisplayItem::HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) {
|
|
// TODO(emilio): special-casing zero is a bit arbitrary... Maybe we should
|
|
// only consider fully opaque items? Or make this configurable somehow?
|
|
if (aBuilder->HitTestIsForVisibility() && mOpacity == 0.0f) {
|
|
return;
|
|
}
|
|
nsDisplayWrapList::HitTest(aBuilder, aRect, aState, aOutFrames);
|
|
}
|
|
|
|
nsRegion nsDisplayOpacity::GetOpaqueRegion(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = false;
|
|
// The only time where mOpacity == 1.0 should be when we have will-change.
|
|
// We could report this as opaque then but when the will-change value starts
|
|
// animating the element would become non opaque and could cause repaints.
|
|
return nsRegion();
|
|
}
|
|
|
|
// nsDisplayOpacity uses layers for rendering
|
|
already_AddRefed<Layer> nsDisplayOpacity::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
ContainerLayerParameters params = aContainerParameters;
|
|
RefPtr<Layer> container = aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, &mList, params, nullptr,
|
|
FrameLayerBuilder::CONTAINER_ALLOW_PULL_BACKGROUND_COLOR);
|
|
if (!container) {
|
|
return nullptr;
|
|
}
|
|
|
|
container->SetOpacity(mOpacity);
|
|
nsDisplayListBuilder::AddAnimationsAndTransitionsToLayer(
|
|
container, aBuilder, this, mFrame, GetType());
|
|
return container.forget();
|
|
}
|
|
|
|
/**
|
|
* This doesn't take into account layer scaling --- the layer may be
|
|
* rendered at a higher (or lower) resolution, affecting the retained layer
|
|
* size --- but this should be good enough.
|
|
*/
|
|
static bool IsItemTooSmallForActiveLayer(nsIFrame* aFrame) {
|
|
nsIntRect visibleDevPixels =
|
|
aFrame->GetVisualOverflowRectRelativeToSelf().ToOutsidePixels(
|
|
aFrame->PresContext()->AppUnitsPerDevPixel());
|
|
return visibleDevPixels.Size() <
|
|
nsIntSize(StaticPrefs::layout_min_active_layer_size(),
|
|
StaticPrefs::layout_min_active_layer_size());
|
|
}
|
|
|
|
/* static */
|
|
bool nsDisplayOpacity::NeedsActiveLayer(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame,
|
|
bool aEnforceMinimumSize) {
|
|
if (EffectCompositor::HasAnimationsForCompositor(
|
|
aFrame, DisplayItemType::TYPE_OPACITY) ||
|
|
(ActiveLayerTracker::IsStyleAnimated(
|
|
aBuilder, aFrame, nsCSSPropertyIDSet::OpacityProperties()) &&
|
|
!(aEnforceMinimumSize && IsItemTooSmallForActiveLayer(aFrame)))) {
|
|
return true;
|
|
}
|
|
return false;
|
|
}
|
|
|
|
void nsDisplayOpacity::ApplyOpacity(nsDisplayListBuilder* aBuilder,
|
|
float aOpacity,
|
|
const DisplayItemClipChain* aClip) {
|
|
NS_ASSERTION(CanApplyOpacity(), "ApplyOpacity should be allowed");
|
|
mOpacity = mOpacity * aOpacity;
|
|
IntersectClip(aBuilder, aClip, false);
|
|
}
|
|
|
|
bool nsDisplayOpacity::CanApplyOpacity() const {
|
|
return !EffectCompositor::HasAnimationsForCompositor(
|
|
mFrame, DisplayItemType::TYPE_OPACITY);
|
|
}
|
|
|
|
// Only try folding our opacity down if we have at most |kOpacityMaxChildCount|
|
|
// children that don't overlap and can all apply the opacity to themselves.
|
|
static const size_t kOpacityMaxChildCount = 3;
|
|
|
|
// |kOpacityMaxListSize| defines an early exit condition for opacity items that
|
|
// are likely have more child items than |kOpacityMaxChildCount|.
|
|
static const size_t kOpacityMaxListSize = kOpacityMaxChildCount * 2;
|
|
|
|
/**
|
|
* Recursively iterates through |aList| and collects at most
|
|
* |kOpacityMaxChildCount| display item pointers to items that return true for
|
|
* CanApplyOpacity(). The item pointers are added to |aArray|.
|
|
*
|
|
* LayerEventRegions and WrapList items are ignored.
|
|
*
|
|
* We need to do this recursively, because the child display items might contain
|
|
* nested nsDisplayWrapLists.
|
|
*
|
|
* Returns false if there are more than |kOpacityMaxChildCount| items, or if an
|
|
* item that returns false for CanApplyOpacity() is encountered.
|
|
* Otherwise returns true.
|
|
*/
|
|
static bool CollectItemsWithOpacity(nsDisplayList* aList,
|
|
nsTArray<nsPaintedDisplayItem*>& aArray) {
|
|
if (aList->Count() > kOpacityMaxListSize) {
|
|
// Exit early, since |aList| will likely contain more than
|
|
// |kOpacityMaxChildCount| items.
|
|
return false;
|
|
}
|
|
|
|
for (nsDisplayItem* i : *aList) {
|
|
const DisplayItemType type = i->GetType();
|
|
|
|
if (type == DisplayItemType::TYPE_COMPOSITOR_HITTEST_INFO) {
|
|
continue;
|
|
}
|
|
|
|
// Descend only into wraplists.
|
|
if (type == DisplayItemType::TYPE_WRAP_LIST ||
|
|
type == DisplayItemType::TYPE_CONTAINER) {
|
|
// The current display item has children, process them first.
|
|
if (!CollectItemsWithOpacity(i->GetChildren(), aArray)) {
|
|
return false;
|
|
}
|
|
|
|
continue;
|
|
}
|
|
|
|
if (aArray.Length() == kOpacityMaxChildCount) {
|
|
return false;
|
|
}
|
|
|
|
auto* item = i->AsPaintedDisplayItem();
|
|
if (!item || !item->CanApplyOpacity()) {
|
|
return false;
|
|
}
|
|
|
|
aArray.AppendElement(item);
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayOpacity::ApplyOpacityToChildren(nsDisplayListBuilder* aBuilder) {
|
|
if (mChildOpacityState == ChildOpacityState::Deferred) {
|
|
return false;
|
|
}
|
|
|
|
// Iterate through the child display list and copy at most
|
|
// |kOpacityMaxChildCount| child display item pointers to a temporary list.
|
|
AutoTArray<nsPaintedDisplayItem*, kOpacityMaxChildCount> items;
|
|
if (!CollectItemsWithOpacity(&mList, items)) {
|
|
mChildOpacityState = ChildOpacityState::Deferred;
|
|
return false;
|
|
}
|
|
|
|
struct {
|
|
nsPaintedDisplayItem* item;
|
|
nsRect bounds;
|
|
} children[kOpacityMaxChildCount];
|
|
|
|
bool snap;
|
|
size_t childCount = 0;
|
|
for (nsPaintedDisplayItem* item : items) {
|
|
children[childCount].item = item;
|
|
children[childCount].bounds = item->GetBounds(aBuilder, &snap);
|
|
childCount++;
|
|
}
|
|
|
|
for (size_t i = 0; i < childCount; i++) {
|
|
for (size_t j = i + 1; j < childCount; j++) {
|
|
if (children[i].bounds.Intersects(children[j].bounds)) {
|
|
mChildOpacityState = ChildOpacityState::Deferred;
|
|
return false;
|
|
}
|
|
}
|
|
}
|
|
|
|
for (uint32_t i = 0; i < childCount; i++) {
|
|
children[i].item->ApplyOpacity(aBuilder, mOpacity, mClipChain);
|
|
}
|
|
|
|
mChildOpacityState = ChildOpacityState::Applied;
|
|
return true;
|
|
}
|
|
|
|
/**
|
|
* Returns true if this nsDisplayOpacity contains only a filter or a mask item
|
|
* that has the same frame as the opacity item. In this case the opacity item
|
|
* can be optimized away.
|
|
*/
|
|
bool nsDisplayOpacity::IsEffectsWrapper() const {
|
|
if (mList.Count() != 1) {
|
|
return false;
|
|
}
|
|
|
|
const nsDisplayItem* item = mList.GetBottom();
|
|
|
|
if (item->Frame() != mFrame) {
|
|
// The effect item needs to have the same frame as the opacity item.
|
|
return false;
|
|
}
|
|
|
|
const DisplayItemType type = item->GetType();
|
|
return type == DisplayItemType::TYPE_MASK ||
|
|
type == DisplayItemType::TYPE_FILTER;
|
|
}
|
|
|
|
bool nsDisplayOpacity::ShouldFlattenAway(nsDisplayListBuilder* aBuilder) {
|
|
if (mFrame->GetPrevContinuation() || mFrame->GetNextContinuation() ||
|
|
mFrame->HasAnyStateBits(NS_FRAME_PART_OF_IBSPLIT)) {
|
|
// If we've been split, then we might need to merge, so
|
|
// don't flatten us away.
|
|
return false;
|
|
}
|
|
|
|
if (mNeedsActiveLayer || mOpacity == 0.0) {
|
|
// If our opacity is zero then we'll discard all descendant display items
|
|
// except for layer event regions, so there's no point in doing this
|
|
// optimization (and if we do do it, then invalidations of those descendants
|
|
// might trigger repainting).
|
|
return false;
|
|
}
|
|
|
|
if (mList.IsEmpty()) {
|
|
return false;
|
|
}
|
|
|
|
if (IsEffectsWrapper()) {
|
|
MOZ_ASSERT(nsSVGIntegrationUtils::UsingEffectsForFrame(mFrame));
|
|
static_cast<nsDisplayEffectsBase*>(mList.GetBottom())->SetHandleOpacity();
|
|
mChildOpacityState = ChildOpacityState::Applied;
|
|
return true;
|
|
}
|
|
|
|
// Return true if we successfully applied opacity to child items, or if
|
|
// WebRender is not in use. In the latter case, the opacity gets flattened and
|
|
// applied during layer building.
|
|
return ApplyOpacityToChildren(aBuilder) || !gfxVars::UseWebRender();
|
|
}
|
|
|
|
nsDisplayItem::LayerState nsDisplayOpacity::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
// If we only created this item so that we'd get correct nsDisplayEventRegions
|
|
// for child frames, then force us to inactive to avoid unnecessary
|
|
// layerization changes for content that won't ever be painted.
|
|
if (mForEventsAndPluginsOnly) {
|
|
MOZ_ASSERT(mOpacity == 0);
|
|
return LayerState::LAYER_INACTIVE;
|
|
}
|
|
|
|
if (mNeedsActiveLayer) {
|
|
// Returns LayerState::LAYER_ACTIVE_FORCE to avoid flatterning the layer for
|
|
// async animations.
|
|
return LayerState::LAYER_ACTIVE_FORCE;
|
|
}
|
|
|
|
return RequiredLayerStateForChildren(aBuilder, aManager, aParameters, mList,
|
|
GetAnimatedGeometryRoot());
|
|
}
|
|
|
|
bool nsDisplayOpacity::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
// Our children are translucent so we should not allow them to subtract
|
|
// area from aVisibleRegion. We do need to find out what is visible under
|
|
// our children in the temporary compositing buffer, because if our children
|
|
// paint our entire bounds opaquely then we don't need an alpha channel in
|
|
// the temporary compositing buffer.
|
|
nsRect bounds = GetClippedBounds(aBuilder);
|
|
nsRegion visibleUnderChildren;
|
|
visibleUnderChildren.And(*aVisibleRegion, bounds);
|
|
return nsDisplayWrapList::ComputeVisibility(aBuilder, &visibleUnderChildren);
|
|
}
|
|
|
|
void nsDisplayOpacity::ComputeInvalidationRegion(
|
|
nsDisplayListBuilder* aBuilder, const nsDisplayItemGeometry* aGeometry,
|
|
nsRegion* aInvalidRegion) const {
|
|
auto* geometry = static_cast<const nsDisplayOpacityGeometry*>(aGeometry);
|
|
|
|
bool snap;
|
|
if (mOpacity != geometry->mOpacity) {
|
|
aInvalidRegion->Or(GetBounds(aBuilder, &snap), geometry->mBounds);
|
|
}
|
|
}
|
|
|
|
void nsDisplayOpacity::WriteDebugInfo(std::stringstream& aStream) {
|
|
aStream << " (opacity " << mOpacity << ")";
|
|
}
|
|
|
|
bool nsDisplayOpacity::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
float* opacityForSC = &mOpacity;
|
|
|
|
uint64_t animationsId = AddAnimationsForWebRender(
|
|
this, aManager, aDisplayListBuilder, aBuilder.GetRenderRoot());
|
|
wr::WrAnimationProperty prop{
|
|
wr::WrAnimationType::Opacity,
|
|
animationsId,
|
|
};
|
|
|
|
wr::StackingContextParams params;
|
|
params.animation = animationsId ? &prop : nullptr;
|
|
params.opacity = opacityForSC;
|
|
params.clip =
|
|
wr::WrStackingContextClip::ClipChain(aBuilder.CurrentClipChainId());
|
|
StackingContextHelper sc(aSc, GetActiveScrolledRoot(), mFrame, this, aBuilder,
|
|
params);
|
|
|
|
aManager->CommandBuilder().CreateWebRenderCommandsFromDisplayList(
|
|
&mList, this, aDisplayListBuilder, sc, aBuilder, aResources);
|
|
return true;
|
|
}
|
|
|
|
nsDisplayBlendMode::nsDisplayBlendMode(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
uint8_t aBlendMode, const ActiveScrolledRoot* aActiveScrolledRoot,
|
|
uint16_t aIndex)
|
|
: nsDisplayWrapList(aBuilder, aFrame, aList, aActiveScrolledRoot, true),
|
|
mBlendMode(aBlendMode),
|
|
mIndex(aIndex) {
|
|
MOZ_COUNT_CTOR(nsDisplayBlendMode);
|
|
}
|
|
|
|
nsRegion nsDisplayBlendMode::GetOpaqueRegion(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = false;
|
|
// We are never considered opaque
|
|
return nsRegion();
|
|
}
|
|
|
|
LayerState nsDisplayBlendMode::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
return LayerState::LAYER_ACTIVE;
|
|
}
|
|
|
|
bool nsDisplayBlendMode::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
wr::StackingContextParams params;
|
|
params.mix_blend_mode =
|
|
wr::ToMixBlendMode(nsCSSRendering::GetGFXBlendMode(mBlendMode));
|
|
params.clip =
|
|
wr::WrStackingContextClip::ClipChain(aBuilder.CurrentClipChainId());
|
|
StackingContextHelper sc(aSc, GetActiveScrolledRoot(), mFrame, this, aBuilder,
|
|
params);
|
|
|
|
return nsDisplayWrapList::CreateWebRenderCommands(
|
|
aBuilder, aResources, sc, aManager, aDisplayListBuilder);
|
|
}
|
|
|
|
// nsDisplayBlendMode uses layers for rendering
|
|
already_AddRefed<Layer> nsDisplayBlendMode::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
ContainerLayerParameters newContainerParameters = aContainerParameters;
|
|
newContainerParameters.mDisableSubpixelAntialiasingInDescendants = true;
|
|
|
|
RefPtr<Layer> container = aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, &mList, newContainerParameters,
|
|
nullptr);
|
|
if (!container) {
|
|
return nullptr;
|
|
}
|
|
|
|
container->SetMixBlendMode(nsCSSRendering::GetGFXBlendMode(mBlendMode));
|
|
|
|
return container.forget();
|
|
}
|
|
|
|
mozilla::gfx::CompositionOp nsDisplayBlendMode::BlendMode() {
|
|
return nsCSSRendering::GetGFXBlendMode(mBlendMode);
|
|
}
|
|
|
|
bool nsDisplayBlendMode::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
// Our children are need their backdrop so we should not allow them to
|
|
// subtract area from aVisibleRegion. We do need to find out what is visible
|
|
// under our children in the temporary compositing buffer, because if our
|
|
// children paint our entire bounds opaquely then we don't need an alpha
|
|
// channel in the temporary compositing buffer.
|
|
nsRect bounds = GetClippedBounds(aBuilder);
|
|
nsRegion visibleUnderChildren;
|
|
visibleUnderChildren.And(*aVisibleRegion, bounds);
|
|
return nsDisplayWrapList::ComputeVisibility(aBuilder, &visibleUnderChildren);
|
|
}
|
|
|
|
bool nsDisplayBlendMode::CanMerge(const nsDisplayItem* aItem) const {
|
|
// Items for the same content element should be merged into a single
|
|
// compositing group.
|
|
if (!HasDifferentFrame(aItem) || !HasSameTypeAndClip(aItem) ||
|
|
!HasSameContent(aItem)) {
|
|
return false;
|
|
}
|
|
|
|
const nsDisplayBlendMode* item =
|
|
static_cast<const nsDisplayBlendMode*>(aItem);
|
|
|
|
if (item->mIndex != 0 || mIndex != 0) {
|
|
// Don't merge background-blend-mode items
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
/* static */
|
|
nsDisplayBlendContainer* nsDisplayBlendContainer::CreateForMixBlendMode(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot) {
|
|
return MakeDisplayItem<nsDisplayBlendContainer>(aBuilder, aFrame, aList,
|
|
aActiveScrolledRoot, false);
|
|
}
|
|
|
|
/* static */
|
|
nsDisplayBlendContainer* nsDisplayBlendContainer::CreateForBackgroundBlendMode(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot) {
|
|
return MakeDisplayItem<nsDisplayBlendContainer>(aBuilder, aFrame, aList,
|
|
aActiveScrolledRoot, true);
|
|
}
|
|
|
|
nsDisplayBlendContainer::nsDisplayBlendContainer(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot, bool aIsForBackground)
|
|
: nsDisplayWrapList(aBuilder, aFrame, aList, aActiveScrolledRoot, true),
|
|
mIsForBackground(aIsForBackground) {
|
|
MOZ_COUNT_CTOR(nsDisplayBlendContainer);
|
|
}
|
|
|
|
// nsDisplayBlendContainer uses layers for rendering
|
|
already_AddRefed<Layer> nsDisplayBlendContainer::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
// turn off anti-aliasing in the parent stacking context because it changes
|
|
// how the group is initialized.
|
|
ContainerLayerParameters newContainerParameters = aContainerParameters;
|
|
newContainerParameters.mDisableSubpixelAntialiasingInDescendants = true;
|
|
|
|
RefPtr<Layer> container = aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, &mList, newContainerParameters,
|
|
nullptr);
|
|
if (!container) {
|
|
return nullptr;
|
|
}
|
|
|
|
container->SetForceIsolatedGroup(true);
|
|
return container.forget();
|
|
}
|
|
|
|
LayerState nsDisplayBlendContainer::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
return RequiredLayerStateForChildren(aBuilder, aManager, aParameters, mList,
|
|
GetAnimatedGeometryRoot());
|
|
}
|
|
|
|
bool nsDisplayBlendContainer::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
wr::StackingContextParams params;
|
|
params.clip =
|
|
wr::WrStackingContextClip::ClipChain(aBuilder.CurrentClipChainId());
|
|
StackingContextHelper sc(aSc, GetActiveScrolledRoot(), mFrame, this, aBuilder,
|
|
params);
|
|
|
|
return nsDisplayWrapList::CreateWebRenderCommands(
|
|
aBuilder, aResources, sc, aManager, aDisplayListBuilder);
|
|
}
|
|
|
|
/* static */
|
|
nsDisplayTableBlendContainer*
|
|
nsDisplayTableBlendContainer::CreateForBackgroundBlendMode(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot, nsIFrame* aAncestorFrame) {
|
|
return MakeDisplayItem<nsDisplayTableBlendContainer>(
|
|
aBuilder, aFrame, aList, aActiveScrolledRoot, true, aAncestorFrame);
|
|
}
|
|
|
|
nsDisplayOwnLayer::nsDisplayOwnLayer(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot,
|
|
nsDisplayOwnLayerFlags aFlags, const ScrollbarData& aScrollbarData,
|
|
bool aForceActive, bool aClearClipChain, uint16_t aIndex)
|
|
: nsDisplayWrapList(aBuilder, aFrame, aList, aActiveScrolledRoot,
|
|
aClearClipChain),
|
|
mFlags(aFlags),
|
|
mScrollbarData(aScrollbarData),
|
|
mForceActive(aForceActive),
|
|
mWrAnimationId(0),
|
|
mIndex(aIndex) {
|
|
MOZ_COUNT_CTOR(nsDisplayOwnLayer);
|
|
|
|
// For scroll thumb layers, override the AGR to be the thumb's AGR rather
|
|
// than the AGR for mFrame (which is the slider frame).
|
|
if (IsScrollThumbLayer()) {
|
|
if (nsIFrame* thumbFrame = nsBox::GetChildXULBox(mFrame)) {
|
|
mAnimatedGeometryRoot = aBuilder->FindAnimatedGeometryRootFor(thumbFrame);
|
|
}
|
|
}
|
|
}
|
|
|
|
LayerState nsDisplayOwnLayer::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
if (mForceActive) {
|
|
return mozilla::LayerState::LAYER_ACTIVE_FORCE;
|
|
}
|
|
|
|
return RequiredLayerStateForChildren(aBuilder, aManager, aParameters, mList,
|
|
mAnimatedGeometryRoot);
|
|
}
|
|
|
|
bool nsDisplayOwnLayer::IsScrollThumbLayer() const {
|
|
return mScrollbarData.mScrollbarLayerType ==
|
|
layers::ScrollbarLayerType::Thumb;
|
|
}
|
|
|
|
bool nsDisplayOwnLayer::IsScrollbarContainer() const {
|
|
return mScrollbarData.mScrollbarLayerType ==
|
|
layers::ScrollbarLayerType::Container;
|
|
}
|
|
|
|
bool nsDisplayOwnLayer::IsZoomingLayer() const {
|
|
return GetType() == DisplayItemType::TYPE_ASYNC_ZOOM;
|
|
}
|
|
|
|
bool nsDisplayOwnLayer::IsFixedPositionLayer() const {
|
|
return GetType() == DisplayItemType::TYPE_FIXED_POSITION;
|
|
}
|
|
|
|
// nsDisplayOpacity uses layers for rendering
|
|
already_AddRefed<Layer> nsDisplayOwnLayer::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
RefPtr<ContainerLayer> layer =
|
|
aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, &mList, aContainerParameters,
|
|
nullptr, FrameLayerBuilder::CONTAINER_ALLOW_PULL_BACKGROUND_COLOR);
|
|
|
|
if (IsScrollThumbLayer() || IsScrollbarContainer()) {
|
|
layer->SetScrollbarData(mScrollbarData);
|
|
}
|
|
|
|
if (mFlags & nsDisplayOwnLayerFlags::GenerateSubdocInvalidations) {
|
|
mFrame->PresContext()->SetNotifySubDocInvalidationData(layer);
|
|
}
|
|
return layer.forget();
|
|
}
|
|
|
|
bool nsDisplayOwnLayer::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc, RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
Maybe<wr::WrAnimationProperty> prop;
|
|
bool needsProp =
|
|
aManager->LayerManager()->AsyncPanZoomEnabled() &&
|
|
(IsScrollThumbLayer() || IsZoomingLayer() || IsFixedPositionLayer());
|
|
|
|
if (needsProp) {
|
|
// APZ is enabled and this is a scroll thumb or zooming layer, so we need
|
|
// to create and set an animation id. That way APZ can adjust the position/
|
|
// zoom of this content asynchronously as needed.
|
|
RefPtr<WebRenderAnimationData> animationData =
|
|
aManager->CommandBuilder()
|
|
.CreateOrRecycleWebRenderUserData<WebRenderAnimationData>(
|
|
this, aBuilder.GetRenderRoot());
|
|
AnimationInfo& animationInfo = animationData->GetAnimationInfo();
|
|
animationInfo.EnsureAnimationsId();
|
|
mWrAnimationId = animationInfo.GetCompositorAnimationsId();
|
|
|
|
prop.emplace();
|
|
prop->id = mWrAnimationId;
|
|
prop->effect_type = wr::WrAnimationType::Transform;
|
|
}
|
|
|
|
wr::StackingContextParams params;
|
|
params.animation = prop.ptrOr(nullptr);
|
|
params.clip =
|
|
wr::WrStackingContextClip::ClipChain(aBuilder.CurrentClipChainId());
|
|
if (IsScrollbarContainer()) {
|
|
params.prim_flags |= wr::PrimitiveFlags::IS_SCROLLBAR_CONTAINER;
|
|
}
|
|
if (IsScrollThumbLayer()) {
|
|
params.prim_flags |= wr::PrimitiveFlags::IS_SCROLLBAR_THUMB;
|
|
}
|
|
StackingContextHelper sc(aSc, GetActiveScrolledRoot(), mFrame, this, aBuilder,
|
|
params);
|
|
|
|
nsDisplayWrapList::CreateWebRenderCommands(aBuilder, aResources, sc, aManager,
|
|
aDisplayListBuilder);
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayOwnLayer::UpdateScrollData(
|
|
mozilla::layers::WebRenderScrollData* aData,
|
|
mozilla::layers::WebRenderLayerScrollData* aLayerData) {
|
|
bool isRelevantToApz = (IsScrollThumbLayer() || IsScrollbarContainer() ||
|
|
IsZoomingLayer() || IsFixedPositionLayer());
|
|
if (!isRelevantToApz) {
|
|
return false;
|
|
}
|
|
|
|
if (!aLayerData) {
|
|
return true;
|
|
}
|
|
|
|
if (IsZoomingLayer()) {
|
|
aLayerData->SetZoomAnimationId(mWrAnimationId);
|
|
return true;
|
|
}
|
|
|
|
if (IsFixedPositionLayer()) {
|
|
aLayerData->SetFixedPositionAnimationId(mWrAnimationId);
|
|
return true;
|
|
}
|
|
|
|
aLayerData->SetScrollbarData(mScrollbarData);
|
|
if (IsScrollThumbLayer()) {
|
|
aLayerData->SetScrollbarAnimationId(mWrAnimationId);
|
|
LayoutDeviceRect bounds = LayoutDeviceIntRect::FromAppUnits(
|
|
mBounds, mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
// We use a resolution of 1.0 because this is a WebRender codepath which
|
|
// always uses containerless scrolling, and so resolution doesn't apply to
|
|
// scrollbars.
|
|
LayerIntRect layerBounds =
|
|
RoundedOut(bounds * LayoutDeviceToLayerScale(1.0f));
|
|
aLayerData->SetVisibleRegion(LayerIntRegion(layerBounds));
|
|
}
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayOwnLayer::WriteDebugInfo(std::stringstream& aStream) {
|
|
aStream << nsPrintfCString(" (flags 0x%x) (scrolltarget %" PRIu64 ")",
|
|
(int)mFlags, mScrollbarData.mTargetViewId)
|
|
.get();
|
|
}
|
|
|
|
nsDisplayRenderRoot::nsDisplayRenderRoot(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot, wr::RenderRoot aRenderRoot)
|
|
: nsDisplayWrapList(aBuilder, aFrame, aList, aActiveScrolledRoot, true),
|
|
mRenderRoot(aRenderRoot),
|
|
mBuiltWRCommands(false) {
|
|
MOZ_ASSERT(aRenderRoot != wr::RenderRoot::Default);
|
|
MOZ_ASSERT(StaticPrefs::gfx_webrender_split_render_roots_AtStartup());
|
|
ExpandDisplayListBuilderRenderRootRect(aBuilder);
|
|
MOZ_COUNT_CTOR(nsDisplayRenderRoot);
|
|
}
|
|
|
|
void nsDisplayRenderRoot::Destroy(nsDisplayListBuilder* aBuilder) {
|
|
if (mBuiltWRCommands && aBuilder) {
|
|
aBuilder->SetNeedsDisplayListBuild(mRenderRoot);
|
|
}
|
|
nsDisplayWrapList::Destroy(aBuilder);
|
|
}
|
|
|
|
void nsDisplayRenderRoot::NotifyUsed(nsDisplayListBuilder* aBuilder) {
|
|
ExpandDisplayListBuilderRenderRootRect(aBuilder);
|
|
nsDisplayWrapList::SetReused(aBuilder);
|
|
}
|
|
|
|
void nsDisplayRenderRoot::InvalidateCachedChildInfo(
|
|
nsDisplayListBuilder* aBuilder) {
|
|
if (mBuiltWRCommands && aBuilder) {
|
|
aBuilder->SetNeedsDisplayListBuild(mRenderRoot);
|
|
mBuiltWRCommands = false;
|
|
}
|
|
}
|
|
|
|
bool nsDisplayRenderRoot::UpdateScrollData(
|
|
mozilla::layers::WebRenderScrollData* aData,
|
|
mozilla::layers::WebRenderLayerScrollData* aLayerData) {
|
|
// Ideally we'd return false if this nsDisplayRenderRoot isn't actually
|
|
// establishing a render root boundary (i.e. if mRenderRoot ==
|
|
// aBuilder.GetRenderRoot()). But we don't have aBuilder here so we can't do
|
|
// that. Returning true unconditionally is suboptimal but still correct.
|
|
if (aLayerData) {
|
|
if (mBoundary) {
|
|
// Scroll data entries that are referents should never have direct
|
|
// descendants. Instead they will refer to a different subtree
|
|
MOZ_ASSERT(aLayerData->GetDescendantCount() == 0);
|
|
aLayerData->SetReferentRenderRoot(*mBoundary);
|
|
}
|
|
}
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayRenderRoot::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc, RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
// It's important to get the userData here even in the early-return case,
|
|
// because this has the important side-effect of marking the user data "used"
|
|
// so it doesn't get discarded at the end of the transaction.
|
|
RefPtr<WebRenderRenderRootData> userData =
|
|
aManager->CommandBuilder()
|
|
.CreateOrRecycleWebRenderUserData<WebRenderRenderRootData>(
|
|
this, aBuilder.GetRenderRoot());
|
|
// Technically the next line is redundant but maybe it will stop people who
|
|
// don't read comments from accidentally moving the above line of code back
|
|
// down below the early-return.
|
|
userData->SetUsed(true);
|
|
|
|
if (!aDisplayListBuilder->GetNeedsDisplayListBuild(mRenderRoot) &&
|
|
mBuiltWRCommands) {
|
|
return true;
|
|
}
|
|
if (aBuilder.GetRenderRoot() == mRenderRoot) {
|
|
nsDisplayWrapList::CreateWebRenderCommands(aBuilder, aResources, aSc,
|
|
aManager, aDisplayListBuilder);
|
|
} else {
|
|
mBoundary = Some(userData->EnsureHasBoundary(mRenderRoot));
|
|
|
|
WebRenderCommandBuilder::ScrollDataBoundaryWrapper wrapper(
|
|
aManager->CommandBuilder(), *mBoundary);
|
|
|
|
aBuilder.SetSendSubBuilderDisplayList(mRenderRoot);
|
|
wr::DisplayListBuilder& builder = aBuilder.SubBuilder(mRenderRoot);
|
|
wr::IpcResourceUpdateQueue& resources = aResources.SubQueue(mRenderRoot);
|
|
|
|
wr::StackingContextParams params;
|
|
params.clip =
|
|
wr::WrStackingContextClip::ClipChain(builder.CurrentClipChainId());
|
|
LayoutDeviceRect rrRect =
|
|
aDisplayListBuilder->GetRenderRootRect(mRenderRoot);
|
|
LayoutDevicePoint scOrigin = aSc.GetOrigin();
|
|
// Subtract the render root rect from this, as it already acts as the
|
|
// origin for our whole display list in WebRender (via SetDocumentView).
|
|
// However, we can't simply ignore the value of aSc.GetOrigin(), even
|
|
// though they will often be the same. This is because there might be
|
|
// multiple cousin nsDisplayRenderRoots in a tree of nsDisplayItems, and
|
|
// the RenderRootRect will be the union of their areas.
|
|
scOrigin.x -= rrRect.x;
|
|
scOrigin.y -= rrRect.y;
|
|
StackingContextHelper sc(
|
|
aManager->CommandBuilder().GetRootStackingContextHelper(mRenderRoot),
|
|
nullptr, nullptr, nullptr, builder, params,
|
|
LayoutDeviceRect(scOrigin, LayoutDeviceSize()));
|
|
|
|
nsDisplayWrapList::CreateWebRenderCommands(builder, resources, sc, aManager,
|
|
aDisplayListBuilder);
|
|
}
|
|
mBuiltWRCommands = true;
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayRenderRoot::ExpandDisplayListBuilderRenderRootRect(
|
|
nsDisplayListBuilder* aBuilder) {
|
|
if (mFrame->GetRect().IsEmpty()) {
|
|
return;
|
|
}
|
|
mozilla::LayoutDeviceRect rect = mozilla::LayoutDeviceRect::FromAppUnits(
|
|
mFrame->GetRectRelativeToSelf() + ToReferenceFrame(),
|
|
mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
aBuilder->ExpandRenderRootRect(rect, mRenderRoot);
|
|
}
|
|
|
|
nsDisplaySubDocument::nsDisplaySubDocument(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame,
|
|
nsSubDocumentFrame* aSubDocFrame,
|
|
nsDisplayList* aList,
|
|
nsDisplayOwnLayerFlags aFlags)
|
|
: nsDisplayOwnLayer(aBuilder, aFrame, aList,
|
|
aBuilder->CurrentActiveScrolledRoot(), aFlags),
|
|
mScrollParentId(aBuilder->GetCurrentScrollParentId()),
|
|
mShouldFlatten(false),
|
|
mSubDocFrame(aSubDocFrame) {
|
|
MOZ_COUNT_CTOR(nsDisplaySubDocument);
|
|
|
|
// The SubDocument display item is conceptually outside the viewport frame,
|
|
// so in cases where the viewport frame is an AGR, the SubDocument's AGR
|
|
// should be not the viewport frame itself, but its parent AGR.
|
|
if (*mAnimatedGeometryRoot == mFrame && mAnimatedGeometryRoot->mParentAGR) {
|
|
mAnimatedGeometryRoot = mAnimatedGeometryRoot->mParentAGR;
|
|
}
|
|
|
|
if (mSubDocFrame && mSubDocFrame != mFrame) {
|
|
mSubDocFrame->AddDisplayItem(this);
|
|
}
|
|
}
|
|
|
|
nsDisplaySubDocument::~nsDisplaySubDocument() {
|
|
MOZ_COUNT_DTOR(nsDisplaySubDocument);
|
|
if (mSubDocFrame) {
|
|
mSubDocFrame->RemoveDisplayItem(this);
|
|
}
|
|
}
|
|
|
|
nsIFrame* nsDisplaySubDocument::FrameForInvalidation() const {
|
|
return mSubDocFrame ? mSubDocFrame : mFrame;
|
|
}
|
|
|
|
void nsDisplaySubDocument::RemoveFrame(nsIFrame* aFrame) {
|
|
if (aFrame == mSubDocFrame) {
|
|
mSubDocFrame = nullptr;
|
|
SetDeletedFrame();
|
|
}
|
|
nsDisplayOwnLayer::RemoveFrame(aFrame);
|
|
}
|
|
|
|
void nsDisplaySubDocument::Disown() {
|
|
if (mFrame) {
|
|
mFrame->RemoveDisplayItem(this);
|
|
RemoveFrame(mFrame);
|
|
}
|
|
if (mSubDocFrame) {
|
|
mSubDocFrame->RemoveDisplayItem(this);
|
|
RemoveFrame(mSubDocFrame);
|
|
}
|
|
}
|
|
|
|
UniquePtr<ScrollMetadata> nsDisplaySubDocument::ComputeScrollMetadata(
|
|
LayerManager* aLayerManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
if (!(mFlags & nsDisplayOwnLayerFlags::GenerateScrollableLayer)) {
|
|
return UniquePtr<ScrollMetadata>(nullptr);
|
|
}
|
|
|
|
nsPresContext* presContext = mFrame->PresContext();
|
|
nsIFrame* rootScrollFrame = presContext->PresShell()->GetRootScrollFrame();
|
|
bool isRootContentDocument = presContext->IsRootContentDocumentCrossProcess();
|
|
PresShell* presShell = presContext->PresShell();
|
|
ContainerLayerParameters params(
|
|
aContainerParameters.mXScale * presShell->GetResolution(),
|
|
aContainerParameters.mYScale * presShell->GetResolution(), nsIntPoint(),
|
|
aContainerParameters);
|
|
|
|
nsRect viewport = mFrame->GetRect() - mFrame->GetPosition() +
|
|
mFrame->GetOffsetToCrossDoc(ReferenceFrame());
|
|
|
|
nsIScrollableFrame* scrollableFrame = rootScrollFrame->GetScrollTargetFrame();
|
|
if (isRootContentDocument) {
|
|
viewport.SizeTo(scrollableFrame->GetScrollPortRect().Size());
|
|
}
|
|
|
|
UniquePtr<ScrollMetadata> metadata =
|
|
MakeUnique<ScrollMetadata>(nsLayoutUtils::ComputeScrollMetadata(
|
|
mFrame, rootScrollFrame, rootScrollFrame->GetContent(),
|
|
ReferenceFrame(), aLayerManager, mScrollParentId, viewport, Nothing(),
|
|
isRootContentDocument, Some(params)));
|
|
if (scrollableFrame) {
|
|
scrollableFrame->NotifyApzTransaction();
|
|
}
|
|
|
|
return metadata;
|
|
}
|
|
|
|
static bool UseDisplayPortForViewport(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame) {
|
|
return aBuilder->IsPaintingToWindow() &&
|
|
nsLayoutUtils::ViewportHasDisplayPort(aFrame->PresContext());
|
|
}
|
|
|
|
nsRect nsDisplaySubDocument::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
bool usingDisplayPort = UseDisplayPortForViewport(aBuilder, mFrame);
|
|
|
|
if ((mFlags & nsDisplayOwnLayerFlags::GenerateScrollableLayer) &&
|
|
usingDisplayPort) {
|
|
*aSnap = false;
|
|
return mFrame->GetRect() + aBuilder->ToReferenceFrame(mFrame);
|
|
}
|
|
|
|
return nsDisplayOwnLayer::GetBounds(aBuilder, aSnap);
|
|
}
|
|
|
|
bool nsDisplaySubDocument::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
bool usingDisplayPort = UseDisplayPortForViewport(aBuilder, mFrame);
|
|
|
|
if (!(mFlags & nsDisplayOwnLayerFlags::GenerateScrollableLayer) ||
|
|
!usingDisplayPort) {
|
|
return nsDisplayWrapList::ComputeVisibility(aBuilder, aVisibleRegion);
|
|
}
|
|
|
|
nsRect displayport;
|
|
nsIFrame* rootScrollFrame = mFrame->PresShell()->GetRootScrollFrame();
|
|
MOZ_ASSERT(rootScrollFrame);
|
|
Unused << nsLayoutUtils::GetDisplayPort(
|
|
rootScrollFrame->GetContent(), &displayport, RelativeTo::ScrollFrame);
|
|
|
|
nsRegion childVisibleRegion;
|
|
// The visible region for the children may be much bigger than the hole we
|
|
// are viewing the children from, so that the compositor process has enough
|
|
// content to asynchronously pan while content is being refreshed.
|
|
childVisibleRegion =
|
|
displayport + mFrame->GetOffsetToCrossDoc(ReferenceFrame());
|
|
|
|
nsRect boundedRect = childVisibleRegion.GetBounds().Intersect(
|
|
mList.GetClippedBoundsWithRespectToASR(aBuilder, mActiveScrolledRoot));
|
|
bool visible = mList.ComputeVisibilityForSublist(
|
|
aBuilder, &childVisibleRegion, boundedRect);
|
|
|
|
// If APZ is enabled then don't allow this computation to influence
|
|
// aVisibleRegion, on the assumption that the layer can be asynchronously
|
|
// scrolled so we'll definitely need all the content under it.
|
|
if (!nsLayoutUtils::UsesAsyncScrolling(mFrame)) {
|
|
bool snap;
|
|
nsRect bounds = GetBounds(aBuilder, &snap);
|
|
nsRegion removed;
|
|
removed.Sub(bounds, childVisibleRegion);
|
|
|
|
aBuilder->SubtractFromVisibleRegion(aVisibleRegion, removed);
|
|
}
|
|
|
|
return visible;
|
|
}
|
|
|
|
nsRegion nsDisplaySubDocument::GetOpaqueRegion(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
bool usingDisplayPort = UseDisplayPortForViewport(aBuilder, mFrame);
|
|
|
|
if ((mFlags & nsDisplayOwnLayerFlags::GenerateScrollableLayer) &&
|
|
usingDisplayPort) {
|
|
*aSnap = false;
|
|
return nsRegion();
|
|
}
|
|
|
|
return nsDisplayOwnLayer::GetOpaqueRegion(aBuilder, aSnap);
|
|
}
|
|
|
|
nsDisplayResolution::nsDisplayResolution(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame,
|
|
nsSubDocumentFrame* aSubDocFrame,
|
|
nsDisplayList* aList,
|
|
nsDisplayOwnLayerFlags aFlags)
|
|
: nsDisplaySubDocument(aBuilder, aFrame, aSubDocFrame, aList, aFlags) {
|
|
MOZ_COUNT_CTOR(nsDisplayResolution);
|
|
}
|
|
|
|
void nsDisplayResolution::HitTest(nsDisplayListBuilder* aBuilder,
|
|
const nsRect& aRect, HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) {
|
|
PresShell* presShell = mFrame->PresShell();
|
|
nsRect rect = aRect.RemoveResolution(presShell->GetResolution());
|
|
mList.HitTest(aBuilder, rect, aState, aOutFrames);
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayResolution::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
float rootLayerResolution = 1.0f;
|
|
ContainerLayerParameters containerParameters(
|
|
rootLayerResolution, rootLayerResolution, nsIntPoint(),
|
|
aContainerParameters);
|
|
|
|
RefPtr<Layer> layer =
|
|
nsDisplaySubDocument::BuildLayer(aBuilder, aManager, containerParameters);
|
|
|
|
return layer.forget();
|
|
}
|
|
|
|
nsDisplayFixedPosition::nsDisplayFixedPosition(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot,
|
|
const ActiveScrolledRoot* aContainerASR)
|
|
: nsDisplayOwnLayer(aBuilder, aFrame, aList, aActiveScrolledRoot),
|
|
mContainerASR(aContainerASR),
|
|
mIndex(0),
|
|
mIsFixedBackground(false) {
|
|
MOZ_COUNT_CTOR(nsDisplayFixedPosition);
|
|
Init(aBuilder);
|
|
}
|
|
|
|
nsDisplayFixedPosition::nsDisplayFixedPosition(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame,
|
|
nsDisplayList* aList,
|
|
uint16_t aIndex)
|
|
: nsDisplayOwnLayer(aBuilder, aFrame, aList,
|
|
aBuilder->CurrentActiveScrolledRoot()),
|
|
mContainerASR(nullptr), // XXX maybe this should be something?
|
|
mIndex(aIndex),
|
|
mIsFixedBackground(true) {
|
|
MOZ_COUNT_CTOR(nsDisplayFixedPosition);
|
|
Init(aBuilder);
|
|
}
|
|
|
|
void nsDisplayFixedPosition::Init(nsDisplayListBuilder* aBuilder) {
|
|
mAnimatedGeometryRootForScrollMetadata = mAnimatedGeometryRoot;
|
|
if (ShouldFixToViewport(aBuilder)) {
|
|
mAnimatedGeometryRoot = aBuilder->FindAnimatedGeometryRootFor(this);
|
|
}
|
|
}
|
|
|
|
/* static */
|
|
nsDisplayFixedPosition* nsDisplayFixedPosition::CreateForFixedBackground(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame,
|
|
nsDisplayBackgroundImage* aImage, uint16_t aIndex) {
|
|
nsDisplayList temp;
|
|
temp.AppendToTop(aImage);
|
|
|
|
return MakeDisplayItem<nsDisplayFixedPosition>(aBuilder, aFrame, &temp,
|
|
aIndex + 1);
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayFixedPosition::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
RefPtr<Layer> layer =
|
|
nsDisplayOwnLayer::BuildLayer(aBuilder, aManager, aContainerParameters);
|
|
|
|
layer->SetIsFixedPosition(true);
|
|
|
|
nsPresContext* presContext = mFrame->PresContext();
|
|
nsIFrame* fixedFrame =
|
|
mIsFixedBackground ? presContext->PresShell()->GetRootFrame() : mFrame;
|
|
|
|
const nsIFrame* viewportFrame = fixedFrame->GetParent();
|
|
// anchorRect will be in the container's coordinate system (aLayer's parent
|
|
// layer). This is the same as the display items' reference frame.
|
|
nsRect anchorRect;
|
|
if (viewportFrame) {
|
|
anchorRect.SizeTo(viewportFrame->GetSize());
|
|
// Fixed position frames are reflowed into the scroll-port size if one has
|
|
// been set.
|
|
if (const ViewportFrame* viewport = do_QueryFrame(viewportFrame)) {
|
|
anchorRect.SizeTo(
|
|
viewport->AdjustViewportSizeForFixedPosition(anchorRect));
|
|
}
|
|
} else {
|
|
// A display item directly attached to the viewport.
|
|
// For background-attachment:fixed items, the anchor point is always the
|
|
// top-left of the viewport currently.
|
|
viewportFrame = fixedFrame;
|
|
}
|
|
// The anchorRect top-left is always the viewport top-left.
|
|
anchorRect.MoveTo(viewportFrame->GetOffsetToCrossDoc(ReferenceFrame()));
|
|
|
|
nsLayoutUtils::SetFixedPositionLayerData(layer, viewportFrame, anchorRect,
|
|
fixedFrame, presContext,
|
|
aContainerParameters);
|
|
|
|
return layer.forget();
|
|
}
|
|
|
|
ViewID nsDisplayFixedPosition::GetScrollTargetId() {
|
|
if (mContainerASR && !nsLayoutUtils::IsReallyFixedPos(mFrame)) {
|
|
return mContainerASR->GetViewId();
|
|
}
|
|
return nsLayoutUtils::ScrollIdForRootScrollFrame(mFrame->PresContext());
|
|
}
|
|
|
|
bool nsDisplayFixedPosition::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
SideBits sides = SideBits::eNone;
|
|
if (!mIsFixedBackground) {
|
|
sides = nsLayoutUtils::GetSideBitsForFixedPositionContent(mFrame);
|
|
}
|
|
|
|
// We install this RAII scrolltarget tracker so that any
|
|
// nsDisplayCompositorHitTestInfo items inside this fixed-pos item (and that
|
|
// share the same ASR as this item) use the correct scroll target. That way
|
|
// attempts to scroll on those items will scroll the root scroll frame.
|
|
mozilla::wr::DisplayListBuilder::FixedPosScrollTargetTracker tracker(
|
|
aBuilder, GetActiveScrolledRoot(), GetScrollTargetId(), sides);
|
|
return nsDisplayOwnLayer::CreateWebRenderCommands(
|
|
aBuilder, aResources, aSc, aManager, aDisplayListBuilder);
|
|
}
|
|
|
|
bool nsDisplayFixedPosition::UpdateScrollData(
|
|
mozilla::layers::WebRenderScrollData* aData,
|
|
mozilla::layers::WebRenderLayerScrollData* aLayerData) {
|
|
if (aLayerData) {
|
|
if (!mIsFixedBackground) {
|
|
aLayerData->SetFixedPositionSides(
|
|
nsLayoutUtils::GetSideBitsForFixedPositionContent(mFrame));
|
|
}
|
|
aLayerData->SetFixedPositionScrollContainerId(GetScrollTargetId());
|
|
}
|
|
return nsDisplayOwnLayer::UpdateScrollData(aData, aLayerData) | true;
|
|
}
|
|
|
|
void nsDisplayFixedPosition::WriteDebugInfo(std::stringstream& aStream) {
|
|
aStream << nsPrintfCString(" (containerASR %s) (scrolltarget %" PRIu64 ")",
|
|
ActiveScrolledRoot::ToString(mContainerASR).get(),
|
|
GetScrollTargetId())
|
|
.get();
|
|
}
|
|
|
|
TableType GetTableTypeFromFrame(nsIFrame* aFrame) {
|
|
if (aFrame->IsTableFrame()) {
|
|
return TableType::Table;
|
|
}
|
|
|
|
if (aFrame->IsTableColFrame()) {
|
|
return TableType::TableCol;
|
|
}
|
|
|
|
if (aFrame->IsTableColGroupFrame()) {
|
|
return TableType::TableColGroup;
|
|
}
|
|
|
|
if (aFrame->IsTableRowFrame()) {
|
|
return TableType::TableRow;
|
|
}
|
|
|
|
if (aFrame->IsTableRowGroupFrame()) {
|
|
return TableType::TableRowGroup;
|
|
}
|
|
|
|
if (aFrame->IsTableCellFrame()) {
|
|
return TableType::TableCell;
|
|
}
|
|
|
|
MOZ_ASSERT_UNREACHABLE("Invalid frame.");
|
|
return TableType::Table;
|
|
}
|
|
|
|
nsDisplayTableFixedPosition::nsDisplayTableFixedPosition(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
uint16_t aIndex, nsIFrame* aAncestorFrame)
|
|
: nsDisplayFixedPosition(aBuilder, aFrame, aList, aIndex),
|
|
mAncestorFrame(aAncestorFrame),
|
|
mTableType(GetTableTypeFromFrame(aAncestorFrame)) {
|
|
if (aBuilder->IsRetainingDisplayList()) {
|
|
mAncestorFrame->AddDisplayItem(this);
|
|
}
|
|
}
|
|
|
|
/* static */
|
|
nsDisplayTableFixedPosition*
|
|
nsDisplayTableFixedPosition::CreateForFixedBackground(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame,
|
|
nsDisplayBackgroundImage* aImage, uint16_t aIndex,
|
|
nsIFrame* aAncestorFrame) {
|
|
nsDisplayList temp;
|
|
temp.AppendToTop(aImage);
|
|
|
|
return MakeDisplayItem<nsDisplayTableFixedPosition>(
|
|
aBuilder, aFrame, &temp, aIndex + 1, aAncestorFrame);
|
|
}
|
|
|
|
nsDisplayStickyPosition::nsDisplayStickyPosition(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot,
|
|
const ActiveScrolledRoot* aContainerASR)
|
|
: nsDisplayOwnLayer(aBuilder, aFrame, aList, aActiveScrolledRoot),
|
|
mContainerASR(aContainerASR) {
|
|
MOZ_COUNT_CTOR(nsDisplayStickyPosition);
|
|
}
|
|
|
|
void nsDisplayStickyPosition::SetClipChain(
|
|
const DisplayItemClipChain* aClipChain, bool aStore) {
|
|
mClipChain = aClipChain;
|
|
mClip = nullptr;
|
|
|
|
MOZ_ASSERT(!mClip,
|
|
"There should never be a clip on this item because no clip moves "
|
|
"with it.");
|
|
|
|
if (aStore) {
|
|
mState.mClipChain = aClipChain;
|
|
mState.mClip = mClip;
|
|
}
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayStickyPosition::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
RefPtr<Layer> layer =
|
|
nsDisplayOwnLayer::BuildLayer(aBuilder, aManager, aContainerParameters);
|
|
|
|
StickyScrollContainer* stickyScrollContainer =
|
|
StickyScrollContainer::GetStickyScrollContainerForFrame(mFrame);
|
|
if (!stickyScrollContainer) {
|
|
return layer.forget();
|
|
}
|
|
|
|
nsIFrame* scrollFrame = do_QueryFrame(stickyScrollContainer->ScrollFrame());
|
|
nsPresContext* presContext = scrollFrame->PresContext();
|
|
|
|
// Sticky position frames whose scroll frame is the root scroll frame are
|
|
// reflowed into the scroll-port size if one has been set.
|
|
nsSize scrollFrameSize = scrollFrame->GetSize();
|
|
if (scrollFrame == presContext->PresShell()->GetRootScrollFrame() &&
|
|
presContext->PresShell()->IsVisualViewportSizeSet()) {
|
|
scrollFrameSize = presContext->PresShell()->GetVisualViewportSize();
|
|
}
|
|
|
|
nsLayoutUtils::SetFixedPositionLayerData(
|
|
layer, scrollFrame,
|
|
nsRect(scrollFrame->GetOffsetToCrossDoc(ReferenceFrame()),
|
|
scrollFrameSize),
|
|
mFrame, presContext, aContainerParameters);
|
|
|
|
ViewID scrollId = nsLayoutUtils::FindOrCreateIDFor(
|
|
stickyScrollContainer->ScrollFrame()->GetScrolledFrame()->GetContent());
|
|
|
|
float factor = presContext->AppUnitsPerDevPixel();
|
|
nsRectAbsolute outer;
|
|
nsRectAbsolute inner;
|
|
stickyScrollContainer->GetScrollRanges(mFrame, &outer, &inner);
|
|
LayerRectAbsolute stickyOuter(
|
|
NSAppUnitsToFloatPixels(outer.X(), factor) * aContainerParameters.mXScale,
|
|
NSAppUnitsToFloatPixels(outer.Y(), factor) * aContainerParameters.mYScale,
|
|
NSAppUnitsToFloatPixels(outer.XMost(), factor) *
|
|
aContainerParameters.mXScale,
|
|
NSAppUnitsToFloatPixels(outer.YMost(), factor) *
|
|
aContainerParameters.mYScale);
|
|
LayerRectAbsolute stickyInner(
|
|
NSAppUnitsToFloatPixels(inner.X(), factor) * aContainerParameters.mXScale,
|
|
NSAppUnitsToFloatPixels(inner.Y(), factor) * aContainerParameters.mYScale,
|
|
NSAppUnitsToFloatPixels(inner.XMost(), factor) *
|
|
aContainerParameters.mXScale,
|
|
NSAppUnitsToFloatPixels(inner.YMost(), factor) *
|
|
aContainerParameters.mYScale);
|
|
layer->SetStickyPositionData(scrollId, stickyOuter, stickyInner);
|
|
|
|
return layer.forget();
|
|
}
|
|
|
|
// Returns the smallest distance from "0" to the range [min, max] where
|
|
// min <= max.
|
|
static nscoord DistanceToRange(nscoord min, nscoord max) {
|
|
MOZ_ASSERT(min <= max);
|
|
if (max < 0) {
|
|
return max;
|
|
}
|
|
if (min > 0) {
|
|
return min;
|
|
}
|
|
MOZ_ASSERT(min <= 0 && max >= 0);
|
|
return 0;
|
|
}
|
|
|
|
bool nsDisplayStickyPosition::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc, RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
StickyScrollContainer* stickyScrollContainer =
|
|
StickyScrollContainer::GetStickyScrollContainerForFrame(mFrame);
|
|
if (stickyScrollContainer) {
|
|
// If there's no ASR for the scrollframe that this sticky item is attached
|
|
// to, then don't create a WR sticky item for it either. Trying to do so
|
|
// will end in sadness because WR will interpret some coordinates as
|
|
// relative to the nearest enclosing scrollframe, which will correspond
|
|
// to the nearest ancestor ASR on the gecko side. That ASR will not be the
|
|
// same as the scrollframe this sticky item is actually supposed to be
|
|
// attached to, thus the sadness.
|
|
// Not sending WR the sticky item is ok, because the enclosing scrollframe
|
|
// will never be asynchronously scrolled. Instead we will always position
|
|
// the sticky items correctly on the gecko side and WR will never need to
|
|
// adjust their position itself.
|
|
if (!stickyScrollContainer->ScrollFrame()
|
|
->IsMaybeAsynchronouslyScrolled()) {
|
|
stickyScrollContainer = nullptr;
|
|
}
|
|
}
|
|
|
|
Maybe<wr::SpaceAndClipChainHelper> saccHelper;
|
|
|
|
if (stickyScrollContainer) {
|
|
float auPerDevPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
|
|
bool snap;
|
|
nsRect itemBounds = GetBounds(aDisplayListBuilder, &snap);
|
|
|
|
Maybe<float> topMargin;
|
|
Maybe<float> rightMargin;
|
|
Maybe<float> bottomMargin;
|
|
Maybe<float> leftMargin;
|
|
wr::StickyOffsetBounds vBounds = {0.0, 0.0};
|
|
wr::StickyOffsetBounds hBounds = {0.0, 0.0};
|
|
nsPoint appliedOffset;
|
|
|
|
nsRectAbsolute outer;
|
|
nsRectAbsolute inner;
|
|
stickyScrollContainer->GetScrollRanges(mFrame, &outer, &inner);
|
|
|
|
nsIFrame* scrollFrame = do_QueryFrame(stickyScrollContainer->ScrollFrame());
|
|
nsPoint offset = scrollFrame->GetOffsetToCrossDoc(ReferenceFrame());
|
|
|
|
// Adjust the scrollPort coordinates to be relative to the reference frame,
|
|
// so that it is in the same space as everything else.
|
|
nsRect scrollPort =
|
|
stickyScrollContainer->ScrollFrame()->GetScrollPortRect();
|
|
scrollPort += offset;
|
|
|
|
// The following computations make more sense upon understanding the
|
|
// semantics of "inner" and "outer", which is explained in the comment on
|
|
// SetStickyPositionData in Layers.h.
|
|
|
|
if (outer.YMost() != inner.YMost()) {
|
|
// Question: How far will itemBounds.y be from the top of the scrollport
|
|
// when we have scrolled from the current scroll position of "0" to
|
|
// reach the range [inner.YMost(), outer.YMost()] where the item gets
|
|
// stuck?
|
|
// Answer: the current distance is "itemBounds.y - scrollPort.y". That
|
|
// needs to be adjusted by the distance to the range. If the distance is
|
|
// negative (i.e. inner.YMost() <= outer.YMost() < 0) then we would be
|
|
// scrolling upwards (decreasing scroll offset) to reach that range,
|
|
// which would increase itemBounds.y and make it farther away from the
|
|
// top of the scrollport. So in that case the adjustment is -distance.
|
|
// If the distance is positive (0 < inner.YMost() <= outer.YMost()) then
|
|
// we would be scrolling downwards, itemBounds.y would decrease, and we
|
|
// again need to adjust by -distance. If we are already in the range
|
|
// then no adjustment is needed and distance is 0 so again using
|
|
// -distance works.
|
|
nscoord distance = DistanceToRange(inner.YMost(), outer.YMost());
|
|
topMargin = Some(NSAppUnitsToFloatPixels(
|
|
itemBounds.y - scrollPort.y - distance, auPerDevPixel));
|
|
// Question: What is the maximum positive ("downward") offset that WR
|
|
// will have to apply to this item in order to prevent the item from
|
|
// visually moving?
|
|
// Answer: Since the item is "sticky" in the range [inner.YMost(),
|
|
// outer.YMost()], the maximum offset will be the size of the range, which
|
|
// is outer.YMost() - inner.YMost().
|
|
vBounds.max =
|
|
NSAppUnitsToFloatPixels(outer.YMost() - inner.YMost(), auPerDevPixel);
|
|
// Question: how much of an offset has layout already applied to the item?
|
|
// Answer: if we are
|
|
// (a) inside the sticky range (inner.YMost() < 0 <= outer.YMost()), or
|
|
// (b) past the sticky range (inner.YMost() < outer.YMost() < 0)
|
|
// then layout has already applied some offset to the position of the
|
|
// item. The amount of the adjustment is |0 - inner.YMost()| in case (a)
|
|
// and |outer.YMost() - inner.YMost()| in case (b).
|
|
if (inner.YMost() < 0) {
|
|
appliedOffset.y = std::min(0, outer.YMost()) - inner.YMost();
|
|
MOZ_ASSERT(appliedOffset.y > 0);
|
|
}
|
|
}
|
|
if (outer.Y() != inner.Y()) {
|
|
// Similar logic as in the previous section, but this time we care about
|
|
// the distance from itemBounds.YMost() to scrollPort.YMost().
|
|
nscoord distance = DistanceToRange(outer.Y(), inner.Y());
|
|
bottomMargin = Some(NSAppUnitsToFloatPixels(
|
|
scrollPort.YMost() - itemBounds.YMost() + distance, auPerDevPixel));
|
|
// And here WR will be moving the item upwards rather than downwards so
|
|
// again things are inverted from the previous block.
|
|
vBounds.min =
|
|
NSAppUnitsToFloatPixels(outer.Y() - inner.Y(), auPerDevPixel);
|
|
// We can't have appliedOffset be both positive and negative, and the top
|
|
// adjustment takes priority. So here we only update appliedOffset.y if
|
|
// it wasn't set by the top-sticky case above.
|
|
if (appliedOffset.y == 0 && inner.Y() > 0) {
|
|
appliedOffset.y = std::max(0, outer.Y()) - inner.Y();
|
|
MOZ_ASSERT(appliedOffset.y < 0);
|
|
}
|
|
}
|
|
// Same as above, but for the x-axis
|
|
if (outer.XMost() != inner.XMost()) {
|
|
nscoord distance = DistanceToRange(inner.XMost(), outer.XMost());
|
|
leftMargin = Some(NSAppUnitsToFloatPixels(
|
|
itemBounds.x - scrollPort.x - distance, auPerDevPixel));
|
|
hBounds.max =
|
|
NSAppUnitsToFloatPixels(outer.XMost() - inner.XMost(), auPerDevPixel);
|
|
if (inner.XMost() < 0) {
|
|
appliedOffset.x = std::min(0, outer.XMost()) - inner.XMost();
|
|
MOZ_ASSERT(appliedOffset.x > 0);
|
|
}
|
|
}
|
|
if (outer.X() != inner.X()) {
|
|
nscoord distance = DistanceToRange(outer.X(), inner.X());
|
|
rightMargin = Some(NSAppUnitsToFloatPixels(
|
|
scrollPort.XMost() - itemBounds.XMost() + distance, auPerDevPixel));
|
|
hBounds.min =
|
|
NSAppUnitsToFloatPixels(outer.X() - inner.X(), auPerDevPixel);
|
|
if (appliedOffset.x == 0 && inner.X() > 0) {
|
|
appliedOffset.x = std::max(0, outer.X()) - inner.X();
|
|
MOZ_ASSERT(appliedOffset.x < 0);
|
|
}
|
|
}
|
|
|
|
LayoutDeviceRect bounds =
|
|
LayoutDeviceRect::FromAppUnits(itemBounds, auPerDevPixel);
|
|
wr::LayoutVector2D applied = {
|
|
NSAppUnitsToFloatPixels(appliedOffset.x, auPerDevPixel),
|
|
NSAppUnitsToFloatPixels(appliedOffset.y, auPerDevPixel)};
|
|
wr::WrSpatialId spatialId = aBuilder.DefineStickyFrame(
|
|
wr::ToLayoutRect(bounds), topMargin.ptrOr(nullptr),
|
|
rightMargin.ptrOr(nullptr), bottomMargin.ptrOr(nullptr),
|
|
leftMargin.ptrOr(nullptr), vBounds, hBounds, applied);
|
|
|
|
saccHelper.emplace(aBuilder, spatialId);
|
|
aManager->CommandBuilder().PushOverrideForASR(mContainerASR, spatialId);
|
|
}
|
|
|
|
{
|
|
wr::StackingContextParams params;
|
|
params.clip =
|
|
wr::WrStackingContextClip::ClipChain(aBuilder.CurrentClipChainId());
|
|
StackingContextHelper sc(aSc, GetActiveScrolledRoot(), mFrame, this,
|
|
aBuilder, params);
|
|
nsDisplayWrapList::CreateWebRenderCommands(aBuilder, aResources, sc,
|
|
aManager, aDisplayListBuilder);
|
|
}
|
|
|
|
if (stickyScrollContainer) {
|
|
aManager->CommandBuilder().PopOverrideForASR(mContainerASR);
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
nsDisplayScrollInfoLayer::nsDisplayScrollInfoLayer(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aScrolledFrame,
|
|
nsIFrame* aScrollFrame)
|
|
: nsDisplayWrapList(aBuilder, aScrollFrame),
|
|
mScrollFrame(aScrollFrame),
|
|
mScrolledFrame(aScrolledFrame),
|
|
mScrollParentId(aBuilder->GetCurrentScrollParentId()) {
|
|
#ifdef NS_BUILD_REFCNT_LOGGING
|
|
MOZ_COUNT_CTOR(nsDisplayScrollInfoLayer);
|
|
#endif
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayScrollInfoLayer::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
// In general for APZ with event-regions we no longer have a need for
|
|
// scrollinfo layers. However, in some cases, there might be content that
|
|
// cannot be layerized, and so needs to scroll synchronously. To handle those
|
|
// cases, we still want to generate scrollinfo layers.
|
|
|
|
return aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, &mList, aContainerParameters, nullptr,
|
|
FrameLayerBuilder::CONTAINER_ALLOW_PULL_BACKGROUND_COLOR);
|
|
}
|
|
|
|
LayerState nsDisplayScrollInfoLayer::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
return LayerState::LAYER_ACTIVE_EMPTY;
|
|
}
|
|
|
|
UniquePtr<ScrollMetadata> nsDisplayScrollInfoLayer::ComputeScrollMetadata(
|
|
LayerManager* aLayerManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
nsRect viewport = mScrollFrame->GetRect() - mScrollFrame->GetPosition() +
|
|
mScrollFrame->GetOffsetToCrossDoc(ReferenceFrame());
|
|
|
|
ScrollMetadata metadata = nsLayoutUtils::ComputeScrollMetadata(
|
|
mScrolledFrame, mScrollFrame, mScrollFrame->GetContent(),
|
|
ReferenceFrame(), aLayerManager, mScrollParentId, viewport, Nothing(),
|
|
false, Some(aContainerParameters));
|
|
metadata.GetMetrics().SetIsScrollInfoLayer(true);
|
|
nsIScrollableFrame* scrollableFrame = mScrollFrame->GetScrollTargetFrame();
|
|
if (scrollableFrame) {
|
|
scrollableFrame->NotifyApzTransaction();
|
|
}
|
|
|
|
return UniquePtr<ScrollMetadata>(new ScrollMetadata(metadata));
|
|
}
|
|
|
|
bool nsDisplayScrollInfoLayer::UpdateScrollData(
|
|
mozilla::layers::WebRenderScrollData* aData,
|
|
mozilla::layers::WebRenderLayerScrollData* aLayerData) {
|
|
if (aLayerData) {
|
|
UniquePtr<ScrollMetadata> metadata =
|
|
ComputeScrollMetadata(aData->GetManager(), ContainerLayerParameters());
|
|
MOZ_ASSERT(aData);
|
|
MOZ_ASSERT(metadata);
|
|
aLayerData->AppendScrollMetadata(*aData, *metadata);
|
|
}
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayScrollInfoLayer::WriteDebugInfo(std::stringstream& aStream) {
|
|
aStream << " (scrollframe " << mScrollFrame << " scrolledFrame "
|
|
<< mScrolledFrame << ")";
|
|
}
|
|
|
|
nsDisplayZoom::nsDisplayZoom(nsDisplayListBuilder* aBuilder, nsIFrame* aFrame,
|
|
nsSubDocumentFrame* aSubDocFrame,
|
|
nsDisplayList* aList, int32_t aAPD,
|
|
int32_t aParentAPD, nsDisplayOwnLayerFlags aFlags)
|
|
: nsDisplaySubDocument(aBuilder, aFrame, aSubDocFrame, aList, aFlags),
|
|
mAPD(aAPD),
|
|
mParentAPD(aParentAPD) {
|
|
MOZ_COUNT_CTOR(nsDisplayZoom);
|
|
}
|
|
|
|
nsRect nsDisplayZoom::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
nsRect bounds = nsDisplaySubDocument::GetBounds(aBuilder, aSnap);
|
|
*aSnap = false;
|
|
return bounds.ScaleToOtherAppUnitsRoundOut(mAPD, mParentAPD);
|
|
}
|
|
|
|
void nsDisplayZoom::HitTest(nsDisplayListBuilder* aBuilder, const nsRect& aRect,
|
|
HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) {
|
|
nsRect rect;
|
|
// A 1x1 rect indicates we are just hit testing a point, so pass down a 1x1
|
|
// rect as well instead of possibly rounding the width or height to zero.
|
|
if (aRect.width == 1 && aRect.height == 1) {
|
|
rect.MoveTo(aRect.TopLeft().ScaleToOtherAppUnits(mParentAPD, mAPD));
|
|
rect.width = rect.height = 1;
|
|
} else {
|
|
rect = aRect.ScaleToOtherAppUnitsRoundOut(mParentAPD, mAPD);
|
|
}
|
|
mList.HitTest(aBuilder, rect, aState, aOutFrames);
|
|
}
|
|
|
|
bool nsDisplayZoom::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
// Convert the passed in visible region to our appunits.
|
|
nsRegion visibleRegion;
|
|
// mVisibleRect has been clipped to GetClippedBounds
|
|
visibleRegion.And(*aVisibleRegion, GetPaintRect());
|
|
visibleRegion = visibleRegion.ScaleToOtherAppUnitsRoundOut(mParentAPD, mAPD);
|
|
nsRegion originalVisibleRegion = visibleRegion;
|
|
|
|
nsRect transformedVisibleRect =
|
|
GetPaintRect().ScaleToOtherAppUnitsRoundOut(mParentAPD, mAPD);
|
|
bool retval;
|
|
// If we are to generate a scrollable layer we call
|
|
// nsDisplaySubDocument::ComputeVisibility to make the necessary adjustments
|
|
// for ComputeVisibility, it does all it's calculations in the child APD.
|
|
bool usingDisplayPort = UseDisplayPortForViewport(aBuilder, mFrame);
|
|
if (!(mFlags & nsDisplayOwnLayerFlags::GenerateScrollableLayer) ||
|
|
!usingDisplayPort) {
|
|
retval = mList.ComputeVisibilityForSublist(aBuilder, &visibleRegion,
|
|
transformedVisibleRect);
|
|
} else {
|
|
retval = nsDisplaySubDocument::ComputeVisibility(aBuilder, &visibleRegion);
|
|
}
|
|
|
|
nsRegion removed;
|
|
// removed = originalVisibleRegion - visibleRegion
|
|
removed.Sub(originalVisibleRegion, visibleRegion);
|
|
// Convert removed region to parent appunits.
|
|
removed = removed.ScaleToOtherAppUnitsRoundIn(mAPD, mParentAPD);
|
|
// aVisibleRegion = aVisibleRegion - removed (modulo any simplifications
|
|
// SubtractFromVisibleRegion does)
|
|
aBuilder->SubtractFromVisibleRegion(aVisibleRegion, removed);
|
|
|
|
return retval;
|
|
}
|
|
|
|
nsDisplayAsyncZoom::nsDisplayAsyncZoom(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot,
|
|
mozilla::layers::FrameMetrics::ViewID aViewID)
|
|
: nsDisplayOwnLayer(aBuilder, aFrame, aList, aActiveScrolledRoot),
|
|
mViewID(aViewID) {
|
|
MOZ_COUNT_CTOR(nsDisplayAsyncZoom);
|
|
}
|
|
|
|
#ifdef NS_BUILD_REFCNT_LOGGING
|
|
nsDisplayAsyncZoom::~nsDisplayAsyncZoom() {
|
|
MOZ_COUNT_DTOR(nsDisplayAsyncZoom);
|
|
}
|
|
#endif
|
|
|
|
already_AddRefed<Layer> nsDisplayAsyncZoom::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
PresShell* presShell = mFrame->PresShell();
|
|
ContainerLayerParameters containerParameters(
|
|
presShell->GetResolution(), presShell->GetResolution(), nsIntPoint(),
|
|
aContainerParameters);
|
|
|
|
RefPtr<Layer> layer =
|
|
nsDisplayOwnLayer::BuildLayer(aBuilder, aManager, containerParameters);
|
|
|
|
layer->SetIsAsyncZoomContainer(Some(mViewID));
|
|
|
|
layer->SetPostScale(1.0f / presShell->GetResolution(),
|
|
1.0f / presShell->GetResolution());
|
|
layer->AsContainerLayer()->SetScaleToResolution(presShell->GetResolution());
|
|
|
|
return layer.forget();
|
|
}
|
|
|
|
bool nsDisplayAsyncZoom::UpdateScrollData(
|
|
mozilla::layers::WebRenderScrollData* aData,
|
|
mozilla::layers::WebRenderLayerScrollData* aLayerData) {
|
|
bool ret = nsDisplayOwnLayer::UpdateScrollData(aData, aLayerData);
|
|
MOZ_ASSERT(ret);
|
|
if (aLayerData) {
|
|
aLayerData->SetAsyncZoomContainerId(mViewID);
|
|
}
|
|
return ret;
|
|
}
|
|
|
|
///////////////////////////////////////////////////
|
|
// nsDisplayTransform Implementation
|
|
//
|
|
|
|
#ifndef DEBUG
|
|
static_assert(sizeof(nsDisplayTransform) < 512, "nsDisplayTransform has grown");
|
|
#endif
|
|
|
|
nsDisplayTransform::nsDisplayTransform(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame, nsDisplayList* aList,
|
|
const nsRect& aChildrenBuildingRect,
|
|
uint16_t aIndex)
|
|
: nsDisplayHitTestInfoBase(aBuilder, aFrame),
|
|
mTransform(Some(Matrix4x4())),
|
|
mTransformGetter(nullptr),
|
|
mAnimatedGeometryRootForChildren(mAnimatedGeometryRoot),
|
|
mAnimatedGeometryRootForScrollMetadata(mAnimatedGeometryRoot),
|
|
mChildrenBuildingRect(aChildrenBuildingRect),
|
|
mIndex(aIndex),
|
|
mIsTransformSeparator(true),
|
|
mAllowAsyncAnimation(false) {
|
|
MOZ_COUNT_CTOR(nsDisplayTransform);
|
|
MOZ_ASSERT(aFrame, "Must have a frame!");
|
|
Init(aBuilder, aList);
|
|
}
|
|
|
|
nsDisplayTransform::nsDisplayTransform(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame, nsDisplayList* aList,
|
|
const nsRect& aChildrenBuildingRect,
|
|
uint16_t aIndex,
|
|
bool aAllowAsyncAnimation)
|
|
: nsDisplayHitTestInfoBase(aBuilder, aFrame),
|
|
mTransformGetter(nullptr),
|
|
mAnimatedGeometryRootForChildren(mAnimatedGeometryRoot),
|
|
mAnimatedGeometryRootForScrollMetadata(mAnimatedGeometryRoot),
|
|
mChildrenBuildingRect(aChildrenBuildingRect),
|
|
mIndex(aIndex),
|
|
mIsTransformSeparator(false),
|
|
mAllowAsyncAnimation(aAllowAsyncAnimation) {
|
|
MOZ_COUNT_CTOR(nsDisplayTransform);
|
|
MOZ_ASSERT(aFrame, "Must have a frame!");
|
|
SetReferenceFrameToAncestor(aBuilder);
|
|
Init(aBuilder, aList);
|
|
}
|
|
|
|
nsDisplayTransform::nsDisplayTransform(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const nsRect& aChildrenBuildingRect, uint16_t aIndex,
|
|
ComputeTransformFunction aTransformGetter)
|
|
: nsDisplayHitTestInfoBase(aBuilder, aFrame),
|
|
mTransformGetter(aTransformGetter),
|
|
mAnimatedGeometryRootForChildren(mAnimatedGeometryRoot),
|
|
mAnimatedGeometryRootForScrollMetadata(mAnimatedGeometryRoot),
|
|
mChildrenBuildingRect(aChildrenBuildingRect),
|
|
mIndex(aIndex),
|
|
mIsTransformSeparator(false),
|
|
mAllowAsyncAnimation(false) {
|
|
MOZ_COUNT_CTOR(nsDisplayTransform);
|
|
MOZ_ASSERT(aFrame, "Must have a frame!");
|
|
Init(aBuilder, aList);
|
|
}
|
|
|
|
void nsDisplayTransform::SetReferenceFrameToAncestor(
|
|
nsDisplayListBuilder* aBuilder) {
|
|
if (mFrame == aBuilder->RootReferenceFrame()) {
|
|
return;
|
|
}
|
|
nsIFrame* outerFrame = nsLayoutUtils::GetCrossDocParentFrame(mFrame);
|
|
mReferenceFrame = aBuilder->FindReferenceFrameFor(outerFrame);
|
|
mToReferenceFrame = mFrame->GetOffsetToCrossDoc(mReferenceFrame);
|
|
if (nsLayoutUtils::IsFixedPosFrameInDisplayPort(mFrame)) {
|
|
// This is an odd special case. If we are both IsFixedPosFrameInDisplayPort
|
|
// and transformed that we are our own AGR parent.
|
|
// We want our frame to be our AGR because FrameLayerBuilder uses our AGR to
|
|
// determine if we are inside a fixed pos subtree. If we use the outer AGR
|
|
// from outside the fixed pos subtree FLB can't tell that we are fixed pos.
|
|
mAnimatedGeometryRoot = mAnimatedGeometryRootForChildren;
|
|
} else if (mFrame->StyleDisplay()->mPosition ==
|
|
StylePositionProperty::Sticky &&
|
|
IsStickyFrameActive(aBuilder, mFrame, nullptr)) {
|
|
// Similar to the IsFixedPosFrameInDisplayPort case we are our own AGR.
|
|
// We are inside the sticky position, so our AGR is the sticky positioned
|
|
// frame, which is our AGR, not the parent AGR.
|
|
mAnimatedGeometryRoot = mAnimatedGeometryRootForChildren;
|
|
} else if (mAnimatedGeometryRoot->mParentAGR) {
|
|
mAnimatedGeometryRootForScrollMetadata = mAnimatedGeometryRoot->mParentAGR;
|
|
if (!MayBeAnimated(aBuilder)) {
|
|
// If we're an animated transform then we want the same AGR as our
|
|
// children so that FrameLayerBuilder knows that this layer moves with the
|
|
// transform and won't compute occlusions. If we're not animated then use
|
|
// our parent AGR so that inactive transform layers can go in the same
|
|
// PaintedLayer as surrounding content.
|
|
mAnimatedGeometryRoot = mAnimatedGeometryRoot->mParentAGR;
|
|
}
|
|
}
|
|
|
|
SetBuildingRect(aBuilder->GetVisibleRect() + mToReferenceFrame);
|
|
}
|
|
|
|
void nsDisplayTransform::Init(nsDisplayListBuilder* aBuilder,
|
|
nsDisplayList* aChildren) {
|
|
mShouldFlatten = false;
|
|
mChildren.AppendToTop(aChildren);
|
|
UpdateBounds(aBuilder);
|
|
}
|
|
|
|
bool nsDisplayTransform::ShouldFlattenAway(nsDisplayListBuilder* aBuilder) {
|
|
if (gfxVars::UseWebRender() ||
|
|
!StaticPrefs::layout_display_list_flatten_transform()) {
|
|
return false;
|
|
}
|
|
|
|
MOZ_ASSERT(!mShouldFlatten);
|
|
mShouldFlatten = GetTransform().Is2D();
|
|
return mShouldFlatten;
|
|
}
|
|
|
|
/* Returns the delta specified by the transform-origin property.
|
|
* This is a positive delta, meaning that it indicates the direction to move
|
|
* to get from (0, 0) of the frame to the transform origin. This function is
|
|
* called off the main thread.
|
|
*/
|
|
/* static */
|
|
Point3D nsDisplayTransform::GetDeltaToTransformOrigin(
|
|
const nsIFrame* aFrame, TransformReferenceBox& aRefBox,
|
|
float aAppUnitsPerPixel) {
|
|
MOZ_ASSERT(aFrame, "Can't get delta for a null frame!");
|
|
MOZ_ASSERT(aFrame->IsTransformed() || aFrame->BackfaceIsHidden() ||
|
|
aFrame->Combines3DTransformWithAncestors(),
|
|
"Shouldn't get a delta for an untransformed frame!");
|
|
|
|
if (!aFrame->IsTransformed()) {
|
|
return Point3D();
|
|
}
|
|
|
|
/* For both of the coordinates, if the value of transform is a
|
|
* percentage, it's relative to the size of the frame. Otherwise, if it's
|
|
* a distance, it's already computed for us!
|
|
*/
|
|
const nsStyleDisplay* display = aFrame->StyleDisplay();
|
|
|
|
const StyleTransformOrigin& transformOrigin = display->mTransformOrigin;
|
|
CSSPoint origin = nsStyleTransformMatrix::Convert2DPosition(
|
|
transformOrigin.horizontal, transformOrigin.vertical, aRefBox);
|
|
|
|
if (aFrame->GetStateBits() & NS_FRAME_SVG_LAYOUT) {
|
|
// SVG frames (unlike other frames) have a reference box that can be (and
|
|
// typically is) offset from the TopLeft() of the frame. We need to account
|
|
// for that here.
|
|
origin.x += CSSPixel::FromAppUnits(aRefBox.X());
|
|
origin.y += CSSPixel::FromAppUnits(aRefBox.Y());
|
|
}
|
|
|
|
float scale = mozilla::AppUnitsPerCSSPixel() / float(aAppUnitsPerPixel);
|
|
float z = transformOrigin.depth._0;
|
|
return Point3D(origin.x * scale, origin.y * scale, z * scale);
|
|
}
|
|
|
|
/* static */
|
|
bool nsDisplayTransform::ComputePerspectiveMatrix(const nsIFrame* aFrame,
|
|
float aAppUnitsPerPixel,
|
|
Matrix4x4& aOutMatrix) {
|
|
MOZ_ASSERT(aFrame, "Can't get delta for a null frame!");
|
|
MOZ_ASSERT(aFrame->IsTransformed() || aFrame->BackfaceIsHidden() ||
|
|
aFrame->Combines3DTransformWithAncestors(),
|
|
"Shouldn't get a delta for an untransformed frame!");
|
|
MOZ_ASSERT(aOutMatrix.IsIdentity(), "Must have a blank output matrix");
|
|
|
|
if (!aFrame->IsTransformed()) {
|
|
return false;
|
|
}
|
|
|
|
/* Find our containing block, which is the element that provides the
|
|
* value for perspective we need to use
|
|
*/
|
|
|
|
// TODO: Is it possible that the cbFrame's bounds haven't been set correctly
|
|
// yet
|
|
// (similar to the aBoundsOverride case for GetResultingTransformMatrix)?
|
|
nsIFrame* cbFrame = aFrame->GetContainingBlock(nsIFrame::SKIP_SCROLLED_FRAME);
|
|
if (!cbFrame) {
|
|
return false;
|
|
}
|
|
|
|
/* Grab the values for perspective and perspective-origin (if present) */
|
|
const nsStyleDisplay* cbDisplay = cbFrame->StyleDisplay();
|
|
if (cbDisplay->mChildPerspective.IsNone()) {
|
|
return false;
|
|
}
|
|
|
|
MOZ_ASSERT(cbDisplay->mChildPerspective.IsLength());
|
|
// TODO(emilio): Seems quite silly to go through app units just to convert to
|
|
// float pixels below.
|
|
nscoord perspective = cbDisplay->mChildPerspective.length._0.ToAppUnits();
|
|
if (perspective < std::numeric_limits<Float>::epsilon()) {
|
|
return true;
|
|
}
|
|
|
|
TransformReferenceBox refBox(cbFrame);
|
|
|
|
Point perspectiveOrigin = nsStyleTransformMatrix::Convert2DPosition(
|
|
cbDisplay->mPerspectiveOrigin.horizontal,
|
|
cbDisplay->mPerspectiveOrigin.vertical, refBox, aAppUnitsPerPixel);
|
|
|
|
/* GetOffsetTo computes the offset required to move from 0,0 in cbFrame to 0,0
|
|
* in aFrame. Although we actually want the inverse of this, it's faster to
|
|
* compute this way.
|
|
*/
|
|
nsPoint frameToCbOffset = -aFrame->GetOffsetTo(cbFrame);
|
|
Point frameToCbGfxOffset(
|
|
NSAppUnitsToFloatPixels(frameToCbOffset.x, aAppUnitsPerPixel),
|
|
NSAppUnitsToFloatPixels(frameToCbOffset.y, aAppUnitsPerPixel));
|
|
|
|
/* Move the perspective origin to be relative to aFrame, instead of relative
|
|
* to the containing block which is how it was specified in the style system.
|
|
*/
|
|
perspectiveOrigin += frameToCbGfxOffset;
|
|
|
|
aOutMatrix._34 =
|
|
-1.0 / NSAppUnitsToFloatPixels(perspective, aAppUnitsPerPixel);
|
|
|
|
aOutMatrix.ChangeBasis(Point3D(perspectiveOrigin.x, perspectiveOrigin.y, 0));
|
|
return true;
|
|
}
|
|
|
|
nsDisplayTransform::FrameTransformProperties::FrameTransformProperties(
|
|
const nsIFrame* aFrame, TransformReferenceBox& aRefBox,
|
|
float aAppUnitsPerPixel)
|
|
: mFrame(aFrame),
|
|
mTranslate(aFrame->StyleDisplay()->mTranslate),
|
|
mRotate(aFrame->StyleDisplay()->mRotate),
|
|
mScale(aFrame->StyleDisplay()->mScale),
|
|
mTransform(aFrame->StyleDisplay()->mTransform),
|
|
mMotion(MotionPathUtils::ResolveMotionPath(aFrame, aRefBox)),
|
|
mToTransformOrigin(
|
|
GetDeltaToTransformOrigin(aFrame, aRefBox, aAppUnitsPerPixel)) {}
|
|
|
|
/* Wraps up the transform matrix in a change-of-basis matrix pair that
|
|
* translates from local coordinate space to transform coordinate space, then
|
|
* hands it back.
|
|
*/
|
|
Matrix4x4 nsDisplayTransform::GetResultingTransformMatrix(
|
|
const FrameTransformProperties& aProperties, TransformReferenceBox& aRefBox,
|
|
const nsPoint& aOrigin, float aAppUnitsPerPixel, uint32_t aFlags) {
|
|
return GetResultingTransformMatrixInternal(aProperties, aRefBox, aOrigin,
|
|
aAppUnitsPerPixel, aFlags);
|
|
}
|
|
|
|
Matrix4x4 nsDisplayTransform::GetResultingTransformMatrix(
|
|
const nsIFrame* aFrame, const nsPoint& aOrigin, float aAppUnitsPerPixel,
|
|
uint32_t aFlags) {
|
|
// mRect before calling FinishAndStoreOverflow so we don't need the override.
|
|
TransformReferenceBox refBox(aFrame);
|
|
FrameTransformProperties props(aFrame, refBox, aAppUnitsPerPixel);
|
|
|
|
return GetResultingTransformMatrixInternal(props, refBox, aOrigin,
|
|
aAppUnitsPerPixel, aFlags);
|
|
}
|
|
|
|
static bool ShouldRoundTransformOrigin(const nsIFrame* aFrame) {
|
|
// An SVG frame should not have its translation rounded.
|
|
// Note it's possible that the SVG frame doesn't have an SVG
|
|
// transform but only has a CSS transform.
|
|
return !aFrame || !aFrame->HasAnyStateBits(NS_FRAME_SVG_LAYOUT) ||
|
|
aFrame->IsSVGOuterSVGAnonChildFrame();
|
|
}
|
|
|
|
Matrix4x4 nsDisplayTransform::GetResultingTransformMatrixInternal(
|
|
const FrameTransformProperties& aProperties, TransformReferenceBox& aRefBox,
|
|
const nsPoint& aOrigin, float aAppUnitsPerPixel, uint32_t aFlags) {
|
|
const nsIFrame* frame = aProperties.mFrame;
|
|
NS_ASSERTION(frame || !(aFlags & INCLUDE_PERSPECTIVE),
|
|
"Must have a frame to compute perspective!");
|
|
|
|
// Get the underlying transform matrix:
|
|
|
|
/* Get the matrix, then change its basis to factor in the origin. */
|
|
Matrix4x4 result;
|
|
// Call IsSVGTransformed() regardless of the value of
|
|
// disp->mSpecifiedTransform, since we still need any
|
|
// parentsChildrenOnlyTransform.
|
|
Matrix svgTransform, parentsChildrenOnlyTransform;
|
|
bool hasSVGTransforms =
|
|
frame &&
|
|
frame->IsSVGTransformed(&svgTransform, &parentsChildrenOnlyTransform);
|
|
|
|
bool shouldRound = ShouldRoundTransformOrigin(frame);
|
|
|
|
/* Transformed frames always have a transform, or are preserving 3d (and might
|
|
* still have perspective!) */
|
|
if (aProperties.HasTransform()) {
|
|
result = nsStyleTransformMatrix::ReadTransforms(
|
|
aProperties.mTranslate, aProperties.mRotate, aProperties.mScale,
|
|
aProperties.mMotion, aProperties.mTransform, aRefBox,
|
|
aAppUnitsPerPixel);
|
|
} else if (hasSVGTransforms) {
|
|
// Correct the translation components for zoom:
|
|
float pixelsPerCSSPx = AppUnitsPerCSSPixel() / aAppUnitsPerPixel;
|
|
svgTransform._31 *= pixelsPerCSSPx;
|
|
svgTransform._32 *= pixelsPerCSSPx;
|
|
result = Matrix4x4::From2D(svgTransform);
|
|
}
|
|
|
|
// Apply any translation due to 'transform-origin' and/or 'transform-box':
|
|
result.ChangeBasis(aProperties.mToTransformOrigin);
|
|
|
|
// See the comment for nsSVGContainerFrame::HasChildrenOnlyTransform for
|
|
// an explanation of what children-only transforms are.
|
|
bool parentHasChildrenOnlyTransform =
|
|
hasSVGTransforms && !parentsChildrenOnlyTransform.IsIdentity();
|
|
|
|
if (parentHasChildrenOnlyTransform) {
|
|
float pixelsPerCSSPx = AppUnitsPerCSSPixel() / aAppUnitsPerPixel;
|
|
parentsChildrenOnlyTransform._31 *= pixelsPerCSSPx;
|
|
parentsChildrenOnlyTransform._32 *= pixelsPerCSSPx;
|
|
|
|
Point3D frameOffset(
|
|
NSAppUnitsToFloatPixels(-frame->GetPosition().x, aAppUnitsPerPixel),
|
|
NSAppUnitsToFloatPixels(-frame->GetPosition().y, aAppUnitsPerPixel), 0);
|
|
Matrix4x4 parentsChildrenOnlyTransform3D =
|
|
Matrix4x4::From2D(parentsChildrenOnlyTransform)
|
|
.ChangeBasis(frameOffset);
|
|
|
|
result *= parentsChildrenOnlyTransform3D;
|
|
}
|
|
|
|
Matrix4x4 perspectiveMatrix;
|
|
bool hasPerspective = aFlags & INCLUDE_PERSPECTIVE;
|
|
if (hasPerspective) {
|
|
if (ComputePerspectiveMatrix(frame, aAppUnitsPerPixel, perspectiveMatrix)) {
|
|
result *= perspectiveMatrix;
|
|
}
|
|
}
|
|
|
|
if ((aFlags & INCLUDE_PRESERVE3D_ANCESTORS) && frame &&
|
|
frame->Combines3DTransformWithAncestors()) {
|
|
// Include the transform set on our parent
|
|
nsIFrame* parentFrame =
|
|
frame->GetClosestFlattenedTreeAncestorPrimaryFrame();
|
|
NS_ASSERTION(parentFrame && parentFrame->IsTransformed() &&
|
|
parentFrame->Extend3DContext(),
|
|
"Preserve3D mismatch!");
|
|
TransformReferenceBox refBox(parentFrame);
|
|
FrameTransformProperties props(parentFrame, refBox, aAppUnitsPerPixel);
|
|
|
|
uint32_t flags =
|
|
aFlags & (INCLUDE_PRESERVE3D_ANCESTORS | INCLUDE_PERSPECTIVE);
|
|
|
|
// If this frame isn't transformed (but we exist for backface-visibility),
|
|
// then we're not a reference frame so no offset to origin will be added.
|
|
// Otherwise we need to manually translate into our parent's coordinate
|
|
// space.
|
|
if (frame->IsTransformed()) {
|
|
nsLayoutUtils::PostTranslate(result, frame->GetPosition(),
|
|
aAppUnitsPerPixel, shouldRound);
|
|
}
|
|
Matrix4x4 parent = GetResultingTransformMatrixInternal(
|
|
props, aRefBox, nsPoint(0, 0), aAppUnitsPerPixel, flags);
|
|
result = result * parent;
|
|
}
|
|
|
|
if (aFlags & OFFSET_BY_ORIGIN) {
|
|
nsLayoutUtils::PostTranslate(result, aOrigin, aAppUnitsPerPixel,
|
|
shouldRound);
|
|
}
|
|
|
|
return result;
|
|
}
|
|
|
|
bool nsDisplayOpacity::CanUseAsyncAnimations(nsDisplayListBuilder* aBuilder) {
|
|
static constexpr nsCSSPropertyIDSet opacitySet =
|
|
nsCSSPropertyIDSet::OpacityProperties();
|
|
if (ActiveLayerTracker::IsStyleAnimated(aBuilder, mFrame, opacitySet)) {
|
|
return true;
|
|
}
|
|
|
|
EffectCompositor::SetPerformanceWarning(
|
|
mFrame, opacitySet,
|
|
AnimationPerformanceWarning(
|
|
AnimationPerformanceWarning::Type::OpacityFrameInactive));
|
|
|
|
return false;
|
|
}
|
|
|
|
bool nsDisplayTransform::CanUseAsyncAnimations(nsDisplayListBuilder* aBuilder) {
|
|
return mAllowAsyncAnimation;
|
|
}
|
|
|
|
bool nsDisplayBackgroundColor::CanUseAsyncAnimations(
|
|
nsDisplayListBuilder* aBuilder) {
|
|
LayerManager* layerManager = aBuilder->GetWidgetLayerManager();
|
|
return layerManager &&
|
|
layerManager->GetBackendType() != layers::LayersBackend::LAYERS_WR;
|
|
}
|
|
|
|
/* static */
|
|
auto nsDisplayTransform::ShouldPrerenderTransformedContent(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsRect* aDirtyRect)
|
|
-> PrerenderDecision {
|
|
// Elements whose transform has been modified recently, or which
|
|
// have a compositor-animated transform, can be prerendered. An element
|
|
// might have only just had its transform animated in which case
|
|
// the ActiveLayerManager may not have been notified yet.
|
|
static constexpr nsCSSPropertyIDSet transformSet =
|
|
nsCSSPropertyIDSet::TransformLikeProperties();
|
|
if (!ActiveLayerTracker::IsTransformMaybeAnimated(aFrame) &&
|
|
!EffectCompositor::HasAnimationsForCompositor(
|
|
aFrame, DisplayItemType::TYPE_TRANSFORM)) {
|
|
EffectCompositor::SetPerformanceWarning(
|
|
aFrame, transformSet,
|
|
AnimationPerformanceWarning(
|
|
AnimationPerformanceWarning::Type::TransformFrameInactive));
|
|
|
|
return NoPrerender;
|
|
}
|
|
|
|
// We should not allow prerender if any ancestor container element has
|
|
// mask/clip-path effects.
|
|
//
|
|
// With prerender and async transform animation, we do not need to restyle an
|
|
// animated element to respect position changes, since that transform is done
|
|
// by layer animation. As a result, the container element is not aware of
|
|
// position change of that containing element and loses the chance to update
|
|
// the content of mask/clip-path.
|
|
//
|
|
// Why do we need to update a mask? This is relative to how we generate a
|
|
// mask layer in ContainerState::SetupMaskLayerForCSSMask. While creating a
|
|
// mask layer, to reduce memory usage, we did not choose the size of the
|
|
// masked element as mask size. Instead, we read the union of bounds of all
|
|
// children display items by nsDisplayWrapList::GetBounds, which is smaller
|
|
// than or equal to the masked element's boundary, and use it as the position
|
|
// size of the mask layer. That union bounds is actually affected by the
|
|
// geometry of the animated element. To keep the content of mask up to date,
|
|
// forbidding of prerender is required.
|
|
for (nsIFrame* container = nsLayoutUtils::GetCrossDocParentFrame(aFrame);
|
|
container;
|
|
container = nsLayoutUtils::GetCrossDocParentFrame(container)) {
|
|
const nsStyleSVGReset* svgReset = container->StyleSVGReset();
|
|
if (svgReset->HasMask() || svgReset->HasClipPath()) {
|
|
return NoPrerender;
|
|
}
|
|
}
|
|
|
|
// If the incoming dirty rect already contains the entire overflow area,
|
|
// we are already rendering the entire content.
|
|
nsRect overflow = aFrame->GetVisualOverflowRectRelativeToSelf();
|
|
if (aDirtyRect->Contains(overflow)) {
|
|
return FullPrerender;
|
|
}
|
|
|
|
float viewportRatioX =
|
|
StaticPrefs::layout_animation_prerender_viewport_ratio_limit_x();
|
|
float viewportRatioY =
|
|
StaticPrefs::layout_animation_prerender_viewport_ratio_limit_y();
|
|
uint32_t absoluteLimitX =
|
|
StaticPrefs::layout_animation_prerender_absolute_limit_x();
|
|
uint32_t absoluteLimitY =
|
|
StaticPrefs::layout_animation_prerender_absolute_limit_y();
|
|
nsSize refSize = aBuilder->RootReferenceFrame()->GetSize();
|
|
// Only prerender if the transformed frame's size is <= a multiple of the
|
|
// reference frame size (~viewport), and less than an absolute limit.
|
|
// Both the ratio and the absolute limit are configurable.
|
|
nsSize relativeLimit(nscoord(refSize.width * viewportRatioX),
|
|
nscoord(refSize.height * viewportRatioY));
|
|
nsSize absoluteLimit(
|
|
aFrame->PresContext()->DevPixelsToAppUnits(absoluteLimitX),
|
|
aFrame->PresContext()->DevPixelsToAppUnits(absoluteLimitY));
|
|
nsSize maxSize = Min(relativeLimit, absoluteLimit);
|
|
|
|
const auto transform = nsLayoutUtils::GetTransformToAncestor(
|
|
aFrame, nsLayoutUtils::GetDisplayRootFrame(aFrame));
|
|
const gfxRect transformedBounds = transform.TransformAndClipBounds(
|
|
gfxRect(overflow.x, overflow.y, overflow.width, overflow.height),
|
|
gfxRect::MaxIntRect());
|
|
const nsSize frameSize =
|
|
nsSize(transformedBounds.width, transformedBounds.height);
|
|
|
|
uint64_t maxLimitArea = uint64_t(maxSize.width) * maxSize.height;
|
|
uint64_t frameArea = uint64_t(frameSize.width) * frameSize.height;
|
|
if (frameArea <= maxLimitArea && frameSize <= absoluteLimit) {
|
|
*aDirtyRect = overflow;
|
|
return FullPrerender;
|
|
}
|
|
|
|
// Frames participating any preserve-3d rendering context are force to
|
|
// full prerender here if we failed the previous if-conditions.
|
|
// If the current frame doesn't have animations and any of its parent has
|
|
// animations, we still hit this path because
|
|
// ActiveLayerTracker::IsTransformMaybeAnimated() catches the existing
|
|
// animations on the parent frames in the preserve-3d rendering context.
|
|
// FIXME: For now, we force all the frame in the 3D rendering context to use
|
|
// FullPrerender, so this may have an issue if one of the frame is too large
|
|
// and it has animations (which may make the calculation of visible and dirty
|
|
// rects not correct.).
|
|
if (aFrame->Extend3DContext() || aFrame->Combines3DTransformWithAncestors()) {
|
|
// Use overflow as dirty and force to use full prerender.
|
|
*aDirtyRect = overflow;
|
|
return FullPrerender;
|
|
}
|
|
|
|
if (StaticPrefs::layout_animation_prerender_partial()) {
|
|
*aDirtyRect = nsLayoutUtils::ComputePartialPrerenderArea(*aDirtyRect,
|
|
overflow, maxSize);
|
|
return PartialPrerender;
|
|
}
|
|
|
|
if (frameArea > maxLimitArea) {
|
|
uint64_t appUnitsPerPixel = AppUnitsPerCSSPixel();
|
|
EffectCompositor::SetPerformanceWarning(
|
|
aFrame, transformSet,
|
|
AnimationPerformanceWarning(
|
|
AnimationPerformanceWarning::Type::ContentTooLargeArea,
|
|
{
|
|
int(frameArea / (appUnitsPerPixel * appUnitsPerPixel)),
|
|
int(maxLimitArea / (appUnitsPerPixel * appUnitsPerPixel)),
|
|
}));
|
|
} else {
|
|
EffectCompositor::SetPerformanceWarning(
|
|
aFrame, transformSet,
|
|
AnimationPerformanceWarning(
|
|
AnimationPerformanceWarning::Type::ContentTooLarge,
|
|
{
|
|
nsPresContext::AppUnitsToIntCSSPixels(frameSize.width),
|
|
nsPresContext::AppUnitsToIntCSSPixels(frameSize.height),
|
|
nsPresContext::AppUnitsToIntCSSPixels(relativeLimit.width),
|
|
nsPresContext::AppUnitsToIntCSSPixels(relativeLimit.height),
|
|
nsPresContext::AppUnitsToIntCSSPixels(absoluteLimit.width),
|
|
nsPresContext::AppUnitsToIntCSSPixels(absoluteLimit.height),
|
|
}));
|
|
}
|
|
|
|
return NoPrerender;
|
|
}
|
|
|
|
/* If the matrix is singular, or a hidden backface is shown, the frame won't be
|
|
* visible or hit. */
|
|
static bool IsFrameVisible(nsIFrame* aFrame, const Matrix4x4& aMatrix) {
|
|
if (aMatrix.IsSingular()) {
|
|
return false;
|
|
}
|
|
if (aFrame->BackfaceIsHidden() && aMatrix.IsBackfaceVisible()) {
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
|
|
const Matrix4x4Flagged& nsDisplayTransform::GetTransform() const {
|
|
if (mTransform) {
|
|
return *mTransform;
|
|
}
|
|
|
|
float scale = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
|
|
if (mTransformGetter) {
|
|
mTransform.emplace(mTransformGetter(mFrame, scale));
|
|
Point3D newOrigin =
|
|
Point3D(NSAppUnitsToFloatPixels(mToReferenceFrame.x, scale),
|
|
NSAppUnitsToFloatPixels(mToReferenceFrame.y, scale), 0.0f);
|
|
mTransform->ChangeBasis(newOrigin.x, newOrigin.y, newOrigin.z);
|
|
} else if (!mIsTransformSeparator) {
|
|
DebugOnly<bool> isReference = mFrame->IsTransformed() ||
|
|
mFrame->Combines3DTransformWithAncestors() ||
|
|
mFrame->Extend3DContext();
|
|
MOZ_ASSERT(isReference);
|
|
mTransform.emplace(
|
|
GetResultingTransformMatrix(mFrame, ToReferenceFrame(), scale,
|
|
INCLUDE_PERSPECTIVE | OFFSET_BY_ORIGIN));
|
|
} else {
|
|
// Use identity matrix
|
|
mTransform.emplace();
|
|
}
|
|
|
|
return *mTransform;
|
|
}
|
|
|
|
const Matrix4x4Flagged& nsDisplayTransform::GetInverseTransform() const {
|
|
if (mInverseTransform) {
|
|
return *mInverseTransform;
|
|
}
|
|
|
|
MOZ_ASSERT(!GetTransform().IsSingular());
|
|
|
|
mInverseTransform.emplace(GetTransform().Inverse());
|
|
|
|
return *mInverseTransform;
|
|
}
|
|
|
|
Matrix4x4 nsDisplayTransform::GetTransformForRendering(
|
|
LayoutDevicePoint* aOutOrigin) const {
|
|
if (!mFrame->HasPerspective() || mTransformGetter || mIsTransformSeparator) {
|
|
if (!mTransformGetter && !mIsTransformSeparator && aOutOrigin) {
|
|
// If aOutOrigin is provided, put the offset to origin into it, because
|
|
// we need to keep it separate for webrender. The combination of
|
|
// *aOutOrigin and the returned matrix here should always be equivalent
|
|
// to what GetTransform() would have returned.
|
|
float scale = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
*aOutOrigin = LayoutDevicePoint::FromAppUnits(ToReferenceFrame(), scale);
|
|
|
|
// The rounding behavior should also be the same as GetTransform().
|
|
if (ShouldRoundTransformOrigin(mFrame)) {
|
|
aOutOrigin->Round();
|
|
}
|
|
return GetResultingTransformMatrix(mFrame, nsPoint(0, 0), scale,
|
|
INCLUDE_PERSPECTIVE);
|
|
}
|
|
return GetTransform().GetMatrix();
|
|
}
|
|
MOZ_ASSERT(!mTransformGetter);
|
|
|
|
float scale = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
// Don't include perspective transform, or the offset to origin, since
|
|
// nsDisplayPerspective will handle both of those.
|
|
return GetResultingTransformMatrix(mFrame, ToReferenceFrame(), scale, 0);
|
|
}
|
|
|
|
const Matrix4x4& nsDisplayTransform::GetAccumulatedPreserved3DTransform(
|
|
nsDisplayListBuilder* aBuilder) {
|
|
MOZ_ASSERT(!mFrame->Extend3DContext() || IsLeafOf3DContext());
|
|
|
|
if (!IsLeafOf3DContext()) {
|
|
return GetTransform().GetMatrix();
|
|
}
|
|
|
|
// XXX: should go back to fix mTransformGetter.
|
|
if (!mTransformPreserves3D) {
|
|
const nsIFrame* establisher; // Establisher of the 3D rendering context.
|
|
for (establisher = mFrame;
|
|
establisher && establisher->Combines3DTransformWithAncestors();
|
|
establisher =
|
|
establisher->GetClosestFlattenedTreeAncestorPrimaryFrame()) {
|
|
}
|
|
const nsIFrame* establisherReference = aBuilder->FindReferenceFrameFor(
|
|
nsLayoutUtils::GetCrossDocParentFrame(establisher));
|
|
|
|
nsPoint offset = establisher->GetOffsetToCrossDoc(establisherReference);
|
|
float scale = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
uint32_t flags =
|
|
INCLUDE_PRESERVE3D_ANCESTORS | INCLUDE_PERSPECTIVE | OFFSET_BY_ORIGIN;
|
|
mTransformPreserves3D = MakeUnique<Matrix4x4>(
|
|
GetResultingTransformMatrix(mFrame, offset, scale, flags));
|
|
}
|
|
|
|
return *mTransformPreserves3D;
|
|
}
|
|
|
|
bool nsDisplayTransform::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc, RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
// We want to make sure we don't pollute the transform property in the WR
|
|
// stacking context by including the position of this frame (relative to the
|
|
// parent reference frame). We need to keep those separate; the position of
|
|
// this frame goes into the stacking context bounds while the transform goes
|
|
// into the transform.
|
|
LayoutDevicePoint position;
|
|
Matrix4x4 newTransformMatrix = GetTransformForRendering(&position);
|
|
|
|
gfx::Matrix4x4* transformForSC = &newTransformMatrix;
|
|
if (newTransformMatrix.IsIdentity()) {
|
|
// If the transform is an identity transform, strip it out so that WR
|
|
// doesn't turn this stacking context into a reference frame, as it
|
|
// affects positioning. Bug 1345577 tracks a better fix.
|
|
transformForSC = nullptr;
|
|
|
|
// In ChooseScaleAndSetTransform, we round the offset from the reference
|
|
// frame used to adjust the transform, if there is no transform, or it
|
|
// is just a translation. We need to do the same here.
|
|
position.Round();
|
|
}
|
|
|
|
// We don't send animations for transform separator display items.
|
|
uint64_t animationsId =
|
|
mIsTransformSeparator
|
|
? 0
|
|
: AddAnimationsForWebRender(this, aManager, aDisplayListBuilder,
|
|
aBuilder.GetRenderRoot());
|
|
wr::WrAnimationProperty prop{
|
|
wr::WrAnimationType::Transform,
|
|
animationsId,
|
|
};
|
|
|
|
Maybe<nsDisplayTransform*> deferredTransformItem;
|
|
if (!mFrame->ChildrenHavePerspective()) {
|
|
// If it has perspective, we create a new scroll data via the
|
|
// UpdateScrollData call because that scenario is more complex. Otherwise
|
|
// we can just stash the transform on the StackingContextHelper and
|
|
// apply it to any scroll data that are created inside this
|
|
// nsDisplayTransform.
|
|
deferredTransformItem = Some(this);
|
|
}
|
|
|
|
// Determine if we're possibly animated (= would need an active layer in FLB).
|
|
bool animated = !mIsTransformSeparator &&
|
|
ActiveLayerTracker::IsTransformMaybeAnimated(Frame());
|
|
|
|
wr::StackingContextParams params;
|
|
params.mBoundTransform = &newTransformMatrix;
|
|
params.animation = animationsId ? &prop : nullptr;
|
|
params.mTransformPtr = transformForSC;
|
|
params.prim_flags = !BackfaceIsHidden()
|
|
? wr::PrimitiveFlags::IS_BACKFACE_VISIBLE
|
|
: wr::PrimitiveFlags{0};
|
|
params.mDeferredTransformItem = deferredTransformItem;
|
|
params.mAnimated = animated;
|
|
// Determine if we would have to rasterize any items in local raster space
|
|
// (i.e. disable subpixel AA). We don't always need to rasterize locally even
|
|
// if the stacking context is possibly animated (at the cost of potentially
|
|
// some false negatives with respect to will-change handling), so we pass in
|
|
// this determination separately to accurately match with when FLB would
|
|
// normally disable subpixel AA.
|
|
params.mRasterizeLocally = animated && Frame()->HasAnimationOfTransform();
|
|
params.SetPreserve3D(mFrame->Extend3DContext() && !mIsTransformSeparator);
|
|
params.clip =
|
|
wr::WrStackingContextClip::ClipChain(aBuilder.CurrentClipChainId());
|
|
|
|
LayoutDeviceSize boundsSize = LayoutDeviceSize::FromAppUnits(
|
|
mChildBounds.Size(), mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
|
|
StackingContextHelper sc(aSc, GetActiveScrolledRoot(), mFrame, this, aBuilder,
|
|
params, LayoutDeviceRect(position, boundsSize));
|
|
|
|
aManager->CommandBuilder().CreateWebRenderCommandsFromDisplayList(
|
|
GetChildren(), this, aDisplayListBuilder, sc, aBuilder, aResources);
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayTransform::UpdateScrollData(
|
|
mozilla::layers::WebRenderScrollData* aData,
|
|
mozilla::layers::WebRenderLayerScrollData* aLayerData) {
|
|
if (!mFrame->ChildrenHavePerspective()) {
|
|
// This case is handled in CreateWebRenderCommands by stashing the transform
|
|
// on the stacking context.
|
|
return false;
|
|
}
|
|
if (aLayerData) {
|
|
aLayerData->SetTransform(GetTransform().GetMatrix());
|
|
aLayerData->SetTransformIsPerspective(true);
|
|
}
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayTransform::ShouldSkipTransform(
|
|
nsDisplayListBuilder* aBuilder) const {
|
|
return (aBuilder->RootReferenceFrame() == mFrame) &&
|
|
aBuilder->IsForGenerateGlyphMask();
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayTransform::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
// While generating a glyph mask, the transform vector of the root frame had
|
|
// been applied into the target context, so stop applying it again here.
|
|
const bool shouldSkipTransform = ShouldSkipTransform(aBuilder);
|
|
|
|
/* For frames without transform, it would not be removed for
|
|
* backface hidden here. But, it would be removed by the init
|
|
* function of nsDisplayTransform.
|
|
*/
|
|
const Matrix4x4 newTransformMatrix =
|
|
shouldSkipTransform ? Matrix4x4() : GetTransformForRendering();
|
|
|
|
uint32_t flags = FrameLayerBuilder::CONTAINER_ALLOW_PULL_BACKGROUND_COLOR;
|
|
RefPtr<ContainerLayer> container =
|
|
aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, GetChildren(), aContainerParameters,
|
|
&newTransformMatrix, flags);
|
|
|
|
if (!container) {
|
|
return nullptr;
|
|
}
|
|
|
|
// Add the preserve-3d flag for this layer, BuildContainerLayerFor clears all
|
|
// flags, so we never need to explicitly unset this flag.
|
|
if (mFrame->Extend3DContext() && !mIsTransformSeparator) {
|
|
container->SetContentFlags(container->GetContentFlags() |
|
|
Layer::CONTENT_EXTEND_3D_CONTEXT);
|
|
} else {
|
|
container->SetContentFlags(container->GetContentFlags() &
|
|
~Layer::CONTENT_EXTEND_3D_CONTEXT);
|
|
}
|
|
|
|
if (mAllowAsyncAnimation) {
|
|
mFrame->SetProperty(nsIFrame::RefusedAsyncAnimationProperty(), false);
|
|
}
|
|
|
|
// We don't send animations for transform separator display items.
|
|
if (!mIsTransformSeparator) {
|
|
nsDisplayListBuilder::AddAnimationsAndTransitionsToLayer(
|
|
container, aBuilder, this, mFrame, GetType());
|
|
}
|
|
|
|
if (mAllowAsyncAnimation && MayBeAnimated(aBuilder)) {
|
|
// Only allow async updates to the transform if we're an animated layer,
|
|
// since that's what triggers us to set the correct AGR in the constructor
|
|
// and makes sure FrameLayerBuilder won't compute occlusions for this layer.
|
|
container->SetUserData(nsIFrame::LayerIsPrerenderedDataKey(),
|
|
/*the value is irrelevant*/ nullptr);
|
|
container->SetContentFlags(container->GetContentFlags() |
|
|
Layer::CONTENT_MAY_CHANGE_TRANSFORM);
|
|
} else {
|
|
container->RemoveUserData(nsIFrame::LayerIsPrerenderedDataKey());
|
|
container->SetContentFlags(container->GetContentFlags() &
|
|
~Layer::CONTENT_MAY_CHANGE_TRANSFORM);
|
|
}
|
|
return container.forget();
|
|
}
|
|
|
|
bool nsDisplayTransform::MayBeAnimated(nsDisplayListBuilder* aBuilder,
|
|
bool aEnforceMinimumSize) const {
|
|
// If EffectCompositor::HasAnimationsForCompositor() is true then we can
|
|
// completely bypass the main thread for this animation, so it is always
|
|
// worthwhile.
|
|
// For ActiveLayerTracker::IsTransformAnimated() cases the main thread is
|
|
// already involved so there is less to be gained.
|
|
// Therefore we check that the *post-transform* bounds of this item are
|
|
// big enough to justify an active layer.
|
|
if (EffectCompositor::HasAnimationsForCompositor(
|
|
mFrame, DisplayItemType::TYPE_TRANSFORM) ||
|
|
(ActiveLayerTracker::IsTransformAnimated(aBuilder, mFrame) &&
|
|
!(aEnforceMinimumSize && IsItemTooSmallForActiveLayer(mFrame)))) {
|
|
return true;
|
|
}
|
|
return false;
|
|
}
|
|
|
|
nsDisplayItem::LayerState nsDisplayTransform::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
// If the transform is 3d, the layer takes part in preserve-3d
|
|
// sorting, or the layer is a separator then we *always* want this
|
|
// to be an active layer.
|
|
// Checking HasPerspective() is needed to handle perspective value 0 when
|
|
// the transform is 2D.
|
|
if (!GetTransform().Is2D() || Combines3DTransformWithAncestors() ||
|
|
mIsTransformSeparator || mFrame->HasPerspective()) {
|
|
return LayerState::LAYER_ACTIVE_FORCE;
|
|
}
|
|
|
|
if (MayBeAnimated(aBuilder)) {
|
|
// Returns LayerState::LAYER_ACTIVE_FORCE to avoid flatterning the layer for
|
|
// async animations.
|
|
return LayerState::LAYER_ACTIVE_FORCE;
|
|
}
|
|
|
|
// Expect the child display items to have this frame as their animated
|
|
// geometry root (since it will be their reference frame). If they have a
|
|
// different animated geometry root, we'll make this an active layer so the
|
|
// animation can be accelerated.
|
|
return RequiredLayerStateForChildren(aBuilder, aManager, aParameters,
|
|
*GetChildren(),
|
|
mAnimatedGeometryRootForChildren);
|
|
}
|
|
|
|
bool nsDisplayTransform::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
// nsDisplayTransform::GetBounds() returns an empty rect in nested 3d context.
|
|
// Calling mStoredList.RecomputeVisibility below for such transform causes the
|
|
// child display items to end up with empty visible rect.
|
|
// We avoid this by bailing out always if we are dealing with a 3d context.
|
|
if (mFrame->Extend3DContext() || Combines3DTransformWithAncestors()) {
|
|
return true;
|
|
}
|
|
|
|
/* As we do this, we need to be sure to
|
|
* untransform the visible rect, since we want everything that's painting to
|
|
* think that it's painting in its original rectangular coordinate space.
|
|
* If we can't untransform, take the entire overflow rect */
|
|
nsRect untransformedVisibleRect;
|
|
if (!UntransformPaintRect(aBuilder, &untransformedVisibleRect)) {
|
|
untransformedVisibleRect = mFrame->GetVisualOverflowRectRelativeToSelf();
|
|
}
|
|
|
|
bool snap;
|
|
const nsRect bounds = GetUntransformedBounds(aBuilder, &snap);
|
|
nsRegion visibleRegion;
|
|
visibleRegion.And(bounds, untransformedVisibleRect);
|
|
GetChildren()->ComputeVisibilityForSublist(aBuilder, &visibleRegion,
|
|
visibleRegion.GetBounds());
|
|
|
|
return true;
|
|
}
|
|
|
|
nsRect nsDisplayTransform::TransformUntransformedBounds(
|
|
nsDisplayListBuilder* aBuilder, const Matrix4x4Flagged& aMatrix) const {
|
|
bool snap;
|
|
const nsRect untransformedBounds = GetUntransformedBounds(aBuilder, &snap);
|
|
// GetTransform always operates in dev pixels.
|
|
const float factor = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
return nsLayoutUtils::MatrixTransformRect(untransformedBounds, aMatrix,
|
|
factor);
|
|
}
|
|
|
|
/**
|
|
* Returns the bounds for this transform. The bounds are calculated during
|
|
* display list building and merging, see |nsDisplayTransform::UpdateBounds()|.
|
|
*/
|
|
nsRect nsDisplayTransform::GetBounds(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = false;
|
|
return mBounds;
|
|
}
|
|
|
|
void nsDisplayTransform::ComputeBounds(nsDisplayListBuilder* aBuilder) {
|
|
MOZ_ASSERT(mFrame->Extend3DContext() || IsLeafOf3DContext());
|
|
|
|
/* Some transforms can get empty bounds in 2D, but might get transformed again
|
|
* and get non-empty bounds. A simple example of this would be a 180 degree
|
|
* rotation getting applied twice.
|
|
* We should not depend on transforming bounds level by level.
|
|
*
|
|
* This function collects the bounds of this transform and stores it in
|
|
* nsDisplayListBuilder. If this is not a leaf of a 3D context, we recurse
|
|
* down and include the bounds of the child transforms.
|
|
* The bounds are transformed with the accumulated transformation matrix up to
|
|
* the 3D context root coordinate space.
|
|
*/
|
|
nsDisplayListBuilder::AutoAccumulateTransform accTransform(aBuilder);
|
|
accTransform.Accumulate(GetTransform().GetMatrix());
|
|
|
|
// Do not dive into another 3D context.
|
|
if (!IsLeafOf3DContext()) {
|
|
for (nsDisplayItem* i : *GetChildren()) {
|
|
i->DoUpdateBoundsPreserves3D(aBuilder);
|
|
}
|
|
}
|
|
|
|
/* The child transforms that extend 3D context further will have empty bounds,
|
|
* so the untransformed bounds here is the bounds of all the non-preserve-3d
|
|
* content under this transform.
|
|
*/
|
|
const nsRect rect = TransformUntransformedBounds(
|
|
aBuilder, accTransform.GetCurrentTransform());
|
|
aBuilder->AccumulateRect(rect);
|
|
}
|
|
|
|
void nsDisplayTransform::DoUpdateBoundsPreserves3D(
|
|
nsDisplayListBuilder* aBuilder) {
|
|
MOZ_ASSERT(mFrame->Combines3DTransformWithAncestors() ||
|
|
IsTransformSeparator());
|
|
// Updating is not going through to child 3D context.
|
|
ComputeBounds(aBuilder);
|
|
}
|
|
|
|
void nsDisplayTransform::UpdateBounds(nsDisplayListBuilder* aBuilder) {
|
|
UpdateUntransformedBounds(aBuilder);
|
|
|
|
if (IsTransformSeparator()) {
|
|
MOZ_ASSERT(GetTransform().IsIdentity());
|
|
mBounds = mChildBounds;
|
|
return;
|
|
}
|
|
|
|
if (mFrame->Extend3DContext()) {
|
|
if (!Combines3DTransformWithAncestors()) {
|
|
// The transform establishes a 3D context. |UpdateBoundsFor3D()| will
|
|
// collect the bounds from the child transforms.
|
|
UpdateBoundsFor3D(aBuilder);
|
|
} else {
|
|
// With nested 3D transforms, the 2D bounds might not be useful.
|
|
mBounds = nsRect();
|
|
}
|
|
|
|
return;
|
|
}
|
|
|
|
MOZ_ASSERT(!mFrame->Extend3DContext());
|
|
|
|
// We would like to avoid calculating 2D bounds here for nested 3D transforms,
|
|
// but mix-blend-mode relies on having bounds set. See bug 1556956.
|
|
|
|
// A stand-alone transform.
|
|
mBounds = TransformUntransformedBounds(aBuilder, GetTransform());
|
|
}
|
|
|
|
void nsDisplayTransform::UpdateBoundsFor3D(nsDisplayListBuilder* aBuilder) {
|
|
MOZ_ASSERT(mFrame->Extend3DContext() &&
|
|
!mFrame->Combines3DTransformWithAncestors() &&
|
|
!IsTransformSeparator());
|
|
|
|
// Always start updating from an establisher of a 3D rendering context.
|
|
nsDisplayListBuilder::AutoAccumulateRect accRect(aBuilder);
|
|
nsDisplayListBuilder::AutoAccumulateTransform accTransform(aBuilder);
|
|
accTransform.StartRoot();
|
|
ComputeBounds(aBuilder);
|
|
mBounds = aBuilder->GetAccumulatedRect();
|
|
}
|
|
|
|
void nsDisplayTransform::UpdateUntransformedBounds(
|
|
nsDisplayListBuilder* aBuilder) {
|
|
mChildBounds = GetChildren()->GetClippedBoundsWithRespectToASR(
|
|
aBuilder, mActiveScrolledRoot);
|
|
}
|
|
|
|
#ifdef DEBUG_HIT
|
|
# include <time.h>
|
|
#endif
|
|
|
|
/* HitTest does some fun stuff with matrix transforms to obtain the answer. */
|
|
void nsDisplayTransform::HitTest(nsDisplayListBuilder* aBuilder,
|
|
const nsRect& aRect, HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) {
|
|
if (aState->mInPreserves3D) {
|
|
GetChildren()->HitTest(aBuilder, aRect, aState, aOutFrames);
|
|
return;
|
|
}
|
|
|
|
/* Here's how this works:
|
|
* 1. Get the matrix. If it's singular, abort (clearly we didn't hit
|
|
* anything).
|
|
* 2. Invert the matrix.
|
|
* 3. Use it to transform the rect into the correct space.
|
|
* 4. Pass that rect down through to the list's version of HitTest.
|
|
*/
|
|
// GetTransform always operates in dev pixels.
|
|
float factor = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
Matrix4x4 matrix = GetAccumulatedPreserved3DTransform(aBuilder);
|
|
|
|
if (!IsFrameVisible(mFrame, matrix)) {
|
|
return;
|
|
}
|
|
|
|
/* We want to go from transformed-space to regular space.
|
|
* Thus we have to invert the matrix, which normally does
|
|
* the reverse operation (e.g. regular->transformed)
|
|
*/
|
|
|
|
/* Now, apply the transform and pass it down the channel. */
|
|
matrix.Invert();
|
|
nsRect resultingRect;
|
|
if (aRect.width == 1 && aRect.height == 1) {
|
|
// Magic width/height indicating we're hit testing a point, not a rect
|
|
Point4D point =
|
|
matrix.ProjectPoint(Point(NSAppUnitsToFloatPixels(aRect.x, factor),
|
|
NSAppUnitsToFloatPixels(aRect.y, factor)));
|
|
if (!point.HasPositiveWCoord()) {
|
|
return;
|
|
}
|
|
|
|
Point point2d = point.As2DPoint();
|
|
|
|
resultingRect =
|
|
nsRect(NSFloatPixelsToAppUnits(float(point2d.x), factor),
|
|
NSFloatPixelsToAppUnits(float(point2d.y), factor), 1, 1);
|
|
|
|
} else {
|
|
Rect originalRect(NSAppUnitsToFloatPixels(aRect.x, factor),
|
|
NSAppUnitsToFloatPixels(aRect.y, factor),
|
|
NSAppUnitsToFloatPixels(aRect.width, factor),
|
|
NSAppUnitsToFloatPixels(aRect.height, factor));
|
|
|
|
bool snap;
|
|
nsRect childBounds = GetUntransformedBounds(aBuilder, &snap);
|
|
Rect childGfxBounds(NSAppUnitsToFloatPixels(childBounds.x, factor),
|
|
NSAppUnitsToFloatPixels(childBounds.y, factor),
|
|
NSAppUnitsToFloatPixels(childBounds.width, factor),
|
|
NSAppUnitsToFloatPixels(childBounds.height, factor));
|
|
|
|
Rect rect = matrix.ProjectRectBounds(originalRect, childGfxBounds);
|
|
|
|
resultingRect =
|
|
nsRect(NSFloatPixelsToAppUnits(float(rect.X()), factor),
|
|
NSFloatPixelsToAppUnits(float(rect.Y()), factor),
|
|
NSFloatPixelsToAppUnits(float(rect.Width()), factor),
|
|
NSFloatPixelsToAppUnits(float(rect.Height()), factor));
|
|
}
|
|
|
|
if (resultingRect.IsEmpty()) {
|
|
return;
|
|
}
|
|
|
|
#ifdef DEBUG_HIT
|
|
printf("Frame: %p\n", dynamic_cast<void*>(mFrame));
|
|
printf(" Untransformed point: (%f, %f)\n", resultingRect.X(),
|
|
resultingRect.Y());
|
|
uint32_t originalFrameCount = aOutFrames.Length();
|
|
#endif
|
|
|
|
GetChildren()->HitTest(aBuilder, resultingRect, aState, aOutFrames);
|
|
|
|
#ifdef DEBUG_HIT
|
|
if (originalFrameCount != aOutFrames.Length())
|
|
printf(" Hit! Time: %f, first frame: %p\n", static_cast<double>(clock()),
|
|
dynamic_cast<void*>(aOutFrames.ElementAt(0)));
|
|
printf("=== end of hit test ===\n");
|
|
#endif
|
|
}
|
|
|
|
float nsDisplayTransform::GetHitDepthAtPoint(nsDisplayListBuilder* aBuilder,
|
|
const nsPoint& aPoint) {
|
|
// GetTransform always operates in dev pixels.
|
|
float factor = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
Matrix4x4 matrix = GetAccumulatedPreserved3DTransform(aBuilder);
|
|
|
|
NS_ASSERTION(IsFrameVisible(mFrame, matrix),
|
|
"We can't have hit a frame that isn't visible!");
|
|
|
|
Matrix4x4 inverse = matrix;
|
|
inverse.Invert();
|
|
Point4D point =
|
|
inverse.ProjectPoint(Point(NSAppUnitsToFloatPixels(aPoint.x, factor),
|
|
NSAppUnitsToFloatPixels(aPoint.y, factor)));
|
|
|
|
Point point2d = point.As2DPoint();
|
|
|
|
Point3D transformed = matrix.TransformPoint(Point3D(point2d.x, point2d.y, 0));
|
|
return transformed.z;
|
|
}
|
|
|
|
/* The transform is opaque iff the transform consists solely of scales and
|
|
* translations and if the underlying content is opaque. Thus if the transform
|
|
* is of the form
|
|
*
|
|
* |a c e|
|
|
* |b d f|
|
|
* |0 0 1|
|
|
*
|
|
* We need b and c to be zero.
|
|
*
|
|
* We also need to check whether the underlying opaque content completely fills
|
|
* our visible rect. We use UntransformRect which expands to the axis-aligned
|
|
* bounding rect, but that's OK since if
|
|
* mStoredList.GetVisibleRect().Contains(untransformedVisible), then it
|
|
* certainly contains the actual (non-axis-aligned) untransformed rect.
|
|
*/
|
|
nsRegion nsDisplayTransform::GetOpaqueRegion(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = false;
|
|
|
|
nsRect untransformedVisible;
|
|
if (!UntransformBuildingRect(aBuilder, &untransformedVisible)) {
|
|
return nsRegion();
|
|
}
|
|
|
|
const Matrix4x4Flagged& matrix = GetTransform();
|
|
Matrix matrix2d;
|
|
if (!matrix.Is2D(&matrix2d) || !matrix2d.PreservesAxisAlignedRectangles()) {
|
|
return nsRegion();
|
|
}
|
|
|
|
nsRegion result;
|
|
|
|
bool tmpSnap;
|
|
const nsRect bounds = GetUntransformedBounds(aBuilder, &tmpSnap);
|
|
const nsRegion opaque = ::GetOpaqueRegion(aBuilder, GetChildren(), bounds);
|
|
|
|
if (opaque.Contains(untransformedVisible)) {
|
|
result = GetBuildingRect().Intersect(GetBounds(aBuilder, &tmpSnap));
|
|
}
|
|
return result;
|
|
}
|
|
|
|
nsRect nsDisplayTransform::GetComponentAlphaBounds(
|
|
nsDisplayListBuilder* aBuilder) const {
|
|
if (GetChildren()->GetComponentAlphaBounds(aBuilder).IsEmpty()) {
|
|
return nsRect();
|
|
}
|
|
|
|
bool snap;
|
|
return GetBounds(aBuilder, &snap);
|
|
}
|
|
|
|
/* TransformRect takes in as parameters a rectangle (in app space) and returns
|
|
* the smallest rectangle (in app space) containing the transformed image of
|
|
* that rectangle. That is, it takes the four corners of the rectangle,
|
|
* transforms them according to the matrix associated with the specified frame,
|
|
* then returns the smallest rectangle containing the four transformed points.
|
|
*
|
|
* @param aUntransformedBounds The rectangle (in app units) to transform.
|
|
* @param aFrame The frame whose transformation should be applied.
|
|
* @param aOrigin The delta from the frame origin to the coordinate space origin
|
|
* @return The smallest rectangle containing the image of the transformed
|
|
* rectangle.
|
|
*/
|
|
nsRect nsDisplayTransform::TransformRect(const nsRect& aUntransformedBounds,
|
|
const nsIFrame* aFrame,
|
|
TransformReferenceBox& aRefBox) {
|
|
MOZ_ASSERT(aFrame, "Can't take the transform based on a null frame!");
|
|
|
|
float factor = aFrame->PresContext()->AppUnitsPerDevPixel();
|
|
|
|
uint32_t flags =
|
|
INCLUDE_PERSPECTIVE | OFFSET_BY_ORIGIN | INCLUDE_PRESERVE3D_ANCESTORS;
|
|
FrameTransformProperties props(aFrame, aRefBox, factor);
|
|
return nsLayoutUtils::MatrixTransformRect(
|
|
aUntransformedBounds,
|
|
GetResultingTransformMatrixInternal(props, aRefBox, nsPoint(0, 0), factor,
|
|
flags),
|
|
factor);
|
|
}
|
|
|
|
bool nsDisplayTransform::UntransformRect(const nsRect& aTransformedBounds,
|
|
const nsRect& aChildBounds,
|
|
const nsIFrame* aFrame,
|
|
nsRect* aOutRect) {
|
|
MOZ_ASSERT(aFrame, "Can't take the transform based on a null frame!");
|
|
|
|
float factor = aFrame->PresContext()->AppUnitsPerDevPixel();
|
|
|
|
uint32_t flags =
|
|
INCLUDE_PERSPECTIVE | OFFSET_BY_ORIGIN | INCLUDE_PRESERVE3D_ANCESTORS;
|
|
|
|
Matrix4x4 transform =
|
|
GetResultingTransformMatrix(aFrame, nsPoint(0, 0), factor, flags);
|
|
if (transform.IsSingular()) {
|
|
return false;
|
|
}
|
|
|
|
RectDouble result(NSAppUnitsToFloatPixels(aTransformedBounds.x, factor),
|
|
NSAppUnitsToFloatPixels(aTransformedBounds.y, factor),
|
|
NSAppUnitsToFloatPixels(aTransformedBounds.width, factor),
|
|
NSAppUnitsToFloatPixels(aTransformedBounds.height, factor));
|
|
|
|
RectDouble childGfxBounds(
|
|
NSAppUnitsToFloatPixels(aChildBounds.x, factor),
|
|
NSAppUnitsToFloatPixels(aChildBounds.y, factor),
|
|
NSAppUnitsToFloatPixels(aChildBounds.width, factor),
|
|
NSAppUnitsToFloatPixels(aChildBounds.height, factor));
|
|
|
|
result = transform.Inverse().ProjectRectBounds(result, childGfxBounds);
|
|
*aOutRect = nsLayoutUtils::RoundGfxRectToAppRect(ThebesRect(result), factor);
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayTransform::UntransformRect(nsDisplayListBuilder* aBuilder,
|
|
const nsRect& aRect,
|
|
nsRect* aOutRect) const {
|
|
if (GetTransform().IsSingular()) {
|
|
return false;
|
|
}
|
|
|
|
// GetTransform always operates in dev pixels.
|
|
float factor = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
RectDouble result(NSAppUnitsToFloatPixels(aRect.x, factor),
|
|
NSAppUnitsToFloatPixels(aRect.y, factor),
|
|
NSAppUnitsToFloatPixels(aRect.width, factor),
|
|
NSAppUnitsToFloatPixels(aRect.height, factor));
|
|
|
|
bool snap;
|
|
nsRect childBounds = GetUntransformedBounds(aBuilder, &snap);
|
|
RectDouble childGfxBounds(
|
|
NSAppUnitsToFloatPixels(childBounds.x, factor),
|
|
NSAppUnitsToFloatPixels(childBounds.y, factor),
|
|
NSAppUnitsToFloatPixels(childBounds.width, factor),
|
|
NSAppUnitsToFloatPixels(childBounds.height, factor));
|
|
|
|
/* We want to untransform the matrix, so invert the transformation first! */
|
|
result = GetInverseTransform().ProjectRectBounds(result, childGfxBounds);
|
|
|
|
*aOutRect = nsLayoutUtils::RoundGfxRectToAppRect(ThebesRect(result), factor);
|
|
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayTransform::WriteDebugInfo(std::stringstream& aStream) {
|
|
AppendToString(aStream, GetTransform().GetMatrix());
|
|
if (IsTransformSeparator()) {
|
|
aStream << " transform-separator";
|
|
}
|
|
if (IsLeafOf3DContext()) {
|
|
aStream << " 3d-context-leaf";
|
|
}
|
|
if (mFrame->Extend3DContext()) {
|
|
aStream << " extends-3d-context";
|
|
}
|
|
if (mFrame->Combines3DTransformWithAncestors()) {
|
|
aStream << " combines-3d-with-ancestors";
|
|
}
|
|
|
|
aStream << " allowAsync(" << (mAllowAsyncAnimation ? "true" : "false") << ")";
|
|
aStream << " childrenBuildingRect" << mChildrenBuildingRect;
|
|
}
|
|
|
|
nsDisplayPerspective::nsDisplayPerspective(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame,
|
|
nsDisplayList* aList)
|
|
: nsDisplayHitTestInfoBase(aBuilder, aFrame) {
|
|
mList.AppendToTop(aList);
|
|
MOZ_ASSERT(mList.Count() == 1);
|
|
MOZ_ASSERT(mList.GetTop()->GetType() == DisplayItemType::TYPE_TRANSFORM);
|
|
mAnimatedGeometryRoot = aBuilder->FindAnimatedGeometryRootFor(
|
|
mFrame->GetContainingBlock(nsIFrame::SKIP_SCROLLED_FRAME));
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayPerspective::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
float appUnitsPerPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
|
|
Matrix4x4 perspectiveMatrix;
|
|
DebugOnly<bool> hasPerspective = nsDisplayTransform::ComputePerspectiveMatrix(
|
|
mFrame, appUnitsPerPixel, perspectiveMatrix);
|
|
MOZ_ASSERT(hasPerspective, "Why did we create nsDisplayPerspective?");
|
|
|
|
/*
|
|
* ClipListToRange can remove our child after we were created.
|
|
*/
|
|
if (!GetChildren()->GetTop()) {
|
|
return nullptr;
|
|
}
|
|
|
|
/*
|
|
* The resulting matrix is still in the coordinate space of the transformed
|
|
* frame. Append a translation to the reference frame coordinates.
|
|
*/
|
|
nsDisplayTransform* transform =
|
|
static_cast<nsDisplayTransform*>(GetChildren()->GetTop());
|
|
|
|
Point3D newOrigin =
|
|
Point3D(NSAppUnitsToFloatPixels(transform->ToReferenceFrame().x,
|
|
appUnitsPerPixel),
|
|
NSAppUnitsToFloatPixels(transform->ToReferenceFrame().y,
|
|
appUnitsPerPixel),
|
|
0.0f);
|
|
Point3D roundedOrigin(NS_round(newOrigin.x), NS_round(newOrigin.y), 0);
|
|
|
|
perspectiveMatrix.PostTranslate(roundedOrigin);
|
|
|
|
RefPtr<ContainerLayer> container =
|
|
aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, GetChildren(), aContainerParameters,
|
|
&perspectiveMatrix, 0);
|
|
|
|
if (!container) {
|
|
return nullptr;
|
|
}
|
|
|
|
// Sort of a lie, but we want to pretend that the perspective layer extends a
|
|
// 3d context so that it gets its transform combined with children. Might need
|
|
// a better name that reflects this use case and isn't specific to
|
|
// preserve-3d.
|
|
container->SetContentFlags(container->GetContentFlags() |
|
|
Layer::CONTENT_EXTEND_3D_CONTEXT);
|
|
container->SetTransformIsPerspective(true);
|
|
|
|
return container.forget();
|
|
}
|
|
|
|
LayerState nsDisplayPerspective::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
return LayerState::LAYER_ACTIVE_FORCE;
|
|
}
|
|
|
|
nsRegion nsDisplayPerspective::GetOpaqueRegion(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
if (!GetChildren()->GetTop()) {
|
|
*aSnap = false;
|
|
return nsRegion();
|
|
}
|
|
|
|
return GetChildren()->GetTop()->GetOpaqueRegion(aBuilder, aSnap);
|
|
}
|
|
|
|
bool nsDisplayPerspective::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc, RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
float appUnitsPerPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
Matrix4x4 perspectiveMatrix;
|
|
DebugOnly<bool> hasPerspective = nsDisplayTransform::ComputePerspectiveMatrix(
|
|
mFrame, appUnitsPerPixel, perspectiveMatrix);
|
|
MOZ_ASSERT(hasPerspective, "Why did we create nsDisplayPerspective?");
|
|
|
|
/*
|
|
* ClipListToRange can remove our child after we were created.
|
|
*/
|
|
if (!GetChildren()->GetTop()) {
|
|
return false;
|
|
}
|
|
|
|
/*
|
|
* The resulting matrix is still in the coordinate space of the transformed
|
|
* frame. Append a translation to the reference frame coordinates.
|
|
*/
|
|
nsDisplayTransform* transform =
|
|
static_cast<nsDisplayTransform*>(GetChildren()->GetTop());
|
|
|
|
Point3D newOrigin =
|
|
Point3D(NSAppUnitsToFloatPixels(transform->ToReferenceFrame().x,
|
|
appUnitsPerPixel),
|
|
NSAppUnitsToFloatPixels(transform->ToReferenceFrame().y,
|
|
appUnitsPerPixel),
|
|
0.0f);
|
|
Point3D roundedOrigin(NS_round(newOrigin.x), NS_round(newOrigin.y), 0);
|
|
|
|
perspectiveMatrix.PostTranslate(roundedOrigin);
|
|
|
|
nsIFrame* perspectiveFrame =
|
|
mFrame->GetContainingBlock(nsIFrame::SKIP_SCROLLED_FRAME);
|
|
|
|
// Passing true here is always correct, since perspective always combines
|
|
// transforms with the descendants. However that'd make WR do a lot of work
|
|
// that it doesn't really need to do if there aren't other transforms forming
|
|
// part of the 3D context.
|
|
//
|
|
// WR knows how to treat perspective in that case, so the only thing we need
|
|
// to do is to ensure we pass true when we're involved in a 3d context in any
|
|
// other way via the transform-style property on either the transformed frame
|
|
// or the perspective frame in order to not confuse WR's preserve-3d code in
|
|
// very awful ways.
|
|
bool preserve3D =
|
|
mFrame->Extend3DContext() || perspectiveFrame->Extend3DContext();
|
|
|
|
wr::StackingContextParams params;
|
|
params.mTransformPtr = &perspectiveMatrix;
|
|
params.reference_frame_kind = wr::WrReferenceFrameKind::Perspective;
|
|
params.prim_flags = !BackfaceIsHidden()
|
|
? wr::PrimitiveFlags::IS_BACKFACE_VISIBLE
|
|
: wr::PrimitiveFlags{0};
|
|
params.SetPreserve3D(preserve3D);
|
|
params.clip =
|
|
wr::WrStackingContextClip::ClipChain(aBuilder.CurrentClipChainId());
|
|
|
|
Maybe<uint64_t> scrollingRelativeTo;
|
|
for (auto* asr = GetActiveScrolledRoot(); asr; asr = asr->mParent) {
|
|
if (nsLayoutUtils::IsAncestorFrameCrossDoc(
|
|
asr->mScrollableFrame->GetScrolledFrame(), perspectiveFrame)) {
|
|
scrollingRelativeTo.emplace(asr->GetViewId());
|
|
break;
|
|
}
|
|
}
|
|
|
|
// We put the perspective reference frame wrapping the transformed frame,
|
|
// even though there may be arbitrarily nested scroll frames in between.
|
|
//
|
|
// We need to know how many ancestor scroll-frames are we nested in, in order
|
|
// for the async scrolling code in WebRender to calculate the right
|
|
// transformation for the perspective contents.
|
|
params.scrolling_relative_to = scrollingRelativeTo.ptrOr(nullptr);
|
|
|
|
StackingContextHelper sc(aSc, GetActiveScrolledRoot(), mFrame, this, aBuilder,
|
|
params);
|
|
|
|
aManager->CommandBuilder().CreateWebRenderCommandsFromDisplayList(
|
|
GetChildren(), this, aDisplayListBuilder, sc, aBuilder, aResources);
|
|
|
|
return true;
|
|
}
|
|
|
|
nsDisplayText::nsDisplayText(nsDisplayListBuilder* aBuilder,
|
|
nsTextFrame* aFrame,
|
|
const Maybe<bool>& aIsSelected)
|
|
: nsPaintedDisplayItem(aBuilder, aFrame),
|
|
mOpacity(1.0f),
|
|
mVisIStartEdge(0),
|
|
mVisIEndEdge(0) {
|
|
MOZ_COUNT_CTOR(nsDisplayText);
|
|
mIsFrameSelected = aIsSelected;
|
|
mBounds = mFrame->GetVisualOverflowRectRelativeToSelf() + ToReferenceFrame();
|
|
// Bug 748228
|
|
mBounds.Inflate(mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
}
|
|
|
|
bool nsDisplayText::CanApplyOpacity() const {
|
|
if (IsSelected()) {
|
|
return false;
|
|
}
|
|
|
|
nsTextFrame* f = static_cast<nsTextFrame*>(mFrame);
|
|
const nsStyleText* textStyle = f->StyleText();
|
|
if (textStyle->HasTextShadow()) {
|
|
return false;
|
|
}
|
|
|
|
nsTextFrame::TextDecorations decorations;
|
|
f->GetTextDecorations(f->PresContext(), nsTextFrame::eResolvedColors,
|
|
decorations);
|
|
if (decorations.HasDecorationLines()) {
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayText::Paint(nsDisplayListBuilder* aBuilder, gfxContext* aCtx) {
|
|
AUTO_PROFILER_LABEL("nsDisplayText::Paint", GRAPHICS);
|
|
|
|
DrawTargetAutoDisableSubpixelAntialiasing disable(aCtx->GetDrawTarget(),
|
|
IsSubpixelAADisabled());
|
|
RenderToContext(aCtx, aBuilder);
|
|
}
|
|
|
|
bool nsDisplayText::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc, RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
auto* f = static_cast<nsTextFrame*>(mFrame);
|
|
auto appUnitsPerDevPixel = f->PresContext()->AppUnitsPerDevPixel();
|
|
|
|
nsRect bounds = f->WebRenderBounds() + ToReferenceFrame();
|
|
// Bug 748228
|
|
bounds.Inflate(appUnitsPerDevPixel);
|
|
|
|
if (bounds.IsEmpty()) {
|
|
return true;
|
|
}
|
|
|
|
gfx::Point deviceOffset =
|
|
LayoutDevicePoint::FromAppUnits(bounds.TopLeft(), appUnitsPerDevPixel)
|
|
.ToUnknownPoint();
|
|
|
|
// Clipping the bounds to the PaintRect (factoring in what's covered by parent
|
|
// frames) let's us early reject a bunch of things, but it can produce
|
|
// incorrect results for shadows, because they can translate things back into
|
|
// view. Also if we're selected we might have some shadows from the
|
|
// ::selected and ::inctive-selected pseudo-selectors. So don't do this
|
|
// optimization if we have shadows or a selection.
|
|
if (!(IsSelected() || f->StyleText()->HasTextShadow())) {
|
|
nsRect visible = GetPaintRect();
|
|
visible.Inflate(3 * appUnitsPerDevPixel);
|
|
bounds = bounds.Intersect(visible);
|
|
}
|
|
|
|
RefPtr<gfxContext> textDrawer = aBuilder.GetTextContext(
|
|
aResources, aSc, aManager, this, bounds, deviceOffset);
|
|
|
|
RenderToContext(textDrawer, aDisplayListBuilder, true);
|
|
|
|
return textDrawer->GetTextDrawer()->Finish();
|
|
}
|
|
|
|
void nsDisplayText::RenderToContext(gfxContext* aCtx,
|
|
nsDisplayListBuilder* aBuilder,
|
|
bool aIsRecording) {
|
|
nsTextFrame* f = static_cast<nsTextFrame*>(mFrame);
|
|
|
|
// Add 1 pixel of dirty area around mVisibleRect to allow us to paint
|
|
// antialiased pixels beyond the measured text extents.
|
|
// This is temporary until we do this in the actual calculation of text
|
|
// extents.
|
|
auto A2D = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
LayoutDeviceRect extraVisible =
|
|
LayoutDeviceRect::FromAppUnits(GetPaintRect(), A2D);
|
|
extraVisible.Inflate(1);
|
|
|
|
gfxRect pixelVisible(extraVisible.x, extraVisible.y, extraVisible.width,
|
|
extraVisible.height);
|
|
pixelVisible.Inflate(2);
|
|
pixelVisible.RoundOut();
|
|
|
|
bool willClip = !aBuilder->IsForGenerateGlyphMask() && !aIsRecording;
|
|
if (willClip) {
|
|
aCtx->NewPath();
|
|
aCtx->Rectangle(pixelVisible);
|
|
aCtx->Clip();
|
|
}
|
|
|
|
NS_ASSERTION(mVisIStartEdge >= 0, "illegal start edge");
|
|
NS_ASSERTION(mVisIEndEdge >= 0, "illegal end edge");
|
|
|
|
gfxContextMatrixAutoSaveRestore matrixSR;
|
|
|
|
nsPoint framePt = ToReferenceFrame();
|
|
if (f->Style()->IsTextCombined()) {
|
|
float scaleFactor = nsTextFrame::GetTextCombineScaleFactor(f);
|
|
if (scaleFactor != 1.0f) {
|
|
if (auto* textDrawer = aCtx->GetTextDrawer()) {
|
|
// WebRender doesn't support scaling text like this yet
|
|
textDrawer->FoundUnsupportedFeature();
|
|
return;
|
|
}
|
|
matrixSR.SetContext(aCtx);
|
|
// Setup matrix to compress text for text-combine-upright if
|
|
// necessary. This is done here because we want selection be
|
|
// compressed at the same time as text.
|
|
gfxPoint pt = nsLayoutUtils::PointToGfxPoint(framePt, A2D);
|
|
gfxMatrix mat = aCtx->CurrentMatrixDouble()
|
|
.PreTranslate(pt)
|
|
.PreScale(scaleFactor, 1.0)
|
|
.PreTranslate(-pt);
|
|
aCtx->SetMatrixDouble(mat);
|
|
}
|
|
}
|
|
nsTextFrame::PaintTextParams params(aCtx);
|
|
params.framePt = gfx::Point(framePt.x, framePt.y);
|
|
params.dirtyRect = extraVisible;
|
|
|
|
if (aBuilder->IsForGenerateGlyphMask()) {
|
|
params.state = nsTextFrame::PaintTextParams::GenerateTextMask;
|
|
} else {
|
|
params.state = nsTextFrame::PaintTextParams::PaintText;
|
|
}
|
|
|
|
f->PaintText(params, mVisIStartEdge, mVisIEndEdge, ToReferenceFrame(),
|
|
IsSelected(), mOpacity);
|
|
|
|
if (willClip) {
|
|
aCtx->PopClip();
|
|
}
|
|
}
|
|
|
|
bool nsDisplayText::IsSelected() const {
|
|
if (mIsFrameSelected.isNothing()) {
|
|
MOZ_ASSERT((nsTextFrame*)do_QueryFrame(mFrame));
|
|
auto* f = static_cast<nsTextFrame*>(mFrame);
|
|
mIsFrameSelected.emplace(f->IsSelected());
|
|
}
|
|
|
|
return mIsFrameSelected.value();
|
|
}
|
|
|
|
class nsDisplayTextGeometry : public nsDisplayItemGenericGeometry {
|
|
public:
|
|
nsDisplayTextGeometry(nsDisplayText* aItem, nsDisplayListBuilder* aBuilder)
|
|
: nsDisplayItemGenericGeometry(aItem, aBuilder),
|
|
mOpacity(aItem->Opacity()),
|
|
mVisIStartEdge(aItem->VisIStartEdge()),
|
|
mVisIEndEdge(aItem->VisIEndEdge()) {
|
|
nsTextFrame* f = static_cast<nsTextFrame*>(aItem->Frame());
|
|
f->GetTextDecorations(f->PresContext(), nsTextFrame::eResolvedColors,
|
|
mDecorations);
|
|
}
|
|
|
|
/**
|
|
* We store the computed text decorations here since they are
|
|
* computed using style data from parent frames. Any changes to these
|
|
* styles will only invalidate the parent frame and not this frame.
|
|
*/
|
|
nsTextFrame::TextDecorations mDecorations;
|
|
float mOpacity;
|
|
nscoord mVisIStartEdge;
|
|
nscoord mVisIEndEdge;
|
|
};
|
|
|
|
nsDisplayItemGeometry* nsDisplayText::AllocateGeometry(
|
|
nsDisplayListBuilder* aBuilder) {
|
|
return new nsDisplayTextGeometry(this, aBuilder);
|
|
}
|
|
|
|
void nsDisplayText::ComputeInvalidationRegion(
|
|
nsDisplayListBuilder* aBuilder, const nsDisplayItemGeometry* aGeometry,
|
|
nsRegion* aInvalidRegion) const {
|
|
const nsDisplayTextGeometry* geometry =
|
|
static_cast<const nsDisplayTextGeometry*>(aGeometry);
|
|
nsTextFrame* f = static_cast<nsTextFrame*>(mFrame);
|
|
|
|
nsTextFrame::TextDecorations decorations;
|
|
f->GetTextDecorations(f->PresContext(), nsTextFrame::eResolvedColors,
|
|
decorations);
|
|
|
|
bool snap;
|
|
const nsRect& newRect = geometry->mBounds;
|
|
nsRect oldRect = GetBounds(aBuilder, &snap);
|
|
if (decorations != geometry->mDecorations ||
|
|
mVisIStartEdge != geometry->mVisIStartEdge ||
|
|
mVisIEndEdge != geometry->mVisIEndEdge ||
|
|
!oldRect.IsEqualInterior(newRect) ||
|
|
!geometry->mBorderRect.IsEqualInterior(GetBorderRect()) ||
|
|
mOpacity != geometry->mOpacity) {
|
|
aInvalidRegion->Or(oldRect, newRect);
|
|
}
|
|
}
|
|
|
|
void nsDisplayText::WriteDebugInfo(std::stringstream& aStream) {
|
|
#ifdef DEBUG
|
|
aStream << " (\"";
|
|
|
|
nsTextFrame* f = static_cast<nsTextFrame*>(mFrame);
|
|
nsCString buf;
|
|
int32_t totalContentLength;
|
|
f->ToCString(buf, &totalContentLength);
|
|
|
|
aStream << buf.get() << "\")";
|
|
#endif
|
|
}
|
|
|
|
nsDisplayEffectsBase::nsDisplayEffectsBase(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot, bool aClearClipChain)
|
|
: nsDisplayWrapList(aBuilder, aFrame, aList, aActiveScrolledRoot,
|
|
aClearClipChain),
|
|
mHandleOpacity(false) {
|
|
MOZ_COUNT_CTOR(nsDisplayEffectsBase);
|
|
}
|
|
|
|
nsDisplayEffectsBase::nsDisplayEffectsBase(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame,
|
|
nsDisplayList* aList)
|
|
: nsDisplayWrapList(aBuilder, aFrame, aList), mHandleOpacity(false) {
|
|
MOZ_COUNT_CTOR(nsDisplayEffectsBase);
|
|
}
|
|
|
|
nsRegion nsDisplayEffectsBase::GetOpaqueRegion(nsDisplayListBuilder* aBuilder,
|
|
bool* aSnap) const {
|
|
*aSnap = false;
|
|
return nsRegion();
|
|
}
|
|
|
|
void nsDisplayEffectsBase::HitTest(nsDisplayListBuilder* aBuilder,
|
|
const nsRect& aRect, HitTestState* aState,
|
|
nsTArray<nsIFrame*>* aOutFrames) {
|
|
nsPoint rectCenter(aRect.x + aRect.width / 2, aRect.y + aRect.height / 2);
|
|
if (nsSVGIntegrationUtils::HitTestFrameForEffects(
|
|
mFrame, rectCenter - ToReferenceFrame())) {
|
|
mList.HitTest(aBuilder, aRect, aState, aOutFrames);
|
|
}
|
|
}
|
|
|
|
gfxRect nsDisplayEffectsBase::BBoxInUserSpace() const {
|
|
return nsSVGUtils::GetBBox(mFrame);
|
|
}
|
|
|
|
gfxPoint nsDisplayEffectsBase::UserSpaceOffset() const {
|
|
return nsSVGUtils::FrameSpaceInCSSPxToUserSpaceOffset(mFrame);
|
|
}
|
|
|
|
void nsDisplayEffectsBase::ComputeInvalidationRegion(
|
|
nsDisplayListBuilder* aBuilder, const nsDisplayItemGeometry* aGeometry,
|
|
nsRegion* aInvalidRegion) const {
|
|
auto* geometry = static_cast<const nsDisplaySVGEffectGeometry*>(aGeometry);
|
|
bool snap;
|
|
nsRect bounds = GetBounds(aBuilder, &snap);
|
|
if (geometry->mFrameOffsetToReferenceFrame != ToReferenceFrame() ||
|
|
geometry->mUserSpaceOffset != UserSpaceOffset() ||
|
|
!geometry->mBBox.IsEqualInterior(BBoxInUserSpace()) ||
|
|
geometry->mOpacity != mFrame->StyleEffects()->mOpacity ||
|
|
geometry->mHandleOpacity != ShouldHandleOpacity()) {
|
|
// Filter and mask output can depend on the location of the frame's user
|
|
// space and on the frame's BBox. We need to invalidate if either of these
|
|
// change relative to the reference frame.
|
|
// Invalidations from our inactive layer manager are not enough to catch
|
|
// some of these cases because filters can produce output even if there's
|
|
// nothing in the filter input.
|
|
aInvalidRegion->Or(bounds, geometry->mBounds);
|
|
}
|
|
}
|
|
|
|
bool nsDisplayEffectsBase::ValidateSVGFrame() {
|
|
const nsIContent* content = mFrame->GetContent();
|
|
bool hasSVGLayout = (mFrame->GetStateBits() & NS_FRAME_SVG_LAYOUT);
|
|
if (hasSVGLayout) {
|
|
nsSVGDisplayableFrame* svgFrame = do_QueryFrame(mFrame);
|
|
if (!svgFrame || !mFrame->GetContent()->IsSVGElement()) {
|
|
NS_ASSERTION(false, "why?");
|
|
return false;
|
|
}
|
|
if (!static_cast<const SVGElement*>(content)->HasValidDimensions()) {
|
|
return false; // The SVG spec says not to draw filters for this
|
|
}
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
typedef nsSVGIntegrationUtils::PaintFramesParams PaintFramesParams;
|
|
|
|
static void ComputeMaskGeometry(PaintFramesParams& aParams) {
|
|
// Properties are added lazily and may have been removed by a restyle, so
|
|
// make sure all applicable ones are set again.
|
|
nsIFrame* firstFrame =
|
|
nsLayoutUtils::FirstContinuationOrIBSplitSibling(aParams.frame);
|
|
|
|
const nsStyleSVGReset* svgReset = firstFrame->StyleSVGReset();
|
|
|
|
nsTArray<nsSVGMaskFrame*> maskFrames;
|
|
// XXX check return value?
|
|
SVGObserverUtils::GetAndObserveMasks(firstFrame, &maskFrames);
|
|
|
|
if (maskFrames.Length() == 0) {
|
|
return;
|
|
}
|
|
|
|
gfxContext& ctx = aParams.ctx;
|
|
nsIFrame* frame = aParams.frame;
|
|
|
|
nsPoint offsetToUserSpace =
|
|
nsLayoutUtils::ComputeOffsetToUserSpace(aParams.builder, aParams.frame);
|
|
|
|
gfxPoint devPixelOffsetToUserSpace = nsLayoutUtils::PointToGfxPoint(
|
|
offsetToUserSpace, frame->PresContext()->AppUnitsPerDevPixel());
|
|
|
|
gfxContextMatrixAutoSaveRestore matSR(&ctx);
|
|
ctx.SetMatrixDouble(
|
|
ctx.CurrentMatrixDouble().PreTranslate(devPixelOffsetToUserSpace));
|
|
|
|
// Convert boaderArea and dirtyRect to user space.
|
|
int32_t appUnitsPerDevPixel = frame->PresContext()->AppUnitsPerDevPixel();
|
|
nsRect userSpaceBorderArea = aParams.borderArea - offsetToUserSpace;
|
|
nsRect userSpaceDirtyRect = aParams.dirtyRect - offsetToUserSpace;
|
|
|
|
// Union all mask layer rectangles in user space.
|
|
gfxRect maskInUserSpace;
|
|
for (size_t i = 0; i < maskFrames.Length(); i++) {
|
|
nsSVGMaskFrame* maskFrame = maskFrames[i];
|
|
gfxRect currentMaskSurfaceRect;
|
|
|
|
if (maskFrame) {
|
|
currentMaskSurfaceRect = maskFrame->GetMaskArea(aParams.frame);
|
|
} else {
|
|
nsCSSRendering::ImageLayerClipState clipState;
|
|
nsCSSRendering::GetImageLayerClip(
|
|
svgReset->mMask.mLayers[i], frame, *frame->StyleBorder(),
|
|
userSpaceBorderArea, userSpaceDirtyRect, false, /* aWillPaintBorder */
|
|
appUnitsPerDevPixel, &clipState);
|
|
currentMaskSurfaceRect = clipState.mDirtyRectInDevPx;
|
|
}
|
|
|
|
maskInUserSpace = maskInUserSpace.Union(currentMaskSurfaceRect);
|
|
}
|
|
|
|
if (!maskInUserSpace.IsEmpty()) {
|
|
aParams.maskRect = Some(ToRect(maskInUserSpace));
|
|
} else {
|
|
aParams.maskRect = Nothing();
|
|
}
|
|
}
|
|
|
|
nsDisplayMasksAndClipPaths::nsDisplayMasksAndClipPaths(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame, nsDisplayList* aList,
|
|
const ActiveScrolledRoot* aActiveScrolledRoot)
|
|
: nsDisplayEffectsBase(aBuilder, aFrame, aList, aActiveScrolledRoot, true) {
|
|
MOZ_COUNT_CTOR(nsDisplayMasksAndClipPaths);
|
|
|
|
nsPresContext* presContext = mFrame->PresContext();
|
|
uint32_t flags =
|
|
aBuilder->GetBackgroundPaintFlags() | nsCSSRendering::PAINTBG_MASK_IMAGE;
|
|
const nsStyleSVGReset* svgReset = aFrame->StyleSVGReset();
|
|
NS_FOR_VISIBLE_IMAGE_LAYERS_BACK_TO_FRONT(i, svgReset->mMask) {
|
|
if (!svgReset->mMask.mLayers[i].mImage.IsResolved()) {
|
|
continue;
|
|
}
|
|
bool isTransformedFixed;
|
|
nsBackgroundLayerState state = nsCSSRendering::PrepareImageLayer(
|
|
presContext, aFrame, flags, mFrame->GetRectRelativeToSelf(),
|
|
mFrame->GetRectRelativeToSelf(), svgReset->mMask.mLayers[i],
|
|
&isTransformedFixed);
|
|
mDestRects.AppendElement(state.mDestArea);
|
|
}
|
|
}
|
|
|
|
static bool CanMergeDisplayMaskFrame(nsIFrame* aFrame) {
|
|
// Do not merge items for box-decoration-break:clone elements,
|
|
// since each box should have its own mask in that case.
|
|
if (aFrame->StyleBorder()->mBoxDecorationBreak ==
|
|
mozilla::StyleBoxDecorationBreak::Clone) {
|
|
return false;
|
|
}
|
|
|
|
// Do not merge if either frame has a mask. Continuation frames should apply
|
|
// the mask independently (just like nsDisplayBackgroundImage).
|
|
if (aFrame->StyleSVGReset()->HasMask()) {
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayMasksAndClipPaths::CanMerge(const nsDisplayItem* aItem) const {
|
|
// Items for the same content element should be merged into a single
|
|
// compositing group.
|
|
if (!HasDifferentFrame(aItem) || !HasSameTypeAndClip(aItem) ||
|
|
!HasSameContent(aItem)) {
|
|
return false;
|
|
}
|
|
|
|
return CanMergeDisplayMaskFrame(mFrame) &&
|
|
CanMergeDisplayMaskFrame(aItem->Frame());
|
|
}
|
|
|
|
bool nsDisplayMasksAndClipPaths::IsValidMask() {
|
|
if (!ValidateSVGFrame()) {
|
|
return false;
|
|
}
|
|
|
|
if (mFrame->StyleEffects()->mOpacity == 0.0f && mHandleOpacity) {
|
|
return false;
|
|
}
|
|
|
|
nsIFrame* firstFrame =
|
|
nsLayoutUtils::FirstContinuationOrIBSplitSibling(mFrame);
|
|
|
|
if (SVGObserverUtils::GetAndObserveClipPath(firstFrame, nullptr) ==
|
|
SVGObserverUtils::eHasRefsSomeInvalid ||
|
|
SVGObserverUtils::GetAndObserveMasks(firstFrame, nullptr) ==
|
|
SVGObserverUtils::eHasRefsSomeInvalid) {
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayMasksAndClipPaths::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
if (!IsValidMask()) {
|
|
return nullptr;
|
|
}
|
|
|
|
RefPtr<ContainerLayer> container =
|
|
aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, &mList, aContainerParameters,
|
|
nullptr);
|
|
|
|
return container.forget();
|
|
}
|
|
|
|
bool nsDisplayMasksAndClipPaths::PaintMask(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aMaskContext,
|
|
bool* aMaskPainted) {
|
|
MOZ_ASSERT(aMaskContext->GetDrawTarget()->GetFormat() == SurfaceFormat::A8);
|
|
|
|
imgDrawingParams imgParams(aBuilder->GetImageDecodeFlags());
|
|
nsRect borderArea = nsRect(ToReferenceFrame(), mFrame->GetSize());
|
|
nsSVGIntegrationUtils::PaintFramesParams params(
|
|
*aMaskContext, mFrame, mBounds, borderArea, aBuilder, nullptr,
|
|
mHandleOpacity, imgParams);
|
|
ComputeMaskGeometry(params);
|
|
bool painted = nsSVGIntegrationUtils::PaintMask(params);
|
|
if (aMaskPainted) {
|
|
*aMaskPainted = painted;
|
|
}
|
|
|
|
nsDisplayMasksAndClipPathsGeometry::UpdateDrawResult(this, imgParams.result);
|
|
|
|
return imgParams.result == ImgDrawResult::SUCCESS ||
|
|
imgParams.result == ImgDrawResult::SUCCESS_NOT_COMPLETE;
|
|
}
|
|
|
|
LayerState nsDisplayMasksAndClipPaths::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
if (CanPaintOnMaskLayer(aManager)) {
|
|
LayerState result = RequiredLayerStateForChildren(
|
|
aBuilder, aManager, aParameters, mList, GetAnimatedGeometryRoot());
|
|
// When we're not active, FrameLayerBuilder will call PaintAsLayer()
|
|
// on us during painting. In that case we don't want a mask layer to
|
|
// be created, because PaintAsLayer() takes care of applying the mask.
|
|
// So we return LayerState::LAYER_SVG_EFFECTS instead of
|
|
// LayerState::LAYER_INACTIVE so that FrameLayerBuilder doesn't set a mask
|
|
// layer on our layer.
|
|
return result == LayerState::LAYER_INACTIVE ? LayerState::LAYER_SVG_EFFECTS
|
|
: result;
|
|
}
|
|
|
|
return LayerState::LAYER_SVG_EFFECTS;
|
|
}
|
|
|
|
bool nsDisplayMasksAndClipPaths::CanPaintOnMaskLayer(LayerManager* aManager) {
|
|
if (!aManager->IsWidgetLayerManager()) {
|
|
return false;
|
|
}
|
|
|
|
if (!nsSVGIntegrationUtils::IsMaskResourceReady(mFrame)) {
|
|
return false;
|
|
}
|
|
|
|
if (StaticPrefs::layers_draw_mask_debug()) {
|
|
return false;
|
|
}
|
|
|
|
// We don't currently support this item creating a mask
|
|
// for both the clip-path, and rounded rect clipping.
|
|
if (GetClip().GetRoundedRectCount() != 0) {
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
bool nsDisplayMasksAndClipPaths::ComputeVisibility(
|
|
nsDisplayListBuilder* aBuilder, nsRegion* aVisibleRegion) {
|
|
// Our children may be made translucent or arbitrarily deformed so we should
|
|
// not allow them to subtract area from aVisibleRegion.
|
|
nsRegion childrenVisible(GetPaintRect());
|
|
nsRect r = GetPaintRect().Intersect(mList.GetClippedBounds(aBuilder));
|
|
mList.ComputeVisibilityForSublist(aBuilder, &childrenVisible, r);
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayMasksAndClipPaths::ComputeInvalidationRegion(
|
|
nsDisplayListBuilder* aBuilder, const nsDisplayItemGeometry* aGeometry,
|
|
nsRegion* aInvalidRegion) const {
|
|
nsDisplayEffectsBase::ComputeInvalidationRegion(aBuilder, aGeometry,
|
|
aInvalidRegion);
|
|
|
|
auto* geometry =
|
|
static_cast<const nsDisplayMasksAndClipPathsGeometry*>(aGeometry);
|
|
bool snap;
|
|
nsRect bounds = GetBounds(aBuilder, &snap);
|
|
|
|
if (mDestRects.Length() != geometry->mDestRects.Length()) {
|
|
aInvalidRegion->Or(bounds, geometry->mBounds);
|
|
} else {
|
|
for (size_t i = 0; i < mDestRects.Length(); i++) {
|
|
if (!mDestRects[i].IsEqualInterior(geometry->mDestRects[i])) {
|
|
aInvalidRegion->Or(bounds, geometry->mBounds);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
if (aBuilder->ShouldSyncDecodeImages() &&
|
|
geometry->ShouldInvalidateToSyncDecodeImages()) {
|
|
const nsStyleSVGReset* svgReset = mFrame->StyleSVGReset();
|
|
NS_FOR_VISIBLE_IMAGE_LAYERS_BACK_TO_FRONT(i, svgReset->mMask) {
|
|
const auto& image = svgReset->mMask.mLayers[i].mImage;
|
|
if (image.IsImageRequestType()) {
|
|
aInvalidRegion->Or(*aInvalidRegion, bounds);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
void nsDisplayMasksAndClipPaths::PaintAsLayer(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx,
|
|
LayerManager* aManager) {
|
|
// Clip the drawing target by mVisibleRect, which contains the visible
|
|
// region of the target frame and its out-of-flow and inflow descendants.
|
|
gfxContext* context = aCtx;
|
|
|
|
Rect bounds = NSRectToRect(GetPaintRect(),
|
|
mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
bounds.RoundOut();
|
|
context->Clip(bounds);
|
|
|
|
imgDrawingParams imgParams(aBuilder->GetImageDecodeFlags());
|
|
nsRect borderArea = nsRect(ToReferenceFrame(), mFrame->GetSize());
|
|
nsSVGIntegrationUtils::PaintFramesParams params(
|
|
*aCtx, mFrame, GetPaintRect(), borderArea, aBuilder, aManager,
|
|
mHandleOpacity, imgParams);
|
|
|
|
ComputeMaskGeometry(params);
|
|
|
|
nsSVGIntegrationUtils::PaintMaskAndClipPath(params);
|
|
|
|
context->PopClip();
|
|
|
|
nsDisplayMasksAndClipPathsGeometry::UpdateDrawResult(this, imgParams.result);
|
|
}
|
|
|
|
void nsDisplayMasksAndClipPaths::PaintWithContentsPaintCallback(
|
|
nsDisplayListBuilder* aBuilder, gfxContext* aCtx,
|
|
const std::function<void()>& aPaintChildren) {
|
|
// Clip the drawing target by mVisibleRect, which contains the visible
|
|
// region of the target frame and its out-of-flow and inflow descendants.
|
|
gfxContext* context = aCtx;
|
|
|
|
Rect bounds = NSRectToRect(GetPaintRect(),
|
|
mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
bounds.RoundOut();
|
|
context->Clip(bounds);
|
|
|
|
imgDrawingParams imgParams(aBuilder->GetImageDecodeFlags());
|
|
nsRect borderArea = nsRect(ToReferenceFrame(), mFrame->GetSize());
|
|
nsSVGIntegrationUtils::PaintFramesParams params(*aCtx, mFrame, GetPaintRect(),
|
|
borderArea, aBuilder, nullptr,
|
|
mHandleOpacity, imgParams);
|
|
|
|
ComputeMaskGeometry(params);
|
|
|
|
nsSVGIntegrationUtils::PaintMaskAndClipPath(params, aPaintChildren);
|
|
|
|
context->PopClip();
|
|
|
|
nsDisplayMasksAndClipPathsGeometry::UpdateDrawResult(this, imgParams.result);
|
|
}
|
|
|
|
static Maybe<wr::WrClipId> CreateSimpleClipRegion(
|
|
const nsDisplayMasksAndClipPaths& aDisplayItem,
|
|
wr::DisplayListBuilder& aBuilder) {
|
|
nsIFrame* frame = aDisplayItem.Frame();
|
|
auto* style = frame->StyleSVGReset();
|
|
MOZ_ASSERT(style->HasClipPath() || style->HasMask());
|
|
if (!nsSVGIntegrationUtils::UsingSimpleClipPathForFrame(frame)) {
|
|
return Nothing();
|
|
}
|
|
|
|
const auto& clipPath = style->mClipPath;
|
|
const auto& shape = *clipPath.AsShape()._0;
|
|
|
|
auto appUnitsPerDevPixel = frame->PresContext()->AppUnitsPerDevPixel();
|
|
const nsRect refBox =
|
|
nsLayoutUtils::ComputeGeometryBox(frame, clipPath.AsShape()._1);
|
|
|
|
AutoTArray<wr::ComplexClipRegion, 1> clipRegions;
|
|
|
|
wr::LayoutRect rect;
|
|
switch (shape.tag) {
|
|
case StyleBasicShape::Tag::Inset: {
|
|
const nsRect insetRect = ShapeUtils::ComputeInsetRect(shape, refBox) +
|
|
aDisplayItem.ToReferenceFrame();
|
|
|
|
nscoord radii[8] = {0};
|
|
|
|
if (ShapeUtils::ComputeInsetRadii(shape, insetRect, refBox, radii)) {
|
|
clipRegions.AppendElement(
|
|
wr::ToComplexClipRegion(insetRect, radii, appUnitsPerDevPixel));
|
|
}
|
|
|
|
rect = wr::ToLayoutRect(
|
|
LayoutDeviceRect::FromAppUnits(insetRect, appUnitsPerDevPixel));
|
|
break;
|
|
}
|
|
case StyleBasicShape::Tag::Ellipse:
|
|
case StyleBasicShape::Tag::Circle: {
|
|
nsPoint center = ShapeUtils::ComputeCircleOrEllipseCenter(shape, refBox);
|
|
|
|
nsSize radii;
|
|
if (shape.IsEllipse()) {
|
|
radii = ShapeUtils::ComputeEllipseRadii(shape, center, refBox);
|
|
} else {
|
|
nscoord radius = ShapeUtils::ComputeCircleRadius(shape, center, refBox);
|
|
radii = {radius, radius};
|
|
}
|
|
|
|
nsRect ellipseRect(aDisplayItem.ToReferenceFrame() + center -
|
|
nsPoint(radii.width, radii.height),
|
|
radii * 2);
|
|
|
|
nscoord ellipseRadii[8];
|
|
for (const auto corner : mozilla::AllPhysicalHalfCorners()) {
|
|
ellipseRadii[corner] =
|
|
HalfCornerIsX(corner) ? radii.width : radii.height;
|
|
}
|
|
|
|
clipRegions.AppendElement(wr::ToComplexClipRegion(
|
|
ellipseRect, ellipseRadii, appUnitsPerDevPixel));
|
|
|
|
rect = wr::ToLayoutRect(
|
|
LayoutDeviceRect::FromAppUnits(ellipseRect, appUnitsPerDevPixel));
|
|
break;
|
|
}
|
|
default:
|
|
// Please don't add more exceptions, try to find a way to define the clip
|
|
// without using a mask image.
|
|
//
|
|
// And if you _really really_ need to add an exception, add it to
|
|
// nsSVGIntegrationUtils::UsingSimpleClipPathForFrame
|
|
MOZ_ASSERT_UNREACHABLE("Unhandled shape id?");
|
|
return Nothing();
|
|
}
|
|
wr::WrClipId clipId =
|
|
aBuilder.DefineClip(Nothing(), rect, &clipRegions, nullptr);
|
|
return Some(clipId);
|
|
}
|
|
|
|
enum class HandleOpacity {
|
|
No,
|
|
Yes,
|
|
};
|
|
|
|
static Maybe<Pair<wr::WrClipId, HandleOpacity>> CreateWRClipPathAndMasks(
|
|
nsDisplayMasksAndClipPaths* aDisplayItem, const LayoutDeviceRect& aBounds,
|
|
wr::IpcResourceUpdateQueue& aResources, wr::DisplayListBuilder& aBuilder,
|
|
const StackingContextHelper& aSc, layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
if (auto clip = CreateSimpleClipRegion(*aDisplayItem, aBuilder)) {
|
|
return Some(MakePair(*clip, HandleOpacity::Yes));
|
|
}
|
|
|
|
Maybe<wr::ImageMask> mask = aManager->CommandBuilder().BuildWrMaskImage(
|
|
aDisplayItem, aBuilder, aResources, aSc, aDisplayListBuilder, aBounds);
|
|
if (!mask) {
|
|
return Nothing();
|
|
}
|
|
|
|
wr::WrClipId clipId = aBuilder.DefineClip(
|
|
Nothing(), wr::ToLayoutRect(aBounds), nullptr, mask.ptr());
|
|
|
|
return Some(MakePair(clipId, HandleOpacity::No));
|
|
}
|
|
|
|
bool nsDisplayMasksAndClipPaths::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
bool snap;
|
|
auto appUnitsPerDevPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
nsRect displayBounds = GetBounds(aDisplayListBuilder, &snap);
|
|
LayoutDeviceRect bounds =
|
|
LayoutDeviceRect::FromAppUnits(displayBounds, appUnitsPerDevPixel);
|
|
|
|
Maybe<Pair<wr::WrClipId, HandleOpacity>> clip = CreateWRClipPathAndMasks(
|
|
this, bounds, aResources, aBuilder, aSc, aManager, aDisplayListBuilder);
|
|
|
|
Maybe<StackingContextHelper> layer;
|
|
const StackingContextHelper* sc = &aSc;
|
|
if (clip) {
|
|
// Create a new stacking context to attach the mask to, ensuring the mask is
|
|
// applied to the aggregate, and not the individual elements.
|
|
|
|
// The stacking context shouldn't have any offset.
|
|
bounds.MoveTo(0, 0);
|
|
|
|
wr::WrClipId clipId = clip->first();
|
|
|
|
Maybe<float> opacity = clip->second() == HandleOpacity::Yes
|
|
? Some(mFrame->StyleEffects()->mOpacity)
|
|
: Nothing();
|
|
|
|
wr::StackingContextParams params;
|
|
params.clip = wr::WrStackingContextClip::ClipId(clipId);
|
|
params.opacity = opacity.ptrOr(nullptr);
|
|
layer.emplace(aSc, GetActiveScrolledRoot(), mFrame, this, aBuilder, params,
|
|
bounds);
|
|
sc = layer.ptr();
|
|
}
|
|
|
|
nsDisplayEffectsBase::CreateWebRenderCommands(aBuilder, aResources, *sc,
|
|
aManager, aDisplayListBuilder);
|
|
|
|
return true;
|
|
}
|
|
|
|
Maybe<nsRect> nsDisplayMasksAndClipPaths::GetClipWithRespectToASR(
|
|
nsDisplayListBuilder* aBuilder, const ActiveScrolledRoot* aASR) const {
|
|
if (const DisplayItemClip* clip =
|
|
DisplayItemClipChain::ClipForASR(GetClipChain(), aASR)) {
|
|
return Some(clip->GetClipRect());
|
|
}
|
|
// This item does not have a clip with respect to |aASR|. However, we
|
|
// might still have finite bounds with respect to |aASR|. Check our
|
|
// children.
|
|
nsDisplayList* childList = GetSameCoordinateSystemChildren();
|
|
if (childList) {
|
|
return Some(childList->GetClippedBoundsWithRespectToASR(aBuilder, aASR));
|
|
}
|
|
#ifdef DEBUG
|
|
MOZ_ASSERT(false, "item should have finite clip with respect to aASR");
|
|
#endif
|
|
return Nothing();
|
|
}
|
|
|
|
#ifdef MOZ_DUMP_PAINTING
|
|
void nsDisplayMasksAndClipPaths::PrintEffects(nsACString& aTo) {
|
|
nsIFrame* firstFrame =
|
|
nsLayoutUtils::FirstContinuationOrIBSplitSibling(mFrame);
|
|
bool first = true;
|
|
aTo += " effects=(";
|
|
if (mHandleOpacity) {
|
|
first = false;
|
|
aTo += nsPrintfCString("opacity(%f)", mFrame->StyleEffects()->mOpacity);
|
|
}
|
|
nsSVGClipPathFrame* clipPathFrame;
|
|
// XXX Check return value?
|
|
SVGObserverUtils::GetAndObserveClipPath(firstFrame, &clipPathFrame);
|
|
if (clipPathFrame) {
|
|
if (!first) {
|
|
aTo += ", ";
|
|
}
|
|
aTo += nsPrintfCString(
|
|
"clip(%s)", clipPathFrame->IsTrivial() ? "trivial" : "non-trivial");
|
|
first = false;
|
|
} else if (mFrame->StyleSVGReset()->HasClipPath()) {
|
|
if (!first) {
|
|
aTo += ", ";
|
|
}
|
|
aTo += "clip(basic-shape)";
|
|
first = false;
|
|
}
|
|
|
|
nsTArray<nsSVGMaskFrame*> masks;
|
|
// XXX check return value?
|
|
SVGObserverUtils::GetAndObserveMasks(firstFrame, &masks);
|
|
if (!masks.IsEmpty() && masks[0]) {
|
|
if (!first) {
|
|
aTo += ", ";
|
|
}
|
|
aTo += "mask";
|
|
}
|
|
aTo += ")";
|
|
}
|
|
#endif
|
|
|
|
already_AddRefed<Layer> nsDisplayBackdropRootContainer::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
RefPtr<Layer> container = aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, &mList, aContainerParameters, nullptr);
|
|
if (!container) {
|
|
return nullptr;
|
|
}
|
|
|
|
return container.forget();
|
|
}
|
|
|
|
LayerState nsDisplayBackdropRootContainer::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
return RequiredLayerStateForChildren(aBuilder, aManager, aParameters, mList,
|
|
GetAnimatedGeometryRoot());
|
|
}
|
|
|
|
bool nsDisplayBackdropRootContainer::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
wr::StackingContextParams params;
|
|
params.is_backdrop_root = true;
|
|
params.clip =
|
|
wr::WrStackingContextClip::ClipChain(aBuilder.CurrentClipChainId());
|
|
StackingContextHelper sc(aSc, GetActiveScrolledRoot(), mFrame, this, aBuilder,
|
|
params);
|
|
|
|
nsDisplayWrapList::CreateWebRenderCommands(aBuilder, aResources, sc, aManager,
|
|
aDisplayListBuilder);
|
|
return true;
|
|
}
|
|
|
|
/* static */
|
|
bool nsDisplayBackdropFilters::CanCreateWebRenderCommands(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aFrame) {
|
|
return nsSVGIntegrationUtils::CanCreateWebRenderFiltersForFrame(aFrame);
|
|
}
|
|
|
|
bool nsDisplayBackdropFilters::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
WrFiltersHolder wrFilters;
|
|
Maybe<nsRect> filterClip;
|
|
auto filterChain = mFrame->StyleEffects()->mBackdropFilters.AsSpan();
|
|
if (!nsSVGIntegrationUtils::CreateWebRenderCSSFilters(filterChain, mFrame,
|
|
wrFilters) &&
|
|
!nsSVGIntegrationUtils::BuildWebRenderFilters(mFrame, filterChain,
|
|
wrFilters, filterClip)) {
|
|
return false;
|
|
}
|
|
|
|
nsCSSRendering::ImageLayerClipState clip;
|
|
nsCSSRendering::GetImageLayerClip(
|
|
mFrame->StyleBackground()->BottomLayer(), mFrame, *mFrame->StyleBorder(),
|
|
mBackdropRect, mBackdropRect, false,
|
|
mFrame->PresContext()->AppUnitsPerDevPixel(), &clip);
|
|
|
|
LayoutDeviceRect bounds = LayoutDeviceRect::FromAppUnits(
|
|
mBackdropRect, mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
|
|
wr::ComplexClipRegion region =
|
|
wr::ToComplexClipRegion(clip.mBGClipArea, clip.mRadii,
|
|
mFrame->PresContext()->AppUnitsPerDevPixel());
|
|
|
|
aBuilder.PushBackdropFilter(wr::ToLayoutRect(bounds), region,
|
|
wrFilters.filters, wrFilters.filter_datas,
|
|
!BackfaceIsHidden());
|
|
|
|
wr::StackingContextParams params;
|
|
params.clip =
|
|
wr::WrStackingContextClip::ClipChain(aBuilder.CurrentClipChainId());
|
|
StackingContextHelper sc(aSc, GetActiveScrolledRoot(), mFrame, this, aBuilder,
|
|
params);
|
|
|
|
nsDisplayWrapList::CreateWebRenderCommands(aBuilder, aResources, sc, aManager,
|
|
aDisplayListBuilder);
|
|
return true;
|
|
}
|
|
|
|
/* static */
|
|
nsDisplayFilters::nsDisplayFilters(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame, nsDisplayList* aList)
|
|
: nsDisplayEffectsBase(aBuilder, aFrame, aList),
|
|
mEffectsBounds(aFrame->GetVisualOverflowRectRelativeToSelf()) {
|
|
MOZ_COUNT_CTOR(nsDisplayFilters);
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayFilters::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
if (!ValidateSVGFrame()) {
|
|
return nullptr;
|
|
}
|
|
|
|
if (mFrame->StyleEffects()->mOpacity == 0.0f && mHandleOpacity) {
|
|
return nullptr;
|
|
}
|
|
|
|
nsIFrame* firstFrame =
|
|
nsLayoutUtils::FirstContinuationOrIBSplitSibling(mFrame);
|
|
|
|
// We may exist for a mix of CSS filter functions and/or references to SVG
|
|
// filters. If we have invalid references to SVG filters then we paint
|
|
// nothing, so no need for a layer.
|
|
if (SVGObserverUtils::GetAndObserveFilters(firstFrame, nullptr) ==
|
|
SVGObserverUtils::eHasRefsSomeInvalid) {
|
|
return nullptr;
|
|
}
|
|
|
|
ContainerLayerParameters newContainerParameters = aContainerParameters;
|
|
newContainerParameters.mDisableSubpixelAntialiasingInDescendants = true;
|
|
|
|
RefPtr<ContainerLayer> container =
|
|
aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, &mList, newContainerParameters,
|
|
nullptr);
|
|
return container.forget();
|
|
}
|
|
|
|
LayerState nsDisplayFilters::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
return LayerState::LAYER_SVG_EFFECTS;
|
|
}
|
|
|
|
bool nsDisplayFilters::ComputeVisibility(nsDisplayListBuilder* aBuilder,
|
|
nsRegion* aVisibleRegion) {
|
|
nsPoint offset = ToReferenceFrame();
|
|
nsRect dirtyRect = nsSVGIntegrationUtils::GetRequiredSourceForInvalidArea(
|
|
mFrame, GetPaintRect() - offset) +
|
|
offset;
|
|
|
|
// Our children may be made translucent or arbitrarily deformed so we should
|
|
// not allow them to subtract area from aVisibleRegion.
|
|
nsRegion childrenVisible(dirtyRect);
|
|
nsRect r = dirtyRect.Intersect(
|
|
mList.GetClippedBoundsWithRespectToASR(aBuilder, mActiveScrolledRoot));
|
|
mList.ComputeVisibilityForSublist(aBuilder, &childrenVisible, r);
|
|
return true;
|
|
}
|
|
|
|
void nsDisplayFilters::ComputeInvalidationRegion(
|
|
nsDisplayListBuilder* aBuilder, const nsDisplayItemGeometry* aGeometry,
|
|
nsRegion* aInvalidRegion) const {
|
|
nsDisplayEffectsBase::ComputeInvalidationRegion(aBuilder, aGeometry,
|
|
aInvalidRegion);
|
|
|
|
auto* geometry = static_cast<const nsDisplayFiltersGeometry*>(aGeometry);
|
|
|
|
if (aBuilder->ShouldSyncDecodeImages() &&
|
|
geometry->ShouldInvalidateToSyncDecodeImages()) {
|
|
bool snap;
|
|
nsRect bounds = GetBounds(aBuilder, &snap);
|
|
aInvalidRegion->Or(*aInvalidRegion, bounds);
|
|
}
|
|
}
|
|
|
|
void nsDisplayFilters::PaintAsLayer(nsDisplayListBuilder* aBuilder,
|
|
gfxContext* aCtx, LayerManager* aManager) {
|
|
imgDrawingParams imgParams(aBuilder->GetImageDecodeFlags());
|
|
nsRect borderArea = nsRect(ToReferenceFrame(), mFrame->GetSize());
|
|
nsSVGIntegrationUtils::PaintFramesParams params(
|
|
*aCtx, mFrame, GetPaintRect(), borderArea, aBuilder, aManager,
|
|
mHandleOpacity, imgParams);
|
|
nsSVGIntegrationUtils::PaintFilter(params);
|
|
nsDisplayFiltersGeometry::UpdateDrawResult(this, imgParams.result);
|
|
}
|
|
|
|
bool nsDisplayFilters::CanCreateWebRenderCommands() {
|
|
return nsSVGIntegrationUtils::CanCreateWebRenderFiltersForFrame(mFrame);
|
|
}
|
|
|
|
bool nsDisplayFilters::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
float auPerDevPixel = mFrame->PresContext()->AppUnitsPerDevPixel();
|
|
|
|
WrFiltersHolder wrFilters;
|
|
Maybe<nsRect> filterClip;
|
|
auto filterChain = mFrame->StyleEffects()->mFilters.AsSpan();
|
|
if (!nsSVGIntegrationUtils::CreateWebRenderCSSFilters(filterChain, mFrame,
|
|
wrFilters) &&
|
|
!nsSVGIntegrationUtils::BuildWebRenderFilters(mFrame, filterChain,
|
|
wrFilters, filterClip)) {
|
|
return false;
|
|
}
|
|
|
|
wr::WrStackingContextClip clip;
|
|
if (filterClip) {
|
|
auto devPxRect = LayoutDeviceRect::FromAppUnits(
|
|
filterClip.value() + ToReferenceFrame(), auPerDevPixel);
|
|
wr::WrClipId clipId =
|
|
aBuilder.DefineClip(Nothing(), wr::ToLayoutRect(devPxRect));
|
|
clip = wr::WrStackingContextClip::ClipId(clipId);
|
|
} else {
|
|
clip = wr::WrStackingContextClip::ClipChain(aBuilder.CurrentClipChainId());
|
|
}
|
|
|
|
float opacity = mFrame->StyleEffects()->mOpacity;
|
|
wr::StackingContextParams params;
|
|
params.mFilters = std::move(wrFilters.filters);
|
|
params.mFilterDatas = std::move(wrFilters.filter_datas);
|
|
params.opacity = opacity != 1.0f && mHandleOpacity ? &opacity : nullptr;
|
|
params.clip = clip;
|
|
StackingContextHelper sc(aSc, GetActiveScrolledRoot(), mFrame, this, aBuilder,
|
|
params);
|
|
|
|
nsDisplayEffectsBase::CreateWebRenderCommands(aBuilder, aResources, sc,
|
|
aManager, aDisplayListBuilder);
|
|
|
|
return true;
|
|
}
|
|
|
|
#ifdef MOZ_DUMP_PAINTING
|
|
void nsDisplayFilters::PrintEffects(nsACString& aTo) {
|
|
nsIFrame* firstFrame =
|
|
nsLayoutUtils::FirstContinuationOrIBSplitSibling(mFrame);
|
|
bool first = true;
|
|
aTo += " effects=(";
|
|
if (mHandleOpacity) {
|
|
first = false;
|
|
aTo += nsPrintfCString("opacity(%f)", mFrame->StyleEffects()->mOpacity);
|
|
}
|
|
// We may exist for a mix of CSS filter functions and/or references to SVG
|
|
// filters. If we have invalid references to SVG filters then we paint
|
|
// nothing, but otherwise we will apply one or more filters.
|
|
if (SVGObserverUtils::GetAndObserveFilters(firstFrame, nullptr) !=
|
|
SVGObserverUtils::eHasRefsSomeInvalid) {
|
|
if (!first) {
|
|
aTo += ", ";
|
|
}
|
|
aTo += "filter";
|
|
}
|
|
aTo += ")";
|
|
}
|
|
#endif
|
|
|
|
nsDisplaySVGWrapper::nsDisplaySVGWrapper(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame, nsDisplayList* aList)
|
|
: nsDisplayWrapList(aBuilder, aFrame, aList) {
|
|
MOZ_COUNT_CTOR(nsDisplaySVGWrapper);
|
|
}
|
|
|
|
LayerState nsDisplaySVGWrapper::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
RefPtr<LayerManager> layerManager = aBuilder->GetWidgetLayerManager();
|
|
if (layerManager &&
|
|
layerManager->GetBackendType() == layers::LayersBackend::LAYERS_WR) {
|
|
return LayerState::LAYER_ACTIVE_FORCE;
|
|
}
|
|
return LayerState::LAYER_NONE;
|
|
}
|
|
|
|
bool nsDisplaySVGWrapper::ShouldFlattenAway(nsDisplayListBuilder* aBuilder) {
|
|
RefPtr<LayerManager> layerManager = aBuilder->GetWidgetLayerManager();
|
|
if (layerManager &&
|
|
layerManager->GetBackendType() == layers::LayersBackend::LAYERS_WR) {
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplaySVGWrapper::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
ContainerLayerParameters newContainerParameters = aContainerParameters;
|
|
newContainerParameters.mDisableSubpixelAntialiasingInDescendants = true;
|
|
|
|
RefPtr<ContainerLayer> container =
|
|
aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, &mList, newContainerParameters,
|
|
nullptr);
|
|
|
|
return container.forget();
|
|
}
|
|
|
|
bool nsDisplaySVGWrapper::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
return nsDisplayWrapList::CreateWebRenderCommands(
|
|
aBuilder, aResources, aSc, aManager, aDisplayListBuilder);
|
|
}
|
|
|
|
nsDisplayForeignObject::nsDisplayForeignObject(nsDisplayListBuilder* aBuilder,
|
|
nsIFrame* aFrame,
|
|
nsDisplayList* aList)
|
|
: nsDisplayWrapList(aBuilder, aFrame, aList) {
|
|
MOZ_COUNT_CTOR(nsDisplayForeignObject);
|
|
}
|
|
|
|
#ifdef NS_BUILD_REFCNT_LOGGING
|
|
nsDisplayForeignObject::~nsDisplayForeignObject() {
|
|
MOZ_COUNT_DTOR(nsDisplayForeignObject);
|
|
}
|
|
#endif
|
|
|
|
LayerState nsDisplayForeignObject::GetLayerState(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aParameters) {
|
|
RefPtr<LayerManager> layerManager = aBuilder->GetWidgetLayerManager();
|
|
if (layerManager &&
|
|
layerManager->GetBackendType() == layers::LayersBackend::LAYERS_WR) {
|
|
return LayerState::LAYER_ACTIVE_FORCE;
|
|
}
|
|
return LayerState::LAYER_NONE;
|
|
}
|
|
|
|
bool nsDisplayForeignObject::ShouldFlattenAway(nsDisplayListBuilder* aBuilder) {
|
|
RefPtr<LayerManager> layerManager = aBuilder->GetWidgetLayerManager();
|
|
if (layerManager &&
|
|
layerManager->GetBackendType() == layers::LayersBackend::LAYERS_WR) {
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
|
|
already_AddRefed<Layer> nsDisplayForeignObject::BuildLayer(
|
|
nsDisplayListBuilder* aBuilder, LayerManager* aManager,
|
|
const ContainerLayerParameters& aContainerParameters) {
|
|
ContainerLayerParameters newContainerParameters = aContainerParameters;
|
|
newContainerParameters.mDisableSubpixelAntialiasingInDescendants = true;
|
|
|
|
RefPtr<ContainerLayer> container =
|
|
aManager->GetLayerBuilder()->BuildContainerLayerFor(
|
|
aBuilder, aManager, mFrame, this, &mList, newContainerParameters,
|
|
nullptr);
|
|
|
|
return container.forget();
|
|
}
|
|
|
|
bool nsDisplayForeignObject::CreateWebRenderCommands(
|
|
mozilla::wr::DisplayListBuilder& aBuilder,
|
|
mozilla::wr::IpcResourceUpdateQueue& aResources,
|
|
const StackingContextHelper& aSc,
|
|
mozilla::layers::RenderRootStateManager* aManager,
|
|
nsDisplayListBuilder* aDisplayListBuilder) {
|
|
AutoRestore<bool> restoreDoGrouping(aManager->CommandBuilder().mDoGrouping);
|
|
aManager->CommandBuilder().mDoGrouping = false;
|
|
return nsDisplayWrapList::CreateWebRenderCommands(
|
|
aBuilder, aResources, aSc, aManager, aDisplayListBuilder);
|
|
}
|
|
|
|
void nsDisplayListCollection::SerializeWithCorrectZOrder(
|
|
nsDisplayList* aOutResultList, nsIContent* aContent) {
|
|
// Sort PositionedDescendants() in CSS 'z-order' order. The list is already
|
|
// in content document order and SortByZOrder is a stable sort which
|
|
// guarantees that boxes produced by the same element are placed together
|
|
// in the sort. Consider a position:relative inline element that breaks
|
|
// across lines and has absolutely positioned children; all the abs-pos
|
|
// children should be z-ordered after all the boxes for the position:relative
|
|
// element itself.
|
|
PositionedDescendants()->SortByZOrder();
|
|
|
|
// Now follow the rules of http://www.w3.org/TR/CSS21/zindex.html
|
|
// 1,2: backgrounds and borders
|
|
aOutResultList->AppendToTop(BorderBackground());
|
|
// 3: negative z-index children.
|
|
for (;;) {
|
|
nsDisplayItem* item = PositionedDescendants()->GetBottom();
|
|
if (item && item->ZIndex() < 0) {
|
|
PositionedDescendants()->RemoveBottom();
|
|
aOutResultList->AppendToTop(item);
|
|
continue;
|
|
}
|
|
break;
|
|
}
|
|
// 4: block backgrounds
|
|
aOutResultList->AppendToTop(BlockBorderBackgrounds());
|
|
// 5: floats
|
|
aOutResultList->AppendToTop(Floats());
|
|
// 7: general content
|
|
aOutResultList->AppendToTop(Content());
|
|
// 7.5: outlines, in content tree order. We need to sort by content order
|
|
// because an element with outline that breaks and has children with outline
|
|
// might have placed child outline items between its own outline items.
|
|
// The element's outline items need to all come before any child outline
|
|
// items.
|
|
if (aContent) {
|
|
Outlines()->SortByContentOrder(aContent);
|
|
}
|
|
aOutResultList->AppendToTop(Outlines());
|
|
// 8, 9: non-negative z-index children
|
|
aOutResultList->AppendToTop(PositionedDescendants());
|
|
}
|
|
|
|
namespace mozilla {
|
|
|
|
uint32_t PaintTelemetry::sPaintLevel = 0;
|
|
uint32_t PaintTelemetry::sMetricLevel = 0;
|
|
EnumeratedArray<PaintTelemetry::Metric, PaintTelemetry::Metric::COUNT, double>
|
|
PaintTelemetry::sMetrics;
|
|
|
|
PaintTelemetry::AutoRecordPaint::AutoRecordPaint() {
|
|
// Don't record nested paints.
|
|
if (sPaintLevel++ > 0) {
|
|
return;
|
|
}
|
|
|
|
// Reset metrics for a new paint.
|
|
for (auto& metric : sMetrics) {
|
|
metric = 0.0;
|
|
}
|
|
mStart = TimeStamp::Now();
|
|
}
|
|
|
|
PaintTelemetry::AutoRecordPaint::~AutoRecordPaint() {
|
|
MOZ_ASSERT(sPaintLevel != 0);
|
|
if (--sPaintLevel > 0) {
|
|
return;
|
|
}
|
|
|
|
// If we're in multi-process mode, don't include paint times for the parent
|
|
// process.
|
|
if (gfxVars::BrowserTabsRemoteAutostart() && XRE_IsParentProcess()) {
|
|
return;
|
|
}
|
|
|
|
double totalMs = (TimeStamp::Now() - mStart).ToMilliseconds();
|
|
|
|
// Record the total time.
|
|
Telemetry::Accumulate(Telemetry::CONTENT_PAINT_TIME,
|
|
static_cast<uint32_t>(totalMs));
|
|
|
|
// Helpers for recording large/small paints.
|
|
auto recordLarge = [=](const nsCString& aKey, double aDurationMs) -> void {
|
|
MOZ_ASSERT(aDurationMs <= totalMs);
|
|
uint32_t amount = static_cast<int32_t>((aDurationMs / totalMs) * 100.0);
|
|
Telemetry::Accumulate(Telemetry::CONTENT_LARGE_PAINT_PHASE_WEIGHT, aKey,
|
|
amount);
|
|
};
|
|
auto recordSmall = [=](const nsCString& aKey, double aDurationMs) -> void {
|
|
MOZ_ASSERT(aDurationMs <= totalMs);
|
|
uint32_t amount = static_cast<int32_t>((aDurationMs / totalMs) * 100.0);
|
|
Telemetry::Accumulate(Telemetry::CONTENT_SMALL_PAINT_PHASE_WEIGHT, aKey,
|
|
amount);
|
|
};
|
|
|
|
double dlMs = sMetrics[Metric::DisplayList];
|
|
double flbMs = sMetrics[Metric::Layerization];
|
|
double frMs = sMetrics[Metric::FlushRasterization];
|
|
double rMs = sMetrics[Metric::Rasterization];
|
|
|
|
// If the total time was >= 16ms, then it's likely we missed a frame due to
|
|
// painting. We bucket these metrics separately.
|
|
if (totalMs >= 16.0) {
|
|
recordLarge(NS_LITERAL_CSTRING("dl"), dlMs);
|
|
recordLarge(NS_LITERAL_CSTRING("flb"), flbMs);
|
|
recordLarge(NS_LITERAL_CSTRING("fr"), frMs);
|
|
recordLarge(NS_LITERAL_CSTRING("r"), rMs);
|
|
} else {
|
|
recordSmall(NS_LITERAL_CSTRING("dl"), dlMs);
|
|
recordSmall(NS_LITERAL_CSTRING("flb"), flbMs);
|
|
recordSmall(NS_LITERAL_CSTRING("fr"), frMs);
|
|
recordSmall(NS_LITERAL_CSTRING("r"), rMs);
|
|
}
|
|
|
|
Telemetry::Accumulate(Telemetry::PAINT_BUILD_LAYERS_TIME, flbMs);
|
|
}
|
|
|
|
PaintTelemetry::AutoRecord::AutoRecord(Metric aMetric) : mMetric(aMetric) {
|
|
// Don't double-record anything nested.
|
|
if (sMetricLevel++ > 0) {
|
|
return;
|
|
}
|
|
|
|
// Don't record inside nested paints, or outside of paints.
|
|
if (sPaintLevel != 1) {
|
|
return;
|
|
}
|
|
|
|
mStart = TimeStamp::Now();
|
|
}
|
|
|
|
PaintTelemetry::AutoRecord::~AutoRecord() {
|
|
MOZ_ASSERT(sMetricLevel != 0);
|
|
|
|
sMetricLevel--;
|
|
if (mStart.IsNull()) {
|
|
return;
|
|
}
|
|
|
|
sMetrics[mMetric] += (TimeStamp::Now() - mStart).ToMilliseconds();
|
|
}
|
|
|
|
} // namespace mozilla
|
|
|
|
static nsIFrame* GetSelfOrPlaceholderFor(nsIFrame* aFrame) {
|
|
if (aFrame->GetStateBits() & NS_FRAME_IS_PUSHED_FLOAT) {
|
|
return aFrame;
|
|
}
|
|
|
|
if ((aFrame->GetStateBits() & NS_FRAME_OUT_OF_FLOW) &&
|
|
!aFrame->GetPrevInFlow()) {
|
|
return aFrame->GetPlaceholderFrame();
|
|
}
|
|
|
|
return aFrame;
|
|
}
|
|
|
|
static nsIFrame* GetAncestorFor(nsIFrame* aFrame) {
|
|
nsIFrame* f = GetSelfOrPlaceholderFor(aFrame);
|
|
MOZ_ASSERT(f);
|
|
return nsLayoutUtils::GetCrossDocParentFrame(f);
|
|
}
|
|
|
|
nsDisplayListBuilder::AutoBuildingDisplayList::AutoBuildingDisplayList(
|
|
nsDisplayListBuilder* aBuilder, nsIFrame* aForChild,
|
|
const nsRect& aVisibleRect, const nsRect& aDirtyRect,
|
|
const bool aIsTransformed)
|
|
: mBuilder(aBuilder),
|
|
mPrevFrame(aBuilder->mCurrentFrame),
|
|
mPrevReferenceFrame(aBuilder->mCurrentReferenceFrame),
|
|
mPrevHitTestArea(aBuilder->mHitTestArea),
|
|
mPrevHitTestInfo(aBuilder->mHitTestInfo),
|
|
mPrevOffset(aBuilder->mCurrentOffsetToReferenceFrame),
|
|
mPrevVisibleRect(aBuilder->mVisibleRect),
|
|
mPrevDirtyRect(aBuilder->mDirtyRect),
|
|
mPrevAGR(aBuilder->mCurrentAGR),
|
|
mPrevAncestorHasApzAwareEventHandler(
|
|
aBuilder->mAncestorHasApzAwareEventHandler),
|
|
mPrevBuildingInvisibleItems(aBuilder->mBuildingInvisibleItems),
|
|
mPrevInInvalidSubtree(aBuilder->mInInvalidSubtree) {
|
|
if (aIsTransformed) {
|
|
aBuilder->mCurrentOffsetToReferenceFrame = nsPoint();
|
|
aBuilder->mCurrentReferenceFrame = aForChild;
|
|
} else if (aBuilder->mCurrentFrame == aForChild->GetParent()) {
|
|
aBuilder->mCurrentOffsetToReferenceFrame += aForChild->GetPosition();
|
|
} else {
|
|
aBuilder->mCurrentReferenceFrame = aBuilder->FindReferenceFrameFor(
|
|
aForChild, &aBuilder->mCurrentOffsetToReferenceFrame);
|
|
}
|
|
|
|
bool isAsync;
|
|
mCurrentAGRState = aBuilder->IsAnimatedGeometryRoot(aForChild, isAsync);
|
|
|
|
if (aBuilder->mCurrentFrame == aForChild->GetParent()) {
|
|
if (mCurrentAGRState == AGR_YES) {
|
|
aBuilder->mCurrentAGR =
|
|
aBuilder->WrapAGRForFrame(aForChild, isAsync, aBuilder->mCurrentAGR);
|
|
}
|
|
} else if (aBuilder->mCurrentFrame != aForChild) {
|
|
aBuilder->mCurrentAGR = aBuilder->FindAnimatedGeometryRootFor(aForChild);
|
|
}
|
|
|
|
MOZ_ASSERT(nsLayoutUtils::IsAncestorFrameCrossDoc(
|
|
aBuilder->RootReferenceFrame(), *aBuilder->mCurrentAGR));
|
|
|
|
// If aForChild is being visited from a frame other than it's ancestor frame,
|
|
// mInInvalidSubtree will need to be recalculated the slow way.
|
|
if (aForChild == mPrevFrame || GetAncestorFor(aForChild) == mPrevFrame) {
|
|
aBuilder->mInInvalidSubtree =
|
|
aBuilder->mInInvalidSubtree || aForChild->IsFrameModified();
|
|
} else {
|
|
aBuilder->mInInvalidSubtree = AnyContentAncestorModified(aForChild);
|
|
}
|
|
|
|
aBuilder->mCurrentFrame = aForChild;
|
|
aBuilder->mVisibleRect = aVisibleRect;
|
|
aBuilder->mDirtyRect =
|
|
aBuilder->mInInvalidSubtree ? aVisibleRect : aDirtyRect;
|
|
}
|