Title: [174367] trunk/Source/WebKit2
- Revision
- 174367
- Author
- [email protected]
- Date
- 2014-10-06 14:52:42 -0700 (Mon, 06 Oct 2014)
Log Message
Don't attempt to paint into zero-sized backing store
https://bugs.webkit.org/show_bug.cgi?id=137465
Reviewed by Tim Horton.
Page scale could cause the backing store for a small composited element to become empty,
in which case we'd try to allocate, and paint into a graphics context with no surface
behind it.
Fix by bailing from RemoteLayerBackingStore::display() when checking the backing store
size after accounting for scale.
* Shared/mac/RemoteLayerBackingStore.h:
* Shared/mac/RemoteLayerBackingStore.mm:
(WebKit::RemoteLayerBackingStore::backingStoreSize):
(WebKit::RemoteLayerBackingStore::swapToValidFrontBuffer):
(WebKit::RemoteLayerBackingStore::display):
Modified Paths
Diff
Modified: trunk/Source/WebKit2/ChangeLog (174366 => 174367)
--- trunk/Source/WebKit2/ChangeLog 2014-10-06 21:37:28 UTC (rev 174366)
+++ trunk/Source/WebKit2/ChangeLog 2014-10-06 21:52:42 UTC (rev 174367)
@@ -1,3 +1,23 @@
+2014-10-06 Simon Fraser <[email protected]>
+
+ Don't attempt to paint into zero-sized backing store
+ https://bugs.webkit.org/show_bug.cgi?id=137465
+
+ Reviewed by Tim Horton.
+
+ Page scale could cause the backing store for a small composited element to become empty,
+ in which case we'd try to allocate, and paint into a graphics context with no surface
+ behind it.
+
+ Fix by bailing from RemoteLayerBackingStore::display() when checking the backing store
+ size after accounting for scale.
+
+ * Shared/mac/RemoteLayerBackingStore.h:
+ * Shared/mac/RemoteLayerBackingStore.mm:
+ (WebKit::RemoteLayerBackingStore::backingStoreSize):
+ (WebKit::RemoteLayerBackingStore::swapToValidFrontBuffer):
+ (WebKit::RemoteLayerBackingStore::display):
+
2014-10-06 Christophe Dumez <[email protected]>
Use is<>() / downcast<>() for ScrollingTree subclasses
Modified: trunk/Source/WebKit2/Shared/mac/RemoteLayerBackingStore.h (174366 => 174367)
--- trunk/Source/WebKit2/Shared/mac/RemoteLayerBackingStore.h 2014-10-06 21:37:28 UTC (rev 174366)
+++ trunk/Source/WebKit2/Shared/mac/RemoteLayerBackingStore.h 2014-10-06 21:52:42 UTC (rev 174367)
@@ -96,6 +96,8 @@
void drawInContext(WebCore::GraphicsContext&, CGImageRef backImage);
void clearBackingStore();
void swapToValidFrontBuffer();
+
+ WebCore::IntSize backingStoreSize() const;
PlatformCALayerRemote* m_layer;
Modified: trunk/Source/WebKit2/Shared/mac/RemoteLayerBackingStore.mm (174366 => 174367)
--- trunk/Source/WebKit2/Shared/mac/RemoteLayerBackingStore.mm 2014-10-06 21:37:28 UTC (rev 174366)
+++ trunk/Source/WebKit2/Shared/mac/RemoteLayerBackingStore.mm 2014-10-06 21:52:42 UTC (rev 174367)
@@ -171,12 +171,17 @@
setNeedsDisplay(IntRect(IntPoint(), expandedIntSize(m_size)));
}
-void RemoteLayerBackingStore::swapToValidFrontBuffer()
+IntSize RemoteLayerBackingStore::backingStoreSize() const
{
FloatSize scaledSize = m_size;
scaledSize.scale(m_scale);
- IntSize expandedScaledSize = roundedIntSize(scaledSize);
+ return roundedIntSize(scaledSize);
+}
+void RemoteLayerBackingStore::swapToValidFrontBuffer()
+{
+ IntSize expandedScaledSize = backingStoreSize();
+
#if USE(IOSURFACE)
if (m_acceleratesDrawing) {
if (!m_backBuffer.surface || m_backBuffer.surface->isInUse()) {
@@ -215,7 +220,9 @@
// Make the previous front buffer non-volatile early, so that we can dirty the whole layer if it comes back empty.
setBufferVolatility(BufferType::Front, false);
- if (m_dirtyRegion.isEmpty() || m_size.isEmpty())
+ IntSize expandedScaledSize = backingStoreSize();
+
+ if (m_dirtyRegion.isEmpty() || expandedScaledSize.isEmpty())
return false;
IntRect layerBounds(IntPoint(), expandedIntSize(m_size));
@@ -227,9 +234,6 @@
m_dirtyRegion.unite(indicatorRect);
}
- FloatSize scaledSize = m_size;
- scaledSize.scale(m_scale);
- IntSize expandedScaledSize = roundedIntSize(scaledSize);
IntRect expandedScaledLayerBounds(IntPoint(), expandedScaledSize);
bool willPaintEntireBackingStore = m_dirtyRegion.contains(layerBounds);
_______________________________________________
webkit-changes mailing list
[email protected]
https://lists.webkit.org/mailman/listinfo/webkit-changes
