1 files changed, 685 insertions, 0 deletions
diff --git a/gfx/layers/client/MultiTiledContentClient.cpp b/gfx/layers/client/MultiTiledContentClient.cpp
new file mode 100644
index 0000000000..2ab5c83e39
--- /dev/null
+++ b/gfx/layers/client/MultiTiledContentClient.cpp
@@ -0,0 +1,685 @@
+/* -*- 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/. */
+
+#include "mozilla/layers/MultiTiledContentClient.h"
+
+#include "ClientTiledPaintedLayer.h"
+#include "mozilla/StaticPrefs_layers.h"
+#include "mozilla/layers/APZUtils.h"
+#include "mozilla/layers/LayerMetricsWrapper.h"
+
+namespace mozilla {
+
+using namespace gfx;
+
+namespace layers {
+
+MultiTiledContentClient::MultiTiledContentClient(
+    ClientTiledPaintedLayer& aPaintedLayer, ClientLayerManager* aManager)
+    : TiledContentClient(aManager, "Multi"),
+      mTiledBuffer(aPaintedLayer, *this, aManager, &mSharedFrameMetricsHelper),
+      mLowPrecisionTiledBuffer(aPaintedLayer, *this, aManager,
+                               &mSharedFrameMetricsHelper) {
+  MOZ_COUNT_CTOR(MultiTiledContentClient);
+  mLowPrecisionTiledBuffer.SetResolution(
+      StaticPrefs::layers_low_precision_resolution());
+  mHasLowPrecision = StaticPrefs::layers_low_precision_buffer();
+}
+
+void MultiTiledContentClient::ClearCachedResources() {
+  CompositableClient::ClearCachedResources();
+  mTiledBuffer.DiscardBuffers();
+  mLowPrecisionTiledBuffer.DiscardBuffers();
+}
+
+void MultiTiledContentClient::UpdatedBuffer(TiledBufferType aType) {
+  ClientMultiTiledLayerBuffer* buffer = aType == LOW_PRECISION_TILED_BUFFER
+                                            ? &mLowPrecisionTiledBuffer
+                                            : &mTiledBuffer;
+
+  MOZ_ASSERT(aType != LOW_PRECISION_TILED_BUFFER || mHasLowPrecision);
+
+  mForwarder->UseTiledLayerBuffer(this, buffer->GetSurfaceDescriptorTiles());
+}
+
+ClientMultiTiledLayerBuffer::ClientMultiTiledLayerBuffer(
+    ClientTiledPaintedLayer& aPaintedLayer,
+    CompositableClient& aCompositableClient, ClientLayerManager* aManager,
+    SharedFrameMetricsHelper* aHelper)
+    : ClientTiledLayerBuffer(aPaintedLayer, aCompositableClient),
+      mManager(aManager),
+      mCallback(nullptr),
+      mCallbackData(nullptr),
+      mSharedFrameMetricsHelper(aHelper) {}
+
+void ClientMultiTiledLayerBuffer::DiscardBuffers() {
+  for (TileClient& tile : mRetainedTiles) {
+    tile.DiscardBuffers();
+  }
+}
+
+SurfaceDescriptorTiles
+ClientMultiTiledLayerBuffer::GetSurfaceDescriptorTiles() {
+  nsTArray<TileDescriptor> tiles;
+
+  for (TileClient& tile : mRetainedTiles) {
+    TileDescriptor tileDesc = tile.GetTileDescriptor();
+    tiles.AppendElement(tileDesc);
+    // Reset the update rect
+    tile.mUpdateRect = IntRect();
+  }
+  return SurfaceDescriptorTiles(
+      mValidRegion, tiles, mTileOrigin, mTileSize, mTiles.mFirst.x,
+      mTiles.mFirst.y, mTiles.mSize.width, mTiles.mSize.height, mResolution,
+      mFrameResolution.xScale, mFrameResolution.yScale,
+      mWasLastPaintProgressive);
+}
+
+void ClientMultiTiledLayerBuffer::PaintThebes(
+    const nsIntRegion& aNewValidRegion, const nsIntRegion& aPaintRegion,
+    const nsIntRegion& aDirtyRegion,
+    LayerManager::DrawPaintedLayerCallback aCallback, void* aCallbackData,
+    TilePaintFlags aFlags) {
+  TILING_LOG("TILING %p: PaintThebes painting region %s\n", &mPaintedLayer,
+             Stringify(aPaintRegion).c_str());
+  TILING_LOG("TILING %p: PaintThebes new valid region %s\n", &mPaintedLayer,
+             Stringify(aNewValidRegion).c_str());
+
+  mCallback = aCallback;
+  mCallbackData = aCallbackData;
+  mWasLastPaintProgressive = !!(aFlags & TilePaintFlags::Progressive);
+
+#ifdef GFX_TILEDLAYER_PREF_WARNINGS
+  long start = PR_IntervalNow();
+#endif
+
+#ifdef GFX_TILEDLAYER_PREF_WARNINGS
+  if (PR_IntervalNow() - start > 30) {
+    const IntRect bounds = aPaintRegion.GetBounds();
+    printf_stderr("Time to draw %i: %i, %i, %i, %i\n", PR_IntervalNow() - start,
+                  bounds.x, bounds.y, bounds.width, bounds.height);
+    if (aPaintRegion.IsComplex()) {
+      printf_stderr("Complex region\n");
+      for (auto iter = aPaintRegion.RectIter(); !iter.Done(); iter.Next()) {
+        const IntRect& rect = iter.Get();
+        printf_stderr(" rect %i, %i, %i, %i\n", rect.x, rect.y, rect.width,
+                      rect.height);
+      }
+    }
+  }
+  start = PR_IntervalNow();
+#endif
+
+  AUTO_PROFILER_LABEL("ClientMultiTiledLayerBuffer::PaintThebes", GRAPHICS);
+
+  mNewValidRegion = aNewValidRegion;
+  Update(aNewValidRegion, aPaintRegion, aDirtyRegion, aFlags);
+
+#ifdef GFX_TILEDLAYER_PREF_WARNINGS
+  if (PR_IntervalNow() - start > 10) {
+    const IntRect bounds = aPaintRegion.GetBounds();
+    printf_stderr("Time to tile %i: %i, %i, %i, %i\n", PR_IntervalNow() - start,
+                  bounds.x, bounds.y, bounds.width, bounds.height);
+  }
+#endif
+
+  mLastPaintContentType = GetContentType(&mLastPaintSurfaceMode);
+  mCallback = nullptr;
+  mCallbackData = nullptr;
+}
+
+void ClientMultiTiledLayerBuffer::MaybeSyncTextures(
+    const nsIntRegion& aPaintRegion, const TilesPlacement& aNewTiles,
+    const IntSize& aScaledTileSize) {
+  if (mManager->AsShadowForwarder()->SupportsTextureDirectMapping()) {
+    AutoTArray<uint64_t, 10> syncTextureSerials;
+    SurfaceMode mode;
+    Unused << GetContentType(&mode);
+
+    // Pre-pass through the tiles (mirroring the filter logic below) to gather
+    // texture IDs that we need to ensure are unused by the GPU before we
+    // continue.
+    if (!aPaintRegion.IsEmpty()) {
+      MOZ_ASSERT(mPaintTasks.IsEmpty());
+      for (size_t i = 0; i < mRetainedTiles.Length(); ++i) {
+        const TileCoordIntPoint tileCoord = aNewTiles.TileCoord(i);
+
+        IntPoint tileOffset = GetTileOffset(tileCoord);
+        nsIntRegion tileDrawRegion = IntRect(tileOffset, aScaledTileSize);
+        tileDrawRegion.AndWith(aPaintRegion);
+
+        if (tileDrawRegion.IsEmpty()) {
+          continue;
+        }
+
+        TileClient& tile = mRetainedTiles[i];
+        tile.GetSyncTextureSerials(mode, syncTextureSerials);
+      }
+    }
+
+    if (syncTextureSerials.Length() > 0) {
+      mManager->AsShadowForwarder()->SyncTextures(syncTextureSerials);
+    }
+  }
+}
+
+void ClientMultiTiledLayerBuffer::Update(const nsIntRegion& newValidRegion,
+                                         const nsIntRegion& aPaintRegion,
+                                         const nsIntRegion& aDirtyRegion,
+                                         TilePaintFlags aFlags) {
+  const IntSize scaledTileSize = GetScaledTileSize();
+  const gfx::IntRect newBounds = newValidRegion.GetBounds();
+
+  const TilesPlacement oldTiles = mTiles;
+  const TilesPlacement newTiles(
+      floor_div(newBounds.X(), scaledTileSize.width),
+      floor_div(newBounds.Y(), scaledTileSize.height),
+      floor_div(
+          GetTileStart(newBounds.X(), scaledTileSize.width) + newBounds.Width(),
+          scaledTileSize.width) +
+          1,
+      floor_div(GetTileStart(newBounds.Y(), scaledTileSize.height) +
+                    newBounds.Height(),
+                scaledTileSize.height) +
+          1);
+
+  const size_t oldTileCount = mRetainedTiles.Length();
+  const size_t newTileCount = newTiles.mSize.width * newTiles.mSize.height;
+
+  nsTArray<TileClient> oldRetainedTiles = std::move(mRetainedTiles);
+  mRetainedTiles.SetLength(newTileCount);
+
+  for (size_t oldIndex = 0; oldIndex < oldTileCount; oldIndex++) {
+    const TileCoordIntPoint tileCoord = oldTiles.TileCoord(oldIndex);
+    const size_t newIndex = newTiles.TileIndex(tileCoord);
+    // First, get the already existing tiles to the right place in the new
+    // array. Leave placeholders (default constructor) where there was no tile.
+    if (newTiles.HasTile(tileCoord)) {
+      mRetainedTiles[newIndex] = oldRetainedTiles[oldIndex];
+    } else {
+      // release tiles that we are not going to reuse before allocating new ones
+      // to avoid allocating unnecessarily.
+      oldRetainedTiles[oldIndex].DiscardBuffers();
+    }
+  }
+
+  oldRetainedTiles.Clear();
+
+  nsIntRegion paintRegion = aPaintRegion;
+  nsIntRegion dirtyRegion = aDirtyRegion;
+
+  MaybeSyncTextures(paintRegion, newTiles, scaledTileSize);
+
+  if (!paintRegion.IsEmpty()) {
+    MOZ_ASSERT(mPaintTasks.IsEmpty());
+
+    for (size_t i = 0; i < newTileCount; ++i) {
+      const TileCoordIntPoint tileCoord = newTiles.TileCoord(i);
+
+      IntPoint tileOffset = GetTileOffset(tileCoord);
+      nsIntRegion tileDrawRegion = IntRect(tileOffset, scaledTileSize);
+      tileDrawRegion.AndWith(paintRegion);
+
+      if (tileDrawRegion.IsEmpty()) {
+        continue;
+      }
+
+      TileClient& tile = mRetainedTiles[i];
+      if (!ValidateTile(tile, GetTileOffset(tileCoord), tileDrawRegion,
+                        aFlags)) {
+        gfxCriticalError() << "ValidateTile failed";
+      }
+
+      // Validating the tile may have required more to be painted.
+      paintRegion.OrWith(tileDrawRegion);
+      dirtyRegion.OrWith(tileDrawRegion);
+    }
+
+    if (!mPaintTiles.IsEmpty()) {
+      // Create a tiled draw target
+      gfx::TileSet tileset;
+      for (size_t i = 0; i < mPaintTiles.Length(); ++i) {
+        mPaintTiles[i].mTileOrigin -= mTilingOrigin;
+      }
+      tileset.mTiles = mPaintTiles.Elements();
+      tileset.mTileCount = mPaintTiles.Length();
+      RefPtr<DrawTarget> drawTarget =
+          gfx::Factory::CreateTiledDrawTarget(tileset);
+      if (!drawTarget || !drawTarget->IsValid()) {
+        gfxDevCrash(LogReason::InvalidContext) << "Invalid tiled draw target";
+        return;
+      }
+      drawTarget->SetTransform(Matrix());
+
+      // Draw into the tiled draw target
+      RefPtr<gfxContext> ctx = gfxContext::CreateOrNull(drawTarget);
+      MOZ_ASSERT(ctx);  // already checked the draw target above
+      ctx->SetMatrix(ctx->CurrentMatrix()
+                         .PreScale(mResolution, mResolution)
+                         .PreTranslate(-mTilingOrigin));
+
+      mCallback(&mPaintedLayer, ctx, paintRegion, dirtyRegion,
+                DrawRegionClip::DRAW, nsIntRegion(), mCallbackData);
+      ctx = nullptr;
+
+      // Edge padding allows us to avoid resampling artifacts
+      if (StaticPrefs::layers_tiles_edge_padding_AtStartup() &&
+          mResolution == 1) {
+        drawTarget->PadEdges(newValidRegion.MovedBy(-mTilingOrigin));
+      }
+
+      // Reset
+      mPaintTiles.Clear();
+      mTilingOrigin = IntPoint(std::numeric_limits<int32_t>::max(),
+                               std::numeric_limits<int32_t>::max());
+    }
+
+    // Dispatch to the paint thread
+    if (aFlags & TilePaintFlags::Async) {
+      bool queuedTask = false;
+
+      while (!mPaintTasks.IsEmpty()) {
+        UniquePtr<PaintTask> task = mPaintTasks.PopLastElement();
+        if (!task->mCapture->IsEmpty()) {
+          PaintThread::Get()->QueuePaintTask(std::move(task));
+          queuedTask = true;
+        }
+      }
+
+      if (queuedTask) {
+        mManager->SetQueuedAsyncPaints();
+      }
+
+      mPaintTasks.Clear();
+    }
+
+    for (uint32_t i = 0; i < mRetainedTiles.Length(); ++i) {
+      TileClient& tile = mRetainedTiles[i];
+      UnlockTile(tile);
+    }
+  }
+
+  mTiles = newTiles;
+  mValidRegion = newValidRegion;
+}
+
+bool ClientMultiTiledLayerBuffer::ValidateTile(TileClient& aTile,
+                                               const nsIntPoint& aTileOrigin,
+                                               nsIntRegion& aDirtyRegion,
+                                               TilePaintFlags aFlags) {
+#ifdef GFX_TILEDLAYER_PREF_WARNINGS
+  if (aDirtyRegion.IsComplex()) {
+    printf_stderr("Complex region\n");
+  }
+#endif
+
+  SurfaceMode mode;
+  gfxContentType content = GetContentType(&mode);
+
+  if (!aTile.mAllocator) {
+    aTile.SetTextureAllocator(
+        mManager->GetCompositorBridgeChild()->GetTexturePool(
+            mManager->AsShadowForwarder(),
+            gfxPlatform::GetPlatform()->Optimal2DFormatForContent(content),
+            TextureFlags::DISALLOW_BIGIMAGE | TextureFlags::IMMEDIATE_UPLOAD |
+                TextureFlags::NON_BLOCKING_READ_LOCK));
+    MOZ_ASSERT(aTile.mAllocator);
+  }
+
+  nsIntRegion tileDirtyRegion = aDirtyRegion.MovedBy(-aTileOrigin);
+  tileDirtyRegion.ScaleRoundOut(mResolution, mResolution);
+
+  nsIntRegion tileVisibleRegion = mNewValidRegion.MovedBy(-aTileOrigin);
+  tileVisibleRegion.ScaleRoundOut(mResolution, mResolution);
+
+  std::vector<RefPtr<TextureClient>> asyncPaintClients;
+
+  Maybe<AcquiredBackBuffer> backBuffer =
+      aTile.AcquireBackBuffer(mCompositableClient, tileDirtyRegion,
+                              tileVisibleRegion, content, mode, aFlags);
+
+  if (!backBuffer) {
+    return false;
+  }
+
+  // Mark the area we need to paint in the back buffer as invalid in the
+  // front buffer as they will become out of sync.
+  aTile.mInvalidFront.OrWith(tileDirtyRegion);
+
+  // Add the backbuffer's invalid region intersected with the visible region to
+  // the dirty region we will be painting. This will be empty if we are able to
+  // copy from the front into the back.
+  nsIntRegion tileInvalidRegion = aTile.mInvalidBack;
+  tileInvalidRegion.AndWith(tileVisibleRegion);
+
+  nsIntRegion invalidRegion = tileInvalidRegion;
+  invalidRegion.MoveBy(aTileOrigin);
+  invalidRegion.ScaleInverseRoundOut(mResolution, mResolution);
+
+  tileDirtyRegion.OrWith(tileInvalidRegion);
+  aDirtyRegion.OrWith(invalidRegion);
+
+  // Mark the region we will be painting and the region we copied from the front
+  // buffer as needing to be uploaded to the compositor
+  aTile.mUpdateRect =
+      tileDirtyRegion.GetBounds().Union(backBuffer->mUpdatedRect);
+
+  // We need to clear the dirty region of the tile before painting
+  // if we are painting non-opaque content
+  if (mode != SurfaceMode::SURFACE_OPAQUE) {
+    for (auto iter = tileDirtyRegion.RectIter(); !iter.Done(); iter.Next()) {
+      const gfx::Rect drawRect(iter.Get().X(), iter.Get().Y(),
+                               iter.Get().Width(), iter.Get().Height());
+      backBuffer->mTarget->ClearRect(drawRect);
+    }
+  }
+
+  gfx::Tile paintTile;
+  paintTile.mTileOrigin = gfx::IntPoint(aTileOrigin.x, aTileOrigin.y);
+  paintTile.mDrawTarget = backBuffer->mTarget;
+  mPaintTiles.AppendElement(paintTile);
+
+  if (aFlags & TilePaintFlags::Async) {
+    UniquePtr<PaintTask> task(new PaintTask());
+    task->mCapture = backBuffer->mCapture;
+    task->mTarget = backBuffer->mBackBuffer;
+    task->mClients = std::move(backBuffer->mTextureClients);
+    mPaintTasks.AppendElement(std::move(task));
+  } else {
+    MOZ_RELEASE_ASSERT(backBuffer->mTarget == backBuffer->mBackBuffer);
+    MOZ_RELEASE_ASSERT(backBuffer->mCapture == nullptr);
+  }
+
+  mTilingOrigin.x = std::min(mTilingOrigin.x, paintTile.mTileOrigin.x);
+  mTilingOrigin.y = std::min(mTilingOrigin.y, paintTile.mTileOrigin.y);
+
+  // The new buffer is now validated, remove the dirty region from it.
+  aTile.mInvalidBack.SubOut(tileDirtyRegion);
+
+  aTile.Flip();
+
+  return true;
+}
+
+/**
+ * This function takes the transform stored in aTransformToCompBounds
+ * (which was generated in GetTransformToAncestorsParentLayer), and
+ * modifies it with the ViewTransform from the compositor side so that
+ * it reflects what the compositor is actually rendering. This operation
+ * basically adds in the layer's async transform.
+ * This function then returns the scroll ancestor's composition bounds,
+ * transformed into the painted layer's LayerPixel coordinates, accounting
+ * for the compositor state.
+ */
+static Maybe<LayerRect> GetCompositorSideCompositionBounds(
+    const LayerMetricsWrapper& aScrollAncestor,
+    const LayerToParentLayerMatrix4x4& aTransformToCompBounds,
+    const AsyncTransform& aAPZTransform, const LayerRect& aClip) {
+  LayerToParentLayerMatrix4x4 transform =
+      aTransformToCompBounds * AsyncTransformComponentMatrix(aAPZTransform);
+
+  return UntransformBy(transform.Inverse(),
+                       aScrollAncestor.Metrics().GetCompositionBounds(), aClip);
+}
+
+bool ClientMultiTiledLayerBuffer::ComputeProgressiveUpdateRegion(
+    const nsIntRegion& aInvalidRegion, const nsIntRegion& aOldValidRegion,
+    nsIntRegion& aRegionToPaint, BasicTiledLayerPaintData* aPaintData,
+    bool aIsRepeated) {
+  aRegionToPaint = aInvalidRegion;
+
+  // If the composition bounds rect is empty, we can't make any sensible
+  // decision about how to update coherently. In this case, just update
+  // everything in one transaction.
+  if (aPaintData->mCompositionBounds.IsEmpty()) {
+    aPaintData->mPaintFinished = true;
+    return false;
+  }
+
+  // If this is a low precision buffer, we force progressive updates. The
+  // assumption is that the contents is less important, so visual coherency
+  // is lower priority than speed.
+  bool drawingLowPrecision = IsLowPrecision();
+
+  // Find out if we have any non-stale content to update.
+  nsIntRegion staleRegion;
+  staleRegion.And(aInvalidRegion, aOldValidRegion);
+
+  TILING_LOG("TILING %p: Progressive update stale region %s\n", &mPaintedLayer,
+             Stringify(staleRegion).c_str());
+
+  LayerMetricsWrapper scrollAncestor;
+  mPaintedLayer.GetAncestorLayers(&scrollAncestor, nullptr, nullptr);
+
+  // Find out the current view transform to determine which tiles to draw
+  // first, and see if we should just abort this paint. Aborting is usually
+  // caused by there being an incoming, more relevant paint.
+  AsyncTransform viewTransform;
+  MOZ_ASSERT(mSharedFrameMetricsHelper);
+
+  bool abortPaint = mSharedFrameMetricsHelper->UpdateFromCompositorFrameMetrics(
+      scrollAncestor, !staleRegion.Contains(aInvalidRegion),
+      drawingLowPrecision, viewTransform);
+
+  TILING_LOG(
+      "TILING %p: Progressive update view transform %s zoom %f abort %d\n",
+      &mPaintedLayer, ToString(viewTransform.mTranslation).c_str(),
+      viewTransform.mScale.scale, abortPaint);
+
+  if (abortPaint) {
+    // We ignore if front-end wants to abort if this is the first,
+    // non-low-precision paint, as in that situation, we're about to override
+    // front-end's page/viewport metrics.
+    if (!aPaintData->mFirstPaint || drawingLowPrecision) {
+      AUTO_PROFILER_LABEL(
+          "ClientMultiTiledLayerBuffer::ComputeProgressiveUpdateRegion",
+          GRAPHICS);
+
+      aRegionToPaint.SetEmpty();
+      return aIsRepeated;
+    }
+  }
+
+  Maybe<LayerRect> transformedCompositionBounds =
+      GetCompositorSideCompositionBounds(
+          scrollAncestor, aPaintData->mTransformToCompBounds, viewTransform,
+          LayerRect(mPaintedLayer.GetVisibleRegion().GetBounds()));
+
+  if (!transformedCompositionBounds) {
+    aPaintData->mPaintFinished = true;
+    return false;
+  }
+
+  TILING_LOG("TILING %p: Progressive update transformed compositor bounds %s\n",
+             &mPaintedLayer, Stringify(*transformedCompositionBounds).c_str());
+
+  // Compute a "coherent update rect" that we should paint all at once in a
+  // single transaction. This is to avoid rendering glitches on animated
+  // page content, and when layers change size/shape.
+  // On Fennec uploads are more expensive because we're not using gralloc, so
+  // we use a coherent update rect that is intersected with the screen at the
+  // time of issuing the draw command. This will paint faster but also
+  // potentially make the progressive paint more visible to the user while
+  // scrolling.
+  IntRect coherentUpdateRect(RoundedOut(
+#ifdef MOZ_WIDGET_ANDROID
+                                 transformedCompositionBounds->Intersect(
+                                     aPaintData->mCompositionBounds)
+#else
+                                 *transformedCompositionBounds
+#endif
+                                     )
+                                 .ToUnknownRect());
+
+  TILING_LOG("TILING %p: Progressive update final coherency rect %s\n",
+             &mPaintedLayer, Stringify(coherentUpdateRect).c_str());
+
+  aRegionToPaint.And(aInvalidRegion, coherentUpdateRect);
+  aRegionToPaint.Or(aRegionToPaint, staleRegion);
+  bool drawingStale = !aRegionToPaint.IsEmpty();
+  if (!drawingStale) {
+    aRegionToPaint = aInvalidRegion;
+  }
+
+  // Prioritise tiles that are currently visible on the screen.
+  bool paintingVisible = false;
+  if (aRegionToPaint.Intersects(coherentUpdateRect)) {
+    aRegionToPaint.And(aRegionToPaint, coherentUpdateRect);
+    paintingVisible = true;
+  }
+
+  TILING_LOG("TILING %p: Progressive update final paint region %s\n",
+             &mPaintedLayer, Stringify(aRegionToPaint).c_str());
+
+  // Paint area that's visible and overlaps previously valid content to avoid
+  // visible glitches in animated elements, such as gifs.
+  bool paintInSingleTransaction =
+      paintingVisible && (drawingStale || aPaintData->mFirstPaint);
+
+  TILING_LOG(
+      "TILING %p: paintingVisible %d drawingStale %d firstPaint %d "
+      "singleTransaction %d\n",
+      &mPaintedLayer, paintingVisible, drawingStale, aPaintData->mFirstPaint,
+      paintInSingleTransaction);
+
+  // The following code decides what order to draw tiles in, based on the
+  // current scroll direction of the primary scrollable layer.
+  NS_ASSERTION(!aRegionToPaint.IsEmpty(), "Unexpectedly empty paint region!");
+  IntRect paintBounds = aRegionToPaint.GetBounds();
+
+  int startX, incX, startY, incY;
+  gfx::IntSize scaledTileSize = GetScaledTileSize();
+  if (aPaintData->mScrollOffset.x >= aPaintData->mLastScrollOffset.x) {
+    startX = RoundDownToTileEdge(paintBounds.X(), scaledTileSize.width);
+    incX = scaledTileSize.width;
+  } else {
+    startX = RoundDownToTileEdge(paintBounds.XMost() - 1, scaledTileSize.width);
+    incX = -scaledTileSize.width;
+  }
+
+  if (aPaintData->mScrollOffset.y >= aPaintData->mLastScrollOffset.y) {
+    startY = RoundDownToTileEdge(paintBounds.Y(), scaledTileSize.height);
+    incY = scaledTileSize.height;
+  } else {
+    startY =
+        RoundDownToTileEdge(paintBounds.YMost() - 1, scaledTileSize.height);
+    incY = -scaledTileSize.height;
+  }
+
+  // Find a tile to draw.
+  IntRect tileBounds(startX, startY, scaledTileSize.width,
+                     scaledTileSize.height);
+  int32_t scrollDiffX =
+      aPaintData->mScrollOffset.x - aPaintData->mLastScrollOffset.x;
+  int32_t scrollDiffY =
+      aPaintData->mScrollOffset.y - aPaintData->mLastScrollOffset.y;
+  // This loop will always terminate, as there is at least one tile area
+  // along the first/last row/column intersecting with regionToPaint, or its
+  // bounds would have been smaller.
+  while (true) {
+    aRegionToPaint.And(aInvalidRegion, tileBounds);
+    if (!aRegionToPaint.IsEmpty()) {
+      if (mResolution != 1) {
+        // Paint the entire tile for low-res. This is aimed to fixing low-res
+        // resampling and to avoid doing costly region accurate painting for a
+        // small area.
+        aRegionToPaint = tileBounds;
+      }
+      break;
+    }
+    if (Abs(scrollDiffY) >= Abs(scrollDiffX)) {
+      tileBounds.MoveByX(incX);
+    } else {
+      tileBounds.MoveByY(incY);
+    }
+  }
+
+  if (!aRegionToPaint.Contains(aInvalidRegion)) {
+    // The region needed to paint is larger then our progressive chunk size
+    // therefore update what we want to paint and ask for a new paint
+    // transaction.
+
+    // If we need to draw more than one tile to maintain coherency, make
+    // sure it happens in the same transaction by requesting this work be
+    // repeated immediately.
+    // If this is unnecessary, the remaining work will be done tile-by-tile in
+    // subsequent transactions. The caller code is responsible for scheduling
+    // the subsequent transactions as long as we don't set the mPaintFinished
+    // flag to true.
+    return (!drawingLowPrecision && paintInSingleTransaction);
+  }
+
+  // We're not repeating painting and we've not requested a repeat transaction,
+  // so the paint is finished. If there's still a separate low precision
+  // paint to do, it will get marked as unfinished later.
+  aPaintData->mPaintFinished = true;
+  return false;
+}
+
+bool ClientMultiTiledLayerBuffer::ProgressiveUpdate(
+    const nsIntRegion& aValidRegion, const nsIntRegion& aInvalidRegion,
+    const nsIntRegion& aOldValidRegion, nsIntRegion& aOutDrawnRegion,
+    BasicTiledLayerPaintData* aPaintData,
+    LayerManager::DrawPaintedLayerCallback aCallback, void* aCallbackData) {
+  TILING_LOG("TILING %p: Progressive update valid region %s\n", &mPaintedLayer,
+             Stringify(aValidRegion).c_str());
+  TILING_LOG("TILING %p: Progressive update invalid region %s\n",
+             &mPaintedLayer, Stringify(aInvalidRegion).c_str());
+  TILING_LOG("TILING %p: Progressive update old valid region %s\n",
+             &mPaintedLayer, Stringify(aOldValidRegion).c_str());
+
+  bool repeat = false;
+  bool isBufferChanged = false;
+  nsIntRegion remainingInvalidRegion = aInvalidRegion;
+  nsIntRegion updatedValidRegion = aValidRegion;
+  do {
+    // Compute the region that should be updated. Repeat as many times as
+    // is required.
+    nsIntRegion regionToPaint;
+    repeat =
+        ComputeProgressiveUpdateRegion(remainingInvalidRegion, aOldValidRegion,
+                                       regionToPaint, aPaintData, repeat);
+
+    TILING_LOG(
+        "TILING %p: Progressive update computed paint region %s repeat %d\n",
+        &mPaintedLayer, Stringify(regionToPaint).c_str(), repeat);
+
+    // There's no further work to be done.
+    if (regionToPaint.IsEmpty()) {
+      break;
+    }
+
+    isBufferChanged = true;
+
+    // Keep track of what we're about to refresh.
+    aOutDrawnRegion.OrWith(regionToPaint);
+    updatedValidRegion.OrWith(regionToPaint);
+
+    // aValidRegion may have been altered by InvalidateRegion, but we still
+    // want to display stale content until it gets progressively updated.
+    // Create a region that includes stale content.
+    nsIntRegion validOrStale;
+    validOrStale.Or(updatedValidRegion, aOldValidRegion);
+
+    // Paint the computed region and subtract it from the invalid region.
+    PaintThebes(validOrStale, regionToPaint, remainingInvalidRegion, aCallback,
+                aCallbackData, TilePaintFlags::Progressive);
+    remainingInvalidRegion.SubOut(regionToPaint);
+  } while (repeat);
+
+  TILING_LOG(
+      "TILING %p: Progressive update final valid region %s buffer changed %d\n",
+      &mPaintedLayer, Stringify(updatedValidRegion).c_str(), isBufferChanged);
+  TILING_LOG("TILING %p: Progressive update final invalid region %s\n",
+             &mPaintedLayer, Stringify(remainingInvalidRegion).c_str());
+
+  // Return false if nothing has been drawn, or give what has been drawn
+  // to the shadow layer to upload.
+  return isBufferChanged;
+}
+
+}  // namespace layers
+}  // namespace mozilla