summaryrefslogtreecommitdiffstats
path: root/cc/trees
Commit message (Collapse)AuthorAgeFilesLines
* Revert of Re-land: cc: Use worker context for one-copy tile initialization. ↵sky2015-08-185-12/+36
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | (patchset #29 id:560001 of https://codereview.chromium.org/1230203007/ ) Reason for revert: Sorry Dave. I'm reverting as this broke mandoline. To see the breakage build mandoline:all, and run with: mandoline --oopif-always-create-new-frame-tree http://phuang.github.io/iframe.html . We do have mandoline tests that run, but they run with the code path gfx::GLSurfaceTestSupport::InitializeOneOff(), which is why they likely still pass. I'll see about making the tests not run with that config. Original issue's description: > Re-land: cc: Use worker context for one-copy tile initialization. > > This moves management of staging resources to > OneCopyTileTaskWorkerPool class. This makes it possible > to use a worker context to issue and detect when copy > operations complete. > > BUG=490295 > TBR=jamesr@chromium.org, sky@chromium.org > CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel > > Committed: https://crrev.com/af0ab609b53c64ce43172be42e20e7ca8beab497 > Cr-Commit-Position: refs/heads/master@{#343947} TBR=jamesr@chromium.org,kbr@chromium.org,piman@chromium.org,primiano@chromium.org,vmiura@chromium.org,vmpstr@chromium.org,reveman@chromium.org NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG=490295 Review URL: https://codereview.chromium.org/1298143003 Cr-Commit-Position: refs/heads/master@{#344042}
* cc: Remove render_pass_test_common, move to render_pass_test_utils.danakj2015-08-182-6/+6
| | | | | | | | | | | | The common file had a single method, while the utils has 6. Move that one method over to the utils file, and drop the TestRenderPass type. R=enne, vmpstr, weiliangc CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1291363003 Cr-Commit-Position: refs/heads/master@{#344014}
* Screen Space Transform Of Render Surface From Property Treesjaydasika2015-08-183-2/+35
| | | | | | | | CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1301803002 Cr-Commit-Position: refs/heads/master@{#344005}
* cc: Convert RS flattening unit test to LayerImplenne2015-08-181-23/+20
| | | | | | | | | | | | | It wants to check screen space transforms and flattening which makes more sense to do on the compositor thread. R=weiliangc@chromium.org BUG=502041 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1293073003 Cr-Commit-Position: refs/heads/master@{#344003}
* cc: Fix LayerTreeRebuildTest to run with only property treesenne2015-08-182-16/+5
| | | | | | | | | | | | | | | | The meta walk wasn't happening for testing in some cases, even though it should. That fixes the unclipped descendant failures with this test. See https://codereview.chromium.org/1298793002 for the reasoning to remove the input handler checking. R=weiliangc@chromium.org BUG=502041 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1293793005 Cr-Commit-Position: refs/heads/master@{#343986}
* cc: Convert DescendantsOfClipChildren to LayerImplenne2015-08-181-46/+22
| | | | | | | | | | R=weiliangc@chromium.org BUG=502041 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1287163004 Cr-Commit-Position: refs/heads/master@{#343974}
* cc: Convert DoNotSkipLayersWithHandlers to LayerImplenne2015-08-181-29/+13
| | | | | | | | | | | | | The main thread property tree version doesn't have to care about handlers, so convert this to be a LayerImpl test. R=weiliangc@chromium.org BUG=502041 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1296933003 Cr-Commit-Position: refs/heads/master@{#343966}
* cc: Convert DrawableContentRectForLayers to LayerImplenne2015-08-181-67/+26
| | | | | | | | | | | | This is a compositor thread-only concept, so convert this test. R=weiliangc@chromium.org BUG=502041 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1297823003 Cr-Commit-Position: refs/heads/master@{#343961}
* cc: Split IsClippedCorrectly unit test into two versionsenne2015-08-181-4/+106
| | | | | | | | | | | | | | | | | | | The main thread doesn't know anything about render surface clipping. However, the main thread is the only place that sets is_clipped on layers, so it seemed worthwhile keeping both versions of this test, even if it's a bit redundant. It's possible that is_clipped can be answered by property trees instead of layer properties in the future, in which case this could be collapsed back into one test. R=weiliangc@chromium.org,ajuma@chromium.org BUG=502041 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1300723003 Cr-Commit-Position: refs/heads/master@{#343959}
* cc: Convert ClippedByScrollParent unittest to LayerImplenne2015-08-181-51/+26
| | | | | | | | | | | | | The main thread property tree approach doesn't compute clip rects explicitly, so convert this to be LayerImpl instead. R=weiliangc@chromium.org BUG=502041 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1299713002 Cr-Commit-Position: refs/heads/master@{#343958}
* Reland of Change LatencyInfo to use Flow API v2. (patchset #1 id:1 of ↵yuhaoz2015-08-182-4/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | https://codereview.chromium.org/1295533005/ ) Reason for revert: This CL didn't cause the memory issue. So per glider@'s advise, revert this revert. Original issue's description: > Revert of Change LatencyInfo to use Flow API v2. (patchset #2 id:20001 of https://codereview.chromium.org/1291243002/ ) > > Reason for revert: > I believe this had caused use-after-free errors in content_unittests (http://build.chromium.org/p/chromium.memory.fyi/builders/Linux%20Tests%20%28valgrind%29%284%29/builds/44367) > See also crbug.com/522049. > > BUG=522049 > > Original issue's description: > > Change LatencyInfo to use Flow API v2. > > > > BUG=https://github.com/catapult-project/catapult/issues/1079 > > CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel > > > > Committed: https://crrev.com/b0bdf54635605d8e639eb0d213ae463e8805cfb0 > > Cr-Commit-Position: refs/heads/master@{#343738} > > TBR=dsinclair@chromium.org,nduca@chromium.org,vmpstr@chromium.org,yufetus@chromium.org,jdduke@chromium.org,sievers@chromium.org,miletus@chromium.org,yuhaoz@google.com > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG=https://github.com/catapult-project/catapult/issues/1079 > > Committed: https://crrev.com/5befb8d4949af90be7ab047b44c7419ef45fd50a > Cr-Commit-Position: refs/heads/master@{#343902} TBR=dsinclair@chromium.org,nduca@chromium.org,vmpstr@chromium.org,yufetus@chromium.org,jdduke@chromium.org,sievers@chromium.org,miletus@chromium.org,glider@chromium.org NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG=522049 Review URL: https://codereview.chromium.org/1288893006 Cr-Commit-Position: refs/heads/master@{#343954}
* Re-land: cc: Use worker context for one-copy tile initialization.reveman2015-08-185-36/+12
| | | | | | | | | | | | | | | This moves management of staging resources to OneCopyTileTaskWorkerPool class. This makes it possible to use a worker context to issue and detect when copy operations complete. BUG=490295 TBR=jamesr@chromium.org, sky@chromium.org CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1230203007 Cr-Commit-Position: refs/heads/master@{#343947}
* cc: Convert RenderSurfaceForBlendMode to LayerImplenne2015-08-181-12/+10
| | | | | | | | | | | | Drawing blend modes are not computed on the main thread. R=weiliangc@chromium.org BUG=502041 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1297853004 Cr-Commit-Position: refs/heads/master@{#343943}
* Rename OpacityTree to EffectTreeweiliangc2015-08-187-55/+53
| | | | | | | | | | | | | Effect tree is going to contain opacity as well as other render effects. Change the name as a first step. BUG=521717 R=enne CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1284103007 Cr-Commit-Position: refs/heads/master@{#343936}
* Enable partial rasterization in the renderer using a persistent GpuMemoryBuffer.jbroman2015-08-181-0/+7
| | | | | | | | | BUG=492754 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1263363004 Cr-Commit-Position: refs/heads/master@{#343935}
* Revert of Change LatencyInfo to use Flow API v2. (patchset #2 id:20001 of ↵glider2015-08-182-7/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | https://codereview.chromium.org/1291243002/ ) Reason for revert: I believe this had caused use-after-free errors in content_unittests (http://build.chromium.org/p/chromium.memory.fyi/builders/Linux%20Tests%20%28valgrind%29%284%29/builds/44367) See also crbug.com/522049. BUG=522049 Original issue's description: > Change LatencyInfo to use Flow API v2. > > BUG=https://github.com/catapult-project/catapult/issues/1079 > CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel > > Committed: https://crrev.com/b0bdf54635605d8e639eb0d213ae463e8805cfb0 > Cr-Commit-Position: refs/heads/master@{#343738} TBR=dsinclair@chromium.org,nduca@chromium.org,vmpstr@chromium.org,yufetus@chromium.org,jdduke@chromium.org,sievers@chromium.org,miletus@chromium.org,yuhaoz@google.com NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG=https://github.com/catapult-project/catapult/issues/1079 Review URL: https://codereview.chromium.org/1295533005 Cr-Commit-Position: refs/heads/master@{#343902}
* Clip Rect of Render Surface From Property Treesjaydasika2015-08-184-2/+33
| | | | | | | | CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1288063005 Cr-Commit-Position: refs/heads/master@{#343899}
* cc: Make FixedPositionWithInterveningRenderSurface use property treesenne2015-08-181-4/+9
| | | | | | | | | | | | | | | | Previously it was looking at draw_transform which doesn't get filled in on the main thread by property trees. Once main thread CDP is removed, Layer::draw_transform can do this work, and this can be reverted. R=weiliangc@chromium.org BUG=502041 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1295203002 Cr-Commit-Position: refs/heads/master@{#343802}
* Change LatencyInfo to use Flow API v2.yuhaoz2015-08-172-4/+7
| | | | | | | | | BUG=https://github.com/catapult-project/catapult/issues/1079 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1291243002 Cr-Commit-Position: refs/heads/master@{#343738}
* Lock video controls' position during fullscreenhugo.holgersson2015-08-172-6/+76
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | Problem: Video controls can be scrolled (moved) when an Android WebView does fullscreen-video. Solution: When visual viewport's scrolling is disabled using setUserScrollable(false, false), do also disable over-scrolling (do not send overscroll-events to the Android WebView). BUG=448738 TEST= 1. Start com.android.browser and go to http://robnyman.github.io/fullscreen/ . 2. Start video playback. 3. Go fullscreen on video. 4. Fling the screen. 5. Notice 1: Video controls cannot move. 6. Notice 2: No "over glow" (compare with Android Chrome or Lollipop's WebView). CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1228423002 Cr-Commit-Position: refs/heads/master@{#343675}
* Compute blend_mode outside CDPjaydasika2015-08-153-4/+8
| | | | | | | | | BUG=497817 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1296713002 Cr-Commit-Position: refs/heads/master@{#343555}
* cc: Make HiDPI devices use 4 MSAA samples for GPU rasterization by default.senorblanco2015-08-142-5/+12
| | | | | | | | | | | | | | | In order to detect the DPI, move the default MSAA samples computation from CompositorUtil to LayerTreeHostImpl. Use -1 to signify "no command line flag specified" (= auto-compute MSAA samples from DPI), so 0 can continue to mean MSAA was explicitly disabled on the command line, or in about:flags. BUG=520243 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1284343003 Cr-Commit-Position: refs/heads/master@{#343506}
* Render Surface is_clipped from Property Treesjaydasika2015-08-147-3/+31
| | | | | | | | | | | Render Surface is_clipped is computed in CDP. This patch computes it using property trees and checks that property tree computation matches CDP computation. CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1292473005 Cr-Commit-Position: refs/heads/master@{#343494}
* Clip Rect From Property Treesjaydasika2015-08-145-10/+72
| | | | | | | | | | | | | Clip Rect is computed in CDP. This patch computes it using property trees and checks that property tree computation matches CDP computation. BUG=497817 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1291653003 Cr-Commit-Position: refs/heads/master@{#343380}
* Fix overscroll glow when page contents are smaller than viewport.bokan2015-08-143-3/+77
| | | | | | | | | | | | | | | | | | | The overscroll glow effect positioning assumed that the content was always at least as big as the viewport. When this assumption was violated, the viewport would appear to "overscroll" its bound, pushing the bottom glow up, since the positioning code attempts to offset the glow so that it appears to be fixed to the page contents (and scrolls with it). This patch artificially expands the content size to be at lesat the size of the viewport. This is done by inflating the contents to the size of the viewport at minimum scale, since that's what the scrolling bounds are. BUG=421652 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1118033002 Cr-Commit-Position: refs/heads/master@{#343322}
* Use regular bubbling scroll for scrolling during pinch gesture.bokan2015-08-122-11/+28
| | | | | | | | | | | | | | | | | Scrolls occuring during a pinch gesture, that is, moving the center of pinch during the gesture, are treated as regular scrolls that bubble from one viewport layer to the next. Note, scrolls during a pinch made in order to keep the pinch point centered remain unchanged: under the --invert-viewport-scroll-order flag, these scrolls will "pan" the viewport, scrolling *only* the inner viewport. BUG=433724 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1288773002 Cr-Commit-Position: refs/heads/master@{#343112}
* cc: Make layer updating optional in BeginMainFrameskyostil2015-08-125-78/+254
| | | | | | | | | | | | | | | | | If no-one has requested a layer update, either by calling Proxy::SetNeedsUpdateLayers() or Proxy::SetNeedsCommit(), skip computing the layer update in BeginMainFrame. This reduces the static overhead of a BeginMainFrame/Commit operation when the layer tree doesn't change. This change improves efficiency, for example, when a page has a requestAnimationFrame callback which does not modify the DOM. BUG=490465 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1255863004 Cr-Commit-Position: refs/heads/master@{#343004}
* Separate RenderSurfaceLayerList Generation CalcDrawPropsweiliangc2015-08-112-185/+201
| | | | | | | | | | | | | | | | | | Use a separate walk to generate RenderSurfaceLayerList that is not tangled with CalcDrawProps. After CalcDrawProps or its PropertyTrees equivalence, remove layers that do not draw content from RenderSurfaceLayerList. RenderSurfaceLayerList is now only generated on compositor thread when we are about to draw. BUG=504466 R=enne CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1266103002 Cr-Commit-Position: refs/heads/master@{#342907}
* cc: Disable smoothness for ScrollDelegateboliu2015-08-111-5/+9
| | | | | | | | | | | | | | | | Webview is not able to reliably determine if the AnimateInput callback will be called. If it is not called, cc risks being stuck in smoothness mode, which causes other problems. Disable smoothness for scrolls and flings from the ScrollDelegate. BUG=516362 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1270343007 Cr-Commit-Position: refs/heads/master@{#342756}
* Drawable Content Rect From Property Treesjaydasika2015-08-106-7/+61
| | | | | | | | | | | | | Drawable content rect is computed in CDP. This patch computes it using property trees and checks that property tree computation matches CDP computation. BUG=497817 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1241223002 Cr-Commit-Position: refs/heads/master@{#342703}
* Reland of Preserve viewport scrolling layer when a fling starts (patchset #1 ↵jdduke2015-08-102-3/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | id:1 of https://codereview.chromium.org/1280113003/ ) Reason for revert: The revert turned out to be unnecessary. Original issue's description: > Revert of Preserve viewport scrolling layer when a fling starts (patchset #2 id:20001 of https://codereview.chromium.org/1225693008/ ) > > Reason for revert: > Causes issues with WebView when the embedder overrides computeScroll. > > Original issue's description: > > Preserve viewport scrolling layer when a fling starts > > > > A touch fling will lock the scrolling layer to the first layer that > > scrolls after the animation begins. This can disrupt viewport scrolling > > if the outer viewport was the originally targetted scroll layer, e.g., > > by resetting bookkeeping for scrollbar fade animations. Prevent this > > by allowing scroll bubbling if the viewport is the locked layer when > > the fling starts. > > > > BUG=507786 > > CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel > > > > Committed: https://crrev.com/89192797b50f7e33967445095cd7fa16b78f68e1 > > Cr-Commit-Position: refs/heads/master@{#337747} > > TBR=aelias@chromium.org > NOPRESUBMIT=true > NOTREECHECKS=true > NOTRY=true > BUG=507786,516362 > > Committed: https://crrev.com/7168c3a89473e240d8d9692fbf21606b58321082 > Cr-Commit-Position: refs/heads/master@{#342514} TBR=aelias@chromium.org,boliu@chromium.org NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG=507786,516362 Review URL: https://codereview.chromium.org/1284683003 Cr-Commit-Position: refs/heads/master@{#342623}
* Revert of Re-land: cc: Use worker context for one-copy tile initialization. ↵kbr2015-08-095-12/+36
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | (patchset #22 id:420001 of https://codereview.chromium.org/1230203007/ ) Reason for revert: I'm sorry, but this patch definitely causes failures of one of the context_lost tests on Windows on multiple bots as described in https://code.google.com/p/chromium/issues/detail?id=490295#c15 . I'm sorry the CQ isn't catching these, and I'm not sure why it isn't -- but it's important to fix these failures rather than leaving them on the tree. Original issue's description: > Re-land: cc: Use worker context for one-copy tile initialization. > > This moves management of staging resources to > OneCopyTileTaskWorkerPool class. This makes it possible > to use a worker context to issue and detect when copy > operations complete. > > BUG=490295 > CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel > > Committed: https://crrev.com/a6b998600660f34bcce9fbc2c75bbc0a3f7f03cc > Cr-Commit-Position: refs/heads/master@{#342567} TBR=piman@chromium.org,primiano@chromium.org,vmiura@chromium.org,vmpstr@chromium.org,reveman@chromium.org NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG=490295 Review URL: https://codereview.chromium.org/1282183002 Cr-Commit-Position: refs/heads/master@{#342568}
* Re-land: cc: Use worker context for one-copy tile initialization.reveman2015-08-095-36/+12
| | | | | | | | | | | | | | This moves management of staging resources to OneCopyTileTaskWorkerPool class. This makes it possible to use a worker context to issue and detect when copy operations complete. BUG=490295 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1230203007 Cr-Commit-Position: refs/heads/master@{#342567}
* Revert of Preserve viewport scrolling layer when a fling starts (patchset #2 ↵jdduke2015-08-082-15/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | id:20001 of https://codereview.chromium.org/1225693008/ ) Reason for revert: Causes issues with WebView when the embedder overrides computeScroll. Original issue's description: > Preserve viewport scrolling layer when a fling starts > > A touch fling will lock the scrolling layer to the first layer that > scrolls after the animation begins. This can disrupt viewport scrolling > if the outer viewport was the originally targetted scroll layer, e.g., > by resetting bookkeeping for scrollbar fade animations. Prevent this > by allowing scroll bubbling if the viewport is the locked layer when > the fling starts. > > BUG=507786 > CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel > > Committed: https://crrev.com/89192797b50f7e33967445095cd7fa16b78f68e1 > Cr-Commit-Position: refs/heads/master@{#337747} TBR=aelias@chromium.org NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG=507786,516362 Review URL: https://codereview.chromium.org/1280113003 Cr-Commit-Position: refs/heads/master@{#342514}
* Change switches::kContentImageTextureTarget to a list of image texture targetsjie.a.chen2015-08-073-13/+12
| | | | | | | | | | | | | | | | Not knowing the GpuMemoryBuffer format that the compostor will use, the host is hard to determine the image texture target properly. This CL changes switches::kContentImageTextureTarget to a list of targets for each format, so that the compositor can decide what format it wants to use without having to worry about the target being wrong. BUG=490362,512665 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1251693003 Cr-Commit-Position: refs/heads/master@{#342339}
* Revert of cc: Use worker context for one-copy tile initialization. (patchset ↵kbr2015-08-075-13/+38
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | #20 id:380001 of https://codereview.chromium.org/1230203007/ ) Reason for revert: Suspected of colliding with another patch and causing Issue 517719. Original issue's description: > cc: Use worker context for one-copy tile initialization. > > This moves management of staging resources to > OneCopyTileTaskWorkerPool class. This makes it possible > to use a worker context to issue and detect when copy > operations complete. > > BUG=490295 > CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel > > Committed: https://crrev.com/d5157ac5afec145b85746b42f2d86383483a8222 > Cr-Commit-Position: refs/heads/master@{#342235} TBR=piman@chromium.org,primiano@chromium.org,vmiura@chromium.org,vmpstr@chromium.org,reveman@chromium.org NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG=490295,517719 Review URL: https://codereview.chromium.org/1273163004 Cr-Commit-Position: refs/heads/master@{#342284}
* Remove SoftwareFrameDatajbauman2015-08-071-2/+2
| | | | | | | | | | We never transmit software frames over IPC anymore, so this is unnecessary. CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1265243004 Cr-Commit-Position: refs/heads/master@{#342265}
* cc: Use worker context for one-copy tile initialization.reveman2015-08-065-38/+13
| | | | | | | | | | | | | | This moves management of staging resources to OneCopyTileTaskWorkerPool class. This makes it possible to use a worker context to issue and detect when copy operations complete. BUG=490295 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1230203007 Cr-Commit-Position: refs/heads/master@{#342235}
* Add ClipNode when Render Surface Inherits Clip.jaydasika2015-08-065-23/+161
| | | | | | | | | | | | | | | | | The differences are because property trees don't create a clip node in the case where a layer : has_render_surface=true, ancestor_clips_subtee=true and data_for_children->ancestor_clips_subtee=false. We need a clip node in this case. This patch adds a clip node for this case and computes the clip required in this case. BUG=510185 TEST=visible rects from CDP and property trees match in http://www.clicktorelease.com/code/css3dclouds with this patch. CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1252313004 Cr-Commit-Position: refs/heads/master@{#342209}
* cc: Fix flaky RSLLMembershipWithScale testenne2015-08-061-27/+37
| | | | | | | | | | | | | | | | | | | | | | Update this test to reflect the removal of twin updating. Instead of having tilings removed, tilings currently just stay stale if their layer owner is not in the RSLL and updated. The test was going to the next step before a pinch was complete, and so would sometimes end up with an extra draw from source frame 0 before the commit from source frame 1 would occur. This is easily repro'd by adding a sleep(1) to the draw function in the test. This is avoided by tracking NotifyReadyToDraw and only going to step #2 after the draw following it. R=danakj@chromium.org,vmpstr@chromium.org BUG=460581 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1276683002 Cr-Commit-Position: refs/heads/master@{#342158}
* cc: Remove two DCHECKs that don't always holdajuma2015-08-053-3/+112
| | | | | | | | | | | | | | | | | | | | | | | | | | CalcDrawProps and property trees skip subtrees rooted by a layer with an uninvertible transform (except when there's a transform animation). In theory, this means means that layers with uninvertible draw transforms should get skipped too. However, floating point math means that multiplying two invertible matrices can produce an uninvertible matrix. So, in fact, layers with uninvertible draw transforms aren't always skipped. CDP makes subtree skipping decisions before computing draw properties, so can't consider draw transform invertibility. On the other hand, property trees make subtree skipping decisions after updating the transform tree, and already store a value ('ancestors_are_invertible') that (despite its name) represents whether the screen space transform is invertible. Once CDP is removed, subtree skipping decisions should use this value. For now (to avoid making property trees and CDP diverge while we still have property tree verification enabled for tests), this CL removes two DCHECKs that expect uninvertible draw transforms. BUG=514540 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1274763004 Cr-Commit-Position: refs/heads/master@{#341984}
* Make LatencyInfo to be classmiletus2015-08-054-11/+10
| | | | | | | | | | | | | LatencyInfo has grown to be more than a pure data holder struct. Make it a class so that we have a clean API surface that can access/modify the data members. BUG=None CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1245753003 Cr-Commit-Position: refs/heads/master@{#341831}
* Don't hide pinch viewport scrollbars near min-scale with device emulation.skobes2015-08-047-18/+33
| | | | | | | | | | | Depends on http://crrev.com/1269813004. BUG=515746 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1267603004 Cr-Commit-Position: refs/heads/master@{#341784}
* All scrolls scroll visual viewport first. (Chromium-side)bokan2015-08-041-0/+58
| | | | | | | | | | | | | | | | Scrolling the viewport involves scrolling two scrollers, the visual and layout viewports. Until now, we would first try to scroll the layout viewport, then, if that was at its maximum scroll extent, we would bubble the scroll to the visual viewport. This CL flips the order of the viewports so that we scroll the visual first when the --invert-viewport-scroll-order flag is set. BUG=443724 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1258953003 Cr-Commit-Position: refs/heads/master@{#341747}
* cc: Keep the transform tree's to_screen_is_animated updatedajuma2015-08-0410-2/+103
| | | | | | | | | | | | | | | The transform tree's to_screen_is_animated is currently set during tree building but is never updated. This CL updates this value as transform animations change, and makes VerifyPropertyTreeValuesForLayer verify that this matches CDP's screen_space_transform_is_animating. BUG=509673 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1227863009 Cr-Commit-Position: refs/heads/master@{#341728}
* SkImage-backed PictureImageLayerfmalita2015-08-042-22/+31
| | | | | | | | | | | | | | | Now that Blink only ever calls SetImage, remove the SetBitmap impls and replace the internal SkBitmap with an SkImage ref. Also update the related unittests to use SkSurface/SkImage. BUG=515935 R=reed@google.com,danakj@chromium.org CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1260913003 Cr-Commit-Position: refs/heads/master@{#341676}
* Adds missing constructor initializationsvarkha2015-08-031-2/+2
| | | | | | | | | BUG=NONE CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1251153002 Cr-Commit-Position: refs/heads/master@{#341485}
* Arrange compositor scrolling into scroll customization format (WIP)tdresser2015-07-313-70/+91
| | | | | | | | | | | | | This splits compositor driven scrolling into applyScroll and distributeScroll methods. Future patches will allow JavaScript on a CompositorWorker to override these methods. BUG=505851 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1215183004 Cr-Commit-Position: refs/heads/master@{#341323}
* cc: Remove target_space_transform_is_animating from DrawPropertiesajuma2015-07-312-21/+59
| | | | | | | | | | | | target_space_transform_is_animating is only used in two places, but in both these places it's makes more sense to use screen_space_transform_is_animating. CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1256333007 Cr-Commit-Position: refs/heads/master@{#341266}
* Assume outer viewport always exists in Viewport::ScrollBy.bokan2015-07-301-108/+31
| | | | | | | | | | | | | | The outer viewport should always exist in the layer hierarchy but this conditional exists because old unit tests were creating hierarchies from pre-virtual-viewport days. This patch fixes these tests and removes the check. BUG=425102 CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel Review URL: https://codereview.chromium.org/1268623003 Cr-Commit-Position: refs/heads/master@{#341128}
generated by cgit v1.1 at 2024-09-14 22:12:21 +0000