GitLab

When calling LayerOwner::RecreateLayer() on the root Layer, the Compositor gets
moved to the new layer after all children are added.

If there are ongoing animations while this occurs they are cancelled. If any
animation observer attempts to access the compositor in response it will crash
Chrome as the compositor is null.

Update LayerOwner::RecreateLayer() to set the Compositor on the new root layer
bfore adding children. This also brings it to the same timing as when the
compositor would be set while performing RecreateLayer on non-root layers.

TEST=LayerOwnerTestWithCompositor.RecreateRootLayerDuringAnimation,
compositor_unittests, ash_unittests, views_unittests
BUG=chrome-os-partner:40032

Review URL: https://codereview.chromium.org/1128103003

Cr-Commit-Position: refs/heads/master@{#329446}

Add NetworkChangeNotifierFactoryCast to create instance of
NetworkChangeNotifierLinux that ignores interfaces that will not be
used to connect to internet.

BUG=469863
R=gunsch@chromium.org

Review URL: https://codereview.chromium.org/1138923002

Cr-Commit-Position: refs/heads/master@{#329445}

In addition to the normal separation of MojoMain() into its own build
target, this patch also renames a class and moves a class into the
surfaces:: namespace. This was done because they were non-unique
symbol names in the resultant binary.

(There were two classes named mojo::ContextProviderMojo and had
different implementations.)

BUG=477435

Review URL: https://codereview.chromium.org/1138293002

Cr-Commit-Position: refs/heads/master@{#329444}

main change: Value::CreateNullValue returns a scoped_ptr

BUG=none

Review URL: https://codereview.chromium.org/1129083003

Cr-Commit-Position: refs/heads/master@{#329443}

For ensuring that the blink is always initialized with WebThread, the
call sites of blink must implement WebThread to expose the task runner.
UtilityThread is the only case where blink is run without exposing the
platform thread. This CL adds new WebThread implementation for
UtilityThread. After this change blink will always initialized with
platform thread and task runner exposed.

BUG=486782

Review URL: https://codereview.chromium.org/1133563007

Cr-Commit-Position: refs/heads/master@{#329442}

Prior to this fix the incoming connection to the content handler saw
an empty requesting url. This change makes the requesting url
correspond to the url of the first app connecting to the content
handler.

BUG=none
TEST=covered by test
R=ben@chromium.org

Review URL: https://codereview.chromium.org/1130353005

Cr-Commit-Position: refs/heads/master@{#329441}

BUG=486916
TEST=See bug comment 1 for repro steps.

Review URL: https://codereview.chromium.org/1134193003

Cr-Commit-Position: refs/heads/master@{#329440}

BUG=486916
TEST=Histogram properly labeled

Review URL: https://codereview.chromium.org/1128993005

Cr-Commit-Position: refs/heads/master@{#329439}

Creates CRWTestJSWindowIdManager to enable force injection,
bypassing hasBeenInjected check in the inject method.

Moves crw_js_window_id_manager_unittest.mm out of public directory.

BUG=465898

Review URL: https://codereview.chromium.org/1129763003

Cr-Commit-Position: refs/heads/master@{#329438}

BatteryManager is offered as a system service and Chrome should not be
instantiating it. Updates the code to retrieve the correct instance.

Review URL: https://codereview.chromium.org/1133253003

Cr-Commit-Position: refs/heads/master@{#329437}

 * Add support for selecting the new OpenGL renderer.
 * Remove the --use-warp flag and merge it into the --use-angle flag.

BUG=485541

Review URL: https://codereview.chromium.org/1123343004

Cr-Commit-Position: refs/heads/master@{#329436}

std::iterators are classes in debug builds in MSVC, so outline
a destructor in a class that has many iterator members in net/.

Make the copy constructor of TilingSetRasterQueueAll private.
Without this, an implicit copy constructor is generated with the
ms abi because the class is marked dllexport, and the implicit
copy constructor needs a copy constructor for TilingIterator
which the plugin also wants to have out-of-line. Making the
outer dllexported class not copyable should fix this.

BUG=467287,483986

Review URL: https://codereview.chromium.org/1131023004

Cr-Commit-Position: refs/heads/master@{#329435}

When run in multi threaded mode, Speech Dispatcher notifies Chrome of significant tts events such as the end of a text utterance. This fixes some long standing issues with users of the tts api such as ChromeVox. Since speechd is indefinitely behind a flag, this change a no real impact except for users who explicitly want speechd.

TEST=run Chrome with speechd enabled, and observe ChromeVox receives end callbacks.

Review URL: https://codereview.chromium.org/1134223003

Cr-Commit-Position: refs/heads/master@{#329434}

The FocusGrid in the extensions list needs to have rows added and
removed to keep it in sync because extensions can now be added without
refreshing the entire list.

BUG=480178

Review URL: https://codereview.chromium.org/1128953004

Cr-Commit-Position: refs/heads/master@{#329433}

This allows the WebIDB* APIs to be covered by Indexed DB OWNERS

Three sided patch:
[1] http://crrev.com/1140573002 Blink - move headers, add temp redirects
[2] This patch - update include paths in chromium
[3] http://crrev.com/1124333009 Blink - remove temp redirects

BUG=453032

Review URL: https://codereview.chromium.org/1133153002

Cr-Commit-Position: refs/heads/master@{#329432}

Rather than having a file that must be kept current with all changes to
third_party/qcms/src highlight in the README.Chromium how the changes
since the import can be readily seen.

R=noel@chromium.org
BUG=None
TEST=None

Review URL: https://codereview.chromium.org/1136283002

Cr-Commit-Position: refs/heads/master@{#329431}

Summary of changes available at:
https://chromium.googlesource.com/chromium/blink/+log/2c4108a..804012f

TBR=dpranke@chromium.org,alancutter@chromium.org

Review URL: https://codereview.chromium.org/1136523009

Cr-Commit-Position: refs/heads/master@{#329430}

Currently the tests of this suite (or, at least,
DownloadHighResAvatarTest) read and write to the default system-wide
location, which is not good and can cause tests flakiness.

Now we override the DIR_USER_DATA to a temporary directory instead.

BUG=486785

Review URL: https://codereview.chromium.org/1138003004

Cr-Commit-Position: refs/heads/master@{#329429}

Currently GetWhispernetClient() internally calls whispernet_client(),
so functionality of whicpernet_client() moved to GetWhispernetClient().
and whispernet_client() removed.

BUG=487175

Review URL: https://codereview.chromium.org/1131173004

Cr-Commit-Position: refs/heads/master@{#329428}

Summary of changes available at:
https://chromium.googlesource.com/skia/+log/0d5fd11..7da2e55

CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
TBR=robertphillips@google.com

Review URL: https://codereview.chromium.org/1136833004

Cr-Commit-Position: refs/heads/master@{#329427}

This avoids a StrictMode policy violation due to I/O on the UI thread.

BUG=473356

Review URL: https://codereview.chromium.org/1124333002

Cr-Commit-Position: refs/heads/master@{#329426}

BUG=none

Review URL: https://codereview.chromium.org/1123343006

Cr-Commit-Position: refs/heads/master@{#329425}

On iOS, having the max at 60 results in the topsites database being
not updated often enough since the app isn't usually running for long
stretches of time. Reduce the value to 5 minutes instead and fix unit
tests.

BUG=485070

Review URL: https://codereview.chromium.org/1133093002

Cr-Commit-Position: refs/heads/master@{#329424}

The overage values should only be applied if the frame is actually
chosen by cadence.  This is most wrong when there is video at some
non zero time and audio at time zero.  If cadence is present the
first frame will have an obscene display count which carries over
and forever wrecks cadence :(

We should also not track render counts until the first frame has
reached its display period.

BUG=439548
TEST=new unittest

Review URL: https://codereview.chromium.org/1124333012

Cr-Commit-Position: refs/heads/master@{#329423}

This change landed in r329325 but was reverted due to an
uninitialized member variable. The issue has been fixed.

Original description: ----------------------------

Previously, cached values in the TouchSelectionController would be reset
to force future selection updates. However, these cached values can
actually be used outside of selection update calls, e.g., when force
showing the selection from the current cached values. Instead of
resetting the cached values, simply set a dirty bit that forces an
update, avoiding issues when dealing with the reset values.

BUG=393025

Review URL: https://codereview.chromium.org/1127383007

Cr-Commit-Position: refs/heads/master@{#329422}

When mandoline is run with --enable-multiprocess, we should run some
services in different processes. This reuses |qualifier| in
Identity (which is already used for content handlers that need to run in
different contexts) to be a process name in contexts where there is a
package alias. This places mojo:network_services in its own process, and
everything else in a different process.

When run without --enable-multiprocess, this change is a noop.

BUG=477435

Review URL: https://codereview.chromium.org/1126393006

Cr-Commit-Position: refs/heads/master@{#329421}

BUG=

Review URL: https://codereview.chromium.org/1135263002

Cr-Commit-Position: refs/heads/master@{#329420}

If the active tab is pinned and you have at least one pinned tab after the
active tab then we would end up opening new link clicks after the
pinned tab regardless of the opener. This is a regression from
https://codereview.chromium.org/836223004 .

BUG=446997, 446998, 476038
TEST=covered by test now.
R=pkasting@chromium.org

Review URL: https://codereview.chromium.org/1131233004

Cr-Commit-Position: refs/heads/master@{#329419}

With crrev.com/1112063002, they're not needed anymore.

BUG=486806

Review URL: https://codereview.chromium.org/1138793003

Cr-Commit-Position: refs/heads/master@{#329418}

This patch introduces a BackgroundSyncProvider for use by the Blink
BackgroundSync module, as well as a proxy object which is used when calling the
provider from off of the main thread. The code in BlinkPlatformImpl will
dispatch to the appropriate object depending on the running thread.

This patch depends on the Mojo service definitions in
https://codereview.chromium.org/1106193002

BUG=479251

Review URL: https://codereview.chromium.org/1104283002

Cr-Commit-Position: refs/heads/master@{#329417}

BUG=

Review URL: https://codereview.chromium.org/1130833005

Cr-Commit-Position: refs/heads/master@{#329416}

This is a speculative fix for crash which happens on calling
respondToSelector: method on dangling web_controller_observer_ pointer.

BUG=486668

Review URL: https://codereview.chromium.org/1136753004

Cr-Commit-Position: refs/heads/master@{#329415}

BUG=369661
TEST=All tests remain green.

Review URL: https://codereview.chromium.org/1137773003

Cr-Commit-Position: refs/heads/master@{#329414}

Revert of Revert of Adjusts dragging logic to be less likely to trigger false positive switch from snapping to docking (patchset #1 id:1 of https://codereview.chromium.org/1128933005/)

Reason for revert:
This revert probably did not fix the flakiness. I suspect that a revert that actually ended up fixing the bots is this: https://codereview.chromium.org/1139943002.

I will re-land my original change.

Original issue's description:
> Revert of Adjusts dragging logic to be less likely to trigger false positive switch from snapping to docking (patchset #5 id:90001 of https://codereview.chromium.org/1127133003/)
>
> Reason for revert:
> I suspect this is causing a browser_tests failure on the bots:
>
> http://build.chromium.org/p/chromium.chromiumos/builders/Linux%20ChromiumOS%20Tests%20%28dbg%29%281%29/builds/2069
> https://build.chromium.org/p/chromium.webkit/builders/Linux%20ChromiumOS%20Tests%20%28dbg%29%281%29/builds/1913
>
> Sorry!
>
> Original issue's description:
> > Adjusts dragging logic to be less likely to trigger false positive switch from
> > snapping to docking.
> >
> > This CL adjusts the logic which selects whether a window will be snapped to
> > half of the workspace width or will be docked at the end of the drag when the
> > window is dragged to the edge of the workspace in the special case of the dock
> > being visible. The CL makes it easier to snap a window to half the workspace
> > width in this special case (and harder to inadvertently dock the window)
> >
> > BUG=484877
> > TEST=None
> >
> > Committed: https://crrev.com/988c697aff35b3fd347dec35f2f2ac159c8328a9
> > Cr-Commit-Position: refs/heads/master@{#329268}
>
> TBR=pkotwicz@chromium.org,varkha@chromium.org
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=484877
>
> Committed: https://crrev.com/0d4f7d597774f7415f1aff374db498c8fbe88a9a
> Cr-Commit-Position: refs/heads/master@{#329301}

TBR=pkotwicz@chromium.org,dpranke@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=484877

Review URL: https://codereview.chromium.org/1140733002

Cr-Commit-Position: refs/heads/master@{#329413}

BUG=483627
TEST=unit_tests

Review URL: https://codereview.chromium.org/1128753003

Cr-Commit-Position: refs/heads/master@{#329412}

The khronos headers are present in GYP, but missing in GN, causing
the build using GN to fail when 'ozone_platform_gbm=true' is used.

BUG=487108

Review URL: https://codereview.chromium.org/1129713003

Cr-Commit-Position: refs/heads/master@{#329411}

Make sure we use the same file in all instances of VaapiWrapper.

Define a refcounted global instance of VADisplayState that is only initialized
(via vaInitialize()) once and destroyed upon release of the last refcount.
This ensures a single drm_intel_bufmgr per fd, and avoids the double-close
problem as seen in bug 464628.

Pipe a PreSandboxInitialization to open the render node before starting sandbox.

After this CL the VEA/VDA unittests need to invoke the PreSandboxInitialization
as well.

BUG=475250
TEST=verify HW video works on link (check histograms) and VDA/VEA unittests

Review URL: https://codereview.chromium.org/1137483002

Cr-Commit-Position: refs/heads/master@{#329410}

Previously it was split between the helper thread and
the main thread. Way back in the day it was only the
helper thread, but there was concern about thread-safety
and part of the process moved to the main thread,
straddling two threads. Now we move all the way back to
the main thread only.

This removes the need to have "WaitFor*Start" in the helper
thread. Otherwise the helper thread would post work to
the main thread, and wait. That makes it very risky w.r.t.
deadlock, since the main thread will join() the helper
thread as part of plugin teardown. We had a bunch of
workarounds like a timeout when waiting on the condvar.

The bad thing about main thread loads is that it blocks
the renderer from doing work (like animation) as the nexe
loads. However, since we bounce work to the main thread
anyway, we block right now without this CL already.
In a later CL we can take advantage of the sel_main_chrome
load status hook to know when the nexe load is about done,
and only then call-back into the plugin when the blocking
is minimized. Hook added here:
https://codereview.chromium.org/1090233003/diff/100001/components/nacl/common/nacl_renderer_messages.h

SRPC client connection is still done on the helper thread.
I'm not sure of the requirement, but it will block
indefinitely if we do that in the main thread so
I left that in the helper thread.

There's still a bunch of pnacl_translate_thread mutexes
and condvars and the *_subprocess_active flags left
in the pnacl_translate_thread. Haven't moved the
activeness tracking out so that's left as is.

BUG= 473474

Review URL: https://codereview.chromium.org/1128943003

Cr-Commit-Position: refs/heads/master@{#329409}

The usual channel for delivering onReceivedTitle (via updating title for
a NavigationEntry) doesn't work in the case of history navigations, because
WebContentsImpl silences no-op updates of entries titles.

Chrome Browser uses another event for updating the title in this case --
via WebContentsDelegate::LoadingStateChanged. This patch enables WebView to
use it as well.

Note that WebChromeClient.onReceivedIcon doesn't need to be re-sent, as
nobody seems to bother checking whether the icon update is a "no-op", so
updates on history navigations just work.

BUG=481570

Review URL: https://codereview.chromium.org/1126383005

Cr-Commit-Position: refs/heads/master@{#329408}

Remove --enable-spdy4 and --enable-npn-http command line flags that used to
force enable and disable both of HTTP/2 and SPDY/3.1, respectively.  They are
obsolete because
 * these protocols are not experimental any more,
 * they are enabled by default,
 * usage of disabling flag is very low and it is unlikely that these flags will
   ever be needed for debugging.

BUG=

Review URL: https://codereview.chromium.org/1130353003

Cr-Commit-Position: refs/heads/master@{#329407}