GitLab

This enables unit tests to run with Bluetooth permission,
by adding the Bluetooth permission to native_test.

Non-test applications will not have the Bluetooth permission,
and unit tests should also verify behavior
when the permission is not given. To enable this
createWithoutPermissionForTesting is added and
results in a state equivalent to when the permission
is not available.

Unit tests will be built in parallel for both when Bluetooth
permission exists and doesn't.

BUG=471536

Committed: https://crrev.com/30e81472dc168ea1c331a8779b3f6a14f684c54b
Cr-Commit-Position: refs/heads/master@{#329212}

Review URL: https://codereview.chromium.org/1129683002

Cr-Commit-Position: refs/heads/master@{#329286}

BUG=464922

Review URL: https://codereview.chromium.org/1129123004

Cr-Commit-Position: refs/heads/master@{#329285}

This CL introduces the 'gaia-icon-button' decorator, which can be applied to any
<button>.
All icon buttons introduced recently are <button is="gaia-icon-button">s now.

BUG=478545,472671
TEST=manual

Review URL: https://codereview.chromium.org/1133733003

Cr-Commit-Position: refs/heads/master@{#329284}

This change was reverted in r322170 due to WebView breakage. The
ApplicationStatus dependency has been made optional, allowing
WebView to opt-out of its use. It was speculatively reverted
again in r327092, but that turned out to be a false alarm.

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

Currently, when an activity is stopped, we explicitly hide the
foreground Tab. This is problematic, as current hiding semantics
might clear the visual front buffer before the window is hidden.
This in turn causes an unpleasant flickering during activity
transitions, e.g., when backgrounding Chrome or locking the screen.

Wire Activity onPause/onResume notifications to WindowAndroidObservers,
allowing the foreground tab to preserve its front buffer while hiding
its web content. If the tab is explicitly hidden, or the root window
is lost, the front buffer will be cleared as usual.

BUG=481450,434401

Review URL: https://codereview.chromium.org/1001573003

Cr-Commit-Position: refs/heads/master@{#329283}

Summary of changes available at:
https://chromium.googlesource.com/chromium/blink/+log/ff93a86..38a0fb8

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

Review URL: https://codereview.chromium.org/1128003004

Cr-Commit-Position: refs/heads/master@{#329282}

http://crrev.com/328649 changed VideoRendererImpl so that it only
posted AttemptRead() when there was space in the queue.  Unfortunately
this causes more dropped frames for demanding content (H.264 4K in
particular) than always posting since the AttemptRead() call may be
delayed for some time:

AttemptRead() TimeFromPostUntilExecuted: 0.076 ms
AttemptRead() TimeFromPostUntilExecuted: 4.613 ms
AttemptRead() TimeFromPostUntilExecuted: 56.158 ms
AttemptRead() TimeFromPostUntilExecuted: 56.197 ms
AttemptRead() TimeFromPostUntilExecuted: 52.214 ms
AttemptRead() TimeFromPostUntilExecuted: 35.555 ms
AttemptRead() TimeFromPostUntilExecuted: 18.917 ms

At the same time, the performance improvements I saw locally on my
MacBook are not reflected by the performance bots.  I further see the
dropped frame regression locally on Windows, so let's revert.

I considered a couple other solutions to avoid always posting, but
none were very glamorous:
- Release lock before calling AttemptRead(), fixes cases where Render()
was blocked on the lock in FrameReady().
- Checking the end time / ideal render count for the current frame and
posting on when we're almost expired...

In both cases, neither resolved all problem cases and both add more
complexity than just posting.  The task will abort trivially if there
is no work to do, so just revert.

BUG=439548
TEST=telemetry dropped frame count is lower.

Review URL: https://codereview.chromium.org/1135143002

Cr-Commit-Position: refs/heads/master@{#329281}

Brings in coverage support for Windows.

Green build with this revision:
http://build.chromium.org/p/chromium.fyi/builders/CrWinAsan%20tester/builds/217/steps/gclient%20runhooks/logs/stdio

TBR=inferno@chromium.org
NOTRY=true

Review URL: https://codereview.chromium.org/1140623002

Cr-Commit-Position: refs/heads/master@{#329280}

BUG=484920

Review URL: https://codereview.chromium.org/1133523004

Cr-Commit-Position: refs/heads/master@{#329279}

Blink tests have been modified to use the new layoutAndPaintAsync()
and layoutAndPaintAsyncThen().

BUG=484706

Review URL: https://codereview.chromium.org/1131803003

Cr-Commit-Position: refs/heads/master@{#329278}

With this change, the Data Reduction Proxy bypass logic will be applied
to any response that either used one of the currently configured Data
Reduction Proxies, or used a different proxy that sent the Data
Reduction Proxy via header.

BUG=476610

Review URL: https://codereview.chromium.org/1135153002

Cr-Commit-Position: refs/heads/master@{#329277}

This was caused by my recent change to allow handles other than STDOUT and STDERR to be shared
with the target. Reason for the crash was copying additional handles to the HANDLE array which had
space for 2 handles only.

Fix is to use scoped_ptr instead and allocate appropriate space for all handles being shared.

BUG=486434
R=cpu

Review URL: https://codereview.chromium.org/1128903006

Cr-Commit-Position: refs/heads/master@{#329276}

ContextualSearchEventFilter has a complex logic to detect
where to propagate events to, whether the Search Panel or
the Search Content View. Besides that, a single stream of
events (a single gesture) might be forwarded to different
targets depending on a number of factors (whether the
Panel was maximized, whether the gesture was horizontal
or vertical, whether the Search Content View is scrolled
at the top, and so on).

Things can get much more complicated when multiple
pointers are involved, like in gestures with multiple
fingers. If we mess with the order of the events, or
don't properly forward events from all pointers to the
Search Content View it will crash.

This CL fixes the problem by:

1) Forwarding all pointer events when multiple finger
gestures are used.

2) Ignoring multitouch events when forwarding horizontal
gestures to the expanded Panel. In this particular case,
we must ignore multitouch events because the presence of
a second pointer will cause the Search Page to scroll,
which is not supposed to happen in this case. So, now
we are properly ignoring ACTION_POINTER_DOWN and
ACTION_POINTER_UP events, while at the same time,
creating MotionEvents with a single pointer in order
to implement the intended "lock horizontally" behavior.

BUG=485008
BUG=486901
BUG=486902

Review URL: https://codereview.chromium.org/1137003003

Cr-Commit-Position: refs/heads/master@{#329275}

While I was at it, I also added the page title as a translated string, as this seems to have been forgotten in the first implementation.

Review URL: https://codereview.chromium.org/1137063002

Cr-Commit-Position: refs/heads/master@{#329274}

The linux_me2me_host script launches the host process, then writes the
JSON config to the child's stdin. If the host process dies before the
data is written to stdin, an IOError (EPIPE) exception is raised. This
CL handles the exception and retries launching the host, instead of
exiting the script.

To test this, I added a time.sleep(5) call between launching the
host and writing the config to stdin. This allowed me enough time to
simulate the error by killing the host process.

BUG=473441

Review URL: https://codereview.chromium.org/1138613002

Cr-Commit-Position: refs/heads/master@{#329273}

This was only used in tests. In that case there's no reason to send the contents to another process, so we can use the simpler cc::SoftwareOutputDevice instead.

Review URL: https://codereview.chromium.org/1136553004

Cr-Commit-Position: refs/heads/master@{#329272}

Make it visually aligned with other buttons from chrome error
pages.  Make it easier to see by giving it more contrast.  Update
the formatting to remove an explicit height being set to give a
nicer look on various orientations of devices.

BUG=486816

Review URL: https://codereview.chromium.org/1136133003

Cr-Commit-Position: refs/heads/master@{#329271}

BUG=480565

Review URL: https://codereview.chromium.org/1097303003

Cr-Commit-Position: refs/heads/master@{#329270}

We recently introduced SchemeIsCryptographic() and IsOriginSecure(),
which are meant to replace SchemeIsSecure().

IsOriginSecure() roughly means "do we trust this content not to be
tampered with before it reaches the user?" [1] This is a higher-level
definition that corresponds to the new "privileged contexts" spec. [2]

SchemeIsCryptographic() [3] is close to the old definition of
SchemeIsSecure(), and literally just checks if the scheme is a
cryptographic scheme (HTTPS or WSS as of right now). The difference is
that SchemeIsCryptographic() will not consider filesystem URLs secure.

[1] https://code.google.com/p/chromium/codesearch#chromium/src/content/public/common/origin_util.h&sq=package:chromium&type=cs&l=19&rcl=143099866
[2] https://www.chromium.org/Home/chromium-security/prefer-secure-origins-for-powerful-new-features and https://w3c.github.io/webappsec/specs/powerfulfeatures/
[3] https://code.google.com/p/chromium/codesearch#chromium/src/url/gurl.h&sq=package:chromium&type=cs&l=250&rcl=1430998666

BUG=362214

Review URL: https://codereview.chromium.org/1128363006

Cr-Commit-Position: refs/heads/master@{#329269}

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

Review URL: https://codereview.chromium.org/1127133003

Cr-Commit-Position: refs/heads/master@{#329268}

I was able to repro the flake with ~1% probability and traced.

It was flaky because earlier window server messages were still pending
when the test started running checks. So the first `ActivateSync` would
actually break out of the message loop wait too soon.

BUG=484836

Review URL: https://codereview.chromium.org/1130683004

Cr-Commit-Position: refs/heads/master@{#329267}

Add a zoomDisabled flag (disabled so that it will default to false
everywhere except where we set it) to WebGestureEvent.

Note that the equivalent blink::Event does not have a corresponding
gesture sub-type that sends pinch scale information, so the
zoomDisabled information is not passed on either.

Blink side:
https://codereview.chromium.org/1127013002/

BUG=478981

Review URL: https://codereview.chromium.org/1129693002

Cr-Commit-Position: refs/heads/master@{#329266}

I was ambitious in r325585 to remove this behavior, asserting that the Port's
owner should clear its own listeners. This is an optimistic expectation, and a
backwards incompatible one at any rate.

BUG=486715
R=rockot@chromium.org

Review URL: https://codereview.chromium.org/1135123002

Cr-Commit-Position: refs/heads/master@{#329265}

BUG=486488
Test=Check some link that has search term, e.g.
https://developer.chrome.com/extensions/browsingData#samples
click on the "on the samples page" link.

Review URL: https://codereview.chromium.org/1124343004

Cr-Commit-Position: refs/heads/master@{#329264}

Also remove unused data reduction pref from preferences API.

BUG=442494

Review URL: https://codereview.chromium.org/1130863002

Cr-Commit-Position: refs/heads/master@{#329263}

Two independent but related fixes here:

1. IPCSupportInitializer could previously outlive its IO
   runner, resulting in Mojo not being shut down in time.
   The EDK posts tasks directly to a MessageLoopForIO
   pointer because it's assumed to remain valid until
   EDK shutdown. This CL limits IPCSupportInitializer's
   lifetime to that of the IO thread, regardless of
   ScopedIPCSupport lifetime.

2. In single-process mode, the child IO thread cannot be
   used to create the client ChannelMojo, because all channels
   in-process must be created on the EDK IO thread. This problem only
   surfaced in practice because #1 above changed some shutdown
   ordering and tickled a DCHECK in EDK shutdown.

   This CL has ChannelMojo hop to its IO runner (only if
   necessary) before creating underlying messaging pipes.

BUG=None
R=morrita@chromium.org

Review URL: https://codereview.chromium.org/1130413002

Cr-Commit-Position: refs/heads/master@{#329262}

Adds support for multiple thumbnail files on Single iframe NTP, fallback to the best one that opens.

BUG=486879

Review URL: https://codereview.chromium.org/1129383003

Cr-Commit-Position: refs/heads/master@{#329261}

BUG=471536

Review URL: https://codereview.chromium.org/1125213005

Cr-Commit-Position: refs/heads/master@{#329260}

Summary of changes available at:
https://chromium.googlesource.com/chromiumos/platform/system_api/+log/789778e..f2d7a9a

BUG=486738
TBR=pneubeck@chromium.org

Review URL: https://codereview.chromium.org/1128803005

Cr-Commit-Position: refs/heads/master@{#329259}

It's not used anymore.

Review URL: https://codereview.chromium.org/1139483002

Cr-Commit-Position: refs/heads/master@{#329258}

Summary of changes available at:
https://chromium.googlesource.com/chromium/blink/+log/f2af598..ff93a86

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

Review URL: https://codereview.chromium.org/1134583004

Cr-Commit-Position: refs/heads/master@{#329257}

* Use the standard integer types from stdint.h instead.
* Use macros.h for the DISALLOW_COPY_AND_ASSIGN macro.

BUG=138542
TEST=sql_unittests
R=shess@chromium.org

Review URL: https://codereview.chromium.org/1133053004

Cr-Commit-Position: refs/heads/master@{#329256}

If rendering code that is shared with GrContext changes any GL state,
then skia caches need to be notified.  This adds InvalidateGrContext()
method that will call private pure virtual method.  InvalidateGrContext
default argument will clear all state in GrContext.

Updated classes that derive from ContextProvider.

R=piman@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1131723002

Cr-Commit-Position: refs/heads/master@{#329255}

BUG=486780

Review URL: https://codereview.chromium.org/1131513003

Cr-Commit-Position: refs/heads/master@{#329254}

BUG=369661
TEST=All navigation tests remain green.

Review URL: https://codereview.chromium.org/1129913004

Cr-Commit-Position: refs/heads/master@{#329253}

R=inferno@chromium.org
BUG=450268

Review URL: https://codereview.chromium.org/1131293005

Cr-Commit-Position: refs/heads/master@{#329252}

Based on the crash dump, the crash occurs while loading a Nacl module in the renderer process. The Nacl translate
thread has a valid file handle which is created by the Nacl host in the browser. It then calls into the Nacl loader
to load the module which fails. The Nacl loading code in LaunchSelLdr function is closing the file handle which is
passed in. Based on comments in the PnaclTranslateThread class, ownership of the file handle is only transferred on success.

Thus when the call returns the PnaclTranslateThread code tries to close the file handle which is already closed. In the
meantime the Windows handle is reused to something else which is tracked by our handle tracker. The second CloseHandle
attempt causes a CHECK to fire because we are closing a handle which is being tracked.

Fix is to not close the file handle in the PnaclTranslateThread class as ownership is transferred on
call to LaunchSelHdr.

BUG=426582, 475872

Review URL: https://codereview.chromium.org/1137833003

Cr-Commit-Position: refs/heads/master@{#329251}

This consists of the following pieces:
. html_viewer now uses ResourceProvider to get resources.
. html_viewer no longer directly uses java.
. The java support for mojo_runner/shell has been mode pluggable:
  . You specify the name of the .so to load by way of a meta-tag in
    the manifest. This way mandoline can specify it's own.
  . Android's main function calls to an InitContext() function that
    mandoline plugs in.
  . The apk can include resources to extract. These are listed in the
    file 'assets_list.' The gn template rule
    generate_mojo_shell_assets_list does this for you given a
    directory.
. Changed around scripts to work with both mojo_shell/runner and
  mandoline.

BUG=481698
TEST=covered by tests
R=ben@chromim.org, jcivelli@chromium.org

Review URL: https://codereview.chromium.org/1130763004

Cr-Commit-Position: refs/heads/master@{#329250}

Such that they are based on the same base class.

BUG=452524

Review URL: https://codereview.chromium.org/1009403002

Cr-Commit-Position: refs/heads/master@{#329249}

Using https://crrev.com/15679025 as template, but with some complication
involving base::CancelableTaskTracker.

BUG=467712

Review URL: https://codereview.chromium.org/1122103003

Cr-Commit-Position: refs/heads/master@{#329248}

Render frames can use this service to verify platform on ChromeOS.

BUG=479836

Review URL: https://codereview.chromium.org/1053773007

Cr-Commit-Position: refs/heads/master@{#329247}