GitLab

When native GPU memory buffers were enabled we noticed many crashes
reporting mmap failing.

This CL adds fd number to crash report when dmabuf mmap fails, because we think
fd limit may cause this crash.
It's follow-up CL of https://codereview.chromium.org/2710183005/

BUG=629521

Review-Url: https://codereview.chromium.org/2725813005
Cr-Commit-Position: refs/heads/master@{#456154}

On Linux, Video encoding is currently erroneously reported on chrome://gpu as "Hardware
accelerated". This CL fixes this by adding a blacklist entry.

BUG=641485
TEST=Open chrome://gpu on Linux and check that Video Encode is reported as
     Software only, hardware acceleration unavailable
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2744683002
Cr-Commit-Position: refs/heads/master@{#456153}

This patch adds a Java wrapper for RenderFrameHost. It also adds a
getMainFrame method to WebContents.

This will be useful for Payment Request which needs to know
the exact (sub)frame issued the payment request.

BUG=620173
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

Review-Url: https://codereview.chromium.org/2681933002
Cr-Original-Commit-Position: refs/heads/master@{#455635}
Committed: https://chromium.googlesource.com/chromium/src/+/67dd9f2795389cc3776481ff07740984b2389f43
Review-Url: https://codereview.chromium.org/2681933002
Cr-Commit-Position: refs/heads/master@{#456152}

See also https://codereview.chromium.org/2641123006.

BUG=700229

Review-Url: https://codereview.chromium.org/2741193003
Cr-Commit-Position: refs/heads/master@{#456151}

This also switches the snappy mirror reference to the new mirror that
tracks the google/snappy GitHub repository. See http://crbug.com/693936
for details.

BUG=533648

Review-Url: https://codereview.chromium.org/2712273002
Cr-Commit-Position: refs/heads/master@{#456150}

No new code, just shuffling things around so WebKit/Source/modules
and content/renderer have a more related structure, also in terms
of bug components and OWNERS.

BUG=none  (no new code)

TBR=jochen@chromium.org
for adding a third_party/skia to DEPS; this dependency was
in content/renderer/media before anywa, and
third_party/skia hasn't got OWNERs itself :?

Review-Url: https://codereview.chromium.org/2739253003
Cr-Commit-Position: refs/heads/master@{#456149}

This component provides one dedicated thread per
SingleThreadTaskRunner.

Reference Change:
https://codereview.chromium.org/2686593003/

BUG=684080, 694823, 697697

Review-Url: https://codereview.chromium.org/2698843006
Cr-Original-Commit-Position: refs/heads/master@{#453162}
Committed: https://chromium.googlesource.com/chromium/src/+/fc203f52e046a8d82e053525d764671f7128f3d4
Review-Url: https://codereview.chromium.org/2698843006
Cr-Commit-Position: refs/heads/master@{#456148}

It is said to be a M57 regression, but no one has touchedthe feedback dialog for a while. Regardless, this CL makes the dialog look like the expected_report.mp4 in the bug. Tested for RTL as well.

TEST=manual
BUG=696913
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2723613005
Cr-Commit-Position: refs/heads/master@{#456147}

As a result of this patch, we have:
- 65 icons of 28x24 size (belong to "large" icons)
- 24 icons of 10x10 size (belong to "small" icons)

The rest of the icons belong to "medium" icons and will
be normalized to be 16x16.

R=dgozman, pfeldman
BUG=700249

Review-Url: https://codereview.chromium.org/2744643003
Cr-Commit-Position: refs/heads/master@{#456146}

We are adding COMPONENT/TEAM information into OWNERS file
(security, etc).

Proposal to add TEAM/COMPONENT information into OWNERS files
http://bit.ly/add-team-component-proposal
Proposal about how to get suggested component for directory.
http://bit.ly/directory-mapping-proposal
TEAM-COMPONENT mapping
http://bit.ly/component-team-mapping

Additional Information:
Component lists
https://bugs.chromium.org/p/chromium/adminComponents

BUG=679905
R=estark@chromium.org

Review-Url: https://codereview.chromium.org/2741833003
Cr-Commit-Position: refs/heads/master@{#456145}

This CL implements the get() and fetch() methods for Background Fetch.
The BackgroundFetchRegistration interface and BackgroundFetchOptions
and IconDefinition dictionary types are added too.

The method has not been hooked up with the browser process yet, as the
Mojo interface has not been implemented yet.

Tests are provided, and will be moved to the WPT directory once it has
been approved upstream:
  https://github.com/w3c/web-platform-tests/pull/5094

BUG=692534

Review-Url: https://codereview.chromium.org/2745543002
Cr-Commit-Position: refs/heads/master@{#456144}

We are adding COMPONENT/TEAM information into OWNERS file.

Proposal to add TEAM/COMPONENT information into OWNERS files
http://bit.ly/add-team-component-proposal
Proposal about how to get suggested component for directory.
http://bit.ly/directory-mapping-proposal
TEAM-COMPONENT mapping
http://bit.ly/component-team-mapping

Additional Information:
Component lists
https://bugs.chromium.org/p/chromium/adminComponents

BUG=679905
R=eranm@chromium.org

Review-Url: https://codereview.chromium.org/2740133004
Cr-Commit-Position: refs/heads/master@{#456143}

On Android, when we'd like to use a SurfaceView for video playback,
we must enable the alpha channel on the CompositorView.  Otherwise,
we cannot draw a hole to see the video's SurfaceView.  We can't keep
transparency enabled all the time since it can cause power
regressions on some hardware.

However, because SurfaceView destroys and recreates the surface when
changing the pixel format, there is a visible flash during the
transition.  This CL removes that.

It causes the CompositorView to have two SurfaceView child views,
rather than extending SurfaceView directly.  When a request is made
to switch the output format, it makes the change to the background
SurfaceView, and swaps it to the front.  It also keeps the outgoing
SurfaceView around until the new one is in use by the compositor.

There are a few interesting bits:

- SurfaceViews are invisible until the first buffer is posted.  So,
  we can continue to see the outgoing view until the compositor is
  ready to use the new one.
- All buffers are released by the outgoing SurfaceView when its
  visibility is set to Gone.  'dumpsys SurfaceFlinger' shows that the
  same amount of gralloc memory is used in steady state.
- No power regression was observed on a Nexus 5.
- Unfortunately, new SurfaceViews are z-ordered behind existing ones,
  which makes it critical that we guess when to delete the outgoing
  (topmost) SurfaceView.  We currently time one frame, and then hide
  it.  Empirically, this works fine.
- This might seem like a more natural extension to
  CompositorViewHolder, but the CompositorView currently owns the
  native compositor.  Since CompositorViewHolder is already fairly
  complicated, it wasn't clear that it would be a good idea to
  refactor that into CVH.

Another approach is to use EGL to change the buffer format to include
an alpha channel, or not.  This avoids the power regression, since
opaque surfaces and buffers without alpha channels are treated the
same by SurfaceFlinger.  However, it causes problems for virtualized
contexts.  In particular, the off-screen contexts will have an alpha
channel at all times, so they cannot be shared with the compositor
context without one.  For some hardware (e.g., QC) that requires the
use of virtualized GL contexts rather than share groups, this may
have a stability regression.  So, we don't try it.

Please see https://goo.gl/aAmQzR for the design doc.

BUG=629130

Review-Url: https://codereview.chromium.org/2201483002
Cr-Commit-Position: refs/heads/master@{#456142}

generate_test_data.sh uses `hdiutil create` and `hdiutil convert`. The former
can overwrite an existing output file, but the latter cannot. This led to the
following error:

  hdiutil: convert failed - File exists

But this message was getting swallowed because the script 1) did not do `set -e`
to stop on errors, and 2) redirect all output to /dev/null. This also fixes both
of those issues.

BUG=696529
R=thakis@chromium.org

Review-Url: https://codereview.chromium.org/2748453002
Cr-Commit-Position: refs/heads/master@{#456141}

Revert: https://codereview.chromium.org/2729733003/

As this is currently crashing on all mash builds, preventing development.

TBR=rockot@chromium.org, jochen@chromium.org, sammc@chromium.org
BUG=700155, 654988

Review-Url: https://codereview.chromium.org/2742523005
Cr-Commit-Position: refs/heads/master@{#456140}

Review-Url: https://codereview.chromium.org/2740173002
Cr-Commit-Position: refs/heads/master@{#456139}

Add unit tests using the new NavigationSimulator infrastructure

BUG=637415

Review-Url: https://codereview.chromium.org/2731013002
Cr-Commit-Position: refs/heads/master@{#456138}

The LayoutManagerChrome can still be null when #onNewTabPageShown()
is called for the first time.

BUG=699935

Review-Url: https://codereview.chromium.org/2741843003
Cr-Commit-Position: refs/heads/master@{#456137}

This CL replaces the Clang 'c++-analyzer' Perl script with logic added to the Python wrapper script. This gives us the hooks we need to run analysis builds with "gomacc" (necessary for distributed builds) and gives us a convenient place to specify analyzer config flags which aren't exposed by scan-build.

Adds a warning suppression rule which prevents the analyzer from raising false positives from stdlib code (code w/namespace "std").

Other changes:
* Modify Goma portions of the GCC toolchain GNI to build paths to
  the Clang static analyzer.
* Create an exception for ASM tool invocations, which don't play
  well with the analysis command line flags.
* Removed 'scan-build' DEP.

BUG=687245
R=thakis@chromium.org,wez@chromium.org

Review-Url: https://codereview.chromium.org/2667853004
Cr-Commit-Position: refs/heads/master@{#456136}

Wrap usage of ContentVideoView in an AndroidOverlay interface.  This
doesn't change the functionality, but starts to move AVDA towards
using AndroidOverlays rather than hard-coding CVV.

Much of Allocate/DeallocateSurface could be moved from
AVDACodecAllocator to ContentVideoViewOverlay.  However, to keep the
size of this change small, it isn't yet.

There shouldn't be any functional difference with this CL.

BUG=667950

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2692863011
Cr-Commit-Position: refs/heads/master@{#456135}

Child frames have some degree of continuity across navigations: the
contents are still embedded in the same document. Let them continue to
propagate user gesture state across navigations. However, main frames
don't have that continuity and should reset their user gesture state
on navigation.

TEST=DocumentUserGestureTokenTest.Navigate
BUG=chromedriver:922

Review-Url: https://codereview.chromium.org/2721423003
Cr-Commit-Position: refs/heads/master@{#456134}

TEST=manual; verify Docs, Sheets, and Slides work with regard to whitespace, key echo, etc when braille mode is off.
BUG=690635,690636,618099
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2740793002
Cr-Commit-Position: refs/heads/master@{#456133}

This CL changes the name of a CSS variable. The checkbox spacing was/is
being used for spacing between a control label and something graphical
such as the checkbox or a policy indicator. The intent is to convey
that it's used for more than checkbox spacing.

BUG=None
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2744763003
Cr-Commit-Position: refs/heads/master@{#456132}

Just like with touch event handlers, at present a sub-frame's
compositor does not receive regions rects for non-fast scrollable
regions that accompany MouseWheel handlers. Until ScrollingCoordinator
is made per-frame, we should mark the entire root layer rect as
non-fast scrollable in the presence of blocking wheel handlers.

The primary changes are all in render_widget_compositor.cc; other
files are touched only to give the function a more inclusive name.

BUG=698195

Review-Url: https://codereview.chromium.org/2740133002
Cr-Commit-Position: refs/heads/master@{#456131}

Implement compositor bypass for WebVR via mailbox copy:

- Create a GPU command buffer GL context in the VrShellGl thread that's
  backed by a native Surface. This is encapsulated inside the new
  MailboxToSurfaceBridge class.

- WebKit's VRDisplay sends a MailboxHolder for each frame (as drawn
  via WebGL on the GPU process) as argument to the SubmitFrame mojo
  call.

- MailboxToSurfaceLayerBridge uses its GPU command buffer GL context
  to draw the mailbox content to a Surface.

- VrShellGl picks up the frame from the corresponding SurfaceTexture
  and  proceeds with rendering as usual with its native GL context
  as needed for the low-level GVR API.

Also implements on-the-fly canvas resizing, the transfer surface and
GVR renderbuffer are resized as needed to match the source canvas.

Added frame number as an argument to various TRACE_EVENTs for timing
analysis.

It includes two separate modes for frame scheduling, the more
aggressive one is conditional on
chrome:://flags#enable-webvr-experimental-rendering since the
framerate seems a bit less stable.

BUG=698643
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:closure_compilation;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2729523002
Cr-Original-Commit-Position: refs/heads/master@{#455592}
Committed: https://chromium.googlesource.com/chromium/src/+/6280ec5b6e3b4c307926eb1bf0167acc2a0d4a74
Review-Url: https://codereview.chromium.org/2729523002
Cr-Commit-Position: refs/heads/master@{#456130}

In ReloadType, NONE means that the navigation is not a reload, so it
makes no sense to call Reload function with ReloadType::NONE as
parameter.

This CL clarifies it by mentioning it in public API comment and add a
DCHECK in the implementation.

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

Review-Url: https://codereview.chromium.org/2740843004
Cr-Commit-Position: refs/heads/master@{#456129}

This adds the targetsAtLeastO() helper function to BuildInfo. I also
updated the check we were using for HTTPS clear text to use this
instead of manually checking targetSdkVersion.

This helper should be updated once the SDK level for O has been
finalized.

BUG=688556

Review-Url: https://codereview.chromium.org/2745663002
Cr-Commit-Position: refs/heads/master@{#456128}

While running LayoutTests, this test timed out. I was curious, looked into it, and realized that the test was completely broken:

1) there was an extra argument getting passed to testType() resulting in assert function being garbage. This caused 1000s of failures which caused test to run slowly.

2) Some other arguments being passed in were also wrong.

My fix is not perfect, but it results in a saner outcome: it runs in fraction of a second, and triggers fewer failures.

Running LayoutTests on my local machine was impossible, so I have not generated new test expectations.

BUG=697646

Review-Url: https://codereview.chromium.org/2721243004
Cr-Commit-Position: refs/heads/master@{#456127}

Once browser initiated nagivations are enabled, we need a secondary map to map framenodeid's to IO thread clients.

BUG=645983

Review-Url: https://codereview.chromium.org/2737663004
Cr-Commit-Position: refs/heads/master@{#456126}

Now that we do direct routing of MouseWheel events to guests,
BrowserPlugin should (1) never mark a MouseWheel as handled (this may
cancel GesturePinch on Mac), and (2) it should never forward the
MouseWheel to the guest. This CL implements this policy.

BUG=695907

Review-Url: https://codereview.chromium.org/2745783002
Cr-Commit-Position: refs/heads/master@{#456125}

Checking hasGetAccountsPermission before calling android.AccountManager.updateCredentials
is irrelevant, because this call never required GET_ACCOUNTS permission.
Added check for MANAGE_ACCOUNTS permission that was required before
Android M. On M+ it doesn't require any special permission.

BUG=699050

Review-Url: https://codereview.chromium.org/2745533002
Cr-Commit-Position: refs/heads/master@{#456124}

In this CL, I just tried to remove the flaky expectations
and rebaseline using `webkit-patch rebaseline-cl` to see
if that works.

BUG=685851

Review-Url: https://codereview.chromium.org/2733453002
Cr-Commit-Position: refs/heads/master@{#456123}

Activate the keychain reauthorization. Reauthorization will be performed
if the users preferences (cfprefs) indicate that the operation has been
attempted less than 2 times previously, and that the operation has not
previously succeeded. The number of attempts is set before the
operation, so that a crashy operation stops running after a finite
number of attempts. After successful completion, the preferences are
updated so that the operation is never run again. This CL also updates
the designated requirement with a new hash.

BUG=629906

Review-Url: https://codereview.chromium.org/2738053007
Cr-Commit-Position: refs/heads/master@{#456122}

There may be existing pointers to the memory block that are being actively
accessed while the zeroing of memory is taking place. While this change
doesn't fix that, it does make it possible for those other threads to recognize
that memory is changing under it and be able to react appropriately.

Also, restore missing kTypeIdAny that was somehow lost.

BUG=620813

Review-Url: https://codereview.chromium.org/2709113003
Cr-Commit-Position: refs/heads/master@{#456121}

Added myself to a new blink_image_capture and
removed/renamed myself from a few categories

Review-Url: https://codereview.chromium.org/2741853004
Cr-Commit-Position: refs/heads/master@{#456120}

This is so that mojo/binding.html can find the generated bindings for
math_calculator.mojom. This layout test was added in r455925, but failed
to add this data dependency.

BUG=700281
TEST=gn desc blink_tests runtime_deps

Review-Url: https://codereview.chromium.org/2742793003
Cr-Commit-Position: refs/heads/master@{#456119}

In general View::RequestFocus() and FocusManager::SetFocusedView() do
the same thing. The only exception to that is the SetFocusedView()
does not check if the view wants focus, where as View::RequestFocus()
does. For the cases I'm changing here RequestFocus() is fine.

BUG=none
TEST=none
R=msw@chromium.org

Review-Url: https://codereview.chromium.org/2738063007
Cr-Commit-Position: refs/heads/master@{#456118}

- Fixed the layout issues with the fullscreen toolbar
- Fixed the hiding animation

BUG=689115

Review-Url: https://codereview.chromium.org/2738043002
Cr-Commit-Position: refs/heads/master@{#456117}

In crrev.com/2354793003, we changed state updates to go through
TextInputManager, but TextInputManager does not propagate state update
when the current and last state was NONE. This can cause IME thread to
hang because it waits for the request to be replied.

BUG=699699

Review-Url: https://codereview.chromium.org/2743933003
Cr-Commit-Position: refs/heads/master@{#456116}

To resolve code review comments from https://codereview.chromium.org/2675033002/diff/280001/chrome/browser/media/router/mojo/media_router.mojom

Create an ip_address.mojom to be used by media_router.mojom

BUG=698423

Review-Url: https://codereview.chromium.org/2745643002
Cr-Commit-Position: refs/heads/master@{#456115}