GitLab

This is a more faithful implementation of more_rbsp_data().

There are two ways this could happen despite being invalid:
  (1) H264Parser does not strip trailing null bytes from NAL units as it
      should. This is generally not a problem because the conversion to
      Annex B won't introduce them either.
  (2) There could be trailing null bytes before the Annex B conversion.
      This isn't valid (at least for PPS NAL units), but the referenced
      bug includes a sample where the PPS in the avcC record includes a
      trailing null byte.

BUG=479229

Review URL: https://codereview.chromium.org/1107593004

Cr-Commit-Position: refs/heads/master@{#381879}

Cr-Commit-Position: refs/heads/master@{#381878}

BUG=595951
TBR=reed@chromium.org

Review URL: https://codereview.chromium.org/1815443003

Cr-Commit-Position: refs/heads/master@{#381877}

For multipart responses, in the first ResourceLoader::didFinishLoadingOnePart(),
(1) the ResourceLoader is moved from |m_loaders| to |m_nonBlockingLoaders|
    in ResourceFetcher::subresourceLoaderFinishedLoadingOnePart(), and
(2) it is removed from |m_nonBlockingLoaders| in
    ResourceFetcher::willTerminateResourceLoader() called in didFinishLoading().
Therefore, if we call window.stop() just after multipart <img>'s onload event,
we cannot stop loading of the multipart response because it is no longer in
|m_loaders| nor |m_nonBlockingLoaders|.

However, in the second didFinishLoadingOnePart(),
(3) In subresourceLoaderFinishedLoadingOnePart(),
    the ResourceLoader is added again to |m_nonBlockingLoaders|, and
(4) We don't call didFinishLoading()/willTerminateResourceLoader().
Therefore, we can stop loading of the image after that by window.stop().

This CL makes the ResourceLoader to be kept in |m_nonBlockingLoaders| after (1)
and to be remove from |m_nonBlockingLoaders| when
ResourceLoader::didFinishLoading() is called.
This CL enables to stop loading of multipart responses at any time, and also
enables https://codereview.chromium.org/1756953002/ to stop dispatching
didFinishLoadingOnePart() for each part without affecting window.stop()
behavior on multipart responses.

BUG=570608

Review URL: https://codereview.chromium.org/1800333003

Cr-Commit-Position: refs/heads/master@{#381876}

Change the original flow:
  invalidateTreeIfNeeded(parentPaintInvalidationState)
  {
    newPaintInvalidationContainer = ...;
    invalidatePaintIfNeeded(parentPaintInvalidationState, newPaintInvalidationContainer);
    PaintInvalidationState childPaintInvalidationState(parentPaintInvalidationState);
    invalidatePaintOfSubtreesIfNeeded(childPaintinvalidationState);
  }

to:
  invalidateTreeIfNeeded(parentPaintInvalidationState)
  {
    newPaintInvalidationState(parentPaintInvalidationState); // Which updates paintInvalidationContainer, but not geometry info for children.
    invalidatePaintIfNeeded(newPaintInvalidationState);
    newPaintInvalidationState.updatePaintOffsetAndClipForChildren();
    invalidatePaintOfSubtreesIfNeeded(newPaintInvalidationState);
  }

This is like how PaintPropertyTreeBuilderContexts are inherited and updated:
copy from parent first, then update in multiple steps. We may use the state
for different purposes after different steps.

The benefits are:
- Avoid additional paintInvalidationContainer in invalidatePaintIfNeeded()
  by letting paintInvalidationState parameter contain the correct
  paintInvalidationContainer for the current object;
- We move the logic to find the paintInvalidationContainer from LayoutObjects
  into PaintInvalidationState which knows more information;
- This is a preparation for the next step works to avoid some
  ForcedHorriblySlowRectMappings because we need to track correct
  paintInvalidationContainer (for normal objects and stacked objects) in
  PaintInvalidationState, instead of being passed in from
  LayoutBoxModelObject::invalidateTreeIfNeeded();

BUG=591199

Review URL: https://codereview.chromium.org/1804963005

Cr-Commit-Position: refs/heads/master@{#381875}

BUG=590825
CQ_INCLUDE_TRYBOTS=tryserver.chromium.win:win_optional_gpu_tests_rel;tryserver.chromium.mac:mac_optional_gpu_tests_rel

Review URL: https://codereview.chromium.org/1805133002

Cr-Commit-Position: refs/heads/master@{#381874}

PAGE_TRANSITION_AUTO_TOPLEVEL is used to indicate a wide variety of
transitions. For example, New page transitions
(https://codereview.chromium.org/1397673004), EmbedContentViewActivity.

BUG=595909

Review URL: https://codereview.chromium.org/1811173002

Cr-Commit-Position: refs/heads/master@{#381873}

and move the 'masters' section to the top of the file, since that seems
like it's probably the more common starting point.

R=estaab@chromium.org

Review URL: https://codereview.chromium.org/1814953002

Cr-Commit-Position: refs/heads/master@{#381872}

Mus server might prevent some windows from becoming system modals. In
this case, SetModal would fail and the client needs to handle the
failure and revert the modality set locally for the window.

BUG=548402

Review URL: https://codereview.chromium.org/1811913004

Cr-Commit-Position: refs/heads/master@{#381871}

The DrMemory update process only runs when it thinks that it needs to
but its heuristic is not accurate. A better thing to do is to always
update.

This change also adds setlocal so that environment variables don't
leak outside. This may also help with making this script more robust
in the future, once it has been in place for long enough for all
machines affected by the environment variable leakage to have rebooted.

BUG=595867
R=scottmg@chromium.org

Review URL: https://codereview.chromium.org/1807013004 .

Cr-Commit-Position: refs/heads/master@{#381870}

BUG=584819

Review URL: https://codereview.chromium.org/1808643006

Cr-Commit-Position: refs/heads/master@{#381869}

On Windows 10 in particular, active and inactive windows may have very different frame colors, and the incognito icon for one may be invisible on the other.

We should probably also consider toggling the TOOLBAR_TOP_SEPARATOR_COLOR.  That also makes me wonder if this is the right way to do this; perhaps both these should have _INACTIVE colors and whatever draws the frame in the correct color should draw these as well.

It's also not clear to me whether the nonclient view really needs the widget destruction overload (as noted in a comment).

BUG=none
TEST=Run Chrome on Windows 10 with window titlebars set to a dark color.  Open an incognito window.  Toggle the window between active and inactive and check that the incognito icon toggles between white and grey.

Review URL: https://codereview.chromium.org/1733293002

Cr-Commit-Position: refs/heads/master@{#381868}

BUG=595789

Review URL: https://codereview.chromium.org/1809253002

Cr-Commit-Position: refs/heads/master@{#381867}

BUG=298951

Review URL: https://codereview.chromium.org/1815573002

Cr-Commit-Position: refs/heads/master@{#381866}

This patch moves the routines for consuming <color> and <position> out
of CSSPropertyParser.cpp and into CSSPropertyParserHelpers. We expect
that we'll require these routines for various Houdini APIs.

BUG=499780

Review URL: https://codereview.chromium.org/1802673003

Cr-Commit-Position: refs/heads/master@{#381865}

Make parsing of <transform> more strict by checking
that something valid follows after a comma.

BUG=593321

Review URL: https://codereview.chromium.org/1811953002

Cr-Commit-Position: refs/heads/master@{#381864}

BUG=

Review URL: https://codereview.chromium.org/1811133002

Cr-Commit-Position: refs/heads/master@{#381863}

This feature will (hopefully) improve the performance of Syzyasan builds.

BUG=

Review URL: https://codereview.chromium.org/1671063002

Cr-Commit-Position: refs/heads/master@{#381862}

MojoAudioDecoder is the client side part of the decoder
that runs in the renderer process.

This CL connects it to the remote service and propagates
the media task runner to it.

BUG=542910

Review URL: https://codereview.chromium.org/1797393007

Cr-Commit-Position: refs/heads/master@{#381861}

The settings breadcrumb feature is being removed.
That location on the title bar will become the search
within settings.  This CL removes the breadcrumb
code.

BUG=589630

Review URL: https://codereview.chromium.org/1815523002

Cr-Commit-Position: refs/heads/master@{#381860}

We use fixed color format (kN32_SkColorType) to decode images. Since the default
preferred format is 16-bit color on low-end devices, we can reduce memory
footprint by storing decoded image in 16-bit color format when we use the
default settings for the preferred format. This CL makes
SoftwareImageDecodeController allocate memory based on the preferred format.

I ran memory.blink_memory_mobile 20 times. The results can be found at [1].
I removed SimulateMemoryPressureNotification() so that I can observe memory usage in
common patterns (still, memory pressure notifications would be sent if memory
usage is high). According to the results, we could reduce ~40% discardable
memory on average.

Note that Blink still uses fixed 32-bit color for decoding. This means that
skia/cc will convert decoded images when we store decoded image in 16-bit color.
We might want to make Blink's decoding format configurable in the future.

[1] https://drive.google.com/file/d/0B6NYyLPujP4TaWItc3lsZXR6TTg/view?usp=sharing

BUG=519146
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel

Review URL: https://codereview.chromium.org/1808633002

Cr-Commit-Position: refs/heads/master@{#381859}

BUG=593910

Review URL: https://codereview.chromium.org/1809283002

Cr-Commit-Position: refs/heads/master@{#381858}

This contains the last fixes necessary for VS2015 compatibility.

TBR=siggi@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1811943004

Cr-Commit-Position: refs/heads/master@{#381857}

Rather than keeping it around forever, we should kill it. Now that the user shell is acting as the login broker for apps run for that user, it's not necessary to keep login too. This simplifies the login app a bit.

BUG=

Review URL: https://codereview.chromium.org/1806203002

Cr-Commit-Position: refs/heads/master@{#381856}

BUG=589194
TEST=Put too many icons on shelf, move shelf to left, click chevron to
open overflow bubble.  Observe that bubble doesn't overlap shelf.

Review URL: https://codereview.chromium.org/1809963003

Cr-Commit-Position: refs/heads/master@{#381855}

This CL extends the ChromeAnimation class to notify
Animatable interfaces that a particular property
animation has finished.

BUG=

Review URL: https://codereview.chromium.org/1801363006

Cr-Commit-Position: refs/heads/master@{#381854}

Review URL: https://codereview.chromium.org/1807093002

Cr-Commit-Position: refs/heads/master@{#381853}

On OSX, NodeChannel::ShutDown() calls into
NodeController::GetMachPortRelay() which acquires |parent_lock_| and
deadlocks due to a recursuve lock acquisition.

BUG=582468

Review URL: https://codereview.chromium.org/1808233003

Cr-Commit-Position: refs/heads/master@{#381852}

This change better handles the non-user-triggered focus event that happens when the dialog loads. The behavior only appears on Mac and when the dialog initially is opened; however, it appears 100% of the time in these cases.

This adds an element with the sole purpose of triggering focus on that element when the dialog is open rather than any of the other header elements. This prevents messing with the tabindex (whether elements are focusable and in what priority) of the other elements. The tabindex of this element is then updated to be unfocusable so that the user cannot accidentally tab to the element.

BUG=594951

Review URL: https://codereview.chromium.org/1807803002

Cr-Commit-Position: refs/heads/master@{#381851}

A performance optimization. This avoids any heap allocations.

BUG=595584
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel

Review URL: https://codereview.chromium.org/1806223002

Cr-Commit-Position: refs/heads/master@{#381850}

Avoid most window->shelf lookup calls; misc cleanup.
Remove some ChromeLauncherController auto-hide functions.
(inline pref lookup function call, context menu ops)

Add Shell::OnShelfAlignmentChanged to revise set pattern.
(support ShelfLayoutManager::SetAlignment, w/o Shell::SetShelfAlignment)

Prerequisite for https://codereview.chromium.org/1760743002

BUG=557406
TEST=No behavior changes or regressions.
R=sky@chromium.org

Review URL: https://codereview.chromium.org/1812013003

Cr-Commit-Position: refs/heads/master@{#381849}

BUG=547073

Review URL: https://codereview.chromium.org/1807703002

Cr-Commit-Position: refs/heads/master@{#381848}

I expect to add 2 more documents in the follow up:

- clusterfuzz-libfuzzer integration documentation (build bots, corpus, status links, reports)
- reference (fuzzer_test reference, fuzzer options, dictionaries, etc.)

BUG=539572

Review URL: https://codereview.chromium.org/1809843002

Cr-Commit-Position: refs/heads/master@{#381847}

BUG=357747

Review URL: https://codereview.chromium.org/1807893002

Cr-Commit-Position: refs/heads/master@{#381846}

Adds a browser proxy object and implements the setup flow. Tests in a
separate CL.

BUG=563721

Review URL: https://codereview.chromium.org/1806123002

Cr-Commit-Position: refs/heads/master@{#381845}

In r380416, we switched to one accessibility tree per frame, whether site
isolation is on or not. On Android we only want the main frame's
BrowserAccessibilityManager to be connected to the ContentViewCore, not
any child frames.

I modified RWHVAndroid::CreateBrowserAccessibilityManager in that change so
that it wouldn't associate child frames with ContentViewCore, but I got my
logic flipped (in a way that made tests pass, but didn't work in practice).

In fixing it here, I realized that it may not be safe to assume that the
main frame is the first one to create a BrowserAccessibilityManager.
A more robust fix is for RFHI to pass a flag indicating whether it wants
a BAM for the main frame or for another frame.

In the future we may want to use that flag on other platforms, too - like
on Mac really only the main frame's BAM should have the cocoa_view_, but
I'll do that in a follow-up.

BUG=595150
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_site_isolation

Review URL: https://codereview.chromium.org/1809793002

Cr-Commit-Position: refs/heads/master@{#381844}

The first time IOSurface 2D Canvas was enabled, there were no test failures over
several days. The second time it was enabled (along with a bug fix, and some new
tests), trace_test.IOSurface2DCanvas started flakily failing.

BUG=595754

Review URL: https://codereview.chromium.org/1813073002

Cr-Commit-Position: refs/heads/master@{#381843}

BUG=NONE
TEST=NONE
R=sky@chromium.org

Review URL: https://codereview.chromium.org/1811203002

Cr-Commit-Position: refs/heads/master@{#381842}

We want to get rid of mojom::EventPtr and use ui::Event instead at some point.
This CL removes mojom::EventPtr from one more layer in mus+ash.

BUG=578206

Review URL: https://codereview.chromium.org/1814533002

Cr-Commit-Position: refs/heads/master@{#381841}

In VS 2015 builds every call to realloc triggers a warning like this:

Error #1: INVALID HEAP ARGUMENT: allocated with Windows API layer, queried with C library layer
  0 replace_malloc_usable_size
  1 _recalloc_base
  2 <lambda_4e60a939b0d047cfe11ddc22648dfba9>::operator()
  3 __crt_seh_guarded_call<>::operator()<>
  4 __acrt_lock_and_call<>
  5 _register_onexit_function
  6 _crt_atexit
  7 _onexit
  8 atexit
  9 testing::`dynamic initializer for 'FLAGS_gmock_verbose''

This mismatch is harmless (because the CRT always uses the Windows heap)
and out of our control. Therefore it should be suppressed.

With this suppression I now get a clean run on this previously failing
test:

tools\valgrind\chrome_tests.bat -t content_browsertests --tool \
    drmemory_light --build-dir=out\Release \
    --gtest_filter=ManifestBrowserTest.DummyManifest

BUG=440500,594808

Review URL: https://codereview.chromium.org/1813063002

Cr-Commit-Position: refs/heads/master@{#381840}