GitLab

Summary of changes available at:
https://chromium.googlesource.com/v8/v8/+log/3721449f..8358e322

Please follow these instructions for assigning/CC'ing issues:
https://github.com/v8/v8/wiki/Triaging%20issues

Please close rolling in case of a roll revert:
https://v8-roll.appspot.com/
This only works with a Google account.

CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_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

TBR=hablich@chromium.org,machenbach@chromium.org,littledan@chromium.org,vogelheim@chromium.org

Review-Url: https://codereview.chromium.org/2364823002
Cr-Commit-Position: refs/heads/master@{#420603}

This patch also removes the Editor::firstRectForRange(const Range*)
overload because it has no call site.

BUG=647219

Review-Url: https://codereview.chromium.org/2352763004
Cr-Commit-Position: refs/heads/master@{#420602}

The introduced 'emplaceAppend()' method provides more efficient code path when a new object instance requiring multiple constructor arguments is appended to a vector.

Note: the existing 'append()' method is already implicitly emplacing new items for a single constructor argument case.
Review-Url: https://codereview.chromium.org/2365533003
Cr-Commit-Position: refs/heads/master@{#420601}

This CL invalidates the serialized page state any chrome:// URLs,
causing the page to get reloaded rather than restored. This will prevent
the upgrade to MD History from breaking things since it lives at a
different physical URL than classic history.

BUG=614758
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2355543003
Cr-Commit-Position: refs/heads/master@{#420600}

The CacheCounter class in chrome/browser/browsing_data could not be
componentized (unlike other counters in components/browsing_data) because
of its usage of StoragePartitionHttpCacheDataRemover that has a dependency
on content/.

This CL implements a separate CacheCounter class for iOS with the same
functionality. The core of the implementation, a series of IO thread
callbacks, is delegated to the private IOThreadCacheCounter class that
mimics StoragePartitionHttpCacheDataRemover.

BUG=620317

Review-Url: https://codereview.chromium.org/2354643002
Cr-Commit-Position: refs/heads/master@{#420599}

This CL is connected to Issue 2312093003, which generates bindings for
IDL callback functions. This CL exteneded implementation to be able to
use not only string or long but also interface as arguments of callback
functions.

BUG=569301

Review-Url: https://codereview.chromium.org/2367543004
Cr-Commit-Position: refs/heads/master@{#420598}

No change in behavior.

BUG=631960
TBR=piman@chromium.org

Review-Url: https://codereview.chromium.org/2352563002
Cr-Commit-Position: refs/heads/master@{#420597}

We report deprecation warnings when the old
motion* property names are used.

Discussed in the offset-path Intent To Ship,
https://groups.google.com/a/chromium.org/d/msg/blink-dev/o1C5NzGf9Q0/sbaLF1MbAgAJ

BUG=637543

Review-Url: https://codereview.chromium.org/2366493002
Cr-Commit-Position: refs/heads/master@{#420596}

Layout inflation on the UI thread triggers StrictMode violations on some
devices. Temporarily allow disk reads, as the inflation should happen on
the UI thread, and the reads are inside the framework's code.

BUG=644243

Review-Url: https://codereview.chromium.org/2355263002
Cr-Commit-Position: refs/heads/master@{#420595}

Review-Url: https://codereview.chromium.org/2358303003
Cr-Commit-Position: refs/heads/master@{#420594}

This is part 1 of removing non-material top chrome support.

BUG=649368

Review-Url: https://codereview.chromium.org/2361153002
Cr-Commit-Position: refs/heads/master@{#420593}

This CL unifies worker thread shutdown sequences to fix a crash and simplify
shutdown.

<Problem>

When termination is requested on the main thread before WorkerThread is
initialized on the worker thread, shutdown sequence runs in a different way from
regular shutdown sequence: initialization sequence seamlessly switches to
shutdown sequence and asks the main thread to destroy WorkerThread
(see WorkerThread::initializeOnWorkerThread).

This causes a crash in a following scenario:

  1) Request to start the worker thread from the main thread.
  2) Post a task to the worker thread from the main thread.
  3) Request to terminate the worker thread from the main thread.
  4) Start initialziation sequence and switch to shutdown sequence on the worker
     thread.
  5) WorkerThread is destroyed on the main thread.
  6) The posted task runs on the worker thread and crashes.

<Solution>

This CL makes the initialization sequence complete regardless of termination
request and defer to the regular shutdown task posted from the main thread.
Other tasks also posted from the main thread(*) are guaranteed to be drained
until the shutdown task runs.

<Appendix>

Regarding (*), you might wonder if tasks posted from/to the worker thread cannot
be guaranteed to be drained until the shutdown tasks run. This case is covered
by other mechanism. Let's consider it in following 2 cases:

  1) An uninitialized worker thread does not post a task to itself, so there
     should be no tasks when termination happens before initialization.
  2) Otherwise, WorkerBackingThread::shutdown drains all tasks.

Therefore, we can ensure that tasks posted from the worker thread also never run
after shutdown.

BUG=632810, 640843

Review-Url: https://codereview.chromium.org/2280523002
Cr-Commit-Position: refs/heads/master@{#420592}

FontFaceSet::document() gets Document from getExecutionContext(),
so the returned value should be checked after r419951.

BUG=610176,649272
NOTRY=true

Review-Url: https://codereview.chromium.org/2364693005
Cr-Commit-Position: refs/heads/master@{#420591}

app-route is a fairly heavy-weight element, supporting complex routing
patterns designed for large pages. This is unnecessary for MD History,
which currently has one routing pattern and one query parameter.

This CL replaces app-route with a custom router, <history-router>, which
is responsible for two-way binding between the page state and the page
URL. This also refactors handling of search terms elsewhere in the app
to ensure that changes to the search term are reflected correctly across
the page.

This CL has a small positive impact on load time (~20ms on a Z620), code
health (all routing is handled in one place) and binary size (app-route
can be deleted from the binary).

BUG=629710
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2352293002
Cr-Commit-Position: refs/heads/master@{#420590}

BUG=614320

Review-Url: https://codereview.chromium.org/2360723002
Cr-Commit-Position: refs/heads/master@{#420589}

Bluetooth needs to use document->frame() in ContextLifecycleObesrver::contextDestroyed().
Hence this CL moves m_frame=nullptr to after dispatching contextDestroyed().

See https://groups.google.com/a/chromium.org/d/topic/blink-dev/sfpRQc4LB_I/discussion for more motivation of this CL.

BUG=610176

Review-Url: https://codereview.chromium.org/2362173002
Cr-Commit-Position: refs/heads/master@{#420588}

BUG=624095

Review-Url: https://codereview.chromium.org/2353333005
Cr-Commit-Position: refs/heads/master@{#420587}

We applied OPTION's style incorrectly.

BUG=649320

Review-Url: https://codereview.chromium.org/2361113003
Cr-Commit-Position: refs/heads/master@{#420586}

This CL make sure the downloads of the entries start with the delay associated
with the entry and retry to download them if the failure is recoverable.

BUG=629771

Review-Url: https://codereview.chromium.org/2351113003
Cr-Commit-Position: refs/heads/master@{#420585}

The original DCHECK was asserting on float equality. This was too strict
and was firing for small negligible differences. This change flattens
specified Lengths into a single float and compares them with a constant
slack allowance.

This check is intended to pick up on differences like -5 vs 0 due to
different clamping behaviour, small differences in floating point
arithmetic is fine.

BUG=649525

Review-Url: https://codereview.chromium.org/2366823003
Cr-Commit-Position: refs/heads/master@{#420584}

This was an abstraction so that tests could substitute in a fake
version of a Mojo interface in tests. However, the test can just
implement a testing version of the Mojo interface and use it
directly, simplifying the code.

BUG=none

Review-Url: https://codereview.chromium.org/2345423002
Cr-Commit-Position: refs/heads/master@{#420583}

BUG=148757

Review-Url: https://codereview.chromium.org/2361253002
Cr-Commit-Position: refs/heads/master@{#420582}

offset-position and offset-anchor are both animatable as
<position>.
https://drafts.fxtf.org/motion-1/#offset-position-property
https://drafts.fxtf.org/motion-1/#offset-anchor-property

BUG=638055

Review-Url: https://codereview.chromium.org/2364503002
Cr-Commit-Position: refs/heads/master@{#420581}

The purpose of this CL is to make the code a little simpler and easier to read.

Review-Url: https://codereview.chromium.org/2367573002
Cr-Commit-Position: refs/heads/master@{#420580}

This patch fixes var() references to registered properties to return
initial values where appropriate. When a registered property is not
explicitly set, var() references to it should result in that property's
initial value. We now store the initial value as a token stream as well
to accommodate this.

Any unregistered property in a var() cycle will continue to compute to
their initial value (i.e. the invalid value). Any registered property
in a var() cycle similar compute to their initial value (I don't think
this is explicitly mentioned in the spec, but for consistency this makes
more sense than the unset value). Valid references to properties in a
var() cycle will resolve to the referenced property'd initial value.

BUG=641877

Review-Url: https://codereview.chromium.org/2358203003
Cr-Commit-Position: refs/heads/master@{#420579}

 * The hash function is not resilient to hash flooding (but neither is the other
   hash functions we already use for e.g. unordered_containers).

 * Add std::hash and WTF::DefaultHash/HashTraits traits for {Inlined}StructPtr.

 * Modify the generated Struct/UnionTraits so they generate const accessors to
   the struct/union fields where possible (i.e. when there are not handle
   fields).

Review-Url: https://codereview.chromium.org/2339413004
Cr-Commit-Position: refs/heads/master@{#420578}

Slimming paint v2 (SPV2) initially had root property tree nodes on the
FrameView. This approach required some contortions when the root-layer-
scrolls codepath was added because LayoutView needed to store the root
nodes.

This patch adds static root paint property tree nodes but only uses
them on FrameView. A followup patch will switch the root-layer-scrolls
codepath to use this approach as well.

Because the root nodes can no longer be looked up off the FrameView,
a cache of the entire property tree state has been added to FrameView
(totalPropertyTreeStateForContents). This is similar to
LocalBorderBoxProperties on ObjectPaintProperties, and a followup patch
will refactor LocalBorderBoxProperties to use PropertyTreeState as well.

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2
BUG=645615

Review-Url: https://codereview.chromium.org/2359063002
Cr-Commit-Position: refs/heads/master@{#420577}

Review-Url: https://codereview.chromium.org/2364543003
Cr-Commit-Position: refs/heads/master@{#420576}

CopyTexImage2D has bug for cube map texture target on Intel Mac, if the
CUBE_MAP_POSITIVE_X texture is not allocated yet.

This CL works around this bug by doing TexImage2D to allocate the CUBE_MAP_POSITIVE_X
texture before CopyTexImage2D if target is one face of a cube map texture.

BUG=648197
TEST=deqp/functional/gles3/texturespecification/basic_copyteximage2d.html
CQ_INCLUDE_TRYBOTS=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/2348243002
Cr-Commit-Position: refs/heads/master@{#420575}

BUG=232040

Review-Url: https://codereview.chromium.org/2351373003
Cr-Commit-Position: refs/heads/master@{#420574}

Update createContextualFragment behavior to latest spec listed under [1].
This means specifically we do not throw an exception when the context node is
not HTML or SVG and if initially the context element is null, we follow step 2
to create an appropriate context element to be used in step 3.

Behavior matches Firefox and Safari.

BUG=616760

[1] https://w3c.github.io/DOM-Parsing/#extensions-to-the-range-interface

Review-Url: https://codereview.chromium.org/2356373002
Cr-Commit-Position: refs/heads/master@{#420573}

There was previously a typo in this line. We should pick the high-resolution
icon when scale >= 1.5f and the low-resolution icon otherwise.

BUG=648998
BUG=b/31596656
TEST=samus verify icon is high resolution in app list before arc opt-in
TBR=xiyuan

Review-Url: https://codereview.chromium.org/2360873004
Cr-Commit-Position: refs/heads/master@{#420572}

The approach is very simple: if the PaintChunk.id's match, then reuse the layer
backing a chunk, and rely on the raster invalidations to invalidate as needed if
content changes or moves relative to the PaintChunk.

BUG=647831
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2345233004
Cr-Commit-Position: refs/heads/master@{#420571}

Skia actually renders to a larger radius than just blur().

BUG=648963,624175

Review-Url: https://codereview.chromium.org/2356213004
Cr-Commit-Position: refs/heads/master@{#420570}

BUG=649539

Review-Url: https://codereview.chromium.org/2358083004
Cr-Commit-Position: refs/heads/master@{#420569}

https://pdfium.googlesource.com/pdfium.git/+log/39ee9dfac1c6..917d24852841

$ git log 39ee9dfac..917d24852 --date=short --no-merges --format='%ad %ae %s'
2016-09-22 tsepez Make CPDF_ColorSpace::Load() return a unique_ptr

TBR=dsinclair@chromium.org

Review-Url: https://codereview.chromium.org/2361333002
Cr-Commit-Position: refs/heads/master@{#420568}

https://boringssl.googlesource.com/boringssl/+log/e34bcc91c07c0bf65ecc53a814d51f5246007150..bd672ae8c7494732be9cf71b9bfdd431580204f1

BUG=none
TBR=svaldez@chromium.org

Review-Url: https://codereview.chromium.org/2368473002
Cr-Commit-Position: refs/heads/master@{#420567}

ChannelMojo between Service process <-> service utility process is
currently disabled because:
  -- Service process starts service utility process with a command line
     containing kMojoChannelToken switch but no kMojoApplicationChannelToken,
     because they are not in ServiceManager world now.
  -- But, service utility process(ChildThreadImpl) will disable ChannelMojo
     because of lack of kMojoApplicationChannelToken switch.

This CL lets ChildThreadImpl enable ChannelMojo also for a kMojoChannelToken
switch.

BUG=646833
TBR=ben@chromium.org

Review-Url: https://codereview.chromium.org/2366443004
Cr-Commit-Position: refs/heads/master@{#420566}

This lets the layer view to be reused without bringing the entire Timeline
panel as a dependency. Also, made all component widgets of the layer viewer
to be web components.

Review-Url: https://codereview.chromium.org/2358253002
Cr-Commit-Position: refs/heads/master@{#420565}

This is the same dialog as shown when external links are opened from
incognito, informing the user that they are leaving incognito mode and
potentially sharing data with an external application.

BUG=645007

Review-Url: https://codereview.chromium.org/2354833002
Cr-Commit-Position: refs/heads/master@{#420564}