GitLab

This change eliminates the property that non-singleton services are
owned by the first service that connects to them. Instead, singleton
and non-singleton services are treated identically by the Service
Manager: they go away either (1) when the service requests so or
(2) when the Service Manager itself goes away.

The motivation for this change is that (a) there are situations where
this "tree ownership" model doesn't make sense (see linked bug), and
(b) there is nothing currently in the codebase that is utilizing the
tree ownership model. If desired, the tree ownership model could be
resurrected on a per-service basis by adding a "session manager" attribute
to a service's manifest, which would denote that that service should own
all of the services that it directly or transitively connects to.

This CL also changes Service Manager's destructor to perform
two-phase shutdown on its owned Instances. This change eliminates
possible deadlock in the following situation:
- An Instance A blocks in its destructor for its Runner to complete
- That Runner blocks completion on the corresponding Service A shutting down
- Service A blocks its shutdown on some distinct Service B shutting down
- Service B is waiting to receive a connection error in order to initiate
  shutdown
- ServiceManager never deletes Instance B since it is blocked waiting for
  the destructor of Instance A to return

An example of such a situation is found in connect_test_package.cc:
ConnectTestService clears its ProvidedService instances in its OnStop() method,
while ProvidedService does not exit its destructor until it receives a
connection error. Before this CL, deadlock was not tickled because the
ProvidedServices were always killed when the first service that
connected to them was killed, which happened to be before ConnectTestService
was killed.

BUG=672863

Review-Url: https://codereview.chromium.org/2572803002
Cr-Commit-Position: refs/heads/master@{#438780}

The main change is to use base::LaunchUnitTestsSerially as a test runner
so it can parse some command-line arguments the isolate test runner
passes in.

BUG=663126
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/2521793002
Cr-Commit-Position: refs/heads/master@{#438779}

JumpListIcons(Old) folders are getting unexpectively huge for some users.
This is caused by file operation failures.

After analyzing those two folders' operation metrics by landing several changes,
we find that there are a few issues that need to be fixed.

1) base::CopyDirectory() in file_util_win.cc has incorrect logic when target
dir string contains source dir string. Please see detailed comments in the code,
2) When it fails to delete folder |JumpListIconOld|, we should not move
folder |JumpListIcon| to |JumpListIconOld|, otherwise |JumpListIconOld| can
get unexpectedly huge eventually.
3) When it fails to move folder |JumpListIcon| to |JumpListIconOld|, we
should still delete |JumpListIcon| to avoid endless file accumulation in it.

To fix the first issue, IsParent() function is used in file_util_win.cc to
replace the old logic. I have tested it in a previously checked-in CL.
The gathered user data confirms its correctness. It can be found in line
370 of the old version of jumplist.cc in this CL. In that version, I copied
and changed a few file operation functions from //base for debugging and
testing purposes.

We still log one kind of user data for verification and further analysis.

This CL also reverts several changes I made for debugging purpose.

BUG=40407,179576
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.win:win10_chromium_x64_rel_ng

Review-Url: https://codereview.chromium.org/2570693002
Cr-Commit-Position: refs/heads/master@{#438778}

https://pdfium.googlesource.com/pdfium.git/+log/37b12ad87319..72a22025ec44

$ git log 37b12ad87..72a22025e --date=short --no-merges --format='%ad %ae %s'
2016-12-14 tsepez Remove unused m_OtfFontData.

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, see:
http://www.chromium.org/developers/tree-sheriffs/sheriff-details-chromium#TOC-Failures-due-to-DEPS-rolls

TBR=dsinclair@chromium.org

Review-Url: https://codereview.chromium.org/2575393002
Cr-Commit-Position: refs/heads/master@{#438777}

"Show Google Now cards" option should populate when it is enabled in field trial
and the default search provider is google.

BUG=614439
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2572963004
Cr-Commit-Position: refs/heads/master@{#438776}

This CL adds blendMode field to EffectPaintPropertyNode, upgrades
PaintPropertyTreeBuilder to create effect nodes if CSS mix-blend-mode is
applied to an element and ensure propery isolation on the parent, and adds
corresponding conversion in PaintArtifactCompositor.

BUG=609937
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2564193002
Cr-Commit-Position: refs/heads/master@{#438775}

From video_decode_accelerator.h, it's not clear if it's valid to
call Reset before NotifyFlushDone. Currently ARC may call Reset
before NotifyFlushDone is back. We need to document this and add
a test for it.

BUG=671944
TEST=None

Review-Url: https://codereview.chromium.org/2556873003
Cr-Commit-Position: refs/heads/master@{#438774}

Fixed typo in quick view for no playback available message to have proper margin between the icon.

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

Review-Url: https://codereview.chromium.org/2562023002
Cr-Commit-Position: refs/heads/master@{#438773}

As a part of "Persistence" plan, the UISourceCode.checkContentUpdated
method should move away from UISourceCode.

This CL is a preparation for this. This patch kills a few clients of the
method, since they are not needed any more:
1. SourcesSearchScope shouldn't be forcing checkContentUpdated during search,
   it can simply request content
2. The UISourceCodeFrame shouldn't request content on focus - the iNotify
   reports changes while devtools are blurred.

BUG=673086
R=dgozman

Review-Url: https://codereview.chromium.org/2567283004
Cr-Commit-Position: refs/heads/master@{#438772}

As part of this, starts allocating a transform node to represent direct
compositing of will-change:transform elements and their stacking children.

BUG=668342
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2571043002
Cr-Commit-Position: refs/heads/master@{#438771}

Summary of changes available at:
https://chromium.googlesource.com/v8/v8/+log/5ba6e0d1..027994b8

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_trusty_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/2580723002
Cr-Commit-Position: refs/heads/master@{#438770}

If user opens a Quick View for a folder, show the folder icon
instead of the generic file icon with no preview available
message.

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

Review-Url: https://codereview.chromium.org/2561633006
Cr-Commit-Position: refs/heads/master@{#438769}

BUG=none

Review-Url: https://codereview.chromium.org/2560043005
Cr-Commit-Position: refs/heads/master@{#438768}

We are now animating offset-rotate, hence
CSSOffsetRotationInterpolationType.

BUG=669782

Review-Url: https://codereview.chromium.org/2579673002
Cr-Commit-Position: refs/heads/master@{#438767}

Now all reloads triggered by user interfaces are replaced by
ReloadMainResource, but Location.reload() and History.go() still call
the traditional Reload.

This patch changes them to call ReloadMainResource insteads under a
runtime flag. This flag will be wired to a field study flag later.

This is discussed at loading-dev@ in the following thread.
https://groups.google.com/a/chromium.org/d/topic/loading-dev/gD-MPRcfwVA/discussion

BUG=670237

Review-Url: https://codereview.chromium.org/2555963003
Cr-Commit-Position: refs/heads/master@{#438766}

In this patch, the class does nothing interesting,
but would have a dynamic instance management in
following changes.

Migration would happen per platform behind a
field study.

See crbug.com/672793 for rough design and plan.

BUG=672793

Review-Url: https://codereview.chromium.org/2566673002
Cr-Commit-Position: refs/heads/master@{#438765}

We pause the animation, and progress using animationDelay.

animation-direction-normal.html now uses testharness.js

BUG=674123

Review-Url: https://codereview.chromium.org/2580703002
Cr-Commit-Position: refs/heads/master@{#438764}

For desktop Linux, Chrome OS and Windows only.

Only active behind the
--enable-experimental-web-platform-features flag.

BUG=668389
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

Review-Url: https://codereview.chromium.org/2545323002
Cr-Commit-Position: refs/heads/master@{#438763}

Accroding to HTML spec
https://html.spec.whatwg.org/multipage/semantics.html#the-link-element, if the
href attribute in link element is empty, then the element does not define a link
and should be ignored.

BUG=669351

Review-Url: https://codereview.chromium.org/2575273002
Cr-Commit-Position: refs/heads/master@{#438762}

This patch sets a default color for icon masks. This should be good
enough to solve concerns in https://crrev.com/2570783006 while supporting
dark theme.

BUG=669323
R=dgozman

Review-Url: https://codereview.chromium.org/2572283002
Cr-Commit-Position: refs/heads/master@{#438761}

BUG=667357

Review-Url: https://codereview.chromium.org/2571603003
Cr-Commit-Position: refs/heads/master@{#438760}

Currently, the persistence only syncs working copy changes. However,
one of the valid change of working copy is its resetting - means that
UISourceCode becomes non-dirty again.

This patch starts resetting working copy of UISourceCode if its counterpart
got into a clean state.

BUG=674332
R=dgozman

Review-Url: https://codereview.chromium.org/2578823002
Cr-Commit-Position: refs/heads/master@{#438759}

Some mount points are virtual, and do not store bytes on the local
file system.

TEST=None
BUG=653593

Review-Url: https://codereview.chromium.org/2556413002
Cr-Commit-Position: refs/heads/master@{#438758}

This patch moves the main logic of WebLocalFrameImpl::replaceSelection to Editor,
so that the web/ layer becomes thinner, and less details of editing are exposed.
No behavior is changed.

This patch is also a preparation for https://codereview.chromium.org/2576903002.

BUG=673789

Review-Url: https://codereview.chromium.org/2575343002
Cr-Commit-Position: refs/heads/master@{#438757}

URLResponseBodyConsumer reads data from a mojo data pipe and dispatches it as a
URLResponseBodyConsumer::ReceivedData. When the dispatched data is destructed,
mojo::EndReadDataRaw is called. Until then, mojo::BeginRead will return
MOJO_RESULT_BUSY. This CL implements the correct handling for such a case.

This mis-handling was the cause of the failure of
virtual/mojo-loading/http/tests/inspector/network/network-fetch.html. This CL
also adds text() call in makeFetch function so that the backpressure mechanism
won't stop the loading.

BUG=669357

Review-Url: https://codereview.chromium.org/2573743002
Cr-Commit-Position: refs/heads/master@{#438756}

BUG=n/a

Review-Url: https://codereview.chromium.org/2579713002
Cr-Commit-Position: refs/heads/master@{#438755}

An action item from this ES6 proposal:
https://groups.google.com/a/chromium.org/forum/#!topic/chromium-dev/iJrC4PVSfoU

R=dpranke@chromium.org
BUG=671426

Review-Url: https://codereview.chromium.org/2576253002
Cr-Commit-Position: refs/heads/master@{#438754}

https://pdfium.googlesource.com/pdfium.git/+log/c7739322e5cd..37b12ad87319

$ git log c7739322e..37b12ad87 --date=short --no-merges --format='%ad %ae %s'
2016-12-14 tsepez Return unique_ptr from CXFA_FFApp::CreateDoc.
2016-12-14 tsepez More unique_ptr returns from DIB methods.

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, see:
http://www.chromium.org/developers/tree-sheriffs/sheriff-details-chromium#TOC-Failures-due-to-DEPS-rolls

TBR=dsinclair@chromium.org

Review-Url: https://codereview.chromium.org/2576983002
Cr-Commit-Position: refs/heads/master@{#438753}

Some layout tests in http/tests/inspector/tracing check |encoded_data_length|
which is not dispatched when the resource is cached because no bytes are
received from the network in such a case. To deflake them, this CL adds a
random query to avoid caching.

BUG=None
R=dgozman@chromium.org

Review-Url: https://codereview.chromium.org/2573393002
Cr-Commit-Position: refs/heads/master@{#438752}

It is a test for a Document leak by Document::m_associatedFormControls
[1]. However [2] removed m_associatedFormControls, and now the test isn't
helpful.

[1] https://chromium.googlesource.com/chromium/src/+/ca06c84dc478d4af2aa0b37026da6fa1a679d02e
[2] https://chromium.googlesource.com/chromium/src/+/c7e3c1f85c4691489409104d9ab53d30b52cd8d9

BUG=674194

Review-Url: https://codereview.chromium.org/2572383002
Cr-Commit-Position: refs/heads/master@{#438751}

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/ab7345371ad0..f87ce97ea1fa

$ git log ab7345371..f87ce97ea --date=short --no-merges --format='%ad %ae %s'
2016-12-14 lpy [Trace Viewer] Show runtime statistics in background parsing.

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, see:
http://www.chromium.org/developers/tree-sheriffs/sheriff-details-chromium#TOC-Failures-due-to-DEPS-rolls

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_optional_gpu_tests_rel
TBR=catapult-sheriff@chromium.org

Review-Url: https://codereview.chromium.org/2577993002
Cr-Commit-Position: refs/heads/master@{#438750}

Fix coming up @ https://codereview.chromium.org/2491613004/ but
suppressing race in mean time to make TSAN bot green again
(issue 672860).

BUG=587199
TEST=TimerTest.OneShotTimer_CustomTaskRunner no longer reported as failure by linux_chromium_tsan_rel_ng.
NO_DEPENDENCY_CHECKS=true

Review-Url: https://codereview.chromium.org/2576503002
Cr-Commit-Position: refs/heads/master@{#438749}

The callers can have a registration id even before storage is initialized,
for example if registration ids are stored in preferences.

BUG=672716

Review-Url: https://codereview.chromium.org/2569353003
Cr-Commit-Position: refs/heads/master@{#438748}

This abstraction is not needed any more. All the clients are migrated
over to the Common.FormatterWorkerPool.parseCSS method.

BUG=none
R=dgozman

Review-Url: https://codereview.chromium.org/2572053002
Cr-Commit-Position: refs/heads/master@{#438747}

Made -webkit-box-direction property independent. Also updated
independent-inheritance-fast-path.html to include webkitBoxDirection.

Also updated make_computed_style_base.py to use 'name_for_methods'
instead of 'upper_camel_name' so that Webkit is removed from the front
of the name when generating method names.

Tested using independent-inheritance-fast-path.html.

BUG=628043

Review-Url: https://codereview.chromium.org/2574063002
Cr-Commit-Position: refs/heads/master@{#438746}

class:
(1) Schedule clean up stale navigation footprint
(2) Backtrace navigation events up to 2 user gesture for a given
    download, and add this info into ClientDownloadRequest proto.
(3) Add UMA metrices to track attribution results

BUG=639467

Review-Url: https://codereview.chromium.org/2538483002
Cr-Commit-Position: refs/heads/master@{#438745}

This CL allows user to open files which are browsable on Chrome
on Quick View. Browsable files include pdf and text.
Screencast: https://bugs.chromium.org/p/chromium/issues/detail?id=640696#c11

BUG=642713,640696
TEST=manually tested the following:
- For pdf and text, preview with 24px of vertical margin appears.
- For image, video and audio, preview with 15% of vertical margin
appears as before.
- For .docx, 'no preview available' is shown as expected, though the
file type is 'document' which is the same as pdf.

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2540863004
Cr-Commit-Position: refs/heads/master@{#438744}

We are going to merge MemoryCoordinator into MemoryCoordinatorImpl
as a part of preparations for adding an pure interface of
MemoryCoordinator in base/. Since they are not simple classes,
just merging them will create a big class which has multiple
responsibilities. Before merging them, factor out state updating
code from MemoryCoordinatorImpl.

BUG=673216

Review-Url: https://codereview.chromium.org/2568393002
Cr-Commit-Position: refs/heads/master@{#438743}

This is needed for chrome.

BUG=663525
TEST=none
R=erg@chromium.org

Review-Url: https://codereview.chromium.org/2581503002
Cr-Commit-Position: refs/heads/master@{#438742}

Adds a QuotaSettings datastructure and changes how the total poolsize
for temporary storage is determined. This CL adds a public content API
that the embedder must use to inform the quota system of the desired
QuotaSettings.

The old computation: [available_space + current_temp_usage] / 3
The new computation: [total_volume_size - os_overhead] / 3

The new computation is simplified to be a function of the size of the
volume, and no longer takes as input how much chrome is currently using.
This also reduces the time take to initialize the quota system.

To defend against filling the drive, as available space gets "too low",
the calculated per-host quota is clamped to the hosts current usage. This
prevents the writing of new data because quota clients query
GetUsageAndQuota() prior to performing a write. Also the eviction logic
deletes existing data to keep |must_remain_available| disk space free.

BUG=520318

Review-Url: https://codereview.chromium.org/1782053004
Cr-Commit-Position: refs/heads/master@{#438741}