GitLab

https://codereview.chromium.org/877293003

Review URL: https://codereview.chromium.org/884243008

Cr-Commit-Position: refs/heads/master@{#314403}

Summary of changes available at:
https://chromium.googlesource.com/skia/+log/46c77f7..c9fed24

CQ_EXTRA_TRYBOTS=tryserver.blink:linux_blink_rel,linux_blink_dbg
TBR=stephana@google.com

Review URL: https://codereview.chromium.org/895953002

Cr-Commit-Position: refs/heads/master@{#314402}

UpdateCacheFromPrefsOnPrefThread and UpdatePrefsOnPrefThread.
Factor out key strings into variables.

Review URL: https://codereview.chromium.org/890093002

Cr-Commit-Position: refs/heads/master@{#314401}

This CL:
- Makes PointerDeviceObserver get notified when a mouse or touchpad is
  plugged in or removed by making PointerDeviceObserver an
  InputDeviceEventObserver on ozone only.
- Adds InputDeviceEventObserver::OnMouseDeviceConfigurationChanged() and
       InputDeviceEventObserver::OnTouchpadDeviceConfigurationChanged().
- Adds the plumbing from EventFactoryEvdev to notify
  InputDeviceEventObservers when a mouse or touchpad is plugged in or
  removed.

In a later CL I will make PointerDeviceObserver an InputDeviceEventObserver on
X11 and add the X11 plumbing

BUG=450899
TEST=Manual, see bug

Review URL: https://codereview.chromium.org/889403003

Cr-Commit-Position: refs/heads/master@{#314400}

Upcoming changes to filename_util are unreviewable if they include
material code changes and formatting changes. This CL just includes
formatting changes and is the result of running:

  'clang-format -i --style=Chromium filename_util_unittest.cc'.

Depends on: https://codereview.chromium.org/869823003
Needed by : https://codereview.chromium.org/869233006

BUG=444102

Review URL: https://codereview.chromium.org/873843002

Cr-Commit-Position: refs/heads/master@{#314399}

This patch adds composite event recording for requested rects into the
frame timing tracker.

The current approach is to record all rects that intersect the visible
rect for that layer.

R=danakj, mpb@chromium.org
BUG=441555

Review URL: https://codereview.chromium.org/884243004

Cr-Commit-Position: refs/heads/master@{#314398}

UploadFileElementReader relies on checking the modified time of files
being uploaded to determine if a sliced file was modified during upload.
Clients of the net stack (in particular Blink) currently pass around the
expected modified time in a manner which cause the timestamp to lose
precision (E.g. converting to and from a double time_t).

As a result the expected timestamp and the current timestamp as returned
by GetFileInfo() will not match exactly. Current code attempted to
compensate for this by converting both timestamps to (integer) time_t.
However, since the conversion rounds down, this check would only succeed
if the loss of precision of the expected timestamp also caused it to
round down. This is not always the case. (E.g. (epoch + 10.999999) will
become 10 when converted to time_t, but the expected timestamp may have
rounded up to (epoch + 11.0) in the meantime.)

This patch compares the timestamps by checking if the magnitude of the
difference is less than one second.

BUG=426465
R=mmenke

Review URL: https://codereview.chromium.org/868253012

Cr-Commit-Position: refs/heads/master@{#314397}

This change also:
 - adds x86/x86_64 in the list of supported ABIs.
 - adjusts the regexp of the REST endpoints to support the emulator:
   the old regex was using \w+ to match the device serial, which
   doesn't work with the cases such as "emulator-1234".

TBR=primiano@chromium.org
BUG=

Review URL: https://codereview.chromium.org/887583004

Cr-Commit-Position: refs/heads/master@{#314396}

This adds commit history to named commits in the CommitTracker. The history can be viewed by requesting /_query_commit/<commit-name>.

This also adds a _reset_commit servlet which will reset the current commit ID of a named commit. This servlet will only obey the request if the given commit ID is present in the named commit's history.

BUG=None
R=kalman@chromium.org
NOTRY=true

Review URL: https://codereview.chromium.org/736773002

Cr-Commit-Position: refs/heads/master@{#314395}

* Parcels out logic from the ShortcutHelper that deals with
  Manifest icons and puts it into the ManifestIconSelector.

* Renames the ShortcutHelperTest to ManifestIconSelectorTest
  since that class entirely tested Manifest icon support,
  anyway.

BUG=453170

Review URL: https://codereview.chromium.org/880203004

Cr-Commit-Position: refs/heads/master@{#314394}

(placeholder images inserted, bug 454009 filed)

Also, move CardUnmaskPrompt.java to proper location.

BUG=454009

Review URL: https://codereview.chromium.org/887153003

Cr-Commit-Position: refs/heads/master@{#314393}

This is a follow-up based on TBR review comments made to this change:
https://codereview.chromium.org/873393004

This moves the warning message to a more-central location so that posting
delayed tasks from additional call points (eg., MessageLoopProxy) will be
checked.

BUG=450045

Review URL: https://codereview.chromium.org/878033005

Cr-Commit-Position: refs/heads/master@{#314392}

|IsWiFiConnection| isn't accurate on some platforms. Whenever SSID is available,
use it and ignore |IsWiFiConnection|. This CL also adds a delegate to
|CaptivePortalBlockingPage| for better testing.

BUG=451272

Review URL: https://codereview.chromium.org/894153003

Cr-Commit-Position: refs/heads/master@{#314391}

UseCounter added for setting AudioBufferSourceNode.buffer more than
once.

BUG=448960

Review URL: https://codereview.chromium.org/885183002

Cr-Commit-Position: refs/heads/master@{#314390}

When running /analyze on all of Chrome there were three warnings about
using the comma operator in a tested expression. One was a genuine bug
and the other two were in UncompressAndPatchChromeArchive. Rewriting
to use nested if statements resolves these warnings (getting the count
to zero) and makes the code clearer.

Resolving the warning is not crucial, but it makes it easier to make a
zero-tolerance policy for that warning and thus avoid future bugs.
Plus, it makes the setup code easier to read.

The warnings are:

src\chrome\installer\setup\setup_main.cc(174) : warning C6319: Use of the comma-operator in a tested expression causes the left argument to be ignored when it has no side-effects.
src\chrome\installer\setup\setup_main.cc(176) : warning C6319: Use of the comma-operator in a tested expression causes the left argument to be ignored when it has no side-effects.

The bug that was found through this warning, which could eventually
have been very serious, was fixed here:
https://codereview.chromium.org/678193003

BUG=427616

Review URL: https://codereview.chromium.org/893703004

Cr-Commit-Position: refs/heads/master@{#314389}

The ReadFile API on Windows invoked by the FileStream::Context class which is used by URLRequestFileJob at times completes synchronously.

The FileStream::Context class is used by the URLRequestFileJob class on the IO thread. The intention here is to open
the file for overlapped IO and thus complete the read and writes asynchonously.
Turns out that there are cases where the ReadFile API completes synchronously which ends up blocking the IO thread.
http://support.microsoft.com/kb/156932

Fix for this is to post the ReadFile call to a worker pool and inform the caller about failures etc on the calling thread.

Changes in this patch are as below:-
1. The FileStream::Context::Read function posts the ReadFile call to the Worker thread pool. The entry point for this task is the static
   function ReadAsync.
   This function posts results back to the originating thread via the function ReadAsyncResult

2. Fixed the FileStreamTest.WriteRead test to not read the data in the completion callback of the stream
   Write function. This causes a DCHECK to fire in the MessageLoopForIO class that we are entering a nested
   loop which is a no no for that message pump type.

   This DCHECK fires because of the change on Windows to execute the ReadFile in a Worker thread pool.
   In any case this change is safe for all platforms.
   The written data is validated by reading it back in the function ValidateWrittenData in the TestWriteReadCompletionCallback class.

   We call this explicitly in the FileStreamTest.WriteRead test.

3. Fix the SyncableFileOperationRunnerTest.CopyAndMove test failures.

This test internally initiates a ReadFile call on Windows via the FileStream::Context class for the Copy and Move
test. It invokes Copy and Move and then calls MessageLoop::RunUntilIdle. This call basically runs the message loop
until all pending tasks and events have been processed. This works on Windows by fluke because the ReadFile call
which completes asynchronously posts IO completion tasks to the IO thread (current thread) which then get pulled out
by the RunUntilIdle call.

There could be cases where the ReadFile call does not post IO completion events before RunUntilIdle returns and we could
see the same failures.

Reason we see this with this patch is because ReadFile is now completed by a worker pool thread which basically guarantees
the above scenario. Fix is to use base::RunLoop::Run() and Quit the loop when the DidFinish callback is received.

I changed all places in the syncable_file_operation_runner_unittest.cc file to use base::RunLoop::Run() or
base::RunLoop::RunUntilIdle() as the MessageLoop equivalents are deprecated.

BUG=423948

Review URL: https://codereview.chromium.org/887863002

Cr-Commit-Position: refs/heads/master@{#314388}

- Allow positively scaled transformations. Overlay hardware can handle this
- Allow the STREAM_VIDEO_QUAD type. Tests added for that type

TEST=manual test, video in overlay

Review URL: https://codereview.chromium.org/855403002

Cr-Commit-Position: refs/heads/master@{#314387}

According to test runs on my machine, the test looks like suspended
after the test body finished, which seems to be waiting for
remaining timer tests. ClearSessionLengthLimit() should be called
at the end.

This also has cosmetic changes like NULL -> nullptr

BUG=454601
R=stevenjb@chromium.org
TEST=build with build_for_tool=memcheck, run ash_unittests with
  valgrind, and confirmed the test finishes in realistic time

Review URL: https://codereview.chromium.org/895863002

Cr-Commit-Position: refs/heads/master@{#314386}

ChromeHistoryClient is no longer a HistoryServiceObserver, so remove
the cleanup code that dealt with that.

Fix error path when HistoryService::Init() fails in TestingProfile by:
- not calling SetTestingFactoryAndUse() with a null testing factory
- returning false when HistoryService::Init() fails.

BUG=373326

Review URL: https://codereview.chromium.org/894783002

Cr-Commit-Position: refs/heads/master@{#314385}

LogcatMonitor provides the same functionality as
AndroidCommands.WaitForLogMatch and the associated functions.

BUG=267773

Review URL: https://codereview.chromium.org/896503002

Cr-Commit-Position: refs/heads/master@{#314384}

Also improved error reporting: if an error occurs, and
chrome.runtime.lastError is not checked, then an error is printed to the
console (to be consistent with error reporting in the other async APIs).

The documentation has been updated to mention that these parameters
were required before Chrome 42, these lines should be removed once
Chrome 42 hits stable (https://crbug.com/452990).

BUG=163750
TEST=./out/Debug/interactive_ui_tests --gtest_filter=NotificationsApiTest.TestBasicUsage
TEST=Manually, called chrome.tabs.create/update/clear without callback
     and observed that the notification is created as expected, and that
     an error is printed to the console when an error occurs (e.g.
     invalid iconUrl).
R=kalman@chromium.org

Review URL: https://codereview.chromium.org/855813002

Cr-Commit-Position: refs/heads/master@{#314383}

It's set whenever the mouse is over an element, which just leads
to flakiness.

BUG=454600
NOPRESUBMIT=true

Review URL: https://codereview.chromium.org/896653002

Cr-Commit-Position: refs/heads/master@{#314382}

Handle -G output_dir=foo as well as -Goutput_dir=foo.
I'm not sure how common this case is but we have
NaCl SDK bot that do this.

Getting this wrong results in windows toolchain being
installed in 'out/' (which doesn't necessarily exist)
rather than the output_dir tree and gyp uses.

Add unittests for the affected parts of gyp_chromium.

BUG=454594

Review URL: https://codereview.chromium.org/896663002

Cr-Commit-Position: refs/heads/master@{#314381}

615082d  Remove OpenJPEG header generator binary from PDFium build.
f3823b2  Add embedder test for fix to bug_452455.

Review URL: https://codereview.chromium.org/899593003

Cr-Commit-Position: refs/heads/master@{#314380}

https://codereview.chromium.org/894473002/)

test_opus_api, test_opus_encode, test_opus_decode, and test_opus_padding
can be run directly, with no command-line options, and they print out
"OK" during the test and an overall status message such as
"All API tests passed." or "Tests completed successfully." at the end.

The celt component has several unit tests. I didn't add build targets
for them because some of them don't print any output, so you'd need a
test harness script to examine their exit status.

R=henrika@chromium.org,sergeyu@chromium.org
BUG=452984

Review URL: https://codereview.chromium.org/891353005

Cr-Commit-Position: refs/heads/master@{#314379}

Original issue https://codereview.chromium.org/880063002

TBR=dtseng@chromium.org

Review URL: https://codereview.chromium.org/895623003

Cr-Commit-Position: refs/heads/master@{#314378}

BUG=450382

Review URL: https://codereview.chromium.org/882153005

Cr-Commit-Position: refs/heads/master@{#314377}

Uses ozone GLImage instead of GLImageLinuxDMABuffer and binds it in the
TextureManager, this lets it be scheduled as an overlay

BUG=370522
TEST=loaded on device (with 812913003, 806413004, 855403002, 810343004), play h264 video, verify that overlays are used and no artifacts are produced)

Review URL: https://codereview.chromium.org/858653002

Cr-Commit-Position: refs/heads/master@{#314376}

This reverts commit 02f7bf13 and
thus effectively relands fa906382. That
commit was suspected of breaking the EME tests but reverting it didn't
unbreak anything.

https://codereview.chromium.org/899463002/

Cr-Commit-Position: refs/heads/master@{#314375}

Summary of changes available at:
https://chromium.googlesource.com/skia/+log/4c93a12..46c77f7

CQ_EXTRA_TRYBOTS=tryserver.blink:linux_blink_rel,linux_blink_dbg
TBR=stephana@google.com

Review URL: https://codereview.chromium.org/900573003

Cr-Commit-Position: refs/heads/master@{#314374}

BUG=454554

Review URL: https://codereview.chromium.org/899553002

Cr-Commit-Position: refs/heads/master@{#314373}

BUG=399775,368298

Review URL: https://codereview.chromium.org/864963003

Cr-Commit-Position: refs/heads/master@{#314372}

BUG=431934
TEST=browser tests

Review URL: https://codereview.chromium.org/860273006

Cr-Commit-Position: refs/heads/master@{#314371}

BUG=369661
TEST=tests stay green

Committed: https://crrev.com/66f3ec8e5ee6c3b528b5a6054cdca85ac0b29699
Reverted: https://crrev.com/96545af0f94ec1ac19a57c86806c95834de5d33c

Committed: https://crrev.com/5fe8124cafd9ea996b99cfc88366315aad0c63b2
Reverted: https://crrev.com/b05344954395b9cbb021057269875d931268dfd0

Review URL: https://codereview.chromium.org/839413004

Cr-Commit-Position: refs/heads/master@{#314370}

in the getrealpan flow.

Also, switch to using RealPanClient instead of fake card numbers (meaning we can not demo the UI until getrealpan works).

BUG=451286

Review URL: https://codereview.chromium.org/875183003

Cr-Commit-Position: refs/heads/master@{#314369}

Disable RepostFormWarningTest#testFormResubmissionCancel and RepostFormWarningTest#testFormResubmissionContinue since they're flaky.

BUG=454834
TBR=ppi

Review URL: https://codereview.chromium.org/894793003

Cr-Commit-Position: refs/heads/master@{#314368}

This also removed the code from content, which doesn't belong there because it's for a feature in Chrome.

In order to implement this in a unified way, I needed to create a public interface around ChildThread. A lot of the change is updating callers of ChildThread to use ChildThreadImpl instead.

BUG=448473

Review URL: https://codereview.chromium.org/897563002

Cr-Commit-Position: refs/heads/master@{#314367}

Also remove references to deleted BrowserMainTest.

BUG=139429

Review URL: https://codereview.chromium.org/891883002

Cr-Commit-Position: refs/heads/master@{#314366}

R=ckehoe@chromium.org, timurrrr@chromium.org
BUG=449198

Review URL: https://codereview.chromium.org/894873002

Cr-Commit-Position: refs/heads/master@{#314365}

This patch lets the task queue manager run more than one posted task
per invocation. This helps reduce the overhead of yielding to and from
the main message loop and can speed up cases where tasks are posted
very frequently. One example is indexeddb, where some operations such as
index building can result in 2500 posted tasks/s (Nexus 7).

This patch also adds accounting for the desired run time of the next
pending delayed task. This information is used to break out of a work
batch if a delayed task should be run instead. Doing this avoids adding
extra delay to delayed tasks.

A potential downside of this change is that it can penalize work that
runs on the message loop without going through the task queue manager.
Based on performance tests[1], almost all tasks on the renderer main
thread are already getting executed by the task queue manager, so I
believe this change shouldn't cause a regression.

Note that this version of the patch still uses a batch size of 1 while
we investigate some mac test failures triggered by larger batch sizes.

[1] https://docs.google.com/a/chromium.org/spreadsheets/d/1IJZpBabW1pr4fb2T8BlkleHcOvHYrjvmCx_dLesxfMA/edit#gid=1492760051

BUG=444764,451593,453898

Review URL: https://codereview.chromium.org/845543004

Cr-Commit-Position: refs/heads/master@{#314364}