GitLab

The cursor is special since movements don't need to be routed to
the UI thread, instead they are directly dispatched to the GPU
process from the IO-thread. During the recovery of a crashed GPU
process, the windowing state needs to be re-created on the new GPU. As
such we need to make sure that the windows are re-created on the GPU
before the cursor can send targeted events (to a specific window).

BUG=468485

Review URL: https://codereview.chromium.org/1024083002

Cr-Commit-Position: refs/heads/master@{#321645}

This fixes an issue found by -fsanitize=cfi-derived-cast caused by code that
casts a WDResult<T> pointer to WDDestroyableResult<T>.

BUG=457523
R=caitkp@chromium.org

Review URL: https://codereview.chromium.org/1021043002

Cr-Commit-Position: refs/heads/master@{#321644}

This removes the ability to save and replay browsing sessions to/from the cache.

BUG=469256

Review URL: https://codereview.chromium.org/1008723006

Cr-Commit-Position: refs/heads/master@{#321643}

Record the time to update top sites through db thread to understand if this is low and how slow it is in the wild.

BUG=223430

Review URL: https://codereview.chromium.org/1018713002

Cr-Commit-Position: refs/heads/master@{#321642}

Summary of changes available at:
https://chromium.googlesource.com/chromium/blink/+log/0be08ff..8dec3386

TBR=yurys@chromium.org,mnissler@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1020383002

Cr-Commit-Position: refs/heads/master@{#321641}

DisplayConfigurator::OnConfigure may get empty framebuffer size when the framebuffer size didn't change. Don't update in that case.

BUG=457395
TEST=covered by unittests

Review URL: https://codereview.chromium.org/1024943003

Cr-Commit-Position: refs/heads/master@{#321640}

This reverts commit a74f4d24.

Reason for revert: Win8 GN failure.
TBR=brettw@chromium.org

Review URL: https://codereview.chromium.org/1020303006

Cr-Commit-Position: refs/heads/master@{#321639}

[Icons NTP] Allow chrome-search:// large-icon and fallback-icon hosts to use <view_id>/<restricted_id> instead of URL.

For chrome-search://thumbnail on NTP we don't allow direct URL queries,
but only allow <view_id>/<restricted_id>. We're adding the same features
to chrome-search://big-icon and chrome-search://fallback-icon . For the
latter we can actually enable both direct URL and ID's (since URL is only
parsed to extract first letter in domain name) -- but we don't need this
feature right now.

BUG=467712

Review URL: https://codereview.chromium.org/1017853002

Cr-Commit-Position: refs/heads/master@{#321638}

BUG=386810

Review URL: https://codereview.chromium.org/1023253002

Cr-Commit-Position: refs/heads/master@{#321637}

Review URL: https://codereview.chromium.org/1021153003

Cr-Commit-Position: refs/heads/master@{#321636}

TEST= none
BUG= none

Review URL: https://codereview.chromium.org/1027883002

Cr-Commit-Position: refs/heads/master@{#321635}

BUG=none

Review URL: https://codereview.chromium.org/1020013003

Cr-Commit-Position: refs/heads/master@{#321634}

BUG=468919, 468922
R=cpu,jschuh@chromium.org,isherman@chromium.org

Review URL: https://codereview.chromium.org/1024463005

Cr-Commit-Position: refs/heads/master@{#321633}

My last roll missed to update Android Support Library. This CL
rolls from 21.0.3 to 22.0.0

https://chromium-review.googlesource.com/#/c/261413/

BUG=440601,469201

Review URL: https://codereview.chromium.org/1022173002

Cr-Commit-Position: refs/heads/master@{#321632}

Revert of Add android N5 bot as 10% experimental on CQ. (patchset #3 id:40001 of https://codereview.chromium.org/980413003/)

Reason for revert:
Experimental bot did it's job. Now it's time to return this bot to the CQ fleet. Here is the Revert CL that will make the android_n5 bot back to a regular CQ bot: https://codereview.chromium.org/1025673003/

Original issue's description:
> Add android N5 bot as 10% experimental on CQ.
>
> BUG=460553
>
> Committed: https://crrev.com/230ec392dae3136340a43236d6a6b3f0afd2d838
> Cr-Commit-Position: refs/heads/master@{#320133}

TBR=phajdan@google.com,friedman@chromium.org,jbudorick@chromium.org,sergiyb@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=460553

Review URL: https://codereview.chromium.org/1028523004

Cr-Commit-Position: refs/heads/master@{#321631}

I had two cases flipped.
BUG=463658

Review URL: https://codereview.chromium.org/1027843003

Cr-Commit-Position: refs/heads/master@{#321630}

Combine the two api functions developerPrivate.inspect and
developerPrivate.openDevTools, and provide a custom binding wrapper to
maintain compatability with developerPrivate.inspect.

BUG=461039

Review URL: https://codereview.chromium.org/1018493002

Cr-Commit-Position: refs/heads/master@{#321629}

R=isherman@chromium.org

Review URL: https://codereview.chromium.org/1027643002

Cr-Commit-Position: refs/heads/master@{#321628}

Tracked the failure down to this function.   Looks like a simple typo,
test passes now.

BUG=internal b/19821810

Review URL: https://codereview.chromium.org/1020633002

Cr-Commit-Position: refs/heads/master@{#321627}

BUG=

Review URL: https://codereview.chromium.org/1026813003

Cr-Commit-Position: refs/heads/master@{#321626}

This allows the scheduler to evolve to a state where it's able to
perform multiple actions of the same type in a given frame. One example
of this is webview where we can stop receiving BeginFrames but still
have to draw any number of times the system asks us to.

BUG=439275

Review URL: https://codereview.chromium.org/1002493006

Cr-Commit-Position: refs/heads/master@{#321625}

This removes the plumbing in //content to pass a custom cookie store to a
request. This reverts (a small portion of)
https://codereview.chromium.org/188693003

BUG=457344

Review URL: https://codereview.chromium.org/1002363002

Cr-Commit-Position: refs/heads/master@{#321624}

BUG=469224

Review URL: https://codereview.chromium.org/1028503005

Cr-Commit-Position: refs/heads/master@{#321623}

* Rename SetAlternateProtocol, WasAlternateProtocolRecentlyBroken,
  ConfirmAlternateProtocol, ClearAlternateProtocol methods to
  *AlternativeService*.
* Rename SetBrokenAlternateProtocol to MarkAlternativeServiceBroken.
* Make MarkAlternativeServiceBroken, WasAlternativeServiceRecentlyBroken, and
  ConfirmAlternativeService take |alternative_service| instead of |origin|.
* Make SetAlternativeService take |alternative_service| instead of separate
  |protocol|, |port| arguments.  |alternative_service.host| is currently
  ignored.

This CL does not rename SetAlternateProtocolProbabilityThreshold and
GetAlternateProtocolProbabilityThreshold methods or
alternate_protocol_probability_threshold_ members, that will come later.

TBR=pauljensen@chromium.org
BUG=392576

Review URL: https://codereview.chromium.org/1018943002

Cr-Commit-Position: refs/heads/master@{#321622}

privetd is a chromeos service that handles device bootstrapping
and setup.
Expose an API to provide events from the privetd dbus.

BUG=467018

Review URL: https://codereview.chromium.org/996013003

Cr-Commit-Position: refs/heads/master@{#321621}

Part 3/3. Once any downstream uses that really wanted the
getActiveDataTypes() behavior are converted in 2/3, this CL can fix
getPreferredDataTypes().

BUG=466223

Review URL: https://codereview.chromium.org/1019733002

Cr-Commit-Position: refs/heads/master@{#321620}

(nothing uses it anymore)

Review URL: https://codereview.chromium.org/1020283002

Cr-Commit-Position: refs/heads/master@{#321619}

A newer version of PDFium is needed to unblock GN work.

Sheriffs: Revert first and ask questions later.
The last time this rolled, the build broke due to an issue that can
not be caught by the current trybots/CQ configuration.  I've applied
a speculative fix based upon the compilation logs for that run,
but can't be 100%.

This pulls in:
d1f792a Fix subtle issues in opj_skip_from_memory and add unit tests.
74b81d9 Revert "TEST that the tree goes RED when pixel/javascript tests fail." This reverts commit de00893874a9d5ecae497e00511e2395fc2f02e8.
de00893 TEST that the tree goes RED when pixel/javascript tests fail.
2c4f5ea Return OS status code from python test driver scripts.
a416173 Revert "Fix a problem that JP2 image is not displayed because index color space is used"
b29e12a Pull our new test case repository via deps.
a7865e9 Speculative fix for breakage in pdfum DEPS roll
5a0e504 Fix a problem that JP2 image is not displayed because index color space is used
cd3c476 Fix a bug that JPX images can't be shown
ae5b85a Fix potential integer overflow in fpdf_render_image.cpp
b8cd18a Revert "Kill CFX_GrowOnlyPool."
090d683 Kill CFX_GrowOnlyPool.
3c676ec Kill remaining sprintfs
1e580a1 Fix multiply defined macro in zlib.h on linux_chromium_clobber_rel_ng builder.
0700106 Make conversions for CPDF_Link explicit.
8842c62 Fix the pdfium component build.
5a370d7 Make conversion between CPDF_Dest and its object explicit.
7348fae Add gmock to .gitignore
624b6c1 Make conversion between CPDF_Action and its dictionary explicit.
61c25ee Merge to Master: Fix a compiling error in fxjs_v8.h
e1671bd Kill off JS_ErrorString type.
b720d0a Return error information from pdfium to JS.
944ccad Do hit tests against Annots in reverse order.
5b47dbf Remove unused IDS_ constants from pdfium.
3d25502 Pull in gmock for standalone pdfium builds.
ec61a85 Upgrade openjpeg to revision 2997.
42a8ada Implement a delegate for EmbedderTests.
57e4c86 Kill off unused IFXJS interface methods.
4351ac2 Fix compilation error on Windows.
f9cfdda Add small flate decode unit test.
e796478 Kill multi-byte comments in fxet_list.h
23f9288 Convert pdfium expected test results to PNG format.
0fd53cf Fix GN build after PNG support backport
ef1e85c Backport PNG output format to origin/master branch.

Review URL: https://codereview.chromium.org/1024703002

Cr-Commit-Position: refs/heads/master@{#321618}

BUG=450861
TEST=existing EME layout tests pass

Review URL: https://codereview.chromium.org/1023003002

Cr-Commit-Position: refs/heads/master@{#321617}

Summary of changes available at:
https://chromium.googlesource.com/skia/+log/b502ee3..6e8cd96

CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
TBR=egdaniel@google.com

Review URL: https://codereview.chromium.org/1020323002

Cr-Commit-Position: refs/heads/master@{#321616}

Making ParsedFallbackIconPath resemble ParsedFaviconPath, to prepare it
for chrome-search:// URL with <view_id>/<restricted_id> . Details:
- Replacing url() with url_string(). Need to update chrome://large-icon
  endpoint and test to accommodate.
- Adding path_index().

BUG=467712

Review URL: https://codereview.chromium.org/1017803003

Cr-Commit-Position: refs/heads/master@{#321615}

Revert of Clean up libjingle.gyp and remove references to libpeer_target_type. (patchset #1 id:1 of https://codereview.chromium.org/1008023002/)

Reason for revert:
See if this caused errors on Android.

Original issue's description:
> Clean up libjingle.gyp and remove references to libpeer_target_type.
> This variable isn't supported any longer.
>
> BUG=446865
>
> Committed: https://crrev.com/fd22585c501bdd46f0a511fa5346fc9be1b85f88
> Cr-Commit-Position: refs/heads/master@{#321371}

TBR=kjellander@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=446865

Review URL: https://codereview.chromium.org/1025003002

Cr-Commit-Position: refs/heads/master@{#321614}

When GPU raster is enabled, the active texture state is regularly set to
something different at the start of this function. This was observed
empirically.

Of note is that, in principle, all GL state much be reset. When we use
the command buffer to draw inside the CAOpenGLLayer, making the virtual
context current will reset all state for us.

BUG=423163

Review URL: https://codereview.chromium.org/1023673006

Cr-Commit-Position: refs/heads/master@{#321613}

This CL adds support for V4L2 MPLANE Capture Api.
Only supported format is YUV420M triplanar.

A new method is added to VideoCaptureDeviceClient,
namely OnIncomingCapturedYuvData(...), which forces
adding MOCKing here and there, and its own
implementation.

V4L2 MMAP API works via user mmap()ing a number of
buffer allocated by V4L2 capture device. If those
buffers are not correctly munmap()ed, bad things (c)
happen. In light of this, the manual buffer lifetime
management is changed to automatic one. Construction
(mmap()ing) of those called BufferTracker is
planarity specific (i.e. there's one such ctor in
each of BufferTracker{S,M}Plane), while the dtor
is generic and the same.

ToT class diagram:
     +------------------------------------+
     | VideoCaptureDeviceLinux            |
     |            +----------------------+|
     | <<ref>> -->| V4L2CaptureDelegate  ||
     |   cnt      |  (struct Buffer)     ||
     |            +----------------------+|
     +------------------------------------+

This CL class scheme:

 +--------------------------+
 | VideoCaptureDeviceLinux  |
 |                          |
 | <<ref_cnt>> ---+         |
 +----------------|---------+
 +----------------v-----------+ v4l2_capture_delegate.{cc,h}
 | +-----------------------+  |
 | |V4L2CaptureDelegate    |  |
 | |  (class BufferTracker)|  |
 | +-----------------------+  |
 +-------^------------------^-+
         |                  |
    +----|-------+ +--------|--+ v4l2_capture_delegate_multi_plane.{cc,h}
    | SPlane     | | MPlane    |
    | (BTSplane) | | (BTMPlane)|
    |            | +-----------+
    +------------+ v4l2_capture_delegate_single_plane.{cc,h}

- VCDevice works on the premise that its calls into
 VCDevice::Client::OnIncomingWhatever() are synchronous.
 That assumption is respected here.

- A bit of cleanup is done in OnIncomingCaptureData(),
 in what regards rotation/crop/odd sizes. A unit test
 is subsequently added.

- VideoCaptureDeviceFactory labels the devices as
 Single or Multi Planar. That labeling capture_api_type()
 needs to ripple through a bunch of files, causing some
 otherwise uninteresting changes in the patchsets.

BUG=441836

TEST= Compile and insmod vivid.ko into a kernel,
with options for supporting MPLANE api (multiplanar=2)
then capture using patched Chromium. Current vivid
does _not_ support any common Mplane format, needs
a patch:

https://github.com/miguelao/linux/tree/adding_yu12_yv12_nv12_nv21___mplane_formats___with_mods_for_kernel_3_13

that needs to be compiled against ubuntu sources 3.13 etc.

For even better coverage, use a normal WebCam and
navigate to http://goo.gl/fUcIiP, then open both
the MPlane camera mentioned in the previous paragraph
and the "normal" webcam (this is,partially, how I
try it).

Review URL: https://codereview.chromium.org/967793002

Cr-Commit-Position: refs/heads/master@{#321612}

This makes sure that when the notifications are being retrieved,
they always have the appropriate ID set.

BUG=447628

Review URL: https://codereview.chromium.org/1025743002

Cr-Commit-Position: refs/heads/master@{#321611}

R=danakj@chromium.org

Review URL: https://codereview.chromium.org/1020333002

Cr-Commit-Position: refs/heads/master@{#321610}

Check if the connected socket id already exists for P2PSocketDispatcherHost::OnAcceptIncomingTcpConnection

BUG=469152

Review URL: https://codereview.chromium.org/1020293002

Cr-Commit-Position: refs/heads/master@{#321609}

Removed access to the native XHR object used by the API.

This is a larger change than one might expect for two reasons: First,
because the native XHR object only allows the response content to be
retrieved while the onreadystatechange handler is executing, and
second, because the unit test for dns_blackhole_checker.js relied on
synchronous semantics which cannot be duplicated with promises because
when a promise is resolved, its "then" handlers are not called until
the next event cycle.

Review URL: https://codereview.chromium.org/1003433002

Cr-Commit-Position: refs/heads/master@{#321608}

Making toolkit-views available for non-browser UI means that this guard
in print_preview.js needs updating to ensure ESC still closes the dialog
on Mac. The webui container is still Cocoa.

Everything is toolkit-views now (except mobile/cast), so just check for
mac.

BUG=469045

Review URL: https://codereview.chromium.org/1025613002

Cr-Commit-Position: refs/heads/master@{#321607}

In preparation of componentization of FaviconTabHelper, introduce a
new target //components/favicon/content (as favicon is a layered
component) and move the function to convert from content::FaviconURL
to favicon::FaviconURL there.

BUG=374281

Review URL: https://codereview.chromium.org/1020213002

Cr-Commit-Position: refs/heads/master@{#321606}