GitLab

Revert of Add new gpu driver bug workaround DISABLE_TRANSPARENT_VISUALS (patchset #17 id:310001 of https://codereview.chromium.org/2124633002/ )

Reason for revert:
Seems to break the Android and Linux GPU fyi bots:
https://build.chromium.org/p/chromium.gpu.fyi/builders/Android%20Release%20%28Nexus%205X%29/builds/1098
https://build.chromium.org/p/chromium.gpu.fyi/builders/Linux%20Release%20%28ATI%29/builds/48671
https://build.chromium.org/p/chromium.gpu.fyi/builders/Android%20Release%20%28Nexus%206P%29/builds/1004
<couple others>

GpuProcess.no_transparent_visuals_gpu_process
Failure: disable_transparent_visuals is missing in Browser process workarounds: [u'broken_egl_image_ref_counting', u'clear_uniforms_before_first_program_use', u'disable_blend_equation_advanced', u'disable_chromium_framebuffer_multisample', u'disable_depth_texture', u'disable_discard_framebuffer', u'disable_framebuffer_cmaa', u'disable_program_cache', u'force_cube_map_positive_x_allocation', u'max_copy_texture_chromium_size_1048576', u'max_texture_size_limit_4096', u'scalarize_vec_and_mat_constructor_args', u'unbind_attachments_on_bound_render_fbo_delete', u'unbind_egl_context_to_flush_driver_caches', u'use_virtualized_gl_contexts', u'wake_up_gpu_before_drawing']

[  FAILED  ] GpuProcess.no_transparent_visuals_gpu_process (5860 ms)

Original issue's description:
> Add new gpu driver bug workaround DISABLE_TRANSPARENT_VISUALS
>
> This is for better polish in the UI but it fixes a drag
> image issue on Linux crbug.com/593256 .
>
> Also add a kGpuDriverBugListJson entry to automatically
> disable transparent visuals on all drivers (especially
> the proprietary NVIDIA driver) except for opensource
> drivers (i.e. Mesa based drivers).
>
> BUG=369209
>
> R=kbr@chromium.org, sadrul@chromium.org, tapted@chromium.org
> 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
>
> Committed: https://crrev.com/71b2517dd4cd620a56ca6f13571129c6aee30b42
> Cr-Commit-Position: refs/heads/master@{#406505}

TBR=kbr@chromium.org,sadrul@chromium.org,tapted@chromium.org,piman@chromium.org,j.isorce@samsung.com
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=369209

Review-Url: https://codereview.chromium.org/2171023002
Cr-Commit-Position: refs/heads/master@{#407047}

Previously WebrtcDataStreamAdapter was implementing MessageChannelFactory,
which means that it was always creating channels asynchronously and in OPEN
state. Refactored it to create channels in CONNECTING state. Significantly
simplifies channel setup code as it's always synchronous. It's also possible
to reject incoming channels before they are connected and handle channels
rejected by the peer before being fully connected.

Review-Url: https://codereview.chromium.org/2164163002
Cr-Commit-Position: refs/heads/master@{#407046}

Flip the remaining CrOS builders to GN:

- chromium.lkgr:
  - ChromiumOS ASAN Release
- chromium.memory:
  - Linux Chromium OS ASan LSan Builder
- chromium.memory.fyi:
  - Chromium Linux ChromeOS MSan Builder
- tryserver.chromium.linux:
  - linux_chromium_chromeos_asan_rel_ng

R=stevenjb@chromium.org
BUG=605732
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_chromium_chromeos_asan_rel_ng

Review-Url: https://codereview.chromium.org/2170703002
Cr-Commit-Position: refs/heads/master@{#407045}

This saves us a branch at each call site since we don't need to check
for NUL characters.

BUG=605792

Review-Url: https://codereview.chromium.org/2167073002
Cr-Commit-Position: refs/heads/master@{#407044}

This patch makes |ApplyStyleCommand::applyBlockStyle()| to use connected
position of |paragraphStart| by checking whether |paragraphStart| is connected
or not after calling |moveParagraphContentsToNewBlockIfNecessary()| which makes
anchor node of |paragraphStart| detached from document.

This patch also introduces |DCHECK(paragraphStart.isConnected()| to catch
other cases in future.

In crbug.com/626730, we call |endOfParagraph()| with disconnected position in
|paragraphStart| then we hit connected position |DCHECK| in |VisiblePosition|
constructor.

BUG=626730
TEST=LayoutTests/editing/execCommand/apply_style/justify_right_ul_br_crash.html

Review-Url: https://codereview.chromium.org/2170823002
Cr-Commit-Position: refs/heads/master@{#407043}

BUG=617887

Review-Url: https://codereview.chromium.org/2149673003
Cr-Commit-Position: refs/heads/master@{#407042}

Chrome has been suffering from not being able to use xml values in the
support library for a while. This CL fixes it by making
chrome_java_resources and android_tools connected.

BUG=622941

Review-Url: https://codereview.chromium.org/2170203002
Cr-Commit-Position: refs/heads/master@{#407041}

Move the timeout for handling scroll from CodeMirrorTextEditor to SourceFrame.

R=lushnikov

Review-Url: https://codereview.chromium.org/2173553002
Cr-Commit-Position: refs/heads/master@{#407040}

BUG=317807
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

Review-Url: https://codereview.chromium.org/2146323002
Cr-Commit-Position: refs/heads/master@{#407039}

BUG=611329
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.perf:android_s5_perf_cq;master.tryserver.chromium.perf:mac_retina_perf_cq

Review-Url: https://codereview.chromium.org/2164913002
Cr-Commit-Position: refs/heads/master@{#407038}

Remove the direct variants from the SINGLE_AND_MULTI_THREAD_TEST_F
macros and friends, so you will only get a DirectRenderer if you
request it via SINGLE_AND_MULTI_THREAD_DIRECT_RENDERER_TEST_F and
similar macros.

Before, cc_unittests has 3366 tests. After, it has 3075.

R=enne
BUG=606056
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_blink_rel

Review-Url: https://codereview.chromium.org/2170183002
Cr-Commit-Position: refs/heads/master@{#407037}

Make Node::hasEditableStyle and Node::layoutObjectIsRichlyEditable global functions.

This is a preparation to move these functions from core/dom to core/editing.

TEST=No change in behavior

Review-Url: https://codereview.chromium.org/2171493003
Cr-Commit-Position: refs/heads/master@{#407036}

Screencast is available for our headless clients and turned on if
remoteFrontend queryparam is set.

This CL also begins using the devtool frontend from the cloud for 2 reasons:
 1. We need the frontend to version match our headless target
 2. The screencast module isn't in the bundled devtools, and only in the
    cloud.

BUG=627541

Review-Url: https://codereview.chromium.org/2144653004
Cr-Commit-Position: refs/heads/master@{#407035}

Summary of changes available at:
https://chromium.googlesource.com/v8/v8/+log/4209816c..cc88d565

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_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,yangguo@chromium.org,vogelheim@chromium.org

Review-Url: https://codereview.chromium.org/2167143002
Cr-Commit-Position: refs/heads/master@{#407034}

This CL simplifies RenderFrameImpl::decidePolicyForNavigation check for
navigating to file: URIs.  The simplification is okay, because if it's
safe to fork for file-to-file navigations when the opener isn't found,
then it should be also safe to do it even when there is an opener.
The simplification is desirable, because it removes code that is not
yet quite compatible with OOPIFs (where top()->document() may be null).

The simplification means that after the CL
fast/events/popup-allowed-from-gesture-initiated-form-submit.html
layout test (which POSTs a form to a file URI) goes through
NavigationControllerImpl::LoadURLWithParams.  This exposes the need
to remove an overagressive DCHECK that expected POST method only
for http (or https) URIs.  Since apparently other parts of the system
work gracefully when POSTing to a non-http URI, the DCHECK is being
removed.

BUG=101395, 466297
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

Review-Url: https://codereview.chromium.org/2166113002
Cr-Commit-Position: refs/heads/master@{#407033}

Remove calls to setHighlighted: for MD bookmark items.
Ensure that the items aren't highlighted when they are not enabled

BUG=627722, 628187, 628204

Review-Url: https://codereview.chromium.org/2169623002
Cr-Commit-Position: refs/heads/master@{#407032}

This caused fast/autoresize/turn-off-autoresize.html to change the screen size
to 0x0, which persisted to later tests. When a subsequent test attempted to
resize the window, it was clamped to the 0x0 screen size, causing (flaky)
failures.

Review-Url: https://codereview.chromium.org/2172083002
Cr-Commit-Position: refs/heads/master@{#407031}

Previously landed as: refs/heads/master@{#406780}

Review-Url: https://codereview.chromium.org/2144533002
Cr-Commit-Position: refs/heads/master@{#407030}

This CL replaces usage of Binding::WaitForIncomingMethodCall() with
a spin runloop waiting for incoming call or a connection error.
Affected test codes are within:
  components/arc/
  services/shell/

BUG=622438

Review-Url: https://codereview.chromium.org/2135223002
Cr-Commit-Position: refs/heads/master@{#407029}

These buffers can add up pretty quickly on memory constrained
Android devices.

Also add a histogram for performance of large uploads.  We don't
seem to have any metric aimed at upload performance, and want to
protect ourselves against a significant regression that only affects
upload performance.

BUG=524258

Review-Url: https://codereview.chromium.org/2092563002
Cr-Commit-Position: refs/heads/master@{#407028}

BUG=none
TBR=pfeldman

Review-Url: https://codereview.chromium.org/2164353002
Cr-Commit-Position: refs/heads/master@{#407027}

GN and GYP have different setup of include directories of gmock and
gtest. GN propagate include_dir of gtest and gmock to all deps, while
GYP doesn't. This CL changes GYP setting to propagate the setting to all
deps for parity to GN.

BUG=630299

Review-Url: https://codereview.chromium.org/2168983002
Cr-Commit-Position: refs/heads/master@{#407026}

This patch fixes unnecessary hyphens appear when soft hyphens are used
and its suffix is too long to fit.

In that case, m_lineBreak is still pointing to the last break
opportunity, which is the beginning of the line, and the character
before is on the previous line.

Also, different code doing the same logic for when the word with soft
hyphen is at the end of a text node and is not are unified.

BUG=627715

Review-Url: https://codereview.chromium.org/2167093002
Cr-Commit-Position: refs/heads/master@{#407025}

BUG=612331
TEST=none
R=jamescook@chromium.org

Review-Url: https://codereview.chromium.org/2170753005
Cr-Commit-Position: refs/heads/master@{#407024}

It is possible to start tracing, ie GPUTracer::IsTracing goes from false
to true, between GPUTracer::BeginDecoding and GPUTracer::Begin. If that
happens, outputter_ is null, which calls a NPE. Fix that.
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/2171553004
Cr-Commit-Position: refs/heads/master@{#407023}

in the setUp of GpuIntegrationTest

Note: I am not certain there is a good way to test the setup
functionality of GpuIntegrationTest in a unittest.  Given that
we are using a Fakes, I have added a hack to simulate throwing
an error in the setup method, but it is not how it would behave
in practice.  Any suggestions for a better way to unittest this
are appreciated.

Dependent on https://codereview.chromium.org/2148283003 landing in telemetry first

BUG=628022
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_optional_gpu_tests_rel;tryserver.chromium.mac:mac_optional_gpu_tests_rel;tryserver.chromium.win:win_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2151983002
Cr-Commit-Position: refs/heads/master@{#407022}

Review-Url: https://codereview.chromium.org/2169153002
Cr-Commit-Position: refs/heads/master@{#407021}

There are cases about the search field where:

- it is showing and focused, which means user is inside the field and forward
  slash key should be regarded as input.
- it is not showing and focused, in which case the forward slash key should be
  run shortcut command.
- it is showing, but not focused, when there are some contents in the field and
  it loses focus, in this case forward slash key should be shortcut.
- it is not showing, but focused, when user uses Tab to focus on the field,
  forward slash key should be input.

Considering all these cases, the forward slash key shortcut should just rely on
focused only.

BUG=630175
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2166263002
Cr-Commit-Position: refs/heads/master@{#407020}

In case frame is detached, do not call setFocusedFrame() as it can't be focused.

BUG=626750
TEST=LayoutTests/fast/dom/Selection/selection-crash.html

Review-Url: https://codereview.chromium.org/2153063003
Cr-Commit-Position: refs/heads/master@{#407019}

https://chromium.googlesource.com/skia.git/+log/81681949d363..901257a3ba1d

$ git log 81681949d..901257a3b --date=short --no-merges --format='%ad %ae %s'
2016-07-21 mtklein GN: remove hooks again
2016-07-21 mtklein No symlink.  The bots are not running hooks anyway.
2016-07-21 bungeman Fix SkScalerContext_FreeType::generateFontMetrics.
2016-07-21 brianosman Add SkColorSpace to GrDrawContext
2016-07-21 herb In the current code, tiling and bilerp sampling are strongly tied together. They can be separated by taking advantage of observation that translating a sample point into filter points in the bilerp stage the filter points will be at most 0.5 outside the tile. This allows simplified repositioning for the various tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0-> max. This allows bilerp to simply treat the filter points that fall off the tile. This allows tiling and bilerp sampling to be totally separate.
2016-07-21 msarett Miscellaneous color space refactors
2016-07-21 mtklein GN: add missing .exe
2016-07-21 bsalomon Reenable support for CHROMIUM_set_uniform_location
2016-07-21 caryclark let line/cubic intersect at four spots
2016-07-21 mtklein Basic standalone GN configs.
2016-07-21 dvonbeck Fixed GM:lightingshader2 size
2016-07-21 robertphillips Retract PipelineBuilder some more
2016-07-21 msarett Drop support for libjpeg-turbo < 1.2 in SkJpegCodec
2016-07-21 msarett Remove unnecessary getColorSpace() API from SkCodec
2016-07-21 egdaniel Setup system in Vulkan to reuse VkDescriptorSet allocations.

CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_blink_rel
TBR=jcgregorio@google.com

Review-Url: https://codereview.chromium.org/2167973005
Cr-Commit-Position: refs/heads/master@{#407018}

This fixes an issue where it was possible to have the menu open
for a history item which is no longer visible.

BUG=626231
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2165473002
Cr-Commit-Position: refs/heads/master@{#407017}

already injected, and injects a script only if it has not been injected.

BUG=248627
TEST= Run target browser_tests with
--gtest_filter="ExtensionApiTest.ContentScript*"

Review-Url: https://codereview.chromium.org/2116613002
Cr-Commit-Position: refs/heads/master@{#407016}

GeolocationProviderImpl is a singleton that directly references
GetContentClient()::browser() on ctor, to retrieve an embedder-
provider GeolocationDelegate. This blocks moving Geolocation
out of content/

This CL changes that mechanism: the GeolocationDelegate
is injected in BrowserMainLoop::PostMainMessageLoopStart()
(so it won't affect startup time IIUC), and kept in a variable
until the singleton construction.

This CL is the last blocking factor before taking Geolocation
out of content/browser and into device.

BUG=612334
TEST=./out/gn/browser_tests --gtest_filter="GeolocationBrowserTest.*"

PS:
A note about shell_content_browser_client.cc: Changing
GeolocationDelegate to injected, causes component's
AutofillRiskFingerprintTest  to start failing because
ShellGeolocationDelegate was initialized too early
and could not get a correct ShellBrowserContext*.
This CL also changes the ShellGeolocationDelegate to get
the browser_context() when is needed.

Review-Url: https://codereview.chromium.org/2127973002
Cr-Commit-Position: refs/heads/master@{#407015}

The Mojo PermissionService can be used directly from Blink. Creating
separate pipes for workers also removes the need to implement an
additional thread-hopping layer.

BUG=561879

Review-Url: https://codereview.chromium.org/2108003002
Cr-Commit-Position: refs/heads/master@{#407014}

Things I wanted clarified while investigating https://crbug.com/625042

BUG=625042

Review-Url: https://codereview.chromium.org/2149273003
Cr-Commit-Position: refs/heads/master@{#407013}

(version 4f58244457f8e785c5fbfa9db2735473f9e74ccf of https://github.com/w3c/webcrypto)

Review-Url: https://codereview.chromium.org/2167883003
Cr-Commit-Position: refs/heads/master@{#407012}

As of CUPS 1.6, CUPS has moved away from PPD attributes to IPP attributes.  This new backend uses the IPP APIs and will be better suited to use with CUPS going forward.  Until we can bump our minumum CUPS version to 1.7, the only client will be Chrome OS where the version is 2.1.3.

BUG=607668

TEST=With --enable-native-cups enabled, print dialog will show available CUPS printers if any are configured.

Review-Url: https://codereview.chromium.org/2105463002
Cr-Commit-Position: refs/heads/master@{#407011}

R=rockot@chromium.org
BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

Review-Url: https://codereview.chromium.org/2166953006
Cr-Commit-Position: refs/heads/master@{#407010}

Hide sign in guide in guest mode and show 'no synced tabs'.

BUG=628102
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2171513004
Cr-Commit-Position: refs/heads/master@{#407009}

Allows content shell to run context menu handling code in
WebContentsViewAura, including touch selection code. This is needed to
be able to properly test long press touch selection behavior.

BUG=594101

Review-Url: https://codereview.chromium.org/2132983002
Cr-Commit-Position: refs/heads/master@{#407008}