GitLab

Also fix the appearance of the "Show All" button on the download shelf when using custom themes (this regressed at some point in M55, as it works properly in my M54 build).

BUG=647047

Review-Url: https://codereview.chromium.org/2400563004
Cr-Commit-Position: refs/heads/master@{#423689}

When the coded size of the video frame is different from the visible
rect, SkCanvasVideoRenderer::Paint tries to scale the video while it
must ignore the pixels outside visible rect. This is addressed by
clipping the SkImage object created from VideoFrame prior to painting
the frame on the canvas.

BUG=611330

Review-Url: https://codereview.chromium.org/2369093002
Cr-Commit-Position: refs/heads/master@{#423688}

Revert of DevTools: archive content shell (patchset #1 id:1 of https://codereview.chromium.org/2383043003/ )

Reason for revert:
This broke the chromium build archive crbug.com/653406

Original issue's description:
> DevTools: archive content shell
>
> Creates an archive of the content shell for each desktop platform.
> This will enable external contributors to use a precompiled
> content shell to run devtools layout tests (and possibly others).
>
> I manually verified these files are needed to run devtools layout tests
> with the content shell.
>
> BUG=none
>
> Committed: https://crrev.com/baf98fcec2132efbdf404f9dff40de42703b65bf
> Cr-Commit-Position: refs/heads/master@{#423195}

TBR=kerz@chromium.org,kerz@google.com
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=none
NOTRY=true
NOPRESUBMIT=true

Review-Url: https://codereview.chromium.org/2399173002
Cr-Commit-Position: refs/heads/master@{#423687}

This CL reverts the following CLs:
1. "Change testing trial to be Disabled for All Bookmarks" (d7e42ebb)
2. Make "All Bookmarks" to be controlled by feature" (b8a3ebb9)
3. "Add an about flag to toggle the visibility of "All bookmarks"" (d37c3af1)
4. "Revert "Remove all bookmark section from bookmark manager"" (837a30e5)

BUG=605614
R=dfalcantara@chromium.org, isherman@chromium.org

Review URL: https://codereview.chromium.org/2394273002 .

Cr-Commit-Position: refs/heads/master@{#423686}

Revert of Use FileDescriptorWatcher in AlarmTimer. (patchset #2 id:20001 of https://codereview.chromium.org/2398753003/ )

Reason for revert:
This CL is causing a memory leak in the unit tests:

https://build.chromium.org/p/chromium.memory/builders/Linux%20Chromium%20OS%20ASan%20LSan%20Tests%20%281%29/builds/16728

failures:
AlarmTimerTest.RetainNonRepeatIsRunning
AlarmTimerTest.MessageLoopShutdown
AlarmTimerTest.NonRepeatIsRunning
AlarmTimerTest.RetainRepeatIsRunning

Sample report:
Direct leak of 152 byte(s) in 1 object(s) allocated from:
    #0 0xab8ebb in operator new(unsigned long) (/b/swarm_slave/w/irlnnUGq/out/Release/components_unittests+0xab8ebb)
    #1 0x9eba8fd in MakeUnique<base::FileDescriptorWatcher::Controller::Watcher, base::WeakPtr<base::FileDescriptorWatcher::Controller>, base::MessageLoopForIO::Mode &, int &> base/memory/ptr_util.h:56:29
    #2 0x9eba8fd in base::FileDescriptorWatcher::Controller::Controller(base::MessageLoopForIO::Mode, int, base::Callback<void (), (base::internal::CopyMode)1, (base::internal::RepeatMode)1> const&) base/files/file_descriptor_watcher_posix.cc:159
    #3 0x9ebb293 in base::FileDescriptorWatcher::WatchReadable(int, base::Callback<void (), (base::internal::CopyMode)1, (base::internal::RepeatMode)1> const&) base/files/file_descriptor_watcher_posix.cc:198:25
    #4 0xe0016a3 in timers::AlarmTimer::Reset() components/timers/alarm_timer_chromeos.cc:103:25
    #5 0x515b488 in timers::AlarmTimerTest_RetainRepeatIsRunning_Test::TestBody() components/timers/alarm_timer_unittest.cc:310:9

Original issue's description:
> Use FileDescriptorWatcher in AlarmTimer.
>
> FileDescriptorWatcher is a new API that replaces
> MessageLoopForIO::WatchFileDescriptor.
>
> This CL also gets rid of the base::Thread created in the anonymous
> namespace of alarm_timer_chromeos.cc. The FileDescriptorWatcher
> API is supported by every TaskScheduler thread. Very soon, most
> tasks in Chrome will run in TaskScheduler. Therefore, there is no
> reason to create a base::Thread to allow AlarmTimer to be used from
> threads that don't support the FileDescriptorWatcher API.
>
> BUG=645114
>
> Committed: https://crrev.com/8e17d7e9031b7e2e58473fe66a399943051357f0
> Cr-Commit-Position: refs/heads/master@{#423617}

TBR=chirantan@chromium.org,fdoray@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=645114

Review-Url: https://codereview.chromium.org/2394333002
Cr-Commit-Position: refs/heads/master@{#423685}

https://chromium.googlesource.com/chromium/src/+/bd1529c0d

BUG=649760
TBR=chrishtr@chromium.org

Review URL: https://codereview.chromium.org/2396203002 .

Cr-Commit-Position: refs/heads/master@{#423684}

https://pdfium.googlesource.com/pdfium.git/+log/ca03f7a1c654..3a0a808ff546

$ git log ca03f7a1c..3a0a808ff --date=short --no-merges --format='%ad %ae %s'
2016-10-06 kcwu Fix assertion failure when decoding malform G4 fax image
2016-10-06 kcwu Reject JBig2 Huffman table with too large shift value
2016-10-06 npm Split m_InstalledTTFonts into two vectors to avoid sketchy logic.
2016-10-06 caryclark switch to new skia blend mode enum class

BUG=653044

TBR=dsinclair@chromium.org

Review-Url: https://codereview.chromium.org/2395263002
Cr-Commit-Position: refs/heads/master@{#423683}

This reverts commit ea63767b, which
represents a feature that was disabled in M54 via field trial.

BUG=620535
TBR=isherman@chromium.org

Review-Url: https://codereview.chromium.org/2393993005
Cr-Commit-Position: refs/heads/master@{#423682}

That event will contain ExecutionMode/SequenceToken and TaskPriority.

It would be better to tack an optional extra arg to the main task event but
tracing doesn't currently allow this (http://crbug.com/652692).

BUG=649084

Review-Url: https://codereview.chromium.org/2392903002
Cr-Commit-Position: refs/heads/master@{#423681}

Add an ApplyAutoMargin function in ng_length_utils and use that from
ng_block_layout_algorithm.

I did not add it to computeMargins because not all callers should take
margin: auto into account. In particular, a child's width: auto computation
needs to treat margin: auto as zero (and, at any rate, does not have a fragment
available yet).

R=glebl@chromium.org,eae@chromium.org
BUG=635619

Review-Url: https://codereview.chromium.org/2399963002
Cr-Commit-Position: refs/heads/master@{#423680}

This removes TestWebKitPort (as suggested by a FIXME in the code)
and replaces each use with different Port subclasses. For the
most part this is straightforward, except for a few test methods.

Notes:

 - Previously, test_path_to_apache_config_file just tested the logic
   in the base Port class, but Port subclasses have different behavior,
   so more specific test methods could be added in test case subclasses.
 - The test_apache_config_file_name_for_platform method was testing
   behavior in the base Port class which didn't make sense to repeat
   for port subclasses, so I moved it to base_unittest.

Review-Url: https://codereview.chromium.org/2392683003
Cr-Commit-Position: refs/heads/master@{#423679}

eyyit.com:
> I jumped the gun on hsts preloading and now have a subdomain that needs to
> read xml feeds from non-https sites.  This HSTS preload including subdomains
> broke that.  Instead of waiting for the feeds to update their site to https,
> I'd rather get the removal process started soon so I have some options.

svallee.fr:
> • home.svallee.fr (many services, not all with HTTPS available)

vjirovsky.cz:
> axure.vjirovsky.cz – it’s branded (CNAME) version of axshare.com and they are
> not able to have certificate for axure.vjirovsky.cz

tradeacademy.in:
> list.tradeacademy.in -> We are using this domain for our newsletter
> application & we don't have Wildcard SSL certificate so whenever we open
> list.tradeacademy.in it goes to HTTPS in chrome.

almeria.fr:
> store.almeria.fr – problem with our website supplier

almeria-si.fr:
> store.almeria-si.fr – this site is managed by a partner, and we are not ready
> to migrate to https.

vozp.cz:
> • maps.vozp.cz – subsite dont using SSL
> • intranet.vozp.cz – internal application
> • helpdesk.vozp.cz – internal application

zen-trader.com:
> We cannot support HTTPS on the following subdomains:
> • status - Not under our control (third party server)
> • email - Not under our control (third party server)
> • clk - Not under our control (third party server)
> • mailsrv - Not under our control (third party server)

bratteng.me:
> dell.bratteng.me - some of the ports other than 80 and 443 does not support
> https yet

soleus.nu:
> We are a non-profit hosting association, and almost all subdomains are
> assigned to members (with their own vps and hosting services).

chrishamper.com:
> I had enabled the HSTS header with the "preload" directive on my domain while
> following an online guide related to HSTS, which didn't explain the meaning or
> repercussions of that directive. It is now causing much trouble when
> attempting to do development work using subdomains I'm spinning up as needed.

elisa.ee:
> tanama.elisa.ee – it needs to be opened with HTTP

mijailovic.net:
> • www.mijailovic.net - I’m moving my website from custom server to GitHub
> pages, but GitHub doesn’t support https on their custom subdomains.

skyo.com:
> • Skyo.com/api/ - Most Clients using API part of our site only support weak
> SSLv3 or dont support SSL/TLS at all. This had a huge impact on our business
> and we have found temporary ways around this until we can get off the preload
> list. We eventually will add the main site, skyo.com (www.skyo.com), back to
> HSTS but our backend Admin and API sections we will avoid the change due to
> too many clients not running modern systems.

mitell.jp:
> • [staging.mitell.jp, test.mitell.jp] - They are for test and our project
> cannot apply SSL for test sites.

callcap.com:
> We have several callcap.com subdomains that are used internally only that
> cannot work with HTTPS. The preload directive was configured by mistake on our
> webservers but has since been removed.

BUG=527947
TBR=palmer@chromium.org

Review URL: https://codereview.chromium.org/2306423004 .

Cr-Commit-Position: refs/heads/master@{#423678}

Review-Url: https://codereview.chromium.org/2398823002
Cr-Commit-Position: refs/heads/master@{#423677}

NSWindowDelegate.

Currently the NSWindowDelegate is cleared beforehand, which
means observers of OnWidgetDestroying() are unable to use
GetWidgetForNativeWindow(). This is a problem during closure of
the browser window because inactive tabs are reattached when they
become active due to the active tab being closed. This results in a
DCHECK in NativeViewHostMac::AttachNativeView().

To fix, clear the NSWindowDelegate in the method
BridgedNativeWidget calls on NativeWidgetMac which deletes the
BridgedNativeWidget instance (i.e. just after emitting
OnNativeWidgetDestroying()).

R=tapted@chromium.org

BUG=604628

Review-Url: https://codereview.chromium.org/2375903002
Cr-Commit-Position: refs/heads/master@{#423676}

BUG=604119
TEST=none

Review-Url: https://codereview.chromium.org/2399163002
Cr-Commit-Position: refs/heads/master@{#423675}

Record a QUIC packet as received once it's been decrypted, instead of waiting for it to be fully processed.  Protected by FLAGS_quic_receive_packet_once_decrypted.
Merge internal change: 134893437
https://codereview.chromium.org/2388153005/

Rename QuicSpdyStream::MarkTrailersDelivered to MarkTrailersConsumed. No behavior change.
Merge internal change: 134828057
https://codereview.chromium.org/2399543002/

Remove unused "bytes" argument from QuicReceivedPacketManager::RecordPacketReceived.  No functional change, not flag protected.
Merge internal change: 134709757
https://codereview.chromium.org/2391933005/

Deprecate FLAGS_quic_send_scfg_ttl
Merge internal change: 134688122
https://codereview.chromium.org/2394903002/

Deprecate FLAGS_quic_better_last_send_for_timeout
Merge internal change: 134622401
https://codereview.chromium.org/2396833002/

Adds QUIC_TOO_MANY_SESSIONS_ON_SERVER error code.
Merge internal change: 134538917
https://codereview.chromium.org/2393943002/

Move the logic from SendAlgorithmInterface::RetransmissionDelay() into QuicSentPacketManager, because it's identical for all send algorithms. No functional change, not flag protected.
Merge internal change: 134467054
https://codereview.chromium.org/2393933002/

StringPiece will be replaced with std::string_view. std::string_view doesn't have starts_with.
Merge internal change: 134445920
https://codereview.chromium.org/2399473003/

short circuit QUIC packets to packet buffer if there is already a CHLO on same connection buffered there. Protected by FLAGS_quic_buffer_packets_after_chlo.
Merge internal change: 134431904
https://codereview.chromium.org/2390773005/

Conversion of a QUIC method to an async signature and resulting fallout. No functional change intended, not flag-protected.
Merge internal change: 134406963
https://codereview.chromium.org/2397513002/

Add a QUIC connection option to use a more conservative handshake retransmission timer.  Protected by ENABLED FLAGS_quic_conservative_handshake_retransmits.
Merge internal change: 134348006
https://codereview.chromium.org/2392093002/

Negotiate token binding version 0.10 in QUIC crypto.
Merge internal change: 134333443
https://codereview.chromium.org/2392053002/
BUG=

Review-Url: https://codereview.chromium.org/2391103003
Cr-Commit-Position: refs/heads/master@{#423674}

Start plumbing it through to actions that require binaries from the Xcode
toolchain.

BUG=651267
TBR=sdefresne@chromium.org

Review-Url: https://codereview.chromium.org/2388063003
Cr-Commit-Position: refs/heads/master@{#423673}

This plumbs through multi advertising capabilities for the
instance. Instances can register and unregister advertisements.

Now with virtual destructors to prevent memory leaks in the
StructTraits conversion.

BUG=637171,653338
TEST=use nrf connect on remote device and verify that packets
  registered from the instance are visible on the remote device,
  run tests with ASAN/LSAN turned on and verify there are no
  memory errors

Review-Url: https://codereview.chromium.org/2394683007
Cr-Commit-Position: refs/heads/master@{#423672}

FrameView and PaintLayerScrollableArea may perform more work than
necessary if hide_scrollbars is true (e.g. trigger unnecessary
layouts), because they think that scrollbar existence has changed
even if it didn't.

For FrameView, we can avoid this by enforcing hide_scrollbars
in computeScrollbarExistence instead. For PLSA, we check for
hide_scrollbars in updateAfterLayout to determine if the
scrollbar existence will change.

Also fixes another potential bug in PLSA, which may incorrectly
assume existence of scrollbars if hide_scrollbars is true.

BUG=652317,639806
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2394163002
Cr-Commit-Position: refs/heads/master@{#423671}

BUG=563793
R=pdr@chromium.org

Review-Url: https://codereview.chromium.org/2391693004
Cr-Commit-Position: refs/heads/master@{#423670}

TBR=zmo@chromium.org
BUG=angleproject:1492
BUG=chromium:634525
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;master.tryserver.chromium.android:android_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2397213002
Cr-Commit-Position: refs/heads/master@{#423669}

BUG=653668
TEST=built the test

Review-Url: https://codereview.chromium.org/2394223003
Cr-Commit-Position: refs/heads/master@{#423668}

This CL implements variations querying and control for the number of
days after an app banner is dismissed or ignored that the banner will
not be shown for that site.

BUG=653339

Review-Url: https://codereview.chromium.org/2397043002
Cr-Commit-Position: refs/heads/master@{#423667}

ScheduleNextFrame() has several entry points. Ensure capture
timer is not set with no valid callback.

BUG=

Review-Url: https://codereview.chromium.org/2398993004
Cr-Commit-Position: refs/heads/master@{#423666}

If the desktop browser is opened to a banner-eligible PWA, and then
"Add to homescreen" is run from the Devtools "Application" tab, a DCHECK
failure in HostContentSettingsMap is triggered. The cause of this is a
null validated_url; this variable is usually set in
AppBannerManager::DidFinishLoad, but this will not run since banners are
disabled by default on desktop and triggering add to homescreen manually
creates the AppBannerManager after DidFinishLoad is run.

This CL fixes the issue by manually setting the validated_url when the
banner pipeline begins if it is null. This bug should only manifest
itself in this exact instance; enabling banners via
chrome://flags#enable-add-to-shelf ensures that the AppBannerManager is
created.

BUG=None
TEST=Disable "Add to shelf" in chrome://flags in a desktop Chrome
instance. Navigate to a banner-eligible site, open Devtools, and trigger
"Add to homescreen" from the "Application" tab. No crash should happen
and the add to shelf banner should be seen.

Review-Url: https://codereview.chromium.org/2392313003
Cr-Commit-Position: refs/heads/master@{#423665}

The external protocol handler dialog currently presents users with
a very large amount of jargon-filled text. This CL simplifies the dialog
down to just a title containing the app name and the existing checkbox.

On Windows 7 and earlier, the registry is queried for an app path,
which is displayed in the dialog. This CL modifies the query to
retrieve a program name from the app path, falling back to just
the executable name if the retrieval doesn't succeed.

A new histogram to log the number of times users check the
checkbox in the dialog is also added.

BUG=601725

Review-Url: https://codereview.chromium.org/2076253002
Cr-Commit-Position: refs/heads/master@{#423664}

This reverts commit bdc8e1e4.
This is a top crasher that's blocking release.

BUG=653403
TBR=horo@chromium.org

Review-Url: https://codereview.chromium.org/2399903003
Cr-Commit-Position: refs/heads/master@{#423663}

This shouldn't result in any change in behavior. The purpose of this
CL is to make it clearer what arguments may be used by
copy-existing-baselines-internal and which aren't.

Review-Url: https://codereview.chromium.org/2395633002
Cr-Commit-Position: refs/heads/master@{#423662}

This reverts commit 64f8e6ee.

BUG=563793
R=ikilpatrick@chromium.org
NOTRY=true

Review-Url: https://codereview.chromium.org/2398623003
Cr-Commit-Position: refs/heads/master@{#423661}

This CL contains an abstract interface for SyncManager and a private
class skeleton SyncManagerImpl with the method bodies stubbed out.

R=steimel@chromium.org,scf@chromium.org
CC=wez@chromium.org
BUG=650469

Review-Url: https://codereview.chromium.org/2377873002
Cr-Commit-Position: refs/heads/master@{#423660}

Review-Url: https://codereview.chromium.org/2400683002
Cr-Commit-Position: refs/heads/master@{#423659}

The real cost in validating signed modules is validating individual certificates. Since many certificates are provided en masse by catalogs, this counts catalogs as providing a single certificate.

BUG=617176
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.win:win10_chromium_x64_rel_ng

Review-Url: https://codereview.chromium.org/2384413004
Cr-Commit-Position: refs/heads/master@{#423658}

The viewport rect is always at 0,0 and the clip rect is always the
viewport rect now that WebView does external viewport/clip things via
embedding a SurfaceDrawQuad into another surface. So express this in
the code.

R=enne@chromium.org
BUG=606056
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Review-Url: https://codereview.chromium.org/2399713003
Cr-Commit-Position: refs/heads/master@{#423657}

As some applications are providing multiple configurations for
compatability with previous versions, only check that at least
one capability is provided on success, as that is likely to fail
once Chrome starts following the EME spec strictly. This also
avoids the problem of counting both success and failure for the
same configuration (if multiple configs were provided).

BUG=616233
TEST=encrypted media layout tests pass

Review-Url: https://codereview.chromium.org/2397543002
Cr-Commit-Position: refs/heads/master@{#423656}

BUG=652467

Review-Url: https://codereview.chromium.org/2399113002
Cr-Commit-Position: refs/heads/master@{#423655}

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/6723c79f0772..a2738b13731a

$ git log 6723c79f0..a2738b137 --date=short --no-merges --format='%ad %ae %s'
2016-10-06 dtu [code health] Move request_handler.py and xsrf.py into common/.

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

Review-Url: https://codereview.chromium.org/2387113007
Cr-Commit-Position: refs/heads/master@{#423654}

BUG=653289
TBR=grt

Review-Url: https://codereview.chromium.org/2394163003
Cr-Commit-Position: refs/heads/master@{#423653}

WebappInfo#create() returns null if either the id or the uri are null

BUG=None

Review-Url: https://codereview.chromium.org/2386383002
Cr-Commit-Position: refs/heads/master@{#423652}

This CL lands server change 135025499 by rogeeff.

BUG=488484

Review-Url: https://codereview.chromium.org/2400463002
Cr-Commit-Position: refs/heads/master@{#423651}

SurfaceSize is no longer needed that the external viewport is always
at 0,0 and no output surfaces override the size returned there except
cc unittests. So make them stop and remove the getter.

This kinda points out that GLFrameData's size is redundant with the
size given to Reshape. For now I'm DCHECKing that they are the same
to document (and confirm) this.

Also adds DCHECKs that the viewport rect is at 0,0 and the clip rect
given to DirectRenderer matches the viewport rect since that is what
Display does now. Followups can remove code to not need DCHECKs.

R=enne@chromium.org
BUG=606056
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Review-Url: https://codereview.chromium.org/2392183003
Cr-Commit-Position: refs/heads/master@{#423650}