f0e1aa75a7 Merge pull request #2 from cclauss/modernize-Python-2-codes
08bedeaf9d Use print() function in both Python 2 and Python 3

git-subtree-dir: css/tools/apiclient
git-subtree-split: f0e1aa75a7113c2df87ab76cdf6734e77dfbaeb7

chromedriver doesn't allow changing Object.prototype to add enumerable
properties, but this test requires setting some values on
Object.prototype.  When Object.prototype.a is set to:

  {b: {c: 'on proto'}}

chromedriver fails with:

    JavascriptErrorException: javascript error (500): Maximum call stack size exceeded
      (Session info: chrome=72.0.3626.121)

    Remote-end stacktrace:

    #0 0x563ff3a32a59 <unknown>
    #1 0x563ff39cb7f3 <unknown>
    #2 0x563ff38fcd7c <unknown>
    #3 0x563ff38ff78c <unknown>
    #4 0x563ff38ff5f7 <unknown>
    #5 0x563ff38ffbe7 <unknown>
    #6 0x563ff38fff1b <unknown>
    #7 0x563ff38a3f7a <unknown>
    #8 0x563ff3899bf2 <unknown>
    #9 0x563ff38a37b7 <unknown>
    #10 0x563ff3899ac3 <unknown>
    #11 0x563ff38782d2 <unknown>
    #12 0x563ff3879112 <unknown>
    #13 0x563ff39fe865 <unknown>
    #14 0x563ff39ff32b <unknown>
    #15 0x563ff39ff70c <unknown>
    #16 0x563ff39d940a <unknown>
    #17 0x563ff39ff997 <unknown>
    #18 0x563ff39e9947 <unknown>
    #19 0x563ff3a1a800 <unknown>
    #20 0x563ff3a3c8be <unknown>
    #21 0x7f3bf4545494 start_thread
    #22 0x7f3bf2d58a8f clone

    Ran 1 tests finished in 2.0 seconds.
      • 0 ran as expected. 0 tests skipped.
      • 1 tests had errors unexpectedly

Work around this problem by cleaning up the test environment so
Object.prototype no longer has the override by the time chromedriver
tries to inspect the test result.

While here, fix the other tests to use the t.add_cleanup() function
so they'll cleanup their test environment in case they exit in
some other way besides reaching t.done().

The underlying chromedriver issue is tracked upstream at
https://crbug.com/chromedriver/2555.

Bug: 934844
Change-Id: Id1b4ab2a908bfbc001e2a2d045eeec3ef01c24d9

We used to commit navigation after receiving the first byte of
document response. This CL moves commit earlier, synchronously done
from CommitNavigation call. The change should not be web-observable,
but some internal assumptions may have been affected.

Test changes:
- ReplacingDocumentLoaderFiresLoadEvent was testing the old behavior,
  which is not applicable anymore.
- MultiChunkWithReentrancy now uses a different method to trigger
  reentrancy (pdf plugin), since we no longer commit after first byte.
- backdrop-object.html and anchor-change-href.svg relied on test finishing
  late enough, now they wait for onload to eliminate a race.
- use-property-synchronization-crash.html now reports an error message
  synchronously and therefore has JS stack and a line number.
- setting-allowpaymentrequest-timing.https.sub.html has a race as
  explained here [1], and now fails even without site isolation.

This corresponds to the step 8.b from the doc linked to the bug.

Difference from attempt #1 (https://chromium-review.googlesource.com/c/1399447):
- PluginDocumentParser and MediaDocumentParser early return if not parsing
  before accessing GetDocument. This is because DocumentLoader calls Finish()
  even after parser was stopped/detached. For example in Document::Abort
  we cancel parsing, but committed DocumentLoader might be still receiving data.
  We should ideally clean up all calls into parser, there are numerous TODOs
  for that.
- pageload-image-in-popup.html relies on small image being parsed in the same
  task as navigation commit. Using onload seems to fix the issue.
- touch-handler-iframe-plugin-assert.js hopes that onload for about:blank
  happens after test has finished, which is racy now.

[1] https://bugs.chromium.org/p/chromium/issues/detail?id=819800#c6

Bug: 855189, 937639, 836242, 937358
Change-Id: I65048a27e6d249a608d4eb61e5c882292386026e
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1506663
Commit-Queue: Dmitry Gozman <dgozman@chromium.org>
Reviewed-by: Daniel Cheng <dcheng@chromium.org>
Reviewed-by: Arthur Sonzogni <arthursonzogni@chromium.org>
Cr-Commit-Position: refs/heads/master@{#639992}

All tests pass, and crashes no longer happen. I believe
that code will not longer crash, but there might be
futher instances of incorrect positioning.

Fix #1
LayoutDescendantCandidates did not sweep newly discovered
candidates. This was done
manually once inside NGOutOfFlowLayoutPart::Run, and
sweep was not performed for LayoutDescendantCandidates
found in Legacy. Fix is to make LayoutDescendantCandidates
perform sweep instead.

Fix #2
fix #1 exposed a bug where duplicate fragments were generated
for a single layout object. This happened when NG was generating
fragments not inside ContainingBlock. Fix one instance of this
inside NGOutOfFlowLayoutPart::IsContainingBlockForDescendant
by making sure that OOF with inline containers are only positioned
inside its ContainingBlock()

Fix #3
NGOutOfFlowLayoutPart::LayoutDescendant offset adjustment.

Bug: 935805
Change-Id: I9f7ebbc7223f40fbbf6ba3739d9385bfd59e3641
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1517093
Commit-Queue: Aleks Totic <atotic@chromium.org>
Reviewed-by: Morten Stenshorne <mstensho@chromium.org>
Reviewed-by: Koji Ishii <kojii@chromium.org>
Cr-Commit-Position: refs/heads/master@{#641628}

…the WPT innerText getter test.

Depends on D45159

Differential Revision: https://phabricator.services.mozilla.com/D46186

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1241631
gecko-commit: 4dc6ff8d58b31c747e519abcbe01270d01d66636
gecko-integration-branch: autoland
gecko-reviewers: mats

It should only be definite if the element already had a definite main
size or if the container has a definite main size.

This is #2 from https://drafts.csswg.org/css-flexbox/#definite-sizes

Bug: 845235
Change-Id: I0230080d22ada93ebc8bae09aeda629d87cf5b6d
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1797442
Reviewed-by: Christian Biesinger <cbiesinger@chromium.org>
Commit-Queue: David Grogan <dgrogan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#698790}

…eb-platform-tests#28617)

Subresource Web Bundles.

The problem is: when Web Bundle fetching fails due to a network error,
Subresource fetch doesn't fail forever.
One such case (subresource-loading-cors-error test) was
timing out previously but passes successfully with this change.

This CL also adds 2 WPT tests:
1. subresource-loading-network-error.https.tentative.sub.html
2. subresource-loading-web-bundle-fetch-failed.https.tentative.html

Test #1 is a scenario with a different network error than the CORS
one, but with the same issue of subresource fetching timing out
without the change. It passes successfully after the change.

Test #2 is a scenario with a Web bundle not found error, which is
not directly influenced by the code added in this CL, but it expands
the test coverage which was found to be lacking the error cases before.

Bug: 1168449

Change-Id: Ia3abb967e36274becc86e317bc51b1272d3ae679
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2826001
Reviewed-by: Tsuyoshi Horo <horo@chromium.org>
Reviewed-by: Hayato Ito <hayato@chromium.org>
Reviewed-by: Kinuko Yasuda <kinuko@chromium.org>
Commit-Queue: Miras Myrzakerey <myrzakereyms@google.com>
Cr-Commit-Position: refs/heads/master@{#875532}

Co-authored-by: Miras Myrzakerey <myrzakereyms@google.com>

1. Use GetWithoutInvalidation() instead of Get() in DCHECKs.
We should never call Get() inside of a DCHECK(), because this can
lead to a different code path depending on whether DCHECKs are enabled.

2. Get() should not cause immediate side effects. At most, it should
queue up an invalidation for later processing.

Fixing #1 and #2 were required in order to get past a first set of
errors introduced by the new test.

3. The actual fix -- avoid infinite loop by calling a special
new SlotAssignmentWillChange(), rather than ChildrenChanged(),
where a minimal GetWithoutInvalidation() is called that does not
lead to IsShadowContentRelevantForAccessibility() => FirstChild() =>
RecalcAssignedNodes() => ChildrenChanged() ... (infinite loop).

A simpler potential fix is in CL:2965317 but requires more
research. It's also mentioned in a TODO comment.

Bug: 1219311
Change-Id: Iafaa289f241a851404ce352715d2970172a2e5f8
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2961158
Reviewed-by: Joey Arhar <jarhar@chromium.org>
Reviewed-by: Mason Freed <masonf@chromium.org>
Reviewed-by: Dominic Mazzoni <dmazzoni@chromium.org>
Commit-Queue: Aaron Leventhal <aleventhal@chromium.org>
Cr-Commit-Position: refs/heads/master@{#892778}

This is a manual reland of
https://chromium-review.googlesource.com/c/chromium/src/+/3247449

The difference from the previous reland is that the browser tests now
include 2 separate timeouts and a double rAF, to ensure that the
presentation timestamp taken is far enough from both the time the first
frame is sent as well as from the time the second frame is sent.
More importantly, the test now actually is looking at the UKM metric,
rather than at the histogram.

Original change's description:
> [LCP] Add animated image support
>
> This CL adds support for better handling of animated images in LCP:
> * A new attribute is exposing the first animated frame's paint time
> (behind a flag).
> * `startTime` is not changed.
> * The PageLoadMetrics reported for LCP are set to that first frame paint
> time for animated images (behind another flag).
> * Entries are not emitted until the image is loaded.
>
> Relevant spec issue:
> w3c/largest-contentful-paint#83

Bug: 1260953
Change-Id: I34070bd90a74ed44281da63b547f13d9669f389b
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3250690
Reviewed-by: Nicolás Peña Moreno <npm@chromium.org>
Commit-Queue: Yoav Weiss <yoavweiss@chromium.org>
Cr-Commit-Position: refs/heads/main@{#936516}

The previous patch:
https://chromium-review.googlesource.com/c/chromium/src/+/3371612/6
checked an AnonymousIframe and an Iframe wasn't sharing the same
partition.

This one test:
- Two sibling same-origin anonymous iframe share the same partition.
- Two same-origin nested anonymous iframe share the same partition.
- Two same-origin anonymous iframe from different popup do not share
  the same partition.

Bug: 1285331,1226469
Change-Id: I7ebc3a5bbb5e1f12d0ceaac9d89c1deb30174a37
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3379159
Reviewed-by: Andrew Williams <awillia@google.com>
Commit-Queue: Arthur Sonzogni <arthursonzogni@chromium.org>
Cr-Commit-Position: refs/heads/main@{#960946}

Co-authored-by: Arthur Sonzogni <arthursonzogni@chromium.org>

By adding new exhaustive tests under ordering/, it was revealed that the ordering between navigatesuccess/navigateerror and the committed/finished promises was not always consistent:

1. Simply adding a currentchange event handler would cause microtasks to run during commit, which changed some ordering.

2. Calling transitionWhile() would take us from the zero-promise case to the 1+-promise case in ScriptPromise::All(). As the new comment explains, both the spec and implementation have an observably-different fast path for the 0-promise case which caused changes in ordering.

In the course of fixing this, I found out that the did_finish_before_commit_ code in app_history_api_navigation.{h,cc} was actually not a mitigation for the case it stated, where promises passed to transitionWhile() would settle faster than the browser-process roundtrip for same-document traversals. That is in fact impossible, since we only fire the navigate event after the browser-process roundtrip has completed. Instead, they were a mitigation for (1).

This commit then ensures consistent ordering, tested with new rather-exhaustive tests, in the following manner:

* We move the firing of currentchange to before resolving the committed promise. This eliminates (1) and allows us to delete the did_finish_before_commit_ tracking.

* We always ensure we pass 1+ promises to ScriptPromise::All(). This eliminates (2).

A consequence of this is that we are now more likely to get rejected finished promises, in cases like

    await appHistory.navigate("#1").committed;
    await appHistory.navigate("#2").committed;

Before, the finished promise for the #1 navigation would go through the fast path per (2), and fulfill before the navigation to #2 canceled it. Now that does not happen, so code like the above will give an unhandled promise rejection for #1's finished promise.

To avoid this, we unconditionally mark finished promises as handled. This follows some web platform precedent, e.g. stream closed promises, where the promise is one of several information channels (in this case the developer might also find out via the AbortSignal or the navigateerror event). We do *not* do this for the committed promise though, as if a commit fails, that's probably something more deeply wrong, and cannot be ignored.

All of these changes will require spec updates.

For the tests, we introduce a new ordering/ directory which contains cross-cutting ordering tests, and we consolidate a few tests into the newly-introduced variant-driven exhaustive ones. A couple of other tests were affected by these changes too or fixed as a drive-by.

Change-Id: I8a50ca28d009e0a8a2c94331cd17f29b0a8dc463
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3405377
Reviewed-by: Nate Chapin <japhet@chromium.org>
Commit-Queue: Domenic Denicola <domenic@chromium.org>
Cr-Commit-Position: refs/heads/main@{#963772}

This test fails with off-by-one values on the green background. This
is attempt #2 to fix that, by adding fuzziness.

Bug: 1323293
Change-Id: I9f51f257ef0064b6cd144a32ae01b148ed881112
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3638193
Reviewed-by: Philip Rogers <pdr@chromium.org>
Auto-Submit: Mason Freed <masonf@chromium.org>
Commit-Queue: Mason Freed <masonf@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1001695}

In the case that a popover contains an invoker that points back to that
invoker, the tab navigation code used to get confused. E.g.:

```
<div id="menu" popover>
  <button autofocus popoverhidetarget="menu">Button #1</button>
  <button popoverhidetarget="menu">Button #2</button>
</div>
```

In this case, trying to tab between the first and second button would
break because the second button appeared to be an invoker for a new
popover, when in reality it was an invoker for the same popover.

Fixed: 1399601
Bug: 1307772
Change-Id: I276370d7c8eee0dd32f0c89da202a0d3777bf911
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4133482
Commit-Queue: Mason Freed <masonf@chromium.org>
Auto-Submit: Mason Freed <masonf@chromium.org>
Commit-Queue: Joey Arhar <jarhar@chromium.org>
Reviewed-by: Joey Arhar <jarhar@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1089080}

…on discrete properties."

Original patch: https://crrev.com/c/4219315
Revert 1: https://crrev.com/c/4305045
Reland 1: https://crrev.com/c/4304846
Revert 2: https://crrev.com/c/4307944

Despite the most relaxed value in TestExpectations for
discrete-no-interpolation.tentative.html, some bots are still breaking
for reasons I don't understand. I am going to remove the test completely
until it stops crashing or failing. This patch will still be useful
despite the lack of tests because other people are doing prototyping
work that builds on this change.

Bug: 1399631, 1413556
Change-Id: Ic08f16d6436e37acc8947a94ca699c1735fe221f
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4308075
Commit-Queue: Joey Arhar <jarhar@chromium.org>
Reviewed-by: Mason Freed <masonf@chromium.org>
Auto-Submit: Joey Arhar <jarhar@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1114026}

…sitions on discrete properties.""

This reverts commit 268250b.

…les transitions on discrete properties.""" (web-platform-tests#38936)

This reverts commit 78f70fd.

Most of the logic now lives on NaviagateEvent itself (in React()
and ReactDone()), which the new NavigateEvent::Reaction is a minimal
callback function wrapper.

Change-Id: Id31152b048fc7166a8ac45bda7e136a2143fd56f
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4369772
Reviewed-by: Domenic Denicola <domenic@chromium.org>
Commit-Queue: Nate Chapin <japhet@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1123074}

…each time of the loop

There are 2 possible scenarios which are not handled by the method.

1. Moving content node to new `<blockquote>` has already been moved to outside
of the editing host.
2. There is no container to insert new `<blockquote>`, e.g., in an inline
editing host.

In the case #1, we should ignore the ex-child node.  In the case #2, we should
abort it.  Note that Chrome inserts `<blockquote>` even if there is no proper
container.  However, such behavior is disagreed in interop-2023.  Therefore,
it's okay just to abort it for now.

Depends on D180781

Differential Revision: https://phabricator.services.mozilla.com/D180782

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1756237
gecko-commit: 42f3f3ab11b47f1d56d8bcd6a128398539dd1f23
gecko-reviewers: m_kato

…eb-platform-tests#40504)

* [wdspec] browsingContext.print: fix rounding error in page.py test

[pytest](https://github.com/web-platform-tests/wpt/blob/7a087d54be8b6c0ca0181a86dc1ff0b28461c383/webdriver/tests/support/image.py)
uses:

    def cm_to_px(cm): return round(cm * 96 / 2.54)

[js](https://github.com/web-platform-tests/wpt/blob/7a087d54be8b6c0ca0181a86dc1ff0b28461c383/tools/wptrunner/wptrunner/print_pdf_runner.html)
uses:

    const viewport = page.getViewport({ scale: 96. / 72. });
    ...
    canvas.height = viewport.height;
    canvas.width = viewport.width;

This produces a rounding error, even though the dimension is correct:

    >       assert cm_to_px(expected_dimensions["height"]) == height
    E       assert 454 == 453
    E         +454
    E         -453

The inconsistency of rounding in both ends becomes clear when we
eliminate "round" in the pytest side:

    >       assert cm_to_px(expected_dimensions["height"]) == height
    E       assert 453.54330708661416 == 453
    E         +453.54330708661416
    E         -453

There are multiple ways to fix this issue.

Option #1: Use "floor" instead of "round" in pytest.

Option #2: Use a range in the assertion comparison, allowing a
difference of up to +-1.0. This is what this PR does.

The comparison is performed in
[`assert_pdf_dimensions`](https://github.com/web-platform-tests/wpt/blob/b6107cc1ac8b9c2800b4c8e58af719b8e4d9b8db/webdriver/tests/support/fixtures_bidi.py#L210).

The problematic part is .96 / .72 which evaluates to 4/3 = 1.333333....

* use floor in cm_to_px instead of round

* compare to floor and to round instead of a range

* Revert "compare to floor and to round instead of a range"

This reverts commit 63f894e.

* Revert "use floor in cm_to_px instead of round"

This reverts commit 7e65d91.

This CL fixes a :has() invalidation bug when the following conditions
are met:

1. A style rule uses a :has() pseudo class. The :has() test result is
   affected by the anchor element's relationship to its sibling
   element at fixed distance. (e.g. '.a:has(+ .b) {}')
2. The :has() pseudo class was tested on an anchor element and it
   didn't matched.
3. If a sibling of the anchor element is removed, the :has() will
   match the anchor element.
   (e.g. '<div class=a></div><div id=target></div><div class=b></div>')
4. Remove a sibling of the anchor element so that the :has() matches
   the anchor element. (e.g. 'target.remove();')

For the removal, StyleEngine have to schedule :has() invalidation
even if the removed element doesn't have any identifier stored in
RuleFeatureSet. But it is not efficient to schedule :has()
invalidation for every element removal.

To avoid unnecessary :has() invalidation, StyleEngine checks whether
its parent has the 'ChildrenAffectedByDirectAdjacentRules' flag set
or not.

Currently, the SelectorChecker sets the flag only when it consumes
a direct adjacent combinator(+). This works most cases but it doesn't
work in this case (condition #2) because the SelectorChecker stops
the :has() argument selector matching before consuming the direct
adjacent combinator. Due to this, the parent of the anchor element
doesn't have the 'ChildrenAffectedByDirectAdjacentRules' flag set
and the StyleEngine doesn't schedule the :has() invalidation for the
removal.

To fix the error, when the SelectorChecker tests a :has() pseudo
class on an anchor element and the :has() is affected by the anchor
element's relationship to a sibling at fixed distance, the
SelectorChecker sets the flag of the parent to indicate that
StyleEngine need to schedule :has() invalidation whenever any child
of the element is removed.

Bug: 1480643
Change-Id: I5ec2e3c1db2773020368415f68bca1503367e669
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4864627
Commit-Queue: Byungwoo Lee <blee@igalia.com>
Reviewed-by: Rune Lillesveen <futhark@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1198137}

…rners

We had two issues:
1.  Before we had fast rounded corners, we always created mask layers
for rounded corner clips, and the mask layer made the scroll begin
unreliable and fall back to the main thread. With fast rounded corners,
the scrolls were treated as reliable without checking if the point is
in or out of the rounded corners.
2. If the scroller has a rounded corner by itself (instead of from an
ancestor), as we only create InnerBorderRadiusClip for the contents,
the compositor doesn't actually know which part of the layer bounds
is transparent to hit test (e.g. if the scroller has a border which
is outside of the InnerBorderRadiusClip). Now with HitTestOpaqueness,
such layers have HitTestOpaqueness::kMixed.

This CL changes the behavior of
LayerTreeImpl::FindLayersUpToFirstOpaqueToHitTest (renamed from
FindLayerUpToFirstScrollableOrOpaqueToHitTest):
- For issue #1: LayerImpl::OpaqueToHitTest() also checks whether the
  layer is affected by any fast rounded corners;
- For issue #2: FindLayerUpToFirstOpaqueToHitTest checks only
  OpaqueToHitTest() (without checking IsScrollerOrScrollbar())
  because a hit test on a scrollable layer is reliable only if it's
  opaque to hit test.

Bug: 40277896
Change-Id: I1acb16f2c6790760661e8239ea1599035f83ea51
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5466909
Commit-Queue: Xianzhu Wang <wangxianzhu@chromium.org>
Reviewed-by: Steve Kobes <skobes@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1291538}

Bug #1: AbstractRange::(Mark|Unmark)Descendants should always use
the shadow tree of web-exposed shadow root, instead of using
light DOM elements of the host.

Bug #2: aRange could possibly create mCrossShadowBoundaryRange
first (due to boundaries are in different tree), and later
moves the boundaries to the same tree. When this happens, we
should remove mCrossShadowBoundaryRange and use the default
range to represent it.

Differential Revision: https://phabricator.services.mozilla.com/D207608

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1891783
gecko-commit: 0f54a84c32d1c22d71ff7307944b824639adbd6f
gecko-reviewers: jjaschke, smaug, dom-core

Since @page border box layout objects aren't in the the layout tree, any
code that wants to walk up the tree to find the containing block will be
in for a surprise.

This would happen if percentage-based @page padding was used [1].
Recomputing padding during painting when we have already done it during
layout is rather pointless anyway. Read it out directly from the
fragment.

[1] #1 blink::LayoutBox::ContainingBlockLogicalWidthForContent()
    #2 blink::LayoutBoxModelObject::ComputedCSSPadding()
    #3 blink::LayoutBoxModelObject::PaddingTop()
    #4 blink::LayoutBoxModelObject::PaddingOutsets()
    #5 blink::BoxPainterBase::PaintFillLayer()
    #6 blink::BoxPainterBase::PaintFillLayers()
    #7 blink::BoxFragmentPainter::PaintBackground()

Bug: 40286153
Change-Id: I1e6e92c2ce1d81aab2673ec9a877eac455534102
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5526469
Commit-Queue: Morten Stenshorne <mstensho@chromium.org>
Reviewed-by: Xianzhu Wang <wangxianzhu@chromium.org>
Reviewed-by: Ian Kilpatrick <ikilpatrick@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1300711}