Grid repeat(auto) with max-width shouldn't pay attention only to that (…

…web-platform-tests#28247)

[Taskcluster] pin Chrome Dev to 91.0.4449.6 (web-platform-tests#28238)

This is a temporary measure while the cause of the crash is being
investigated. This follows a similar approach taken before:
web-platform-tests#19360

Mitigation (not fix) for web-platform-tests#28209.

WPT: Increase timeout for slow XFO test

The WPT external/wpt/x-frame-options/multiple.html is slow because it
navigates a lot of subframes. Sometimes, it ends up hitting the
timeout.

Change-Id: I55e2f3e839bd3d20e41ebca71300294fe53bdabb
Bug: 626703
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2784869
Reviewed-by: Arthur Sonzogni <[email protected]>
Commit-Queue: Antonio Sartori <[email protected]>
Cr-Commit-Position: refs/heads/master@{#866527}

Merge color-space directory with wide gamut

I just noticed that we already have a directory for color-space related
tests. Merge the test I added with the existing directory.

Bug: 1178707

Change-Id: I509e845219d339d6a7f7f4414eab27425cfe37f3
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2786303
Reviewed-by: ccameron <[email protected]>
Commit-Queue: Yi Xu <[email protected]>
Cr-Commit-Position: refs/heads/master@{#866492}

[TablesNG] Colspanned cell uses table grid size defined by COL

Colspanned cells cannot extend beyond size of the table grid.
NGTableBorders::ComputeTableBorders did not use COLs
to compute grid size, and cells would up too narrow.

Bug: 1192304
Change-Id: I7e8e54a341c6f8ba99212e5076939c1e27601d30
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2785495
Commit-Queue: Aleks Totic <[email protected]>
Reviewed-by: Ian Kilpatrick <[email protected]>
Cr-Commit-Position: refs/heads/master@{#866448}

Disable NavigatorPluginsEmpty feature

This feature was enabled by default [1] to check web compat, and after
M90 went to Beta, a number of issues were raised ([2],[3]). As a
result, this feature is being disabled. This will likely need to
be merged back to M90 Beta.

This CL removes the WPT that tests for empty arrays, and reinstates
the prior layout test which enforces non-empty arrays.

[1] https://chromium-review.googlesource.com/c/chromium/src/+/2629990
[2] b/183094994
[3] crbug.com/1191639

Bug: 1164635, 1191639
Change-Id: Ieed5de8ece2332b5b830e9e295c9163e21e9c4a0
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2783393
Reviewed-by: Joey Arhar <[email protected]>
Auto-Submit: Mason Freed <[email protected]>
Commit-Queue: Mason Freed <[email protected]>
Cr-Commit-Position: refs/heads/master@{#866271}

Deflakes a background-color transition layout test

The fix here follows the same strategy as:
https://chromium-review.googlesource.com/c/chromium/src/+/2785141

Bug: 1187461
Change-Id: I335d5994bb4837ef81bd4ba0e12f7f0f0e155d20
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2785560
Commit-Queue: Xida Chen <[email protected]>
Reviewed-by: Kevin Ellis <[email protected]>
Cr-Commit-Position: refs/heads/master@{#866282}

Deflake a background-color animation layout test

It seems that adding a document.getAnimations()[0].ready fixes
the problem. Locally I have run the test 1000 times with -f
and no flakes at all.

Bug: 1190211
Change-Id: Idc4f3fc429b775c2553a37d9a98c88322204939a
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2785141
Commit-Queue: Xida Chen <[email protected]>
Reviewed-by: Kevin Ellis <[email protected]>
Cr-Commit-Position: refs/heads/master@{#866212}

Reland "Create canvasFilter class and interface"

Using the NewCanvas2DAPI flag on an idl file caused MSAN failures because
that flag was not explicitly initialized on the WebPreferences
constructor. This CL solves that problem
https://chromium-review.googlesource.com/c/chromium/src/+/2783641
I'll also run this CL through the MSan bots to confirm.

This is a reland of 5d00b2d8b1b3a3431ce093a990524a8b25a2bcac
It was reverted here https://chromium-review.googlesource.com/c/chromium/src/+/2775779

Original change's description:
> Create canvasFilter class and interface
>
> Create a "mongo-db" like interface for filters. Implement blur, but only
> blur.
>
> Bug: 1169216
> Change-Id: I0dc045fa70f0d1e443cf4f8ff53632056cca6924
> Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2741994
> Reviewed-by: Aaron Krajeski <[email protected]>
> Reviewed-by: Fernando Serboncini <[email protected]>
> Reviewed-by: Jeremy Roman <[email protected]>
> Reviewed-by: Xida Chen <[email protected]>
> Reviewed-by: Juanmi Huertas <[email protected]>
> Reviewed-by: Yi Xu <[email protected]>
> Commit-Queue: Aaron Krajeski <[email protected]>
> Cr-Commit-Position: refs/heads/master@{#864863}

Bug: 1169216
Change-Id: I6e5badf27d41f6dbdc1759ac518d502d1983ebd2
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2780336
Reviewed-by: Jeremy Roman <[email protected]>
Commit-Queue: Aaron Krajeski <[email protected]>
Cr-Commit-Position: refs/heads/master@{#866266}

[TablesNG] Fix dynamic writing mode invalidation

Our fancy new DCHECK uncovered a bug in TablesNG code.
Changing only the writing mode did not invalidate
logical borders properly.

The is because we were using BorderVisuallyEqual to
compare old and new borders. This function operates
in physical, not logical space.

A unittest was trying to test if borders are good
after writing mode change:
LayoutTableCellTest::BorderWidthsWithCollapsedBorders

The test was testing the wrong thing. I've tested
results in the browser, and Chrome was not definitely
getting the right borders. With the fix, expectations
for TablesNG and Legacy are very different, one is
buggy, the other one is not.

Change-Id: Ice7e20fb3b2ab9b449e2ee9bb15f316c4acd8161
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2783627
Commit-Queue: Aleks Totic <[email protected]>
Reviewed-by: Ian Kilpatrick <[email protected]>
Cr-Commit-Position: refs/heads/master@{#866306}