I'll try to rebaseline it again and failing that mark it as failing.

Any update on this? I want to get this test to stop failing so if there's no rebaseline today then I'll just mark it as failing.

Looks like it's passing on Linux and failing on Windows so not trivial to just rebaseline anyway.

Marked as failing:
https://codereview.chromium.org/1919013004

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/e4a2d6aa3557d266a83da63bc8875d4a6d877aa1

commit e4a2d6aa3557d266a83da63bc8875d4a6d877aa1
Author: mgiuca <mgiuca@chromium.org>
Date: Tue Apr 26 05:43:45 2016

Mark 3d-point-mapping-preserve-3d layout test as failing.

Failing on Win7 bots. Image diff shows slightly different borders, but
test is still passing on Linux so can't be trivially rebaselined.

TBR=eae@chromium.org
BUG= 606302 

Review URL: https://codereview.chromium.org/1919013004

Cr-Commit-Position: refs/heads/master@{#389705}

[modify] https://crrev.com/e4a2d6aa3557d266a83da63bc8875d4a6d877aa1/third_party/WebKit/LayoutTests/TestExpectations

Flaky rather than failing, looks like a skia antialiasing issue, perhaps due to difference in rounding?

This test is now failing on:
  WebKit Linux (dbg)
  WebKit Mac10.11 (dbg)
  WebKit Win7
  WebKit Win7 (dbg)

Also, transforms/3d/point-mapping/3d-point-mapping-deep.html is now failing sometimes on WebKit Win7 (dbg).

It started failing since the rebaseline bot made commit b1c1df42d68ba1b52ee63a28a2addef58475cd83, which was done in response to tobiasjs' commit b1dcc9b1e28e6de16347ca3e183efb847db50f80 ("Compute line-plane intersections in a more numerically robust manner.")

It appears to be failing consistently on these bots since this morning.

Maybe the screenshots are slightly different between debug and release versions?

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/541e709d9f9d021320789ad7afccf65ebf2cbf4c

commit 541e709d9f9d021320789ad7afccf65ebf2cbf4c
Author: amistry <amistry@chromium.org>
Date: Thu May 19 04:24:13 2016

Mark 3d-point-mapping-deep and 3d-point-mapping-preserve-3d as failing.

BUG= 606302 
TBR=tobiasjs@chromium.org

Review-Url: https://codereview.chromium.org/1992633004
Cr-Commit-Position: refs/heads/master@{#394669}

[modify] https://crrev.com/541e709d9f9d021320789ad7afccf65ebf2cbf4c/third_party/WebKit/LayoutTests/TestExpectations

Thanks for the interim fix @qyearsley, the Win7 failures were expected from before my CL (although I hoped that maybe the change would fix them). Will investigate why it's expanded to Linux/Mac debug, and see if there's anything that can be done to address that.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/b73c6641c1a82d7627558036a9424bc22114159b

commit b73c6641c1a82d7627558036a9424bc22114159b
Author: hiroshige <hiroshige@chromium.org>
Date: Thu May 19 23:28:24 2016

Mark 3d-point-mapping-preserve-3d.html as failing on Win7 Release

TBR=amistry@chromium.org
NOTRY=true
BUG= 606302 

Review-Url: https://codereview.chromium.org/1990193002
Cr-Commit-Position: refs/heads/master@{#394902}

[modify] https://crrev.com/b73c6641c1a82d7627558036a9424bc22114159b/third_party/WebKit/LayoutTests/TestExpectations

FYI
transforms/3d/point-mapping/3d-point-mapping-preserve-3d.html is start failing since r394893:r394901 on Linux Release builds, probably due to another cause.
I created  Issue 613497  for the newer regression.

Has this been disabled on Mac yet?  It's continuing to prevent some CLs from landing in the CQ (e.g., https://codereview.chromium.org/1907443006/, https://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_rel_ng/builds/231566).

In many cases, the CL will land anyway if the test fails without the patch (though more slowly).  In cases like this, though, retries don't work: "retries without patch disabled (build config changes detected) "

Can this be resolved soon?

Given that this is clearly causing pain at the moment, and it appears to be a bunch of pixel tests failing causing everyone pain vs. a unittest failing affecting N5X bots, maybe the best approach is to revert everything. The other alternative would be to mark those tests as failing everywhere in the interim, and I'll resolve it (either with a revert or a new CL) early next week.

Investigating why this might result in tiny differences in rendering between different platforms and Debug/Release is taking a bit of time. I spent some time today, and couldn't reproduce any of the differences. My tests showed the plane line intersection code generating consistent results across Debug and Release on Linux, at least. I'll have to expand to look at different platforms.

I presume that the baseline images are generated on Linux Release, and are expected to be the same on all other platforms. Is that correct?

The baseline images are pulled from the results of the bots running on https://build.chromium.org/p/chromium.webkit/builders which use Release builds; they're generated on different platforms (e.g. Win7, Win10, Mac 10.10, mac 10.11, Linux Trusty, Linux Precise), and then the debug build results are expected to be the same as the release build results.

 Issue 613497  has been merged into this issue.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/b86f78b2d23aed1e32214a69d148e021d783aad3

commit b86f78b2d23aed1e32214a69d148e021d783aad3
Author: lukasza <lukasza@chromium.org>
Date: Fri May 20 18:26:33 2016

Disabling consistently failing test - 3d-point-mapping-preserve-3d.html

The test consistently fails on mac_chromium_rel_ng, and is preventing
unrelated CLs from landing via CQ (which is not retrying without patch
because "retries without patch disabled (build config changes detected)").
Example:
https://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_rel_ng/builds/231566

The test also consistently fails when I tried running it locally, on my
stock Goobuntu machine:
$ cat out/gn_no_asan/args.gn
dcheck_always_on = true
is_component_build = true
is_debug = false
use_goma = true

BUG= 606302 
NOTRY=true

Review-Url: https://codereview.chromium.org/1999793003
Cr-Commit-Position: refs/heads/master@{#395121}

[modify] https://crrev.com/b86f78b2d23aed1e32214a69d148e021d783aad3/third_party/WebKit/LayoutTests/TestExpectations

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/97f34fffe2696482fb16ea7945e4aab2ff28a179

commit 97f34fffe2696482fb16ea7945e4aab2ff28a179
Author: ksakamoto <ksakamoto@chromium.org>
Date: Mon Jul 04 05:12:02 2016

Sheriffing: disable three failing tests

The following layout test is failing on Win7 debug:

compositing/perpendicular-layer-sorting.html
transforms/3d/point-mapping/3d-point-mapping-deep.html
transforms/3d/point-mapping/3d-point-mapping-preserve-3d.html

3d-point-mapping-preserve-3d.html is also failing on Mac and Linux.

These are failing since r403584, which is auto-rebaseline for tobiasjs' r403559.

BUG= 606302 
TBR=tobiasjs@chromium.org
NOTRY=TRUE

Review-Url: https://codereview.chromium.org/2120893002
Cr-Commit-Position: refs/heads/master@{#403635}

[modify] https://crrev.com/97f34fffe2696482fb16ea7945e4aab2ff28a179/third_party/WebKit/LayoutTests/TestExpectations

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/bac5297e0e9cf8f4c7b737032d140c3673daa3df

commit bac5297e0e9cf8f4c7b737032d140c3673daa3df
Author: schenney <schenney@chromium.org>
Date: Wed Oct 12 22:35:58 2016

Move LayoutTests that fail only on Win7 Debug to NeverFixTests

TBR=fmalita@chromium.org
BUG= 484370 , 552433 , 606302 

Review-Url: https://codereview.chromium.org/2418553002
Cr-Commit-Position: refs/heads/master@{#424887}

[modify] https://crrev.com/bac5297e0e9cf8f4c7b737032d140c3673daa3df/third_party/WebKit/LayoutTests/NeverFixTests
[modify] https://crrev.com/bac5297e0e9cf8f4c7b737032d140c3673daa3df/third_party/WebKit/LayoutTests/TestExpectations

These tests are now failing on all Mac bots, see e.g. here:
https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Mac10.9/builds/39589/steps/webkit_tests/logs/stdio

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/0589d1e77106779929525de308943c2f3c477c43

commit 0589d1e77106779929525de308943c2f3c477c43
Author: msramek <msramek@chromium.org>
Date: Thu Nov 17 13:43:12 2016

Disable the failing 3d-point-mapping-deep.html

transforms/3d/point-mapping/3d-point-mapping-deep.html is failing on all
Linux, Mac, and Windows Webkit bots.

NOTRY=True
TBR=tobiasjs@chromium.org
BUG= 606302 

Review-Url: https://codereview.chromium.org/2512683002
Cr-Commit-Position: refs/heads/master@{#432858}

[modify] https://crrev.com/0589d1e77106779929525de308943c2f3c477c43/third_party/WebKit/LayoutTests/TestExpectations

Disabled them also on Linux and Windows, as they're actually failing on all of those platforms as well...

I can't see any Webkit CLs in the range where this started failing, but there is one CL which matches the keyword "3d" and that is https://codereview.chromium.org/2504583003 

danakj@, could it have broken 3d-point-mapping-deep.html ?

Looking at the CL range that broke the mac bots, the most likely suspect is:

https://codereview.chromium.org/2504583003

But I think that the correct response to that is not to disable the tests, but to rebaseline.

Dana, does this sound right to you?

If that CL changed visual results they are more correct now and we should rebaseline yes.

I'm failing at finding a bot run with layout test results I can look at to verify visually what changed.

Run the various [platform]_blink_rel bots with a patch that removes the TestExpectation line. You can then use the webkit-patch rebaseline-cl command to get new baselines.

See, for example, https://codereview.chromium.org/2505163002/

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/9127c895e679e4ae159bee651b3ce9373b37e891

commit 9127c895e679e4ae159bee651b3ce9373b37e891
Author: schenney <schenney@chromium.org>
Date: Fri Nov 18 18:03:42 2016

Remove expectation for fixed test.

The test fast/transforms/transform-update-frame-overflow.html was
de-flaked in https://codereview.chromium.org/2509843003/ and
the flakiness dashboard indicates it is indeed no longer flaky.

Also add a tracking date to a newly added expectation line.

TBR=chrishtr@chromium.org
BUG= 655955 , 606302 

Review-Url: https://codereview.chromium.org/2511883004
Cr-Commit-Position: refs/heads/master@{#433236}

[modify] https://crrev.com/9127c895e679e4ae159bee651b3ce9373b37e891/third_party/WebKit/LayoutTests/TestExpectations

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/daebb5887b27f4a4f0a09b9920cb1db86b379459

commit daebb5887b27f4a4f0a09b9920cb1db86b379459
Author: schenney <schenney@chromium.org>
Date: Tue Apr 04 18:18:08 2017

Rebaseline 2 transforms/3d tests

Both
transforms/3d/point-mapping/3d-point-mapping-preserve-3d.html
and
transforms/3d/point-mapping/3d-point-mapping-deep.html
are marked as flakey or failing, but we can't tell if that's the case
without rebaselining. So do that and see what happens.

TBR=danakj@chromium.org
BUG= 606302 

Review-Url: https://codereview.chromium.org/2788213002
Cr-Commit-Position: refs/heads/master@{#461771}

[modify] https://crrev.com/daebb5887b27f4a4f0a09b9920cb1db86b379459/third_party/WebKit/LayoutTests/TestExpectations
[modify] https://crrev.com/daebb5887b27f4a4f0a09b9920cb1db86b379459/third_party/WebKit/LayoutTests/platform/linux/transforms/3d/point-mapping/3d-point-mapping-deep-expected.png
[modify] https://crrev.com/daebb5887b27f4a4f0a09b9920cb1db86b379459/third_party/WebKit/LayoutTests/platform/linux/transforms/3d/point-mapping/3d-point-mapping-deep-expected.txt
[modify] https://crrev.com/daebb5887b27f4a4f0a09b9920cb1db86b379459/third_party/WebKit/LayoutTests/platform/linux/transforms/3d/point-mapping/3d-point-mapping-preserve-3d-expected.png
[modify] https://crrev.com/daebb5887b27f4a4f0a09b9920cb1db86b379459/third_party/WebKit/LayoutTests/platform/linux/transforms/3d/point-mapping/3d-point-mapping-preserve-3d-expected.txt
[modify] https://crrev.com/daebb5887b27f4a4f0a09b9920cb1db86b379459/third_party/WebKit/LayoutTests/platform/mac-mac10.9/transforms/3d/point-mapping/3d-point-mapping-deep-expected.png
[modify] https://crrev.com/daebb5887b27f4a4f0a09b9920cb1db86b379459/third_party/WebKit/LayoutTests/platform/mac/transforms/3d/point-mapping/3d-point-mapping-deep-expected.png
[modify] https://crrev.com/daebb5887b27f4a4f0a09b9920cb1db86b379459/third_party/WebKit/LayoutTests/platform/win/transforms/3d/point-mapping/3d-point-mapping-deep-expected.png
[modify] https://crrev.com/daebb5887b27f4a4f0a09b9920cb1db86b379459/third_party/WebKit/LayoutTests/platform/win/transforms/3d/point-mapping/3d-point-mapping-deep-expected.txt

https://codereview.chromium.org/2793423004 re-added expectations for Win Debug. They need to be moved to NeverFixTests.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/858bfb659e50d2d2debebd92847c59b884848ae4

commit 858bfb659e50d2d2debebd92847c59b884848ae4
Author: schenney <schenney@chromium.org>
Date: Wed Apr 05 21:52:46 2017

Move Paint Team TestExpectations lines into appropriate sections

This completes a review of all the Paint Team owned layout test
failures. With this patch tests are grouped so we know which ones
might be worth investigating in the future and so we can more
easily incorporate new failures.

There is one outstanding patch to be comnitted rebasing one set
of tests.

TBR=chrishtr@chromium.org
BUG=309675,504613, 517449 , 549742 , 602110 , 606302 ,614910, 676561 

Review-Url: https://codereview.chromium.org/2803633005
Cr-Commit-Position: refs/heads/master@{#462231}

[modify] https://crrev.com/858bfb659e50d2d2debebd92847c59b884848ae4/third_party/WebKit/LayoutTests/NeverFixTests
[modify] https://crrev.com/858bfb659e50d2d2debebd92847c59b884848ae4/third_party/WebKit/LayoutTests/TestExpectations

Now the test is in NeverFixTests.