Reduction attached. Hover over the text to see it disappear.
Works correctly in Firefox / OS X. Broken on Chrome Canary
60.0.3074.0

Deleted: test.html 493 bytes

test.html 493 bytes View Download

FWIW, confirmed that it does indeed require the transform. It's not a function of the final static state.

Is this really a P3? I would go for M-60 P1 at least until we know what caused it. We know it worked in M-58.

Pri-1 sounds good. Stephen, sending to you for next steps.

Bisect comes back with 
https://chromium.googlesource.com/chromium/src/+log/d2122a68e4106aabcde2b9a4ac766350ed68de2c..1459bd0dc4107bcde17077387bd5705097d51c44

Which includes:
https://chromium.googlesource.com/chromium/src/+/9dde8ce672f222dc77bd460ac703a2bef6910bc7

which is
commit	9dde8ce672f222dc77bd460ac703a2bef6910bc7	[log] [tgz]
author	chrishtr <chrishtr@chromium.org>	Fri Mar 31 03:33:32 2017
committer	Commit bot <commit-bot@chromium.org>	Fri Mar 31 03:33:32 2017
tree	66f84e67a1669102714bda3ae8ee45bf05921acd
parent	904a97549b34407f69820f1ff114c861f9c35bf6 [diff]
Account for filter bounds for backgroundRect during PaintLayer painting.

BUG= 701654 
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2781863005
Cr-Commit-Position: refs/heads/master@{#461014}

I'll try reverting that and see what happens.

Marking as an M59 blocker.

Revert fails due to all the changes that have gone in since.

Bask to Chris to see if it is indeed the correct blame patch.

I found the root cause. It's a bug only in M59, and only with
SPInvalidaiton. Therefore M58 is fine.

https://codereview.chromium.org/2830013007

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/fb28c1b7511ceeb4d47ed42dbeb0676591469fe6

commit fb28c1b7511ceeb4d47ed42dbeb0676591469fe6
Author: trchen <trchen@chromium.org>
Date: Wed Apr 26 03:09:06 2017

[blink] Skip ancestor clip in PaintLayerPainter for squashed layers.

The squashing layer passed a differrent painting root that is not the same
as the backing's space when painting squashed layers, thus a wrong clip
is applied when squashed layers are painted.

This was not a problem prior to GeometryMapper-based PaintLayerClipper
because the non-GM PaintLayerClipper skipped visual rect optimization
in this particular corner case.

BUG= 712783 
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2830013007
Cr-Commit-Position: refs/heads/master@{#467220}

[add] https://crrev.com/fb28c1b7511ceeb4d47ed42dbeb0676591469fe6/third_party/WebKit/LayoutTests/compositing/squashing/squash-transformed-with-clip-expected.html
[add] https://crrev.com/fb28c1b7511ceeb4d47ed42dbeb0676591469fe6/third_party/WebKit/LayoutTests/compositing/squashing/squash-transformed-with-clip.html
[modify] https://crrev.com/fb28c1b7511ceeb4d47ed42dbeb0676591469fe6/third_party/WebKit/Source/core/paint/PaintLayerPainter.cpp

Your change meets the bar and is auto-approved for M59. Please go ahead and merge the CL to branch 3071 manually. Please contact milestone owner if you have questions.
Owners: amineer@(Android), cmasso@(iOS), gkihumba@(ChromeOS), Abdul Syed@(Desktop)

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Verified this fixes the bug originally reported on  https://crbug.com/709081#c18 

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/840427fa341fe9f413fdf5c7df1941f08e152380

commit 840427fa341fe9f413fdf5c7df1941f08e152380
Author: Tien-Ren Chen <trchen@chromium.org>
Date: Wed Apr 26 23:50:53 2017

[blink] Skip ancestor clip in PaintLayerPainter for squashed layers.

The squashing layer passed a differrent painting root that is not the same
as the backing's space when painting squashed layers, thus a wrong clip
is applied when squashed layers are painted.

This was not a problem prior to GeometryMapper-based PaintLayerClipper
because the non-GM PaintLayerClipper skipped visual rect optimization
in this particular corner case.

BUG= 712783 
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2830013007
Cr-Commit-Position: refs/heads/master@{#467220}
(cherry picked from commit fb28c1b7511ceeb4d47ed42dbeb0676591469fe6)

Review-Url: https://codereview.chromium.org/2843223003 .
Cr-Commit-Position: refs/branch-heads/3071@{#242}
Cr-Branched-From: a106f0abbf69dad349d4aaf4bcc4f5d376dd2377-refs/heads/master@{#464641}

[add] https://crrev.com/840427fa341fe9f413fdf5c7df1941f08e152380/third_party/WebKit/LayoutTests/compositing/squashing/squash-transformed-with-clip-expected.html
[add] https://crrev.com/840427fa341fe9f413fdf5c7df1941f08e152380/third_party/WebKit/LayoutTests/compositing/squashing/squash-transformed-with-clip.html
[modify] https://crrev.com/840427fa341fe9f413fdf5c7df1941f08e152380/third_party/WebKit/Source/core/paint/PaintLayerPainter.cpp

Tested the issue on Latest Chrome Dev# 60.0.3088.3 on Windows, Mac and Linux and is found to be fixed.
Hovering on the text makes the text to shift down on hovering. Hence adding TE-Verified Labels.
Thank You.