1176969 - part of 3d transform animated chessboard updates only when mouse moves, if OMTA (off-main-thread/async animations) enabled

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[arni2033]

Attachment: Async Animation - The spin animation twitches on this page.webm

1. Open URL above
2. See the choppy animation

Note 1:  If layers.offmainthreadcomposition.async-animations = false then animation is fine

Note 2:  If I start constantly moving mouse, animation becomes almost as smooth as on Release version (that has async-animations = false).

Comment 1

[Markus Stange [:mstange]]

I can reproduce this.

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: simple testcase

Simple testcase, based on poking around at the demo in inspector.

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: variant testcase

This is a variant with an OMTA-able animation on the child.

In this variant the child never moves at all.

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

from #layout, just now:
[2015-06-29 15:02:31 -0700] <dbaron> mattwoodrow, I'm trying to remember how preserve-3d works, especially as it relates to the layer tree and transforms set on layers
[2015-06-29 15:03:11 -0700] <dbaron> mattwoodrow, is it correct that if there are multiple elements in a preserve-3d scene (say, an element with preserve-3d that's the root of the 3d scene and a child of that element)
[2015-06-29 15:03:29 -0700] <mattwoodrow> dbaron: nsDisplayTransform accumulates the transforms from the ‘leaf’ element in the scene up to the root
[2015-06-29 15:03:31 -0700] <dbaron> mattwoodrow, then if both of those elements have layers, the layers will not be parent and child
[2015-06-29 15:03:40 -0700] <mattwoodrow> yes
[2015-06-29 15:03:48 -0700] <mattwoodrow> thinker is working on changing that though
[2015-06-29 15:04:12 -0700] <dbaron> mattwoodrow, so do you agree that this implies that we can't do OMT animations of transform on either the parent or the child
[2015-06-29 15:04:39 -0700] <dbaron> mattwoodrow, since if we do an OMT animation of transform on the parent, we'd fail to update the child during the animation
[2015-06-29 15:04:58 -0700] <dbaron> mattwoodrow, and if we do an OMT animation of transform on the child, we'll simply give it the wrong transform because we won't do the accumulation
[2015-06-29 15:05:09 -0700] <mattwoodrow> dbaron: yes, exactly
[2015-06-29 15:05:19 -0700] <dbaron> mattwoodrow, ok, good
[2015-06-29 15:05:34 -0700] <dbaron> mattwoodrow, I think I need to change the && to an ||, then, and that fixes the bug I'm looking at
[2015-06-29 15:05:49 -0700] <dbaron> mattwoodrow, hopefully there isn't something in B2G depending on this for perf...
[2015-06-29 15:07:02 -0700] <dbaron> (this is 1176969, if you're interested)
[2015-06-29 15:07:52 -0700] <mattwoodrow> dbaron: Which bit of code are you looking at?
[2015-06-29 15:08:24 -0700] <dbaron> mattwoodrow, in layout/style/AnimationCommon.cpp, AnimationCollection::CanAnimatePropertyOnCompositor
[2015-06-29 15:08:45 -0700] <dbaron> currently frame->Preserves3D() && frame->Preserves3DChildren()
[2015-06-29 15:10:12 -0700] <mattwoodrow> I agree that that is the right fix :)
[2015-06-29 15:10:56 -0700] <dbaron> ok, I'll write a reftest and then upload...

Comment 5

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=6b16fa5413a0

Comment 6

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: Disable OMT animation for any frame in a preserve-3d scene rather than only frames whose parent and child are in a preserve-3d scene

frame->Preserves3D() is whether the frame's parent has transform-style:
preserve-3d, which means that the frame is part of the same 3-D scene as
its parent.  frame->Preserves3DChildren() is whether the frame itself
has transform-style: preserve-3d, which means that the frame is part of
the same 3-D scene as its children.

Neither of these are valid cases for doing off-main-thread (OMT)
animation because all of the layers in a preserve-3d scene are currently
siblings of each other, rather than preserving ancestor/descendant
relationships.  This means that it's not valid to animate transform of
the parent on the compositor because the compositor animation won't
update any of its children that have layers.  Likewise, it's not valid
to animate transform of the child on the compositor because the code
that sends transform information to the compositor doesn't handle the
accumulation of transforms needed to get the "right" transform for the
child (i.e., with the transforms of its ancestors up to the top of the
3-D scene merged in).

This means that we do OMT animation for slightly fewer cases with the
patch than we did without the patch.  This means it's pretty low risk in
terms of correctness, although there's a chance it might regress
performance on one of the (somewhat limited) set of cases where the
optimization was valid.  (Bug 779598 covers doing OMT animation for
preserve-3d cases, and depends on the work ongoing in bug 1097464.)

The animate-preserve3d-parent.html reftest doesn't fail without the
patch, since something seems to invalidate in the test; it was testing
the testcase that showed correct behavior when the mouse was moving, so
this isn't incredibly surprising (although that invalidation from mouse
movement is itself worth debugging).  The animate-preserve3d-child.html
test does fail without the patch, though.

(With an initial transform of none instead of the 30deg transform, both
tests also show an invalidation bug without the patch.)

Both tests require a small amount of intermittent fuzz with the patch;
this should be investigated.

Comment 7

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

(I'll need to adjust the test to deal with the intermittents on Android -- or maybe just give up.  It seems to be a little tricky to write a reliable automated test for this.)

Comment 8

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/ed913a061dd0

Comment 9

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/58274d19225e

Comment 10

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/ed913a061dd0
https://hg.mozilla.org/mozilla-central/rev/58274d19225e

Comment 11

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Comment on attachment 8627443 [details] [diff] [review]
Disable OMT animation for any frame in a preserve-3d scene rather than only frames whose parent and child are in a preserve-3d scene

Approval Request Comment
[Feature/regressing bug #]: OMT Animations
[User impact if declined]: animations that are wildly incorrect
[Describe test coverage new/current, TreeHerder]: reftest in patch
[Risks and why]: low risk; it disables OMT animation in a case where it's not yet valid to use, by slightly broadening the preserve-3d cases in which it's disabled
[String/UUID change made/needed]: no

Comment 12

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/a93bd6a0b911

Comment 13

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/cb4cdfca64a2

Comment 14

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/a93bd6a0b911
https://hg.mozilla.org/mozilla-central/rev/cb4cdfca64a2

Comment 15

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

After disabling the reftest last night, I'm going to reenable it based on this try run:
https://treeherder.mozilla.org/#/jobs?repo=try&revision=88819f29a560
(I'm fixing the one orange in that run by bumping the fuzz numbers on mac.)

Comment 16

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/2f29a1c8f49e

Comment 17

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/2f29a1c8f49e

Comment 18

[Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)]

Comment on attachment 8627443 [details] [diff] [review]
Disable OMT animation for any frame in a preserve-3d scene rather than only frames whose parent and child are in a preserve-3d scene

Approving for uplift to Aurora. I see that we added tests to verify this fix which is great! Also, this patch has been in m-c for a few days now so it seems stable enough to uplift to Aurora.

Comment 19

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/dff3b0967443