Bug 199506

Summary:

[ContentChangeObserver] didFinishTransition triggers a nested style recalc via isConsideredClickable

Product:

WebKit

Reporter:

zalan <zalan>

Component:

Layout and Rendering

Assignee:

zalan <zalan>

Status:

RESOLVED FIXED

Severity:

Normal

CC:

bfulgham, commit-queue, koivisto, rniwa, simon.fraser, webkit-bug-importer, zalan

Priority:

Keywords:

InRadar

Version:

WebKit Nightly Build

Hardware:

Unspecified

OS:

Unspecified

Attachments:

Description	Flags
Patch	none
Patch	none
Patch	none

Description zalan 2019-07-04 16:19:44 PDT

<rdar://problem/52656221>

Comment 1 zalan 2019-07-04 16:57:01 PDT

Created attachment 373475 [details]
Patch

Comment 2 zalan 2019-07-04 20:54:12 PDT

Created attachment 373476 [details]
Patch

Comment 3 zalan 2019-07-04 21:06:29 PDT

Created attachment 373477 [details]
Patch

Comment 4 zalan 2019-07-04 21:06:51 PDT

^^more reliable test case

Comment 5 Antti Koivisto 2019-07-04 21:42:04 PDT

Another option would be to add a parameter to willRespondToMouseClickEvents that prevents style update. computeEditability() where the nested recalc is entered already takes such parameter.

Comment 6 zalan 2019-07-04 21:50:31 PDT

(In reply to Antti Koivisto from comment #5)
> Another option would be to add a parameter to willRespondToMouseClickEvents
> that prevents style update. computeEditability() where the nested recalc is
> entered already takes such parameter.
There's another isConsideredClickable() callsite (which I can't make async) and I was planning to do exactly that in there.

Comment 7 WebKit Commit Bot 2019-07-04 22:44:23 PDT

Comment on attachment 373477 [details]
Patch

Clearing flags on attachment: 373477

Committed r247147: <https://trac.webkit.org/changeset/247147>

Comment 8 WebKit Commit Bot 2019-07-04 22:44:25 PDT

All reviewed patches have been landed.  Closing bug.

Comment 9 Radar WebKit Bug Importer 2019-07-04 22:45:18 PDT

<rdar://problem/52669516>

Comment 10 Ryosuke Niwa 2019-07-05 01:10:27 PDT

(In reply to zalan from comment #6)
> (In reply to Antti Koivisto from comment #5)
> > Another option would be to add a parameter to willRespondToMouseClickEvents
> > that prevents style update. computeEditability() where the nested recalc is
> > entered already takes such parameter.
> There's another isConsideredClickable() callsite (which I can't make async)
> and I was planning to do exactly that in there.

Is that guaranteed to be correct? computeEditability needs the up-to-date style. In cases where computeEditability is called and we don't end up updating the style due to this flag, is the style computeEditability uses guaranteed to be up-to-date?

Comment 11 Antti Koivisto 2019-07-05 04:42:43 PDT

Presumably the heuristics would trigger when the style actually changes, if needed.

Comment 12 zalan 2019-07-06 06:14:29 PDT

Actually the other callsite is at RenderTreeUpdater::updateElementRenderer() (ContentChangeObserver::StyleChangeScope -> ~StyleChangeScope). I assume by the time we get here all the styles are resolved and it's safe to call computeEditability. Antti?

Comment 13 Ryosuke Niwa 2019-07-08 19:36:41 PDT

(In reply to zalan from comment #12)
> Actually the other callsite is at RenderTreeUpdater::updateElementRenderer()
> (ContentChangeObserver::StyleChangeScope -> ~StyleChangeScope). I assume by
> the time we get here all the styles are resolved and it's safe to call
> computeEditability. Antti?

Why can't the code in StyleChangeScope::~StyleChangeScope() be pushed as a post style update task??

Comment 14 Antti Koivisto 2019-07-09 05:09:22 PDT

(In reply to zalan from comment #12)
> Actually the other callsite is at RenderTreeUpdater::updateElementRenderer()
> (ContentChangeObserver::StyleChangeScope -> ~StyleChangeScope). I assume by
> the time we get here all the styles are resolved and it's safe to call
> computeEditability. Antti?

Yes, it should be safe. Style has just been updated by updateElementRenderer when  ~StyleChangeScope runs. Cutting close though.