1357869 - Assertion failure: !(mHintsHandledByAncestors & nsChangeHint_ReconstructFrame) (why restyle descendants if we are reconstructing the frame for an ancestor?), at layout/base/GeckoRestyleManager.cpp:1179

Status: RESOLVED FIXED

(Core :: CSS Parsing and Computation, defect)

Description

[Bob Clary [:bc] (inactive)]

1. http://www.warhammer-forum.com/index.php?/topic/70920-comtes-vampires-perso-le-duc-rouge/&page=3#comment-3051481

2. Assertion failure: !(mHintsHandledByAncestors & nsChangeHint_ReconstructFrame) (why restyle descendants if we are reconstructing the frame for an ancestor?), at /home/worker/workspace/build/src/layout/base/GeckoRestyleManager.cpp:1179
#01: mozilla::ElementRestyler::RestyleSelf [layout/base/GeckoRestyleManager.cpp:2463]
#02: mozilla::ElementRestyler::Restyle [layout/base/GeckoRestyleManager.cpp:1844]
#03: mozilla::ElementRestyler::RestyleContentChildren [layout/base/GeckoRestyleManager.cpp:3448]
#04: mozilla::ElementRestyler::RestyleChildren [layout/base/GeckoRestyleManager.cpp:2973]
#05: mozilla::ElementRestyler::Restyle [layout/base/GeckoRestyleManager.cpp:1999]
...

Windows/Linux Nightly.

Assertion introduced in Bug 1301258

Comment 1

[Cameron McCormack (:heycam)]

I was able to reproduce this one locally, intermittently.  The issue is that the assertions check that we don't waste time restyling descendants of elements that we already recorded a ReconstructFrame change hint for.  But the current code does do that if ElementRestyler::Restyle picks a ReconstructFrame hint out of the RestyleTracker (as opposed to generating it from a CalcDifference call) and then proceeds to call RestyleSelf, which then restyles a child provider frame.

In theory we should be able to avoid calling RestyleSelf altogether in this case, but for now I'll just remove the assertion rather than spend much effort on Gecko's restyle code.

Comment 3

[Cameron McCormack (:heycam)]

Attachment: Bug 1357869 - Don't assert when we restyle a child provider frame when an ancestor already handled nsChangeHint_ReconstructFrame.

Review commit: https://reviewboard.mozilla.org/r/131960/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/131960/

Comment 4

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

Comment on attachment 8859903 [details]
Bug 1357869 - Don't assert when we restyle a child provider frame when an ancestor already handled nsChangeHint_ReconstructFrame.

https://reviewboard.mozilla.org/r/131960/#review135550

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/d2a6304ad598
Don't assert when we restyle a child provider frame when an ancestor already handled nsChangeHint_ReconstructFrame. r=dbaron

Comment 6

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/d2a6304ad598