Noticed this while trying to construct a test-case for bug 172753. In Element::resolveComputedStyle: elementsRequiringComputedStyle.prepend(&ancestor); if (auto* existingStyle = ancestor.existingComputedStyle()) { computedStyle = existingStyle; break; } I think that we resolve style when not needed for that ancestor we find an existing computed style for... Also, I think, that we resolve the first element inheriting from itself, which seems wrong (though I don't think it's observable, fwiw)... Will post a patch here to run through CI, and will update for review when needed.
Created attachment 314594 [details] Patch
Nice find, r=me
Comment on attachment 314594 [details] Patch Clearing flags on attachment: 314594 Committed r219138: <http://trac.webkit.org/changeset/219138>
All reviewed patches have been landed. Closing bug.