909603 - Bogus assertion in nsLayoutUtils::PaintFrame

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[Robert O'Callahan (:roc) (email my personal email if necessary)]

if (ignoreViewportScrolling) {
    nsIDocument* doc = aFrame->GetContent() ?
      aFrame->GetContent()->GetCurrentDoc() : nullptr;
    NS_ASSERTION(!aFrame->GetParent() ||
                 (doc && doc->IsBeingUsedAsImage()),
                 "Only expecting ignoreViewportScrolling for root frames and "
                 "for image documents.");


This assertion is bogus. Features like -moz-element can cause nsLayoutUtils::PaintFrame to be called on any frame while we're doing a drawWindow that paints ignoring viewport scrolling, and there's nothing wrong with that.

Comment 1

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: fix

Comment 2

[Timothy Nikkel (:tnikkel)]

Comment on attachment 795786 [details] [diff] [review]
fix

Nice catch.

Perhaps we should only set ignoreViewportScrolling to true if we are actually going to change our behaviour based on it? (ie if we are painting a root frame)

You also have a hunk adding the PAINT_IGNORE_VIEWPORT_SCROLLING flag to the header, but you don't need it in this patch. Not sure why.

It looks like we also have a similar problem in this function with the usingDisplayPort variable, except that effects behaviour and not just asserts.

Comment 3

[Timothy Nikkel (:tnikkel)]

(In reply to Timothy Nikkel (:tn) from comment #2)
> It looks like we also have a similar problem in this function with the
> usingDisplayPort variable, except that effects behaviour and not just
> asserts.

I created bug 910529 for this.

Comment 4

[Robert O'Callahan (:roc) (email my personal email if necessary)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/82570baff149

Comment 5

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/82570baff149