Jonathan Watt [:jwatt] Assignee
	Description • 19 years ago

The call to GetContent could return null, in which case NS_RELEASE dereference null.

	Boris Zbarsky [:bzbarsky]
	Comment 2 • 19 years ago

> The call to GetContent could return null

Not really.  The code is:

 186                       if (GetContent()) {
 187                         nsIContent* oldVal = GetContent();

Now perhaps we want to only call GetContent() once and then just test oldVal?  Would probably make coverity happier and mean only two conditional tests there instead of the current 3....

I'd really have expected coverity to not flag this, though.  Is there a way to report to them that this is bogus?  Does flagging the report as "FALSE" in scan.coverity.com do that?

	Jonathan Watt [:jwatt] Assignee
	Comment 3 • 19 years ago

True, I should have noticed that. :-/ Nevertheless, there's still a leak on OOM that we should fix.

	Jonathan Watt [:jwatt] Assignee
	Comment 4 • 19 years ago

I don't know if there's a special way to report false positives to coverity. I'd hope they keep looking at the reports marked FALSE though to improve their code.

	Boris Zbarsky [:bzbarsky]
	Comment 5 • 19 years ago

Comment on attachment 218133 [details] [diff] [review]
patch for leak

SetContent addrefs, so this will double-addref the content and leak.

	Boris Zbarsky [:bzbarsky]
	Comment 6 • 19 years ago

Comment on attachment 218133 [details] [diff] [review]
patch for leak

Never mind me.  I need to learn to read.

	timeless
	Comment 7 • 19 years ago

fwiw they do read through FALSEs, but if there's a specific pattern we want them to learn, we're supposed to email something.

	Jonathan Watt [:jwatt] Assignee
	Comment 8 • 19 years ago

Checked in.