Did a cursory examination of this, and it looks like mTransparentSurface should not ever really be null here like the stack trace is implying. EnsureTransparentSurface() unconditionally/infallibly allocates a gfxSurface and assigns it. The best guess I have here is that maybe we have either some sort of race condition from OMTP occurring here, or a bad ref-count, or both. CC'ing various people who have stomped around this area of the code and/or are working on OMTP.
OS: Unspecified → Windows 7
Status: NEW → RESOLVED
Last Resolved: 22 hours ago
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.