6 Mar
2010
6 Mar
'10
6:50 p.m.
No, that was not the debug I expected. I think I confused this with the GC_CYCLE_DEBUG, but that one won't be helpful here. I'm sorry for the confusion.
Looking around in the code some more I actually think the C stack might provide some clues (didn't think so before because of the queue system in the mark pass). Could you please post a couple of gdb backtraces, preferably but not necessarily together with the stderr log?