On Sun, Apr 10, 2011 at 8:22 PM, Laurynas Biveinis <laurynas.bivei...@gmail.com> wrote: > It is certainly true that moving away from GC will make some kinds of > bugs possible again, but I hope that not enough to be an unmanageable > concern.
One thing that may help, is to poison parts of released obstacks, instead of actually releasing the memory. A bit like the poison mechanism of GGC. An obstack memory checking mechanism similar to gcac would also be nice. Something that walks all RTL and verifies that RTX objects are allocated on the obstacks where you'd expect them to be (via _obstack_allocated_p). I don't think GCC had any checking like that 20 years ago... Ciao! Steven
