[Mono-dev] Ephemeron support for sgen

[Mark Probst]

On Tue, Apr 27, 2010 at 5:41 PM, Rodrigo Kumpera <kumpera at gmail.com> wrote:
> The problem with old gen arrays is due to their content not been marked it
> might end up pointing to freed memory.

Maybe I misunderstand, but by scanning the remsets like usual all
nursery objects reachable from old gen arrays would be considered live
and thus promoted.

> But it makes sense that during minor gc we scan them only once and just
> assume that they point to reachable memory.

That's the point of the write barriers - no special casing needed.

Mark