Segfaults building Rakudo

Vasily Chekalkin bacek at
Sat Apr 16 02:03:00 UTC 2011


On Sat, Apr 16, 2011 at 3:21 AM, Moritz Lenz <moritz at> wrote:
> FWIW on my machine, all myrakudo spectest runs based on that branch have
> been successful, with and without --optimize.

Ok. Branch is kind of finished in ready for extensive testing and codereview.

Main changes:
1. Change Parrot_PackFile typedef to use PMC.
2. Change functions to work with new typedef.

Current behavior in master:
1. We create PackFile*.
2. Wrap it in PtrObj.
3. Store raw pointer inside interp->initial_pf.
4. Update it directly without writebarriering of PtrObj.
5. Segfaults and fireworks.

Behavior in branch:
1. Create PackFile*
2. Wrap it in PtrObj.
3. Store in interp->current_pf.
4. Always WB on access.
5. (probably) Slow down everything.


More information about the parrot-dev mailing list