[PATCH,RFC] CSE path following on basic blocks

[Steven Bosscher]

On 11/28/06, Eric Botcazou <ebotcazou@libertysurf.fr> wrote:
> > Because I don't understand why we would write past the end of
> > qty_table.  It leaves me with a feeling that, given the right test
> > case, we could even trigger that problem if we use the 15 year old
> > trick.
> >
> > But maybe I'm being too conservative.  It's a known shortcoming in
> > most aerospace engineers ;-)
>
> It would seem to me that keeping a 15-year old trick is second only to
> replacing it with a 16-year old trick, as far as conservatism is concerned,
> no? :-)

Actually, not in this case.

The reason for the 15 year old trick is not to prevent memory
corruption, but to allow CSE to continue traversing the insns list
after beyond the last insn that was scanned by cse_end_of_basic_block,
when CSE did something on the path that allows it to continue
(ignoring e.g. the label, or whatever).

The reason for my new trick is that I can be fairly sure that there
are never more QTYs than nsets*MAX_RECOG_OPERANDS.

I could revert to the old trick, but don't hold me responsible if the
bug ever shows up on a real test case! :-D

Gr.
Steven