ssa bootstrap problem on x86 (cmpstrsi_1 pattern)

[Richard Henderson]

On Thu, Jul 27, 2000 at 03:28:21PM -0700, Mark Mitchell wrote:
> ... but are we willing to forbid this kind of usage in all machine
> descriptions?

I am.  There are also other alternatives, mostly involving 
encapsulating all the block move/compare/whatever in big
unspecs until we get past the early optimization passes.

The only alternative is to give up on SSA on RTL entirely,
and instead do it on a higher-level IL.  Which might not
be a bad idea anyway...


r~