ssa bootstrap problem on x86 (cmpstrsi_1 pattern)

[Richard Henderson]

On Thu, Jul 27, 2000 at 01:10:11PM -0700, Mark Mitchell wrote:
> I think we need to bite the bullet and get the CLOBBERs out of there;
> otherwise, we are forced to do some kind of dataflow anlaysis in SSA...

Not at all.  We have a register set, once, to an unspecified value.
That unspecified value is "valid" everywhere, just as with a register
set to a useful value.



r~