Your flow.c change on 4/8

[Richard Henderson]

On Mon, Apr 10, 2000 at 06:07:13PM -0400, Richard Kenner wrote:
> It looks wrong to me and is causing problems.  Suppose we have an insn
> that sets a single register but uses a multi-word register including
> one of those.  It will return 1 in this case and prevent a REG_DEAD
> note from being made.

Yes, I know.  Geoff's PPC regression testing showed this up 
pretty well (though it took me a day and a half to get a PPC
cross-compiler built).

I'll get it fixed this evening.


r~