On Fri, 2011-11-04 at 12:25 -0600, Jeff Law wrote: > The only way I can think of to have two pseudos assigned the same hard > reg at the same point in the insn stream is if the two pseudos are > known to have the same value.
Having the same value is the more common way two overlapping pseudos don't conflict, but if one or both of the pseudos are undefined (ie, no definition point), then they won't conflict either. > I'm don't think IRA implements this particular optimization -- I vaguely > recall Vlad talking about it and mentioning that it wasn't worth the > implementation effort. I'm surprised and saddened to head we don't at least treat copies specially wrt computing conflicts. That said, if one or both of the pseudos are undefined, then that could explain why two pseudos are allocated to the same hard register at the same time even with the code we have now. Peter
