direct return optimization

[Richard Henderson]

On Tue, Feb 08, 2000 at 05:13:05PM -0700, Jeffrey A Law wrote:
> A decrement-and-branch with the return register can certainly happen on
> the PA.

True, but the case under consideration is a forward conditional branch
to the next instruction.  I think it unlikely that we'll ever generate
a forward decr-and-branch.  Not impossible, mind you, which is why I
acked Clint's patch, but unlikely.


r~