Re: RFA: Fix rtl-optimization/22258

[Bernd Schmidt <bernds_cb1 at t-online dot de>]

Joern RENNECKE wrote:
> Bernd Schmidt wrote:
>
> >    Example?  Why isn't this caught by cant_combine_insn_p?
>
>
>
> Look at the rtl dumps in the PR.  none of the combined insns mentions
> a CLASS_LIKELY_SPILLED_P register.  i3 is insn 31, which is
> preceded by insn 30.  insn30 is not included in the combination attempt,
> put it sets fr0, which is CLASS_LIKLEY_SPILLED_P.  insn 30 and
> insn 31 (and to some extent the clobber in insn 29) together implement
> the return value copy.

Ok, I see.  I must say I'm not very fond of introducing hard reg 
references into complex insns during combine; I'd prefer 
cant_combine_insn_p to just reject any moves involving hardregs.  My 
initial version of that function did that, but in
   http://gcc.gnu.org/ml/gcc-patches/2002-05/msg01216.html
Dale Johannesen claims that restricting the test to 
CLASS_LIKELY_SPILLED_P is a win on PPC.  I'm curious why.


Bernd