This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: rtx_equal_function_value_matters
- To: "David S. Miller" <davem at dm dot cobaltmicro dot com>
- Subject: Re: rtx_equal_function_value_matters
- From: Jeffrey A Law <law at cygnus dot com>
- Date: Wed, 25 Nov 1998 00:26:45 -0700
- cc: rth at cygnus dot com, jj at sunsite dot ms dot mff dot cuni dot cz, egcs at cygnus dot com, crux at pool dot informatik dot rwth-aachen dot de
- Reply-To: law at cygnus dot com
In message <199811250721.XAA31712@dm.cobaltmicro.com>you write:
> Date: Tue, 24 Nov 1998 11:37:45 -0700
> From: Jeffrey A Law <law@hurl.cygnus.com>
>
> I suspect creating new pseudos after cse_not_expected doesn't win anythi
> ng
> at all except for the cases where we create them during regclass.
>
> See my other mail, consider instruction scheduling.
That is precisely the case I was considering -- I was assuming the pseudos
created via the gen_move_insn calls in regclass were going to be useful in
some manner that helped scheduling.
I can't think of any other cases where creating a pseudo after cse_not_expected
is true could be a win. Or more correctly there aren't cases I'm aware of
between the time we set cse_not_expected and the start of the first life pass
where we create new pseudos, except for the regclass stuff.
What cases are you thinking of?
jeff