Re: [m68k] Ping: Add ffs pattern for ColdFire 528x using the ff1insn

[Bernardo Innocenti <bernie at develer dot com>]

Richard Henderson wrote:
> On Mon, Aug 23, 2004 at 08:09:08PM +0200, Bernardo Innocenti wrote:
>
> > http://gcc.gnu.org/ml/gcc-patches/2004-06/msg01592.html
>
>
> I'd prefer these patterns be reworked so that you use "bfffo"
> when appropriate.

Thank you, I've added Peter to the Cc list so he also knows.


Also, here are these other unreviewed patches I'd like to
see in 3.5.  This is a missing piece of a patch that already
went into mainline:

 [m68k] Add 'R' constraint for ColdFire V4 mov3q.l instruction
 http://gcc.gnu.org/ml/gcc-patches/2004-07/msg02607.html


Those I've not tested, but look like useful fixes/improvements
to me:

 m68k: fix 20020720-1
 http://gcc.gnu.org/ml/gcc-patches/2004-07/msg01350.html

 m68k: fix builtins/memset.c
 http://gcc.gnu.org/ml/gcc-patches/2004-07/msg01351.html

 m68k: improve zero_extend patterns
 http://gcc.gnu.org/ml/gcc-patches/2004-07/msg02238.html

 m68k: various cleanups
 http://gcc.gnu.org/ml/gcc-patches/2004-07/msg02237.html

I've tested this one, but it causes an ICE with m68k-uclinux
(see my reply in the thread).  I'd very much like to see the hand-made
prologue/epilogue code go away from the m68k backend:

 m68k: add prologue/epilogue rtl exapnder
 http://gcc.gnu.org/ml/gcc-patches/2004-08/msg00564.html


This one you have already reviewed, but there was an issue
with the peephole2 part.  I've asked Peter to resubmit it
because, without it, the uClinux toolchain wouldn't bootstrap:

 http://gcc.gnu.org/ml/gcc-patches/2004-05/msg00182.html
 gcc-3.4.0 fails for ColdFire(does not satisfy constraints)

--
 // Bernardo Innocenti - Develer S.r.l., R&D dept.
\X/  http://www.develer.com/