Jeffrey Law - Re: [m68k 05/13] improve 64bit shift handling (original) (raw)

This is the mail archive of the gcc-patches@gcc.gnu.orgmailing list for the GCC project.

• From: Jeffrey Law
• To: Roman Zippel
• Cc: gcc-patches at gcc dot gnu dot org
• Date: Mon, 12 Feb 2007 10:19:37 -0700
• Subject: Re: [m68k 05/13] improve 64bit shift handling
• References: 20070130112615.782382000@linux-m68k.org 20070130114606.866698000@linux-m68k.org <1170786455.4134.70.camel@sweet.slc.redhat.com> <Pine.LNX.4.64.0702072248210.14457@scrub.home>
• Reply-to: law at redhat dot com

On Wed, 2007-02-07 at 23:04 +0100, Roman Zippel wrote:

Hi,

On Tue, 6 Feb 2007, Jeffrey Law wrote:

1. You need a function comment before split_di. Also in split_di s/refuse/refuses/

Hmm, I copied that from i386. :) Whomever added it to i386.c should have included a comment. When you cobble a comment together, could you please add it to i386.c as well (consider it pre-approved).

1. You modified extendsidi2, but it's not mentioned in the ChangeLog. DId you mean to include it? (Arguably it could have been in its own patch).

Yes, I'll add it. (It sort of belongs to ashrdi_const32, as generates the same code sequence.) OK. Thanks.

1. Do you need a CC_STATUS_INIT in your new *ashrdi3_const1? Similarly for the other shift by one patterns.

2. Are the CC status bits correct when we split something like a 32bit shift into moves?

All CC status bits for any shift operation are always cleared by notice_update_cc(). OK.

Based on your replies, I think you can go ahead and install the patch (after adding the missed comment and ChangeLog entry of course.)

Thanks, jeff

• References:
  • Re: [m68k 05/13] improve 64bit shift handling
    * From: Jeffrey Law
  • Re: [m68k 05/13] improve 64bit shift handling
    * From: Roman Zippel