This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: mandatory named patterns
- To: Alan Lehotsky <apl at alum dot mit dot edu>
- Subject: Re: mandatory named patterns
- From: Michael Meissner <meissner at cygnus dot com>
- Date: Wed, 13 Jun 2001 00:34:59 -0400
- Cc: Bahman Sistany <bsistany at zucotto dot com>, gcc at gcc dot gnu dot org
- References: <3B268238.D407D282@zucotto.com> <p0433010ab74c681cb0e9@[192.168.1.254]>
On Tue, Jun 12, 2001 at 08:48:49PM -0400, Alan Lehotsky wrote:
> At 4:57 PM -0400 6/12/01, Bahman Sistany wrote:
>
> >When doing a port certain named patterns like the movM patterns are
> >mandatory to define. Is there a list of them somewhere? (don't see it in
> >the "Using and porting" doc)
> >Also, what is meant by mandatory? will the compiler abort if they are
> >not defined? (my assumption)
>
> In many cases, the compiler won't COMPILE. There will be places that
> access the operator optables without a HAVE_xxxx guard.
>
> I know for a fact that the compiler won't compile without one of the shift operators, (arithmetic
> right shift, if I remember correctly).
>
> I also doubt the compiler will work if there's not an addM3 pattern, for at least some mode.
And call/call_value/jump/indirect_jump/b<cc> off the top of my head.
--
Michael Meissner, Red Hat, Inc. (GCC group)
PMB 198, 174 Littleton Road #3, Westford, Massachusetts 01886, USA
Work: meissner@redhat.com phone: +1 978-486-9304
Non-work: meissner@spectacle-pond.org fax: +1 978-692-4482