This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Adding constants to LO_SUM
- To: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
- Subject: Re: Adding constants to LO_SUM
- From: Geoff Keating <geoffk at geoffk dot org>
- Date: 09 Jul 2001 14:50:04 -0700
- CC: gcc-patches at gcc dot gnu dot org
- References: <10107092129.AA07895@vlsi1.ultra.nyu.edu>
kenner@vlsi1.ultra.nyu.edu (Richard Kenner) writes:
> I wouldn't describe GCC's incredibly complex alignment calculations as
> "an object of a mode is aligned to the mode boundary". See
> GET_MODE_ALIGNMENT (which is overridable, although not overriden by
> any contributed port), get_mode_alignment for the default, and
> BIGGEST_ALIGNMENT.
>
> Well, sure, but what I said is still true for nearly all practical purposes.
> For the most part, the only modes for which GET_MODE_ALIGNMENT and
> GET_MODE_BITSIZE differ are the complex modes, and you don't normally see
> MEMs of that mode.
I am working now on a port where BIGGEST_ALIGNMENT is 16, so I'm
pretty sure it covers more than just complex modes. (This port
doesn't use LO_SUM yet, but it could very easily.)
> Then you have to worry about ports where LO_SUM applies to the low 2
> bits of the address, because they're word-addressed...
>
> Do we have such?
No idea.
--
- Geoffrey Keating <geoffk@geoffk.org>