This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] gcse.c: Fix the header ordering.
- From: Zack Weinberg <zack at codesourcery dot com>
- To: Roger Sayle <roger at www dot eyesopen dot com>
- Cc: Kazu Hirata <kazu at cs dot umass dot edu>, <graham dot stott at btinternet dot com>, <gcc-patches at gcc dot gnu dot org>
- Date: Fri, 31 Jan 2003 23:09:04 -0800
- Subject: Re: [patch] gcse.c: Fix the header ordering.
- References: <Pine.LNX.4.44.0301312300450.12006-100000@www.eyesopen.com>
Roger Sayle <roger@www.eyesopen.com> writes:
> On Fri, 31 Jan 2003, Zack Weinberg wrote:
>> rtl.h is supposed to be machine independent - that is, it should have
>> no dependencies on tm.h or any insn-*.h. The CC0_P macro needs to be
>> moved to a more appropriate location.
>
> Whilst I agree with your ideal of making rtl.h machine independent,
> the current situation is (and has been for a while) that rtl.h has
> multiple dependencies on target headers, and that these dependencies
> aren't explicit is causing real failures in the compiler.
Sure - but let's not make it worse, okay?
> For example, if you look at the definition of the enumeration
> global_rtx_index, you'll notice that its littered with
> "#if FRAME_POINTER_REGNUM == ARG_POINTER_REGNUM", and so on.
This shouldn't be the same category of problem, since both these
macros are normally defined in tm.h, not insn-config.h. I want them
out of there too, though.
zw