This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [cond-optab] Convert frv
- From: Michael Meissner <meissner at linux dot vnet dot ibm dot com>
- To: Paolo Bonzini <bonzini at gnu dot org>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Mon, 13 Apr 2009 16:55:03 -0400
- Subject: Re: [cond-optab] Convert frv
- References: <E1Lml40-0002Nm-Fs@fencepost.gnu.org>
On Thu, Mar 26, 2009 at 04:35:48AM -0400, Paolo Bonzini wrote:
> FRV is rather complex for an embedded target, but didn't pose
> any problem.
>
> Paolo
>
> * config/frv/frv.md (cbranchsi4, cbranchsf4, cbranchdf4,
> cstoresi4, cstoresf4, cstoredf4): New.
> (cmpdi, cmpsi, cmpsf, cmpdf, bCC, sCC): Remove.
> * config/frv/frv-protos.h (frv_emit_cbranch, frv_emit_scc):
> Receive the entire operands array.
> * config/frv/frv.h (frv_compare_op0,
> frv_compare_op1): Delete.
> * config/frv/frv.c (frv_compare_op0,
> frv_compare_op1): Delete.
> * config/frv/frv-protos.h (frv_emit_cbranch, frv_emit_scc):
> Get test/op0/op1 from the operands array.
> (frv_emit_cond_move): Get test/op0/op1 from the test_rtx.
Your patches look rather straight forward.
I recall when working on the FR-V a few years ago, one of the problems with the
chip was when reload wanted to spill the CR registers. It might be a concern
if your patches make reload more prone to extend the lifetime of cr registers.
--
Michael Meissner, IBM
4 Technology Place Drive, MS 2203A, Westford, MA, 01886, USA
meissner@linux.vnet.ibm.com