This is the mail archive of the
mailing list for the GCC project.
Re: [RFC] type promotion pass
- From: Segher Boessenkool <segher at kernel dot crashing dot org>
- To: Michael Clark <michaeljclark at mac dot com>
- Cc: Prathamesh Kulkarni <prathamesh dot kulkarni at linaro dot org>, David Edelsohn <dje dot gcc at gmail dot com>, "William J. Schmidt" <wschmidt at linux dot vnet dot ibm dot com>, Kugan <kugan dot vivekanandarajah at linaro dot org>, Richard Biener <rguenther at suse dot de>, "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>
- Date: Fri, 15 Sep 2017 15:59:43 -0500
- Subject: Re: [RFC] type promotion pass
- Authentication-results: sourceware.org; auth=none
- References: <CAAgBjM=4tTO9G-4AxTJ3cs6L-EHCEeAB-PP8OKN7Z6a=d9Ly1A@mail.gmail.com> <CAGWvnynp5gNrU+WLMNOKZX5R69-PS-oKxOQ9t4QOB6Au+B7qtg@mail.gmail.com> <7B5FDF38-049C-4535-94AC-0EA28C369735@mac.com>
On Sat, Sep 16, 2017 at 08:47:03AM +1200, Michael Clark wrote:
> RISC-V defines promote_mode on RV64 to promote SImode to signed DImode subregisters. I did an experiment on RISC-V to not promote SImode to DImode and it improved codegen for many of my regression test cases, but unfortunately it breaks the RISC-V ABI.
It sounds like you should implement TARGET_PROMOTE_FUNCTION_MODE as well?