This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: New __atomic builtins generating an unwanted/unneeded stack slot
- From: Richard Henderson <rth at redhat dot com>
- To: Richard Biener <richard dot guenther at gmail dot com>
- Cc: Frederic Riss <frederic dot riss at gmail dot com>, gcc <gcc at gcc dot gnu dot org>
- Date: Tue, 29 Oct 2013 09:08:47 -0700
- Subject: Re: New __atomic builtins generating an unwanted/unneeded stack slot
- Authentication-results: sourceware.org; auth=none
- References: <CAKk_3s8cfPM+jsD2zudN+rHQXNs2T4xW21507WeV0c_=kMHrpg at mail dot gmail dot com> <526EAE14 dot 40603 at redhat dot com> <CAFiYyc2kSDSDnyvy7SNrVhVSymjV8PX-C8e3Fwq-Y5qSgyULYg at mail dot gmail dot com>
On 10/29/2013 03:06 AM, Richard Biener wrote:
> On Mon, Oct 28, 2013 at 7:33 PM, Richard Henderson <rth@redhat.com> wrote:
>> On 10/28/2013 02:25 AM, Frederic Riss wrote:
>>> Is there a clean way to have the compiler discard the unneeded stack slot?
>>
>> Not yet. There is a rewrite of the atomic support in gcc to
>> move away from using builtins, which will allow two outputs
>> to be ssa allocacted. But this will not be complete for 4.9.
>
> Though if the expander is ok with not having the stack slot and the
> result is not required we could fold this argument to NULL ...
FWIW, the md interface returns two registers.
r~