This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH], PR 71977/70568/78823: Improve PowerPC code that uses SFmode in unions
- From: Michael Meissner <meissner at linux dot vnet dot ibm dot com>
- To: David Edelsohn <dje dot gcc at gmail dot com>
- Cc: Michael Meissner <meissner at linux dot vnet dot ibm dot com>, GCC Patches <gcc-patches at gcc dot gnu dot org>, Segher Boessenkool <segher at kernel dot crashing dot org>, Bill Schmidt <wschmidt at linux dot vnet dot ibm dot com>
- Date: Wed, 4 Jan 2017 16:00:14 -0500
- Subject: Re: [PATCH], PR 71977/70568/78823: Improve PowerPC code that uses SFmode in unions
- Authentication-results: sourceware.org; auth=none
- References: <20161230002406.GA15372@ibm-tiger.the-meissners.org> <CAGWvny=mOURYWSdRUVXXGvUNvJBhyU01u4qiUjm4i9RKadL9nQ@mail.gmail.com>
On Wed, Jan 04, 2017 at 02:17:16PM -0500, David Edelsohn wrote:
> The change to rs6000_emit_move() really should have been in a helper
> function. We have to stop adding to the complexity of the function. I
> won't insist that you split it out, but any addition of more than a
> few lines in rs6000_emit_move(), prologue or epilogue should be placed
> into a separate function.
I can certainly move it into a static helper function (the code in
rs6000_emit_move is only 20 lines once you eliminate the comments), but did you
want me to move some of the other special case parts that are in the same
region to their own helper functions? I suspect this should be a GCC 8 work
item, but since I'm already moving stuff around, it would be simple to do it
now while I'm in the code.
These include support for block moves, thread local moves, splitting up 128-bit
IBM double double constants, and all of the decimal reload gunk.
> Okay with those changes. This change is too invasive for GCC 6
> backport at this late phase of GCC 6 release.
>
> We'll see if Segher catches anything additional when he wants some
> light reading. :-)
>
> Thanks, David
>
--
Michael Meissner, IBM
IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA
email: meissner@linux.vnet.ibm.com, phone: +1 (978) 899-4797