<div dir="ltr"><br><div class="gmail_extra"><br><div class="gmail_quote">On Tue, May 16, 2017 at 8:28 AM, Sebastian Huber <span dir="ltr"><<a href="mailto:sebastian.huber@embedded-brains.de" target="_blank">sebastian.huber@embedded-brains.de</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">FYI<br>
<br>
I guess for RTEMS we should use "powerpcspe-rtems*-*". Is it possible to use the RTEMS "powerpc" directories with such a target? We had an "arm-rtemseabi*" maybe due to some configure/automake limitations. So, maybe "powerpc-rtemsspe*"?<br></blockquote><div><br></div><div>I personally think the GCC discussions which put SPE as part of the OS name are</div><div>horribly incorrect. SPE is an architecture variant and the pattern for configure </div><div>triples is very well defined. The pattern is ARCHITECTURE-VENDOR-OS[version]</div><div><br></div><div>I would prefer powerpcspe-rtemsVERSION.</div><div><br></div><div>On the sharing the code issue, how much gets shared? How do you envision</div><div>this impacting the RTEMS tree? Split the PowerPC port like GCC? Or just </div><div>somehow magically build the same powerpc directories two different ways?</div><div><br></div><div>--joel</div><div><br></div><div>--joel </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
-------- Forwarded Message --------<br>
Subject: [PATCH 0/3] Split off powerpcspe from rs6000 port<br>
Date: Mon, 15 May 2017 20:51:49 +0000<br>
From: Segher Boessenkool <<a href="mailto:segher@kernel.crashing.org" target="_blank">segher@kernel.crashing.org</a>><br>
To: <a href="mailto:gcc-patches@gcc.gnu.org" target="_blank">gcc-patches@gcc.gnu.org</a><br>
CC: Segher Boessenkool <<a href="mailto:segher@kernel.crashing.org" target="_blank">segher@kernel.crashing.org</a>>, David Edelsohn <<a href="mailto:dje.gcc@gmail.com" target="_blank">dje.gcc@gmail.com</a>>, Andrew Jenner <<a href="mailto:andrew@codesourcery.com" target="_blank">andrew@codesourcery.com</a>>, Arnaud Charlet <<a href="mailto:charlet@adacore.com" target="_blank">charlet@adacore.com</a>>, Joel Brobecker <<a href="mailto:brobecker@adacore.com" target="_blank">brobecker@adacore.com</a>>, Joel Sherrill <<a href="mailto:joel.sherrill@oarcorp.com" target="_blank">joel.sherrill@oarcorp.com</a>>, Joseph Myers <<a href="mailto:joseph@codesourcery.com" target="_blank">joseph@codesourcery.com</a>>, Olivier Hainque <<a href="mailto:hainque@adacore.com" target="_blank">hainque@adacore.com</a>>, Sandra Loosemore <<a href="mailto:sandra@codesourcery.com" target="_blank">sandra@codesourcery.com</a>><br>
<br>
<br>
<br>
Hi!<br>
<br>
As discussed before, here is a series to split powerpcspe from the<br>
rs6000 port. This series does not yet make any real changes to either<br>
port: it is a copy of rs6000/ to powerpcspe/, with some renames and<br>
some necessary changes to the port file, and slightly more involved<br>
changes to config.gcc .<br>
<br>
This was tested on powerpc64-linux {-m32,-m64}, and it was build-tested<br>
on powerpc-linux-gnuspe (and the resulting compiler was tested to be<br>
functional: it can build various Linux defconfigs for SPE systems).<br>
<br>
I have tried to see how much the powerpcspe port can be simplified<br>
after this, and found it can lose 80% of the code without big problems.<br>
You may however not want all that, for example, I removed all 64-bit<br>
support in that test. Getting rid of all VMX/VSX support is a big part<br>
of it already, as is removing "classic" floating point (and paired<br>
single, and xilinx fpu, and all newer ISA features, etc.)<br>
<br>
For the rs6000 port the low-hanging fruits are much more modest, only<br>
5% or a bit more; but in pretty gnarly code. For example, some current<br>
pain points are the SPE ABI (for separate shrink-wrapping), and how<br>
isel is handled.<br>
<br>
This won't be the final series... I have a few questions:<br>
<br>
-- This uses powerpc-*-rtems*spe*; do we want powerpcspe-*-rtems*<br>
instead? Or both?<br>
-- This uses powerpc-wrs-vxworksspe; do we want powerpcspe-wrs-vxworks<br>
instead? Both? What about the ae and mils variants?<br>
-- Does powerpc*-*-freebsd*spe* exist?<br>
-- Does powerpc-*-netbsd*spe* exist?<br>
-- Does powerpc-*-eabisim*spe* exist?<br>
-- Does powerpcle-*-*spe* exist?<br>
<br>
Also, testing is needed :-) You can get better testing by removing<br>
the rs6000/ directories completely, btw.; otherwise files from rs6000/<br>
can accidentally be picked up instead of the corresponfing file from<br>
powerpcspe/, which will currently work because there are no big<br>
differences yet, but things will diverge later (and then break).<br>
<br>
<br>
Segher<br>
<br>
<br>
<br>
______________________________<wbr>_________________<br>
devel mailing list<br>
<a href="mailto:devel@rtems.org" target="_blank">devel@rtems.org</a><br>
<a href="http://lists.rtems.org/mailman/listinfo/devel" rel="noreferrer" target="_blank">http://lists.rtems.org/mailman<wbr>/listinfo/devel</a><br>
</blockquote></div><br></div></div>