Fw: [PATCH 1/2] BZ#13743: PowerPC - Add a new header for platform specific functions

[Carlos O'Donell]

On Fri, Mar 9, 2012 at 5:46 PM, Steven Munroe
<munroesj@linux.vnet.ibm.com> wrote:
> Ok, as I understand that GCC could be a home for intrinsics for
> computation and synchronization but not a catch-all. This is very close
> to my assumption above. But you may being disagreeing with Michael on
> the timebase?
>
> I would assume that things like timebase, and program priority register
> are more strongly associated with multi-core and threading APIs (ie
> GLIBC)
>
> How do you feel about the data stream control register and getting a
> more complete set (than __sync_synchronize()) of memory barriers? *GCC
> or GLIBC)?

Given the rationale I've written up here, I'd say "GCC."
http://sourceware.org/glibc/wiki/PlatformHeaders

It sounds like the GCC *and* GLIBC concensus is that the
current function would most benefit your users as a GCC
builtin.

Would it be wrong of me to suggest the following course
of action?

Start with:

* Rework the patch to add a sys/platform/ppc.h header.
* Document this in the glibc manual as the place for
  platform macros, and document the current Power macros.

In parallel:

* Work to migrate any macros from sys/platform/ppc.h into
  GCC as built-ins (where appropriate) keeping the same
  name for convenience.
* Document the built-ins in the GCC manual.

At any given point in the time the user must always include
sys/platform/ppc.h, but they must check the documentation
to see which function is a builtin and which is provided
as macro.

Cheers,
Carlos.