On Tue, 2018-06-05 at 16:45 -0500, Bill Schmidt wrote:
> Hi Carl,
>
> That looks like a typo in the ABI document to me. The return type
> should match the
> argument types like it does for the other variants. Sorry -- I'll
> open a bug against
> the ABI doc.
So, the ABI doc currently says:
vector unsigned char vec_permxor (vector signed char, vector signed char,
vector signed char);
vector unsigned char vec_permxor (vector unsigned char, vector unsigned
char, vector unsigned char);
and we want it to read:
vector signed char vec_permxor (vector signed char, vector signed char,
vector signed char);
vector unsigned char vec_permxor (vector unsigned char, vector unsigned
char, vector unsigned char);
If so, we only want the changes to vec_insert4b in builtins-7-p9-runnable.c.
I will re-spin the patch. Thanks.
Carl Love
> vec_insert4b
> Thanks! Good catch, Segher.
>
> -- Bill
>
> Bill Schmidt, Ph.D.
> STSM, GCC Architect for Linux on Power
> IBM Linux Technology Center
> wschm...@linux.vnet.ibm.com
>
> > On Jun 5, 2018, at 4:32 PM, Segher Boessenkool <segher@kernel.crash
> > ing.org> wrote:
> >
> > Hi Carl,
> >
> > On Tue, Jun 05, 2018 at 01:57:52PM -0700, Carl Love wrote:
> > > The following patch fixes the return type for the existing
> > > vec_permxor
> > > builtin to match the ABI specification. The test case for the
> > > builtin
> > > is also updates.
> >
> > Hrm, is that a bug in the ABI doc though? Bill? Most older
> > builtins
> > return their source type; some newer ones always return
> > unsigned? Is this
> > on purpose?
> >
> > > Secondly, the first argument of the vec_insert4b() builtin test
> > > case is
> > > fixed to match the ABI specification for the builtin.
> >
> > The patch is fine for trunk if the ABI doc is correct. Thanks,
> >
> >
> > Segher
> >
>
>
