On Thu, Jul 18, 2024 at 09:53:05AM -0500, Peter Bergner wrote:
> On 7/18/24 8:23 AM, Segher Boessenkool wrote:
> > Everything in gcc.target/powerpc is only run for target powerpc*-*-*
> > anyway, so make this just
> > 
> > /* { dg-do compile } */
> > 
> > please.  (Or nothing, it is the default after all, but you may want to
> > have it explicit).
> 
> Personally, I do like seeing the /* { dg-do compile } */ even though
> that is the default in this testsuite directory.

It depends :-)  For testsuites that often do something else as well
(like in gcc.target), it can be nice to be verbose, yes.

> >> +/* { dg-require-effective-target powerpc_vsx_ok } */
> > 
> > Why is this needed?  It needs a comment saying that.
> 
> After Kewen's testsuite patch, powerpc_vsx_ok is pretty useless as it only
> checks whether the assembler knows about vsx, and what people normally want
> is powerpc_vsx which verifies that the compiler options used to compile the
> test support generating VSX insns.

That, sure, but even then it does not make any sense.  Each of the
funcs uses its own -mcpu= so either will have VSX or not, and it doesn't
matter at all what the default compiler target does.

And, absolutely nothing in this test case uses VSX at all.

> > Saying that this is "to test if can use the target attr" is nonsense.
> > That does not need Linux either btw.  Target selection might not work
> > correctly currently on some other systems, but this is not a run test!
> 
> Agreed.  Also, if the clones stuff really doesn't work for those
> targets even in a compile test, then rather than diabling this way,
> I think I'd like to see a target-supports clones_ok test or similar.

Yup.


Segher

Reply via email to