Re: [PATCH] rs6000: Fix 2 for PR92661, Do not define builtins that overload disabled builtins

[Segher Boessenkool <segher@kernel.crashing.org>]

On Wed, Dec 04, 2019 at 08:47:49PM +0000, Iain Sandoe wrote:
> Peter Bergner <bergner@linux.ibm.com> wrote:
> >On 12/4/19 1:16 PM, Segher Boessenkool wrote:
> >>This isn't run from powerpc.exp, so it needs to still do that first check.
> >>And it's up to the Darwin maintainers whether they want that second part
> >>(there are many more tests and testsuites that disable *-darwin* while
> >>that isn't really necessary).
> 
> As Peter mentions below, it produces a lot of meaningless FAILs if we run
> the tests on Darwin, so one way or another, I’d like to skip them  
> (unless/until
> we have a situation that DFP is supported on some hardware running
> Darwin)...

Sure.  The question is, should separate tests have a
dg-require-effective-target clause, or do we want that in dfp.exp?  Either
way the separate tests can be updated to remove other resstrictions after
this (never running on aix, for example).

> >Well, yes.  I saw those tests being run for my --disable-decimal-float
> >runs, which resulted in FAILs for all of those tests.  They had ICE's
> >on unpatched trunk and FAILed gracefully using my patch, but they all
> >still FAILed, since these are DFP tests and DFP is disabled.
> >There's no sense in running these tests when DFP is disabled, either
> >manually due to --disable-decimal-float or implicitly because of the
> >specific target.
> >
> >Why isn't just testing check_effective_target_dfp enough to disable the
> >tests on Darwin, --disable-decimal-float, etc.?
> 
> … It should be a better solution - I will confirm this.

Yes, this would be ideal.  The fewer tests we disable for specific OSes,
the better.


Segher