[Bug target/113276] gcc.dg/torture/fp-double-convert-float-1.c fails at execution on arm cortex-m33

[azoff at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113276

--- Comment #5 from Torbjorn SVENSSON <azoff at gcc dot gnu.org> ---
(In reply to richard.earnshaw from comment #4)
> On 27/03/2026 14:54, azoff at gcc dot gnu.org via Gcc-bugs wrote:
> > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113276
> > 
> > --- Comment #3 from Torbjorn SVENSSON <azoff at gcc dot gnu.org> ---
> > Could we perhaps change
> > 
> > /* { dg-do run } */
> > 
> > to
> > 
> > /* { dg-do run { xfail { arm-none-eabi && { ! arm_fp_dp_ok } } } } */
> > 
> > in order to mark the test as having expectancy to fail unless there is DP HW
> > support? Or would this have any side effects that I'm nor aware of?
> 
> Not quite, since unfortunately that test tries some different combinations
> of flags which we won't later add.  The test isn't suitable for a 'run'
> target.
> 
> But otherwise, in principle this looks like the right approach.

Would it be better to simply add this then?

/* { dg-skip-if "PR113276" { arm-none-eabi && { ! arm_fp_dp_ok } } } */

To me, it feels like a bad situation to have tests marked as fail that would
never be fixed. At some point, the noise would be too high so that real issues
would be missed.