This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [AArch64] Fix simd intrinsics bug on float vminnm/vmaxnm
On 7 July 2016 at 14:54, Jiong Wang <jiong.wang@foss.arm.com> wrote:
>
>
> On 07/07/16 12:36, Christophe Lyon wrote:
>>
>> On 7 July 2016 at 11:16, Jiong Wang <jiong.wang@foss.arm.com> wrote:
>>>
>>>
>>> I was using dg-xfail-if, (the description is still using "marked as
>>> XFAIL"...),
>>> but later found it's actually broken under advsimd-intrinsics,
>>> UNRESOLVEDs
>>> are
>>> given at the same time instead of clean XFAIL, I suspect those dg-do-what
>>> overriding broken dejangu internal variable, Christophe, do you mind have
>>> a
>>> look?
>>>
>> I've made a quick attempt (replacing an existing dg-skip-if with
>> dg-xfail-if in vcvt_high-1.c)
>> and I do see XFAIL and UNRESOLVED.
>> But this seems normal in this case, because:
>> - when using dg-skip-if, the test was not compiled (skipped)
>> - when using dg-xfail-if, the test is actually compiled, leading to
>> compilation errors which are reported as UNRESOLVED
>
>
> My test was:
>
> cp k.c gcc/testsuite/gcc.target/aarch64
> cp k.c gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/
>
> cat k.c
> ===
> /* { dg-xfail-if "" { aarch64*-*-* } } */
> int c (
>
> make check RUNTESTFLAGS="advsimd-intrinsics.exp=k.c"
>
> # of expected failures 8
> # of unresolved testcases 8
>
> make check RUNTESTFLAGS="aarch64.exp=k.c"
>
> # of expected failures 1
>
> The iteration number doesn't matter, but there wan't extra unresolved
> failures when it's driven by aarch64.exp.
>
>
in aarch64.exp, the default dg-do-what is "compile", while in
advsimd-intrinsics.exp, it's "run" unless is-effective-target
arm_neon_hw on arm targets.
So the XFAIL comes from the compilation error messages, while the
UNRESOLVED part comes from the impossible execution
Christophe