https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88104
--- Comment #3 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot
Uni-Bielefeld.DE> ---
> --- Comment #2 from Martin Sebor <msebor at gcc dot gnu.org> ---
> The failing tests are compile-only (don't involve the assembler), and would
> pass even with a default-configured GCC. The reason why they fail is that the
> large_long_double DejaGnu configuration test both compiles and assembles the
> test, and the latter step fails. The following fixes it:
>
> @@ -2664,7 +2664,7 @@ proc check_effective_target_large_long_double { }
> # 0 otherwise.
>
> proc check_effective_target_large_double { } {
> - return [check_no_compiler_messages large_double object {
> + return [check_no_compiler_messages large_double assembly {
> int dummy[sizeof(double) > sizeof(float) ? 1 : -1];
> }]
> }
Indeed. However, this is only one instance of the same problem and I'd
rather not change one random instance only, especially given the easy
workaround. I hope to go through the effective-target tests with an eye
for this problems.
