[PATCH] LoongArch: fix missing trap for enabled exceptions on narrowing operation

[Florian Weimer]

* Xi Ruoyao:

> On Thu, 2026-03-19 at 10:14 +0100, Florian Weimer wrote:
>> * Xi Ruoyao:
>> 
>> > diff --git a/math/test-narrowing-trap.c b/math/test-narrowing-trap.c
>> > new file mode 100644
>> > index 0000000000..110ac15f62
>> 
>> > +  if (!EXCEPTION_ENABLE_SUPPORTED (FE_INVALID))
>> > +    {
>> > +      puts ("feenableexcept (FE_INVALID) not supported, cannot test.");
>> > +      return 77;
>> > +    }
>> 
>> You can use FAIL_UNSUPPORTED from <support/check.h>.
>> 
>> > +  pid = fork ();
>> > +  if (pid == 0)
>> > +    {
>> > +#ifdef RLIMIT_CORE
>> > +      /* Try to avoid dumping core.  */
>> > +      struct rlimit core_limit = {};
>> > +      setrlimit (RLIMIT_CORE, &core_limit);
>> > +#endif
>> 
>> The test framework (<support/test-driver.c>) deals with that.
>> 
>> We also have error-checking variants xfork, xwaitpid, and TEST_VERIFY
>> and TEST_COMPARE for checking specific conditions.
>
> test-fenv.c uses fork() and check if it returns ENOSYS (if so it skips
> the test).  Is doing so really necessary or may I just use xfork()
> instead?

No, this is just a very, very old test.  We don't support testing glibc
without a working fork.

Thanks,
Florian