The following testcase fails when compiled with -march=2.0:
/* Test front-end conversions, optimizer conversions, and run-time
conversions between different arithmetic types.
Constants are specified in a non-obvious way to make them work for
any word size. Their value on a 32-bit machine is indicated in the
comments.
Note that this code is NOT intended for testing of accuracy of fp
conversions. */
unsigned long long int
f2ull(float f)
{
return (unsigned long long int) f;
}
test_float_to_longlong_integer()
{
if (f2ull((float) ((~0ULL) >> 1)) != (~0ULL) >> 1 && /* 0x7fffffff */
f2ull((float) ((~0ULL) >> 1)) != ((~0ULL) >> 1) + 1)
abort();
}
main()
{
test_float_to_longlong_integer();
exit(0);
}
The problem is the fcnv,t,sgl,udw instruction. It appears it isn't emulated
correctly. Test doesn't fail on hpux.
Dave
--
J. David Anglin dave.anglin@xxxxxxxxxxxxxx
National Research Council of Canada (613) 990-0752 (FAX: 952-6602)
--
To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
