This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Patch for PR target/20924
- From: Steve Ellcey <sje at cup dot hp dot com>
- To: rasky at develer dot com
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Tue, 12 Apr 2005 11:08:26 -0700 (PDT)
- Subject: Re: Patch for PR target/20924
> Any specific reason why the testcase is not converted to DejaGnu format and
> included in this patch?
> --
> Giovanni Bajo
Because I didn't think of it? :-) I guess that's not much of a reason
so I will go ahead and make a proper test.
It looks like fetestexcept is part of the C99 standard so I can make
this a generic test that should work on all platforms that support C99.
I also see that gcc.c-torture/execute has an ieee subdirectory. I
wonder if I should put the test there instead of under gcc.dg. Or
should gcc.dg also have a ieee subdirectory? I guess I can use the
HAVE_C99_RUNTIME macro to determine if fetestexcept exists or not when
creating the test.
Anyone care to comment on where new ieee math tests should go?
My inclination would be to create an ieee subdirectory under gcc.dg and
include this test under it as div.c and expand it to test various
division results like overflow, underflow, inexact, etc. I know enough
about floating point issues that getting such a test right is tricky but
I think that a general test would be more valuable than an IA64 specific
test for this particular bug fix.
Steve Ellcey
sje@cup.hp.com