This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: [PATCH][fortran] Handle c99 builtins via .def mechanism
On Tue, Dec 20, 2005 at 12:03:42PM -0500, Andrew Pinski wrote:
>
> On Dec 6, 2005, at 2:54 PM, Richard Guenther wrote:
>
> >
> >This adds another file with builtin definitions, c99builtins.def,
> >where C99 builtins used by the middle-end or for optimization
> >purposes can be put.
> >
> >Build and tested on x86_64-unknown-linux-gnu.
> >
> >Ok for mainline?
>
> I would like this approved so it makes it easier to get
> large_real_kind_2.F90
> fixed on powerpc-darwin.
>
I object to this patch and its companion. For whatever
reason, the companion patch leads to over 112 regressions
in the testsuite due to broken GC. This is on an amd64-*-freebsd.
system with 12 GB of memory. The last time I tried tracking down
the problem, I needed
Index: params.def
===================================================================
--- params.def (revision 108854)
+++ params.def (working copy)
@@ -451,7 +451,7 @@ DEFPARAM(PARAM_MAX_FLOW_MEMORY_LOCATIONS
# define GGC_MIN_HEAPSIZE_DEFAULT 0
#else
# define GGC_MIN_EXPAND_DEFAULT 30
-# define GGC_MIN_HEAPSIZE_DEFAULT 4096
+# define GGC_MIN_HEAPSIZE_DEFAULT 32768
to reduce the failures to 7. I asked on IRC what the implications
of the above change would be and got ZERO response.
--
Steve