* [Bug fortran/31202] wrong code generated with gfortran
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
@ 2007-03-16 20:42 ` fxcoudert at gcc dot gnu dot org
2007-03-16 20:55 ` fxcoudert at gcc dot gnu dot org
` (10 subsequent siblings)
11 siblings, 0 replies; 13+ messages in thread
From: fxcoudert at gcc dot gnu dot org @ 2007-03-16 20:42 UTC (permalink / raw)
To: gcc-bugs
------- Comment #1 from fxcoudert at gcc dot gnu dot org 2007-03-16 20:42 -------
Confirmed on x86_64-linux, but somehow it doesn't happen on i386.
The reason why it fails and the way to fix it are given in the ML link given by
the reporter.
--
fxcoudert at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
Keywords| |wrong-code
Known to fail| |4.1.3 4.2.0 4.3.0
Last reconfirmed|0000-00-00 00:00:00 |2007-03-16 20:42:28
date| |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Bug fortran/31202] wrong code generated with gfortran
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
2007-03-16 20:42 ` [Bug fortran/31202] " fxcoudert at gcc dot gnu dot org
@ 2007-03-16 20:55 ` fxcoudert at gcc dot gnu dot org
2007-03-19 1:50 ` [Bug fortran/31202] Incorrect rounding generated for NINT fxcoudert at gcc dot gnu dot org
` (9 subsequent siblings)
11 siblings, 0 replies; 13+ messages in thread
From: fxcoudert at gcc dot gnu dot org @ 2007-03-16 20:55 UTC (permalink / raw)
To: gcc-bugs
------- Comment #2 from fxcoudert at gcc dot gnu dot org 2007-03-16 20:55 -------
*** Bug 31208 has been marked as a duplicate of this bug. ***
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Bug fortran/31202] Incorrect rounding generated for NINT
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
2007-03-16 20:42 ` [Bug fortran/31202] " fxcoudert at gcc dot gnu dot org
2007-03-16 20:55 ` fxcoudert at gcc dot gnu dot org
@ 2007-03-19 1:50 ` fxcoudert at gcc dot gnu dot org
2007-05-08 22:08 ` fxcoudert at gcc dot gnu dot org
` (8 subsequent siblings)
11 siblings, 0 replies; 13+ messages in thread
From: fxcoudert at gcc dot gnu dot org @ 2007-03-19 1:50 UTC (permalink / raw)
To: gcc-bugs
------- Comment #3 from fxcoudert at gcc dot gnu dot org 2007-03-19 01:50 -------
Created an attachment (id=13228)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=13228&action=view)
Patch for this bug
This is a patch following exactly the very good explanation done in the
mailing-list post pointed by the testcase. I need to work on a good testcase
exercing this with differents kinds. And there also is a question for
middle-end maintainers about how to get the longest floating point type in a
systematic way.
--
fxcoudert at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
AssignedTo|unassigned at gcc dot gnu |fxcoudert at gcc dot gnu dot
|dot org |org
Status|NEW |ASSIGNED
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Bug fortran/31202] Incorrect rounding generated for NINT
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
` (2 preceding siblings ...)
2007-03-19 1:50 ` [Bug fortran/31202] Incorrect rounding generated for NINT fxcoudert at gcc dot gnu dot org
@ 2007-05-08 22:08 ` fxcoudert at gcc dot gnu dot org
2007-06-29 11:35 ` jv244 at cam dot ac dot uk
` (7 subsequent siblings)
11 siblings, 0 replies; 13+ messages in thread
From: fxcoudert at gcc dot gnu dot org @ 2007-05-08 22:08 UTC (permalink / raw)
To: gcc-bugs
------- Comment #4 from fxcoudert at gcc dot gnu dot org 2007-05-08 23:08 -------
Created an attachment (id=13532)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=13532&action=view)
New patch
Here's a new patch from a completely different perspective, using C99 lround
and llround functions (and their float and long double variants).
--
fxcoudert at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Attachment #13228|0 |1
is obsolete| |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Bug fortran/31202] Incorrect rounding generated for NINT
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
` (3 preceding siblings ...)
2007-05-08 22:08 ` fxcoudert at gcc dot gnu dot org
@ 2007-06-29 11:35 ` jv244 at cam dot ac dot uk
2007-06-29 11:45 ` fxcoudert at gcc dot gnu dot org
` (6 subsequent siblings)
11 siblings, 0 replies; 13+ messages in thread
From: jv244 at cam dot ac dot uk @ 2007-06-29 11:35 UTC (permalink / raw)
To: gcc-bugs
------- Comment #5 from jv244 at cam dot ac dot uk 2007-06-29 11:34 -------
is this patch still OK ?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Bug fortran/31202] Incorrect rounding generated for NINT
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
` (4 preceding siblings ...)
2007-06-29 11:35 ` jv244 at cam dot ac dot uk
@ 2007-06-29 11:45 ` fxcoudert at gcc dot gnu dot org
2007-06-29 16:10 ` kargl at gcc dot gnu dot org
` (5 subsequent siblings)
11 siblings, 0 replies; 13+ messages in thread
From: fxcoudert at gcc dot gnu dot org @ 2007-06-29 11:45 UTC (permalink / raw)
To: gcc-bugs
------- Comment #6 from fxcoudert at gcc dot gnu dot org 2007-06-29 11:45 -------
(In reply to comment #5)
> is this patch still OK ?
The lround patch should be OK on C99 targets, but it would probably break
things on non-C99 targets, which is why I didn't submit it for formal review. I
should have some time to look into it later, but the last weeks have been a
complete chaos (PhD defence and finishing every little project I had in my
group).
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Bug fortran/31202] Incorrect rounding generated for NINT
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
` (5 preceding siblings ...)
2007-06-29 11:45 ` fxcoudert at gcc dot gnu dot org
@ 2007-06-29 16:10 ` kargl at gcc dot gnu dot org
2007-08-03 21:26 ` fxcoudert at gcc dot gnu dot org
` (4 subsequent siblings)
11 siblings, 0 replies; 13+ messages in thread
From: kargl at gcc dot gnu dot org @ 2007-06-29 16:10 UTC (permalink / raw)
To: gcc-bugs
------- Comment #7 from kargl at gcc dot gnu dot org 2007-06-29 16:10 -------
(In reply to comment #6)
> (In reply to comment #5)
> > is this patch still OK ?
>
> The lround patch should be OK on C99 targets, but it would probably break
> things on non-C99 targets, which is why I didn't submit it for formal review. I
On non-c99 targets, gfortran supplies roundf and roundl. You could
do (GFC_INTEGER_?) roundl() for lroundl().
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Bug fortran/31202] Incorrect rounding generated for NINT
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
` (6 preceding siblings ...)
2007-06-29 16:10 ` kargl at gcc dot gnu dot org
@ 2007-08-03 21:26 ` fxcoudert at gcc dot gnu dot org
2007-08-03 21:27 ` fxcoudert at gcc dot gnu dot org
` (3 subsequent siblings)
11 siblings, 0 replies; 13+ messages in thread
From: fxcoudert at gcc dot gnu dot org @ 2007-08-03 21:26 UTC (permalink / raw)
To: gcc-bugs
------- Comment #8 from fxcoudert at gcc dot gnu dot org 2007-08-03 21:26 -------
Subject: Bug 31202
Author: fxcoudert
Date: Fri Aug 3 21:26:10 2007
New Revision: 127185
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=127185
Log:
PR fortran/31202
* f95-lang.c (gfc_init_builtin_functions): Defin builtins for
lround{f,,l} and llround{f,,l}.
* trans-intrinsic.c (build_fix_expr): Generate calls to the
{l,}round{f,,l} functions.
* intrinsics/c99_functions.c (roundl,lroundf,lround,lroundl,
llroundf,llround,llroundl): New functions.
* c99_protos.h (roundl,lroundf,lround,lroundl,llroundf,llround,
llroundl): New prototypes.
* configure.ac: Check for lroundf, lround, lroundl, llroundf,
llround and llroundl.
* configure: Regenerate.
* Makefile.in: Regenerate.
* config.h.in: Regenerate.
* gfortran.dg/nint_2.f90: New test.
Added:
trunk/gcc/testsuite/gfortran.dg/nint_2.f90
Modified:
trunk/gcc/fortran/ChangeLog
trunk/gcc/fortran/f95-lang.c
trunk/gcc/fortran/trans-intrinsic.c
trunk/gcc/testsuite/ChangeLog
trunk/libgfortran/ChangeLog
trunk/libgfortran/Makefile.in
trunk/libgfortran/c99_protos.h
trunk/libgfortran/config.h.in
trunk/libgfortran/configure
trunk/libgfortran/configure.ac
trunk/libgfortran/intrinsics/c99_functions.c
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Bug fortran/31202] Incorrect rounding generated for NINT
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
` (7 preceding siblings ...)
2007-08-03 21:26 ` fxcoudert at gcc dot gnu dot org
@ 2007-08-03 21:27 ` fxcoudert at gcc dot gnu dot org
2007-08-05 13:12 ` cato at df dot lth dot se
` (2 subsequent siblings)
11 siblings, 0 replies; 13+ messages in thread
From: fxcoudert at gcc dot gnu dot org @ 2007-08-03 21:27 UTC (permalink / raw)
To: gcc-bugs
------- Comment #9 from fxcoudert at gcc dot gnu dot org 2007-08-03 21:27 -------
Well, we'll see if I broke any non-C99 target :)
--
fxcoudert at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|ASSIGNED |RESOLVED
Resolution| |FIXED
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Bug fortran/31202] Incorrect rounding generated for NINT
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
` (8 preceding siblings ...)
2007-08-03 21:27 ` fxcoudert at gcc dot gnu dot org
@ 2007-08-05 13:12 ` cato at df dot lth dot se
2007-08-05 22:05 ` fxcoudert at gcc dot gnu dot org
2007-08-05 22:14 ` fxcoudert at gcc dot gnu dot org
11 siblings, 0 replies; 13+ messages in thread
From: cato at df dot lth dot se @ 2007-08-05 13:12 UTC (permalink / raw)
To: gcc-bugs
------- Comment #10 from cato at df dot lth dot se 2007-08-05 13:12 -------
Subject: Re: Incorrect rounding generated for NINT
On Fri, 3 Aug 2007, fxcoudert at gcc dot gnu dot org wrote:
> Well, we'll see if I broke any non-C99 target :)
Yes, you did. :)
NetBSD does not have lroundl (and not roundl, so lroundl is not added by
intrinsics/c99_functions.c) with the result that linking with libgfortran
fails with "undefined reference to `lroundl'".
/Krister
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Bug fortran/31202] Incorrect rounding generated for NINT
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
` (9 preceding siblings ...)
2007-08-05 13:12 ` cato at df dot lth dot se
@ 2007-08-05 22:05 ` fxcoudert at gcc dot gnu dot org
2007-08-05 22:14 ` fxcoudert at gcc dot gnu dot org
11 siblings, 0 replies; 13+ messages in thread
From: fxcoudert at gcc dot gnu dot org @ 2007-08-05 22:05 UTC (permalink / raw)
To: gcc-bugs
------- Comment #11 from fxcoudert at gcc dot gnu dot org 2007-08-05 22:05 -------
Index: intrinsics/c99_functions.c
===================================================================
--- intrinsics/c99_functions.c (revision 127224)
+++ intrinsics/c99_functions.c (working copy)
@@ -500,8 +500,9 @@ powf(float x, float y)
/* Algorithm by Steven G. Kargl. */
-#if !defined(HAVE_ROUNDL) && defined(HAVE_CEILL)
+#if !defined(HAVE_ROUNDL)
#define HAVE_ROUNDL 1
+#if defined(HAVE_CEILL)
/* Round to nearest integral value. If the argument is halfway between two
integral values then round away from zero. */
@@ -527,6 +528,27 @@ roundl(long double x)
return (-t);
}
}
+#else
+
+/* Poor version of roundl for system that don't have ceill. */
+long double
+roundl(long double x)
+{
+ if (x > DBL_MAX || x < -DBL_MAX)
+ {
+#ifdef HAVE_NEXTAFTERL
+ static long double prechalf = nexafterl (0.5L, LDBL_MAX);
+#else
+ static long double prechalf = 0.5L;
+#endif
+ return (GFC_INTEGER_LARGEST) (x + (x > 0 ? prechalf : -prechalf));
+ }
+ else
+ /* Use round(). */
+ return round((double) x);
+}
+
+#endif
#endif
#ifndef HAVE_ROUND
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread
* [Bug fortran/31202] Incorrect rounding generated for NINT
2007-03-16 11:27 [Bug fortran/31202] New: wrong code generated with gfortran jv244 at cam dot ac dot uk
` (10 preceding siblings ...)
2007-08-05 22:05 ` fxcoudert at gcc dot gnu dot org
@ 2007-08-05 22:14 ` fxcoudert at gcc dot gnu dot org
11 siblings, 0 replies; 13+ messages in thread
From: fxcoudert at gcc dot gnu dot org @ 2007-08-05 22:14 UTC (permalink / raw)
To: gcc-bugs
------- Comment #12 from fxcoudert at gcc dot gnu dot org 2007-08-05 22:14 -------
Subject: Bug 31202
Author: fxcoudert
Date: Sun Aug 5 22:14:34 2007
New Revision: 127227
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=127227
Log:
2007-08-05 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
PR fortran/31202
* intrinsics/c99_functions.c (roundl): Provide fallback
implementation for systems without ceill.
* c99_protos.h (roundl): Define prototype in all cases.
Modified:
trunk/libgfortran/ChangeLog
trunk/libgfortran/c99_protos.h
trunk/libgfortran/intrinsics/c99_functions.c
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31202
^ permalink raw reply [flat|nested] 13+ messages in thread