The testcase:
--cut here--
#include stdio.h
double
__attribute__ ((noinline))
test (double x)
{
return x = 0.0 ? x : -x;
}
double
__attribute__ ((always_inline))
test_inlined (double x)
{
return x = 0.0 ? x : -x;
}
int main()
{
double x = -0.0;
printf (%f %f %f\n, x, test(x),
--- Comment #1 from ubizjak at gmail dot com 2008-03-04 08:40 ---
Hm, -0.0 = 0.0 is folded to TRUE, as confirmed by:
printf (%i\n, -0.0 = 0.0);
This produces 1.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35456
Sent from my iPhone
On Mar 4, 2008, at 0:40, ubizjak at gmail dot com [EMAIL PROTECTED]
wrote:
--- Comment #1 from ubizjak at gmail dot com 2008-03-04 08:40
---
Hm, -0.0 = 0.0 is folded to TRUE, as confirmed by:
printf (%i\n, -0.0 = 0.0);
This produces 1.
That is
--- Comment #2 from pinskia at gmail dot com 2008-03-04 08:49 ---
Subject: Re: Different results for inlined vs. non-inlined function
Sent from my iPhone
On Mar 4, 2008, at 0:40, ubizjak at gmail dot com [EMAIL PROTECTED]
wrote:
--- Comment #1 from ubizjak at gmail dot
--- Comment #11 from ebotcazou at gcc dot gnu dot org 2008-03-04 09:14
---
The original motivation for this report was that I was trying to reproduce the
occasional problem of error being called (a this should never happen sanity
check) at this place in the code. I thought I had
--- Comment #1 from rguenth at gcc dot gnu dot org 2008-03-04 09:31 ---
Patch welcome.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35449
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Target Milestone|--- |4.3.0
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35454
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Component|other |target
Target Milestone|--- |4.2.4
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
--- Comment #3 from ubizjak at gmail dot com 2008-03-04 10:01 ---
Working on a fix.
--
ubizjak at gmail dot com changed:
What|Removed |Added
I get the following error message trying to build trunk r132854 on CELL
SPU:
-O2 ' 'CPPFLAGS=' 'LDFLAGS=' 'build_alias=powerpc64-unknown-linux-gnu'
'host_alias=spu' 'target_alias=spu' --cache-file=.././config.cache
--srcdir=../../../../gcc/newlib/libc
configure: loading cache .././config.cache
--- Comment #17 from pcarlini at suse dot de 2008-03-04 11:04 ---
(In reply to comment #16)
Sorry, I had two versions of patch and managed to commit the wrong copy.
Sent correct one to ML. It should be fixed now.
Indeed, it's fixed! Many, many thanks!
The point I wanted to make is
--- Comment #4 from ubizjak at gmail dot com 2008-03-04 12:41 ---
Patch at http://gcc.gnu.org/ml/gcc-patches/2008-03/msg00207.html.
--
ubizjak at gmail dot com changed:
What|Removed |Added
--- Comment #3 from irar at il dot ibm dot com 2008-03-04 13:30 ---
It fails in initialize_matrix_A() when called with chrec {0, +, {0, +, 4}_1}_2:
int_cst_value (CHREC_RIGHT (chrec)) ICEs, since CHREC_RIGHT (chrec) here is {0,
+, 4}_1.
Ira
--
irar at il dot ibm dot com changed:
Dependency generation (-M) does not quote '#' in filenames.
--
Summary: Dependency generation (-M) does not quote '#' in
filenames
Product: gcc
Version: 4.3.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
--- Comment #1 from markus dot milleder at generali dot at 2008-03-04
13:37 ---
Created an attachment (id=15259)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=15259action=view)
Patch for mkdeps.c + testcase
This trivial patch cures the problem. I do not have a copyright
--- Comment #2 from rwild at gcc dot gnu dot org 2008-03-04 13:53 ---
Please note that quoting # in makefiles is not portable, but only supported
by some make implementations. For example, with
foo: a\#b
bar: c#d
AIX make will assume 'foo' depends on 'a\', and 'bar' on 'c'.
--- Comment #5 from uros at gcc dot gnu dot org 2008-03-04 13:58 ---
Subject: Bug 35456
Author: uros
Date: Tue Mar 4 13:57:27 2008
New Revision: 132863
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=132863
Log:
PR middle-end/35456
* fold-const.c
--- Comment #3 from markus dot milleder at generali dot at 2008-03-04
14:10 ---
(In reply to comment #2)
Please note that quoting # in makefiles is not portable, but only supported
by some make implementations. For example, with
foo: a\#b
bar: c#d
AIX make will assume
--- Comment #10 from fxcoudert at gcc dot gnu dot org 2008-03-04 14:21
---
A bit of experimentation reveals that we can use the asymptotic expression at
http://en.wikipedia.org/wiki/Error_function#Asymptotic_expansion for arguments
larger than 7, and it does converge with less than 100
--- Comment #11 from fxcoudert at gcc dot gnu dot org 2008-03-04 14:26
---
On ppc longdouble, convergence requires argument = 8.6 and, again, less than
100 terms. Why does convergence radius depend on the format of the
floating-point type, I wonder... For the record, here's the small C
--- Comment #16 from uweigand at gcc dot gnu dot org 2008-03-04 14:51
---
Hi Jakub,
we need the same changes in both .eh_frame and .dwarf_frame;
does the gas .cfi_ support both sections?
I'm wondering how save restore should work across two
different FDEs -- in the new FDE, we'd
--- Comment #17 from jakub at gcc dot gnu dot org 2008-03-04 15:10 ---
.cfi_* doesn't create .dwarf_frame, but perhaps it could be taught to do that
optionally (some flag on .cfi_startproc that would switch on additional
creation of .dwarf_frame).
The plan with save/restore was that you
ICE segmentaion fault when compiling. Apologies for the source included but I
had to reduce it down from proprietary code...
Command used: gfortran -v -save-temps -c ice_bug.f90
Output:
Using built-in specs.
Target: i386-redhat-linux
Configured with: ../configure --prefix=/usr
--- Comment #1 from s dot binnie at ucl dot ac dot uk 2008-03-04 15:15
---
Created an attachment (id=15260)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=15260action=view)
testcase for above bug
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35459
--- Comment #4 from irar at il dot ibm dot com 2008-03-04 15:44 ---
Isn't the same problem as in pr34635?
Ira
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35428
--- Comment #1 from ismail at namtrac dot org 2008-03-04 17:14 ---
Confirming on i686-apple-darwin9 with m64.
--
ismail at namtrac dot org changed:
What|Removed |Added
--- Comment #2 from kargl at gcc dot gnu dot org 2008-03-04 17:15 ---
This compiles with 4.2.3 and 4.3.0 (aka trunk). In that you are using
4.1.2, I doubt anyone with fix this bug in that branch. I'll suggest
that an upgrade to a newer version of gfortran is in order.
This bug should
--- Comment #23 from wilson at tuliptree dot org 2008-03-04 17:56 ---
Subject: Re: [4.2/4.3/4.4 regression] undefined reference
to `_Unwind_GetIPInfo'
ubizjak at gmail dot com wrote:
--- Comment #22 from ubizjak at gmail dot com 2008-02-20 18:39 ---
Critical P2 bug and the
When I attempt to compile the snapshot of 2/29/08 on OpenBSD 4.2 I get the
following message:
libcpp.a(errors.o)(.text+0x258): In function `cpp_error':
/home/mrichmon/gcc-4.4-20080229/libcpp/errors.c:141: undefined reference to
`__builtin_stdarg_start'
libcpp.a(errors.o)(.text+0x3a0): In function
--- Comment #11 from hp at gcc dot gnu dot org 2008-03-04 18:12 ---
Patches at http://gcc.gnu.org/ml/gcc-patches/2008-03/msg00225.html and
http://gcc.gnu.org/ml/gcc-patches/2008-03/msg00226.html.
--
hp at gcc dot gnu dot org changed:
What|Removed
--- Comment #1 from rguenth at gcc dot gnu dot org 2008-03-04 18:22 ---
Support for __builtin_stdarg has been removed. I guess this is still
referenced
in OpenBSD headers?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35460
--- Comment #1 from hjl at gcc dot gnu dot org 2008-03-04 19:18 ---
Subject: Bug 35453
Author: hjl
Date: Tue Mar 4 19:17:35 2008
New Revision: 132867
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=132867
Log:
gcc/
2008-03-04 H.J. Lu [EMAIL PROTECTED]
PR target/35453
--- Comment #3 from hjl dot tools at gmail dot com 2008-03-04 19:24 ---
Fixed for both gcc 4.3/4.4.
--
hjl dot tools at gmail dot com changed:
What|Removed |Added
--- Comment #2 from hjl at gcc dot gnu dot org 2008-03-04 19:24 ---
Subject: Bug 35453
Author: hjl
Date: Tue Mar 4 19:23:22 2008
New Revision: 132868
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=132868
Log:
gcc/
2008-03-04 H.J. Lu [EMAIL PROTECTED]
Backport from
the following is accepted by at least xlf, pgf and ifort, but not by
gcc version 4.3.0 20080213 (experimental) [trunk revision 132283] (GCC)
!$OMP PARALLEL DO DEFAULT(PRIVATE) SHARED(I,
!$J)
DO K=1,10
IF (I-J-K==0) I=K
ENDDO
END
--- Comment #1 from rguenth at gcc dot gnu dot org 2008-03-04 19:59 ---
I happen to have a fix.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
typedef struct {
int x;
} mystruct;
mystruct m;
should either have no name or a name of mystruct in C++. I see:
.ascii anonymous struct\0 ; DW_AT_name
and
.ascii $_0\0 ; external name
This works as expected in C.
Still fails in 20080308. I think this is target
typedef struct {
int x;
} mystruct;
mystruct m;
is missing any debug information for mystruct. If one compiles as C, mystruct
is present.
Still broken in 20080308 compiler. Should be target independent.
--
Summary: typedef missing in debug information with -gdwarf-2 for
--- Comment #1 from mrs at apple dot com 2008-03-04 20:14 ---
radr://5070293
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35462
--- Comment #1 from mrs at apple dot com 2008-03-04 20:15 ---
radr://5070293
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35463
--- Comment #1 from burnus at gcc dot gnu dot org 2008-03-04 20:26 ---
I think this is invalid if I read 2.1.2 Free Source Form Directives in
http://www.openmp.org/mp-documents/spec25.pdf correctly.
I think it should be:
!$OMP PARALLEL DO DEFAULT(PRIVATE) SHARED(I,
!$OMP
--- Comment #17 from manu at gcc dot gnu dot org 2008-03-04 20:29 ---
Subject: Bug 28322
Author: manu
Date: Tue Mar 4 20:28:52 2008
New Revision: 132870
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=132870
Log:
2008-03-04 Manuel Lopez-Ibanez [EMAIL PROTECTED]
PR
--- Comment #2 from jakub at gcc dot gnu dot org 2008-03-04 20:41 ---
Yeah, you're right. In free form ror !$omp directives the continuation lines
must start with !$omp (and previous line must end with ).
--
jakub at gcc dot gnu dot org changed:
What|Removed
--- Comment #4 from rguenth at gcc dot gnu dot org 2008-03-04 21:00 ---
http://gcc.gnu.org/ml/gcc-patches/2008-03/msg00243.html
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #3 from rguenth at gcc dot gnu dot org 2008-03-04 20:35 ---
Actually alias_sets_conflict_p is the cuplrit, as we have in the alias subset
of struct YY alias set zero (entered for the char member, accesses conflict
with accesses through char *):
/* See if the first alias
With -gnatwa, GNAT emits very useful warnings; one of them is condition is
always False. However, GNAT does not emit this warning when it applies inside
a generic, or a generic instantiation. The following test case demonstrates
the problem.
procedure Condition_Is_Always_False is
type T is
--- Comment #5 from dann at godzilla dot ics dot uci dot edu 2008-03-04
21:19 ---
(In reply to comment #4)
http://gcc.gnu.org/ml/gcc-patches/2008-03/msg00243.html
Thanks for working on this!
Have you looked at the impact?
Probably the generated code won't too different because the
--- Comment #6 from rguenth at gcc dot gnu dot org 2008-03-04 21:22 ---
Actually RTL alias is just using the same routines. The IL difference is
# SMT.4_6 = VDEF SMT.4_4(D)
# SMT.5_7 = VDEF SMT.5_5(D)
x_1(D)-x = 0;
# SMT.5_8 = VDEF SMT.5_7
y_2(D)-y = 1;
vs.
# SMT.18_5 =
--- Comment #7 from dann at godzilla dot ics dot uci dot edu 2008-03-04
21:32 ---
(In reply to comment #6)
Actually RTL alias is just using the same routines.
Might be, but the RTL level code that optimizes away the abort() in both
testcases (if I remember well
--- Comment #8 from rguenther at suse dot de 2008-03-04 21:35 ---
Subject: Re: adding unused char field inhibits
optimization
On Tue, 4 Mar 2008, dann at godzilla dot ics dot uci dot edu wrote:
--- Comment #7 from dann at godzilla dot ics dot uci dot edu 2008-03-04
21:32
--- Comment #9 from dann at godzilla dot ics dot uci dot edu 2008-03-04
21:43 ---
(In reply to comment #8)
Subject: Re: adding unused char field inhibits
optimization
On Tue, 4 Mar 2008, dann at godzilla dot ics dot uci dot edu wrote:
--- Comment #7 from dann at godzilla
--- Comment #7 from dominiq at lps dot ens dot fr 2008-03-04 21:45 ---
I have done some additional tests: compiled under 10.4.11, passed, then
executed under 10.5.2 failed.
Compiled on 10.5.2 with -S, then did the assembly and link under 10.4.11 (using
gfortran) then the executable
--- Comment #3 from burnus at gcc dot gnu dot org 2008-03-04 21:48 ---
First of all, I would like to thank your for the bug report.
While the bug is present in 4.1, it has been fixed meanwhile and 4.2.3, 4.3.0
and 4.4.0 work. The GCC policy is essentially only to backport fixes to
--- Comment #8 from fxcoudert at gcc dot gnu dot org 2008-03-04 21:58
---
(In reply to comment #7)
It would be nice to have a C test (if possible using gcc 4.0 from apple).
It'll be rather easy if you can do what I suggested in comment #6: trim it down
a bit by printing the various
--- Comment #9 from dominiq at lps dot ens dot fr 2008-03-04 22:20 ---
(In reply to comment #8)
Sorry, for not replying to your comments. I have been out of town for four days
(remote place without connction to the net) and I am trying to clean up the
backlog.
To answer the question,
--- Comment #7 from burnus at gcc dot gnu dot org 2008-03-04 22:20 ---
Intel documentation in PDF format:
http://softwarecommunity.intel.com/isn/downloads/softwareproducts/pdfs/347685.pdf
Directives (!DEC$) start at page 335
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34112
--- Comment #6 from danglin at gcc dot gnu dot org 2008-03-04 23:31 ---
Subject: Bug 35222
Author: danglin
Date: Tue Mar 4 23:30:47 2008
New Revision: 132876
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=132876
Log:
PR target/35222
* configure.ac
--- Comment #7 from danglin at gcc dot gnu dot org 2008-03-04 23:39 ---
Subject: Bug 35222
Author: danglin
Date: Tue Mar 4 23:39:11 2008
New Revision: 132877
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=132877
Log:
PR target/35222
* configure.ac
--- Comment #13 from fxcoudert at gcc dot gnu dot org 2008-03-04 23:47
---
It will probably be useless to others, who have already dug into this PR, but I
did have to research a bit to understand it, so I paste here the results of my
research... in case it can help.
Bootstrap fails at rev. 132875 with:
/opt/gcc/darwin_buildw/./prev-gcc/xgcc -B/opt/gcc/darwin_buildw/./prev-gcc/
-B/opt/gcc/gcc4.4w/powerpc-apple-darwin9/bin/ -c -g -O2 -mdynamic-no-pic
-DIN_GCC -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes
-Wold-style-definition
--- Comment #8 from danglin at gcc dot gnu dot org 2008-03-05 01:11 ---
Patch to force use of sjlj applied to trunk and 4.3.
--
danglin at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #12 from hp at gcc dot gnu dot org 2008-03-05 01:51 ---
Subject: Bug 35293
Author: hp
Date: Wed Mar 5 01:50:33 2008
New Revision: 132888
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=132888
Log:
PR libfortran/35293
* io/unix.c (fd_truncate): Fold
--- Comment #13 from hp at gcc dot gnu dot org 2008-03-05 01:56 ---
Subject: Bug 35293
Author: hp
Date: Wed Mar 5 01:55:27 2008
New Revision: 132889
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=132889
Log:
PR libfortran/35293
* lib/lib/target-supports.exp
--- Comment #14 from hp at gcc dot gnu dot org 2008-03-05 01:58 ---
Patches committed, case closed.
--
hp at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #6 from mckelvey at maskull dot com 2008-03-05 02:50 ---
(In reply to comment #5)
(In reply to comment #4)
A 65,000 line testcase? Seriously?
I'll try to make a smaller one. But it most likely won't be small.
I think the bug is caused by a VERY large
--- Comment #2 from bangerth at dealii dot org 2008-03-05 02:55 ---
This is called two-stage name lookup: when parsing a template all occurrences
of things that do not depend on the template parameter are bound to their
global definitions. Thus, here...
template class Reason
--- Comment #2 from pinskia at gcc dot gnu dot org 2008-03-05 03:18 ---
This is a bug in the openbsd headers. We most likely should fixincludes them
to use __builtin_va_start instead.
See PR 26264 and http://gcc.gnu.org/ml/gcc-patches/2002-07/msg00737.html for
more information on why
.size test, .-test
.section.rodata
.align 16
.LC0:
.long 84673010
.long 1025550150
.long 66901964
.long 807927808
.ident GCC: (GNU) 4.4.0 20080304 (experimental) [trunk revision
132852]
.section.note.GNU-stack
--- Comment #1 from hjl dot tools at gmail dot com 2008-03-05 04:27 ---
The difference comes from tree_expand_cfg pass:
[EMAIL PROTECTED] stage1-gcc]$ diff -up 32/x.i.132r.expand 64
--- 32/x.i.132r.expand 2008-03-04 20:25:21.0 -0800
+++ 64/x.i.132r.expand 2008-03-04
--- Comment #1 from zhouyi04 at ios dot cn 2008-03-05 07:03 ---
(In reply to comment #0)
Sorry there are some problem with 1st patch because the edge e
maybe removed after redirection.
24
| \ /
\ 5
\ |
6
In function redirect_edge_succ_nodup, the edge from 5 to 6 will
be
--- Comment #2 from pinskia at gcc dot gnu dot org 2008-03-05 07:49 ---
This is a HWI issue. I don't know how many times I tell people that x86 should
default to 64bit HWI if it uses TImode.
--
pinskia at gcc dot gnu dot org changed:
What|Removed
The code below demonstrates a missed optimization opportunity that
can have an impact on code.
If arrays are allocated by new(), the static alias analysis is unable to
prove independency:
#include stdlib.h
int
foo (void)
{
int *a = new int [1000];
int *b = new int [1000];
--- Comment #2 from pinskia at gcc dot gnu dot org 2008-03-05 07:48 ---
*** Bug 35467 has been marked as a duplicate of this bug. ***
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #1 from pinskia at gcc dot gnu dot org 2008-03-05 07:48 ---
I guess I have almost all the missed optimization bugs already filed or
something. Anyways this is an exact duplicate of bug 23383 (which I filed :)
).
*** This bug has been marked as a duplicate of 23383 ***
--
--
steven at gcc dot gnu dot org changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
Last
76 matches
Mail list logo