--- Comment #4 from irar at il dot ibm dot com 2007-10-24 07:48 ---
In this testcase we perform vectorization of strided accesses in a loop with
multiple types. In vectorizable_store() we gather all the scalar operands of
the stores in the strided accesses group and later call
vect_get_v
It would be really convenient if the C preprocessor exposed the number of
variable arguments supplied to a variadic macro as an integer. I tried to come
up with ways of working around this missing feature, but I can't figure out how
to do it without evaluating the macro arguments twice or causing
The following stripped down code shows pure virtual method definitions for both
a normal base class and a templated base class. To my surprise, the templated
class' body is not generated, leading to a "pure virtual method called"
termination in my actual threaded code. Leaving out the "= 0" in the
--- Comment #9 from irar at gcc dot gnu dot org 2007-10-24 09:35 ---
Subject: Bug 33804
Author: irar
Date: Wed Oct 24 09:35:00 2007
New Revision: 129599
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=129599
Log:
PR tree-optimization/33804
* tree-vect-transform.c
--- Comment #2 from fxcoudert at gcc dot gnu dot org 2007-10-24 09:40
---
No plan to work on this one, see
http://gcc.gnu.org/ml/fortran/2007-10/msg00263.html.
--
fxcoudert at gcc dot gnu dot org changed:
What|Removed |Added
--
--- Comment #7 from fxcoudert at gcc dot gnu dot org 2007-10-24 09:42
---
(In reply to comment #6)
> FX, you confirmed this enhancement: is do-as-the-others-do really good enough
> a
> reason to implement this?
No (see http://gcc.gnu.org/ml/fortran/2007-10/msg00263.html). The SIGNAL
i
--- Comment #1 from fxcoudert at gcc dot gnu dot org 2007-10-24 09:44
---
We should document the current situation and close this
(http://gcc.gnu.org/ml/fortran/2007-10/msg00263.html).
--
fxcoudert at gcc dot gnu dot org changed:
What|Removed |Add
--- Comment #12 from pault at gcc dot gnu dot org 2007-10-24 10:00 ---
I have prototype fix for this which works OK and does not break anything. It
copies 'p' to a temporary before the FORALL and uses the temporary for the
references. This method will also cure the problem with charact
--- Comment #1 from fxcoudert at gcc dot gnu dot org 2007-10-24 10:00
---
(In reply to comment #0)
> +{".ftn", "@f77", 0, 0, 0},
> +{".FTN", "@f77", 0, 0, 0},
Question is: do both case variants actually exist, and does the uppercase one
requires preprocessing?
--
fxcoudert at gcc d
--- Comment #3 from fxcoudert at gcc dot gnu dot org 2007-10-24 10:09
---
Hi Peter,
The following:
> f951 strauss.f90 -fPIC -quiet -dumpbase strauss.f90 -mmacosx-version-min=10.4
> -mtune=generic -auxbase strauss -version -ffixed-form -ffixed-line-length-132
> -fbounds-check -fmax-er
--- Comment #12 from rguenth at gcc dot gnu dot org 2007-10-24 10:13
---
One working patch/workaround:
http://gcc.gnu.org/ml/gcc-patches/2007-10/msg01377.html
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33870
--- Comment #13 from wvangulik at xs4all dot nl 2007-10-24 11:16 ---
(In reply to comment #10)
Something like this is smaller, faster and works for all registers (no need for
LD_regs). And could easily be writtin in to a insn:
; rOut: output register
; rIn: input register
; rIn, Z, N
--- Comment #5 from dominiq at lps dot ens dot fr 2007-10-24 11:36 ---
With the patch in comment #4, the code in #2 yields the same results as other
compilers without regression.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33759
While trying to update to revision 129600, I got the following error:
gcc -c -g -fkeep-inline-functions -DIN_GCC -W -Wall -Wwrite-strings
-Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition
-Wmissing-format-attribute -fno-common -DHAVE_CONFIG_H -DGENERATOR_FILE -I.
-Ibuild -I../.
--- Comment #13 from rguenth at gcc dot gnu dot org 2007-10-24 11:56
---
And another one:
http://gcc.gnu.org/ml/gcc-patches/2007-10/msg01398.html
unassigning - I neither feel responsible nor competent enough to fix this
design
issue.
--
rguenth at gcc dot gnu dot org changed:
--- Comment #15 from rguenth at gcc dot gnu dot org 2007-10-24 12:01
---
Btw, I can no longer reproduce the problem. var-tracking is not fast, but
even a cc1 compiled with checking and -O0 does not take multiple hours to
compile this (but just a minute):
variable tracking : 11.3
--- Comment #16 from zadeck at naturalbridge dot com 2007-10-24 12:14
---
Subject: Re: inf loop/long compile time, time spent in var-tracking.c
rguenth at gcc dot gnu dot org wrote:
> --- Comment #15 from rguenth at gcc dot gnu dot org 2007-10-24 12:01
> ---
> Btw, I can no
This is with both 4.2 and 4.3 on x86_64/Linux.
gfortran-4.3 -fopenmp a.f90
a.f90: In function 'MAIN__':
a.f90:3: internal compiler error: in extract_omp_for_data, at omp-low.c:162
Test case (by Xavier Andrade):
--
program testbug
integer :: ii
!$omp parallel do
do ii
--- Comment #31 from rguenth at gcc dot gnu dot org 2007-10-24 12:23
---
This is a partitioning issue. We allocate the same MPT to loads and stores
in a loop. Fixed with for example --param max-aliased-vops=1.
Diego, this is yours. Avoiding false aliasing of loads and stores sho
--- Comment #1 from andreasmeier80 at gmx dot de 2007-10-24 12:28 ---
The last commit of Olga was not complete. The two new files weren't commited
--
andreasmeier80 at gmx dot de changed:
What|Removed |Added
--- Comment #4 from matz at gcc dot gnu dot org 2007-10-24 12:54 ---
Subject: Bug 33868
Author: matz
Date: Wed Oct 24 12:54:24 2007
New Revision: 129602
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=129602
Log:
PR debug/33868
* var-tracking.c (variable_union): D
--- Comment #5 from matz at gcc dot gnu dot org 2007-10-24 12:55 ---
Fixed.
--
matz at gcc dot gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--
ubizjak at gmail dot com changed:
What|Removed |Added
URL||http://gcc.gnu.org/ml/gcc-
||
--- Comment #2 from dominiq at lps dot ens dot fr 2007-10-24 13:20 ---
I am now at stage2 with revision 129601. So this PR seems fixed.
Thanks
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33879
--- Comment #1 from jakub at gcc dot gnu dot org 2007-10-24 13:23 ---
/* PR middle-end/33880 */
/* { dg-do compile } */
/* { dg-options "-O2 -fopenmp" } */
int
foo (void)
{
int i = 0;
void bar (void) { i++; }
bar ();
#pragma omp parallel for
for (i = 0; i < 10; i++)
;
--- Comment #3 from ubizjak at gmail dot com 2007-10-24 13:46 ---
Closed as fixed.
--
ubizjak at gmail dot com changed:
What|Removed |Added
Status|UNCONFIRMED
With this program I get a runtime error message
Fortran runtime error: Attempt to allocate a negative amount of memory.
The error occurs at line 33. I suspect the "allocation" is the constant array
actual argument (/name/).
This program is a little longer than I like my bug reports to be, but any
--- Comment #1 from william dot mitchell at nist dot gov 2007-10-24 14:10
---
Created an attachment (id=14405)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14405&action=view)
this is the program
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33881
--- Comment #32 from rguenth at gcc dot gnu dot org 2007-10-24 15:02
---
I have a patch to "fix" the heuristics, but it doesn't have effect as nobody
hoists PHI nodes that have become invariant apperantly. After PRE I see
:
# HEAP.202_828 = PHI
:
# SMT.222_639 = PHI
# dtvol
http://ffmpeg.mplayerhq.hu/download.html and in libavcodec.
With gcc-4.3.0-alpha20071019:
gcc -I"/usr/src/CVS/ffmpeg-gcc43"/libavcodec -DHAVE_AV_CONFIG_H
-I"/usr/src/CVS/ffmpeg-gcc43" -I"/usr/src/CVS/ffmpeg-gcc43"/libavutil -O3 -o
elbg.o elbg.c
elbg.c: In function âtry_shift_candidateâ:
elbg.
--- Comment #33 from rguenth at gcc dot gnu dot org 2007-10-24 15:18
---
DOM does, so scheduling another dominator pass after PRE (where alias is
re-run)
fixes the problem. One problem with partitioning is that we don't collect
statistics for symbols inside a partition, which makes all
--- Comment #34 from rguenth at gcc dot gnu dot org 2007-10-24 15:19
---
Created an attachment (id=14406)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14406&action=view)
partition heuristics change
The idea is to change the statistics to make the values in the fields
match their
--- Comment #2 from burnus at gcc dot gnu dot org 2007-10-24 15:33 ---
> Fortran runtime error: Attempt to allocate a negative amount of memory.
This error usually means that you allocate that much memory that an integer(4)
variable overflows. I cannot reproduce the error here with -m32
--- Comment #1 from pinskia at gcc dot gnu dot org 2007-10-24 15:37 ---
*** This bug has been marked as a duplicate of 33680 ***
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--
--- Comment #19 from pinskia at gcc dot gnu dot org 2007-10-24 15:37
---
*** Bug 33882 has been marked as a duplicate of this bug. ***
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
-
--
jason at gcc dot gnu dot org changed:
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |jason at gcc dot gnu dot org
|dot org
--- Comment #3 from william dot mitchell at nist dot gov 2007-10-24 15:48
---
(In reply to comment #2)
Sorry I forgot to give the computer information. This is a 32 bit CentOS 5
Linux distribution on an Intel Core 2 Duo T7700
[EMAIL PROTECTED]> gfortran -v
Using built-in specs.
Target
This is the same error message reported in bug fortran/33058 on August 13th,
2007, however, the suggestion provided in that thread did not solve the problem
for me. Just wondering if there is something else that might be the problem.
When I try to compile gcc-4.2.2 on Linux to create the gfortran
--- Comment #5 from janis at gcc dot gnu dot org 2007-10-24 17:23 ---
Rob, thanks for investigating this and finding new values. No one seems to
have paid any attention here, so please send this as a patch to
[EMAIL PROTECTED] (let me know privately if you have questions about
that) and
--- Comment #15 from rsandifo at gcc dot gnu dot org 2007-10-24 17:52
---
Subject: Bug 33755
Author: rsandifo
Date: Wed Oct 24 17:52:16 2007
New Revision: 129606
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=129606
Log:
gcc/
PR target/33755
* config/mips/mips.c
--- Comment #16 from rsandifo at gcc dot gnu dot org 2007-10-24 17:54
---
Subject: Bug 33755
Author: rsandifo
Date: Wed Oct 24 17:54:40 2007
New Revision: 129607
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=129607
Log:
gcc/
PR target/33755
* config/mips/mips.c
--- Comment #17 from rsandifo at gcc dot gnu dot org 2007-10-24 17:55
---
Patches applied to mainline and 4.2.
--
rsandifo at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #4 from ubizjak at gmail dot com 2007-10-24 17:58 ---
I'm reopening this bug, because now we fail on x86_64:
cc1: warnings being treated as errors
../../gcc-svn/trunk/gcc/ipa-struct-reorg.c: In function create_new_field_acc:
../../gcc-svn/trunk/gcc/ipa-struct-reorg.c:960:
--
ubizjak at gmail dot com changed:
What|Removed |Added
CC||ubizjak at gmail dot com
Status|UNCONFIRMED |NE
--
olga at gcc dot gnu dot org changed:
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |olga at gcc dot gnu dot org
|dot org
$ cat data.f
subroutine foo
real a
data a /1.3/
end
$ gfortran -c -Wall -O3 data.f
$ g77 -c -Wall -O3 data.f
data.f: In subroutine `foo':
In file included from data.f:0:
data.f:2: warning: unused variable 'a'
I'd expect a warning, like g77.
--
Summary: data-i
--- Comment #6 from tkoenig at gcc dot gnu dot org 2007-10-24 19:30 ---
I'll do this together with PR 33698.
--
tkoenig at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #35 from jason at gcc dot gnu dot org 2007-10-24 19:48 ---
Created an attachment (id=14407)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14407&action=view)
updated patch
Here's a cleaned up version of Liu Guanwei's patch that passes regression
testing with current 4.3
--
jason at gcc dot gnu dot org changed:
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |jason at gcc dot gnu dot org
|dot org
class A{
//
};
class B{
public:
explicit B(const A& a)
: i_a(a)
{
}
~B()
{
}
private:
const A& i_a;
};
A returnA( const char* arg )
{
return A(/*arg*/);
}
foo()
{
const A& aRef = returnA("FirstObject"); // . (1)
{
const B b(returnA("SecondObject"));/
--- Comment #1 from gzljg at hotmail dot com 2007-10-24 20:15 ---
Created an attachment (id=14408)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14408&action=view)
a compilable C source
compile and run it to see the actual outcome.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?
--- Comment #5 from olga at gcc dot gnu dot org 2007-10-24 22:06 ---
Two patches were provided by Samuel Tardieu:
http://gcc.gnu.org/ml/gcc-patches/2007-10/msg01427.html
http://gcc.gnu.org/ml/gcc-patches/2007-10/msg01426.html
Slightly modified:
http://gcc.gnu.org/ml/gcc-patches/2007-10
--
olga at gcc dot gnu dot org changed:
What|Removed |Added
CC||sam at rfc1149 dot net
Status|ASSIGNED|W
--- Comment #1 from jvdelisle at gcc dot gnu dot org 2007-10-24 22:57
---
Try reversing the order here:
./configure --enable-languages=fortran --with-gmp=/home/local/gmp-4.2.2
--with-mpfr=/home/local/mpfr-2.3.0
to:
./configure --enable-languages=fortran --with-mpfr=/home/local/mpfr-
--- Comment #2 from jvdelisle at gcc dot gnu dot org 2007-10-24 23:15
---
Also, you must enable c as well: This is biggest problem
--enable-languages=c,fortran
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33883
--- Comment #35 from pthaugen at gcc dot gnu dot org 2007-10-24 23:15
---
I reran leslie3d on PowerPC with --param max-aliased-vops=1. The result was
a 90% improvement over my prior revision 129550 run (218% improvement over the
original mainline run).
--
http://gcc.gnu.org/bug
While compiling OpenOffice.org with gcc 4.3 at -O2 or higher (-O1 and -O0 are
ok):
/usr/src/ark/BUILD/ooo-build/build/oog680-m7/i18npool/source/nativenumber/nativenumbersupplier.cxx:85:
internal compiler error: canonical types differ for identical types sal_Unicode
[31] and sal_uInt16 [31]
Please
--- Comment #1 from bero at arklinux dot org 2007-10-24 23:25 ---
Created an attachment (id=14409)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14409&action=view)
bzip2-ed preprocessed source
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33886
This C++ test case passes a bitfield value to a function which expects a const
reference. When optimizing, the function gets the wrong value: it gets the
value 0x100, which is an acceptable value for a 24-bit bitfield, but is the
wrong value for extending the bitfield out to a 32-bit integer.
--- Comment #1 from pinskia at gcc dot gnu dot org 2007-10-24 23:51 ---
As far as I know this code is invlaid as you should not able take the address
of the bitfield.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33887
The following causes an ICE:
$ cat ftn95bug.f90
program ftn95bug
implicit none
character(8) :: indata(4) = (/ &
'12344321', '98766789', 'abcdefgh', 'ABCDEFGH' &
/)
call process (myfunc (indata)) ! <- This causes a gfortran ICE !
contains
elemental function myfunc (s)
chara
--- Comment #37 from danglin at gcc dot gnu dot org 2007-10-25 02:31
---
Test is failing on hppa-unknown-linux-gnu:
Executing on host:
/home/dave/gnu/gcc-4.3/objdir/gcc/testsuite/gfortran/../../gf
ortran -B/home/dave/gnu/gcc-4.3/objdir/gcc/testsuite/gfortran/../../
/home/dave/
gnu/gcc-
--- Comment #38 from dave at hiauly1 dot hia dot nrc dot ca 2007-10-25
02:39 ---
Subject: Re: wrong types in character array/scalar binop
Tree dump attached.
Dave
--- Comment #39 from dave at hiauly1 dot hia dot nrc dot ca 2007-10-25
02:39 ---
Created an attachment (id=14
For this simple code
package P is
type T is limited interface;
function F return not null access T'Class;
end P;
package body P is
function X return not null access T'Class;
pragma Import (C, X);
function F return not null access T'Class is
begin
return Result : not null a
ix
gcc version 4.3.0 20071024 (experimental) (GCC)
I append the preprocessed source code.
--
Summary: internal compiler error: in tsubst_copy
Product: gcc
Version: 4.3.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
--- Comment #1 from schnetter at aei dot mpg dot de 2007-10-25 04:31
---
Created an attachment (id=14411)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14411&action=view)
Failing preprocessed source code
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33890
[forwarded from http://bugs.debian.org/445930]
this is regression against the classpath generices backport for 4.1/4.2 as used
by some distributions, found in 4.3, fixed in the classpath-0.96.1 merge.
import javax.swing.*;
import java.awt.*;
public class Exo1 extends JFrame{
public Exo1 (){
--- Comment #2 from ian at airs dot com 2007-10-25 05:07 ---
Nothing is taking the address of a bitfield. It's a const reference, which
should get initialized with the value. If the reference is not const, then the
compiler gives an error.
In any case, this code fails the same way, an
--- Comment #3 from pinskia at gcc dot gnu dot org 2007-10-25 05:21 ---
int f2 = sv.f2;
That should sign extend as far as I know.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33887
--- Comment #1 from burnus at gcc dot gnu dot org 2007-10-25 06:30 ---
Valgrind:
==28333== Invalid read of size 8
==28333==at 0x73B006: get_frame_type (tree-nested.c:198)
==28333==by 0x73B387: get_chain_decl (tree-nested.c:304)
==28333==by 0x73C55C: get_nonlocal_debug_decl (
--- Comment #40 from burnus at gcc dot gnu dot org 2007-10-25 06:36 ---
(In reply to comment #37)
> Test is failing on hppa-unknown-linux-gnu:
> PASS: gfortran.dg/char_cast_1.f90 -O (test for excess errors)
> FAIL: gfortran.dg/char_cast_1.f90 -O scan-tree-dump-times \[S.5\]\[1\] 2
W
71 matches
Mail list logo
