The attached code generates the subject error. The subroutine in question
compiles perfectly if taken outside of the module. Might be related to PR19546?
The message from the compiler refers to a different internal, though.
[EMAIL PROTECTED] F90]$ gfortran -v -c tpc.f90
Using built-in specs.
--- Additional Comments From sfilippone at uniroma2 dot it 2005-02-15
08:05 ---
Created an attachment (id=8199)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=8199action=view)
testcase
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19971
When returning local function pointer, it doesn't produce any warnings.
For example:
int *h() {
int z;
return z;
}
int (*apply (int (*f) (const void *, const void *), void *a))(const void *)
{
int zozo(const void * b) { return f(a,b); };
return zozo;
}
Compile with -O2 -Wall, it
--- Additional Comments From dorit at il dot ibm dot com 2005-02-15 08:44
---
This happens during tree-if-conversion pass (before vectorization) (tree-if-
conversion is enabled by -ftree-vectorize)
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19952
--- Additional Comments From ddonovan at latentzero dot com 2005-02-15
09:30 ---
Subject: RE: error: incomplete type `Environment' used in nested name specifier
Thank you for putting your time into this but if you look at the
attached file, it is basically a simplified program of what
--- Additional Comments From Thomas dot Koenig at online dot de 2005-02-15
10:29 ---
And in fact this only can happen with -funsafe-math-optimizations (or maybe
with -fno-trapping-
math because a+0.0 can trap.
Hmm...
if b is complex and has the value (0., signalling NaN) and a is
ice.cpp:8: internal compiler error: in tree_low_cst, at tree.c:3315
The problem happens with gcc-3.3 also.
Environment:
System: Linux grajagop-lnx 2.6.10-1-686 #1 Tue Jan 18 04:34:19 EST 2005 i686
GNU/Linux
Architecture: i686
host: i486-pc-linux-gnu
build:
--- Additional Comments From jakub at gcc dot gnu dot org 2005-02-15 11:42
---
Patch here: http://gcc.gnu.org/ml/gcc-patches/2005-02/msg00810.html
--
What|Removed |Added
--- Additional Comments From uweigand at gcc dot gnu dot org 2005-02-15
12:18 ---
The smaller test case also still ICEs on s390. Thanks!
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19865
--- Additional Comments From reichelt at gcc dot gnu dot org 2005-02-15
12:32 ---
Not fixed on 3.4 branch.
--
What|Removed |Added
Status|RESOLVED
--- Additional Comments From reichelt at gcc dot gnu dot org 2005-02-15
12:33 ---
Not fixed on 3.3 and 3.4 branch.
--
What|Removed |Added
Status|RESOLVED
--- Additional Comments From reichelt at gcc dot gnu dot org 2005-02-15
12:56 ---
Zdenek, are you going to add a testcase for the problem on mainline?
Is this a latent problem on other branches?
If not, can this PR be closed?
--
What|Removed |Added
--- Additional Comments From dorit at il dot ibm dot com 2005-02-15 13:23
---
I'm testing this patch (it solves this PR):
Index: tree-vectorizer.c
===
RCS file: /cvs/gcc/gcc/gcc/tree-vectorizer.c,v
retrieving revision
--- Additional Comments From rmathew at gcc dot gnu dot org 2005-02-15
13:25 ---
Subject: Re: Java Unnecessarily Disabled for MinGW
[I'm far from a GCC demigod, but...]
Q: As a (new/cautious) target co-maintainer, is this within my domain to fix
without seeking approval?
I would
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
13:29 ---
You are using the C++ front-end and not the C front-end.
Anyways this is a duplicate of bug 18327.
*** This bug has been marked as a duplicate of 18327 ***
--
What|Removed
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
13:29 ---
*** Bug 19973 has been marked as a duplicate of this bug. ***
--
What|Removed |Added
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
14:14 ---
Confirmed.
--
What|Removed |Added
Severity|enhancement |minor
--- Additional Comments From jakub at gcc dot gnu dot org 2005-02-15 14:14
---
The bug is present even with -O2 -finline-functions -finline-limit=604
-fno-strict-aliasing.
strcmp is passed main::b._M_t._M_impl._M_node_count
as one of its arguments.
Looking into it.
--
What
--- Additional Comments From falk at debian dot org 2005-02-15 14:14
---
(In reply to comment #2)
Thank you for your answer. I have installed R on a LINUX machine
instead.
Hmm, if there is no way to reproduce it, I guess we can only close this report.
--
What
general info
gcc --version:
gcc (GCC) 3.3.4
uname -a:
Linux trinita 2.4.26 #6 Mon Jun 14 19:07:27 PDT 2004 i686 unknown unknown
GNU/Linux
compiled with:
gcc bug.c -O1 -save-temps -o bug
The preprocessed files are in the attachment.
Description
---
when compiling w/out
On Feb 15, 2005, at 10:45 AM, Giorgenes Gelatti wrote:
general info
gcc --version:
gcc (GCC) 3.3.4
uname -a:
Linux trinita 2.4.26 #6 Mon Jun 14 19:07:27 PDT 2004 i686 unknown
unknown GNU/Linux
compiled with:
gcc bug.c -O1 -save-temps -o bug
mpixel = (double)((unsigned int)h
--- Additional Comments From jakub at gcc dot gnu dot org 2005-02-15 15:53
---
The bug is the infamous RTX_UNCHANGING_P, apparently still not dead.
Well, now under the MEM_READONLY_P name.
The problem is that set_mem_attributes_minus_bitpos:
1532 tree base = get_base_address
--- Additional Comments From jakub at gcc dot gnu dot org 2005-02-15 15:55
---
s/target hook/langhook/.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19813
--
What|Removed |Added
AssignedTo|nobody at gcc dot gnu dot |unassigned at gcc dot gnu
|org |dot org
Status|ASSIGNED
Checking Lapack with flag_complex_method = 2 on
ia64-unknown-linux-gnu with -O0, I got a hang in xeigtstc
when processing sep.in, in the routine slarrb.f.
The compiler was the 20050213 snapshot.
This may not have anything to do with the setting of
flag_complex_method, because the routine in
--
What|Removed |Added
OtherBugsDependingO||5900
nThis||
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19974
--
What|Removed |Added
Keywords||wrong-code
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19974
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
17:24 ---
The fix was applied so closing.
--
What|Removed |Added
Status|NEW
--- Additional Comments From rth at gcc dot gnu dot org 2005-02-15 17:24
---
The front end isn't supposed to set TREE_READONLY for variables that need
constructing at runtime. That's been the way we've handled this case from
day one.
--
What|Removed
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
17:25 ---
By the way here is the changelog:
* Makefile.am (libgcj_la_SOURCES): Move all sources ...
(libgcj0_convenience_la_SOURCES): ... here.
(libgcj_la_LIBADD): Add libgcj0_convenience.la.
//Here is the original source code describing the problem:
#include cstdio
#include new
using namespace std;
struct Simple
{
static void* operator new[](std::size_t, const std::nothrow_t) throw()
{ return 0; }
static void operator delete[](void* p, std::size_t) throw()
{} //
--
What|Removed |Added
Keywords||wrong-code
Known to fail||3.0.4 3.2.3 3.3.3
Known to work|
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
17:43 ---
This is a dup of bug 13215 which is fixed for 3.4.0.
*** This bug has been marked as a duplicate of 13215 ***
--
What|Removed |Added
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
17:44 ---
*** Bug 19975 has been marked as a duplicate of this bug. ***
--
What|Removed |Added
--- Additional Comments From jrydberg at gnu dot org 2005-02-15 18:30
---
Still happens on 4.0 snapshot of 20050213 :
$ /gnu/bin/gcc -c pr10676.c
pr10676.c:9: error: unknown field 'foo' specified in initializer
$ /gnu/bin/gcc -v
...
gcc version 4.0.0 20050213 (experimental)
--
The test
enum e { E = 0 * (1 / 0) };
should yield an error with -pedantic-errors. Not a regression.
--
Summary: integer division by zero in subexpression should be
overflow
Product: gcc
Version: 4.0.0
Status: UNCONFIRMED
The test
#include limits.h
void
f (void)
{
int c = INT_MAX + 1;
}
should yield only warnings, not an error, with -pedantic-errors.
Not a regression.
--
Summary: overflow in non-static initializer should not be pedwarn
Product: gcc
Version: 4.0.0
The test
#include limits.h
int
f (void)
{
return INT_MAX + 1 - INT_MAX;
}
should yield only a single integer overflow in expression warning,
but yields two. Not a regression.
--
Summary: overflow in expression of constants should not cause
multiple warnings
Sometimes the %sil/%dil AMD64 registers are allocated for inline assembler. The
smallest reproducible test case I found is the following code:
int test()
{
register unsigned char a = 1, b = 2, c = 3, d = 4, e = 5, f = 6;
__asm__ __volatile__ (
movb %0, %1\n
movb %2, %3\n
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
19:10 ---
I'm of the opinion that the source is invalid. Using a =q constraint
instead of an =r constraint properly shows that the compiler must choose
one of the four low byte registers.
This is a dup of bug
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
19:10 ---
*** Bug 19979 has been marked as a duplicate of this bug. ***
--
What|Removed |Added
The following invalid code snippet causes an ICE on mainline:
int foo;
templateint void foo() {}
The error message reads:
bug.cc:2: error: 'templateint anonymous void foo()' redeclared as
different kind of symbol
bug.cc:1: error:
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-02-15
19:39 ---
Subject: Bug 17788
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-02-15 19:39:31
Modified files:
gcc/cp : ChangeLog class.c
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
19:50 ---
Confirmed:
: Search converges between 2004-02-01-trunk (#445) and 2004-03-01-trunk (#446).
--
What|Removed |Added
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
19:53 ---
Fixed.
--
What|Removed |Added
Status|ASSIGNED|RESOLVED
--- Additional Comments From Thomas dot Koenig at online dot de 2005-02-15
20:14 ---
This does not happen on an Athlon-xp with -march=athlon-xp
-mfmath=sse. Might be target or 64-bit specific.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19974
I am trying to build gcc-3.4.3 on solaris 8. I get to the make install point
and then there is a problem. It appears the INSTALL variable is not getting
set right for the install step in the various directories.
I have the source in /gcc-3.4.3 and the build dir is /gcc-3.4.3_sun5.8
This is a bug report for RHIDE 1.5.0.1 (Apr 8 2003 18:05:46)
I am running MS-DOS 8.0 (Windows 95)
Important environment variables:
PATH=C:\Arquivos de programas\Analog Devices\VisualDSP\System;C:\Arquivos de
programas\Analog
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
20:42 ---
A call for testing patch is here:
http://gcc.gnu.org/ml/gcc-patches/2005-02/msg00858.html.
--
What|Removed |Added
I am trying to build gcc-3.4.3 on solaris 8. I get to the make install point
and then there is a problem. It appears the INSTALL variable is not getting
set right for the install step in the various directories.
I have the source in /gcc-3.4.3 and the build dir is /gcc-3.4.3_sun5.8
Here
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
20:51 ---
*** Bug 19981 has been marked as a duplicate of this bug. ***
--
What|Removed |Added
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
20:51 ---
It has nothing to do with nested structs at all but an array's whos size is not
specified. This is valid C99
but not C++. Though we should accept it for C++ also but we don't which is a
bug. Also note
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-02-15
20:52 ---
Subject: Bug 14940
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-02-15 20:52:14
Modified files:
gcc: ChangeLog
gcc/config :
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-02-15
20:52 ---
Subject: Bug 19300
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-02-15 20:52:14
Modified files:
gcc: ChangeLog
gcc/config :
This is PR 13993 which only effects if you configure with a relative
patch and
only 3.4.x, it is fixed on the mainline already.
-- Pinski
--- Additional Comments From phython at gcc dot gnu dot org 2005-02-15
20:56 ---
Fixed with the previous commit
--
What|Removed |Added
Status|ASSIGNED
--
What|Removed |Added
Target Milestone|--- |4.0.0
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19300
Sourc code for t.cpp:
int mymain()
{
int i = 1;
int j = 2;
const volatile int ri = i;
const_castint *(ri) = j; // error - lvalue required for =
return 0;
}
The __builtin_nan functions should allow 0X as well as 0x in their arguments.
real.c:real_nan only allows 0x at present. In
double n1 = __builtin_nan(0x1);
double n2 = __builtin_nan(0X1);
only the first initializer is accepted, but both should be.
--
Summary: __builtin_nan should
The functions nan, nanf, nanl should be C99 built-in functions rather
than just being built-in in the __builtin_ form.
They were wrongly C90 built-ins, changed to be __builtin_ only to fix
bug 14635. At that time bug 14649 meant they shouldn't be C99 built-ins,
but bug 14649 has been fixed so
I am trying to build gcc-3.4.3 on solaris 8. I get to the make install
point and then there is a problem. It appears the INSTALL variable is not
getting set right for the install step in the various directories.
I have the source in /gcc-3.4.3 and the build dir is
/gcc-3.4.3_sun5.8
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
21:10 ---
Confirmed but this is minor as this is an ICE after an error. Also note on the
mainline we get one error
message and no ICE (or confused by earlier errors, bailing out).
--
What|Removed
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
21:13 ---
Confirmed.
--
What|Removed |Added
Status|UNCONFIRMED |NEW
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
21:14 ---
Confirmed.
--
What|Removed |Added
Status|UNCONFIRMED |NEW
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
21:15 ---
Can you try with 3.4.3?
Also can you provide the output of gcc -v?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18513
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
21:18 ---
Confirmed.
--
What|Removed |Added
Status|UNCONFIRMED |NEW
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
21:20 ---
Confirmed.
--
What|Removed |Added
Status|UNCONFIRMED |NEW
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
21:28 ---
The bug is very obvious from the code:
if (*str == '0')
{
if (*++str == 'x')
str++, base = 16;
else
base = 8;
}
Note no check for the
I created a small test I will attach to this bug report that shows that
executables created with -fprofile-arcs -ftest-coverage crash in gcov_exit ()
if they dynamically load functions from libraries.
I tested with gcc 3.4.1 and 4.0-20050213. Both show the same problem.
Here's the gdb
--
What|Removed |Added
CC||hubicka at gcc dot gnu dot
||org
Component|c++
--- Additional Comments From lothar at xcerla dot com 2005-02-15 21:46
---
Created an attachment (id=8201)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=8201action=view)
test case to reproduce the segfault
This is a project of a small test case that reproduces the segmentation
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
21:55 ---
Hmm, this looks like we don't deal with dlclose correctly.
Honza could you comment on this if there is a way to deal with dlclose?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19985
--- Additional Comments From phython at gcc dot gnu dot org 2005-02-15
21:58 ---
Fixed.
--
What|Removed |Added
Status|ASSIGNED|RESOLVED
Just a meta-bug for all the missing fold which are missing which are in the RTL
version of simplify_*.
--
Summary: [meta-bug] fold missing optimizations (compared to RTL)
Product: gcc
Version: 4.0.0
Status: UNCONFIRMED
Keywords:
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
22:14 ---
Confirmed, there might be others which I had forgot.
--
What|Removed |Added
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
22:22 ---
As I said before I was not going to fix a fold problem (for the first testcase).
--
What|Removed |Added
--
What|Removed |Added
OtherBugsDependingO||19987
nThis||
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19986
Just a meta-bug for missing fold optimizations.
--
Summary: [meta-bug] fold missing optimizations in general
Product: gcc
Version: 4.0.0
Status: UNCONFIRMED
Keywords: missed-optimization, meta-bug
Severity: enhancement
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
22:31 ---
Confirmed, added the most I know of.
--
What|Removed |Added
BugsThisDependsOn|
When I compile the following code with 'gcc -O3 --save-temps -c':
double foo(double x, double y)
{
return ((x + 0.1234 * y) * (x - 0.1234 * y));
}
gcc 3.x gives one load of the constant 0.1234, one multiplication
0.1234 * y, one addition, one subtraction, and the final
multiplication:
The following testcase shows some inconsistency w.r.t zero-size arrays.
Without -pedantic we diagnose the array(s) whose dimension is found via
dependent lookup.
==
templateint struct A
{
static const int i = 0;
};
templateint N struct B
{
int
Internal compiler error in compiling a f90-module of the HSL library
on RedHat Fedora Core 3 Linux, i686, IBM Thinkpad, Kernel 2.6.10
version informations
./gcc -v
Using built-in specs.
Target: i686-pc-linux-gnu
Configured with: ../gcc/configure --enable-languages=c,f95
The following valid code snippet is rejected since gcc 3.4.0:
=
enum { e = 1 };
templatetypename struct A
{
static const int i = e;
char a[i];
};
=
The error message is:
bug.cc:6: error: size of array `a' is not an
--
What|Removed |Added
Component|c |middle-end
Keywords||missed-optimization
Target Milestone|---
--- Additional Comments From reichelt at gcc dot gnu dot org 2005-02-15
23:16 ---
Sorry, this is related to PR19883, of course.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19991
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-02-15
23:17 ---
Subject: Bug 19019
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-02-15 23:16:49
Modified files:
gcc: ChangeLog
: ChangeLog macro.c
Added files:
gcc/testsuite/gcc.dg/cpp: 20050215-1.c
Log message:
2005-02-15 Eric Christopher [EMAIL PROTECTED]
PR preprocessor/19077
* macro.c (cpp_macro_definition): Move handling of whitespace
to PREV_WHITE conditional. Remove
--- Additional Comments From echristo at redhat dot com 2005-02-15 23:18
---
Fixed.
--
What|Removed |Added
Status|ASSIGNED|RESOLVED
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
23:20 ---
Confirmed.
--
What|Removed |Added
CC||pinskia
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
23:23 ---
Patch here: http://gcc.gnu.org/ml/gcc-patches/2005-02/msg00885.html.
--
What|Removed |Added
--- Additional Comments From bjoern dot m dot haase at web dot de
2005-02-15 23:24 ---
FYI: I have just tested the patch posted in
http://gcc.gnu.org/ml/gcc-patches/2005-02/msg00858.html
After applying this patch, the build for the avr target again succeeds.
Yours,
Björn
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-02-15
23:29 ---
Subject: Bug 19955
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-02-15 23:29:52
Modified files:
libstdc++-v3 : ChangeLog
C++ has the concepts of hosted and freestanding implementations, so the
-fhosted and -ffreestanding flags should be supported for C++ rather
than being C-only.
--
Summary: -fhosted/-ffreestanding for C++
Product: gcc
Version: 4.0.0
Status: UNCONFIRMED
--- Additional Comments From reichelt at gcc dot gnu dot org 2005-02-15
23:36 ---
Not fixed on the 3.4 branch.
--
What|Removed |Added
Status|RESOLVED
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
23:42 ---
Almost want to say this is caused by:
2004-07-11 Roger Sayle [EMAIL PROTECTED]
* fold-const.c (fold) PLUS_EXPR: Canonicalize X + -C as X - C for
floating point additions, to keep real
--- Additional Comments From echristo at redhat dot com 2005-02-15 23:45
---
I suppose I could apply the patch there too.
--
What|Removed |Added
Status|REOPENED
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-02-15
23:48 ---
Confirmed.
--
What|Removed |Added
Status|UNCONFIRMED |NEW
--- Additional Comments From dwatkins at tascsystems dot com 2005-02-16
00:04 ---
Regarding this bug and AVRFreaks Forum topic:
http://www.avrfreaks.com/index.php?
name=PNphpBB2file=viewtopict=23436highlight=sid=e1dbd9f208947a5efc679075b4af
f237
in which the debug info for code past
--- Additional Comments From amodra at bigpond dot net dot au 2005-02-16
00:28 ---
This has been fixed, most likely by
2004-10-13 Richard Henderson [EMAIL PROTECTED]
* crtstuff.c (__FRAME_END__): Remove mode attribute. Find 32-bit
integer from internal limits macros.
--- Additional Comments From amodra at bigpond dot net dot au 2005-02-16
01:03 ---
Changing to target since this is due to Aldy's altivec.h implementation.
--
What|Removed |Added
1 - 100 of 152 matches
Mail list logo