--- Comment #18 from rakdver at atrey dot karlin dot mff dot cuni dot cz
2006-02-01 08:18 ---
Subject: Re: [4.0/4.1/4.2 Regression] ICE on throw code
Hello,
Zdenek, have you submitted the patch yet for mainline?
no, I was waiting for reactions on my questions, so that I am sure
--- Comment #19 from mark at codesourcery dot com 2006-02-01 08:21 ---
Subject: Re: [4.0/4.1/4.2 Regression] ICE on throw code
rakdver at atrey dot karlin dot mff dot cuni dot cz wrote:
Zdenek, have you submitted the patch yet for mainline?
no, I was waiting for reactions on my
--- Comment #7 from mmitchel at gcc dot gnu dot org 2006-02-01 08:51
---
The problem is that determine_specialization decides that the out-of-class
declaration is a specialization of the in-class declaration, even though it's
not. We should probably observe that the out-of-class
Following program works differently - if compiled with flag -g or without:
--
#include iostream
using namespace std;
int main(){
unsigned char a[10];
a[0] = 'A';
a[2] = 'B';
a[0] ^= a[2] ^= a[0] ^= a[2];
cout a0=a[0] a2=a[2]endl;
}
==
--- Comment #3 from jakub at gcc dot gnu dot org 2006-02-01 10:17 ---
For the first problem, you can use something like:
+#if FFI_TRAMPOLINE_SIZE == 10
+/* Precalculated for crappy assemblers. */
+#define RAW_CLOSURE_CIF_OFFSET 12
+#define RAW_CLOSURE_FUN_OFFSET 16
+#define
--- Comment #1 from pluto at agmk dot net 2006-02-01 10:23 ---
It's not a gcc bug. The code relies on the results of intermediate
subexpressions. According to Stroustrup, The C++ Programming Language, section
6.2.2, The order of evaluation of subexpressions within
an expression is
--- Comment #2 from rguenth at gcc dot gnu dot org 2006-02-01 10:30 ---
Subject: Bug 26055
Author: rguenth
Date: Wed Feb 1 10:30:43 2006
New Revision: 110469
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=110469
Log:
2006-02-01 Richard Guenther [EMAIL PROTECTED]
PR
--- Comment #2 from rguenth at gcc dot gnu dot org 2006-02-01 10:32 ---
Subject: Bug 26059
Author: rguenth
Date: Wed Feb 1 10:32:22 2006
New Revision: 110470
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=110470
Log:
2006-02-01 Richard Guenther [EMAIL PROTECTED]
PR
--- Comment #2 from schwab at suse dot de 2006-02-01 10:34 ---
*** This bug has been marked as a duplicate of 11751 ***
--
schwab at suse dot de changed:
What|Removed |Added
--- Comment #58 from schwab at suse dot de 2006-02-01 10:34 ---
*** Bug 26060 has been marked as a duplicate of this bug. ***
--
schwab at suse dot de changed:
What|Removed |Added
I think it would be nice if gcc would give an error/warning count at the end of
its output. Something like:
$ gcc -Wall test.c
test.c:3: error: conflicting types for 'p'
test.c:2: error: previous declaration of 'p' was here
test.c:3: error: conflicting types for 'p'
test.c:2: error: previous
--- Comment #5 from dtemirbulatov at gmail dot com 2006-02-01 11:01 ---
no such regression on mainline
$ ./gcc-exec/bin/x86_64-unknown-linux-gnu-g++ -v
Using built-in specs.
Target: x86_64-unknown-linux-gnu
Configured with: ../gcc-20051026/configure --target=x86_64-unknown-linux-gnu
--- Comment #3 from vitaly at siliconds dot com 2006-02-01 11:08 ---
Thank you very much.
--
vitaly at siliconds dot com changed:
What|Removed |Added
--- Comment #4 from eedelman at gcc dot gnu dot org 2006-02-01 11:29
---
(In reply to comment #2)
How about this?
As far as I can see, this latter testcase is valid code. Confirmed.
An additional comment:
If I change the call to the generic subroutine bar_ in xxx to a call to
--- Comment #3 from rguenth at gcc dot gnu dot org 2006-02-01 11:36 ---
This is fixed/worked around on the mainline. A completer fix was posted here
http://gcc.gnu.org/ml/gcc-patches/2006-02/msg00026.html
for testing.
--
rguenth at gcc dot gnu dot org changed:
What
--- Comment #3 from rguenth at gcc dot gnu dot org 2006-02-01 11:36 ---
Fixed.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Status|NEW
--- Comment #3 from rguenth at gcc dot gnu dot org 2006-02-01 11:38 ---
Can you provide the patch that fixed the issue for you?
Thanks,
Richard.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #13 from reichelt at gcc dot gnu dot org 2006-02-01 11:46
---
Subject: Bug 24278
Author: reichelt
Date: Wed Feb 1 11:46:53 2006
New Revision: 110471
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=110471
Log:
Backport:
2005-12-19 Mark Mitchell
--- Comment #14 from reichelt at gcc dot gnu dot org 2006-02-01 11:48
---
Now also fixed on the 3.4 branch.
--
reichelt at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #9 from reichelt at gcc dot gnu dot org 2006-02-01 11:52
---
Subject: Bug 24915
Author: reichelt
Date: Wed Feb 1 11:52:56 2006
New Revision: 110472
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=110472
Log:
Backport:
2005-12-19 Mark Mitchell [EMAIL
--- Comment #10 from reichelt at gcc dot gnu dot org 2006-02-01 11:54
---
Now also fixed on the 3.4 branch.
--
reichelt at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #16 from reichelt at gcc dot gnu dot org 2006-02-01 11:57
---
Subject: Bug 25369
Author: reichelt
Date: Wed Feb 1 11:57:42 2006
New Revision: 110473
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=110473
Log:
Backport:
2005-12-22 Mark Mitchell
--- Comment #17 from reichelt at gcc dot gnu dot org 2006-02-01 11:58
---
Now also fixed on the 3.4 branch.
--
reichelt at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #4 from ro at techfak dot uni-bielefeld dot de 2006-02-01
12:18 ---
Subject: Re: [4.2 Regression] Use of u_int32_t in libgcc-math breaks bootstrap
on Solaris 10/x86
rguenth at gcc dot gnu dot org writes:
Can you provide the patch that fixed the issue for you?
Sure, but
--- Comment #5 from ro at gcc dot gnu dot org 2006-02-01 12:19 ---
Created an attachment (id=10769)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10769action=view)
Trivial patch to avoid use non-standard u_int32_t
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26050
--- Comment #2 from reichelt at gcc dot gnu dot org 2006-02-01 12:20
---
The following testcase shows the same problem, but *without* using
GCC's extension of anonymous structs:
==
class A {
union {
union {
int i;
};
};
};
int j=A().i;
--- Comment #6 from rguenth at gcc dot gnu dot org 2006-02-01 12:23 ---
Oh, I thought you did the config/stdint.m4 fallback, too. I'll take care of
that.
Thanks anyway.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26050
--- Comment #7 from reichelt at gcc dot gnu dot org 2006-02-01 12:25
---
Shorter testcase:
struct A
{
A();
virtual A foo() const;
};
void bar()
{
const A a=A();
a.foo();
}
--
reichelt at gcc dot gnu
--- Comment #7 from ro at techfak dot uni-bielefeld dot de 2006-02-01
12:31 ---
Subject: Re: [4.2 Regression] Use of u_int32_t in libgcc-math breaks bootstrap
on Solaris 10/x86
rguenth at gcc dot gnu dot org writes:
Oh, I thought you did the config/stdint.m4 fallback, too. I'll
--- Comment #2 from reichelt at gcc dot gnu dot org 2006-02-01 12:34
---
This looks like a duplicate of PR 25990 to me since I can trigger
the bug also at -O0 with appropriate --param ggc-min-expand settings.
If I change the parameter the ICE happens at different places.
--
--- Comment #7 from reichelt at gcc dot gnu dot org 2006-02-01 12:38
---
Hi Diego,
did you have a look at this one?
This makes -fopenmp almost unusable for the C++ frontend.
PR 26032 is probably a duplicate of this one.
I stumbled over the bug independently, too.
--
--- Comment #1 from pinskia at gcc dot gnu dot org 2006-02-01 12:49 ---
Confirmed, I really don't know if this is useful or not.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #8 from dnovillo at gcc dot gnu dot org 2006-02-01 12:56
---
Mine.
--
dnovillo at gcc dot gnu dot org changed:
What|Removed |Added
The following testcase
#include iostream
bool flag = true;
class guardian
{
public:
guardian() { flag = false; }
~guardian() { flag = true; }
bool get() { return flag; }
};
int main()
{
guardian guard();
std::cout guard.get() std::endl;
std::cout flag std::endl;
}
--- Comment #1 from pinskia at gcc dot gnu dot org 2006-02-01 13:23 ---
guardian guard();
That is a declaration of a function and not a variable. In fact the error
message tells you that.
Use guardian guard; instead.
This is a dup of bug 9217.
*** This bug has been marked as
--- Comment #6 from pinskia at gcc dot gnu dot org 2006-02-01 13:23 ---
*** Bug 26062 has been marked as a duplicate of this bug. ***
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
My patch for PR 24616 introduced a memory leak in _Jv_Linker::link_symbol_table
because some ffi structures and a String are not reachable by the garbage
collector.
It was decided to accept this problem because the above mentioned patch got
bigger and bigger.
--
Summary: memory leak
--- Comment #29 from thebohemian at gmx dot net 2006-02-01 13:45 ---
Fixed by: http://gcc.gnu.org/ml/java-patches/2006-q1/msg00124.html
--
thebohemian at gmx dot net changed:
What|Removed |Added
--- Comment #2 from thebohemian at gmx dot net 2006-02-01 13:49 ---
The patch for PR 24616 contains variants of the methods I was talking here that
do not throw any exception and instead make the caller responsible for throwing
the correct exception or error:
/configure --enable-shared
--prefix=/home/[...]/20060201 --enable-languages=c,c++,fortran --enable-threads
--enable-__cxa_atexit
Thread model: posix
gcc version 4.2.0 20060201 (experimental)
/home/[...]/20060201/bin/../libexec/gcc/i686-pc-linux-gnu/4.2.0/f951
regression.f90 -quiet -dumpbase
--- Comment #8 from rguenth at gcc dot gnu dot org 2006-02-01 14:27 ---
Patch posted that includes generating a fallback _stdint.h header.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #9 from tbptbp at gmail dot com 2006-02-01 14:28 ---
And you can add PR 25983 on top of it :)
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25990
--- Comment #1 from pinskia at gcc dot gnu dot org 2006-02-01 14:43 ---
Confirmed.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #1 from pinskia at gcc dot gnu dot org 2006-02-01 14:45 ---
This looks related to PR 26041.
Can you try the patch at
http://gcc.gnu.org/ml/gcc-patches/2006-01/msg02261.html
?
--
pinskia at gcc dot gnu dot org changed:
What|Removed
--- Comment #5 from pinskia at gcc dot gnu dot org 2006-02-01 14:46 ---
Patch posted:
http://gcc.gnu.org/ml/gcc-patches/2006-01/msg02261.html
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #15 from hp at gcc dot gnu dot org 2006-02-01 15:32 ---
(In reply to comment #14)
I'm not surprised that decl_overlaps_hard_reg_set_p is only used in
config/cris/cris.c or that it may seem limited in its implementation.
It's just a convenience function invented for it's
--- Comment #10 from dnovillo at gcc dot gnu dot org 2006-02-01 16:05
---
I can't reproduced this with [EMAIL PROTECTED] nor [EMAIL PROTECTED] Could you
please
try again?
--
dnovillo at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #11 from reichelt at gcc dot gnu dot org 2006-02-01 16:19
---
I still get a segfault for the testcase in comment #6.
This is revision 110467 on x86_64-unknown-linux-gnu.
Note, that this is with the C frontend:
gcc -fopenmp --param ggc-min-expand=0 --param
--- Comment #2 from stephen at marenka dot net 2006-02-01 16:28 ---
Also fails with gcc-4.1 (debian 4.1-0exp4).
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25514
--- Comment #2 from sven dot buijssen at math dot uni-dortmund dot de
2006-02-01 16:36 ---
The patch fixes the regression, but introduces a new internal compiler error
for the following testcase:
--- cut here ---
% cat ice.f90 EOF
module ice
implicit none
contains
subroutine
sylvain dot joyeux at m4x dot org wrote:
The following testcase
#include iostream
bool flag = true;
class guardian
{
public:
guardian() { flag = false; }
~guardian() { flag = true; }
bool get() { return flag; }
};
int main()
{
guardian guard();
std::cout guard.get()
--- Comment #2 from sebor at roguewave dot com 2006-02-01 16:47 ---
Subject: Re: New: Class object(); is not interpreted as
a call to default constructor
sylvain dot joyeux at m4x dot org wrote:
The following testcase
#include iostream
bool flag = true;
class guardian
{
--- Comment #2 from thebohemian at gmx dot net 2006-02-01 16:52 ---
Added aph to get some ideas on how to solve this.
Some ideas:
The code that uses the ffi structure is so complicated because it is neccessary
to prepare a call that takes one argument (a class name). I plan to put
--- Comment #6 from hjl at lucon dot org 2006-02-01 16:49 ---
*** Bug 26064 has been marked as a duplicate of this bug. ***
--
hjl at lucon dot org changed:
What|Removed |Added
--- Comment #3 from hjl at lucon dot org 2006-02-01 16:49 ---
It is the same bug.
*** This bug has been marked as a duplicate of 26041 ***
--
hjl at lucon dot org changed:
What|Removed |Added
--- Comment #3 from aph at gcc dot gnu dot org 2006-02-01 16:57 ---
Working around the lack of libffi support on ARM is Just Plain Stupid. We
should either fix it or cease to build on ARM. Without libffi libgcj cannot
work in any reasonable way.
The memory leak thing can be fixed by
--- Comment #4 from hjl at lucon dot org 2006-02-01 17:00 ---
I didn't see it:
[EMAIL PROTECTED] cat xxx.f90
module ice
implicit none
contains
subroutine foo()
contains
subroutine bar(baz)
integer :: baz
if (present(baz)) then
endif
--- Comment #12 from pinskia at gcc dot gnu dot org 2006-02-01 18:06
---
I can still confirm it too.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #3 from pinskia at gcc dot gnu dot org 2006-02-01 18:15 ---
Related closely to PR 13699.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #4 from tromey at gcc dot gnu dot org 2006-02-01 18:32 ---
FWIW, some embedded users want to be able to remove ffi --
not for its own sake but because they also want to remove
the interpreter, reflection info (see David Daney's recent
proposal). It doesn't seem that hard to
--- Comment #12 from ro at techfak dot uni-bielefeld dot de 2006-02-01
19:00 ---
Subject: Re: [4.1/4.2 regression] Many Solaris 10/x86 testsuite failures with
native as: use of .word
I've now tried this patch on the 4.1 branch, and many testsuite failures
due to missing .word support
--- Comment #13 from pinskia at gcc dot gnu dot org 2006-02-01 19:12
---
Testing the fix on x86_64-linux-gnu. Thanks for confirming it fixes the
problems.
Sorry for breaking this in the first place.
--
pinskia at gcc dot gnu dot org changed:
What|Removed
I tried to build gcc 4.0.2 on my FC4 machine. I configured it with the
following options:
../configure --prefix=/tmp/gccbuildtest/gcc
--mandir=/tmp/gccbuildtest/gcc/share/man
--infodir=/tmp/gccbuildtest/gcc/share/info --enable-shared --disable-threads
--disable-checking --with-system-zlib
--- Comment #1 from polite at itd dot nrl dot navy dot mil 2006-02-01
19:29 ---
Created an attachment (id=10770)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10770action=view)
'make bootstrap' output (warnings, errors, etc.)
--
--- Comment #5 from sven dot buijssen at math dot uni-dortmund dot de
2006-02-01 20:01 ---
I'm sorry. I had been experimenting which circumstances trigger the ice and
accidentally deleted the optional argument. To be valid the code needs to be
like
% cat ice.f90 EOF
module ice
--- Comment #7 from pinskia at gcc dot gnu dot org 2006-02-01 20:02 ---
Fixed.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
Status|NEW
--- Comment #6 from hjl at lucon dot org 2006-02-01 21:20 ---
I got
Breakpoint 3, gfc_sym_type (sym=0xb5b800)
at /net/gnu-13/export/gnu/src/gcc-last/gcc/gcc/fortran/trans-types.c:1301
1301 type = gfc_typenode_for_spec (sym-ts);
$8 = {name = 0xb51b46 present, module = 0x0,
--- Comment #16 from amylaar at gcc dot gnu dot org 2006-02-01 21:56
---
(In reply to comment #15)
I'm not surprised that decl_overlaps_hard_reg_set_p is only used in
config/cris/cris.c or that it may seem limited in its implementation.
It's just a convenience function invented for
--- Comment #11 from paolo at gcc dot gnu dot org 2006-02-01 21:57 ---
Subject: Bug 21554
Author: paolo
Date: Wed Feb 1 21:57:48 2006
New Revision: 110482
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=110482
Log:
2006-02-01 Paolo Carlini [EMAIL PROTECTED]
PR
--- Comment #12 from pcarlini at suse dot de 2006-02-01 22:01 ---
Fixed for 4.1.0 and mainline. Will not fix in 4_0-branch.
--
pcarlini at suse dot de changed:
What|Removed |Added
--- Comment #13 from dnovillo at gcc dot gnu dot org 2006-02-01 22:36
---
(In reply to comment #11)
I still get a segfault for the testcase in comment #6.
This is revision 110467 on x86_64-unknown-linux-gnu.
Ah, yes. I had failed to notice the failure is on x86_64. Sorry about
--- Comment #5 from sje at cup dot hp dot com 2006-02-02 00:08 ---
The defect number was not put in the ChangeLog file but this was fixed in GCC
subversion revision 110476. The change was to trunk/gcc/gcc.c.
--
sje at cup dot hp dot com changed:
What|Removed
--- Comment #2 from sje at gcc dot gnu dot org 2006-02-02 00:13 ---
Subject: Bug 24901
Author: sje
Date: Thu Feb 2 00:13:39 2006
New Revision: 110487
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=110487
Log:
PR middle-end/24901
* fold-const.c
Acccording to C++ standard 7.5/7 A linkage-specification directly containing a
single declaration shall not specify a storage class. They also give the
following example as invalid:
extern C static void foo();
But g++ version 4.0.2 accepts this.
--
Summary: extern C static
--- Comment #1 from pinskia at gcc dot gnu dot org 2006-02-02 00:23 ---
Confirmed, a regression from 3.0.4:
t.cc:1: multiple storage classes in declaration of `foo'
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #6 from dave at hiauly1 dot hia dot nrc dot ca 2006-02-02
01:12 ---
Subject: Re: [4.2 Regression] gcc.c:3866: warning: comparison between signed
and unsigned
The defect number was not put in the ChangeLog file but this was fixed in GCC
subversion revision 110476. The
--- Comment #7 from hjl at lucon dot org 2006-02-02 01:18 ---
An updated patch is posted at
http://gcc.gnu.org/ml/gcc-patches/2006-02/msg00102.html
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26041
--- Comment #7 from hjl at lucon dot org 2006-02-02 01:19 ---
Please try this one
http://gcc.gnu.org/ml/gcc-patches/2006-02/msg00102.html
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26064
--- Comment #4 from amodra at bigpond dot net dot au 2006-02-02 04:30
---
I don't think this bug should have been closed so quickly. Fixing __gcc_qadd
to properly handle a -0 result is trivial, and only costs one extra compare and
branch. The multiply and divide support in
bool tell_endian() {
unsigned x = 1;
return *(char*)x;
}
g++ 3.4.2 produces:
_Z11tell_endianv:
0: 55 push %ebp
1: 89 e5 mov%esp,%ebp
3: b8 01 00 00 00 mov$0x1,%eax
8: c9 leave
--- Comment #1 from atgraham at gmail dot com 2006-02-02 05:02 ---
I've also confirmed the following:
The optimization also works on gcc 2.95.2
It does not work on gcc 4.0.2
--
atgraham at gmail dot com changed:
What|Removed |Added
--- Comment #2 from pinskia at gcc dot gnu dot org 2006-02-02 05:19 ---
If you used GCC's extension of using an union as not violating aliasing rules,
it works there but does not with using the portable ways:
bool tell_endian() {
unsigned x = 1;
return *(char*)x;
}
bool
--- Comment #5 from amodra at bigpond dot net dot au 2006-02-02 06:18
---
* config/rs6000/darwin-ldouble.c (__gcc_qadd): Preserve -0.0 result.
Index: gcc/config/rs6000/darwin-ldouble.c
===
---
83 matches
Mail list logo
