--- Comment #5 from bonzini at gnu dot org 2010-05-30 06:42 ---
Richi, I think we're saying the same thing from two different directions.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44321
--- Comment #20 from raj dot khem at gmail dot com 2010-05-30 06:54 ---
Created an attachment (id=20777)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20777action=view)
includes fix for all arms
I see the problem after reading Richard's comment #10, here is some more
additions to
I have enountered a strange difference in behavior with gcc optimizations. I am
testing this with gcc-4.5.0 on a x86_64 target. Given the following code which
defines to template functions. The goal is to have a find_first and
find_last for a bitset. It will return the size of the bitset if no
--- Comment #1 from eteran at alum dot rit dot edu 2010-05-30 07:07 ---
I just noticed that the documentation for __builtin_clz and __builtin_ffsl say
the result is undefined if x is 0. Therefore, this bug is invalid, sorry for
the false bug report.
--
eteran at alum dot rit dot edu
--- Comment #2 from jay dot krell at cornell dot edu 2010-05-30 08:06
---
I understand it is ok.
Question is: Do you want to be warning free?
At what cost?
I think you do, at some finite cost, and I think the change is worthwhile here.
This isn't adding symbols at random, it is adding
Fixincludes has two ways of doing things: oneprocess and twoprocess.
It appears that oneprocess is more often used, presumably slightly faster, but
that twoprocess is more portable and would work everywhere oneprocess is used.
Suggest just keeping one portable form and removing the other.
Less
--- Comment #1 from jay dot krell at cornell dot edu 2010-05-30 08:46
---
Created an attachment (id=20778)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20778action=view)
Something like the attached, which is strictly removal of code (plus inlining
of twoprocess in Makefile.in)
--- Comment #3 from jay dot krell at cornell dot edu 2010-05-30 08:55
---
Hm. There are more. Perhaps better to not compile/lib/link files that can be
easily predicted to have no symbols? Maybe just ignore it. Maybe get Apple to
change ranlib. Maybe they already have.
ranlib: file:
--- Comment #1 from eblot dot ml at gmail dot com 2010-05-30 10:36 ---
Update: This is a C++ only bug, the same code in C language does not exhibit
the bug:
# --- Building as C language ---
$ arm-eabi-gcc -c -O2 -mcpu=arm926ej-s -o fopen.o -x c fopen.c -Wextra
-save-temps cat fopen.s
--- Comment #13 from iains at gcc dot gnu dot org 2010-05-30 12:25 ---
In a bid to remove cases where the FE is bypassing varpool/cgraph and going
directly to asm
I've replaced the (2) instances of mark_decl_referenced () with DECL_PRESERVE_P
() =1 (Honza's suggestion)
I've
--- Comment #6 from redi at gcc dot gnu dot org 2010-05-30 12:39 ---
that's fine I'll be able to update the docs from that info - thanks
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44268
--- Comment #4 from jay dot krell at cornell dot edu 2010-05-30 13:24
---
It's MacOSX for local console, remote to Debian, Solaris, OpenBSD, etc., I
can't keep track of when it works and when it doesn't, but it often doesn't.
Local console works fine.
As the bug says in many
Compiler output:
$ gcc -mno-sse2 testcase.c
testcase.c: In function 'foo':
testcase.c:6:6: internal compiler error: in copy_to_mode_reg, at explow.c:635
Please submit a full bug report,
with preprocessed source if appropriate.
See http://gcc.gnu.org/bugs.html for instructions.
Tested revisions:
--- Comment #14 from iains at gcc dot gnu dot org 2010-05-30 13:42 ---
Created an attachment (id=20779)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20779action=view)
patch to add LTO tests to OobjC
This adds an LTO directory to objc.dg and a single test (trivial_0.m) which is
--- Comment #2 from mikpe at it dot uu dot se 2010-05-30 14:50 ---
I can confirm this wrong-code when gcc 4.4/4.5 targets arm-unknown-eabi.
However, a 4.4/4.5 running natively on arm-unknown-linux-gnueabi does not
exhibit this behaviour. There there's no 'comparison always true'
--- Comment #15 from iains at gcc dot gnu dot org 2010-05-30 14:51 ---
we emit the meta-data [mostly tabular] for the runtime on the basis of
DECL_PRESERVE_P (or mark_decl_referenced () in the trunk version).
The initlist for these decls contains references to the classes and methods.
--- Comment #16 from hjl dot tools at gmail dot com 2010-05-30 14:58
---
(In reply to comment #13)
HJ - would you prefer this to be closed and for me to open a new bug to track
the real issue?
I opened this bug to report
objc.dg/torture/tls/thr-init-3.m
Since it is XFAILED, I
--- Comment #13 from itosre at trash-mail dot com 2010-05-30 15:09 ---
(In reply to comment #12)
I'm assuming this is current Linus git (post 2.6.34).
I'm guessing from
Tested on several kernels: 2.6.32.8, 2.6.33.4, 2.6.34-rc7
That this isn't the case. Strangely enough, it WFT
--- Comment #1 from rguenth at gcc dot gnu dot org 2010-05-30 15:21 ---
Well, doctor it hurts when I do that ...
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44332
I would really like this to be possible, would help me a lot in getting rid of
cstddef uses in the library, and it is indeed possible with ICC and Comeau,
and SunStudio, but GCC doesn't like it, for some reason.
typedef unsigned my_size_t;
namespace my
{
typedef unsigned my_size_t;
}
using
--- Comment #1 from paolo dot carlini at oracle dot com 2010-05-30 15:54
---
And replacing both:
using namespace my;
with
using my::my_size_t;
instead works. This is crazy (IMHO ;) and would not be a workaround for me)
--
paolo dot carlini at oracle dot com changed:
--- Comment #2 from steven at gcc dot gnu dot org 2010-05-30 16:55 ---
Should not ICE, though.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44332
--- Comment #2 from rguenth at gcc dot gnu dot org 2010-05-30 16:57 ---
Fixed.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Keywords||missed-optimization, ra
Priority|P3
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Priority|P3 |P2
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44290
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Priority|P3 |P2
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44301
After revision 159852
Author: pault
Date: Wed May 26 05:11:04 2010 UTC (4 days, 12 hours ago)
Changed paths: 4
Log Message:
2010-05-26 Paul Thomas pa...@gcc.gnu.org
PR fortran/40011
* resolve.c (resolve_global_procedure): Resolve the gsymbol's
namespace before
--- Comment #3 from eblot dot ml at gmail dot com 2010-05-30 17:36 ---
I can confirm this wrong-code when gcc 4.4/4.5 targets arm-unknown-eabi.
However, a 4.4/4.5 running natively on arm-unknown-linux-gnueabi does not
exhibit this behaviour.
IIRC, Linux ABI does not follow the ARM
--- Comment #1 from dominiq at lps dot ens dot fr 2010-05-30 18:06 ---
I'll attach the assembly generated with -O3 -ffast-math -funroll-loops
-fomit-frame-pointer -flto for revisions 159851 and 159852. It is the same
with/without -fwhole-program (probably obvious), however when
--- Comment #2 from rguenth at gcc dot gnu dot org 2010-05-30 18:09 ---
Insufficient analysis. This more sounds like a dup of profile-estimate
messed up by inlining.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #3 from dominiq at lps dot ens dot fr 2010-05-30 18:10 ---
Created an attachment (id=20780)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20780action=view)
Assembly generated with -O3 -ffast-math -funroll-loops -fomit-frame-pointer
-flto and revision 159851
--
--- Comment #4 from dominiq at lps dot ens dot fr 2010-05-30 18:12 ---
Created an attachment (id=20781)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20781action=view)
Assembly generated with -O3 -ffast-math -funroll-loops -fomit-frame-pointer
-flto and revision 159852
--
--- Comment #5 from dominiq at lps dot ens dot fr 2010-05-30 18:30 ---
Output of gprof on darwin:
Revision 159851:
called/total parents
index %timeself descendents called+selfname index
--- Comment #6 from rguenth at gcc dot gnu dot org 2010-05-30 18:48 ---
0.0 0.00 0.00 5572994 0.00 0.00 _xerbla_ [154]
eh? that's the blas error handler. something is fishy with your setup.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44334
--- Comment #7 from dominiq at lps dot ens dot fr 2010-05-30 18:55 ---
Insufficient analysis. This more sounds like a dup of profile-estimate
messed up by inlining.
Do you mean a dup of pr40106? Or is there others I am not aware of?
eh? that's the blas error handler. something
--- Comment #3 from rguenth at gcc dot gnu dot org 2010-05-30 19:53 ---
Fixed.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--- Comment #2 from rguenth at gcc dot gnu dot org 2010-05-30 19:53 ---
Subject: Bug 42975
Author: rguenth
Date: Sun May 30 19:52:53 2010
New Revision: 160059
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=160059
Log:
2010-05-30 Richard Guenther rguent...@suse.de
PR
Attempting to bootstrap gcc-4.6-20100529 on armv5tel-unknown-linux-gnueabi with
java enabled fails:
...
In file included from /home/mikpe/gcc-4.6-20100529/gcc/java/except.c:33:0:
/home/mikpe/gcc-4.6-20100529/gcc/except.h:322:6: error: #error
EH_RETURN_DATA_REGNO required
--- Comment #1 from rguenth at gcc dot gnu dot org 2010-05-30 20:45 ---
Should be fixed with
2010-05-30 Steven Bosscher ste...@gcc.gnu.org
* except.c: Include tm.h.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Tested revisions:
r160052 - crash
r159523 - segfault (release checking)
r159387 - OK
r158969 - OK
r158095 - OK
Compiler output:
$ gcc -O1 -fipa-struct-reorg -fwhole-program testcase.c
testcase.c: In function 'main':
testcase.c:4:3: warning: return makes integer from pointer without a cast
Compiler output:
$ gcc -O -fno-tree-dce -fno-tree-dse testcase.c
testcase.c: In function #8216;foo#8217;:
testcase.c:4:8: internal compiler error: in expand_assignment, at expr.c:4280
Please submit a full bug report,
with preprocessed source if appropriate.
See http://gcc.gnu.org/bugs.html for
Tested revisions:
r160052 - crash
Compiler output:
$ gcc -mno-fused-madd /mnt/svn/gcc-trunk/gcc/testsuite/gcc.target/i386/sse-23.c
In file included from
/mnt/svn/gcc-trunk/binary-160052-lto-fortran/lib/gcc/x86_64-unknown-linux-gnu/4.6.0/include/x86intrin.h:69:0,
from
When I try to compile a std::set std::weak_ptr int , it fails. However,
when I try to compile the exact same code but using boost::weak_ptr instead, it
works.
--
Summary: Usage of std::weak_ptr in ordered stl container (C++0x)
Product: gcc
Version: 4.4.3
--- Comment #1 from pcayouette at spoluck dot ca 2010-05-31 00:18 ---
Created an attachment (id=20782)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20782action=view)
Comparison between std case and boost case.
Uncomment the line
//#define USING_STD_SHARED_PTR
to expose the
GNAT bug box generated:
gcc-4.4 -c -g -gnatq -gnatQ test_tab.adb
+===GNAT BUG DETECTED==+
| 4.4.3 (i486-pc-linux-gnu) GCC error: |
| in gnat_to_gnu_entity, at ada/gcc-interface/decl.c:348 |
|
--- Comment #2 from pcayouette at spoluck dot ca 2010-05-31 00:23 ---
(From update of attachment 20782)
Must have boost to compile the file.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44339
--- Comment #3 from pcayouette at spoluck dot ca 2010-05-31 00:25 ---
Compile command:
g++ -O0 -g3 -Wall -std=c++0x -c -fmessage-length=0 -MMD -MP
-MFsrc/TestCpp0x.d -MTsrc/TestCpp0x.d -osrc/TestCpp0x.o
../src/TestCpp0x.cpp
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44339
--- Comment #4 from pcayouette at spoluck dot ca 2010-05-31 00:26 ---
Error when #define line uncommented in example attached:
In file included from /usr/include/c++/4.4/memory:66,
from ../src/TestCpp0x.cpp:1:
/usr/include/c++/4.4/bits/shared_ptr.h: In member function
--- Comment #14 from bdubbs at linuxfromscratch dot org 2010-05-31 00:41
---
(In reply to comment #13)
(In reply to comment #12)
I'm assuming this is current Linus git (post 2.6.34).
I'm guessing from
Tested on several kernels: 2.6.32.8, 2.6.33.4, 2.6.34-rc7
That this
--- Comment #5 from paolo dot carlini at oracle dot com 2010-05-31 00:52
---
I'm adding Jon in CC, but I think it's safe to close this PR as invalid: if you
look at n3090 (the latest working draft), it's clear that std::weak_ptr is
*not* LessThanComparable. However, if you change the
Java enabled cross build for sh4-unknown-linux-gnu on i686-pc-linux-gnu
host fails during linking libjavamath.so:
libtool: link: /exp/ldroot/dodes/xsh-gcc-orig/./gcc/xgcc
-B/exp/ldroot/dodes/xsh-gcc-orig/./gcc/ -B/usr/local/sh4-unknown-linux-gnu/bin/
-B/usr/local/sh4-unknown-linux-gnu/lib/
--- Comment #15 from hpa at zytor dot com 2010-05-31 01:04 ---
OK, thanks for confirming that it is not related to asm goto.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44129
--- Comment #6 from paolo dot carlini at oracle dot com 2010-05-31 01:24
---
... and I think the correct way to achieve what you want now would be:
std::setstd::weak_ptrint, std::owner_lessstd::weak_ptrint wStdSet;
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44339
--- Comment #3 from catphive at catphive dot net 2010-05-31 03:43 ---
I've also run into this bug on Ubuntu with gcc version: gcc (Ubuntu
4.4.1-4ubuntu9) 4.4.1
The program that triggers it for me is this:
#include vector
using std::vector;
int main(int argc, char* argv[]) {
55 matches
Mail list logo