I was recompiling my MMX code on GCC 4 and was surprised to see that none of
the moves from general regs to MMX (and other way around) were happening
directly but through intermediate memory location and I could not find a
combination of optimization options which would disable this behaviour.
--- Additional Comments From vahur dot sinijarv at cydonia dot ee
2005-08-31 17:24 ---
Created an attachment (id=9637)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=9637action=view)
Preprocessed source
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23660
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-08-31
17:28 ---
Subject: Bug 23570
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-08-31 17:27:54
Modified files:
gcc: ChangeLog
gcc/config/i386:
--- Additional Comments From vahur dot sinijarv at cydonia dot ee
2005-08-31 17:34 ---
Created an attachment (id=9638)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=9638action=view)
4.1.0 vs 3.4.4
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23660
--
What|Removed |Added
Target Milestone|--- |4.0.2
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23632
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
19:18 ---
The issue is more complicated than you think.
This is a dup of bug 22076 which has about the same code and issue.
*** This bug has been marked as a duplicate of 22076 ***
--
What|Removed
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
19:18 ---
*** Bug 23660 has been marked as a duplicate of this bug. ***
--
What|Removed |Added
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
19:22 ---
This looks more like a target bug as this is being called from ia64's reorg.
--
What|Removed |Added
--
What|Removed |Added
CC||pinskia at gcc dot gnu dot
||org
Status|UNCONFIRMED
--- Additional Comments From redi at gcc dot gnu dot org 2005-08-31 19:27
---
As DR39 has been ruled a defect, this should be re-opened.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13590
--
What|Removed |Added
Status|SUSPENDED |NEW
Last reconfirmed|2005-07-12 22:07:51 |2005-08-31 19:34:41
date|
--- Additional Comments From tobi at gcc dot gnu dot org 2005-08-31 19:35
---
Probably Paul Thomas' character fixes play a role in making those two different
bugs.
--
What|Removed |Added
test case:
$ cat print_fmt.f95
PROGRAM print_fmt_test
IMPLICIT NONE
character(len=5) :: fmt = ( a)
integer:: i
print fmt, (=, i=1, 2)
END PROGRAM print_fmt_test
$ gfortran -c print_fmt.f95
In file print_fmt.f95:5
print fmt, (=, i=1, 2)
1
Error: Unclassifiable statement at
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
19:42 ---
Note this was caused as an extension to the C++ standard:
http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_closed.html#150
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9737
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
19:49 ---
Confirmed.
--
What|Removed |Added
Status|UNCONFIRMED |NEW
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-08-31
19:51 ---
Subject: Bug 21255
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-08-31 19:50:29
Modified files:
gcc/config/sh : sh.c
gcc/testsuite :
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-08-31
19:52 ---
Subject: Bug 21255
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-08-31 19:52:35
Modified files:
gcc: ChangeLog
Log message:
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
19:56 ---
http://www.open-std.org/JTC1/SC22/WG14/www/docs/dr_289.htm
This was decided that this was a defect at least as far as I can see.
--
What|Removed |Added
--- Additional Comments From ngmlinux at gmail dot com 2005-08-31 20:18
---
I now have a stacktrace from someone on this issue from gdb:
They used the following configuration parameters to build arm-linux-gcc cross
compilers:
The following was tested with gcc-3.4.3 and gcc-4.0.1
--- Additional Comments From bangerth at dealii dot org 2005-08-31 20:34
---
In the discussion on the duplicate PR 23660, rth explained part of this here:
http://gcc.gnu.org/ml/gcc/2005-08/msg00934.html
W.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22076
--- Additional Comments From rsandifo at gcc dot gnu dot org 2005-08-31
20:42 ---
Working on a patch.
--
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu
A similar bug was reported against 3.1 (# 5487). A new bug is being opened as
that bug had been misreported as fixed against 3.1 and has since been closed.
I've been able to reproduce the same exact issue in arm-linux-gcc-4.0.1 as
reported in Bug # 5487 against gcc-3.1 which I build from source
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
20:59 ---
The fix would be change choose_personality_routine but that would mean you
cannot mix C++ and
Objc exceptions in one TU. So far there are no tests for this. I might
implement the change to the
ObjC
--- Additional Comments From tobi at gcc dot gnu dot org 2005-08-31 20:59
---
This is the same bug as PR23460
--
What|Removed |Added
BugsThisDependsOn|
--- Additional Comments From ngmlinux at gmail dot com 2005-08-31 21:00
---
I now have a stacktrace from someone on this issue from gdb:
They used the following configuration parameters to build arm-linux-gcc cross
compilers:
The following was tested with gcc-3.4.3 and gcc-4.0.1
--- Additional Comments From tobi at gcc dot gnu dot org 2005-08-31 21:00
---
well, essentially the same: the compiler doesn't deal the temporary file
correctly. Probable workaround: use the -pipe commandline option.
--
What|Removed |Added
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
21:03 ---
From your back trace this is in the GC and not really in an issue for the
front-end or class library.
Also from the looks of it, you using gcj 4.0.0 but using 3.4.3's libgcj, this
is not supported at
--- Additional Comments From tobi at gcc dot gnu dot org 2005-08-31 21:04
---
(In reply to comment #1)
Gfortran works as expected.
Depends on your definition of 'as expected', see PR20811, as pointed out by Paul
Brook here: http://gcc.gnu.org/ml/fortran/2005-08/msg00310.html
--
--- Additional Comments From redi at gcc dot gnu dot org 2005-08-31 21:04
---
Is this a dup of Bug 14258 ?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15260
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
21:07 ---
The last known good test results from arm-linux:
http://gcc.gnu.org/ml/gcc-testresults/2003-06/msg00168.html
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23662
--- Additional Comments From redi at gcc dot gnu dot org 2005-08-31 21:10
---
The resolution of DR460 makes this code illegal and GCC's behaviour correct.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16468
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
21:14 ---
Actually there is a later one:
http://gcc.gnu.org/ml/gcc-testresults/2004-04/msg00322.html
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23662
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
21:21 ---
This is a bug in ifcvt:
#0 internal_error (gmsgid=0xb1eb1f in %s, at %s:%d) at
/home/pinskia/src/onetest/gcc/gcc/
diagnostic.c:534
#1 0x0060c3f2 in fancy_abort (file=0xbe7b20
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
21:24 ---
It also ICEs on powerpc-darwin.
--
What|Removed |Added
GCC target
The following code does not compile
function ibug()
ibug=0
return
entry ierror()
ierror=0
ierror=max(ierror,1)
return
end
error is
[/Users/fca] /opt/gcc-4_0/bin/gfortran -c junk.f
In file junk.f:6
ierror=max(ierror,1)
--
What|Removed |Added
CC||alfredo dot ferrari at cern
||dot ch
--- Additional Comments From steven at gcc dot gnu dot org 2005-08-31
21:38 ---
Can the alias gurus please give their view on this one?
--
What|Removed |Added
--- Additional Comments From dnovillo at gcc dot gnu dot org 2005-08-31
21:41 ---
Working on it.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23659
--- Additional Comments From fitzsim at redhat dot com 2005-08-31 21:57
---
Fixed in GNU Classpath. Closing.
--
What|Removed |Added
Status|ASSIGNED
--
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |mark at codesourcery dot com
|dot org |
Status|NEW
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
22:10 ---
Confirmed.
--
What|Removed |Added
Status|UNCONFIRMED |NEW
--- Additional Comments From qrczak at knm dot org dot pl 2005-08-31 22:11
---
I think this is fixed by the patch for PR 23509:
http://gcc.gnu.org/ml/gcc-patches/2005-08/msg01647.html
Indeed, applying this patch and recompiling gcc fixed the bug. Thanks.
--
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-08-31
22:14 ---
(In reply to comment #2)
Indeed, applying this patch and recompiling gcc fixed the bug. Thanks.
Thanks for trying the patch. I only thought about this patch because of the
symptom was about the
same as
--- Additional Comments From qrczak at knm dot org dot pl 2005-08-31 22:23
---
*** This bug has been marked as a duplicate of 23509 ***
--
What|Removed |Added
--- Additional Comments From qrczak at knm dot org dot pl 2005-08-31 22:23
---
*** Bug 23657 has been marked as a duplicate of this bug. ***
--
What|Removed |Added
--- Additional Comments From tobi at gcc dot gnu dot org 2005-08-31 23:10
---
What happens is this:
In match_io we have the following code
if (gfc_match_char ('(') == MATCH_NO)
{
/* Treat the non-standard case of PRINT namelist. */
if (k == M_PRINT (gfc_match_name
--- Additional Comments From tobi at gcc dot gnu dot org 2005-08-31 23:11
---
(I understood this between my mail to the list, and writing the previous
comment, in case you're wondering about the discrepancy :-)
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23661
--- Additional Comments From xiaoyi_wu at yahoo dot com 2005-08-31 23:34
---
Hi, this is the original reporter of this bug.
But a day later, I can not reproduce it at all on the same machine. I didn't
even log out of my account. Strange!
Anyway, I have changed the resolution to
These two functions should produce the same asm:
int f(int a, int b)
{
return (a0xF)+(b0xF0);
}
int f1(int a, int b)
{
return (a0xF)|(b0xF0);
}
For PPC, we get:
_f:
rlwinm r4,r4,0,24,27
rlwinm r3,r3,0,28,31
add r3,r3,r4
blr
_f1:
rlwinm r4,r4,0,24,27
--
What|Removed |Added
BugsThisDependsOn||23664
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19987
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-09-01
00:42 ---
I should give credit to LLVM's instruction combine pass which have this.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23664
Note that the following code invokes undefined behavior. I haven't been
able to reproduce this bug on platforms other than Cygwin.
% cat gcc-bug.c
double func();
int main(void)
{
double d = func(42);
return 0;
}
double func(double arg)
{
return arg * 2.0;
}
% uname -a
CYGWIN_NT-5.1
--- Additional Comments From ngmlinux at gmail dot com 2005-09-01 00:57
---
(In reply to comment #2)
From your back trace this is in the GC and not really in an issue for the
front-end or class library.
Also from the looks of it, you using gcj 4.0.0 but using 3.4.3's libgcj, this
is
--- Additional Comments From ngmlinux at gmail dot com 2005-09-01 01:02
---
Is it neccissary for the host-to-host version of gcc to be the exact same
version as the host-to-target gcc version being built? I doubt this is the case
as this would present major imcompatibility issues for
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-09-01
01:03 ---
Confirmed, only a 3.4 regression.
--
What|Removed |Added
Status|UNCONFIRMED
--
What|Removed |Added
CC||pinskia at gcc dot gnu dot
||org
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-09-01
01:08 ---
First don't add me to the CC, I read gcc-bugs.
(In reply to comment #6)
Is it neccissary for the host-to-host version of gcc to be the exact same
version as the host-to-target gcc version being built? I
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-09-01
01:32 ---
Fixed for 4.1.0.
--
What|Removed |Added
Status|ASSIGNED
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-09-01
01:32 ---
Subject: Bug 23306
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-09-01 01:32:11
Modified files:
gcc/objc : ChangeLog objc-act.c
--- Additional Comments From ngmlinux at gmail dot com 2005-09-01 01:39
---
(In reply to comment #7)
First don't add me to the CC, I read gcc-bugs.
I appologize.
As I said this is a bug in boehm-gc, you might to see if you can use the
latest version of boehm-gc on
your target.
The asm for the following two functions should be the same.
int f(int a)
{
return 10 - ~a;
}
int f1(int a)
{
return a + 11;
}
Found while reading LLVM's instruction combiner.
--
Summary: Fold does not reduce C - ~a into a + (C+1)
Product: gcc
Version: 4.1.0
--
What|Removed |Added
BugsThisDependsOn||23666
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19987
Between 20050829 and 20050830
FAIL: tr1/6_containers/unordered/hashtable/23465.cc execution test
appeared on both mainline and 4.0 branch, on hppa64-hp-hpux11.11 and
arm-none-eabi; this failure is a test which formerly reliably passed now
consistently timing out; gcc-testresults shows this
Between 20050830 and 20050831 the following FAILs of tests which previously
passed appeared on mainline on i686-pc-linux-gnu.
FAIL: gcc.dg/i386-sse-11.c (test for excess errors)
FAIL: gcc.target/i386/pr13366.c (test for excess errors)
/scratch/gcc/nightly-2005-08-31-mainline/src/gcc-mainline/gcc
--- Additional Comments From jsm28 at gcc dot gnu dot org 2005-09-01 01:51
---
The testcase regex was fixed some time ago.
--
What|Removed |Added
Status|NEW
--- Additional Comments From jsm28 at gcc dot gnu dot org 2005-09-01 01:54
---
Appears fixed at around same time as 23463, at least on HP-UX.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23464
--
What|Removed |Added
BugsThisDependsOn||23669
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19987
Another idea from LLVM.
The following two functions should be the same asm:
int f(int a)
{
return (-a)/10;
}
int f1(int a)
{
return a/-10;
}
We currently get on PPC:
_f:
lis r0,0x
neg r3,r3
ori r0,r0,26215
mulhw r0,r3,r0
srawi r3,r3,31
--
What|Removed |Added
BugsThisDependsOn||23670
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19986
Example:
int f(int a, int b)
{
return (a|9)9;
}
int f1(int a, int b)
{
return (a|9)9;
}
This is folded at the rtl level but not at the tree level.
--
Summary: Fold does not fold (a|b)b to b likewise for (ab) | b
to b
Product: gcc
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-09-01
02:15 ---
Oh, this was from LLVM again.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23670
/home/dave/gcc-4.1/objdir/./gcc/xgcc -shared-libgcc -B/home/dave/gcc-4.1/objdir/
./gcc -nostdinc++ -L/home/dave/gcc-4.1/objdir/hppa-linux/libstdc++-v3/src -L/hom
e/dave/gcc-4.1/objdir/hppa-linux/libstdc++-v3/src/.libs -B/home/dave/opt/gnu/gcc
/gcc-4.1.0/hppa-linux/bin/
--
What|Removed |Added
BugsThisDependsOn||23672
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19986
These two functions should be the same at the tree level but are not:
int f(int a, int b)
{
return (a^b)^a;
}
int f1(int a, int b)
{
return b;
}
---
Another one from LLVM.
--
Summary: Fold does not fold (a^b)^a to b
Product: gcc
Version: 4.1.0
--- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca
2005-09-01 02:28 ---
Subject: Re: [4.1 Regression] compat tests fail
--- Additional Comments From jsm28 at gcc dot gnu dot org 2005-09-01
01:54 ---
Appears fixed at around same time as 23463, at least on
These two functions should be the same on the tree level but are not:
int f(int a, int b)
{
return (a^b) != 0;
}
int f1(int a, int b)
{
return a != b;
}
They do result in the same asm though.
Another one from LLVM.
--
Summary: fold does not fold (a^b) != 0 to a != b
--
What|Removed |Added
CC||rth at gcc dot gnu dot org
GCC build triplet|hppa-unknown-linux-gnu |
GCC host
--
What|Removed |Added
Keywords||ice-on-valid-code, ssemmx
Target Milestone|--- |4.1.0
--
What|Removed |Added
Severity|normal |enhancement
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23669
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-09-01
02:33 ---
Fixed then since that (PR 23463) was my bug.
--
What|Removed |Added
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-09-01
02:48 ---
Subject: Bug 23649
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-09-01 02:48:02
Modified files:
gcc: ChangeLog
--- Additional Comments From amodra at bigpond dot net dot au 2005-09-01
02:49 ---
Fixed
--
What|Removed |Added
URL|
--- Additional Comments From amodra at bigpond dot net dot au 2005-09-01
03:47 ---
Fails on powerpc-linux -m32 -O2 too. Passes with -m64 -O2.
--
What|Removed |Added
GCC
--- Additional Comments From rth at gcc dot gnu dot org 2005-09-01 05:03
---
I know you're going to attach a preprocessed file for non-pa owners...
--
What|Removed |Added
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-09-01
05:29 ---
Subject: Bug 23478
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-09-01 05:29:04
Modified files:
gcc: ChangeLog global.c local-alloc.c
101 - 185 of 185 matches
Mail list logo
