--- Comment #3 from pinskia at gcc dot gnu dot org 2007-10-23 06:09 ---
This is most likely just a latent bug that showed up wth Ollie's patch, looking
into it further.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #17 from pault at gcc dot gnu dot org 2007-10-23 06:19 ---
(In reply to comment #11)
I'm adding Paul to the CC list, as perhaps he immediately knows what's
happening (Paul, see comment #5). Otherwise I will investigate tomorrow
evening or Saturday.
I'm pretty
--- Comment #4 from pinskia at gcc dot gnu dot org 2007-10-23 06:38 ---
So when out of SSA (TER) combines:
iftmp.1D.2042_25 = (derivedD.2000:: *) somemethodD.1996;
# SMT.36D.2106_36 = VDEF SMT.36D.2106_34(D) { SMT.36D.2106 }
iftmp.1_25 (0B);
It does not change the case where it
--- Comment #5 from pinskia at gcc dot gnu dot org 2007-10-23 06:53 ---
Here is a semi undefined case which shows the same issue:
void f(void*) throw();
void somefunction()
{
try {
void (*g)(void*) = (void (*)(void*))f;
void (*g2)(int*) = (void (*)(int*))g;
g2(0);
} catch
--- Comment #5 from victork at gcc dot gnu dot org 2007-10-23 06:53 ---
Created an attachment (id=14394)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14394action=view)
vectorizer dump
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33319
--- Comment #6 from victork at gcc dot gnu dot org 2007-10-23 06:57 ---
I think this bug is duplicate of pr31081.
Probably versioning which is done by vectorizer exposes the problem in inliner.
I've attached the vectorizer dump for analysys (pr27549.C.103t.vect)
% gdb --args
--- Comment #7 from pinskia at gcc dot gnu dot org 2007-10-23 07:02 ---
Actually I doubt it is related to the other bug as the other one only happens
on the trunk as there is no SSA inliner in 4.2 so changing back to NEW.
--
pinskia at gcc dot gnu dot org changed:
What
--- Comment #7 from burnus at gcc dot gnu dot org 2007-10-23 08:20 ---
(In reply to comment #6)
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=129539
Log:
2007-10-21 Paul Thomas [EMAIL PROTECTED]
PR fortran/33749
Paul, can this bug be closed as FIXED or is
(sid)30431:[EMAIL PROTECTED]: ~] /usr/lib/gcc-snapshot/bin/gcc -c -O1
-ftree-vectorize
hdf5-tselect.c
hdf5-tselect.c: In function 'test_select_fill_hyper_simple':
hdf5-tselect.c:12: internal compiler error: in vect_get_vec_def_for_stmt_copy,
at tree-vect-transform.c:1937
Please submit a full bug
--- Comment #1 from tbm at cyrius dot com 2007-10-23 08:28 ---
Created an attachment (id=14395)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14395action=view)
preprocessed source
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33866
--- Comment #2 from tbm at cyrius dot com 2007-10-23 08:29 ---
/* Testcase by Martin Michlmayr [EMAIL PROTECTED] */
typedef struct
{
long *coords;
}
fill_iter_info;
extern H5Diterate (fill_iter_info *);
void test_select_fill_hyper_simple (long *offset)
{
long start[2];
int
--- Comment #3 from tbm at cyrius dot com 2007-10-23 08:29 ---
Breakpoint 1, fancy_abort (file=0xc4e238 gcc/tree-vect-transform.c,
line=1937, function=0xc4f890 vect_get_vec_def_for_stmt_copy)
at gcc/diagnostic.c:659
659 {
(gdb) where
#0 fancy_abort (file=0xc4e238
(sid)30567:[EMAIL PROTECTED]: ~] /usr/lib/gcc-snapshot/bin/gcc -c -O
crystalspace-polyclip.cc
crystalspace-polyclip.cc: In member function 'virtual void
csBoxClipper::Clip()':
crystalspace-polyclip.cc:51: error: verify_flow_info: Wrong frequency of block
7 -1247
crystalspace-polyclip.cc:51: error:
--- Comment #1 from tbm at cyrius dot com 2007-10-23 08:54 ---
Created an attachment (id=14396)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14396action=view)
preprocessed source
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33867
--- Comment #4 from rguenth at gcc dot gnu dot org 2007-10-23 08:55 ---
Fixed.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #2 from tbm at cyrius dot com 2007-10-23 08:55 ---
/* Testcase by Martin Michlmayr [EMAIL PROTECTED] */
extern void *memcpy (void *, __const void *, long unsigned int);
class csVector2
{
public:float x;
};
class csBox2
{
};
class csBoxClipper
{
csBox2 region;
virtual
--- Comment #5 from rguenth at gcc dot gnu dot org 2007-10-23 08:55 ---
Subject: Bug 33608
Author: rguenth
Date: Tue Oct 23 08:54:56 2007
New Revision: 129576
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=129576
Log:
2007-10-23 Richard Guenther [EMAIL PROTECTED]
PR
--- Comment #3 from tbm at cyrius dot com 2007-10-23 08:56 ---
Performing interprocedural optimizations
visibility early_local_cleanups inline static-var
pure-constAssembling functions:
virtual void csBoxClipper::Clip()
/home/tbm/crystalspace-polyclip.cc: In member function âvirtual
--- Comment #9 from tbm at cyrius dot com 2007-10-23 09:06 ---
(In reply to comment #8)
Can you try the patch that is in PR 33755 ?
Richard, can you do this? Since you wrote that patch I'm sure you have
a GCC with it applied (whereas I don't at the moment).
--
with 20071023-r129576 (the current trunk I used to
report this bug was before PR33620 got fixed.)
*** This bug has been marked as a duplicate of 33620 ***
--
tbm at cyrius dot com changed:
What|Removed |Added
--- Comment #7 from tbm at cyrius dot com 2007-10-23 09:36 ---
*** Bug 33865 has been marked as a duplicate of this bug. ***
--
tbm at cyrius dot com changed:
What|Removed |Added
The glibc testsuite runs into a var-tracking issue where we consume loads of
memory (2GB) to compile test-tgmath2.i. Probably related to PR31412.
./cc1 -O -o /dev/null test-tgmath2.i -std=gnu99 -fgnu89-inline
-fmerge-all-constants -fno-builtin -g
--
Summary: [4.3 Regression] Gross
--- Comment #1 from rguenth at gcc dot gnu dot org 2007-10-23 09:57 ---
Created an attachment (id=14397)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14397action=view)
testcase (unreduced)
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33868
--- Comment #2 from rguenth at gcc dot gnu dot org 2007-10-23 10:05 ---
We don't finish vt_find_locations () and have loads of variables allocated:
750 new_var = pool_alloc (var_pool);
(gdb) print *var_pool
$3 = {name = 0xf3d05a variable_def pool, id = 22681, elts_per_block = 64,
I see the following on IA64. It doesn't fail on x86_64.
(sid)[EMAIL PROTECTED]:~$ /usr/lib/gcc-snapshot/bin/g++ -c -O3
mercator-TerrainMod.cc
mercator-TerrainMod.cc: In member function 'AxisBox2
ShapeTerrainModShape::bbox() const [with Shape = RotBox2]':
mercator-TerrainMod.cc:59: error: missing
--- Comment #3 from tbm at cyrius dot com 2007-10-23 10:49 ---
This didn't ICE doesn't happen with 20070820 but happens now with 20071020.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33869
--- Comment #2 from tbm at cyrius dot com 2007-10-23 10:47 ---
/* Testcase by Martin Michlmayr [EMAIL PROTECTED] */
template const int dim class Point
{
public:
Point ()
{
}
void operator= (const Point rhs);
private:
float m_elem[dim];
};
template const int
--- Comment #1 from tbm at cyrius dot com 2007-10-23 10:47 ---
Created an attachment (id=14398)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14398action=view)
preprocessed source
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33869
--- Comment #4 from tbm at cyrius dot com 2007-10-23 11:11 ---
Also works with 20070902.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33869
--- Comment #5 from tbm at cyrius dot com 2007-10-23 11:12 ---
... and works with 20070916.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33869
--- Comment #6 from rguenth at gcc dot gnu dot org 2007-10-23 11:18 ---
Does disabling the vectorizer fix it?
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #7 from tbm at cyrius dot com 2007-10-23 11:32 ---
(In reply to comment #6)
Does disabling the vectorizer fix it?
Yes, that works, and it also fails with -O2 -ftree-vectorize.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33869
Consider this C code:
#include stdio.h
typedef long long int int64;
typedef long long unsigned int uint64;
struct pointer {
uint64 data : 58;
uint64 tag : 6;
};
union value {
int64 i;
struct pointer p;
};
typedef union value Value;
void test(Value x) {
printf(%llu\n, x.p.data);
}
--- Comment #1 from pinskia at gcc dot gnu dot org 2007-10-23 19:05 ---
See PR 21962 on the explanation on why this diagnostic is correct.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #24 from geir at cray dot com 2007-10-23 19:11 ---
State-Changed-From-To: open-suspended
What is the status of this bug? Will the proposed patches be implemented?
(Note: http://gcc.gnu.org/bugzilla/page.cgi?id=fields.html#status does not
describe SUSPENDED status).
--
--- Comment #5 from dorit at gcc dot gnu dot org 2007-10-23 19:50 ---
Subject: Bug 33860
Author: dorit
Date: Tue Oct 23 19:50:18 2007
New Revision: 129587
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=129587
Log:
PR tree-optimization/33860
* tree-vect-transform.c
--- Comment #25 from dave at hiauly1 dot hia dot nrc dot ca 2007-10-23
20:11 ---
Subject: Re: Bad reference to build directory in libstdc++.la
What is the status of this bug? Will the proposed patches be implemented?
(Note:
--- Comment #6 from uros at gcc dot gnu dot org 2007-10-23 20:15 ---
Subject: Bug 33846
Author: uros
Date: Tue Oct 23 20:15:04 2007
New Revision: 129589
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=129589
Log:
PR rtl-optimization/33846
* combine.c
--- Comment #7 from ubizjak at gmail dot com 2007-10-23 20:16 ---
Fixed.
--
ubizjak at gmail dot com changed:
What|Removed |Added
URL|
# On g++-sjlj.exe (GCC) 4.2.1-sjlj (mingw32 sjlj-unwind)
# g++ -v
Using built-in specs.
Target: mingw32
Configured with: ../gcc-4.2.1/configure --with-gcc --enable-libgomp
--host=mingw32 --build=mingw32 --target=mingw32 --program-suffix=-sjlj
--with-arch=i486 --with-tune=generic --disable-werror
--- Comment #29 from pthaugen at gcc dot gnu dot org 2007-10-23 20:29
---
Found another example on PowerPC in the same benchmark that is not fixed by
the checked in patches. Compiled with -m32 -O2. From the loop in procedure
FLUXI:
revision 126325:
.L47:
lfd
--- Comment #30 from pthaugen at gcc dot gnu dot org 2007-10-23 20:30
---
Created an attachment (id=14402)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14402action=view)
Testcase
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32921
--- Comment #1 from pinskia at gcc dot gnu dot org 2007-10-23 20:48 ---
*** This bug has been marked as a duplicate of 19774 ***
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #4 from pinskia at gcc dot gnu dot org 2007-10-23 20:48 ---
*** Bug 33874 has been marked as a duplicate of this bug. ***
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #8 from rguenth at gcc dot gnu dot org 2007-10-23 21:26 ---
For the parent variable of the missing VOP SFT.6 there is no SFT with offset
zero in pD.1574_36 NMTs aliases. So the following triggers:
! EXECUTE_IF_SET_IN_BITMAP (aliases, 0, i, bi)
{
! al
--- Comment #9 from rguenth at gcc dot gnu dot org 2007-10-23 21:27 ---
Created an attachment (id=14403)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14403action=view)
full checking patch
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33870
--- Comment #10 from rguenth at gcc dot gnu dot org 2007-10-23 21:28
---
And sadly the following doesn't help.
Index: tree-ssa-structalias.c
===
--- tree-ssa-structalias.c (revision 129582)
+++ tree-ssa-structalias.c
--- Comment #11 from rguenth at gcc dot gnu dot org 2007-10-23 21:52
---
Now it is memory paritioning again ;)
static void
rewrite_alias_set_for (tree tag, bitmap new_aliases)
{
bitmap_iterator bi;
unsigned i;
tree mpt, sym;
EXECUTE_IF_SET_IN_BITMAP (MTAG_ALIASES (tag), 0, i,
--- Comment #7 from burnus at gcc dot gnu dot org 2007-10-23 21:57 ---
(In reply to comment #6)
The test case generic_actual_arg.f90 fails because the dg-error has not been
adjusted:
/opt/gcc/_gcc-clean/gcc/testsuite/gfortran.dg/generic_actual_arg.f90:21.7:
CALL F(CALCULATION) ! {
--- Comment #16 from dhaliK at jla dot rutgers dot edu 2007-10-23 22:07
---
Ok, so I'm back on the track of what's going on here. I built it with multilib
as you said, and it builds 32/64 fine on Solaris, but when it goes to link
internally it chokes on the 64-bitness. In other words,
When local-pure_const_state in ipa-pure-const.c/check_decl has been set to
IPA_NEITHER, it cannot be modified in the rest of the function, so it is
possible to return immediately. This should bring a tiny performance
improvement while not impairing the code readability.
Patch follows, against SVN
--- Comment #1 from sam at rfc1149 dot net 2007-10-23 22:24 ---
Created an attachment (id=14404)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14404action=view)
Small performance improvement
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33875
const static data members of an integral type, which are initialized in the
class declaration can not be used in templated functions if these functions
take references as arguments. The linker reports an undefined reference to that
member.
The problem occured e.g. with std::min() and std:max()
--- Comment #1 from pinskia at gcc dot gnu dot org 2007-10-23 23:20 ---
And this is correct behavior. The C++ standard says they have to be declared
still, you just have a definition.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #2 from michael dot zillich at gmx dot net 2007-10-23 23:43
---
(In reply to comment #1)
And this is correct behavior. The C++ standard says they have to be declared
still, you just have a definition.
Ok, got You.
But then why the different bahaviour for
a) optimization
--- Comment #3 from pinskia at gcc dot gnu dot org 2007-10-24 00:29 ---
Subject: Re: in-class initialization of const static data members and template
functions
But then why the different bahaviour for
Because the standard does not require a diagnostic for this invalid code.
--
--- Comment #8 from jvdelisle at gcc dot gnu dot org 2007-10-24 03:29
---
Yes, regression tests help, I forgot to fix the testcase
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33849
--- Comment #9 from jvdelisle at gcc dot gnu dot org 2007-10-24 03:39
---
Subject: Bug 33849
Author: jvdelisle
Date: Wed Oct 24 03:39:03 2007
New Revision: 129595
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=129595
Log:
2007-10-23 Jerry DeLisle [EMAIL PROTECTED]
PR
--- Comment #27 from jason at gcc dot gnu dot org 2007-10-24 03:45 ---
Subject: Bug 25950
Author: jason
Date: Wed Oct 24 03:45:37 2007
New Revision: 129596
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=129596
Log:
PR c++/25950 (DR 391)
* call.c (struct
--- Comment #2 from ubizjak at gmail dot com 2007-10-24 05:21 ---
(In reply to comment #1)
Created an attachment (id=14404)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14404action=view) [edit]
Small performance improvement
Please post this patch (together with comment in
--- Comment #17 from ebotcazou at gcc dot gnu dot org 2007-10-24 05:26
---
In other words, gcc looks like it requires gnu ld in order to build 64bit
on Solaris and that was why we had been building two flavors of it.
That's wrong.
The file I'm showing above had been build with gcc
61 matches
Mail list logo
