[Bug c/21461] New: [ia64] miscompilation at -O2 in 3.4.x, 4.x
The SPASS theorem prover (see below) is miscompiled at -O2 by: gcc version 3.4.3 gcc version 3.4.4 20050314 (prerelease) (Debian 3.4.3-13) gcc version 4.0.0 (release) gcc version 4.1.0 20050509 (experimental) it is compiled correctly by: gcc version 3.3.6 (Debian 1:3.3.6-3) gcc version 3.3.5 Intel(R) C Itanium(R) Compiler for Itanium(R)-based applications Version 8.1 LLVM and the problematic compilers mentioned above at -O1. TO REPRODUCE: 1) find yourself an ia64 box 2) unpack http://kinoko.c.u-tokyo.ac.jp/~duraid/spassbug.tgz (sorry) 3) make (you'll want to change the gcc-3.4 -O2 that appears in the Makefile) 4) ./a.out bug_trigger (this segfaults fairly rapidly, it should run to completion in ~20s on an itanium 2 system) -- Summary: [ia64] miscompilation at -O2 in 3.4.x, 4.x Product: gcc Version: 3.4.3 Status: UNCONFIRMED Severity: normal Priority: P2 Component: c AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: duraid at octopus dot com dot au CC: gcc-bugs at gcc dot gnu dot org GCC build triplet: ia64-linux-unknown GCC host triplet: ia64-linux-unknown GCC target triplet: ia64-linux-unknown http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21461
[Bug c/21461] miscompilation at -O2 in 3.4.x, 4.x
--- Additional Comments From duraid at octopus dot com dot au 2005-05-09 06:44 --- oops, my mistake. this bug appears on x86 too, at least: gcc version 3.4.4 20050314 (prerelease) (Debian 3.4.3-13) fails (at -O2), but: gcc version 3.3.6 (Debian 1:3.3.6-3) works. Perhaps this is a bug in SPASS, but the program does appear to behave as far as valgrind is concerned. Not sure what sort of legal optimization could break a program in this way. I'm pretty sure there's no floating-point code affecting control flow here. -- What|Removed |Added GCC build triplet|ia64-linux-unknown | GCC host triplet|ia64-linux-unknown | GCC target triplet|ia64-linux-unknown | Summary|[ia64] miscompilation at -O2|miscompilation at -O2 in |in 3.4.x, 4.x |3.4.x, 4.x http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21461
[Bug target/19087] Overflowed address in dwarf debug line information
--- Additional Comments From tsandnes at norway dot atmel dot com 2005-05-09 07:01 --- Subject: Re: Overflowed address in dwarf debug line information bjoern dot m dot haase at web dot de wrote: --- Additional Comments From bjoern dot m dot haase at web dot de 2005-05-06 14:12 --- Hi Torleif, I have just had a look at PR19885 and I am having one additional question: IIUC, the lable references, that eventually cause the overflow problems refer to memory in form of byte addresses and not word addresses: I.e. the offsets and address information you are finding in the elf files all are byte addresses. Am I right? Yes. This is what I have observed. avr-readelf with the -w switch is very useful for viewing the debug information. In the asm outputs I have investigated so far, I realized that for the lable references in the dwarf sections avr-gcc is allocating two bytes only. So in case that these two bytes are filled with byte addresses instead of word addresses, it is obvious that an overflow could not be avoided. The cleanest solution then would be to redefine the debugging format such that within the debugging sections for the dwarf symbols, gcc allocates four bytes for the pointers. I fear, however, that this would be a major challenge since one would probably need to rewrite a considerable portion of the dwarf-output-genenerator in a target-specific fashion. If I'm not too wrong, we probably won't have a short term solution for this PR. I think your explanation makes sense. But how can target specific code be added to the dwarf output generator? AFAIK everything AVR related must be put in the avr.c, avr.h or avr.md in the gcc/config directory. Must it not rather be rewritten to work for all targets *and* the AVR port? Thanks for taking the time to look into this. Torleif -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19087
[Bug target/21397] -march selects wrong arm_tune flags
--- Additional Comments From trauscher at loytec dot com 2005-05-09 07:04 --- The problem is that -march=xxx -mtune=yyy doesn't work. The arm_override_option function scans the options in the order -mtune, -march, -mcpu. So when -mtune is given, 'arm_tune' first set by -mtune and then overwritten with the wrong tune settings from -march. I think, this problem was introduced between gcc-3.4.3 and gcc-4.0.0. In 3.4.3 only -mtune sets 'arm_tune' where in 4.0.0 all three flags affect 'arm_tune'. -- What|Removed |Added Status|RESOLVED|UNCONFIRMED Resolution|INVALID | http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21397
[Bug c/16676] [3.4 Regression] ICE with nested functions and -g1, blocks glibc
--- Additional Comments From giovannibajo at libero dot it 2005-05-09 07:07 --- Zack, this is a regression of part of your c-decl stuff. Can you possibly give it a look? It breaks builds of glibc on primary platforms. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16676
[Bug c/16676] [3.4 Regression] ICE with nested functions and -g1, blocks glibc
--- Additional Comments From benh at kernel dot crashing dot org 2005-05-09 07:11 --- Ben Elliston just produced a patch for it that I tested. It fixed building of glibc on debian powerpc with -g1 (used by debian rules for nptl). The patch is on it's way to the patch list (which didn't catch up yet it seems). The patch is very simple, it just causes gcc to emit more debug infos in the case of nested functions (like the -g2 case) instead of trying to be smart. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16676
[Bug c++/18604] [3.4/4.0/4.1 Regression] Strong using lookup conflicts
--- Additional Comments From giovannibajo at libero dot it 2005-05-09
07:25 ---
I analyzed this PR again. The problem seems not related to attribute strong
itself, but rather the way v3 namespaces work. Basically, without debug
support, the failing code looks like:
-
namespace std {
struct A {};
};
struct A;
namespace std {
A a;// std::A, OK!
};
-
Instead, with debug support, we end up with code like this:
-
namespace std {
struct A {};
};
namespace __gnu_norm {
using namespace std;
};
struct A;
namespace __gnu_norm {
A a;// ambiguous!
};
-
hence the failure. There are at least two ways to fix this in v3:
- Fully qualify all references to std names from code within namespace
_GLIBCXX_STD (which expands to __gnu_norm in debug mode). This is hard because
it's impossible to fully check it, and it's pervasive. E.g. bits/stl_set.h:69,
use std::less instead of less.
- Define namespace __gnu_norm within namespace std. This could be easier
because you probably have to touch *only* namespace definitions.
Comments?
--
What|Removed |Added
CC||bkoz at gcc dot gnu dot org,
||mmitchel at gcc dot gnu dot
||org, pcarlini at suse dot de
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18604
[Bug tree-optimization/21462] New: if statement not converted to MIN_EXPR
Consider the following short program:
#include algorithm
void Tst1(short* __restrict__ SrcP, short* __restrict__ MinP, int Len)
{
for (int x=0; xLen; x++)
MinP[x] = SrcP[x] ? MinP[x];
}
void Tst2(short* __restrict__ SrcP, short* __restrict__ MinP, int Len)
{
for (int x=0; xLen; x++)
MinP[x] = (SrcP[x]MinP[x]) ? SrcP[x] : MinP[x];
}
void Tst3(short* __restrict__ SrcP, short* __restrict__ MinP, int Len)
{
for (int x=0; xLen; x++)
MinP[x] = (SrcP[x]MinP[x]) ? (short)SrcP[x] : MinP[x];
}
void Tst4(short* __restrict__ SrcP, short* __restrict__ MinP, int Len)
{
for (int x=0; xLen; x++)
MinP[x] = std::min(SrcP[x], MinP[x]);
}
void Tst5(short* __restrict__ SrcP, short* __restrict__ MinP, int Len)
{
for (int x=0; xLen; x++)
if (SrcP[x]MinP[x]) MinP[x]=SrcP[x];
}
If I compile it with
gcc41 -O2 -ftree-vectorize -ftree-vectorizer-verbose=5
function Tst1/Tst2/Tst3 gets vectorized but Tst4/Tst5 not.
The reason for this is the first three funtions result in a MIN_EXPR while
std::min and the if statement generate conditional code which the vectorizer
does not recognize.
Michael Cieslinski
--
Summary: if statement not converted to MIN_EXPR
Product: gcc
Version: 4.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: tree-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: micis at gmx dot de
CC: gcc-bugs at gcc dot gnu dot org
GCC build triplet: x86_64-unknown-linux-gnu
GCC host triplet: x86_64-unknown-linux-gnu
GCC target triplet: x86_64-unknown-linux-gnu
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21462
[Bug tree-optimization/19716] Segfault with -ftree-vectorize
--- Additional Comments From micis at gmx dot de 2005-05-09 08:24 --- I cannot reproduce this bug any longer. Maybe the fix for pr20122 also helps here. Michael Cieslinski -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19716
[Bug regression/21463] New: min/max and references
I don't quite know how to characterize this one, so i'll let it up to those in
the know to fix the summary/description.
Primo, AFAIK, gcc has always struggled to get this right but 4.1 is setting a
new record; so i'll qualify that as a regression vs 3.3/3.4.
Secundo, i haven't found any related bugreports but one about PPC.
So here we go.
Consider this testcase:
#include algorithm
templateclass T static inline T max(const T a, const T b) { return ab ? b :
a; }
templateclass T static inline T min(const T a, const T b) { return ab ? a :
b; }
templateclass T static inline const T ref_max(const T a, const T b) {
return ab ? b : a; }
templateclass T static inline const T ref_min(const T a, const T b) {
return ab ? a : b; }
templateclass T struct foo_t {
T a0, a1;
T bar(const T b, const T c) {
return max(min(a0, c), min(max(a1, c), b));
}
T bar_ref(const T b, const T c) {
return ref_max(ref_min(a0, c), ref_min(ref_max(a1, c), b));
}
T bar_stl(const T b, const T c) {
return std::max(std::min(a0, c), std::min(std::max(a1, c), b));
}
};
template struct foo_tint;
template struct foo_tfloat;
int main() { return 0; }
With g++-4120050501 -O3 i get such creative entertaining code as:
00400610 foo_tfloat::bar_ref(float, float):
400610: ucomiss 0x4(%rdi),%xmm1
400614: lea0x4(%rdi),%rax
400618: lea0xfff8(%rsp),%rdx
40061d: movss %xmm0,0xfffc(%rsp)
400623: movss %xmm1,0xfff8(%rsp)
400629: movaps %xmm1,%xmm2
40062c: cmova %rdx,%rax
400630: movss (%rax),%xmm1
400634: ucomiss %xmm1,%xmm0
400637: ja 400641 foo_tfloat::bar_ref(float, float)+0x31
400639: lea0xfffc(%rsp),%rax
40063e: movaps %xmm0,%xmm1
400641: ucomiss (%rdi),%xmm2
400644: cmova %rdi,%rdx
400648: movss (%rdx),%xmm0
40064c: ucomiss %xmm0,%xmm1
40064f: jbe400655 foo_tfloat::bar_ref(float, float)+0x45
400651: movss (%rax),%xmm0
400655: repz retq
Compare that to what g++-3.4.4-20050314 gives (g++ 3.3.6 is similar):
00400610 foo_tfloat::bar_ref(float, float):
400610: ucomiss (%rdi),%xmm1
400613: lea0xfffc(%rsp),%rsi
400618: mov%rdi,%rcx
40061b: lea0x4(%rdi),%rdx
40061f: movss %xmm0,0xfff8(%rsp)
400625: lea0xfff8(%rsp),%rax
40062a: movss %xmm1,0xfffc(%rsp)
400630: cmovbe %rsi,%rcx
400634: ucomiss 0x4(%rdi),%xmm1
400638: cmova %rsi,%rdx
40063c: ucomiss (%rdx),%xmm0
40063f: cmova %rdx,%rax
400643: movss (%rax),%xmm0
400647: ucomiss (%rcx),%xmm0
40064a: cmovbe %rcx,%rax
40064e: movss (%rax),%xmm0
400652: retq
Certainly not optimal, and in fact quite ugly, but at least there's no branch.
Happens as soon as references are used and enough min/max are piled up, and that
means that unless you use your own min/max instead of the STL version, you're
doomed.
--
Summary: min/max and references
Product: gcc
Version: 4.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: regression
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: tbptbp at gmail dot com
CC: gcc-bugs at gcc dot gnu dot org
GCC host triplet: x86*
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21463
[Bug rtl-optimization/19680] sub-optimial register allocation with sse
--- Additional Comments From tbptbp at gmail dot com 2005-05-09 08:46 --- I'm going to ping this bugreport because there's still some very bad interaction with gcse in current gcc. Just compile the 'packet_intersection.cpp' testcase with ie g++-4.1-4120050501 for ia32 to convince yourself (x86-64 is apparently immune, thanks to the lower register pressure). It takes a bit of luck, or the use of -fno-gcse, to dodge that issue which really kills performance. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19680
[Bug regression/21463] min/max and references
--- Additional Comments From tbptbp at gmail dot com 2005-05-09 08:50 --- I forgot to say that using ternary operators or if/else while changing the codegen slightly doesn't make much difference. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21463
[Bug c++/18604] [3.4/4.0/4.1 Regression] Strong using lookup conflicts
--- Additional Comments From pcarlini at suse dot de 2005-05-09 09:04 --- - Fully qualify all references to std names from code within namespace _GLIBCXX_STD (which expands to __gnu_norm in debug mode). This is hard because it's impossible to fully check it, and it's pervasive. E.g. bits/stl_set.h:69, use std::less instead of less. Actually, this is already the case (modulo mistakes): we did this work a lot of time ago, *before* the strong using work. The only remaining occurrences of unqualified names are fixed as we find them (Jonathan is doing a great job here, and in fact he uses debug mode) or are there for a reason: if I remember correctly, this is the case of less. I'm pretty sure Gaby said something. Gaby? In case fixing that would be trivial, of course. -- What|Removed |Added CC||gdr at integrable-solutions ||dot net http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18604
[Bug c++/18604] [3.4/4.0/4.1 Regression] Strong using lookup conflicts
--- Additional Comments From pcarlini at suse dot de 2005-05-09 09:16 --- I found something in the archive: http://gcc.gnu.org/ml/libstdc++/2002-12/msg9.html That's why we didn't qualify less co... In mainline (vs v7-branch), where we still don't have anything special for those function objects, maybe we should add the qualifications? -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18604
[Bug c/21464] New: assembler code is not generated for uint64_t comparison with -O
input:
== a.c =
#include stdio.h
#include inttypes.h
int main() {
uint64_t pos = 1;
pos = 1;
if (pos 1 || pos 0xULL) {
printf(fail\n);
return 1;
}
printf(ok\n);
return 0;
}
= end of a.c =
command lines:
$ gcc a.c
$ ./a.out
ok
$ gcc -O a.c
$ ./a.out
fail
assembler code for comparison of pos is not generated correctly.
it should test, at least, pos 1,
but generated code act like if the condition was true.
if pos 0xUL is not present, it runs successfully.
definitely, optimization bug while preevaluating the condition at compile time.
my gcc version information is:
$ gcc -v
Reading specs from /usr/lib/gcc/i686-pc-linux-gnu/3.4.3-20050110/specs
Configured with: /var/tmp/portage/gcc-3.4.3.20050110-r2/work/gcc-3.4.3/configure
--enable-version-specific-runtime-libs --prefix=/usr
--bindir=/usr/i686-pc-linux-gnu/gcc-bin/3.4.3-20050110
--includedir=/usr/lib/gcc/i686-pc-linux-gnu/3.4.3-20050110/include
--datadir=/usr/share/gcc-data/i686-pc-linux-gnu/3.4.3-20050110
--mandir=/usr/share/gcc-data/i686-pc-linux-gnu/3.4.3-20050110/man
--infodir=/usr/share/gcc-data/i686-pc-linux-gnu/3.4.3-20050110/info
--with-gxx-include-dir=/usr/lib/gcc/i686-pc-linux-gnu/3.4.3-20050110/include/g++-v3
--host=i686-pc-linux-gnu --disable-altivec --enable-nls
--without-included-gettext --with-system-zlib --disable-checking
--disable-werror --disable-libunwind-exceptions --disable-multilib
--disable-libgcj --enable-languages=c,c++,f77 --enable-shared
--enable-threads=posix --enable-__cxa_atexit --enable-clocale=gnu
Thread model: posix
gcc version 3.4.3-20050110 (Gentoo Linux 3.4.3.20050110-r2,
ssp-3.4.3.20050110-0, pie-8.7.7)
--
Summary: assembler code is not generated for uint64_t comparison
with -O
Product: gcc
Version: 3.4.3
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: c
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: moonz at kaist dot ac dot kr
CC: gcc-bugs at gcc dot gnu dot org
GCC build triplet: i686-pc-linux-gnu
GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21464
[Bug java/21442] problem with imports and multifile builds
-- What|Removed |Added OtherBugsDependingO||21418 nThis|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21442
[Bug java/21442] problem with imports and multifile builds
--- Additional Comments From bonzini at gcc dot gnu dot org 2005-05-09 10:03 --- The problem is that, while looking for AbstractInterruptibleChannel (a superclass of SelectableChannel), we do not look into the imports of SelectableChannel. *** This bug has been marked as a duplicate of 21436 *** -- What|Removed |Added BugsThisDependsOn|21436 | Status|UNCONFIRMED |RESOLVED Resolution||DUPLICATE http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21442
[Bug java/21436] imports not remembered for previously compiled files
--- Additional Comments From bonzini at gcc dot gnu dot org 2005-05-09 10:03 --- *** Bug 21442 has been marked as a duplicate of this bug. *** -- What|Removed |Added OtherBugsDependingO|21442 | nThis|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21436
[Bug java/21418] Order of source files matters when compiling
-- Bug 21418 depends on bug 21442, which changed state. Bug 21442 Summary: problem with imports and multifile builds http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21442 What|Old Value |New Value Status|UNCONFIRMED |RESOLVED Resolution||DUPLICATE http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21418
[Bug tree-optimization/21465] New: autovectorisation conflicts with -ffast-math
Consider the following short function:
void Tst1(float* __restrict__ SrcP, float* __restrict__ DstP, int Len)
{
for (int x=0; xLen; x++)
{
DstP[x] = SrcP[x] * SrcP[x];
}
}
If this function is compiled with:
g++41 -O -S TstPow.cpp -ftree-vectorizer-verbose=5 -ftree-vectorize
it is vectorized, but if you compile with:
g++41 -O -ffast-math -S TstPow.cpp -ftree-vectorizer-verbose=5
-ftree-vectorize
it is not with the following output:
note: Vectorizing an unaligned access.
note: Alignment of access forced using peeling.
note: not vectorized: stmt not supported: D.2052_15 = __builtin_powf
(D.2051_14, 2.0e+0)
note: vectorized 0 loops in function.
Michael Cieslinski
--
Summary: autovectorisation conflicts with -ffast-math
Product: gcc
Version: 4.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: tree-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: micis at gmx dot de
CC: gcc-bugs at gcc dot gnu dot org
GCC build triplet: x86_64-unknown-linux-gnu
GCC host triplet: x86_64-unknown-linux-gnu
GCC target triplet: x86_64-unknown-linux-gnu
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21465
[Bug libgcj/21414] thread lib posix = runtime gij segfaults on Debian G/L Woody
--- Additional Comments From vaclav dot ovsik at i dot cz 2005-05-09 10:47 --- Subject: Re: thread lib posix = runtime gij segfaults on Debian G/L Woody On Fri, May 06, 2005 at 01:45:49PM -, pinskia at gcc dot gnu dot org wrote: This sounds like recursive mutexes are not working on your machine. Also you need a newer binutils, from http://gcc.gnu.org/install/specific.html: As of GCC 3.3, binutils 2.13.1 or later is required for this platform. See bug 10877 for more information. Can you first try with a newer binutils? I compiled and installed binutils-2.16 and rebuild gcc-4.0.0 (default thread file posix). There are no segfaults anymore. Please, excuse my ignorance in reading http://gcc.gnu.org/install/specific.html, I overlooked requirement of binutils ver. = 2.13.1. Problem can be closed. Best Regards -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21414
[Bug tree-optimization/21466] New: sqrt() function not vectorised
The following function is not vectorised while there is a sqrt vector function
available.
#include math.h
void Tst1(float* __restrict__ SrcP, float* __restrict__ DstP, int Len)
{
for (int x=0; xLen; x++)
DstP[x] = sqrt(SrcP[x]);
}
Compiled with
g++41 -O -ffast-math -S TstSqrt.cpp -ftree-vectorizer-verbose=5
-ftree-vectorize -march=opteron
gives following output:
TstSqrt.cpp:5: note: Vectorizing an unaligned access.
TstSqrt.cpp:5: note: Alignment of access forced using peeling.
TstSqrt.cpp:5: note: not vectorized: stmt not supported: D.3171_15 =
__builtin_sqrtf (D.3170_14)
TstSqrt.cpp:3: note: vectorized 0 loops in function.
Michael Cieslinski
--
Summary: sqrt() function not vectorised
Product: gcc
Version: 4.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: tree-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: micis at gmx dot de
CC: gcc-bugs at gcc dot gnu dot org
GCC build triplet: x86_64-unknown-linux-gnu
GCC host triplet: x86_64-unknown-linux-gnu
GCC target triplet: x86_64-unknown-linux-gnu
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21466
[Bug libfortran/21459] strings of different length in a single character(len=*) declaration
--- Additional Comments From fxcoudert at gcc dot gnu dot org 2005-05-09 11:15 --- Reduced testcase is following. The root of the problem is that when we declare strings of different length in the same character(len=*) declaration, all strings are truncated to the length of the first one. Example: $ cat pr21459.f90 character(len=*) :: c1='12345', c2='123456789' print *, c1 print *, c2 end $ gfortran pr21459.f90 ./a.out 12345 12345 -- What|Removed |Added CC||fxcoudert at gcc dot gnu dot ||org OtherBugsDependingO||19276 nThis|| Status|UNCONFIRMED |NEW Ever Confirmed||1 Last reconfirmed|-00-00 00:00:00 |2005-05-09 11:15:34 date|| Summary|Fortran runtime error: |strings of different length |Unexpected end of format|in a single character(len=*) |string |declaration http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21459
[Bug target/21397] arm: -march overrides -mtune
--- Additional Comments From rearnsha at gcc dot gnu dot org 2005-05-09 11:18 --- Ah! I understand now. Sorry, your initial report wasn't clear. Yes, I agree this is incorrect. -- What|Removed |Added Status|UNCONFIRMED |NEW Ever Confirmed||1 Last reconfirmed|-00-00 00:00:00 |2005-05-09 11:18:37 date|| Summary|-march selects wrong|arm: -march overrides -mtune |arm_tune flags | http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21397
[Bug libfortran/19155] blanks not treated as zeros in 'E' format read (NIST FM110.FOR)
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 11:27 --- Subject: Bug 19155 CVSROOT:/cvs/gcc Module name:gcc Changes by: [EMAIL PROTECTED] 2005-05-09 11:21:02 Modified files: libgfortran/io : unix.c libgfortran: ChangeLog gcc/testsuite : ChangeLog Added files: gcc/testsuite/gfortran.dg: pr19155.f Log message: PR libfortran/19155 * io/read.c (read_f): Accept 'e', 'E', 'd' and 'D' as first non-blank characters of a real number. * gfortran.dg/pr19155.f: New test. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libgfortran/io/unix.c.diff?cvsroot=gccr1=1.22r2=1.23 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libgfortran/ChangeLog.diff?cvsroot=gccr1=1.208r2=1.209 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gccr1=1.5448r2=1.5449 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/pr19155.f.diff?cvsroot=gccr1=NONEr2=1.1 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19155
[Bug libfortran/21376] libfortran E output format causes FPE
-- What|Removed |Added AssignedTo|unassigned at gcc dot gnu |fxcoudert at gcc dot gnu dot |dot org |org Status|UNCONFIRMED |ASSIGNED Ever Confirmed||1 GCC build triplet|i686-pc-linux-gnu | GCC host triplet|i686-pc-linux-gnu | GCC target triplet|i686-pc-linux-gnu | Last reconfirmed|-00-00 00:00:00 |2005-05-09 11:29:26 date|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21376
[Bug rtl-optimization/21461] [3.4/4.0 regression] miscompilation at -O2 in 3.4.x, 4.x
--- Additional Comments From belyshev at depni dot sinp dot msu dot ru 2005-05-09 11:30 --- Could you please post backtrace of segfault? with both gcc-4.0 and gcc-3.4 if it is different. -- What|Removed |Added Component|c |rtl-optimization Keywords||wrong-code Known to work||3.3.6 Summary|miscompilation at -O2 in|[3.4/4.0 regression] |3.4.x, 4.x |miscompilation at -O2 in ||3.4.x, 4.x http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21461
[Bug libfortran/19155] blanks not treated as zeros in 'E' format read (NIST FM110.FOR)
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 11:31 --- Subject: Bug 19155 CVSROOT:/cvs/gcc Module name:gcc Branch: gcc-4_0-branch Changes by: [EMAIL PROTECTED] 2005-05-09 11:27:06 Modified files: gcc/testsuite : ChangeLog libgfortran: ChangeLog libgfortran/io : read.c Added files: gcc/testsuite/gfortran.dg: pr19155.f Log message: PR libfortran/19155 * io/read.c (read_f): Accept 'e', 'E', 'd' and 'D' as first non-blank characters of a real number. * gfortran.dg/pr19155.f: New test. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.5084.2.166r2=1.5084.2.167 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gfortran.dg/pr19155.f.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=NONEr2=1.1.2.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libgfortran/ChangeLog.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.163.2.24r2=1.163.2.25 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libgfortran/io/read.c.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.8r2=1.8.10.1 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19155
[Bug libfortran/19155] blanks not treated as zeros in 'E' format read (NIST FM110.FOR)
--- Additional Comments From fxcoudert at gcc dot gnu dot org 2005-05-09 11:37 --- I don't close this one since NIST test FM110 is not fixed with this patch. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19155
[Bug c++/21427] [3.4/4.0 only] Failure in make_thunk, cp/method.c:145 when compiling with multiply-inherited members
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 11:53 --- Subject: Bug 21427 CVSROOT:/cvs/gcc Module name:gcc Branch: gcc-3_4-branch Changes by: [EMAIL PROTECTED] 2005-05-09 11:48:12 Modified files: gcc/cp : ChangeLog class.c cp-tree.h method.c gcc/testsuite : ChangeLog Added files: gcc/testsuite/g++.dg/inherit: covariant12.C covariant13.C Log message: cp: PR c++/21427 Backport 2005-03-01 Nathan Sidwell [EMAIL PROTECTED] * class.c (update_vtable_entry_for_fn): Don't crash on invalid covariancy. * cp-tree.h (THUNK_TARGET): Expand comment. * method.c (use_thunk): Make sure we also use the target, if that is a thunk. Backport 2005-02-11 Nathan Sidwell [EMAIL PROTECTED] * class.c (update_vtable_entry_for_fn): Walk the covariant's binfo chain rather than using lookup_base. testsuite: PR c++/21427 * g++.dg/inherit/covariant13.C: New. Backport 2005-03-01 Nathan Sidwell [EMAIL PROTECTED] * g++.dg/inherit/covariant12.C: New. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/ChangeLog.diff?cvsroot=gcconly_with_tag=gcc-3_4-branchr1=1.3892.2.217r2=1.3892.2.218 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/class.c.diff?cvsroot=gcconly_with_tag=gcc-3_4-branchr1=1.595.4.10r2=1.595.4.11 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/cp-tree.h.diff?cvsroot=gcconly_with_tag=gcc-3_4-branchr1=1.946.4.19r2=1.946.4.20 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/method.c.diff?cvsroot=gcconly_with_tag=gcc-3_4-branchr1=1.275.4.5r2=1.275.4.6 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcconly_with_tag=gcc-3_4-branchr1=1.3389.2.391r2=1.3389.2.392 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/inherit/covariant12.C.diff?cvsroot=gcconly_with_tag=gcc-3_4-branchr1=NONEr2=1.1.20.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/inherit/covariant13.C.diff?cvsroot=gcconly_with_tag=gcc-3_4-branchr1=NONEr2=1.1.2.1 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21427
[Bug c++/21427] [3.4/4.0 only] Failure in make_thunk, cp/method.c:145 when compiling with multiply-inherited members
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 11:56 --- Subject: Bug 21427 CVSROOT:/cvs/gcc Module name:gcc Changes by: [EMAIL PROTECTED] 2005-05-09 11:52:45 Modified files: gcc/testsuite : ChangeLog Added files: gcc/testsuite/g++.dg/inherit: covariant13.C Log message: PR c++/21427 * g++.dg/inherit/covariant13.C: New. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gccr1=1.5449r2=1.5450 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/inherit/covariant13.C.diff?cvsroot=gccr1=1.1r2=1.2 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21427
[Bug rtl-optimization/21461] [3.4/4.0 regression] miscompilation at -O2 in 3.4.x, 4.x
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 12:01 --- Does -fno-strict-aliasing fix the problem? Also is there any warnings from -Wstrict-aliasing? If so this might not be a bug in gcc. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21461
[Bug fortran/15080] Forall bounds not calculated correctly (forall_3.f90)
--- Additional Comments From tobi at gcc dot gnu dot org 2005-05-09 12:05 --- Are you going to review that patch, Paul? I don't think anybody else is qualified. -- What|Removed |Added CC||tobi at gcc dot gnu dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15080
[Bug libfortran/15160] Generated files don't regenerate
--- Additional Comments From tobi at gcc dot gnu dot org 2005-05-09 12:06 --- I don't think so. Several people have used it, and noone complained about weird things happening. -- What|Removed |Added Status|NEW |RESOLVED Resolution||WORKSFORME http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15160
[Bug c++/18604] [3.4/4.0/4.1 Regression] Strong using lookup conflicts
--- Additional Comments From giovannibajo at libero dot it 2005-05-09 12:18 --- There are others unqualified names as default template parmaters, for instance allocator. A good testcase would be something like: -- struct less; struct allocator; struct vector; struct set; struct pair; struct map; /**/ #include set #include vector #include map #include functional /**/ -- which should compile without errors. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18604
[Bug libgcj/21414] thread lib posix = runtime gij segfaults on Debian G/L Woody
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 12:20 --- Closing as works for me. -- What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution||WORKSFORME http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21414
[Bug rtl-optimization/21461] [3.4/4.0 regression] miscompilation at -O2 in 3.4.x, 4.x
--- Additional Comments From duraid at octopus dot com dot au 2005-05-09 12:20 --- (In reply to comment #3) Does -fno-strict-aliasing fix the problem? Yes, oops. Also is there any warnings from -Wstrict-aliasing? No. If so this might not be a bug in gcc. Indeed. Sorry! -- What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution||INVALID http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21461
[Bug c++/21427] [3.4/4.0 only] Failure in make_thunk, cp/method.c:145 when compiling with multiply-inherited members
--- Additional Comments From nathan at gcc dot gnu dot org 2005-05-09 12:21 --- 2005-05-08 Nathan Sidwell [EMAIL PROTECTED] PR c++/21427 Backport 2005-03-01 Nathan Sidwell [EMAIL PROTECTED] * class.c (update_vtable_entry_for_fn): Don't crash on invalid covariancy. * cp-tree.h (THUNK_TARGET): Expand comment. * method.c (use_thunk): Make sure we also use the target, if that is a thunk. Backport 2005-02-11 Nathan Sidwell [EMAIL PROTECTED] * class.c (update_vtable_entry_for_fn): Walk the covariant's binfo chain rather than using lookup_base. -- What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21427
[Bug tree-optimization/21466] sqrt() function not vectorized
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 12:21 --- Confirmed. -- What|Removed |Added Severity|normal |enhancement Status|UNCONFIRMED |NEW Ever Confirmed||1 Keywords||missed-optimization Last reconfirmed|-00-00 00:00:00 |2005-05-09 12:21:28 date|| Summary|sqrt() function not |sqrt() function not |vectorised |vectorized http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21466
[Bug tree-optimization/21465] autovectorisation conflicts with -ffast-math
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 12:21 --- Confirmed. -- What|Removed |Added Severity|normal |minor Status|UNCONFIRMED |NEW Ever Confirmed||1 Keywords||missed-optimization Last reconfirmed|-00-00 00:00:00 |2005-05-09 12:21:51 date|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21465
[Bug rtl-optimization/21461] [3.4/4.0 regression] miscompilation at -O2 in 3.4.x, 4.x
--- Additional Comments From duraid at octopus dot com dot au 2005-05-09 12:24 --- Actually, I shouldn't have closed this so hastily. The code _is_ pretty dirty but I'm not sure GCC is really doing something legal at -O2. -- What|Removed |Added Status|RESOLVED|UNCONFIRMED Resolution|INVALID | http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21461
[Bug libstdc++/18604] [3.4/4.0/4.1 Regression] Use of unqualified names conflict with names in global namespace
--- Additional Comments From pcarlini at suse dot de 2005-05-09 12:29 --- I see, thanks. Indeed, I don't remember having paid attention to default template arguments. Let's wait a bit for Gaby's opinion on the whole issue, and, in case, let's quickly fix those problems. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18604
[Bug c++/21467] New: ICE for template function if missing required default argument
The following code makes g++ 3.4.4 and g++ 3.3.6 (Debian) crash on my system.
Adding the missing default argument for arg1 fixes the crash
templatetypename F
void test(F function, bool arg0 = false, bool arg1)
{
}
int main()
{
test(main);
}
Detailed G++ versions:
g++-3.3 (GCC) 3.3.6 (Debian 1:3.3.6-3)
Reading specs from /usr/lib/gcc-lib/powerpc-linux/3.3.6/specs
Configured with: ../src/configure -v
--enable-languages=c,c++,java,f77,pascal,objc,ada --prefix=/usr
--mandir=/usr/share/man --infodir=/usr/share/info
--with-gxx-include-dir=/usr/include/c++/3.3 --enable-shared
--enable-__cxa_atexit --with-system-zlib --enable-nls
--without-included-gettext --enable-clocale=gnu --enable-debug
--enable-java-gc=boehm --enable-java-awt=xlib --enable-objc-gc
--disable-multilib powerpc-linux
Thread model: posix
gcc version 3.3.6 (Debian 1:3.3.6-3)
g++-3.4 (GCC) 3.4.4 20050314 (prerelease) (Debian 3.4.3-13)
Reading specs from /usr/lib/gcc/powerpc-linux/3.4.4/specs
Configured with: ../src/configure -v
--enable-languages=c,c++,java,f77,pascal,objc,ada --prefix=/usr
--libexecdir=/usr/lib --with-gxx-include-dir=/usr/include/c++/3.4
--enable-shared --with-system-zlib --enable-nls --without-included-gettext
--program-suffix=-3.4 --enable-__cxa_atexit --enable-libstdcxx-allocator=mt
--enable-clocale=gnu --enable-libstdcxx-debug --enable-java-gc=boehm
--enable-java-awt=gtk --disable-multilib --disable-softfloat powerpc-linux
Thread model: posix
gcc version 3.4.4 20050314 (prerelease) (Debian 3.4.3-13)
--
Summary: ICE for template function if missing required default
argument
Product: gcc
Version: 3.4.4
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: c++
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: sylvain dot joyeux at m4x dot org
CC: gcc-bugs at gcc dot gnu dot org
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21467
[Bug c++/21467] ICE for template function if missing required default argument
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 12:53 --- Confirmed and not a regression. Here is the backtrace: #0 convert_default_arg (type=0xb7bf9804, arg=0x0, fn=0xb7c8c948, parmnum=2) at /home/peshtigo/pinskia/src/gnu/gcc/src/gcc/cp/call.c:4513 #1 0x0810a7c0 in build_function_call (function=0xb7c8c948, params=0x0) at /home/peshtigo/pinskia/src/gnu/gcc/src/gcc/cp/typeck.c:2618 #2 0x0805c8b9 in build_new_function_call (fn=0xb7c68e40, args=0xb7c68f30) at /home/peshtigo/pinskia/src/gnu/gcc/src/gcc/cp/call.c:2763 #3 0x08131ea0 in finish_call_expr (fn=0xb7c68e40, args=0xb7c68f30, disallow_virtual=0 '\0', koenig_p=1 '\001') at /home/peshtigo/pinskia/src/gnu/gcc/src/gcc/cp/semantics.c:1846 #4 0x080ef725 in cp_parser_unary_expression (parser=0xb7c8b2d8, address_p=0 '\0', cast_p=0 '\0') at /home/peshtigo/pinskia/src/gnu/gcc/src/gcc/cp/parser.c:4158 #5 0x080f0192 in cp_parser_assignment_expression (parser=0xb7c8b2d8, cast_p=0 '\0') at /home/peshtigo/pinskia/src/gnu/gcc/src/gcc/cp/parser.c:5413 #6 0x080f1939 in cp_parser_expression (parser=0xb7c8b2d8, cast_p=Variable cast_p is not available. ) -- What|Removed |Added Status|UNCONFIRMED |NEW Ever Confirmed||1 Keywords||ice-on-invalid-code Known to fail||2.95.3 3.0.4 3.2.3 3.3.3 ||3.4.0 4.0.0 Last reconfirmed|-00-00 00:00:00 |2005-05-09 12:53:37 date|| Summary|ICE for template function if|ICE for template function if |missing required default|missing required default |argument|argument http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21467
[Bug ada/14997] ncurses build fails with Ada
--- Additional Comments From ebotcazou at gcc dot gnu dot org 2005-05-09 13:00 --- Looking into it for AdaCore. -- What|Removed |Added AssignedTo|unassigned at gcc dot gnu |ebotcazou at gcc dot gnu dot |dot org |org Status|NEW |ASSIGNED Last reconfirmed|2005-02-23 23:42:44 |2005-05-09 13:00:15 date|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14997
[Bug rtl-optimization/21461] [3.4/4.0 regression] miscompilation at -O2 in 3.4.x, 4.x
--- Additional Comments From belyshev at depni dot sinp dot msu dot ru 2005-05-09 13:23 --- yes, please do not close this bug as i can reproduce it even with -fno-strict-aliasing, but it seems it breaks at least in four files (dfgparser.c, list.c, sharing.c, subst.c) so it could take some time to reduce. I'd like to see your backtrace of segfault to compare with. (compile it with -O2 -g, and do gdb --args ./a.out bug_trigger and then 'run' and 'bt') -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21461
[Bug fortran/16939] Pointers not passed as subroutine arguments
--- Additional Comments From tobi at gcc dot gnu dot org 2005-05-09 13:40
---
What is wrong is this from the .t02.original dump:
create (temp_ptr, _temp_ptr)
{
char[1:15] * new_item;
{
void * * ptr.0;
ptr.0 = (void * *) new_item;
_gfortran_allocate (ptr.0, 15, 0);
}
_gfortran_copy_string (15, new_item, 14, A NEW STRING );
temp_ptr = (char[1:15] * ) new_item;
^
_gfortran_filename = pr16939.f90;
_gfortran_line = 19;
_gfortran_ioparm.unit = 6;
_gfortran_ioparm.list_format = 1;
_gfortran_st_write ();
_gfortran_transfer_character (temp_ptr, 15);
_gfortran_st_write_done ();
goto __return_create;
__return_create:;
}
The underlined line should be
temp_ptr = new_item
unless I misread something. temp_ptr should point to the same character string
new_item points to. The underlined statement makes it point to new_item
instead.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16939
[Bug rtl-optimization/21461] [3.4/4.0 regression] miscompilation at -O2 in 3.4.x, 4.x
--- Additional Comments From duraid at octopus dot com dot au 2005-05-09 13:41 --- Building on ia64 with the 3.4.4 compiler mentioned above, I get: #0 red_ReduceInput (Search=0x600ac338, ClauseList=0x60112e18) at clause.h:525 #1 0x4010dd90 in top_ProofSearch (Search=0x600ac338, ProblemClauses=0x60123f68, InputFlags=0x600cd948, UserPrecedence=0x0, BoundApplied=0x6fcf3768) at top.c:913 #2 0x40111350 in main (argc=2, argv=0x0) at top.c:1457 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21461
[Bug java/19285] Interfaces not initialized by static field access
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 13:41 --- Subject: Bug 19285 CVSROOT:/cvs/gcc Module name:gcc Branch: gcc-4_0-branch Changes by: [EMAIL PROTECTED] 2005-05-09 13:32:30 Modified files: libjava: ChangeLog prims.cc link.cc libjava/include: jvm.h Log message: 2005-04-27 Andrew Haley [EMAIL PROTECTED] PR java/19285 * prims.cc (_Jv_ResolvePoolEntry): New function. * include/jvm.h (_Jv_Linker::find_field): New arg: found_class. * link.cc (_Jv_Linker::find_field): New arg: found_class. (resolve_pool_entry): Initialize the class in which a field is found. (link_symbol_table): Pass new arg to found_class. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libjava/ChangeLog.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.3391.2.59r2=1.3391.2.60 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libjava/prims.cc.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.104.2.3r2=1.104.2.4 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libjava/link.cc.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.10.2.2r2=1.10.2.3 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libjava/include/jvm.h.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.77.2.3r2=1.77.2.4 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19285
[Bug libffi/21285] gij fails to handle NullPointerException exception
--- Additional Comments From rth at gcc dot gnu dot org 2005-05-09 13:44 --- Fixed. -- What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED Target Milestone|--- |3.4.4 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21285
[Bug target/21412] ICE loading TLS address
--- Additional Comments From rth at gcc dot gnu dot org 2005-05-09 13:46 --- Powerpc fixed; ia64 still broken. -- What|Removed |Added Summary|[PowerPC] ICE loading TLS |ICE loading TLS address |address | http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21412
[Bug java/19285] Interfaces not initialized by static field access
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 13:48 --- Subject: Bug 19285 CVSROOT:/cvs/gcc Module name:gcc Branch: gcc-4_0-branch Changes by: [EMAIL PROTECTED] 2005-05-09 13:34:33 Modified files: libjava/java/lang: Class.h Log message: 2005-04-27 Andrew Haley [EMAIL PROTECTED] PR java/19285 * prims.cc (_Jv_ResolvePoolEntry): New function. * include/jvm.h (_Jv_Linker::find_field): New arg: found_class. * link.cc (_Jv_Linker::find_field): New arg: found_class. (resolve_pool_entry): Initialize the class in which a field is found. (link_symbol_table): Pass new arg to found_class. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libjava/java/lang/Class.h.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.78.2.1r2=1.78.2.2 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19285
[Bug libffi/21285] gij fails to handle NullPointerException exception
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 13:54 --- Subject: Bug 21285 CVSROOT:/cvs/gcc Module name:gcc Branch: gcc-4_0-branch Changes by: [EMAIL PROTECTED] 2005-05-09 13:39:35 Modified files: libffi : ChangeLog libffi/src/alpha: osf.S Log message: PR libffi/21285 * src/alpha/osf.S: Update unwind into to match code. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libffi/ChangeLog.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.222.2.1r2=1.222.2.2 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libffi/src/alpha/osf.S.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.9r2=1.9.72.1 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21285
[Bug libffi/21285] gij fails to handle NullPointerException exception
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 13:55 --- Subject: Bug 21285 CVSROOT:/cvs/gcc Module name:gcc Branch: gcc-3_4-branch Changes by: [EMAIL PROTECTED] 2005-05-09 13:40:16 Modified files: libffi : ChangeLog libffi/src/alpha: osf.S Log message: PR libffi/21285 * src/alpha/osf.S: Update unwind into to match code. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libffi/ChangeLog.diff?cvsroot=gcconly_with_tag=gcc-3_4-branchr1=1.153.4.12r2=1.153.4.13 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libffi/src/alpha/osf.S.diff?cvsroot=gcconly_with_tag=gcc-3_4-branchr1=1.9r2=1.9.10.1 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21285
[Bug tree-optimization/21407] [4.1 Regression] wrong code with upcast in C++
--- Additional Comments From dberlin at gcc dot gnu dot org 2005-05-09
13:58 ---
This is a call clobbering bug.
We don't pass the address of c to the call, we pass the address of some
substructure.
As a result, we don't think foo can touch c.b, when it can beause of the upcast.
Whee.
I'd imagine the following will fix it, though i haven't tested it yet
Index: tree-ssa-operands.c
===
RCS file: /cvs/gcc/gcc/gcc/tree-ssa-operands.c,v
retrieving revision 2.78
diff -u -p -r2.78 tree-ssa-operands.c
--- tree-ssa-operands.c 21 Apr 2005 18:05:26 - 2.78
+++ tree-ssa-operands.c 9 May 2005 13:57:44 -
@@ -1936,7 +1936,7 @@ note_addressable (tree var, stmt_ann_t s
Otherwise, we take the address of all the subvariables, plus the real
ones. */
- if (var TREE_CODE (var) == COMPONENT_REF
+ if (0 var TREE_CODE (var) == COMPONENT_REF
(ref = okay_component_ref_for_subvars (var, offset, size)))
{
subvar_t sv;
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21407
[Bug java/19285] Interfaces not initialized by static field access
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 14:00 --- Subject: Bug 19285 CVSROOT:/cvs/gcc Module name:gcc Branch: gcc-4_0-branch Changes by: [EMAIL PROTECTED] 2005-05-09 13:50:01 Modified files: gcc/java : ChangeLog java-tree.h expr.c builtins.c decl.c class.c constants.c Log message: 2005-04-28 Andrew Haley [EMAIL PROTECTED] PR java/19285 * java-tree.h (soft_resolvepoolentry_node): New. (alloc_constant_fieldref): Declare. * expr.c (expand_java_field_op): Don't call class_init for accesses to static fields with indirect dispatch. * builtins.c (initialize_builtins): Add __builtin_expect. * decl.c (soft_resolvepoolentry_node): New variable. (java_init_decl_processing): Create a decl for _Jv_ResolvePoolEntry. * class.c (build_fieldref_cache_entry): New function. * class.c (build_static_field_ref): New function. (build_static_field_ref): Rewrite for indirect dispatch. * constants.c (find_name_and_type_constant_tree): New function. (alloc_constant_fieldref): Likewise. (build_constants_constructor): Handle CONSTANT_Fieldref and CONSTANT_NameAndType. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/ChangeLog.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.1556.2.18r2=1.1556.2.19 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/java-tree.h.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.226r2=1.226.8.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/expr.c.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.218.4.3r2=1.218.4.4 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/builtins.c.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.29r2=1.29.4.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/decl.c.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.209.4.2r2=1.209.4.3 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/class.c.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.220.8.1r2=1.220.8.2 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/constants.c.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.40r2=1.40.34.1 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19285
[Bug c++/21467] ICE for template function if missing required default argument
--- Additional Comments From giovannibajo at libero dot it 2005-05-09
14:03 ---
The real bug is this accepts-invalid:
templatetypename F
void test(F function, bool arg0 = false, bool arg1)
{}
The ICE when you call the function is just a fallout of not having properly
validated the parameter declarations.
--
What|Removed |Added
Keywords||accepts-invalid
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21467
[Bug target/21397] [4.0/4.1 regression] arm: -march overrides -mtune
--- Additional Comments From rearnsha at gcc dot gnu dot org 2005-05-09 14:04 --- This is, of course, a regression from 3.4.x -- What|Removed |Added AssignedTo|unassigned at gcc dot gnu |rearnsha at gcc dot gnu dot |dot org |org Status|NEW |ASSIGNED Summary|arm: -march overrides -mtune|[4.0/4.1 regression] arm: - ||march overrides -mtune http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21397
[Bug target/21397] [4.0/4.1 regression] arm: -march overrides -mtune
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 14:20 --- Subject: Bug 21397 CVSROOT:/cvs/gcc Module name:gcc Changes by: [EMAIL PROTECTED] 2005-05-09 14:15:51 Modified files: gcc: ChangeLog gcc/config/arm : arm.c Log message: PR target/21397 * arm.c (ARM_OPT_SET_CPU, ARM_OPT_SET_ARCH, ARM_OPT_SET_TUNE): Define. (arm_override_options): Use them instead of manifest constants. Don't allow -march to override the -mtune setting. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gccr1=2.8670r2=2.8671 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/config/arm/arm.c.diff?cvsroot=gccr1=1.451r2=1.452 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21397
[Bug c++/21467] ICE for template function if missing required default argument
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 14:22 --- *** This bug has been marked as a duplicate of 16829 *** -- What|Removed |Added Status|NEW |RESOLVED Resolution||DUPLICATE http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21467
[Bug c++/16829] default parameter can be not one of the last in template function
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 14:22 --- *** Bug 21467 has been marked as a duplicate of this bug. *** -- What|Removed |Added CC||sylvain dot joyeux at m4x ||dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16829
[Bug libfortran/21468] New: vectorizing libfortran
Libgfortran doesn't vectorize very well at the moment. I added some vectorization options to the Makefile on ia64-unknown-linux-gnu, ran make under script and grepped for VECTORIZED. This is what I got: ../../../gcc-4.1-20050508/libgfortran/intrinsics/date_and_time.c:235: note: LOOP VECTORIZED. None of the matmul, dotprod, ... intrinsics were vectorized. This would probably require further improvements in gcc's vectorization capabilities, plus special-casing stride 1 in many cases. -- Summary: vectorizing libfortran Product: gcc Version: 4.1.0 Status: UNCONFIRMED Keywords: missed-optimization Severity: enhancement Priority: P2 Component: libfortran AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: tkoenig at gcc dot gnu dot org CC: gcc-bugs at gcc dot gnu dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21468
[Bug rtl-optimization/21469] New: nternal compiler error: in spill_failure, at reload1.c:1897
Compiling the following code with gcc -march=i386 -O2 gives and ICE.
-march=i686 doesn't have the same failure.
gcc version 4.1.0 20050507 (experimental)
3.3 and 3.4 both compile the code OK.
typedef unsigned char uint8_t;
typedef unsigned short uint16_t;
typedef union {
uint8_t _b[8];
uint16_t _w[2];
} MMXReg;
register MMXReg *env asm(ebp);
register long T0 asm(ebx);
register long T1 asm(esi);
register long T2 asm(edi);
void foo (int order) {
MMXReg r;
r._w[0] = env-_w[order];
*env = r;
}
t2.c:16: error: unable to find a register to spill in class 'GENERAL_REGS'
t2.c:16: error: this is the insn:
(insn:HI 14 28 16 0 (set (strict_low_part (subreg:HI (reg/v:DI 59 [ r ]) 0))
(mem/s:HI (plus:SI (mult:SI (reg/v:SI 60 [ order ])
(const_int 2 [0x2]))
(reg:SI 58 [ env.0 ])) [0 variable._w S2 A16])) 42
{*movstricthi_1} (insn_list:REG_DEP_TRUE 12 (insn_list:REG_DEP_TRUE 6
(insn_list:REG_DEP_TRUE 26 (nil
(expr_list:REG_DEAD (reg/v:SI 60 [ order ])
(nil)))
t2.c:16: internal compiler error: in spill_failure, at reload1.c:1897
--
Summary: nternal compiler error: in spill_failure, at
reload1.c:1897
Product: gcc
Version: 4.1.0
Status: UNCONFIRMED
Keywords: ice-on-valid-code
Severity: normal
Priority: P2
Component: rtl-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: pbrook at gcc dot gnu dot org
CC: gcc-bugs at gcc dot gnu dot org
GCC build triplet: i386-pc-linux-gnu
GCC host triplet: i386-pc-linux-gnu
GCC target triplet: i386-pc-linux-gnu
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21469
[Bug middle-end/21237] [4.0/4.1 Regression] unnecessary __cmpdi2 usage?
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 14:46 --- Subject: Bug 21237 CVSROOT:/cvs/gcc Module name:gcc Changes by: [EMAIL PROTECTED] 2005-05-09 14:45:59 Modified files: gcc: ChangeLog Log message: PR middle-end/21237 * dojump.c (do_jump, MINUS_EXPR): Build NE_EXPR and fall through to NE_EXPR case. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gccr1=2.8673r2=2.8674 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21237
[Bug fortran/21470] New: mark subroutine arguments __restrict__
Fortran subroutine arguments should be marked as __restrict__. Currently, the following code isn't vectorized: $ cat loopsum.f90 subroutine foo(a,b,c,n) real, dimension(n), intent(in) :: a,b real, dimension(n), intent(out) :: c do i=1,n c(i) = a(i)+b(i) end do end $ gfortran -O -S -ftree-vectorize -ftree-vectorizer-verbose=5 loopsum.f90 loopsum.f90:4: note: not vectorized: can't determine dependence between: (*a_18) [D.527_17] and (*c_25)[D.527_17] loopsum.f90:1: note: vectorized 0 loops in function. $ gfortran -v Using built-in specs. Target: ia64-unknown-linux-gnu Configured with: ../gcc-4.1-20050508/configure --prefix=/home/zfkts --enable- languages=c,f95 Thread model: posix gcc version 4.1.0 20050508 (experimental) -- Summary: mark subroutine arguments __restrict__ Product: gcc Version: 4.1.0 Status: UNCONFIRMED Keywords: missed-optimization Severity: normal Priority: P2 Component: fortran AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: tkoenig at gcc dot gnu dot org CC: gcc-bugs at gcc dot gnu dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21470
[Bug fortran/21470] mark subroutine arguments __restrict__
-- What|Removed |Added Severity|normal |enhancement http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21470
[Bug fortran/21471] New: 'POSITION = APPEND' doesn't seem to work
Hi! I am under the impression that the gfortran compiler doesn't (at least in some cases) handle well the APPEND file access, and rather overwrites the contents of the file... Can you help me with that? Thanks! Philippe PS: gfortran -vUsing built-in specs. Target: i686-pc-linux-gnu Configured with: /USER/philippe/Irix/Compilation/gcc/Sources/Release_4_0_0/gcc/configure --prefix=/usr1/MICRESS/Philippe/Tools/Gcc --with-mpfr=/usr1/MICRESS/Philippe/Tools/Mpfr --with-gmp=/usr1/MICRESS/Philippe/Tools/Gmp Thread model: posix gcc version 4.1.0 20050503 (experimental) PPS: the command I use to run this test: gfortran --static -o Hello_World Hello_World.f90 ; rm -f File ; touch File ; ./Hello_World ; cat File PPPS: a small example I put together: File should contains 13 times Hello_World, but in fact only has the last one !==! ! ** ! ** ! *T e s t * ! ** ! ** !==! ! !==! ! Declaration !==! ! PROGRAM Test ! INTEGER:: Counter ! CHARACTER ( LEN = 250 ), PARAMETER:: String = Hello world! ! !==! ! Begin !==! ! ! WRITE ( 6, (/A) ) Beginning of screen output DO Counter = 1, 13 OPEN ( UNIT = 13, FILE = File, FORM = FORMATTED, POSITION = APPEND, STATUS = OLD ) WRITE ( UNIT = 13, FMT = (A,I5) ) TRIM(String), Counter CLOSE ( UNIT = 13 ) WRITE ( 6, (A,I3) ) TRIM(String), Counter END DO WRITE ( 6, (A/) ) End of screen output ! ! !==! ! End !==! END PROGRAM Test -- Summary: 'POSITION = APPEND' doesn't seem to work Product: gcc Version: 4.1.0 Status: UNCONFIRMED Severity: normal Priority: P2 Component: fortran AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: P dot Schaffnit at access dot rwth-aachen dot de CC: gcc-bugs at gcc dot gnu dot org GCC target triplet: i686-pc-linux-gnu http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21471
[Bug rtl-optimization/13565] Register spill problem in x86_64 compiler
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 14:50 --- *** This bug has been marked as a duplicate of 9085 *** -- What|Removed |Added Status|NEW |RESOLVED Resolution||DUPLICATE http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13565
[Bug rtl-optimization/9085] Unable to find register to spill when optimizing
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 14:51 --- *** Bug 13565 has been marked as a duplicate of this bug. *** -- What|Removed |Added CC||marc at mrousseau dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9085
[Bug rtl-optimization/21469] nternal compiler error: in spill_failure, at reload1.c:1897
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 14:51 --- Why do people use global registers :(. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21469
[Bug target/21397] [4.0/4.1 regression] arm: -march overrides -mtune
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 14:53 --- Subject: Bug 21397 CVSROOT:/cvs/gcc Module name:gcc Branch: gcc-4_0-branch Changes by: [EMAIL PROTECTED] 2005-05-09 14:53:20 Modified files: gcc: ChangeLog gcc/config/arm : arm.c Log message: PR target/21397 * arm.c (ARM_OPT_SET_CPU, ARM_OPT_SET_ARCH, ARM_OPT_SET_TUNE): Define. (arm_override_options): Use them instead of manifest constants. Don't allow -march to override the -mtune setting. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=2.7592.2.232r2=2.7592.2.233 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/config/arm/arm.c.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.430.2.1r2=1.430.2.2 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21397
[Bug tree-optimization/21470] -fargument-noalias-global has no effect for the vectorizer
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 14:53 --- Actually -fargument-noalias-global is set to true for fortran and should be respected instead of marking them as restrict. -- What|Removed |Added Status|UNCONFIRMED |NEW Component|fortran |tree-optimization Ever Confirmed||1 Last reconfirmed|-00-00 00:00:00 |2005-05-09 14:53:51 date|| Summary|mark subroutine arguments |-fargument-noalias-global |__restrict__|has no effect for the ||vectorizer http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21470
[Bug target/21397] [4.0/4.1 regression] arm: -march overrides -mtune
--- Additional Comments From rearnsha at gcc dot gnu dot org 2005-05-09 15:04 --- fixed -- What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED Target Milestone|--- |4.0.1 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21397
[Bug c++/21308] [3.4/4.0/4.1 Regression] Very high compile time
--- Additional Comments From jakub at gcc dot gnu dot org 2005-05-09 15:26 --- Do we really have to call walk_subobject_offsets on every single array element? Couldn't we delay that to the time anybody would be looking at that info? Say extend the offsets splay tree, so that there would not only be INTEGER_CST key nodes representing particular offsets, but also nodes representing an offset interval, with value a TREE_VEC that could then be used to recreate that information (i.e. contain enough info to call walk_subobject_offsets in a loop with record_subobject_offset fn). Instead of calling walk_subobject_offsets here 1 21 times, we could e.g. call it on the first and last array element only and represent the middle of the array (if any) with an interval. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21308
[Bug tree-optimization/21470] -fargument-noalias-global has no effect for the vectorizer
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 15:54 --- If the aliasing information is correct on the VOPS, I would assume that the vectorizer would not have problems with it. Note there is another bug about this option, see PR 17064. -- What|Removed |Added CC||pinskia at gcc dot gnu dot ||org BugsThisDependsOn||17064 Keywords||alias http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21470
[Bug fortran/16939] Pointers not passed as subroutine arguments
--- Additional Comments From paulthomas2 at wanadoo dot fr 2005-05-09 16:06 --- That's right Tobi The patch that fixes this bug and regtests is: Index: trans-expr.c === RCS file: /cvsroot/gcc/gcc/gcc/fortran/trans-expr.c,v retrieving revision 1.42 diff -p -c -3 -r1.42 trans-expr.c *** trans-expr.c29 Apr 2005 15:31:33 - 1.42 --- trans-expr.c9 May 2005 14:41:23 - *** gfc_conv_variable (gfc_se * se, gfc_expr *** 360,365 --- 360,372 if (sym-attr.dummy sym-ts.type != BT_CHARACTER !sym-attr.dimension) + se-expr = gfc_build_indirect_ref (se-expr); + + /*Dereference dummy args that are character pointers. */ + if (sym-attr.dummy + sym-attr.pointer + sym-ts.type == BT_CHARACTER + !sym-attr.dimension) se-expr = gfc_build_indirect_ref (se-expr); /* Dereference pointer variables. */ I will do an update, bootstrap and regtest tonight before submitting this to the lists. It should be posted tonight. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16939
[Bug fortran/16939] Pointers not passed as subroutine arguments
--- Additional Comments From paulthomas2 at wanadoo dot fr 2005-05-09 16:08 --- Subject: Re: Pointers not passed as subroutine arguments The following fixes PR16939 and regtests. I'll do a full synchronisatio, bootstrap and regtest before posting it on the lists: Best regards Paul Index: trans-expr.c === RCS file: /cvsroot/gcc/gcc/gcc/fortran/trans-expr.c,v retrieving revision 1.42 diff -p -c -3 -r1.42 trans-expr.c *** trans-expr.c29 Apr 2005 15:31:33 - 1.42 --- trans-expr.c9 May 2005 14:41:23 - *** gfc_conv_variable (gfc_se * se, gfc_expr *** 360,365 --- 360,372 if (sym-attr.dummy sym-ts.type != BT_CHARACTER !sym-attr.dimension) + se-expr = gfc_build_indirect_ref (se-expr); + + /*Dereference dummy args that are character pointers. */ + if (sym-attr.dummy + sym-attr.pointer + sym-ts.type == BT_CHARACTER + !sym-attr.dimension) se-expr = gfc_build_indirect_ref (se-expr); /* Dereference pointer variables. */ -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16939
[Bug target/21472] New: Gcc generates CALL16 reloc that ld objects to
Compiling lib/csu/common_elf/crtbegin.c in the NetBSD 1.6.2 source tree with gcc 4.0.0 and binutils 2.16, I get complaints from ld: crtbeginS.o: CALL16 reloc at 0xfc not against global symbol The call16 is in the .s file generated by gcc, but I can't tell why it's any different from the others. Will attach .i and .s files (it would be nice if you could do that when entering a bug...) -- Summary: Gcc generates CALL16 reloc that ld objects to Product: gcc Version: 4.0.0 Status: UNCONFIRMED Severity: normal Priority: P2 Component: target AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: pkoning at equallogic dot com CC: gcc-bugs at gcc dot gnu dot org GCC build triplet: i386-netbsdelf GCC host triplet: i386-netbsdelf GCC target triplet: mipsel-netbsdelf http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21472
[Bug target/21472] Gcc generates CALL16 reloc that ld objects to
--- Additional Comments From pkoning at equallogic dot com 2005-05-09 16:30 --- Created an attachment (id=8845) -- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=8845action=view) Preprocessed source file of the code that shows the problem -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21472
[Bug target/21472] Gcc generates CALL16 reloc that ld objects to
--- Additional Comments From pkoning at equallogic dot com 2005-05-09 16:31 --- Created an attachment (id=8846) -- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=8846action=view) Assembly file generated by gcc 4.0.0 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21472
warning and _Complex_I
I have not found in the bug list of the GCC Bugzilla something about an invalid warning: warning: imaginary constants are a GCC extension when _Complex_I is used with -std=c99 -pedantic. Did I missed something? Laurent.
[Bug target/14981] [3.4 Regression] ICE in _mm_xor_pd for SSE2 with -O1
--- Additional Comments From jakub at gcc dot gnu dot org 2005-05-09 16:50
---
This patch creates regressions on i?86:
FAIL: gcc.dg/20020523-1.c (test for excess errors)
WARNING: gcc.dg/20020523-1.c compilation failed to produce executable
FAIL: gcc.dg/20020523-2.c (test for excess errors)
WARNING: gcc.dg/20020523-2.c compilation failed to produce executable
./xgcc -B ./ -v -m32 -march=pentium3 -msse -ffast-math -O2 20020523-1.c
Reading specs from ./specs
Configured with: ../configure --enable-languages=c,c++
Thread model: posix
gcc version 3.4.4 20050509 (prerelease)
./cc1 -quiet -v -iprefix
/usr/src/gcc-3.4/obj/gcc/../lib/gcc/x86_64-unknown-linux-gnu/3.4.4/ -isystem
./include 20020523-1.c -quiet -dumpbase 20020523-1.c -m32 -march=pentium3 -msse
-auxbase 20020523-1 -O2 -version -ffast-math -o /tmp/ccwFAGHO.s
ignoring nonexistent directory
/usr/src/gcc-3.4/obj/gcc/../lib/gcc/x86_64-unknown-linux-gnu/3.4.4/include
ignoring nonexistent directory
/usr/src/gcc-3.4/obj/gcc/../lib/gcc/x86_64-unknown-linux-gnu/3.4.4/../../../../x86_64-unknown-linux-gnu/include
ignoring nonexistent directory NONE/include
ignoring nonexistent directory
/usr/local/lib/gcc/x86_64-unknown-linux-gnu/3.4.4/include
ignoring nonexistent directory
/usr/local/lib/../x86_64-unknown-linux-gnu/include
#include ... search starts here:
#include ... search starts here:
./include
/usr/local/include
/usr/include
End of search list.
GNU C version 3.4.4 20050509 (prerelease) (x86_64-unknown-linux-gnu)
compiled by GNU C version 3.4.3 20050227 (Red Hat 3.4.3-22.fc3).
GGC heuristics: --param ggc-min-expand=98 --param ggc-min-heapsize=128053
20020523-1.c: In function `main':
20020523-1.c:65: error: could not split insn
(insn:TI 108 123 112 (set (reg:SF 22 xmm1)
(if_then_else:SF (gt (reg:SF 22 xmm1)
(mem/u/f:SF (symbol_ref/u:SI (*.LC2) [flags 0x2]) [2 S4 A32]))
(reg:SF 21 xmm0 [76])
(const_double:SF 0.0 [0x0.0p+0]))) 663 {*sse_movsfcc_const0_1}
(insn_list 123 (insn_list 122 (insn_list:REG_DEP_ANTI 106 (nil
(expr_list:REG_DEAD (reg:SF 21 xmm0 [76])
(expr_list:REG_EQUIV (mem/f:SF (reg/f:SI 7 sp) [0 S4 A32])
(nil
20020523-1.c:65: internal compiler error: in final_scan_insn, at final.c:2429
Please submit a full bug report,
with preprocessed source if appropriate.
See URL:http://gcc.gnu.org/bugs.html for instructions.
Reverting this patch fixes it.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14981
[Bug c/21473] New: Number Representation - Octal
This bug was reported back in 1998/07/19 by Alexander Favorov msg00504.html . It is well described by him. The bug still exists as of gcc 3.3.3 . Its a simple parsing error. My understanding of C is: '\10' is ten (decimal) '\010' is eight (octal) '\0x10' is sixteen (hexadecimal) For a number to be octal, it must be preceeded by a zero, otherwise it is decimal. Book Reference: C: The Complete Reference 3rd Edition Author: Schildt, Herbert Page 37. -- Summary: Number Representation - Octal Product: gcc Version: 3.3.3 Status: UNCONFIRMED Severity: normal Priority: P2 Component: c AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: robert dot amodeo at ntlworld dot com CC: gcc-bugs at gcc dot gnu dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21473
[Bug c/21473] Number Representation - Octal
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 17:07 --- Not in charactors. '/xx' is octal aways. -- What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution||INVALID http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21473
[Bug target/21472] Gcc generates CALL16 reloc that ld objects to
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 17:15 --- lw $25,%call16(fini_fallthru)($28) I want to say the above is the problem and it is a local symbol but to GCC it is a global one. -- What|Removed |Added Keywords||build http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21472
[Bug libfortran/21471] 'POSITION = APPEND' doesn't seem to work
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 17:21 --- It is a bug in the library as it is being passed correctly to it: _gfortran_ioparm.position = APPEND; _gfortran_ioparm.position_len = 6; -- What|Removed |Added Component|fortran |libfortran http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21471
[Bug c/21474] New: missed optimizations when comparing address to NULL
The program below gives some expressions which gcc could, but does not, evaluate
to true. E.g. gcc considers p-a true, even when a is the first element
of the struct (is this a bug?), but does not consider p-b[3] to be true.
struct foo {int a, b[10];};
int subr(int i, struct foo *p)
{
int x[10];
#if 0
// gcc folds this
if (p-a) return 1;
#else
// but not these
if (p-b[3]) return 1;
if (x[3] != 0) return 1;
if (x[i] != 0) return 1; // not sure if this one is safe to fold
#endif
return 0;
}
--
Summary: missed optimizations when comparing address to NULL
Product: gcc
Version: 4.1.0
Status: UNCONFIRMED
Severity: enhancement
Priority: P2
Component: c
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: trt at acm dot org
CC: gcc-bugs at gcc dot gnu dot org
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21474
[Bug middle-end/21474] missed optimizations when comparing address to NULL
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 17:26 --- Actually I don't think it is safe to fold any of these. -- What|Removed |Added Component|c |middle-end http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21474
[Bug target/20695] sh64-*-* port deos not handle 32 / 64 bit conversions properly
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 17:44 --- Subject: Bug 20695 CVSROOT:/cvs/gcc Module name:gcc Changes by: [EMAIL PROTECTED] 2005-05-09 17:42:56 Modified files: gcc/doc: invoke.texi gcc/config/sh : crt1.asm elf.h lib1funcs.asm libgcc-excl.ver linux.h netbsd-elf.h sh-modes.def sh-protos.h sh.c sh.h sh.md shmedia.md sshmedia.h t-linux t-sh64 ushmedia.h gcc/testsuite : ChangeLog gcc/testsuite/gcc.dg: 20021029-1.c pr15784-3.c Added files: gcc/config/sh : divtab.c newlib.h predicates.md superh.h superh64.h t-superh Log message: gcc: 2005-05-09 Jorn Rennecke [EMAIL PROTECTED] * config/sh/sh.h (OVERRIDE_OPTIONS): Don't set flag_finite_math_only if flag_signaling_nans is set. For TARGET_SH2E, if flag_finite_math_only is not set, set IEEE_BIT. * doc/invoke.texi (SH -mieee): Document relation to -ffinite-math-only. 2005-05-06 Jorn Rennecke [EMAIL PROTECTED] Merge of sh-elf specific patches from sh-elf-4_1-branch: 2005-05-05 Kaz Kojima [EMAIL PROTECTED] * config/sh/sh.h (ASM_OUTPUT_REG_PUSH): Provide SHMEDIA version. (ASM_OUTPUT_REG_POP): Likewise. 2005-05-05 Jorn Rennecke [EMAIL PROTECTED] Kaz Kojima [EMAIL PROTECTED] * config/sh/sh.c (sh_builtin_saveregs): Use copy_to_mode_reg and plus_constant. 2005-05-04 Kaz Kojima [EMAIL PROTECTED] * config/sh/sh.c (sh_div_strategy): Initialize with SH_DIV_STRATEGY_DEFAULT. * config/sh/sh.c (SH_DIV_STR_FOR_SIZE): Define. (SH_DIV_STRATEGY_DEFAULT): Likewise. (OPTIMIZATION_OPTIONS): Set sh_div_str to SH_DIV_STR_FOR_SIZE when optimized for size. * config/sh/linux.h (SH_DIV_STRATEGY_DEFAULT): Redefine. (SH_DIV_STR_FOR_SIZE): Likewise. * config/sh/netbsd-elf.h (SH_DIV_STRATEGY_DEFAULT): Likewise. (SH_DIV_STR_FOR_SIZE): Likewise. 2005-05-04 Jorn Rennecke [EMAIL PROTECTED] * config/sh/sh-modes.def (PDImode): Add. * config/sh/sh-protos.h (shmedia_prepare_call_address): Declare. * config/sh/sh.c (print_operand): Handle IF_THEN_ELSE. (target_reg_operand): Allow PDImode. (sh_register_move_cost): If neither sh_gettrcost_str nor TARGET_PT_FIXED is set, assume gettr costs 100. (shmedia_prepare_call_address): New function. (sh_gettrcost_str): Initialize to empty string. (sh_divsi3_libfunc): New variable. * config/sh/sh.h (PT_FIXED_BIT, TARGET_INVALID_SYMBOLS): Define. (TARGET_SWITCH_SH5_32_ANY_EXTRA): Likewise. (TARGET_SWITCH_SH5_MEDIA_ANY_EXTRA): Likewise. (TARGET_SWITCHES): Use TARGET_SWITCH_SH5_32_ANY_EXTRA and TARGET_SWITCH_SH5_MEDIA_ANY_EXTRA. (TARGET_OPTIONS): Add -mdivsi3_libfunc. (OVERRIDE_OPTIONS): Set sh_divsi3_libfunc if it hasn't been set by the user. Also set flag_no_function_cse for (TARGET_SHMEDIA !TARGET_PT_FIXED). (HARD_REGNO_MODE_OK): Allow TARGET_REGS in PDImode. (CONSTRAINT_LEN): Remove debug version. (SECONDARY_INOUT_RELOAD_CLASS:) Break out of (SECONDARY_OUTPUT_RELOAD_CLASS). Use EXTRA_CONSTRAINT_Csy for check if a target register needs a secondary reload through GENERAL_REGS. (SECONDARY_INPUT_RELOAD_CLASS): Use SECONDARY_INOUT_RELOAD_CLASS. (sh_divsi3_libfunc): Declare. (FUNCTION_PROFILER): Provide SHMEDIA version. * config/sh/predicates.md: New file. * config/sh/sh.md (predicates.md): Include. (divsi_inv_call_combine, divsi3): Use sh_divsi3_libfunc. (reload_insi): Fix predicates and constraints. (ptabs): New expander. (*extendsipdi_media, *truncdipdi_media): New insns. (call, call_value, sibcall): Use shmedia_prepare_call_address. * doc/invoke.texi (-multcost, -mdiv): Document new SH options. (-mdivsi3_libfunc, -madjust-unroll, -mindexed-addressing): Likewise. (-mgettrcost, -mpt-fixed, -minvalid-symbols): Likewise. 2005-04-11 Jorn Rennecke [EMAIL PROTECTED] * sh.c (print_operand): Remove sh_rep_vec extraction. (sh_output_mi_thunk): Make i unsigned. * sh.c (TARGET_ADJUST_UNROLL_MAX): Only redefine if already defined. (sh_adjust_unroll_max): Only define if TARGET_ADJUST_UNROLL_MAX is defined. Update label detection code and iteration lookup, enable basic functionality, but without IV analysis. 2005-04-11 Jorn Rennecke [EMAIL PROTECTED] * sh.h (OPTIMIZATION_OPTIONS): Don't make setting of
[Bug rtl-optimization/21461] [3.4/4.0 regression] miscompilation at -O2 in 3.4.x, 4.x
--- Additional Comments From belyshev at depni dot sinp dot msu dot ru 2005-05-09 17:45 --- This code: memory.h:413: *(POINTER *)Freepointer = memory_ARRAY[Size]-free; memory_ARRAY[Size]-free = Freepointer; is invalid, because you are changing object of some type via pointer of type (void **). Either it should be rewritten using unions, or -fno-strict-aliasing option should be used. More on this here: http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fstrict_002daliasing-524 -- What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution||INVALID http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21461
[Bug rtl-optimization/21138] wrong code in sixtrack for -fmodulo-sched
--- Additional Comments From janis at gcc dot gnu dot org 2005-05-09 19:32 --- sixtrack works fine for me with -m64 -O2 with both SPEC's test input and its ref output, for all of my daily builds since 20050503, while with the addition of -fmodulo-sched it gets miscompares. lucas is now passing again, but apsi is still gettting miscompares with -m64 -O2 -fmodulo-sched. Every week or so I run SPEC CPU2000 with GCC mainline built the previous night and run the tests using test input, for each of about 40 sets of GCC command line options with -m32 and then with -m64. This has found several regressions (21272, 21155, 21054, 21048, 21030, 20813 in the last month). Almost all of the failures I see are ICEs; the ones reported here with -fmodulo-sched are the only runtime failures I've seen for a long time. I'm running the tests now with the 20050508 compiler (last night's build failed) and will let you know if failures with other options look related to this one. The only known problem right now is PR 21155, intermittent ICEs with -ftree-vectorize. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21138
[Bug middle-end/21460] Internal compiler error in calc_dfs_tree on valid code
-- What|Removed |Added Component|tree-optimization |middle-end Keywords||ice-on-valid-code http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21460
[Bug libstdc++/21238] conflicting typedef __numpunct_cache_CharT __cache_type; in std::numpunctCharT makes specialization impossible
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 19:59 --- Subject: Bug 21238 CVSROOT:/cvs/gcc Module name:gcc Branch: gcc-4_0-branch Changes by: [EMAIL PROTECTED] 2005-05-09 19:59:17 Modified files: libstdc++-v3 : ChangeLog libstdc++-v3/include/bits: locale_facets.tcc libstdc++-v3/testsuite: testsuite_character.h testsuite_character.cc Added files: libstdc++-v3/testsuite/22_locale/num_get: 3.cc libstdc++-v3/testsuite/22_locale/num_put: 3.cc libstdc++-v3/testsuite/22_locale/money_get: 3.cc libstdc++-v3/testsuite/22_locale/money_put: 3.cc Log message: 2005-05-09 Paolo Carlini [EMAIL PROTECTED] PR libstdc++/21238 * include/bits/locale_facets.tcc (num_get::_M_extract_float, num_get::_M_extract_int, num_get::do_get(bool), num_put::_M_insert_int, num_put::_M_insert_float, num_put::do_put(bool), money_get::_M_extract, money_put::_M_insert): Adjust the __cache_type typedef not to forward to a numpunct/moneypunct typedef. * testsuite/testsuite_character.h: Add pod_uint and its numpunct and moneypunct specializations. * testsuite/testsuite_character.cc: Add numpunctpod_uint::id and moneypunctpod_uint::id. * testsuite/22_locale/num_get/3.cc: New. * testsuite/22_locale/num_put/3.cc: Likewise. * testsuite/22_locale/money_get/3.cc: Likewise. * testsuite/22_locale/money_put/3.cc: Likewise. * include/bits/locale_facets.tcc (money_put::_M_insert, time_get::_M_extract_name): Prefer operator== to operator!= on char_types. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/ChangeLog.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.2917.2.43r2=1.2917.2.44 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/include/bits/locale_facets.tcc.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.211.12.3r2=1.211.12.4 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/testsuite_character.h.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.2.54.2r2=1.2.54.3 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/testsuite_character.cc.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=1.3.4.2r2=1.3.4.3 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/22_locale/num_get/3.cc.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=NONEr2=1.1.4.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/22_locale/num_put/3.cc.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=NONEr2=1.1.4.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/22_locale/money_get/3.cc.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=NONEr2=1.1.4.1 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libstdc++-v3/testsuite/22_locale/money_put/3.cc.diff?cvsroot=gcconly_with_tag=gcc-4_0-branchr1=NONEr2=1.1.4.1 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21238
[Bug libstdc++/21238] conflicting typedef __numpunct_cache_CharT __cache_type; in std::numpunctCharT makes specialization impossible
--- Additional Comments From pcarlini at suse dot de 2005-05-09 20:01 --- Fixed for 4.0.1. -- What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED Target Milestone|--- |4.0.1 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21238
[Bug rtl-optimization/21469] nternal compiler error: in spill_failure, at reload1.c:1897
--- Additional Comments From pbrook at gcc dot gnu dot org 2005-05-09
20:04 ---
Annother testcase. This one fails at -O2 with gcc 4.1.0 20050507 and 4.0.1
20050430.
This is a regression from 3.4.
typedef unsigned long long uint64_t;
register int *env asm (ebp);
register long T0 asm (ebx);
register long T1 asm (esi);
register long T2 asm (edi);
void foo(void);
void
bar (long physaddr, uint64_t v)
{
if (*env)
{
foo ();
*(uint64_t *) (physaddr + *env) = v;
}
}
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21469
[Bug target/18655] Incorrect data in .debug_frame section for PowerPC
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-05-09 20:20 --- Subject: Bug 18655 CVSROOT:/cvs/gcc Module name:gcc Changes by: [EMAIL PROTECTED] 2005-05-09 20:20:26 Modified files: gcc: ChangeLog dwarf2out.c Log message: PR 18655 * dwarf2out.c (output_call_frame_info): Use DWARF2_FRAME_REG_OUT before outputting DWARF_FRAME_RETURN_COLUMN. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gccr1=2.8692r2=2.8693 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/dwarf2out.c.diff?cvsroot=gccr1=1.592r2=1.593 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18655
[Bug target/21351] internal compiler error with sse
--- Additional Comments From js at jeannot dot org 2005-05-09 20:21 --- The bug is still present gcc version 3.4.4 20050506. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21351
[Bug rtl-optimization/21469] nternal compiler error: in spill_failure, at reload1.c:1897
--- Additional Comments From belyshev at depni dot sinp dot msu dot ru 2005-05-09 20:28 --- *** This bug has been marked as a duplicate of 16185 *** -- What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution||DUPLICATE http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21469
[Bug target/16185] ICE: in spill_failure, at reload1.c:1892, global registers and long long
--- Additional Comments From belyshev at depni dot sinp dot msu dot ru 2005-05-09 20:28 --- *** Bug 21469 has been marked as a duplicate of this bug. *** -- What|Removed |Added CC||pbrook at gcc dot gnu dot ||org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16185
[Bug rtl-optimization/21469] nternal compiler error: in spill_failure, at reload1.c:1897
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-05-09 20:30 --- Hmm, I think only the second example is a dup. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21469
