--- Comment #5 from dannysmith at users dot sourceforge dot net 2007-10-05
07:03 ---
(In reply to comment #4)
>
> However, for some reasons, the path to ld is detected as:
>
> c:/Docume~1/gdr/Desktop/sandbox/eval-build/i686-pc-mingw32/libstdc++-v3/c:/mingw/bin/../lib/gcc/mingw32/3.4.5
--- Comment #6 from fxcoudert at gcc dot gnu dot org 2007-10-05 07:52
---
(In reply to comment #5)
> Ah yes, this is the old "moves of cost 2 don't need reloads" thing.
> Could you try the attached patch?
Bootstrap completed (C + Fortran) with this patch applied. Thanks.
--
fxcoude
--- Comment #7 from rsandifo at gcc dot gnu dot org 2007-10-05 08:34
---
Subject: Bug 33635
Author: rsandifo
Date: Fri Oct 5 08:34:01 2007
New Revision: 129031
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=129031
Log:
gcc/
PR target/33635
* config/mips/mips.c
--- Comment #8 from rsandifo at gcc dot gnu dot org 2007-10-05 08:42
---
Thanks for testing. I included some more general cleanups in the
patch and committed to mainline.
--
rsandifo at gcc dot gnu dot org changed:
What|Removed |Added
---
Experienced with 4.1.1, 4.1.2 and 4.2.1. Below is the minimum test case that I
can narrow it down to while being reproducible across these versions.
$ g++ -O3 -Wall test.cpp -o test
test.cpp: In function int main():
test.cpp:19: warning: this.11 is used uninitialized in this function
test.cpp
--- Comment #7 from manfred99 at gmx dot ch 2007-10-05 09:36 ---
Ok, I managed to produce a testcase: a wrapper around the miscompiled
function and all the missing routines. Input data is read from a binary file,
so the program has to be run on a x86 machine (littleendian).
I attach a t
--- Comment #8 from manfred99 at gmx dot ch 2007-10-05 09:39 ---
Created an attachment (id=14300)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14300&action=view)
test case
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33638
--- Comment #5 from manfred99 at gmx dot ch 2007-10-05 09:37 ---
Created an attachment (id=14299)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14299&action=view)
test case
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32382
--- Comment #6 from manfred99 at gmx dot ch 2007-10-05 09:38 ---
(From update of attachment 14299)
sorry, for wrong bug
--
manfred99 at gmx dot ch changed:
What|Removed |Added
gfortran crashes on the compile of the following (invalid) Fortran module.
There are two obvious faults with the program:
1. there is no intent
2. there is no dimension to N
--
Summary: crash on invalid program
Product: gcc
Version: 4.3.0
Status: UNC
--- Comment #1 from holst at matmech dot com 2007-10-05 09:51 ---
Created an attachment (id=14301)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14301&action=view)
invalid program which crashes latest trunk (gcc (GCC) 4.3.0 20070927
(experimental) [trunk revision 128829])
test.f:0
--- Comment #2 from gerald at pfeifer dot com 2007-10-05 10:30 ---
I don't remember getting an answer to the following, but let's close this.
From: Gerald Pfeifer <[EMAIL PROTECTED]>
To: Jan Hubicka <[EMAIL PROTECTED]>
Cc: Andrew Pinski <[EMAIL PROTECTED]>, <[EMAIL PROTECTED]>
Date: Su
--
zadeck at naturalbridge dot com changed:
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |zadeck at naturalbridge dot
|dot org
--- Comment #1 from bernhard at ratte dot dhs dot org 2007-10-05 11:40
---
Created an attachment (id=14303)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14303&action=view)
stn.cc the file that crashes gcc
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33668
every time I compile stn.cc gcc crashes with the following message:
g++ -march=prescott -O0 -pipe -pg -ggdb3 -Wall -c -o stn.o stn.cc
stn.cc:1129: internal compiler error: in reference_to_unused, at
dwarf2out.c:10010
Please submit a full bug report,
with preprocessed source if appropriate.
See ht
--- Comment #9 from ubizjak at gmail dot com 2007-10-05 10:47 ---
> Hope this helps.
Sure, I've got the problem. The problem is actually in RTL optimization, where
dse1 pass removes wrong insn.
Suprisingly, the problem is in line 61 of comunpack.f:
--> bscale = 2.0**real(idrstmpl(2))
--- Comment #3 from rguenth at gcc dot gnu dot org 2007-10-05 12:14 ---
*** This bug has been marked as a duplicate of 31899 ***
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
--
--- Comment #2 from fxcoudert at gcc dot gnu dot org 2007-10-05 11:15
---
Confirmed. Reduced testcase:
$ cat a.f90
module test
contains
subroutine func_1(u,n)
integer :: n, u(n(1))
end subroutine
end module test
$ gfortran a.f90
f951: internal compiler error: Segmentation fault
--- Comment #1 from rguenth at gcc dot gnu dot org 2007-10-05 12:00 ---
I have a patch.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
AssignedTo|
--- Comment #1 from jakub at gcc dot gnu dot org 2007-10-05 10:54 ---
Isn't this a dup of PR33454 ?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33657
--- Comment #2 from rguenth at gcc dot gnu dot org 2007-10-05 10:55 ---
Err, yes.
*** This bug has been marked as a duplicate of 33454 ***
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
-
--- Comment #2 from rguenth at gcc dot gnu dot org 2007-10-05 10:55 ---
*** Bug 33657 has been marked as a duplicate of this bug. ***
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33454
--- Comment #11 from rguenth at gcc dot gnu dot org 2007-10-05 12:36
---
But powf is pure/const, so the call is not a use.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33638
The following fails with 32bit pointers and 64bit long long if the compiler is
built with type-checking enabled.
void __lock_get_list(void *dp)
{
if (((unsigned int)dp + 1) & ~1ULL)
;
}
The original trees look ok:
if (((long long unsigned int) (dp + 1) & 0x0fffe) != 0)
{
--- Comment #1 from rguenth at gcc dot gnu dot org 2007-10-05 10:15 ---
Can you attach preprocessed source if this still happens?
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
---
gcc -DHAVE_CONFIG_H -I. -I.. -DMAGIC=\"/etc/magic:/usr/share/misc/magic\"
-march=i586 -mtune=i686 -fmessage-length=0 -D_FORTIFY_SOURCE=2 -O2
-D_FILE_OFFSET_BITS=64 -DHOWMANY=69632 -D_LARGEFILE_SOURCE -D_GNU_SOURCE -fPIC
-MT softmagic.lo -MD -MP -MF .deps/softmagic.Tpo -c softmagic.c -fPIC -DPIC -o
--- Comment #1 from rguenth at gcc dot gnu dot org 2007-10-05 10:17 ---
Created an attachment (id=14302)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14302&action=view)
testcase (unreduced)
Reducing.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33667
--- Comment #2 from rguenth at gcc dot gnu dot org 2007-10-05 10:56 ---
Reduced testcase, fails with -O:
typedef unsigned int size_t;
typedef unsigned char uint8_t;
typedef unsigned short int uint16_t;
typedef unsigned long long int uint64_t;
struct magic {
uint8_t mask_op;
union {
--- Comment #4 from jakub at gcc dot gnu dot org 2007-10-05 10:58 ---
If __builtin_expect is the only problem, then we could certainly allow
__builtin_expect to be replaceable (perhaps add a few other builtin calls
that never result in any actual calls, like __builtin_object_size).
--
--- Comment #13 from rguenth at gcc dot gnu dot org 2007-10-05 12:14
---
*** Bug 33668 has been marked as a duplicate of this bug. ***
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
-
--
fxcoudert at gcc dot gnu dot org changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
Ke
--- Comment #10 from ubizjak at gmail dot com 2007-10-05 11:40 ---
(In reply to comment #9)
> This is wrong. dse1 pass should _not_ remove (insn 1312). (insn 1322) is
> storing to the same address as (insn 1312), but we have a call to powf in
> between, so new store insn doesn't kill pr
--- Comment #2 from bernhard at ratte dot dhs dot org 2007-10-05 11:41
---
Created an attachment (id=14304)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14304&action=view)
the precompiled version of stn.cc (gzipped)
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33668
--
fxcoudert at gcc dot gnu dot org changed:
What|Removed |Added
CC||fxcoudert at gcc dot gnu dot
|
--
fxcoudert at gcc dot gnu dot org changed:
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |fxcoudert at gcc dot gnu dot
|dot org
--- Comment #12 from zadeck at naturalbridge dot com 2007-10-05 13:02
---
Subject: Re: [4.3 regression]: wrong code with
-fforce-addr
rguenth at gcc dot gnu dot org wrote:
> --- Comment #11 from rguenth at gcc dot gnu dot org 2007-10-05 12:36
> ---
> But powf is pure/const,
--
fxcoudert at gcc dot gnu dot org changed:
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |fxcoudert at gcc dot gnu dot
|dot org
--- Comment #3 from matz at gcc dot gnu dot org 2007-10-05 14:36 ---
It's lower-subreg.
--
matz at gcc dot gnu dot org changed:
What|Removed |Added
Status|UNC
On Linux/ia32, revision 128957 miscompiles 481.wrf in SPEC CPU 2006:
Compiling Binaries
Building 481.wrf test base o2 default: (build_base_o2.)
Build successes: 481.wrf(base)
Parsing Flags
Looking at 481.wrf base o2 default: done
Flag Parsing Complete
Setting Up Run Directories
Settin
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
CC||rguenth at gcc dot gnu dot
|
--- Comment #4 from matz at gcc dot gnu dot org 2007-10-05 15:06 ---
Mine. Have patch.
--
matz at gcc dot gnu dot org changed:
What|Removed |Added
AssignedTo|unas
--
aoliva at gcc dot gnu dot org changed:
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |aoliva at gcc dot gnu dot
|dot org
--- Comment #6 from matz at gcc dot gnu dot org 2007-10-05 15:36 ---
Subject: Bug 33600
Author: matz
Date: Fri Oct 5 15:36:16 2007
New Revision: 129035
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=129035
Log:
gcc/
PR inline-asm/33600
* function.c (match_asm_constraints_1):
--- Comment #7 from matz at gcc dot gnu dot org 2007-10-05 15:38 ---
Fixed now.
--
matz at gcc dot gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--- Comment #10 from tobi at gcc dot gnu dot org 2007-10-05 15:55 ---
I'm down to 14 differences. Looking into the remaining ones.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32147
--- Comment #2 from rguenth at gcc dot gnu dot org 2007-10-05 16:33 ---
Subject: Bug 33666
Author: rguenth
Date: Fri Oct 5 16:33:25 2007
New Revision: 129036
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=129036
Log:
2007-10-05 Richard Guenther <[EMAIL PROTECTED]>
PR
--- Comment #3 from rguenth at gcc dot gnu dot org 2007-10-05 16:33 ---
Fixed.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--- Comment #5 from pinskia at gmail dot com 2007-10-05 17:05 ---
Subject: Re: [4.1/4.2/4.3 Regression] TER breaks some inline-asm code (again)
On 5 Oct 2007 10:58:10 -, jakub at gcc dot gnu dot org
<[EMAIL PROTECTED]> wrote:
>
>
> --- Comment #4 from jakub at gcc dot gnu dot o
--- Comment #11 from tobi at gcc dot gnu dot org 2007-10-05 17:33 ---
Created an attachment (id=14305)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14305&action=view)
A patch
Wolfgang, this patch works for your testcase, but it is somewhat dependent on
the history of memory alloc
--- Comment #21 from fxcoudert at gcc dot gnu dot org 2007-10-05 17:41
---
Yet another one:
print *, transfer(achar([0]), 0_1)
end
Reducing this testcase has opened Pandora's box, I'll try to fix them one after
another.
--
fxcoudert at gcc dot gnu dot org changed:
--
aoliva at gcc dot gnu dot org changed:
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |aoliva at gcc dot gnu dot
|dot org
--- Comment #7 from aoliva at gcc dot gnu dot org 2007-10-05 17:44 ---
Possible fix at http://gcc.gnu.org/ml/gcc-patches/2007-10/msg00302.html
--
aoliva at gcc dot gnu dot org changed:
What|Removed |Added
---
--- Comment #3 from aoliva at gcc dot gnu dot org 2007-10-05 17:45 ---
Testing patch at http://gcc.gnu.org/ml/gcc-patches/2007-10/msg00298.html
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33655
--- Comment #13 from kargl at gcc dot gnu dot org 2007-10-05 17:50 ---
(In reply to comment #9)
> > Hope this helps.
> Sure, I've got the problem. The problem is actually in RTL optimization, where
> dse1 pass removes wrong insn.
>
> Suprisingly, the problem is in line 61 of comunpack.f
--- Comment #5 from matz at gcc dot gnu dot org 2007-10-05 17:57 ---
Subject: Bug 33667
Author: matz
Date: Fri Oct 5 17:56:52 2007
New Revision: 129038
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=129038
Log:
gcc/
PR middle-end/33667
* lower-subreg.c (decompose_multiword_su
--- Comment #6 from matz at gcc dot gnu dot org 2007-10-05 18:03 ---
Fixed now.
--
matz at gcc dot gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--- Comment #8 from aoliva at gcc dot gnu dot org 2007-10-05 18:17 ---
Actually, Dave tells me in one of the "Ignored" entries that he posted by
e-mail that the earlier patch at
http://gcc.gnu.org/ml/gcc-patches/2007-10/msg00186.html had already fixed the
bootstrap problem, but that he s
--- Comment #9 from dave at hiauly1 dot hia dot nrc dot ca 2007-10-05
19:02 ---
Subject: Re: bootstrap with ada failed
> --- Comment #8 from aoliva at gcc dot gnu dot org 2007-10-05 18:17
> ---
> Actually, Dave tells me in one of the "Ignored" entries that he posted by
> e-m
--- Comment #11 from tobi at gcc dot gnu dot org 2007-10-05 19:51 ---
Wonderful, I just had a look at this bug. Implementing transformational
functions in the compiler will be a lot of joy. See gfc_simplify_reshape for
the delights this brings.
--
tobi at gcc dot gnu dot org change
Benchmark twolf in SPEC CPU2000 fails to compile on powerpc64-linux with -O2
-fsched-stalled-insns=[01234] (didn't try other values). The following
testcase also fails for powerpc*-*-* and ia64-linux cross cc1:
typedef struct binbox { int p; int n; } B;
extern B ***binptr;
extern int ablock
--- Comment #14 from rguenth at gcc dot gnu dot org 2007-10-05 20:16
---
The testcase works for me, that is, it produces the expected output good.out.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33638
--- Comment #15 from zadeck at naturalbridge dot com 2007-10-05 20:17
---
Subject: Re: [4.3 regression]: wrong code with
-fforce-addr
kargl at gcc dot gnu dot org wrote:
> --- Comment #13 from kargl at gcc dot gnu dot org 2007-10-05 17:50
> ---
> (In reply to comment #9)
>
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
CC||pinskia at gcc dot gnu dot
|
--- Comment #1 from hjl at lucon dot org 2007-10-05 21:17 ---
netcdf/posixio.c is miscompiled.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33669
--- Comment #2 from tobi at gcc dot gnu dot org 2007-10-05 21:40 ---
Perhaps I should add that my hack seems to hold up against the testsuite.
--
tobi at gcc dot gnu dot org changed:
What|Removed |Added
-
Summary: Incorrect code is generated by gcc 4.1.1 when compiling vectorized
code on a PS3 running yellowdog linux.
Other platforms: the code compiles correctly on a ppc macintosh, and compiles
correctly on the PS3 if sony's ppu-gcc is used.
Discussion: It seems to be an ABI issue: the caller ass
--- Comment #1 from pinskia at gcc dot gnu dot org 2007-10-05 21:44 ---
You forgot to include -mabi=altivec.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #2 from pinskia at gcc dot gnu dot org 2007-10-05 21:45 ---
Yep, including -mabi=altivec fixes the issue so this is not a real issue.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
---
--- Comment #3 from pinskia at gcc dot gnu dot org 2007-10-05 21:46 ---
[EMAIL PROTECTED] ~]$ gcc -O3 t1212.c -maltivec
[EMAIL PROTECTED] ~]$ ./a.out
1.00 1.00 1.00 1.00
0.00 0.00 0.00 0.00
[EMAIL PROTECTED] ~]$ gcc -O3 t1212.c -maltivec -mabi=altivec
[EMA
--- Comment #3 from tobi at gcc dot gnu dot org 2007-10-05 21:47 ---
Sorry if I appear talking to myself, alternative fixes that I pondered, and
that may be cleaner, would be the following:
- record both the place where a symbol is given a type and where it is first
used. Then verify t
--- Comment #2 from hjl at lucon dot org 2007-10-05 21:49 ---
px_get in netcdf/posixio.c is miscompiled at -O2.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33669
--- Comment #3 from hjl at lucon dot org 2007-10-05 21:52 ---
px_get in netcdf/posixio.c is miscompiled at -O1 and -O0 is OK.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33669
--- Comment #1 from tobi at gcc dot gnu dot org 2007-10-05 21:38 ---
Created an attachment (id=14306)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14306&action=view)
A patch
Here's a quick hack to fix this problem. It is not complete (INTRINSIC_USER is
not handled correctly), an
--- Comment #22 from fxcoudert at gcc dot gnu dot org 2007-10-05 21:59
---
Created an attachment (id=14307)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14307&action=view)
Patch for 4 testcases of this PR
The following three cases are fixed by the patch attached (fix resolution
--- Comment #4 from j dot m dot taylor at dur dot ac dot uk 2007-10-05
22:06 ---
> Yep, including -mabi=altivec fixes the issue so this is not a real issue.
Thankyou for your prompt reply. I was not aware that I needed to do this. I am
glad to hear it's a compiler option issue rather th
--- Comment #23 from fxcoudert at gcc dot gnu dot org 2007-10-05 22:02
---
(In reply to comment #22)
> Created an attachment (id=14307)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14307&action=view) [edit]
> Patch for 4 testcases of this PR
It also fixes that one:
contains
C
--- Comment #4 from tobi at gcc dot gnu dot org 2007-10-05 22:33 ---
Ok, last comment for tonight: the second option would fail for the perfectly
legal
subroutine a(arr)
implicit none
dimension arr(5)
real arr
end
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29813
--- Comment #2 from bkoz at gcc dot gnu dot org 2007-10-05 23:01 ---
This is an include snafu, caused by algorithm bits using std::partial_sum
without an include of .
Of course, I had to wade through a couple things to get here
1) added testsuite for algo/numeric functions
2) fixu
--- Comment #3 from bkoz at gcc dot gnu dot org 2007-10-05 23:01 ---
Created an attachment (id=14308)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14308&action=view)
patch
Patch.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33487
--- Comment #1 from ludovic at ludovic-brenta dot org 2007-10-05 23:09
---
Reduced test case:
package pak1 is
package pak2 is
type T2 is tagged null record;
end pak2;
type T1 is record
a: access pak2.T2'Class;
end record;
end pak1;
gcc-4.1 -c -gnat05 pak1.ads
+=
--- Comment #2 from ludovic at ludovic-brenta dot org 2007-10-05 23:11
---
Also in 4.1.1 on i686:
gcc -c -gnat05 pak1.ads
+===GNAT BUG DETECTED==+
| 4.1.1 (i686-unknown-linux-gnu) Assert_Failure sinfo.adb:351 |
| Error de
--- Comment #4 from hjl at lucon dot org 2007-10-05 23:14 ---
#include
#include
typedef struct foo_t {
size_t blksz;
size_t bf_extent;
size_t bf_cnt;
void*bf_base;
} foo_t;
#define _RNDUP(x, unit) x) + (unit) - 1) / (unit)) \
--- Comment #3 from ludovic at ludovic-brenta dot org 2007-10-05 23:15
---
The bug seems not to be present in GCC 4.2.1; tested with gnat-4.2 4.2.1-7
(Debian).
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33071
As discussed on comp.lang.fortran. The test case namelist_15.f90 has three
examples of invalid derived type specifiers. Only one none-zero rank object is
allowed.
> write (10, '(A)') " x%i = , ,-3, -4"
The line above is illegal. It represents an array of arrays (i and m), which
is not allowe
--- Comment #2 from janis at gcc dot gnu dot org 2007-10-06 00:11 ---
In response to comment #1, the GCC Manual doesn't say anything about
-freorder-blocks-and-partition needing profile feedback although it does say
that for -freorder-functions.
--
http://gcc.gnu.org/bugzilla/show_b
Benchmark 176.gcc from SPEC CPU2000 fails with an ICE in verify_flow_info and a
message about a missing barrier when compiled on powerpc-linux with "-Os
-fno-forward-propagate -fno-guess-branch-probability -fno-move-loop-invariants
-fno-tree-dominator-opts -fno-tree-loop-optimize". It's an odd set
--- Comment #1 from janis at gcc dot gnu dot org 2007-10-06 01:23 ---
Created an attachment (id=14309)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14309&action=view)
minimized testcase
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33673
--- Comment #2 from pinskia at gcc dot gnu dot org 2007-10-06 01:30 ---
Even though the patch was for Tree level, the patch exposed a latent bug in the
rtl optimizers (missing barrier after can only happen on the rtl level).
--
pinskia at gcc dot gnu dot org changed:
What
--- Comment #5 from hjl at lucon dot org 2007-10-06 02:07 ---
Kenny, does your patch
http://gcc.gnu.org/ml/gcc-patches/2007-10/msg00124.html
handle cases where number of consecutive hard regs needed to hold some mode > 1
correctly? IA32 needs 2 hard registers to hold long long and your
--- Comment #6 from hjl at lucon dot org 2007-10-06 02:31 ---
A smaller testcase:
bash-3.2$ cat x.c
extern void abort (void);
typedef struct foo_t
{
unsigned int blksz;
unsigned int bf_cnt;
} foo_t;
#define _RNDUP(x, unit) x) + (unit) - 1) / (unit)) * (unit))
#define _RNDDO
--- Comment #7 from zadeck at naturalbridge dot com 2007-10-06 04:11
---
Subject: Re: [4.3 Regression] Revision 128957
miscompiles 481.wrf
hjl at lucon dot org wrote:
> --- Comment #5 from hjl at lucon dot org 2007-10-06 02:07 ---
> Kenny, does your patch
>
> http://gcc.gnu
--- Comment #16 from ubizjak at gmail dot com 2007-10-06 06:49 ---
(In reply to comment #14)
> The testcase works for me, that is, it produces the expected output good.out.
Uh, you have to un-comment the line 315 of the comunpack.f test. The testcase,
as attached, produces good code. Un
92 matches
Mail list logo
