|
From: konstantin s. <kon...@gm...> - 2005-12-30 09:07:42
|
Hi all.
I was trying to use Intel C/C++ compiler with valgrind on x86 linux.
Valgrind 301 worked fine for small programs, but it didn't work for large.
Now, version 310 does work for larger programs, but it does not understand
debug information provided by Intel compiler (301 did).
Namely, valgrind 310 does not show line numbers if the program is compiled
with Intel compiler.
This is how I run the test:
--------------------------------------------------------
uname -a
cat -n testfile.c
for compiler in gcc icc; do
$compiler -g testfile.c -o $compiler.exe
for vlg in 301 310 ; do
echo --- $compiler with valgrind $vlg
./valgrind$vlg/bin/valgrind -q ./$compiler.exe 2>&1 | head -5;
done
done
--------------------------------------------------------
And this is what I get:
--------------------------------------------------------
Linux msticlxl1 2.4.21-15.ELsmp #1 SMP Thu Apr 22 00:18:24 EDT 2004 i686
i686 i386 GNU/Linux
1 #include <stdio.h>
2 int main(int argc)
3 {
4 int TTTT;
5 if(argc > 1)
6 TTTT =3D 777;
7 printf("%d\n", TTTT);
8 return 0;
9 }
--- gcc with valgrind 301
=3D=3D15671=3D=3D Use of uninitialised value of size 4
=3D=3D15671=3D=3D at 0x1B95BDC4: vfprintf (in /lib/tls/libc-2.3.2.so)
=3D=3D15671=3D=3D by 0x1B962E51: printf (in /lib/tls/libc-2.3.2.so)
=3D=3D15671=3D=3D by 0x8048370: main (testfile.c:7)
^^^^^^^^^^^^
=3D=3D15671=3D=3D
--- gcc with valgrind 310
=3D=3D15673=3D=3D Use of uninitialised value of size 4
=3D=3D15673=3D=3D at 0x4077DC4: vfprintf (in /lib/tls/libc-2.3.2.so)
=3D=3D15673=3D=3D by 0x407EE51: printf (in /lib/tls/libc-2.3.2.so)
=3D=3D15673=3D=3D by 0x8048370: main (testfile.c:7)
^^^^^^^^^^^^
=3D=3D15673=3D=3D
--- icc with valgrind 301
=3D=3D15681=3D=3D Use of uninitialised value of size 4
=3D=3D15681=3D=3D at 0x1B987DC4: vfprintf (in /lib/tls/libc-2.3.2.so)
=3D=3D15681=3D=3D by 0x1B98EE51: printf (in /lib/tls/libc-2.3.2.so)
=3D=3D15681=3D=3D by 0x80483C1: main (testfile.c:7)
^^^^^^^^^^^^
=3D=3D15681=3D=3D
--- icc with valgrind 310
=3D=3D15683=3D=3D Use of uninitialised value of size 4
=3D=3D15683=3D=3D at 0x40A3DC4: vfprintf (in /lib/tls/libc-2.3.2.so)
=3D=3D15683=3D=3D by 0x40AAE51: printf (in /lib/tls/libc-2.3.2.so)
=3D=3D15683=3D=3D by 0x80483C1: main (in /export/users/ksserebr/send/icc=
.exe)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
=3D=3D15683=3D=3D
--------------------------------------------------------
In combination 'icc + valgrind310' I do not get line numbers.
Three other combinations give correct line numbers.
I was trying icc 8.1 and 9.0
I would appreciate any suggestion or help.
Thanks,
Kostya
|
|
From: Nicholas N. <nj...@cs...> - 2005-12-30 20:18:36
|
On Fri, 30 Dec 2005, konstantin serebryany wrote: > I was trying to use Intel C/C++ compiler with valgrind on x86 linux. > > Valgrind 301 worked fine for small programs, but it didn't work for large. > > Now, version 310 does work for larger programs, but it does not understand > debug information provided by Intel compiler (301 did). This is a bug in 3.1.0, I think it has been fixed in the repository. To try the repository code, see http://www.valgrind.org/downloads/repository.html. Nick |
|
From: Konstantin 'K. S. <kon...@gm...> - 2005-12-31 12:35:45
|
I was able to download current version (5465) from repository, but my problem still exist. :( On 12/30/05, Nicholas Nethercote <nj...@cs...> wrote: > > > This is a bug in 3.1.0, I think it has been fixed in the repository. To > try the repository code, see > http://www.valgrind.org/downloads/repository.html. > > Nick > |
|
From: Julian S. <js...@ac...> - 2005-12-31 13:36:42
|
On Saturday 31 December 2005 12:35, Konstantin 'KCC' Serebryany wrote: > I was able to download current version (5465) from repository, but my > problem still exist. :( Yes - I'm not surprised. You're the second person to report this problem, and the change Nick refers to does not help the first person either. I will investigate. J |
|
From: Nicholas N. <nj...@cs...> - 2005-12-30 22:40:49
|
On Sat, 31 Dec 2005, Konstantin 'KCC' Serebryany wrote: > Thanks a lot! > Is there a way to download the latest version as tar ball? > I'll have a hard time using svn over my firewall.... I'm not sure, others might know... Nick |
|
From: Julian S. <js...@ac...> - 2006-01-01 23:12:47
|
Ok, svn update (make sure you get at least r5470) and try again. This time I think it is a proper fix :-) Let me know success/failure. (Gregory: see http://www.valgrind.org/downloads/repository.html for details how to build from svn.) J On Saturday 31 December 2005 12:35, Konstantin 'KCC' Serebryany wrote: > I was able to download current version (5465) from repository, but my > problem still exist. :( > > On 12/30/05, Nicholas Nethercote <nj...@cs...> wrote: > > This is a bug in 3.1.0, I think it has been fixed in the repository. To > > try the repository code, see > > http://www.valgrind.org/downloads/repository.html. > > > > Nick |
|
From: Konstantin 'K. S. <kon...@gm...> - 2006-01-02 09:19:03
|
r5470 works excellent for me! Thank you! --kcc On 1/2/06, Julian Seward <js...@ac...> wrote: > > > Ok, svn update (make sure you get at least r5470) and try again. > This time I think it is a proper fix :-) Let me know success/failure. > > (Gregory: see http://www.valgrind.org/downloads/repository.html for > details how to build from svn.) > > J > > On Saturday 31 December 2005 12:35, Konstantin 'KCC' Serebryany wrote: > > I was able to download current version (5465) from repository, but my > > problem still exist. :( > > > > On 12/30/05, Nicholas Nethercote <nj...@cs...> wrote: > > > This is a bug in 3.1.0, I think it has been fixed in the > repository. To > > > try the repository code, see > > > http://www.valgrind.org/downloads/repository.html. > > > > > > Nick > |