|
From: Fernandez, D. <df...@qu...> - 2004-08-02 21:26:20
|
Hi there, I've been trying to get valgrind to work on a multi-threaded application and keep getting the following error midway. My platform is a RH Linux AS 3 Update 2 , IBM x-server dual CPU Error=3D=3D=3D valgrind's libpthread.so: VG_N_RWLOCKS is too low. Increase and recompile. Please report this bug at: valgrind.kde.org =3D=3D32706=3D=3D =3D=3D32706=3D=3D ERROR SUMMARY: 3 errors from 3 contexts (suppressed: 0 = from 0) =3D=3D32706=3D=3D malloc/free: in use at exit: 24475064 bytes in 25258 = blocks. =3D=3D32706=3D=3D malloc/free: 25826 allocs, 569 frees, 24844538 bytes allocated. =3D=3D32706=3D=3D For a detailed leak analysis, rerun with: = --leak-check=3Dyes =3D=3D32706=3D=3D For counts of detected errors, rerun with: -v I tried building valgrind with different values in in coregrind/vg_include.h up to 25000 and still get the same error.=20 Any ideas what the right value should be and how to circumvent this problem? Thanks in advance, David |
|
From: Tom H. <th...@cy...> - 2004-08-02 23:25:22
|
In message <6805AEBFF646BD4FB58D21469872D8767EE6D0@NAEX02.na.qualcomm.com>
"Fernandez, Dave" <df...@qu...> wrote:
> I tried building valgrind with different values in in
> coregrind/vg_include.h up to 25000 and still get the same error.
> Any ideas what the right value should be and how to circumvent this
> problem?
That seems awfully large - what would you expect to be the
maximum number of read-write locks created by your application?
Tom
--
Tom Hughes (th...@cy...)
Software Engineer, Cyberscience Corporation
http://www.cyberscience.com/
|
|
From: David F. <df...@qu...> - 2004-08-03 00:20:25
|
Tom Hughes <thh <at> cyberscience.com> writes: > > In message <6805AEBFF646BD4FB58D21469872D8767EE6D0 <at> NAEX02.na.qualcomm.com> > "Fernandez, Dave" <dfern <at> qualcomm.com> wrote: > > > I tried building valgrind with different values in in > > coregrind/vg_include.h up to 25000 and still get the same error. > > Any ideas what the right value should be and how to circumvent this > > problem? > > That seems awfully large - what would you expect to be the > maximum number of read-write locks created by your application? > > Tom > I agree the number is quite large but I started with 100, 1000,10000 and so on and I still get the same error. The application starts to come up but then valgrind aborts right after printing the VG_N_RWLOCKS error. My application should not require more than 200 RW locks and maybe that's still high but is a good estimate. I'm trying with valgrind 2.1.2 Even when running valgrind with -v I don't see any meaningful debug info to know what might be wrong. Any suggestions on how to keep valgrind happy? Thanks, DF |
|
From: Tom H. <th...@cy...> - 2004-08-03 06:16:23
|
In message <loo...@po...>
David Fernandez <df...@qu...> wrote:
> I agree the number is quite large but I started with 100, 1000,10000 and so
> on and I still get the same error. The application starts to come up but
> then valgrind aborts right after printing the VG_N_RWLOCKS error.
>
> My application should not require more than 200 RW locks and maybe that's
> still high but is a good estimate.
Sounds like there is a bug in valgrind then, if what you say about your
application is right. If you only need about 200 then setting VG_N_RWLOCKS
to 1000 should obviously be plenty.
> Any suggestions on how to keep valgrind happy?
Well clearly there must be a bug in valgrind. I suggest you try and
create a simple test case and then file a bug report.
Tom
--
Tom Hughes (th...@cy...)
Software Engineer, Cyberscience Corporation
http://www.cyberscience.com/
|