|
From: Eric <eh...@al...> - 2011-03-17 20:47:59
|
So I spoke to a friend who has experience with valgrind and it turns out I had the commands in the wrong order. He told me I wanted to run valgrind on my process and not on mpiexec. So now that I'm running the commands in the following order, it's working "mpiexec -np 5 valgrind spk_mac_mpi < inputfile.txt". This may have been obvious to everyone else, but I thought I'd post the solution all the same. On 3/17/11 9:23 AM, "Eric" <eh...@al...> wrote: I'm a new user of valgrind and am attempting to debug an open-source package called spparks (available at http://www.cs.sandia.gov/~sjplimp/spparks.html). I have installed valgrind and it works when I run a simple commands like "valgrind ls -l". However, when I run the spparks package within valgrind, I get an error. This is being run on a Mac Pro 64-bit workstation and runs under MPI. I am fairly confident that the spparks package does not have any memory issues for the simple case I am testing (I plan to later debug some personalized code I added). So when I execute the command "valgrind mpiexec -np 5 spk_mac_mpi < inputfile.txt" It starts out like it is going to work and then continuously prints out the following two lines of text over and over --- ==37273== Warning: ignored attempt to set SIGUSR2 handler in sigaction(); ==37273== the SIGUSR2 signal is used internally by Valgrind --- I have looked through the valgrind source code and found the following text in "memcheck/tests/sigkill.stderr.exp-darwin" --- setting signal 31: Warning: ignored attempt to set SIGUSR2 handler in sigaction(); the SIGUSR2 signal is used internally by Valgrind Invalid argument getting signal 31: Success --- and it looks like the signal 31 is set in the following files on the following lines: --- include/vki/vki-ppc32-aix5.h:224:#define VKI_SIGUSR2 31 include/vki/vki-ppc64-aix5.h:225:#define VKI_SIGUSR2 31 --- now I may be missing something, but it seems to me that this is the only time that these are the only files that the signal 31 is set for SIGUSR2 and this happens to be for the ppc files and I'm running this on an Intel machine. When I run the configure command my Primary build target (and only build target as I enabled only 64 bit) is AMD64_DARWIN. So going on this, if I look at the include file for darwin I see that it is --- include/vki/vki-darwin.h:482:#define VKI_SIGUSR2 SIGUSR2 --- and isn't set to a specific number unless SIGUSR2 is set to 31 elsewhere. So I may have gone down the wrong path entirely, but I'm wondering if there could be an error for the AMD64_DARWIN build. Or the more likely result is that I am doing something wrong, but I don't know what that is. Thanks for your help in the matter. Eric------------------------------------------------------------------------------ Colocation vs. Managed Hosting A question and answer guide to determining the best fit for your organization - today and in the future. http://p.sf.net/sfu/internap-sfd2d_______________________________________________ Valgrind-users mailing list Val...@li... https://lists.sourceforge.net/lists/listinfo/valgrind-users |