|
From: Pol v. A. <mac...@po...> - 2008-10-20 22:54:22
|
Hello, When running an application under valgrind, any signal that normally would trigger a core dump is caught by valgrind. This is as expected, however, the valgrind docs state that valgrind will produce a vgcore.%pid% dump when this happens. However, no such thing happens in my case. The command used to run valgrind is valgrind -v --tool=memcheck --leak-check=yes --num-callers=10 --time-stamp=yes --log-file=/opt/inspircd/valgrindlogs/valdebug.%p $binpath/$executable -nofork -debug -nolog >> valdebug.insp.out 2>> valdebug.insp.err & The valgrind debug log entries for the specific signal are ==404== ==404== Process terminating with default action of signal 6 (SIGABRT) ==404== at 0x419C5D5: raise (in /lib/i686/libc-2.7.so) Normally, SIGABRT triggers a core dump according to the man pages. So why doesn't valgrind dump the core? Regards, Pol van Aubel P.S. Apologies in advance for any double post that might ensue from posting with a wrong e-mail address first. |
|
From: Julian S. <js...@ac...> - 2008-10-20 23:03:08
|
> Normally, SIGABRT triggers a core dump according to the man pages. So > why doesn't valgrind dump the core? Um, I don't know. What architecture are you running on? Can you send a small test case program that shows the problem? J |
|
From: Pol v. A. <mac...@po...> - 2008-10-20 23:55:51
|
My apologies, it appears I have found the cause of the problem. Since the software when run by itself generated core dumps just fine I assumed core ulimit was set to something other than 0. Turns out my assumption was incorrect. After changing it to unlimited, valgrind acts as desired. Again, my apologies for the inconvenience. Pol Julian Seward wrote: >> Normally, SIGABRT triggers a core dump according to the man pages. So >> why doesn't valgrind dump the core? > > Um, I don't know. What architecture are you running on? Can you send a > small test case program that shows the problem? > > J |