Frederick,
I think the problem is in the handler. It reminds me of a discussion
we have had earlier this year
with Mark Krentel.
The problem is related to your assumption about which thread handles
the signal for which context.
You are assuming that each thread processes its own samples. There is
no such guarantee with async
signals. So I think the reason you program hangs sometimes is simply
because you are not reading
from the correct context.
The only information that is trustworthy is the file descriptor
carried by the siginfo (info->si_fd), assuming
you've specified SA_SIGINFO in the sigaction structure.. You need to
use info->si_fd to identify the context
for which you have to process the signal. Using gettid(), or
pthread_self() won't work. And yes, a thread may
be used to process another thread's samples.
Obviously, I don't have control over POSIX signal semantics and
threads, so there is not much we can do
about this. One thing I do recall, though, is that if you use a
real-time signal instead (sig > 32) , then things
improve.
> void my_handler(int n, struct siginfo *info, struct sigcontext *sc) {
> int tid = syscall(SYS_gettid);
> int cpu = tid - getpid() - 1;
>
> printf("cpu: %i\n", cpu);
> if (ibs_handle_samples(ctx_fd[cpu], &pd[cpu], buffer[cpu]) != IBS_SUCCESS)
> exit(-1);
> }
-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
perfmon2-devel mailing list
perfmon2-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
