Hi, On Mon, Sep 8, 2008 at 11:43 PM, Andrej van der Zee <[EMAIL PROTECTED]> wrote: > Hi, > >> You certainly don't need to modify the code of the child processes. >> In the handler you can read the counter of the child processes. But >> for the read to succeed, you need to stop the child. This can currently >> be accomplished using ptrace and in particular PTRACE_ATTACH. >> So you issue this call, wait (WUNTRACED) until this is effective, then you >> call pfm_read_pmds(). When you are done, simply PTRACE_DETACH >> the child. >> > > Thanks. > > I guess the pfmon source code for the --follow-all option is a good > example. The problem is that in my case I cannot stop the child with > ptrace() because the handler-thread forks in a hook-method that is not > implemented by me. So I have no access to the child. > To operate on a thread, and especially its machine state which the PMU is a part of, it needs to be stopped. Ptrace is the mechanism we currently use until utrace gets into the kernel.
> Another option that I though of is to attach to the *thread* after it > is created and then implement pfmon's --follow-all option. But I guess > attach() does not work for threads, right? > I think it does work. With pfmon for instance, with the --attach-task, you can specify a thread id and not just a process id. ------------------------------------------------------------------------- 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
