On Sun, Apr 19, 2015 at 9:35 AM, Johannes Lorenz <johannes89@mailueberfall.de> wrote:

> It's realtime safe as long as you know that those system calls pass
> control to the thread you want them to pass control to and not some
> other random thread that has nothing to do with JACK.

Ok, but how does JACK keep the OS from executing a completely different process as the next job?

because the relevant threads in the server AND in the clients are all scheduled in the SCHED_FIFO class, with (relatively) high priority.

Also, as you say, by using e.g., malloc(), the realtime app indeed passes the control to the kernel thread, which has to do a lot with JACK (it does exactly what the rt app need to get process). So why is calling malloc() not realtime then?

**if** malloc() ends up with a call to the system call brk(), then the kernel needs to memory allocation on behalf of the calling process. this is not realtime safe.