> On Fri, 11 Jun 1999, Ingo Molnar wrote:
>
> > ... FYI, i have a patch in the queue that makes nr_tasks a /proc/sys
> > runtime configurable option and removes the GDT limitation in the x86
> > case. I've run a kernel with 11000 processes. So i think it's not worth
> > improving the configurability of NR_TASKS (in 2.3).
>
> Please, mail-me when this patch are done. It can be very useful in my
> future beowulf projects. And I can test it for you.
I'd actually like to see a 64-bit pid_t. First of all, I've already once
being biten by a bug caused by a re-used PID resulting in the process
re-using getting killed. Second never re-using PIDs will make get_pid()
trivial like
static inline pid_t get_pid(unsigned long flags)
{
if (flags & CLONE_PID)
return current->pid;
return atomic_inc(last_pid);
}
Further optimizations are possible for certain special applications like
clusters.
(I think some of the realworld problems with different pid assignment
strategies have already been discussed on l-k before but I don't remember
the details anymore.)
Ralf
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/