Re: Linux 2.3.39 has 32bit uid. What about 32bit pid?

From: Andries.Brouwer@cwi.nl
Date: Sat Jan 15 2000 - 18:26:54 EST


    From khim.sch57.msk.ru!linux-kernel@khim.sch57.msk.ru Sat Jan 15 23:05:57 2000

>> /proc/sys/kern/max_pid ?

> I do not think that would be meaningful.

    It WOULD BE. Think about clusters. With 32bit pids you can use 8bit to identify
    node and 24bit to identify task on that node. To do this you'll need
    max_pid==16777215. IMHO max_pid is exactly right here.

Maybe. But it looks like you didnt understand my point of view.
Knobs are harmful.

What happens if we have /proc/sys/kern/max_pid?
People will want to add code to the kernel to do
the right thing when max_pid is decreased.
People will add code in userspace to find the right value for max_pid.
People will report mysterious crashes that
turn out to be caused by max_pid=64.
We will have a max_pid HOWTO.

That is what I want to avoid. If there is a compile time constant
that is 2^31 by default, 2^15 for people that ask for backward
compatibility, 2^23 for people with clusters, then everybody is happy.
No need to change this value dynamically.

Anyway, all this is unimportant. Someone must look at proc.
Maybe I already did something a month or so ago, but then
Alexander also was changing proc. Hope to have time next week
to look at the details if nobody else does it first.

Andries

-
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/



This archive was generated by hypermail 2b29 : Sun Jan 23 2000 - 21:00:13 EST