Why not just have two kinds of pids, namespace local and global.
Local pids look like current pids, global ones have the high bit set. Each process can now be
"named" in two ways (namespace relative or absolute).
Clearly some things (like kill(2)) have to verify that a pid referenced through a global
identifier is in the same namespace (or has right to affect the other namespace), but the
global id is useful for things like the robust futexes.
Posted Nov 9, 2007 22:31 UTC (Fri) by khim (subscriber, #9252)
[Link]
Suppose I have 400'000-500'000 computers (Google actually does) and I want to use globally unique IDs. And one bit to separate global vs local PIDs. I can only have ~4000 processes on one computer! In all containers! Combined! Plus I'll need complex system to keep all these tables around somewhere and do a lot of other things - just to make it possible to do some insane things...