I have been _slowly_ working on some CLONE_PID changes (tids, proc, sharing
pids,uid,gid,groups, ...) but there are a lot of little details that make
things much more intricate than I thought it would be. Even if I finish this
stuff before 2.2 I don't think it should go in as it needs lots of testing.
It looks like the _NSIG==32 limit is now gone with the new signal handling
code, so we should be able remove the linuxthreads dependence on SIGUSR[12],
providing the libc support is there (glibc).
I think that MT debugging should probably be possible right now with some
user level gdb work but I don't know for sure.
Peeter