Re: [RFC][v8][PATCH 10/10]: Document clone3() syscall

[Sukadev Bhattiprolu <sukadev-23VcF4HTsmIX0ybBhKVfKdBPR1lH4CV8@public.gmane.org>]

Arnd Bergmann [arndbergmann-gM/Ye1E23mwN+BqQ9rBEUg@public.gmane.org] wrote:
| On Tuesday 13 October 2009, Sukadev Bhattiprolu wrote:
| > +clone3(struct clone_struct * __user clone_args, pid_t * __user pids)
| > +
| > +       In addition to doing everything that clone() system call does,
| > +       the clone3() system call:
| > +
| > +               - allows additional clone flags (all 32 bits in the flags
| > +                 parameter to clone() are in use)
| > +
| > +               - allows user to specify a pid for the child process in its
| > +                 active and ancestor pid name spaces.
| 
| Someone (sorry, can't find the old mail) pointed out last time that
| the 'pid_t *__user tidptr' argument needs to be an independent pointer,
| in order to allow the same use patterns with CLONE_CHILD_SETTID and
| CLONE_CHILD_CLEARTID that you can do with the current clone
| implementation.

I think it was Oren Laadan.

| 
| Moving that argument from clone_struct into the argument list would
| also make this a three-argument syscall, which solves the naming problem.

How about we pull 'nr_pids' out of clone_args struct and make that
a separate parameter. It seems the odd man out and controls the pid_t
list parameter.
| 
| 	Arnd <><