Re: [RFC][v6][PATCH 0/9] clone_with_pids() syscall

[Arnd Bergmann <arnd@arndb.de>]

On Friday 11 September 2009, Peter Zijlstra wrote:
> Why not have something like:
> 
> struct clone_struct {
>         u32 size;
>         u32 __reserved;
>         u64 flags;
>         u64 child_stack;
>         u32 child_tid;
>         u32 parent_tid;
> };
> 
> struct clone_pid_struct {
>         u32 nr;
>         pid_t pids[];
> };
> 
> int clone2(struct clone_struct *cs, struct clone_pid_struct *cps);
> 
> If you then get passed a longer clone_struct than you know about, all is
> well IFF the tail is 0, otherwise fail with -E2BIG.
> 
> If you get passed a short clone_struct, zero out the tail.

I would leave out the size argument. We can put a few reserved fields
and flag bits in there for possible extensions, but if we ever run out
of these, just define a new syscall.

Also, if you're passing a struct, why not put nr_pids in there, and
replace clone_pid_struct with a simple array? That would give us

struct clone_struct {
        u64 flags;
        u64 child_stack;
        u32 child_tid;
        u32 parent_tid;
	u32 nr_pids;
	u32 reserved1;
	u64 reserved2;
};
 
int clone2(struct clone_struct *cs, pid_t *pids);

	Arnd <><