Re: [PATCH] rework/fix is_single_threaded()

[David Howells <dhowells@redhat.com>]

Oleg Nesterov <oleg@redhat.com> wrote:

> - Fix the comment, is_single_threaded(p) actually means that nobody shares
>   ->mm with p.
> 
>   I think this helper should be renamed,

What we want to know when we ask this function is whether or not a process is
single-threaded, hence the name.  The fact that because:

	CLONE_THREAD => CLONE_SIGHAND => CLONE_VM

we can work this out purely by checking that there aren't any processes that
share VM space with us is immaterial.

> and it should not have arguments.  With or without this patch it must not be
> used unless p == current, otherwise we can't safely use p->signal or p->mm.

Well, I can live with that, but you need to check with the SELinux people too.
Whilst they do currently limit the selinux_setprocattr() to current only, they
still hand the task pointer that function is given around.

> - Use down_write(mm->mmap_sem) + rcu_read_lock() instead of tasklist_lock
>   to iterate over the process list. If there is another CLONE_VM process
>   it can't pass exit_mm() which takes the same mm->mmap_sem. We can miss
>   a freshly forked CLONE_VM task, but this doesn't matter because we must
>   see its parent and return false.

Hmmm...  I'd quite like to avoid using down_write() if possible.  Why do we
need to do this?  Is it just to stop processes that might cease using mm from
doing so until we've finished?

David