Re: [RFC] fs, proc: Introduce /proc/<pid>/task/<tid>/children entry v6

[Oleg Nesterov <oleg@redhat.com>]

On 01/19, Oleg Nesterov wrote:
>
> On 01/18, Cyrill Gorcunov wrote:
> >
> > I suppose it might be something like below. I've updated comment and
> > quoted your comment there just I wont forget this next time I'll be
> > reading the source. Thanks!
>
> I believe the patrch is correct.
>
> But... Cyrill, I am wondering how much will you hate me if I make
> yet another attempt to delay this patch.

Cough... and another attempt...

> > +static int children_seq_open(struct inode *inode, struct file *file)
> > +{
> > +	struct proc_pid_children_iter *iter = NULL;
> > +	struct task_struct *task = NULL;
> > +	int ret = 0;
> > +
> > +	task = get_proc_task(inode);
> > +	if (!task) {
> > +		ret = -ENOENT;
> > +		goto err;
> > +	}
>
> For what??
>
> > +	if (!ret) {
> > +		struct seq_file *m = file->private_data;
> > +		m->private = iter;
> > +
> > +		iter->pid_start = get_pid(task_pid(task));
>
> This is what we need, right? So can't we remove this "task_struct *task"
> and simply do
>
> 		iter->pid_start = get_ppid(proc_pid(inode));
>
> ?
>
> And while this is absolutely cosmetic probably ->parent_pid is
> a bit better name, but this is up to you.

Thinking more... I am not sure, but do we really need
proc_pid_children_iter at all??

It is very possibly I missed something, but we can get both
parent_pid and pid_ns from inode, right? so can't we just remember
inode in seq->private?

Oleg.