Re: [PATCH RFC 1/2] Add polling support to pidfd

[Joel Fernandes]

On Tue, Apr 16, 2019 at 02:04:31PM +0200, Oleg Nesterov wrote:
> On 04/11, Joel Fernandes (Google) wrote:
> >
> > +static unsigned int proc_tgid_base_poll(struct file *file, struct poll_table_struct *pts)
> > +{
> > +	int poll_flags = 0;
> > +	struct task_struct *task;
> > +	struct pid *pid;
> > +
> > +	task = get_proc_task(file->f_path.dentry->d_inode);
> > +
> > +	WARN_ON_ONCE(task && !thread_group_leader(task));
> > +
> > +	/*
> > +	 * tasklist_lock must be held because to avoid racing with
> > +	 * changes in exit_state and wake up. Basically to avoid:
> > +	 *
> > +	 * P0: read exit_state = 0
> > +	 * P1: write exit_state = EXIT_DEAD
> > +	 * P1: Do a wake up - wq is empty, so do nothing
> > +	 * P0: Queue for polling - wait forever.
> > +	 */
> > +	read_lock(&tasklist_lock);
> > +	if (!task)
> > +		poll_flags = POLLIN | POLLRDNORM | POLLERR;
> > +	else if (task->exit_state == EXIT_DEAD)
> > +		poll_flags = POLLIN | POLLRDNORM;
> > +	else if (task->exit_state == EXIT_ZOMBIE && thread_group_empty(task))
> > +		poll_flags = POLLIN | POLLRDNORM;
> > +
> > +	if (!poll_flags) {
> > +		pid = proc_pid(file->f_path.dentry->d_inode);
> > +		poll_wait(file, &pid->wait_pidfd, pts);
> > +	}
> 
> can't understand...
> 
> Could you explain when it should return POLLIN? When the whole process exits?

It returns POLLIN when the task is dead or doesn't exist anymore, or when it
is in a zombie state and there's no other thread in the thread group.

> Then all you need is
> 
> 	!task || task->exit_state && thread_group_empty(task)

Yes this works as well, all the tests pass with your suggestion so I'll
change it to that. Although I will the be giving up returing EPOLLERR if the
task_struct doesn't exit. We don't need that, but I thought it was cool to
return it anyway.

> Please do not use EXIT_DEAD/EXIT_ZOMBIE. And ->wait_pidfd should probably
> live in task->signal_struct.

About wait_pidfd living in signal_struct, that wont work since the waitqueue
has to survive for the duration of the poll system call. Linus also confirmed
this: https://lore.kernel.org/patchwork/patch/1060650/#1257371

Also the waitqueue living in struct pid solves the de_thread() issue I
mentioned later in the following thread and in the commit message:
https://lore.kernel.org/patchwork/comment/1257175/

thanks,

 - Joel