Re: [PATCH] rlimit: locking tidy ups

[Oleg Nesterov <oleg@redhat.com>]

On 05/04, Eric W. Biederman wrote:
>
> Cc'd Oleg as he tends to be deeply involved with this class of locking.
>
> Mateusz Guzik <mguzik@redhat.com> writes:
>
> > proc_pid_limits takes ->sighand lock prior to accessing rlimits, but it
> > serves no purpose as it does not prevent modifications.

Well. I agree this all needs cleanups or at least additional comments, but

> > @@ -618,14 +618,12 @@ static int proc_pid_limits(struct seq_file *m, struct pid_namespace *ns,
> >  			   struct pid *pid, struct task_struct *task)
> >  {
> >  	unsigned int i;
> > -	unsigned long flags;
> >  
> >  	struct rlimit rlim[RLIM_NLIMITS];
> >  
> > -	if (!lock_task_sighand(task, &flags))
> > -		return 0;
> > +	task_lock(task->group_leader);

This is already unsafe. ->group_leader can point to nowhere if this threads
exits. lock_task_sighand() ensures that this can't happen.

> > -	/* protect tsk->signal and tsk->sighand from disappearing */
> > -	read_lock(&tasklist_lock);
> > -	if (!tsk->sighand) {
> > -		retval = -ESRCH;
> > -		goto out;
> > +	task_lock(tsk->group_leader);

The same, but yes the comment is misleading.

Oleg.