Re: [PATCH v10 4/9] cgroup: cgroup v2 freezer

[Oleg Nesterov]

On 04/05, Roman Gushchin wrote:
>
> +void cgroup_leave_frozen(bool always_leave)
> +{
> +     struct cgroup *cgrp;
> +
> +     spin_lock_irq(&css_set_lock);
> +     cgrp = task_dfl_cgroup(current);
> +     if (always_leave || !test_bit(CGRP_FREEZE, &cgrp->flags)) {
> +             cgroup_dec_frozen_cnt(cgrp);
> +             cgroup_update_frozen(cgrp);
> +             WARN_ON_ONCE(!current->frozen);
> +             current->frozen = false;
> +     }
> +     spin_unlock_irq(&css_set_lock);
> +
> +     if (unlikely(current->frozen)) {
> +             /*
> +              * If the task remained in the frozen state,
> +              * make sure it won't reach userspace without
> +              * entering the signal handling loop.
> +              */
> +             spin_lock_irq(&current->sighand->siglock);
> +             recalc_sigpending();
> +             spin_unlock_irq(&current->sighand->siglock);

I still can't understand this logic.

Once again, suppose we race with CGRP_FREEZE. If JOBCTL_TRAP_FREEZE is already
set then signal_pending() must be already T and we do not need 
recalc_sigpending?
If JOBCTL_TRAP_FREEZE is not set yet, how can recalc_sigpending() help?

> +static void cgroup_freeze_task(struct task_struct *task, bool freeze)
> +{
> +     unsigned long flags;
> +
> +     /* If the task is about to die, don't bother with freezing it. */
> +     if (!lock_task_sighand(task, &flags))
> +             return;
> +
> +     if (freeze) {
> +             task->jobctl |= JOBCTL_TRAP_FREEZE;
> +             signal_wake_up(task, false);
> +     } else {
> +             task->jobctl &= ~JOBCTL_TRAP_FREEZE;
> +             wake_up_process(task);

wake_up_interruptible() ?

>  static int ptrace_signal(int signr, kernel_siginfo_t *info)
>  {
>       /*
> @@ -2442,6 +2483,10 @@ bool get_signal(struct ksignal *ksig)
>               ksig->info.si_signo = signr = SIGKILL;
>               sigdelset(&current->pending.signal, SIGKILL);
>               recalc_sigpending();
> +             current->jobctl &= ~JOBCTL_TRAP_FREEZE;
> +             spin_unlock_irq(&sighand->siglock);
> +             if (unlikely(cgroup_task_frozen(current)))
> +                     cgroup_leave_frozen(true);

Oh, and another leave_frozen below...

I feel this must be simplified somehow, but nothing comes to my mind right now.

> +             /*
> +              * If the task is leaving the frozen state, let's update
> +              * cgroup counters and reset the frozen bit.
> +              */
> +             if (unlikely(cgroup_task_frozen(current))) {
>                       spin_unlock_irq(&sighand->siglock);
> +                     cgroup_leave_frozen(true);
>                       goto relock;
>               }

afaics cgroup_leave_frozen(false) makes more sense here.

Oleg.