[PATCH v3 1/2] livepatch: send a fake signal to all blocking tasks

Miroslav Benes mbenes at suse.cz
Thu Nov 2 02:06:19 AEDT 2017


> +/*
> + * Sends a fake signal to all non-kthread tasks with TIF_PATCH_PENDING set.
> + * Kthreads with TIF_PATCH_PENDING set are woken up. Only admin can request this
> + * action currently.
> + */
> +void klp_force_signals(void)
> +{
> +	struct task_struct *g, *task;
> +
> +	pr_notice("signaling remaining tasks\n");
> +
> +	read_lock(&tasklist_lock);
> +	for_each_process_thread(g, task) {
> +		if (!klp_patch_pending(task))
> +			continue;
> +
> +		/*
> +		 * There is a small race here. We could see TIF_PATCH_PENDING
> +		 * set and decide to wake up a kthread or send a fake signal.
> +		 * Meanwhile the task could migrate itself and the action
> +		 * would be meaningless. It is not serious though.
> +		 */
> +		if (task->flags & PF_KTHREAD) {
> +			/*
> +			 * Wake up a kthread which still has not been migrated.
> +			 */
> +			wake_up_process(task);

So this is not as safe as one would hope. It tries to wake all TASK_NORMAL 
tasks, which could cause headaches. Let's make it

			wake_up_state(task, TASK_INTERRUPTIBLE);

to wake only kthreads sleeping interruptedly.

Thanks Petr for spotting this (offline).

Miroslav

> +		} else {
> +			/*
> +			 * Send fake signal to all non-kthread tasks which are
> +			 * still not migrated.
> +			 */
> +			spin_lock_irq(&task->sighand->siglock);
> +			signal_wake_up(task, 0);
> +			spin_unlock_irq(&task->sighand->siglock);
> +		}
> +	}
> +	read_unlock(&tasklist_lock);
> +}


More information about the Linuxppc-dev mailing list