linux-kernel - Re: [RFC][PATCH] exec: Conceal the other threads from wakeups during exec

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200731062804.GA26171@redhat.com>
Date:   Fri, 31 Jul 2020 08:28:05 +0200
From:   Oleg Nesterov <oleg@...hat.com>
To:     "Eric W. Biederman" <ebiederm@...ssion.com>
Cc:     Linus Torvalds <torvalds@...ux-foundation.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Kees Cook <keescook@...omium.org>, Pavel Machek <pavel@....cz>,
        "Rafael J. Wysocki" <rjw@...ysocki.net>,
        linux-fsdevel <linux-fsdevel@...r.kernel.org>,
        Linux PM <linux-pm@...r.kernel.org>
Subject: Re: [RFC][PATCH] exec: Conceal the other threads from wakeups during
 exec

Eric, I won't comment the intent, but I too do not understand this idea.

On 07/30, Eric W. Biederman wrote:
>
> [This change requires more work to handle TASK_STOPPED and TASK_TRACED]

Yes. And it is not clear to me how can you solve this.

> [This adds a new lock ordering dependency siglock -> pi_lock -> rq_lock ]

Not really, ttwu() can be safely called with siglock held and it takes
pi_lock + rq_lock. Say, signal_wake_up().

> +int make_task_wakekill(struct task_struct *p)
> +{
> +	unsigned long flags;
> +	int cpu, success = 0;
> +	struct rq_flags rf;
> +	struct rq *rq;
> +	long state;
> +
> +	/* Assumes p != current */
> +	preempt_disable();
> +	/*
> +	 * If we are going to change a thread waiting for CONDITION we
> +	 * need to ensure that CONDITION=1 done by the caller can not be
> +	 * reordered with p->state check below. This pairs with mb() in
> +	 * set_current_state() the waiting thread does.
> +	 */
> +	raw_spin_lock_irqsave(&p->pi_lock, flags);
> +	smp_mb__after_spinlock();
> +	state = p->state;
> +
> +	/* FIXME handle TASK_STOPPED and TASK_TRACED */
> +	if ((state == TASK_KILLABLE) ||
> +	    (state == TASK_INTERRUPTIBLE)) {
> +		success = 1;
> +		cpu = task_cpu(p);
> +		rq = cpu_rq(cpu);
> +		rq_lock(rq, &rf);
> +		p->state = TASK_WAKEKILL;

You can only do this if the task was already deactivated. Just suppose it
is preempted or does something like

	set_current_sate(TASK_INTERRUPTIBLE);

	if (CONDITION) {
		// make_task_wakekill() sets state = TASK_WAKEKILL
		__set_current_state(TASK_RUNNING);
		return;
	}

	schedule();

Oleg.