Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 21 Jun 2018 22:21:44 +0200 (CEST)
From:   Thomas Gleixner <tglx@linutronix.de>
To:     Lukas Wunner <lukas@wunner.de>
cc:     Bjorn Helgaas <bhelgaas@google.com>,
        Mika Westerberg <mika.westerberg@linux.intel.com>,
        Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
        linux-kernel@vger.kernel.org, linux-pci@vger.kernel.org
Subject: Re: [PATCH] genirq: Synchronize only with single thread on
 free_irq()
In-Reply-To: <8f770886632640321592873e4c902218d42c436b.1527194314.git.lukas@wunner.de>
Message-ID: <alpine.DEB.2.21.1806212217571.1591@nanos.tec.linutronix.de>
References: <8f770886632640321592873e4c902218d42c436b.1527194314.git.lukas@wunner.de>
User-Agent: Alpine 2.21 (DEB 202 2017-01-01)
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, 24 May 2018, Lukas Wunner wrote:
>  static int irq_wait_for_interrupt(struct irqaction *action)
>  {
> -	set_current_state(TASK_INTERRUPTIBLE);
> +	for (;;) {
> +		set_current_state(TASK_INTERRUPTIBLE);
>  
> -	while (!kthread_should_stop()) {
> +		if (kthread_should_stop()) {
> +			/* may need to run one last time. */
> +			if (test_and_clear_bit(IRQTF_RUNTHREAD,
> +					       &action->thread_flags)) {
> +				__set_current_state(TASK_RUNNING);
> +				return 0;
> +			}
> +			__set_current_state(TASK_RUNNING);
> +			return -1;
> +		}
>  
>  		if (test_and_clear_bit(IRQTF_RUNTHREAD,
>  				       &action->thread_flags)) {
> @@ -766,10 +776,7 @@ static int irq_wait_for_interrupt(struct irqaction *action)
>  			return 0;
>  		}
>  		schedule();
> -		set_current_state(TASK_INTERRUPTIBLE);
>  	}
> -	__set_current_state(TASK_RUNNING);
> -	return -1;
>  }
>  
>  /*
> @@ -990,7 +997,7 @@ static int irq_thread(void *data)
>  	/*
>  	 * This is the regular exit path. __free_irq() is stopping the
>  	 * thread via kthread_stop() after calling
> -	 * synchronize_irq(). So neither IRQTF_RUNTHREAD nor the
> +	 * synchronize_hardirq(). So neither IRQTF_RUNTHREAD nor the
>  	 * oneshot mask bit can be set. We cannot verify that as we
>  	 * cannot touch the oneshot mask at this point anymore as
>  	 * __setup_irq() might have given out currents thread_mask
> @@ -1595,7 +1602,7 @@ static struct irqaction *__free_irq(struct irq_desc *desc, void *dev_id)
>  	unregister_handler_proc(irq, action);
>  
>  	/* Make sure it's not being used on another CPU: */
> -	synchronize_irq(irq);
> +	synchronize_hardirq(irq);

So after that, action can be freed and when the thread above tries to
access it. Nice Use After Free. That needs more thought.

Thanks,

	tglx