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;
Subject: Re: [PATCH v3 3/3] uio: fix crash after the device is unregistered
To:     Mike Christie <mchristi@redhat.com>, gregkh@linuxfoundation.org,
        linux-kernel@vger.kernel.org
Cc:     hamish.martin@alliedtelesis.co.nz, jannh@google.com,
        pkalever@redhat.com, pkarampu@redhat.com, atumball@redhat.com,
        sabose@redhat.com
References: <1530845836-49101-1-git-send-email-xiubli@redhat.com>
 <1530845836-49101-4-git-send-email-xiubli@redhat.com>
 <5B3FB3B0.7010105@redhat.com>
From:   Xiubo Li <xiubli@redhat.com>
Message-ID: <1d3a72f6-2d48-1e36-82e4-c764c5359765@redhat.com>
Date:   Sat, 7 Jul 2018 09:28:45 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:52.0) Gecko/20100101
 Thunderbird/52.9.0
MIME-Version: 1.0
In-Reply-To: <5B3FB3B0.7010105@redhat.com>
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer-Encoding: 7bit
Content-Language: en-US
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 2018/7/7 2:23, Mike Christie wrote:
> On 07/05/2018 09:57 PM, xiubli@redhat.com wrote:
>>   static irqreturn_t uio_interrupt(int irq, void *dev_id)
>>   {
>>   	struct uio_device *idev = (struct uio_device *)dev_id;
>> -	irqreturn_t ret = idev->info->handler(irq, idev->info);
>> +	irqreturn_t ret;
>> +
>> +	mutex_lock(&idev->info_lock);
>> +	if (!idev->info) {
>> +		ret = IRQ_NONE;
>> +		goto out;
>> +	}
>>   
>> +	ret = idev->info->handler(irq, idev->info);
>>   	if (ret == IRQ_HANDLED)
>>   		uio_event_notify(idev->info);
>>   
>> +out:
>> +	mutex_unlock(&idev->info_lock);
>>   	return ret;
>>   }
>
> Do you need the interrupt related changes in this patch and the first
> one?
Actually, the NULL checking is not a must, we can remove this. But the 
lock/unlock is needed.
>   When we do uio_unregister_device -> free_irq does free_irq return
> when there are no longer running interrupt handlers that we requested?
>
> If that is not the case then I think we can hit a similar bug. We do:
>
> __uio_register_device -> device_register -> device's refcount goes to
> zero so we do -> uio_device_release -> kfree(idev)
>
> and if it is possible the interrupt handler could still run after
> free_irq then we would end up doing:
>
> uio_interrupt -> mutex_lock(&idev->info_lock) -> idev access freed memory.

I think this shouldn't happen. Because the free_irq function does not 
return until any executing interrupts for this IRQ have completed.

Thanks,
BRs