Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH v9 16/17] vduse: Introduce VDUSE - vDPA Device in
 Userspace
To:     Greg KH <gregkh@linuxfoundation.org>
Cc:     "Michael S. Tsirkin" <mst@redhat.com>,
        Xie Yongji <xieyongji@bytedance.com>, stefanha@redhat.com,
        sgarzare@redhat.com, parav@nvidia.com, hch@infradead.org,
        christian.brauner@canonical.com, rdunlap@infradead.org,
        willy@infradead.org, viro@zeniv.linux.org.uk, axboe@kernel.dk,
        bcrl@kvack.org, corbet@lwn.net, mika.penttila@nextfour.com,
        dan.carpenter@oracle.com, joro@8bytes.org, zhe.he@windriver.com,
        xiaodong.liu@intel.com, songmuchun@bytedance.com,
        virtualization@lists.linux-foundation.org, netdev@vger.kernel.org,
        kvm@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        iommu@lists.linux-foundation.org, linux-kernel@vger.kernel.org
References: <20210713084656.232-1-xieyongji@bytedance.com>
 <20210713084656.232-17-xieyongji@bytedance.com>
 <26116714-f485-eeab-4939-71c4c10c30de@redhat.com>
 <20210714014817-mutt-send-email-mst@kernel.org>
 <0565ed6c-88e2-6d93-7cc6-7b4afaab599c@redhat.com>
 <YO6IiDIMUjQsA2LS@kroah.com>
From:   Jason Wang <jasowang@redhat.com>
Message-ID: <a493bc51-9a5c-05f1-2eac-54cb9e6c9d9b@redhat.com>
Date:   Wed, 14 Jul 2021 16:56:55 +0800
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:78.0)
 Gecko/20100101 Thunderbird/78.11.0
MIME-Version: 1.0
In-Reply-To: <YO6IiDIMUjQsA2LS@kroah.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Content-Language: en-US
Precedence: bulk


在 2021/7/14 下午2:47, Greg KH 写道:
> On Wed, Jul 14, 2021 at 02:02:50PM +0800, Jason Wang wrote:
>> 在 2021/7/14 下午1:54, Michael S. Tsirkin 写道:
>>> On Wed, Jul 14, 2021 at 01:45:39PM +0800, Jason Wang wrote:
>>>>> +static int vduse_dev_msg_sync(struct vduse_dev *dev,
>>>>> +			      struct vduse_dev_msg *msg)
>>>>> +{
>>>>> +	int ret;
>>>>> +
>>>>> +	init_waitqueue_head(&msg->waitq);
>>>>> +	spin_lock(&dev->msg_lock);
>>>>> +	msg->req.request_id = dev->msg_unique++;
>>>>> +	vduse_enqueue_msg(&dev->send_list, msg);
>>>>> +	wake_up(&dev->waitq);
>>>>> +	spin_unlock(&dev->msg_lock);
>>>>> +
>>>>> +	wait_event_killable_timeout(msg->waitq, msg->completed,
>>>>> +				    VDUSE_REQUEST_TIMEOUT * HZ);
>>>>> +	spin_lock(&dev->msg_lock);
>>>>> +	if (!msg->completed) {
>>>>> +		list_del(&msg->list);
>>>>> +		msg->resp.result = VDUSE_REQ_RESULT_FAILED;
>>>>> +	}
>>>>> +	ret = (msg->resp.result == VDUSE_REQ_RESULT_OK) ? 0 : -EIO;
>>>> I think we should mark the device as malfunction when there is a timeout and
>>>> forbid any userspace operations except for the destroy aftwards for safety.
>>> This looks like if one tried to run gdb on the program the behaviour
>>> will change completely because kernel wants it to respond within
>>> specific time. Looks like a receipe for heisenbugs.
>>>
>>> Let's not build interfaces with arbitrary timeouts like that.
>>> Interruptible wait exists for this very reason.
>>
>> The problem is. Do we want userspace program like modprobe to be stuck for
>> indefinite time and expect the administrator to kill that?
> Why would modprobe be stuck for forever?
>
> Is this on the module probe path?


Yes, it is called in the device probing path where the kernel forwards 
the device configuration request to userspace and wait for its response.

If it turns out to be tricky, we can implement the whole device inside 
the kernel and leave only the datapath in the userspace (as what TUN did).

Thanks


>