Re: [Qemu-devel] frame reordering in qemu_net_queue_send() ?

[Jan Kiszka]

On 2012-05-31 16:10, Zhi Yong Wu wrote:
> On Thu, May 31, 2012 at 9:48 PM, Jan Kiszka <address@hidden> wrote:
>> On 2012-05-31 15:45, Zhi Yong Wu wrote:
>>> On Thu, May 31, 2012 at 5:18 PM, Stefan Hajnoczi <address@hidden> wrote:
>>>> On Wed, May 30, 2012 at 8:59 AM, Luigi Rizzo <address@hidden> wrote:
>>>>> Hi,
>>>>> while investigating rx performance for emulated network devices
>>>>> (i am looking at the userspace version, relying on net=tap
>>>>> or similar approaches) i noticed the code
>>>>> in net/queue.c :: qemu_net_queue_send()
>>>>> which look strange to me (same goes for the iov version).
>>>>>
>>>>> The whole function is below, just for reference.
>>>>> My impression is that the call to qemu_net_queue_flush()
>>>>> is misplaced, and it should be before qemu_net_queue_deliver()
>>>>> otherwise the current packet is pushed out before anything
>>>>> was already in the queue.
>>>>>
>>>>> Does it make sense ?
>>>>>
>>>>> cheers
>>>>> luigi
>>>>>
>>>>>    ssize_t qemu_net_queue_send(NetQueue *queue,
>>>>>                                VLANClientState *sender,
>>>>>                                unsigned flags,
>>>>>                                const uint8_t *data,
>>>>>                                size_t size,
>>>>>                                NetPacketSent *sent_cb)
>>>>>    {
>>>>>        ssize_t ret;
>>>>>
>>>>>        if (queue->delivering) {
>>>>>            return qemu_net_queue_append(queue, sender, flags, data, size, 
>>>>> NULL);
>>>>>        }
>>>>>
>>>>>        ret = qemu_net_queue_deliver(queue, sender, flags, data, size);
>>>>>        if (ret == 0) {
>>>>>            qemu_net_queue_append(queue, sender, flags, data, size, 
>>>>> sent_cb);
>>>>>            return 0;
>>>>>        }
>>>>>
>>>>>        qemu_net_queue_flush(queue);
>>>>
>>>> This of the case where delivering a packet causes additional
>>>> (re-entrant) qemu_net_queue_send() calls to this queue.  We'll be in
>>> If qemu_net_queue_send() are re-entranced, what issue or badness will
>>> it introduce?
>>> I haven't found what issue will take place when queue_flush() is moved
>>> before queue_deliver().
>>>
>>
>> Delayed packets that are hold back during delivering==1. Having a flush
> Sorry, i don't get what it mean? You know, english is not my first
> language.:) You mean that delayed packets may be reenqueued?
>> before and after may be fine - if we really need it before.
>>

I might have been to brief as well: The purpose of the flush after
delivery is the case when we re-entered qemu_net_queue_send while
delivering. If queue->delivering is 1 on entry, we don't deliver but
queue the packet. Once the returned from qemu_net_queue_deliver to the
root qemu_net_queue_send, the queue of hold back packets has to be
flushed to avoid delays.

Jan

-- 
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux