[dpdk-stable] [Suspected-Phishing][PATCH] net/vhost: fix redundant queue state event

[Maxime Coquelin]

On 6/26/19 10:03 AM, Noa Ezra wrote:
> 
> 
>> -----Original Message-----
>> From: Maxime Coquelin [mailto:maxime.coquelin at redhat.com]
>> Sent: Wednesday, June 26, 2019 10:01 AM
>> To: Noa Ezra <noae at mellanox.com>
>> Cc: Matan Azrad <matan at mellanox.com>; dev at dpdk.org; stable at dpdk.org
>> Subject: Re: [Suspected-Phishing][PATCH] net/vhost: fix redundant queue
>> state event
>>
>>
>>
>> On 6/26/19 8:37 AM, Noa Ezra wrote:
>>> Hi,
>>>
>>>> -----Original Message-----
>>>> From: Maxime Coquelin [mailto:maxime.coquelin at redhat.com]
>>>> Sent: Tuesday, June 25, 2019 11:24 AM
>>>> To: Noa Ezra <noae at mellanox.com>
>>>> Cc: Matan Azrad <matan at mellanox.com>; dev at dpdk.org;
>> stable at dpdk.org
>>>> Subject: Re: [Suspected-Phishing][PATCH] net/vhost: fix redundant
>>>> queue state event
>>>>
>>>>
>>>>
>>>> On 6/25/19 9:04 AM, Noa Ezra wrote:
>>>>> Hi,
>>>>> What do you think about this patch?
>>>>>
>>>>> Thanks,
>>>>> Noa.
>>>>>
>>>>>> -----Original Message-----
>>>>>> From: Noa Ezra [mailto:noae at mellanox.com]
>>>>>> Sent: Wednesday, June 19, 2019 9:16 AM
>>>>>> To: maxime.coquelin at redhat.com
>>>>>> Cc: Matan Azrad <matan at mellanox.com>; dev at dpdk.org; Noa Ezra
>>>>>> <noae at mellanox.com>; stable at dpdk.org
>>>>>> Subject: [Suspected-Phishing][PATCH] net/vhost: fix redundant queue
>>>>>> state event
>>>>>>
>>>>>> In some situations, when a virtual machine is starting,
>>>>>> vring_state_changed can be called while there was no change in the
>>>>>> queue state. This fix makes sure that there was really a change in
>>>>>> the queue state before calling the callback for EVENT_QUEUE_STATE.
>>>>>>
>>>>>> Fixes: ee584e9710b9 ("vhost: add driver on top of the library")
>>>>>> Cc: stable at dpdk.org
>>>>>> Reviewed-by: Matan Azrad <matan at mellanox.com>
>>>>>> ---
>>>>>>     drivers/net/vhost/rte_eth_vhost.c | 4 ++++
>>>>>>     1 file changed, 4 insertions(+)
>>>>>>
>>>>>> diff --git a/drivers/net/vhost/rte_eth_vhost.c
>>>>>> b/drivers/net/vhost/rte_eth_vhost.c
>>>>>> index cad1e5c..fbe7a37 100644
>>>>>> --- a/drivers/net/vhost/rte_eth_vhost.c
>>>>>> +++ b/drivers/net/vhost/rte_eth_vhost.c
>>>>>> @@ -855,6 +855,10 @@ struct vhost_xstats_name_off {
>>>>>>     	/* won't be NULL */
>>>>>>     	state = vring_states[eth_dev->data->port_id];
>>>>>>     	rte_spinlock_lock(&state->lock);
>>>>>> +	if (state->cur[vring] == enable) {
>>>>>> +		rte_spinlock_unlock(&state->lock);
>>>>>> +		return 0;
>>>>>> +	}
>>>>>>
>>>>>>     	state->cur[vring] = enable;
>>>>>>     	state->max_vring = RTE_MAX(vring, state->max_vring);
>>>>
>>>> Maybe the application would want to be notified a new queue is
>>>> available, even if it is disabled?
>>>
>>> Can you please look again? As I understand it, "enable" is the "new state"
>> parameter (can be enable/disable).
>>> In this fix I make sure that there was really a change in the state before
>> calling EVENT_QUEUE_STATE (with no change in the state).
>>
>> Ok, I get it.
>>
>> Maybe we would want to update state->max_vring even if we don't sent the
>> event?
> 
> Ok, so do you think we should move the check (if there was a change) to after we update the max_vring?

Yes, that's what I meant.
Looking at the code, it seems it does not change much the behavior,
but I think it is cleaner.

What we want with the patch is only to not send the event.

Thanks,
Maxime

>> Thanks,
>> Maxime
>>>>
>>>>>> --
>>>>>> 1.8.3.1
>>>>>