[dpdk-dev] [PATCHv4 3/9] null: extend number of virtual queues

[Tetsuya Mukawa]

On 2015/09/29 18:46, Kulasek, TomaszX wrote:
>> -----Original Message-----
>> From: Tetsuya Mukawa [mailto:mukawa at igel.co.jp]
>> Sent: Tuesday, September 29, 2015 04:25
>> To: Kulasek, TomaszX; dev at dpdk.org
>> Subject: Re: [dpdk-dev] [PATCHv4 3/9] null: extend number of virtual
>> queues
>>
>> On 2015/07/16 2:26, Tomasz Kulasek wrote:
>>> This patch adds a possibility to configure more than one queue on null
>>> device.
>>>
>>> Signed-off-by: Tomasz Kulasek <tomaszx.kulasek at intel.com>
>>> ---
>>>  drivers/net/null/rte_eth_null.c |   12 ++++++------
>>>  1 file changed, 6 insertions(+), 6 deletions(-)
>>>
>>> diff --git a/drivers/net/null/rte_eth_null.c
>>> b/drivers/net/null/rte_eth_null.c index a8b3191..39ffcde 100644
>>> --- a/drivers/net/null/rte_eth_null.c
>>> +++ b/drivers/net/null/rte_eth_null.c
>>> @@ -71,8 +71,8 @@ struct pmd_internals {
>>>  	unsigned nb_rx_queues;
>>>  	unsigned nb_tx_queues;
>>>
>>> -	struct null_queue rx_null_queues[1];
>>> -	struct null_queue tx_null_queues[1];
>>> +	struct null_queue rx_null_queues[RTE_MAX_QUEUES_PER_PORT];
>>> +	struct null_queue tx_null_queues[RTE_MAX_QUEUES_PER_PORT];
>>>  };
>>>
>>>
>>> @@ -213,7 +213,7 @@ eth_rx_queue_setup(struct rte_eth_dev *dev,
>> uint16_t rx_queue_id,
>>>  	if ((dev == NULL) || (mb_pool == NULL))
>>>  		return -EINVAL;
>>>
>>> -	if (rx_queue_id != 0)
>>> +	if (rx_queue_id >= dev->data->nb_rx_queues)
>>>  		return -ENODEV;
>>>
>>>  	internals = dev->data->dev_private;
>>> @@ -246,7 +246,7 @@ eth_tx_queue_setup(struct rte_eth_dev *dev,
>> uint16_t tx_queue_id,
>>>  	if (dev == NULL)
>>>  		return -EINVAL;
>>>
>>> -	if (tx_queue_id != 0)
>>> +	if (tx_queue_id >= dev->data->nb_tx_queues)
>>>  		return -ENODEV;
>>>
>>>  	internals = dev->data->dev_private;
>>> @@ -279,8 +279,8 @@ eth_dev_info(struct rte_eth_dev *dev,
>>>  	dev_info->driver_name = drivername;
>>>  	dev_info->max_mac_addrs = 1;
>>>  	dev_info->max_rx_pktlen = (uint32_t)-1;
>>> -	dev_info->max_rx_queues = (uint16_t)internals->nb_rx_queues;
>>> -	dev_info->max_tx_queues = (uint16_t)internals->nb_tx_queues;
>>> +	dev_info->max_rx_queues = RTE_DIM(internals->rx_null_queues);
>>> +	dev_info->max_tx_queues = RTE_DIM(internals->tx_null_queues);
>>>  	dev_info->min_rx_bufsize = 0;
>>>  	dev_info->pci_dev = NULL;
>>>  }
>> Hi Thomasz,
>>
>> To do like above, should we change below variables also?
>>
>> static int
>> eth_dev_null_create(const char *name,
>>                 const unsigned numa_node,
>>                 unsigned packet_size,
>>                 unsigned packet_copy)
>> {
>>         const unsigned nb_rx_queues = 1;
>>         const unsigned nb_tx_queues = 1;
>>
>>         (snip)
>>
>> }
>>
>> Tetsuya
> Hi Tetsuya,
>
> The number of requested queues are configured later, through the rte eth_dev_configure API. Here these values mean the default number of queues after null pmd creation.
>
> Tomasz.
>

Hi Tomasz,

I guess we may need to change below variables when a queue is configured.
 - internals->nb_rx_queues;
 - internals->nb_tx_queues;
Without changing, 'num_stats' in eth_stats_get() might be wrong value.

Tetsuya