Re: [dpdk-dev] [PATCH 1/3] app/testpmd: update Rx offload after setting MTU sccessfully

["Wei Hu (Xavier)" <huwei013@chinasoftinc.com>]

Hi, Ferruh Yigit

On 2020/1/28 19:27, Ferruh Yigit wrote:
> On 1/21/2020 11:44 AM, Wei Hu (Xavier) wrote:
>> From: "Wei Hu (Xavier)" <xavier.huwei@huawei.com>
>>
>> Currently, Rx offload capabilities and max_rx_pkt_len in the struct
>> variable named rte_port are not updated after setting mtu successfully
>> in port_mtu_set function by 'port config mtu <port_id> <value>' command.
>> This may lead to reconfig mtu to the initial value in the driver when
>> recalling rte_eth_dev_configure API interface.
>>
>> This patch updates Rx offload capabilities and max_rx_pkt_len after
>> setting mtu successfully when configuring mtu.
>>
>> Fixes: ae03d0d18adf ("app/testpmd: command to configure MTU")
>> Cc: stable@dpdk.org
>>
>> Signed-off-by: Huisong Li <lihuisong@huawei.com>
>> Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
>> ---
>>   app/test-pmd/config.c | 18 +++++++++++++++++-
>>   1 file changed, 17 insertions(+), 1 deletion(-)
>>
>> diff --git a/app/test-pmd/config.c b/app/test-pmd/config.c
>> index 9669cbd4c..09a1579f5 100644
>> --- a/app/test-pmd/config.c
>> +++ b/app/test-pmd/config.c
>> @@ -1216,7 +1216,9 @@ void
>>   port_mtu_set(portid_t port_id, uint16_t mtu)
>>   {
>>   	int diag;
>> +	struct rte_port *rte_port = &ports[port_id];
>>   	struct rte_eth_dev_info dev_info;
>> +	uint16_t eth_overhead;
>>   	int ret;
>>   
>>   	if (port_id_is_invalid(port_id, ENABLED_WARN))
>> @@ -1232,8 +1234,22 @@ port_mtu_set(portid_t port_id, uint16_t mtu)
>>   		return;
>>   	}
>>   	diag = rte_eth_dev_set_mtu(port_id, mtu);
>> -	if (diag == 0)
>> +	if (diag == 0) {
>> +		/*
>> +		 * Ether overhead in driver is equal to the difference of
>> +		 * max_rx_pktlen and max_mtu in rte_eth_dev_info.
>> +		 */
>> +		eth_overhead = dev_info.max_rx_pktlen - dev_info.max_mtu;
>> +		if (mtu > RTE_ETHER_MAX_LEN - eth_overhead)
>> +			rte_port->dev_conf.rxmode.offloads |=
>> +						DEV_RX_OFFLOAD_JUMBO_FRAME;
>> +		else
>> +			rte_port->dev_conf.rxmode.offloads &=
>> +						~DEV_RX_OFFLOAD_JUMBO_FRAME;
> 
> If the jumbo frame capability supported or not should be tested before setting it.
> 
>> +		rte_port->dev_conf.rxmode.max_rx_pkt_len = mtu + eth_overhead;
> 
> May need to check against 'dev_info.max_rx_pktlen', if the 'max_rx_pkt_len' is
> bigger than this, it will fail in next configure.
> 
> Also some divers already does this in PMD code, should we clean that code or not
> is a question.
> 
The snippset is adjusted as follows:

if (mtu > RTE_ETHER_MAX_LEN - eth_overhead && dev_info.rx_offload_capa & 
DEV_RX_OFFLOAD_JUMBO_FRAME) {
	rte_port->dev_conf.rxmode.offloads |= 			
                                          DEV_RX_OFFLOAD_JUMBO_FRAME;
	rte_port->dev_conf.rxmode.max_rx_pkt_len = mtu + eth_overhead;
} else
	rte_port->dev_conf.rxmode.offloads &=
					~DEV_RX_OFFLOAD_JUMBO_FRAME;

We only modifies the internal variables of testpmd, don't impact on the 
implementation of the driver.

Thanks for more suggestions.

Xavier