On 21/10/2019 03:44, Xiao Zhang wrote:
> Vlan packets with ip length bigger then 1496 will not be received by
> i40e/i40evf due to wrong packets size checking. This patch fixes the issue
> by correcting the maximum frame size during checking.
>
> Fixes: 43e5488c0ac6 ("net/i40e: support MTU configuration")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Xiao Zhang <xiao.zh...@intel.com>
> ---
> v2
> add fix for i40evf and correct the checking when using the max_pkt_len.
> ---
> drivers/net/i40e/i40e_ethdev.c | 2 +-
> drivers/net/i40e/i40e_ethdev_vf.c | 8 +++++---
> drivers/net/i40e/i40e_rxtx.c | 6 ++++--
> 3 files changed, 10 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
> index 2ca14da..156d67b 100644
> --- a/drivers/net/i40e/i40e_ethdev.c
> +++ b/drivers/net/i40e/i40e_ethdev.c
> @@ -12103,7 +12103,7 @@ i40e_dev_mtu_set(struct rte_eth_dev *dev, uint16_t
> mtu)
> return -EBUSY;
> }
>
> - if (frame_size > RTE_ETHER_MAX_LEN)
> + if (frame_size > RTE_ETHER_MAX_LEN + I40E_VLAN_TAG_SIZE * 2)
> dev_data->dev_conf.rxmode.offloads |=
> DEV_RX_OFFLOAD_JUMBO_FRAME;
> else
> diff --git a/drivers/net/i40e/i40e_ethdev_vf.c
> b/drivers/net/i40e/i40e_ethdev_vf.c
> index 5dba092..bc825e2 100644
> --- a/drivers/net/i40e/i40e_ethdev_vf.c
> +++ b/drivers/net/i40e/i40e_ethdev_vf.c
> @@ -1782,12 +1782,14 @@ i40evf_rxq_init(struct rte_eth_dev *dev, struct
> i40e_rx_queue *rxq)
> }
Size check in jumbo on case needs to be updated also
> } else {
> if (rxq->max_pkt_len < RTE_ETHER_MIN_LEN ||
> - rxq->max_pkt_len > RTE_ETHER_MAX_LEN) {
> + rxq->max_pkt_len > RTE_ETHER_MAX_LEN +
> + I40E_VLAN_TAG_SIZE * 2) {
> PMD_DRV_LOG(ERR, "maximum packet length must be "
> "larger than %u and smaller than %u, as jumbo "
> "frame is disabled",
> (uint32_t)RTE_ETHER_MIN_LEN,
> - (uint32_t)RTE_ETHER_MAX_LEN);
> + (uint32_t)RTE_ETHER_MAX_LEN +
> + I40E_VLAN_TAG_SIZE * 2);
> return I40E_ERR_CONFIG;
> }
> }
> @@ -2747,7 +2749,7 @@ i40evf_dev_mtu_set(struct rte_eth_dev *dev, uint16_t
> mtu)
> return -EBUSY;
> }
>
> - if (frame_size > RTE_ETHER_MAX_LEN)
> + if (frame_size > RTE_ETHER_MAX_LEN + I40E_VLAN_TAG_SIZE * 2)
> dev_data->dev_conf.rxmode.offloads |=
> DEV_RX_OFFLOAD_JUMBO_FRAME;
> else
> diff --git a/drivers/net/i40e/i40e_rxtx.c b/drivers/net/i40e/i40e_rxtx.c
> index ff6eb4a..0c178e3 100644
> --- a/drivers/net/i40e/i40e_rxtx.c
> +++ b/drivers/net/i40e/i40e_rxtx.c
> @@ -2634,12 +2634,14 @@ i40e_rx_queue_config(struct i40e_rx_queue *rxq)
> }
Size check in jumbo on case needs to be updated also
> } else {
> if (rxq->max_pkt_len < RTE_ETHER_MIN_LEN ||
> - rxq->max_pkt_len > RTE_ETHER_MAX_LEN) {
> + rxq->max_pkt_len > RTE_ETHER_MAX_LEN +
> + I40E_VLAN_TAG_SIZE * 2) {
> PMD_DRV_LOG(ERR, "maximum packet length must be "
> "larger than %u and smaller than %u, "
> "as jumbo frame is disabled",
> (uint32_t)RTE_ETHER_MIN_LEN,
> - (uint32_t)RTE_ETHER_MAX_LEN);
> + (uint32_t)RTE_ETHER_MAX_LEN +
> + I40E_VLAN_TAG_SIZE * 2);
> return I40E_ERR_CONFIG;
> }
> }
>
Doing quick search for RTE_ETHER_MAX_LEN,
- in i40e_fdir_rx_queue_init(), is 'rx_ctx.rxmax = RTE_ETHER_MAX_LEN;'
still correct now?
- Looking at rte_eth_dev_configure()
https://git.dpdk.org/dpdk/tree/lib/librte_ethdev/rte_ethdev.c#n1242
Does it mean an MTU not allowed without jumbo frames during
configuration is allowed later with mtu_set?
Request is to review related code and see if there are any impacts with
this change.
@Beilei Xing @Qi Zhang, this patch needs careful review by i40e maintainer.
thanks,
Kevin.
