[dpdk-dev] [PATCH v2 05/11] net/e1000: parse n-tuple filter
Zhao1, Wei
wei.zhao1 at intel.com
Mon Jun 5 03:55:10 CEST 2017
HI, wenzhuo
> -----Original Message-----
> From: Lu, Wenzhuo
> Sent: Monday, June 5, 2017 9:21 AM
> To: Zhao1, Wei <wei.zhao1 at intel.com>; dev at dpdk.org
> Subject: RE: [PATCH v2 05/11] net/e1000: parse n-tuple filter
>
> Hi Wei,
>
> > -----Original Message-----
> > From: Zhao1, Wei
> > Sent: Friday, June 2, 2017 2:36 PM
> > To: dev at dpdk.org
> > Cc: Lu, Wenzhuo; Zhao1, Wei
> > Subject: [PATCH v2 05/11] net/e1000: parse n-tuple filter
> >
> > Add rule validate function and check if the rule is a n-tuple rule,
> > and get the n-tuple info.
> >
> > Signed-off-by: Wei Zhao <wei.zhao1 at intel.com>
> > ---
> > drivers/net/e1000/Makefile | 1 +
> > drivers/net/e1000/e1000_ethdev.h | 9 +
> > drivers/net/e1000/igb_ethdev.c | 14 +-
> > drivers/net/e1000/igb_flow.c | 505
> > +++++++++++++++++++++++++++++++++++++++
> > 4 files changed, 521 insertions(+), 8 deletions(-) create mode
> > 100644
>
> > +
> > + if (hw->mac.type == e1000_82576) {
> > + if (filter->queue >= IGB_MAX_RX_QUEUE_NUM_82576) {
> > + memset(filter, 0, sizeof(struct
> rte_eth_ntuple_filter));
> > + rte_flow_error_set(error, EINVAL,
> > + RTE_FLOW_ERROR_TYPE_ITEM,
> > + NULL, "queue number not "
> > + "supported by ntuple filter");
> > + return -rte_errno;
> > + }
> > + filter->flags |= RTE_5TUPLE_FLAGS;
> To my opinion, that the filter is 5-tuple or 2-tuple should depend on the input
> pattern. I don't understand why it's set based on the NIC type.
This is because i350 and 82580 nic only support 2 tuple filter,
But 82576 support 5 tuple filter.
This is list in function igb_add_del_ntuple_filter().
>
> > + } else {
> > + if (filter->queue >= IGB_MAX_RX_QUEUE_NUM) {
> > + memset(filter, 0, sizeof(struct
> rte_eth_ntuple_filter));
> > + rte_flow_error_set(error, EINVAL,
> > + RTE_FLOW_ERROR_TYPE_ITEM,
> > + NULL, "queue number not "
> > + "supported by ntuple filter");
> > + return -rte_errno;
> > + }
> > + filter->flags |= RTE_2TUPLE_FLAGS;
> > + }
More information about the dev
mailing list