[dpdk-dev] [RFC] ethdev: add VLAN attributes to ETH item

Dekel Peled dekelp at mellanox.com
Wed Aug 5 08:53:42 CEST 2020


Thanks, PSB.

> -----Original Message-----
> From: Eli Britstein <elibr at mellanox.com>
> Sent: Tuesday, August 4, 2020 6:47 PM
> To: Dekel Peled <dekelp at mellanox.com>; ferruh.yigit at intel.com;
> arybchenko at solarflare.com; Ori Kam <orika at mellanox.com>; Thomas
> Monjalon <thomas at monjalon.net>
> Cc: Asaf Penso <asafp at mellanox.com>; Matan Azrad
> <matan at mellanox.com>; dev at dpdk.org
> Subject: Re: [RFC] ethdev: add VLAN attributes to ETH item
> 
> 
> On 8/4/2020 6:36 PM, Dekel Peled wrote:
> > In existing code the match on tagged/untagged packets is not explicit.
> > Recent documentation update [1] describes the different patterns and
> > clarifies the intended use of different patterns.
> >
> > This patch proposes an update to ETH item struct, to clearly define
> > the required characteristic of a packet, and enable precise match criteria.
> >
> > [1] http://mails.dpdk.org/archives/dev/2020-May/166257.html
> >
> > Signed-off-by: Dekel Peled <dekelp at mellanox.com>
> > ---
> >   lib/librte_ethdev/rte_flow.h | 9 +++++++++
> >   1 file changed, 9 insertions(+)
> >
> > diff --git a/lib/librte_ethdev/rte_flow.h
> > b/lib/librte_ethdev/rte_flow.h index cf0eccb..345feb5 100644
> > --- a/lib/librte_ethdev/rte_flow.h
> > +++ b/lib/librte_ethdev/rte_flow.h
> > @@ -726,11 +726,20 @@ struct rte_flow_item_raw {
> >    * If the @p ETH item is the only item in the pattern, and the @p type field
> >    * is not specified, then both tagged and untagged packets will match the
> >    * pattern.
> > + * The fields @p cvlan_exist and @p svlan_exist can be used to match
> > + specific
> > + * packet types, instead of using the @p type field. This can be used
> > + to match
> > + * on packets that do/don't contain either cvlan, svlan, or both.
> > + * The field @p num_of_vlans can be used to match packets by the
> > + exact number
> > + * of VLANs in header.
> >    */
> >   struct rte_flow_item_eth {
> >   	struct rte_ether_addr dst; /**< Destination MAC. */
> >   	struct rte_ether_addr src; /**< Source MAC. */
> >   	rte_be16_t type; /**< EtherType or TPID. */
> > +	uint32_t cvlan_exist:1; /**< C-tag VLAN exist in header. */
> > +	uint32_t svlan_exist:1; /**< S-tag VLAN exist in header. */
> > +	uint32_t reserved:14; /**< Reserved, must be zero. */
> > +	uint32_t num_of_vlans:16; /**< Number of VLANs in header. */
> We can deduct from num_of_vlans the values of cvlan_exist/svlan_exist, so
> those are redundant fields. Keeping them introduce a conflicting match. For
> example num_of_vlans=0 and cvlan_exist=1.

Such conflict is simple to validate and reject.
Even if num_of_vlans is removed, we can still get conflict svlan_exist=1, cvlan_exist=0.
The different fields are proposed to allow flexible match on different VLAN attributes.
Every PMD can choose to support any or none of them.

> >   };
> >
> >   /** Default mask for RTE_FLOW_ITEM_TYPE_ETH. */


More information about the dev mailing list