Re: [PATCH v2 1/2] mbuf: introduce new Tx offload flag for MPLS-in-UDP

[Olivier Matz <olivier.matz@6wind.com>]

On Thu, 8 Jun 2017 21:46:00 +0000, "Patil, Harish" <Harish.Patil@cavium.com> wrote:
> >Hi Rasesh,
> >
> >On Wed, 7 Jun 2017 00:43:48 -0700, Rasesh Mody <rasesh.mody@cavium.com>
> >wrote:  
> >> From: Harish Patil <harish.patil@cavium.com>
> >> 
> >> Some PMDs need to know the tunnel type in order to handle advance TX
> >> features. This patch adds a new TX offload flag for MPLS-in-UDP packets.
> >> 
> >> Signed-off-by: Harish Patil <harish.patil@cavium.com>
> >> ---
> >>  lib/librte_mbuf/rte_mbuf.c |    2 ++
> >>  lib/librte_mbuf/rte_mbuf.h |   17 ++++++++++-------
> >>  2 files changed, 12 insertions(+), 7 deletions(-)
> >> 
> >> diff --git a/lib/librte_mbuf/rte_mbuf.c b/lib/librte_mbuf/rte_mbuf.c
> >> index 0e3e36a..c2793fb 100644
> >> --- a/lib/librte_mbuf/rte_mbuf.c
> >> +++ b/lib/librte_mbuf/rte_mbuf.c
> >> @@ -410,6 +410,7 @@ const char *rte_get_tx_ol_flag_name(uint64_t mask)
> >>  	case PKT_TX_TUNNEL_IPIP: return "PKT_TX_TUNNEL_IPIP";
> >>  	case PKT_TX_TUNNEL_GENEVE: return "PKT_TX_TUNNEL_GENEVE";
> >>  	case PKT_TX_MACSEC: return "PKT_TX_MACSEC";
> >> +	case PKT_TX_TUNNEL_MPLSINUDP: return "PKT_TX_TUNNEL_MPLSINUDP";
> >>  	default: return NULL;
> >>  	}
> >>  }
> >> @@ -441,6 +442,7 @@ const char *rte_get_tx_ol_flag_name(uint64_t mask)
> >>  		{ PKT_TX_TUNNEL_GENEVE, PKT_TX_TUNNEL_MASK,
> >>  		  "PKT_TX_TUNNEL_NONE" },
> >>  		{ PKT_TX_MACSEC, PKT_TX_MACSEC, NULL },
> >> +		{ PKT_TX_TUNNEL_MPLSINUDP, PKT_TX_TUNNEL_MPLSINUDP, NULL },
> >>  	};
> >>  	const char *name;
> >>  	unsigned int i;
> >> diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
> >> index 1cb0310..27ad421 100644
> >> --- a/lib/librte_mbuf/rte_mbuf.h
> >> +++ b/lib/librte_mbuf/rte_mbuf.h
> >> @@ -197,19 +197,22 @@
> >>   * Offload the MACsec. This flag must be set by the application to
> >>enable
> >>   * this offload feature for a packet to be transmitted.
> >>   */
> >> -#define PKT_TX_MACSEC        (1ULL << 44)
> >> +#define PKT_TX_MACSEC        (1ULL << 43)  
> >
> >I'm not sure it is suitable to change the value of an existing
> >flag, since it breaks the ABI.
> >
> >  
> >>  /**
> >> - * Bits 45:48 used for the tunnel type.
> >> + * Bits 44:48 used for the tunnel type.
> >>   * When doing Tx offload like TSO or checksum, the HW needs to
> >>configure the
> >>   * tunnel type into the HW descriptors.
> >>   */
> >> -#define PKT_TX_TUNNEL_VXLAN   (0x1ULL << 45)
> >> -#define PKT_TX_TUNNEL_GRE     (0x2ULL << 45)
> >> -#define PKT_TX_TUNNEL_IPIP    (0x3ULL << 45)
> >> -#define PKT_TX_TUNNEL_GENEVE  (0x4ULL << 45)
> >> +/**< TX packet with MPLS-in-UDP RFC 7510 header. */
> >> +#define PKT_TX_TUNNEL_MPLSINUDP (0x1ULL << 44)
> >> +
> >> +#define PKT_TX_TUNNEL_VXLAN   (0x2ULL << 44)
> >> +#define PKT_TX_TUNNEL_GRE     (0x3ULL << 44)
> >> +#define PKT_TX_TUNNEL_IPIP    (0x4ULL << 44)
> >> +#define PKT_TX_TUNNEL_GENEVE  (0x5ULL << 45)
> >>  /* add new TX TUNNEL type here */
> >> -#define PKT_TX_TUNNEL_MASK    (0xFULL << 45)
> >> +#define PKT_TX_TUNNEL_MASK    (0x1FULL << 44)
> >>  
> >>  /**
> >>   * Second VLAN insertion (QinQ) flag.  
> >
> >I dont understand why adding
> >#define PKT_TX_TUNNEL_MPLSINUDP (0x5ULL << 45)
> >wouldn't do the job?
> >
> >Currently, the tunnel mask is 0xF << 45, which gives 16 possible values.  
> 
> [Harish]  Hi Olivier,
> Not too sure whether I understand your comment.
> My understanding is that those are bitmapped values for each Tx tunnel
> type in the range [48:45].
> They are not values. So defining PKT_TX_TUNNEL_MPLSINUDP (0x5ULL << 45)
> won’t work.

Currently, we have:

#define PKT_TX_TUNNEL_VXLAN   (0x1ULL << 45)
in binary:     000..000[0001]000..000

#define PKT_TX_TUNNEL_GRE     (0x2ULL << 45)
in binary:     000..000[0010]000..000

#define PKT_TX_TUNNEL_IPIP    (0x3ULL << 45)
in binary:     000..000[0011]000..000

#define PKT_TX_TUNNEL_GENEVE  (0x4ULL << 45)
in binary:     000..000[0100]000..000

So, I'm still saying there's a room for 11 more values.



Olivier