Re: [dpdk-dev] [PATCH v3 5/5] ixgbe: support VxLAN & NVGRE TX checksum off-load

["Lu, Wenzhuo" <wenzhuo.lu@intel.com>]

Hi Konstantin,


> -----Original Message-----
> From: Ananyev, Konstantin
> Sent: Thursday, February 25, 2016 10:53 PM
> To: Lu, Wenzhuo; dev@dpdk.org
> Subject: RE: [dpdk-dev] [PATCH v3 5/5] ixgbe: support VxLAN & NVGRE TX
> checksum off-load
> 
> Hi Wenzhuo,
> 
> > -----Original Message-----
> > From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Wenzhuo Lu
> > Sent: Thursday, February 18, 2016 3:18 AM
> > To: dev@dpdk.org
> > Subject: [dpdk-dev] [PATCH v3 5/5] ixgbe: support VxLAN & NVGRE TX
> > checksum off-load
> >
> > The patch add VxLAN & NVGRE TX checksum off-load. When the flag of
> > outer IP header checksum offload is set, we'll set the context
> > descriptor to enable this checksum off-load.
> >
> > Also update release note for VxLAN & NVGRE checksum off-load support.
> >
> > Signed-off-by: Wenzhuo Lu <wenzhuo.lu@intel.com>
> > ---
> >  doc/guides/rel_notes/release_2_3.rst |  8 ++++++++
> >  drivers/net/ixgbe/ixgbe_ethdev.c     |  4 ++++
> >  drivers/net/ixgbe/ixgbe_rxtx.c       | 18 ++++++++++++++++--
> >  drivers/net/ixgbe/ixgbe_rxtx.h       |  1 +
> >  4 files changed, 29 insertions(+), 2 deletions(-)
> >
> > diff --git a/doc/guides/rel_notes/release_2_3.rst
> > b/doc/guides/rel_notes/release_2_3.rst
> > index 99de186..5dce7fb 100644
> > --- a/doc/guides/rel_notes/release_2_3.rst
> > +++ b/doc/guides/rel_notes/release_2_3.rst
> > @@ -4,6 +4,14 @@ DPDK Release 2.3
> >  New Features
> >  ------------
> >
> > +* **Added support for VxLAN & NVGRE checksum off-load on X550.**
> > +
> > +  * Added support for VxLAN & NVGRE RX/TX checksum off-load on
> > +    X550. RX/TX checksum off-load is provided on both inner and
> > +    outer IP header and TCP header.
> > +  * Added functions to support VxLAN port configuration. The
> > +    default VxLAN port number is 4789 but this can be updated
> > +    programmatically.
> >
> >  Resolved Issues
> >  ---------------
> > diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c
> > b/drivers/net/ixgbe/ixgbe_ethdev.c
> > index fd9751e..3c3aa72 100644
> > --- a/drivers/net/ixgbe/ixgbe_ethdev.c
> > +++ b/drivers/net/ixgbe/ixgbe_ethdev.c
> > @@ -2811,6 +2811,10 @@ ixgbe_dev_info_get(struct rte_eth_dev *dev,
> struct rte_eth_dev_info *dev_info)
> >  		DEV_TX_OFFLOAD_SCTP_CKSUM  |
> >  		DEV_TX_OFFLOAD_TCP_TSO;
> >
> > +	if (hw->mac.type == ixgbe_mac_X550 ||
> > +	    hw->mac.type == ixgbe_mac_X550EM_x)
> > +		dev_info->tx_offload_capa |=
> DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM;
> > +
> >  	dev_info->default_rxconf = (struct rte_eth_rxconf) {
> >  		.rx_thresh = {
> >  			.pthresh = IXGBE_DEFAULT_RX_PTHRESH, diff --git
> > a/drivers/net/ixgbe/ixgbe_rxtx.c b/drivers/net/ixgbe/ixgbe_rxtx.c
> > index 4ed9ddd..8eab768 100644
> > --- a/drivers/net/ixgbe/ixgbe_rxtx.c
> > +++ b/drivers/net/ixgbe/ixgbe_rxtx.c
> > @@ -85,7 +85,8 @@
> >  		PKT_TX_VLAN_PKT |		 \
> >  		PKT_TX_IP_CKSUM |		 \
> >  		PKT_TX_L4_MASK |		 \
> > -		PKT_TX_TCP_SEG)
> > +		PKT_TX_TCP_SEG |		 \
> > +		PKT_TX_OUTER_IP_CKSUM)
> >
> >  static inline struct rte_mbuf *
> >  rte_rxmbuf_alloc(struct rte_mempool *mp) @@ -364,6 +365,7 @@
> > ixgbe_set_xmit_ctx(struct ixgbe_tx_queue *txq,
> >  	uint32_t ctx_idx;
> >  	uint32_t vlan_macip_lens;
> >  	union ixgbe_tx_offload tx_offload_mask;
> > +	uint32_t seqnum_seed = 0;
> >
> >  	ctx_idx = txq->ctx_curr;
> >  	tx_offload_mask.data = 0;
> > @@ -430,6 +432,15 @@ ixgbe_set_xmit_ctx(struct ixgbe_tx_queue *txq,
> >  		}
> >  	}
> >
> > +	if (ol_flags & PKT_TX_OUTER_IP_CKSUM) {
> > +		tx_offload_mask.outer_l3_len |= ~0;
> > +		tx_offload_mask.l2_len |= ~0;
> > +		seqnum_seed |= tx_offload.outer_l3_len
> > +			       << IXGBE_ADVTXD_OUTER_IPLEN;
> > +		seqnum_seed |= tx_offload.l2_len
> > +			       << IXGBE_ADVTXD_TUNNEL_LEN;
> > +	}
> > +
> >  	txq->ctx_cache[ctx_idx].flags = ol_flags;
> >  	txq->ctx_cache[ctx_idx].tx_offload.data  =
> >  		tx_offload_mask.data & tx_offload.data; @@ -441,7 +452,7
> @@
> > ixgbe_set_xmit_ctx(struct ixgbe_tx_queue *txq,
> >  	vlan_macip_lens |= ((uint32_t)tx_offload.vlan_tci <<
> IXGBE_ADVTXD_VLAN_SHIFT);
> >  	ctx_txd->vlan_macip_lens = rte_cpu_to_le_32(vlan_macip_lens);
> >  	ctx_txd->mss_l4len_idx   = rte_cpu_to_le_32(mss_l4len_idx);
> > -	ctx_txd->seqnum_seed     = 0;
> > +	ctx_txd->seqnum_seed     = seqnum_seed;
> >  }
> >
> >  /*
> > @@ -492,6 +503,8 @@ tx_desc_ol_flags_to_cmdtype(uint64_t ol_flags)
> >  		cmdtype |= IXGBE_ADVTXD_DCMD_VLE;
> >  	if (ol_flags & PKT_TX_TCP_SEG)
> >  		cmdtype |= IXGBE_ADVTXD_DCMD_TSE;
> > +	if (ol_flags & PKT_TX_OUTER_IP_CKSUM)
> > +		cmdtype |= (1 << IXGBE_ADVTXD_OUTERIPCS_SHIFT);
> >  	return cmdtype;
> >  }
> >
> > @@ -623,6 +636,7 @@ ixgbe_xmit_pkts(void *tx_queue, struct rte_mbuf
> **tx_pkts,
> >  			tx_offload.l4_len = tx_pkt->l4_len;
> >  			tx_offload.vlan_tci = tx_pkt->vlan_tci;
> >  			tx_offload.tso_segsz = tx_pkt->tso_segsz;
> > +			tx_offload.outer_l3_len = tx_pkt->outer_l3_len;
> >
> >  			/* If new context need be built or reuse the exist ctx. */
> >  			ctx = what_advctx_update(txq, tx_ol_req, diff --git
> > a/drivers/net/ixgbe/ixgbe_rxtx.h b/drivers/net/ixgbe/ixgbe_rxtx.h
> > index 475a800..26ff35f 100644
> > --- a/drivers/net/ixgbe/ixgbe_rxtx.h
> > +++ b/drivers/net/ixgbe/ixgbe_rxtx.h
> > @@ -168,6 +168,7 @@ union ixgbe_tx_offload {
> >  		uint64_t l2_len:7; /**< L2 (MAC) Header Length. */
> >  		uint64_t l3_len:9; /**< L3 (IP) Header Length. */
> >  		uint64_t l4_len:8; /**< L4 (TCP/UDP) Header Length. */
> > +		uint64_t outer_l3_len:8; /**< Outer L3 (IP) Hdr Length. */
> 
> I think you are missing outer_l2_len here.
> Right now, for tunnel packet you setup:
> TUNNELLEN=mbuf->l2_len;
> MACLEN=mbuf->l2_len;
> Though as I could read x550 spec it should be:
> MACLEN=mbuf->outer_l2_len;
> TUNNELLEN=mbuf->l2_len;
> 
> Konstantin
Yes, you're right. I cannot set the value of MACLEN to l2_len. I still need outer_l2_len. Thanks for the comments. I'll correct this mistake.

> 
> >  		uint64_t tso_segsz:16; /**< TCP TSO segment size */
> >  		uint64_t vlan_tci:16;
> >  		/**< VLAN Tag Control Identifier (CPU order). */
> > --
> > 1.9.3