Revert ixgbe driver vRX removal of packet type recognition

VPP still needs to use packet type recognition in vRX path of the
ixgbe driver so ip packet can bypass ethernet-input node processing
for better performance. The shortcoming of packet type recognition
in the vRX path does not affect VPP.

Change-Id: Ibc96021c50402a75a8bc4f80125c664714276cef
Signed-off-by: John Lo <loj@cisco.com>

1 files changed, 128 insertions, 0 deletions

diff --git a/dpdk/dpdk-16.04_patches/0013-Revert-ixgbe-fix-packet-type-from-vector-Rx.patch b/dpdk/dpdk-16.04_patches/0013-Revert-ixgbe-fix-packet-type-from-vector-Rx.patch
new file mode 100755
index 00000000..fca07a6f
--- /dev/null
+++ b/dpdk/dpdk-16.04_patches/0013-Revert-ixgbe-fix-packet-type-from-vector-Rx.patch
@@ -0,0 +1,128 @@
+From 379d9246e60eb289edd911fbc27929cf73a038ac Mon Sep 17 00:00:00 2001
+From: Damjan Marion <damarion@cisco.com>
+Date: Fri, 29 Apr 2016 19:51:35 +0200
+Subject: [PATCH] Revert "ixgbe: fix packet type from vector Rx"
+
+This reverts commit d9a2009a81089093645fea2e04b51dd37edf3e6f.
+---
+ drivers/net/ixgbe/ixgbe_ethdev.c   |  4 +++-
+ drivers/net/ixgbe/ixgbe_rxtx_vec.c | 34 +++++++++++++++++++++++-----------
+ 2 files changed, 26 insertions(+), 12 deletions(-)
+
+diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
+index eec607c..b24cdbd 100644
+--- a/drivers/net/ixgbe/ixgbe_ethdev.c
++++ b/drivers/net/ixgbe/ixgbe_ethdev.c
+@@ -2996,7 +2996,9 @@ ixgbe_dev_supported_ptypes_get(struct rte_eth_dev *dev)
+ 	if (dev->rx_pkt_burst == ixgbe_recv_pkts ||
+ 	    dev->rx_pkt_burst == ixgbe_recv_pkts_lro_single_alloc ||
+ 	    dev->rx_pkt_burst == ixgbe_recv_pkts_lro_bulk_alloc ||
+-	    dev->rx_pkt_burst == ixgbe_recv_pkts_bulk_alloc)
++	    dev->rx_pkt_burst == ixgbe_recv_pkts_bulk_alloc ||
++	    dev->rx_pkt_burst == ixgbe_recv_pkts_vec ||
++	    dev->rx_pkt_burst == ixgbe_recv_scattered_pkts_vec)
+ 		return ptypes;
+ 	return NULL;
+ }
+diff --git a/drivers/net/ixgbe/ixgbe_rxtx_vec.c b/drivers/net/ixgbe/ixgbe_rxtx_vec.c
+index 5040704..ccd93c7 100644
+--- a/drivers/net/ixgbe/ixgbe_rxtx_vec.c
++++ b/drivers/net/ixgbe/ixgbe_rxtx_vec.c
+@@ -220,6 +220,8 @@ _recv_raw_pkts_vec(struct ixgbe_rx_queue *rxq, struct rte_mbuf **rx_pkts,
+ 				0, 0            /* ignore pkt_type field */
+ 			);
+ 	__m128i dd_check, eop_check;
++	__m128i desc_mask = _mm_set_epi32(0xFFFFFFFF, 0xFFFFFFFF,
++					  0xFFFFFFFF, 0xFFFF07F0);
+ 
+ 	/* nb_pkts shall be less equal than RTE_IXGBE_MAX_RX_BURST */
+ 	nb_pkts = RTE_MIN(nb_pkts, RTE_IXGBE_MAX_RX_BURST);
+@@ -257,8 +259,9 @@ _recv_raw_pkts_vec(struct ixgbe_rx_queue *rxq, struct rte_mbuf **rx_pkts,
+ 		13, 12,      /* octet 12~13, 16 bits data_len */
+ 		0xFF, 0xFF,  /* skip high 16 bits pkt_len, zero out */
+ 		13, 12,      /* octet 12~13, low 16 bits pkt_len */
+-		0xFF, 0xFF,  /* skip 32 bit pkt_type */
+-		0xFF, 0xFF
++		0xFF, 0xFF,  /* skip high 16 bits pkt_type */
++		1,           /* octet 1, 8 bits pkt_type field */
++		0            /* octet 0, 4 bits offset 4 pkt_type field */
+ 		);
+ 
+ 	/* Cache is empty -> need to scan the buffer rings, but first move
+@@ -275,6 +278,7 @@ _recv_raw_pkts_vec(struct ixgbe_rx_queue *rxq, struct rte_mbuf **rx_pkts,
+ 	for (pos = 0, nb_pkts_recd = 0; pos < nb_pkts;
+ 			pos += RTE_IXGBE_DESCS_PER_LOOP,
+ 			rxdp += RTE_IXGBE_DESCS_PER_LOOP) {
++		__m128i descs0[RTE_IXGBE_DESCS_PER_LOOP];
+ 		__m128i descs[RTE_IXGBE_DESCS_PER_LOOP];
+ 		__m128i pkt_mb1, pkt_mb2, pkt_mb3, pkt_mb4;
+ 		__m128i zero, staterr, sterr_tmp1, sterr_tmp2;
+@@ -285,7 +289,7 @@ _recv_raw_pkts_vec(struct ixgbe_rx_queue *rxq, struct rte_mbuf **rx_pkts,
+ 
+ 		/* Read desc statuses backwards to avoid race condition */
+ 		/* A.1 load 4 pkts desc */
+-		descs[3] = _mm_loadu_si128((__m128i *)(rxdp + 3));
++		descs0[3] = _mm_loadu_si128((__m128i *)(rxdp + 3));
+ 
+ 		/* B.2 copy 2 mbuf point into rx_pkts  */
+ 		_mm_storeu_si128((__m128i *)&rx_pkts[pos], mbp1);
+@@ -293,10 +297,10 @@ _recv_raw_pkts_vec(struct ixgbe_rx_queue *rxq, struct rte_mbuf **rx_pkts,
+ 		/* B.1 load 1 mbuf point */
+ 		mbp2 = _mm_loadu_si128((__m128i *)&sw_ring[pos+2]);
+ 
+-		descs[2] = _mm_loadu_si128((__m128i *)(rxdp + 2));
++		descs0[2] = _mm_loadu_si128((__m128i *)(rxdp + 2));
+ 		/* B.1 load 2 mbuf point */
+-		descs[1] = _mm_loadu_si128((__m128i *)(rxdp + 1));
+-		descs[0] = _mm_loadu_si128((__m128i *)(rxdp));
++		descs0[1] = _mm_loadu_si128((__m128i *)(rxdp + 1));
++		descs0[0] = _mm_loadu_si128((__m128i *)(rxdp));
+ 
+ 		/* B.2 copy 2 mbuf point into rx_pkts  */
+ 		_mm_storeu_si128((__m128i *)&rx_pkts[pos+2], mbp2);
+@@ -308,6 +312,14 @@ _recv_raw_pkts_vec(struct ixgbe_rx_queue *rxq, struct rte_mbuf **rx_pkts,
+ 			rte_prefetch0(&rx_pkts[pos + 3]->cacheline1);
+ 		}
+ 
++		/* A* mask out 0~3 bits RSS type */
++		descs[3] = _mm_and_si128(descs0[3], desc_mask);
++		descs[2] = _mm_and_si128(descs0[2], desc_mask);
++
++		/* A* mask out 0~3 bits RSS type */
++		descs[1] = _mm_and_si128(descs0[1], desc_mask);
++		descs[0] = _mm_and_si128(descs0[0], desc_mask);
++
+ 		/* avoid compiler reorder optimization */
+ 		rte_compiler_barrier();
+ 
+@@ -315,22 +327,22 @@ _recv_raw_pkts_vec(struct ixgbe_rx_queue *rxq, struct rte_mbuf **rx_pkts,
+ 		pkt_mb4 = _mm_shuffle_epi8(descs[3], shuf_msk);
+ 		pkt_mb3 = _mm_shuffle_epi8(descs[2], shuf_msk);
+ 
+-		/* D.1 pkt 1,2 convert format from desc to pktmbuf */
+-		pkt_mb2 = _mm_shuffle_epi8(descs[1], shuf_msk);
+-		pkt_mb1 = _mm_shuffle_epi8(descs[0], shuf_msk);
+-
+ 		/* C.1 4=>2 filter staterr info only */
+ 		sterr_tmp2 = _mm_unpackhi_epi32(descs[3], descs[2]);
+ 		/* C.1 4=>2 filter staterr info only */
+ 		sterr_tmp1 = _mm_unpackhi_epi32(descs[1], descs[0]);
+ 
+ 		/* set ol_flags with vlan packet type */
+-		desc_to_olflags_v(descs, &rx_pkts[pos]);
++		desc_to_olflags_v(descs0, &rx_pkts[pos]);
+ 
+ 		/* D.2 pkt 3,4 set in_port/nb_seg and remove crc */
+ 		pkt_mb4 = _mm_add_epi16(pkt_mb4, crc_adjust);
+ 		pkt_mb3 = _mm_add_epi16(pkt_mb3, crc_adjust);
+ 
++		/* D.1 pkt 1,2 convert format from desc to pktmbuf */
++		pkt_mb2 = _mm_shuffle_epi8(descs[1], shuf_msk);
++		pkt_mb1 = _mm_shuffle_epi8(descs[0], shuf_msk);
++
+ 		/* C.2 get 4 pkts staterr value  */
+ 		zero = _mm_xor_si128(dd_check, dd_check);
+ 		staterr = _mm_unpacklo_epi32(sterr_tmp1, sterr_tmp2);
+-- 
+2.7.4
+