DPDK patches and discussions
 help / color / mirror / Atom feed
From: Feifei Wang <feifei.wang2@arm.com>
To: Jerin Jacob <jerinj@marvell.com>,
	Ruifeng Wang <ruifeng.wang@arm.com>, Jeff Guo <jia.guo@intel.com>,
	Haiyue Wang <haiyue.wang@intel.com>
Cc: dev@dpdk.org, nd@arm.com, Feifei Wang <feifei.wang2@arm.com>
Subject: [dpdk-dev] [PATCH v2 2/4] net/ixgbe: support bad checksum flag for NEON vector
Date: Wed, 18 Nov 2020 04:48:57 -0600
Message-ID: <20201118104859.29047-3-feifei.wang2@arm.com> (raw)
In-Reply-To: <20201118104859.29047-1-feifei.wang2@arm.com>

Updated desc_to_olflags_v() to support PKT_RX_IP_CKSUM_BAD and
PKT_RX_L4_CKSUM_BAD in the ol_flags of the mbuf.

And then the NEON vector RX function can be called with hw_ip_checksum
enabled.

Signed-off-by: Feifei Wang <feifei.wang2@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
---
 drivers/net/ixgbe/ixgbe_rxtx_vec_neon.c | 47 +++++++++++++++++++------
 1 file changed, 36 insertions(+), 11 deletions(-)

diff --git a/drivers/net/ixgbe/ixgbe_rxtx_vec_neon.c b/drivers/net/ixgbe/ixgbe_rxtx_vec_neon.c
index e6d877af9..4d6f057e7 100644
--- a/drivers/net/ixgbe/ixgbe_rxtx_vec_neon.c
+++ b/drivers/net/ixgbe/ixgbe_rxtx_vec_neon.c
@@ -87,6 +87,8 @@ desc_to_olflags_v(uint8x16x2_t sterr_tmp1, uint8x16x2_t sterr_tmp2,
 {
 	uint8x16_t ptype;
 	uint8x16_t vtag;
+	uint8x16_t temp_csum;
+	uint32x4_t csum = {0, 0, 0, 0};
 
 	union {
 		uint8_t e[4];
@@ -105,26 +107,51 @@ desc_to_olflags_v(uint8x16x2_t sterr_tmp1, uint8x16x2_t sterr_tmp2,
 			PKT_RX_RSS_HASH, 0, 0, 0,
 			0, 0, 0, PKT_RX_FDIR};
 
-	const uint8x16_t vlan_msk = {
+	/* mask everything except vlan present and l4/ip csum error */
+	const uint8x16_t vlan_csum_msk = {
 			IXGBE_RXD_STAT_VP, IXGBE_RXD_STAT_VP,
 			IXGBE_RXD_STAT_VP, IXGBE_RXD_STAT_VP,
 			0, 0, 0, 0,
 			0, 0, 0, 0,
-			0, 0, 0, 0};
-
-	const uint8x16_t vlan_map = {
+			(IXGBE_RXDADV_ERR_TCPE | IXGBE_RXDADV_ERR_IPE) >> 24,
+			(IXGBE_RXDADV_ERR_TCPE | IXGBE_RXDADV_ERR_IPE) >> 24,
+			(IXGBE_RXDADV_ERR_TCPE | IXGBE_RXDADV_ERR_IPE) >> 24,
+			(IXGBE_RXDADV_ERR_TCPE | IXGBE_RXDADV_ERR_IPE) >> 24};
+
+	/* map vlan present (0x8), IPE (0x2), L4E (0x1) to ol_flags */
+	const uint8x16_t vlan_csum_map = {
+			0,
+			PKT_RX_L4_CKSUM_BAD,
+			PKT_RX_IP_CKSUM_BAD,
+			PKT_RX_IP_CKSUM_BAD | PKT_RX_L4_CKSUM_BAD,
 			0, 0, 0, 0,
-			0, 0, 0, 0,
-			vlan_flags, 0, 0, 0,
+			vlan_flags,
+			vlan_flags | PKT_RX_L4_CKSUM_BAD,
+			vlan_flags | PKT_RX_IP_CKSUM_BAD,
+			vlan_flags | PKT_RX_IP_CKSUM_BAD | PKT_RX_L4_CKSUM_BAD,
 			0, 0, 0, 0};
 
 	ptype = vzipq_u8(sterr_tmp1.val[0], sterr_tmp2.val[0]).val[0];
 	ptype = vandq_u8(ptype, rsstype_msk);
 	ptype = vqtbl1q_u8(rss_flags, ptype);
 
-	/* extract vlan_flags from IXGBE_RXD_STAT_VP bits of staterr */
-	vtag = vandq_u8(staterr, vlan_msk);
-	vtag = vqtbl1q_u8(vlan_map, vtag);
+	/* extract vlan_flags and csum_error from staterr */
+	vtag = vandq_u8(staterr, vlan_csum_msk);
+
+	/* csum bits are in the most significant, to use shuffle we need to
+	 * shift them. Change mask from 0xc0 to 0x03.
+	 */
+	temp_csum = vshrq_n_u8(vtag, 6);
+
+	/* 'OR' the most significant 32 bits containing the checksum
+	 * flags with the vlan present flags
+	 * Then bits layout of each lane(8bits) will be 'xxxx,VP,x,IPE,L4E'
+	 */
+	csum = vsetq_lane_u32(vgetq_lane_u32(vreinterpretq_u32_u8(temp_csum), 3), csum, 0);
+	vtag = vorrq_u8(vreinterpretq_u8_u32(csum), vtag);
+
+	/* convert VP, IPE, L4E to ol_flags */
+	vtag = vqtbl1q_u8(vlan_csum_map, vtag);
 	vtag = vorrq_u8(ptype, vtag);
 
 	vol.word = vgetq_lane_u32(vreinterpretq_u32_u8(vtag), 0);
@@ -391,7 +418,6 @@ _recv_raw_pkts_vec(struct ixgbe_rx_queue *rxq, struct rte_mbuf **rx_pkts,
  * Notice:
  * - nb_pkts < RTE_IXGBE_DESCS_PER_LOOP, just return no packet
  * - floor align nb_pkts to a RTE_IXGBE_DESC_PER_LOOP power-of-two
- * - don't support ol_flags for rss and csum err
  */
 uint16_t
 ixgbe_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
@@ -404,7 +430,6 @@ ixgbe_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
  * vPMD receive routine that reassembles scattered packets
  *
  * Notice:
- * - don't support ol_flags for rss and csum err
  * - nb_pkts < RTE_IXGBE_DESCS_PER_LOOP, just return no packet
  * - floor align nb_pkts to a RTE_IXGBE_DESC_PER_LOOP power-of-two
  */
-- 
2.17.1


  parent reply	other threads:[~2020-11-18 10:50 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-11-18  7:35 [dpdk-dev] [PATCH v1 0/4] Enable Checksum Offloading " Feifei Wang
2020-11-18  7:35 ` [dpdk-dev] [PATCH v1 1/4] net/ixgbe: add VLAN stripping support for Arm Feifei Wang
2020-11-18  8:34   ` Wang, Haiyue
2020-11-18 10:51     ` [dpdk-dev] 回复: " Feifei Wang
2020-11-18  7:35 ` [dpdk-dev] [PATCH v1 2/4] net/ixgbe: support checksum flags for NEON vector Feifei Wang
2020-11-18  7:35 ` [dpdk-dev] [PATCH v1 3/4] net/ixgbe: implement good checksum flag " Feifei Wang
2020-11-18  7:35 ` [dpdk-dev] [PATCH v1 4/4] net/ixgbe: enable IXGBE NEON vector PMD when CHECKSUM is set Feifei Wang
2020-11-18 10:48 ` [dpdk-dev] [PATCH v2 0/4] Enable Checksum Offloading for NEON vector Feifei Wang
2020-11-18 10:48   ` [dpdk-dev] [PATCH v2 1/4] net/ixgbe: add new flag of stripped VLAN " Feifei Wang
2020-11-18 11:51     ` Wang, Haiyue
2020-11-18 10:48   ` Feifei Wang [this message]
2020-11-18 11:51     ` [dpdk-dev] [PATCH v2 2/4] net/ixgbe: support bad checksum flag " Wang, Haiyue
2020-11-18 10:48   ` [dpdk-dev] [PATCH v2 3/4] net/ixgbe: support good " Feifei Wang
2020-11-18 11:51     ` Wang, Haiyue
2020-11-18 10:48   ` [dpdk-dev] [PATCH v2 4/4] net/ixgbe: enable IXGBE NEON vector when need to checksum Feifei Wang
2020-11-18 11:51     ` Wang, Haiyue
2020-12-11 12:06   ` [dpdk-dev] [PATCH v2 0/4] Enable Checksum Offloading for NEON vector Ferruh Yigit

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20201118104859.29047-3-feifei.wang2@arm.com \
    --to=feifei.wang2@arm.com \
    --cc=dev@dpdk.org \
    --cc=haiyue.wang@intel.com \
    --cc=jerinj@marvell.com \
    --cc=jia.guo@intel.com \
    --cc=nd@arm.com \
    --cc=ruifeng.wang@arm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

DPDK patches and discussions

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://inbox.dpdk.org/dev/0 dev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 dev dev/ https://inbox.dpdk.org/dev \
		dev@dpdk.org
	public-inbox-index dev

Example config snippet for mirrors.
Newsgroup available over NNTP:
	nntp://inbox.dpdk.org/inbox.dpdk.dev


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git