[dpdk-dev] [PATCH 21/32] net/dpaa2: add rss flow distribution
Ferruh Yigit
ferruh.yigit at intel.com
Tue Dec 6 20:50:05 CET 2016
On 12/4/2016 6:17 PM, Hemant Agrawal wrote:
> Signed-off-by: Hemant Agrawal <hemant.agrawal at nxp.com>
> ---
> doc/guides/nics/features/dpaa2.ini | 1 +
> drivers/net/dpaa2/Makefile | 3 +-
> drivers/net/dpaa2/base/dpaa2_hw_dpni.c | 287 +++++++++++++++++++++++++++++++++
> drivers/net/dpaa2/base/dpaa2_hw_dpni.h | 15 ++
> drivers/net/dpaa2/base/dpaa2_hw_pvt.h | 2 +
> drivers/net/dpaa2/dpaa2_ethdev.c | 31 +++-
> 6 files changed, 334 insertions(+), 5 deletions(-)
> create mode 100644 drivers/net/dpaa2/base/dpaa2_hw_dpni.c
>
> diff --git a/doc/guides/nics/features/dpaa2.ini b/doc/guides/nics/features/dpaa2.ini
> index 0b59725..20152a0 100644
> --- a/doc/guides/nics/features/dpaa2.ini
> +++ b/doc/guides/nics/features/dpaa2.ini
> @@ -5,6 +5,7 @@
> ;
> [Features]
> Queue start/stop = Y
> +RSS hash = Y
> Linux VFIO = Y
> ARMv8 = Y
> Usage doc = Y
> diff --git a/drivers/net/dpaa2/Makefile b/drivers/net/dpaa2/Makefile
> index 45e28d2..4ed0de3 100644
> --- a/drivers/net/dpaa2/Makefile
> +++ b/drivers/net/dpaa2/Makefile
> @@ -57,10 +57,11 @@ LIBABIVER := 1
>
> SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += base/dpaa2_hw_dpio.c
> SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += base/dpaa2_hw_dpbp.c
> +SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += base/dpaa2_hw_dpni.c
>
> SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += dpaa2_vfio.c
> -# Interfaces with DPDK
> SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += dpaa2_bus.c
> +# Interfaces with DPDK
These kind of things can be fixed where it is introduced at first place.
> SRCS-$(CONFIG_RTE_LIBRTE_DPAA2_PMD) += dpaa2_ethdev.c
>
> # library dependencies
> diff --git a/drivers/net/dpaa2/base/dpaa2_hw_dpni.c b/drivers/net/dpaa2/base/dpaa2_hw_dpni.c
> new file mode 100644
> index 0000000..b26d5a7
> --- /dev/null
> +++ b/drivers/net/dpaa2/base/dpaa2_hw_dpni.c
> @@ -0,0 +1,287 @@
> +/*-
> + * BSD LICENSE
> + *
> + * Copyright (c) 2016 Freescale Semiconductor, Inc. All rights reserved.
> + * Copyright (c) 2016 NXP. All rights reserved.
> + *
> + * Redistribution and use in source and binary forms, with or without
> + * modification, are permitted provided that the following conditions
> + * are met:
> + *
> + * * Redistributions of source code must retain the above copyright
> + * notice, this list of conditions and the following disclaimer.
> + * * Redistributions in binary form must reproduce the above copyright
> + * notice, this list of conditions and the following disclaimer in
> + * the documentation and/or other materials provided with the
> + * distribution.
> + * * Neither the name of Freescale Semiconductor, Inc nor the names of its
> + * contributors may be used to endorse or promote products derived
> + * from this software without specific prior written permission.
> + *
> + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
> + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
> + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
> + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
> + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
> + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
> + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
> + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
> + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
> + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
> + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
> + */
> +
> +#include <time.h>
> +#include <net/if.h>
> +
> +#include <rte_mbuf.h>
> +#include <rte_ethdev.h>
> +#include <rte_malloc.h>
> +#include <rte_memcpy.h>
> +#include <rte_string_fns.h>
> +#include <rte_cycles.h>
> +#include <rte_kvargs.h>
> +#include <rte_dev.h>
> +#include <rte_ethdev.h>
> +
> +#include "dpaa2_logs.h"
> +
> +#include <base/dpaa2_hw_pvt.h>
> +#include <base/dpaa2_hw_dpni.h>
> +
> +static void
> +dpaa2_distset_to_dpkg_profile_cfg(
> + uint32_t req_dist_set,
> + struct dpkg_profile_cfg *kg_cfg);
> +
> +int
> +dpaa2_setup_flow_dist(struct rte_eth_dev *eth_dev,
> + uint32_t req_dist_set)
> +{
> + struct dpaa2_dev_priv *priv = eth_dev->data->dev_private;
> + struct fsl_mc_io *dpni = priv->hw;
> + struct dpni_rx_tc_dist_cfg tc_cfg;
> + struct dpkg_profile_cfg kg_cfg;
> + void *p_params;
> + int ret, tc_index = 0;
> +
> + p_params = rte_malloc(
> + NULL, DIST_PARAM_IOVA_SIZE, RTE_CACHE_LINE_SIZE);
> + if (!p_params) {
> + RTE_LOG(ERR, PMD, "Memory unavaialble\n");
> + return -ENOMEM;
> + }
> + memset(p_params, 0, DIST_PARAM_IOVA_SIZE);
> + memset(&tc_cfg, 0, sizeof(struct dpni_rx_tc_dist_cfg));
> +
> + dpaa2_distset_to_dpkg_profile_cfg(req_dist_set, &kg_cfg);
> + tc_cfg.key_cfg_iova = (uint64_t)(DPAA2_VADDR_TO_IOVA(p_params));
> + tc_cfg.dist_size = eth_dev->data->nb_rx_queues;
> + tc_cfg.dist_mode = DPNI_DIST_MODE_HASH;
> +
> + ret = dpni_prepare_key_cfg(&kg_cfg, p_params);
> + if (ret) {
> + RTE_LOG(ERR, PMD, "Unable to prepare extract parameters\n");
> + rte_free(p_params);
> + return ret;
> + }
> +
> + ret = dpni_set_rx_tc_dist(dpni, CMD_PRI_LOW, priv->token, tc_index,
> + &tc_cfg);
> + rte_free(p_params);
> + if (ret) {
> + RTE_LOG(ERR, PMD, "Setting distribution for Rx failed with"
> + " err code: %d\n", ret);
> + return ret;
> + }
> +
> + return 0;
> +}
> +
> +int dpaa2_remove_flow_dist(
> + struct rte_eth_dev *eth_dev,
> + uint8_t tc_index)
> +{
> + struct dpaa2_dev_priv *priv = eth_dev->data->dev_private;
> + struct fsl_mc_io *dpni = priv->hw;
> + struct dpni_rx_tc_dist_cfg tc_cfg;
> + struct dpkg_profile_cfg kg_cfg;
> + void *p_params;
> + int ret;
> +
> + p_params = rte_malloc(
> + NULL, DIST_PARAM_IOVA_SIZE, RTE_CACHE_LINE_SIZE);
> + if (!p_params) {
> + RTE_LOG(ERR, PMD, "Memory unavaialble\n");
> + return -ENOMEM;
> + }
> + memset(p_params, 0, DIST_PARAM_IOVA_SIZE);
> + memset(&tc_cfg, 0, sizeof(struct dpni_rx_tc_dist_cfg));
> +
> + tc_cfg.key_cfg_iova = (uint64_t)(DPAA2_VADDR_TO_IOVA(p_params));
> + tc_cfg.dist_size = 0;
> + tc_cfg.dist_mode = DPNI_DIST_MODE_NONE;
> +
> + ret = dpni_prepare_key_cfg(&kg_cfg, p_params);
> + if (ret) {
> + RTE_LOG(ERR, PMD, "Unable to prepare extract parameters\n");
> + rte_free(p_params);
> + return ret;
> + }
> +
> + ret = dpni_set_rx_tc_dist(dpni, CMD_PRI_LOW, priv->token, tc_index,
> + &tc_cfg);
> + rte_free(p_params);
> + if (ret) {
> + RTE_LOG(ERR, PMD, "Setting distribution for Rx failed with"
> + " err code: %d\n", ret);
> + return ret;
> + }
> + return ret;
> +}
> +
> +static void
> +dpaa2_distset_to_dpkg_profile_cfg(
> + uint32_t req_dist_set,
> + struct dpkg_profile_cfg *kg_cfg)
> +{
> + uint32_t loop = 0, i = 0, dist_field = 0;
> + int l2_configured = 0, l3_configured = 0;
> + int l4_configured = 0, sctp_configured = 0;
> +
> + memset(kg_cfg, 0, sizeof(struct dpkg_profile_cfg));
> + while (req_dist_set) {
> + if (req_dist_set % 2 != 0) {
> + dist_field = 1U << loop;
> + switch (dist_field) {
> + case ETH_RSS_L2_PAYLOAD:
> +
> + if (l2_configured)
> + break;
> + l2_configured = 1;
> +
> + kg_cfg->extracts[i].extract.from_hdr.prot =
> + NET_PROT_ETH;
> + kg_cfg->extracts[i].extract.from_hdr.field =
> + NH_FLD_ETH_TYPE;
> + kg_cfg->extracts[i].type =
> + DPKG_EXTRACT_FROM_HDR;
> + kg_cfg->extracts[i].extract.from_hdr.type =
> + DPKG_FULL_FIELD;
> + i++;
> + break;
> +
> + case ETH_RSS_IPV4:
> + case ETH_RSS_FRAG_IPV4:
> + case ETH_RSS_NONFRAG_IPV4_OTHER:
> + case ETH_RSS_IPV6:
> + case ETH_RSS_FRAG_IPV6:
> + case ETH_RSS_NONFRAG_IPV6_OTHER:
> + case ETH_RSS_IPV6_EX:
> +
> + if (l3_configured)
> + break;
> + l3_configured = 1;
> +
> + kg_cfg->extracts[i].extract.from_hdr.prot =
> + NET_PROT_IP;
> + kg_cfg->extracts[i].extract.from_hdr.field =
> + NH_FLD_IP_SRC;
> + kg_cfg->extracts[i].type =
> + DPKG_EXTRACT_FROM_HDR;
> + kg_cfg->extracts[i].extract.from_hdr.type =
> + DPKG_FULL_FIELD;
> + i++;
> +
> + kg_cfg->extracts[i].extract.from_hdr.prot =
> + NET_PROT_IP;
> + kg_cfg->extracts[i].extract.from_hdr.field =
> + NH_FLD_IP_DST;
> + kg_cfg->extracts[i].type =
> + DPKG_EXTRACT_FROM_HDR;
> + kg_cfg->extracts[i].extract.from_hdr.type =
> + DPKG_FULL_FIELD;
> + i++;
> +
> + kg_cfg->extracts[i].extract.from_hdr.prot =
> + NET_PROT_IP;
> + kg_cfg->extracts[i].extract.from_hdr.field =
> + NH_FLD_IP_PROTO;
> + kg_cfg->extracts[i].type =
> + DPKG_EXTRACT_FROM_HDR;
> + kg_cfg->extracts[i].extract.from_hdr.type =
> + DPKG_FULL_FIELD;
> + kg_cfg->num_extracts++;
> + i++;
> + break;
> +
> + case ETH_RSS_NONFRAG_IPV4_TCP:
> + case ETH_RSS_NONFRAG_IPV6_TCP:
> + case ETH_RSS_NONFRAG_IPV4_UDP:
> + case ETH_RSS_NONFRAG_IPV6_UDP:
> + case ETH_RSS_IPV6_TCP_EX:
> + case ETH_RSS_IPV6_UDP_EX:
> +
> + if (l4_configured)
> + break;
> + l4_configured = 1;
> +
> + kg_cfg->extracts[i].extract.from_hdr.prot =
> + NET_PROT_TCP;
> + kg_cfg->extracts[i].extract.from_hdr.field =
> + NH_FLD_TCP_PORT_SRC;
> + kg_cfg->extracts[i].type =
> + DPKG_EXTRACT_FROM_HDR;
> + kg_cfg->extracts[i].extract.from_hdr.type =
> + DPKG_FULL_FIELD;
> + i++;
> +
> + kg_cfg->extracts[i].extract.from_hdr.prot =
> + NET_PROT_TCP;
> + kg_cfg->extracts[i].extract.from_hdr.field =
> + NH_FLD_TCP_PORT_SRC;
> + kg_cfg->extracts[i].type =
> + DPKG_EXTRACT_FROM_HDR;
> + kg_cfg->extracts[i].extract.from_hdr.type =
> + DPKG_FULL_FIELD;
> + i++;
> + break;
> +
> + case ETH_RSS_NONFRAG_IPV4_SCTP:
> + case ETH_RSS_NONFRAG_IPV6_SCTP:
> +
> + if (sctp_configured)
> + break;
> + sctp_configured = 1;
> +
> + kg_cfg->extracts[i].extract.from_hdr.prot =
> + NET_PROT_SCTP;
> + kg_cfg->extracts[i].extract.from_hdr.field =
> + NH_FLD_SCTP_PORT_SRC;
> + kg_cfg->extracts[i].type =
> + DPKG_EXTRACT_FROM_HDR;
> + kg_cfg->extracts[i].extract.from_hdr.type =
> + DPKG_FULL_FIELD;
> + i++;
> +
> + kg_cfg->extracts[i].extract.from_hdr.prot =
> + NET_PROT_SCTP;
> + kg_cfg->extracts[i].extract.from_hdr.field =
> + NH_FLD_SCTP_PORT_DST;
> + kg_cfg->extracts[i].type =
> + DPKG_EXTRACT_FROM_HDR;
> + kg_cfg->extracts[i].extract.from_hdr.type =
> + DPKG_FULL_FIELD;
> + i++;
> + break;
> +
> + default:
> + PMD_DRV_LOG(WARNING, "Bad flow distribution"
> + " option %x\n", dist_field);
> + }
> + }
> + req_dist_set = req_dist_set >> 1;
> + loop++;
> + }
> + kg_cfg->num_extracts = i;
> +}
> diff --git a/drivers/net/dpaa2/base/dpaa2_hw_dpni.h b/drivers/net/dpaa2/base/dpaa2_hw_dpni.h
> index 197fd28..c109396 100644
> --- a/drivers/net/dpaa2/base/dpaa2_hw_dpni.h
> +++ b/drivers/net/dpaa2/base/dpaa2_hw_dpni.h
> @@ -37,15 +37,22 @@
> #include <fsl_dpni.h>
> #include <fsl_mc_sys.h>
>
> +#define MAX_TCS DPNI_MAX_TC
> #define MAX_RX_QUEUES 16
> #define MAX_TX_QUEUES 16
>
> +/*! Maximum number of flow distributions per traffic class */
> +#define MAX_DIST_PER_TC 16
> +
> /*default tc to be used for ,congestion, distribution etc configuration. */
> #define DPAA2_DEF_TC 0
>
> /*! Global MCP list */
> extern void *(*mcp_ptr_list);
>
> +/* Size of the input SMMU mapped memory required by MC */
> +#define DIST_PARAM_IOVA_SIZE 256
> +
> struct dpaa2_dev_priv {
> void *hw;
> int32_t hw_id;
> @@ -56,7 +63,15 @@ struct dpaa2_dev_priv {
> void *rx_vq[MAX_RX_QUEUES];
> void *tx_vq[MAX_TX_QUEUES];
>
> + uint16_t num_dist_per_tc[MAX_TCS];
> uint8_t num_tc;
> uint8_t flags; /*dpaa2 config flags */
> };
> +
> +int dpaa2_setup_flow_dist(struct rte_eth_dev *eth_dev,
> + uint32_t req_dist_set);
> +
> +int dpaa2_remove_flow_dist(struct rte_eth_dev *eth_dev,
> + uint8_t tc_index);
> +
> #endif /* _DPAA2_DPNI_H_ */
> diff --git a/drivers/net/dpaa2/base/dpaa2_hw_pvt.h b/drivers/net/dpaa2/base/dpaa2_hw_pvt.h
> index 867611f..abc70ac 100644
> --- a/drivers/net/dpaa2/base/dpaa2_hw_pvt.h
> +++ b/drivers/net/dpaa2/base/dpaa2_hw_pvt.h
> @@ -93,4 +93,6 @@ struct dpaa2_queue {
> /*! Global MCP list */
> extern void *(*mcp_ptr_list);
>
> +#define DPAA2_VADDR_TO_IOVA(_vaddr) (_vaddr)
> +
> #endif
> diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
> index 45c3f8f..094296a 100644
> --- a/drivers/net/dpaa2/dpaa2_ethdev.c
> +++ b/drivers/net/dpaa2/dpaa2_ethdev.c
> @@ -113,7 +113,8 @@
> }
>
> vq_id = 0;
> - for (dist_idx = 0; dist_idx < priv->nb_rx_queues; dist_idx++) {
> + for (dist_idx = 0; dist_idx < priv->num_dist_per_tc[DPAA2_DEF_TC];
> + dist_idx++) {
> mcq = (struct dpaa2_queue *)priv->rx_vq[vq_id];
> mcq->tc_index = DPAA2_DEF_TC;
> mcq->flow_id = dist_idx;
> @@ -139,6 +140,7 @@
> {
> struct rte_eth_dev_data *data = dev->data;
> struct rte_eth_conf *eth_conf = &data->dev_conf;
> + int ret;
>
> PMD_INIT_FUNC_TRACE();
>
> @@ -150,6 +152,18 @@
> return -1;
> }
>
> + if (eth_conf->rxmode.mq_mode == ETH_MQ_RX_RSS) {
> + /* Return in case number of Rx queues is 1 */
> + if (data->nb_rx_queues == 1)
> + return 0;
> + ret = dpaa2_setup_flow_dist(dev,
> + eth_conf->rx_adv_conf.rss_conf.rss_hf);
> + if (ret) {
> + PMD_INIT_LOG(ERR, "unable to set flow distribution."
> + "please check queue config\n");
> + return ret;
> + }
> + }
> return 0;
> }
>
> @@ -181,7 +195,7 @@
> dpaa2_q->mb_pool = mb_pool; /**< mbuf pool to populate RX ring. */
>
> /*Get the tc id and flow id from given VQ id*/
> - flow_id = rx_queue_id;
> + flow_id = rx_queue_id % priv->num_dist_per_tc[dpaa2_q->tc_index];
> memset(&cfg, 0, sizeof(struct dpni_queue));
>
> options = options | DPNI_QUEUE_OPT_USER_CTX;
> @@ -371,7 +385,7 @@
> struct fsl_mc_io *dpni_dev;
> struct dpni_attr attr;
> struct dpaa2_dev_priv *priv = eth_dev->data->dev_private;
> - int ret, hw_id;
> + int i, ret, hw_id;
>
> PMD_INIT_FUNC_TRACE();
>
> @@ -413,7 +427,16 @@
> }
>
> priv->num_tc = attr.num_tcs;
> - priv->nb_rx_queues = attr.num_queues;
> + for (i = 0; i < attr.num_tcs; i++) {
> + priv->num_dist_per_tc[i] = attr.num_queues;
> + break;
> + }
> +
> + /* Distribution is per Tc only,
> + * so choosing RX queues from default TC only
> + */
> + priv->nb_rx_queues = priv->num_dist_per_tc[DPAA2_DEF_TC];
> +
> priv->nb_tx_queues = attr.num_queues;
>
> eth_dev->data->nb_rx_queues = priv->nb_rx_queues;
>
More information about the dev
mailing list