Hi Jingjing,
> -----Original Message-----
> From: Wu, Jingjing <jingjing...@intel.com>
> Sent: Saturday, December 19, 2020 10:38 PM
> To: dev@dpdk.org
> Cc: Wu, Jingjing <jingjing...@intel.com>; Xing, Beilei
> <beilei.x...@intel.com>;
> Xia, Chenbo <chenbo....@intel.com>; Lu, Xiuchun <xiuchun...@intel.com>
> Subject: [PATCH v1 2/2] net/iavf: introduce iavf driver on vfio-user client
>
> This patch add a new net driver based on vdev abstraction,
> i.e. iavf_client_ethdev.c. It is using common iavf functions
> to talk with Emulated pci interfaces based on vfio-user.
>
> ----------------------
> | ------------------ |
> | | iavf driver | |----> (iavf_client_ethdev.c)
> | ------------------ |
> | ------------------ |
> | | device emulate | |
> | | vfio-user adapt| |
> | ------------------ |
> ----------------------
> |
> |
> ----------------------
> | vfio-user |
> | client |
> ----------------------
>
> Signed-off-by: Jingjing Wu <jingjing...@intel.com>
> ---
> drivers/common/iavf/iavf_prototype.h | 1 +
> drivers/common/iavf/version.map | 1 +
> drivers/net/iavf/iavf.h | 18 +-
> drivers/net/iavf/iavf_client_ethdev.c | 298 ++++++++++++++++++++++++++
> drivers/net/iavf/iavf_ethdev.c | 26 +--
> drivers/net/iavf/iavf_rxtx.c | 23 +-
> drivers/net/iavf/meson.build | 1 +
> 7 files changed, 347 insertions(+), 21 deletions(-)
> create mode 100644 drivers/net/iavf/iavf_client_ethdev.c
>
> diff --git a/drivers/common/iavf/iavf_prototype.h
> b/drivers/common/iavf/iavf_prototype.h
> index f34e77db0f..3998d26dc0 100644
> --- a/drivers/common/iavf/iavf_prototype.h
> +++ b/drivers/common/iavf/iavf_prototype.h
> @@ -83,6 +83,7 @@ void iavf_destroy_spinlock(struct iavf_spinlock *sp);
> __rte_internal
> void iavf_vf_parse_hw_config(struct iavf_hw *hw,
> struct virtchnl_vf_resource *msg);
> +__rte_internal
> enum iavf_status iavf_vf_reset(struct iavf_hw *hw);
> __rte_internal
> enum iavf_status iavf_aq_send_msg_to_pf(struct iavf_hw *hw,
> diff --git a/drivers/common/iavf/version.map b/drivers/common/iavf/version.map
> index 8808779ab7..4dc2d42196 100644
> --- a/drivers/common/iavf/version.map
> +++ b/drivers/common/iavf/version.map
> @@ -7,6 +7,7 @@ INTERNAL {
> iavf_set_mac_type;
> iavf_shutdown_adminq;
> iavf_vf_parse_hw_config;
> + iavf_vf_reset;
> client_vfio_user_setup;
> client_vfio_user_get_bar_addr;
> iavf_write_addr;
> diff --git a/drivers/net/iavf/iavf.h b/drivers/net/iavf/iavf.h
> index 6d5912d8c1..c34f971721 100644
> --- a/drivers/net/iavf/iavf.h
> +++ b/drivers/net/iavf/iavf.h
> @@ -195,7 +195,10 @@ struct iavf_adapter {
> struct iavf_hw hw;
> struct rte_eth_dev *eth_dev;
> struct iavf_info vf;
> -
> +#ifdef RTE_LIBRTE_IAVF_CLIENT
> + /* used for avf_client driver */
> + struct vfio_device *user_dev;
> +#endif
> bool rx_bulk_alloc_allowed;
> /* For vector PMD */
> bool rx_vec_allowed;
> @@ -231,6 +234,16 @@ iavf_init_adminq_parameter(struct iavf_hw *hw)
> hw->aq.asq_buf_size = IAVF_AQ_BUF_SZ;
> }
>
> +static inline void
> +iavf_disable_irq0(struct iavf_hw *hw)
> +{
> + /* Disable all interrupt types */
> + IAVF_WRITE_REG(hw, IAVF_VFINT_ICR0_ENA1, 0);
> + IAVF_WRITE_REG(hw, IAVF_VFINT_DYN_CTL01,
> + IAVF_VFINT_DYN_CTL01_ITR_INDX_MASK);
> + IAVF_WRITE_FLUSH(hw);
> +}
> +
> static inline uint16_t
> iavf_calc_itr_interval(int16_t interval)
> {
> @@ -284,6 +297,9 @@ _atomic_set_cmd(struct iavf_info *vf, enum virtchnl_ops
> ops)
> return !ret;
> }
>
> +extern const struct eth_dev_ops iavf_eth_dev_ops;
> +
> +int iavf_init_vf(struct rte_eth_dev *dev);
> int iavf_check_api_version(struct iavf_adapter *adapter);
> int iavf_get_vf_resource(struct iavf_adapter *adapter);
> void iavf_handle_virtchnl_msg(struct rte_eth_dev *dev);
> diff --git a/drivers/net/iavf/iavf_client_ethdev.c
> b/drivers/net/iavf/iavf_client_ethdev.c
> new file mode 100644
> index 0000000000..03f759c761
> --- /dev/null
> +++ b/drivers/net/iavf/iavf_client_ethdev.c
> @@ -0,0 +1,298 @@
> +/* SPDX-License-Identifier: BSD-3-Clause
> + * Copyright(c) 2017 Intel Corporation
> + */
> +
> +#include <errno.h>
> +#include <stdint.h>
> +#include <string.h>
> +
> +#include <rte_common.h>
> +#include <rte_ether.h>
> +#include <rte_ethdev_driver.h>
> +#include <rte_ethdev_vdev.h>
> +#include <rte_alarm.h>
> +#include <rte_bus_vdev.h>
> +#include <rte_malloc.h>
> +#include <vfio_user/vfio_user_pci.h>
> +
> +#include "iavf.h"
> +#include "iavf_rxtx.h"
> +
> +static int iavf_client_dev_close(struct rte_eth_dev *dev);
> +static int iavf_client_dev_reset(struct rte_eth_dev *dev);
> +
> +/* set iavf_client_dev_ops to iavf's by default */
> +static struct eth_dev_ops iavf_client_eth_dev_ops;
> +
> +static const char *valid_args[] = {
> +#define AVF_CLIENT_ARG_PATH "path"
> + AVF_CLIENT_ARG_PATH,
> + NULL
> +};
> +
> +/* set up vfio_device for iavf_client*/
> +static int
> +iavf_client_vfio_user_setup(struct rte_eth_dev *dev, const char *path)
> +{
> + struct iavf_adapter *adapter =
> + IAVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
> + struct iavf_hw *hw = IAVF_DEV_PRIVATE_TO_HW(adapter);
> + struct vfio_device *vfio_dev;
> + int max_fds, i;
> +
> + vfio_dev = client_vfio_user_setup(path, dev->device->numa_node);
> + if (vfio_dev == NULL) {
> + printf("Error to create vfio_device for iavf_client\n");
> + return -1;
> + }
> + hw->bus.type = iavf_bus_type_vfio_user;
> +
> + /* Use hw_addr to record dev ptr */
> + hw->hw_addr = (uint8_t *)vfio_dev;
> +
> + hw->back = IAVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
> +
> + if (!dev->intr_handle) {
> + dev->intr_handle = malloc(sizeof(*dev->intr_handle));
> + if (!dev->intr_handle) {
> + PMD_INIT_LOG(ERR, "fail to allocate intr_handle");
> + return -1;
> + }
> + memset(dev->intr_handle, 0, sizeof(*dev->intr_handle));
> + }
> +
> + max_fds = RTE_MIN(RTE_MAX_RXTX_INTR_VEC_ID,
> adapter->vf.num_queue_pairs);
> +
> + /* FD for control has been enabled */
> + dev->intr_handle->fd = -1;
> + dev->intr_handle->type = RTE_INTR_HANDLE_VDEV;
> + dev->intr_handle->max_intr = max_fds + 1;
> + dev->intr_handle->nb_efd = max_fds;
> + for (i = 0; i < max_fds; ++i)
> + dev->intr_handle->efds[i] = vfio_dev->irqfds[i];
> + dev->intr_handle->efd_counter_size = 0;
> +
> + return 0;
> +}
> +
> +
> +static inline void
> +avf_client_init_eth_ops(void)
> +{
> + iavf_client_eth_dev_ops = iavf_eth_dev_ops;
> + /* keep other unchanged */
> + iavf_client_eth_dev_ops.dev_close = iavf_client_dev_close,
> + iavf_client_eth_dev_ops.dev_reset = iavf_client_dev_reset,
> + iavf_client_eth_dev_ops.dev_supported_ptypes_get = NULL;
> + iavf_client_eth_dev_ops.reta_update = NULL;
> + iavf_client_eth_dev_ops.reta_query = NULL;
> + iavf_client_eth_dev_ops.rss_hash_update = NULL;
> + iavf_client_eth_dev_ops.rss_hash_conf_get = NULL;
> + iavf_client_eth_dev_ops.rx_queue_intr_enable = NULL;
> + iavf_client_eth_dev_ops.rx_queue_intr_disable = NULL;
> +}
> +
> +#define IAVF_CLIENT_ALARM_INTERVAL 50000 /* us */
> +static void
> +iavf_client_dev_alarm_handler(void *param)
> +{
> + struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
> +
> + iavf_handle_virtchnl_msg(dev);
> +
> + rte_eal_alarm_set(IAVF_CLIENT_ALARM_INTERVAL,
> + iavf_client_dev_alarm_handler, dev);
> +}
> +
> +/* init ethdev for the avf client device*/
> +static int
> +iavf_client_eth_init(struct rte_eth_dev *eth_dev)
> +{
> + struct iavf_adapter *adapter =
> + IAVF_DEV_PRIVATE_TO_ADAPTER(eth_dev->data->dev_private);
> + struct iavf_hw *hw = IAVF_DEV_PRIVATE_TO_HW(adapter);
> +
> + /* update eth_dev_op by assigning ops func pointer */
> + avf_client_init_eth_ops();
> + eth_dev->dev_ops = (const struct eth_dev_ops
> *)(&iavf_client_eth_dev_ops);
> +
> + eth_dev->rx_pkt_burst = &iavf_recv_pkts;
> + eth_dev->tx_pkt_burst = &iavf_xmit_pkts;
> + eth_dev->tx_pkt_prepare = &iavf_prep_pkts;
> +
> + hw->back = IAVF_DEV_PRIVATE_TO_ADAPTER(eth_dev->data->dev_private);
> + adapter->eth_dev = eth_dev;
> + adapter->stopped = 1;
> +
I think the ptype table (adapter->ptype_tbl) should be set during init.
Maybe using iavf_get_default_ptype_table().
Thanks!
Chenbo
> + if (iavf_init_vf(eth_dev) != 0) {
> + PMD_INIT_LOG(ERR, "Init vf failed");
> + return -1;
> + }
> +
> + /* copy mac addr */
> + eth_dev->data->mac_addrs = rte_zmalloc(
> + "iavf_client_mac",
> + RTE_ETHER_ADDR_LEN * IAVF_NUM_MACADDR_MAX, 0);
> + if (!eth_dev->data->mac_addrs) {
> + PMD_INIT_LOG(ERR, "Failed to allocate %d bytes needed to"
> + " store MAC addresses",
> + RTE_ETHER_ADDR_LEN * IAVF_NUM_MACADDR_MAX);
> + return -ENOMEM;
> + }
> + /* If the MAC address is not configured by host,
> + * generate a random one.
> + */
> + if (!rte_is_valid_assigned_ether_addr(
> + (struct rte_ether_addr *)hw->mac.addr))
> + rte_eth_random_addr(hw->mac.addr);
> + rte_ether_addr_copy((struct rte_ether_addr *)hw->mac.addr,
> + ð_dev->data->mac_addrs[0]);
> +
> + rte_eal_alarm_set(IAVF_CLIENT_ALARM_INTERVAL,
> + iavf_client_dev_alarm_handler, eth_dev);
> + return 0;
> +}
> +
> +static int
> +iavf_client_dev_reset(struct rte_eth_dev *dev)
> +{
> + struct iavf_hw *hw = IAVF_DEV_PRIVATE_TO_HW(dev->data->dev_private);
> + int ret;
> +
> + rte_eal_alarm_cancel(iavf_client_dev_alarm_handler, dev);
> +
> + iavf_shutdown_adminq(hw);
> + ret = iavf_init_vf(dev);
> +
> + /* send reset msg to PF */
> + iavf_vf_reset(hw);
> + rte_eal_alarm_set(IAVF_CLIENT_ALARM_INTERVAL,
> + iavf_client_dev_alarm_handler, dev);
> +
> + return ret;
> +}
> +
> +static int
> +iavf_client_dev_close(struct rte_eth_dev *dev)
> +{
> + struct iavf_adapter *adapter =
> + IAVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
> + struct iavf_hw *hw = IAVF_DEV_PRIVATE_TO_HW(dev->data->dev_private);
> + struct rte_intr_handle *intr_handle = dev->intr_handle;
> +
> + if (!adapter->stopped) {
> + iavf_stop_queues(dev);
> +
> + if (intr_handle) {
> + /* Disable the interrupt for Rx */
> + rte_intr_efd_disable(intr_handle);
> + /* Rx interrupt vector mapping free */
> + if (intr_handle->intr_vec) {
> + rte_free(intr_handle->intr_vec);
> + intr_handle->intr_vec = NULL;
> + }
> + }
> + /* Remove all mac addrs */
> + iavf_add_del_all_mac_addr(adapter, false);
> + adapter->stopped = 1;
> + }
> + iavf_shutdown_adminq(hw);
> + iavf_disable_irq0(hw);
> + rte_eal_alarm_cancel(iavf_client_dev_alarm_handler, dev);
> +
> + return 0;
> +}
> +
> +static int
> +iavf_client_get_string_arg(const char *key __rte_unused,
> + const char *value, void *extra_args)
> +{
> + if (!value || !extra_args)
> + return -EINVAL;
> +
> + *(char **)extra_args = strdup(value);
> +
> + if (!*(char **)extra_args)
> + return -ENOMEM;
> +
> + return 0;
> +}
> +
> +static int
> +avf_client_pmd_probe(struct rte_vdev_device *vdev)
> +{
> + struct rte_kvargs *kvlist = NULL;
> + struct rte_eth_dev *eth_dev;
> + struct iavf_adapter *adapter;
> + char *path = NULL;
> + int ret;
> +
> + kvlist = rte_kvargs_parse(rte_vdev_device_args(vdev), valid_args);
> + if (!kvlist) {
> + PMD_INIT_LOG(ERR, "error when parsing param");
> + return -EINVAL;
> + }
> +
> + if (rte_kvargs_count(kvlist, AVF_CLIENT_ARG_PATH) == 1) {
> + if (rte_kvargs_process(kvlist, AVF_CLIENT_ARG_PATH,
> + &iavf_client_get_string_arg, &path) < 0)
> {
> + PMD_INIT_LOG(ERR, "error to parse %s",
> + AVF_CLIENT_ARG_PATH);
> + return -EINVAL;
> + }
> + } else {
> + PMD_INIT_LOG(ERR, "arg %s is mandatory for virtio_user",
> + AVF_CLIENT_ARG_PATH);
> + return -EINVAL;
> + }
> +
> + eth_dev = rte_eth_vdev_allocate(vdev, sizeof(*adapter));
> +
> + ret = iavf_client_vfio_user_setup(eth_dev, path);
> + if (ret) {
> + goto err;
> + }
> +
> + ret = iavf_client_eth_init(eth_dev);
> + if (ret) {
> + goto err;
> + }
> +
> + rte_eth_dev_probing_finish(eth_dev);
> + rte_kvargs_free(kvlist);
> +
> + return 0;
> +err:
> + rte_eth_dev_release_port(eth_dev);
> + rte_kvargs_free(kvlist);
> + return ret;
> +}
> +
> +
> +static int
> +avf_client_pmd_remove(struct rte_vdev_device *vdev)
> +{
> + struct rte_eth_dev *eth_dev = NULL;
> +
> + if (vdev == NULL)
> + return -1;
> +
> + /* find the ethdev entry */
> + eth_dev = rte_eth_dev_allocated(rte_vdev_device_name(vdev));
> + if (eth_dev == NULL)
> + return 0;
> +
> + iavf_client_dev_close(eth_dev);
> + rte_eth_dev_release_port(eth_dev);
> + return 0;
> +}
> +
> +static struct rte_vdev_driver iavf_client_driver = {
> + .probe = avf_client_pmd_probe,
> + .remove = avf_client_pmd_remove,
> +};
> +
> +RTE_PMD_REGISTER_VDEV(net_iavf_client, iavf_client_driver);
> +RTE_PMD_REGISTER_ALIAS(net_iavf_client, iavf_client);
> +RTE_PMD_REGISTER_PARAM_STRING(net_iavf_client,
> + "path=<path>");
> diff --git a/drivers/net/iavf/iavf_ethdev.c b/drivers/net/iavf/iavf_ethdev.c
> index 7e3c26a94e..6b5e47adf2 100644
> --- a/drivers/net/iavf/iavf_ethdev.c
> +++ b/drivers/net/iavf/iavf_ethdev.c
> @@ -154,7 +154,7 @@ static const struct rte_iavf_xstats_name_off
> rte_iavf_stats_strings[] = {
> #define IAVF_NB_XSTATS (sizeof(rte_iavf_stats_strings) / \
> sizeof(rte_iavf_stats_strings[0]))
>
> -static const struct eth_dev_ops iavf_eth_dev_ops = {
> +const struct eth_dev_ops iavf_eth_dev_ops = {
> .dev_configure = iavf_dev_configure,
> .dev_start = iavf_dev_start,
> .dev_stop = iavf_dev_stop,
> @@ -1780,7 +1780,7 @@ iavf_init_proto_xtr(struct rte_eth_dev *dev)
> }
> }
>
> -static int
> +int
> iavf_init_vf(struct rte_eth_dev *dev)
> {
> int err, bufsz;
> @@ -1789,12 +1789,6 @@ iavf_init_vf(struct rte_eth_dev *dev)
> struct iavf_hw *hw = IAVF_DEV_PRIVATE_TO_HW(dev->data->dev_private);
> struct iavf_info *vf = IAVF_DEV_PRIVATE_TO_VF(dev->data->dev_private);
>
> - err = iavf_parse_devargs(dev);
> - if (err) {
> - PMD_INIT_LOG(ERR, "Failed to parse devargs");
> - goto err;
> - }
> -
> err = iavf_set_mac_type(hw);
> if (err) {
> PMD_INIT_LOG(ERR, "set_mac_type failed: %d", err);
> @@ -1891,16 +1885,6 @@ iavf_enable_irq0(struct iavf_hw *hw)
> IAVF_WRITE_FLUSH(hw);
> }
>
> -static inline void
> -iavf_disable_irq0(struct iavf_hw *hw)
> -{
> - /* Disable all interrupt types */
> - IAVF_WRITE_REG(hw, IAVF_VFINT_ICR0_ENA1, 0);
> - IAVF_WRITE_REG(hw, IAVF_VFINT_DYN_CTL01,
> - IAVF_VFINT_DYN_CTL01_ITR_INDX_MASK);
> - IAVF_WRITE_FLUSH(hw);
> -}
> -
> static void
> iavf_dev_interrupt_handler(void *param)
> {
> @@ -1986,6 +1970,12 @@ iavf_dev_init(struct rte_eth_dev *eth_dev)
> adapter->eth_dev = eth_dev;
> adapter->stopped = 1;
>
> + ret = iavf_parse_devargs(eth_dev);
> + if (ret) {
> + PMD_INIT_LOG(ERR, "Failed to parse devargs");
> + return ret;
> + }
> +
> if (iavf_init_vf(eth_dev) != 0) {
> PMD_INIT_LOG(ERR, "Init vf failed");
> return -1;
> diff --git a/drivers/net/iavf/iavf_rxtx.c b/drivers/net/iavf/iavf_rxtx.c
> index 21d508b3f4..d8192f3675 100644
> --- a/drivers/net/iavf/iavf_rxtx.c
> +++ b/drivers/net/iavf/iavf_rxtx.c
> @@ -24,6 +24,9 @@
> #include <rte_ip.h>
> #include <rte_net.h>
> #include <rte_vect.h>
> +#ifdef RTE_LIBRTE_IAVF_CLIENT
> +#include <vfio_user/vfio_user_pci.h>
> +#endif
>
> #include "iavf.h"
> #include "iavf_rxtx.h"
> @@ -595,7 +598,15 @@ iavf_dev_rx_queue_setup(struct rte_eth_dev *dev, uint16_t
> queue_idx,
> reset_rx_queue(rxq);
> rxq->q_set = true;
> dev->data->rx_queues[queue_idx] = rxq;
> - rxq->qrx_tail = hw->hw_addr + IAVF_QRX_TAIL1(rxq->queue_id);
> +
> +#ifdef RTE_LIBRTE_IAVF_CLIENT
> + if (hw->bus.type == iavf_bus_type_vfio_user)
> + rxq->qrx_tail = client_vfio_user_get_bar_addr(
> + (struct vfio_device *)hw->hw_addr, 0,
> + IAVF_QRX_TAIL1(rxq->queue_id), 4);
> + else
> +#endif
> + rxq->qrx_tail = hw->hw_addr + IAVF_QRX_TAIL1(rxq->queue_id);
> rxq->ops = &def_rxq_ops;
>
> if (check_rx_bulk_allow(rxq) == true) {
> @@ -705,7 +716,15 @@ iavf_dev_tx_queue_setup(struct rte_eth_dev *dev,
> reset_tx_queue(txq);
> txq->q_set = true;
> dev->data->tx_queues[queue_idx] = txq;
> - txq->qtx_tail = hw->hw_addr + IAVF_QTX_TAIL1(queue_idx);
> +#ifdef RTE_LIBRTE_IAVF_CLIENT
> + if (hw->bus.type == iavf_bus_type_vfio_user)
> + txq->qtx_tail = client_vfio_user_get_bar_addr(
> + (struct vfio_device *)hw->hw_addr, 0,
> + IAVF_QTX_TAIL1(queue_idx), 4);
> + else
> +#endif
> + txq->qtx_tail = hw->hw_addr + IAVF_QTX_TAIL1(queue_idx);
> +
> txq->ops = &def_txq_ops;
>
> if (check_tx_vec_allow(txq) == false) {
> diff --git a/drivers/net/iavf/meson.build b/drivers/net/iavf/meson.build
> index 26c02c4401..580307c462 100644
> --- a/drivers/net/iavf/meson.build
> +++ b/drivers/net/iavf/meson.build
> @@ -13,6 +13,7 @@ sources = files(
> 'iavf_generic_flow.c',
> 'iavf_fdir.c',
> 'iavf_hash.c',
> + 'iavf_client_ethdev.c',
> )
>
> if arch_subdir == 'x86'
> --
> 2.21.1
