vdpa/sfc: introduce Xilinx vDPA driver

Add new vDPA PMD to support vDPA operations of Xilinx devices.
This patch implements probe and remove functions.

Signed-off-by: Vijay Kumar Srivastava <vsrivast@xilinx.com>
Acked-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

MAINTAINERS

@@ -1240,6 +1240,12 @@ F: drivers/vdpa/mlx5/
 F: doc/guides/vdpadevs/mlx5.rst
 F: doc/guides/vdpadevs/features/mlx5.ini
 
+Xilinx sfc vDPA
+M: Vijay Kumar Srivastava <vsrivast@xilinx.com>
+F: drivers/vdpa/sfc/
+F: doc/guides/vdpadevs/sfc.rst
+F: doc/guides/vdpadevs/features/sfc.ini
+
 
 Eventdev Drivers
 ----------------

doc/guides/rel_notes/release_21_11.rst

@@ -233,6 +233,11 @@ New Features
   * Added flow control.
   * Added IEEE 1588.
 
+* **Added new vDPA PMD based on Xilinx devices.**
+
+  Added a new Xilinx vDPA  (``sfc_vdpa``) PMD.
+  See the :doc:`../vdpadevs/sfc` guide for more details on this driver.
+
 * **Added telemetry callbacks to cryptodev library.**
 
   Added telemetry callback functions which allow a list of crypto devices,

doc/guides/vdpadevs/features/sfc.ini

@@ -0,0 +1,9 @@
+;
+; Supported features of the 'sfc' vDPA driver.
+;
+; Refer to default.ini for the full list of available driver features.
+;
+[Features]
+Linux			= Y
+x86-64			= Y
+Usage doc		= Y

doc/guides/vdpadevs/index.rst

@@ -14,3 +14,4 @@ which can be used from an application through vhost API.
     features_overview
     ifc
     mlx5
+    sfc

doc/guides/vdpadevs/sfc.rst

@@ -0,0 +1,97 @@
+..  SPDX-License-Identifier: BSD-3-Clause
+    Copyright(c) 2021 Xilinx Corporation.
+
+Xilinx vDPA driver
+==================
+
+The Xilinx vDPA (vhost data path acceleration) driver (**librte_pmd_sfc_vdpa**)
+provides support for the Xilinx SN1022 SmartNICs family of 10/25/40/50/100 Gbps
+adapters that have support for latest Linux and FreeBSD operating systems.
+
+More information can be found at Xilinx website https://www.xilinx.com.
+
+
+Xilinx vDPA implementation
+--------------------------
+
+ef100 device can be configured in the net device or vDPA mode.
+Adding "class=vdpa" parameter helps to specify that this
+device is to be used in vDPA mode. If this parameter is not specified, device
+will be probed by net/sfc driver and will used as a net device.
+
+This PMD uses libefx (common/sfc_efx) code to access the device firmware.
+
+
+Supported NICs
+--------------
+
+- Xilinx SN1022 SmartNICs
+
+
+Features
+--------
+
+Features of the Xilinx vDPA driver are:
+
+- Compatibility with virtio 0.95 and 1.0
+
+
+Non-supported Features
+----------------------
+
+- Control Queue
+- Multi queue
+- Live Migration
+
+
+Prerequisites
+-------------
+
+Requires firmware version: v1.0.7.0 or higher
+
+Visit `Xilinx Support Downloads <https://www.xilinx.com/support.html>`_
+to get Xilinx Utilities with the latest firmware.
+Follow instructions from Alveo SN1000 SmartNICs User Guide to
+update firmware and configure the adapter.
+
+
+Per-Device Parameters
+~~~~~~~~~~~~~~~~~~~~~
+
+The following per-device parameters can be passed via EAL PCI device
+allowlist option like "-a 02:00.0,arg1=value1,...".
+
+Case-insensitive 1/y/yes/on or 0/n/no/off may be used to specify
+boolean parameters value.
+
+- ``class`` [net|vdpa] (default **net**)
+
+  Choose the mode of operation of ef100 device.
+  **net** device will work as network device and will be probed by net/sfc driver.
+  **vdpa** device will work as vdpa device and will be probed by vdpa/sfc driver.
+  If this parameter is not specified then ef100 device will operate as network device.
+
+
+Dynamic Logging Parameters
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+One may leverage EAL option "--log-level" to change default levels
+for the log types supported by the driver. The option is used with
+an argument typically consisting of two parts separated by a colon.
+
+Level value is the last part which takes a symbolic name (or integer).
+Log type is the former part which may shell match syntax.
+Depending on the choice of the expression, the given log level may
+be used either for some specific log type or for a subset of types.
+
+SFC vDPA PMD provides the following log types available for control:
+
+- ``pmd.vdpa.sfc.driver`` (default level is **notice**)
+
+  Affects driver-wide messages unrelated to any particular devices.
+
+- ``pmd.vdpa.sfc.main`` (default level is **notice**)
+
+  Matches a subset of per-port log types registered during runtime.
+  A full name for a particular type may be obtained by appending a
+  dot and a PCI device identifier (``XXXX:XX:XX.X``) to the prefix.

drivers/vdpa/meson.build

@@ -8,6 +8,7 @@ endif
 drivers = [
         'ifc',
         'mlx5',
+        'sfc',
 ]
 std_deps = ['bus_pci', 'kvargs']
 std_deps += ['vhost']

drivers/vdpa/sfc/meson.build

@@ -0,0 +1,23 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2020-2021 Xilinx, Inc.
+
+if ((arch_subdir != 'x86' or not dpdk_conf.get('RTE_ARCH_64')) and
+        (arch_subdir != 'arm' or not
+            host_machine.cpu_family().startswith('aarch64')))
+    build = false
+    reason = 'only supported on x86_64 and aarch64'
+endif
+
+fmt_name = 'sfc_vdpa'
+extra_flags = []
+
+foreach flag: extra_flags
+    if cc.has_argument(flag)
+        cflags += flag
+    endif
+endforeach
+
+deps += ['common_sfc_efx', 'bus_pci']
+sources = files(
+        'sfc_vdpa.c',
+)

drivers/vdpa/sfc/sfc_vdpa.c

@@ -0,0 +1,281 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2020-2021 Xilinx, Inc.
+ */
+
+#include <stdbool.h>
+#include <stdint.h>
+#include <sys/queue.h>
+
+#include <rte_common.h>
+#include <rte_errno.h>
+#include <rte_string_fns.h>
+#include <rte_vfio.h>
+#include <rte_vhost.h>
+
+#include "efx.h"
+#include "sfc_efx.h"
+#include "sfc_vdpa.h"
+
+TAILQ_HEAD(sfc_vdpa_adapter_list_head, sfc_vdpa_adapter);
+static struct sfc_vdpa_adapter_list_head sfc_vdpa_adapter_list =
+	TAILQ_HEAD_INITIALIZER(sfc_vdpa_adapter_list);
+
+static pthread_mutex_t sfc_vdpa_adapter_list_lock = PTHREAD_MUTEX_INITIALIZER;
+
+struct sfc_vdpa_adapter *
+sfc_vdpa_get_adapter_by_dev(struct rte_pci_device *pdev)
+{
+	bool found = false;
+	struct sfc_vdpa_adapter *sva;
+
+	pthread_mutex_lock(&sfc_vdpa_adapter_list_lock);
+
+	TAILQ_FOREACH(sva, &sfc_vdpa_adapter_list, next) {
+		if (pdev == sva->pdev) {
+			found = true;
+			break;
+		}
+	}
+
+	pthread_mutex_unlock(&sfc_vdpa_adapter_list_lock);
+
+	return found ? sva : NULL;
+}
+
+static int
+sfc_vdpa_vfio_setup(struct sfc_vdpa_adapter *sva)
+{
+	struct rte_pci_device *dev = sva->pdev;
+	char dev_name[RTE_DEV_NAME_MAX_LEN] = {0};
+	int rc;
+
+	rte_pci_device_name(&dev->addr, dev_name, RTE_DEV_NAME_MAX_LEN);
+
+	sva->vfio_container_fd = rte_vfio_container_create();
+	if (sva->vfio_container_fd < 0)	{
+		sfc_vdpa_err(sva, "failed to create VFIO container");
+		goto fail_container_create;
+	}
+
+	rc = rte_vfio_get_group_num(rte_pci_get_sysfs_path(), dev_name,
+				    &sva->iommu_group_num);
+	if (rc <= 0) {
+		sfc_vdpa_err(sva, "failed to get IOMMU group for %s : %s",
+			     dev_name, rte_strerror(-rc));
+		goto fail_get_group_num;
+	}
+
+	sva->vfio_group_fd =
+		rte_vfio_container_group_bind(sva->vfio_container_fd,
+					      sva->iommu_group_num);
+	if (sva->vfio_group_fd < 0) {
+		sfc_vdpa_err(sva,
+			     "failed to bind IOMMU group %d to container %d",
+			     sva->iommu_group_num, sva->vfio_container_fd);
+		goto fail_group_bind;
+	}
+
+	if (rte_pci_map_device(dev) != 0) {
+		sfc_vdpa_err(sva, "failed to map PCI device %s : %s",
+			     dev_name, rte_strerror(rte_errno));
+		goto fail_pci_map_device;
+	}
+
+	sva->vfio_dev_fd = rte_intr_dev_fd_get(dev->intr_handle);
+
+	return 0;
+
+fail_pci_map_device:
+	if (rte_vfio_container_group_unbind(sva->vfio_container_fd,
+					sva->iommu_group_num) != 0) {
+		sfc_vdpa_err(sva,
+			     "failed to unbind IOMMU group %d from container %d",
+			     sva->iommu_group_num, sva->vfio_container_fd);
+	}
+
+fail_group_bind:
+fail_get_group_num:
+	if (rte_vfio_container_destroy(sva->vfio_container_fd) != 0) {
+		sfc_vdpa_err(sva, "failed to destroy container %d",
+			     sva->vfio_container_fd);
+	}
+
+fail_container_create:
+	return -1;
+}
+
+static void
+sfc_vdpa_vfio_teardown(struct sfc_vdpa_adapter *sva)
+{
+	rte_pci_unmap_device(sva->pdev);
+
+	if (rte_vfio_container_group_unbind(sva->vfio_container_fd,
+					    sva->iommu_group_num) != 0) {
+		sfc_vdpa_err(sva,
+			     "failed to unbind IOMMU group %d from container %d",
+			     sva->iommu_group_num, sva->vfio_container_fd);
+	}
+
+	if (rte_vfio_container_destroy(sva->vfio_container_fd) != 0) {
+		sfc_vdpa_err(sva,
+			     "failed to destroy container %d",
+			     sva->vfio_container_fd);
+	}
+}
+
+static int
+sfc_vdpa_set_log_prefix(struct sfc_vdpa_adapter *sva)
+{
+	struct rte_pci_device *pci_dev = sva->pdev;
+	int ret;
+
+	ret = snprintf(sva->log_prefix, sizeof(sva->log_prefix),
+		       "PMD: sfc_vdpa " PCI_PRI_FMT " : ",
+		       pci_dev->addr.domain, pci_dev->addr.bus,
+		       pci_dev->addr.devid, pci_dev->addr.function);
+
+	if (ret < 0 || ret >= (int)sizeof(sva->log_prefix)) {
+		SFC_VDPA_GENERIC_LOG(ERR,
+			"reserved log prefix is too short for " PCI_PRI_FMT,
+			pci_dev->addr.domain, pci_dev->addr.bus,
+			pci_dev->addr.devid, pci_dev->addr.function);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+uint32_t
+sfc_vdpa_register_logtype(const struct rte_pci_addr *pci_addr,
+			  const char *lt_prefix_str, uint32_t ll_default)
+{
+	size_t lt_prefix_str_size = strlen(lt_prefix_str);
+	size_t lt_str_size_max;
+	char *lt_str = NULL;
+	int ret;
+
+	if (SIZE_MAX - PCI_PRI_STR_SIZE - 1 > lt_prefix_str_size) {
+		++lt_prefix_str_size; /* Reserve space for prefix separator */
+		lt_str_size_max = lt_prefix_str_size + PCI_PRI_STR_SIZE + 1;
+	} else {
+		return RTE_LOGTYPE_PMD;
+	}
+
+	lt_str = rte_zmalloc("logtype_str", lt_str_size_max, 0);
+	if (lt_str == NULL)
+		return RTE_LOGTYPE_PMD;
+
+	strncpy(lt_str, lt_prefix_str, lt_prefix_str_size);
+	lt_str[lt_prefix_str_size - 1] = '.';
+	rte_pci_device_name(pci_addr, lt_str + lt_prefix_str_size,
+			    lt_str_size_max - lt_prefix_str_size);
+	lt_str[lt_str_size_max - 1] = '\0';
+
+	ret = rte_log_register_type_and_pick_level(lt_str, ll_default);
+	rte_free(lt_str);
+
+	return ret < 0 ? RTE_LOGTYPE_PMD : ret;
+}
+
+static struct rte_pci_id pci_id_sfc_vdpa_efx_map[] = {
+	{ RTE_PCI_DEVICE(EFX_PCI_VENID_XILINX, EFX_PCI_DEVID_RIVERHEAD_VF) },
+	{ .vendor_id = 0, /* sentinel */ },
+};
+
+static int
+sfc_vdpa_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
+	struct rte_pci_device *pci_dev)
+{
+	struct sfc_vdpa_adapter *sva = NULL;
+	uint32_t logtype_main;
+	int ret = 0;
+
+	if (sfc_efx_dev_class_get(pci_dev->device.devargs) !=
+			SFC_EFX_DEV_CLASS_VDPA) {
+		SFC_VDPA_GENERIC_LOG(INFO,
+			"Incompatible device class: skip probing, should be probed by other sfc driver.");
+			return 1;
+	}
+
+	/*
+	 * It will not be probed in the secondary process. As device class
+	 * is vdpa so return 0 to avoid probe by other sfc driver
+	 */
+	if (rte_eal_process_type() != RTE_PROC_PRIMARY)
+		return 0;
+
+	logtype_main = sfc_vdpa_register_logtype(&pci_dev->addr,
+						 SFC_VDPA_LOGTYPE_MAIN_STR,
+						 RTE_LOG_NOTICE);
+
+	sva = rte_zmalloc("sfc_vdpa", sizeof(struct sfc_vdpa_adapter), 0);
+	if (sva == NULL)
+		goto fail_zmalloc;
+
+	sva->pdev = pci_dev;
+	sva->logtype_main = logtype_main;
+
+	ret = sfc_vdpa_set_log_prefix(sva);
+	if (ret != 0)
+		goto fail_set_log_prefix;
+
+	sfc_vdpa_log_init(sva, "entry");
+
+	sfc_vdpa_log_init(sva, "vfio init");
+	if (sfc_vdpa_vfio_setup(sva) < 0) {
+		sfc_vdpa_err(sva, "failed to setup device %s", pci_dev->name);
+		goto fail_vfio_setup;
+	}
+
+	pthread_mutex_lock(&sfc_vdpa_adapter_list_lock);
+	TAILQ_INSERT_TAIL(&sfc_vdpa_adapter_list, sva, next);
+	pthread_mutex_unlock(&sfc_vdpa_adapter_list_lock);
+
+	sfc_vdpa_log_init(sva, "done");
+
+	return 0;
+
+fail_vfio_setup:
+fail_set_log_prefix:
+	rte_free(sva);
+
+fail_zmalloc:
+	return -1;
+}
+
+static int
+sfc_vdpa_pci_remove(struct rte_pci_device *pci_dev)
+{
+	struct sfc_vdpa_adapter *sva = NULL;
+
+	if (rte_eal_process_type() != RTE_PROC_PRIMARY)
+		return -1;
+
+	sva = sfc_vdpa_get_adapter_by_dev(pci_dev);
+	if (sva == NULL) {
+		sfc_vdpa_info(sva, "invalid device: %s", pci_dev->name);
+		return -1;
+	}
+
+	pthread_mutex_lock(&sfc_vdpa_adapter_list_lock);
+	TAILQ_REMOVE(&sfc_vdpa_adapter_list, sva, next);
+	pthread_mutex_unlock(&sfc_vdpa_adapter_list_lock);
+
+	sfc_vdpa_vfio_teardown(sva);
+
+	rte_free(sva);
+
+	return 0;
+}
+
+static struct rte_pci_driver rte_sfc_vdpa = {
+	.id_table = pci_id_sfc_vdpa_efx_map,
+	.drv_flags = 0,
+	.probe = sfc_vdpa_pci_probe,
+	.remove = sfc_vdpa_pci_remove,
+};
+
+RTE_PMD_REGISTER_PCI(net_sfc_vdpa, rte_sfc_vdpa);
+RTE_PMD_REGISTER_PCI_TABLE(net_sfc_vdpa, pci_id_sfc_vdpa_efx_map);
+RTE_PMD_REGISTER_KMOD_DEP(net_sfc_vdpa, "* vfio-pci");
+RTE_LOG_REGISTER_SUFFIX(sfc_vdpa_logtype_driver, driver, NOTICE);

drivers/vdpa/sfc/sfc_vdpa.h

@@ -0,0 +1,38 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2020-2021 Xilinx, Inc.
+ */
+
+#ifndef _SFC_VDPA_H
+#define _SFC_VDPA_H
+
+#include <stdint.h>
+#include <sys/queue.h>
+
+#include <rte_bus_pci.h>
+
+#include "sfc_vdpa_log.h"
+
+/* Adapter private data */
+struct sfc_vdpa_adapter {
+	TAILQ_ENTRY(sfc_vdpa_adapter)	next;
+	struct rte_pci_device		*pdev;
+
+	char				log_prefix[SFC_VDPA_LOG_PREFIX_MAX];
+	uint32_t			logtype_main;
+
+	int				vfio_group_fd;
+	int				vfio_dev_fd;
+	int				vfio_container_fd;
+	int				iommu_group_num;
+};
+
+uint32_t
+sfc_vdpa_register_logtype(const struct rte_pci_addr *pci_addr,
+			  const char *lt_prefix_str,
+			  uint32_t ll_default);
+
+struct sfc_vdpa_adapter *
+sfc_vdpa_get_adapter_by_dev(struct rte_pci_device *pdev);
+
+#endif  /* _SFC_VDPA_H */
+

drivers/vdpa/sfc/sfc_vdpa_log.h

@@ -0,0 +1,55 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2020-2021 Xilinx, Inc.
+ */
+
+#ifndef _SFC_VDPA_LOG_H_
+#define _SFC_VDPA_LOG_H_
+
+/** Generic driver log type */
+extern int sfc_vdpa_logtype_driver;
+
+/** Common log type name prefix */
+#define SFC_VDPA_LOGTYPE_PREFIX	"pmd.vdpa.sfc."
+
+/** Log PMD generic message, add a prefix and a line break */
+#define SFC_VDPA_GENERIC_LOG(level, ...) \
+	rte_log(RTE_LOG_ ## level, sfc_vdpa_logtype_driver,		\
+		RTE_FMT("PMD: " RTE_FMT_HEAD(__VA_ARGS__ ,) "\n",	\
+			RTE_FMT_TAIL(__VA_ARGS__ ,)))
+
+/** Name prefix for the per-device log type used to report basic information */
+#define SFC_VDPA_LOGTYPE_MAIN_STR	SFC_VDPA_LOGTYPE_PREFIX "main"
+
+#define SFC_VDPA_LOG_PREFIX_MAX	32
+
+/* Log PMD message, automatically add prefix and \n */
+#define SFC_VDPA_LOG(sva, level, ...) \
+	do {								\
+		const struct sfc_vdpa_adapter *_sva = (sva);		\
+									\
+		rte_log(RTE_LOG_ ## level, _sva->logtype_main,		\
+			RTE_FMT("%s" RTE_FMT_HEAD(__VA_ARGS__ ,) "\n",	\
+				_sva->log_prefix,			\
+				RTE_FMT_TAIL(__VA_ARGS__ ,)));		\
+	} while (0)
+
+#define sfc_vdpa_err(sva, ...) \
+	SFC_VDPA_LOG(sva, ERR, __VA_ARGS__)
+
+#define sfc_vdpa_warn(sva, ...) \
+	SFC_VDPA_LOG(sva, WARNING, __VA_ARGS__)
+
+#define sfc_vdpa_notice(sva, ...) \
+	SFC_VDPA_LOG(sva, NOTICE, __VA_ARGS__)
+
+#define sfc_vdpa_info(sva, ...) \
+	SFC_VDPA_LOG(sva, INFO, __VA_ARGS__)
+
+#define sfc_vdpa_log_init(sva, ...) \
+	SFC_VDPA_LOG(sva, INFO,				\
+		RTE_FMT("%s(): "			\
+			RTE_FMT_HEAD(__VA_ARGS__ ,),	\
+			__func__,			\
+			RTE_FMT_TAIL(__VA_ARGS__ ,)))
+
+#endif /* _SFC_VDPA_LOG_H_ */

drivers/vdpa/sfc/version.map

@@ -0,0 +1,3 @@
+DPDK_22 {
+	local: *;
+};