* [patch net-next v2 1/9] Introduce devlink infrastructure
2016-02-23 15:51 [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it Jiri Pirko
@ 2016-02-23 15:51 ` Jiri Pirko
2016-02-23 21:19 ` Jarod Wilson
2016-02-23 15:51 ` [patch net-next v2 2/9] mlx4: Implement devlink interface Jiri Pirko
` (8 subsequent siblings)
9 siblings, 1 reply; 19+ messages in thread
From: Jiri Pirko @ 2016-02-23 15:51 UTC (permalink / raw)
To: netdev
Cc: davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
From: Jiri Pirko <jiri@mellanox.com>
Introduce devlink infrastructure for drivers to register and expose to
userspace via generic Netlink interface.
There are two basic objects defined:
devlink - one instance for every "parent device", for example switch ASIC
devlink port - one instance for every physical port of the device.
This initial portion implements basic get/dump of objects to userspace.
Also, port splitter and port type setting is implemented.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
v1->v2:
-removed no longer used "devlink_dev" helper
-fix couple of typos and misspells
---
MAINTAINERS | 8 +
include/net/devlink.h | 151 ++++++++
include/uapi/linux/devlink.h | 73 ++++
net/Kconfig | 7 +
net/core/Makefile | 1 +
net/core/devlink.c | 887 +++++++++++++++++++++++++++++++++++++++++++
6 files changed, 1127 insertions(+)
create mode 100644 include/net/devlink.h
create mode 100644 include/uapi/linux/devlink.h
create mode 100644 net/core/devlink.c
diff --git a/MAINTAINERS b/MAINTAINERS
index 27393cf..0890b69 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -3499,6 +3499,14 @@ F: include/linux/device-mapper.h
F: include/linux/dm-*.h
F: include/uapi/linux/dm-*.h
+DEVLINK
+M: Jiri Pirko <jiri@mellanox.com>
+L: netdev@vger.kernel.org
+S: Supported
+F: net/core/devlink.c
+F: include/net/devlink.h
+F: include/uapi/linux/devlink.h
+
DIALOG SEMICONDUCTOR DRIVERS
M: Support Opensource <support.opensource@diasemi.com>
W: http://www.dialog-semiconductor.com/products
diff --git a/include/net/devlink.h b/include/net/devlink.h
new file mode 100644
index 0000000..98e79d9
--- /dev/null
+++ b/include/net/devlink.h
@@ -0,0 +1,151 @@
+/*
+ * include/net/devlink.h - Network physical device Netlink interface
+ * Copyright (c) 2016 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2016 Jiri Pirko <jiri@mellanox.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+#ifndef _NET_DEVLINK_H_
+#define _NET_DEVLINK_H_
+
+#include <linux/device.h>
+#include <linux/slab.h>
+#include <linux/gfp.h>
+#include <linux/list.h>
+#include <linux/netdevice.h>
+#include <net/net_namespace.h>
+#include <uapi/linux/devlink.h>
+
+struct devlink_ops;
+
+struct devlink {
+ struct list_head list;
+ struct list_head port_list;
+ int index;
+ const struct devlink_ops *ops;
+ struct device dev;
+ possible_net_t _net;
+ char priv[0] __aligned(NETDEV_ALIGN);
+};
+
+struct devlink_port {
+ struct list_head list;
+ struct devlink *devlink;
+ unsigned index;
+ bool registered;
+ enum devlink_port_type type;
+ enum devlink_port_type desired_type;
+ void *type_dev;
+ bool split;
+ u32 split_group;
+};
+
+struct devlink_ops {
+ size_t priv_size;
+ int (*port_type_set)(struct devlink_port *devlink_port,
+ enum devlink_port_type port_type);
+ int (*port_split)(struct devlink *devlink, unsigned int port_index,
+ unsigned int count);
+ int (*port_unsplit)(struct devlink *devlink, unsigned int port_index);
+};
+
+static inline void *devlink_priv(struct devlink *devlink)
+{
+ BUG_ON(!devlink);
+ return &devlink->priv;
+}
+
+static inline struct devlink *priv_to_devlink(void *priv)
+{
+ BUG_ON(!priv);
+ return container_of(priv, struct devlink, priv);
+}
+
+static inline void set_devlink_dev(struct devlink *devlink, struct device *dev)
+{
+ devlink->dev.parent = dev;
+}
+
+static inline const char *devlink_name(const struct devlink *devlink)
+{
+ return dev_name(&devlink->dev);
+}
+
+struct ib_device;
+
+#if IS_ENABLED(CONFIG_NET_DEVLINK)
+
+struct devlink *devlink_alloc(const struct devlink_ops *ops, size_t priv_size);
+int devlink_register(struct devlink *devlink);
+void devlink_unregister(struct devlink *devlink);
+void devlink_free(struct devlink *devlink);
+int devlink_port_register(struct devlink *devlink,
+ struct devlink_port *devlink_port,
+ unsigned int port_index);
+void devlink_port_unregister(struct devlink_port *devlink_port);
+void devlink_port_type_eth_set(struct devlink_port *devlink_port,
+ struct net_device *netdev);
+void devlink_port_type_ib_set(struct devlink_port *devlink_port,
+ struct ib_device *ibdev);
+void devlink_port_type_clear(struct devlink_port *devlink_port);
+void devlink_port_split_set(struct devlink_port *devlink_port,
+ u32 split_group);
+
+#else
+
+static inline struct devlink *devlink_alloc(const struct devlink_ops *ops,
+ size_t priv_size)
+{
+ return kzalloc(sizeof(struct devlink) + priv_size, GFP_KERNEL);
+}
+
+static inline int devlink_register(struct devlink *devlink)
+{
+ return 0;
+}
+
+static inline void devlink_unregister(struct devlink *devlink)
+{
+}
+
+static inline void devlink_free(struct devlink *devlink)
+{
+ kfree(devlink);
+}
+
+static inline int devlink_port_register(struct devlink *devlink,
+ struct devlink_port *devlink_port,
+ unsigned int port_index)
+{
+ return 0;
+}
+
+static inline void devlink_port_unregister(struct devlink_port *devlink_port)
+{
+}
+
+static inline void devlink_port_type_eth_set(struct devlink_port *devlink_port,
+ struct net_device *netdev)
+{
+}
+
+static inline void devlink_port_type_ib_set(struct devlink_port *devlink_port,
+ struct ib_device *ibdev)
+{
+}
+
+static inline void devlink_port_type_clear(struct devlink_port *devlink_port)
+{
+}
+
+static inline void devlink_port_split_set(struct devlink_port *devlink_port,
+ u32 split_group)
+{
+}
+
+#endif
+
+#endif /* _NET_DEVLINK_H_ */
diff --git a/include/uapi/linux/devlink.h b/include/uapi/linux/devlink.h
new file mode 100644
index 0000000..04e0376
--- /dev/null
+++ b/include/uapi/linux/devlink.h
@@ -0,0 +1,73 @@
+/*
+ * include/uapi/linux/devlink.h - Network physical device Netlink interface
+ * Copyright (c) 2016 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2016 Jiri Pirko <jiri@mellanox.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#ifndef _UAPI_LINUX_DEVLINK_H_
+#define _UAPI_LINUX_DEVLINK_H_
+
+#define DEVLINK_GENL_NAME "devlink"
+#define DEVLINK_GENL_VERSION 0x1
+#define DEVLINK_GENL_MCGRP_CONFIG_NAME "config"
+
+enum devlink_command {
+ /* don't change the order or add anything between, this is ABI! */
+ DEVLINK_CMD_UNSPEC,
+
+ DEVLINK_CMD_GET, /* can dump */
+ DEVLINK_CMD_SET,
+ DEVLINK_CMD_NEW,
+ DEVLINK_CMD_DEL,
+
+ DEVLINK_CMD_PORT_GET, /* can dump */
+ DEVLINK_CMD_PORT_SET,
+ DEVLINK_CMD_PORT_NEW,
+ DEVLINK_CMD_PORT_DEL,
+
+ DEVLINK_CMD_PORT_SPLIT,
+ DEVLINK_CMD_PORT_UNSPLIT,
+
+ /* add new commands above here */
+
+ __DEVLINK_CMD_MAX,
+ DEVLINK_CMD_MAX = __DEVLINK_CMD_MAX - 1
+};
+
+enum devlink_port_type {
+ DEVLINK_PORT_TYPE_NOTSET,
+ DEVLINK_PORT_TYPE_AUTO,
+ DEVLINK_PORT_TYPE_ETH,
+ DEVLINK_PORT_TYPE_IB,
+};
+
+enum devlink_attr {
+ /* don't change the order or add anything between, this is ABI! */
+ DEVLINK_ATTR_UNSPEC,
+
+ DEVLINK_ATTR_INDEX, /* u32 */
+#define DEVLINK_ATTR_NAME_MAX_LEN 32
+ DEVLINK_ATTR_NAME, /* string */
+ DEVLINK_ATTR_BUS_NAME, /* string */
+ DEVLINK_ATTR_DEV_NAME, /* string */
+ DEVLINK_ATTR_PORT_INDEX, /* u32 */
+ DEVLINK_ATTR_PORT_TYPE, /* u16 */
+ DEVLINK_ATTR_PORT_DESIRED_TYPE, /* u16 */
+ DEVLINK_ATTR_PORT_NETDEV_IFINDEX, /* u32 */
+ DEVLINK_ATTR_PORT_NETDEV_NAME, /* string */
+ DEVLINK_ATTR_PORT_IBDEV_NAME, /* string */
+ DEVLINK_ATTR_PORT_SPLIT_COUNT, /* u32 */
+ DEVLINK_ATTR_PORT_SPLIT_GROUP, /* u32 */
+
+ /* add new attributes above here, update the policy in devlink.c */
+
+ __DEVLINK_ATTR_MAX,
+ DEVLINK_ATTR_MAX = __DEVLINK_ATTR_MAX - 1
+};
+
+#endif /* _UAPI_LINUX_DEVLINK_H_ */
diff --git a/net/Kconfig b/net/Kconfig
index b80efec..6c9cfb0 100644
--- a/net/Kconfig
+++ b/net/Kconfig
@@ -396,6 +396,13 @@ config DST_CACHE
bool "dst cache"
default n
+config NET_DEVLINK
+ tristate "Network physical/parent device Netlink interface"
+ help
+ Network physical/parent device Netlink interface provides
+ infrastructure to support access to physical chip-wide config and
+ monitoring.
+
endif # if NET
# Used by archs to tell that they support BPF_JIT
diff --git a/net/core/Makefile b/net/core/Makefile
index 7a8fb8a..014422e 100644
--- a/net/core/Makefile
+++ b/net/core/Makefile
@@ -25,3 +25,4 @@ obj-$(CONFIG_CGROUP_NET_PRIO) += netprio_cgroup.o
obj-$(CONFIG_CGROUP_NET_CLASSID) += netclassid_cgroup.o
obj-$(CONFIG_LWTUNNEL) += lwtunnel.o
obj-$(CONFIG_DST_CACHE) += dst_cache.o
+obj-$(CONFIG_NET_DEVLINK) += devlink.o
diff --git a/net/core/devlink.c b/net/core/devlink.c
new file mode 100644
index 0000000..3f9aea6
--- /dev/null
+++ b/net/core/devlink.c
@@ -0,0 +1,887 @@
+/*
+ * net/core/devlink.c - Network physical/parent device Netlink interface
+ *
+ * Heavily inspired by net/wireless/
+ * Copyright (c) 2016 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2016 Jiri Pirko <jiri@mellanox.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/gfp.h>
+#include <linux/device.h>
+#include <linux/list.h>
+#include <linux/netdevice.h>
+#include <linux/sysfs.h>
+#include <linux/bitops.h>
+#include <rdma/ib_verbs.h>
+#include <net/netlink.h>
+#include <net/genetlink.h>
+#include <net/rtnetlink.h>
+#include <net/net_namespace.h>
+#include <net/sock.h>
+#include <net/devlink.h>
+
+static LIST_HEAD(devlink_list);
+
+/* devlink_mutex
+ *
+ * An overall lock guarding every operation coming from userspace.
+ * It also guards devlink devices list and it is taken when
+ * driver registers/unregisters it.
+ */
+static DEFINE_MUTEX(devlink_mutex);
+
+/* devlink_port_mutex
+ *
+ * Shared lock to guard lists of ports in all devlink devices.
+ */
+static DEFINE_MUTEX(devlink_port_mutex);
+
+static struct net *devlink_net(const struct devlink *devlink)
+{
+ return read_pnet(&devlink->_net);
+}
+
+static void devlink_net_set(struct devlink *devlink, struct net *net)
+{
+ write_pnet(&devlink->_net, net);
+}
+
+static bool devlink_name_exists(struct net *net, const char *name)
+{
+ struct devlink *devlink;
+
+ list_for_each_entry(devlink, &devlink_list, list) {
+ if (strcmp(devlink_name(devlink), name) == 0 &&
+ net_eq(devlink_net(devlink), net))
+ return true;
+ }
+ return false;
+}
+
+static struct devlink *devlink_get_by_index(struct net *net, int index)
+{
+ struct devlink *devlink;
+
+ list_for_each_entry(devlink, &devlink_list, list) {
+ if (devlink->index == index &&
+ net_eq(devlink_net(devlink), net))
+ return devlink;
+ }
+ return NULL;
+}
+
+static struct devlink *devlink_get_from_attrs(struct net *net,
+ struct nlattr **attrs)
+{
+ if (attrs[DEVLINK_ATTR_INDEX]) {
+ u32 index = nla_get_u32(attrs[DEVLINK_ATTR_INDEX]);
+ struct devlink *devlink;
+
+ devlink = devlink_get_by_index(net, index);
+ if (!devlink)
+ return ERR_PTR(-ENODEV);
+ return devlink;
+ }
+ return ERR_PTR(-EINVAL);
+}
+
+static struct devlink *devlink_get_from_info(struct genl_info *info)
+{
+ return devlink_get_from_attrs(genl_info_net(info), info->attrs);
+}
+
+static struct devlink_port *devlink_port_get_by_index(struct devlink *devlink,
+ int port_index)
+{
+ struct devlink_port *devlink_port;
+
+ list_for_each_entry(devlink_port, &devlink->port_list, list) {
+ if (devlink_port->index == port_index)
+ return devlink_port;
+ }
+ return NULL;
+}
+
+static bool devlink_port_index_exists(struct devlink *devlink, int port_index)
+{
+ return devlink_port_get_by_index(devlink, port_index);
+}
+
+static struct devlink_port *devlink_port_get_from_attrs(struct devlink *devlink,
+ struct nlattr **attrs)
+{
+ if (attrs[DEVLINK_ATTR_PORT_INDEX]) {
+ u32 port_index = nla_get_u32(attrs[DEVLINK_ATTR_PORT_INDEX]);
+ struct devlink_port *devlink_port;
+
+ devlink_port = devlink_port_get_by_index(devlink, port_index);
+ if (!devlink_port)
+ return ERR_PTR(-ENODEV);
+ return devlink_port;
+ }
+ return ERR_PTR(-EINVAL);
+}
+
+static struct devlink_port *devlink_port_get_from_info(struct devlink *devlink,
+ struct genl_info *info)
+{
+ return devlink_port_get_from_attrs(devlink, info->attrs);
+}
+
+#define DEVLINK_NL_FLAG_NEED_PORT BIT(0)
+
+static int devlink_nl_pre_doit(const struct genl_ops *ops,
+ struct sk_buff *skb, struct genl_info *info)
+{
+ struct devlink *devlink;
+
+ mutex_lock(&devlink_mutex);
+ devlink = devlink_get_from_info(info);
+ if (IS_ERR(devlink)) {
+ mutex_unlock(&devlink_mutex);
+ return PTR_ERR(devlink);
+ }
+ info->user_ptr[0] = devlink;
+ if (ops->internal_flags & DEVLINK_NL_FLAG_NEED_PORT) {
+ struct devlink_port *devlink_port;
+
+ mutex_lock(&devlink_port_mutex);
+ devlink_port = devlink_port_get_from_info(devlink, info);
+ if (IS_ERR(devlink_port)) {
+ mutex_unlock(&devlink_port_mutex);
+ mutex_unlock(&devlink_mutex);
+ return PTR_ERR(devlink_port);
+ }
+ info->user_ptr[1] = devlink_port;
+ }
+ return 0;
+}
+
+static void devlink_nl_post_doit(const struct genl_ops *ops,
+ struct sk_buff *skb, struct genl_info *info)
+{
+ if (ops->internal_flags & DEVLINK_NL_FLAG_NEED_PORT)
+ mutex_unlock(&devlink_port_mutex);
+ mutex_unlock(&devlink_mutex);
+}
+
+static struct genl_family devlink_nl_family = {
+ .id = GENL_ID_GENERATE,
+ .name = DEVLINK_GENL_NAME,
+ .version = DEVLINK_GENL_VERSION,
+ .maxattr = DEVLINK_ATTR_MAX,
+ .netnsok = true,
+ .pre_doit = devlink_nl_pre_doit,
+ .post_doit = devlink_nl_post_doit,
+};
+
+enum devlink_multicast_groups {
+ DEVLINK_MCGRP_CONFIG,
+};
+
+static const struct genl_multicast_group devlink_nl_mcgrps[] = {
+ [DEVLINK_MCGRP_CONFIG] = { .name = DEVLINK_GENL_MCGRP_CONFIG_NAME },
+};
+
+static int devlink_nl_fill(struct sk_buff *msg, struct devlink *devlink,
+ enum devlink_command cmd, u32 portid,
+ u32 seq, int flags)
+{
+ void *hdr;
+
+ hdr = genlmsg_put(msg, portid, seq, &devlink_nl_family, flags, cmd);
+ if (!hdr)
+ return -EMSGSIZE;
+
+ if (nla_put_u32(msg, DEVLINK_ATTR_INDEX, devlink->index))
+ goto nla_put_failure;
+ if (nla_put_string(msg, DEVLINK_ATTR_NAME, devlink_name(devlink)))
+ goto nla_put_failure;
+
+ if (devlink->dev.parent) {
+ struct device *dev = devlink->dev.parent;
+
+ if (nla_put_string(msg, DEVLINK_ATTR_BUS_NAME, dev->bus->name))
+ goto nla_put_failure;
+ if (nla_put_string(msg, DEVLINK_ATTR_DEV_NAME, dev_name(dev)))
+ goto nla_put_failure;
+ }
+
+ genlmsg_end(msg, hdr);
+ return 0;
+
+nla_put_failure:
+ genlmsg_cancel(msg, hdr);
+ return -EMSGSIZE;
+}
+
+static void devlink_notify(struct devlink *devlink, enum devlink_command cmd)
+{
+ struct sk_buff *msg;
+ int err;
+
+ WARN_ON(cmd != DEVLINK_CMD_NEW && cmd != DEVLINK_CMD_DEL);
+
+ msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
+ if (!msg)
+ return;
+
+ err = devlink_nl_fill(msg, devlink, cmd, 0, 0, 0);
+ if (err) {
+ nlmsg_free(msg);
+ return;
+ }
+
+ genlmsg_multicast_netns(&devlink_nl_family, devlink_net(devlink),
+ msg, 0, DEVLINK_MCGRP_CONFIG, GFP_KERNEL);
+}
+
+static int devlink_nl_port_fill(struct sk_buff *msg, struct devlink *devlink,
+ struct devlink_port *devlink_port,
+ enum devlink_command cmd, u32 portid,
+ u32 seq, int flags)
+{
+ void *hdr;
+
+ hdr = genlmsg_put(msg, portid, seq, &devlink_nl_family, flags, cmd);
+ if (!hdr)
+ return -EMSGSIZE;
+
+ if (nla_put_u32(msg, DEVLINK_ATTR_INDEX, devlink->index))
+ goto nla_put_failure;
+ if (nla_put_u32(msg, DEVLINK_ATTR_PORT_INDEX, devlink_port->index))
+ goto nla_put_failure;
+ if (nla_put_u16(msg, DEVLINK_ATTR_PORT_TYPE, devlink_port->type))
+ goto nla_put_failure;
+ if (devlink_port->desired_type != DEVLINK_PORT_TYPE_NOTSET &&
+ nla_put_u16(msg, DEVLINK_ATTR_PORT_DESIRED_TYPE,
+ devlink_port->desired_type))
+ goto nla_put_failure;
+ if (devlink_port->type == DEVLINK_PORT_TYPE_ETH) {
+ struct net_device *netdev = devlink_port->type_dev;
+
+ if (netdev &&
+ (nla_put_u32(msg, DEVLINK_ATTR_PORT_NETDEV_IFINDEX,
+ netdev->ifindex) ||
+ nla_put_string(msg, DEVLINK_ATTR_PORT_NETDEV_NAME,
+ netdev->name)))
+ goto nla_put_failure;
+ }
+ if (devlink_port->type == DEVLINK_PORT_TYPE_IB) {
+ struct ib_device *ibdev = devlink_port->type_dev;
+
+ if (ibdev &&
+ nla_put_string(msg, DEVLINK_ATTR_PORT_IBDEV_NAME,
+ ibdev->name))
+ goto nla_put_failure;
+ }
+ if (devlink_port->split &&
+ nla_put_u32(msg, DEVLINK_ATTR_PORT_SPLIT_GROUP,
+ devlink_port->split_group))
+ goto nla_put_failure;
+
+ genlmsg_end(msg, hdr);
+ return 0;
+
+nla_put_failure:
+ genlmsg_cancel(msg, hdr);
+ return -EMSGSIZE;
+}
+
+static void devlink_port_notify(struct devlink_port *devlink_port,
+ enum devlink_command cmd)
+{
+ struct devlink *devlink = devlink_port->devlink;
+ struct sk_buff *msg;
+ int err;
+
+ if (!devlink_port->registered)
+ return;
+
+ WARN_ON(cmd != DEVLINK_CMD_PORT_NEW && cmd != DEVLINK_CMD_PORT_DEL);
+
+ msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
+ if (!msg)
+ return;
+
+ err = devlink_nl_port_fill(msg, devlink, devlink_port, cmd, 0, 0, 0);
+ if (err) {
+ nlmsg_free(msg);
+ return;
+ }
+
+ genlmsg_multicast_netns(&devlink_nl_family, devlink_net(devlink),
+ msg, 0, DEVLINK_MCGRP_CONFIG, GFP_KERNEL);
+}
+
+static int devlink_nl_cmd_get_doit(struct sk_buff *skb, struct genl_info *info)
+{
+ struct devlink *devlink = info->user_ptr[0];
+ struct sk_buff *msg;
+ int err;
+
+ msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
+ if (!msg)
+ return -ENOMEM;
+
+ err = devlink_nl_fill(msg, devlink, DEVLINK_CMD_NEW,
+ info->snd_portid, info->snd_seq, 0);
+ if (err) {
+ nlmsg_free(msg);
+ return err;
+ }
+
+ return genlmsg_reply(msg, info);
+}
+
+static int devlink_nl_cmd_get_dumpit(struct sk_buff *msg,
+ struct netlink_callback *cb)
+{
+ struct devlink *devlink;
+ int start = cb->args[0];
+ int idx = 0;
+ int err;
+
+ mutex_lock(&devlink_mutex);
+ list_for_each_entry(devlink, &devlink_list, list) {
+ if (!net_eq(devlink_net(devlink), sock_net(msg->sk)))
+ continue;
+ if (idx < start) {
+ idx++;
+ continue;
+ }
+ err = devlink_nl_fill(msg, devlink, DEVLINK_CMD_NEW,
+ NETLINK_CB(cb->skb).portid,
+ cb->nlh->nlmsg_seq, NLM_F_MULTI);
+ if (err)
+ goto out;
+ idx++;
+ }
+out:
+ mutex_unlock(&devlink_mutex);
+
+ cb->args[0] = idx;
+ return msg->len;
+}
+
+static int devlink_rename(struct devlink *devlink, const char *newname)
+{
+ int err;
+
+ if (strcmp(newname, devlink_name(devlink)) == 0)
+ return 0;
+ if (devlink_name_exists(devlink_net(devlink), newname))
+ return -EINVAL;
+ if (!dev_valid_name(newname))
+ return -EINVAL;
+ err = device_rename(&devlink->dev, newname);
+ if (err)
+ return err;
+ devlink_notify(devlink, DEVLINK_CMD_NEW);
+ return 0;
+}
+
+static int devlink_nl_cmd_set_doit(struct sk_buff *skb, struct genl_info *info)
+{
+ struct devlink *devlink = info->user_ptr[0];
+ int err;
+
+ if (info->attrs[DEVLINK_ATTR_NAME]) {
+ err = devlink_rename(devlink,
+ nla_data(info->attrs[DEVLINK_ATTR_NAME]));
+ if (err)
+ return err;
+ }
+ return 0;
+}
+
+static int devlink_nl_cmd_port_get_doit(struct sk_buff *skb,
+ struct genl_info *info)
+{
+ struct devlink *devlink = info->user_ptr[0];
+ struct devlink_port *devlink_port = info->user_ptr[1];
+ struct sk_buff *msg;
+ int err;
+
+ msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
+ if (!msg)
+ return -ENOMEM;
+
+ err = devlink_nl_port_fill(msg, devlink, devlink_port,
+ DEVLINK_CMD_PORT_NEW,
+ info->snd_portid, info->snd_seq, 0);
+ if (err) {
+ nlmsg_free(msg);
+ return err;
+ }
+
+ return genlmsg_reply(msg, info);
+}
+
+static int devlink_nl_cmd_port_get_dumpit(struct sk_buff *msg,
+ struct netlink_callback *cb)
+{
+ struct devlink *devlink;
+ struct devlink_port *devlink_port;
+ int start = cb->args[0];
+ int idx = 0;
+ int err;
+
+ mutex_lock(&devlink_mutex);
+ mutex_lock(&devlink_port_mutex);
+ list_for_each_entry(devlink, &devlink_list, list) {
+ if (!net_eq(devlink_net(devlink), sock_net(msg->sk)))
+ continue;
+ list_for_each_entry(devlink_port, &devlink->port_list, list) {
+ if (idx < start) {
+ idx++;
+ continue;
+ }
+ err = devlink_nl_port_fill(msg, devlink, devlink_port,
+ DEVLINK_CMD_NEW,
+ NETLINK_CB(cb->skb).portid,
+ cb->nlh->nlmsg_seq,
+ NLM_F_MULTI);
+ if (err)
+ goto out;
+ idx++;
+ }
+ }
+out:
+ mutex_unlock(&devlink_port_mutex);
+ mutex_unlock(&devlink_mutex);
+
+ cb->args[0] = idx;
+ return msg->len;
+}
+
+static int devlink_port_type_set(struct devlink *devlink,
+ struct devlink_port *devlink_port,
+ enum devlink_port_type port_type)
+
+{
+ int err;
+
+ if (devlink->ops && devlink->ops->port_type_set) {
+ if (port_type == DEVLINK_PORT_TYPE_NOTSET)
+ return -EINVAL;
+ err = devlink->ops->port_type_set(devlink_port, port_type);
+ if (err)
+ return err;
+ devlink_port->desired_type = port_type;
+ devlink_port_notify(devlink_port, DEVLINK_CMD_PORT_NEW);
+ return 0;
+ }
+ return -EOPNOTSUPP;
+}
+
+static int devlink_nl_cmd_port_set_doit(struct sk_buff *skb,
+ struct genl_info *info)
+{
+ struct devlink *devlink = info->user_ptr[0];
+ struct devlink_port *devlink_port = info->user_ptr[1];
+ int err;
+
+ if (info->attrs[DEVLINK_ATTR_PORT_TYPE]) {
+ enum devlink_port_type port_type;
+
+ port_type = nla_get_u16(info->attrs[DEVLINK_ATTR_PORT_TYPE]);
+ err = devlink_port_type_set(devlink, devlink_port, port_type);
+ if (err)
+ return err;
+ }
+ return 0;
+}
+
+static int devlink_port_split(struct devlink *devlink,
+ u32 port_index, u32 count)
+
+{
+ if (devlink->ops && devlink->ops->port_split)
+ return devlink->ops->port_split(devlink, port_index, count);
+ return -EOPNOTSUPP;
+}
+
+static int devlink_nl_cmd_port_split_doit(struct sk_buff *skb,
+ struct genl_info *info)
+{
+ struct devlink *devlink = info->user_ptr[0];
+ u32 port_index;
+ u32 count;
+
+ if (!info->attrs[DEVLINK_ATTR_PORT_INDEX] ||
+ !info->attrs[DEVLINK_ATTR_PORT_SPLIT_COUNT])
+ return -EINVAL;
+
+ port_index = nla_get_u32(info->attrs[DEVLINK_ATTR_PORT_INDEX]);
+ count = nla_get_u32(info->attrs[DEVLINK_ATTR_PORT_SPLIT_COUNT]);
+ return devlink_port_split(devlink, port_index, count);
+}
+
+static int devlink_port_unsplit(struct devlink *devlink, u32 port_index)
+
+{
+ if (devlink->ops && devlink->ops->port_unsplit)
+ return devlink->ops->port_unsplit(devlink, port_index);
+ return -EOPNOTSUPP;
+}
+
+static int devlink_nl_cmd_port_unsplit_doit(struct sk_buff *skb,
+ struct genl_info *info)
+{
+ struct devlink *devlink = info->user_ptr[0];
+ u32 port_index;
+
+ if (!info->attrs[DEVLINK_ATTR_PORT_INDEX])
+ return -EINVAL;
+
+ port_index = nla_get_u32(info->attrs[DEVLINK_ATTR_PORT_INDEX]);
+ return devlink_port_unsplit(devlink, port_index);
+}
+
+static const struct nla_policy devlink_nl_policy[DEVLINK_ATTR_MAX + 1] = {
+ [DEVLINK_ATTR_INDEX] = { .type = NLA_U32 },
+ [DEVLINK_ATTR_NAME] = { .type = NLA_NUL_STRING,
+ .len = DEVLINK_ATTR_NAME_MAX_LEN },
+ [DEVLINK_ATTR_PORT_INDEX] = { .type = NLA_U32 },
+ [DEVLINK_ATTR_PORT_TYPE] = { .type = NLA_U16 },
+ [DEVLINK_ATTR_PORT_SPLIT_COUNT] = { .type = NLA_U32 },
+};
+
+static const struct genl_ops devlink_nl_ops[] = {
+ {
+ .cmd = DEVLINK_CMD_GET,
+ .doit = devlink_nl_cmd_get_doit,
+ .dumpit = devlink_nl_cmd_get_dumpit,
+ .policy = devlink_nl_policy,
+ /* can be retrieved by unprivileged users */
+ },
+ {
+ .cmd = DEVLINK_CMD_SET,
+ .doit = devlink_nl_cmd_set_doit,
+ .policy = devlink_nl_policy,
+ .flags = GENL_ADMIN_PERM,
+ },
+ {
+ .cmd = DEVLINK_CMD_PORT_GET,
+ .doit = devlink_nl_cmd_port_get_doit,
+ .dumpit = devlink_nl_cmd_port_get_dumpit,
+ .policy = devlink_nl_policy,
+ .internal_flags = DEVLINK_NL_FLAG_NEED_PORT,
+ /* can be retrieved by unprivileged users */
+ },
+ {
+ .cmd = DEVLINK_CMD_PORT_SET,
+ .doit = devlink_nl_cmd_port_set_doit,
+ .policy = devlink_nl_policy,
+ .flags = GENL_ADMIN_PERM,
+ .internal_flags = DEVLINK_NL_FLAG_NEED_PORT,
+ },
+ {
+ .cmd = DEVLINK_CMD_PORT_SPLIT,
+ .doit = devlink_nl_cmd_port_split_doit,
+ .policy = devlink_nl_policy,
+ .flags = GENL_ADMIN_PERM,
+ },
+ {
+ .cmd = DEVLINK_CMD_PORT_UNSPLIT,
+ .doit = devlink_nl_cmd_port_unsplit_doit,
+ .policy = devlink_nl_policy,
+ .flags = GENL_ADMIN_PERM,
+ },
+};
+
+static struct class devlink_class;
+
+/**
+ * devlink_alloc - Allocate new devlink instance resources
+ *
+ * @ops: ops
+ * @priv_size: size of user private data
+ *
+ * Allocate new devlink instance resources, including devlink index
+ * and name.
+ */
+struct devlink *devlink_alloc(const struct devlink_ops *ops, size_t priv_size)
+{
+ static atomic_t dev_counter = ATOMIC_INIT(0);
+ struct devlink *devlink;
+
+ devlink = kzalloc(sizeof(*devlink) + priv_size, GFP_KERNEL);
+ if (!devlink)
+ return NULL;
+ devlink->ops = ops;
+ devlink_net_set(devlink, &init_net);
+
+different_name:
+ devlink->index = atomic_inc_return(&dev_counter);
+ if (devlink->index < 0) {
+ /* wrapped */
+ atomic_dec(&dev_counter);
+ kfree(devlink);
+ return NULL;
+ }
+ /* atomic_inc_return makes it start at 1, make it start at 0 */
+ devlink->index--;
+
+ dev_set_name(&devlink->dev, DEVLINK_GENL_NAME "%d", devlink->index);
+ if (devlink_name_exists(devlink_net(devlink), devlink_name(devlink)))
+ goto different_name;
+
+ INIT_LIST_HEAD(&devlink->port_list);
+ device_initialize(&devlink->dev);
+ devlink->dev.class = &devlink_class;
+ devlink->dev.platform_data = devlink;
+ return devlink;
+}
+EXPORT_SYMBOL_GPL(devlink_alloc);
+
+/**
+ * devlink_register - Register devlink instance
+ *
+ * @devlink: devlink
+ */
+int devlink_register(struct devlink *devlink)
+{
+ int err;
+
+ mutex_lock(&devlink_mutex);
+ err = device_add(&devlink->dev);
+ if (err)
+ goto unlock;
+ list_add_tail(&devlink->list, &devlink_list);
+ devlink_notify(devlink, DEVLINK_CMD_NEW);
+unlock:
+ mutex_unlock(&devlink_mutex);
+ return err;
+}
+EXPORT_SYMBOL_GPL(devlink_register);
+
+/**
+ * devlink_unregister - Unregister devlink instance
+ *
+ * @devlink: devlink
+ */
+void devlink_unregister(struct devlink *devlink)
+{
+ mutex_lock(&devlink_mutex);
+ devlink_notify(devlink, DEVLINK_CMD_DEL);
+ list_del(&devlink->list);
+ device_del(&devlink->dev);
+ mutex_unlock(&devlink_mutex);
+}
+EXPORT_SYMBOL_GPL(devlink_unregister);
+
+/**
+ * devlink_free - Free devlink instance resources
+ *
+ * @devlink: devlink
+ */
+void devlink_free(struct devlink *devlink)
+{
+ put_device(&devlink->dev);
+}
+EXPORT_SYMBOL_GPL(devlink_free);
+
+/**
+ * devlink_port_register - Register devlink port
+ *
+ * @devlink: devlink
+ * @devlink_port: devlink port
+ * @port_index
+ *
+ * Register devlink port with provided port index. User can use
+ * any indexing, even hw-related one. devlink_port structure
+ * is convenient to be embedded inside user driver private structure.
+ * Note that the caller should take care of zeroing the devlink_port
+ * structure.
+ */
+int devlink_port_register(struct devlink *devlink,
+ struct devlink_port *devlink_port,
+ unsigned int port_index)
+{
+ mutex_lock(&devlink_port_mutex);
+ if (devlink_port_index_exists(devlink, port_index)) {
+ mutex_unlock(&devlink_port_mutex);
+ return -EEXIST;
+ }
+ devlink_port->devlink = devlink;
+ devlink_port->index = port_index;
+ devlink_port->type = DEVLINK_PORT_TYPE_NOTSET;
+ devlink_port->registered = true;
+ list_add_tail(&devlink_port->list, &devlink->port_list);
+ mutex_unlock(&devlink_port_mutex);
+ devlink_port_notify(devlink_port, DEVLINK_CMD_PORT_NEW);
+ return 0;
+}
+EXPORT_SYMBOL_GPL(devlink_port_register);
+
+/**
+ * devlink_port_unregister - Unregister devlink port
+ *
+ * @devlink_port: devlink port
+ */
+void devlink_port_unregister(struct devlink_port *devlink_port)
+{
+ devlink_port_notify(devlink_port, DEVLINK_CMD_PORT_DEL);
+ mutex_lock(&devlink_port_mutex);
+ list_del(&devlink_port->list);
+ mutex_unlock(&devlink_port_mutex);
+}
+EXPORT_SYMBOL_GPL(devlink_port_unregister);
+
+static void __devlink_port_type_set(struct devlink_port *devlink_port,
+ enum devlink_port_type type,
+ void *type_dev)
+{
+ devlink_port->type = type;
+ devlink_port->type_dev = type_dev;
+ devlink_port_notify(devlink_port, DEVLINK_CMD_PORT_NEW);
+}
+
+/**
+ * devlink_port_type_eth_set - Set port type to Ethernet
+ *
+ * @devlink_port: devlink port
+ * @netdev: related netdevice
+ */
+void devlink_port_type_eth_set(struct devlink_port *devlink_port,
+ struct net_device *netdev)
+{
+ return __devlink_port_type_set(devlink_port,
+ DEVLINK_PORT_TYPE_ETH, netdev);
+}
+EXPORT_SYMBOL_GPL(devlink_port_type_eth_set);
+
+/**
+ * devlink_port_type_ib_set - Set port type to InfiniBand
+ *
+ * @devlink_port: devlink port
+ * @ibdev: related IB device
+ */
+void devlink_port_type_ib_set(struct devlink_port *devlink_port,
+ struct ib_device *ibdev)
+{
+ return __devlink_port_type_set(devlink_port,
+ DEVLINK_PORT_TYPE_IB, ibdev);
+}
+EXPORT_SYMBOL_GPL(devlink_port_type_ib_set);
+
+/**
+ * devlink_port_type_clear - Clear port type
+ *
+ * @devlink_port: devlink port
+ */
+void devlink_port_type_clear(struct devlink_port *devlink_port)
+{
+ return __devlink_port_type_set(devlink_port,
+ DEVLINK_PORT_TYPE_NOTSET, NULL);
+}
+EXPORT_SYMBOL_GPL(devlink_port_type_clear);
+
+/**
+ * devlink_port_split_set - Set port is split
+ *
+ * @devlink_port: devlink port
+ * @split_group: split group - identifies group split port is part of
+ */
+void devlink_port_split_set(struct devlink_port *devlink_port,
+ u32 split_group)
+{
+ devlink_port->split = true;
+ devlink_port->split_group = split_group;
+ devlink_port_notify(devlink_port, DEVLINK_CMD_PORT_NEW);
+}
+EXPORT_SYMBOL_GPL(devlink_port_split_set);
+
+static void __devlink_free(struct devlink *devlink)
+{
+ kfree(devlink);
+}
+
+static struct devlink *dev_to_devlink(struct device *dev)
+{
+ return container_of(dev, struct devlink, dev);
+}
+
+static ssize_t index_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ return sprintf(buf, "%d\n", dev_to_devlink(dev)->index);
+}
+static DEVICE_ATTR_RO(index);
+
+static ssize_t name_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ return sprintf(buf, "%s\n", dev_name(dev));
+}
+static DEVICE_ATTR_RO(name);
+
+static struct attribute *devlink_attrs[] = {
+ &dev_attr_index.attr,
+ &dev_attr_name.attr,
+ NULL,
+};
+ATTRIBUTE_GROUPS(devlink);
+
+static void devlink_dev_release(struct device *dev)
+{
+ __devlink_free(dev_to_devlink(dev));
+}
+
+static const void *devlink_namespace(struct device *dev)
+{
+ return devlink_net(dev_to_devlink(dev));
+}
+
+static struct class devlink_class = {
+ .name = DEVLINK_GENL_NAME,
+ .owner = THIS_MODULE,
+ .dev_release = devlink_dev_release,
+ .dev_groups = devlink_groups,
+ .ns_type = &net_ns_type_operations,
+ .namespace = devlink_namespace,
+};
+
+static int __init devlink_module_init(void)
+{
+ int err;
+
+ err = class_register(&devlink_class);
+ if (err)
+ return err;
+ err = genl_register_family_with_ops_groups(&devlink_nl_family,
+ devlink_nl_ops,
+ devlink_nl_mcgrps);
+ if (err)
+ goto err_genl_register;
+ return 0;
+
+err_genl_register:
+ class_unregister(&devlink_class);
+ return err;
+}
+
+static void __exit devlink_module_exit(void)
+{
+ genl_unregister_family(&devlink_nl_family);
+ class_unregister(&devlink_class);
+}
+
+module_init(devlink_module_init);
+module_exit(devlink_module_exit);
+
+MODULE_LICENSE("GPL v2");
+MODULE_AUTHOR("Jiri Pirko <jiri@mellanox.com>");
+MODULE_DESCRIPTION("Network physical device Netlink interface");
+MODULE_ALIAS_GENL_FAMILY(DEVLINK_GENL_NAME);
--
2.5.0
^ permalink raw reply related [flat|nested] 19+ messages in thread* Re: [patch net-next v2 1/9] Introduce devlink infrastructure
2016-02-23 15:51 ` [patch net-next v2 1/9] Introduce devlink infrastructure Jiri Pirko
@ 2016-02-23 21:19 ` Jarod Wilson
2016-02-24 7:19 ` Jiri Pirko
0 siblings, 1 reply; 19+ messages in thread
From: Jarod Wilson @ 2016-02-23 21:19 UTC (permalink / raw)
To: Jiri Pirko
Cc: netdev, davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
On Tue, Feb 23, 2016 at 04:51:26PM +0100, Jiri Pirko wrote:
> From: Jiri Pirko <jiri@mellanox.com>
>
> Introduce devlink infrastructure for drivers to register and expose to
> userspace via generic Netlink interface.
>
> There are two basic objects defined:
> devlink - one instance for every "parent device", for example switch ASIC
> devlink port - one instance for every physical port of the device.
>
> This initial portion implements basic get/dump of objects to userspace.
> Also, port splitter and port type setting is implemented.
...
> +/**
> + * devlink_alloc - Allocate new devlink instance resources
> + *
> + * @ops: ops
> + * @priv_size: size of user private data
> + *
> + * Allocate new devlink instance resources, including devlink index
> + * and name.
> + */
> +struct devlink *devlink_alloc(const struct devlink_ops *ops, size_t priv_size)
> +{
> + static atomic_t dev_counter = ATOMIC_INIT(0);
> + struct devlink *devlink;
> +
> + devlink = kzalloc(sizeof(*devlink) + priv_size, GFP_KERNEL);
> + if (!devlink)
> + return NULL;
> + devlink->ops = ops;
> + devlink_net_set(devlink, &init_net);
> +
> +different_name:
> + devlink->index = atomic_inc_return(&dev_counter);
> + if (devlink->index < 0) {
> + /* wrapped */
> + atomic_dec(&dev_counter);
> + kfree(devlink);
> + return NULL;
> + }
> + /* atomic_inc_return makes it start at 1, make it start at 0 */
> + devlink->index--;
> +
> + dev_set_name(&devlink->dev, DEVLINK_GENL_NAME "%d", devlink->index);
> + if (devlink_name_exists(devlink_net(devlink), devlink_name(devlink)))
> + goto different_name;
> +
> + INIT_LIST_HEAD(&devlink->port_list);
> + device_initialize(&devlink->dev);
> + devlink->dev.class = &devlink_class;
> + devlink->dev.platform_data = devlink;
> + return devlink;
> +}
> +EXPORT_SYMBOL_GPL(devlink_alloc);
The increment then decrement bit feels... Clunky and a little confusing.
Why not just atomic_read(&dev_counter) before different_name:, then an
atomic_inc_return(&dev_counter) under the if clause before the goto, and
eliminate the devlink->index-- bit?
--
Jarod Wilson
jarod@redhat.com
^ permalink raw reply [flat|nested] 19+ messages in thread* Re: [patch net-next v2 1/9] Introduce devlink infrastructure
2016-02-23 21:19 ` Jarod Wilson
@ 2016-02-24 7:19 ` Jiri Pirko
2016-02-24 16:12 ` Jarod Wilson
0 siblings, 1 reply; 19+ messages in thread
From: Jiri Pirko @ 2016-02-24 7:19 UTC (permalink / raw)
To: Jarod Wilson
Cc: netdev, davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
Tue, Feb 23, 2016 at 10:19:48PM CET, jarod@redhat.com wrote:
>On Tue, Feb 23, 2016 at 04:51:26PM +0100, Jiri Pirko wrote:
>> From: Jiri Pirko <jiri@mellanox.com>
>>
>> Introduce devlink infrastructure for drivers to register and expose to
>> userspace via generic Netlink interface.
>>
>> There are two basic objects defined:
>> devlink - one instance for every "parent device", for example switch ASIC
>> devlink port - one instance for every physical port of the device.
>>
>> This initial portion implements basic get/dump of objects to userspace.
>> Also, port splitter and port type setting is implemented.
>...
>> +/**
>> + * devlink_alloc - Allocate new devlink instance resources
>> + *
>> + * @ops: ops
>> + * @priv_size: size of user private data
>> + *
>> + * Allocate new devlink instance resources, including devlink index
>> + * and name.
>> + */
>> +struct devlink *devlink_alloc(const struct devlink_ops *ops, size_t priv_size)
>> +{
>> + static atomic_t dev_counter = ATOMIC_INIT(0);
>> + struct devlink *devlink;
>> +
>> + devlink = kzalloc(sizeof(*devlink) + priv_size, GFP_KERNEL);
>> + if (!devlink)
>> + return NULL;
>> + devlink->ops = ops;
>> + devlink_net_set(devlink, &init_net);
>> +
>> +different_name:
>> + devlink->index = atomic_inc_return(&dev_counter);
>> + if (devlink->index < 0) {
>> + /* wrapped */
>> + atomic_dec(&dev_counter);
>> + kfree(devlink);
>> + return NULL;
>> + }
>> + /* atomic_inc_return makes it start at 1, make it start at 0 */
>> + devlink->index--;
>> +
>> + dev_set_name(&devlink->dev, DEVLINK_GENL_NAME "%d", devlink->index);
>> + if (devlink_name_exists(devlink_net(devlink), devlink_name(devlink)))
>> + goto different_name;
>> +
>> + INIT_LIST_HEAD(&devlink->port_list);
>> + device_initialize(&devlink->dev);
>> + devlink->dev.class = &devlink_class;
>> + devlink->dev.platform_data = devlink;
>> + return devlink;
>> +}
>> +EXPORT_SYMBOL_GPL(devlink_alloc);
>
>The increment then decrement bit feels... Clunky and a little confusing.
>Why not just atomic_read(&dev_counter) before different_name:, then an
>atomic_inc_return(&dev_counter) under the if clause before the goto, and
>eliminate the devlink->index-- bit?
If I split it it would not be atomic op anymore. But the
devlink->index-- is just to adjust the start, nothing else.
This is copied from nl80211 code.
^ permalink raw reply [flat|nested] 19+ messages in thread* Re: [patch net-next v2 1/9] Introduce devlink infrastructure
2016-02-24 7:19 ` Jiri Pirko
@ 2016-02-24 16:12 ` Jarod Wilson
0 siblings, 0 replies; 19+ messages in thread
From: Jarod Wilson @ 2016-02-24 16:12 UTC (permalink / raw)
To: Jiri Pirko
Cc: netdev, davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
On Wed, Feb 24, 2016 at 08:19:58AM +0100, Jiri Pirko wrote:
> Tue, Feb 23, 2016 at 10:19:48PM CET, jarod@redhat.com wrote:
> >On Tue, Feb 23, 2016 at 04:51:26PM +0100, Jiri Pirko wrote:
> >> From: Jiri Pirko <jiri@mellanox.com>
> >>
> >> Introduce devlink infrastructure for drivers to register and expose to
> >> userspace via generic Netlink interface.
> >>
> >> There are two basic objects defined:
> >> devlink - one instance for every "parent device", for example switch ASIC
> >> devlink port - one instance for every physical port of the device.
> >>
> >> This initial portion implements basic get/dump of objects to userspace.
> >> Also, port splitter and port type setting is implemented.
> >...
> >> +/**
> >> + * devlink_alloc - Allocate new devlink instance resources
> >> + *
> >> + * @ops: ops
> >> + * @priv_size: size of user private data
> >> + *
> >> + * Allocate new devlink instance resources, including devlink index
> >> + * and name.
> >> + */
> >> +struct devlink *devlink_alloc(const struct devlink_ops *ops, size_t priv_size)
> >> +{
> >> + static atomic_t dev_counter = ATOMIC_INIT(0);
> >> + struct devlink *devlink;
> >> +
> >> + devlink = kzalloc(sizeof(*devlink) + priv_size, GFP_KERNEL);
> >> + if (!devlink)
> >> + return NULL;
> >> + devlink->ops = ops;
> >> + devlink_net_set(devlink, &init_net);
> >> +
> >> +different_name:
> >> + devlink->index = atomic_inc_return(&dev_counter);
> >> + if (devlink->index < 0) {
> >> + /* wrapped */
> >> + atomic_dec(&dev_counter);
> >> + kfree(devlink);
> >> + return NULL;
> >> + }
> >> + /* atomic_inc_return makes it start at 1, make it start at 0 */
> >> + devlink->index--;
> >> +
> >> + dev_set_name(&devlink->dev, DEVLINK_GENL_NAME "%d", devlink->index);
> >> + if (devlink_name_exists(devlink_net(devlink), devlink_name(devlink)))
> >> + goto different_name;
> >> +
> >> + INIT_LIST_HEAD(&devlink->port_list);
> >> + device_initialize(&devlink->dev);
> >> + devlink->dev.class = &devlink_class;
> >> + devlink->dev.platform_data = devlink;
> >> + return devlink;
> >> +}
> >> +EXPORT_SYMBOL_GPL(devlink_alloc);
> >
> >The increment then decrement bit feels... Clunky and a little confusing.
> >Why not just atomic_read(&dev_counter) before different_name:, then an
> >atomic_inc_return(&dev_counter) under the if clause before the goto, and
> >eliminate the devlink->index-- bit?
>
> If I split it it would not be atomic op anymore. But the
> devlink->index-- is just to adjust the start, nothing else.
> This is copied from nl80211 code.
Argh, I'm a moron. You need to increment dev_counter, can't just read it's
current value. Still fugly, but okay, does actually serve a purpose this
way.
--
Jarod Wilson
jarod@redhat.com
^ permalink raw reply [flat|nested] 19+ messages in thread
* [patch net-next v2 2/9] mlx4: Implement devlink interface
2016-02-23 15:51 [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it Jiri Pirko
2016-02-23 15:51 ` [patch net-next v2 1/9] Introduce devlink infrastructure Jiri Pirko
@ 2016-02-23 15:51 ` Jiri Pirko
2016-02-23 15:51 ` [patch net-next v2 3/9] mlx4: Implement port type setting via " Jiri Pirko
` (7 subsequent siblings)
9 siblings, 0 replies; 19+ messages in thread
From: Jiri Pirko @ 2016-02-23 15:51 UTC (permalink / raw)
To: netdev
Cc: davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
From: Jiri Pirko <jiri@mellanox.com>
Implement newly introduced devlink interface. Add devlink port instances
for every port and set the port types accordingly.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
drivers/infiniband/hw/mlx4/main.c | 7 ++++
drivers/net/ethernet/mellanox/mlx4/en_netdev.c | 8 ++++-
drivers/net/ethernet/mellanox/mlx4/intf.c | 9 ++++++
drivers/net/ethernet/mellanox/mlx4/main.c | 45 +++++++++++++++++++-------
drivers/net/ethernet/mellanox/mlx4/mlx4.h | 2 ++
include/linux/mlx4/driver.h | 3 ++
6 files changed, 61 insertions(+), 13 deletions(-)
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 1c7ab6c..a15a7b3 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -41,6 +41,7 @@
#include <linux/if_vlan.h>
#include <net/ipv6.h>
#include <net/addrconf.h>
+#include <net/devlink.h>
#include <rdma/ib_smi.h>
#include <rdma/ib_user_verbs.h>
@@ -2519,6 +2520,9 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
}
ibdev->ib_active = true;
+ mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_IB)
+ devlink_port_type_ib_set(mlx4_get_devlink_port(dev, i),
+ &ibdev->ib_dev);
if (mlx4_is_mfunc(ibdev->dev))
init_pkeys(ibdev);
@@ -2643,7 +2647,10 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr)
{
struct mlx4_ib_dev *ibdev = ibdev_ptr;
int p;
+ int i;
+ mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_IB)
+ devlink_port_type_clear(mlx4_get_devlink_port(dev, i));
ibdev->ib_active = false;
flush_workqueue(wq);
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
index 96d95cb..e26b110 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
@@ -40,6 +40,7 @@
#include <net/ip.h>
#include <net/busy_poll.h>
#include <net/vxlan.h>
+#include <net/devlink.h>
#include <linux/mlx4/driver.h>
#include <linux/mlx4/device.h>
@@ -2033,8 +2034,11 @@ void mlx4_en_destroy_netdev(struct net_device *dev)
en_dbg(DRV, priv, "Destroying netdev on port:%d\n", priv->port);
/* Unregister device - this will close the port if it was up */
- if (priv->registered)
+ if (priv->registered) {
+ devlink_port_type_clear(mlx4_get_devlink_port(mdev->dev,
+ priv->port));
unregister_netdev(dev);
+ }
if (priv->allocated)
mlx4_free_hwq_res(mdev->dev, &priv->res, MLX4_EN_PAGE_SIZE);
@@ -3051,6 +3055,8 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port,
}
priv->registered = 1;
+ devlink_port_type_eth_set(mlx4_get_devlink_port(mdev->dev, priv->port),
+ dev);
return 0;
diff --git a/drivers/net/ethernet/mellanox/mlx4/intf.c b/drivers/net/ethernet/mellanox/mlx4/intf.c
index 0472941..dec77d6 100644
--- a/drivers/net/ethernet/mellanox/mlx4/intf.c
+++ b/drivers/net/ethernet/mellanox/mlx4/intf.c
@@ -34,6 +34,7 @@
#include <linux/slab.h>
#include <linux/export.h>
#include <linux/errno.h>
+#include <net/devlink.h>
#include "mlx4.h"
@@ -249,3 +250,11 @@ void *mlx4_get_protocol_dev(struct mlx4_dev *dev, enum mlx4_protocol proto, int
return result;
}
EXPORT_SYMBOL_GPL(mlx4_get_protocol_dev);
+
+struct devlink_port *mlx4_get_devlink_port(struct mlx4_dev *dev, int port)
+{
+ struct mlx4_port_info *info = &mlx4_priv(dev)->port[port];
+
+ return &info->devlink_port;
+}
+EXPORT_SYMBOL_GPL(mlx4_get_devlink_port);
diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
index 2cc3c62..ca88f38 100644
--- a/drivers/net/ethernet/mellanox/mlx4/main.c
+++ b/drivers/net/ethernet/mellanox/mlx4/main.c
@@ -42,6 +42,7 @@
#include <linux/io-mapping.h>
#include <linux/delay.h>
#include <linux/kmod.h>
+#include <net/devlink.h>
#include <linux/mlx4/device.h>
#include <linux/mlx4/doorbell.h>
@@ -2881,8 +2882,13 @@ no_msi:
static int mlx4_init_port_info(struct mlx4_dev *dev, int port)
{
+ struct devlink *devlink = priv_to_devlink(mlx4_priv(dev));
struct mlx4_port_info *info = &mlx4_priv(dev)->port[port];
- int err = 0;
+ int err;
+
+ err = devlink_port_register(devlink, &info->devlink_port, port);
+ if (err)
+ return err;
info->dev = dev;
info->port = port;
@@ -2907,6 +2913,7 @@ static int mlx4_init_port_info(struct mlx4_dev *dev, int port)
err = device_create_file(&dev->persist->pdev->dev, &info->port_attr);
if (err) {
mlx4_err(dev, "Failed to create file for port %d\n", port);
+ devlink_port_unregister(&info->devlink_port);
info->port = -1;
}
@@ -3680,21 +3687,23 @@ err_disable_pdev:
static int mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
{
+ struct devlink *devlink;
struct mlx4_priv *priv;
struct mlx4_dev *dev;
int ret;
printk_once(KERN_INFO "%s", mlx4_version);
- priv = kzalloc(sizeof(*priv), GFP_KERNEL);
- if (!priv)
+ devlink = devlink_alloc(NULL, sizeof(*priv));
+ if (!devlink)
return -ENOMEM;
+ priv = devlink_priv(devlink);
dev = &priv->dev;
dev->persist = kzalloc(sizeof(*dev->persist), GFP_KERNEL);
if (!dev->persist) {
- kfree(priv);
- return -ENOMEM;
+ ret = -ENOMEM;
+ goto err_devlink_free;
}
dev->persist->pdev = pdev;
dev->persist->dev = dev;
@@ -3703,14 +3712,24 @@ static int mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
mutex_init(&dev->persist->device_state_mutex);
mutex_init(&dev->persist->interface_state_mutex);
+ set_devlink_dev(devlink, &pdev->dev);
+ ret = devlink_register(devlink);
+ if (ret)
+ goto err_persist_free;
+
ret = __mlx4_init_one(pdev, id->driver_data, priv);
- if (ret) {
- kfree(dev->persist);
- kfree(priv);
- } else {
- pci_save_state(pdev);
- }
+ if (ret)
+ goto err_devlink_unregister;
+ pci_save_state(pdev);
+ return 0;
+
+err_devlink_unregister:
+ devlink_unregister(devlink);
+err_persist_free:
+ kfree(dev->persist);
+err_devlink_free:
+ devlink_free(devlink);
return ret;
}
@@ -3811,6 +3830,7 @@ static void mlx4_remove_one(struct pci_dev *pdev)
struct mlx4_dev_persistent *persist = pci_get_drvdata(pdev);
struct mlx4_dev *dev = persist->dev;
struct mlx4_priv *priv = mlx4_priv(dev);
+ struct devlink *devlink = priv_to_devlink(priv);
int active_vfs = 0;
mutex_lock(&persist->interface_state_mutex);
@@ -3841,8 +3861,9 @@ static void mlx4_remove_one(struct pci_dev *pdev)
pci_release_regions(pdev);
pci_disable_device(pdev);
+ devlink_unregister(devlink);
kfree(dev->persist);
- kfree(priv);
+ devlink_free(devlink);
pci_set_drvdata(pdev, NULL);
}
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
index 7baef52..ef96831 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
@@ -45,6 +45,7 @@
#include <linux/workqueue.h>
#include <linux/interrupt.h>
#include <linux/spinlock.h>
+#include <net/devlink.h>
#include <linux/mlx4/device.h>
#include <linux/mlx4/driver.h>
@@ -828,6 +829,7 @@ struct mlx4_port_info {
struct mlx4_roce_gid_table gid_table;
int base_qpn;
struct cpu_rmap *rmap;
+ struct devlink_port devlink_port;
};
struct mlx4_sense {
diff --git a/include/linux/mlx4/driver.h b/include/linux/mlx4/driver.h
index 2e8af00..bd0e707 100644
--- a/include/linux/mlx4/driver.h
+++ b/include/linux/mlx4/driver.h
@@ -33,6 +33,7 @@
#ifndef MLX4_DRIVER_H
#define MLX4_DRIVER_H
+#include <net/devlink.h>
#include <linux/mlx4/device.h>
struct mlx4_dev;
@@ -89,6 +90,8 @@ int mlx4_port_map_set(struct mlx4_dev *dev, struct mlx4_port_map *v2p);
void *mlx4_get_protocol_dev(struct mlx4_dev *dev, enum mlx4_protocol proto, int port);
+struct devlink_port *mlx4_get_devlink_port(struct mlx4_dev *dev, int port);
+
static inline u64 mlx4_mac_to_u64(u8 *addr)
{
u64 mac = 0;
--
2.5.0
^ permalink raw reply related [flat|nested] 19+ messages in thread* [patch net-next v2 3/9] mlx4: Implement port type setting via devlink interface
2016-02-23 15:51 [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it Jiri Pirko
2016-02-23 15:51 ` [patch net-next v2 1/9] Introduce devlink infrastructure Jiri Pirko
2016-02-23 15:51 ` [patch net-next v2 2/9] mlx4: Implement devlink interface Jiri Pirko
@ 2016-02-23 15:51 ` Jiri Pirko
2016-02-23 15:51 ` [patch net-next v2 4/9] mlxsw: Implement " Jiri Pirko
` (6 subsequent siblings)
9 siblings, 0 replies; 19+ messages in thread
From: Jiri Pirko @ 2016-02-23 15:51 UTC (permalink / raw)
To: netdev
Cc: davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
From: Jiri Pirko <jiri@mellanox.com>
So far, there has been an mlx4-specific sysfs file allowing user to
change port type to either Ethernet of InfiniBand. This is very
inconvenient.
Allow to expose the same ability to set port type in a generic way
using devlink interface.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/main.c | 86 +++++++++++++++++++++++--------
1 file changed, 65 insertions(+), 21 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
index ca88f38..522d49e 100644
--- a/drivers/net/ethernet/mellanox/mlx4/main.c
+++ b/drivers/net/ethernet/mellanox/mlx4/main.c
@@ -1082,36 +1082,20 @@ static ssize_t show_port_type(struct device *dev,
return strlen(buf);
}
-static ssize_t set_port_type(struct device *dev,
- struct device_attribute *attr,
- const char *buf, size_t count)
+static int __set_port_type(struct mlx4_port_info *info,
+ enum mlx4_port_type port_type)
{
- struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info,
- port_attr);
struct mlx4_dev *mdev = info->dev;
struct mlx4_priv *priv = mlx4_priv(mdev);
enum mlx4_port_type types[MLX4_MAX_PORTS];
enum mlx4_port_type new_types[MLX4_MAX_PORTS];
- static DEFINE_MUTEX(set_port_type_mutex);
int i;
int err = 0;
- mutex_lock(&set_port_type_mutex);
-
- if (!strcmp(buf, "ib\n"))
- info->tmp_type = MLX4_PORT_TYPE_IB;
- else if (!strcmp(buf, "eth\n"))
- info->tmp_type = MLX4_PORT_TYPE_ETH;
- else if (!strcmp(buf, "auto\n"))
- info->tmp_type = MLX4_PORT_TYPE_AUTO;
- else {
- mlx4_err(mdev, "%s is not supported port type\n", buf);
- err = -EINVAL;
- goto err_out;
- }
-
mlx4_stop_sense(mdev);
mutex_lock(&priv->port_mutex);
+ info->tmp_type = port_type;
+
/* Possible type is always the one that was delivered */
mdev->caps.possible_type[info->port] = info->tmp_type;
@@ -1153,6 +1137,37 @@ static ssize_t set_port_type(struct device *dev,
out:
mlx4_start_sense(mdev);
mutex_unlock(&priv->port_mutex);
+
+ return err;
+}
+
+static ssize_t set_port_type(struct device *dev,
+ struct device_attribute *attr,
+ const char *buf, size_t count)
+{
+ struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info,
+ port_attr);
+ struct mlx4_dev *mdev = info->dev;
+ enum mlx4_port_type port_type;
+ static DEFINE_MUTEX(set_port_type_mutex);
+ int err;
+
+ mutex_lock(&set_port_type_mutex);
+
+ if (!strcmp(buf, "ib\n")) {
+ port_type = MLX4_PORT_TYPE_IB;
+ } else if (!strcmp(buf, "eth\n")) {
+ port_type = MLX4_PORT_TYPE_ETH;
+ } else if (!strcmp(buf, "auto\n")) {
+ port_type = MLX4_PORT_TYPE_AUTO;
+ } else {
+ mlx4_err(mdev, "%s is not supported port type\n", buf);
+ err = -EINVAL;
+ goto err_out;
+ }
+
+ err = __set_port_type(info, port_type);
+
err_out:
mutex_unlock(&set_port_type_mutex);
@@ -3685,6 +3700,35 @@ err_disable_pdev:
return err;
}
+static int mlx4_devlink_port_type_set(struct devlink_port *devlink_port,
+ enum devlink_port_type port_type)
+{
+ struct mlx4_port_info *info = container_of(devlink_port,
+ struct mlx4_port_info,
+ devlink_port);
+ enum mlx4_port_type mlx4_port_type;
+
+ switch (port_type) {
+ case DEVLINK_PORT_TYPE_AUTO:
+ mlx4_port_type = MLX4_PORT_TYPE_AUTO;
+ break;
+ case DEVLINK_PORT_TYPE_ETH:
+ mlx4_port_type = MLX4_PORT_TYPE_ETH;
+ break;
+ case DEVLINK_PORT_TYPE_IB:
+ mlx4_port_type = MLX4_PORT_TYPE_IB;
+ break;
+ default:
+ return -EOPNOTSUPP;
+ }
+
+ return __set_port_type(info, mlx4_port_type);
+}
+
+static const struct devlink_ops mlx4_devlink_ops = {
+ .port_type_set = mlx4_devlink_port_type_set,
+};
+
static int mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
{
struct devlink *devlink;
@@ -3694,7 +3738,7 @@ static int mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
printk_once(KERN_INFO "%s", mlx4_version);
- devlink = devlink_alloc(NULL, sizeof(*priv));
+ devlink = devlink_alloc(&mlx4_devlink_ops, sizeof(*priv));
if (!devlink)
return -ENOMEM;
priv = devlink_priv(devlink);
--
2.5.0
^ permalink raw reply related [flat|nested] 19+ messages in thread* [patch net-next v2 4/9] mlxsw: Implement devlink interface
2016-02-23 15:51 [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it Jiri Pirko
` (2 preceding siblings ...)
2016-02-23 15:51 ` [patch net-next v2 3/9] mlx4: Implement port type setting via " Jiri Pirko
@ 2016-02-23 15:51 ` Jiri Pirko
2016-02-23 15:51 ` [patch net-next v2 5/9] mlxsw: core: Add devlink port splitter callbacks Jiri Pirko
` (5 subsequent siblings)
9 siblings, 0 replies; 19+ messages in thread
From: Jiri Pirko @ 2016-02-23 15:51 UTC (permalink / raw)
To: netdev
Cc: davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
From: Jiri Pirko <jiri@mellanox.com>
Implement newly introduced devlink interface. Add devlink port instances
for every port and set the port types accordingly.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
v1->v2:
-removed SET_NETDEV_DEV set to devlink dev
---
drivers/net/ethernet/mellanox/mlxsw/core.c | 24 ++++++++++++++++++------
drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 19 +++++++++++++++++++
drivers/net/ethernet/mellanox/mlxsw/spectrum.h | 2 ++
drivers/net/ethernet/mellanox/mlxsw/switchx2.c | 20 ++++++++++++++++++++
4 files changed, 59 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core.c b/drivers/net/ethernet/mellanox/mlxsw/core.c
index 22379eb..57d9655 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/core.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/core.c
@@ -56,6 +56,7 @@
#include <linux/rcupdate.h>
#include <linux/slab.h>
#include <asm/byteorder.h>
+#include <net/devlink.h>
#include "core.h"
#include "item.h"
@@ -791,6 +792,7 @@ int mlxsw_core_bus_device_register(const struct mlxsw_bus_info *mlxsw_bus_info,
const char *device_kind = mlxsw_bus_info->device_kind;
struct mlxsw_core *mlxsw_core;
struct mlxsw_driver *mlxsw_driver;
+ struct devlink *devlink;
size_t alloc_size;
int err;
@@ -798,12 +800,13 @@ int mlxsw_core_bus_device_register(const struct mlxsw_bus_info *mlxsw_bus_info,
if (!mlxsw_driver)
return -EINVAL;
alloc_size = sizeof(*mlxsw_core) + mlxsw_driver->priv_size;
- mlxsw_core = kzalloc(alloc_size, GFP_KERNEL);
- if (!mlxsw_core) {
+ devlink = devlink_alloc(NULL, alloc_size);
+ if (!devlink) {
err = -ENOMEM;
- goto err_core_alloc;
+ goto err_devlink_alloc;
}
+ mlxsw_core = devlink_priv(devlink);
INIT_LIST_HEAD(&mlxsw_core->rx_listener_list);
INIT_LIST_HEAD(&mlxsw_core->event_listener_list);
mlxsw_core->driver = mlxsw_driver;
@@ -841,6 +844,11 @@ int mlxsw_core_bus_device_register(const struct mlxsw_bus_info *mlxsw_bus_info,
if (err)
goto err_hwmon_init;
+ set_devlink_dev(devlink, mlxsw_bus_info->dev);
+ err = devlink_register(devlink);
+ if (err)
+ goto err_devlink_register;
+
err = mlxsw_driver->init(mlxsw_core->driver_priv, mlxsw_core,
mlxsw_bus_info);
if (err)
@@ -855,6 +863,8 @@ int mlxsw_core_bus_device_register(const struct mlxsw_bus_info *mlxsw_bus_info,
err_debugfs_init:
mlxsw_core->driver->fini(mlxsw_core->driver_priv);
err_driver_init:
+ devlink_unregister(devlink);
+err_devlink_register:
err_hwmon_init:
mlxsw_emad_fini(mlxsw_core);
err_emad_init:
@@ -864,8 +874,8 @@ err_bus_init:
err_alloc_lag_mapping:
free_percpu(mlxsw_core->pcpu_stats);
err_alloc_stats:
- kfree(mlxsw_core);
-err_core_alloc:
+ devlink_free(devlink);
+err_devlink_alloc:
mlxsw_core_driver_put(device_kind);
return err;
}
@@ -874,14 +884,16 @@ EXPORT_SYMBOL(mlxsw_core_bus_device_register);
void mlxsw_core_bus_device_unregister(struct mlxsw_core *mlxsw_core)
{
const char *device_kind = mlxsw_core->bus_info->device_kind;
+ struct devlink *devlink = priv_to_devlink(mlxsw_core);
mlxsw_core_debugfs_fini(mlxsw_core);
mlxsw_core->driver->fini(mlxsw_core->driver_priv);
+ devlink_unregister(devlink);
mlxsw_emad_fini(mlxsw_core);
mlxsw_core->bus->fini(mlxsw_core->bus_priv);
kfree(mlxsw_core->lag.mapping);
free_percpu(mlxsw_core->pcpu_stats);
- kfree(mlxsw_core);
+ devlink_free(devlink);
mlxsw_core_driver_put(device_kind);
}
EXPORT_SYMBOL(mlxsw_core_bus_device_unregister);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 09ce451..20e6783 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -49,6 +49,7 @@
#include <linux/jiffies.h>
#include <linux/bitops.h>
#include <linux/list.h>
+#include <net/devlink.h>
#include <net/switchdev.h>
#include <generated/utsrelease.h>
@@ -1351,7 +1352,9 @@ static const struct ethtool_ops mlxsw_sp_port_ethtool_ops = {
static int mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port)
{
+ struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
struct mlxsw_sp_port *mlxsw_sp_port;
+ struct devlink_port *devlink_port;
struct net_device *dev;
bool usable;
size_t bytes;
@@ -1417,6 +1420,14 @@ static int mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port)
goto port_not_usable;
}
+ devlink_port = &mlxsw_sp_port->devlink_port;
+ err = devlink_port_register(devlink, devlink_port, local_port);
+ if (err) {
+ dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to register devlink port\n",
+ mlxsw_sp_port->local_port);
+ goto err_devlink_port_register;
+ }
+
err = mlxsw_sp_port_system_port_mapping_set(mlxsw_sp_port);
if (err) {
dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to set system port mapping\n",
@@ -1457,6 +1468,8 @@ static int mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port)
goto err_register_netdev;
}
+ devlink_port_type_eth_set(devlink_port, dev);
+
err = mlxsw_sp_port_vlan_init(mlxsw_sp_port);
if (err)
goto err_port_vlan_init;
@@ -1472,6 +1485,8 @@ err_port_admin_status_set:
err_port_mtu_set:
err_port_swid_set:
err_port_system_port_mapping_set:
+ devlink_port_unregister(&mlxsw_sp_port->devlink_port);
+err_devlink_port_register:
port_not_usable:
err_port_module_check:
err_dev_addr_init:
@@ -1505,10 +1520,14 @@ static void mlxsw_sp_port_vports_fini(struct mlxsw_sp_port *mlxsw_sp_port)
static void mlxsw_sp_port_remove(struct mlxsw_sp *mlxsw_sp, u8 local_port)
{
struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp->ports[local_port];
+ struct devlink_port *devlink_port;
if (!mlxsw_sp_port)
return;
+ devlink_port = &mlxsw_sp_port->devlink_port;
+ devlink_port_type_clear(devlink_port);
unregister_netdev(mlxsw_sp_port->dev); /* This calls ndo_stop */
+ devlink_port_unregister(devlink_port);
mlxsw_sp_port_vports_fini(mlxsw_sp_port);
mlxsw_sp_port_switchdev_fini(mlxsw_sp_port);
free_percpu(mlxsw_sp_port->pcpu_stats);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index 3b89ed2..6bf6dac 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -43,6 +43,7 @@
#include <linux/if_vlan.h>
#include <linux/list.h>
#include <net/switchdev.h>
+#include <net/devlink.h>
#include "port.h"
#include "core.h"
@@ -162,6 +163,7 @@ struct mlxsw_sp_port {
unsigned long *untagged_vlans;
/* VLAN interfaces */
struct list_head vports_list;
+ struct devlink_port devlink_port;
};
static inline struct mlxsw_sp_port *
diff --git a/drivers/net/ethernet/mellanox/mlxsw/switchx2.c b/drivers/net/ethernet/mellanox/mlxsw/switchx2.c
index d85960c..7a60a26 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/switchx2.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/switchx2.c
@@ -43,6 +43,7 @@
#include <linux/device.h>
#include <linux/skbuff.h>
#include <linux/if_vlan.h>
+#include <net/devlink.h>
#include <net/switchdev.h>
#include <generated/utsrelease.h>
@@ -78,6 +79,7 @@ struct mlxsw_sx_port {
struct mlxsw_sx_port_pcpu_stats __percpu *pcpu_stats;
struct mlxsw_sx *mlxsw_sx;
u8 local_port;
+ struct devlink_port devlink_port;
};
/* tx_hdr_version
@@ -953,7 +955,9 @@ mlxsw_sx_port_mac_learning_mode_set(struct mlxsw_sx_port *mlxsw_sx_port,
static int mlxsw_sx_port_create(struct mlxsw_sx *mlxsw_sx, u8 local_port)
{
+ struct devlink *devlink = priv_to_devlink(mlxsw_sx->core);
struct mlxsw_sx_port *mlxsw_sx_port;
+ struct devlink_port *devlink_port;
struct net_device *dev;
bool usable;
int err;
@@ -1007,6 +1011,14 @@ static int mlxsw_sx_port_create(struct mlxsw_sx *mlxsw_sx, u8 local_port)
goto port_not_usable;
}
+ devlink_port = &mlxsw_sx_port->devlink_port;
+ err = devlink_port_register(devlink, devlink_port, local_port);
+ if (err) {
+ dev_err(mlxsw_sx->bus_info->dev, "Port %d: Failed to register devlink port\n",
+ mlxsw_sx_port->local_port);
+ goto err_devlink_port_register;
+ }
+
err = mlxsw_sx_port_system_port_mapping_set(mlxsw_sx_port);
if (err) {
dev_err(mlxsw_sx->bus_info->dev, "Port %d: Failed to set system port mapping\n",
@@ -1064,6 +1076,8 @@ static int mlxsw_sx_port_create(struct mlxsw_sx *mlxsw_sx, u8 local_port)
goto err_register_netdev;
}
+ devlink_port_type_eth_set(devlink_port, dev);
+
mlxsw_sx->ports[local_port] = mlxsw_sx_port;
return 0;
@@ -1075,6 +1089,8 @@ err_port_mtu_set:
err_port_speed_set:
err_port_swid_set:
err_port_system_port_mapping_set:
+ devlink_port_unregister(&mlxsw_sx_port->devlink_port);
+err_devlink_port_register:
port_not_usable:
err_port_module_check:
err_dev_addr_get:
@@ -1087,11 +1103,15 @@ err_alloc_stats:
static void mlxsw_sx_port_remove(struct mlxsw_sx *mlxsw_sx, u8 local_port)
{
struct mlxsw_sx_port *mlxsw_sx_port = mlxsw_sx->ports[local_port];
+ struct devlink_port *devlink_port;
if (!mlxsw_sx_port)
return;
+ devlink_port = &mlxsw_sx_port->devlink_port;
+ devlink_port_type_clear(devlink_port);
unregister_netdev(mlxsw_sx_port->dev); /* This calls ndo_stop */
mlxsw_sx_port_swid_set(mlxsw_sx_port, MLXSW_PORT_SWID_DISABLED_PORT);
+ devlink_port_unregister(devlink_port);
free_percpu(mlxsw_sx_port->pcpu_stats);
free_netdev(mlxsw_sx_port->dev);
}
--
2.5.0
^ permalink raw reply related [flat|nested] 19+ messages in thread* [patch net-next v2 5/9] mlxsw: core: Add devlink port splitter callbacks
2016-02-23 15:51 [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it Jiri Pirko
` (3 preceding siblings ...)
2016-02-23 15:51 ` [patch net-next v2 4/9] mlxsw: Implement " Jiri Pirko
@ 2016-02-23 15:51 ` Jiri Pirko
2016-02-23 15:51 ` [patch net-next v2 6/9] mlxsw: spectrum: Unmap local port from module during teardown Jiri Pirko
` (4 subsequent siblings)
9 siblings, 0 replies; 19+ messages in thread
From: Jiri Pirko @ 2016-02-23 15:51 UTC (permalink / raw)
To: netdev
Cc: davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
From: Jiri Pirko <jiri@mellanox.com>
Add middle layer in mlxsw core code to forward port split/unsplit calls
into specific ASIC drivers.
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/core.c | 34 +++++++++++++++++++++++++++++-
drivers/net/ethernet/mellanox/mlxsw/core.h | 2 ++
2 files changed, 35 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core.c b/drivers/net/ethernet/mellanox/mlxsw/core.c
index 57d9655..e9d3040 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/core.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/core.c
@@ -785,6 +785,38 @@ static void mlxsw_core_debugfs_fini(struct mlxsw_core *mlxsw_core)
debugfs_remove_recursive(mlxsw_core->dbg_dir);
}
+static int mlxsw_devlink_port_split(struct devlink *devlink,
+ unsigned int port_index,
+ unsigned int count)
+{
+ struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+
+ if (port_index >= MLXSW_PORT_MAX_PORTS)
+ return -EINVAL;
+ if (!mlxsw_core->driver->port_split)
+ return -EOPNOTSUPP;
+ return mlxsw_core->driver->port_split(mlxsw_core->driver_priv,
+ port_index, count);
+}
+
+static int mlxsw_devlink_port_unsplit(struct devlink *devlink,
+ unsigned int port_index)
+{
+ struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+
+ if (port_index >= MLXSW_PORT_MAX_PORTS)
+ return -EINVAL;
+ if (!mlxsw_core->driver->port_unsplit)
+ return -EOPNOTSUPP;
+ return mlxsw_core->driver->port_unsplit(mlxsw_core->driver_priv,
+ port_index);
+}
+
+static const struct devlink_ops mlxsw_devlink_ops = {
+ .port_split = mlxsw_devlink_port_split,
+ .port_unsplit = mlxsw_devlink_port_unsplit,
+};
+
int mlxsw_core_bus_device_register(const struct mlxsw_bus_info *mlxsw_bus_info,
const struct mlxsw_bus *mlxsw_bus,
void *bus_priv)
@@ -800,7 +832,7 @@ int mlxsw_core_bus_device_register(const struct mlxsw_bus_info *mlxsw_bus_info,
if (!mlxsw_driver)
return -EINVAL;
alloc_size = sizeof(*mlxsw_core) + mlxsw_driver->priv_size;
- devlink = devlink_alloc(NULL, alloc_size);
+ devlink = devlink_alloc(&mlxsw_devlink_ops, alloc_size);
if (!devlink) {
err = -ENOMEM;
goto err_devlink_alloc;
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core.h b/drivers/net/ethernet/mellanox/mlxsw/core.h
index a017236..c73d1c0 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/core.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/core.h
@@ -186,6 +186,8 @@ struct mlxsw_driver {
int (*init)(void *driver_priv, struct mlxsw_core *mlxsw_core,
const struct mlxsw_bus_info *mlxsw_bus_info);
void (*fini)(void *driver_priv);
+ int (*port_split)(void *driver_priv, u8 local_port, unsigned int count);
+ int (*port_unsplit)(void *driver_priv, u8 local_port);
void (*txhdr_construct)(struct sk_buff *skb,
const struct mlxsw_tx_info *tx_info);
u8 txhdr_len;
--
2.5.0
^ permalink raw reply related [flat|nested] 19+ messages in thread* [patch net-next v2 6/9] mlxsw: spectrum: Unmap local port from module during teardown
2016-02-23 15:51 [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it Jiri Pirko
` (4 preceding siblings ...)
2016-02-23 15:51 ` [patch net-next v2 5/9] mlxsw: core: Add devlink port splitter callbacks Jiri Pirko
@ 2016-02-23 15:51 ` Jiri Pirko
2016-02-23 15:51 ` [patch net-next v2 7/9] mlxsw: spectrum: Store local port to module mapping during init Jiri Pirko
` (3 subsequent siblings)
9 siblings, 0 replies; 19+ messages in thread
From: Jiri Pirko @ 2016-02-23 15:51 UTC (permalink / raw)
To: netdev
Cc: davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
From: Ido Schimmel <idosch@mellanox.com>
When splitting a port we replace it with 2 or 4 other ports. To be able
to do that we need to remove the original port netdev and unmap it from
its module. However, we first mark it as disabled, as active ports
cannot be unmapped.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 11 +++++++++++
1 file changed, 11 insertions(+)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 20e6783..0dd72f8 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -320,6 +320,15 @@ static int mlxsw_sp_port_module_check(struct mlxsw_sp_port *mlxsw_sp_port,
return 0;
}
+static int mlxsw_sp_port_module_unmap(struct mlxsw_sp *mlxsw_sp, u8 local_port)
+{
+ char pmlp_pl[MLXSW_REG_PMLP_LEN];
+
+ mlxsw_reg_pmlp_pack(pmlp_pl, local_port);
+ mlxsw_reg_pmlp_width_set(pmlp_pl, 0);
+ return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pmlp), pmlp_pl);
+}
+
static int mlxsw_sp_port_open(struct net_device *dev)
{
struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
@@ -1530,6 +1539,8 @@ static void mlxsw_sp_port_remove(struct mlxsw_sp *mlxsw_sp, u8 local_port)
devlink_port_unregister(devlink_port);
mlxsw_sp_port_vports_fini(mlxsw_sp_port);
mlxsw_sp_port_switchdev_fini(mlxsw_sp_port);
+ mlxsw_sp_port_swid_set(mlxsw_sp_port, MLXSW_PORT_SWID_DISABLED_PORT);
+ mlxsw_sp_port_module_unmap(mlxsw_sp, mlxsw_sp_port->local_port);
free_percpu(mlxsw_sp_port->pcpu_stats);
kfree(mlxsw_sp_port->untagged_vlans);
kfree(mlxsw_sp_port->active_vlans);
--
2.5.0
^ permalink raw reply related [flat|nested] 19+ messages in thread* [patch net-next v2 7/9] mlxsw: spectrum: Store local port to module mapping during init
2016-02-23 15:51 [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it Jiri Pirko
` (5 preceding siblings ...)
2016-02-23 15:51 ` [patch net-next v2 6/9] mlxsw: spectrum: Unmap local port from module during teardown Jiri Pirko
@ 2016-02-23 15:51 ` Jiri Pirko
2016-02-23 15:51 ` [patch net-next v2 8/9] mlxsw: spectrum: Mark unused ports using NULL Jiri Pirko
` (2 subsequent siblings)
9 siblings, 0 replies; 19+ messages in thread
From: Jiri Pirko @ 2016-02-23 15:51 UTC (permalink / raw)
To: netdev
Cc: davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
From: Ido Schimmel <idosch@mellanox.com>
The port netdevs are each associated with a different local port number
in the device. These local ports are grouped into groups of 4 (e.g.
(1-4), (5-8)) called clusters. The cluster constitutes the one of two
possible modules they can be mapped to. This mapping is board-specific
and done by the device's firmware during init.
When splitting a port by 4, the device requires us to first unmap all
the ports in the cluster and then map each to a single lane in the module
associated with the port netdev used as the handle for the operation.
This means that two port netdevs will disappear, as only 100Gb/s (4
lanes) ports can be split and we are guaranteed to have two of these
((1, 3), (5, 7) etc.) in a cluster.
When unsplit occurs we need to reinstantiate the two original 100Gb/s
ports and map each to its origianl module. Therefore, during driver init
store the initial local port to module mapping, so it can be used later
during unsplitting.
Note that a by 2 split doesn't require us to store the mapping, as we
only need to reinstantiate one port whose module is known.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 36 +++++++++++---------------
drivers/net/ethernet/mellanox/mlxsw/spectrum.h | 1 +
2 files changed, 16 insertions(+), 21 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 0dd72f8..240881c 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -305,18 +305,19 @@ mlxsw_sp_port_system_port_mapping_set(struct mlxsw_sp_port *mlxsw_sp_port)
return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sspr), sspr_pl);
}
-static int mlxsw_sp_port_module_check(struct mlxsw_sp_port *mlxsw_sp_port,
- bool *p_usable)
+static int mlxsw_sp_port_module_info_get(struct mlxsw_sp *mlxsw_sp,
+ u8 local_port, u8 *p_module,
+ u8 *p_width)
{
- struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
char pmlp_pl[MLXSW_REG_PMLP_LEN];
int err;
- mlxsw_reg_pmlp_pack(pmlp_pl, mlxsw_sp_port->local_port);
+ mlxsw_reg_pmlp_pack(pmlp_pl, local_port);
err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(pmlp), pmlp_pl);
if (err)
return err;
- *p_usable = mlxsw_reg_pmlp_width_get(pmlp_pl) ? true : false;
+ *p_module = mlxsw_reg_pmlp_module_get(pmlp_pl, 0);
+ *p_width = mlxsw_reg_pmlp_width_get(pmlp_pl);
return 0;
}
@@ -1365,7 +1366,6 @@ static int mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port)
struct mlxsw_sp_port *mlxsw_sp_port;
struct devlink_port *devlink_port;
struct net_device *dev;
- bool usable;
size_t bytes;
int err;
@@ -1416,19 +1416,6 @@ static int mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port)
*/
dev->hard_header_len += MLXSW_TXHDR_LEN;
- err = mlxsw_sp_port_module_check(mlxsw_sp_port, &usable);
- if (err) {
- dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to check module\n",
- mlxsw_sp_port->local_port);
- goto err_port_module_check;
- }
-
- if (!usable) {
- dev_dbg(mlxsw_sp->bus_info->dev, "Port %d: Not usable, skipping initialization\n",
- mlxsw_sp_port->local_port);
- goto port_not_usable;
- }
-
devlink_port = &mlxsw_sp_port->devlink_port;
err = devlink_port_register(devlink, devlink_port, local_port);
if (err) {
@@ -1496,8 +1483,6 @@ err_port_swid_set:
err_port_system_port_mapping_set:
devlink_port_unregister(&mlxsw_sp_port->devlink_port);
err_devlink_port_register:
-port_not_usable:
-err_port_module_check:
err_dev_addr_init:
free_percpu(mlxsw_sp_port->pcpu_stats);
err_alloc_stats:
@@ -1559,6 +1544,7 @@ static void mlxsw_sp_ports_remove(struct mlxsw_sp *mlxsw_sp)
static int mlxsw_sp_ports_create(struct mlxsw_sp *mlxsw_sp)
{
size_t alloc_size;
+ u8 module, width;
int i;
int err;
@@ -1568,6 +1554,13 @@ static int mlxsw_sp_ports_create(struct mlxsw_sp *mlxsw_sp)
return -ENOMEM;
for (i = 1; i < MLXSW_PORT_MAX_PORTS; i++) {
+ err = mlxsw_sp_port_module_info_get(mlxsw_sp, i, &module,
+ &width);
+ if (err)
+ goto err_port_module_info_get;
+ if (!width)
+ continue;
+ mlxsw_sp->port_to_module[i] = module;
err = mlxsw_sp_port_create(mlxsw_sp, i);
if (err)
goto err_port_create;
@@ -1575,6 +1568,7 @@ static int mlxsw_sp_ports_create(struct mlxsw_sp *mlxsw_sp)
return 0;
err_port_create:
+err_port_module_info_get:
for (i--; i >= 1; i--)
mlxsw_sp_port_remove(mlxsw_sp, i);
kfree(mlxsw_sp->ports);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index 6bf6dac..a7d86ac 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -123,6 +123,7 @@ struct mlxsw_sp {
u32 ageing_time;
struct mlxsw_sp_upper master_bridge;
struct mlxsw_sp_upper lags[MLXSW_SP_LAG_MAX];
+ u8 port_to_module[MLXSW_PORT_MAX_PORTS];
};
static inline struct mlxsw_sp_upper *
--
2.5.0
^ permalink raw reply related [flat|nested] 19+ messages in thread* [patch net-next v2 8/9] mlxsw: spectrum: Mark unused ports using NULL
2016-02-23 15:51 [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it Jiri Pirko
` (6 preceding siblings ...)
2016-02-23 15:51 ` [patch net-next v2 7/9] mlxsw: spectrum: Store local port to module mapping during init Jiri Pirko
@ 2016-02-23 15:51 ` Jiri Pirko
2016-02-23 15:51 ` [patch net-next v2 9/9] mlxsw: spectrum: Introduce port splitting Jiri Pirko
2016-02-25 20:12 ` [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it David Miller
9 siblings, 0 replies; 19+ messages in thread
From: Jiri Pirko @ 2016-02-23 15:51 UTC (permalink / raw)
To: netdev
Cc: davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
From: Ido Schimmel <idosch@mellanox.com>
When splitting and unsplitting we'll destroy usable ports on the fly, so
mark them using a NULL pointer to indicate that their local port number
is free and can be re-used.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 240881c..926019e 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -1518,6 +1518,7 @@ static void mlxsw_sp_port_remove(struct mlxsw_sp *mlxsw_sp, u8 local_port)
if (!mlxsw_sp_port)
return;
+ mlxsw_sp->ports[local_port] = NULL;
devlink_port = &mlxsw_sp_port->devlink_port;
devlink_port_type_clear(devlink_port);
unregister_netdev(mlxsw_sp_port->dev); /* This calls ndo_stop */
--
2.5.0
^ permalink raw reply related [flat|nested] 19+ messages in thread* [patch net-next v2 9/9] mlxsw: spectrum: Introduce port splitting
2016-02-23 15:51 [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it Jiri Pirko
` (7 preceding siblings ...)
2016-02-23 15:51 ` [patch net-next v2 8/9] mlxsw: spectrum: Mark unused ports using NULL Jiri Pirko
@ 2016-02-23 15:51 ` Jiri Pirko
2016-02-25 20:12 ` [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it David Miller
9 siblings, 0 replies; 19+ messages in thread
From: Jiri Pirko @ 2016-02-23 15:51 UTC (permalink / raw)
To: netdev
Cc: davem, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
From: Ido Schimmel <idosch@mellanox.com>
Allow a user to split or unsplit a port using the newly introduced
devlink ops.
Once split, the original netdev is destroyed and 2 or 4 others are
created, according to user configuration. The new ports are like any
other port, with the sole difference of supporting a lower maximum
speed. When unsplit, the reverse process takes place.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
drivers/net/ethernet/mellanox/mlxsw/port.h | 2 +
drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 178 ++++++++++++++++++++++++-
drivers/net/ethernet/mellanox/mlxsw/spectrum.h | 5 +-
3 files changed, 182 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/port.h b/drivers/net/ethernet/mellanox/mlxsw/port.h
index ae65b99..f33b997 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/port.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/port.h
@@ -59,6 +59,8 @@
#define MLXSW_PORT_DONT_CARE (MLXSW_PORT_MAX_PORTS)
+#define MLXSW_PORT_MODULE_MAX_WIDTH 4
+
enum mlxsw_port_admin_status {
MLXSW_PORT_ADMIN_STATUS_UP = 1,
MLXSW_PORT_ADMIN_STATUS_DOWN = 2,
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 926019e..1dcd3bc 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -321,6 +321,22 @@ static int mlxsw_sp_port_module_info_get(struct mlxsw_sp *mlxsw_sp,
return 0;
}
+static int mlxsw_sp_port_module_map(struct mlxsw_sp *mlxsw_sp, u8 local_port,
+ u8 module, u8 width, u8 lane)
+{
+ char pmlp_pl[MLXSW_REG_PMLP_LEN];
+ int i;
+
+ mlxsw_reg_pmlp_pack(pmlp_pl, local_port);
+ mlxsw_reg_pmlp_width_set(pmlp_pl, width);
+ for (i = 0; i < width; i++) {
+ mlxsw_reg_pmlp_module_set(pmlp_pl, i, module);
+ mlxsw_reg_pmlp_tx_lane_set(pmlp_pl, i, lane + i); /* Rx & Tx */
+ }
+
+ return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pmlp), pmlp_pl);
+}
+
static int mlxsw_sp_port_module_unmap(struct mlxsw_sp *mlxsw_sp, u8 local_port)
{
char pmlp_pl[MLXSW_REG_PMLP_LEN];
@@ -1360,7 +1376,8 @@ static const struct ethtool_ops mlxsw_sp_port_ethtool_ops = {
.set_settings = mlxsw_sp_port_set_settings,
};
-static int mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port)
+static int __mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port,
+ bool split, u8 module)
{
struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
struct mlxsw_sp_port *mlxsw_sp_port;
@@ -1376,6 +1393,7 @@ static int mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port)
mlxsw_sp_port->dev = dev;
mlxsw_sp_port->mlxsw_sp = mlxsw_sp;
mlxsw_sp_port->local_port = local_port;
+ mlxsw_sp_port->split = split;
bytes = DIV_ROUND_UP(VLAN_N_VID, BITS_PER_BYTE);
mlxsw_sp_port->active_vlans = kzalloc(bytes, GFP_KERNEL);
if (!mlxsw_sp_port->active_vlans) {
@@ -1417,6 +1435,8 @@ static int mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port)
dev->hard_header_len += MLXSW_TXHDR_LEN;
devlink_port = &mlxsw_sp_port->devlink_port;
+ if (mlxsw_sp_port->split)
+ devlink_port_split_set(devlink_port, module);
err = devlink_port_register(devlink, devlink_port, local_port);
if (err) {
dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to register devlink port\n",
@@ -1494,6 +1514,27 @@ err_port_active_vlans_alloc:
return err;
}
+static int mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port,
+ bool split, u8 module, u8 width, u8 lane)
+{
+ int err;
+
+ err = mlxsw_sp_port_module_map(mlxsw_sp, local_port, module, width,
+ lane);
+ if (err)
+ return err;
+
+ err = __mlxsw_sp_port_create(mlxsw_sp, local_port, split, module);
+ if (err)
+ goto err_port_create;
+
+ return 0;
+
+err_port_create:
+ mlxsw_sp_port_module_unmap(mlxsw_sp, local_port);
+ return err;
+}
+
static void mlxsw_sp_port_vports_fini(struct mlxsw_sp_port *mlxsw_sp_port)
{
struct net_device *dev = mlxsw_sp_port->dev;
@@ -1562,7 +1603,7 @@ static int mlxsw_sp_ports_create(struct mlxsw_sp *mlxsw_sp)
if (!width)
continue;
mlxsw_sp->port_to_module[i] = module;
- err = mlxsw_sp_port_create(mlxsw_sp, i);
+ err = __mlxsw_sp_port_create(mlxsw_sp, i, false, module);
if (err)
goto err_port_create;
}
@@ -1576,6 +1617,137 @@ err_port_module_info_get:
return err;
}
+static u8 mlxsw_sp_cluster_base_port_get(u8 local_port)
+{
+ u8 offset = (local_port - 1) % MLXSW_SP_PORTS_PER_CLUSTER_MAX;
+
+ return local_port - offset;
+}
+
+static int mlxsw_sp_port_split(void *priv, u8 local_port, unsigned int count)
+{
+ struct mlxsw_sp *mlxsw_sp = priv;
+ struct mlxsw_sp_port *mlxsw_sp_port;
+ u8 width = MLXSW_PORT_MODULE_MAX_WIDTH / count;
+ u8 module, cur_width, base_port;
+ int i;
+ int err;
+
+ mlxsw_sp_port = mlxsw_sp->ports[local_port];
+ if (!mlxsw_sp_port) {
+ dev_err(mlxsw_sp->bus_info->dev, "Port number \"%d\" does not exist\n",
+ local_port);
+ return -EINVAL;
+ }
+
+ if (count != 2 && count != 4) {
+ netdev_err(mlxsw_sp_port->dev, "Port can only be split into 2 or 4 ports\n");
+ return -EINVAL;
+ }
+
+ err = mlxsw_sp_port_module_info_get(mlxsw_sp, local_port, &module,
+ &cur_width);
+ if (err) {
+ netdev_err(mlxsw_sp_port->dev, "Failed to get port's width\n");
+ return err;
+ }
+
+ if (cur_width != MLXSW_PORT_MODULE_MAX_WIDTH) {
+ netdev_err(mlxsw_sp_port->dev, "Port cannot be split further\n");
+ return -EINVAL;
+ }
+
+ /* Make sure we have enough slave (even) ports for the split. */
+ if (count == 2) {
+ base_port = local_port;
+ if (mlxsw_sp->ports[base_port + 1]) {
+ netdev_err(mlxsw_sp_port->dev, "Invalid split configuration\n");
+ return -EINVAL;
+ }
+ } else {
+ base_port = mlxsw_sp_cluster_base_port_get(local_port);
+ if (mlxsw_sp->ports[base_port + 1] ||
+ mlxsw_sp->ports[base_port + 3]) {
+ netdev_err(mlxsw_sp_port->dev, "Invalid split configuration\n");
+ return -EINVAL;
+ }
+ }
+
+ for (i = 0; i < count; i++)
+ mlxsw_sp_port_remove(mlxsw_sp, base_port + i);
+
+ for (i = 0; i < count; i++) {
+ err = mlxsw_sp_port_create(mlxsw_sp, base_port + i, true,
+ module, width, i * width);
+ if (err) {
+ dev_err(mlxsw_sp->bus_info->dev, "Failed to create split port\n");
+ goto err_port_create;
+ }
+ }
+
+ return 0;
+
+err_port_create:
+ for (i--; i >= 0; i--)
+ mlxsw_sp_port_remove(mlxsw_sp, base_port + i);
+ for (i = 0; i < count / 2; i++) {
+ module = mlxsw_sp->port_to_module[base_port + i * 2];
+ mlxsw_sp_port_create(mlxsw_sp, base_port + i * 2, false,
+ module, MLXSW_PORT_MODULE_MAX_WIDTH, 0);
+ }
+ return err;
+}
+
+static int mlxsw_sp_port_unsplit(void *priv, u8 local_port)
+{
+ struct mlxsw_sp *mlxsw_sp = priv;
+ struct mlxsw_sp_port *mlxsw_sp_port;
+ u8 module, cur_width, base_port;
+ unsigned int count;
+ int i;
+ int err;
+
+ mlxsw_sp_port = mlxsw_sp->ports[local_port];
+ if (!mlxsw_sp_port) {
+ dev_err(mlxsw_sp->bus_info->dev, "Port number \"%d\" does not exist\n",
+ local_port);
+ return -EINVAL;
+ }
+
+ if (!mlxsw_sp_port->split) {
+ netdev_err(mlxsw_sp_port->dev, "Port wasn't split\n");
+ return -EINVAL;
+ }
+
+ err = mlxsw_sp_port_module_info_get(mlxsw_sp, local_port, &module,
+ &cur_width);
+ if (err) {
+ netdev_err(mlxsw_sp_port->dev, "Failed to get port's width\n");
+ return err;
+ }
+ count = cur_width == 1 ? 4 : 2;
+
+ base_port = mlxsw_sp_cluster_base_port_get(local_port);
+
+ /* Determine which ports to remove. */
+ if (count == 2 && local_port >= base_port + 2)
+ base_port = base_port + 2;
+
+ for (i = 0; i < count; i++)
+ mlxsw_sp_port_remove(mlxsw_sp, base_port + i);
+
+ for (i = 0; i < count / 2; i++) {
+ module = mlxsw_sp->port_to_module[base_port + i * 2];
+ err = mlxsw_sp_port_create(mlxsw_sp, base_port + i * 2, false,
+ module, MLXSW_PORT_MODULE_MAX_WIDTH,
+ 0);
+ if (err)
+ dev_err(mlxsw_sp->bus_info->dev, "Failed to reinstantiate port\n");
+ }
+
+ return 0;
+}
+
static void mlxsw_sp_pude_event_func(const struct mlxsw_reg_info *reg,
char *pude_pl, void *priv)
{
@@ -1999,6 +2171,8 @@ static struct mlxsw_driver mlxsw_sp_driver = {
.priv_size = sizeof(struct mlxsw_sp),
.init = mlxsw_sp_init,
.fini = mlxsw_sp_fini,
+ .port_split = mlxsw_sp_port_split,
+ .port_unsplit = mlxsw_sp_port_unsplit,
.txhdr_construct = mlxsw_sp_txhdr_construct,
.txhdr_len = MLXSW_TXHDR_LEN,
.profile = &mlxsw_sp_config_profile,
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index a7d86ac..02cb275 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -58,6 +58,8 @@
#define MLXSW_SP_MID_MAX 7000
+#define MLXSW_SP_PORTS_PER_CLUSTER_MAX 4
+
struct mlxsw_sp_port;
struct mlxsw_sp_upper {
@@ -151,7 +153,8 @@ struct mlxsw_sp_port {
learning_sync:1,
uc_flood:1,
bridged:1,
- lagged:1;
+ lagged:1,
+ split:1;
u16 pvid;
u16 lag_id;
struct {
--
2.5.0
^ permalink raw reply related [flat|nested] 19+ messages in thread* Re: [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it
2016-02-23 15:51 [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it Jiri Pirko
` (8 preceding siblings ...)
2016-02-23 15:51 ` [patch net-next v2 9/9] mlxsw: spectrum: Introduce port splitting Jiri Pirko
@ 2016-02-25 20:12 ` David Miller
2016-02-25 20:44 ` Hannes Frederic Sowa
2016-02-25 21:50 ` Andy Gospodarek
9 siblings, 2 replies; 19+ messages in thread
From: David Miller @ 2016-02-25 20:12 UTC (permalink / raw)
To: jiri
Cc: netdev, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes, gospo
From: Jiri Pirko <jiri@resnulli.us>
Date: Tue, 23 Feb 2016 16:51:25 +0100
> There a is need for some userspace API that would allow to expose things
> that are not directly related to any device class like net_device of
> ib_device, but rather chip-wide/switch-ASIC-wide stuff.
>
> Use cases:
> 1) get/set of port type (Ethernet/InfiniBand)
> 2) setting up port splitters - split port into multiple ones and squash again,
> enables usage of splitter cable
> 3) setting up shared buffers - shared among multiple ports within
> one chip (work in progress)
> 4) configuration of switch wide properties - resources division etc - This will
> allow to pass configuration that is unacceptable to be passed as
> a module option.
>
> First patch of this set introduces a new generic Netlink based interface,
> called "devlink". It is similar to nl80211 model and it is heavily
> influenced by it, including the API definition. The devlink introduction patch
> implements use cases 1) and 2). Other 2 are in development atm and will
> be addressed by follow-ups.
>
> It is very convenient for drivers to use devlink, as you can see in other
> patches in this set.
>
> Counterpart for devlink is userspace tool for now called "dl". Command line
> interface and outputs are derived from "ip" tool so it should be easy
> for users to get used to it.
I am very close to applying this series as-is.
The clincher for me is that there is precendence in the nl80211 stuff,
so obviously whatever userland infrastructure sits on top of that has
found a way to deal whatever perceived shortcomings devlink has.
If all that people can come up with is "device names! omg UDEV!" and
"multiple tools are hard to use!" That's awesome, because it means
nobody has any real substantial objection to this facility. :-)
^ permalink raw reply [flat|nested] 19+ messages in thread* Re: [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it
2016-02-25 20:12 ` [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it David Miller
@ 2016-02-25 20:44 ` Hannes Frederic Sowa
2016-02-25 21:12 ` Jiri Pirko
2016-02-25 21:50 ` Andy Gospodarek
1 sibling, 1 reply; 19+ messages in thread
From: Hannes Frederic Sowa @ 2016-02-25 20:44 UTC (permalink / raw)
To: David Miller, jiri
Cc: netdev, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
gospo
On 25.02.2016 21:12, David Miller wrote:
> From: Jiri Pirko <jiri@resnulli.us>
> Date: Tue, 23 Feb 2016 16:51:25 +0100
>
>> There a is need for some userspace API that would allow to expose things
>> that are not directly related to any device class like net_device of
>> ib_device, but rather chip-wide/switch-ASIC-wide stuff.
>>
>> Use cases:
>> 1) get/set of port type (Ethernet/InfiniBand)
>> 2) setting up port splitters - split port into multiple ones and squash again,
>> enables usage of splitter cable
>> 3) setting up shared buffers - shared among multiple ports within
>> one chip (work in progress)
>> 4) configuration of switch wide properties - resources division etc - This will
>> allow to pass configuration that is unacceptable to be passed as
>> a module option.
>>
>> First patch of this set introduces a new generic Netlink based interface,
>> called "devlink". It is similar to nl80211 model and it is heavily
>> influenced by it, including the API definition. The devlink introduction patch
>> implements use cases 1) and 2). Other 2 are in development atm and will
>> be addressed by follow-ups.
>>
>> It is very convenient for drivers to use devlink, as you can see in other
>> patches in this set.
>>
>> Counterpart for devlink is userspace tool for now called "dl". Command line
>> interface and outputs are derived from "ip" tool so it should be easy
>> for users to get used to it.
>
> I am very close to applying this series as-is.
>
> The clincher for me is that there is precendence in the nl80211 stuff,
> so obviously whatever userland infrastructure sits on top of that has
> found a way to deal whatever perceived shortcomings devlink has.
Actually nl80211 phy interfaces aren't really managed by wpa_supplicant
nor NetworkManager, but they use net_device names to discover those
later on. In devlink we don't necessarily have netdev names, thus my
only objection to this series is to switch to stable topology identifiers.
> If all that people can come up with is "device names! omg UDEV!" and
> "multiple tools are hard to use!" That's awesome, because it means
> nobody has any real substantial objection to this facility. :-)
I do think this is important if a lot of devlinks are visible and they
don't yet provide a net_device. One has to use e.g. pci identifiers or
depend on module loading order. Especially I have no idea how this
should work with devices where the module is not yet loaded (and most
device drivers don't split between pci layer and net_device). This is
even more complicated then the udev stuff we have nowadays.
Bye,
Hannes
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it
2016-02-25 20:44 ` Hannes Frederic Sowa
@ 2016-02-25 21:12 ` Jiri Pirko
2016-02-25 21:53 ` Hannes Frederic Sowa
0 siblings, 1 reply; 19+ messages in thread
From: Jiri Pirko @ 2016-02-25 21:12 UTC (permalink / raw)
To: Hannes Frederic Sowa
Cc: David Miller, netdev, idosch, eladr, yotamg, ogerlitz, yishaih,
dledford, sean.hefty, hal.rosenstock, eugenia, roopa, nikolay,
hadarh, jhs, john.fastabend, jeffrey.t.kirsher, brouer, ivecera,
rami.rosen, gospo
Thu, Feb 25, 2016 at 09:44:43PM CET, hannes@stressinduktion.org wrote:
>On 25.02.2016 21:12, David Miller wrote:
>>From: Jiri Pirko <jiri@resnulli.us>
>>Date: Tue, 23 Feb 2016 16:51:25 +0100
>>
>>>There a is need for some userspace API that would allow to expose things
>>>that are not directly related to any device class like net_device of
>>>ib_device, but rather chip-wide/switch-ASIC-wide stuff.
>>>
>>>Use cases:
>>>1) get/set of port type (Ethernet/InfiniBand)
>>>2) setting up port splitters - split port into multiple ones and squash again,
>>> enables usage of splitter cable
>>>3) setting up shared buffers - shared among multiple ports within
>>> one chip (work in progress)
>>>4) configuration of switch wide properties - resources division etc - This will
>>> allow to pass configuration that is unacceptable to be passed as
>>> a module option.
>>>
>>>First patch of this set introduces a new generic Netlink based interface,
>>>called "devlink". It is similar to nl80211 model and it is heavily
>>>influenced by it, including the API definition. The devlink introduction patch
>>>implements use cases 1) and 2). Other 2 are in development atm and will
>>>be addressed by follow-ups.
>>>
>>>It is very convenient for drivers to use devlink, as you can see in other
>>>patches in this set.
>>>
>>>Counterpart for devlink is userspace tool for now called "dl". Command line
>>>interface and outputs are derived from "ip" tool so it should be easy
>>>for users to get used to it.
>>
>>I am very close to applying this series as-is.
>>
>>The clincher for me is that there is precendence in the nl80211 stuff,
>>so obviously whatever userland infrastructure sits on top of that has
>>found a way to deal whatever perceived shortcomings devlink has.
>
>Actually nl80211 phy interfaces aren't really managed by wpa_supplicant nor
>NetworkManager, but they use net_device names to discover those later on. In
>devlink we don't necessarily have netdev names, thus my only objection to
>this series is to switch to stable topology identifiers.
Hannes, as I mentioned earlier in one of my replies, you can choose to
use devlink name or pci (or other) address for your commands. So you
have your stable names even before udev takes care of renaming. It's up
to you as a user what handle to use.
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it
2016-02-25 21:12 ` Jiri Pirko
@ 2016-02-25 21:53 ` Hannes Frederic Sowa
2016-02-25 22:58 ` Jiri Pirko
0 siblings, 1 reply; 19+ messages in thread
From: Hannes Frederic Sowa @ 2016-02-25 21:53 UTC (permalink / raw)
To: Jiri Pirko
Cc: David Miller, netdev, idosch, eladr, yotamg, ogerlitz, yishaih,
dledford, sean.hefty, hal.rosenstock, eugenia, roopa, nikolay,
hadarh, jhs, john.fastabend, jeffrey.t.kirsher, brouer, ivecera,
rami.rosen, gospo
On 25.02.2016 22:12, Jiri Pirko wrote:
> Thu, Feb 25, 2016 at 09:44:43PM CET, hannes@stressinduktion.org wrote:
>> On 25.02.2016 21:12, David Miller wrote:
>>> From: Jiri Pirko <jiri@resnulli.us>
>>> Date: Tue, 23 Feb 2016 16:51:25 +0100
>>>
>>>> There a is need for some userspace API that would allow to expose things
>>>> that are not directly related to any device class like net_device of
>>>> ib_device, but rather chip-wide/switch-ASIC-wide stuff.
>>>>
>>>> Use cases:
>>>> 1) get/set of port type (Ethernet/InfiniBand)
>>>> 2) setting up port splitters - split port into multiple ones and squash again,
>>>> enables usage of splitter cable
>>>> 3) setting up shared buffers - shared among multiple ports within
>>>> one chip (work in progress)
>>>> 4) configuration of switch wide properties - resources division etc - This will
>>>> allow to pass configuration that is unacceptable to be passed as
>>>> a module option.
>>>>
>>>> First patch of this set introduces a new generic Netlink based interface,
>>>> called "devlink". It is similar to nl80211 model and it is heavily
>>>> influenced by it, including the API definition. The devlink introduction patch
>>>> implements use cases 1) and 2). Other 2 are in development atm and will
>>>> be addressed by follow-ups.
>>>>
>>>> It is very convenient for drivers to use devlink, as you can see in other
>>>> patches in this set.
>>>>
>>>> Counterpart for devlink is userspace tool for now called "dl". Command line
>>>> interface and outputs are derived from "ip" tool so it should be easy
>>>> for users to get used to it.
>>>
>>> I am very close to applying this series as-is.
>>>
>>> The clincher for me is that there is precendence in the nl80211 stuff,
>>> so obviously whatever userland infrastructure sits on top of that has
>>> found a way to deal whatever perceived shortcomings devlink has.
>>
>> Actually nl80211 phy interfaces aren't really managed by wpa_supplicant nor
>> NetworkManager, but they use net_device names to discover those later on. In
>> devlink we don't necessarily have netdev names, thus my only objection to
>> this series is to switch to stable topology identifiers.
>
> Hannes, as I mentioned earlier in one of my replies, you can choose to
> use devlink name or pci (or other) address for your commands. So you
> have your stable names even before udev takes care of renaming. It's up
> to you as a user what handle to use.
I understood that part from the beginning. In my opinion we should
simply keep APIs simple and clean, reduced to the minimum and let user
space build upon this.
If an entity in the kernel doesn't need a name and/or a naming database,
I would not bother implementing it. Not everyone is a kernel developer
who uses Linux and can infer the problems from dl --help. I bet most
users will use names at some point in time and they simply can't match
even with the udev quirks in place when setting up netbooting or
configuration must happen from the currently provided initramfs. This is
my point.
If names are fragile to use simply don't provide it if not necessary.
devlink is an interface of its own and the names won't be referenced by
other subsystems like iptables -i <name> or iproute. So why bother with
names in the kernel?
On the rest with netlink etc. I agree already. I just want to eliminate
possible mistakes by users.
Hope that makes my point more clear,
Hannes
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it
2016-02-25 21:53 ` Hannes Frederic Sowa
@ 2016-02-25 22:58 ` Jiri Pirko
0 siblings, 0 replies; 19+ messages in thread
From: Jiri Pirko @ 2016-02-25 22:58 UTC (permalink / raw)
To: Hannes Frederic Sowa
Cc: David Miller, netdev, idosch, eladr, yotamg, ogerlitz, yishaih,
dledford, sean.hefty, hal.rosenstock, eugenia, roopa, nikolay,
hadarh, jhs, john.fastabend, jeffrey.t.kirsher, brouer, ivecera,
rami.rosen, gospo
Thu, Feb 25, 2016 at 10:53:30PM CET, hannes@stressinduktion.org wrote:
>On 25.02.2016 22:12, Jiri Pirko wrote:
>>Thu, Feb 25, 2016 at 09:44:43PM CET, hannes@stressinduktion.org wrote:
>>>On 25.02.2016 21:12, David Miller wrote:
>>>>From: Jiri Pirko <jiri@resnulli.us>
>>>>Date: Tue, 23 Feb 2016 16:51:25 +0100
>>>>
>>>>>There a is need for some userspace API that would allow to expose things
>>>>>that are not directly related to any device class like net_device of
>>>>>ib_device, but rather chip-wide/switch-ASIC-wide stuff.
>>>>>
>>>>>Use cases:
>>>>>1) get/set of port type (Ethernet/InfiniBand)
>>>>>2) setting up port splitters - split port into multiple ones and squash again,
>>>>> enables usage of splitter cable
>>>>>3) setting up shared buffers - shared among multiple ports within
>>>>> one chip (work in progress)
>>>>>4) configuration of switch wide properties - resources division etc - This will
>>>>> allow to pass configuration that is unacceptable to be passed as
>>>>> a module option.
>>>>>
>>>>>First patch of this set introduces a new generic Netlink based interface,
>>>>>called "devlink". It is similar to nl80211 model and it is heavily
>>>>>influenced by it, including the API definition. The devlink introduction patch
>>>>>implements use cases 1) and 2). Other 2 are in development atm and will
>>>>>be addressed by follow-ups.
>>>>>
>>>>>It is very convenient for drivers to use devlink, as you can see in other
>>>>>patches in this set.
>>>>>
>>>>>Counterpart for devlink is userspace tool for now called "dl". Command line
>>>>>interface and outputs are derived from "ip" tool so it should be easy
>>>>>for users to get used to it.
>>>>
>>>>I am very close to applying this series as-is.
>>>>
>>>>The clincher for me is that there is precendence in the nl80211 stuff,
>>>>so obviously whatever userland infrastructure sits on top of that has
>>>>found a way to deal whatever perceived shortcomings devlink has.
>>>
>>>Actually nl80211 phy interfaces aren't really managed by wpa_supplicant nor
>>>NetworkManager, but they use net_device names to discover those later on. In
>>>devlink we don't necessarily have netdev names, thus my only objection to
>>>this series is to switch to stable topology identifiers.
>>
>>Hannes, as I mentioned earlier in one of my replies, you can choose to
>>use devlink name or pci (or other) address for your commands. So you
>>have your stable names even before udev takes care of renaming. It's up
>>to you as a user what handle to use.
>
>I understood that part from the beginning. In my opinion we should simply
>keep APIs simple and clean, reduced to the minimum and let user space build
>upon this.
>
>If an entity in the kernel doesn't need a name and/or a naming database, I
>would not bother implementing it. Not everyone is a kernel developer who uses
>Linux and can infer the problems from dl --help. I bet most users will use
>names at some point in time and they simply can't match even with the udev
>quirks in place when setting up netbooting or configuration must happen from
>the currently provided initramfs. This is my point.
>
>If names are fragile to use simply don't provide it if not necessary.
>
>devlink is an interface of its own and the names won't be referenced by other
>subsystems like iptables -i <name> or iproute. So why bother with names in
>the kernel?
>
>On the rest with netlink etc. I agree already. I just want to eliminate
>possible mistakes by users.
>
>Hope that makes my point more clear,
Okay. You convinced me. Will send v3 removing in-kernel naming and
indexing of devlinks and will only use bus type and bus name as a handle.
Thanks.
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it
2016-02-25 20:12 ` [patch net-next v2 0/9] Introduce devlink interface and first drivers to use it David Miller
2016-02-25 20:44 ` Hannes Frederic Sowa
@ 2016-02-25 21:50 ` Andy Gospodarek
1 sibling, 0 replies; 19+ messages in thread
From: Andy Gospodarek @ 2016-02-25 21:50 UTC (permalink / raw)
To: David Miller
Cc: jiri, netdev, idosch, eladr, yotamg, ogerlitz, yishaih, dledford,
sean.hefty, hal.rosenstock, eugenia, roopa, nikolay, hadarh, jhs,
john.fastabend, jeffrey.t.kirsher, brouer, ivecera, rami.rosen,
hannes
On Thu, Feb 25, 2016 at 03:12:47PM -0500, David Miller wrote:
> From: Jiri Pirko <jiri@resnulli.us>
> Date: Tue, 23 Feb 2016 16:51:25 +0100
>
> > There a is need for some userspace API that would allow to expose things
> > that are not directly related to any device class like net_device of
> > ib_device, but rather chip-wide/switch-ASIC-wide stuff.
> >
> > Use cases:
> > 1) get/set of port type (Ethernet/InfiniBand)
> > 2) setting up port splitters - split port into multiple ones and squash again,
> > enables usage of splitter cable
> > 3) setting up shared buffers - shared among multiple ports within
> > one chip (work in progress)
> > 4) configuration of switch wide properties - resources division etc - This will
> > allow to pass configuration that is unacceptable to be passed as
> > a module option.
> >
> > First patch of this set introduces a new generic Netlink based interface,
> > called "devlink". It is similar to nl80211 model and it is heavily
> > influenced by it, including the API definition. The devlink introduction patch
> > implements use cases 1) and 2). Other 2 are in development atm and will
> > be addressed by follow-ups.
> >
> > It is very convenient for drivers to use devlink, as you can see in other
> > patches in this set.
> >
> > Counterpart for devlink is userspace tool for now called "dl". Command line
> > interface and outputs are derived from "ip" tool so it should be easy
> > for users to get used to it.
>
> I am very close to applying this series as-is.
>
> The clincher for me is that there is precendence in the nl80211 stuff,
> so obviously whatever userland infrastructure sits on top of that has
> found a way to deal whatever perceived shortcomings devlink has.
>
> If all that people can come up with is "device names! omg UDEV!" and
> "multiple tools are hard to use!" That's awesome, because it means
> nobody has any real substantial objection to this facility. :-)
Generally speaking I had no issue with the patch after being *ahem*
corrected to realize that the set operations only applied to the devlink
devices not specifically to netdevs.
I find this interface useful for some other items we have discussed in
the past (resource limits for routes, etc), so I'm happy to see some
interface for general chip maintenance included.
^ permalink raw reply [flat|nested] 19+ messages in thread