public inbox for linux-rdma@vger.kernel.org
 help / color / mirror / Atom feed
From: Jiri Pirko <jiri@resnulli.us>
To: netdev@vger.kernel.org
Cc: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com,
	edumazet@google.com, tariqt@nvidia.com, moshe@nvidia.com,
	saeedm@nvidia.com, linux-rdma@vger.kernel.org
Subject: [patch net-next v2 05/13] net: devlink: track netdev with devlink_port assigned
Date: Mon,  3 Oct 2022 12:51:56 +0200	[thread overview]
Message-ID: <20221003105204.3315337-6-jiri@resnulli.us> (raw)
In-Reply-To: <20221003105204.3315337-1-jiri@resnulli.us>

From: Jiri Pirko <jiri@nvidia.com>

Currently, ethernet drivers are using devlink_port_type_eth_set() and
devlink_port_type_clear() to set devlink port type and link to related
netdev.

Instead of calling them directly, let the driver use
SET_NETDEV_DEVLINK_PORT macro to assign devlink_port pointer and let
devlink to track it. Note the devlink port pointer is static during
the time netdevice is registered.

In devlink code, use per-namespace netdev notifier to track
the netdevices with devlink_port assigned and change the internal
devlink_port type and related type pointer accordingly.

Signed-off-by: Jiri Pirko <jiri@nvidia.com>
---
v1->v2:
- added kdoc for devlink_port struct field
---
 include/linux/netdevice.h | 19 ++++++++++
 net/core/dev.c            | 11 +++---
 net/core/devlink.c        | 75 ++++++++++++++++++++++++++++++++++++---
 3 files changed, 97 insertions(+), 8 deletions(-)

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index eddf8ee270e7..bbb93be49063 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1999,6 +1999,11 @@ enum netdev_ml_priv_type {
  *					registered
  *	@offload_xstats_l3:	L3 HW stats for this netdevice.
  *
+ *	@devlink_port:	Pointer to related devlink port structure.
+ *			Assigned by a driver before netdev registration using
+ *			SET_NETDEV_DEVLINK_PORT macro. This pointer is static
+ *			during the time netdevice is registered.
+ *
  *	FIXME: cleanup struct net_device such that network protocol info
  *	moves out.
  */
@@ -2349,9 +2354,22 @@ struct net_device {
 	netdevice_tracker	watchdog_dev_tracker;
 	netdevice_tracker	dev_registered_tracker;
 	struct rtnl_hw_stats64	*offload_xstats_l3;
+
+	struct devlink_port	*devlink_port;
 };
 #define to_net_dev(d) container_of(d, struct net_device, dev)
 
+/*
+ * Driver should use this to assign devlink port instance to a netdevice
+ * before it registers the netdevice. Therefore devlink_port is static
+ * during the netdev lifetime after it is registered.
+ */
+#define SET_NETDEV_DEVLINK_PORT(dev, _devlink_port)		\
+({								\
+	WARN_ON(dev->reg_state != NETREG_UNINITIALIZED);	\
+	((dev)->devlink_port = (_devlink_port));		\
+})
+
 static inline bool netif_elide_gro(const struct net_device *dev)
 {
 	if (!(dev->features & NETIF_F_GRO) || dev->xdp_prog)
@@ -2785,6 +2803,7 @@ enum netdev_cmd {
 	NETDEV_PRE_TYPE_CHANGE,
 	NETDEV_POST_TYPE_CHANGE,
 	NETDEV_POST_INIT,
+	NETDEV_PRE_UNINIT,
 	NETDEV_RELEASE,
 	NETDEV_NOTIFY_PEERS,
 	NETDEV_JOIN,
diff --git a/net/core/dev.c b/net/core/dev.c
index fa53830d0683..1b45aa5c976e 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1621,10 +1621,10 @@ const char *netdev_cmd_to_name(enum netdev_cmd cmd)
 	N(UP) N(DOWN) N(REBOOT) N(CHANGE) N(REGISTER) N(UNREGISTER)
 	N(CHANGEMTU) N(CHANGEADDR) N(GOING_DOWN) N(CHANGENAME) N(FEAT_CHANGE)
 	N(BONDING_FAILOVER) N(PRE_UP) N(PRE_TYPE_CHANGE) N(POST_TYPE_CHANGE)
-	N(POST_INIT) N(RELEASE) N(NOTIFY_PEERS) N(JOIN) N(CHANGEUPPER)
-	N(RESEND_IGMP) N(PRECHANGEMTU) N(CHANGEINFODATA) N(BONDING_INFO)
-	N(PRECHANGEUPPER) N(CHANGELOWERSTATE) N(UDP_TUNNEL_PUSH_INFO)
-	N(UDP_TUNNEL_DROP_INFO) N(CHANGE_TX_QUEUE_LEN)
+	N(POST_INIT) N(PRE_UNINIT) N(RELEASE) N(NOTIFY_PEERS) N(JOIN)
+	N(CHANGEUPPER) N(RESEND_IGMP) N(PRECHANGEMTU) N(CHANGEINFODATA)
+	N(BONDING_INFO) N(PRECHANGEUPPER) N(CHANGELOWERSTATE)
+	N(UDP_TUNNEL_PUSH_INFO) N(UDP_TUNNEL_DROP_INFO) N(CHANGE_TX_QUEUE_LEN)
 	N(CVLAN_FILTER_PUSH_INFO) N(CVLAN_FILTER_DROP_INFO)
 	N(SVLAN_FILTER_PUSH_INFO) N(SVLAN_FILTER_DROP_INFO)
 	N(PRE_CHANGEADDR) N(OFFLOAD_XSTATS_ENABLE) N(OFFLOAD_XSTATS_DISABLE)
@@ -10103,6 +10103,7 @@ int register_netdevice(struct net_device *dev)
 	return ret;
 
 err_uninit:
+	call_netdevice_notifiers(NETDEV_PRE_UNINIT, dev);
 	if (dev->netdev_ops->ndo_uninit)
 		dev->netdev_ops->ndo_uninit(dev);
 	if (dev->priv_destructor)
@@ -10856,6 +10857,8 @@ void unregister_netdevice_many(struct list_head *head)
 		netdev_name_node_alt_flush(dev);
 		netdev_name_node_free(dev->name_node);
 
+		call_netdevice_notifiers(NETDEV_PRE_UNINIT, dev);
+
 		if (dev->netdev_ops->ndo_uninit)
 			dev->netdev_ops->ndo_uninit(dev);
 
diff --git a/net/core/devlink.c b/net/core/devlink.c
index ee14520d4690..87aa39bc481e 100644
--- a/net/core/devlink.c
+++ b/net/core/devlink.c
@@ -71,6 +71,7 @@ struct devlink {
 	refcount_t refcount;
 	struct completion comp;
 	struct rcu_head rcu;
+	struct notifier_block netdevice_nb;
 	char priv[] __aligned(NETDEV_ALIGN);
 };
 
@@ -9615,6 +9616,9 @@ void devlink_set_features(struct devlink *devlink, u64 features)
 }
 EXPORT_SYMBOL_GPL(devlink_set_features);
 
+static int devlink_netdevice_event(struct notifier_block *nb,
+				   unsigned long event, void *ptr);
+
 /**
  *	devlink_alloc_ns - Allocate new devlink instance resources
  *	in specific namespace
@@ -9645,10 +9649,13 @@ struct devlink *devlink_alloc_ns(const struct devlink_ops *ops,
 
 	ret = xa_alloc_cyclic(&devlinks, &devlink->index, devlink, xa_limit_31b,
 			      &last_id, GFP_KERNEL);
-	if (ret < 0) {
-		kfree(devlink);
-		return NULL;
-	}
+	if (ret < 0)
+		goto err_xa_alloc;
+
+	devlink->netdevice_nb.notifier_call = devlink_netdevice_event;
+	ret = register_netdevice_notifier_net(net, &devlink->netdevice_nb);
+	if (ret)
+		goto err_register_netdevice_notifier;
 
 	devlink->dev = dev;
 	devlink->ops = ops;
@@ -9675,6 +9682,12 @@ struct devlink *devlink_alloc_ns(const struct devlink_ops *ops,
 	init_completion(&devlink->comp);
 
 	return devlink;
+
+err_register_netdevice_notifier:
+	xa_erase(&devlinks, devlink->index);
+err_xa_alloc:
+	kfree(devlink);
+	return NULL;
 }
 EXPORT_SYMBOL_GPL(devlink_alloc_ns);
 
@@ -9828,6 +9841,10 @@ void devlink_free(struct devlink *devlink)
 	WARN_ON(!list_empty(&devlink->port_list));
 
 	xa_destroy(&devlink->snapshot_ids);
+
+	unregister_netdevice_notifier_net(devlink_net(devlink),
+					  &devlink->netdevice_nb);
+
 	xa_erase(&devlinks, devlink->index);
 
 	kfree(devlink);
@@ -10121,6 +10138,56 @@ void devlink_port_type_clear(struct devlink_port *devlink_port)
 }
 EXPORT_SYMBOL_GPL(devlink_port_type_clear);
 
+static int devlink_netdevice_event(struct notifier_block *nb,
+				   unsigned long event, void *ptr)
+{
+	struct net_device *netdev = netdev_notifier_info_to_dev(ptr);
+	struct devlink_port *devlink_port = netdev->devlink_port;
+	struct devlink *devlink;
+
+	devlink = container_of(nb, struct devlink, netdevice_nb);
+
+	if (!devlink_port || devlink_port->devlink != devlink)
+		return NOTIFY_OK;
+
+	switch (event) {
+	case NETDEV_POST_INIT:
+		/* Set the type but not netdev pointer. It is going to be set
+		 * later on by NETDEV_REGISTER event. Happens once during
+		 * netdevice register
+		 */
+		__devlink_port_type_set(devlink_port, DEVLINK_PORT_TYPE_ETH,
+					NULL, true);
+		break;
+	case NETDEV_REGISTER:
+		/* Set the netdev on top of previously set type. Note this
+		 * event happens also during net namespace change so here
+		 * we take into account netdev pointer appearing in this
+		 * namespace.
+		 */
+		__devlink_port_type_set(devlink_port, DEVLINK_PORT_TYPE_ETH,
+					netdev, true);
+		break;
+	case NETDEV_UNREGISTER:
+		/* Clear netdev pointer, but not the type. This event happens
+		 * also during net namespace change so we need to clear
+		 * pointer to netdev that is going to another net namespace.
+		 */
+		__devlink_port_type_set(devlink_port, DEVLINK_PORT_TYPE_ETH,
+					NULL, true);
+		break;
+	case NETDEV_PRE_UNINIT:
+		/* Clear the type and the netdev pointer. Happens one during
+		 * netdevice unregister.
+		 */
+		__devlink_port_type_set(devlink_port, DEVLINK_PORT_TYPE_NOTSET,
+					NULL, true);
+		break;
+	}
+
+	return NOTIFY_OK;
+}
+
 static int __devlink_port_attrs_set(struct devlink_port *devlink_port,
 				    enum devlink_port_flavour flavour)
 {
-- 
2.37.1


  parent reply	other threads:[~2022-10-03 10:52 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-10-03 10:51 [patch net-next v2 00/13] net: fix netdev to devlink_port linkage and expose to user Jiri Pirko
2022-10-03 10:51 ` [patch net-next v2 01/13] net: devlink: convert devlink port type-specific pointers to union Jiri Pirko
2022-10-03 10:51 ` [patch net-next v2 02/13] net: devlink: move port_type_warn_schedule() call to __devlink_port_type_set() Jiri Pirko
2022-10-03 10:51 ` [patch net-next v2 03/13] net: devlink: move port_type_netdev_checks() " Jiri Pirko
2022-10-03 10:51 ` [patch net-next v2 04/13] net: devlink: take RTNL in port_fill() function only if it is not held Jiri Pirko
2022-10-03 10:51 ` Jiri Pirko [this message]
2022-10-03 10:51 ` [patch net-next v2 06/13] net: make drivers to use SET_NETDEV_DEVLINK_PORT to set devlink_port Jiri Pirko
2022-10-03 16:46   ` Jakub Kicinski
2022-10-03 10:51 ` [patch net-next v2 07/13] net: devlink: remove netdev arg from devlink_port_type_eth_set() Jiri Pirko
2022-10-03 10:51 ` [patch net-next v2 08/13] net: devlink: remove net namespace check from devlink_nl_port_fill() Jiri Pirko
2022-10-03 10:52 ` [patch net-next v2 09/13] net: devlink: store copy netdevice ifindex and ifname to allow port_fill() without RTNL held Jiri Pirko
2022-10-03 10:52 ` [patch net-next v2 10/13] net: devlink: add not cleared type warning to port unregister Jiri Pirko
2022-10-03 10:52 ` [patch net-next v2 11/13] net: devlink: use devlink_port pointer instead of ndo_get_devlink_port Jiri Pirko
2022-10-03 10:52 ` [patch net-next v2 12/13] net: remove unused ndo_get_devlink_port Jiri Pirko
2022-10-03 10:52 ` [patch net-next v2 13/13] net: expose devlink port over rtnetlink Jiri Pirko
2022-10-03 16:45 ` [patch net-next v2 00/13] net: fix netdev to devlink_port linkage and expose to user Jakub Kicinski
2022-10-04  6:56   ` Jiri Pirko

Reply instructions:

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

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

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

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

  git send-email \
    --in-reply-to=20221003105204.3315337-6-jiri@resnulli.us \
    --to=jiri@resnulli.us \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=kuba@kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=moshe@nvidia.com \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=saeedm@nvidia.com \
    --cc=tariqt@nvidia.com \
    /path/to/YOUR_REPLY

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

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox