[patch net-next v2 0/9] mlxsw: Enable VRF offload

STALE3426d

17 messages, 3 authors, 2017-03-15 · open the first message on its own page

[patch net-next v2 0/9] mlxsw: Enable VRF offload

From: Jiri Pirko <jiri@resnulli.us>
Date: 2017-03-15 11:05:16

From: Jiri Pirko <redacted>

Ido says:

Packets received from netdevs enslaved to different VRF devices are
forwarded using different FIB tables. In the Spectrum ASIC this is
achieved by binding different router interfaces (RIFs) to different
virtual routers (VRs). Each RIF represents an enslaved netdev and each
VR has its own FIB table according to which packets are forwarded.

The first three patches add helpers to check if the FIB rule's selector
matches all packets and extend the FIB notification chain to include the
rule's info as part of the RULE_{ADD,DEL} events. This allows offloading
drivers to sanitize the rules they don't support and flush their tables.

The fourth patch introduces a small change in the VRF driver to allow
capable drivers to more easily offload VRFs.

Finally, the last patches gradually add support for VRFs in the mlxsw
driver. First, on top of port netdevs, stacked LAG and VLAN devices and
then on top of bridges.

Some limitations I would like to point out:

1) The old model where 'oif' / 'iif' rules were programmed for each L3
master device isn't supported. Upon insertion of these rules the driver
will flush its tables and forwarding will be done by the kernel instead.
It's inferior in every way to the single 'l3mdev' rule, so this shouldn't
be an issue.

2) Inter-VRF routes pointing to a VRF device aren't offloaded. Packets
hitting these routes will be forwarded by the kernel. Inter-VRF routes
pointing to netdevs enslaved to a different VRF are offloaded.

3) There's a small discrepancy between the kernel's datapath and the
device's. By default, packets forwarded by the kernel first do a lookup
in the LOCAL table and then in the VRF's table (assuming no match). In
the device, lookup is done only in the VRF's table, which is probably
the intended behavior. Changes in v2 allow user to properly re-order the
default rules without triggering the abort mechanism.

Changes in v2:
* Drop default rule indication and allow re-ordering of default rules
  (David Ahern).
* Remove ifdef around 'struct fib_rule_notifier_info' and drop redundant
  dependency on IP_MULTIPLE_TABLES from rocker and mlxsw.
* Add David's Acked-by to the fourth patch.
* Remove netif_is_vrf_master() and use netif_is_l3_master() instead
  (David Ahern).

Ido Schimmel (9):
  net: fib_rules: Check if selector matches all packets
  ipv4: fib_rules: Add notifier info to FIB rules notifications
  ipv4: fib_rules: Dump FIB rules when registering FIB notifier
  net: vrf: Set slave's private flag before linking
  mlxsw: spectrum_router: Associate RIFs with correct VR
  mlxsw: spectrum_router: Don't destroy RIF if L3 slave
  mlxsw: spectrum_router: Add support for VRFs
  mlxsw: spectrum_router: Add support for VRFs on top of bridges
  mlxsw: spectrum_router: Don't abort on l3mdev rules

 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     | 71 +++++++++++++++-
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  8 ++
 .../net/ethernet/mellanox/mlxsw/spectrum_router.c  | 99 +++++++++++++++++++++-
 drivers/net/ethernet/rocker/rocker_main.c          | 26 +++++-
 drivers/net/vrf.c                                  |  8 +-
 include/net/fib_rules.h                            |  1 +
 include/net/ip_fib.h                               | 12 +++
 net/core/fib_rules.c                               | 14 +++
 net/ipv4/fib_rules.c                               | 42 +++++++--
 9 files changed, 261 insertions(+), 20 deletions(-)

-- 
2.7.4

[patch net-next v2 2/9] ipv4: fib_rules: Add notifier info to FIB rules notifications

From: Jiri Pirko <jiri@resnulli.us>
Date: 2017-03-15 11:05:19

From: Ido Schimmel <redacted>

Whenever a FIB rule is added or removed, a notification is sent in the
FIB notification chain. However, listeners don't have a way to tell
which rule was added or removed.

This is problematic as we would like to give listeners the ability to
decide which action to execute based on the notified rule. Specifically,
offloading drivers should be able to determine if they support the
reflection of the notified FIB rule and flush their LPM tables in case
they don't.

Do that by adding a notifier info to these notifications and embed the
common FIB rule struct in it.

Signed-off-by: Ido Schimmel <redacted>
Signed-off-by: Jiri Pirko <redacted>
---
 include/net/ip_fib.h |  5 +++++
 net/ipv4/fib_rules.c | 13 ++++++++-----
 2 files changed, 13 insertions(+), 5 deletions(-)

diff --git a/include/net/ip_fib.h b/include/net/ip_fib.h
index ec03e0f..319de77 100644
--- a/include/net/ip_fib.h
+++ b/include/net/ip_fib.h

@@ -213,6 +213,11 @@ struct fib_entry_notifier_info {
 	u32 tb_id;
 };
 
+struct fib_rule_notifier_info {
+	struct fib_notifier_info info; /* must be first */
+	struct fib_rule *rule;
+};
+
 struct fib_nh_notifier_info {
 	struct fib_notifier_info info; /* must be first */
 	struct fib_nh *fib_nh;

diff --git a/net/ipv4/fib_rules.c b/net/ipv4/fib_rules.c
index 7a941a5..d9cd3be 100644
--- a/net/ipv4/fib_rules.c
+++ b/net/ipv4/fib_rules.c

@@ -175,11 +175,14 @@ static struct fib_table *fib_empty_table(struct net *net)
 }
 
 static int call_fib_rule_notifiers(struct net *net,
-				   enum fib_event_type event_type)
+				   enum fib_event_type event_type,
+				   struct fib_rule *rule)
 {
-	struct fib_notifier_info info;
+	struct fib_rule_notifier_info info = {
+		.rule = rule,
+	};
 
-	return call_fib_notifiers(net, event_type, &info);
+	return call_fib_notifiers(net, event_type, &info.info);
 }
 
 void fib_rules_notify(struct net *net, struct notifier_block *nb)

@@ -246,7 +249,7 @@ static int fib4_rule_configure(struct fib_rule *rule, struct sk_buff *skb,
 	rule4->tos = frh->tos;
 
 	net->ipv4.fib_has_custom_rules = true;
-	call_fib_rule_notifiers(net, FIB_EVENT_RULE_ADD);
+	call_fib_rule_notifiers(net, FIB_EVENT_RULE_ADD, rule);
 
 	err = 0;
 errout:

@@ -268,7 +271,7 @@ static int fib4_rule_delete(struct fib_rule *rule)
 		net->ipv4.fib_num_tclassid_users--;
 #endif
 	net->ipv4.fib_has_custom_rules = true;
-	call_fib_rule_notifiers(net, FIB_EVENT_RULE_DEL);
+	call_fib_rule_notifiers(net, FIB_EVENT_RULE_DEL, rule);
 errout:
 	return err;
 }

-- 
2.7.4

Re: [patch net-next v2 2/9] ipv4: fib_rules: Add notifier info to FIB rules notifications

From: David Ahern <hidden>
Date: 2017-03-15 15:16:51

On 3/15/17 5:05 AM, Jiri Pirko wrote:

From: Ido Schimmel <redacted>

Whenever a FIB rule is added or removed, a notification is sent in the
FIB notification chain. However, listeners don't have a way to tell
which rule was added or removed.

This is problematic as we would like to give listeners the ability to
decide which action to execute based on the notified rule. Specifically,
offloading drivers should be able to determine if they support the
reflection of the notified FIB rule and flush their LPM tables in case
they don't.

Do that by adding a notifier info to these notifications and embed the
common FIB rule struct in it.

Signed-off-by: Ido Schimmel <redacted>
Signed-off-by: Jiri Pirko <redacted>
---
 include/net/ip_fib.h |  5 +++++
 net/ipv4/fib_rules.c | 13 ++++++++-----
 2 files changed, 13 insertions(+), 5 deletions(-)

Acked-by: David Ahern <redacted>

[patch net-next v2 1/9] net: fib_rules: Check if selector matches all packets

From: Jiri Pirko <jiri@resnulli.us>
Date: 2017-03-15 11:05:23

From: Ido Schimmel <redacted>

Currently, when non-default (custom) FIB rules are used, devices capable
of layer 3 offloading flush their tables and let the kernel do the
forwarding instead.

When these devices' drivers are loaded they register to the FIB
notification chain, which lets them know about the existence of any
custom FIB rules. This is done by sending a RULE_ADD notification based
on the value of 'net->ipv4.fib_has_custom_rules'.

This approach is problematic when VRF offload is taken into account, as
upon the creation of the first VRF netdev, a l3mdev rule is programmed
to direct skbs to the VRF's table.

Instead of merely reading the above value and sending a single RULE_ADD
notification, we should iterate over all the FIB rules and send a
detailed notification for each, thereby allowing offloading drivers to
sanitize the rules they don't support and potentially flush their
tables.

While l3mdev rules are uniquely marked, the default rules are not.
Therefore, when they are being notified they might invoke offloading
drivers to unnecessarily flush their tables.

Solve this by adding helpers that check whether a FIB rule's selector
matches all packets and is thus essentially a NOP. Unlike the selector,
the rule's action is completely visible to potential listeners and can
therefore be sanitized by them.

As noted by David Ahern, uniquely marking the default rules is
insufficient. When using VRFs, it's common to avoid false hits by moving
the rule for the local table to just before the main table:

Default configuration:
$ ip rule show
0:      from all lookup local
32766:  from all lookup main
32767:  from all lookup default

Common configuration with VRFs:
$ ip rule show
1000:   from all lookup [l3mdev-table]
32765:  from all lookup local
32766:  from all lookup main
32767:  from all lookup default

Signed-off-by: Ido Schimmel <redacted>
Signed-off-by: Jiri Pirko <redacted>
---
 include/net/fib_rules.h |  1 +
 include/net/ip_fib.h    |  7 +++++++
 net/core/fib_rules.c    | 14 ++++++++++++++
 net/ipv4/fib_rules.c    | 10 ++++++++++
 4 files changed, 32 insertions(+)

diff --git a/include/net/fib_rules.h b/include/net/fib_rules.h
index 8dbfdf7..1243b9c 100644
--- a/include/net/fib_rules.h
+++ b/include/net/fib_rules.h

@@ -141,6 +141,7 @@ int fib_rules_lookup(struct fib_rules_ops *, struct flowi *, int flags,
 		     struct fib_lookup_arg *);
 int fib_default_rule_add(struct fib_rules_ops *, u32 pref, u32 table,
 			 u32 flags);
+bool fib_rule_matchall(const struct fib_rule *rule);
 
 int fib_nl_newrule(struct sk_buff *skb, struct nlmsghdr *nlh);
 int fib_nl_delrule(struct sk_buff *skb, struct nlmsghdr *nlh);

diff --git a/include/net/ip_fib.h b/include/net/ip_fib.h
index d9cee96..ec03e0f 100644
--- a/include/net/ip_fib.h
+++ b/include/net/ip_fib.h

@@ -311,6 +311,11 @@ static inline int fib_lookup(struct net *net, const struct flowi4 *flp,
 	return err;
 }
 
+static inline bool fib4_rule_matchall(const struct fib_rule *rule)
+{
+	return true;
+}
+
 #else /* CONFIG_IP_MULTIPLE_TABLES */
 int __net_init fib4_rules_init(struct net *net);
 void __net_exit fib4_rules_exit(struct net *net);

@@ -355,6 +360,8 @@ static inline int fib_lookup(struct net *net, struct flowi4 *flp,
 	return err;
 }
 
+bool fib4_rule_matchall(const struct fib_rule *rule);
+
 #endif /* CONFIG_IP_MULTIPLE_TABLES */
 
 /* Exported by fib_frontend.c */

diff --git a/net/core/fib_rules.c b/net/core/fib_rules.c
index b6791d9..4ed475d 100644
--- a/net/core/fib_rules.c
+++ b/net/core/fib_rules.c

@@ -23,6 +23,20 @@ static const struct fib_kuid_range fib_kuid_range_unset = {
 	KUIDT_INIT(~0),
 };
 
+bool fib_rule_matchall(const struct fib_rule *rule)
+{
+	if (rule->iifindex || rule->oifindex || rule->mark || rule->l3mdev ||
+	    rule->tun_id || rule->flags)
+		return false;
+	if (rule->suppress_ifgroup != -1 || rule->suppress_prefixlen != -1)
+		return false;
+	if (!uid_eq(rule->uid_range.start, fib_kuid_range_unset.start) ||
+	    !uid_eq(rule->uid_range.end, fib_kuid_range_unset.end))
+		return false;
+	return true;
+}
+EXPORT_SYMBOL_GPL(fib_rule_matchall);
+
 int fib_default_rule_add(struct fib_rules_ops *ops,
 			 u32 pref, u32 table, u32 flags)
 {

diff --git a/net/ipv4/fib_rules.c b/net/ipv4/fib_rules.c
index 2892109..7a941a5 100644
--- a/net/ipv4/fib_rules.c
+++ b/net/ipv4/fib_rules.c

@@ -47,6 +47,16 @@ struct fib4_rule {
 #endif
 };
 
+bool fib4_rule_matchall(const struct fib_rule *rule)
+{
+	struct fib4_rule *r = (struct fib4_rule *) rule;
+
+	if (r->dst_len || r->src_len || r->tos)
+		return false;
+	return fib_rule_matchall(rule);
+}
+EXPORT_SYMBOL_GPL(fib4_rule_matchall);
+
 int __fib_lookup(struct net *net, struct flowi4 *flp,
 		 struct fib_result *res, unsigned int flags)
 {

-- 
2.7.4

Re: [patch net-next v2 1/9] net: fib_rules: Check if selector matches all packets

From: David Ahern <hidden>
Date: 2017-03-15 15:15:46

On 3/15/17 5:05 AM, Jiri Pirko wrote:

quoted hunk

diff --git a/net/core/fib_rules.c b/net/core/fib_rules.c
index b6791d9..4ed475d 100644
--- a/net/core/fib_rules.c
+++ b/net/core/fib_rules.c

@@ -23,6 +23,20 @@ static const struct fib_kuid_range fib_kuid_range_unset = {
 	KUIDT_INIT(~0),
 };
 
+bool fib_rule_matchall(const struct fib_rule *rule)
+{
+	if (rule->iifindex || rule->oifindex || rule->mark || rule->l3mdev ||

l3mdev should not be in that list. Setting l3mdev is functionally
equivalent to setting rule->table. The difference is that l3mdev means
go get the table from the device. It has no bearing on a 'matchall' intent.

quoted hunk

+	    rule->tun_id || rule->flags)
+		return false;
+	if (rule->suppress_ifgroup != -1 || rule->suppress_prefixlen != -1)
+		return false;
+	if (!uid_eq(rule->uid_range.start, fib_kuid_range_unset.start) ||
+	    !uid_eq(rule->uid_range.end, fib_kuid_range_unset.end))
+		return false;
+	return true;
+}
+EXPORT_SYMBOL_GPL(fib_rule_matchall);
+
 int fib_default_rule_add(struct fib_rules_ops *ops,
 			 u32 pref, u32 table, u32 flags)
 {

diff --git a/net/ipv4/fib_rules.c b/net/ipv4/fib_rules.c
index 2892109..7a941a5 100644
--- a/net/ipv4/fib_rules.c
+++ b/net/ipv4/fib_rules.c

@@ -47,6 +47,16 @@ struct fib4_rule {
 #endif
 };
 
+bool fib4_rule_matchall(const struct fib_rule *rule)
+{
+	struct fib4_rule *r = (struct fib4_rule *) rule;

Use container_of instead of typecast.

Re: [patch net-next v2 1/9] net: fib_rules: Check if selector matches all packets

From: Ido Schimmel <hidden>
Date: 2017-03-15 15:30:34

On Wed, Mar 15, 2017 at 09:15:43AM -0600, David Ahern wrote:

On 3/15/17 5:05 AM, Jiri Pirko wrote:

quoted

diff --git a/net/core/fib_rules.c b/net/core/fib_rules.c
index b6791d9..4ed475d 100644
--- a/net/core/fib_rules.c
+++ b/net/core/fib_rules.c

@@ -23,6 +23,20 @@ static const struct fib_kuid_range fib_kuid_range_unset = {
 	KUIDT_INIT(~0),
 };
 
+bool fib_rule_matchall(const struct fib_rule *rule)
+{
+	if (rule->iifindex || rule->oifindex || rule->mark || rule->l3mdev ||

l3mdev should not be in that list. Setting l3mdev is functionally
equivalent to setting rule->table. The difference is that l3mdev means
go get the table from the device. It has no bearing on a 'matchall' intent.

Good idea. Will drop this and the last patch.

quoted

+	    rule->tun_id || rule->flags)
+		return false;
+	if (rule->suppress_ifgroup != -1 || rule->suppress_prefixlen != -1)
+		return false;
+	if (!uid_eq(rule->uid_range.start, fib_kuid_range_unset.start) ||
+	    !uid_eq(rule->uid_range.end, fib_kuid_range_unset.end))
+		return false;
+	return true;
+}
+EXPORT_SYMBOL_GPL(fib_rule_matchall);
+
 int fib_default_rule_add(struct fib_rules_ops *ops,
 			 u32 pref, u32 table, u32 flags)
 {

diff --git a/net/ipv4/fib_rules.c b/net/ipv4/fib_rules.c
index 2892109..7a941a5 100644
--- a/net/ipv4/fib_rules.c
+++ b/net/ipv4/fib_rules.c

@@ -47,6 +47,16 @@ struct fib4_rule {
 #endif
 };
 
+bool fib4_rule_matchall(const struct fib_rule *rule)
+{
+	struct fib4_rule *r = (struct fib4_rule *) rule;

Use container_of instead of typecast.

There are multiple conversions from 'fib_rule' to 'fib4_rule' in this
file, all use typecast.

Thanks David.

Re: [patch net-next v2 1/9] net: fib_rules: Check if selector matches all packets

From: David Ahern <hidden>
Date: 2017-03-15 15:33:18

On 3/15/17 9:30 AM, Ido Schimmel wrote:

quoted

diff --git a/net/ipv4/fib_rules.c b/net/ipv4/fib_rules.c
index 2892109..7a941a5 100644
--- a/net/ipv4/fib_rules.c
+++ b/net/ipv4/fib_rules.c

@@ -47,6 +47,16 @@ struct fib4_rule {
 #endif
 };
 
+bool fib4_rule_matchall(const struct fib_rule *rule)
+{
+	struct fib4_rule *r = (struct fib4_rule *) rule;

Use container_of instead of typecast.

There are multiple conversions from 'fib_rule' to 'fib4_rule' in this
file, all use typecast.

understood. dst to rtable/rt6_info has the same problem. Rather than
continuing with the typecast, new code should use container_of and older
code can be converted in time.

Re: [patch net-next v2 1/9] net: fib_rules: Check if selector matches all packets

From: Ido Schimmel <hidden>
Date: 2017-03-15 15:37:20

On Wed, Mar 15, 2017 at 09:33:14AM -0600, David Ahern wrote:

On 3/15/17 9:30 AM, Ido Schimmel wrote:

quoted

diff --git a/net/ipv4/fib_rules.c b/net/ipv4/fib_rules.c
index 2892109..7a941a5 100644
--- a/net/ipv4/fib_rules.c
+++ b/net/ipv4/fib_rules.c

@@ -47,6 +47,16 @@ struct fib4_rule {
 #endif
 };
 
+bool fib4_rule_matchall(const struct fib_rule *rule)
+{
+	struct fib4_rule *r = (struct fib4_rule *) rule;

Use container_of instead of typecast.

There are multiple conversions from 'fib_rule' to 'fib4_rule' in this
file, all use typecast.

understood. dst to rtable/rt6_info has the same problem. Rather than
continuing with the typecast, new code should use container_of and older
code can be converted in time.

OK. Will convert the others in follow-up.

[patch net-next v2 3/9] ipv4: fib_rules: Dump FIB rules when registering FIB notifier

From: Jiri Pirko <jiri@resnulli.us>
Date: 2017-03-15 11:05:25

From: Ido Schimmel <redacted>

In commit c3852ef7f2f8 ("ipv4: fib: Replay events when registering FIB
notifier") we dumped the FIB tables and replayed the events to the
passed notification block.

However, we merely sent a RULE_ADD notification in case custom rules
were in use. As explained in previous patches, this approach won't work
anymore. Instead, we should notify the caller about all the FIB rules
and let it act accordingly.

Upon registration to the FIB notification chain, replay a RULE_ADD
notification for each programmed FIB rule, custom or not. The integrity
of the dump is ensured by the mechanism introduced in the above
mentioned commit.

Prevent regressions by making sure current listeners correctly sanitize
the notified rules.

Signed-off-by: Ido Schimmel <redacted>
Signed-off-by: Jiri Pirko <redacted>
---
 .../net/ethernet/mellanox/mlxsw/spectrum_router.c  | 22 +++++++++++++++++-
 drivers/net/ethernet/rocker/rocker_main.c          | 26 ++++++++++++++++++++--
 net/ipv4/fib_rules.c                               | 19 +++++++++++++---
 3 files changed, 61 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index 80345a1..7a714d4 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c

@@ -45,6 +45,7 @@
 #include <net/neighbour.h>
 #include <net/arp.h>
 #include <net/ip_fib.h>
+#include <net/fib_rules.h>
 
 #include "spectrum.h"
 #include "core.h"

@@ -2510,10 +2511,20 @@ static void mlxsw_sp_router_fib4_abort(struct mlxsw_sp *mlxsw_sp)
 		dev_warn(mlxsw_sp->bus_info->dev, "Failed to set abort trap.\n");
 }
 
+static bool mlxsw_sp_fib4_rule_default(const struct fib_rule *rule)
+{
+	if (!fib4_rule_matchall(rule) || rule->action != FR_ACT_TO_TBL ||
+	    (rule->table != RT_TABLE_LOCAL && rule->table != RT_TABLE_MAIN &&
+	     rule->table != RT_TABLE_DEFAULT))
+		return false;
+	return true;
+}
+
 struct mlxsw_sp_fib_event_work {
 	struct work_struct work;
 	union {
 		struct fib_entry_notifier_info fen_info;
+		struct fib_rule_notifier_info fr_info;
 		struct fib_nh_notifier_info fnh_info;
 	};
 	struct mlxsw_sp *mlxsw_sp;

@@ -2525,6 +2536,7 @@ static void mlxsw_sp_router_fib_event_work(struct work_struct *work)
 	struct mlxsw_sp_fib_event_work *fib_work =
 		container_of(work, struct mlxsw_sp_fib_event_work, work);
 	struct mlxsw_sp *mlxsw_sp = fib_work->mlxsw_sp;
+	struct fib_rule *rule;
 	bool replace, append;
 	int err;

@@ -2548,7 +2560,10 @@ static void mlxsw_sp_router_fib_event_work(struct work_struct *work)
 		break;
 	case FIB_EVENT_RULE_ADD: /* fall through */
 	case FIB_EVENT_RULE_DEL:
-		mlxsw_sp_router_fib4_abort(mlxsw_sp);
+		rule = fib_work->fr_info.rule;
+		if (!mlxsw_sp_fib4_rule_default(rule))
+			mlxsw_sp_router_fib4_abort(mlxsw_sp);
+		fib_rule_put(rule);
 		break;
 	case FIB_EVENT_NH_ADD: /* fall through */
 	case FIB_EVENT_NH_DEL:

@@ -2591,6 +2606,11 @@ static int mlxsw_sp_router_fib_event(struct notifier_block *nb,
 		 */
 		fib_info_hold(fib_work->fen_info.fi);
 		break;
+	case FIB_EVENT_RULE_ADD: /* fall through */
+	case FIB_EVENT_RULE_DEL:
+		memcpy(&fib_work->fr_info, ptr, sizeof(fib_work->fr_info));
+		fib_rule_get(fib_work->fr_info.rule);
+		break;
 	case FIB_EVENT_NH_ADD: /* fall through */
 	case FIB_EVENT_NH_DEL:
 		memcpy(&fib_work->fnh_info, ptr, sizeof(fib_work->fnh_info));

diff --git a/drivers/net/ethernet/rocker/rocker_main.c b/drivers/net/ethernet/rocker/rocker_main.c
index b712ec2..88aaf6e 100644
--- a/drivers/net/ethernet/rocker/rocker_main.c
+++ b/drivers/net/ethernet/rocker/rocker_main.c

@@ -33,6 +33,7 @@
 #include <net/rtnetlink.h>
 #include <net/netevent.h>
 #include <net/arp.h>
+#include <net/fib_rules.h>
 #include <linux/io-64-nonatomic-lo-hi.h>
 #include <generated/utsrelease.h>

@@ -1746,6 +1747,15 @@ static void rocker_world_fib4_abort(struct rocker *rocker)
 		wops->fib4_abort(rocker);
 }
 
+static bool rocker_world_fib4_rule_default(const struct fib_rule *rule)
+{
+	if (!fib4_rule_matchall(rule) || rule->action != FR_ACT_TO_TBL ||
+	    (rule->table != RT_TABLE_LOCAL && rule->table != RT_TABLE_MAIN &&
+	     rule->table != RT_TABLE_DEFAULT))
+		return false;
+	return true;
+}
+
 /*****************
  * Net device ops
  *****************/

@@ -2175,7 +2185,10 @@ static const struct switchdev_ops rocker_port_switchdev_ops = {
 
 struct rocker_fib_event_work {
 	struct work_struct work;
-	struct fib_entry_notifier_info fen_info;
+	union {
+		struct fib_entry_notifier_info fen_info;
+		struct fib_rule_notifier_info fr_info;
+	};
 	struct rocker *rocker;
 	unsigned long event;
 };

@@ -2185,6 +2198,7 @@ static void rocker_router_fib_event_work(struct work_struct *work)
 	struct rocker_fib_event_work *fib_work =
 		container_of(work, struct rocker_fib_event_work, work);
 	struct rocker *rocker = fib_work->rocker;
+	struct fib_rule *rule;
 	int err;
 
 	/* Protect internal structures from changes */

@@ -2202,7 +2216,10 @@ static void rocker_router_fib_event_work(struct work_struct *work)
 		break;
 	case FIB_EVENT_RULE_ADD: /* fall through */
 	case FIB_EVENT_RULE_DEL:
-		rocker_world_fib4_abort(rocker);
+		rule = fib_work->fr_info.rule;
+		if (!rocker_world_fib4_rule_default(rule))
+			rocker_world_fib4_abort(rocker);
+		fib_rule_put(rule);
 		break;
 	}
 	rtnl_unlock();

@@ -2233,6 +2250,11 @@ static int rocker_router_fib_event(struct notifier_block *nb,
 		 */
 		fib_info_hold(fib_work->fen_info.fi);
 		break;
+	case FIB_EVENT_RULE_ADD: /* fall through */
+	case FIB_EVENT_RULE_DEL:
+		memcpy(&fib_work->fr_info, ptr, sizeof(fib_work->fr_info));
+		fib_rule_get(fib_work->fr_info.rule);
+		break;
 	}
 
 	queue_work(rocker->rocker_owq, &fib_work->work);

diff --git a/net/ipv4/fib_rules.c b/net/ipv4/fib_rules.c
index d9cd3be..f4afc9f 100644
--- a/net/ipv4/fib_rules.c
+++ b/net/ipv4/fib_rules.c

@@ -174,6 +174,17 @@ static struct fib_table *fib_empty_table(struct net *net)
 	return NULL;
 }
 
+static int call_fib_rule_notifier(struct notifier_block *nb, struct net *net,
+				  enum fib_event_type event_type,
+				  struct fib_rule *rule)
+{
+	struct fib_rule_notifier_info info = {
+		.rule = rule,
+	};
+
+	return call_fib_notifier(nb, net, event_type, &info.info);
+}
+
 static int call_fib_rule_notifiers(struct net *net,
 				   enum fib_event_type event_type,
 				   struct fib_rule *rule)

@@ -185,12 +196,14 @@ static int call_fib_rule_notifiers(struct net *net,
 	return call_fib_notifiers(net, event_type, &info.info);
 }
 
+/* Called with rcu_read_lock() */
 void fib_rules_notify(struct net *net, struct notifier_block *nb)
 {
-	struct fib_notifier_info info;
+	struct fib_rules_ops *ops = net->ipv4.rules_ops;
+	struct fib_rule *rule;
 
-	if (net->ipv4.fib_has_custom_rules)
-		call_fib_notifier(nb, net, FIB_EVENT_RULE_ADD, &info);
+	list_for_each_entry_rcu(rule, &ops->rules_list, list)
+		call_fib_rule_notifier(nb, net, FIB_EVENT_RULE_ADD, rule);
 }
 
 static const struct nla_policy fib4_rule_policy[FRA_MAX+1] = {

-- 
2.7.4

Re: [patch net-next v2 3/9] ipv4: fib_rules: Dump FIB rules when registering FIB notifier

From: David Ahern <hidden>
Date: 2017-03-15 15:22:14

On 3/15/17 5:05 AM, Jiri Pirko wrote:

quoted hunk

@@ -2510,10 +2511,20 @@ static void mlxsw_sp_router_fib4_abort(struct mlxsw_sp *mlxsw_sp)
 		dev_warn(mlxsw_sp->bus_info->dev, "Failed to set abort trap.\n");
 }
 
+static bool mlxsw_sp_fib4_rule_default(const struct fib_rule *rule)
+{
+	if (!fib4_rule_matchall(rule) || rule->action != FR_ACT_TO_TBL ||
+	    (rule->table != RT_TABLE_LOCAL && rule->table != RT_TABLE_MAIN &&
+	     rule->table != RT_TABLE_DEFAULT))
+		return false;
+	return true;
+}
+

Since a default rule is defined by fib_default_rule_add(), and its
callers, IPv4 and IPv6 should have the above as an exported helper.

Re: [patch net-next v2 3/9] ipv4: fib_rules: Dump FIB rules when registering FIB notifier

From: Ido Schimmel <hidden>
Date: 2017-03-15 15:36:21

On Wed, Mar 15, 2017 at 09:22:10AM -0600, David Ahern wrote:

On 3/15/17 5:05 AM, Jiri Pirko wrote:

quoted

@@ -2510,10 +2511,20 @@ static void mlxsw_sp_router_fib4_abort(struct mlxsw_sp *mlxsw_sp)
 		dev_warn(mlxsw_sp->bus_info->dev, "Failed to set abort trap.\n");
 }
 
+static bool mlxsw_sp_fib4_rule_default(const struct fib_rule *rule)
+{
+	if (!fib4_rule_matchall(rule) || rule->action != FR_ACT_TO_TBL ||
+	    (rule->table != RT_TABLE_LOCAL && rule->table != RT_TABLE_MAIN &&
+	     rule->table != RT_TABLE_DEFAULT))
+		return false;
+	return true;
+}
+

Since a default rule is defined by fib_default_rule_add(), and its
callers, IPv4 and IPv6 should have the above as an exported helper.

OK. Will add this in IPv4 code. When we add support for IPv6 I'll do the
same there.

Thanks

[patch net-next v2 4/9] net: vrf: Set slave's private flag before linking

From: Jiri Pirko <jiri@resnulli.us>
Date: 2017-03-15 11:05:26

From: Ido Schimmel <redacted>

Allow listeners of the subsequent CHANGEUPPER notification to retrieve
the VRF's table ID by calling l3mdev_fib_table() with the slave netdev.
Without this change, the netdev won't be considered an L3 slave and the
function would return 0.

This is consistent with other master device such as bridge and bond that
set the slave's private flag before linking. It also makes
do_vrf_{add,del}_slave() symmetric.

Signed-off-by: Ido Schimmel <redacted>
Acked-by: David Ahern <redacted>
Signed-off-by: Jiri Pirko <redacted>
---
 drivers/net/vrf.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/net/vrf.c b/drivers/net/vrf.c
index 22379da..c693267 100644
--- a/drivers/net/vrf.c
+++ b/drivers/net/vrf.c

@@ -746,14 +746,18 @@ static int do_vrf_add_slave(struct net_device *dev, struct net_device *port_dev)
 {
 	int ret;
 
+	port_dev->priv_flags |= IFF_L3MDEV_SLAVE;
 	ret = netdev_master_upper_dev_link(port_dev, dev, NULL, NULL);
 	if (ret < 0)
-		return ret;
+		goto err;
 
-	port_dev->priv_flags |= IFF_L3MDEV_SLAVE;
 	cycle_netdev(port_dev);
 
 	return 0;
+
+err:
+	port_dev->priv_flags &= ~IFF_L3MDEV_SLAVE;
+	return ret;
 }
 
 static int vrf_add_slave(struct net_device *dev, struct net_device *port_dev)

-- 
2.7.4

[patch net-next v2 5/9] mlxsw: spectrum_router: Associate RIFs with correct VR

From: Jiri Pirko <jiri@resnulli.us>
Date: 2017-03-15 11:05:27

From: Ido Schimmel <redacted>

When a router interface (RIF) is created due to a netdev being enslaved
to a VRF master, then it should be associated with the appropriate
virtual router (VR) and not the default one.

If netdev is a VRF slave, lookup the VR based on the VRF's table ID.
Otherwise default to the MAIN table.

Signed-off-by: Ido Schimmel <redacted>
Signed-off-by: Jiri Pirko <redacted>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index 7a714d4..c29713a 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c

@@ -46,6 +46,7 @@
 #include <net/arp.h>
 #include <net/ip_fib.h>
 #include <net/fib_rules.h>
+#include <net/l3mdev.h>
 
 #include "spectrum.h"
 #include "core.h"

@@ -2771,6 +2772,7 @@ mlxsw_sp_vport_rif_sp_create(struct mlxsw_sp_port *mlxsw_sp_vport,
 			     struct net_device *l3_dev)
 {
 	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_vport->mlxsw_sp;
+	u32 tb_id = l3mdev_fib_table(l3_dev);
 	struct mlxsw_sp_vr *vr;
 	struct mlxsw_sp_fid *f;
 	struct mlxsw_sp_rif *r;

@@ -2781,7 +2783,7 @@ mlxsw_sp_vport_rif_sp_create(struct mlxsw_sp_port *mlxsw_sp_vport,
 	if (rif == MLXSW_SP_INVALID_RIF)
 		return ERR_PTR(-ERANGE);
 
-	vr = mlxsw_sp_vr_get(mlxsw_sp, RT_TABLE_MAIN);
+	vr = mlxsw_sp_vr_get(mlxsw_sp, tb_id ? : RT_TABLE_MAIN);
 	if (IS_ERR(vr))
 		return ERR_CAST(vr);

@@ -3019,6 +3021,7 @@ static int mlxsw_sp_rif_bridge_create(struct mlxsw_sp *mlxsw_sp,
 				      struct net_device *l3_dev,
 				      struct mlxsw_sp_fid *f)
 {
+	u32 tb_id = l3mdev_fib_table(l3_dev);
 	struct mlxsw_sp_vr *vr;
 	struct mlxsw_sp_rif *r;
 	u16 rif;

@@ -3028,7 +3031,7 @@ static int mlxsw_sp_rif_bridge_create(struct mlxsw_sp *mlxsw_sp,
 	if (rif == MLXSW_SP_INVALID_RIF)
 		return -ERANGE;
 
-	vr = mlxsw_sp_vr_get(mlxsw_sp, RT_TABLE_MAIN);
+	vr = mlxsw_sp_vr_get(mlxsw_sp, tb_id ? : RT_TABLE_MAIN);
 	if (IS_ERR(vr))
 		return PTR_ERR(vr);

-- 
2.7.4

[patch net-next v2 9/9] mlxsw: spectrum_router: Don't abort on l3mdev rules

From: Jiri Pirko <jiri@resnulli.us>
Date: 2017-03-15 11:05:27

From: Ido Schimmel <redacted>

Before we allow port netdevs to be enslaved to a VRF master we need to
make sure the device's routing tables won't be flushed upon the
insertion of a l3mdev rule.

Note that we assume the notified l3mdev rule is a simple rule as used by
the VRF master. We don't check for the presence of other selectors such
as 'iif' and 'oif'.

Signed-off-by: Ido Schimmel <redacted>
Signed-off-by: Jiri Pirko <redacted>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index 4924d9f..04c8c56 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c

@@ -2563,7 +2563,7 @@ static void mlxsw_sp_router_fib_event_work(struct work_struct *work)
 	case FIB_EVENT_RULE_ADD: /* fall through */
 	case FIB_EVENT_RULE_DEL:
 		rule = fib_work->fr_info.rule;
-		if (!mlxsw_sp_fib4_rule_default(rule))
+		if (!mlxsw_sp_fib4_rule_default(rule) && !rule->l3mdev)
 			mlxsw_sp_router_fib4_abort(mlxsw_sp);
 		fib_rule_put(rule);
 		break;

-- 
2.7.4

[patch net-next v2 6/9] mlxsw: spectrum_router: Don't destroy RIF if L3 slave

From: Jiri Pirko <jiri@resnulli.us>
Date: 2017-03-15 11:05:33

From: Ido Schimmel <redacted>

We usually destroy the netdev's router interface (RIF) when the last IP
address is removed from it.

However, we shouldn't do that if it's enslaved to an L3 master device.

Signed-off-by: Ido Schimmel <redacted>
Signed-off-by: Jiri Pirko <redacted>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index c29713a..ff3cf55 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c

@@ -41,6 +41,7 @@
 #include <linux/in6.h>
 #include <linux/notifier.h>
 #include <linux/inetdevice.h>
+#include <linux/netdevice.h>
 #include <net/netevent.h>
 #include <net/neighbour.h>
 #include <net/arp.h>

@@ -2669,7 +2670,7 @@ static bool mlxsw_sp_rif_should_config(struct mlxsw_sp_rif *r,
 			return true;
 		return false;
 	case NETDEV_DOWN:
-		if (r && !in_dev->ifa_list)
+		if (r && !in_dev->ifa_list && !netif_is_l3_slave(r->dev))
 			return true;
 		/* It is possible we already removed the RIF ourselves
 		 * if it was assigned to a netdev that is now a bridge

-- 
2.7.4

[patch net-next v2 8/9] mlxsw: spectrum_router: Add support for VRFs on top of bridges

From: Jiri Pirko <jiri@resnulli.us>
Date: 2017-03-15 11:05:36

From: Ido Schimmel <redacted>

In a similar fashion to the previous patch, allow bridges and VLAN
devices on top of bridges to be enslaved to a VRF master device.

Signed-off-by: Ido Schimmel <redacted>
Signed-off-by: Jiri Pirko <redacted>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     | 52 +++++++++++++++++++++-
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  4 ++
 .../net/ethernet/mellanox/mlxsw/spectrum_router.c  | 26 +++++++++++
 3 files changed, 81 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 659df32..28019f8 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c

@@ -4111,7 +4111,7 @@ static int mlxsw_sp_netdevice_bridge_event(struct net_device *br_dev,
 	switch (event) {
 	case NETDEV_PRECHANGEUPPER:
 		upper_dev = info->upper_dev;
-		if (!is_vlan_dev(upper_dev))
+		if (!is_vlan_dev(upper_dev) && !netif_is_l3_master(upper_dev))
 			return -EINVAL;
 		if (is_vlan_dev(upper_dev) &&
 		    br_dev != mlxsw_sp->master_bridge.dev)

@@ -4126,6 +4126,12 @@ static int mlxsw_sp_netdevice_bridge_event(struct net_device *br_dev,
 			else
 				mlxsw_sp_master_bridge_vlan_unlink(mlxsw_sp,
 								   upper_dev);
+		} else if (netif_is_l3_master(upper_dev)) {
+			if (info->linking)
+				err = mlxsw_sp_bridge_vrf_join(mlxsw_sp,
+							       br_dev);
+			else
+				mlxsw_sp_bridge_vrf_leave(mlxsw_sp, br_dev);
 		} else {
 			err = -EINVAL;
 			WARN_ON(1);

@@ -4415,6 +4421,47 @@ static int mlxsw_sp_netdevice_lag_vport_event(struct net_device *lag_dev,
 	return 0;
 }
 
+static int mlxsw_sp_netdevice_bridge_vlan_event(struct net_device *vlan_dev,
+						unsigned long event, void *ptr)
+{
+	struct netdev_notifier_changeupper_info *info;
+	struct mlxsw_sp *mlxsw_sp;
+	int err = 0;
+
+	mlxsw_sp = mlxsw_sp_lower_get(vlan_dev);
+	if (!mlxsw_sp)
+		return 0;
+
+	info = ptr;
+
+	switch (event) {
+	case NETDEV_PRECHANGEUPPER:
+		/* VLAN devices are only allowed on top of the
+		 * VLAN-aware bridge.
+		 */
+		if (WARN_ON(vlan_dev_real_dev(vlan_dev) !=
+			    mlxsw_sp->master_bridge.dev))
+			return -EINVAL;
+		if (!netif_is_l3_master(info->upper_dev))
+			return -EINVAL;
+		break;
+	case NETDEV_CHANGEUPPER:
+		if (netif_is_l3_master(info->upper_dev)) {
+			if (info->linking)
+				err = mlxsw_sp_bridge_vrf_join(mlxsw_sp,
+							       vlan_dev);
+			else
+				mlxsw_sp_bridge_vrf_leave(mlxsw_sp, vlan_dev);
+		} else {
+			err = -EINVAL;
+			WARN_ON(1);
+		}
+		break;
+	}
+
+	return err;
+}
+
 static int mlxsw_sp_netdevice_vlan_event(struct net_device *vlan_dev,
 					 unsigned long event, void *ptr)
 {

@@ -4427,6 +4474,9 @@ static int mlxsw_sp_netdevice_vlan_event(struct net_device *vlan_dev,
 	else if (netif_is_lag_master(real_dev))
 		return mlxsw_sp_netdevice_lag_vport_event(real_dev, event, ptr,
 							  vid);
+	else if (netif_is_bridge_master(real_dev))
+		return mlxsw_sp_netdevice_bridge_vlan_event(vlan_dev, event,
+							    ptr);
 
 	return 0;
 }

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index 60004d9..0e223f6 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h

@@ -582,6 +582,10 @@ int mlxsw_sp_vport_vrf_join(struct mlxsw_sp_port *mlxsw_sp_vport);
 void mlxsw_sp_vport_vrf_leave(struct mlxsw_sp_port *mlxsw_sp_vport);
 int mlxsw_sp_port_vrf_join(struct mlxsw_sp_port *mlxsw_sp_port);
 void mlxsw_sp_port_vrf_leave(struct mlxsw_sp_port *mlxsw_sp_port);
+int mlxsw_sp_bridge_vrf_join(struct mlxsw_sp *mlxsw_sp,
+			     struct net_device *l3_dev);
+void mlxsw_sp_bridge_vrf_leave(struct mlxsw_sp *mlxsw_sp,
+			       struct net_device *l3_dev);
 
 int mlxsw_sp_kvdl_alloc(struct mlxsw_sp *mlxsw_sp, unsigned int entry_count);
 void mlxsw_sp_kvdl_free(struct mlxsw_sp *mlxsw_sp, int entry_index);

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index cefa6c6..4924d9f 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c

@@ -3276,6 +3276,32 @@ void mlxsw_sp_port_vrf_leave(struct mlxsw_sp_port *mlxsw_sp_port)
 	mlxsw_sp_vport_vrf_leave(mlxsw_sp_vport);
 }
 
+int mlxsw_sp_bridge_vrf_join(struct mlxsw_sp *mlxsw_sp,
+			     struct net_device *l3_dev)
+{
+	struct mlxsw_sp_fid *f;
+
+	f = mlxsw_sp_bridge_fid_get(mlxsw_sp, l3_dev);
+	if (WARN_ON(!f))
+		return -EINVAL;
+
+	if (f->r)
+		mlxsw_sp_rif_bridge_destroy(mlxsw_sp, f->r);
+
+	return mlxsw_sp_rif_bridge_create(mlxsw_sp, l3_dev, f);
+}
+
+void mlxsw_sp_bridge_vrf_leave(struct mlxsw_sp *mlxsw_sp,
+			       struct net_device *l3_dev)
+{
+	struct mlxsw_sp_fid *f;
+
+	f = mlxsw_sp_bridge_fid_get(mlxsw_sp, l3_dev);
+	if (WARN_ON(!f))
+		return;
+	mlxsw_sp_rif_bridge_destroy(mlxsw_sp, f->r);
+}
+
 static void mlxsw_sp_router_fib_dump_flush(struct notifier_block *nb)
 {
 	struct mlxsw_sp *mlxsw_sp = container_of(nb, struct mlxsw_sp, fib_nb);

-- 
2.7.4

[patch net-next v2 7/9] mlxsw: spectrum_router: Add support for VRFs

From: Jiri Pirko <jiri@resnulli.us>
Date: 2017-03-15 11:05:39

From: Ido Schimmel <redacted>

Allow port netdevs, LAG and VLAN devices stacked on top of these to be
enslaved to a VRF master device.

Upon enslavement, create a router interface (RIF) for the enslaved
netdev and associate it with a virtual router (VR) based on the VRF's
table ID.

If a RIF already exists for the netdev (f.e., due to the existence of an
IP address), then it's deleted and a new one is created with the
appropriate VR binding.

Signed-off-by: Ido Schimmel <redacted>
Signed-off-by: Jiri Pirko <redacted>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     | 19 ++++++++--
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  4 +++
 .../net/ethernet/mellanox/mlxsw/spectrum_router.c  | 41 ++++++++++++++++++++++
 3 files changed, 61 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 475499b..659df32 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c

@@ -3951,7 +3951,8 @@ static int mlxsw_sp_netdevice_port_upper_event(struct net_device *dev,
 		upper_dev = info->upper_dev;
 		if (!is_vlan_dev(upper_dev) &&
 		    !netif_is_lag_master(upper_dev) &&
-		    !netif_is_bridge_master(upper_dev))
+		    !netif_is_bridge_master(upper_dev) &&
+		    !netif_is_l3_master(upper_dev))
 			return -EINVAL;
 		if (!info->linking)
 			break;

@@ -3991,6 +3992,11 @@ static int mlxsw_sp_netdevice_port_upper_event(struct net_device *dev,
 			else
 				mlxsw_sp_port_lag_leave(mlxsw_sp_port,
 							upper_dev);
+		} else if (netif_is_l3_master(upper_dev)) {
+			if (info->linking)
+				err = mlxsw_sp_port_vrf_join(mlxsw_sp_port);
+			else
+				mlxsw_sp_port_vrf_leave(mlxsw_sp_port);
 		} else {
 			err = -EINVAL;
 			WARN_ON(1);

@@ -4353,14 +4359,16 @@ static int mlxsw_sp_netdevice_vport_event(struct net_device *dev,
 	switch (event) {
 	case NETDEV_PRECHANGEUPPER:
 		upper_dev = info->upper_dev;
-		if (!netif_is_bridge_master(upper_dev))
+		if (!netif_is_bridge_master(upper_dev) &&
+		    !netif_is_l3_master(upper_dev))
 			return -EINVAL;
 		if (!info->linking)
 			break;
 		/* We can't have multiple VLAN interfaces configured on
 		 * the same port and being members in the same bridge.
 		 */
-		if (!mlxsw_sp_port_master_bridge_check(mlxsw_sp_port,
+		if (netif_is_bridge_master(upper_dev) &&
+		    !mlxsw_sp_port_master_bridge_check(mlxsw_sp_port,
 						       upper_dev))
 			return -EINVAL;
 		break;

@@ -4372,6 +4380,11 @@ static int mlxsw_sp_netdevice_vport_event(struct net_device *dev,
 								 upper_dev);
 			else
 				mlxsw_sp_vport_bridge_leave(mlxsw_sp_vport);
+		} else if (netif_is_l3_master(upper_dev)) {
+			if (info->linking)
+				err = mlxsw_sp_vport_vrf_join(mlxsw_sp_vport);
+			else
+				mlxsw_sp_vport_vrf_leave(mlxsw_sp_vport);
 		} else {
 			err = -EINVAL;
 			WARN_ON(1);

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index 5502232..60004d9 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h

@@ -578,6 +578,10 @@ int mlxsw_sp_inetaddr_event(struct notifier_block *unused,
 			    unsigned long event, void *ptr);
 void mlxsw_sp_rif_bridge_destroy(struct mlxsw_sp *mlxsw_sp,
 				 struct mlxsw_sp_rif *r);
+int mlxsw_sp_vport_vrf_join(struct mlxsw_sp_port *mlxsw_sp_vport);
+void mlxsw_sp_vport_vrf_leave(struct mlxsw_sp_port *mlxsw_sp_vport);
+int mlxsw_sp_port_vrf_join(struct mlxsw_sp_port *mlxsw_sp_port);
+void mlxsw_sp_port_vrf_leave(struct mlxsw_sp_port *mlxsw_sp_port);
 
 int mlxsw_sp_kvdl_alloc(struct mlxsw_sp *mlxsw_sp, unsigned int entry_count);
 void mlxsw_sp_kvdl_free(struct mlxsw_sp *mlxsw_sp, int entry_index);

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index ff3cf55..cefa6c6 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c

@@ -3235,6 +3235,47 @@ int mlxsw_sp_netdevice_router_port_event(struct net_device *dev)
 	return err;
 }
 
+int mlxsw_sp_vport_vrf_join(struct mlxsw_sp_port *mlxsw_sp_vport)
+{
+	struct mlxsw_sp_fid *f = mlxsw_sp_vport_fid_get(mlxsw_sp_vport);
+	struct net_device *dev = mlxsw_sp_vport->dev;
+
+	/* In case vPort already has a RIF, then we need to drop it.
+	 * A new one will be created using the VRF's VR.
+	 */
+	if (f && f->r)
+		mlxsw_sp_vport_rif_sp_leave(mlxsw_sp_vport);
+
+	return mlxsw_sp_vport_rif_sp_join(mlxsw_sp_vport, dev);
+}
+
+void mlxsw_sp_vport_vrf_leave(struct mlxsw_sp_port *mlxsw_sp_vport)
+{
+	mlxsw_sp_vport_rif_sp_leave(mlxsw_sp_vport);
+}
+
+int mlxsw_sp_port_vrf_join(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp_port *mlxsw_sp_vport;
+
+	mlxsw_sp_vport = mlxsw_sp_port_vport_find(mlxsw_sp_port, 1);
+	if (WARN_ON(!mlxsw_sp_vport))
+		return -EINVAL;
+
+	return mlxsw_sp_vport_vrf_join(mlxsw_sp_vport);
+}
+
+void mlxsw_sp_port_vrf_leave(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp_port *mlxsw_sp_vport;
+
+	mlxsw_sp_vport = mlxsw_sp_port_vport_find(mlxsw_sp_port, 1);
+	if (WARN_ON(!mlxsw_sp_vport))
+		return;
+
+	mlxsw_sp_vport_vrf_leave(mlxsw_sp_vport);
+}
+
 static void mlxsw_sp_router_fib_dump_flush(struct notifier_block *nb)
 {
 	struct mlxsw_sp *mlxsw_sp = container_of(nb, struct mlxsw_sp, fib_nb);

-- 
2.7.4

`h`	back out one level
`j`	next message in thread
`k`	previous message in thread
`l`	drill in
`Esc`	close help / fold thread tree
`?`	toggle this help