Ido Schimmel
2023-Jan-09 11:43 UTC
[Bridge] [PATCH v5 net-next 01/15] net: bridge: mst: Multiple Spanning Tree (MST) mode
On Mon, Jan 09, 2023 at 12:02:36PM +0200, Vladimir Oltean wrote:> On Mon, Jan 09, 2023 at 10:05:53AM +0200, Ido Schimmel wrote: > > > + if (on) > > > + static_branch_enable(&br_mst_used); > > > + else > > > + static_branch_disable(&br_mst_used); > > > > Hi, > > > > I'm not actually using MST, but I ran into this code and was wondering > > if the static key usage is correct. The static key is global (not > > per-bridge), so what happens when two bridges have MST enabled and then > > it is disabled on one? I believe it would be disabled for both. If so, > > maybe use static_branch_inc() / static_branch_dec() instead? > > Sounds about right. FWIW, br_switchdev_tx_fwd_offload does use > static_branch_inc() / static_branch_dec().OK, thanks for confirming. Will send a patch later this week if Tobias won't take care of it by then. First patch will probably be [1] to make sure we dump the correct MST state to user space. It will also make it easier to show the problem and validate the fix. [1] diff --git a/net/bridge/br.c b/net/bridge/br.c index 4f5098d33a46..f02a1ad589de 100644 --- a/net/bridge/br.c +++ b/net/bridge/br.c @@ -286,7 +286,7 @@ int br_boolopt_get(const struct net_bridge *br, enum br_boolopt_id opt) case BR_BOOLOPT_MCAST_VLAN_SNOOPING: return br_opt_get(br, BROPT_MCAST_VLAN_SNOOPING_ENABLED); case BR_BOOLOPT_MST_ENABLE: - return br_opt_get(br, BROPT_MST_ENABLED); + return br_mst_is_enabled(br); default: /* shouldn't be called with unsupported options */ WARN_ON(1); diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h index 75aff9bbf17e..7f0475f62d45 100644 --- a/net/bridge/br_private.h +++ b/net/bridge/br_private.h @@ -1827,7 +1827,7 @@ static inline bool br_vlan_state_allowed(u8 state, bool learn_allow) /* br_mst.c */ #ifdef CONFIG_BRIDGE_VLAN_FILTERING DECLARE_STATIC_KEY_FALSE(br_mst_used); -static inline bool br_mst_is_enabled(struct net_bridge *br) +static inline bool br_mst_is_enabled(const struct net_bridge *br) { return static_branch_unlikely(&br_mst_used) && br_opt_get(br, BROPT_MST_ENABLED); @@ -1845,7 +1845,7 @@ int br_mst_fill_info(struct sk_buff *skb, int br_mst_process(struct net_bridge_port *p, const struct nlattr *mst_attr, struct netlink_ext_ack *extack); #else -static inline bool br_mst_is_enabled(struct net_bridge *br) +static inline bool br_mst_is_enabled(const struct net_bridge *br) { return false; }
Nikolay Aleksandrov
2023-Jan-09 11:51 UTC
[Bridge] [PATCH v5 net-next 01/15] net: bridge: mst: Multiple Spanning Tree (MST) mode
On 09/01/2023 13:43, Ido Schimmel wrote:> On Mon, Jan 09, 2023 at 12:02:36PM +0200, Vladimir Oltean wrote: >> On Mon, Jan 09, 2023 at 10:05:53AM +0200, Ido Schimmel wrote: >>>> + if (on) >>>> + static_branch_enable(&br_mst_used); >>>> + else >>>> + static_branch_disable(&br_mst_used); >>> >>> Hi, >>> >>> I'm not actually using MST, but I ran into this code and was wondering >>> if the static key usage is correct. The static key is global (not >>> per-bridge), so what happens when two bridges have MST enabled and then >>> it is disabled on one? I believe it would be disabled for both. If so, >>> maybe use static_branch_inc() / static_branch_dec() instead? >> >> Sounds about right. FWIW, br_switchdev_tx_fwd_offload does use >> static_branch_inc() / static_branch_dec(). > > OK, thanks for confirming. Will send a patch later this week if Tobias > won't take care of it by then. First patch will probably be [1] to make > sure we dump the correct MST state to user space. It will also make it > easier to show the problem and validate the fix. > > [1] > diff --git a/net/bridge/br.c b/net/bridge/br.c > index 4f5098d33a46..f02a1ad589de 100644 > --- a/net/bridge/br.c > +++ b/net/bridge/br.c > @@ -286,7 +286,7 @@ int br_boolopt_get(const struct net_bridge *br, enum br_boolopt_id opt) > case BR_BOOLOPT_MCAST_VLAN_SNOOPING: > return br_opt_get(br, BROPT_MCAST_VLAN_SNOOPING_ENABLED); > case BR_BOOLOPT_MST_ENABLE: > - return br_opt_get(br, BROPT_MST_ENABLED); > + return br_mst_is_enabled(br); > default: > /* shouldn't be called with unsupported options */ > WARN_ON(1); > diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h > index 75aff9bbf17e..7f0475f62d45 100644 > --- a/net/bridge/br_private.h > +++ b/net/bridge/br_private.h > @@ -1827,7 +1827,7 @@ static inline bool br_vlan_state_allowed(u8 state, bool learn_allow) > /* br_mst.c */ > #ifdef CONFIG_BRIDGE_VLAN_FILTERING > DECLARE_STATIC_KEY_FALSE(br_mst_used); > -static inline bool br_mst_is_enabled(struct net_bridge *br) > +static inline bool br_mst_is_enabled(const struct net_bridge *br) > { > return static_branch_unlikely(&br_mst_used) && > br_opt_get(br, BROPT_MST_ENABLED); > @@ -1845,7 +1845,7 @@ int br_mst_fill_info(struct sk_buff *skb, > int br_mst_process(struct net_bridge_port *p, const struct nlattr *mst_attr, > struct netlink_ext_ack *extack); > #else > -static inline bool br_mst_is_enabled(struct net_bridge *br) > +static inline bool br_mst_is_enabled(const struct net_bridge *br) > { > return false; > }Ack, good catch. This should've been _inc/_dec indeed. Thanks, Nik
Vladimir Oltean
2023-Jan-09 11:56 UTC
[Bridge] [PATCH v5 net-next 01/15] net: bridge: mst: Multiple Spanning Tree (MST) mode
On Mon, Jan 09, 2023 at 01:43:46PM +0200, Ido Schimmel wrote:> OK, thanks for confirming. Will send a patch later this week if Tobias > won't take care of it by then. First patch will probably be [1] to make > sure we dump the correct MST state to user space. It will also make it > easier to show the problem and validate the fix. > > [1] > diff --git a/net/bridge/br.c b/net/bridge/br.c > index 4f5098d33a46..f02a1ad589de 100644 > --- a/net/bridge/br.c > +++ b/net/bridge/br.c > @@ -286,7 +286,7 @@ int br_boolopt_get(const struct net_bridge *br, enum br_boolopt_id opt) > case BR_BOOLOPT_MCAST_VLAN_SNOOPING: > return br_opt_get(br, BROPT_MCAST_VLAN_SNOOPING_ENABLED); > case BR_BOOLOPT_MST_ENABLE: > - return br_opt_get(br, BROPT_MST_ENABLED); > + return br_mst_is_enabled(br);Well, this did report the correct MST state despite the incorrect static branch state, no? The users of br_mst_is_enabled(br) are broken, not those of br_opt_get(br, BROPT_MST_ENABLED). Anyway, I see there's a br_mst_is_enabled() and also a br_mst_enabled()?! One is used in the fast path and the other in the slow path. They should probably be merged, I guess. They both exist probably because somebody thought that the "if (!netif_is_bridge_master(dev))" test is redundant in the fast path.> default: > /* shouldn't be called with unsupported options */ > WARN_ON(1); > diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h > index 75aff9bbf17e..7f0475f62d45 100644 > --- a/net/bridge/br_private.h > +++ b/net/bridge/br_private.h > @@ -1827,7 +1827,7 @@ static inline bool br_vlan_state_allowed(u8 state, bool learn_allow) > /* br_mst.c */ > #ifdef CONFIG_BRIDGE_VLAN_FILTERING > DECLARE_STATIC_KEY_FALSE(br_mst_used); > -static inline bool br_mst_is_enabled(struct net_bridge *br) > +static inline bool br_mst_is_enabled(const struct net_bridge *br) > { > return static_branch_unlikely(&br_mst_used) && > br_opt_get(br, BROPT_MST_ENABLED); > @@ -1845,7 +1845,7 @@ int br_mst_fill_info(struct sk_buff *skb, > int br_mst_process(struct net_bridge_port *p, const struct nlattr *mst_attr, > struct netlink_ext_ack *extack); > #else > -static inline bool br_mst_is_enabled(struct net_bridge *br) > +static inline bool br_mst_is_enabled(const struct net_bridge *br) > { > return false; > }