| Message ID | 20190413012822.30931-13-olteanv@gmail.com |
|---|---|
| State | Changes Requested |
| Delegated to: | David Miller |
| Headers | show |
| Series | NXP SJA1105 DSA driver | expand |
On Sat, Apr 13, 2019 at 04:28:10AM +0300, Vladimir Oltean wrote: > The current behavior is not as obvious as one would assume (which is > that, if the driver set vlan_filtering_is_global = 1, then checking any > dp->vlan_filtering would yield the same result). Only the ports which > are actively enslaved into a bridge would have vlan_filtering set. > > This makes it tricky for drivers to check what the global state is. > Moreover, the most obvious place to check for this setting, the CPU > port, is not populated since it's not being enslaved to the bridge. > So fix this and make the CPU port hold the global state of VLAN > filtering on this switch. > > Signed-off-by: Vladimir Oltean <olteanv@gmail.com> > --- > Changes in v3: > Patch is new. > > net/dsa/port.c | 8 ++++++++ > 1 file changed, 8 insertions(+) > > diff --git a/net/dsa/port.c b/net/dsa/port.c > index c8eb2cbcea6e..acb4ed1f9929 100644 > --- a/net/dsa/port.c > +++ b/net/dsa/port.c > @@ -190,6 +190,8 @@ static bool dsa_port_can_apply_vlan_filtering(struct dsa_port *dp, > int dsa_port_vlan_filtering(struct dsa_port *dp, bool vlan_filtering, > struct switchdev_trans *trans) > { > + /* Violate a const pointer here */ > + struct dsa_port *cpu_dp = (struct dsa_port *)dp->cpu_dp; Hi Vladimir As compilers get more picky, i expect that is going to result in a warning. Since this is a switch global attribute, putting it in dsa_switch would be better, next to vlan_filteris_is_global. Andrew
On Sat, 13 Apr 2019 at 18:23, Andrew Lunn <andrew@lunn.ch> wrote: > > On Sat, Apr 13, 2019 at 04:28:10AM +0300, Vladimir Oltean wrote: > > The current behavior is not as obvious as one would assume (which is > > that, if the driver set vlan_filtering_is_global = 1, then checking any > > dp->vlan_filtering would yield the same result). Only the ports which > > are actively enslaved into a bridge would have vlan_filtering set. > > > > This makes it tricky for drivers to check what the global state is. > > Moreover, the most obvious place to check for this setting, the CPU > > port, is not populated since it's not being enslaved to the bridge. > > So fix this and make the CPU port hold the global state of VLAN > > filtering on this switch. > > > > Signed-off-by: Vladimir Oltean <olteanv@gmail.com> > > --- > > Changes in v3: > > Patch is new. > > > > net/dsa/port.c | 8 ++++++++ > > 1 file changed, 8 insertions(+) > > > > diff --git a/net/dsa/port.c b/net/dsa/port.c > > index c8eb2cbcea6e..acb4ed1f9929 100644 > > --- a/net/dsa/port.c > > +++ b/net/dsa/port.c > > @@ -190,6 +190,8 @@ static bool dsa_port_can_apply_vlan_filtering(struct dsa_port *dp, > > int dsa_port_vlan_filtering(struct dsa_port *dp, bool vlan_filtering, > > struct switchdev_trans *trans) > > { > > + /* Violate a const pointer here */ > > + struct dsa_port *cpu_dp = (struct dsa_port *)dp->cpu_dp; > > Hi Vladimir > > As compilers get more picky, i expect that is going to result in a > warning. > > Since this is a switch global attribute, putting it in dsa_switch > would be better, next to vlan_filteris_is_global. > > Andrew Hi Andrew, Creating a bool ds->vlan_filtering wouldn't make a lot of sense for the majority of drivers. Additionally in my sja1105_filter() function, that would require me to pass through one more pointer (dev->dsa_ptr->vlan_filtering vs dev->dsa_ptr->*ds->*vlan_filtering) to reach the same information. I don't think that keeping it in cpu_dp->vlan_filtering has any semantical overlap with anything else that might appear in the future. And I don't know why the cpu_dp pointer is const. In the dsa_switch_tree it isn't. Thanks, -Vladimir
diff --git a/net/dsa/port.c b/net/dsa/port.c index c8eb2cbcea6e..acb4ed1f9929 100644 --- a/net/dsa/port.c +++ b/net/dsa/port.c @@ -190,6 +190,8 @@ static bool dsa_port_can_apply_vlan_filtering(struct dsa_port *dp, int dsa_port_vlan_filtering(struct dsa_port *dp, bool vlan_filtering, struct switchdev_trans *trans) { + /* Violate a const pointer here */ + struct dsa_port *cpu_dp = (struct dsa_port *)dp->cpu_dp; struct dsa_switch *ds = dp->ds; int err; @@ -209,6 +211,12 @@ int dsa_port_vlan_filtering(struct dsa_port *dp, bool vlan_filtering, return err; dp->vlan_filtering = vlan_filtering; + /* In case of switches where VLAN filtering is not per-port, + * also put the setting in the most unambiguous place to + * retrieve it from. + */ + if (ds->vlan_filtering_is_global) + cpu_dp->vlan_filtering = vlan_filtering; return 0; }
The current behavior is not as obvious as one would assume (which is that, if the driver set vlan_filtering_is_global = 1, then checking any dp->vlan_filtering would yield the same result). Only the ports which are actively enslaved into a bridge would have vlan_filtering set. This makes it tricky for drivers to check what the global state is. Moreover, the most obvious place to check for this setting, the CPU port, is not populated since it's not being enslaved to the bridge. So fix this and make the CPU port hold the global state of VLAN filtering on this switch. Signed-off-by: Vladimir Oltean <olteanv@gmail.com> --- Changes in v3: Patch is new. net/dsa/port.c | 8 ++++++++ 1 file changed, 8 insertions(+)