2019-05-27 14:55:01 +08:00
|
|
|
// SPDX-License-Identifier: GPL-2.0-or-later
|
2017-05-20 05:00:44 +08:00
|
|
|
/*
|
|
|
|
|
* Handling of a single switch port
|
|
|
|
|
*
|
|
|
|
|
* Copyright (c) 2017 Savoir-faire Linux Inc.
|
|
|
|
|
* Vivien Didelot <vivien.didelot@savoirfairelinux.com>
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
#include <linux/if_bridge.h>
|
2022-09-11 09:06:59 +08:00
|
|
|
#include <linux/netdevice.h>
|
2017-05-20 05:00:45 +08:00
|
|
|
#include <linux/notifier.h>
|
2017-10-26 22:50:07 +08:00
|
|
|
#include <linux/of_mdio.h>
|
|
|
|
|
#include <linux/of_net.h>
|
2017-05-20 05:00:44 +08:00
|
|
|
|
2022-11-21 21:55:52 +08:00
|
|
|
#include "dsa.h"
|
2022-11-21 21:55:44 +08:00
|
|
|
#include "port.h"
|
2022-11-21 21:55:48 +08:00
|
|
|
#include "switch.h"
|
2022-11-21 21:55:54 +08:00
|
|
|
#include "tag_8021q.h"
|
2023-10-24 02:17:28 +08:00
|
|
|
#include "user.h"
|
2017-05-20 05:00:44 +08:00
|
|
|
|
2021-01-29 09:00:04 +08:00
|
|
|
/**
|
|
|
|
|
* dsa_port_notify - Notify the switching fabric of changes to a port
|
|
|
|
|
* @dp: port on which change occurred
|
|
|
|
|
* @e: event, must be of type DSA_NOTIFIER_*
|
|
|
|
|
* @v: event-specific value.
|
|
|
|
|
*
|
|
|
|
|
* Notify all switches in the DSA tree that this port's switch belongs to,
|
|
|
|
|
* including this switch itself, of an event. Allows the other switches to
|
|
|
|
|
* reconfigure themselves for cross-chip operations. Can also be used to
|
|
|
|
|
* reconfigure ports without net_devices (CPU ports, DSA links) whenever
|
|
|
|
|
* a user port's state changes.
|
|
|
|
|
*/
|
2017-11-10 06:11:01 +08:00
|
|
|
static int dsa_port_notify(const struct dsa_port *dp, unsigned long e, void *v)
|
2017-05-20 05:00:45 +08:00
|
|
|
{
|
2021-01-29 09:00:04 +08:00
|
|
|
return dsa_tree_notify(dp->ds->dst, e, v);
|
2017-05-20 05:00:45 +08:00
|
|
|
}
|
|
|
|
|
|
2022-03-16 23:08:54 +08:00
|
|
|
static void dsa_port_notify_bridge_fdb_flush(const struct dsa_port *dp, u16 vid)
|
net: dsa: flush the dynamic FDB of the software bridge when fast ageing a port
Currently, when DSA performs fast ageing on a port, 'bridge fdb' shows
us that the 'self' entries (corresponding to the hardware bridge, as
printed by dsa_slave_fdb_dump) are deleted, but the 'master' entries
(corresponding to the software bridge) aren't.
Indeed, searching through the bridge driver, neither the
brport_attr_learning handler nor the IFLA_BRPORT_LEARNING handler call
br_fdb_delete_by_port. However, br_stp_disable_port does, which is one
of the paths which DSA uses to trigger a fast ageing process anyway.
There is, however, one other very promising caller of
br_fdb_delete_by_port, and that is the bridge driver's handler of the
SWITCHDEV_FDB_FLUSH_TO_BRIDGE atomic notifier. Currently the s390/qeth
HiperSockets card driver is the only user of this.
I can't say I understand that driver's architecture or interaction with
the bridge, but it appears to not be a switchdev driver in the traditional
sense of the word. Nonetheless, the mechanism it provides is a useful
way for DSA to express the fact that it performs fast ageing too, in a
way that does not change the existing behavior for other drivers.
Cc: Alexandra Winter <wintera@linux.ibm.com>
Cc: Julian Wiedmann <jwi@linux.ibm.com>
Cc: Roopa Prabhu <roopa@nvidia.com>
Cc: Nikolay Aleksandrov <nikolay@nvidia.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 22:35:25 +08:00
|
|
|
{
|
|
|
|
|
struct net_device *brport_dev = dsa_port_to_bridge_port(dp);
|
|
|
|
|
struct switchdev_notifier_fdb_info info = {
|
2022-03-16 23:08:54 +08:00
|
|
|
.vid = vid,
|
net: dsa: flush the dynamic FDB of the software bridge when fast ageing a port
Currently, when DSA performs fast ageing on a port, 'bridge fdb' shows
us that the 'self' entries (corresponding to the hardware bridge, as
printed by dsa_slave_fdb_dump) are deleted, but the 'master' entries
(corresponding to the software bridge) aren't.
Indeed, searching through the bridge driver, neither the
brport_attr_learning handler nor the IFLA_BRPORT_LEARNING handler call
br_fdb_delete_by_port. However, br_stp_disable_port does, which is one
of the paths which DSA uses to trigger a fast ageing process anyway.
There is, however, one other very promising caller of
br_fdb_delete_by_port, and that is the bridge driver's handler of the
SWITCHDEV_FDB_FLUSH_TO_BRIDGE atomic notifier. Currently the s390/qeth
HiperSockets card driver is the only user of this.
I can't say I understand that driver's architecture or interaction with
the bridge, but it appears to not be a switchdev driver in the traditional
sense of the word. Nonetheless, the mechanism it provides is a useful
way for DSA to express the fact that it performs fast ageing too, in a
way that does not change the existing behavior for other drivers.
Cc: Alexandra Winter <wintera@linux.ibm.com>
Cc: Julian Wiedmann <jwi@linux.ibm.com>
Cc: Roopa Prabhu <roopa@nvidia.com>
Cc: Nikolay Aleksandrov <nikolay@nvidia.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 22:35:25 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/* When the port becomes standalone it has already left the bridge.
|
|
|
|
|
* Don't notify the bridge in that case.
|
|
|
|
|
*/
|
|
|
|
|
if (!brport_dev)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
call_switchdev_notifiers(SWITCHDEV_FDB_FLUSH_TO_BRIDGE,
|
|
|
|
|
brport_dev, &info.info, NULL);
|
|
|
|
|
}
|
|
|
|
|
|
net: dsa: centralize fast ageing when address learning is turned off
Currently DSA leaves it down to device drivers to fast age the FDB on a
port when address learning is disabled on it. There are 2 reasons for
doing that in the first place:
- when address learning is disabled by user space, through
IFLA_BRPORT_LEARNING or the brport_attr_learning sysfs, what user
space typically wants to achieve is to operate in a mode with no
dynamic FDB entry on that port. But if the port is already up, some
addresses might have been already learned on it, and it seems silly to
wait for 5 minutes for them to expire until something useful can be
done.
- when a port leaves a bridge and becomes standalone, DSA turns off
address learning on it. This also has the nice side effect of flushing
the dynamically learned bridge FDB entries on it, which is a good idea
because standalone ports should not have bridge FDB entries on them.
We let drivers manage fast ageing under this condition because if DSA
were to do it, it would need to track each port's learning state, and
act upon the transition, which it currently doesn't.
But there are 2 reasons why doing it is better after all:
- drivers might get it wrong and not do it (see b53_port_set_learning)
- we would like to flush the dynamic entries from the software bridge
too, and letting drivers do that would be another pain point
So track the port learning state and trigger a fast age process
automatically within DSA.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 22:35:23 +08:00
|
|
|
static void dsa_port_fast_age(const struct dsa_port *dp)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->port_fast_age)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
ds->ops->port_fast_age(ds, dp->index);
|
net: dsa: flush the dynamic FDB of the software bridge when fast ageing a port
Currently, when DSA performs fast ageing on a port, 'bridge fdb' shows
us that the 'self' entries (corresponding to the hardware bridge, as
printed by dsa_slave_fdb_dump) are deleted, but the 'master' entries
(corresponding to the software bridge) aren't.
Indeed, searching through the bridge driver, neither the
brport_attr_learning handler nor the IFLA_BRPORT_LEARNING handler call
br_fdb_delete_by_port. However, br_stp_disable_port does, which is one
of the paths which DSA uses to trigger a fast ageing process anyway.
There is, however, one other very promising caller of
br_fdb_delete_by_port, and that is the bridge driver's handler of the
SWITCHDEV_FDB_FLUSH_TO_BRIDGE atomic notifier. Currently the s390/qeth
HiperSockets card driver is the only user of this.
I can't say I understand that driver's architecture or interaction with
the bridge, but it appears to not be a switchdev driver in the traditional
sense of the word. Nonetheless, the mechanism it provides is a useful
way for DSA to express the fact that it performs fast ageing too, in a
way that does not change the existing behavior for other drivers.
Cc: Alexandra Winter <wintera@linux.ibm.com>
Cc: Julian Wiedmann <jwi@linux.ibm.com>
Cc: Roopa Prabhu <roopa@nvidia.com>
Cc: Nikolay Aleksandrov <nikolay@nvidia.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 22:35:25 +08:00
|
|
|
|
2022-03-16 23:08:54 +08:00
|
|
|
/* flush all VLANs */
|
|
|
|
|
dsa_port_notify_bridge_fdb_flush(dp, 0);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int dsa_port_vlan_fast_age(const struct dsa_port *dp, u16 vid)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->port_vlan_fast_age)
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
|
|
|
|
|
err = ds->ops->port_vlan_fast_age(ds, dp->index, vid);
|
|
|
|
|
|
|
|
|
|
if (!err)
|
|
|
|
|
dsa_port_notify_bridge_fdb_flush(dp, vid);
|
|
|
|
|
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int dsa_port_msti_fast_age(const struct dsa_port *dp, u16 msti)
|
|
|
|
|
{
|
|
|
|
|
DECLARE_BITMAP(vids, VLAN_N_VID) = { 0 };
|
|
|
|
|
int err, vid;
|
|
|
|
|
|
|
|
|
|
err = br_mst_get_info(dsa_port_bridge_dev_get(dp), msti, vids);
|
|
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
|
|
|
|
|
for_each_set_bit(vid, vids, VLAN_N_VID) {
|
|
|
|
|
err = dsa_port_vlan_fast_age(dp, vid);
|
|
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
net: dsa: centralize fast ageing when address learning is turned off
Currently DSA leaves it down to device drivers to fast age the FDB on a
port when address learning is disabled on it. There are 2 reasons for
doing that in the first place:
- when address learning is disabled by user space, through
IFLA_BRPORT_LEARNING or the brport_attr_learning sysfs, what user
space typically wants to achieve is to operate in a mode with no
dynamic FDB entry on that port. But if the port is already up, some
addresses might have been already learned on it, and it seems silly to
wait for 5 minutes for them to expire until something useful can be
done.
- when a port leaves a bridge and becomes standalone, DSA turns off
address learning on it. This also has the nice side effect of flushing
the dynamically learned bridge FDB entries on it, which is a good idea
because standalone ports should not have bridge FDB entries on them.
We let drivers manage fast ageing under this condition because if DSA
were to do it, it would need to track each port's learning state, and
act upon the transition, which it currently doesn't.
But there are 2 reasons why doing it is better after all:
- drivers might get it wrong and not do it (see b53_port_set_learning)
- we would like to flush the dynamic entries from the software bridge
too, and letting drivers do that would be another pain point
So track the port learning state and trigger a fast age process
automatically within DSA.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 22:35:23 +08:00
|
|
|
}
|
|
|
|
|
|
2021-08-09 06:56:48 +08:00
|
|
|
static bool dsa_port_can_configure_learning(struct dsa_port *dp)
|
|
|
|
|
{
|
|
|
|
|
struct switchdev_brport_flags flags = {
|
|
|
|
|
.mask = BR_LEARNING,
|
|
|
|
|
};
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->port_bridge_flags || !ds->ops->port_pre_bridge_flags)
|
|
|
|
|
return false;
|
|
|
|
|
|
|
|
|
|
err = ds->ops->port_pre_bridge_flags(ds, dp->index, flags, NULL);
|
|
|
|
|
return !err;
|
|
|
|
|
}
|
|
|
|
|
|
2023-04-02 20:37:54 +08:00
|
|
|
bool dsa_port_supports_hwtstamp(struct dsa_port *dp)
|
2022-11-12 05:10:20 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
2023-04-02 20:37:54 +08:00
|
|
|
struct ifreq ifr = {};
|
2022-11-12 05:10:20 +08:00
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->port_hwtstamp_get || !ds->ops->port_hwtstamp_set)
|
|
|
|
|
return false;
|
|
|
|
|
|
|
|
|
|
/* "See through" shim implementations of the "get" method.
|
2023-04-02 20:37:54 +08:00
|
|
|
* Since we can't cook up a complete ioctl request structure, this will
|
|
|
|
|
* fail in copy_to_user() with -EFAULT, which hopefully is enough to
|
|
|
|
|
* detect a valid implementation.
|
2022-11-12 05:10:20 +08:00
|
|
|
*/
|
2023-04-02 20:37:54 +08:00
|
|
|
err = ds->ops->port_hwtstamp_get(ds, dp->index, &ifr);
|
2022-11-12 05:10:20 +08:00
|
|
|
return err != -EOPNOTSUPP;
|
|
|
|
|
}
|
|
|
|
|
|
net: dsa: don't fast age standalone ports
DSA drives the procedure to flush dynamic FDB entries from a port based
on the change of STP state: whenever we go from a state where address
learning is enabled (LEARNING, FORWARDING) to a state where it isn't
(LISTENING, BLOCKING, DISABLED), we need to flush the existing dynamic
entries.
However, there are cases when this is not needed. Internally, when a
DSA switch interface is not under a bridge, DSA still keeps it in the
"FORWARDING" STP state. And when that interface joins a bridge, the
bridge will meticulously iterate that port through all STP states,
starting with BLOCKING and ending with FORWARDING. Because there is a
state transition from the standalone version of FORWARDING into the
temporary BLOCKING bridge port state, DSA calls the fast age procedure.
Since commit 5e38c15856e9 ("net: dsa: configure better brport flags when
ports leave the bridge"), DSA asks standalone ports to disable address
learning. Therefore, there can be no dynamic FDB entries on a standalone
port. Therefore, it does not make sense to flush dynamic FDB entries on
one.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 19:16:37 +08:00
|
|
|
int dsa_port_set_state(struct dsa_port *dp, u8 state, bool do_fast_age)
|
2017-05-20 05:00:44 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
int port = dp->index;
|
|
|
|
|
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
if (!ds->ops->port_stp_state_set)
|
|
|
|
|
return -EOPNOTSUPP;
|
2017-05-20 05:00:44 +08:00
|
|
|
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
ds->ops->port_stp_state_set(ds, port, state);
|
2017-05-20 05:00:44 +08:00
|
|
|
|
2021-08-09 06:56:48 +08:00
|
|
|
if (!dsa_port_can_configure_learning(dp) ||
|
|
|
|
|
(do_fast_age && dp->learning)) {
|
2017-05-20 05:00:44 +08:00
|
|
|
/* Fast age FDB entries or flush appropriate forwarding database
|
|
|
|
|
* for the given port, if we are moving it from Learning or
|
|
|
|
|
* Forwarding state, to Disabled or Blocking or Listening state.
|
net: dsa: don't fast age standalone ports
DSA drives the procedure to flush dynamic FDB entries from a port based
on the change of STP state: whenever we go from a state where address
learning is enabled (LEARNING, FORWARDING) to a state where it isn't
(LISTENING, BLOCKING, DISABLED), we need to flush the existing dynamic
entries.
However, there are cases when this is not needed. Internally, when a
DSA switch interface is not under a bridge, DSA still keeps it in the
"FORWARDING" STP state. And when that interface joins a bridge, the
bridge will meticulously iterate that port through all STP states,
starting with BLOCKING and ending with FORWARDING. Because there is a
state transition from the standalone version of FORWARDING into the
temporary BLOCKING bridge port state, DSA calls the fast age procedure.
Since commit 5e38c15856e9 ("net: dsa: configure better brport flags when
ports leave the bridge"), DSA asks standalone ports to disable address
learning. Therefore, there can be no dynamic FDB entries on a standalone
port. Therefore, it does not make sense to flush dynamic FDB entries on
one.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 19:16:37 +08:00
|
|
|
* Ports that were standalone before the STP state change don't
|
|
|
|
|
* need to fast age the FDB, since address learning is off in
|
|
|
|
|
* standalone mode.
|
2017-05-20 05:00:44 +08:00
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
if ((dp->stp_state == BR_STATE_LEARNING ||
|
|
|
|
|
dp->stp_state == BR_STATE_FORWARDING) &&
|
|
|
|
|
(state == BR_STATE_DISABLED ||
|
|
|
|
|
state == BR_STATE_BLOCKING ||
|
|
|
|
|
state == BR_STATE_LISTENING))
|
net: dsa: centralize fast ageing when address learning is turned off
Currently DSA leaves it down to device drivers to fast age the FDB on a
port when address learning is disabled on it. There are 2 reasons for
doing that in the first place:
- when address learning is disabled by user space, through
IFLA_BRPORT_LEARNING or the brport_attr_learning sysfs, what user
space typically wants to achieve is to operate in a mode with no
dynamic FDB entry on that port. But if the port is already up, some
addresses might have been already learned on it, and it seems silly to
wait for 5 minutes for them to expire until something useful can be
done.
- when a port leaves a bridge and becomes standalone, DSA turns off
address learning on it. This also has the nice side effect of flushing
the dynamically learned bridge FDB entries on it, which is a good idea
because standalone ports should not have bridge FDB entries on them.
We let drivers manage fast ageing under this condition because if DSA
were to do it, it would need to track each port's learning state, and
act upon the transition, which it currently doesn't.
But there are 2 reasons why doing it is better after all:
- drivers might get it wrong and not do it (see b53_port_set_learning)
- we would like to flush the dynamic entries from the software bridge
too, and letting drivers do that would be another pain point
So track the port learning state and trigger a fast age process
automatically within DSA.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 22:35:23 +08:00
|
|
|
dsa_port_fast_age(dp);
|
2017-05-20 05:00:44 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
dp->stp_state = state;
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
net: dsa: don't fast age standalone ports
DSA drives the procedure to flush dynamic FDB entries from a port based
on the change of STP state: whenever we go from a state where address
learning is enabled (LEARNING, FORWARDING) to a state where it isn't
(LISTENING, BLOCKING, DISABLED), we need to flush the existing dynamic
entries.
However, there are cases when this is not needed. Internally, when a
DSA switch interface is not under a bridge, DSA still keeps it in the
"FORWARDING" STP state. And when that interface joins a bridge, the
bridge will meticulously iterate that port through all STP states,
starting with BLOCKING and ending with FORWARDING. Because there is a
state transition from the standalone version of FORWARDING into the
temporary BLOCKING bridge port state, DSA calls the fast age procedure.
Since commit 5e38c15856e9 ("net: dsa: configure better brport flags when
ports leave the bridge"), DSA asks standalone ports to disable address
learning. Therefore, there can be no dynamic FDB entries on a standalone
port. Therefore, it does not make sense to flush dynamic FDB entries on
one.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 19:16:37 +08:00
|
|
|
static void dsa_port_set_state_now(struct dsa_port *dp, u8 state,
|
|
|
|
|
bool do_fast_age)
|
2017-05-20 05:00:44 +08:00
|
|
|
{
|
2022-08-17 04:14:45 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
2017-05-20 05:00:44 +08:00
|
|
|
int err;
|
|
|
|
|
|
net: dsa: don't fast age standalone ports
DSA drives the procedure to flush dynamic FDB entries from a port based
on the change of STP state: whenever we go from a state where address
learning is enabled (LEARNING, FORWARDING) to a state where it isn't
(LISTENING, BLOCKING, DISABLED), we need to flush the existing dynamic
entries.
However, there are cases when this is not needed. Internally, when a
DSA switch interface is not under a bridge, DSA still keeps it in the
"FORWARDING" STP state. And when that interface joins a bridge, the
bridge will meticulously iterate that port through all STP states,
starting with BLOCKING and ending with FORWARDING. Because there is a
state transition from the standalone version of FORWARDING into the
temporary BLOCKING bridge port state, DSA calls the fast age procedure.
Since commit 5e38c15856e9 ("net: dsa: configure better brport flags when
ports leave the bridge"), DSA asks standalone ports to disable address
learning. Therefore, there can be no dynamic FDB entries on a standalone
port. Therefore, it does not make sense to flush dynamic FDB entries on
one.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 19:16:37 +08:00
|
|
|
err = dsa_port_set_state(dp, state, do_fast_age);
|
2022-08-17 04:14:45 +08:00
|
|
|
if (err && err != -EOPNOTSUPP) {
|
|
|
|
|
dev_err(ds->dev, "port %d failed to set STP state %u: %pe\n",
|
|
|
|
|
dp->index, state, ERR_PTR(err));
|
|
|
|
|
}
|
2017-05-20 05:00:44 +08:00
|
|
|
}
|
2017-05-20 05:00:45 +08:00
|
|
|
|
2022-03-16 23:08:54 +08:00
|
|
|
int dsa_port_set_mst_state(struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_mst_state *state,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
u8 prev_state;
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->port_mst_state_set)
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
|
|
|
|
|
err = br_mst_get_state(dsa_port_to_bridge_port(dp), state->msti,
|
|
|
|
|
&prev_state);
|
|
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
|
|
|
|
|
err = ds->ops->port_mst_state_set(ds, dp->index, state);
|
|
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
|
|
|
|
|
if (!(dp->learning &&
|
|
|
|
|
(prev_state == BR_STATE_LEARNING ||
|
|
|
|
|
prev_state == BR_STATE_FORWARDING) &&
|
|
|
|
|
(state->state == BR_STATE_DISABLED ||
|
|
|
|
|
state->state == BR_STATE_BLOCKING ||
|
|
|
|
|
state->state == BR_STATE_LISTENING)))
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
err = dsa_port_msti_fast_age(dp, state->msti);
|
|
|
|
|
if (err)
|
|
|
|
|
NL_SET_ERR_MSG_MOD(extack,
|
|
|
|
|
"Unable to flush associated VLANs");
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-03 23:01:46 +08:00
|
|
|
int dsa_port_enable_rt(struct dsa_port *dp, struct phy_device *phy)
|
2017-09-23 07:01:56 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
int port = dp->index;
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
if (ds->ops->port_enable) {
|
|
|
|
|
err = ds->ops->port_enable(ds, port, phy);
|
|
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
if (!dp->bridge)
|
net: dsa: don't fast age standalone ports
DSA drives the procedure to flush dynamic FDB entries from a port based
on the change of STP state: whenever we go from a state where address
learning is enabled (LEARNING, FORWARDING) to a state where it isn't
(LISTENING, BLOCKING, DISABLED), we need to flush the existing dynamic
entries.
However, there are cases when this is not needed. Internally, when a
DSA switch interface is not under a bridge, DSA still keeps it in the
"FORWARDING" STP state. And when that interface joins a bridge, the
bridge will meticulously iterate that port through all STP states,
starting with BLOCKING and ending with FORWARDING. Because there is a
state transition from the standalone version of FORWARDING into the
temporary BLOCKING bridge port state, DSA calls the fast age procedure.
Since commit 5e38c15856e9 ("net: dsa: configure better brport flags when
ports leave the bridge"), DSA asks standalone ports to disable address
learning. Therefore, there can be no dynamic FDB entries on a standalone
port. Therefore, it does not make sense to flush dynamic FDB entries on
one.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 19:16:37 +08:00
|
|
|
dsa_port_set_state_now(dp, BR_STATE_FORWARDING, false);
|
2017-09-23 07:01:56 +08:00
|
|
|
|
2020-03-03 23:01:46 +08:00
|
|
|
if (dp->pl)
|
|
|
|
|
phylink_start(dp->pl);
|
|
|
|
|
|
2017-09-23 07:01:56 +08:00
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-03 23:01:46 +08:00
|
|
|
int dsa_port_enable(struct dsa_port *dp, struct phy_device *phy)
|
|
|
|
|
{
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
rtnl_lock();
|
|
|
|
|
err = dsa_port_enable_rt(dp, phy);
|
|
|
|
|
rtnl_unlock();
|
|
|
|
|
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void dsa_port_disable_rt(struct dsa_port *dp)
|
2017-09-23 07:01:56 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
int port = dp->index;
|
|
|
|
|
|
2020-03-03 23:01:46 +08:00
|
|
|
if (dp->pl)
|
|
|
|
|
phylink_stop(dp->pl);
|
|
|
|
|
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
if (!dp->bridge)
|
net: dsa: don't fast age standalone ports
DSA drives the procedure to flush dynamic FDB entries from a port based
on the change of STP state: whenever we go from a state where address
learning is enabled (LEARNING, FORWARDING) to a state where it isn't
(LISTENING, BLOCKING, DISABLED), we need to flush the existing dynamic
entries.
However, there are cases when this is not needed. Internally, when a
DSA switch interface is not under a bridge, DSA still keeps it in the
"FORWARDING" STP state. And when that interface joins a bridge, the
bridge will meticulously iterate that port through all STP states,
starting with BLOCKING and ending with FORWARDING. Because there is a
state transition from the standalone version of FORWARDING into the
temporary BLOCKING bridge port state, DSA calls the fast age procedure.
Since commit 5e38c15856e9 ("net: dsa: configure better brport flags when
ports leave the bridge"), DSA asks standalone ports to disable address
learning. Therefore, there can be no dynamic FDB entries on a standalone
port. Therefore, it does not make sense to flush dynamic FDB entries on
one.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 19:16:37 +08:00
|
|
|
dsa_port_set_state_now(dp, BR_STATE_DISABLED, false);
|
2017-09-23 07:01:56 +08:00
|
|
|
|
|
|
|
|
if (ds->ops->port_disable)
|
2019-02-25 03:44:43 +08:00
|
|
|
ds->ops->port_disable(ds, port);
|
2017-09-23 07:01:56 +08:00
|
|
|
}
|
|
|
|
|
|
2020-03-03 23:01:46 +08:00
|
|
|
void dsa_port_disable(struct dsa_port *dp)
|
|
|
|
|
{
|
|
|
|
|
rtnl_lock();
|
|
|
|
|
dsa_port_disable_rt(dp);
|
|
|
|
|
rtnl_unlock();
|
|
|
|
|
}
|
|
|
|
|
|
2022-04-15 23:46:21 +08:00
|
|
|
static void dsa_port_reset_vlan_filtering(struct dsa_port *dp,
|
|
|
|
|
struct dsa_bridge bridge)
|
|
|
|
|
{
|
|
|
|
|
struct netlink_ext_ack extack = {0};
|
|
|
|
|
bool change_vlan_filtering = false;
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
2022-07-15 23:16:59 +08:00
|
|
|
struct dsa_port *other_dp;
|
2022-04-15 23:46:21 +08:00
|
|
|
bool vlan_filtering;
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
if (ds->needs_standalone_vlan_filtering &&
|
|
|
|
|
!br_vlan_enabled(bridge.dev)) {
|
|
|
|
|
change_vlan_filtering = true;
|
|
|
|
|
vlan_filtering = true;
|
|
|
|
|
} else if (!ds->needs_standalone_vlan_filtering &&
|
|
|
|
|
br_vlan_enabled(bridge.dev)) {
|
|
|
|
|
change_vlan_filtering = true;
|
|
|
|
|
vlan_filtering = false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* If the bridge was vlan_filtering, the bridge core doesn't trigger an
|
2023-10-24 02:17:28 +08:00
|
|
|
* event for changing vlan_filtering setting upon user ports leaving
|
2022-04-15 23:46:21 +08:00
|
|
|
* it. That is a good thing, because that lets us handle it and also
|
|
|
|
|
* handle the case where the switch's vlan_filtering setting is global
|
|
|
|
|
* (not per port). When that happens, the correct moment to trigger the
|
|
|
|
|
* vlan_filtering callback is only when the last port leaves the last
|
|
|
|
|
* VLAN-aware bridge.
|
|
|
|
|
*/
|
|
|
|
|
if (change_vlan_filtering && ds->vlan_filtering_is_global) {
|
2022-07-15 23:16:59 +08:00
|
|
|
dsa_switch_for_each_port(other_dp, ds) {
|
|
|
|
|
struct net_device *br = dsa_port_bridge_dev_get(other_dp);
|
2022-04-15 23:46:21 +08:00
|
|
|
|
|
|
|
|
if (br && br_vlan_enabled(br)) {
|
|
|
|
|
change_vlan_filtering = false;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (!change_vlan_filtering)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
err = dsa_port_vlan_filtering(dp, vlan_filtering, &extack);
|
|
|
|
|
if (extack._msg) {
|
|
|
|
|
dev_err(ds->dev, "port %d: %s\n", dp->index,
|
|
|
|
|
extack._msg);
|
|
|
|
|
}
|
|
|
|
|
if (err && err != -EOPNOTSUPP) {
|
|
|
|
|
dev_err(ds->dev,
|
|
|
|
|
"port %d failed to reset VLAN filtering to %d: %pe\n",
|
|
|
|
|
dp->index, vlan_filtering, ERR_PTR(err));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2021-03-23 07:51:49 +08:00
|
|
|
static int dsa_port_inherit_brport_flags(struct dsa_port *dp,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
net: dsa: configure better brport flags when ports leave the bridge
For a DSA switch port operating in standalone mode, address learning
doesn't make much sense since that is a bridge function. In fact,
address learning even breaks setups such as this one:
+---------------------------------------------+
| |
| +-------------------+ |
| | br0 | send receive |
| +--------+-+--------+ +--------+ +--------+ |
| | | | | | | | | |
| | swp0 | | swp1 | | swp2 | | swp3 | |
| | | | | | | | | |
+-+--------+-+--------+-+--------+-+--------+-+
| ^ | ^
| | | |
| +-----------+ |
| |
+--------------------------------+
because if the switch has a single FDB (can offload a single bridge)
then source address learning on swp3 can "steal" the source MAC address
of swp2 from br0's FDB, because learning frames coming from swp2 will be
done twice: first on the swp1 ingress port, second on the swp3 ingress
port. So the hardware FDB will become out of sync with the software
bridge, and when swp2 tries to send one more packet towards swp1, the
ASIC will attempt to short-circuit the forwarding path and send it
directly to swp3 (since that's the last port it learned that address on),
which it obviously can't, because swp3 operates in standalone mode.
So DSA drivers operating in standalone mode should still configure a
list of bridge port flags even when they are standalone. Currently DSA
attempts to call dsa_port_bridge_flags with 0, which disables egress
flooding of unknown unicast and multicast, something which doesn't make
much sense. For the switches that implement .port_egress_floods - b53
and mv88e6xxx, it probably doesn't matter too much either, since they
can possibly inject traffic from the CPU into a standalone port,
regardless of MAC DA, even if egress flooding is turned off for that
port, but certainly not all DSA switches can do that - sja1105, for
example, can't. So it makes sense to use a better common default there,
such as "flood everything".
It should also be noted that what DSA calls "dsa_port_bridge_flags()"
is a degenerate name for just calling .port_egress_floods(), since
nothing else is implemented - not learning, in particular. But disabling
address learning, something that this driver is also coding up for, will
be supported by individual drivers once .port_egress_floods is replaced
with a more generic .port_bridge_flags.
Previous attempts to code up this logic have been in the common bridge
layer, but as pointed out by Ido Schimmel, there are corner cases that
are missed when doing that:
https://patchwork.kernel.org/project/netdevbpf/patch/20210209151936.97382-5-olteanv@gmail.com/
So, at least for now, let's leave DSA in charge of setting port flags
before and after the bridge join and leave.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:54 +08:00
|
|
|
{
|
2021-03-23 07:51:49 +08:00
|
|
|
const unsigned long mask = BR_LEARNING | BR_FLOOD | BR_MCAST_FLOOD |
|
2022-02-23 18:16:48 +08:00
|
|
|
BR_BCAST_FLOOD | BR_PORT_LOCKED;
|
2021-03-23 07:51:49 +08:00
|
|
|
struct net_device *brport_dev = dsa_port_to_bridge_port(dp);
|
|
|
|
|
int flag, err;
|
net: dsa: configure better brport flags when ports leave the bridge
For a DSA switch port operating in standalone mode, address learning
doesn't make much sense since that is a bridge function. In fact,
address learning even breaks setups such as this one:
+---------------------------------------------+
| |
| +-------------------+ |
| | br0 | send receive |
| +--------+-+--------+ +--------+ +--------+ |
| | | | | | | | | |
| | swp0 | | swp1 | | swp2 | | swp3 | |
| | | | | | | | | |
+-+--------+-+--------+-+--------+-+--------+-+
| ^ | ^
| | | |
| +-----------+ |
| |
+--------------------------------+
because if the switch has a single FDB (can offload a single bridge)
then source address learning on swp3 can "steal" the source MAC address
of swp2 from br0's FDB, because learning frames coming from swp2 will be
done twice: first on the swp1 ingress port, second on the swp3 ingress
port. So the hardware FDB will become out of sync with the software
bridge, and when swp2 tries to send one more packet towards swp1, the
ASIC will attempt to short-circuit the forwarding path and send it
directly to swp3 (since that's the last port it learned that address on),
which it obviously can't, because swp3 operates in standalone mode.
So DSA drivers operating in standalone mode should still configure a
list of bridge port flags even when they are standalone. Currently DSA
attempts to call dsa_port_bridge_flags with 0, which disables egress
flooding of unknown unicast and multicast, something which doesn't make
much sense. For the switches that implement .port_egress_floods - b53
and mv88e6xxx, it probably doesn't matter too much either, since they
can possibly inject traffic from the CPU into a standalone port,
regardless of MAC DA, even if egress flooding is turned off for that
port, but certainly not all DSA switches can do that - sja1105, for
example, can't. So it makes sense to use a better common default there,
such as "flood everything".
It should also be noted that what DSA calls "dsa_port_bridge_flags()"
is a degenerate name for just calling .port_egress_floods(), since
nothing else is implemented - not learning, in particular. But disabling
address learning, something that this driver is also coding up for, will
be supported by individual drivers once .port_egress_floods is replaced
with a more generic .port_bridge_flags.
Previous attempts to code up this logic have been in the common bridge
layer, but as pointed out by Ido Schimmel, there are corner cases that
are missed when doing that:
https://patchwork.kernel.org/project/netdevbpf/patch/20210209151936.97382-5-olteanv@gmail.com/
So, at least for now, let's leave DSA in charge of setting port flags
before and after the bridge join and leave.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:54 +08:00
|
|
|
|
2021-03-23 07:51:49 +08:00
|
|
|
for_each_set_bit(flag, &mask, 32) {
|
|
|
|
|
struct switchdev_brport_flags flags = {0};
|
net: dsa: configure better brport flags when ports leave the bridge
For a DSA switch port operating in standalone mode, address learning
doesn't make much sense since that is a bridge function. In fact,
address learning even breaks setups such as this one:
+---------------------------------------------+
| |
| +-------------------+ |
| | br0 | send receive |
| +--------+-+--------+ +--------+ +--------+ |
| | | | | | | | | |
| | swp0 | | swp1 | | swp2 | | swp3 | |
| | | | | | | | | |
+-+--------+-+--------+-+--------+-+--------+-+
| ^ | ^
| | | |
| +-----------+ |
| |
+--------------------------------+
because if the switch has a single FDB (can offload a single bridge)
then source address learning on swp3 can "steal" the source MAC address
of swp2 from br0's FDB, because learning frames coming from swp2 will be
done twice: first on the swp1 ingress port, second on the swp3 ingress
port. So the hardware FDB will become out of sync with the software
bridge, and when swp2 tries to send one more packet towards swp1, the
ASIC will attempt to short-circuit the forwarding path and send it
directly to swp3 (since that's the last port it learned that address on),
which it obviously can't, because swp3 operates in standalone mode.
So DSA drivers operating in standalone mode should still configure a
list of bridge port flags even when they are standalone. Currently DSA
attempts to call dsa_port_bridge_flags with 0, which disables egress
flooding of unknown unicast and multicast, something which doesn't make
much sense. For the switches that implement .port_egress_floods - b53
and mv88e6xxx, it probably doesn't matter too much either, since they
can possibly inject traffic from the CPU into a standalone port,
regardless of MAC DA, even if egress flooding is turned off for that
port, but certainly not all DSA switches can do that - sja1105, for
example, can't. So it makes sense to use a better common default there,
such as "flood everything".
It should also be noted that what DSA calls "dsa_port_bridge_flags()"
is a degenerate name for just calling .port_egress_floods(), since
nothing else is implemented - not learning, in particular. But disabling
address learning, something that this driver is also coding up for, will
be supported by individual drivers once .port_egress_floods is replaced
with a more generic .port_bridge_flags.
Previous attempts to code up this logic have been in the common bridge
layer, but as pointed out by Ido Schimmel, there are corner cases that
are missed when doing that:
https://patchwork.kernel.org/project/netdevbpf/patch/20210209151936.97382-5-olteanv@gmail.com/
So, at least for now, let's leave DSA in charge of setting port flags
before and after the bridge join and leave.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:54 +08:00
|
|
|
|
2021-03-23 07:51:49 +08:00
|
|
|
flags.mask = BIT(flag);
|
net: switchdev: pass flags and mask to both {PRE_,}BRIDGE_FLAGS attributes
This switchdev attribute offers a counterproductive API for a driver
writer, because although br_switchdev_set_port_flag gets passed a
"flags" and a "mask", those are passed piecemeal to the driver, so while
the PRE_BRIDGE_FLAGS listener knows what changed because it has the
"mask", the BRIDGE_FLAGS listener doesn't, because it only has the final
value. But certain drivers can offload only certain combinations of
settings, like for example they cannot change unicast flooding
independently of multicast flooding - they must be both on or both off.
The way the information is passed to switchdev makes drivers not
expressive enough, and unable to reject this request ahead of time, in
the PRE_BRIDGE_FLAGS notifier, so they are forced to reject it during
the deferred BRIDGE_FLAGS attribute, where the rejection is currently
ignored.
This patch also changes drivers to make use of the "mask" field for edge
detection when possible.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Grygorii Strashko <grygorii.strashko@ti.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:55 +08:00
|
|
|
|
2021-03-23 07:51:49 +08:00
|
|
|
if (br_port_flag_is_set(brport_dev, BIT(flag)))
|
|
|
|
|
flags.val = BIT(flag);
|
net: dsa: configure better brport flags when ports leave the bridge
For a DSA switch port operating in standalone mode, address learning
doesn't make much sense since that is a bridge function. In fact,
address learning even breaks setups such as this one:
+---------------------------------------------+
| |
| +-------------------+ |
| | br0 | send receive |
| +--------+-+--------+ +--------+ +--------+ |
| | | | | | | | | |
| | swp0 | | swp1 | | swp2 | | swp3 | |
| | | | | | | | | |
+-+--------+-+--------+-+--------+-+--------+-+
| ^ | ^
| | | |
| +-----------+ |
| |
+--------------------------------+
because if the switch has a single FDB (can offload a single bridge)
then source address learning on swp3 can "steal" the source MAC address
of swp2 from br0's FDB, because learning frames coming from swp2 will be
done twice: first on the swp1 ingress port, second on the swp3 ingress
port. So the hardware FDB will become out of sync with the software
bridge, and when swp2 tries to send one more packet towards swp1, the
ASIC will attempt to short-circuit the forwarding path and send it
directly to swp3 (since that's the last port it learned that address on),
which it obviously can't, because swp3 operates in standalone mode.
So DSA drivers operating in standalone mode should still configure a
list of bridge port flags even when they are standalone. Currently DSA
attempts to call dsa_port_bridge_flags with 0, which disables egress
flooding of unknown unicast and multicast, something which doesn't make
much sense. For the switches that implement .port_egress_floods - b53
and mv88e6xxx, it probably doesn't matter too much either, since they
can possibly inject traffic from the CPU into a standalone port,
regardless of MAC DA, even if egress flooding is turned off for that
port, but certainly not all DSA switches can do that - sja1105, for
example, can't. So it makes sense to use a better common default there,
such as "flood everything".
It should also be noted that what DSA calls "dsa_port_bridge_flags()"
is a degenerate name for just calling .port_egress_floods(), since
nothing else is implemented - not learning, in particular. But disabling
address learning, something that this driver is also coding up for, will
be supported by individual drivers once .port_egress_floods is replaced
with a more generic .port_bridge_flags.
Previous attempts to code up this logic have been in the common bridge
layer, but as pointed out by Ido Schimmel, there are corner cases that
are missed when doing that:
https://patchwork.kernel.org/project/netdevbpf/patch/20210209151936.97382-5-olteanv@gmail.com/
So, at least for now, let's leave DSA in charge of setting port flags
before and after the bridge join and leave.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:54 +08:00
|
|
|
|
2021-03-23 07:51:49 +08:00
|
|
|
err = dsa_port_bridge_flags(dp, flags, extack);
|
|
|
|
|
if (err && err != -EOPNOTSUPP)
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void dsa_port_clear_brport_flags(struct dsa_port *dp)
|
|
|
|
|
{
|
|
|
|
|
const unsigned long val = BR_FLOOD | BR_MCAST_FLOOD | BR_BCAST_FLOOD;
|
|
|
|
|
const unsigned long mask = BR_LEARNING | BR_FLOOD | BR_MCAST_FLOOD |
|
2022-02-23 18:16:48 +08:00
|
|
|
BR_BCAST_FLOOD | BR_PORT_LOCKED;
|
2021-03-23 07:51:49 +08:00
|
|
|
int flag, err;
|
|
|
|
|
|
|
|
|
|
for_each_set_bit(flag, &mask, 32) {
|
|
|
|
|
struct switchdev_brport_flags flags = {0};
|
|
|
|
|
|
|
|
|
|
flags.mask = BIT(flag);
|
|
|
|
|
flags.val = val & BIT(flag);
|
|
|
|
|
|
|
|
|
|
err = dsa_port_bridge_flags(dp, flags, NULL);
|
|
|
|
|
if (err && err != -EOPNOTSUPP)
|
|
|
|
|
dev_err(dp->ds->dev,
|
|
|
|
|
"failed to clear bridge port flag %lu: %pe\n",
|
|
|
|
|
flags.val, ERR_PTR(err));
|
net: dsa: configure better brport flags when ports leave the bridge
For a DSA switch port operating in standalone mode, address learning
doesn't make much sense since that is a bridge function. In fact,
address learning even breaks setups such as this one:
+---------------------------------------------+
| |
| +-------------------+ |
| | br0 | send receive |
| +--------+-+--------+ +--------+ +--------+ |
| | | | | | | | | |
| | swp0 | | swp1 | | swp2 | | swp3 | |
| | | | | | | | | |
+-+--------+-+--------+-+--------+-+--------+-+
| ^ | ^
| | | |
| +-----------+ |
| |
+--------------------------------+
because if the switch has a single FDB (can offload a single bridge)
then source address learning on swp3 can "steal" the source MAC address
of swp2 from br0's FDB, because learning frames coming from swp2 will be
done twice: first on the swp1 ingress port, second on the swp3 ingress
port. So the hardware FDB will become out of sync with the software
bridge, and when swp2 tries to send one more packet towards swp1, the
ASIC will attempt to short-circuit the forwarding path and send it
directly to swp3 (since that's the last port it learned that address on),
which it obviously can't, because swp3 operates in standalone mode.
So DSA drivers operating in standalone mode should still configure a
list of bridge port flags even when they are standalone. Currently DSA
attempts to call dsa_port_bridge_flags with 0, which disables egress
flooding of unknown unicast and multicast, something which doesn't make
much sense. For the switches that implement .port_egress_floods - b53
and mv88e6xxx, it probably doesn't matter too much either, since they
can possibly inject traffic from the CPU into a standalone port,
regardless of MAC DA, even if egress flooding is turned off for that
port, but certainly not all DSA switches can do that - sja1105, for
example, can't. So it makes sense to use a better common default there,
such as "flood everything".
It should also be noted that what DSA calls "dsa_port_bridge_flags()"
is a degenerate name for just calling .port_egress_floods(), since
nothing else is implemented - not learning, in particular. But disabling
address learning, something that this driver is also coding up for, will
be supported by individual drivers once .port_egress_floods is replaced
with a more generic .port_bridge_flags.
Previous attempts to code up this logic have been in the common bridge
layer, but as pointed out by Ido Schimmel, there are corner cases that
are missed when doing that:
https://patchwork.kernel.org/project/netdevbpf/patch/20210209151936.97382-5-olteanv@gmail.com/
So, at least for now, let's leave DSA in charge of setting port flags
before and after the bridge join and leave.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:54 +08:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
net: bridge: move the switchdev object replay helpers to "push" mode
Starting with commit 4f2673b3a2b6 ("net: bridge: add helper to replay
port and host-joined mdb entries"), DSA has introduced some bridge
helpers that replay switchdev events (FDB/MDB/VLAN additions and
deletions) that can be lost by the switchdev drivers in a variety of
circumstances:
- an IP multicast group was host-joined on the bridge itself before any
switchdev port joined the bridge, leading to the host MDB entries
missing in the hardware database.
- during the bridge creation process, the MAC address of the bridge was
added to the FDB as an entry pointing towards the bridge device
itself, but with no switchdev ports being part of the bridge yet, this
local FDB entry would remain unknown to the switchdev hardware
database.
- a VLAN/FDB/MDB was added to a bridge port that is a LAG interface,
before any switchdev port joined that LAG, leading to the hardware
database missing those entries.
- a switchdev port left a LAG that is a bridge port, while the LAG
remained part of the bridge, and all FDB/MDB/VLAN entries remained
installed in the hardware database of the switchdev port.
Also, since commit 0d2cfbd41c4a ("net: bridge: ignore switchdev events
for LAG ports which didn't request replay"), DSA introduced a method,
based on a const void *ctx, to ensure that two switchdev ports under the
same LAG that is a bridge port do not see the same MDB/VLAN entry being
replayed twice by the bridge, once for every bridge port that joins the
LAG.
With so many ordering corner cases being possible, it seems unreasonable
to expect a switchdev driver writer to get it right from the first try.
Therefore, now that DSA has experimented with the bridge replay helpers
for a little bit, we can move the code to the bridge driver where it is
more readily available to all switchdev drivers.
To convert the switchdev object replay helpers from "pull mode" (where
the driver asks for them) to a "push mode" (where the bridge offers them
automatically), the biggest problem is that the bridge needs to be aware
when a switchdev port joins and leaves, even when the switchdev is only
indirectly a bridge port (for example when the bridge port is a LAG
upper of the switchdev).
Luckily, we already have a hook for that, in the form of the newly
introduced switchdev_bridge_port_offload() and
switchdev_bridge_port_unoffload() calls. These offer a natural place for
hooking the object addition and deletion replays.
Extend the above 2 functions with:
- pointers to the switchdev atomic notifier (for FDB replays) and the
blocking notifier (for MDB and VLAN replays).
- the "const void *ctx" argument required for drivers to be able to
disambiguate between which port is targeted, when multiple ports are
lowers of the same LAG that is a bridge port. Most of the drivers pass
NULL to this argument, except the ones that support LAG offload and have
the proper context check already in place in the switchdev blocking
notifier handler.
Also unexport the replay helpers, since nobody except the bridge calls
them directly now.
Note that:
(a) we abuse the terminology slightly, because FDB entries are not
"switchdev objects", but we count them as objects nonetheless.
With no direct way to prove it, I think they are not modeled as
switchdev objects because those can only be installed by the bridge
to the hardware (as opposed to FDB entries which can be propagated
in the other direction too). This is merely an abuse of terms, FDB
entries are replayed too, despite not being objects.
(b) the bridge does not attempt to sync port attributes to newly joined
ports, just the countable stuff (the objects). The reason for this
is simple: no universal and symmetric way to sync and unsync them is
known. For example, VLAN filtering: what to do on unsync, disable or
leave it enabled? Similarly, STP state, ageing timer, etc etc. What
a switchdev port does when it becomes standalone again is not really
up to the bridge's competence, and the driver should deal with it.
On the other hand, replaying deletions of switchdev objects can be
seen a matter of cleanup and therefore be treated by the bridge,
hence this patch.
We make the replay helpers opt-in for drivers, because they might not
bring immediate benefits for them:
- nbp_vlan_init() is called _after_ netdev_master_upper_dev_link(),
so br_vlan_replay() should not do anything for the new drivers on
which we call it. The existing drivers where there was even a slight
possibility for there to exist a VLAN on a bridge port before they
join it are already guarded against this: mlxsw and prestera deny
joining LAG interfaces that are members of a bridge.
- br_fdb_replay() should now notify of local FDB entries, but I patched
all drivers except DSA to ignore these new entries in commit
2c4eca3ef716 ("net: bridge: switchdev: include local flag in FDB
notifications"). Driver authors can lift this restriction as they
wish, and when they do, they can also opt into the FDB replay
functionality.
- br_mdb_replay() should fix a real issue which is described in commit
4f2673b3a2b6 ("net: bridge: add helper to replay port and host-joined
mdb entries"). However most drivers do not offload the
SWITCHDEV_OBJ_ID_HOST_MDB to see this issue: only cpsw and am65_cpsw
offload this switchdev object, and I don't completely understand the
way in which they offload this switchdev object anyway. So I'll leave
it up to these drivers' respective maintainers to opt into
br_mdb_replay().
So most of the drivers pass NULL notifier blocks for the replay helpers,
except:
- dpaa2-switch which was already acked/regression-tested with the
helpers enabled (and there isn't much of a downside in having them)
- ocelot which already had replay logic in "pull" mode
- DSA which already had replay logic in "pull" mode
An important observation is that the drivers which don't currently
request bridge event replays don't even have the
switchdev_bridge_port_{offload,unoffload} calls placed in proper places
right now. This was done to avoid unnecessary rework for drivers which
might never even add support for this. For driver writers who wish to
add replay support, this can be used as a tentative placement guide:
https://patchwork.kernel.org/project/netdevbpf/patch/20210720134655.892334-11-vladimir.oltean@nxp.com/
Cc: Vadym Kochan <vkochan@marvell.com>
Cc: Taras Chornyi <tchornyi@marvell.com>
Cc: Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Cc: Claudiu Manoil <claudiu.manoil@nxp.com>
Cc: Alexandre Belloni <alexandre.belloni@bootlin.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 00:24:03 +08:00
|
|
|
static int dsa_port_switchdev_sync_attrs(struct dsa_port *dp,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
2021-03-23 07:51:49 +08:00
|
|
|
{
|
2021-03-23 07:51:50 +08:00
|
|
|
struct net_device *brport_dev = dsa_port_to_bridge_port(dp);
|
2021-12-07 00:57:52 +08:00
|
|
|
struct net_device *br = dsa_port_bridge_dev_get(dp);
|
2021-03-23 07:51:49 +08:00
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
err = dsa_port_inherit_brport_flags(dp, extack);
|
|
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
|
net: dsa: don't fast age standalone ports
DSA drives the procedure to flush dynamic FDB entries from a port based
on the change of STP state: whenever we go from a state where address
learning is enabled (LEARNING, FORWARDING) to a state where it isn't
(LISTENING, BLOCKING, DISABLED), we need to flush the existing dynamic
entries.
However, there are cases when this is not needed. Internally, when a
DSA switch interface is not under a bridge, DSA still keeps it in the
"FORWARDING" STP state. And when that interface joins a bridge, the
bridge will meticulously iterate that port through all STP states,
starting with BLOCKING and ending with FORWARDING. Because there is a
state transition from the standalone version of FORWARDING into the
temporary BLOCKING bridge port state, DSA calls the fast age procedure.
Since commit 5e38c15856e9 ("net: dsa: configure better brport flags when
ports leave the bridge"), DSA asks standalone ports to disable address
learning. Therefore, there can be no dynamic FDB entries on a standalone
port. Therefore, it does not make sense to flush dynamic FDB entries on
one.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 19:16:37 +08:00
|
|
|
err = dsa_port_set_state(dp, br_port_get_stp_state(brport_dev), false);
|
2021-03-23 07:51:50 +08:00
|
|
|
if (err && err != -EOPNOTSUPP)
|
|
|
|
|
return err;
|
|
|
|
|
|
|
|
|
|
err = dsa_port_vlan_filtering(dp, br_vlan_enabled(br), extack);
|
|
|
|
|
if (err && err != -EOPNOTSUPP)
|
|
|
|
|
return err;
|
|
|
|
|
|
|
|
|
|
err = dsa_port_ageing_time(dp, br_get_ageing_time(br));
|
|
|
|
|
if (err && err != -EOPNOTSUPP)
|
|
|
|
|
return err;
|
|
|
|
|
|
2021-06-27 19:54:29 +08:00
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2022-04-15 23:46:21 +08:00
|
|
|
static void dsa_port_switchdev_unsync_attrs(struct dsa_port *dp,
|
|
|
|
|
struct dsa_bridge bridge)
|
2021-03-23 07:51:49 +08:00
|
|
|
{
|
|
|
|
|
/* Configure the port for standalone mode (no address learning,
|
|
|
|
|
* flood everything).
|
|
|
|
|
* The bridge only emits SWITCHDEV_ATTR_ID_PORT_BRIDGE_FLAGS events
|
|
|
|
|
* when the user requests it through netlink or sysfs, but not
|
|
|
|
|
* automatically at port join or leave, so we need to handle resetting
|
|
|
|
|
* the brport flags ourselves. But we even prefer it that way, because
|
|
|
|
|
* otherwise, some setups might never get the notification they need,
|
|
|
|
|
* for example, when a port leaves a LAG that offloads the bridge,
|
|
|
|
|
* it becomes standalone, but as far as the bridge is concerned, no
|
|
|
|
|
* port ever left.
|
|
|
|
|
*/
|
|
|
|
|
dsa_port_clear_brport_flags(dp);
|
|
|
|
|
|
|
|
|
|
/* Port left the bridge, put in BR_STATE_DISABLED by the bridge layer,
|
|
|
|
|
* so allow it to be in BR_STATE_FORWARDING to be kept functional
|
|
|
|
|
*/
|
net: dsa: don't fast age standalone ports
DSA drives the procedure to flush dynamic FDB entries from a port based
on the change of STP state: whenever we go from a state where address
learning is enabled (LEARNING, FORWARDING) to a state where it isn't
(LISTENING, BLOCKING, DISABLED), we need to flush the existing dynamic
entries.
However, there are cases when this is not needed. Internally, when a
DSA switch interface is not under a bridge, DSA still keeps it in the
"FORWARDING" STP state. And when that interface joins a bridge, the
bridge will meticulously iterate that port through all STP states,
starting with BLOCKING and ending with FORWARDING. Because there is a
state transition from the standalone version of FORWARDING into the
temporary BLOCKING bridge port state, DSA calls the fast age procedure.
Since commit 5e38c15856e9 ("net: dsa: configure better brport flags when
ports leave the bridge"), DSA asks standalone ports to disable address
learning. Therefore, there can be no dynamic FDB entries on a standalone
port. Therefore, it does not make sense to flush dynamic FDB entries on
one.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 19:16:37 +08:00
|
|
|
dsa_port_set_state_now(dp, BR_STATE_FORWARDING, true);
|
2021-03-23 07:51:50 +08:00
|
|
|
|
2022-04-15 23:46:21 +08:00
|
|
|
dsa_port_reset_vlan_filtering(dp, bridge);
|
2021-03-23 07:51:50 +08:00
|
|
|
|
|
|
|
|
/* Ageing time may be global to the switch chip, so don't change it
|
|
|
|
|
* here because we have no good reason (or value) to change it to.
|
|
|
|
|
*/
|
2021-03-23 07:51:49 +08:00
|
|
|
}
|
|
|
|
|
|
2021-12-07 00:57:48 +08:00
|
|
|
static int dsa_port_bridge_create(struct dsa_port *dp,
|
|
|
|
|
struct net_device *br,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
struct dsa_bridge *bridge;
|
2021-12-07 00:57:48 +08:00
|
|
|
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
bridge = dsa_tree_bridge_find(ds->dst, br);
|
|
|
|
|
if (bridge) {
|
|
|
|
|
refcount_inc(&bridge->refcount);
|
|
|
|
|
dp->bridge = bridge;
|
2021-12-07 00:57:48 +08:00
|
|
|
return 0;
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
bridge = kzalloc(sizeof(*bridge), GFP_KERNEL);
|
|
|
|
|
if (!bridge)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
refcount_set(&bridge->refcount, 1);
|
|
|
|
|
|
|
|
|
|
bridge->dev = br;
|
2021-12-07 00:57:48 +08:00
|
|
|
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
bridge->num = dsa_bridge_num_get(br, ds->max_num_bridges);
|
|
|
|
|
if (ds->max_num_bridges && !bridge->num) {
|
2021-12-07 00:57:48 +08:00
|
|
|
NL_SET_ERR_MSG_MOD(extack,
|
|
|
|
|
"Range of offloadable bridges exceeded");
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
kfree(bridge);
|
2021-12-07 00:57:48 +08:00
|
|
|
return -EOPNOTSUPP;
|
net: dsa: add support for bridge TX forwarding offload
For a DSA switch, to offload the forwarding process of a bridge device
means to send the packets coming from the software bridge as data plane
packets. This is contrary to everything that DSA has done so far,
because the current taggers only know to send control packets (ones that
target a specific destination port), whereas data plane packets are
supposed to be forwarded according to the FDB lookup, much like packets
ingressing on any regular ingress port. If the FDB lookup process
returns multiple destination ports (flooding, multicast), then
replication is also handled by the switch hardware - the bridge only
sends a single packet and avoids the skb_clone().
DSA keeps for each bridge port a zero-based index (the number of the
bridge). Multiple ports performing TX forwarding offload to the same
bridge have the same dp->bridge_num value, and ports not offloading the
TX data plane of a bridge have dp->bridge_num = -1.
The tagger can check if the packet that is being transmitted on has
skb->offload_fwd_mark = true or not. If it does, it can be sure that the
packet belongs to the data plane of a bridge, further information about
which can be obtained based on dp->bridge_dev and dp->bridge_num.
It can then compose a DSA tag for injecting a data plane packet into
that bridge number.
For the switch driver side, we offer two new dsa_switch_ops methods,
called .port_bridge_fwd_offload_{add,del}, which are modeled after
.port_bridge_{join,leave}.
These methods are provided in case the driver needs to configure the
hardware to treat packets coming from that bridge software interface as
data plane packets. The switchdev <-> bridge interaction happens during
the netdev_master_upper_dev_link() call, so to switch drivers, the
effect is that the .port_bridge_fwd_offload_add() method is called
immediately after .port_bridge_join().
If the bridge number exceeds the number of bridges for which the switch
driver can offload the TX data plane (and this includes the case where
the driver can offload none), DSA falls back to simply returning
tx_fwd_offload = false in the switchdev_bridge_port_offload() call.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 23:55:40 +08:00
|
|
|
}
|
|
|
|
|
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
dp->bridge = bridge;
|
2021-12-07 00:57:48 +08:00
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void dsa_port_bridge_destroy(struct dsa_port *dp,
|
|
|
|
|
const struct net_device *br)
|
|
|
|
|
{
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
struct dsa_bridge *bridge = dp->bridge;
|
|
|
|
|
|
|
|
|
|
dp->bridge = NULL;
|
2021-12-07 00:57:48 +08:00
|
|
|
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
if (!refcount_dec_and_test(&bridge->refcount))
|
|
|
|
|
return;
|
2021-12-07 00:57:48 +08:00
|
|
|
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
if (bridge->num)
|
|
|
|
|
dsa_bridge_num_put(br, bridge->num);
|
2021-12-07 00:57:48 +08:00
|
|
|
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
kfree(bridge);
|
net: dsa: add support for bridge TX forwarding offload
For a DSA switch, to offload the forwarding process of a bridge device
means to send the packets coming from the software bridge as data plane
packets. This is contrary to everything that DSA has done so far,
because the current taggers only know to send control packets (ones that
target a specific destination port), whereas data plane packets are
supposed to be forwarded according to the FDB lookup, much like packets
ingressing on any regular ingress port. If the FDB lookup process
returns multiple destination ports (flooding, multicast), then
replication is also handled by the switch hardware - the bridge only
sends a single packet and avoids the skb_clone().
DSA keeps for each bridge port a zero-based index (the number of the
bridge). Multiple ports performing TX forwarding offload to the same
bridge have the same dp->bridge_num value, and ports not offloading the
TX data plane of a bridge have dp->bridge_num = -1.
The tagger can check if the packet that is being transmitted on has
skb->offload_fwd_mark = true or not. If it does, it can be sure that the
packet belongs to the data plane of a bridge, further information about
which can be obtained based on dp->bridge_dev and dp->bridge_num.
It can then compose a DSA tag for injecting a data plane packet into
that bridge number.
For the switch driver side, we offer two new dsa_switch_ops methods,
called .port_bridge_fwd_offload_{add,del}, which are modeled after
.port_bridge_{join,leave}.
These methods are provided in case the driver needs to configure the
hardware to treat packets coming from that bridge software interface as
data plane packets. The switchdev <-> bridge interaction happens during
the netdev_master_upper_dev_link() call, so to switch drivers, the
effect is that the .port_bridge_fwd_offload_add() method is called
immediately after .port_bridge_join().
If the bridge number exceeds the number of bridges for which the switch
driver can offload the TX data plane (and this includes the case where
the driver can offload none), DSA falls back to simply returning
tx_fwd_offload = false in the switchdev_bridge_port_offload() call.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 23:55:40 +08:00
|
|
|
}
|
|
|
|
|
|
2022-03-16 23:08:52 +08:00
|
|
|
static bool dsa_port_supports_mst(struct dsa_port *dp)
|
|
|
|
|
{
|
2022-03-16 23:08:53 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
|
|
|
|
return ds->ops->vlan_msti_set &&
|
2022-03-16 23:08:54 +08:00
|
|
|
ds->ops->port_mst_state_set &&
|
|
|
|
|
ds->ops->port_vlan_fast_age &&
|
2022-03-16 23:08:53 +08:00
|
|
|
dsa_port_can_configure_learning(dp);
|
2022-03-16 23:08:52 +08:00
|
|
|
}
|
|
|
|
|
|
2021-03-23 07:51:48 +08:00
|
|
|
int dsa_port_bridge_join(struct dsa_port *dp, struct net_device *br,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
2017-05-20 05:00:45 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_notifier_bridge_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2022-02-25 17:22:23 +08:00
|
|
|
.extack = extack,
|
2017-05-20 05:00:45 +08:00
|
|
|
};
|
2023-10-24 02:17:28 +08:00
|
|
|
struct net_device *dev = dp->user;
|
net: bridge: switchdev: let drivers inform which bridge ports are offloaded
On reception of an skb, the bridge checks if it was marked as 'already
forwarded in hardware' (checks if skb->offload_fwd_mark == 1), and if it
is, it assigns the source hardware domain of that skb based on the
hardware domain of the ingress port. Then during forwarding, it enforces
that the egress port must have a different hardware domain than the
ingress one (this is done in nbp_switchdev_allowed_egress).
Non-switchdev drivers don't report any physical switch id (neither
through devlink nor .ndo_get_port_parent_id), therefore the bridge
assigns them a hardware domain of 0, and packets coming from them will
always have skb->offload_fwd_mark = 0. So there aren't any restrictions.
Problems appear due to the fact that DSA would like to perform software
fallback for bonding and team interfaces that the physical switch cannot
offload.
+-- br0 ---+
/ / | \
/ / | \
/ | | bond0
/ | | / \
swp0 swp1 swp2 swp3 swp4
There, it is desirable that the presence of swp3 and swp4 under a
non-offloaded LAG does not preclude us from doing hardware bridging
beteen swp0, swp1 and swp2. The bandwidth of the CPU is often times high
enough that software bridging between {swp0,swp1,swp2} and bond0 is not
impractical.
But this creates an impossible paradox given the current way in which
port hardware domains are assigned. When the driver receives a packet
from swp0 (say, due to flooding), it must set skb->offload_fwd_mark to
something.
- If we set it to 0, then the bridge will forward it towards swp1, swp2
and bond0. But the switch has already forwarded it towards swp1 and
swp2 (not to bond0, remember, that isn't offloaded, so as far as the
switch is concerned, ports swp3 and swp4 are not looking up the FDB,
and the entire bond0 is a destination that is strictly behind the
CPU). But we don't want duplicated traffic towards swp1 and swp2, so
it's not ok to set skb->offload_fwd_mark = 0.
- If we set it to 1, then the bridge will not forward the skb towards
the ports with the same switchdev mark, i.e. not to swp1, swp2 and
bond0. Towards swp1 and swp2 that's ok, but towards bond0? It should
have forwarded the skb there.
So the real issue is that bond0 will be assigned the same hardware
domain as {swp0,swp1,swp2}, because the function that assigns hardware
domains to bridge ports, nbp_switchdev_add(), recurses through bond0's
lower interfaces until it finds something that implements devlink (calls
dev_get_port_parent_id with bool recurse = true). This is a problem
because the fact that bond0 can be offloaded by swp3 and swp4 in our
example is merely an assumption.
A solution is to give the bridge explicit hints as to what hardware
domain it should use for each port.
Currently, the bridging offload is very 'silent': a driver registers a
netdevice notifier, which is put on the netns's notifier chain, and
which sniffs around for NETDEV_CHANGEUPPER events where the upper is a
bridge, and the lower is an interface it knows about (one registered by
this driver, normally). Then, from within that notifier, it does a bunch
of stuff behind the bridge's back, without the bridge necessarily
knowing that there's somebody offloading that port. It looks like this:
ip link set swp0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v
call_netdevice_notifiers
|
v
dsa_slave_netdevice_event
|
v
oh, hey! it's for me!
|
v
.port_bridge_join
What we do to solve the conundrum is to be less silent, and change the
switchdev drivers to present themselves to the bridge. Something like this:
ip link set swp0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge: Aye! I'll use this
call_netdevice_notifiers ^ ppid as the
| | hardware domain for
v | this port, and zero
dsa_slave_netdevice_event | if I got nothing.
| |
v |
oh, hey! it's for me! |
| |
v |
.port_bridge_join |
| |
+------------------------+
switchdev_bridge_port_offload(swp0, swp0)
Then stacked interfaces (like bond0 on top of swp3/swp4) would be
treated differently in DSA, depending on whether we can or cannot
offload them.
The offload case:
ip link set bond0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge: Aye! I'll use this
call_netdevice_notifiers ^ ppid as the
| | switchdev mark for
v | bond0.
dsa_slave_netdevice_event | Coincidentally (or not),
| | bond0 and swp0, swp1, swp2
v | all have the same switchdev
hmm, it's not quite for me, | mark now, since the ASIC
but my driver has already | is able to forward towards
called .port_lag_join | all these ports in hw.
for it, because I have |
a port with dp->lag_dev == bond0. |
| |
v |
.port_bridge_join |
for swp3 and swp4 |
| |
+------------------------+
switchdev_bridge_port_offload(bond0, swp3)
switchdev_bridge_port_offload(bond0, swp4)
And the non-offload case:
ip link set bond0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge waiting:
call_netdevice_notifiers ^ huh, switchdev_bridge_port_offload
| | wasn't called, okay, I'll use a
v | hwdom of zero for this one.
dsa_slave_netdevice_event : Then packets received on swp0 will
| : not be software-forwarded towards
v : swp1, but they will towards bond0.
it's not for me, but
bond0 is an upper of swp3
and swp4, but their dp->lag_dev
is NULL because they couldn't
offload it.
Basically we can draw the conclusion that the lowers of a bridge port
can come and go, so depending on the configuration of lowers for a
bridge port, it can dynamically toggle between offloaded and unoffloaded.
Therefore, we need an equivalent switchdev_bridge_port_unoffload too.
This patch changes the way any switchdev driver interacts with the
bridge. From now on, everybody needs to call switchdev_bridge_port_offload
and switchdev_bridge_port_unoffload, otherwise the bridge will treat the
port as non-offloaded and allow software flooding to other ports from
the same ASIC.
Note that these functions lay the ground for a more complex handshake
between switchdev drivers and the bridge in the future.
For drivers that will request a replay of the switchdev objects when
they offload and unoffload a bridge port (DSA, dpaa2-switch, ocelot), we
place the call to switchdev_bridge_port_unoffload() strategically inside
the NETDEV_PRECHANGEUPPER notifier's code path, and not inside
NETDEV_CHANGEUPPER. This is because the switchdev object replay helpers
need the netdev adjacency lists to be valid, and that is only true in
NETDEV_PRECHANGEUPPER.
Cc: Vadym Kochan <vkochan@marvell.com>
Cc: Taras Chornyi <tchornyi@marvell.com>
Cc: Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Cc: Claudiu Manoil <claudiu.manoil@nxp.com>
Cc: Alexandre Belloni <alexandre.belloni@bootlin.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Tested-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch: regression
Acked-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch
Tested-by: Horatiu Vultur <horatiu.vultur@microchip.com> # ocelot-switch
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 00:24:01 +08:00
|
|
|
struct net_device *brport_dev;
|
2017-05-20 05:00:45 +08:00
|
|
|
int err;
|
|
|
|
|
|
2022-03-16 23:08:52 +08:00
|
|
|
if (br_mst_enabled(br) && !dsa_port_supports_mst(dp))
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
|
net: dsa: enable flooding for bridge ports
Switches work by learning the MAC address for each attached station by
monitoring traffic from each station. When a station sends a packet,
the switch records which port the MAC address is connected to.
With IPv4 networking, before communication commences with a neighbour,
an ARP packet is broadcasted to all stations asking for the MAC address
corresponding with the IPv4. The desired station responds with an ARP
reply, and the ARP reply causes the switch to learn which port the
station is connected to.
With IPv6 networking, the situation is rather different. Rather than
broadcasting ARP packets, a "neighbour solicitation" is multicasted
rather than broadcasted. This multicast needs to reach the intended
station in order for the neighbour to be discovered.
Once a neighbour has been discovered, and entered into the sending
stations neighbour cache, communication can restart at a point later
without sending a new neighbour solicitation, even if the entry in
the neighbour cache is marked as stale. This can be after the MAC
address has expired from the forwarding cache of the DSA switch -
when that occurs, there is a long pause in communication.
Our DSA implementation for mv88e6xxx switches disables flooding of
multicast and unicast frames for bridged ports. As per the above
description, this is fine for IPv4 networking, since the broadcasted
ARP queries will be sent to and received by all stations on the same
network. However, this breaks IPv6 very badly - blocking neighbour
solicitations and later causing connections to stall.
The defaults that the Linux bridge code expect from bridges are for
unknown unicast and unknown multicast frames to be flooded to all ports
on the bridge, which is at odds to the defaults adopted by our DSA
implementation for mv88e6xxx switches.
This commit enables by default flooding of both unknown unicast and
unknown multicast frames whenever a port is added to a bridge, and
disables the flooding when a port leaves the bridge. This means that
mv88e6xxx DSA switches now behave as per the bridge(8) man page, and
IPv6 works flawlessly through such a switch.
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Vivien Didelot <vivien.didelot@gmail.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-02-21 07:35:06 +08:00
|
|
|
/* Here the interface is already bridged. Reflect the current
|
|
|
|
|
* configuration so that drivers can program their chips accordingly.
|
2017-05-20 05:00:45 +08:00
|
|
|
*/
|
2021-12-07 00:57:48 +08:00
|
|
|
err = dsa_port_bridge_create(dp, br, extack);
|
|
|
|
|
if (err)
|
|
|
|
|
return err;
|
2017-05-20 05:00:45 +08:00
|
|
|
|
net: bridge: switchdev: let drivers inform which bridge ports are offloaded
On reception of an skb, the bridge checks if it was marked as 'already
forwarded in hardware' (checks if skb->offload_fwd_mark == 1), and if it
is, it assigns the source hardware domain of that skb based on the
hardware domain of the ingress port. Then during forwarding, it enforces
that the egress port must have a different hardware domain than the
ingress one (this is done in nbp_switchdev_allowed_egress).
Non-switchdev drivers don't report any physical switch id (neither
through devlink nor .ndo_get_port_parent_id), therefore the bridge
assigns them a hardware domain of 0, and packets coming from them will
always have skb->offload_fwd_mark = 0. So there aren't any restrictions.
Problems appear due to the fact that DSA would like to perform software
fallback for bonding and team interfaces that the physical switch cannot
offload.
+-- br0 ---+
/ / | \
/ / | \
/ | | bond0
/ | | / \
swp0 swp1 swp2 swp3 swp4
There, it is desirable that the presence of swp3 and swp4 under a
non-offloaded LAG does not preclude us from doing hardware bridging
beteen swp0, swp1 and swp2. The bandwidth of the CPU is often times high
enough that software bridging between {swp0,swp1,swp2} and bond0 is not
impractical.
But this creates an impossible paradox given the current way in which
port hardware domains are assigned. When the driver receives a packet
from swp0 (say, due to flooding), it must set skb->offload_fwd_mark to
something.
- If we set it to 0, then the bridge will forward it towards swp1, swp2
and bond0. But the switch has already forwarded it towards swp1 and
swp2 (not to bond0, remember, that isn't offloaded, so as far as the
switch is concerned, ports swp3 and swp4 are not looking up the FDB,
and the entire bond0 is a destination that is strictly behind the
CPU). But we don't want duplicated traffic towards swp1 and swp2, so
it's not ok to set skb->offload_fwd_mark = 0.
- If we set it to 1, then the bridge will not forward the skb towards
the ports with the same switchdev mark, i.e. not to swp1, swp2 and
bond0. Towards swp1 and swp2 that's ok, but towards bond0? It should
have forwarded the skb there.
So the real issue is that bond0 will be assigned the same hardware
domain as {swp0,swp1,swp2}, because the function that assigns hardware
domains to bridge ports, nbp_switchdev_add(), recurses through bond0's
lower interfaces until it finds something that implements devlink (calls
dev_get_port_parent_id with bool recurse = true). This is a problem
because the fact that bond0 can be offloaded by swp3 and swp4 in our
example is merely an assumption.
A solution is to give the bridge explicit hints as to what hardware
domain it should use for each port.
Currently, the bridging offload is very 'silent': a driver registers a
netdevice notifier, which is put on the netns's notifier chain, and
which sniffs around for NETDEV_CHANGEUPPER events where the upper is a
bridge, and the lower is an interface it knows about (one registered by
this driver, normally). Then, from within that notifier, it does a bunch
of stuff behind the bridge's back, without the bridge necessarily
knowing that there's somebody offloading that port. It looks like this:
ip link set swp0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v
call_netdevice_notifiers
|
v
dsa_slave_netdevice_event
|
v
oh, hey! it's for me!
|
v
.port_bridge_join
What we do to solve the conundrum is to be less silent, and change the
switchdev drivers to present themselves to the bridge. Something like this:
ip link set swp0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge: Aye! I'll use this
call_netdevice_notifiers ^ ppid as the
| | hardware domain for
v | this port, and zero
dsa_slave_netdevice_event | if I got nothing.
| |
v |
oh, hey! it's for me! |
| |
v |
.port_bridge_join |
| |
+------------------------+
switchdev_bridge_port_offload(swp0, swp0)
Then stacked interfaces (like bond0 on top of swp3/swp4) would be
treated differently in DSA, depending on whether we can or cannot
offload them.
The offload case:
ip link set bond0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge: Aye! I'll use this
call_netdevice_notifiers ^ ppid as the
| | switchdev mark for
v | bond0.
dsa_slave_netdevice_event | Coincidentally (or not),
| | bond0 and swp0, swp1, swp2
v | all have the same switchdev
hmm, it's not quite for me, | mark now, since the ASIC
but my driver has already | is able to forward towards
called .port_lag_join | all these ports in hw.
for it, because I have |
a port with dp->lag_dev == bond0. |
| |
v |
.port_bridge_join |
for swp3 and swp4 |
| |
+------------------------+
switchdev_bridge_port_offload(bond0, swp3)
switchdev_bridge_port_offload(bond0, swp4)
And the non-offload case:
ip link set bond0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge waiting:
call_netdevice_notifiers ^ huh, switchdev_bridge_port_offload
| | wasn't called, okay, I'll use a
v | hwdom of zero for this one.
dsa_slave_netdevice_event : Then packets received on swp0 will
| : not be software-forwarded towards
v : swp1, but they will towards bond0.
it's not for me, but
bond0 is an upper of swp3
and swp4, but their dp->lag_dev
is NULL because they couldn't
offload it.
Basically we can draw the conclusion that the lowers of a bridge port
can come and go, so depending on the configuration of lowers for a
bridge port, it can dynamically toggle between offloaded and unoffloaded.
Therefore, we need an equivalent switchdev_bridge_port_unoffload too.
This patch changes the way any switchdev driver interacts with the
bridge. From now on, everybody needs to call switchdev_bridge_port_offload
and switchdev_bridge_port_unoffload, otherwise the bridge will treat the
port as non-offloaded and allow software flooding to other ports from
the same ASIC.
Note that these functions lay the ground for a more complex handshake
between switchdev drivers and the bridge in the future.
For drivers that will request a replay of the switchdev objects when
they offload and unoffload a bridge port (DSA, dpaa2-switch, ocelot), we
place the call to switchdev_bridge_port_unoffload() strategically inside
the NETDEV_PRECHANGEUPPER notifier's code path, and not inside
NETDEV_CHANGEUPPER. This is because the switchdev object replay helpers
need the netdev adjacency lists to be valid, and that is only true in
NETDEV_PRECHANGEUPPER.
Cc: Vadym Kochan <vkochan@marvell.com>
Cc: Taras Chornyi <tchornyi@marvell.com>
Cc: Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Cc: Claudiu Manoil <claudiu.manoil@nxp.com>
Cc: Alexandre Belloni <alexandre.belloni@bootlin.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Tested-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch: regression
Acked-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch
Tested-by: Horatiu Vultur <horatiu.vultur@microchip.com> # ocelot-switch
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 00:24:01 +08:00
|
|
|
brport_dev = dsa_port_to_bridge_port(dp);
|
|
|
|
|
|
net: dsa: keep the bridge_dev and bridge_num as part of the same structure
The main desire behind this is to provide coherent bridge information to
the fast path without locking.
For example, right now we set dp->bridge_dev and dp->bridge_num from
separate code paths, it is theoretically possible for a packet
transmission to read these two port properties consecutively and find a
bridge number which does not correspond with the bridge device.
Another desire is to start passing more complex bridge information to
dsa_switch_ops functions. For example, with FDB isolation, it is
expected that drivers will need to be passed the bridge which requested
an FDB/MDB entry to be offloaded, and along with that bridge_dev, the
associated bridge_num should be passed too, in case the driver might
want to implement an isolation scheme based on that number.
We already pass the {bridge_dev, bridge_num} pair to the TX forwarding
offload switch API, however we'd like to remove that and squash it into
the basic bridge join/leave API. So that means we need to pass this
pair to the bridge join/leave API.
During dsa_port_bridge_leave, first we unset dp->bridge_dev, then we
call the driver's .port_bridge_leave with what used to be our
dp->bridge_dev, but provided as an argument.
When bridge_dev and bridge_num get folded into a single structure, we
need to preserve this behavior in dsa_port_bridge_leave: we need a copy
of what used to be in dp->bridge.
Switch drivers check bridge membership by comparing dp->bridge_dev with
the provided bridge_dev, but now, if we provide the struct dsa_bridge as
a pointer, they cannot keep comparing dp->bridge to the provided
pointer, since this only points to an on-stack copy. To make this
obvious and prevent driver writers from forgetting and doing stupid
things, in this new API, the struct dsa_bridge is provided as a full
structure (not very large, contains an int and a pointer) instead of a
pointer. An explicit comparison function needs to be used to determine
bridge membership: dsa_port_offloads_bridge().
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-12-07 00:57:56 +08:00
|
|
|
info.bridge = *dp->bridge;
|
net: dsa: permit cross-chip bridging between all trees in the system
One way of utilizing DSA is by cascading switches which do not all have
compatible taggers. Consider the following real-life topology:
+---------------------------------------------------------------+
| LS1028A |
| +------------------------------+ |
| | DSA master for Felix | |
| |(internal ENETC port 2: eno2))| |
| +------------+------------------------------+-------------+ |
| | Felix embedded L2 switch | |
| | | |
| | +--------------+ +--------------+ +--------------+ | |
| | |DSA master for| |DSA master for| |DSA master for| | |
| | | SJA1105 1 | | SJA1105 2 | | SJA1105 3 | | |
| | |(Felix port 1)| |(Felix port 2)| |(Felix port 3)| | |
+--+-+--------------+---+--------------+---+--------------+--+--+
+-----------------------+ +-----------------------+ +-----------------------+
| SJA1105 switch 1 | | SJA1105 switch 2 | | SJA1105 switch 3 |
+-----+-----+-----+-----+ +-----+-----+-----+-----+ +-----+-----+-----+-----+
|sw1p0|sw1p1|sw1p2|sw1p3| |sw2p0|sw2p1|sw2p2|sw2p3| |sw3p0|sw3p1|sw3p2|sw3p3|
+-----+-----+-----+-----+ +-----+-----+-----+-----+ +-----+-----+-----+-----+
The above can be described in the device tree as follows (obviously not
complete):
mscc_felix {
dsa,member = <0 0>;
ports {
port@4 {
ethernet = <&enetc_port2>;
};
};
};
sja1105_switch1 {
dsa,member = <1 1>;
ports {
port@4 {
ethernet = <&mscc_felix_port1>;
};
};
};
sja1105_switch2 {
dsa,member = <2 2>;
ports {
port@4 {
ethernet = <&mscc_felix_port2>;
};
};
};
sja1105_switch3 {
dsa,member = <3 3>;
ports {
port@4 {
ethernet = <&mscc_felix_port3>;
};
};
};
Basically we instantiate one DSA switch tree for every hardware switch
in the system, but we still give them globally unique switch IDs (will
come back to that later). Having 3 disjoint switch trees makes the
tagger drivers "just work", because net devices are registered for the
3 Felix DSA master ports, and they are also DSA slave ports to the ENETC
port. So packets received on the ENETC port are stripped of their
stacked DSA tags one by one.
Currently, hardware bridging between ports on the same sja1105 chip is
possible, but switching between sja1105 ports on different chips is
handled by the software bridge. This is fine, but we can do better.
In fact, the dsa_8021q tag used by sja1105 is compatible with cascading.
In other words, a sja1105 switch can correctly parse and route a packet
containing a dsa_8021q tag. So if we could enable hardware bridging on
the Felix DSA master ports, cross-chip bridging could be completely
offloaded.
Such as system would be used as follows:
ip link add dev br0 type bridge && ip link set dev br0 up
for port in sw0p0 sw0p1 sw0p2 sw0p3 \
sw1p0 sw1p1 sw1p2 sw1p3 \
sw2p0 sw2p1 sw2p2 sw2p3; do
ip link set dev $port master br0
done
The above makes switching between ports on the same row be performed in
hardware, and between ports on different rows in software. Now assume
the Felix switch ports are called swp0, swp1, swp2. By running the
following extra commands:
ip link add dev br1 type bridge && ip link set dev br1 up
for port in swp0 swp1 swp2; do
ip link set dev $port master br1
done
the CPU no longer sees packets which traverse sja1105 switch boundaries
and can be forwarded directly by Felix. The br1 bridge would not be used
for any sort of traffic termination.
For this to work, we need to give drivers an opportunity to listen for
bridging events on DSA trees other than their own, and pass that other
tree index as argument. I have made the assumption, for the moment, that
the other existing DSA notifiers don't need to be broadcast to other
trees. That assumption might turn out to be incorrect. But in the
meantime, introduce a dsa_broadcast function, similar in purpose to
dsa_port_notify, which is used only by the bridging notifiers.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2020-05-11 00:37:41 +08:00
|
|
|
err = dsa_broadcast(DSA_NOTIFIER_BRIDGE_JOIN, &info);
|
2021-03-23 07:51:49 +08:00
|
|
|
if (err)
|
|
|
|
|
goto out_rollback;
|
2017-05-20 05:00:45 +08:00
|
|
|
|
2021-12-07 00:57:58 +08:00
|
|
|
/* Drivers which support bridge TX forwarding should set this */
|
|
|
|
|
dp->bridge->tx_fwd_offload = info.tx_fwd_offload;
|
net: dsa: add support for bridge TX forwarding offload
For a DSA switch, to offload the forwarding process of a bridge device
means to send the packets coming from the software bridge as data plane
packets. This is contrary to everything that DSA has done so far,
because the current taggers only know to send control packets (ones that
target a specific destination port), whereas data plane packets are
supposed to be forwarded according to the FDB lookup, much like packets
ingressing on any regular ingress port. If the FDB lookup process
returns multiple destination ports (flooding, multicast), then
replication is also handled by the switch hardware - the bridge only
sends a single packet and avoids the skb_clone().
DSA keeps for each bridge port a zero-based index (the number of the
bridge). Multiple ports performing TX forwarding offload to the same
bridge have the same dp->bridge_num value, and ports not offloading the
TX data plane of a bridge have dp->bridge_num = -1.
The tagger can check if the packet that is being transmitted on has
skb->offload_fwd_mark = true or not. If it does, it can be sure that the
packet belongs to the data plane of a bridge, further information about
which can be obtained based on dp->bridge_dev and dp->bridge_num.
It can then compose a DSA tag for injecting a data plane packet into
that bridge number.
For the switch driver side, we offer two new dsa_switch_ops methods,
called .port_bridge_fwd_offload_{add,del}, which are modeled after
.port_bridge_{join,leave}.
These methods are provided in case the driver needs to configure the
hardware to treat packets coming from that bridge software interface as
data plane packets. The switchdev <-> bridge interaction happens during
the netdev_master_upper_dev_link() call, so to switch drivers, the
effect is that the .port_bridge_fwd_offload_add() method is called
immediately after .port_bridge_join().
If the bridge number exceeds the number of bridges for which the switch
driver can offload the TX data plane (and this includes the case where
the driver can offload none), DSA falls back to simply returning
tx_fwd_offload = false in the switchdev_bridge_port_offload() call.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 23:55:40 +08:00
|
|
|
|
net: bridge: move the switchdev object replay helpers to "push" mode
Starting with commit 4f2673b3a2b6 ("net: bridge: add helper to replay
port and host-joined mdb entries"), DSA has introduced some bridge
helpers that replay switchdev events (FDB/MDB/VLAN additions and
deletions) that can be lost by the switchdev drivers in a variety of
circumstances:
- an IP multicast group was host-joined on the bridge itself before any
switchdev port joined the bridge, leading to the host MDB entries
missing in the hardware database.
- during the bridge creation process, the MAC address of the bridge was
added to the FDB as an entry pointing towards the bridge device
itself, but with no switchdev ports being part of the bridge yet, this
local FDB entry would remain unknown to the switchdev hardware
database.
- a VLAN/FDB/MDB was added to a bridge port that is a LAG interface,
before any switchdev port joined that LAG, leading to the hardware
database missing those entries.
- a switchdev port left a LAG that is a bridge port, while the LAG
remained part of the bridge, and all FDB/MDB/VLAN entries remained
installed in the hardware database of the switchdev port.
Also, since commit 0d2cfbd41c4a ("net: bridge: ignore switchdev events
for LAG ports which didn't request replay"), DSA introduced a method,
based on a const void *ctx, to ensure that two switchdev ports under the
same LAG that is a bridge port do not see the same MDB/VLAN entry being
replayed twice by the bridge, once for every bridge port that joins the
LAG.
With so many ordering corner cases being possible, it seems unreasonable
to expect a switchdev driver writer to get it right from the first try.
Therefore, now that DSA has experimented with the bridge replay helpers
for a little bit, we can move the code to the bridge driver where it is
more readily available to all switchdev drivers.
To convert the switchdev object replay helpers from "pull mode" (where
the driver asks for them) to a "push mode" (where the bridge offers them
automatically), the biggest problem is that the bridge needs to be aware
when a switchdev port joins and leaves, even when the switchdev is only
indirectly a bridge port (for example when the bridge port is a LAG
upper of the switchdev).
Luckily, we already have a hook for that, in the form of the newly
introduced switchdev_bridge_port_offload() and
switchdev_bridge_port_unoffload() calls. These offer a natural place for
hooking the object addition and deletion replays.
Extend the above 2 functions with:
- pointers to the switchdev atomic notifier (for FDB replays) and the
blocking notifier (for MDB and VLAN replays).
- the "const void *ctx" argument required for drivers to be able to
disambiguate between which port is targeted, when multiple ports are
lowers of the same LAG that is a bridge port. Most of the drivers pass
NULL to this argument, except the ones that support LAG offload and have
the proper context check already in place in the switchdev blocking
notifier handler.
Also unexport the replay helpers, since nobody except the bridge calls
them directly now.
Note that:
(a) we abuse the terminology slightly, because FDB entries are not
"switchdev objects", but we count them as objects nonetheless.
With no direct way to prove it, I think they are not modeled as
switchdev objects because those can only be installed by the bridge
to the hardware (as opposed to FDB entries which can be propagated
in the other direction too). This is merely an abuse of terms, FDB
entries are replayed too, despite not being objects.
(b) the bridge does not attempt to sync port attributes to newly joined
ports, just the countable stuff (the objects). The reason for this
is simple: no universal and symmetric way to sync and unsync them is
known. For example, VLAN filtering: what to do on unsync, disable or
leave it enabled? Similarly, STP state, ageing timer, etc etc. What
a switchdev port does when it becomes standalone again is not really
up to the bridge's competence, and the driver should deal with it.
On the other hand, replaying deletions of switchdev objects can be
seen a matter of cleanup and therefore be treated by the bridge,
hence this patch.
We make the replay helpers opt-in for drivers, because they might not
bring immediate benefits for them:
- nbp_vlan_init() is called _after_ netdev_master_upper_dev_link(),
so br_vlan_replay() should not do anything for the new drivers on
which we call it. The existing drivers where there was even a slight
possibility for there to exist a VLAN on a bridge port before they
join it are already guarded against this: mlxsw and prestera deny
joining LAG interfaces that are members of a bridge.
- br_fdb_replay() should now notify of local FDB entries, but I patched
all drivers except DSA to ignore these new entries in commit
2c4eca3ef716 ("net: bridge: switchdev: include local flag in FDB
notifications"). Driver authors can lift this restriction as they
wish, and when they do, they can also opt into the FDB replay
functionality.
- br_mdb_replay() should fix a real issue which is described in commit
4f2673b3a2b6 ("net: bridge: add helper to replay port and host-joined
mdb entries"). However most drivers do not offload the
SWITCHDEV_OBJ_ID_HOST_MDB to see this issue: only cpsw and am65_cpsw
offload this switchdev object, and I don't completely understand the
way in which they offload this switchdev object anyway. So I'll leave
it up to these drivers' respective maintainers to opt into
br_mdb_replay().
So most of the drivers pass NULL notifier blocks for the replay helpers,
except:
- dpaa2-switch which was already acked/regression-tested with the
helpers enabled (and there isn't much of a downside in having them)
- ocelot which already had replay logic in "pull" mode
- DSA which already had replay logic in "pull" mode
An important observation is that the drivers which don't currently
request bridge event replays don't even have the
switchdev_bridge_port_{offload,unoffload} calls placed in proper places
right now. This was done to avoid unnecessary rework for drivers which
might never even add support for this. For driver writers who wish to
add replay support, this can be used as a tentative placement guide:
https://patchwork.kernel.org/project/netdevbpf/patch/20210720134655.892334-11-vladimir.oltean@nxp.com/
Cc: Vadym Kochan <vkochan@marvell.com>
Cc: Taras Chornyi <tchornyi@marvell.com>
Cc: Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Cc: Claudiu Manoil <claudiu.manoil@nxp.com>
Cc: Alexandre Belloni <alexandre.belloni@bootlin.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 00:24:03 +08:00
|
|
|
err = switchdev_bridge_port_offload(brport_dev, dev, dp,
|
2023-10-24 02:17:28 +08:00
|
|
|
&dsa_user_switchdev_notifier,
|
|
|
|
|
&dsa_user_switchdev_blocking_notifier,
|
2021-12-07 00:57:58 +08:00
|
|
|
dp->bridge->tx_fwd_offload, extack);
|
2021-03-23 07:51:49 +08:00
|
|
|
if (err)
|
|
|
|
|
goto out_rollback_unbridge;
|
|
|
|
|
|
net: bridge: move the switchdev object replay helpers to "push" mode
Starting with commit 4f2673b3a2b6 ("net: bridge: add helper to replay
port and host-joined mdb entries"), DSA has introduced some bridge
helpers that replay switchdev events (FDB/MDB/VLAN additions and
deletions) that can be lost by the switchdev drivers in a variety of
circumstances:
- an IP multicast group was host-joined on the bridge itself before any
switchdev port joined the bridge, leading to the host MDB entries
missing in the hardware database.
- during the bridge creation process, the MAC address of the bridge was
added to the FDB as an entry pointing towards the bridge device
itself, but with no switchdev ports being part of the bridge yet, this
local FDB entry would remain unknown to the switchdev hardware
database.
- a VLAN/FDB/MDB was added to a bridge port that is a LAG interface,
before any switchdev port joined that LAG, leading to the hardware
database missing those entries.
- a switchdev port left a LAG that is a bridge port, while the LAG
remained part of the bridge, and all FDB/MDB/VLAN entries remained
installed in the hardware database of the switchdev port.
Also, since commit 0d2cfbd41c4a ("net: bridge: ignore switchdev events
for LAG ports which didn't request replay"), DSA introduced a method,
based on a const void *ctx, to ensure that two switchdev ports under the
same LAG that is a bridge port do not see the same MDB/VLAN entry being
replayed twice by the bridge, once for every bridge port that joins the
LAG.
With so many ordering corner cases being possible, it seems unreasonable
to expect a switchdev driver writer to get it right from the first try.
Therefore, now that DSA has experimented with the bridge replay helpers
for a little bit, we can move the code to the bridge driver where it is
more readily available to all switchdev drivers.
To convert the switchdev object replay helpers from "pull mode" (where
the driver asks for them) to a "push mode" (where the bridge offers them
automatically), the biggest problem is that the bridge needs to be aware
when a switchdev port joins and leaves, even when the switchdev is only
indirectly a bridge port (for example when the bridge port is a LAG
upper of the switchdev).
Luckily, we already have a hook for that, in the form of the newly
introduced switchdev_bridge_port_offload() and
switchdev_bridge_port_unoffload() calls. These offer a natural place for
hooking the object addition and deletion replays.
Extend the above 2 functions with:
- pointers to the switchdev atomic notifier (for FDB replays) and the
blocking notifier (for MDB and VLAN replays).
- the "const void *ctx" argument required for drivers to be able to
disambiguate between which port is targeted, when multiple ports are
lowers of the same LAG that is a bridge port. Most of the drivers pass
NULL to this argument, except the ones that support LAG offload and have
the proper context check already in place in the switchdev blocking
notifier handler.
Also unexport the replay helpers, since nobody except the bridge calls
them directly now.
Note that:
(a) we abuse the terminology slightly, because FDB entries are not
"switchdev objects", but we count them as objects nonetheless.
With no direct way to prove it, I think they are not modeled as
switchdev objects because those can only be installed by the bridge
to the hardware (as opposed to FDB entries which can be propagated
in the other direction too). This is merely an abuse of terms, FDB
entries are replayed too, despite not being objects.
(b) the bridge does not attempt to sync port attributes to newly joined
ports, just the countable stuff (the objects). The reason for this
is simple: no universal and symmetric way to sync and unsync them is
known. For example, VLAN filtering: what to do on unsync, disable or
leave it enabled? Similarly, STP state, ageing timer, etc etc. What
a switchdev port does when it becomes standalone again is not really
up to the bridge's competence, and the driver should deal with it.
On the other hand, replaying deletions of switchdev objects can be
seen a matter of cleanup and therefore be treated by the bridge,
hence this patch.
We make the replay helpers opt-in for drivers, because they might not
bring immediate benefits for them:
- nbp_vlan_init() is called _after_ netdev_master_upper_dev_link(),
so br_vlan_replay() should not do anything for the new drivers on
which we call it. The existing drivers where there was even a slight
possibility for there to exist a VLAN on a bridge port before they
join it are already guarded against this: mlxsw and prestera deny
joining LAG interfaces that are members of a bridge.
- br_fdb_replay() should now notify of local FDB entries, but I patched
all drivers except DSA to ignore these new entries in commit
2c4eca3ef716 ("net: bridge: switchdev: include local flag in FDB
notifications"). Driver authors can lift this restriction as they
wish, and when they do, they can also opt into the FDB replay
functionality.
- br_mdb_replay() should fix a real issue which is described in commit
4f2673b3a2b6 ("net: bridge: add helper to replay port and host-joined
mdb entries"). However most drivers do not offload the
SWITCHDEV_OBJ_ID_HOST_MDB to see this issue: only cpsw and am65_cpsw
offload this switchdev object, and I don't completely understand the
way in which they offload this switchdev object anyway. So I'll leave
it up to these drivers' respective maintainers to opt into
br_mdb_replay().
So most of the drivers pass NULL notifier blocks for the replay helpers,
except:
- dpaa2-switch which was already acked/regression-tested with the
helpers enabled (and there isn't much of a downside in having them)
- ocelot which already had replay logic in "pull" mode
- DSA which already had replay logic in "pull" mode
An important observation is that the drivers which don't currently
request bridge event replays don't even have the
switchdev_bridge_port_{offload,unoffload} calls placed in proper places
right now. This was done to avoid unnecessary rework for drivers which
might never even add support for this. For driver writers who wish to
add replay support, this can be used as a tentative placement guide:
https://patchwork.kernel.org/project/netdevbpf/patch/20210720134655.892334-11-vladimir.oltean@nxp.com/
Cc: Vadym Kochan <vkochan@marvell.com>
Cc: Taras Chornyi <tchornyi@marvell.com>
Cc: Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Cc: Claudiu Manoil <claudiu.manoil@nxp.com>
Cc: Alexandre Belloni <alexandre.belloni@bootlin.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 00:24:03 +08:00
|
|
|
err = dsa_port_switchdev_sync_attrs(dp, extack);
|
net: bridge: switchdev: let drivers inform which bridge ports are offloaded
On reception of an skb, the bridge checks if it was marked as 'already
forwarded in hardware' (checks if skb->offload_fwd_mark == 1), and if it
is, it assigns the source hardware domain of that skb based on the
hardware domain of the ingress port. Then during forwarding, it enforces
that the egress port must have a different hardware domain than the
ingress one (this is done in nbp_switchdev_allowed_egress).
Non-switchdev drivers don't report any physical switch id (neither
through devlink nor .ndo_get_port_parent_id), therefore the bridge
assigns them a hardware domain of 0, and packets coming from them will
always have skb->offload_fwd_mark = 0. So there aren't any restrictions.
Problems appear due to the fact that DSA would like to perform software
fallback for bonding and team interfaces that the physical switch cannot
offload.
+-- br0 ---+
/ / | \
/ / | \
/ | | bond0
/ | | / \
swp0 swp1 swp2 swp3 swp4
There, it is desirable that the presence of swp3 and swp4 under a
non-offloaded LAG does not preclude us from doing hardware bridging
beteen swp0, swp1 and swp2. The bandwidth of the CPU is often times high
enough that software bridging between {swp0,swp1,swp2} and bond0 is not
impractical.
But this creates an impossible paradox given the current way in which
port hardware domains are assigned. When the driver receives a packet
from swp0 (say, due to flooding), it must set skb->offload_fwd_mark to
something.
- If we set it to 0, then the bridge will forward it towards swp1, swp2
and bond0. But the switch has already forwarded it towards swp1 and
swp2 (not to bond0, remember, that isn't offloaded, so as far as the
switch is concerned, ports swp3 and swp4 are not looking up the FDB,
and the entire bond0 is a destination that is strictly behind the
CPU). But we don't want duplicated traffic towards swp1 and swp2, so
it's not ok to set skb->offload_fwd_mark = 0.
- If we set it to 1, then the bridge will not forward the skb towards
the ports with the same switchdev mark, i.e. not to swp1, swp2 and
bond0. Towards swp1 and swp2 that's ok, but towards bond0? It should
have forwarded the skb there.
So the real issue is that bond0 will be assigned the same hardware
domain as {swp0,swp1,swp2}, because the function that assigns hardware
domains to bridge ports, nbp_switchdev_add(), recurses through bond0's
lower interfaces until it finds something that implements devlink (calls
dev_get_port_parent_id with bool recurse = true). This is a problem
because the fact that bond0 can be offloaded by swp3 and swp4 in our
example is merely an assumption.
A solution is to give the bridge explicit hints as to what hardware
domain it should use for each port.
Currently, the bridging offload is very 'silent': a driver registers a
netdevice notifier, which is put on the netns's notifier chain, and
which sniffs around for NETDEV_CHANGEUPPER events where the upper is a
bridge, and the lower is an interface it knows about (one registered by
this driver, normally). Then, from within that notifier, it does a bunch
of stuff behind the bridge's back, without the bridge necessarily
knowing that there's somebody offloading that port. It looks like this:
ip link set swp0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v
call_netdevice_notifiers
|
v
dsa_slave_netdevice_event
|
v
oh, hey! it's for me!
|
v
.port_bridge_join
What we do to solve the conundrum is to be less silent, and change the
switchdev drivers to present themselves to the bridge. Something like this:
ip link set swp0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge: Aye! I'll use this
call_netdevice_notifiers ^ ppid as the
| | hardware domain for
v | this port, and zero
dsa_slave_netdevice_event | if I got nothing.
| |
v |
oh, hey! it's for me! |
| |
v |
.port_bridge_join |
| |
+------------------------+
switchdev_bridge_port_offload(swp0, swp0)
Then stacked interfaces (like bond0 on top of swp3/swp4) would be
treated differently in DSA, depending on whether we can or cannot
offload them.
The offload case:
ip link set bond0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge: Aye! I'll use this
call_netdevice_notifiers ^ ppid as the
| | switchdev mark for
v | bond0.
dsa_slave_netdevice_event | Coincidentally (or not),
| | bond0 and swp0, swp1, swp2
v | all have the same switchdev
hmm, it's not quite for me, | mark now, since the ASIC
but my driver has already | is able to forward towards
called .port_lag_join | all these ports in hw.
for it, because I have |
a port with dp->lag_dev == bond0. |
| |
v |
.port_bridge_join |
for swp3 and swp4 |
| |
+------------------------+
switchdev_bridge_port_offload(bond0, swp3)
switchdev_bridge_port_offload(bond0, swp4)
And the non-offload case:
ip link set bond0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge waiting:
call_netdevice_notifiers ^ huh, switchdev_bridge_port_offload
| | wasn't called, okay, I'll use a
v | hwdom of zero for this one.
dsa_slave_netdevice_event : Then packets received on swp0 will
| : not be software-forwarded towards
v : swp1, but they will towards bond0.
it's not for me, but
bond0 is an upper of swp3
and swp4, but their dp->lag_dev
is NULL because they couldn't
offload it.
Basically we can draw the conclusion that the lowers of a bridge port
can come and go, so depending on the configuration of lowers for a
bridge port, it can dynamically toggle between offloaded and unoffloaded.
Therefore, we need an equivalent switchdev_bridge_port_unoffload too.
This patch changes the way any switchdev driver interacts with the
bridge. From now on, everybody needs to call switchdev_bridge_port_offload
and switchdev_bridge_port_unoffload, otherwise the bridge will treat the
port as non-offloaded and allow software flooding to other ports from
the same ASIC.
Note that these functions lay the ground for a more complex handshake
between switchdev drivers and the bridge in the future.
For drivers that will request a replay of the switchdev objects when
they offload and unoffload a bridge port (DSA, dpaa2-switch, ocelot), we
place the call to switchdev_bridge_port_unoffload() strategically inside
the NETDEV_PRECHANGEUPPER notifier's code path, and not inside
NETDEV_CHANGEUPPER. This is because the switchdev object replay helpers
need the netdev adjacency lists to be valid, and that is only true in
NETDEV_PRECHANGEUPPER.
Cc: Vadym Kochan <vkochan@marvell.com>
Cc: Taras Chornyi <tchornyi@marvell.com>
Cc: Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Cc: Claudiu Manoil <claudiu.manoil@nxp.com>
Cc: Alexandre Belloni <alexandre.belloni@bootlin.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Tested-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch: regression
Acked-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch
Tested-by: Horatiu Vultur <horatiu.vultur@microchip.com> # ocelot-switch
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 00:24:01 +08:00
|
|
|
if (err)
|
|
|
|
|
goto out_rollback_unoffload;
|
|
|
|
|
|
2021-03-23 07:51:49 +08:00
|
|
|
return 0;
|
2017-05-20 05:00:45 +08:00
|
|
|
|
net: bridge: switchdev: let drivers inform which bridge ports are offloaded
On reception of an skb, the bridge checks if it was marked as 'already
forwarded in hardware' (checks if skb->offload_fwd_mark == 1), and if it
is, it assigns the source hardware domain of that skb based on the
hardware domain of the ingress port. Then during forwarding, it enforces
that the egress port must have a different hardware domain than the
ingress one (this is done in nbp_switchdev_allowed_egress).
Non-switchdev drivers don't report any physical switch id (neither
through devlink nor .ndo_get_port_parent_id), therefore the bridge
assigns them a hardware domain of 0, and packets coming from them will
always have skb->offload_fwd_mark = 0. So there aren't any restrictions.
Problems appear due to the fact that DSA would like to perform software
fallback for bonding and team interfaces that the physical switch cannot
offload.
+-- br0 ---+
/ / | \
/ / | \
/ | | bond0
/ | | / \
swp0 swp1 swp2 swp3 swp4
There, it is desirable that the presence of swp3 and swp4 under a
non-offloaded LAG does not preclude us from doing hardware bridging
beteen swp0, swp1 and swp2. The bandwidth of the CPU is often times high
enough that software bridging between {swp0,swp1,swp2} and bond0 is not
impractical.
But this creates an impossible paradox given the current way in which
port hardware domains are assigned. When the driver receives a packet
from swp0 (say, due to flooding), it must set skb->offload_fwd_mark to
something.
- If we set it to 0, then the bridge will forward it towards swp1, swp2
and bond0. But the switch has already forwarded it towards swp1 and
swp2 (not to bond0, remember, that isn't offloaded, so as far as the
switch is concerned, ports swp3 and swp4 are not looking up the FDB,
and the entire bond0 is a destination that is strictly behind the
CPU). But we don't want duplicated traffic towards swp1 and swp2, so
it's not ok to set skb->offload_fwd_mark = 0.
- If we set it to 1, then the bridge will not forward the skb towards
the ports with the same switchdev mark, i.e. not to swp1, swp2 and
bond0. Towards swp1 and swp2 that's ok, but towards bond0? It should
have forwarded the skb there.
So the real issue is that bond0 will be assigned the same hardware
domain as {swp0,swp1,swp2}, because the function that assigns hardware
domains to bridge ports, nbp_switchdev_add(), recurses through bond0's
lower interfaces until it finds something that implements devlink (calls
dev_get_port_parent_id with bool recurse = true). This is a problem
because the fact that bond0 can be offloaded by swp3 and swp4 in our
example is merely an assumption.
A solution is to give the bridge explicit hints as to what hardware
domain it should use for each port.
Currently, the bridging offload is very 'silent': a driver registers a
netdevice notifier, which is put on the netns's notifier chain, and
which sniffs around for NETDEV_CHANGEUPPER events where the upper is a
bridge, and the lower is an interface it knows about (one registered by
this driver, normally). Then, from within that notifier, it does a bunch
of stuff behind the bridge's back, without the bridge necessarily
knowing that there's somebody offloading that port. It looks like this:
ip link set swp0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v
call_netdevice_notifiers
|
v
dsa_slave_netdevice_event
|
v
oh, hey! it's for me!
|
v
.port_bridge_join
What we do to solve the conundrum is to be less silent, and change the
switchdev drivers to present themselves to the bridge. Something like this:
ip link set swp0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge: Aye! I'll use this
call_netdevice_notifiers ^ ppid as the
| | hardware domain for
v | this port, and zero
dsa_slave_netdevice_event | if I got nothing.
| |
v |
oh, hey! it's for me! |
| |
v |
.port_bridge_join |
| |
+------------------------+
switchdev_bridge_port_offload(swp0, swp0)
Then stacked interfaces (like bond0 on top of swp3/swp4) would be
treated differently in DSA, depending on whether we can or cannot
offload them.
The offload case:
ip link set bond0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge: Aye! I'll use this
call_netdevice_notifiers ^ ppid as the
| | switchdev mark for
v | bond0.
dsa_slave_netdevice_event | Coincidentally (or not),
| | bond0 and swp0, swp1, swp2
v | all have the same switchdev
hmm, it's not quite for me, | mark now, since the ASIC
but my driver has already | is able to forward towards
called .port_lag_join | all these ports in hw.
for it, because I have |
a port with dp->lag_dev == bond0. |
| |
v |
.port_bridge_join |
for swp3 and swp4 |
| |
+------------------------+
switchdev_bridge_port_offload(bond0, swp3)
switchdev_bridge_port_offload(bond0, swp4)
And the non-offload case:
ip link set bond0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge waiting:
call_netdevice_notifiers ^ huh, switchdev_bridge_port_offload
| | wasn't called, okay, I'll use a
v | hwdom of zero for this one.
dsa_slave_netdevice_event : Then packets received on swp0 will
| : not be software-forwarded towards
v : swp1, but they will towards bond0.
it's not for me, but
bond0 is an upper of swp3
and swp4, but their dp->lag_dev
is NULL because they couldn't
offload it.
Basically we can draw the conclusion that the lowers of a bridge port
can come and go, so depending on the configuration of lowers for a
bridge port, it can dynamically toggle between offloaded and unoffloaded.
Therefore, we need an equivalent switchdev_bridge_port_unoffload too.
This patch changes the way any switchdev driver interacts with the
bridge. From now on, everybody needs to call switchdev_bridge_port_offload
and switchdev_bridge_port_unoffload, otherwise the bridge will treat the
port as non-offloaded and allow software flooding to other ports from
the same ASIC.
Note that these functions lay the ground for a more complex handshake
between switchdev drivers and the bridge in the future.
For drivers that will request a replay of the switchdev objects when
they offload and unoffload a bridge port (DSA, dpaa2-switch, ocelot), we
place the call to switchdev_bridge_port_unoffload() strategically inside
the NETDEV_PRECHANGEUPPER notifier's code path, and not inside
NETDEV_CHANGEUPPER. This is because the switchdev object replay helpers
need the netdev adjacency lists to be valid, and that is only true in
NETDEV_PRECHANGEUPPER.
Cc: Vadym Kochan <vkochan@marvell.com>
Cc: Taras Chornyi <tchornyi@marvell.com>
Cc: Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Cc: Claudiu Manoil <claudiu.manoil@nxp.com>
Cc: Alexandre Belloni <alexandre.belloni@bootlin.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Tested-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch: regression
Acked-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch
Tested-by: Horatiu Vultur <horatiu.vultur@microchip.com> # ocelot-switch
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 00:24:01 +08:00
|
|
|
out_rollback_unoffload:
|
net: bridge: move the switchdev object replay helpers to "push" mode
Starting with commit 4f2673b3a2b6 ("net: bridge: add helper to replay
port and host-joined mdb entries"), DSA has introduced some bridge
helpers that replay switchdev events (FDB/MDB/VLAN additions and
deletions) that can be lost by the switchdev drivers in a variety of
circumstances:
- an IP multicast group was host-joined on the bridge itself before any
switchdev port joined the bridge, leading to the host MDB entries
missing in the hardware database.
- during the bridge creation process, the MAC address of the bridge was
added to the FDB as an entry pointing towards the bridge device
itself, but with no switchdev ports being part of the bridge yet, this
local FDB entry would remain unknown to the switchdev hardware
database.
- a VLAN/FDB/MDB was added to a bridge port that is a LAG interface,
before any switchdev port joined that LAG, leading to the hardware
database missing those entries.
- a switchdev port left a LAG that is a bridge port, while the LAG
remained part of the bridge, and all FDB/MDB/VLAN entries remained
installed in the hardware database of the switchdev port.
Also, since commit 0d2cfbd41c4a ("net: bridge: ignore switchdev events
for LAG ports which didn't request replay"), DSA introduced a method,
based on a const void *ctx, to ensure that two switchdev ports under the
same LAG that is a bridge port do not see the same MDB/VLAN entry being
replayed twice by the bridge, once for every bridge port that joins the
LAG.
With so many ordering corner cases being possible, it seems unreasonable
to expect a switchdev driver writer to get it right from the first try.
Therefore, now that DSA has experimented with the bridge replay helpers
for a little bit, we can move the code to the bridge driver where it is
more readily available to all switchdev drivers.
To convert the switchdev object replay helpers from "pull mode" (where
the driver asks for them) to a "push mode" (where the bridge offers them
automatically), the biggest problem is that the bridge needs to be aware
when a switchdev port joins and leaves, even when the switchdev is only
indirectly a bridge port (for example when the bridge port is a LAG
upper of the switchdev).
Luckily, we already have a hook for that, in the form of the newly
introduced switchdev_bridge_port_offload() and
switchdev_bridge_port_unoffload() calls. These offer a natural place for
hooking the object addition and deletion replays.
Extend the above 2 functions with:
- pointers to the switchdev atomic notifier (for FDB replays) and the
blocking notifier (for MDB and VLAN replays).
- the "const void *ctx" argument required for drivers to be able to
disambiguate between which port is targeted, when multiple ports are
lowers of the same LAG that is a bridge port. Most of the drivers pass
NULL to this argument, except the ones that support LAG offload and have
the proper context check already in place in the switchdev blocking
notifier handler.
Also unexport the replay helpers, since nobody except the bridge calls
them directly now.
Note that:
(a) we abuse the terminology slightly, because FDB entries are not
"switchdev objects", but we count them as objects nonetheless.
With no direct way to prove it, I think they are not modeled as
switchdev objects because those can only be installed by the bridge
to the hardware (as opposed to FDB entries which can be propagated
in the other direction too). This is merely an abuse of terms, FDB
entries are replayed too, despite not being objects.
(b) the bridge does not attempt to sync port attributes to newly joined
ports, just the countable stuff (the objects). The reason for this
is simple: no universal and symmetric way to sync and unsync them is
known. For example, VLAN filtering: what to do on unsync, disable or
leave it enabled? Similarly, STP state, ageing timer, etc etc. What
a switchdev port does when it becomes standalone again is not really
up to the bridge's competence, and the driver should deal with it.
On the other hand, replaying deletions of switchdev objects can be
seen a matter of cleanup and therefore be treated by the bridge,
hence this patch.
We make the replay helpers opt-in for drivers, because they might not
bring immediate benefits for them:
- nbp_vlan_init() is called _after_ netdev_master_upper_dev_link(),
so br_vlan_replay() should not do anything for the new drivers on
which we call it. The existing drivers where there was even a slight
possibility for there to exist a VLAN on a bridge port before they
join it are already guarded against this: mlxsw and prestera deny
joining LAG interfaces that are members of a bridge.
- br_fdb_replay() should now notify of local FDB entries, but I patched
all drivers except DSA to ignore these new entries in commit
2c4eca3ef716 ("net: bridge: switchdev: include local flag in FDB
notifications"). Driver authors can lift this restriction as they
wish, and when they do, they can also opt into the FDB replay
functionality.
- br_mdb_replay() should fix a real issue which is described in commit
4f2673b3a2b6 ("net: bridge: add helper to replay port and host-joined
mdb entries"). However most drivers do not offload the
SWITCHDEV_OBJ_ID_HOST_MDB to see this issue: only cpsw and am65_cpsw
offload this switchdev object, and I don't completely understand the
way in which they offload this switchdev object anyway. So I'll leave
it up to these drivers' respective maintainers to opt into
br_mdb_replay().
So most of the drivers pass NULL notifier blocks for the replay helpers,
except:
- dpaa2-switch which was already acked/regression-tested with the
helpers enabled (and there isn't much of a downside in having them)
- ocelot which already had replay logic in "pull" mode
- DSA which already had replay logic in "pull" mode
An important observation is that the drivers which don't currently
request bridge event replays don't even have the
switchdev_bridge_port_{offload,unoffload} calls placed in proper places
right now. This was done to avoid unnecessary rework for drivers which
might never even add support for this. For driver writers who wish to
add replay support, this can be used as a tentative placement guide:
https://patchwork.kernel.org/project/netdevbpf/patch/20210720134655.892334-11-vladimir.oltean@nxp.com/
Cc: Vadym Kochan <vkochan@marvell.com>
Cc: Taras Chornyi <tchornyi@marvell.com>
Cc: Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Cc: Claudiu Manoil <claudiu.manoil@nxp.com>
Cc: Alexandre Belloni <alexandre.belloni@bootlin.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 00:24:03 +08:00
|
|
|
switchdev_bridge_port_unoffload(brport_dev, dp,
|
2023-10-24 02:17:28 +08:00
|
|
|
&dsa_user_switchdev_notifier,
|
|
|
|
|
&dsa_user_switchdev_blocking_notifier);
|
2022-05-07 21:45:50 +08:00
|
|
|
dsa_flush_workqueue();
|
2021-03-23 07:51:49 +08:00
|
|
|
out_rollback_unbridge:
|
|
|
|
|
dsa_broadcast(DSA_NOTIFIER_BRIDGE_LEAVE, &info);
|
|
|
|
|
out_rollback:
|
2021-12-07 00:57:48 +08:00
|
|
|
dsa_port_bridge_destroy(dp, br);
|
2017-05-20 05:00:45 +08:00
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
net: bridge: move the switchdev object replay helpers to "push" mode
Starting with commit 4f2673b3a2b6 ("net: bridge: add helper to replay
port and host-joined mdb entries"), DSA has introduced some bridge
helpers that replay switchdev events (FDB/MDB/VLAN additions and
deletions) that can be lost by the switchdev drivers in a variety of
circumstances:
- an IP multicast group was host-joined on the bridge itself before any
switchdev port joined the bridge, leading to the host MDB entries
missing in the hardware database.
- during the bridge creation process, the MAC address of the bridge was
added to the FDB as an entry pointing towards the bridge device
itself, but with no switchdev ports being part of the bridge yet, this
local FDB entry would remain unknown to the switchdev hardware
database.
- a VLAN/FDB/MDB was added to a bridge port that is a LAG interface,
before any switchdev port joined that LAG, leading to the hardware
database missing those entries.
- a switchdev port left a LAG that is a bridge port, while the LAG
remained part of the bridge, and all FDB/MDB/VLAN entries remained
installed in the hardware database of the switchdev port.
Also, since commit 0d2cfbd41c4a ("net: bridge: ignore switchdev events
for LAG ports which didn't request replay"), DSA introduced a method,
based on a const void *ctx, to ensure that two switchdev ports under the
same LAG that is a bridge port do not see the same MDB/VLAN entry being
replayed twice by the bridge, once for every bridge port that joins the
LAG.
With so many ordering corner cases being possible, it seems unreasonable
to expect a switchdev driver writer to get it right from the first try.
Therefore, now that DSA has experimented with the bridge replay helpers
for a little bit, we can move the code to the bridge driver where it is
more readily available to all switchdev drivers.
To convert the switchdev object replay helpers from "pull mode" (where
the driver asks for them) to a "push mode" (where the bridge offers them
automatically), the biggest problem is that the bridge needs to be aware
when a switchdev port joins and leaves, even when the switchdev is only
indirectly a bridge port (for example when the bridge port is a LAG
upper of the switchdev).
Luckily, we already have a hook for that, in the form of the newly
introduced switchdev_bridge_port_offload() and
switchdev_bridge_port_unoffload() calls. These offer a natural place for
hooking the object addition and deletion replays.
Extend the above 2 functions with:
- pointers to the switchdev atomic notifier (for FDB replays) and the
blocking notifier (for MDB and VLAN replays).
- the "const void *ctx" argument required for drivers to be able to
disambiguate between which port is targeted, when multiple ports are
lowers of the same LAG that is a bridge port. Most of the drivers pass
NULL to this argument, except the ones that support LAG offload and have
the proper context check already in place in the switchdev blocking
notifier handler.
Also unexport the replay helpers, since nobody except the bridge calls
them directly now.
Note that:
(a) we abuse the terminology slightly, because FDB entries are not
"switchdev objects", but we count them as objects nonetheless.
With no direct way to prove it, I think they are not modeled as
switchdev objects because those can only be installed by the bridge
to the hardware (as opposed to FDB entries which can be propagated
in the other direction too). This is merely an abuse of terms, FDB
entries are replayed too, despite not being objects.
(b) the bridge does not attempt to sync port attributes to newly joined
ports, just the countable stuff (the objects). The reason for this
is simple: no universal and symmetric way to sync and unsync them is
known. For example, VLAN filtering: what to do on unsync, disable or
leave it enabled? Similarly, STP state, ageing timer, etc etc. What
a switchdev port does when it becomes standalone again is not really
up to the bridge's competence, and the driver should deal with it.
On the other hand, replaying deletions of switchdev objects can be
seen a matter of cleanup and therefore be treated by the bridge,
hence this patch.
We make the replay helpers opt-in for drivers, because they might not
bring immediate benefits for them:
- nbp_vlan_init() is called _after_ netdev_master_upper_dev_link(),
so br_vlan_replay() should not do anything for the new drivers on
which we call it. The existing drivers where there was even a slight
possibility for there to exist a VLAN on a bridge port before they
join it are already guarded against this: mlxsw and prestera deny
joining LAG interfaces that are members of a bridge.
- br_fdb_replay() should now notify of local FDB entries, but I patched
all drivers except DSA to ignore these new entries in commit
2c4eca3ef716 ("net: bridge: switchdev: include local flag in FDB
notifications"). Driver authors can lift this restriction as they
wish, and when they do, they can also opt into the FDB replay
functionality.
- br_mdb_replay() should fix a real issue which is described in commit
4f2673b3a2b6 ("net: bridge: add helper to replay port and host-joined
mdb entries"). However most drivers do not offload the
SWITCHDEV_OBJ_ID_HOST_MDB to see this issue: only cpsw and am65_cpsw
offload this switchdev object, and I don't completely understand the
way in which they offload this switchdev object anyway. So I'll leave
it up to these drivers' respective maintainers to opt into
br_mdb_replay().
So most of the drivers pass NULL notifier blocks for the replay helpers,
except:
- dpaa2-switch which was already acked/regression-tested with the
helpers enabled (and there isn't much of a downside in having them)
- ocelot which already had replay logic in "pull" mode
- DSA which already had replay logic in "pull" mode
An important observation is that the drivers which don't currently
request bridge event replays don't even have the
switchdev_bridge_port_{offload,unoffload} calls placed in proper places
right now. This was done to avoid unnecessary rework for drivers which
might never even add support for this. For driver writers who wish to
add replay support, this can be used as a tentative placement guide:
https://patchwork.kernel.org/project/netdevbpf/patch/20210720134655.892334-11-vladimir.oltean@nxp.com/
Cc: Vadym Kochan <vkochan@marvell.com>
Cc: Taras Chornyi <tchornyi@marvell.com>
Cc: Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Cc: Claudiu Manoil <claudiu.manoil@nxp.com>
Cc: Alexandre Belloni <alexandre.belloni@bootlin.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 00:24:03 +08:00
|
|
|
void dsa_port_pre_bridge_leave(struct dsa_port *dp, struct net_device *br)
|
2021-06-27 19:54:29 +08:00
|
|
|
{
|
net: bridge: switchdev: let drivers inform which bridge ports are offloaded
On reception of an skb, the bridge checks if it was marked as 'already
forwarded in hardware' (checks if skb->offload_fwd_mark == 1), and if it
is, it assigns the source hardware domain of that skb based on the
hardware domain of the ingress port. Then during forwarding, it enforces
that the egress port must have a different hardware domain than the
ingress one (this is done in nbp_switchdev_allowed_egress).
Non-switchdev drivers don't report any physical switch id (neither
through devlink nor .ndo_get_port_parent_id), therefore the bridge
assigns them a hardware domain of 0, and packets coming from them will
always have skb->offload_fwd_mark = 0. So there aren't any restrictions.
Problems appear due to the fact that DSA would like to perform software
fallback for bonding and team interfaces that the physical switch cannot
offload.
+-- br0 ---+
/ / | \
/ / | \
/ | | bond0
/ | | / \
swp0 swp1 swp2 swp3 swp4
There, it is desirable that the presence of swp3 and swp4 under a
non-offloaded LAG does not preclude us from doing hardware bridging
beteen swp0, swp1 and swp2. The bandwidth of the CPU is often times high
enough that software bridging between {swp0,swp1,swp2} and bond0 is not
impractical.
But this creates an impossible paradox given the current way in which
port hardware domains are assigned. When the driver receives a packet
from swp0 (say, due to flooding), it must set skb->offload_fwd_mark to
something.
- If we set it to 0, then the bridge will forward it towards swp1, swp2
and bond0. But the switch has already forwarded it towards swp1 and
swp2 (not to bond0, remember, that isn't offloaded, so as far as the
switch is concerned, ports swp3 and swp4 are not looking up the FDB,
and the entire bond0 is a destination that is strictly behind the
CPU). But we don't want duplicated traffic towards swp1 and swp2, so
it's not ok to set skb->offload_fwd_mark = 0.
- If we set it to 1, then the bridge will not forward the skb towards
the ports with the same switchdev mark, i.e. not to swp1, swp2 and
bond0. Towards swp1 and swp2 that's ok, but towards bond0? It should
have forwarded the skb there.
So the real issue is that bond0 will be assigned the same hardware
domain as {swp0,swp1,swp2}, because the function that assigns hardware
domains to bridge ports, nbp_switchdev_add(), recurses through bond0's
lower interfaces until it finds something that implements devlink (calls
dev_get_port_parent_id with bool recurse = true). This is a problem
because the fact that bond0 can be offloaded by swp3 and swp4 in our
example is merely an assumption.
A solution is to give the bridge explicit hints as to what hardware
domain it should use for each port.
Currently, the bridging offload is very 'silent': a driver registers a
netdevice notifier, which is put on the netns's notifier chain, and
which sniffs around for NETDEV_CHANGEUPPER events where the upper is a
bridge, and the lower is an interface it knows about (one registered by
this driver, normally). Then, from within that notifier, it does a bunch
of stuff behind the bridge's back, without the bridge necessarily
knowing that there's somebody offloading that port. It looks like this:
ip link set swp0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v
call_netdevice_notifiers
|
v
dsa_slave_netdevice_event
|
v
oh, hey! it's for me!
|
v
.port_bridge_join
What we do to solve the conundrum is to be less silent, and change the
switchdev drivers to present themselves to the bridge. Something like this:
ip link set swp0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge: Aye! I'll use this
call_netdevice_notifiers ^ ppid as the
| | hardware domain for
v | this port, and zero
dsa_slave_netdevice_event | if I got nothing.
| |
v |
oh, hey! it's for me! |
| |
v |
.port_bridge_join |
| |
+------------------------+
switchdev_bridge_port_offload(swp0, swp0)
Then stacked interfaces (like bond0 on top of swp3/swp4) would be
treated differently in DSA, depending on whether we can or cannot
offload them.
The offload case:
ip link set bond0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge: Aye! I'll use this
call_netdevice_notifiers ^ ppid as the
| | switchdev mark for
v | bond0.
dsa_slave_netdevice_event | Coincidentally (or not),
| | bond0 and swp0, swp1, swp2
v | all have the same switchdev
hmm, it's not quite for me, | mark now, since the ASIC
but my driver has already | is able to forward towards
called .port_lag_join | all these ports in hw.
for it, because I have |
a port with dp->lag_dev == bond0. |
| |
v |
.port_bridge_join |
for swp3 and swp4 |
| |
+------------------------+
switchdev_bridge_port_offload(bond0, swp3)
switchdev_bridge_port_offload(bond0, swp4)
And the non-offload case:
ip link set bond0 master br0
|
v
br_add_if() calls netdev_master_upper_dev_link()
|
v bridge waiting:
call_netdevice_notifiers ^ huh, switchdev_bridge_port_offload
| | wasn't called, okay, I'll use a
v | hwdom of zero for this one.
dsa_slave_netdevice_event : Then packets received on swp0 will
| : not be software-forwarded towards
v : swp1, but they will towards bond0.
it's not for me, but
bond0 is an upper of swp3
and swp4, but their dp->lag_dev
is NULL because they couldn't
offload it.
Basically we can draw the conclusion that the lowers of a bridge port
can come and go, so depending on the configuration of lowers for a
bridge port, it can dynamically toggle between offloaded and unoffloaded.
Therefore, we need an equivalent switchdev_bridge_port_unoffload too.
This patch changes the way any switchdev driver interacts with the
bridge. From now on, everybody needs to call switchdev_bridge_port_offload
and switchdev_bridge_port_unoffload, otherwise the bridge will treat the
port as non-offloaded and allow software flooding to other ports from
the same ASIC.
Note that these functions lay the ground for a more complex handshake
between switchdev drivers and the bridge in the future.
For drivers that will request a replay of the switchdev objects when
they offload and unoffload a bridge port (DSA, dpaa2-switch, ocelot), we
place the call to switchdev_bridge_port_unoffload() strategically inside
the NETDEV_PRECHANGEUPPER notifier's code path, and not inside
NETDEV_CHANGEUPPER. This is because the switchdev object replay helpers
need the netdev adjacency lists to be valid, and that is only true in
NETDEV_PRECHANGEUPPER.
Cc: Vadym Kochan <vkochan@marvell.com>
Cc: Taras Chornyi <tchornyi@marvell.com>
Cc: Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Cc: Claudiu Manoil <claudiu.manoil@nxp.com>
Cc: Alexandre Belloni <alexandre.belloni@bootlin.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Tested-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch: regression
Acked-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch
Tested-by: Horatiu Vultur <horatiu.vultur@microchip.com> # ocelot-switch
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 00:24:01 +08:00
|
|
|
struct net_device *brport_dev = dsa_port_to_bridge_port(dp);
|
|
|
|
|
|
2021-08-24 05:22:55 +08:00
|
|
|
/* Don't try to unoffload something that is not offloaded */
|
|
|
|
|
if (!brport_dev)
|
|
|
|
|
return;
|
|
|
|
|
|
net: bridge: move the switchdev object replay helpers to "push" mode
Starting with commit 4f2673b3a2b6 ("net: bridge: add helper to replay
port and host-joined mdb entries"), DSA has introduced some bridge
helpers that replay switchdev events (FDB/MDB/VLAN additions and
deletions) that can be lost by the switchdev drivers in a variety of
circumstances:
- an IP multicast group was host-joined on the bridge itself before any
switchdev port joined the bridge, leading to the host MDB entries
missing in the hardware database.
- during the bridge creation process, the MAC address of the bridge was
added to the FDB as an entry pointing towards the bridge device
itself, but with no switchdev ports being part of the bridge yet, this
local FDB entry would remain unknown to the switchdev hardware
database.
- a VLAN/FDB/MDB was added to a bridge port that is a LAG interface,
before any switchdev port joined that LAG, leading to the hardware
database missing those entries.
- a switchdev port left a LAG that is a bridge port, while the LAG
remained part of the bridge, and all FDB/MDB/VLAN entries remained
installed in the hardware database of the switchdev port.
Also, since commit 0d2cfbd41c4a ("net: bridge: ignore switchdev events
for LAG ports which didn't request replay"), DSA introduced a method,
based on a const void *ctx, to ensure that two switchdev ports under the
same LAG that is a bridge port do not see the same MDB/VLAN entry being
replayed twice by the bridge, once for every bridge port that joins the
LAG.
With so many ordering corner cases being possible, it seems unreasonable
to expect a switchdev driver writer to get it right from the first try.
Therefore, now that DSA has experimented with the bridge replay helpers
for a little bit, we can move the code to the bridge driver where it is
more readily available to all switchdev drivers.
To convert the switchdev object replay helpers from "pull mode" (where
the driver asks for them) to a "push mode" (where the bridge offers them
automatically), the biggest problem is that the bridge needs to be aware
when a switchdev port joins and leaves, even when the switchdev is only
indirectly a bridge port (for example when the bridge port is a LAG
upper of the switchdev).
Luckily, we already have a hook for that, in the form of the newly
introduced switchdev_bridge_port_offload() and
switchdev_bridge_port_unoffload() calls. These offer a natural place for
hooking the object addition and deletion replays.
Extend the above 2 functions with:
- pointers to the switchdev atomic notifier (for FDB replays) and the
blocking notifier (for MDB and VLAN replays).
- the "const void *ctx" argument required for drivers to be able to
disambiguate between which port is targeted, when multiple ports are
lowers of the same LAG that is a bridge port. Most of the drivers pass
NULL to this argument, except the ones that support LAG offload and have
the proper context check already in place in the switchdev blocking
notifier handler.
Also unexport the replay helpers, since nobody except the bridge calls
them directly now.
Note that:
(a) we abuse the terminology slightly, because FDB entries are not
"switchdev objects", but we count them as objects nonetheless.
With no direct way to prove it, I think they are not modeled as
switchdev objects because those can only be installed by the bridge
to the hardware (as opposed to FDB entries which can be propagated
in the other direction too). This is merely an abuse of terms, FDB
entries are replayed too, despite not being objects.
(b) the bridge does not attempt to sync port attributes to newly joined
ports, just the countable stuff (the objects). The reason for this
is simple: no universal and symmetric way to sync and unsync them is
known. For example, VLAN filtering: what to do on unsync, disable or
leave it enabled? Similarly, STP state, ageing timer, etc etc. What
a switchdev port does when it becomes standalone again is not really
up to the bridge's competence, and the driver should deal with it.
On the other hand, replaying deletions of switchdev objects can be
seen a matter of cleanup and therefore be treated by the bridge,
hence this patch.
We make the replay helpers opt-in for drivers, because they might not
bring immediate benefits for them:
- nbp_vlan_init() is called _after_ netdev_master_upper_dev_link(),
so br_vlan_replay() should not do anything for the new drivers on
which we call it. The existing drivers where there was even a slight
possibility for there to exist a VLAN on a bridge port before they
join it are already guarded against this: mlxsw and prestera deny
joining LAG interfaces that are members of a bridge.
- br_fdb_replay() should now notify of local FDB entries, but I patched
all drivers except DSA to ignore these new entries in commit
2c4eca3ef716 ("net: bridge: switchdev: include local flag in FDB
notifications"). Driver authors can lift this restriction as they
wish, and when they do, they can also opt into the FDB replay
functionality.
- br_mdb_replay() should fix a real issue which is described in commit
4f2673b3a2b6 ("net: bridge: add helper to replay port and host-joined
mdb entries"). However most drivers do not offload the
SWITCHDEV_OBJ_ID_HOST_MDB to see this issue: only cpsw and am65_cpsw
offload this switchdev object, and I don't completely understand the
way in which they offload this switchdev object anyway. So I'll leave
it up to these drivers' respective maintainers to opt into
br_mdb_replay().
So most of the drivers pass NULL notifier blocks for the replay helpers,
except:
- dpaa2-switch which was already acked/regression-tested with the
helpers enabled (and there isn't much of a downside in having them)
- ocelot which already had replay logic in "pull" mode
- DSA which already had replay logic in "pull" mode
An important observation is that the drivers which don't currently
request bridge event replays don't even have the
switchdev_bridge_port_{offload,unoffload} calls placed in proper places
right now. This was done to avoid unnecessary rework for drivers which
might never even add support for this. For driver writers who wish to
add replay support, this can be used as a tentative placement guide:
https://patchwork.kernel.org/project/netdevbpf/patch/20210720134655.892334-11-vladimir.oltean@nxp.com/
Cc: Vadym Kochan <vkochan@marvell.com>
Cc: Taras Chornyi <tchornyi@marvell.com>
Cc: Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: Steen Hegelund <Steen.Hegelund@microchip.com>
Cc: UNGLinuxDriver@microchip.com
Cc: Claudiu Manoil <claudiu.manoil@nxp.com>
Cc: Alexandre Belloni <alexandre.belloni@bootlin.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Ioana Ciornei <ioana.ciornei@nxp.com> # dpaa2-switch
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-22 00:24:03 +08:00
|
|
|
switchdev_bridge_port_unoffload(brport_dev, dp,
|
2023-10-24 02:17:28 +08:00
|
|
|
&dsa_user_switchdev_notifier,
|
|
|
|
|
&dsa_user_switchdev_blocking_notifier);
|
net: dsa: flush switchdev workqueue when leaving the bridge
DSA is preparing to offer switch drivers an API through which they can
associate each FDB entry with a struct net_device *bridge_dev. This can
be used to perform FDB isolation (the FDB lookup performed on the
ingress of a standalone, or bridged port, should not find an FDB entry
that is present in the FDB of another bridge).
In preparation of that work, DSA needs to ensure that by the time we
call the switch .port_fdb_add and .port_fdb_del methods, the
dp->bridge_dev pointer is still valid, i.e. the port is still a bridge
port.
This is not guaranteed because the SWITCHDEV_FDB_{ADD,DEL}_TO_DEVICE API
requires drivers that must have sleepable context to handle those events
to schedule the deferred work themselves. DSA does this through the
dsa_owq.
It can happen that a port leaves a bridge, del_nbp() flushes the FDB on
that port, SWITCHDEV_FDB_DEL_TO_DEVICE is notified in atomic context,
DSA schedules its deferred work, but del_nbp() finishes unlinking the
bridge as a master from the port before DSA's deferred work is run.
Fundamentally, the port must not be unlinked from the bridge until all
FDB deletion deferred work items have been flushed. The bridge must wait
for the completion of these hardware accesses.
An attempt has been made to address this issue centrally in switchdev by
making SWITCHDEV_FDB_DEL_TO_DEVICE deferred (=> blocking) at the switchdev
level, which would offer implicit synchronization with del_nbp:
https://patchwork.kernel.org/project/netdevbpf/cover/20210820115746.3701811-1-vladimir.oltean@nxp.com/
but it seems that any attempt to modify switchdev's behavior and make
the events blocking there would introduce undesirable side effects in
other switchdev consumers.
The most undesirable behavior seems to be that
switchdev_deferred_process_work() takes the rtnl_mutex itself, which
would be worse off than having the rtnl_mutex taken individually from
drivers which is what we have now (except DSA which has removed that
lock since commit 0faf890fc519 ("net: dsa: drop rtnl_lock from
dsa_slave_switchdev_event_work")).
So to offer the needed guarantee to DSA switch drivers, I have come up
with a compromise solution that does not require switchdev rework:
we already have a hook at the last moment in time when the bridge is
still an upper of ours: the NETDEV_PRECHANGEUPPER handler. We can flush
the dsa_owq manually from there, which makes all FDB deletions
synchronous.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-10-26 17:25:55 +08:00
|
|
|
|
|
|
|
|
dsa_flush_workqueue();
|
2021-06-27 19:54:29 +08:00
|
|
|
}
|
|
|
|
|
|
2017-05-20 05:00:45 +08:00
|
|
|
void dsa_port_bridge_leave(struct dsa_port *dp, struct net_device *br)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_notifier_bridge_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2017-05-20 05:00:45 +08:00
|
|
|
};
|
|
|
|
|
int err;
|
|
|
|
|
|
2022-02-22 04:35:38 +08:00
|
|
|
/* If the port could not be offloaded to begin with, then
|
|
|
|
|
* there is nothing to do.
|
|
|
|
|
*/
|
|
|
|
|
if (!dp->bridge)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
info.bridge = *dp->bridge;
|
|
|
|
|
|
2017-05-20 05:00:45 +08:00
|
|
|
/* Here the port is already unbridged. Reflect the current configuration
|
|
|
|
|
* so that drivers can program their chips accordingly.
|
|
|
|
|
*/
|
2021-12-07 00:57:48 +08:00
|
|
|
dsa_port_bridge_destroy(dp, br);
|
2017-05-20 05:00:45 +08:00
|
|
|
|
net: dsa: permit cross-chip bridging between all trees in the system
One way of utilizing DSA is by cascading switches which do not all have
compatible taggers. Consider the following real-life topology:
+---------------------------------------------------------------+
| LS1028A |
| +------------------------------+ |
| | DSA master for Felix | |
| |(internal ENETC port 2: eno2))| |
| +------------+------------------------------+-------------+ |
| | Felix embedded L2 switch | |
| | | |
| | +--------------+ +--------------+ +--------------+ | |
| | |DSA master for| |DSA master for| |DSA master for| | |
| | | SJA1105 1 | | SJA1105 2 | | SJA1105 3 | | |
| | |(Felix port 1)| |(Felix port 2)| |(Felix port 3)| | |
+--+-+--------------+---+--------------+---+--------------+--+--+
+-----------------------+ +-----------------------+ +-----------------------+
| SJA1105 switch 1 | | SJA1105 switch 2 | | SJA1105 switch 3 |
+-----+-----+-----+-----+ +-----+-----+-----+-----+ +-----+-----+-----+-----+
|sw1p0|sw1p1|sw1p2|sw1p3| |sw2p0|sw2p1|sw2p2|sw2p3| |sw3p0|sw3p1|sw3p2|sw3p3|
+-----+-----+-----+-----+ +-----+-----+-----+-----+ +-----+-----+-----+-----+
The above can be described in the device tree as follows (obviously not
complete):
mscc_felix {
dsa,member = <0 0>;
ports {
port@4 {
ethernet = <&enetc_port2>;
};
};
};
sja1105_switch1 {
dsa,member = <1 1>;
ports {
port@4 {
ethernet = <&mscc_felix_port1>;
};
};
};
sja1105_switch2 {
dsa,member = <2 2>;
ports {
port@4 {
ethernet = <&mscc_felix_port2>;
};
};
};
sja1105_switch3 {
dsa,member = <3 3>;
ports {
port@4 {
ethernet = <&mscc_felix_port3>;
};
};
};
Basically we instantiate one DSA switch tree for every hardware switch
in the system, but we still give them globally unique switch IDs (will
come back to that later). Having 3 disjoint switch trees makes the
tagger drivers "just work", because net devices are registered for the
3 Felix DSA master ports, and they are also DSA slave ports to the ENETC
port. So packets received on the ENETC port are stripped of their
stacked DSA tags one by one.
Currently, hardware bridging between ports on the same sja1105 chip is
possible, but switching between sja1105 ports on different chips is
handled by the software bridge. This is fine, but we can do better.
In fact, the dsa_8021q tag used by sja1105 is compatible with cascading.
In other words, a sja1105 switch can correctly parse and route a packet
containing a dsa_8021q tag. So if we could enable hardware bridging on
the Felix DSA master ports, cross-chip bridging could be completely
offloaded.
Such as system would be used as follows:
ip link add dev br0 type bridge && ip link set dev br0 up
for port in sw0p0 sw0p1 sw0p2 sw0p3 \
sw1p0 sw1p1 sw1p2 sw1p3 \
sw2p0 sw2p1 sw2p2 sw2p3; do
ip link set dev $port master br0
done
The above makes switching between ports on the same row be performed in
hardware, and between ports on different rows in software. Now assume
the Felix switch ports are called swp0, swp1, swp2. By running the
following extra commands:
ip link add dev br1 type bridge && ip link set dev br1 up
for port in swp0 swp1 swp2; do
ip link set dev $port master br1
done
the CPU no longer sees packets which traverse sja1105 switch boundaries
and can be forwarded directly by Felix. The br1 bridge would not be used
for any sort of traffic termination.
For this to work, we need to give drivers an opportunity to listen for
bridging events on DSA trees other than their own, and pass that other
tree index as argument. I have made the assumption, for the moment, that
the other existing DSA notifiers don't need to be broadcast to other
trees. That assumption might turn out to be incorrect. But in the
meantime, introduce a dsa_broadcast function, similar in purpose to
dsa_port_notify, which is used only by the bridging notifiers.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2020-05-11 00:37:41 +08:00
|
|
|
err = dsa_broadcast(DSA_NOTIFIER_BRIDGE_LEAVE, &info);
|
2017-05-20 05:00:45 +08:00
|
|
|
if (err)
|
2021-08-11 21:46:05 +08:00
|
|
|
dev_err(dp->ds->dev,
|
|
|
|
|
"port %d failed to notify DSA_NOTIFIER_BRIDGE_LEAVE: %pe\n",
|
|
|
|
|
dp->index, ERR_PTR(err));
|
2017-05-20 05:00:45 +08:00
|
|
|
|
2022-04-15 23:46:21 +08:00
|
|
|
dsa_port_switchdev_unsync_attrs(dp, info.bridge);
|
2017-05-20 05:00:45 +08:00
|
|
|
}
|
2017-05-20 05:00:46 +08:00
|
|
|
|
2021-01-13 16:42:53 +08:00
|
|
|
int dsa_port_lag_change(struct dsa_port *dp,
|
|
|
|
|
struct netdev_lag_lower_state_info *linfo)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_notifier_lag_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2021-01-13 16:42:53 +08:00
|
|
|
};
|
|
|
|
|
bool tx_enabled;
|
|
|
|
|
|
2022-02-23 22:00:49 +08:00
|
|
|
if (!dp->lag)
|
2021-01-13 16:42:53 +08:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
/* On statically configured aggregates (e.g. loadbalance
|
|
|
|
|
* without LACP) ports will always be tx_enabled, even if the
|
|
|
|
|
* link is down. Thus we require both link_up and tx_enabled
|
|
|
|
|
* in order to include it in the tx set.
|
|
|
|
|
*/
|
|
|
|
|
tx_enabled = linfo->link_up && linfo->tx_enabled;
|
|
|
|
|
|
|
|
|
|
if (tx_enabled == dp->lag_tx_enabled)
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
dp->lag_tx_enabled = tx_enabled;
|
|
|
|
|
|
|
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_LAG_CHANGE, &info);
|
|
|
|
|
}
|
|
|
|
|
|
2022-02-23 22:00:49 +08:00
|
|
|
static int dsa_port_lag_create(struct dsa_port *dp,
|
|
|
|
|
struct net_device *lag_dev)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
struct dsa_lag *lag;
|
|
|
|
|
|
|
|
|
|
lag = dsa_tree_lag_find(ds->dst, lag_dev);
|
|
|
|
|
if (lag) {
|
|
|
|
|
refcount_inc(&lag->refcount);
|
|
|
|
|
dp->lag = lag;
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
lag = kzalloc(sizeof(*lag), GFP_KERNEL);
|
|
|
|
|
if (!lag)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
refcount_set(&lag->refcount, 1);
|
net: dsa: support FDB events on offloaded LAG interfaces
This change introduces support for installing static FDB entries towards
a bridge port that is a LAG of multiple DSA switch ports, as well as
support for filtering towards the CPU local FDB entries emitted for LAG
interfaces that are bridge ports.
Conceptually, host addresses on LAG ports are identical to what we do
for plain bridge ports. Whereas FDB entries _towards_ a LAG can't simply
be replicated towards all member ports like we do for multicast, or VLAN.
Instead we need new driver API. Hardware usually considers a LAG to be a
"logical port", and sets the entire LAG as the forwarding destination.
The physical egress port selection within the LAG is made by hashing
policy, as usual.
To represent the logical port corresponding to the LAG, we pass by value
a copy of the dsa_lag structure to all switches in the tree that have at
least one port in that LAG.
To illustrate why a refcounted list of FDB entries is needed in struct
dsa_lag, it is enough to say that:
- a LAG may be a bridge port and may therefore receive FDB events even
while it isn't yet offloaded by any DSA interface
- DSA interfaces may be removed from a LAG while that is a bridge port;
we don't want FDB entries lingering around, but we don't want to
remove entries that are still in use, either
For all the cases below to work, the idea is to always keep an FDB entry
on a LAG with a reference count equal to the DSA member ports. So:
- if a port joins a LAG, it requests the bridge to replay the FDB, and
the FDB entries get created, or their refcount gets bumped by one
- if a port leaves a LAG, the FDB replay deletes or decrements refcount
by one
- if an FDB is installed towards a LAG with ports already present, that
entry is created (if it doesn't exist) and its refcount is bumped by
the amount of ports already present in the LAG
echo "Adding FDB entry to bond with existing ports"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond, then removing ports one by one"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link set swp1 nomaster
ip link set swp2 nomaster
ip link del br0
ip link del bond0
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-23 22:00:53 +08:00
|
|
|
mutex_init(&lag->fdb_lock);
|
|
|
|
|
INIT_LIST_HEAD(&lag->fdbs);
|
2022-02-23 22:00:49 +08:00
|
|
|
lag->dev = lag_dev;
|
|
|
|
|
dsa_lag_map(ds->dst, lag);
|
|
|
|
|
dp->lag = lag;
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void dsa_port_lag_destroy(struct dsa_port *dp)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_lag *lag = dp->lag;
|
|
|
|
|
|
|
|
|
|
dp->lag = NULL;
|
|
|
|
|
dp->lag_tx_enabled = false;
|
|
|
|
|
|
|
|
|
|
if (!refcount_dec_and_test(&lag->refcount))
|
|
|
|
|
return;
|
|
|
|
|
|
net: dsa: support FDB events on offloaded LAG interfaces
This change introduces support for installing static FDB entries towards
a bridge port that is a LAG of multiple DSA switch ports, as well as
support for filtering towards the CPU local FDB entries emitted for LAG
interfaces that are bridge ports.
Conceptually, host addresses on LAG ports are identical to what we do
for plain bridge ports. Whereas FDB entries _towards_ a LAG can't simply
be replicated towards all member ports like we do for multicast, or VLAN.
Instead we need new driver API. Hardware usually considers a LAG to be a
"logical port", and sets the entire LAG as the forwarding destination.
The physical egress port selection within the LAG is made by hashing
policy, as usual.
To represent the logical port corresponding to the LAG, we pass by value
a copy of the dsa_lag structure to all switches in the tree that have at
least one port in that LAG.
To illustrate why a refcounted list of FDB entries is needed in struct
dsa_lag, it is enough to say that:
- a LAG may be a bridge port and may therefore receive FDB events even
while it isn't yet offloaded by any DSA interface
- DSA interfaces may be removed from a LAG while that is a bridge port;
we don't want FDB entries lingering around, but we don't want to
remove entries that are still in use, either
For all the cases below to work, the idea is to always keep an FDB entry
on a LAG with a reference count equal to the DSA member ports. So:
- if a port joins a LAG, it requests the bridge to replay the FDB, and
the FDB entries get created, or their refcount gets bumped by one
- if a port leaves a LAG, the FDB replay deletes or decrements refcount
by one
- if an FDB is installed towards a LAG with ports already present, that
entry is created (if it doesn't exist) and its refcount is bumped by
the amount of ports already present in the LAG
echo "Adding FDB entry to bond with existing ports"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond, then removing ports one by one"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link set swp1 nomaster
ip link set swp2 nomaster
ip link del br0
ip link del bond0
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-23 22:00:53 +08:00
|
|
|
WARN_ON(!list_empty(&lag->fdbs));
|
2022-02-23 22:00:49 +08:00
|
|
|
dsa_lag_unmap(dp->ds->dst, lag);
|
|
|
|
|
kfree(lag);
|
|
|
|
|
}
|
|
|
|
|
|
2022-02-23 22:00:44 +08:00
|
|
|
int dsa_port_lag_join(struct dsa_port *dp, struct net_device *lag_dev,
|
2021-03-23 07:51:48 +08:00
|
|
|
struct netdev_lag_upper_info *uinfo,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
2021-01-13 16:42:53 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_notifier_lag_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2021-01-13 16:42:53 +08:00
|
|
|
.info = uinfo,
|
2022-09-11 09:07:03 +08:00
|
|
|
.extack = extack,
|
2021-01-13 16:42:53 +08:00
|
|
|
};
|
2021-03-23 07:51:47 +08:00
|
|
|
struct net_device *bridge_dev;
|
2021-01-13 16:42:53 +08:00
|
|
|
int err;
|
|
|
|
|
|
2022-02-23 22:00:49 +08:00
|
|
|
err = dsa_port_lag_create(dp, lag_dev);
|
|
|
|
|
if (err)
|
|
|
|
|
goto err_lag_create;
|
2021-01-13 16:42:53 +08:00
|
|
|
|
2022-02-23 22:00:49 +08:00
|
|
|
info.lag = *dp->lag;
|
2021-01-13 16:42:53 +08:00
|
|
|
err = dsa_port_notify(dp, DSA_NOTIFIER_LAG_JOIN, &info);
|
2021-03-23 07:51:47 +08:00
|
|
|
if (err)
|
|
|
|
|
goto err_lag_join;
|
2021-01-13 16:42:53 +08:00
|
|
|
|
2022-02-23 22:00:44 +08:00
|
|
|
bridge_dev = netdev_master_upper_dev_get(lag_dev);
|
2021-03-23 07:51:47 +08:00
|
|
|
if (!bridge_dev || !netif_is_bridge_master(bridge_dev))
|
|
|
|
|
return 0;
|
|
|
|
|
|
2021-03-23 07:51:48 +08:00
|
|
|
err = dsa_port_bridge_join(dp, bridge_dev, extack);
|
2021-03-23 07:51:47 +08:00
|
|
|
if (err)
|
|
|
|
|
goto err_bridge_join;
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
err_bridge_join:
|
|
|
|
|
dsa_port_notify(dp, DSA_NOTIFIER_LAG_LEAVE, &info);
|
|
|
|
|
err_lag_join:
|
2022-02-23 22:00:49 +08:00
|
|
|
dsa_port_lag_destroy(dp);
|
|
|
|
|
err_lag_create:
|
2021-01-13 16:42:53 +08:00
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
2022-02-23 22:00:44 +08:00
|
|
|
void dsa_port_pre_lag_leave(struct dsa_port *dp, struct net_device *lag_dev)
|
2021-06-27 19:54:29 +08:00
|
|
|
{
|
2021-12-07 00:57:52 +08:00
|
|
|
struct net_device *br = dsa_port_bridge_dev_get(dp);
|
|
|
|
|
|
|
|
|
|
if (br)
|
|
|
|
|
dsa_port_pre_bridge_leave(dp, br);
|
2021-06-27 19:54:29 +08:00
|
|
|
}
|
|
|
|
|
|
2022-02-23 22:00:44 +08:00
|
|
|
void dsa_port_lag_leave(struct dsa_port *dp, struct net_device *lag_dev)
|
2021-01-13 16:42:53 +08:00
|
|
|
{
|
2021-12-07 00:57:52 +08:00
|
|
|
struct net_device *br = dsa_port_bridge_dev_get(dp);
|
2021-01-13 16:42:53 +08:00
|
|
|
struct dsa_notifier_lag_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2021-01-13 16:42:53 +08:00
|
|
|
};
|
|
|
|
|
int err;
|
|
|
|
|
|
2022-02-23 22:00:49 +08:00
|
|
|
if (!dp->lag)
|
2021-01-13 16:42:53 +08:00
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
/* Port might have been part of a LAG that in turn was
|
|
|
|
|
* attached to a bridge.
|
|
|
|
|
*/
|
2021-12-07 00:57:52 +08:00
|
|
|
if (br)
|
|
|
|
|
dsa_port_bridge_leave(dp, br);
|
2021-01-13 16:42:53 +08:00
|
|
|
|
2022-02-23 22:00:49 +08:00
|
|
|
info.lag = *dp->lag;
|
|
|
|
|
|
|
|
|
|
dsa_port_lag_destroy(dp);
|
2021-01-13 16:42:53 +08:00
|
|
|
|
|
|
|
|
err = dsa_port_notify(dp, DSA_NOTIFIER_LAG_LEAVE, &info);
|
|
|
|
|
if (err)
|
2021-08-11 21:46:05 +08:00
|
|
|
dev_err(dp->ds->dev,
|
|
|
|
|
"port %d failed to notify DSA_NOTIFIER_LAG_LEAVE: %pe\n",
|
|
|
|
|
dp->index, ERR_PTR(err));
|
2021-01-13 16:42:53 +08:00
|
|
|
}
|
|
|
|
|
|
2020-09-21 08:10:28 +08:00
|
|
|
/* Must be called under rcu_read_lock() */
|
2019-04-29 02:45:44 +08:00
|
|
|
static bool dsa_port_can_apply_vlan_filtering(struct dsa_port *dp,
|
2021-02-14 04:43:19 +08:00
|
|
|
bool vlan_filtering,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
2019-04-29 02:45:44 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
2021-10-21 01:49:50 +08:00
|
|
|
struct dsa_port *other_dp;
|
|
|
|
|
int err;
|
2020-09-21 08:10:28 +08:00
|
|
|
|
|
|
|
|
/* VLAN awareness was off, so the question is "can we turn it on".
|
|
|
|
|
* We may have had 8021q uppers, those need to go. Make sure we don't
|
|
|
|
|
* enter an inconsistent state: deny changing the VLAN awareness state
|
|
|
|
|
* as long as we have 8021q uppers.
|
|
|
|
|
*/
|
2021-10-21 01:49:52 +08:00
|
|
|
if (vlan_filtering && dsa_port_is_user(dp)) {
|
2021-12-07 00:57:52 +08:00
|
|
|
struct net_device *br = dsa_port_bridge_dev_get(dp);
|
2023-10-24 02:17:28 +08:00
|
|
|
struct net_device *upper_dev, *user = dp->user;
|
2020-09-21 08:10:28 +08:00
|
|
|
struct list_head *iter;
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
netdev_for_each_upper_dev_rcu(user, upper_dev, iter) {
|
2020-09-21 08:10:28 +08:00
|
|
|
struct bridge_vlan_info br_info;
|
|
|
|
|
u16 vid;
|
|
|
|
|
|
|
|
|
|
if (!is_vlan_dev(upper_dev))
|
|
|
|
|
continue;
|
|
|
|
|
|
|
|
|
|
vid = vlan_dev_vlan_id(upper_dev);
|
|
|
|
|
|
|
|
|
|
/* br_vlan_get_info() returns -EINVAL or -ENOENT if the
|
|
|
|
|
* device, respectively the VID is not found, returning
|
|
|
|
|
* 0 means success, which is a failure for us here.
|
|
|
|
|
*/
|
|
|
|
|
err = br_vlan_get_info(br, vid, &br_info);
|
|
|
|
|
if (err == 0) {
|
2021-02-14 04:43:19 +08:00
|
|
|
NL_SET_ERR_MSG_MOD(extack,
|
|
|
|
|
"Must first remove VLAN uppers having VIDs also present in bridge");
|
2020-09-21 08:10:28 +08:00
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
2019-04-29 02:45:44 +08:00
|
|
|
|
|
|
|
|
if (!ds->vlan_filtering_is_global)
|
|
|
|
|
return true;
|
|
|
|
|
|
|
|
|
|
/* For cases where enabling/disabling VLAN awareness is global to the
|
|
|
|
|
* switch, we need to handle the case where multiple bridges span
|
|
|
|
|
* different ports of the same switch device and one of them has a
|
|
|
|
|
* different setting than what is being requested.
|
|
|
|
|
*/
|
2021-10-21 01:49:50 +08:00
|
|
|
dsa_switch_for_each_port(other_dp, ds) {
|
2021-12-07 00:57:52 +08:00
|
|
|
struct net_device *other_br = dsa_port_bridge_dev_get(other_dp);
|
2019-04-29 02:45:44 +08:00
|
|
|
|
|
|
|
|
/* If it's the same bridge, it also has same
|
|
|
|
|
* vlan_filtering setting => no need to check
|
|
|
|
|
*/
|
2021-12-07 00:57:52 +08:00
|
|
|
if (!other_br || other_br == dsa_port_bridge_dev_get(dp))
|
2019-04-29 02:45:44 +08:00
|
|
|
continue;
|
2021-12-07 00:57:52 +08:00
|
|
|
|
|
|
|
|
if (br_vlan_enabled(other_br) != vlan_filtering) {
|
2021-02-14 04:43:19 +08:00
|
|
|
NL_SET_ERR_MSG_MOD(extack,
|
|
|
|
|
"VLAN filtering is a global setting");
|
2019-04-29 02:45:44 +08:00
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
2021-02-14 04:43:19 +08:00
|
|
|
int dsa_port_vlan_filtering(struct dsa_port *dp, bool vlan_filtering,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
2017-05-20 05:00:46 +08:00
|
|
|
{
|
net: dsa: don't advertise 'rx-vlan-filter' when not needed
There have been multiple independent reports about
dsa_slave_vlan_rx_add_vid being called (and consequently calling the
drivers' .port_vlan_add) when it isn't needed, and sometimes (not
always) causing problems in the process.
Case 1:
mv88e6xxx_port_vlan_prepare is stubborn and only accepts VLANs on
bridged ports. That is understandably so, because standalone mv88e6xxx
ports are VLAN-unaware, and VTU entries are said to be a scarce
resource.
Otherwise said, the following fails lamentably on mv88e6xxx:
ip link add br0 type bridge vlan_filtering 1
ip link set lan3 master br0
ip link add link lan10 name lan10.1 type vlan id 1
[485256.724147] mv88e6085 d0032004.mdio-mii:12: p10: hw VLAN 1 already used by port 3 in br0
RTNETLINK answers: Operation not supported
This has become a worse issue since commit 9b236d2a69da ("net: dsa:
Advertise the VLAN offload netdev ability only if switch supports it").
Up to that point, the driver was returning -EOPNOTSUPP and DSA was
reconverting that error to 0, making the 8021q upper think all is ok
(but obviously the error message was there even prior to this change).
After that change the -EOPNOTSUPP is propagated to vlan_vid_add, and it
is a hard error.
Case 2:
Ports that don't offload the Linux bridge (have a dp->bridge_dev = NULL
because they don't implement .port_bridge_{join,leave}). Understandably,
a standalone port should not offload VLANs either, it should remain VLAN
unaware and any VLAN should be a software VLAN (as long as the hardware
is not quirky, that is).
In fact, dsa_slave_port_obj_add does do the right thing and rejects
switchdev VLAN objects coming from the bridge when that bridge is not
offloaded:
case SWITCHDEV_OBJ_ID_PORT_VLAN:
if (!dsa_port_offloads_bridge_port(dp, obj->orig_dev))
return -EOPNOTSUPP;
err = dsa_slave_vlan_add(dev, obj, extack);
But it seems that the bridge is able to trick us. The __vlan_vid_add
from br_vlan.c has:
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q add.
*/
err = br_switchdev_port_vlan_add(dev, v->vid, flags, extack);
if (err == -EOPNOTSUPP)
return vlan_vid_add(dev, br->vlan_proto, v->vid);
So it says "no, no, you need this VLAN in your life!". And we, naive as
we are, say "oh, this comes from the vlan_vid_add code path, it must be
an 8021q upper, sure, I'll take that". And we end up with that bridge
VLAN installed on our port anyway. But this time, it has the wrong flags:
if the bridge was trying to install VLAN 1 as a pvid/untagged VLAN,
failed via switchdev, retried via vlan_vid_add, we have this comment:
/* This API only allows programming tagged, non-PVID VIDs */
So what we do makes absolutely no sense.
Backtracing a bit, we see the common pattern. We allow the network stack
to think that our standalone ports are VLAN-aware, but they aren't, for
the vast majority of switches. The quirky ones should not dictate the
norm. The dsa_slave_vlan_rx_add_vid and dsa_slave_vlan_rx_kill_vid
methods exist for drivers that need the 'rx-vlan-filter: on' feature in
ethtool -k, which can be due to any of the following reasons:
1. vlan_filtering_is_global = true, and some ports are under a
VLAN-aware bridge while others are standalone, and the standalone
ports would otherwise drop VLAN-tagged traffic. This is described in
commit 061f6a505ac3 ("net: dsa: Add ndo_vlan_rx_{add, kill}_vid
implementation").
2. the ports that are under a VLAN-aware bridge should also set this
feature, for 8021q uppers having a VID not claimed by the bridge.
In this case, the driver will essentially not even know that the VID
is coming from the 8021q layer and not the bridge.
3. Hellcreek. This driver needs it because in standalone mode, it uses
unique VLANs per port to ensure separation. For separation of untagged
traffic, it uses different PVIDs for each port, and for separation of
VLAN-tagged traffic, it never accepts 8021q uppers with the same vid
on two ports.
If a driver does not fall under any of the above 3 categories, there is
no reason why it should advertise the 'rx-vlan-filter' feature, therefore
no reason why it should offload the VLANs added through vlan_vid_add.
This commit fixes the problem by removing the 'rx-vlan-filter' feature
from the slave devices when they operate in standalone mode, and when
they offload a VLAN-unaware bridge.
The way it works is that vlan_vid_add will now stop its processing here:
vlan_add_rx_filter_info:
if (!vlan_hw_filter_capable(dev, proto))
return 0;
So the VLAN will still be saved in the interface's VLAN RX filtering
list, but because it does not declare VLAN filtering in its features,
the 8021q module will return zero without committing that VLAN to
hardware.
This gives the drivers what they want, since it keeps the 8021q VLANs
away from the VLAN table until VLAN awareness is enabled (point at which
the ports are no longer standalone, hence in the mv88e6xxx case, the
check in mv88e6xxx_port_vlan_prepare passes).
Since the issue predates the existence of the hellcreek driver, case 3
will be dealt with in a separate patch.
The main change that this patch makes is to no longer set
NETIF_F_HW_VLAN_CTAG_FILTER unconditionally, but toggle it dynamically
(for most switches, never).
The second part of the patch addresses an issue that the first part
introduces: because the 'rx-vlan-filter' feature is now dynamically
toggled, and our .ndo_vlan_rx_add_vid does not get called when
'rx-vlan-filter' is off, we need to avoid bugs such as the following by
replaying the VLANs from 8021q uppers every time we enable VLAN
filtering:
ip link add link lan0 name lan0.100 type vlan id 100
ip addr add 192.168.100.1/24 dev lan0.100
ping 192.168.100.2 # should work
ip link add br0 type bridge vlan_filtering 0
ip link set lan0 master br0
ping 192.168.100.2 # should still work
ip link set br0 type bridge vlan_filtering 1
ping 192.168.100.2 # should still work but doesn't
As reported by Florian, some drivers look at ds->vlan_filtering in
their .port_vlan_add() implementation. So this patch also makes sure
that ds->vlan_filtering is committed before calling the driver. This is
the reason why it is first committed, then restored on the failure path.
Reported-by: Tobias Waldekranz <tobias@waldekranz.com>
Reported-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-24 05:22:57 +08:00
|
|
|
bool old_vlan_filtering = dsa_port_is_vlan_filtering(dp);
|
2017-05-20 05:00:46 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
bool apply;
|
2019-04-29 02:45:43 +08:00
|
|
|
int err;
|
2017-05-20 05:00:46 +08:00
|
|
|
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
if (!ds->ops->port_vlan_filtering)
|
|
|
|
|
return -EOPNOTSUPP;
|
2020-09-21 08:10:28 +08:00
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
/* We are called from dsa_user_switchdev_blocking_event(),
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
* which is not under rcu_read_lock(), unlike
|
2023-10-24 02:17:28 +08:00
|
|
|
* dsa_user_switchdev_event().
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
*/
|
|
|
|
|
rcu_read_lock();
|
2021-02-14 04:43:19 +08:00
|
|
|
apply = dsa_port_can_apply_vlan_filtering(dp, vlan_filtering, extack);
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
rcu_read_unlock();
|
|
|
|
|
if (!apply)
|
|
|
|
|
return -EINVAL;
|
2019-04-29 02:45:44 +08:00
|
|
|
|
2019-04-29 02:45:51 +08:00
|
|
|
if (dsa_port_is_vlan_filtering(dp) == vlan_filtering)
|
|
|
|
|
return 0;
|
|
|
|
|
|
2021-02-14 04:43:19 +08:00
|
|
|
err = ds->ops->port_vlan_filtering(ds, dp->index, vlan_filtering,
|
|
|
|
|
extack);
|
2019-04-29 02:45:44 +08:00
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
|
net: dsa: don't advertise 'rx-vlan-filter' when not needed
There have been multiple independent reports about
dsa_slave_vlan_rx_add_vid being called (and consequently calling the
drivers' .port_vlan_add) when it isn't needed, and sometimes (not
always) causing problems in the process.
Case 1:
mv88e6xxx_port_vlan_prepare is stubborn and only accepts VLANs on
bridged ports. That is understandably so, because standalone mv88e6xxx
ports are VLAN-unaware, and VTU entries are said to be a scarce
resource.
Otherwise said, the following fails lamentably on mv88e6xxx:
ip link add br0 type bridge vlan_filtering 1
ip link set lan3 master br0
ip link add link lan10 name lan10.1 type vlan id 1
[485256.724147] mv88e6085 d0032004.mdio-mii:12: p10: hw VLAN 1 already used by port 3 in br0
RTNETLINK answers: Operation not supported
This has become a worse issue since commit 9b236d2a69da ("net: dsa:
Advertise the VLAN offload netdev ability only if switch supports it").
Up to that point, the driver was returning -EOPNOTSUPP and DSA was
reconverting that error to 0, making the 8021q upper think all is ok
(but obviously the error message was there even prior to this change).
After that change the -EOPNOTSUPP is propagated to vlan_vid_add, and it
is a hard error.
Case 2:
Ports that don't offload the Linux bridge (have a dp->bridge_dev = NULL
because they don't implement .port_bridge_{join,leave}). Understandably,
a standalone port should not offload VLANs either, it should remain VLAN
unaware and any VLAN should be a software VLAN (as long as the hardware
is not quirky, that is).
In fact, dsa_slave_port_obj_add does do the right thing and rejects
switchdev VLAN objects coming from the bridge when that bridge is not
offloaded:
case SWITCHDEV_OBJ_ID_PORT_VLAN:
if (!dsa_port_offloads_bridge_port(dp, obj->orig_dev))
return -EOPNOTSUPP;
err = dsa_slave_vlan_add(dev, obj, extack);
But it seems that the bridge is able to trick us. The __vlan_vid_add
from br_vlan.c has:
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q add.
*/
err = br_switchdev_port_vlan_add(dev, v->vid, flags, extack);
if (err == -EOPNOTSUPP)
return vlan_vid_add(dev, br->vlan_proto, v->vid);
So it says "no, no, you need this VLAN in your life!". And we, naive as
we are, say "oh, this comes from the vlan_vid_add code path, it must be
an 8021q upper, sure, I'll take that". And we end up with that bridge
VLAN installed on our port anyway. But this time, it has the wrong flags:
if the bridge was trying to install VLAN 1 as a pvid/untagged VLAN,
failed via switchdev, retried via vlan_vid_add, we have this comment:
/* This API only allows programming tagged, non-PVID VIDs */
So what we do makes absolutely no sense.
Backtracing a bit, we see the common pattern. We allow the network stack
to think that our standalone ports are VLAN-aware, but they aren't, for
the vast majority of switches. The quirky ones should not dictate the
norm. The dsa_slave_vlan_rx_add_vid and dsa_slave_vlan_rx_kill_vid
methods exist for drivers that need the 'rx-vlan-filter: on' feature in
ethtool -k, which can be due to any of the following reasons:
1. vlan_filtering_is_global = true, and some ports are under a
VLAN-aware bridge while others are standalone, and the standalone
ports would otherwise drop VLAN-tagged traffic. This is described in
commit 061f6a505ac3 ("net: dsa: Add ndo_vlan_rx_{add, kill}_vid
implementation").
2. the ports that are under a VLAN-aware bridge should also set this
feature, for 8021q uppers having a VID not claimed by the bridge.
In this case, the driver will essentially not even know that the VID
is coming from the 8021q layer and not the bridge.
3. Hellcreek. This driver needs it because in standalone mode, it uses
unique VLANs per port to ensure separation. For separation of untagged
traffic, it uses different PVIDs for each port, and for separation of
VLAN-tagged traffic, it never accepts 8021q uppers with the same vid
on two ports.
If a driver does not fall under any of the above 3 categories, there is
no reason why it should advertise the 'rx-vlan-filter' feature, therefore
no reason why it should offload the VLANs added through vlan_vid_add.
This commit fixes the problem by removing the 'rx-vlan-filter' feature
from the slave devices when they operate in standalone mode, and when
they offload a VLAN-unaware bridge.
The way it works is that vlan_vid_add will now stop its processing here:
vlan_add_rx_filter_info:
if (!vlan_hw_filter_capable(dev, proto))
return 0;
So the VLAN will still be saved in the interface's VLAN RX filtering
list, but because it does not declare VLAN filtering in its features,
the 8021q module will return zero without committing that VLAN to
hardware.
This gives the drivers what they want, since it keeps the 8021q VLANs
away from the VLAN table until VLAN awareness is enabled (point at which
the ports are no longer standalone, hence in the mv88e6xxx case, the
check in mv88e6xxx_port_vlan_prepare passes).
Since the issue predates the existence of the hellcreek driver, case 3
will be dealt with in a separate patch.
The main change that this patch makes is to no longer set
NETIF_F_HW_VLAN_CTAG_FILTER unconditionally, but toggle it dynamically
(for most switches, never).
The second part of the patch addresses an issue that the first part
introduces: because the 'rx-vlan-filter' feature is now dynamically
toggled, and our .ndo_vlan_rx_add_vid does not get called when
'rx-vlan-filter' is off, we need to avoid bugs such as the following by
replaying the VLANs from 8021q uppers every time we enable VLAN
filtering:
ip link add link lan0 name lan0.100 type vlan id 100
ip addr add 192.168.100.1/24 dev lan0.100
ping 192.168.100.2 # should work
ip link add br0 type bridge vlan_filtering 0
ip link set lan0 master br0
ping 192.168.100.2 # should still work
ip link set br0 type bridge vlan_filtering 1
ping 192.168.100.2 # should still work but doesn't
As reported by Florian, some drivers look at ds->vlan_filtering in
their .port_vlan_add() implementation. So this patch also makes sure
that ds->vlan_filtering is committed before calling the driver. This is
the reason why it is first committed, then restored on the failure path.
Reported-by: Tobias Waldekranz <tobias@waldekranz.com>
Reported-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-24 05:22:57 +08:00
|
|
|
if (ds->vlan_filtering_is_global) {
|
2021-10-21 01:49:50 +08:00
|
|
|
struct dsa_port *other_dp;
|
net: dsa: don't advertise 'rx-vlan-filter' when not needed
There have been multiple independent reports about
dsa_slave_vlan_rx_add_vid being called (and consequently calling the
drivers' .port_vlan_add) when it isn't needed, and sometimes (not
always) causing problems in the process.
Case 1:
mv88e6xxx_port_vlan_prepare is stubborn and only accepts VLANs on
bridged ports. That is understandably so, because standalone mv88e6xxx
ports are VLAN-unaware, and VTU entries are said to be a scarce
resource.
Otherwise said, the following fails lamentably on mv88e6xxx:
ip link add br0 type bridge vlan_filtering 1
ip link set lan3 master br0
ip link add link lan10 name lan10.1 type vlan id 1
[485256.724147] mv88e6085 d0032004.mdio-mii:12: p10: hw VLAN 1 already used by port 3 in br0
RTNETLINK answers: Operation not supported
This has become a worse issue since commit 9b236d2a69da ("net: dsa:
Advertise the VLAN offload netdev ability only if switch supports it").
Up to that point, the driver was returning -EOPNOTSUPP and DSA was
reconverting that error to 0, making the 8021q upper think all is ok
(but obviously the error message was there even prior to this change).
After that change the -EOPNOTSUPP is propagated to vlan_vid_add, and it
is a hard error.
Case 2:
Ports that don't offload the Linux bridge (have a dp->bridge_dev = NULL
because they don't implement .port_bridge_{join,leave}). Understandably,
a standalone port should not offload VLANs either, it should remain VLAN
unaware and any VLAN should be a software VLAN (as long as the hardware
is not quirky, that is).
In fact, dsa_slave_port_obj_add does do the right thing and rejects
switchdev VLAN objects coming from the bridge when that bridge is not
offloaded:
case SWITCHDEV_OBJ_ID_PORT_VLAN:
if (!dsa_port_offloads_bridge_port(dp, obj->orig_dev))
return -EOPNOTSUPP;
err = dsa_slave_vlan_add(dev, obj, extack);
But it seems that the bridge is able to trick us. The __vlan_vid_add
from br_vlan.c has:
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q add.
*/
err = br_switchdev_port_vlan_add(dev, v->vid, flags, extack);
if (err == -EOPNOTSUPP)
return vlan_vid_add(dev, br->vlan_proto, v->vid);
So it says "no, no, you need this VLAN in your life!". And we, naive as
we are, say "oh, this comes from the vlan_vid_add code path, it must be
an 8021q upper, sure, I'll take that". And we end up with that bridge
VLAN installed on our port anyway. But this time, it has the wrong flags:
if the bridge was trying to install VLAN 1 as a pvid/untagged VLAN,
failed via switchdev, retried via vlan_vid_add, we have this comment:
/* This API only allows programming tagged, non-PVID VIDs */
So what we do makes absolutely no sense.
Backtracing a bit, we see the common pattern. We allow the network stack
to think that our standalone ports are VLAN-aware, but they aren't, for
the vast majority of switches. The quirky ones should not dictate the
norm. The dsa_slave_vlan_rx_add_vid and dsa_slave_vlan_rx_kill_vid
methods exist for drivers that need the 'rx-vlan-filter: on' feature in
ethtool -k, which can be due to any of the following reasons:
1. vlan_filtering_is_global = true, and some ports are under a
VLAN-aware bridge while others are standalone, and the standalone
ports would otherwise drop VLAN-tagged traffic. This is described in
commit 061f6a505ac3 ("net: dsa: Add ndo_vlan_rx_{add, kill}_vid
implementation").
2. the ports that are under a VLAN-aware bridge should also set this
feature, for 8021q uppers having a VID not claimed by the bridge.
In this case, the driver will essentially not even know that the VID
is coming from the 8021q layer and not the bridge.
3. Hellcreek. This driver needs it because in standalone mode, it uses
unique VLANs per port to ensure separation. For separation of untagged
traffic, it uses different PVIDs for each port, and for separation of
VLAN-tagged traffic, it never accepts 8021q uppers with the same vid
on two ports.
If a driver does not fall under any of the above 3 categories, there is
no reason why it should advertise the 'rx-vlan-filter' feature, therefore
no reason why it should offload the VLANs added through vlan_vid_add.
This commit fixes the problem by removing the 'rx-vlan-filter' feature
from the slave devices when they operate in standalone mode, and when
they offload a VLAN-unaware bridge.
The way it works is that vlan_vid_add will now stop its processing here:
vlan_add_rx_filter_info:
if (!vlan_hw_filter_capable(dev, proto))
return 0;
So the VLAN will still be saved in the interface's VLAN RX filtering
list, but because it does not declare VLAN filtering in its features,
the 8021q module will return zero without committing that VLAN to
hardware.
This gives the drivers what they want, since it keeps the 8021q VLANs
away from the VLAN table until VLAN awareness is enabled (point at which
the ports are no longer standalone, hence in the mv88e6xxx case, the
check in mv88e6xxx_port_vlan_prepare passes).
Since the issue predates the existence of the hellcreek driver, case 3
will be dealt with in a separate patch.
The main change that this patch makes is to no longer set
NETIF_F_HW_VLAN_CTAG_FILTER unconditionally, but toggle it dynamically
(for most switches, never).
The second part of the patch addresses an issue that the first part
introduces: because the 'rx-vlan-filter' feature is now dynamically
toggled, and our .ndo_vlan_rx_add_vid does not get called when
'rx-vlan-filter' is off, we need to avoid bugs such as the following by
replaying the VLANs from 8021q uppers every time we enable VLAN
filtering:
ip link add link lan0 name lan0.100 type vlan id 100
ip addr add 192.168.100.1/24 dev lan0.100
ping 192.168.100.2 # should work
ip link add br0 type bridge vlan_filtering 0
ip link set lan0 master br0
ping 192.168.100.2 # should still work
ip link set br0 type bridge vlan_filtering 1
ping 192.168.100.2 # should still work but doesn't
As reported by Florian, some drivers look at ds->vlan_filtering in
their .port_vlan_add() implementation. So this patch also makes sure
that ds->vlan_filtering is committed before calling the driver. This is
the reason why it is first committed, then restored on the failure path.
Reported-by: Tobias Waldekranz <tobias@waldekranz.com>
Reported-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-24 05:22:57 +08:00
|
|
|
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
ds->vlan_filtering = vlan_filtering;
|
net: dsa: don't advertise 'rx-vlan-filter' when not needed
There have been multiple independent reports about
dsa_slave_vlan_rx_add_vid being called (and consequently calling the
drivers' .port_vlan_add) when it isn't needed, and sometimes (not
always) causing problems in the process.
Case 1:
mv88e6xxx_port_vlan_prepare is stubborn and only accepts VLANs on
bridged ports. That is understandably so, because standalone mv88e6xxx
ports are VLAN-unaware, and VTU entries are said to be a scarce
resource.
Otherwise said, the following fails lamentably on mv88e6xxx:
ip link add br0 type bridge vlan_filtering 1
ip link set lan3 master br0
ip link add link lan10 name lan10.1 type vlan id 1
[485256.724147] mv88e6085 d0032004.mdio-mii:12: p10: hw VLAN 1 already used by port 3 in br0
RTNETLINK answers: Operation not supported
This has become a worse issue since commit 9b236d2a69da ("net: dsa:
Advertise the VLAN offload netdev ability only if switch supports it").
Up to that point, the driver was returning -EOPNOTSUPP and DSA was
reconverting that error to 0, making the 8021q upper think all is ok
(but obviously the error message was there even prior to this change).
After that change the -EOPNOTSUPP is propagated to vlan_vid_add, and it
is a hard error.
Case 2:
Ports that don't offload the Linux bridge (have a dp->bridge_dev = NULL
because they don't implement .port_bridge_{join,leave}). Understandably,
a standalone port should not offload VLANs either, it should remain VLAN
unaware and any VLAN should be a software VLAN (as long as the hardware
is not quirky, that is).
In fact, dsa_slave_port_obj_add does do the right thing and rejects
switchdev VLAN objects coming from the bridge when that bridge is not
offloaded:
case SWITCHDEV_OBJ_ID_PORT_VLAN:
if (!dsa_port_offloads_bridge_port(dp, obj->orig_dev))
return -EOPNOTSUPP;
err = dsa_slave_vlan_add(dev, obj, extack);
But it seems that the bridge is able to trick us. The __vlan_vid_add
from br_vlan.c has:
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q add.
*/
err = br_switchdev_port_vlan_add(dev, v->vid, flags, extack);
if (err == -EOPNOTSUPP)
return vlan_vid_add(dev, br->vlan_proto, v->vid);
So it says "no, no, you need this VLAN in your life!". And we, naive as
we are, say "oh, this comes from the vlan_vid_add code path, it must be
an 8021q upper, sure, I'll take that". And we end up with that bridge
VLAN installed on our port anyway. But this time, it has the wrong flags:
if the bridge was trying to install VLAN 1 as a pvid/untagged VLAN,
failed via switchdev, retried via vlan_vid_add, we have this comment:
/* This API only allows programming tagged, non-PVID VIDs */
So what we do makes absolutely no sense.
Backtracing a bit, we see the common pattern. We allow the network stack
to think that our standalone ports are VLAN-aware, but they aren't, for
the vast majority of switches. The quirky ones should not dictate the
norm. The dsa_slave_vlan_rx_add_vid and dsa_slave_vlan_rx_kill_vid
methods exist for drivers that need the 'rx-vlan-filter: on' feature in
ethtool -k, which can be due to any of the following reasons:
1. vlan_filtering_is_global = true, and some ports are under a
VLAN-aware bridge while others are standalone, and the standalone
ports would otherwise drop VLAN-tagged traffic. This is described in
commit 061f6a505ac3 ("net: dsa: Add ndo_vlan_rx_{add, kill}_vid
implementation").
2. the ports that are under a VLAN-aware bridge should also set this
feature, for 8021q uppers having a VID not claimed by the bridge.
In this case, the driver will essentially not even know that the VID
is coming from the 8021q layer and not the bridge.
3. Hellcreek. This driver needs it because in standalone mode, it uses
unique VLANs per port to ensure separation. For separation of untagged
traffic, it uses different PVIDs for each port, and for separation of
VLAN-tagged traffic, it never accepts 8021q uppers with the same vid
on two ports.
If a driver does not fall under any of the above 3 categories, there is
no reason why it should advertise the 'rx-vlan-filter' feature, therefore
no reason why it should offload the VLANs added through vlan_vid_add.
This commit fixes the problem by removing the 'rx-vlan-filter' feature
from the slave devices when they operate in standalone mode, and when
they offload a VLAN-unaware bridge.
The way it works is that vlan_vid_add will now stop its processing here:
vlan_add_rx_filter_info:
if (!vlan_hw_filter_capable(dev, proto))
return 0;
So the VLAN will still be saved in the interface's VLAN RX filtering
list, but because it does not declare VLAN filtering in its features,
the 8021q module will return zero without committing that VLAN to
hardware.
This gives the drivers what they want, since it keeps the 8021q VLANs
away from the VLAN table until VLAN awareness is enabled (point at which
the ports are no longer standalone, hence in the mv88e6xxx case, the
check in mv88e6xxx_port_vlan_prepare passes).
Since the issue predates the existence of the hellcreek driver, case 3
will be dealt with in a separate patch.
The main change that this patch makes is to no longer set
NETIF_F_HW_VLAN_CTAG_FILTER unconditionally, but toggle it dynamically
(for most switches, never).
The second part of the patch addresses an issue that the first part
introduces: because the 'rx-vlan-filter' feature is now dynamically
toggled, and our .ndo_vlan_rx_add_vid does not get called when
'rx-vlan-filter' is off, we need to avoid bugs such as the following by
replaying the VLANs from 8021q uppers every time we enable VLAN
filtering:
ip link add link lan0 name lan0.100 type vlan id 100
ip addr add 192.168.100.1/24 dev lan0.100
ping 192.168.100.2 # should work
ip link add br0 type bridge vlan_filtering 0
ip link set lan0 master br0
ping 192.168.100.2 # should still work
ip link set br0 type bridge vlan_filtering 1
ping 192.168.100.2 # should still work but doesn't
As reported by Florian, some drivers look at ds->vlan_filtering in
their .port_vlan_add() implementation. So this patch also makes sure
that ds->vlan_filtering is committed before calling the driver. This is
the reason why it is first committed, then restored on the failure path.
Reported-by: Tobias Waldekranz <tobias@waldekranz.com>
Reported-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-24 05:22:57 +08:00
|
|
|
|
2021-10-21 01:49:50 +08:00
|
|
|
dsa_switch_for_each_user_port(other_dp, ds) {
|
2023-10-24 02:17:28 +08:00
|
|
|
struct net_device *user = other_dp->user;
|
net: dsa: don't advertise 'rx-vlan-filter' when not needed
There have been multiple independent reports about
dsa_slave_vlan_rx_add_vid being called (and consequently calling the
drivers' .port_vlan_add) when it isn't needed, and sometimes (not
always) causing problems in the process.
Case 1:
mv88e6xxx_port_vlan_prepare is stubborn and only accepts VLANs on
bridged ports. That is understandably so, because standalone mv88e6xxx
ports are VLAN-unaware, and VTU entries are said to be a scarce
resource.
Otherwise said, the following fails lamentably on mv88e6xxx:
ip link add br0 type bridge vlan_filtering 1
ip link set lan3 master br0
ip link add link lan10 name lan10.1 type vlan id 1
[485256.724147] mv88e6085 d0032004.mdio-mii:12: p10: hw VLAN 1 already used by port 3 in br0
RTNETLINK answers: Operation not supported
This has become a worse issue since commit 9b236d2a69da ("net: dsa:
Advertise the VLAN offload netdev ability only if switch supports it").
Up to that point, the driver was returning -EOPNOTSUPP and DSA was
reconverting that error to 0, making the 8021q upper think all is ok
(but obviously the error message was there even prior to this change).
After that change the -EOPNOTSUPP is propagated to vlan_vid_add, and it
is a hard error.
Case 2:
Ports that don't offload the Linux bridge (have a dp->bridge_dev = NULL
because they don't implement .port_bridge_{join,leave}). Understandably,
a standalone port should not offload VLANs either, it should remain VLAN
unaware and any VLAN should be a software VLAN (as long as the hardware
is not quirky, that is).
In fact, dsa_slave_port_obj_add does do the right thing and rejects
switchdev VLAN objects coming from the bridge when that bridge is not
offloaded:
case SWITCHDEV_OBJ_ID_PORT_VLAN:
if (!dsa_port_offloads_bridge_port(dp, obj->orig_dev))
return -EOPNOTSUPP;
err = dsa_slave_vlan_add(dev, obj, extack);
But it seems that the bridge is able to trick us. The __vlan_vid_add
from br_vlan.c has:
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q add.
*/
err = br_switchdev_port_vlan_add(dev, v->vid, flags, extack);
if (err == -EOPNOTSUPP)
return vlan_vid_add(dev, br->vlan_proto, v->vid);
So it says "no, no, you need this VLAN in your life!". And we, naive as
we are, say "oh, this comes from the vlan_vid_add code path, it must be
an 8021q upper, sure, I'll take that". And we end up with that bridge
VLAN installed on our port anyway. But this time, it has the wrong flags:
if the bridge was trying to install VLAN 1 as a pvid/untagged VLAN,
failed via switchdev, retried via vlan_vid_add, we have this comment:
/* This API only allows programming tagged, non-PVID VIDs */
So what we do makes absolutely no sense.
Backtracing a bit, we see the common pattern. We allow the network stack
to think that our standalone ports are VLAN-aware, but they aren't, for
the vast majority of switches. The quirky ones should not dictate the
norm. The dsa_slave_vlan_rx_add_vid and dsa_slave_vlan_rx_kill_vid
methods exist for drivers that need the 'rx-vlan-filter: on' feature in
ethtool -k, which can be due to any of the following reasons:
1. vlan_filtering_is_global = true, and some ports are under a
VLAN-aware bridge while others are standalone, and the standalone
ports would otherwise drop VLAN-tagged traffic. This is described in
commit 061f6a505ac3 ("net: dsa: Add ndo_vlan_rx_{add, kill}_vid
implementation").
2. the ports that are under a VLAN-aware bridge should also set this
feature, for 8021q uppers having a VID not claimed by the bridge.
In this case, the driver will essentially not even know that the VID
is coming from the 8021q layer and not the bridge.
3. Hellcreek. This driver needs it because in standalone mode, it uses
unique VLANs per port to ensure separation. For separation of untagged
traffic, it uses different PVIDs for each port, and for separation of
VLAN-tagged traffic, it never accepts 8021q uppers with the same vid
on two ports.
If a driver does not fall under any of the above 3 categories, there is
no reason why it should advertise the 'rx-vlan-filter' feature, therefore
no reason why it should offload the VLANs added through vlan_vid_add.
This commit fixes the problem by removing the 'rx-vlan-filter' feature
from the slave devices when they operate in standalone mode, and when
they offload a VLAN-unaware bridge.
The way it works is that vlan_vid_add will now stop its processing here:
vlan_add_rx_filter_info:
if (!vlan_hw_filter_capable(dev, proto))
return 0;
So the VLAN will still be saved in the interface's VLAN RX filtering
list, but because it does not declare VLAN filtering in its features,
the 8021q module will return zero without committing that VLAN to
hardware.
This gives the drivers what they want, since it keeps the 8021q VLANs
away from the VLAN table until VLAN awareness is enabled (point at which
the ports are no longer standalone, hence in the mv88e6xxx case, the
check in mv88e6xxx_port_vlan_prepare passes).
Since the issue predates the existence of the hellcreek driver, case 3
will be dealt with in a separate patch.
The main change that this patch makes is to no longer set
NETIF_F_HW_VLAN_CTAG_FILTER unconditionally, but toggle it dynamically
(for most switches, never).
The second part of the patch addresses an issue that the first part
introduces: because the 'rx-vlan-filter' feature is now dynamically
toggled, and our .ndo_vlan_rx_add_vid does not get called when
'rx-vlan-filter' is off, we need to avoid bugs such as the following by
replaying the VLANs from 8021q uppers every time we enable VLAN
filtering:
ip link add link lan0 name lan0.100 type vlan id 100
ip addr add 192.168.100.1/24 dev lan0.100
ping 192.168.100.2 # should work
ip link add br0 type bridge vlan_filtering 0
ip link set lan0 master br0
ping 192.168.100.2 # should still work
ip link set br0 type bridge vlan_filtering 1
ping 192.168.100.2 # should still work but doesn't
As reported by Florian, some drivers look at ds->vlan_filtering in
their .port_vlan_add() implementation. So this patch also makes sure
that ds->vlan_filtering is committed before calling the driver. This is
the reason why it is first committed, then restored on the failure path.
Reported-by: Tobias Waldekranz <tobias@waldekranz.com>
Reported-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-24 05:22:57 +08:00
|
|
|
|
|
|
|
|
/* We might be called in the unbind path, so not
|
2023-10-24 02:17:28 +08:00
|
|
|
* all user devices might still be registered.
|
net: dsa: don't advertise 'rx-vlan-filter' when not needed
There have been multiple independent reports about
dsa_slave_vlan_rx_add_vid being called (and consequently calling the
drivers' .port_vlan_add) when it isn't needed, and sometimes (not
always) causing problems in the process.
Case 1:
mv88e6xxx_port_vlan_prepare is stubborn and only accepts VLANs on
bridged ports. That is understandably so, because standalone mv88e6xxx
ports are VLAN-unaware, and VTU entries are said to be a scarce
resource.
Otherwise said, the following fails lamentably on mv88e6xxx:
ip link add br0 type bridge vlan_filtering 1
ip link set lan3 master br0
ip link add link lan10 name lan10.1 type vlan id 1
[485256.724147] mv88e6085 d0032004.mdio-mii:12: p10: hw VLAN 1 already used by port 3 in br0
RTNETLINK answers: Operation not supported
This has become a worse issue since commit 9b236d2a69da ("net: dsa:
Advertise the VLAN offload netdev ability only if switch supports it").
Up to that point, the driver was returning -EOPNOTSUPP and DSA was
reconverting that error to 0, making the 8021q upper think all is ok
(but obviously the error message was there even prior to this change).
After that change the -EOPNOTSUPP is propagated to vlan_vid_add, and it
is a hard error.
Case 2:
Ports that don't offload the Linux bridge (have a dp->bridge_dev = NULL
because they don't implement .port_bridge_{join,leave}). Understandably,
a standalone port should not offload VLANs either, it should remain VLAN
unaware and any VLAN should be a software VLAN (as long as the hardware
is not quirky, that is).
In fact, dsa_slave_port_obj_add does do the right thing and rejects
switchdev VLAN objects coming from the bridge when that bridge is not
offloaded:
case SWITCHDEV_OBJ_ID_PORT_VLAN:
if (!dsa_port_offloads_bridge_port(dp, obj->orig_dev))
return -EOPNOTSUPP;
err = dsa_slave_vlan_add(dev, obj, extack);
But it seems that the bridge is able to trick us. The __vlan_vid_add
from br_vlan.c has:
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q add.
*/
err = br_switchdev_port_vlan_add(dev, v->vid, flags, extack);
if (err == -EOPNOTSUPP)
return vlan_vid_add(dev, br->vlan_proto, v->vid);
So it says "no, no, you need this VLAN in your life!". And we, naive as
we are, say "oh, this comes from the vlan_vid_add code path, it must be
an 8021q upper, sure, I'll take that". And we end up with that bridge
VLAN installed on our port anyway. But this time, it has the wrong flags:
if the bridge was trying to install VLAN 1 as a pvid/untagged VLAN,
failed via switchdev, retried via vlan_vid_add, we have this comment:
/* This API only allows programming tagged, non-PVID VIDs */
So what we do makes absolutely no sense.
Backtracing a bit, we see the common pattern. We allow the network stack
to think that our standalone ports are VLAN-aware, but they aren't, for
the vast majority of switches. The quirky ones should not dictate the
norm. The dsa_slave_vlan_rx_add_vid and dsa_slave_vlan_rx_kill_vid
methods exist for drivers that need the 'rx-vlan-filter: on' feature in
ethtool -k, which can be due to any of the following reasons:
1. vlan_filtering_is_global = true, and some ports are under a
VLAN-aware bridge while others are standalone, and the standalone
ports would otherwise drop VLAN-tagged traffic. This is described in
commit 061f6a505ac3 ("net: dsa: Add ndo_vlan_rx_{add, kill}_vid
implementation").
2. the ports that are under a VLAN-aware bridge should also set this
feature, for 8021q uppers having a VID not claimed by the bridge.
In this case, the driver will essentially not even know that the VID
is coming from the 8021q layer and not the bridge.
3. Hellcreek. This driver needs it because in standalone mode, it uses
unique VLANs per port to ensure separation. For separation of untagged
traffic, it uses different PVIDs for each port, and for separation of
VLAN-tagged traffic, it never accepts 8021q uppers with the same vid
on two ports.
If a driver does not fall under any of the above 3 categories, there is
no reason why it should advertise the 'rx-vlan-filter' feature, therefore
no reason why it should offload the VLANs added through vlan_vid_add.
This commit fixes the problem by removing the 'rx-vlan-filter' feature
from the slave devices when they operate in standalone mode, and when
they offload a VLAN-unaware bridge.
The way it works is that vlan_vid_add will now stop its processing here:
vlan_add_rx_filter_info:
if (!vlan_hw_filter_capable(dev, proto))
return 0;
So the VLAN will still be saved in the interface's VLAN RX filtering
list, but because it does not declare VLAN filtering in its features,
the 8021q module will return zero without committing that VLAN to
hardware.
This gives the drivers what they want, since it keeps the 8021q VLANs
away from the VLAN table until VLAN awareness is enabled (point at which
the ports are no longer standalone, hence in the mv88e6xxx case, the
check in mv88e6xxx_port_vlan_prepare passes).
Since the issue predates the existence of the hellcreek driver, case 3
will be dealt with in a separate patch.
The main change that this patch makes is to no longer set
NETIF_F_HW_VLAN_CTAG_FILTER unconditionally, but toggle it dynamically
(for most switches, never).
The second part of the patch addresses an issue that the first part
introduces: because the 'rx-vlan-filter' feature is now dynamically
toggled, and our .ndo_vlan_rx_add_vid does not get called when
'rx-vlan-filter' is off, we need to avoid bugs such as the following by
replaying the VLANs from 8021q uppers every time we enable VLAN
filtering:
ip link add link lan0 name lan0.100 type vlan id 100
ip addr add 192.168.100.1/24 dev lan0.100
ping 192.168.100.2 # should work
ip link add br0 type bridge vlan_filtering 0
ip link set lan0 master br0
ping 192.168.100.2 # should still work
ip link set br0 type bridge vlan_filtering 1
ping 192.168.100.2 # should still work but doesn't
As reported by Florian, some drivers look at ds->vlan_filtering in
their .port_vlan_add() implementation. So this patch also makes sure
that ds->vlan_filtering is committed before calling the driver. This is
the reason why it is first committed, then restored on the failure path.
Reported-by: Tobias Waldekranz <tobias@waldekranz.com>
Reported-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-24 05:22:57 +08:00
|
|
|
*/
|
2023-10-24 02:17:28 +08:00
|
|
|
if (!user)
|
net: dsa: don't advertise 'rx-vlan-filter' when not needed
There have been multiple independent reports about
dsa_slave_vlan_rx_add_vid being called (and consequently calling the
drivers' .port_vlan_add) when it isn't needed, and sometimes (not
always) causing problems in the process.
Case 1:
mv88e6xxx_port_vlan_prepare is stubborn and only accepts VLANs on
bridged ports. That is understandably so, because standalone mv88e6xxx
ports are VLAN-unaware, and VTU entries are said to be a scarce
resource.
Otherwise said, the following fails lamentably on mv88e6xxx:
ip link add br0 type bridge vlan_filtering 1
ip link set lan3 master br0
ip link add link lan10 name lan10.1 type vlan id 1
[485256.724147] mv88e6085 d0032004.mdio-mii:12: p10: hw VLAN 1 already used by port 3 in br0
RTNETLINK answers: Operation not supported
This has become a worse issue since commit 9b236d2a69da ("net: dsa:
Advertise the VLAN offload netdev ability only if switch supports it").
Up to that point, the driver was returning -EOPNOTSUPP and DSA was
reconverting that error to 0, making the 8021q upper think all is ok
(but obviously the error message was there even prior to this change).
After that change the -EOPNOTSUPP is propagated to vlan_vid_add, and it
is a hard error.
Case 2:
Ports that don't offload the Linux bridge (have a dp->bridge_dev = NULL
because they don't implement .port_bridge_{join,leave}). Understandably,
a standalone port should not offload VLANs either, it should remain VLAN
unaware and any VLAN should be a software VLAN (as long as the hardware
is not quirky, that is).
In fact, dsa_slave_port_obj_add does do the right thing and rejects
switchdev VLAN objects coming from the bridge when that bridge is not
offloaded:
case SWITCHDEV_OBJ_ID_PORT_VLAN:
if (!dsa_port_offloads_bridge_port(dp, obj->orig_dev))
return -EOPNOTSUPP;
err = dsa_slave_vlan_add(dev, obj, extack);
But it seems that the bridge is able to trick us. The __vlan_vid_add
from br_vlan.c has:
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q add.
*/
err = br_switchdev_port_vlan_add(dev, v->vid, flags, extack);
if (err == -EOPNOTSUPP)
return vlan_vid_add(dev, br->vlan_proto, v->vid);
So it says "no, no, you need this VLAN in your life!". And we, naive as
we are, say "oh, this comes from the vlan_vid_add code path, it must be
an 8021q upper, sure, I'll take that". And we end up with that bridge
VLAN installed on our port anyway. But this time, it has the wrong flags:
if the bridge was trying to install VLAN 1 as a pvid/untagged VLAN,
failed via switchdev, retried via vlan_vid_add, we have this comment:
/* This API only allows programming tagged, non-PVID VIDs */
So what we do makes absolutely no sense.
Backtracing a bit, we see the common pattern. We allow the network stack
to think that our standalone ports are VLAN-aware, but they aren't, for
the vast majority of switches. The quirky ones should not dictate the
norm. The dsa_slave_vlan_rx_add_vid and dsa_slave_vlan_rx_kill_vid
methods exist for drivers that need the 'rx-vlan-filter: on' feature in
ethtool -k, which can be due to any of the following reasons:
1. vlan_filtering_is_global = true, and some ports are under a
VLAN-aware bridge while others are standalone, and the standalone
ports would otherwise drop VLAN-tagged traffic. This is described in
commit 061f6a505ac3 ("net: dsa: Add ndo_vlan_rx_{add, kill}_vid
implementation").
2. the ports that are under a VLAN-aware bridge should also set this
feature, for 8021q uppers having a VID not claimed by the bridge.
In this case, the driver will essentially not even know that the VID
is coming from the 8021q layer and not the bridge.
3. Hellcreek. This driver needs it because in standalone mode, it uses
unique VLANs per port to ensure separation. For separation of untagged
traffic, it uses different PVIDs for each port, and for separation of
VLAN-tagged traffic, it never accepts 8021q uppers with the same vid
on two ports.
If a driver does not fall under any of the above 3 categories, there is
no reason why it should advertise the 'rx-vlan-filter' feature, therefore
no reason why it should offload the VLANs added through vlan_vid_add.
This commit fixes the problem by removing the 'rx-vlan-filter' feature
from the slave devices when they operate in standalone mode, and when
they offload a VLAN-unaware bridge.
The way it works is that vlan_vid_add will now stop its processing here:
vlan_add_rx_filter_info:
if (!vlan_hw_filter_capable(dev, proto))
return 0;
So the VLAN will still be saved in the interface's VLAN RX filtering
list, but because it does not declare VLAN filtering in its features,
the 8021q module will return zero without committing that VLAN to
hardware.
This gives the drivers what they want, since it keeps the 8021q VLANs
away from the VLAN table until VLAN awareness is enabled (point at which
the ports are no longer standalone, hence in the mv88e6xxx case, the
check in mv88e6xxx_port_vlan_prepare passes).
Since the issue predates the existence of the hellcreek driver, case 3
will be dealt with in a separate patch.
The main change that this patch makes is to no longer set
NETIF_F_HW_VLAN_CTAG_FILTER unconditionally, but toggle it dynamically
(for most switches, never).
The second part of the patch addresses an issue that the first part
introduces: because the 'rx-vlan-filter' feature is now dynamically
toggled, and our .ndo_vlan_rx_add_vid does not get called when
'rx-vlan-filter' is off, we need to avoid bugs such as the following by
replaying the VLANs from 8021q uppers every time we enable VLAN
filtering:
ip link add link lan0 name lan0.100 type vlan id 100
ip addr add 192.168.100.1/24 dev lan0.100
ping 192.168.100.2 # should work
ip link add br0 type bridge vlan_filtering 0
ip link set lan0 master br0
ping 192.168.100.2 # should still work
ip link set br0 type bridge vlan_filtering 1
ping 192.168.100.2 # should still work but doesn't
As reported by Florian, some drivers look at ds->vlan_filtering in
their .port_vlan_add() implementation. So this patch also makes sure
that ds->vlan_filtering is committed before calling the driver. This is
the reason why it is first committed, then restored on the failure path.
Reported-by: Tobias Waldekranz <tobias@waldekranz.com>
Reported-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-24 05:22:57 +08:00
|
|
|
continue;
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
err = dsa_user_manage_vlan_filtering(user,
|
|
|
|
|
vlan_filtering);
|
net: dsa: don't advertise 'rx-vlan-filter' when not needed
There have been multiple independent reports about
dsa_slave_vlan_rx_add_vid being called (and consequently calling the
drivers' .port_vlan_add) when it isn't needed, and sometimes (not
always) causing problems in the process.
Case 1:
mv88e6xxx_port_vlan_prepare is stubborn and only accepts VLANs on
bridged ports. That is understandably so, because standalone mv88e6xxx
ports are VLAN-unaware, and VTU entries are said to be a scarce
resource.
Otherwise said, the following fails lamentably on mv88e6xxx:
ip link add br0 type bridge vlan_filtering 1
ip link set lan3 master br0
ip link add link lan10 name lan10.1 type vlan id 1
[485256.724147] mv88e6085 d0032004.mdio-mii:12: p10: hw VLAN 1 already used by port 3 in br0
RTNETLINK answers: Operation not supported
This has become a worse issue since commit 9b236d2a69da ("net: dsa:
Advertise the VLAN offload netdev ability only if switch supports it").
Up to that point, the driver was returning -EOPNOTSUPP and DSA was
reconverting that error to 0, making the 8021q upper think all is ok
(but obviously the error message was there even prior to this change).
After that change the -EOPNOTSUPP is propagated to vlan_vid_add, and it
is a hard error.
Case 2:
Ports that don't offload the Linux bridge (have a dp->bridge_dev = NULL
because they don't implement .port_bridge_{join,leave}). Understandably,
a standalone port should not offload VLANs either, it should remain VLAN
unaware and any VLAN should be a software VLAN (as long as the hardware
is not quirky, that is).
In fact, dsa_slave_port_obj_add does do the right thing and rejects
switchdev VLAN objects coming from the bridge when that bridge is not
offloaded:
case SWITCHDEV_OBJ_ID_PORT_VLAN:
if (!dsa_port_offloads_bridge_port(dp, obj->orig_dev))
return -EOPNOTSUPP;
err = dsa_slave_vlan_add(dev, obj, extack);
But it seems that the bridge is able to trick us. The __vlan_vid_add
from br_vlan.c has:
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q add.
*/
err = br_switchdev_port_vlan_add(dev, v->vid, flags, extack);
if (err == -EOPNOTSUPP)
return vlan_vid_add(dev, br->vlan_proto, v->vid);
So it says "no, no, you need this VLAN in your life!". And we, naive as
we are, say "oh, this comes from the vlan_vid_add code path, it must be
an 8021q upper, sure, I'll take that". And we end up with that bridge
VLAN installed on our port anyway. But this time, it has the wrong flags:
if the bridge was trying to install VLAN 1 as a pvid/untagged VLAN,
failed via switchdev, retried via vlan_vid_add, we have this comment:
/* This API only allows programming tagged, non-PVID VIDs */
So what we do makes absolutely no sense.
Backtracing a bit, we see the common pattern. We allow the network stack
to think that our standalone ports are VLAN-aware, but they aren't, for
the vast majority of switches. The quirky ones should not dictate the
norm. The dsa_slave_vlan_rx_add_vid and dsa_slave_vlan_rx_kill_vid
methods exist for drivers that need the 'rx-vlan-filter: on' feature in
ethtool -k, which can be due to any of the following reasons:
1. vlan_filtering_is_global = true, and some ports are under a
VLAN-aware bridge while others are standalone, and the standalone
ports would otherwise drop VLAN-tagged traffic. This is described in
commit 061f6a505ac3 ("net: dsa: Add ndo_vlan_rx_{add, kill}_vid
implementation").
2. the ports that are under a VLAN-aware bridge should also set this
feature, for 8021q uppers having a VID not claimed by the bridge.
In this case, the driver will essentially not even know that the VID
is coming from the 8021q layer and not the bridge.
3. Hellcreek. This driver needs it because in standalone mode, it uses
unique VLANs per port to ensure separation. For separation of untagged
traffic, it uses different PVIDs for each port, and for separation of
VLAN-tagged traffic, it never accepts 8021q uppers with the same vid
on two ports.
If a driver does not fall under any of the above 3 categories, there is
no reason why it should advertise the 'rx-vlan-filter' feature, therefore
no reason why it should offload the VLANs added through vlan_vid_add.
This commit fixes the problem by removing the 'rx-vlan-filter' feature
from the slave devices when they operate in standalone mode, and when
they offload a VLAN-unaware bridge.
The way it works is that vlan_vid_add will now stop its processing here:
vlan_add_rx_filter_info:
if (!vlan_hw_filter_capable(dev, proto))
return 0;
So the VLAN will still be saved in the interface's VLAN RX filtering
list, but because it does not declare VLAN filtering in its features,
the 8021q module will return zero without committing that VLAN to
hardware.
This gives the drivers what they want, since it keeps the 8021q VLANs
away from the VLAN table until VLAN awareness is enabled (point at which
the ports are no longer standalone, hence in the mv88e6xxx case, the
check in mv88e6xxx_port_vlan_prepare passes).
Since the issue predates the existence of the hellcreek driver, case 3
will be dealt with in a separate patch.
The main change that this patch makes is to no longer set
NETIF_F_HW_VLAN_CTAG_FILTER unconditionally, but toggle it dynamically
(for most switches, never).
The second part of the patch addresses an issue that the first part
introduces: because the 'rx-vlan-filter' feature is now dynamically
toggled, and our .ndo_vlan_rx_add_vid does not get called when
'rx-vlan-filter' is off, we need to avoid bugs such as the following by
replaying the VLANs from 8021q uppers every time we enable VLAN
filtering:
ip link add link lan0 name lan0.100 type vlan id 100
ip addr add 192.168.100.1/24 dev lan0.100
ping 192.168.100.2 # should work
ip link add br0 type bridge vlan_filtering 0
ip link set lan0 master br0
ping 192.168.100.2 # should still work
ip link set br0 type bridge vlan_filtering 1
ping 192.168.100.2 # should still work but doesn't
As reported by Florian, some drivers look at ds->vlan_filtering in
their .port_vlan_add() implementation. So this patch also makes sure
that ds->vlan_filtering is committed before calling the driver. This is
the reason why it is first committed, then restored on the failure path.
Reported-by: Tobias Waldekranz <tobias@waldekranz.com>
Reported-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-24 05:22:57 +08:00
|
|
|
if (err)
|
|
|
|
|
goto restore;
|
|
|
|
|
}
|
|
|
|
|
} else {
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
dp->vlan_filtering = vlan_filtering;
|
2020-10-03 06:06:46 +08:00
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
err = dsa_user_manage_vlan_filtering(dp->user,
|
|
|
|
|
vlan_filtering);
|
net: dsa: don't advertise 'rx-vlan-filter' when not needed
There have been multiple independent reports about
dsa_slave_vlan_rx_add_vid being called (and consequently calling the
drivers' .port_vlan_add) when it isn't needed, and sometimes (not
always) causing problems in the process.
Case 1:
mv88e6xxx_port_vlan_prepare is stubborn and only accepts VLANs on
bridged ports. That is understandably so, because standalone mv88e6xxx
ports are VLAN-unaware, and VTU entries are said to be a scarce
resource.
Otherwise said, the following fails lamentably on mv88e6xxx:
ip link add br0 type bridge vlan_filtering 1
ip link set lan3 master br0
ip link add link lan10 name lan10.1 type vlan id 1
[485256.724147] mv88e6085 d0032004.mdio-mii:12: p10: hw VLAN 1 already used by port 3 in br0
RTNETLINK answers: Operation not supported
This has become a worse issue since commit 9b236d2a69da ("net: dsa:
Advertise the VLAN offload netdev ability only if switch supports it").
Up to that point, the driver was returning -EOPNOTSUPP and DSA was
reconverting that error to 0, making the 8021q upper think all is ok
(but obviously the error message was there even prior to this change).
After that change the -EOPNOTSUPP is propagated to vlan_vid_add, and it
is a hard error.
Case 2:
Ports that don't offload the Linux bridge (have a dp->bridge_dev = NULL
because they don't implement .port_bridge_{join,leave}). Understandably,
a standalone port should not offload VLANs either, it should remain VLAN
unaware and any VLAN should be a software VLAN (as long as the hardware
is not quirky, that is).
In fact, dsa_slave_port_obj_add does do the right thing and rejects
switchdev VLAN objects coming from the bridge when that bridge is not
offloaded:
case SWITCHDEV_OBJ_ID_PORT_VLAN:
if (!dsa_port_offloads_bridge_port(dp, obj->orig_dev))
return -EOPNOTSUPP;
err = dsa_slave_vlan_add(dev, obj, extack);
But it seems that the bridge is able to trick us. The __vlan_vid_add
from br_vlan.c has:
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q add.
*/
err = br_switchdev_port_vlan_add(dev, v->vid, flags, extack);
if (err == -EOPNOTSUPP)
return vlan_vid_add(dev, br->vlan_proto, v->vid);
So it says "no, no, you need this VLAN in your life!". And we, naive as
we are, say "oh, this comes from the vlan_vid_add code path, it must be
an 8021q upper, sure, I'll take that". And we end up with that bridge
VLAN installed on our port anyway. But this time, it has the wrong flags:
if the bridge was trying to install VLAN 1 as a pvid/untagged VLAN,
failed via switchdev, retried via vlan_vid_add, we have this comment:
/* This API only allows programming tagged, non-PVID VIDs */
So what we do makes absolutely no sense.
Backtracing a bit, we see the common pattern. We allow the network stack
to think that our standalone ports are VLAN-aware, but they aren't, for
the vast majority of switches. The quirky ones should not dictate the
norm. The dsa_slave_vlan_rx_add_vid and dsa_slave_vlan_rx_kill_vid
methods exist for drivers that need the 'rx-vlan-filter: on' feature in
ethtool -k, which can be due to any of the following reasons:
1. vlan_filtering_is_global = true, and some ports are under a
VLAN-aware bridge while others are standalone, and the standalone
ports would otherwise drop VLAN-tagged traffic. This is described in
commit 061f6a505ac3 ("net: dsa: Add ndo_vlan_rx_{add, kill}_vid
implementation").
2. the ports that are under a VLAN-aware bridge should also set this
feature, for 8021q uppers having a VID not claimed by the bridge.
In this case, the driver will essentially not even know that the VID
is coming from the 8021q layer and not the bridge.
3. Hellcreek. This driver needs it because in standalone mode, it uses
unique VLANs per port to ensure separation. For separation of untagged
traffic, it uses different PVIDs for each port, and for separation of
VLAN-tagged traffic, it never accepts 8021q uppers with the same vid
on two ports.
If a driver does not fall under any of the above 3 categories, there is
no reason why it should advertise the 'rx-vlan-filter' feature, therefore
no reason why it should offload the VLANs added through vlan_vid_add.
This commit fixes the problem by removing the 'rx-vlan-filter' feature
from the slave devices when they operate in standalone mode, and when
they offload a VLAN-unaware bridge.
The way it works is that vlan_vid_add will now stop its processing here:
vlan_add_rx_filter_info:
if (!vlan_hw_filter_capable(dev, proto))
return 0;
So the VLAN will still be saved in the interface's VLAN RX filtering
list, but because it does not declare VLAN filtering in its features,
the 8021q module will return zero without committing that VLAN to
hardware.
This gives the drivers what they want, since it keeps the 8021q VLANs
away from the VLAN table until VLAN awareness is enabled (point at which
the ports are no longer standalone, hence in the mv88e6xxx case, the
check in mv88e6xxx_port_vlan_prepare passes).
Since the issue predates the existence of the hellcreek driver, case 3
will be dealt with in a separate patch.
The main change that this patch makes is to no longer set
NETIF_F_HW_VLAN_CTAG_FILTER unconditionally, but toggle it dynamically
(for most switches, never).
The second part of the patch addresses an issue that the first part
introduces: because the 'rx-vlan-filter' feature is now dynamically
toggled, and our .ndo_vlan_rx_add_vid does not get called when
'rx-vlan-filter' is off, we need to avoid bugs such as the following by
replaying the VLANs from 8021q uppers every time we enable VLAN
filtering:
ip link add link lan0 name lan0.100 type vlan id 100
ip addr add 192.168.100.1/24 dev lan0.100
ping 192.168.100.2 # should work
ip link add br0 type bridge vlan_filtering 0
ip link set lan0 master br0
ping 192.168.100.2 # should still work
ip link set br0 type bridge vlan_filtering 1
ping 192.168.100.2 # should still work but doesn't
As reported by Florian, some drivers look at ds->vlan_filtering in
their .port_vlan_add() implementation. So this patch also makes sure
that ds->vlan_filtering is committed before calling the driver. This is
the reason why it is first committed, then restored on the failure path.
Reported-by: Tobias Waldekranz <tobias@waldekranz.com>
Reported-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-24 05:22:57 +08:00
|
|
|
if (err)
|
|
|
|
|
goto restore;
|
|
|
|
|
}
|
|
|
|
|
|
2017-05-20 05:00:46 +08:00
|
|
|
return 0;
|
net: dsa: don't advertise 'rx-vlan-filter' when not needed
There have been multiple independent reports about
dsa_slave_vlan_rx_add_vid being called (and consequently calling the
drivers' .port_vlan_add) when it isn't needed, and sometimes (not
always) causing problems in the process.
Case 1:
mv88e6xxx_port_vlan_prepare is stubborn and only accepts VLANs on
bridged ports. That is understandably so, because standalone mv88e6xxx
ports are VLAN-unaware, and VTU entries are said to be a scarce
resource.
Otherwise said, the following fails lamentably on mv88e6xxx:
ip link add br0 type bridge vlan_filtering 1
ip link set lan3 master br0
ip link add link lan10 name lan10.1 type vlan id 1
[485256.724147] mv88e6085 d0032004.mdio-mii:12: p10: hw VLAN 1 already used by port 3 in br0
RTNETLINK answers: Operation not supported
This has become a worse issue since commit 9b236d2a69da ("net: dsa:
Advertise the VLAN offload netdev ability only if switch supports it").
Up to that point, the driver was returning -EOPNOTSUPP and DSA was
reconverting that error to 0, making the 8021q upper think all is ok
(but obviously the error message was there even prior to this change).
After that change the -EOPNOTSUPP is propagated to vlan_vid_add, and it
is a hard error.
Case 2:
Ports that don't offload the Linux bridge (have a dp->bridge_dev = NULL
because they don't implement .port_bridge_{join,leave}). Understandably,
a standalone port should not offload VLANs either, it should remain VLAN
unaware and any VLAN should be a software VLAN (as long as the hardware
is not quirky, that is).
In fact, dsa_slave_port_obj_add does do the right thing and rejects
switchdev VLAN objects coming from the bridge when that bridge is not
offloaded:
case SWITCHDEV_OBJ_ID_PORT_VLAN:
if (!dsa_port_offloads_bridge_port(dp, obj->orig_dev))
return -EOPNOTSUPP;
err = dsa_slave_vlan_add(dev, obj, extack);
But it seems that the bridge is able to trick us. The __vlan_vid_add
from br_vlan.c has:
/* Try switchdev op first. In case it is not supported, fallback to
* 8021q add.
*/
err = br_switchdev_port_vlan_add(dev, v->vid, flags, extack);
if (err == -EOPNOTSUPP)
return vlan_vid_add(dev, br->vlan_proto, v->vid);
So it says "no, no, you need this VLAN in your life!". And we, naive as
we are, say "oh, this comes from the vlan_vid_add code path, it must be
an 8021q upper, sure, I'll take that". And we end up with that bridge
VLAN installed on our port anyway. But this time, it has the wrong flags:
if the bridge was trying to install VLAN 1 as a pvid/untagged VLAN,
failed via switchdev, retried via vlan_vid_add, we have this comment:
/* This API only allows programming tagged, non-PVID VIDs */
So what we do makes absolutely no sense.
Backtracing a bit, we see the common pattern. We allow the network stack
to think that our standalone ports are VLAN-aware, but they aren't, for
the vast majority of switches. The quirky ones should not dictate the
norm. The dsa_slave_vlan_rx_add_vid and dsa_slave_vlan_rx_kill_vid
methods exist for drivers that need the 'rx-vlan-filter: on' feature in
ethtool -k, which can be due to any of the following reasons:
1. vlan_filtering_is_global = true, and some ports are under a
VLAN-aware bridge while others are standalone, and the standalone
ports would otherwise drop VLAN-tagged traffic. This is described in
commit 061f6a505ac3 ("net: dsa: Add ndo_vlan_rx_{add, kill}_vid
implementation").
2. the ports that are under a VLAN-aware bridge should also set this
feature, for 8021q uppers having a VID not claimed by the bridge.
In this case, the driver will essentially not even know that the VID
is coming from the 8021q layer and not the bridge.
3. Hellcreek. This driver needs it because in standalone mode, it uses
unique VLANs per port to ensure separation. For separation of untagged
traffic, it uses different PVIDs for each port, and for separation of
VLAN-tagged traffic, it never accepts 8021q uppers with the same vid
on two ports.
If a driver does not fall under any of the above 3 categories, there is
no reason why it should advertise the 'rx-vlan-filter' feature, therefore
no reason why it should offload the VLANs added through vlan_vid_add.
This commit fixes the problem by removing the 'rx-vlan-filter' feature
from the slave devices when they operate in standalone mode, and when
they offload a VLAN-unaware bridge.
The way it works is that vlan_vid_add will now stop its processing here:
vlan_add_rx_filter_info:
if (!vlan_hw_filter_capable(dev, proto))
return 0;
So the VLAN will still be saved in the interface's VLAN RX filtering
list, but because it does not declare VLAN filtering in its features,
the 8021q module will return zero without committing that VLAN to
hardware.
This gives the drivers what they want, since it keeps the 8021q VLANs
away from the VLAN table until VLAN awareness is enabled (point at which
the ports are no longer standalone, hence in the mv88e6xxx case, the
check in mv88e6xxx_port_vlan_prepare passes).
Since the issue predates the existence of the hellcreek driver, case 3
will be dealt with in a separate patch.
The main change that this patch makes is to no longer set
NETIF_F_HW_VLAN_CTAG_FILTER unconditionally, but toggle it dynamically
(for most switches, never).
The second part of the patch addresses an issue that the first part
introduces: because the 'rx-vlan-filter' feature is now dynamically
toggled, and our .ndo_vlan_rx_add_vid does not get called when
'rx-vlan-filter' is off, we need to avoid bugs such as the following by
replaying the VLANs from 8021q uppers every time we enable VLAN
filtering:
ip link add link lan0 name lan0.100 type vlan id 100
ip addr add 192.168.100.1/24 dev lan0.100
ping 192.168.100.2 # should work
ip link add br0 type bridge vlan_filtering 0
ip link set lan0 master br0
ping 192.168.100.2 # should still work
ip link set br0 type bridge vlan_filtering 1
ping 192.168.100.2 # should still work but doesn't
As reported by Florian, some drivers look at ds->vlan_filtering in
their .port_vlan_add() implementation. So this patch also makes sure
that ds->vlan_filtering is committed before calling the driver. This is
the reason why it is first committed, then restored on the failure path.
Reported-by: Tobias Waldekranz <tobias@waldekranz.com>
Reported-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-24 05:22:57 +08:00
|
|
|
|
|
|
|
|
restore:
|
|
|
|
|
ds->ops->port_vlan_filtering(ds, dp->index, old_vlan_filtering, NULL);
|
|
|
|
|
|
|
|
|
|
if (ds->vlan_filtering_is_global)
|
|
|
|
|
ds->vlan_filtering = old_vlan_filtering;
|
|
|
|
|
else
|
|
|
|
|
dp->vlan_filtering = old_vlan_filtering;
|
|
|
|
|
|
|
|
|
|
return err;
|
2017-05-20 05:00:46 +08:00
|
|
|
}
|
2017-05-20 05:00:47 +08:00
|
|
|
|
2020-05-13 01:20:25 +08:00
|
|
|
/* This enforces legacy behavior for switch drivers which assume they can't
|
2023-10-24 02:17:28 +08:00
|
|
|
* receive VLAN configuration when joining a bridge with vlan_filtering=0
|
2020-05-13 01:20:25 +08:00
|
|
|
*/
|
|
|
|
|
bool dsa_port_skip_vlan_configuration(struct dsa_port *dp)
|
|
|
|
|
{
|
2021-12-07 00:57:52 +08:00
|
|
|
struct net_device *br = dsa_port_bridge_dev_get(dp);
|
2020-05-13 01:20:25 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
2021-12-07 00:57:52 +08:00
|
|
|
if (!br)
|
2020-05-13 01:20:25 +08:00
|
|
|
return false;
|
|
|
|
|
|
2021-12-07 00:57:52 +08:00
|
|
|
return !ds->configure_vlan_while_not_filtering && !br_vlan_enabled(br);
|
2020-05-13 01:20:25 +08:00
|
|
|
}
|
|
|
|
|
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
int dsa_port_ageing_time(struct dsa_port *dp, clock_t ageing_clock)
|
2017-05-20 05:00:47 +08:00
|
|
|
{
|
|
|
|
|
unsigned long ageing_jiffies = clock_t_to_jiffies(ageing_clock);
|
|
|
|
|
unsigned int ageing_time = jiffies_to_msecs(ageing_jiffies);
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
struct dsa_notifier_ageing_time_info info;
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
info.ageing_time = ageing_time;
|
2017-05-20 05:00:47 +08:00
|
|
|
|
net: switchdev: remove the transaction structure from port attributes
Since the introduction of the switchdev API, port attributes were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
attribute notifier structures, and converts drivers to not look at this
member.
In part, this patch contains a revert of my previous commit 2e554a7a5d8a
("net: dsa: propagate switchdev vlan_filtering prepare phase to
drivers").
For the most part, the conversion was trivial except for:
- Rocker's world implementation based on Broadcom OF-DPA had an odd
implementation of ofdpa_port_attr_bridge_flags_set. The conversion was
done mechanically, by pasting the implementation twice, then only
keeping the code that would get executed during prepare phase on top,
then only keeping the code that gets executed during the commit phase
on bottom, then simplifying the resulting code until this was obtained.
- DSA's offloading of STP state, bridge flags, VLAN filtering and
multicast router could be converted right away. But the ageing time
could not, so a shim was introduced and this was left for a further
commit.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek
Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:50 +08:00
|
|
|
err = dsa_port_notify(dp, DSA_NOTIFIER_AGEING_TIME, &info);
|
|
|
|
|
if (err)
|
|
|
|
|
return err;
|
2017-05-20 05:00:47 +08:00
|
|
|
|
|
|
|
|
dp->ageing_time = ageing_time;
|
|
|
|
|
|
2021-01-09 08:01:51 +08:00
|
|
|
return 0;
|
2017-05-20 05:00:47 +08:00
|
|
|
}
|
2017-05-20 05:00:48 +08:00
|
|
|
|
2022-03-16 23:08:52 +08:00
|
|
|
int dsa_port_mst_enable(struct dsa_port *dp, bool on,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
|
|
|
|
{
|
2022-03-16 23:08:54 +08:00
|
|
|
if (on && !dsa_port_supports_mst(dp)) {
|
2022-03-16 23:08:52 +08:00
|
|
|
NL_SET_ERR_MSG_MOD(extack, "Hardware does not support MST");
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
net: switchdev: pass flags and mask to both {PRE_,}BRIDGE_FLAGS attributes
This switchdev attribute offers a counterproductive API for a driver
writer, because although br_switchdev_set_port_flag gets passed a
"flags" and a "mask", those are passed piecemeal to the driver, so while
the PRE_BRIDGE_FLAGS listener knows what changed because it has the
"mask", the BRIDGE_FLAGS listener doesn't, because it only has the final
value. But certain drivers can offload only certain combinations of
settings, like for example they cannot change unicast flooding
independently of multicast flooding - they must be both on or both off.
The way the information is passed to switchdev makes drivers not
expressive enough, and unable to reject this request ahead of time, in
the PRE_BRIDGE_FLAGS notifier, so they are forced to reject it during
the deferred BRIDGE_FLAGS attribute, where the rejection is currently
ignored.
This patch also changes drivers to make use of the "mask" field for edge
detection when possible.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Grygorii Strashko <grygorii.strashko@ti.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:55 +08:00
|
|
|
int dsa_port_pre_bridge_flags(const struct dsa_port *dp,
|
net: dsa: act as passthrough for bridge port flags
There are multiple ways in which a PORT_BRIDGE_FLAGS attribute can be
expressed by the bridge through switchdev, and not all of them can be
emulated by DSA mid-layer API at the same time.
One possible configuration is when the bridge offloads the port flags
using a mask that has a single bit set - therefore only one feature
should change. However, DSA currently groups together unicast and
multicast flooding in the .port_egress_floods method, which limits our
options when we try to add support for turning off broadcast flooding:
do we extend .port_egress_floods with a third parameter which b53 and
mv88e6xxx will ignore? But that means that the DSA layer, which
currently implements the PRE_BRIDGE_FLAGS attribute all by itself, will
see that .port_egress_floods is implemented, and will report that all 3
types of flooding are supported - not necessarily true.
Another configuration is when the user specifies more than one flag at
the same time, in the same netlink message. If we were to create one
individual function per offloadable bridge port flag, we would limit the
expressiveness of the switch driver of refusing certain combinations of
flag values. For example, a switch may not have an explicit knob for
flooding of unknown multicast, just for flooding in general. In that
case, the only correct thing to do is to allow changes to BR_FLOOD and
BR_MCAST_FLOOD in tandem, and never allow mismatched values. But having
a separate .port_set_unicast_flood and .port_set_multicast_flood would
not allow the driver to possibly reject that.
Also, DSA doesn't consider it necessary to inform the driver that a
SWITCHDEV_ATTR_ID_BRIDGE_MROUTER attribute was offloaded, because it
just calls .port_egress_floods for the CPU port. When we'll add support
for the plain SWITCHDEV_ATTR_ID_PORT_MROUTER, that will become a real
problem because the flood settings will need to be held statefully in
the DSA middle layer, otherwise changing the mrouter port attribute will
impact the flooding attribute. And that's _assuming_ that the underlying
hardware doesn't have anything else to do when a multicast router
attaches to a port than flood unknown traffic to it. If it does, there
will need to be a dedicated .port_set_mrouter anyway.
So we need to let the DSA drivers see the exact form that the bridge
passes this switchdev attribute in, otherwise we are standing in the
way. Therefore we also need to use this form of language when
communicating to the driver that it needs to configure its initial
(before bridge join) and final (after bridge leave) port flags.
The b53 and mv88e6xxx drivers are converted to the passthrough API and
their implementation of .port_egress_floods is split into two: a
function that configures unicast flooding and another for multicast.
The mv88e6xxx implementation is quite hairy, and it turns out that
the implementations of unknown unicast flooding are actually the same
for 6185 and for 6352:
behind the confusing names actually lie two individual bits:
NO_UNKNOWN_MC -> FLOOD_UC = 0x4 = BIT(2)
NO_UNKNOWN_UC -> FLOOD_MC = 0x8 = BIT(3)
so there was no reason to entangle them in the first place.
Whereas the 6185 writes to MV88E6185_PORT_CTL0_FORWARD_UNKNOWN of
PORT_CTL0, which has the exact same bit index. I have left the
implementations separate though, for the only reason that the names are
different enough to confuse me, since I am not able to double-check with
a user manual. The multicast flooding setting for 6185 is in a different
register than for 6352 though.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:56 +08:00
|
|
|
struct switchdev_brport_flags flags,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
2019-02-21 08:58:22 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
net: dsa: act as passthrough for bridge port flags
There are multiple ways in which a PORT_BRIDGE_FLAGS attribute can be
expressed by the bridge through switchdev, and not all of them can be
emulated by DSA mid-layer API at the same time.
One possible configuration is when the bridge offloads the port flags
using a mask that has a single bit set - therefore only one feature
should change. However, DSA currently groups together unicast and
multicast flooding in the .port_egress_floods method, which limits our
options when we try to add support for turning off broadcast flooding:
do we extend .port_egress_floods with a third parameter which b53 and
mv88e6xxx will ignore? But that means that the DSA layer, which
currently implements the PRE_BRIDGE_FLAGS attribute all by itself, will
see that .port_egress_floods is implemented, and will report that all 3
types of flooding are supported - not necessarily true.
Another configuration is when the user specifies more than one flag at
the same time, in the same netlink message. If we were to create one
individual function per offloadable bridge port flag, we would limit the
expressiveness of the switch driver of refusing certain combinations of
flag values. For example, a switch may not have an explicit knob for
flooding of unknown multicast, just for flooding in general. In that
case, the only correct thing to do is to allow changes to BR_FLOOD and
BR_MCAST_FLOOD in tandem, and never allow mismatched values. But having
a separate .port_set_unicast_flood and .port_set_multicast_flood would
not allow the driver to possibly reject that.
Also, DSA doesn't consider it necessary to inform the driver that a
SWITCHDEV_ATTR_ID_BRIDGE_MROUTER attribute was offloaded, because it
just calls .port_egress_floods for the CPU port. When we'll add support
for the plain SWITCHDEV_ATTR_ID_PORT_MROUTER, that will become a real
problem because the flood settings will need to be held statefully in
the DSA middle layer, otherwise changing the mrouter port attribute will
impact the flooding attribute. And that's _assuming_ that the underlying
hardware doesn't have anything else to do when a multicast router
attaches to a port than flood unknown traffic to it. If it does, there
will need to be a dedicated .port_set_mrouter anyway.
So we need to let the DSA drivers see the exact form that the bridge
passes this switchdev attribute in, otherwise we are standing in the
way. Therefore we also need to use this form of language when
communicating to the driver that it needs to configure its initial
(before bridge join) and final (after bridge leave) port flags.
The b53 and mv88e6xxx drivers are converted to the passthrough API and
their implementation of .port_egress_floods is split into two: a
function that configures unicast flooding and another for multicast.
The mv88e6xxx implementation is quite hairy, and it turns out that
the implementations of unknown unicast flooding are actually the same
for 6185 and for 6352:
behind the confusing names actually lie two individual bits:
NO_UNKNOWN_MC -> FLOOD_UC = 0x4 = BIT(2)
NO_UNKNOWN_UC -> FLOOD_MC = 0x8 = BIT(3)
so there was no reason to entangle them in the first place.
Whereas the 6185 writes to MV88E6185_PORT_CTL0_FORWARD_UNKNOWN of
PORT_CTL0, which has the exact same bit index. I have left the
implementations separate though, for the only reason that the names are
different enough to confuse me, since I am not able to double-check with
a user manual. The multicast flooding setting for 6185 is in a different
register than for 6352 though.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:56 +08:00
|
|
|
if (!ds->ops->port_pre_bridge_flags)
|
2019-02-21 08:58:22 +08:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
net: dsa: act as passthrough for bridge port flags
There are multiple ways in which a PORT_BRIDGE_FLAGS attribute can be
expressed by the bridge through switchdev, and not all of them can be
emulated by DSA mid-layer API at the same time.
One possible configuration is when the bridge offloads the port flags
using a mask that has a single bit set - therefore only one feature
should change. However, DSA currently groups together unicast and
multicast flooding in the .port_egress_floods method, which limits our
options when we try to add support for turning off broadcast flooding:
do we extend .port_egress_floods with a third parameter which b53 and
mv88e6xxx will ignore? But that means that the DSA layer, which
currently implements the PRE_BRIDGE_FLAGS attribute all by itself, will
see that .port_egress_floods is implemented, and will report that all 3
types of flooding are supported - not necessarily true.
Another configuration is when the user specifies more than one flag at
the same time, in the same netlink message. If we were to create one
individual function per offloadable bridge port flag, we would limit the
expressiveness of the switch driver of refusing certain combinations of
flag values. For example, a switch may not have an explicit knob for
flooding of unknown multicast, just for flooding in general. In that
case, the only correct thing to do is to allow changes to BR_FLOOD and
BR_MCAST_FLOOD in tandem, and never allow mismatched values. But having
a separate .port_set_unicast_flood and .port_set_multicast_flood would
not allow the driver to possibly reject that.
Also, DSA doesn't consider it necessary to inform the driver that a
SWITCHDEV_ATTR_ID_BRIDGE_MROUTER attribute was offloaded, because it
just calls .port_egress_floods for the CPU port. When we'll add support
for the plain SWITCHDEV_ATTR_ID_PORT_MROUTER, that will become a real
problem because the flood settings will need to be held statefully in
the DSA middle layer, otherwise changing the mrouter port attribute will
impact the flooding attribute. And that's _assuming_ that the underlying
hardware doesn't have anything else to do when a multicast router
attaches to a port than flood unknown traffic to it. If it does, there
will need to be a dedicated .port_set_mrouter anyway.
So we need to let the DSA drivers see the exact form that the bridge
passes this switchdev attribute in, otherwise we are standing in the
way. Therefore we also need to use this form of language when
communicating to the driver that it needs to configure its initial
(before bridge join) and final (after bridge leave) port flags.
The b53 and mv88e6xxx drivers are converted to the passthrough API and
their implementation of .port_egress_floods is split into two: a
function that configures unicast flooding and another for multicast.
The mv88e6xxx implementation is quite hairy, and it turns out that
the implementations of unknown unicast flooding are actually the same
for 6185 and for 6352:
behind the confusing names actually lie two individual bits:
NO_UNKNOWN_MC -> FLOOD_UC = 0x4 = BIT(2)
NO_UNKNOWN_UC -> FLOOD_MC = 0x8 = BIT(3)
so there was no reason to entangle them in the first place.
Whereas the 6185 writes to MV88E6185_PORT_CTL0_FORWARD_UNKNOWN of
PORT_CTL0, which has the exact same bit index. I have left the
implementations separate though, for the only reason that the names are
different enough to confuse me, since I am not able to double-check with
a user manual. The multicast flooding setting for 6185 is in a different
register than for 6352 though.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:56 +08:00
|
|
|
return ds->ops->port_pre_bridge_flags(ds, dp->index, flags, extack);
|
2019-02-21 08:58:22 +08:00
|
|
|
}
|
|
|
|
|
|
net: dsa: centralize fast ageing when address learning is turned off
Currently DSA leaves it down to device drivers to fast age the FDB on a
port when address learning is disabled on it. There are 2 reasons for
doing that in the first place:
- when address learning is disabled by user space, through
IFLA_BRPORT_LEARNING or the brport_attr_learning sysfs, what user
space typically wants to achieve is to operate in a mode with no
dynamic FDB entry on that port. But if the port is already up, some
addresses might have been already learned on it, and it seems silly to
wait for 5 minutes for them to expire until something useful can be
done.
- when a port leaves a bridge and becomes standalone, DSA turns off
address learning on it. This also has the nice side effect of flushing
the dynamically learned bridge FDB entries on it, which is a good idea
because standalone ports should not have bridge FDB entries on them.
We let drivers manage fast ageing under this condition because if DSA
were to do it, it would need to track each port's learning state, and
act upon the transition, which it currently doesn't.
But there are 2 reasons why doing it is better after all:
- drivers might get it wrong and not do it (see b53_port_set_learning)
- we would like to flush the dynamic entries from the software bridge
too, and letting drivers do that would be another pain point
So track the port learning state and trigger a fast age process
automatically within DSA.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 22:35:23 +08:00
|
|
|
int dsa_port_bridge_flags(struct dsa_port *dp,
|
net: dsa: act as passthrough for bridge port flags
There are multiple ways in which a PORT_BRIDGE_FLAGS attribute can be
expressed by the bridge through switchdev, and not all of them can be
emulated by DSA mid-layer API at the same time.
One possible configuration is when the bridge offloads the port flags
using a mask that has a single bit set - therefore only one feature
should change. However, DSA currently groups together unicast and
multicast flooding in the .port_egress_floods method, which limits our
options when we try to add support for turning off broadcast flooding:
do we extend .port_egress_floods with a third parameter which b53 and
mv88e6xxx will ignore? But that means that the DSA layer, which
currently implements the PRE_BRIDGE_FLAGS attribute all by itself, will
see that .port_egress_floods is implemented, and will report that all 3
types of flooding are supported - not necessarily true.
Another configuration is when the user specifies more than one flag at
the same time, in the same netlink message. If we were to create one
individual function per offloadable bridge port flag, we would limit the
expressiveness of the switch driver of refusing certain combinations of
flag values. For example, a switch may not have an explicit knob for
flooding of unknown multicast, just for flooding in general. In that
case, the only correct thing to do is to allow changes to BR_FLOOD and
BR_MCAST_FLOOD in tandem, and never allow mismatched values. But having
a separate .port_set_unicast_flood and .port_set_multicast_flood would
not allow the driver to possibly reject that.
Also, DSA doesn't consider it necessary to inform the driver that a
SWITCHDEV_ATTR_ID_BRIDGE_MROUTER attribute was offloaded, because it
just calls .port_egress_floods for the CPU port. When we'll add support
for the plain SWITCHDEV_ATTR_ID_PORT_MROUTER, that will become a real
problem because the flood settings will need to be held statefully in
the DSA middle layer, otherwise changing the mrouter port attribute will
impact the flooding attribute. And that's _assuming_ that the underlying
hardware doesn't have anything else to do when a multicast router
attaches to a port than flood unknown traffic to it. If it does, there
will need to be a dedicated .port_set_mrouter anyway.
So we need to let the DSA drivers see the exact form that the bridge
passes this switchdev attribute in, otherwise we are standing in the
way. Therefore we also need to use this form of language when
communicating to the driver that it needs to configure its initial
(before bridge join) and final (after bridge leave) port flags.
The b53 and mv88e6xxx drivers are converted to the passthrough API and
their implementation of .port_egress_floods is split into two: a
function that configures unicast flooding and another for multicast.
The mv88e6xxx implementation is quite hairy, and it turns out that
the implementations of unknown unicast flooding are actually the same
for 6185 and for 6352:
behind the confusing names actually lie two individual bits:
NO_UNKNOWN_MC -> FLOOD_UC = 0x4 = BIT(2)
NO_UNKNOWN_UC -> FLOOD_MC = 0x8 = BIT(3)
so there was no reason to entangle them in the first place.
Whereas the 6185 writes to MV88E6185_PORT_CTL0_FORWARD_UNKNOWN of
PORT_CTL0, which has the exact same bit index. I have left the
implementations separate though, for the only reason that the names are
different enough to confuse me, since I am not able to double-check with
a user manual. The multicast flooding setting for 6185 is in a different
register than for 6352 though.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:56 +08:00
|
|
|
struct switchdev_brport_flags flags,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
2019-02-21 07:35:04 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
net: dsa: centralize fast ageing when address learning is turned off
Currently DSA leaves it down to device drivers to fast age the FDB on a
port when address learning is disabled on it. There are 2 reasons for
doing that in the first place:
- when address learning is disabled by user space, through
IFLA_BRPORT_LEARNING or the brport_attr_learning sysfs, what user
space typically wants to achieve is to operate in a mode with no
dynamic FDB entry on that port. But if the port is already up, some
addresses might have been already learned on it, and it seems silly to
wait for 5 minutes for them to expire until something useful can be
done.
- when a port leaves a bridge and becomes standalone, DSA turns off
address learning on it. This also has the nice side effect of flushing
the dynamically learned bridge FDB entries on it, which is a good idea
because standalone ports should not have bridge FDB entries on them.
We let drivers manage fast ageing under this condition because if DSA
were to do it, it would need to track each port's learning state, and
act upon the transition, which it currently doesn't.
But there are 2 reasons why doing it is better after all:
- drivers might get it wrong and not do it (see b53_port_set_learning)
- we would like to flush the dynamic entries from the software bridge
too, and letting drivers do that would be another pain point
So track the port learning state and trigger a fast age process
automatically within DSA.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 22:35:23 +08:00
|
|
|
int err;
|
2019-02-21 07:35:04 +08:00
|
|
|
|
net: dsa: act as passthrough for bridge port flags
There are multiple ways in which a PORT_BRIDGE_FLAGS attribute can be
expressed by the bridge through switchdev, and not all of them can be
emulated by DSA mid-layer API at the same time.
One possible configuration is when the bridge offloads the port flags
using a mask that has a single bit set - therefore only one feature
should change. However, DSA currently groups together unicast and
multicast flooding in the .port_egress_floods method, which limits our
options when we try to add support for turning off broadcast flooding:
do we extend .port_egress_floods with a third parameter which b53 and
mv88e6xxx will ignore? But that means that the DSA layer, which
currently implements the PRE_BRIDGE_FLAGS attribute all by itself, will
see that .port_egress_floods is implemented, and will report that all 3
types of flooding are supported - not necessarily true.
Another configuration is when the user specifies more than one flag at
the same time, in the same netlink message. If we were to create one
individual function per offloadable bridge port flag, we would limit the
expressiveness of the switch driver of refusing certain combinations of
flag values. For example, a switch may not have an explicit knob for
flooding of unknown multicast, just for flooding in general. In that
case, the only correct thing to do is to allow changes to BR_FLOOD and
BR_MCAST_FLOOD in tandem, and never allow mismatched values. But having
a separate .port_set_unicast_flood and .port_set_multicast_flood would
not allow the driver to possibly reject that.
Also, DSA doesn't consider it necessary to inform the driver that a
SWITCHDEV_ATTR_ID_BRIDGE_MROUTER attribute was offloaded, because it
just calls .port_egress_floods for the CPU port. When we'll add support
for the plain SWITCHDEV_ATTR_ID_PORT_MROUTER, that will become a real
problem because the flood settings will need to be held statefully in
the DSA middle layer, otherwise changing the mrouter port attribute will
impact the flooding attribute. And that's _assuming_ that the underlying
hardware doesn't have anything else to do when a multicast router
attaches to a port than flood unknown traffic to it. If it does, there
will need to be a dedicated .port_set_mrouter anyway.
So we need to let the DSA drivers see the exact form that the bridge
passes this switchdev attribute in, otherwise we are standing in the
way. Therefore we also need to use this form of language when
communicating to the driver that it needs to configure its initial
(before bridge join) and final (after bridge leave) port flags.
The b53 and mv88e6xxx drivers are converted to the passthrough API and
their implementation of .port_egress_floods is split into two: a
function that configures unicast flooding and another for multicast.
The mv88e6xxx implementation is quite hairy, and it turns out that
the implementations of unknown unicast flooding are actually the same
for 6185 and for 6352:
behind the confusing names actually lie two individual bits:
NO_UNKNOWN_MC -> FLOOD_UC = 0x4 = BIT(2)
NO_UNKNOWN_UC -> FLOOD_MC = 0x8 = BIT(3)
so there was no reason to entangle them in the first place.
Whereas the 6185 writes to MV88E6185_PORT_CTL0_FORWARD_UNKNOWN of
PORT_CTL0, which has the exact same bit index. I have left the
implementations separate though, for the only reason that the names are
different enough to confuse me, since I am not able to double-check with
a user manual. The multicast flooding setting for 6185 is in a different
register than for 6352 though.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-02-12 23:15:56 +08:00
|
|
|
if (!ds->ops->port_bridge_flags)
|
2021-04-21 21:05:40 +08:00
|
|
|
return -EOPNOTSUPP;
|
2019-02-21 07:35:04 +08:00
|
|
|
|
net: dsa: centralize fast ageing when address learning is turned off
Currently DSA leaves it down to device drivers to fast age the FDB on a
port when address learning is disabled on it. There are 2 reasons for
doing that in the first place:
- when address learning is disabled by user space, through
IFLA_BRPORT_LEARNING or the brport_attr_learning sysfs, what user
space typically wants to achieve is to operate in a mode with no
dynamic FDB entry on that port. But if the port is already up, some
addresses might have been already learned on it, and it seems silly to
wait for 5 minutes for them to expire until something useful can be
done.
- when a port leaves a bridge and becomes standalone, DSA turns off
address learning on it. This also has the nice side effect of flushing
the dynamically learned bridge FDB entries on it, which is a good idea
because standalone ports should not have bridge FDB entries on them.
We let drivers manage fast ageing under this condition because if DSA
were to do it, it would need to track each port's learning state, and
act upon the transition, which it currently doesn't.
But there are 2 reasons why doing it is better after all:
- drivers might get it wrong and not do it (see b53_port_set_learning)
- we would like to flush the dynamic entries from the software bridge
too, and letting drivers do that would be another pain point
So track the port learning state and trigger a fast age process
automatically within DSA.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 22:35:23 +08:00
|
|
|
err = ds->ops->port_bridge_flags(ds, dp->index, flags, extack);
|
|
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
|
|
|
|
|
if (flags.mask & BR_LEARNING) {
|
|
|
|
|
bool learning = flags.val & BR_LEARNING;
|
|
|
|
|
|
|
|
|
|
if (learning == dp->learning)
|
|
|
|
|
return 0;
|
|
|
|
|
|
net: dsa: avoid fast ageing twice when port leaves a bridge
Drivers that support both the toggling of address learning and dynamic
FDB flushing (mv88e6xxx, b53, sja1105) currently need to fast-age a port
twice when it leaves a bridge:
- once, when del_nbp() calls br_stp_disable_port() which puts the port
in the BLOCKING state
- twice, when dsa_port_switchdev_unsync_attrs() calls
dsa_port_clear_brport_flags() which disables address learning
The knee-jerk reaction might be to say "dsa_port_clear_brport_flags does
not need to fast-age the port at all", but the thing is, we still need
both code paths to flush the dynamic FDB entries in different situations.
When a DSA switch port leaves a bonding/team interface that is (still) a
bridge port, no del_nbp() will be called, so we rely on
dsa_port_clear_brport_flags() function to restore proper standalone port
functionality with address learning disabled.
So the solution is just to avoid double the work when both code paths
are called in series. Luckily, DSA already caches the STP port state, so
we can skip flushing the dynamic FDB when we disable address learning
and the STP state is one where no address learning takes place at all.
Under that condition, not flushing the FDB is safe because there is
supposed to not be any dynamic FDB entry at all (they were flushed
during the transition towards that state, and none were learned in the
meanwhile).
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-09 06:56:49 +08:00
|
|
|
if ((dp->learning && !learning) &&
|
|
|
|
|
(dp->stp_state == BR_STATE_LEARNING ||
|
|
|
|
|
dp->stp_state == BR_STATE_FORWARDING))
|
net: dsa: centralize fast ageing when address learning is turned off
Currently DSA leaves it down to device drivers to fast age the FDB on a
port when address learning is disabled on it. There are 2 reasons for
doing that in the first place:
- when address learning is disabled by user space, through
IFLA_BRPORT_LEARNING or the brport_attr_learning sysfs, what user
space typically wants to achieve is to operate in a mode with no
dynamic FDB entry on that port. But if the port is already up, some
addresses might have been already learned on it, and it seems silly to
wait for 5 minutes for them to expire until something useful can be
done.
- when a port leaves a bridge and becomes standalone, DSA turns off
address learning on it. This also has the nice side effect of flushing
the dynamically learned bridge FDB entries on it, which is a good idea
because standalone ports should not have bridge FDB entries on them.
We let drivers manage fast ageing under this condition because if DSA
were to do it, it would need to track each port's learning state, and
act upon the transition, which it currently doesn't.
But there are 2 reasons why doing it is better after all:
- drivers might get it wrong and not do it (see b53_port_set_learning)
- we would like to flush the dynamic entries from the software bridge
too, and letting drivers do that would be another pain point
So track the port learning state and trigger a fast age process
automatically within DSA.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-08 22:35:23 +08:00
|
|
|
dsa_port_fast_age(dp);
|
|
|
|
|
|
|
|
|
|
dp->learning = learning;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
2019-02-21 07:35:04 +08:00
|
|
|
}
|
|
|
|
|
|
net: dsa: felix: manage host flooding using a specific driver callback
At the time - commit 7569459a52c9 ("net: dsa: manage flooding on the CPU
ports") - not introducing a dedicated switch callback for host flooding
made sense, because for the only user, the felix driver, there was
nothing different to do for the CPU port than set the flood flags on the
CPU port just like on any other bridge port.
There are 2 reasons why this approach is not good enough, however.
(1) Other drivers, like sja1105, support configuring flooding as a
function of {ingress port, egress port}, whereas the DSA
->port_bridge_flags() function only operates on an egress port.
So with that driver we'd have useless host flooding from user ports
which don't need it.
(2) Even with the felix driver, support for multiple CPU ports makes it
difficult to piggyback on ->port_bridge_flags(). The way in which
the felix driver is going to support host-filtered addresses with
multiple CPU ports is that it will direct these addresses towards
both CPU ports (in a sort of multicast fashion), then restrict the
forwarding to only one of the two using the forwarding masks.
Consequently, flooding will also be enabled towards both CPU ports.
However, ->port_bridge_flags() gets passed the index of a single CPU
port, and that leaves the flood settings out of sync between the 2
CPU ports.
This is to say, it's better to have a specific driver method for host
flooding, which takes the user port as argument. This solves problem (1)
by allowing the driver to do different things for different user ports,
and problem (2) by abstracting the operation and letting the driver do
whatever, rather than explicitly making the DSA core point to the CPU
port it thinks needs to be touched.
This new method also creates a problem, which is that cross-chip setups
are not handled. However I don't have hardware right now where I can
test what is the proper thing to do, and there isn't hardware compatible
with multi-switch trees that supports host flooding. So it remains a
problem to be tackled in the future.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-05-11 17:50:17 +08:00
|
|
|
void dsa_port_set_host_flood(struct dsa_port *dp, bool uc, bool mc)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
|
|
|
|
if (ds->ops->port_set_host_flood)
|
|
|
|
|
ds->ops->port_set_host_flood(ds, dp->index, uc, mc);
|
|
|
|
|
}
|
|
|
|
|
|
2022-03-16 23:08:53 +08:00
|
|
|
int dsa_port_vlan_msti(struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_vlan_msti *msti)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->vlan_msti_set)
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
|
|
|
|
|
return ds->ops->vlan_msti_set(ds, *dp->bridge, msti);
|
|
|
|
|
}
|
|
|
|
|
|
2022-04-15 23:46:26 +08:00
|
|
|
int dsa_port_mtu_change(struct dsa_port *dp, int new_mtu)
|
net: dsa: configure the MTU for switch ports
It is useful be able to configure port policers on a switch to accept
frames of various sizes:
- Increase the MTU for better throughput from the default of 1500 if it
is known that there is no 10/100 Mbps device in the network.
- Decrease the MTU to limit the latency of high-priority frames under
congestion, or work around various network segments that add extra
headers to packets which can't be fragmented.
For DSA slave ports, this is mostly a pass-through callback, called
through the regular ndo ops and at probe time (to ensure consistency
across all supported switches).
The CPU port is called with an MTU equal to the largest configured MTU
of the slave ports. The assumption is that the user might want to
sustain a bidirectional conversation with a partner over any switch
port.
The DSA master is configured the same as the CPU port, plus the tagger
overhead. Since the MTU is by definition L2 payload (sans Ethernet
header), it is up to each individual driver to figure out if it needs to
do anything special for its frame tags on the CPU port (it shouldn't
except in special cases). So the MTU does not contain the tagger
overhead on the CPU port.
However the MTU of the DSA master, minus the tagger overhead, is used as
a proxy for the MTU of the CPU port, which does not have a net device.
This is to avoid uselessly calling the .change_mtu function on the CPU
port when nothing should change.
So it is safe to assume that the DSA master and the CPU port MTUs are
apart by exactly the tagger's overhead in bytes.
Some changes were made around dsa_master_set_mtu(), function which was
now removed, for 2 reasons:
- dev_set_mtu() already calls dev_validate_mtu(), so it's redundant to
do the same thing in DSA
- __dev_set_mtu() returns 0 if ops->ndo_change_mtu is an absent method
That is to say, there's no need for this function in DSA, we can safely
call dev_set_mtu() directly, take the rtnl lock when necessary, and just
propagate whatever errors get reported (since the user probably wants to
be informed).
Some inspiration (mainly in the MTU DSA notifier) was taken from a
vaguely similar patch from Murali and Florian, who are credited as
co-developers down below.
Co-developed-by: Murali Krishna Policharla <murali.policharla@broadcom.com>
Signed-off-by: Murali Krishna Policharla <murali.policharla@broadcom.com>
Co-developed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-03-28 03:55:42 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_notifier_mtu_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
net: dsa: configure the MTU for switch ports
It is useful be able to configure port policers on a switch to accept
frames of various sizes:
- Increase the MTU for better throughput from the default of 1500 if it
is known that there is no 10/100 Mbps device in the network.
- Decrease the MTU to limit the latency of high-priority frames under
congestion, or work around various network segments that add extra
headers to packets which can't be fragmented.
For DSA slave ports, this is mostly a pass-through callback, called
through the regular ndo ops and at probe time (to ensure consistency
across all supported switches).
The CPU port is called with an MTU equal to the largest configured MTU
of the slave ports. The assumption is that the user might want to
sustain a bidirectional conversation with a partner over any switch
port.
The DSA master is configured the same as the CPU port, plus the tagger
overhead. Since the MTU is by definition L2 payload (sans Ethernet
header), it is up to each individual driver to figure out if it needs to
do anything special for its frame tags on the CPU port (it shouldn't
except in special cases). So the MTU does not contain the tagger
overhead on the CPU port.
However the MTU of the DSA master, minus the tagger overhead, is used as
a proxy for the MTU of the CPU port, which does not have a net device.
This is to avoid uselessly calling the .change_mtu function on the CPU
port when nothing should change.
So it is safe to assume that the DSA master and the CPU port MTUs are
apart by exactly the tagger's overhead in bytes.
Some changes were made around dsa_master_set_mtu(), function which was
now removed, for 2 reasons:
- dev_set_mtu() already calls dev_validate_mtu(), so it's redundant to
do the same thing in DSA
- __dev_set_mtu() returns 0 if ops->ndo_change_mtu is an absent method
That is to say, there's no need for this function in DSA, we can safely
call dev_set_mtu() directly, take the rtnl lock when necessary, and just
propagate whatever errors get reported (since the user probably wants to
be informed).
Some inspiration (mainly in the MTU DSA notifier) was taken from a
vaguely similar patch from Murali and Florian, who are credited as
co-developers down below.
Co-developed-by: Murali Krishna Policharla <murali.policharla@broadcom.com>
Signed-off-by: Murali Krishna Policharla <murali.policharla@broadcom.com>
Co-developed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2020-03-28 03:55:42 +08:00
|
|
|
.mtu = new_mtu,
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_MTU, &info);
|
|
|
|
|
}
|
|
|
|
|
|
2017-08-06 21:15:41 +08:00
|
|
|
int dsa_port_fdb_add(struct dsa_port *dp, const unsigned char *addr,
|
|
|
|
|
u16 vid)
|
2017-05-20 05:00:48 +08:00
|
|
|
{
|
2017-05-20 05:00:53 +08:00
|
|
|
struct dsa_notifier_fdb_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2017-08-06 21:15:41 +08:00
|
|
|
.addr = addr,
|
|
|
|
|
.vid = vid,
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
.db = {
|
|
|
|
|
.type = DSA_DB_BRIDGE,
|
|
|
|
|
.bridge = *dp->bridge,
|
|
|
|
|
},
|
2017-05-20 05:00:53 +08:00
|
|
|
};
|
2017-05-20 05:00:48 +08:00
|
|
|
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
/* Refcounting takes bridge.num as a key, and should be global for all
|
|
|
|
|
* bridges in the absence of FDB isolation, and per bridge otherwise.
|
|
|
|
|
* Force the bridge.num to zero here in the absence of FDB isolation.
|
|
|
|
|
*/
|
|
|
|
|
if (!dp->ds->fdb_isolation)
|
|
|
|
|
info.db.bridge.num = 0;
|
|
|
|
|
|
2017-05-20 05:00:53 +08:00
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_FDB_ADD, &info);
|
2017-05-20 05:00:48 +08:00
|
|
|
}
|
|
|
|
|
|
2017-08-06 21:15:41 +08:00
|
|
|
int dsa_port_fdb_del(struct dsa_port *dp, const unsigned char *addr,
|
|
|
|
|
u16 vid)
|
2017-05-20 05:00:48 +08:00
|
|
|
{
|
2017-05-20 05:00:53 +08:00
|
|
|
struct dsa_notifier_fdb_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2017-08-06 21:15:41 +08:00
|
|
|
.addr = addr,
|
|
|
|
|
.vid = vid,
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
.db = {
|
|
|
|
|
.type = DSA_DB_BRIDGE,
|
|
|
|
|
.bridge = *dp->bridge,
|
|
|
|
|
},
|
2017-05-20 05:00:53 +08:00
|
|
|
};
|
2017-05-20 05:00:48 +08:00
|
|
|
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
if (!dp->ds->fdb_isolation)
|
|
|
|
|
info.db.bridge.num = 0;
|
|
|
|
|
|
2017-05-20 05:00:53 +08:00
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_FDB_DEL, &info);
|
2017-05-20 05:00:48 +08:00
|
|
|
}
|
|
|
|
|
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
static int dsa_port_host_fdb_add(struct dsa_port *dp,
|
|
|
|
|
const unsigned char *addr, u16 vid,
|
|
|
|
|
struct dsa_db db)
|
2021-06-29 22:06:51 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_notifier_fdb_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2021-06-29 22:06:51 +08:00
|
|
|
.addr = addr,
|
|
|
|
|
.vid = vid,
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
.db = db,
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_HOST_FDB_ADD, &info);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_standalone_host_fdb_add(struct dsa_port *dp,
|
|
|
|
|
const unsigned char *addr, u16 vid)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_db db = {
|
|
|
|
|
.type = DSA_DB_PORT,
|
|
|
|
|
.dp = dp,
|
2021-06-29 22:06:51 +08:00
|
|
|
};
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
|
|
|
|
|
return dsa_port_host_fdb_add(dp, addr, vid, db);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_bridge_host_fdb_add(struct dsa_port *dp,
|
|
|
|
|
const unsigned char *addr, u16 vid)
|
|
|
|
|
{
|
2023-10-24 02:17:28 +08:00
|
|
|
struct net_device *conduit = dsa_port_to_conduit(dp);
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
struct dsa_db db = {
|
|
|
|
|
.type = DSA_DB_BRIDGE,
|
|
|
|
|
.bridge = *dp->bridge,
|
|
|
|
|
};
|
2021-06-29 22:06:53 +08:00
|
|
|
int err;
|
|
|
|
|
|
net: dsa: fix db type confusion in host fdb/mdb add/del
We have the following code paths:
Host FDB (unicast RX filtering):
dsa_port_standalone_host_fdb_add() dsa_port_bridge_host_fdb_add()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_fdb_add()
dsa_port_standalone_host_fdb_del() dsa_port_bridge_host_fdb_del()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_fdb_del()
Host MDB (multicast RX filtering):
dsa_port_standalone_host_mdb_add() dsa_port_bridge_host_mdb_add()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_mdb_add()
dsa_port_standalone_host_mdb_del() dsa_port_bridge_host_mdb_del()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_mdb_del()
The logic added by commit 5e8a1e03aa4d ("net: dsa: install secondary
unicast and multicast addresses as host FDB/MDB") zeroes out
db.bridge.num if the switch doesn't support ds->fdb_isolation
(the majority doesn't). This is done for a reason explained in commit
c26933639b54 ("net: dsa: request drivers to perform FDB isolation").
Taking a single code path as example - dsa_port_host_fdb_add() - the
others are similar - the problem is that this function handles:
- DSA_DB_PORT databases, when called from
dsa_port_standalone_host_fdb_add()
- DSA_DB_BRIDGE databases, when called from
dsa_port_bridge_host_fdb_add()
So, if dsa_port_host_fdb_add() were to make any change on the
"bridge.num" attribute of the database, this would only be correct for a
DSA_DB_BRIDGE, and a type confusion for a DSA_DB_PORT bridge.
However, this bug is without consequences, for 2 reasons:
- dsa_port_standalone_host_fdb_add() is only called from code which is
(in)directly guarded by dsa_switch_supports_uc_filtering(ds), and that
function only returns true if ds->fdb_isolation is set. So, the code
only executed for DSA_DB_BRIDGE databases.
- Even if the code was not dead for DSA_DB_PORT, we have the following
memory layout:
struct dsa_bridge {
struct net_device *dev;
unsigned int num;
bool tx_fwd_offload;
refcount_t refcount;
};
struct dsa_db {
enum dsa_db_type type;
union {
const struct dsa_port *dp; // DSA_DB_PORT
struct dsa_lag lag;
struct dsa_bridge bridge; // DSA_DB_BRIDGE
};
};
So, the zeroization of dsa_db :: bridge :: num on a dsa_db structure of
type DSA_DB_PORT would access memory which is unused, because we only
use dsa_db :: dp for DSA_DB_PORT, and this is mapped at the same address
with dsa_db :: dev for DSA_DB_BRIDGE, thanks to the union definition.
It is correct to fix up dsa_db :: bridge :: num only from code paths
that come from the bridge / switchdev, so move these there.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230329133819.697642-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-03-29 21:38:19 +08:00
|
|
|
if (!dp->ds->fdb_isolation)
|
|
|
|
|
db.bridge.num = 0;
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
/* Avoid a call to __dev_set_promiscuity() on the conduit, which
|
net: dsa: avoid call to __dev_set_promiscuity() while rtnl_mutex isn't held
If the DSA master doesn't support IFF_UNICAST_FLT, then the following
call path is possible:
dsa_slave_switchdev_event_work
-> dsa_port_host_fdb_add
-> dev_uc_add
-> __dev_set_rx_mode
-> __dev_set_promiscuity
Since the blamed commit, dsa_slave_switchdev_event_work() no longer
holds rtnl_lock(), which triggers the ASSERT_RTNL() from
__dev_set_promiscuity().
Taking rtnl_lock() around dev_uc_add() is impossible, because all the
code paths that call dsa_flush_workqueue() do so from contexts where the
rtnl_mutex is already held - so this would lead to an instant deadlock.
dev_uc_add() in itself doesn't require the rtnl_mutex for protection.
There is this comment in __dev_set_rx_mode() which assumes so:
/* Unicast addresses changes may only happen under the rtnl,
* therefore calling __dev_set_promiscuity here is safe.
*/
but it is from commit 4417da668c00 ("[NET]: dev: secondary unicast
address support") dated June 2007, and in the meantime, commit
f1f28aa3510d ("netdev: Add addr_list_lock to struct net_device."), dated
July 2008, has added &dev->addr_list_lock to protect this instead of the
global rtnl_mutex.
Nonetheless, __dev_set_promiscuity() does assume rtnl_mutex protection,
but it is the uncommon path of what we typically expect dev_uc_add()
to do. So since only the uncommon path requires rtnl_lock(), just check
ahead of time whether dev_uc_add() would result into a call to
__dev_set_promiscuity(), and handle that condition separately.
DSA already configures the master interface to be promiscuous if the
tagger requires this. We can extend this to also cover the case where
the master doesn't handle dev_uc_add() (doesn't support IFF_UNICAST_FLT),
and on the premise that we'd end up making it promiscuous during
operation anyway, either if a DSA slave has a non-inherited MAC address,
or if the bridge notifies local FDB entries for its own MAC address, the
address of a station learned on a foreign port, etc.
Fixes: 0faf890fc519 ("net: dsa: drop rtnl_lock from dsa_slave_switchdev_event_work")
Reported-by: Oleksij Rempel <o.rempel@pengutronix.de>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-18 20:13:02 +08:00
|
|
|
* requires rtnl_lock(), since we can't guarantee that is held here,
|
|
|
|
|
* and we can't take it either.
|
|
|
|
|
*/
|
2023-10-24 02:17:28 +08:00
|
|
|
if (conduit->priv_flags & IFF_UNICAST_FLT) {
|
|
|
|
|
err = dev_uc_add(conduit, addr);
|
net: dsa: avoid call to __dev_set_promiscuity() while rtnl_mutex isn't held
If the DSA master doesn't support IFF_UNICAST_FLT, then the following
call path is possible:
dsa_slave_switchdev_event_work
-> dsa_port_host_fdb_add
-> dev_uc_add
-> __dev_set_rx_mode
-> __dev_set_promiscuity
Since the blamed commit, dsa_slave_switchdev_event_work() no longer
holds rtnl_lock(), which triggers the ASSERT_RTNL() from
__dev_set_promiscuity().
Taking rtnl_lock() around dev_uc_add() is impossible, because all the
code paths that call dsa_flush_workqueue() do so from contexts where the
rtnl_mutex is already held - so this would lead to an instant deadlock.
dev_uc_add() in itself doesn't require the rtnl_mutex for protection.
There is this comment in __dev_set_rx_mode() which assumes so:
/* Unicast addresses changes may only happen under the rtnl,
* therefore calling __dev_set_promiscuity here is safe.
*/
but it is from commit 4417da668c00 ("[NET]: dev: secondary unicast
address support") dated June 2007, and in the meantime, commit
f1f28aa3510d ("netdev: Add addr_list_lock to struct net_device."), dated
July 2008, has added &dev->addr_list_lock to protect this instead of the
global rtnl_mutex.
Nonetheless, __dev_set_promiscuity() does assume rtnl_mutex protection,
but it is the uncommon path of what we typically expect dev_uc_add()
to do. So since only the uncommon path requires rtnl_lock(), just check
ahead of time whether dev_uc_add() would result into a call to
__dev_set_promiscuity(), and handle that condition separately.
DSA already configures the master interface to be promiscuous if the
tagger requires this. We can extend this to also cover the case where
the master doesn't handle dev_uc_add() (doesn't support IFF_UNICAST_FLT),
and on the premise that we'd end up making it promiscuous during
operation anyway, either if a DSA slave has a non-inherited MAC address,
or if the bridge notifies local FDB entries for its own MAC address, the
address of a station learned on a foreign port, etc.
Fixes: 0faf890fc519 ("net: dsa: drop rtnl_lock from dsa_slave_switchdev_event_work")
Reported-by: Oleksij Rempel <o.rempel@pengutronix.de>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-18 20:13:02 +08:00
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
}
|
2021-06-29 22:06:51 +08:00
|
|
|
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
return dsa_port_host_fdb_add(dp, addr, vid, db);
|
2021-06-29 22:06:51 +08:00
|
|
|
}
|
|
|
|
|
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
static int dsa_port_host_fdb_del(struct dsa_port *dp,
|
|
|
|
|
const unsigned char *addr, u16 vid,
|
|
|
|
|
struct dsa_db db)
|
2021-06-29 22:06:51 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_notifier_fdb_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2021-06-29 22:06:51 +08:00
|
|
|
.addr = addr,
|
|
|
|
|
.vid = vid,
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
.db = db,
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_HOST_FDB_DEL, &info);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_standalone_host_fdb_del(struct dsa_port *dp,
|
|
|
|
|
const unsigned char *addr, u16 vid)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_db db = {
|
|
|
|
|
.type = DSA_DB_PORT,
|
|
|
|
|
.dp = dp,
|
2021-06-29 22:06:51 +08:00
|
|
|
};
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
|
|
|
|
|
return dsa_port_host_fdb_del(dp, addr, vid, db);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_bridge_host_fdb_del(struct dsa_port *dp,
|
|
|
|
|
const unsigned char *addr, u16 vid)
|
|
|
|
|
{
|
2023-10-24 02:17:28 +08:00
|
|
|
struct net_device *conduit = dsa_port_to_conduit(dp);
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
struct dsa_db db = {
|
|
|
|
|
.type = DSA_DB_BRIDGE,
|
|
|
|
|
.bridge = *dp->bridge,
|
|
|
|
|
};
|
2021-06-29 22:06:53 +08:00
|
|
|
int err;
|
|
|
|
|
|
net: dsa: fix db type confusion in host fdb/mdb add/del
We have the following code paths:
Host FDB (unicast RX filtering):
dsa_port_standalone_host_fdb_add() dsa_port_bridge_host_fdb_add()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_fdb_add()
dsa_port_standalone_host_fdb_del() dsa_port_bridge_host_fdb_del()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_fdb_del()
Host MDB (multicast RX filtering):
dsa_port_standalone_host_mdb_add() dsa_port_bridge_host_mdb_add()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_mdb_add()
dsa_port_standalone_host_mdb_del() dsa_port_bridge_host_mdb_del()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_mdb_del()
The logic added by commit 5e8a1e03aa4d ("net: dsa: install secondary
unicast and multicast addresses as host FDB/MDB") zeroes out
db.bridge.num if the switch doesn't support ds->fdb_isolation
(the majority doesn't). This is done for a reason explained in commit
c26933639b54 ("net: dsa: request drivers to perform FDB isolation").
Taking a single code path as example - dsa_port_host_fdb_add() - the
others are similar - the problem is that this function handles:
- DSA_DB_PORT databases, when called from
dsa_port_standalone_host_fdb_add()
- DSA_DB_BRIDGE databases, when called from
dsa_port_bridge_host_fdb_add()
So, if dsa_port_host_fdb_add() were to make any change on the
"bridge.num" attribute of the database, this would only be correct for a
DSA_DB_BRIDGE, and a type confusion for a DSA_DB_PORT bridge.
However, this bug is without consequences, for 2 reasons:
- dsa_port_standalone_host_fdb_add() is only called from code which is
(in)directly guarded by dsa_switch_supports_uc_filtering(ds), and that
function only returns true if ds->fdb_isolation is set. So, the code
only executed for DSA_DB_BRIDGE databases.
- Even if the code was not dead for DSA_DB_PORT, we have the following
memory layout:
struct dsa_bridge {
struct net_device *dev;
unsigned int num;
bool tx_fwd_offload;
refcount_t refcount;
};
struct dsa_db {
enum dsa_db_type type;
union {
const struct dsa_port *dp; // DSA_DB_PORT
struct dsa_lag lag;
struct dsa_bridge bridge; // DSA_DB_BRIDGE
};
};
So, the zeroization of dsa_db :: bridge :: num on a dsa_db structure of
type DSA_DB_PORT would access memory which is unused, because we only
use dsa_db :: dp for DSA_DB_PORT, and this is mapped at the same address
with dsa_db :: dev for DSA_DB_BRIDGE, thanks to the union definition.
It is correct to fix up dsa_db :: bridge :: num only from code paths
that come from the bridge / switchdev, so move these there.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230329133819.697642-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-03-29 21:38:19 +08:00
|
|
|
if (!dp->ds->fdb_isolation)
|
|
|
|
|
db.bridge.num = 0;
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
if (conduit->priv_flags & IFF_UNICAST_FLT) {
|
|
|
|
|
err = dev_uc_del(conduit, addr);
|
net: dsa: avoid call to __dev_set_promiscuity() while rtnl_mutex isn't held
If the DSA master doesn't support IFF_UNICAST_FLT, then the following
call path is possible:
dsa_slave_switchdev_event_work
-> dsa_port_host_fdb_add
-> dev_uc_add
-> __dev_set_rx_mode
-> __dev_set_promiscuity
Since the blamed commit, dsa_slave_switchdev_event_work() no longer
holds rtnl_lock(), which triggers the ASSERT_RTNL() from
__dev_set_promiscuity().
Taking rtnl_lock() around dev_uc_add() is impossible, because all the
code paths that call dsa_flush_workqueue() do so from contexts where the
rtnl_mutex is already held - so this would lead to an instant deadlock.
dev_uc_add() in itself doesn't require the rtnl_mutex for protection.
There is this comment in __dev_set_rx_mode() which assumes so:
/* Unicast addresses changes may only happen under the rtnl,
* therefore calling __dev_set_promiscuity here is safe.
*/
but it is from commit 4417da668c00 ("[NET]: dev: secondary unicast
address support") dated June 2007, and in the meantime, commit
f1f28aa3510d ("netdev: Add addr_list_lock to struct net_device."), dated
July 2008, has added &dev->addr_list_lock to protect this instead of the
global rtnl_mutex.
Nonetheless, __dev_set_promiscuity() does assume rtnl_mutex protection,
but it is the uncommon path of what we typically expect dev_uc_add()
to do. So since only the uncommon path requires rtnl_lock(), just check
ahead of time whether dev_uc_add() would result into a call to
__dev_set_promiscuity(), and handle that condition separately.
DSA already configures the master interface to be promiscuous if the
tagger requires this. We can extend this to also cover the case where
the master doesn't handle dev_uc_add() (doesn't support IFF_UNICAST_FLT),
and on the premise that we'd end up making it promiscuous during
operation anyway, either if a DSA slave has a non-inherited MAC address,
or if the bridge notifies local FDB entries for its own MAC address, the
address of a station learned on a foreign port, etc.
Fixes: 0faf890fc519 ("net: dsa: drop rtnl_lock from dsa_slave_switchdev_event_work")
Reported-by: Oleksij Rempel <o.rempel@pengutronix.de>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-18 20:13:02 +08:00
|
|
|
if (err)
|
|
|
|
|
return err;
|
|
|
|
|
}
|
2021-06-29 22:06:51 +08:00
|
|
|
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
return dsa_port_host_fdb_del(dp, addr, vid, db);
|
2021-06-29 22:06:51 +08:00
|
|
|
}
|
|
|
|
|
|
net: dsa: support FDB events on offloaded LAG interfaces
This change introduces support for installing static FDB entries towards
a bridge port that is a LAG of multiple DSA switch ports, as well as
support for filtering towards the CPU local FDB entries emitted for LAG
interfaces that are bridge ports.
Conceptually, host addresses on LAG ports are identical to what we do
for plain bridge ports. Whereas FDB entries _towards_ a LAG can't simply
be replicated towards all member ports like we do for multicast, or VLAN.
Instead we need new driver API. Hardware usually considers a LAG to be a
"logical port", and sets the entire LAG as the forwarding destination.
The physical egress port selection within the LAG is made by hashing
policy, as usual.
To represent the logical port corresponding to the LAG, we pass by value
a copy of the dsa_lag structure to all switches in the tree that have at
least one port in that LAG.
To illustrate why a refcounted list of FDB entries is needed in struct
dsa_lag, it is enough to say that:
- a LAG may be a bridge port and may therefore receive FDB events even
while it isn't yet offloaded by any DSA interface
- DSA interfaces may be removed from a LAG while that is a bridge port;
we don't want FDB entries lingering around, but we don't want to
remove entries that are still in use, either
For all the cases below to work, the idea is to always keep an FDB entry
on a LAG with a reference count equal to the DSA member ports. So:
- if a port joins a LAG, it requests the bridge to replay the FDB, and
the FDB entries get created, or their refcount gets bumped by one
- if a port leaves a LAG, the FDB replay deletes or decrements refcount
by one
- if an FDB is installed towards a LAG with ports already present, that
entry is created (if it doesn't exist) and its refcount is bumped by
the amount of ports already present in the LAG
echo "Adding FDB entry to bond with existing ports"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond, then removing ports one by one"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link set swp1 nomaster
ip link set swp2 nomaster
ip link del br0
ip link del bond0
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-23 22:00:53 +08:00
|
|
|
int dsa_port_lag_fdb_add(struct dsa_port *dp, const unsigned char *addr,
|
|
|
|
|
u16 vid)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_notifier_lag_fdb_info info = {
|
|
|
|
|
.lag = dp->lag,
|
|
|
|
|
.addr = addr,
|
|
|
|
|
.vid = vid,
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
.db = {
|
|
|
|
|
.type = DSA_DB_BRIDGE,
|
|
|
|
|
.bridge = *dp->bridge,
|
|
|
|
|
},
|
net: dsa: support FDB events on offloaded LAG interfaces
This change introduces support for installing static FDB entries towards
a bridge port that is a LAG of multiple DSA switch ports, as well as
support for filtering towards the CPU local FDB entries emitted for LAG
interfaces that are bridge ports.
Conceptually, host addresses on LAG ports are identical to what we do
for plain bridge ports. Whereas FDB entries _towards_ a LAG can't simply
be replicated towards all member ports like we do for multicast, or VLAN.
Instead we need new driver API. Hardware usually considers a LAG to be a
"logical port", and sets the entire LAG as the forwarding destination.
The physical egress port selection within the LAG is made by hashing
policy, as usual.
To represent the logical port corresponding to the LAG, we pass by value
a copy of the dsa_lag structure to all switches in the tree that have at
least one port in that LAG.
To illustrate why a refcounted list of FDB entries is needed in struct
dsa_lag, it is enough to say that:
- a LAG may be a bridge port and may therefore receive FDB events even
while it isn't yet offloaded by any DSA interface
- DSA interfaces may be removed from a LAG while that is a bridge port;
we don't want FDB entries lingering around, but we don't want to
remove entries that are still in use, either
For all the cases below to work, the idea is to always keep an FDB entry
on a LAG with a reference count equal to the DSA member ports. So:
- if a port joins a LAG, it requests the bridge to replay the FDB, and
the FDB entries get created, or their refcount gets bumped by one
- if a port leaves a LAG, the FDB replay deletes or decrements refcount
by one
- if an FDB is installed towards a LAG with ports already present, that
entry is created (if it doesn't exist) and its refcount is bumped by
the amount of ports already present in the LAG
echo "Adding FDB entry to bond with existing ports"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond, then removing ports one by one"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link set swp1 nomaster
ip link set swp2 nomaster
ip link del br0
ip link del bond0
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-23 22:00:53 +08:00
|
|
|
};
|
|
|
|
|
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
if (!dp->ds->fdb_isolation)
|
|
|
|
|
info.db.bridge.num = 0;
|
|
|
|
|
|
net: dsa: support FDB events on offloaded LAG interfaces
This change introduces support for installing static FDB entries towards
a bridge port that is a LAG of multiple DSA switch ports, as well as
support for filtering towards the CPU local FDB entries emitted for LAG
interfaces that are bridge ports.
Conceptually, host addresses on LAG ports are identical to what we do
for plain bridge ports. Whereas FDB entries _towards_ a LAG can't simply
be replicated towards all member ports like we do for multicast, or VLAN.
Instead we need new driver API. Hardware usually considers a LAG to be a
"logical port", and sets the entire LAG as the forwarding destination.
The physical egress port selection within the LAG is made by hashing
policy, as usual.
To represent the logical port corresponding to the LAG, we pass by value
a copy of the dsa_lag structure to all switches in the tree that have at
least one port in that LAG.
To illustrate why a refcounted list of FDB entries is needed in struct
dsa_lag, it is enough to say that:
- a LAG may be a bridge port and may therefore receive FDB events even
while it isn't yet offloaded by any DSA interface
- DSA interfaces may be removed from a LAG while that is a bridge port;
we don't want FDB entries lingering around, but we don't want to
remove entries that are still in use, either
For all the cases below to work, the idea is to always keep an FDB entry
on a LAG with a reference count equal to the DSA member ports. So:
- if a port joins a LAG, it requests the bridge to replay the FDB, and
the FDB entries get created, or their refcount gets bumped by one
- if a port leaves a LAG, the FDB replay deletes or decrements refcount
by one
- if an FDB is installed towards a LAG with ports already present, that
entry is created (if it doesn't exist) and its refcount is bumped by
the amount of ports already present in the LAG
echo "Adding FDB entry to bond with existing ports"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond, then removing ports one by one"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link set swp1 nomaster
ip link set swp2 nomaster
ip link del br0
ip link del bond0
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-23 22:00:53 +08:00
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_LAG_FDB_ADD, &info);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_lag_fdb_del(struct dsa_port *dp, const unsigned char *addr,
|
|
|
|
|
u16 vid)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_notifier_lag_fdb_info info = {
|
|
|
|
|
.lag = dp->lag,
|
|
|
|
|
.addr = addr,
|
|
|
|
|
.vid = vid,
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
.db = {
|
|
|
|
|
.type = DSA_DB_BRIDGE,
|
|
|
|
|
.bridge = *dp->bridge,
|
|
|
|
|
},
|
net: dsa: support FDB events on offloaded LAG interfaces
This change introduces support for installing static FDB entries towards
a bridge port that is a LAG of multiple DSA switch ports, as well as
support for filtering towards the CPU local FDB entries emitted for LAG
interfaces that are bridge ports.
Conceptually, host addresses on LAG ports are identical to what we do
for plain bridge ports. Whereas FDB entries _towards_ a LAG can't simply
be replicated towards all member ports like we do for multicast, or VLAN.
Instead we need new driver API. Hardware usually considers a LAG to be a
"logical port", and sets the entire LAG as the forwarding destination.
The physical egress port selection within the LAG is made by hashing
policy, as usual.
To represent the logical port corresponding to the LAG, we pass by value
a copy of the dsa_lag structure to all switches in the tree that have at
least one port in that LAG.
To illustrate why a refcounted list of FDB entries is needed in struct
dsa_lag, it is enough to say that:
- a LAG may be a bridge port and may therefore receive FDB events even
while it isn't yet offloaded by any DSA interface
- DSA interfaces may be removed from a LAG while that is a bridge port;
we don't want FDB entries lingering around, but we don't want to
remove entries that are still in use, either
For all the cases below to work, the idea is to always keep an FDB entry
on a LAG with a reference count equal to the DSA member ports. So:
- if a port joins a LAG, it requests the bridge to replay the FDB, and
the FDB entries get created, or their refcount gets bumped by one
- if a port leaves a LAG, the FDB replay deletes or decrements refcount
by one
- if an FDB is installed towards a LAG with ports already present, that
entry is created (if it doesn't exist) and its refcount is bumped by
the amount of ports already present in the LAG
echo "Adding FDB entry to bond with existing ports"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond, then removing ports one by one"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link set swp1 nomaster
ip link set swp2 nomaster
ip link del br0
ip link del bond0
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-23 22:00:53 +08:00
|
|
|
};
|
|
|
|
|
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
if (!dp->ds->fdb_isolation)
|
|
|
|
|
info.db.bridge.num = 0;
|
|
|
|
|
|
net: dsa: support FDB events on offloaded LAG interfaces
This change introduces support for installing static FDB entries towards
a bridge port that is a LAG of multiple DSA switch ports, as well as
support for filtering towards the CPU local FDB entries emitted for LAG
interfaces that are bridge ports.
Conceptually, host addresses on LAG ports are identical to what we do
for plain bridge ports. Whereas FDB entries _towards_ a LAG can't simply
be replicated towards all member ports like we do for multicast, or VLAN.
Instead we need new driver API. Hardware usually considers a LAG to be a
"logical port", and sets the entire LAG as the forwarding destination.
The physical egress port selection within the LAG is made by hashing
policy, as usual.
To represent the logical port corresponding to the LAG, we pass by value
a copy of the dsa_lag structure to all switches in the tree that have at
least one port in that LAG.
To illustrate why a refcounted list of FDB entries is needed in struct
dsa_lag, it is enough to say that:
- a LAG may be a bridge port and may therefore receive FDB events even
while it isn't yet offloaded by any DSA interface
- DSA interfaces may be removed from a LAG while that is a bridge port;
we don't want FDB entries lingering around, but we don't want to
remove entries that are still in use, either
For all the cases below to work, the idea is to always keep an FDB entry
on a LAG with a reference count equal to the DSA member ports. So:
- if a port joins a LAG, it requests the bridge to replay the FDB, and
the FDB entries get created, or their refcount gets bumped by one
- if a port leaves a LAG, the FDB replay deletes or decrements refcount
by one
- if an FDB is installed towards a LAG with ports already present, that
entry is created (if it doesn't exist) and its refcount is bumped by
the amount of ports already present in the LAG
echo "Adding FDB entry to bond with existing ports"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link del br0
ip link del bond0
echo "Adding FDB entry to empty bond, then removing ports one by one"
ip link del bond0
ip link add bond0 type bond mode 802.3ad
ip link del br0
ip link add br0 type bridge
ip link set bond0 master br0
bridge fdb add dev bond0 00:01:02:03:04:05 master static
ip link set swp1 down && ip link set swp1 master bond0 && ip link set swp1 up
ip link set swp2 down && ip link set swp2 master bond0 && ip link set swp2 up
ip link set swp1 nomaster
ip link set swp2 nomaster
ip link del br0
ip link del bond0
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-23 22:00:53 +08:00
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_LAG_FDB_DEL, &info);
|
|
|
|
|
}
|
|
|
|
|
|
2017-09-21 07:32:14 +08:00
|
|
|
int dsa_port_fdb_dump(struct dsa_port *dp, dsa_fdb_dump_cb_t *cb, void *data)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
int port = dp->index;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->port_fdb_dump)
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
|
|
|
|
|
return ds->ops->port_fdb_dump(ds, port, cb, data);
|
|
|
|
|
}
|
|
|
|
|
|
2017-11-10 06:11:01 +08:00
|
|
|
int dsa_port_mdb_add(const struct dsa_port *dp,
|
net: switchdev: remove the transaction structure from port object notifiers
Since the introduction of the switchdev API, port objects were
transmitted to drivers for offloading using a two-step transactional
model, with a prepare phase that was supposed to catch all errors, and a
commit phase that was supposed to never fail.
Some classes of failures can never be avoided, like hardware access, or
memory allocation. In the latter case, merely attempting to move the
memory allocation to the preparation phase makes it impossible to avoid
memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused
transaction item queue") which has removed the unused mechanism of
passing on the allocated memory between one phase and another.
It is time we admit that separating the preparation from the commit
phase is something that is best left for the driver to decide, and not
something that should be baked into the API, especially since there are
no switchdev callers that depend on this.
This patch removes the struct switchdev_trans member from switchdev port
object notifier structures, and converts drivers to not look at this
member.
Where driver conversion is trivial (like in the case of the Marvell
Prestera driver, NXP DPAA2 switch, TI CPSW, and Rocker drivers), it is
done in this patch.
Where driver conversion needs more attention (DSA, Mellanox Spectrum),
the conversion is left for subsequent patches and here we only fake the
prepare/commit phases at a lower level, just not in the switchdev
notifier itself.
Where the code has a natural structure that is best left alone as a
preparation and a commit phase (as in the case of the Ocelot switch),
that structure is left in place, just made to not depend upon the
switchdev transactional model.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 08:01:48 +08:00
|
|
|
const struct switchdev_obj_port_mdb *mdb)
|
2017-05-20 05:00:49 +08:00
|
|
|
{
|
2017-05-20 05:00:54 +08:00
|
|
|
struct dsa_notifier_mdb_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2017-05-20 05:00:54 +08:00
|
|
|
.mdb = mdb,
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
.db = {
|
|
|
|
|
.type = DSA_DB_BRIDGE,
|
|
|
|
|
.bridge = *dp->bridge,
|
|
|
|
|
},
|
2017-05-20 05:00:54 +08:00
|
|
|
};
|
2017-05-20 05:00:49 +08:00
|
|
|
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
if (!dp->ds->fdb_isolation)
|
|
|
|
|
info.db.bridge.num = 0;
|
|
|
|
|
|
2017-05-20 05:00:54 +08:00
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_MDB_ADD, &info);
|
2017-05-20 05:00:49 +08:00
|
|
|
}
|
|
|
|
|
|
2017-11-10 06:11:01 +08:00
|
|
|
int dsa_port_mdb_del(const struct dsa_port *dp,
|
2017-05-20 05:00:49 +08:00
|
|
|
const struct switchdev_obj_port_mdb *mdb)
|
|
|
|
|
{
|
2017-05-20 05:00:54 +08:00
|
|
|
struct dsa_notifier_mdb_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2017-05-20 05:00:54 +08:00
|
|
|
.mdb = mdb,
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
.db = {
|
|
|
|
|
.type = DSA_DB_BRIDGE,
|
|
|
|
|
.bridge = *dp->bridge,
|
|
|
|
|
},
|
2017-05-20 05:00:54 +08:00
|
|
|
};
|
2017-05-20 05:00:49 +08:00
|
|
|
|
net: dsa: request drivers to perform FDB isolation
For DSA, to encourage drivers to perform FDB isolation simply means to
track which bridge does each FDB and MDB entry belong to. It then
becomes the driver responsibility to use something that makes the FDB
entry from one bridge not match the FDB lookup of ports from other
bridges.
The top-level functions where the bridge is determined are:
- dsa_port_fdb_{add,del}
- dsa_port_host_fdb_{add,del}
- dsa_port_mdb_{add,del}
- dsa_port_host_mdb_{add,del}
aka the pre-crosschip-notifier functions.
Changing the API to pass a reference to a bridge is not superfluous, and
looking at the passed bridge argument is not the same as having the
driver look at dsa_to_port(ds, port)->bridge from the ->port_fdb_add()
method.
DSA installs FDB and MDB entries on shared (CPU and DSA) ports as well,
and those do not have any dp->bridge information to retrieve, because
they are not in any bridge - they are merely the pipes that serve the
user ports that are in one or multiple bridges.
The struct dsa_bridge associated with each FDB/MDB entry is encapsulated
in a larger "struct dsa_db" database. Although only databases associated
to bridges are notified for now, this API will be the starting point for
implementing IFF_UNICAST_FLT in DSA. There, the idea is to install FDB
entries on the CPU port which belong to the corresponding user port's
port database. These are supposed to match only when the port is
standalone.
It is better to introduce the API in its expected final form than to
introduce it for bridges first, then to have to change drivers which may
have made one or more assumptions.
Drivers can use the provided bridge.num, but they can also use a
different numbering scheme that is more convenient.
DSA must perform refcounting on the CPU and DSA ports by also taking
into account the bridge number. So if two bridges request the same local
address, DSA must notify the driver twice, once for each bridge.
In fact, if the driver supports FDB isolation, DSA must perform
refcounting per bridge, but if the driver doesn't, DSA must refcount
host addresses across all bridges, otherwise it would be telling the
driver to delete an FDB entry for a bridge and the driver would delete
it for all bridges. So introduce a bool fdb_isolation in drivers which
would make all bridge databases passed to the cross-chip notifier have
the same number (0). This makes dsa_mac_addr_find() -> dsa_db_equal()
say that all bridge databases are the same database - which is
essentially the legacy behavior.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-25 17:22:22 +08:00
|
|
|
if (!dp->ds->fdb_isolation)
|
|
|
|
|
info.db.bridge.num = 0;
|
|
|
|
|
|
2017-05-20 05:00:54 +08:00
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_MDB_DEL, &info);
|
2017-05-20 05:00:49 +08:00
|
|
|
}
|
|
|
|
|
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
static int dsa_port_host_mdb_add(const struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_port_mdb *mdb,
|
|
|
|
|
struct dsa_db db)
|
net: dsa: introduce a separate cross-chip notifier type for host MDBs
Commit abd49535c380 ("net: dsa: execute dsa_switch_mdb_add only for
routing port in cross-chip topologies") does a surprisingly good job
even for the SWITCHDEV_OBJ_ID_HOST_MDB use case, where DSA simply
translates a switchdev object received on dp into a cross-chip notifier
for dp->cpu_dp.
To visualize how that works, imagine the daisy chain topology below and
consider a SWITCHDEV_OBJ_ID_HOST_MDB object emitted on sw2p0. How does
the cross-chip notifier know to match on all the right ports (sw0p4, the
dedicated CPU port, sw1p4, an upstream DSA link, and sw2p4, another
upstream DSA link)?
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ user ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
The answer is simple: the dedicated CPU port of sw2p0 is sw0p4, and
dsa_routing_port returns the upstream port for all switches.
That is fine, but there are other topologies where this does not work as
well. There are trees with "H" topologies in the wild, where there are 2
or more switches with DSA links between them, but every switch has its
dedicated CPU port. For these topologies, it seems stupid for the neighbor
switches to install an MDB entry on the routing port, since these
multicast addresses are fundamentally different than the usual ones we
support (and that is the justification for this patch, to introduce the
concept of a termination plane multicast MAC address, as opposed to a
forwarding plane multicast MAC address).
For example, when a SWITCHDEV_OBJ_ID_HOST_MDB would get added to sw0p0,
without this patch, it would get treated as a regular port MDB on sw0p2
and it would match on the ports below (including the sw1p3 routing port).
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ x ] [ ] [ ] [ ]
With the patch, the host MDB notifier on sw0p0 matches only on the local
switch, which is what we want for a termination plane address.
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ ] [ ] [ ] [ ]
Name this new matching function "dsa_switch_host_address_match" since we
will be reusing it soon for host FDB entries as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-06-29 22:06:49 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_notifier_mdb_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
net: dsa: introduce a separate cross-chip notifier type for host MDBs
Commit abd49535c380 ("net: dsa: execute dsa_switch_mdb_add only for
routing port in cross-chip topologies") does a surprisingly good job
even for the SWITCHDEV_OBJ_ID_HOST_MDB use case, where DSA simply
translates a switchdev object received on dp into a cross-chip notifier
for dp->cpu_dp.
To visualize how that works, imagine the daisy chain topology below and
consider a SWITCHDEV_OBJ_ID_HOST_MDB object emitted on sw2p0. How does
the cross-chip notifier know to match on all the right ports (sw0p4, the
dedicated CPU port, sw1p4, an upstream DSA link, and sw2p4, another
upstream DSA link)?
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ user ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
The answer is simple: the dedicated CPU port of sw2p0 is sw0p4, and
dsa_routing_port returns the upstream port for all switches.
That is fine, but there are other topologies where this does not work as
well. There are trees with "H" topologies in the wild, where there are 2
or more switches with DSA links between them, but every switch has its
dedicated CPU port. For these topologies, it seems stupid for the neighbor
switches to install an MDB entry on the routing port, since these
multicast addresses are fundamentally different than the usual ones we
support (and that is the justification for this patch, to introduce the
concept of a termination plane multicast MAC address, as opposed to a
forwarding plane multicast MAC address).
For example, when a SWITCHDEV_OBJ_ID_HOST_MDB would get added to sw0p0,
without this patch, it would get treated as a regular port MDB on sw0p2
and it would match on the ports below (including the sw1p3 routing port).
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ x ] [ ] [ ] [ ]
With the patch, the host MDB notifier on sw0p0 matches only on the local
switch, which is what we want for a termination plane address.
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ ] [ ] [ ] [ ]
Name this new matching function "dsa_switch_host_address_match" since we
will be reusing it soon for host FDB entries as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-06-29 22:06:49 +08:00
|
|
|
.mdb = mdb,
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
.db = db,
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_HOST_MDB_ADD, &info);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_standalone_host_mdb_add(const struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_port_mdb *mdb)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_db db = {
|
|
|
|
|
.type = DSA_DB_PORT,
|
|
|
|
|
.dp = dp,
|
net: dsa: introduce a separate cross-chip notifier type for host MDBs
Commit abd49535c380 ("net: dsa: execute dsa_switch_mdb_add only for
routing port in cross-chip topologies") does a surprisingly good job
even for the SWITCHDEV_OBJ_ID_HOST_MDB use case, where DSA simply
translates a switchdev object received on dp into a cross-chip notifier
for dp->cpu_dp.
To visualize how that works, imagine the daisy chain topology below and
consider a SWITCHDEV_OBJ_ID_HOST_MDB object emitted on sw2p0. How does
the cross-chip notifier know to match on all the right ports (sw0p4, the
dedicated CPU port, sw1p4, an upstream DSA link, and sw2p4, another
upstream DSA link)?
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ user ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
The answer is simple: the dedicated CPU port of sw2p0 is sw0p4, and
dsa_routing_port returns the upstream port for all switches.
That is fine, but there are other topologies where this does not work as
well. There are trees with "H" topologies in the wild, where there are 2
or more switches with DSA links between them, but every switch has its
dedicated CPU port. For these topologies, it seems stupid for the neighbor
switches to install an MDB entry on the routing port, since these
multicast addresses are fundamentally different than the usual ones we
support (and that is the justification for this patch, to introduce the
concept of a termination plane multicast MAC address, as opposed to a
forwarding plane multicast MAC address).
For example, when a SWITCHDEV_OBJ_ID_HOST_MDB would get added to sw0p0,
without this patch, it would get treated as a regular port MDB on sw0p2
and it would match on the ports below (including the sw1p3 routing port).
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ x ] [ ] [ ] [ ]
With the patch, the host MDB notifier on sw0p0 matches only on the local
switch, which is what we want for a termination plane address.
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ ] [ ] [ ] [ ]
Name this new matching function "dsa_switch_host_address_match" since we
will be reusing it soon for host FDB entries as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-06-29 22:06:49 +08:00
|
|
|
};
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
|
|
|
|
|
return dsa_port_host_mdb_add(dp, mdb, db);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_bridge_host_mdb_add(const struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_port_mdb *mdb)
|
|
|
|
|
{
|
2023-10-24 02:17:28 +08:00
|
|
|
struct net_device *conduit = dsa_port_to_conduit(dp);
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
struct dsa_db db = {
|
|
|
|
|
.type = DSA_DB_BRIDGE,
|
|
|
|
|
.bridge = *dp->bridge,
|
|
|
|
|
};
|
2021-06-29 22:06:53 +08:00
|
|
|
int err;
|
|
|
|
|
|
net: dsa: fix db type confusion in host fdb/mdb add/del
We have the following code paths:
Host FDB (unicast RX filtering):
dsa_port_standalone_host_fdb_add() dsa_port_bridge_host_fdb_add()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_fdb_add()
dsa_port_standalone_host_fdb_del() dsa_port_bridge_host_fdb_del()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_fdb_del()
Host MDB (multicast RX filtering):
dsa_port_standalone_host_mdb_add() dsa_port_bridge_host_mdb_add()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_mdb_add()
dsa_port_standalone_host_mdb_del() dsa_port_bridge_host_mdb_del()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_mdb_del()
The logic added by commit 5e8a1e03aa4d ("net: dsa: install secondary
unicast and multicast addresses as host FDB/MDB") zeroes out
db.bridge.num if the switch doesn't support ds->fdb_isolation
(the majority doesn't). This is done for a reason explained in commit
c26933639b54 ("net: dsa: request drivers to perform FDB isolation").
Taking a single code path as example - dsa_port_host_fdb_add() - the
others are similar - the problem is that this function handles:
- DSA_DB_PORT databases, when called from
dsa_port_standalone_host_fdb_add()
- DSA_DB_BRIDGE databases, when called from
dsa_port_bridge_host_fdb_add()
So, if dsa_port_host_fdb_add() were to make any change on the
"bridge.num" attribute of the database, this would only be correct for a
DSA_DB_BRIDGE, and a type confusion for a DSA_DB_PORT bridge.
However, this bug is without consequences, for 2 reasons:
- dsa_port_standalone_host_fdb_add() is only called from code which is
(in)directly guarded by dsa_switch_supports_uc_filtering(ds), and that
function only returns true if ds->fdb_isolation is set. So, the code
only executed for DSA_DB_BRIDGE databases.
- Even if the code was not dead for DSA_DB_PORT, we have the following
memory layout:
struct dsa_bridge {
struct net_device *dev;
unsigned int num;
bool tx_fwd_offload;
refcount_t refcount;
};
struct dsa_db {
enum dsa_db_type type;
union {
const struct dsa_port *dp; // DSA_DB_PORT
struct dsa_lag lag;
struct dsa_bridge bridge; // DSA_DB_BRIDGE
};
};
So, the zeroization of dsa_db :: bridge :: num on a dsa_db structure of
type DSA_DB_PORT would access memory which is unused, because we only
use dsa_db :: dp for DSA_DB_PORT, and this is mapped at the same address
with dsa_db :: dev for DSA_DB_BRIDGE, thanks to the union definition.
It is correct to fix up dsa_db :: bridge :: num only from code paths
that come from the bridge / switchdev, so move these there.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230329133819.697642-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-03-29 21:38:19 +08:00
|
|
|
if (!dp->ds->fdb_isolation)
|
|
|
|
|
db.bridge.num = 0;
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
err = dev_mc_add(conduit, mdb->addr);
|
2021-06-29 22:06:53 +08:00
|
|
|
if (err)
|
|
|
|
|
return err;
|
net: dsa: introduce a separate cross-chip notifier type for host MDBs
Commit abd49535c380 ("net: dsa: execute dsa_switch_mdb_add only for
routing port in cross-chip topologies") does a surprisingly good job
even for the SWITCHDEV_OBJ_ID_HOST_MDB use case, where DSA simply
translates a switchdev object received on dp into a cross-chip notifier
for dp->cpu_dp.
To visualize how that works, imagine the daisy chain topology below and
consider a SWITCHDEV_OBJ_ID_HOST_MDB object emitted on sw2p0. How does
the cross-chip notifier know to match on all the right ports (sw0p4, the
dedicated CPU port, sw1p4, an upstream DSA link, and sw2p4, another
upstream DSA link)?
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ user ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
The answer is simple: the dedicated CPU port of sw2p0 is sw0p4, and
dsa_routing_port returns the upstream port for all switches.
That is fine, but there are other topologies where this does not work as
well. There are trees with "H" topologies in the wild, where there are 2
or more switches with DSA links between them, but every switch has its
dedicated CPU port. For these topologies, it seems stupid for the neighbor
switches to install an MDB entry on the routing port, since these
multicast addresses are fundamentally different than the usual ones we
support (and that is the justification for this patch, to introduce the
concept of a termination plane multicast MAC address, as opposed to a
forwarding plane multicast MAC address).
For example, when a SWITCHDEV_OBJ_ID_HOST_MDB would get added to sw0p0,
without this patch, it would get treated as a regular port MDB on sw0p2
and it would match on the ports below (including the sw1p3 routing port).
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ x ] [ ] [ ] [ ]
With the patch, the host MDB notifier on sw0p0 matches only on the local
switch, which is what we want for a termination plane address.
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ ] [ ] [ ] [ ]
Name this new matching function "dsa_switch_host_address_match" since we
will be reusing it soon for host FDB entries as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-06-29 22:06:49 +08:00
|
|
|
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
return dsa_port_host_mdb_add(dp, mdb, db);
|
net: dsa: introduce a separate cross-chip notifier type for host MDBs
Commit abd49535c380 ("net: dsa: execute dsa_switch_mdb_add only for
routing port in cross-chip topologies") does a surprisingly good job
even for the SWITCHDEV_OBJ_ID_HOST_MDB use case, where DSA simply
translates a switchdev object received on dp into a cross-chip notifier
for dp->cpu_dp.
To visualize how that works, imagine the daisy chain topology below and
consider a SWITCHDEV_OBJ_ID_HOST_MDB object emitted on sw2p0. How does
the cross-chip notifier know to match on all the right ports (sw0p4, the
dedicated CPU port, sw1p4, an upstream DSA link, and sw2p4, another
upstream DSA link)?
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ user ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
The answer is simple: the dedicated CPU port of sw2p0 is sw0p4, and
dsa_routing_port returns the upstream port for all switches.
That is fine, but there are other topologies where this does not work as
well. There are trees with "H" topologies in the wild, where there are 2
or more switches with DSA links between them, but every switch has its
dedicated CPU port. For these topologies, it seems stupid for the neighbor
switches to install an MDB entry on the routing port, since these
multicast addresses are fundamentally different than the usual ones we
support (and that is the justification for this patch, to introduce the
concept of a termination plane multicast MAC address, as opposed to a
forwarding plane multicast MAC address).
For example, when a SWITCHDEV_OBJ_ID_HOST_MDB would get added to sw0p0,
without this patch, it would get treated as a regular port MDB on sw0p2
and it would match on the ports below (including the sw1p3 routing port).
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ x ] [ ] [ ] [ ]
With the patch, the host MDB notifier on sw0p0 matches only on the local
switch, which is what we want for a termination plane address.
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ ] [ ] [ ] [ ]
Name this new matching function "dsa_switch_host_address_match" since we
will be reusing it soon for host FDB entries as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-06-29 22:06:49 +08:00
|
|
|
}
|
|
|
|
|
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
static int dsa_port_host_mdb_del(const struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_port_mdb *mdb,
|
|
|
|
|
struct dsa_db db)
|
net: dsa: introduce a separate cross-chip notifier type for host MDBs
Commit abd49535c380 ("net: dsa: execute dsa_switch_mdb_add only for
routing port in cross-chip topologies") does a surprisingly good job
even for the SWITCHDEV_OBJ_ID_HOST_MDB use case, where DSA simply
translates a switchdev object received on dp into a cross-chip notifier
for dp->cpu_dp.
To visualize how that works, imagine the daisy chain topology below and
consider a SWITCHDEV_OBJ_ID_HOST_MDB object emitted on sw2p0. How does
the cross-chip notifier know to match on all the right ports (sw0p4, the
dedicated CPU port, sw1p4, an upstream DSA link, and sw2p4, another
upstream DSA link)?
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ user ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
The answer is simple: the dedicated CPU port of sw2p0 is sw0p4, and
dsa_routing_port returns the upstream port for all switches.
That is fine, but there are other topologies where this does not work as
well. There are trees with "H" topologies in the wild, where there are 2
or more switches with DSA links between them, but every switch has its
dedicated CPU port. For these topologies, it seems stupid for the neighbor
switches to install an MDB entry on the routing port, since these
multicast addresses are fundamentally different than the usual ones we
support (and that is the justification for this patch, to introduce the
concept of a termination plane multicast MAC address, as opposed to a
forwarding plane multicast MAC address).
For example, when a SWITCHDEV_OBJ_ID_HOST_MDB would get added to sw0p0,
without this patch, it would get treated as a regular port MDB on sw0p2
and it would match on the ports below (including the sw1p3 routing port).
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ x ] [ ] [ ] [ ]
With the patch, the host MDB notifier on sw0p0 matches only on the local
switch, which is what we want for a termination plane address.
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ ] [ ] [ ] [ ]
Name this new matching function "dsa_switch_host_address_match" since we
will be reusing it soon for host FDB entries as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-06-29 22:06:49 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_notifier_mdb_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
net: dsa: introduce a separate cross-chip notifier type for host MDBs
Commit abd49535c380 ("net: dsa: execute dsa_switch_mdb_add only for
routing port in cross-chip topologies") does a surprisingly good job
even for the SWITCHDEV_OBJ_ID_HOST_MDB use case, where DSA simply
translates a switchdev object received on dp into a cross-chip notifier
for dp->cpu_dp.
To visualize how that works, imagine the daisy chain topology below and
consider a SWITCHDEV_OBJ_ID_HOST_MDB object emitted on sw2p0. How does
the cross-chip notifier know to match on all the right ports (sw0p4, the
dedicated CPU port, sw1p4, an upstream DSA link, and sw2p4, another
upstream DSA link)?
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ user ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
The answer is simple: the dedicated CPU port of sw2p0 is sw0p4, and
dsa_routing_port returns the upstream port for all switches.
That is fine, but there are other topologies where this does not work as
well. There are trees with "H" topologies in the wild, where there are 2
or more switches with DSA links between them, but every switch has its
dedicated CPU port. For these topologies, it seems stupid for the neighbor
switches to install an MDB entry on the routing port, since these
multicast addresses are fundamentally different than the usual ones we
support (and that is the justification for this patch, to introduce the
concept of a termination plane multicast MAC address, as opposed to a
forwarding plane multicast MAC address).
For example, when a SWITCHDEV_OBJ_ID_HOST_MDB would get added to sw0p0,
without this patch, it would get treated as a regular port MDB on sw0p2
and it would match on the ports below (including the sw1p3 routing port).
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ x ] [ ] [ ] [ ]
With the patch, the host MDB notifier on sw0p0 matches only on the local
switch, which is what we want for a termination plane address.
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ ] [ ] [ ] [ ]
Name this new matching function "dsa_switch_host_address_match" since we
will be reusing it soon for host FDB entries as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-06-29 22:06:49 +08:00
|
|
|
.mdb = mdb,
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
.db = db,
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_HOST_MDB_DEL, &info);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_standalone_host_mdb_del(const struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_port_mdb *mdb)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_db db = {
|
|
|
|
|
.type = DSA_DB_PORT,
|
|
|
|
|
.dp = dp,
|
net: dsa: introduce a separate cross-chip notifier type for host MDBs
Commit abd49535c380 ("net: dsa: execute dsa_switch_mdb_add only for
routing port in cross-chip topologies") does a surprisingly good job
even for the SWITCHDEV_OBJ_ID_HOST_MDB use case, where DSA simply
translates a switchdev object received on dp into a cross-chip notifier
for dp->cpu_dp.
To visualize how that works, imagine the daisy chain topology below and
consider a SWITCHDEV_OBJ_ID_HOST_MDB object emitted on sw2p0. How does
the cross-chip notifier know to match on all the right ports (sw0p4, the
dedicated CPU port, sw1p4, an upstream DSA link, and sw2p4, another
upstream DSA link)?
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ user ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
The answer is simple: the dedicated CPU port of sw2p0 is sw0p4, and
dsa_routing_port returns the upstream port for all switches.
That is fine, but there are other topologies where this does not work as
well. There are trees with "H" topologies in the wild, where there are 2
or more switches with DSA links between them, but every switch has its
dedicated CPU port. For these topologies, it seems stupid for the neighbor
switches to install an MDB entry on the routing port, since these
multicast addresses are fundamentally different than the usual ones we
support (and that is the justification for this patch, to introduce the
concept of a termination plane multicast MAC address, as opposed to a
forwarding plane multicast MAC address).
For example, when a SWITCHDEV_OBJ_ID_HOST_MDB would get added to sw0p0,
without this patch, it would get treated as a regular port MDB on sw0p2
and it would match on the ports below (including the sw1p3 routing port).
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ x ] [ ] [ ] [ ]
With the patch, the host MDB notifier on sw0p0 matches only on the local
switch, which is what we want for a termination plane address.
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ ] [ ] [ ] [ ]
Name this new matching function "dsa_switch_host_address_match" since we
will be reusing it soon for host FDB entries as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-06-29 22:06:49 +08:00
|
|
|
};
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
|
|
|
|
|
return dsa_port_host_mdb_del(dp, mdb, db);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_bridge_host_mdb_del(const struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_port_mdb *mdb)
|
|
|
|
|
{
|
2023-10-24 02:17:28 +08:00
|
|
|
struct net_device *conduit = dsa_port_to_conduit(dp);
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
struct dsa_db db = {
|
|
|
|
|
.type = DSA_DB_BRIDGE,
|
|
|
|
|
.bridge = *dp->bridge,
|
|
|
|
|
};
|
2021-06-29 22:06:53 +08:00
|
|
|
int err;
|
|
|
|
|
|
net: dsa: fix db type confusion in host fdb/mdb add/del
We have the following code paths:
Host FDB (unicast RX filtering):
dsa_port_standalone_host_fdb_add() dsa_port_bridge_host_fdb_add()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_fdb_add()
dsa_port_standalone_host_fdb_del() dsa_port_bridge_host_fdb_del()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_fdb_del()
Host MDB (multicast RX filtering):
dsa_port_standalone_host_mdb_add() dsa_port_bridge_host_mdb_add()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_mdb_add()
dsa_port_standalone_host_mdb_del() dsa_port_bridge_host_mdb_del()
| |
+--------------+ +------------+
| |
v v
dsa_port_host_mdb_del()
The logic added by commit 5e8a1e03aa4d ("net: dsa: install secondary
unicast and multicast addresses as host FDB/MDB") zeroes out
db.bridge.num if the switch doesn't support ds->fdb_isolation
(the majority doesn't). This is done for a reason explained in commit
c26933639b54 ("net: dsa: request drivers to perform FDB isolation").
Taking a single code path as example - dsa_port_host_fdb_add() - the
others are similar - the problem is that this function handles:
- DSA_DB_PORT databases, when called from
dsa_port_standalone_host_fdb_add()
- DSA_DB_BRIDGE databases, when called from
dsa_port_bridge_host_fdb_add()
So, if dsa_port_host_fdb_add() were to make any change on the
"bridge.num" attribute of the database, this would only be correct for a
DSA_DB_BRIDGE, and a type confusion for a DSA_DB_PORT bridge.
However, this bug is without consequences, for 2 reasons:
- dsa_port_standalone_host_fdb_add() is only called from code which is
(in)directly guarded by dsa_switch_supports_uc_filtering(ds), and that
function only returns true if ds->fdb_isolation is set. So, the code
only executed for DSA_DB_BRIDGE databases.
- Even if the code was not dead for DSA_DB_PORT, we have the following
memory layout:
struct dsa_bridge {
struct net_device *dev;
unsigned int num;
bool tx_fwd_offload;
refcount_t refcount;
};
struct dsa_db {
enum dsa_db_type type;
union {
const struct dsa_port *dp; // DSA_DB_PORT
struct dsa_lag lag;
struct dsa_bridge bridge; // DSA_DB_BRIDGE
};
};
So, the zeroization of dsa_db :: bridge :: num on a dsa_db structure of
type DSA_DB_PORT would access memory which is unused, because we only
use dsa_db :: dp for DSA_DB_PORT, and this is mapped at the same address
with dsa_db :: dev for DSA_DB_BRIDGE, thanks to the union definition.
It is correct to fix up dsa_db :: bridge :: num only from code paths
that come from the bridge / switchdev, so move these there.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230329133819.697642-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-03-29 21:38:19 +08:00
|
|
|
if (!dp->ds->fdb_isolation)
|
|
|
|
|
db.bridge.num = 0;
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
err = dev_mc_del(conduit, mdb->addr);
|
2021-06-29 22:06:53 +08:00
|
|
|
if (err)
|
|
|
|
|
return err;
|
net: dsa: introduce a separate cross-chip notifier type for host MDBs
Commit abd49535c380 ("net: dsa: execute dsa_switch_mdb_add only for
routing port in cross-chip topologies") does a surprisingly good job
even for the SWITCHDEV_OBJ_ID_HOST_MDB use case, where DSA simply
translates a switchdev object received on dp into a cross-chip notifier
for dp->cpu_dp.
To visualize how that works, imagine the daisy chain topology below and
consider a SWITCHDEV_OBJ_ID_HOST_MDB object emitted on sw2p0. How does
the cross-chip notifier know to match on all the right ports (sw0p4, the
dedicated CPU port, sw1p4, an upstream DSA link, and sw2p4, another
upstream DSA link)?
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ user ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
The answer is simple: the dedicated CPU port of sw2p0 is sw0p4, and
dsa_routing_port returns the upstream port for all switches.
That is fine, but there are other topologies where this does not work as
well. There are trees with "H" topologies in the wild, where there are 2
or more switches with DSA links between them, but every switch has its
dedicated CPU port. For these topologies, it seems stupid for the neighbor
switches to install an MDB entry on the routing port, since these
multicast addresses are fundamentally different than the usual ones we
support (and that is the justification for this patch, to introduce the
concept of a termination plane multicast MAC address, as opposed to a
forwarding plane multicast MAC address).
For example, when a SWITCHDEV_OBJ_ID_HOST_MDB would get added to sw0p0,
without this patch, it would get treated as a regular port MDB on sw0p2
and it would match on the ports below (including the sw1p3 routing port).
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ x ] [ ] [ ] [ ]
With the patch, the host MDB notifier on sw0p0 matches only on the local
switch, which is what we want for a termination plane address.
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ ] [ ] [ ] [ ]
Name this new matching function "dsa_switch_host_address_match" since we
will be reusing it soon for host FDB entries as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-06-29 22:06:49 +08:00
|
|
|
|
net: dsa: install secondary unicast and multicast addresses as host FDB/MDB
In preparation of disabling flooding towards the CPU in standalone ports
mode, identify the addresses requested by upper interfaces and use the
new API for DSA FDB isolation to request the hardware driver to offload
these as FDB or MDB objects. The objects belong to the user port's
database, and are installed pointing towards the CPU port.
Because dev_uc_add()/dev_mc_add() is VLAN-unaware, we offload to the
port standalone database addresses with VID 0 (also VLAN-unaware).
So this excludes switches with global VLAN filtering from supporting
unicast filtering, because there, it is possible for a port of a switch
to join a VLAN-aware bridge, and this changes the VLAN awareness of
standalone ports, requiring VLAN-aware standalone host FDB entries.
For the same reason, hellcreek, which requires VLAN awareness in
standalone mode, is also exempted from unicast filtering.
We create "standalone" variants of dsa_port_host_fdb_add() and
dsa_port_host_mdb_add() (and the _del coresponding functions).
We also create a separate work item type for handling deferred
standalone host FDB/MDB entries compared to the switchdev one.
This is done for the purpose of clarity - the procedure for offloading a
bridge FDB entry is different than offloading a standalone one, and
the switchdev event work handles only FDBs anyway, not MDBs.
Deferral is needed for standalone entries because ndo_set_rx_mode runs
in atomic context. We could probably optimize things a little by first
queuing up all entries that need to be offloaded, and scheduling the
work item just once, but the data structures that we can pass through
__dev_uc_sync() and __dev_mc_sync() are limiting (there is nothing like
a void *priv), so we'd have to keep the list of queued events somewhere
in struct dsa_switch, and possibly a lock for it. Too complicated for
now.
Adding the address to the master is handled by dev_uc_sync(), adding it
to the hardware is handled by __dev_uc_sync(). So this is the reason why
dsa_port_standalone_host_fdb_add() does not call dev_uc_add(). Not that
it had the rtnl_mutex anyway - ndo_set_rx_mode has it, but is atomic.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-03-03 03:14:10 +08:00
|
|
|
return dsa_port_host_mdb_del(dp, mdb, db);
|
net: dsa: introduce a separate cross-chip notifier type for host MDBs
Commit abd49535c380 ("net: dsa: execute dsa_switch_mdb_add only for
routing port in cross-chip topologies") does a surprisingly good job
even for the SWITCHDEV_OBJ_ID_HOST_MDB use case, where DSA simply
translates a switchdev object received on dp into a cross-chip notifier
for dp->cpu_dp.
To visualize how that works, imagine the daisy chain topology below and
consider a SWITCHDEV_OBJ_ID_HOST_MDB object emitted on sw2p0. How does
the cross-chip notifier know to match on all the right ports (sw0p4, the
dedicated CPU port, sw1p4, an upstream DSA link, and sw2p4, another
upstream DSA link)?
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ user ] [ dsa ]
[ ] [ ] [ ] [ ] [ x ]
The answer is simple: the dedicated CPU port of sw2p0 is sw0p4, and
dsa_routing_port returns the upstream port for all switches.
That is fine, but there are other topologies where this does not work as
well. There are trees with "H" topologies in the wild, where there are 2
or more switches with DSA links between them, but every switch has its
dedicated CPU port. For these topologies, it seems stupid for the neighbor
switches to install an MDB entry on the routing port, since these
multicast addresses are fundamentally different than the usual ones we
support (and that is the justification for this patch, to introduce the
concept of a termination plane multicast MAC address, as opposed to a
forwarding plane multicast MAC address).
For example, when a SWITCHDEV_OBJ_ID_HOST_MDB would get added to sw0p0,
without this patch, it would get treated as a regular port MDB on sw0p2
and it would match on the ports below (including the sw1p3 routing port).
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ x ] [ ] [ ] [ ]
With the patch, the host MDB notifier on sw0p0 matches only on the local
switch, which is what we want for a termination plane address.
| |
sw0p0 sw0p1 sw0p2 sw0p3 sw1p3 sw1p2 sw1p1 sw1p0
[ user ] [ user ] [ cpu ] [ dsa ] [ dsa ] [ cpu ] [ user ] [ user ]
[ ] [ ] [ x ] [ ] ---- [ ] [ ] [ ] [ ]
Name this new matching function "dsa_switch_host_address_match" since we
will be reusing it soon for host FDB entries as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-06-29 22:06:49 +08:00
|
|
|
}
|
|
|
|
|
|
2017-05-20 05:00:50 +08:00
|
|
|
int dsa_port_vlan_add(struct dsa_port *dp,
|
2021-02-14 04:43:18 +08:00
|
|
|
const struct switchdev_obj_port_vlan *vlan,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
2017-05-20 05:00:50 +08:00
|
|
|
{
|
2017-05-20 05:00:55 +08:00
|
|
|
struct dsa_notifier_vlan_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2017-05-20 05:00:55 +08:00
|
|
|
.vlan = vlan,
|
2021-02-14 04:43:18 +08:00
|
|
|
.extack = extack,
|
2017-05-20 05:00:55 +08:00
|
|
|
};
|
2017-05-20 05:00:50 +08:00
|
|
|
|
2019-08-26 01:25:18 +08:00
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_VLAN_ADD, &info);
|
2017-05-20 05:00:50 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_vlan_del(struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_port_vlan *vlan)
|
|
|
|
|
{
|
2017-05-20 05:00:55 +08:00
|
|
|
struct dsa_notifier_vlan_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
2017-05-20 05:00:55 +08:00
|
|
|
.vlan = vlan,
|
|
|
|
|
};
|
2017-05-20 05:00:50 +08:00
|
|
|
|
2019-08-26 01:25:18 +08:00
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_VLAN_DEL, &info);
|
2017-05-20 05:00:50 +08:00
|
|
|
}
|
2017-10-26 22:50:07 +08:00
|
|
|
|
net: dsa: add explicit support for host bridge VLANs
Currently, DSA programs VLANs on shared (DSA and CPU) ports each time it
does so on user ports. This is good for basic functionality but has
several limitations:
- the VLAN group which must reach the CPU may be radically different
from the VLAN group that must be autonomously forwarded by the switch.
In other words, the admin may want to isolate noisy stations and avoid
traffic from them going to the control processor of the switch, where
it would just waste useless cycles. The bridge already supports
independent control of VLAN groups on bridge ports and on the bridge
itself, and when VLAN-aware, it will drop packets in software anyway
if their VID isn't added as a 'self' entry towards the bridge device.
- Replaying host FDB entries may depend, for some drivers like mv88e6xxx,
on replaying the host VLANs as well. The 2 VLAN groups are
approximately the same in most regular cases, but there are corner
cases when timing matters, and DSA's approximation of replicating
VLANs on shared ports simply does not work.
- If a user makes the bridge (implicitly the CPU port) join a VLAN by
accident, there is no way for the CPU port to isolate itself from that
noisy VLAN except by rebooting the system. This is because for each
VLAN added on a user port, DSA will add it on shared ports too, but
for each VLAN deletion on a user port, it will remain installed on
shared ports, since DSA has no good indication of whether the VLAN is
still in use or not.
Now that the bridge driver emits well-balanced SWITCHDEV_OBJ_ID_PORT_VLAN
addition and removal events, DSA has a simple and straightforward task
of separating the bridge port VLANs (these have an orig_dev which is a
DSA slave interface, or a LAG interface) from the host VLANs (these have
an orig_dev which is a bridge interface), and to keep a simple reference
count of each VID on each shared port.
Forwarding VLANs must be installed on the bridge ports and on all DSA
ports interconnecting them. We don't have a good view of the exact
topology, so we simply install forwarding VLANs on all DSA ports, which
is what has been done until now.
Host VLANs must be installed primarily on the dedicated CPU port of each
bridge port. More subtly, they must also be installed on upstream-facing
and downstream-facing DSA ports that are connecting the bridge ports and
the CPU. This ensures that the mv88e6xxx's problem (VID of host FDB
entry may be absent from VTU) is still addressed even if that switch is
in a cross-chip setup, and it has no local CPU port.
Therefore:
- user ports contain only bridge port (forwarding) VLANs, and no
refcounting is necessary
- DSA ports contain both forwarding and host VLANs. Refcounting is
necessary among these 2 types.
- CPU ports contain only host VLANs. Refcounting is also necessary.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-16 01:02:17 +08:00
|
|
|
int dsa_port_host_vlan_add(struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_port_vlan *vlan,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
|
|
|
|
{
|
2023-10-24 02:17:28 +08:00
|
|
|
struct net_device *conduit = dsa_port_to_conduit(dp);
|
net: dsa: add explicit support for host bridge VLANs
Currently, DSA programs VLANs on shared (DSA and CPU) ports each time it
does so on user ports. This is good for basic functionality but has
several limitations:
- the VLAN group which must reach the CPU may be radically different
from the VLAN group that must be autonomously forwarded by the switch.
In other words, the admin may want to isolate noisy stations and avoid
traffic from them going to the control processor of the switch, where
it would just waste useless cycles. The bridge already supports
independent control of VLAN groups on bridge ports and on the bridge
itself, and when VLAN-aware, it will drop packets in software anyway
if their VID isn't added as a 'self' entry towards the bridge device.
- Replaying host FDB entries may depend, for some drivers like mv88e6xxx,
on replaying the host VLANs as well. The 2 VLAN groups are
approximately the same in most regular cases, but there are corner
cases when timing matters, and DSA's approximation of replicating
VLANs on shared ports simply does not work.
- If a user makes the bridge (implicitly the CPU port) join a VLAN by
accident, there is no way for the CPU port to isolate itself from that
noisy VLAN except by rebooting the system. This is because for each
VLAN added on a user port, DSA will add it on shared ports too, but
for each VLAN deletion on a user port, it will remain installed on
shared ports, since DSA has no good indication of whether the VLAN is
still in use or not.
Now that the bridge driver emits well-balanced SWITCHDEV_OBJ_ID_PORT_VLAN
addition and removal events, DSA has a simple and straightforward task
of separating the bridge port VLANs (these have an orig_dev which is a
DSA slave interface, or a LAG interface) from the host VLANs (these have
an orig_dev which is a bridge interface), and to keep a simple reference
count of each VID on each shared port.
Forwarding VLANs must be installed on the bridge ports and on all DSA
ports interconnecting them. We don't have a good view of the exact
topology, so we simply install forwarding VLANs on all DSA ports, which
is what has been done until now.
Host VLANs must be installed primarily on the dedicated CPU port of each
bridge port. More subtly, they must also be installed on upstream-facing
and downstream-facing DSA ports that are connecting the bridge ports and
the CPU. This ensures that the mv88e6xxx's problem (VID of host FDB
entry may be absent from VTU) is still addressed even if that switch is
in a cross-chip setup, and it has no local CPU port.
Therefore:
- user ports contain only bridge port (forwarding) VLANs, and no
refcounting is necessary
- DSA ports contain both forwarding and host VLANs. Refcounting is
necessary among these 2 types.
- CPU ports contain only host VLANs. Refcounting is also necessary.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-16 01:02:17 +08:00
|
|
|
struct dsa_notifier_vlan_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
net: dsa: add explicit support for host bridge VLANs
Currently, DSA programs VLANs on shared (DSA and CPU) ports each time it
does so on user ports. This is good for basic functionality but has
several limitations:
- the VLAN group which must reach the CPU may be radically different
from the VLAN group that must be autonomously forwarded by the switch.
In other words, the admin may want to isolate noisy stations and avoid
traffic from them going to the control processor of the switch, where
it would just waste useless cycles. The bridge already supports
independent control of VLAN groups on bridge ports and on the bridge
itself, and when VLAN-aware, it will drop packets in software anyway
if their VID isn't added as a 'self' entry towards the bridge device.
- Replaying host FDB entries may depend, for some drivers like mv88e6xxx,
on replaying the host VLANs as well. The 2 VLAN groups are
approximately the same in most regular cases, but there are corner
cases when timing matters, and DSA's approximation of replicating
VLANs on shared ports simply does not work.
- If a user makes the bridge (implicitly the CPU port) join a VLAN by
accident, there is no way for the CPU port to isolate itself from that
noisy VLAN except by rebooting the system. This is because for each
VLAN added on a user port, DSA will add it on shared ports too, but
for each VLAN deletion on a user port, it will remain installed on
shared ports, since DSA has no good indication of whether the VLAN is
still in use or not.
Now that the bridge driver emits well-balanced SWITCHDEV_OBJ_ID_PORT_VLAN
addition and removal events, DSA has a simple and straightforward task
of separating the bridge port VLANs (these have an orig_dev which is a
DSA slave interface, or a LAG interface) from the host VLANs (these have
an orig_dev which is a bridge interface), and to keep a simple reference
count of each VID on each shared port.
Forwarding VLANs must be installed on the bridge ports and on all DSA
ports interconnecting them. We don't have a good view of the exact
topology, so we simply install forwarding VLANs on all DSA ports, which
is what has been done until now.
Host VLANs must be installed primarily on the dedicated CPU port of each
bridge port. More subtly, they must also be installed on upstream-facing
and downstream-facing DSA ports that are connecting the bridge ports and
the CPU. This ensures that the mv88e6xxx's problem (VID of host FDB
entry may be absent from VTU) is still addressed even if that switch is
in a cross-chip setup, and it has no local CPU port.
Therefore:
- user ports contain only bridge port (forwarding) VLANs, and no
refcounting is necessary
- DSA ports contain both forwarding and host VLANs. Refcounting is
necessary among these 2 types.
- CPU ports contain only host VLANs. Refcounting is also necessary.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-16 01:02:17 +08:00
|
|
|
.vlan = vlan,
|
|
|
|
|
.extack = extack,
|
|
|
|
|
};
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
err = dsa_port_notify(dp, DSA_NOTIFIER_HOST_VLAN_ADD, &info);
|
|
|
|
|
if (err && err != -EOPNOTSUPP)
|
|
|
|
|
return err;
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
vlan_vid_add(conduit, htons(ETH_P_8021Q), vlan->vid);
|
net: dsa: add explicit support for host bridge VLANs
Currently, DSA programs VLANs on shared (DSA and CPU) ports each time it
does so on user ports. This is good for basic functionality but has
several limitations:
- the VLAN group which must reach the CPU may be radically different
from the VLAN group that must be autonomously forwarded by the switch.
In other words, the admin may want to isolate noisy stations and avoid
traffic from them going to the control processor of the switch, where
it would just waste useless cycles. The bridge already supports
independent control of VLAN groups on bridge ports and on the bridge
itself, and when VLAN-aware, it will drop packets in software anyway
if their VID isn't added as a 'self' entry towards the bridge device.
- Replaying host FDB entries may depend, for some drivers like mv88e6xxx,
on replaying the host VLANs as well. The 2 VLAN groups are
approximately the same in most regular cases, but there are corner
cases when timing matters, and DSA's approximation of replicating
VLANs on shared ports simply does not work.
- If a user makes the bridge (implicitly the CPU port) join a VLAN by
accident, there is no way for the CPU port to isolate itself from that
noisy VLAN except by rebooting the system. This is because for each
VLAN added on a user port, DSA will add it on shared ports too, but
for each VLAN deletion on a user port, it will remain installed on
shared ports, since DSA has no good indication of whether the VLAN is
still in use or not.
Now that the bridge driver emits well-balanced SWITCHDEV_OBJ_ID_PORT_VLAN
addition and removal events, DSA has a simple and straightforward task
of separating the bridge port VLANs (these have an orig_dev which is a
DSA slave interface, or a LAG interface) from the host VLANs (these have
an orig_dev which is a bridge interface), and to keep a simple reference
count of each VID on each shared port.
Forwarding VLANs must be installed on the bridge ports and on all DSA
ports interconnecting them. We don't have a good view of the exact
topology, so we simply install forwarding VLANs on all DSA ports, which
is what has been done until now.
Host VLANs must be installed primarily on the dedicated CPU port of each
bridge port. More subtly, they must also be installed on upstream-facing
and downstream-facing DSA ports that are connecting the bridge ports and
the CPU. This ensures that the mv88e6xxx's problem (VID of host FDB
entry may be absent from VTU) is still addressed even if that switch is
in a cross-chip setup, and it has no local CPU port.
Therefore:
- user ports contain only bridge port (forwarding) VLANs, and no
refcounting is necessary
- DSA ports contain both forwarding and host VLANs. Refcounting is
necessary among these 2 types.
- CPU ports contain only host VLANs. Refcounting is also necessary.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-16 01:02:17 +08:00
|
|
|
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_host_vlan_del(struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_port_vlan *vlan)
|
|
|
|
|
{
|
2023-10-24 02:17:28 +08:00
|
|
|
struct net_device *conduit = dsa_port_to_conduit(dp);
|
net: dsa: add explicit support for host bridge VLANs
Currently, DSA programs VLANs on shared (DSA and CPU) ports each time it
does so on user ports. This is good for basic functionality but has
several limitations:
- the VLAN group which must reach the CPU may be radically different
from the VLAN group that must be autonomously forwarded by the switch.
In other words, the admin may want to isolate noisy stations and avoid
traffic from them going to the control processor of the switch, where
it would just waste useless cycles. The bridge already supports
independent control of VLAN groups on bridge ports and on the bridge
itself, and when VLAN-aware, it will drop packets in software anyway
if their VID isn't added as a 'self' entry towards the bridge device.
- Replaying host FDB entries may depend, for some drivers like mv88e6xxx,
on replaying the host VLANs as well. The 2 VLAN groups are
approximately the same in most regular cases, but there are corner
cases when timing matters, and DSA's approximation of replicating
VLANs on shared ports simply does not work.
- If a user makes the bridge (implicitly the CPU port) join a VLAN by
accident, there is no way for the CPU port to isolate itself from that
noisy VLAN except by rebooting the system. This is because for each
VLAN added on a user port, DSA will add it on shared ports too, but
for each VLAN deletion on a user port, it will remain installed on
shared ports, since DSA has no good indication of whether the VLAN is
still in use or not.
Now that the bridge driver emits well-balanced SWITCHDEV_OBJ_ID_PORT_VLAN
addition and removal events, DSA has a simple and straightforward task
of separating the bridge port VLANs (these have an orig_dev which is a
DSA slave interface, or a LAG interface) from the host VLANs (these have
an orig_dev which is a bridge interface), and to keep a simple reference
count of each VID on each shared port.
Forwarding VLANs must be installed on the bridge ports and on all DSA
ports interconnecting them. We don't have a good view of the exact
topology, so we simply install forwarding VLANs on all DSA ports, which
is what has been done until now.
Host VLANs must be installed primarily on the dedicated CPU port of each
bridge port. More subtly, they must also be installed on upstream-facing
and downstream-facing DSA ports that are connecting the bridge ports and
the CPU. This ensures that the mv88e6xxx's problem (VID of host FDB
entry may be absent from VTU) is still addressed even if that switch is
in a cross-chip setup, and it has no local CPU port.
Therefore:
- user ports contain only bridge port (forwarding) VLANs, and no
refcounting is necessary
- DSA ports contain both forwarding and host VLANs. Refcounting is
necessary among these 2 types.
- CPU ports contain only host VLANs. Refcounting is also necessary.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-16 01:02:17 +08:00
|
|
|
struct dsa_notifier_vlan_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
net: dsa: add explicit support for host bridge VLANs
Currently, DSA programs VLANs on shared (DSA and CPU) ports each time it
does so on user ports. This is good for basic functionality but has
several limitations:
- the VLAN group which must reach the CPU may be radically different
from the VLAN group that must be autonomously forwarded by the switch.
In other words, the admin may want to isolate noisy stations and avoid
traffic from them going to the control processor of the switch, where
it would just waste useless cycles. The bridge already supports
independent control of VLAN groups on bridge ports and on the bridge
itself, and when VLAN-aware, it will drop packets in software anyway
if their VID isn't added as a 'self' entry towards the bridge device.
- Replaying host FDB entries may depend, for some drivers like mv88e6xxx,
on replaying the host VLANs as well. The 2 VLAN groups are
approximately the same in most regular cases, but there are corner
cases when timing matters, and DSA's approximation of replicating
VLANs on shared ports simply does not work.
- If a user makes the bridge (implicitly the CPU port) join a VLAN by
accident, there is no way for the CPU port to isolate itself from that
noisy VLAN except by rebooting the system. This is because for each
VLAN added on a user port, DSA will add it on shared ports too, but
for each VLAN deletion on a user port, it will remain installed on
shared ports, since DSA has no good indication of whether the VLAN is
still in use or not.
Now that the bridge driver emits well-balanced SWITCHDEV_OBJ_ID_PORT_VLAN
addition and removal events, DSA has a simple and straightforward task
of separating the bridge port VLANs (these have an orig_dev which is a
DSA slave interface, or a LAG interface) from the host VLANs (these have
an orig_dev which is a bridge interface), and to keep a simple reference
count of each VID on each shared port.
Forwarding VLANs must be installed on the bridge ports and on all DSA
ports interconnecting them. We don't have a good view of the exact
topology, so we simply install forwarding VLANs on all DSA ports, which
is what has been done until now.
Host VLANs must be installed primarily on the dedicated CPU port of each
bridge port. More subtly, they must also be installed on upstream-facing
and downstream-facing DSA ports that are connecting the bridge ports and
the CPU. This ensures that the mv88e6xxx's problem (VID of host FDB
entry may be absent from VTU) is still addressed even if that switch is
in a cross-chip setup, and it has no local CPU port.
Therefore:
- user ports contain only bridge port (forwarding) VLANs, and no
refcounting is necessary
- DSA ports contain both forwarding and host VLANs. Refcounting is
necessary among these 2 types.
- CPU ports contain only host VLANs. Refcounting is also necessary.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-16 01:02:17 +08:00
|
|
|
.vlan = vlan,
|
|
|
|
|
};
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
err = dsa_port_notify(dp, DSA_NOTIFIER_HOST_VLAN_DEL, &info);
|
|
|
|
|
if (err && err != -EOPNOTSUPP)
|
|
|
|
|
return err;
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
vlan_vid_del(conduit, htons(ETH_P_8021Q), vlan->vid);
|
net: dsa: add explicit support for host bridge VLANs
Currently, DSA programs VLANs on shared (DSA and CPU) ports each time it
does so on user ports. This is good for basic functionality but has
several limitations:
- the VLAN group which must reach the CPU may be radically different
from the VLAN group that must be autonomously forwarded by the switch.
In other words, the admin may want to isolate noisy stations and avoid
traffic from them going to the control processor of the switch, where
it would just waste useless cycles. The bridge already supports
independent control of VLAN groups on bridge ports and on the bridge
itself, and when VLAN-aware, it will drop packets in software anyway
if their VID isn't added as a 'self' entry towards the bridge device.
- Replaying host FDB entries may depend, for some drivers like mv88e6xxx,
on replaying the host VLANs as well. The 2 VLAN groups are
approximately the same in most regular cases, but there are corner
cases when timing matters, and DSA's approximation of replicating
VLANs on shared ports simply does not work.
- If a user makes the bridge (implicitly the CPU port) join a VLAN by
accident, there is no way for the CPU port to isolate itself from that
noisy VLAN except by rebooting the system. This is because for each
VLAN added on a user port, DSA will add it on shared ports too, but
for each VLAN deletion on a user port, it will remain installed on
shared ports, since DSA has no good indication of whether the VLAN is
still in use or not.
Now that the bridge driver emits well-balanced SWITCHDEV_OBJ_ID_PORT_VLAN
addition and removal events, DSA has a simple and straightforward task
of separating the bridge port VLANs (these have an orig_dev which is a
DSA slave interface, or a LAG interface) from the host VLANs (these have
an orig_dev which is a bridge interface), and to keep a simple reference
count of each VID on each shared port.
Forwarding VLANs must be installed on the bridge ports and on all DSA
ports interconnecting them. We don't have a good view of the exact
topology, so we simply install forwarding VLANs on all DSA ports, which
is what has been done until now.
Host VLANs must be installed primarily on the dedicated CPU port of each
bridge port. More subtly, they must also be installed on upstream-facing
and downstream-facing DSA ports that are connecting the bridge ports and
the CPU. This ensures that the mv88e6xxx's problem (VID of host FDB
entry may be absent from VTU) is still addressed even if that switch is
in a cross-chip setup, and it has no local CPU port.
Therefore:
- user ports contain only bridge port (forwarding) VLANs, and no
refcounting is necessary
- DSA ports contain both forwarding and host VLANs. Refcounting is
necessary among these 2 types.
- CPU ports contain only host VLANs. Refcounting is also necessary.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-16 01:02:17 +08:00
|
|
|
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
2021-02-17 05:42:04 +08:00
|
|
|
int dsa_port_mrp_add(const struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_mrp *mrp)
|
|
|
|
|
{
|
2022-01-05 21:18:12 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->port_mrp_add)
|
|
|
|
|
return -EOPNOTSUPP;
|
2021-02-17 05:42:04 +08:00
|
|
|
|
2022-01-05 21:18:12 +08:00
|
|
|
return ds->ops->port_mrp_add(ds, dp->index, mrp);
|
2021-02-17 05:42:04 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_mrp_del(const struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_mrp *mrp)
|
|
|
|
|
{
|
2022-01-05 21:18:12 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->port_mrp_del)
|
|
|
|
|
return -EOPNOTSUPP;
|
2021-02-17 05:42:04 +08:00
|
|
|
|
2022-01-05 21:18:12 +08:00
|
|
|
return ds->ops->port_mrp_del(ds, dp->index, mrp);
|
2021-02-17 05:42:04 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_mrp_add_ring_role(const struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_ring_role_mrp *mrp)
|
|
|
|
|
{
|
2022-01-05 21:18:12 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->port_mrp_add_ring_role)
|
|
|
|
|
return -EOPNOTSUPP;
|
2021-02-17 05:42:04 +08:00
|
|
|
|
2022-01-05 21:18:12 +08:00
|
|
|
return ds->ops->port_mrp_add_ring_role(ds, dp->index, mrp);
|
2021-02-17 05:42:04 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int dsa_port_mrp_del_ring_role(const struct dsa_port *dp,
|
|
|
|
|
const struct switchdev_obj_ring_role_mrp *mrp)
|
|
|
|
|
{
|
2022-01-05 21:18:12 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->port_mrp_del_ring_role)
|
|
|
|
|
return -EOPNOTSUPP;
|
2021-02-17 05:42:04 +08:00
|
|
|
|
2022-01-05 21:18:12 +08:00
|
|
|
return ds->ops->port_mrp_del_ring_role(ds, dp->index, mrp);
|
2021-02-17 05:42:04 +08:00
|
|
|
}
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
static int dsa_port_assign_conduit(struct dsa_port *dp,
|
|
|
|
|
struct net_device *conduit,
|
|
|
|
|
struct netlink_ext_ack *extack,
|
|
|
|
|
bool fail_on_err)
|
2022-09-11 09:06:59 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
int port = dp->index, err;
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
err = ds->ops->port_change_conduit(ds, port, conduit, extack);
|
2022-09-11 09:06:59 +08:00
|
|
|
if (err && !fail_on_err)
|
2023-10-24 02:17:28 +08:00
|
|
|
dev_err(ds->dev, "port %d failed to assign conduit %s: %pe\n",
|
|
|
|
|
port, conduit->name, ERR_PTR(err));
|
2022-09-11 09:06:59 +08:00
|
|
|
|
|
|
|
|
if (err && fail_on_err)
|
|
|
|
|
return err;
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
dp->cpu_dp = conduit->dsa_ptr;
|
|
|
|
|
dp->cpu_port_in_lag = netif_is_lag_master(conduit);
|
2022-09-11 09:06:59 +08:00
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Change the dp->cpu_dp affinity for a user port. Note that both cross-chip
|
|
|
|
|
* notifiers and drivers have implicit assumptions about user-to-CPU-port
|
|
|
|
|
* mappings, so we unfortunately cannot delay the deletion of the objects
|
|
|
|
|
* (switchdev, standalone addresses, standalone VLANs) on the old CPU port
|
|
|
|
|
* until the new CPU port has been set up. So we need to completely tear down
|
|
|
|
|
* the old CPU port before changing it, and restore it on errors during the
|
|
|
|
|
* bringup of the new one.
|
|
|
|
|
*/
|
2023-10-24 02:17:28 +08:00
|
|
|
int dsa_port_change_conduit(struct dsa_port *dp, struct net_device *conduit,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
2022-09-11 09:06:59 +08:00
|
|
|
{
|
|
|
|
|
struct net_device *bridge_dev = dsa_port_bridge_dev_get(dp);
|
2023-10-24 02:17:28 +08:00
|
|
|
struct net_device *old_conduit = dsa_port_to_conduit(dp);
|
|
|
|
|
struct net_device *dev = dp->user;
|
2022-09-11 09:06:59 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
bool vlan_filtering;
|
|
|
|
|
int err, tmp;
|
|
|
|
|
|
|
|
|
|
/* Bridges may hold host FDB, MDB and VLAN objects. These need to be
|
|
|
|
|
* migrated, so dynamically unoffload and later reoffload the bridge
|
|
|
|
|
* port.
|
|
|
|
|
*/
|
|
|
|
|
if (bridge_dev) {
|
|
|
|
|
dsa_port_pre_bridge_leave(dp, bridge_dev);
|
|
|
|
|
dsa_port_bridge_leave(dp, bridge_dev);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* The port might still be VLAN filtering even if it's no longer
|
|
|
|
|
* under a bridge, either due to ds->vlan_filtering_is_global or
|
|
|
|
|
* ds->needs_standalone_vlan_filtering. In turn this means VLANs
|
|
|
|
|
* on the CPU port.
|
|
|
|
|
*/
|
|
|
|
|
vlan_filtering = dsa_port_is_vlan_filtering(dp);
|
|
|
|
|
if (vlan_filtering) {
|
2023-10-24 02:17:28 +08:00
|
|
|
err = dsa_user_manage_vlan_filtering(dev, false);
|
2022-09-11 09:06:59 +08:00
|
|
|
if (err) {
|
|
|
|
|
NL_SET_ERR_MSG_MOD(extack,
|
|
|
|
|
"Failed to remove standalone VLANs");
|
|
|
|
|
goto rewind_old_bridge;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Standalone addresses, and addresses of upper interfaces like
|
|
|
|
|
* VLAN, LAG, HSR need to be migrated.
|
|
|
|
|
*/
|
2023-10-24 02:17:28 +08:00
|
|
|
dsa_user_unsync_ha(dev);
|
2022-09-11 09:06:59 +08:00
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
err = dsa_port_assign_conduit(dp, conduit, extack, true);
|
2022-09-11 09:06:59 +08:00
|
|
|
if (err)
|
|
|
|
|
goto rewind_old_addrs;
|
|
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
dsa_user_sync_ha(dev);
|
2022-09-11 09:06:59 +08:00
|
|
|
|
|
|
|
|
if (vlan_filtering) {
|
2023-10-24 02:17:28 +08:00
|
|
|
err = dsa_user_manage_vlan_filtering(dev, true);
|
2022-09-11 09:06:59 +08:00
|
|
|
if (err) {
|
|
|
|
|
NL_SET_ERR_MSG_MOD(extack,
|
|
|
|
|
"Failed to restore standalone VLANs");
|
|
|
|
|
goto rewind_new_addrs;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (bridge_dev) {
|
|
|
|
|
err = dsa_port_bridge_join(dp, bridge_dev, extack);
|
|
|
|
|
if (err && err == -EOPNOTSUPP) {
|
|
|
|
|
NL_SET_ERR_MSG_MOD(extack,
|
|
|
|
|
"Failed to reoffload bridge");
|
|
|
|
|
goto rewind_new_vlan;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
rewind_new_vlan:
|
|
|
|
|
if (vlan_filtering)
|
2023-10-24 02:17:28 +08:00
|
|
|
dsa_user_manage_vlan_filtering(dev, false);
|
2022-09-11 09:06:59 +08:00
|
|
|
|
|
|
|
|
rewind_new_addrs:
|
2023-10-24 02:17:28 +08:00
|
|
|
dsa_user_unsync_ha(dev);
|
2022-09-11 09:06:59 +08:00
|
|
|
|
2023-10-24 02:17:28 +08:00
|
|
|
dsa_port_assign_conduit(dp, old_conduit, NULL, false);
|
2022-09-11 09:06:59 +08:00
|
|
|
|
|
|
|
|
/* Restore the objects on the old CPU port */
|
|
|
|
|
rewind_old_addrs:
|
2023-10-24 02:17:28 +08:00
|
|
|
dsa_user_sync_ha(dev);
|
2022-09-11 09:06:59 +08:00
|
|
|
|
|
|
|
|
if (vlan_filtering) {
|
2023-10-24 02:17:28 +08:00
|
|
|
tmp = dsa_user_manage_vlan_filtering(dev, true);
|
2022-09-11 09:06:59 +08:00
|
|
|
if (tmp) {
|
|
|
|
|
dev_err(ds->dev,
|
|
|
|
|
"port %d failed to restore standalone VLANs: %pe\n",
|
|
|
|
|
dp->index, ERR_PTR(tmp));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
rewind_old_bridge:
|
|
|
|
|
if (bridge_dev) {
|
|
|
|
|
tmp = dsa_port_bridge_join(dp, bridge_dev, extack);
|
|
|
|
|
if (tmp) {
|
|
|
|
|
dev_err(ds->dev,
|
|
|
|
|
"port %d failed to rejoin bridge %s: %pe\n",
|
|
|
|
|
dp->index, bridge_dev->name, ERR_PTR(tmp));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
net: dsa: allow changing the tag protocol via the "tagging" device attribute
Currently DSA exposes the following sysfs:
$ cat /sys/class/net/eno2/dsa/tagging
ocelot
which is a read-only device attribute, introduced in the kernel as
commit 98cdb4807123 ("net: dsa: Expose tagging protocol to user-space"),
and used by libpcap since its commit 993db3800d7d ("Add support for DSA
link-layer types").
It would be nice if we could extend this device attribute by making it
writable:
$ echo ocelot-8021q > /sys/class/net/eno2/dsa/tagging
This is useful with DSA switches that can make use of more than one
tagging protocol. It may be useful in dsa_loop in the future too, to
perform offline testing of various taggers, or for changing between dsa
and edsa on Marvell switches, if that is desirable.
In terms of implementation, drivers can support this feature by
implementing .change_tag_protocol, which should always leave the switch
in a consistent state: either with the new protocol if things went well,
or with the old one if something failed. Teardown of the old protocol,
if necessary, must be handled by the driver.
Some things remain as before:
- The .get_tag_protocol is currently only called at probe time, to load
the initial tagging protocol driver. Nonetheless, new drivers should
report the tagging protocol in current use now.
- The driver should manage by itself the initial setup of tagging
protocol, no later than the .setup() method, as well as destroying
resources used by the last tagger in use, no earlier than the
.teardown() method.
For multi-switch DSA trees, error handling is a bit more complicated,
since e.g. the 5th out of 7 switches may fail to change the tag
protocol. When that happens, a revert to the original tag protocol is
attempted, but that may fail too, leaving the tree in an inconsistent
state despite each individual switch implementing .change_tag_protocol
transactionally. Since the intersection between drivers that implement
.change_tag_protocol and drivers that support D in DSA is currently the
empty set, the possibility for this error to happen is ignored for now.
Testing:
$ insmod mscc_felix.ko
[ 79.549784] mscc_felix 0000:00:00.5: Adding to iommu group 14
[ 79.565712] mscc_felix 0000:00:00.5: Failed to register DSA switch: -517
$ insmod tag_ocelot.ko
$ rmmod mscc_felix.ko
$ insmod mscc_felix.ko
[ 97.261724] libphy: VSC9959 internal MDIO bus: probed
[ 97.267363] mscc_felix 0000:00:00.5: Found PCS at internal MDIO address 0
[ 97.274998] mscc_felix 0000:00:00.5: Found PCS at internal MDIO address 1
[ 97.282561] mscc_felix 0000:00:00.5: Found PCS at internal MDIO address 2
[ 97.289700] mscc_felix 0000:00:00.5: Found PCS at internal MDIO address 3
[ 97.599163] mscc_felix 0000:00:00.5 swp0 (uninitialized): PHY [0000:00:00.3:10] driver [Microsemi GE VSC8514 SyncE] (irq=POLL)
[ 97.862034] mscc_felix 0000:00:00.5 swp1 (uninitialized): PHY [0000:00:00.3:11] driver [Microsemi GE VSC8514 SyncE] (irq=POLL)
[ 97.950731] mscc_felix 0000:00:00.5 swp0: configuring for inband/qsgmii link mode
[ 97.964278] 8021q: adding VLAN 0 to HW filter on device swp0
[ 98.146161] mscc_felix 0000:00:00.5 swp2 (uninitialized): PHY [0000:00:00.3:12] driver [Microsemi GE VSC8514 SyncE] (irq=POLL)
[ 98.238649] mscc_felix 0000:00:00.5 swp1: configuring for inband/qsgmii link mode
[ 98.251845] 8021q: adding VLAN 0 to HW filter on device swp1
[ 98.433916] mscc_felix 0000:00:00.5 swp3 (uninitialized): PHY [0000:00:00.3:13] driver [Microsemi GE VSC8514 SyncE] (irq=POLL)
[ 98.485542] mscc_felix 0000:00:00.5: configuring for fixed/internal link mode
[ 98.503584] mscc_felix 0000:00:00.5: Link is Up - 2.5Gbps/Full - flow control rx/tx
[ 98.527948] device eno2 entered promiscuous mode
[ 98.544755] DSA: tree 0 setup
$ ping 10.0.0.1
PING 10.0.0.1 (10.0.0.1): 56 data bytes
64 bytes from 10.0.0.1: seq=0 ttl=64 time=2.337 ms
64 bytes from 10.0.0.1: seq=1 ttl=64 time=0.754 ms
^C
- 10.0.0.1 ping statistics -
2 packets transmitted, 2 packets received, 0% packet loss
round-trip min/avg/max = 0.754/1.545/2.337 ms
$ cat /sys/class/net/eno2/dsa/tagging
ocelot
$ cat ./test_ocelot_8021q.sh
#!/bin/bash
ip link set swp0 down
ip link set swp1 down
ip link set swp2 down
ip link set swp3 down
ip link set swp5 down
ip link set eno2 down
echo ocelot-8021q > /sys/class/net/eno2/dsa/tagging
ip link set eno2 up
ip link set swp0 up
ip link set swp1 up
ip link set swp2 up
ip link set swp3 up
ip link set swp5 up
$ ./test_ocelot_8021q.sh
./test_ocelot_8021q.sh: line 9: echo: write error: Protocol not available
$ rmmod tag_ocelot.ko
rmmod: can't unload module 'tag_ocelot': Resource temporarily unavailable
$ insmod tag_ocelot_8021q.ko
$ ./test_ocelot_8021q.sh
$ cat /sys/class/net/eno2/dsa/tagging
ocelot-8021q
$ rmmod tag_ocelot.ko
$ rmmod tag_ocelot_8021q.ko
rmmod: can't unload module 'tag_ocelot_8021q': Resource temporarily unavailable
$ ping 10.0.0.1
PING 10.0.0.1 (10.0.0.1): 56 data bytes
64 bytes from 10.0.0.1: seq=0 ttl=64 time=0.953 ms
64 bytes from 10.0.0.1: seq=1 ttl=64 time=0.787 ms
64 bytes from 10.0.0.1: seq=2 ttl=64 time=0.771 ms
$ rmmod mscc_felix.ko
[ 645.544426] mscc_felix 0000:00:00.5: Link is Down
[ 645.838608] DSA: tree 0 torn down
$ rmmod tag_ocelot_8021q.ko
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-29 09:00:06 +08:00
|
|
|
void dsa_port_set_tag_protocol(struct dsa_port *cpu_dp,
|
|
|
|
|
const struct dsa_device_ops *tag_ops)
|
|
|
|
|
{
|
|
|
|
|
cpu_dp->rcv = tag_ops->rcv;
|
|
|
|
|
cpu_dp->tag_ops = tag_ops;
|
|
|
|
|
}
|
|
|
|
|
|
2018-04-26 03:12:51 +08:00
|
|
|
static struct phy_device *dsa_port_get_phy_device(struct dsa_port *dp)
|
2018-01-23 23:03:46 +08:00
|
|
|
{
|
|
|
|
|
struct device_node *phy_dn;
|
|
|
|
|
struct phy_device *phydev;
|
|
|
|
|
|
2018-04-26 03:12:51 +08:00
|
|
|
phy_dn = of_parse_phandle(dp->dn, "phy-handle", 0);
|
2018-01-23 23:03:46 +08:00
|
|
|
if (!phy_dn)
|
2018-04-26 03:12:51 +08:00
|
|
|
return NULL;
|
2018-01-23 23:03:46 +08:00
|
|
|
|
|
|
|
|
phydev = of_phy_find_device(phy_dn);
|
|
|
|
|
if (!phydev) {
|
2018-04-26 03:12:51 +08:00
|
|
|
of_node_put(phy_dn);
|
|
|
|
|
return ERR_PTR(-EPROBE_DEFER);
|
2018-01-23 23:03:46 +08:00
|
|
|
}
|
|
|
|
|
|
2019-02-25 15:22:19 +08:00
|
|
|
of_node_put(phy_dn);
|
2018-04-26 03:12:51 +08:00
|
|
|
return phydev;
|
|
|
|
|
}
|
|
|
|
|
|
2022-02-18 02:30:35 +08:00
|
|
|
static struct phylink_pcs *
|
|
|
|
|
dsa_port_phylink_mac_select_pcs(struct phylink_config *config,
|
|
|
|
|
phy_interface_t interface)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_port *dp = container_of(config, struct dsa_port, pl_config);
|
2022-02-22 01:10:52 +08:00
|
|
|
struct phylink_pcs *pcs = ERR_PTR(-EOPNOTSUPP);
|
2022-02-18 02:30:35 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
|
|
|
|
if (ds->ops->phylink_mac_select_pcs)
|
|
|
|
|
pcs = ds->ops->phylink_mac_select_pcs(ds, dp->index, interface);
|
|
|
|
|
|
|
|
|
|
return pcs;
|
|
|
|
|
}
|
|
|
|
|
|
2023-05-25 18:38:44 +08:00
|
|
|
static int dsa_port_phylink_mac_prepare(struct phylink_config *config,
|
|
|
|
|
unsigned int mode,
|
|
|
|
|
phy_interface_t interface)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_port *dp = container_of(config, struct dsa_port, pl_config);
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
int err = 0;
|
|
|
|
|
|
|
|
|
|
if (ds->ops->phylink_mac_prepare)
|
|
|
|
|
err = ds->ops->phylink_mac_prepare(ds, dp->index, mode,
|
|
|
|
|
interface);
|
|
|
|
|
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
2019-12-17 02:32:47 +08:00
|
|
|
static void dsa_port_phylink_mac_config(struct phylink_config *config,
|
|
|
|
|
unsigned int mode,
|
|
|
|
|
const struct phylink_link_state *state)
|
2019-05-29 01:38:15 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_port *dp = container_of(config, struct dsa_port, pl_config);
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->phylink_mac_config)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
ds->ops->phylink_mac_config(ds, dp->index, mode, state);
|
|
|
|
|
}
|
|
|
|
|
|
2023-05-25 18:38:44 +08:00
|
|
|
static int dsa_port_phylink_mac_finish(struct phylink_config *config,
|
|
|
|
|
unsigned int mode,
|
|
|
|
|
phy_interface_t interface)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_port *dp = container_of(config, struct dsa_port, pl_config);
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
int err = 0;
|
|
|
|
|
|
|
|
|
|
if (ds->ops->phylink_mac_finish)
|
|
|
|
|
err = ds->ops->phylink_mac_finish(ds, dp->index, mode,
|
|
|
|
|
interface);
|
|
|
|
|
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
2019-12-17 02:32:47 +08:00
|
|
|
static void dsa_port_phylink_mac_link_down(struct phylink_config *config,
|
|
|
|
|
unsigned int mode,
|
|
|
|
|
phy_interface_t interface)
|
2019-05-29 01:38:15 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_port *dp = container_of(config, struct dsa_port, pl_config);
|
2019-05-29 01:38:16 +08:00
|
|
|
struct phy_device *phydev = NULL;
|
2019-05-29 01:38:15 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
2021-10-21 01:49:52 +08:00
|
|
|
if (dsa_port_is_user(dp))
|
2023-10-24 02:17:28 +08:00
|
|
|
phydev = dp->user->phydev;
|
2019-05-29 01:38:16 +08:00
|
|
|
|
2019-05-29 01:38:15 +08:00
|
|
|
if (!ds->ops->phylink_mac_link_down) {
|
2019-05-29 01:38:16 +08:00
|
|
|
if (ds->ops->adjust_link && phydev)
|
|
|
|
|
ds->ops->adjust_link(ds, dp->index, phydev);
|
2019-05-29 01:38:15 +08:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ds->ops->phylink_mac_link_down(ds, dp->index, mode, interface);
|
|
|
|
|
}
|
|
|
|
|
|
2019-12-17 02:32:47 +08:00
|
|
|
static void dsa_port_phylink_mac_link_up(struct phylink_config *config,
|
2020-02-26 18:23:41 +08:00
|
|
|
struct phy_device *phydev,
|
2019-12-17 02:32:47 +08:00
|
|
|
unsigned int mode,
|
|
|
|
|
phy_interface_t interface,
|
2020-02-26 18:23:41 +08:00
|
|
|
int speed, int duplex,
|
|
|
|
|
bool tx_pause, bool rx_pause)
|
2019-05-29 01:38:15 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_port *dp = container_of(config, struct dsa_port, pl_config);
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
|
|
|
|
if (!ds->ops->phylink_mac_link_up) {
|
2019-05-29 01:38:16 +08:00
|
|
|
if (ds->ops->adjust_link && phydev)
|
|
|
|
|
ds->ops->adjust_link(ds, dp->index, phydev);
|
2019-05-29 01:38:15 +08:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
2020-02-26 18:23:46 +08:00
|
|
|
ds->ops->phylink_mac_link_up(ds, dp->index, mode, interface, phydev,
|
|
|
|
|
speed, duplex, tx_pause, rx_pause);
|
2019-05-29 01:38:15 +08:00
|
|
|
}
|
|
|
|
|
|
2021-11-30 21:09:55 +08:00
|
|
|
static const struct phylink_mac_ops dsa_port_phylink_mac_ops = {
|
2022-02-18 02:30:35 +08:00
|
|
|
.mac_select_pcs = dsa_port_phylink_mac_select_pcs,
|
2023-05-25 18:38:44 +08:00
|
|
|
.mac_prepare = dsa_port_phylink_mac_prepare,
|
2019-05-29 01:38:15 +08:00
|
|
|
.mac_config = dsa_port_phylink_mac_config,
|
2023-05-25 18:38:44 +08:00
|
|
|
.mac_finish = dsa_port_phylink_mac_finish,
|
2019-05-29 01:38:15 +08:00
|
|
|
.mac_link_down = dsa_port_phylink_mac_link_down,
|
|
|
|
|
.mac_link_up = dsa_port_phylink_mac_link_up,
|
|
|
|
|
};
|
|
|
|
|
|
2021-11-30 21:09:55 +08:00
|
|
|
int dsa_port_phylink_create(struct dsa_port *dp)
|
|
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
phy_interface_t mode;
|
net: dsa: don't leave dangling pointers in dp->pl when failing
There is a desire to simplify the dsa_port registration path with
devlink, and this involves reworking a bit how user ports which fail to
connect to their PHY (because it's missing) get reinitialized as UNUSED
devlink ports.
The desire is for the change to look something like this; basically
dsa_port_setup() has failed, we just change dp->type and call
dsa_port_setup() again.
-/* Destroy the current devlink port, and create a new one which has the UNUSED
- * flavour.
- */
-static int dsa_port_reinit_as_unused(struct dsa_port *dp)
+static int dsa_port_setup_as_unused(struct dsa_port *dp)
{
- dsa_port_devlink_teardown(dp);
dp->type = DSA_PORT_TYPE_UNUSED;
- return dsa_port_devlink_setup(dp);
+ return dsa_port_setup(dp);
}
For an UNUSED port, dsa_port_setup() mostly only calls dsa_port_devlink_setup()
anyway, so we could get away with calling just that. But if we call the
full blown dsa_port_setup(dp) (which will be needed to properly set
dp->setup = true), the callee will have the tendency to go through this
code block too, and call dsa_port_disable(dp):
switch (dp->type) {
case DSA_PORT_TYPE_UNUSED:
dsa_port_disable(dp);
break;
That is not very good, because dsa_port_disable() has this hidden inside
of it:
if (dp->pl)
phylink_stop(dp->pl);
Fact is, we are not prepared to handle a call to dsa_port_disable() with
a struct dsa_port that came from a previous (and failed) call to
dsa_port_setup(). We do not clean up dp->pl, and this will make the
second call to dsa_port_setup() call phylink_stop() on a dangling dp->pl
pointer.
Solve this by creating an API for phylink destruction which is symmetric
to the phylink creation, and never leave dp->pl set to anything except
NULL or a valid phylink structure.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-09-29 15:29:00 +08:00
|
|
|
struct phylink *pl;
|
2021-11-30 21:09:55 +08:00
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
err = of_get_phy_mode(dp->dn, &mode);
|
|
|
|
|
if (err)
|
|
|
|
|
mode = PHY_INTERFACE_MODE_NA;
|
|
|
|
|
|
2023-07-26 22:45:16 +08:00
|
|
|
if (ds->ops->phylink_get_caps) {
|
2021-11-30 21:10:01 +08:00
|
|
|
ds->ops->phylink_get_caps(ds, dp->index, &dp->pl_config);
|
2023-07-26 22:45:16 +08:00
|
|
|
} else {
|
|
|
|
|
/* For legacy drivers */
|
2023-08-08 19:12:16 +08:00
|
|
|
if (mode != PHY_INTERFACE_MODE_NA) {
|
|
|
|
|
__set_bit(mode, dp->pl_config.supported_interfaces);
|
|
|
|
|
} else {
|
|
|
|
|
__set_bit(PHY_INTERFACE_MODE_INTERNAL,
|
|
|
|
|
dp->pl_config.supported_interfaces);
|
|
|
|
|
__set_bit(PHY_INTERFACE_MODE_GMII,
|
|
|
|
|
dp->pl_config.supported_interfaces);
|
|
|
|
|
}
|
2023-07-26 22:45:16 +08:00
|
|
|
}
|
2021-11-30 21:09:55 +08:00
|
|
|
|
net: dsa: don't leave dangling pointers in dp->pl when failing
There is a desire to simplify the dsa_port registration path with
devlink, and this involves reworking a bit how user ports which fail to
connect to their PHY (because it's missing) get reinitialized as UNUSED
devlink ports.
The desire is for the change to look something like this; basically
dsa_port_setup() has failed, we just change dp->type and call
dsa_port_setup() again.
-/* Destroy the current devlink port, and create a new one which has the UNUSED
- * flavour.
- */
-static int dsa_port_reinit_as_unused(struct dsa_port *dp)
+static int dsa_port_setup_as_unused(struct dsa_port *dp)
{
- dsa_port_devlink_teardown(dp);
dp->type = DSA_PORT_TYPE_UNUSED;
- return dsa_port_devlink_setup(dp);
+ return dsa_port_setup(dp);
}
For an UNUSED port, dsa_port_setup() mostly only calls dsa_port_devlink_setup()
anyway, so we could get away with calling just that. But if we call the
full blown dsa_port_setup(dp) (which will be needed to properly set
dp->setup = true), the callee will have the tendency to go through this
code block too, and call dsa_port_disable(dp):
switch (dp->type) {
case DSA_PORT_TYPE_UNUSED:
dsa_port_disable(dp);
break;
That is not very good, because dsa_port_disable() has this hidden inside
of it:
if (dp->pl)
phylink_stop(dp->pl);
Fact is, we are not prepared to handle a call to dsa_port_disable() with
a struct dsa_port that came from a previous (and failed) call to
dsa_port_setup(). We do not clean up dp->pl, and this will make the
second call to dsa_port_setup() call phylink_stop() on a dangling dp->pl
pointer.
Solve this by creating an API for phylink destruction which is symmetric
to the phylink creation, and never leave dp->pl set to anything except
NULL or a valid phylink structure.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-09-29 15:29:00 +08:00
|
|
|
pl = phylink_create(&dp->pl_config, of_fwnode_handle(dp->dn),
|
|
|
|
|
mode, &dsa_port_phylink_mac_ops);
|
|
|
|
|
if (IS_ERR(pl)) {
|
2022-10-08 16:39:42 +08:00
|
|
|
pr_err("error creating PHYLINK: %ld\n", PTR_ERR(pl));
|
net: dsa: don't leave dangling pointers in dp->pl when failing
There is a desire to simplify the dsa_port registration path with
devlink, and this involves reworking a bit how user ports which fail to
connect to their PHY (because it's missing) get reinitialized as UNUSED
devlink ports.
The desire is for the change to look something like this; basically
dsa_port_setup() has failed, we just change dp->type and call
dsa_port_setup() again.
-/* Destroy the current devlink port, and create a new one which has the UNUSED
- * flavour.
- */
-static int dsa_port_reinit_as_unused(struct dsa_port *dp)
+static int dsa_port_setup_as_unused(struct dsa_port *dp)
{
- dsa_port_devlink_teardown(dp);
dp->type = DSA_PORT_TYPE_UNUSED;
- return dsa_port_devlink_setup(dp);
+ return dsa_port_setup(dp);
}
For an UNUSED port, dsa_port_setup() mostly only calls dsa_port_devlink_setup()
anyway, so we could get away with calling just that. But if we call the
full blown dsa_port_setup(dp) (which will be needed to properly set
dp->setup = true), the callee will have the tendency to go through this
code block too, and call dsa_port_disable(dp):
switch (dp->type) {
case DSA_PORT_TYPE_UNUSED:
dsa_port_disable(dp);
break;
That is not very good, because dsa_port_disable() has this hidden inside
of it:
if (dp->pl)
phylink_stop(dp->pl);
Fact is, we are not prepared to handle a call to dsa_port_disable() with
a struct dsa_port that came from a previous (and failed) call to
dsa_port_setup(). We do not clean up dp->pl, and this will make the
second call to dsa_port_setup() call phylink_stop() on a dangling dp->pl
pointer.
Solve this by creating an API for phylink destruction which is symmetric
to the phylink creation, and never leave dp->pl set to anything except
NULL or a valid phylink structure.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-09-29 15:29:00 +08:00
|
|
|
return PTR_ERR(pl);
|
2021-11-30 21:09:55 +08:00
|
|
|
}
|
|
|
|
|
|
net: dsa: don't leave dangling pointers in dp->pl when failing
There is a desire to simplify the dsa_port registration path with
devlink, and this involves reworking a bit how user ports which fail to
connect to their PHY (because it's missing) get reinitialized as UNUSED
devlink ports.
The desire is for the change to look something like this; basically
dsa_port_setup() has failed, we just change dp->type and call
dsa_port_setup() again.
-/* Destroy the current devlink port, and create a new one which has the UNUSED
- * flavour.
- */
-static int dsa_port_reinit_as_unused(struct dsa_port *dp)
+static int dsa_port_setup_as_unused(struct dsa_port *dp)
{
- dsa_port_devlink_teardown(dp);
dp->type = DSA_PORT_TYPE_UNUSED;
- return dsa_port_devlink_setup(dp);
+ return dsa_port_setup(dp);
}
For an UNUSED port, dsa_port_setup() mostly only calls dsa_port_devlink_setup()
anyway, so we could get away with calling just that. But if we call the
full blown dsa_port_setup(dp) (which will be needed to properly set
dp->setup = true), the callee will have the tendency to go through this
code block too, and call dsa_port_disable(dp):
switch (dp->type) {
case DSA_PORT_TYPE_UNUSED:
dsa_port_disable(dp);
break;
That is not very good, because dsa_port_disable() has this hidden inside
of it:
if (dp->pl)
phylink_stop(dp->pl);
Fact is, we are not prepared to handle a call to dsa_port_disable() with
a struct dsa_port that came from a previous (and failed) call to
dsa_port_setup(). We do not clean up dp->pl, and this will make the
second call to dsa_port_setup() call phylink_stop() on a dangling dp->pl
pointer.
Solve this by creating an API for phylink destruction which is symmetric
to the phylink creation, and never leave dp->pl set to anything except
NULL or a valid phylink structure.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-09-29 15:29:00 +08:00
|
|
|
dp->pl = pl;
|
|
|
|
|
|
2021-11-30 21:09:55 +08:00
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
net: dsa: don't leave dangling pointers in dp->pl when failing
There is a desire to simplify the dsa_port registration path with
devlink, and this involves reworking a bit how user ports which fail to
connect to their PHY (because it's missing) get reinitialized as UNUSED
devlink ports.
The desire is for the change to look something like this; basically
dsa_port_setup() has failed, we just change dp->type and call
dsa_port_setup() again.
-/* Destroy the current devlink port, and create a new one which has the UNUSED
- * flavour.
- */
-static int dsa_port_reinit_as_unused(struct dsa_port *dp)
+static int dsa_port_setup_as_unused(struct dsa_port *dp)
{
- dsa_port_devlink_teardown(dp);
dp->type = DSA_PORT_TYPE_UNUSED;
- return dsa_port_devlink_setup(dp);
+ return dsa_port_setup(dp);
}
For an UNUSED port, dsa_port_setup() mostly only calls dsa_port_devlink_setup()
anyway, so we could get away with calling just that. But if we call the
full blown dsa_port_setup(dp) (which will be needed to properly set
dp->setup = true), the callee will have the tendency to go through this
code block too, and call dsa_port_disable(dp):
switch (dp->type) {
case DSA_PORT_TYPE_UNUSED:
dsa_port_disable(dp);
break;
That is not very good, because dsa_port_disable() has this hidden inside
of it:
if (dp->pl)
phylink_stop(dp->pl);
Fact is, we are not prepared to handle a call to dsa_port_disable() with
a struct dsa_port that came from a previous (and failed) call to
dsa_port_setup(). We do not clean up dp->pl, and this will make the
second call to dsa_port_setup() call phylink_stop() on a dangling dp->pl
pointer.
Solve this by creating an API for phylink destruction which is symmetric
to the phylink creation, and never leave dp->pl set to anything except
NULL or a valid phylink structure.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-09-29 15:29:00 +08:00
|
|
|
void dsa_port_phylink_destroy(struct dsa_port *dp)
|
|
|
|
|
{
|
|
|
|
|
phylink_destroy(dp->pl);
|
|
|
|
|
dp->pl = NULL;
|
|
|
|
|
}
|
|
|
|
|
|
2022-08-18 19:54:59 +08:00
|
|
|
static int dsa_shared_port_setup_phy_of(struct dsa_port *dp, bool enable)
|
2018-04-26 03:12:51 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
struct phy_device *phydev;
|
|
|
|
|
int port = dp->index;
|
|
|
|
|
int err = 0;
|
|
|
|
|
|
|
|
|
|
phydev = dsa_port_get_phy_device(dp);
|
|
|
|
|
if (!phydev)
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
if (IS_ERR(phydev))
|
|
|
|
|
return PTR_ERR(phydev);
|
|
|
|
|
|
2018-01-23 23:03:46 +08:00
|
|
|
if (enable) {
|
|
|
|
|
err = genphy_resume(phydev);
|
|
|
|
|
if (err < 0)
|
|
|
|
|
goto err_put_dev;
|
|
|
|
|
|
|
|
|
|
err = genphy_read_status(phydev);
|
|
|
|
|
if (err < 0)
|
|
|
|
|
goto err_put_dev;
|
|
|
|
|
} else {
|
|
|
|
|
err = genphy_suspend(phydev);
|
|
|
|
|
if (err < 0)
|
|
|
|
|
goto err_put_dev;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (ds->ops->adjust_link)
|
|
|
|
|
ds->ops->adjust_link(ds, port, phydev);
|
|
|
|
|
|
|
|
|
|
dev_dbg(ds->dev, "enabled port's phy: %s", phydev_name(phydev));
|
|
|
|
|
|
|
|
|
|
err_put_dev:
|
|
|
|
|
put_device(&phydev->mdio.dev);
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
2022-08-18 19:54:59 +08:00
|
|
|
static int dsa_shared_port_fixed_link_register_of(struct dsa_port *dp)
|
2017-10-26 22:50:07 +08:00
|
|
|
{
|
|
|
|
|
struct device_node *dn = dp->dn;
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
struct phy_device *phydev;
|
|
|
|
|
int port = dp->index;
|
net: of_get_phy_mode: Change API to solve int/unit warnings
Before this change of_get_phy_mode() returned an enum,
phy_interface_t. On error, -ENODEV etc, is returned. If the result of
the function is stored in a variable of type phy_interface_t, and the
compiler has decided to represent this as an unsigned int, comparision
with -ENODEV etc, is a signed vs unsigned comparision.
Fix this problem by changing the API. Make the function return an
error, or 0 on success, and pass a pointer, of type phy_interface_t,
where the phy mode should be stored.
v2:
Return with *interface set to PHY_INTERFACE_MODE_NA on error.
Add error checks to all users of of_get_phy_mode()
Fixup a few reverse christmas tree errors
Fixup a few slightly malformed reverse christmas trees
v3:
Fix 0-day reported errors.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-04 09:40:33 +08:00
|
|
|
phy_interface_t mode;
|
2017-10-26 22:50:07 +08:00
|
|
|
int err;
|
|
|
|
|
|
2018-01-23 23:03:46 +08:00
|
|
|
err = of_phy_register_fixed_link(dn);
|
|
|
|
|
if (err) {
|
|
|
|
|
dev_err(ds->dev,
|
|
|
|
|
"failed to register the fixed PHY of port %d\n",
|
|
|
|
|
port);
|
|
|
|
|
return err;
|
|
|
|
|
}
|
2017-10-26 22:50:07 +08:00
|
|
|
|
2018-01-23 23:03:46 +08:00
|
|
|
phydev = of_phy_find_device(dn);
|
2017-10-26 22:50:07 +08:00
|
|
|
|
net: of_get_phy_mode: Change API to solve int/unit warnings
Before this change of_get_phy_mode() returned an enum,
phy_interface_t. On error, -ENODEV etc, is returned. If the result of
the function is stored in a variable of type phy_interface_t, and the
compiler has decided to represent this as an unsigned int, comparision
with -ENODEV etc, is a signed vs unsigned comparision.
Fix this problem by changing the API. Make the function return an
error, or 0 on success, and pass a pointer, of type phy_interface_t,
where the phy mode should be stored.
v2:
Return with *interface set to PHY_INTERFACE_MODE_NA on error.
Add error checks to all users of of_get_phy_mode()
Fixup a few reverse christmas tree errors
Fixup a few slightly malformed reverse christmas trees
v3:
Fix 0-day reported errors.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-04 09:40:33 +08:00
|
|
|
err = of_get_phy_mode(dn, &mode);
|
|
|
|
|
if (err)
|
2018-01-23 23:03:46 +08:00
|
|
|
mode = PHY_INTERFACE_MODE_NA;
|
|
|
|
|
phydev->interface = mode;
|
2017-10-26 22:50:07 +08:00
|
|
|
|
2018-01-23 23:03:46 +08:00
|
|
|
genphy_read_status(phydev);
|
2017-10-26 22:50:07 +08:00
|
|
|
|
2018-01-23 23:03:46 +08:00
|
|
|
if (ds->ops->adjust_link)
|
|
|
|
|
ds->ops->adjust_link(ds, port, phydev);
|
2017-10-26 22:50:07 +08:00
|
|
|
|
2018-01-23 23:03:46 +08:00
|
|
|
put_device(&phydev->mdio.dev);
|
2017-10-26 22:50:07 +08:00
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2022-08-18 19:54:59 +08:00
|
|
|
static int dsa_shared_port_phylink_register(struct dsa_port *dp)
|
2019-05-29 01:38:16 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
struct device_node *port_dn = dp->dn;
|
net: of_get_phy_mode: Change API to solve int/unit warnings
Before this change of_get_phy_mode() returned an enum,
phy_interface_t. On error, -ENODEV etc, is returned. If the result of
the function is stored in a variable of type phy_interface_t, and the
compiler has decided to represent this as an unsigned int, comparision
with -ENODEV etc, is a signed vs unsigned comparision.
Fix this problem by changing the API. Make the function return an
error, or 0 on success, and pass a pointer, of type phy_interface_t,
where the phy mode should be stored.
v2:
Return with *interface set to PHY_INTERFACE_MODE_NA on error.
Add error checks to all users of of_get_phy_mode()
Fixup a few reverse christmas tree errors
Fixup a few slightly malformed reverse christmas trees
v3:
Fix 0-day reported errors.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-11-04 09:40:33 +08:00
|
|
|
int err;
|
2019-05-29 01:38:16 +08:00
|
|
|
|
|
|
|
|
dp->pl_config.dev = ds->dev;
|
|
|
|
|
dp->pl_config.type = PHYLINK_DEV;
|
|
|
|
|
|
2021-11-30 21:09:55 +08:00
|
|
|
err = dsa_port_phylink_create(dp);
|
|
|
|
|
if (err)
|
|
|
|
|
return err;
|
2019-05-29 01:38:16 +08:00
|
|
|
|
|
|
|
|
err = phylink_of_phy_connect(dp->pl, port_dn, 0);
|
2019-06-11 03:31:49 +08:00
|
|
|
if (err && err != -ENODEV) {
|
2019-05-29 01:38:16 +08:00
|
|
|
pr_err("could not attach to PHY: %d\n", err);
|
|
|
|
|
goto err_phy_connect;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
err_phy_connect:
|
net: dsa: don't leave dangling pointers in dp->pl when failing
There is a desire to simplify the dsa_port registration path with
devlink, and this involves reworking a bit how user ports which fail to
connect to their PHY (because it's missing) get reinitialized as UNUSED
devlink ports.
The desire is for the change to look something like this; basically
dsa_port_setup() has failed, we just change dp->type and call
dsa_port_setup() again.
-/* Destroy the current devlink port, and create a new one which has the UNUSED
- * flavour.
- */
-static int dsa_port_reinit_as_unused(struct dsa_port *dp)
+static int dsa_port_setup_as_unused(struct dsa_port *dp)
{
- dsa_port_devlink_teardown(dp);
dp->type = DSA_PORT_TYPE_UNUSED;
- return dsa_port_devlink_setup(dp);
+ return dsa_port_setup(dp);
}
For an UNUSED port, dsa_port_setup() mostly only calls dsa_port_devlink_setup()
anyway, so we could get away with calling just that. But if we call the
full blown dsa_port_setup(dp) (which will be needed to properly set
dp->setup = true), the callee will have the tendency to go through this
code block too, and call dsa_port_disable(dp):
switch (dp->type) {
case DSA_PORT_TYPE_UNUSED:
dsa_port_disable(dp);
break;
That is not very good, because dsa_port_disable() has this hidden inside
of it:
if (dp->pl)
phylink_stop(dp->pl);
Fact is, we are not prepared to handle a call to dsa_port_disable() with
a struct dsa_port that came from a previous (and failed) call to
dsa_port_setup(). We do not clean up dp->pl, and this will make the
second call to dsa_port_setup() call phylink_stop() on a dangling dp->pl
pointer.
Solve this by creating an API for phylink destruction which is symmetric
to the phylink creation, and never leave dp->pl set to anything except
NULL or a valid phylink structure.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-09-29 15:29:00 +08:00
|
|
|
dsa_port_phylink_destroy(dp);
|
2019-05-29 01:38:16 +08:00
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
net: dsa: make phylink-related OF properties mandatory on DSA and CPU ports
Early DSA drivers were kind of simplistic in that they assumed a fairly
narrow hardware layout. User ports would have integrated PHYs at an
internal MDIO address that is derivable from the port number, and shared
(DSA and CPU) ports would have an MII-style (serial or parallel)
connection to another MAC. Phylib and then phylink were used to drive
the internal PHYs, and this needed little to no description through the
platform data structures. Bringing up the shared ports at the maximum
supported link speed was the responsibility of the drivers.
As a result of this, when these early drivers were converted from
platform data to the new DSA OF bindings, there was no link information
translated into the first DT bindings.
https://lore.kernel.org/all/YtXFtTsf++AeDm1l@lunn.ch/
Later, phylink was adopted for shared ports as well, and today we have a
workaround in place, introduced by commit a20f997010c4 ("net: dsa: Don't
instantiate phylink for CPU/DSA ports unless needed"). There, DSA checks
for the presence of phy-handle/fixed-link/managed OF properties, and if
missing, phylink registration would be skipped. This is because phylink
is optional for some drivers (the shared ports already work without it),
but the process of starting to register a port with phylink is
irreversible: if phylink_create() fails to find the fwnode properties it
needs, it bails out and it leaves the ports inoperational (because
phylink expects ports to be initially down, so DSA necessarily takes
them down, and doesn't know how to put them back up again).
DSA being a common framework, new drivers opt into this workaround
willy-nilly, but the ideal behavior from the DSA core's side would have
been to not interfere with phylink's process of failing at all. This
isn't possible because of regression concerns with pre-phylink DT blobs,
but at least DSA should put a stop to the proliferation of more of such
cases that rely on the workaround to skip phylink registration, and
sanitize the environment that new drivers work in.
To that end, create a list of compatible strings for which the
workaround is preserved, and don't apply the workaround for any drivers
outside that list (this includes new drivers).
In some cases, we make the assumption that even existing drivers don't
rely on DSA's workaround, and we do this by looking at the device trees
in which they appear. We can't fully know what is the situation with
downstream DT blobs, but we can guess the overall trend by studying the
DT blobs that were submitted upstream. If there are upstream blobs that
have lacking descriptions, we take it as very likely that there are many
more downstream blobs that do so too. If all upstream blobs have
complete descriptions, we take that as a hint that the driver is a
candidate for enforcing strict DT bindings (considering that most
bindings are copy-pasted). If there are no upstream DT blobs, we take
the conservative route of allowing the workaround, unless the driver
maintainer instructs us otherwise.
The driver situation is as follows:
ar9331
~~~~~~
compatible strings:
- qca,ar9331-switch
1 occurrence in mainline device trees, part of SoC dtsi
(arch/mips/boot/dts/qca/ar9331.dtsi), description is not problematic.
Verdict: opt into strict DT bindings and out of workarounds.
b53
~~~
compatible strings:
- brcm,bcm5325
- brcm,bcm53115
- brcm,bcm53125
- brcm,bcm53128
- brcm,bcm5365
- brcm,bcm5389
- brcm,bcm5395
- brcm,bcm5397
- brcm,bcm5398
- brcm,bcm53010-srab
- brcm,bcm53011-srab
- brcm,bcm53012-srab
- brcm,bcm53018-srab
- brcm,bcm53019-srab
- brcm,bcm5301x-srab
- brcm,bcm11360-srab
- brcm,bcm58522-srab
- brcm,bcm58525-srab
- brcm,bcm58535-srab
- brcm,bcm58622-srab
- brcm,bcm58623-srab
- brcm,bcm58625-srab
- brcm,bcm88312-srab
- brcm,cygnus-srab
- brcm,nsp-srab
- brcm,omega-srab
- brcm,bcm3384-switch
- brcm,bcm6328-switch
- brcm,bcm6368-switch
- brcm,bcm63xx-switch
I've found at least these mainline DT blobs with problems:
arch/arm/boot/dts/bcm47094-linksys-panamera.dts
- lacks phy-mode
arch/arm/boot/dts/bcm47189-tenda-ac9.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47081-luxul-xap-1410.dts
arch/arm/boot/dts/bcm47081-luxul-xwr-1200.dts
arch/arm/boot/dts/bcm47081-buffalo-wzr-600dhp2.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47094-luxul-xbr-4500.dts
arch/arm/boot/dts/bcm4708-smartrg-sr400ac.dts
arch/arm/boot/dts/bcm4708-luxul-xap-1510.dts
arch/arm/boot/dts/bcm953012er.dts
arch/arm/boot/dts/bcm4708-netgear-r6250.dts
arch/arm/boot/dts/bcm4708-buffalo-wzr-1166dhp-common.dtsi
arch/arm/boot/dts/bcm4708-luxul-xwc-1000.dts
arch/arm/boot/dts/bcm47094-luxul-abr-4500.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm53016-meraki-mr32.dts
- lacks phy-mode
Verdict: opt into DSA workarounds.
bcm_sf2
~~~~~~~
compatible strings:
- brcm,bcm4908-switch
- brcm,bcm7445-switch-v4.0
- brcm,bcm7278-switch-v4.0
- brcm,bcm7278-switch-v4.8
A single occurrence in mainline
(arch/arm64/boot/dts/broadcom/bcm4908/bcm4908.dtsi), part of a SoC
dtsi, valid description. Florian Fainelli explains that most of the
bcm_sf2 device trees lack a full description for the internal IMP
ports.
Verdict: opt the BCM4908 into strict DT bindings, and opt the rest
into the workarounds. Note that even though BCM4908 has strict DT
bindings, it still does not register with phylink on the IMP port
due to it implementing ->adjust_link().
hellcreek
~~~~~~~~~
compatible strings:
- hirschmann,hellcreek-de1soc-r1
No occurrence in mainline device trees. Kurt Kanzenbach explains
that the downstream device trees lacked phy-mode and fixed link, and
needed work, but were fixed in the meantime.
Verdict: opt into strict DT bindings and out of workarounds.
lan9303
~~~~~~~
compatible strings:
- smsc,lan9303-mdio
- smsc,lan9303-i2c
1 occurrence in mainline device trees:
arch/arm/boot/dts/imx53-kp-hsc.dts
- no phy-mode, no fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
lantiq_gswip
~~~~~~~~~~~~
compatible strings:
- lantiq,xrx200-gswip
- lantiq,xrx300-gswip
- lantiq,xrx330-gswip
No occurrences in mainline device trees. Martin Blumenstingl
confirms that the downstream OpenWrt device trees lack a proper
fixed-link and need work, and that the incomplete description can
even be seen in the example from
Documentation/devicetree/bindings/net/dsa/lantiq-gswip.txt.
Verdict: opt out of strict DT bindings and into workarounds.
microchip ksz
~~~~~~~~~~~~~
compatible strings:
- microchip,ksz8765
- microchip,ksz8794
- microchip,ksz8795
- microchip,ksz8863
- microchip,ksz8873
- microchip,ksz9477
- microchip,ksz9897
- microchip,ksz9893
- microchip,ksz9563
- microchip,ksz8563
- microchip,ksz9567
- microchip,lan9370
- microchip,lan9371
- microchip,lan9372
- microchip,lan9373
- microchip,lan9374
5 occurrences in mainline device trees, all descriptions are valid.
But we had a snafu for the ksz8795 and ksz9477 drivers where the
phy-mode property would be expected to be located directly under the
'switch' node rather than under a port OF node. It was fixed by
commit edecfa98f602 ("net: dsa: microchip: look for phy-mode in port
nodes"). The driver still has compatibility with the old DT blobs.
The lan937x support was added later than the above snafu was fixed,
and even though it has support for the broken DT blobs by virtue of
sharing a common probing function, I'll take it that its DT blobs
are correct.
Verdict: opt lan937x into strict DT bindings, and the others out.
mt7530
~~~~~~
compatible strings
- mediatek,mt7621
- mediatek,mt7530
- mediatek,mt7531
Multiple occurrences in mainline device trees, one is part of an SoC
dtsi (arch/mips/boot/dts/ralink/mt7621.dtsi), all descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
mv88e6060
~~~~~~~~~
compatible string:
- marvell,mv88e6060
no occurrences in mainline, nobody knows anybody who uses it.
Verdict: opt out of strict DT bindings and into workarounds.
mv88e6xxx
~~~~~~~~~
compatible strings:
- marvell,mv88e6085
- marvell,mv88e6190
- marvell,mv88e6250
Device trees that have incomplete descriptions of CPU or DSA ports:
arch/arm64/boot/dts/freescale/imx8mq-zii-ultra.dtsi
- lacks phy-mode
arch/arm64/boot/dts/marvell/cn9130-crb.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-spu3.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-mv88f6281gtw-ge.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-spb4.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-cfu1.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-dev-rev-c.dts
- lacks phy-mode on CPU port, fixed-link on DSA ports
arch/arm/boot/dts/vf610-zii-dev-rev-b.dts
- lacks phy-mode on CPU port
arch/arm/boot/dts/armada-381-netgear-gs110emx.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-scu4-aib.dts
- lacks fixed-link on xgmii DSA ports and/or in-band-status on
2500base-x DSA ports, and phy-mode on CPU port
arch/arm/boot/dts/imx6qdl-gw5904.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-385-clearfog-gtr-l8.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-dtu.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-dir665.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-rd88f6281.dtsi
- lacks phy-mode
arch/arm/boot/dts/orion5x-netgear-wnr854t.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-388-clearfog.dts
- lacks phy-mode
arch/arm/boot/dts/armada-xp-linksys-mamba.dts
- lacks phy-mode
arch/arm/boot/dts/armada-385-linksys.dtsi
- lacks phy-mode
arch/arm/boot/dts/imx6q-b450v3.dts
arch/arm/boot/dts/imx6q-b850v3.dts
- has a phy-handle but not a phy-mode?
arch/arm/boot/dts/armada-370-rd.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-linksys-viper.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-rdu1.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-scu2-mezz.dts
- lacks phy-mode
arch/arm/boot/dts/imx6qdl-zii-rdu2.dtsi
- lacks phy-mode
arch/arm/boot/dts/armada-385-clearfog-gtr-s4.dts
- lacks phy-mode and fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
ocelot
~~~~~~
compatible strings:
- mscc,vsc9953-switch
- felix (arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi) is a PCI
device, has no compatible string
2 occurrences in mainline, both are part of SoC dtsi and complete.
Verdict: opt into strict DT bindings and out of workarounds.
qca8k
~~~~~
compatible strings:
- qca,qca8327
- qca,qca8328
- qca,qca8334
- qca,qca8337
5 occurrences in mainline device trees, none of the descriptions are
problematic.
Verdict: opt into strict DT bindings and out of workarounds.
realtek
~~~~~~~
compatible strings:
- realtek,rtl8366rb
- realtek,rtl8365mb
2 occurrences in mainline, both descriptions are fine, additionally
rtl8365mb.c has a comment "The device tree firmware should also
specify the link partner of the extension port - either via a
fixed-link or other phy-handle."
Verdict: opt into strict DT bindings and out of workarounds.
rzn1_a5psw
~~~~~~~~~~
compatible strings:
- renesas,rzn1-a5psw
One single occurrence, part of SoC dtsi
(arch/arm/boot/dts/r9a06g032.dtsi), description is fine.
Verdict: opt into strict DT bindings and out of workarounds.
sja1105
~~~~~~~
Driver already validates its port OF nodes in
sja1105_parse_ports_node().
Verdict: opt into strict DT bindings and out of workarounds.
vsc73xx
~~~~~~~
compatible strings:
- vitesse,vsc7385
- vitesse,vsc7388
- vitesse,vsc7395
- vitesse,vsc7398
2 occurrences in mainline device trees, both descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
xrs700x
~~~~~~~
compatible strings:
- arrow,xrs7003e
- arrow,xrs7003f
- arrow,xrs7004e
- arrow,xrs7004f
no occurrences in mainline, we don't know.
Verdict: opt out of strict DT bindings and into workarounds.
Because there is a pattern where newly added switches reuse existing
drivers more often than introducing new ones, I've opted for deciding
who gets to opt into the workaround based on an OF compatible match
table in the DSA core. The alternative would have been to add another
boolean property to struct dsa_switch, like configure_vlan_while_not_filtering.
But this avoids situations where sometimes driver maintainers obfuscate
what goes on by sharing a common probing function, and therefore making
new switches inherit old quirks.
Side note, we also warn about missing properties for drivers that rely
on the workaround. This isn't an indication that we'll break
compatibility with those DT blobs any time soon, but is rather done to
raise awareness about the change, for future DT blob authors.
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Frank Rowand <frowand.list@gmail.com>
Acked-by: Alvin Šipraga <alsi@bang-olufsen.dk> # realtek
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-18 19:55:00 +08:00
|
|
|
/* During the initial DSA driver migration to OF, port nodes were sometimes
|
|
|
|
|
* added to device trees with no indication of how they should operate from a
|
|
|
|
|
* link management perspective (phy-handle, fixed-link, etc). Additionally, the
|
|
|
|
|
* phy-mode may be absent. The interpretation of these port OF nodes depends on
|
|
|
|
|
* their type.
|
|
|
|
|
*
|
|
|
|
|
* User ports with no phy-handle or fixed-link are expected to connect to an
|
2023-10-24 02:17:28 +08:00
|
|
|
* internal PHY located on the ds->user_mii_bus at an MDIO address equal to
|
net: dsa: make phylink-related OF properties mandatory on DSA and CPU ports
Early DSA drivers were kind of simplistic in that they assumed a fairly
narrow hardware layout. User ports would have integrated PHYs at an
internal MDIO address that is derivable from the port number, and shared
(DSA and CPU) ports would have an MII-style (serial or parallel)
connection to another MAC. Phylib and then phylink were used to drive
the internal PHYs, and this needed little to no description through the
platform data structures. Bringing up the shared ports at the maximum
supported link speed was the responsibility of the drivers.
As a result of this, when these early drivers were converted from
platform data to the new DSA OF bindings, there was no link information
translated into the first DT bindings.
https://lore.kernel.org/all/YtXFtTsf++AeDm1l@lunn.ch/
Later, phylink was adopted for shared ports as well, and today we have a
workaround in place, introduced by commit a20f997010c4 ("net: dsa: Don't
instantiate phylink for CPU/DSA ports unless needed"). There, DSA checks
for the presence of phy-handle/fixed-link/managed OF properties, and if
missing, phylink registration would be skipped. This is because phylink
is optional for some drivers (the shared ports already work without it),
but the process of starting to register a port with phylink is
irreversible: if phylink_create() fails to find the fwnode properties it
needs, it bails out and it leaves the ports inoperational (because
phylink expects ports to be initially down, so DSA necessarily takes
them down, and doesn't know how to put them back up again).
DSA being a common framework, new drivers opt into this workaround
willy-nilly, but the ideal behavior from the DSA core's side would have
been to not interfere with phylink's process of failing at all. This
isn't possible because of regression concerns with pre-phylink DT blobs,
but at least DSA should put a stop to the proliferation of more of such
cases that rely on the workaround to skip phylink registration, and
sanitize the environment that new drivers work in.
To that end, create a list of compatible strings for which the
workaround is preserved, and don't apply the workaround for any drivers
outside that list (this includes new drivers).
In some cases, we make the assumption that even existing drivers don't
rely on DSA's workaround, and we do this by looking at the device trees
in which they appear. We can't fully know what is the situation with
downstream DT blobs, but we can guess the overall trend by studying the
DT blobs that were submitted upstream. If there are upstream blobs that
have lacking descriptions, we take it as very likely that there are many
more downstream blobs that do so too. If all upstream blobs have
complete descriptions, we take that as a hint that the driver is a
candidate for enforcing strict DT bindings (considering that most
bindings are copy-pasted). If there are no upstream DT blobs, we take
the conservative route of allowing the workaround, unless the driver
maintainer instructs us otherwise.
The driver situation is as follows:
ar9331
~~~~~~
compatible strings:
- qca,ar9331-switch
1 occurrence in mainline device trees, part of SoC dtsi
(arch/mips/boot/dts/qca/ar9331.dtsi), description is not problematic.
Verdict: opt into strict DT bindings and out of workarounds.
b53
~~~
compatible strings:
- brcm,bcm5325
- brcm,bcm53115
- brcm,bcm53125
- brcm,bcm53128
- brcm,bcm5365
- brcm,bcm5389
- brcm,bcm5395
- brcm,bcm5397
- brcm,bcm5398
- brcm,bcm53010-srab
- brcm,bcm53011-srab
- brcm,bcm53012-srab
- brcm,bcm53018-srab
- brcm,bcm53019-srab
- brcm,bcm5301x-srab
- brcm,bcm11360-srab
- brcm,bcm58522-srab
- brcm,bcm58525-srab
- brcm,bcm58535-srab
- brcm,bcm58622-srab
- brcm,bcm58623-srab
- brcm,bcm58625-srab
- brcm,bcm88312-srab
- brcm,cygnus-srab
- brcm,nsp-srab
- brcm,omega-srab
- brcm,bcm3384-switch
- brcm,bcm6328-switch
- brcm,bcm6368-switch
- brcm,bcm63xx-switch
I've found at least these mainline DT blobs with problems:
arch/arm/boot/dts/bcm47094-linksys-panamera.dts
- lacks phy-mode
arch/arm/boot/dts/bcm47189-tenda-ac9.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47081-luxul-xap-1410.dts
arch/arm/boot/dts/bcm47081-luxul-xwr-1200.dts
arch/arm/boot/dts/bcm47081-buffalo-wzr-600dhp2.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47094-luxul-xbr-4500.dts
arch/arm/boot/dts/bcm4708-smartrg-sr400ac.dts
arch/arm/boot/dts/bcm4708-luxul-xap-1510.dts
arch/arm/boot/dts/bcm953012er.dts
arch/arm/boot/dts/bcm4708-netgear-r6250.dts
arch/arm/boot/dts/bcm4708-buffalo-wzr-1166dhp-common.dtsi
arch/arm/boot/dts/bcm4708-luxul-xwc-1000.dts
arch/arm/boot/dts/bcm47094-luxul-abr-4500.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm53016-meraki-mr32.dts
- lacks phy-mode
Verdict: opt into DSA workarounds.
bcm_sf2
~~~~~~~
compatible strings:
- brcm,bcm4908-switch
- brcm,bcm7445-switch-v4.0
- brcm,bcm7278-switch-v4.0
- brcm,bcm7278-switch-v4.8
A single occurrence in mainline
(arch/arm64/boot/dts/broadcom/bcm4908/bcm4908.dtsi), part of a SoC
dtsi, valid description. Florian Fainelli explains that most of the
bcm_sf2 device trees lack a full description for the internal IMP
ports.
Verdict: opt the BCM4908 into strict DT bindings, and opt the rest
into the workarounds. Note that even though BCM4908 has strict DT
bindings, it still does not register with phylink on the IMP port
due to it implementing ->adjust_link().
hellcreek
~~~~~~~~~
compatible strings:
- hirschmann,hellcreek-de1soc-r1
No occurrence in mainline device trees. Kurt Kanzenbach explains
that the downstream device trees lacked phy-mode and fixed link, and
needed work, but were fixed in the meantime.
Verdict: opt into strict DT bindings and out of workarounds.
lan9303
~~~~~~~
compatible strings:
- smsc,lan9303-mdio
- smsc,lan9303-i2c
1 occurrence in mainline device trees:
arch/arm/boot/dts/imx53-kp-hsc.dts
- no phy-mode, no fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
lantiq_gswip
~~~~~~~~~~~~
compatible strings:
- lantiq,xrx200-gswip
- lantiq,xrx300-gswip
- lantiq,xrx330-gswip
No occurrences in mainline device trees. Martin Blumenstingl
confirms that the downstream OpenWrt device trees lack a proper
fixed-link and need work, and that the incomplete description can
even be seen in the example from
Documentation/devicetree/bindings/net/dsa/lantiq-gswip.txt.
Verdict: opt out of strict DT bindings and into workarounds.
microchip ksz
~~~~~~~~~~~~~
compatible strings:
- microchip,ksz8765
- microchip,ksz8794
- microchip,ksz8795
- microchip,ksz8863
- microchip,ksz8873
- microchip,ksz9477
- microchip,ksz9897
- microchip,ksz9893
- microchip,ksz9563
- microchip,ksz8563
- microchip,ksz9567
- microchip,lan9370
- microchip,lan9371
- microchip,lan9372
- microchip,lan9373
- microchip,lan9374
5 occurrences in mainline device trees, all descriptions are valid.
But we had a snafu for the ksz8795 and ksz9477 drivers where the
phy-mode property would be expected to be located directly under the
'switch' node rather than under a port OF node. It was fixed by
commit edecfa98f602 ("net: dsa: microchip: look for phy-mode in port
nodes"). The driver still has compatibility with the old DT blobs.
The lan937x support was added later than the above snafu was fixed,
and even though it has support for the broken DT blobs by virtue of
sharing a common probing function, I'll take it that its DT blobs
are correct.
Verdict: opt lan937x into strict DT bindings, and the others out.
mt7530
~~~~~~
compatible strings
- mediatek,mt7621
- mediatek,mt7530
- mediatek,mt7531
Multiple occurrences in mainline device trees, one is part of an SoC
dtsi (arch/mips/boot/dts/ralink/mt7621.dtsi), all descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
mv88e6060
~~~~~~~~~
compatible string:
- marvell,mv88e6060
no occurrences in mainline, nobody knows anybody who uses it.
Verdict: opt out of strict DT bindings and into workarounds.
mv88e6xxx
~~~~~~~~~
compatible strings:
- marvell,mv88e6085
- marvell,mv88e6190
- marvell,mv88e6250
Device trees that have incomplete descriptions of CPU or DSA ports:
arch/arm64/boot/dts/freescale/imx8mq-zii-ultra.dtsi
- lacks phy-mode
arch/arm64/boot/dts/marvell/cn9130-crb.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-spu3.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-mv88f6281gtw-ge.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-spb4.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-cfu1.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-dev-rev-c.dts
- lacks phy-mode on CPU port, fixed-link on DSA ports
arch/arm/boot/dts/vf610-zii-dev-rev-b.dts
- lacks phy-mode on CPU port
arch/arm/boot/dts/armada-381-netgear-gs110emx.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-scu4-aib.dts
- lacks fixed-link on xgmii DSA ports and/or in-band-status on
2500base-x DSA ports, and phy-mode on CPU port
arch/arm/boot/dts/imx6qdl-gw5904.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-385-clearfog-gtr-l8.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-dtu.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-dir665.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-rd88f6281.dtsi
- lacks phy-mode
arch/arm/boot/dts/orion5x-netgear-wnr854t.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-388-clearfog.dts
- lacks phy-mode
arch/arm/boot/dts/armada-xp-linksys-mamba.dts
- lacks phy-mode
arch/arm/boot/dts/armada-385-linksys.dtsi
- lacks phy-mode
arch/arm/boot/dts/imx6q-b450v3.dts
arch/arm/boot/dts/imx6q-b850v3.dts
- has a phy-handle but not a phy-mode?
arch/arm/boot/dts/armada-370-rd.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-linksys-viper.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-rdu1.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-scu2-mezz.dts
- lacks phy-mode
arch/arm/boot/dts/imx6qdl-zii-rdu2.dtsi
- lacks phy-mode
arch/arm/boot/dts/armada-385-clearfog-gtr-s4.dts
- lacks phy-mode and fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
ocelot
~~~~~~
compatible strings:
- mscc,vsc9953-switch
- felix (arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi) is a PCI
device, has no compatible string
2 occurrences in mainline, both are part of SoC dtsi and complete.
Verdict: opt into strict DT bindings and out of workarounds.
qca8k
~~~~~
compatible strings:
- qca,qca8327
- qca,qca8328
- qca,qca8334
- qca,qca8337
5 occurrences in mainline device trees, none of the descriptions are
problematic.
Verdict: opt into strict DT bindings and out of workarounds.
realtek
~~~~~~~
compatible strings:
- realtek,rtl8366rb
- realtek,rtl8365mb
2 occurrences in mainline, both descriptions are fine, additionally
rtl8365mb.c has a comment "The device tree firmware should also
specify the link partner of the extension port - either via a
fixed-link or other phy-handle."
Verdict: opt into strict DT bindings and out of workarounds.
rzn1_a5psw
~~~~~~~~~~
compatible strings:
- renesas,rzn1-a5psw
One single occurrence, part of SoC dtsi
(arch/arm/boot/dts/r9a06g032.dtsi), description is fine.
Verdict: opt into strict DT bindings and out of workarounds.
sja1105
~~~~~~~
Driver already validates its port OF nodes in
sja1105_parse_ports_node().
Verdict: opt into strict DT bindings and out of workarounds.
vsc73xx
~~~~~~~
compatible strings:
- vitesse,vsc7385
- vitesse,vsc7388
- vitesse,vsc7395
- vitesse,vsc7398
2 occurrences in mainline device trees, both descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
xrs700x
~~~~~~~
compatible strings:
- arrow,xrs7003e
- arrow,xrs7003f
- arrow,xrs7004e
- arrow,xrs7004f
no occurrences in mainline, we don't know.
Verdict: opt out of strict DT bindings and into workarounds.
Because there is a pattern where newly added switches reuse existing
drivers more often than introducing new ones, I've opted for deciding
who gets to opt into the workaround based on an OF compatible match
table in the DSA core. The alternative would have been to add another
boolean property to struct dsa_switch, like configure_vlan_while_not_filtering.
But this avoids situations where sometimes driver maintainers obfuscate
what goes on by sharing a common probing function, and therefore making
new switches inherit old quirks.
Side note, we also warn about missing properties for drivers that rely
on the workaround. This isn't an indication that we'll break
compatibility with those DT blobs any time soon, but is rather done to
raise awareness about the change, for future DT blob authors.
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Frank Rowand <frowand.list@gmail.com>
Acked-by: Alvin Šipraga <alsi@bang-olufsen.dk> # realtek
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-18 19:55:00 +08:00
|
|
|
* the port number. This description is still actively supported.
|
|
|
|
|
*
|
|
|
|
|
* Shared (CPU and DSA) ports with no phy-handle or fixed-link are expected to
|
|
|
|
|
* operate at the maximum speed that their phy-mode is capable of. If the
|
|
|
|
|
* phy-mode is absent, they are expected to operate using the phy-mode
|
|
|
|
|
* supported by the port that gives the highest link speed. It is unspecified
|
|
|
|
|
* if the port should use flow control or not, half duplex or full duplex, or
|
|
|
|
|
* if the phy-mode is a SERDES link, whether in-band autoneg is expected to be
|
|
|
|
|
* enabled or not.
|
|
|
|
|
*
|
|
|
|
|
* In the latter case of shared ports, omitting the link management description
|
|
|
|
|
* from the firmware node is deprecated and strongly discouraged. DSA uses
|
|
|
|
|
* phylink, which rejects the firmware nodes of these ports for lacking
|
|
|
|
|
* required properties.
|
|
|
|
|
*
|
|
|
|
|
* For switches in this table, DSA will skip enforcing validation and will
|
|
|
|
|
* later omit registering a phylink instance for the shared ports, if they lack
|
|
|
|
|
* a fixed-link, a phy-handle, or a managed = "in-band-status" property.
|
|
|
|
|
* It becomes the responsibility of the driver to ensure that these ports
|
|
|
|
|
* operate at the maximum speed (whatever this means) and will interoperate
|
2023-10-24 02:17:28 +08:00
|
|
|
* with the DSA conduit or other cascade port, since phylink methods will not be
|
net: dsa: make phylink-related OF properties mandatory on DSA and CPU ports
Early DSA drivers were kind of simplistic in that they assumed a fairly
narrow hardware layout. User ports would have integrated PHYs at an
internal MDIO address that is derivable from the port number, and shared
(DSA and CPU) ports would have an MII-style (serial or parallel)
connection to another MAC. Phylib and then phylink were used to drive
the internal PHYs, and this needed little to no description through the
platform data structures. Bringing up the shared ports at the maximum
supported link speed was the responsibility of the drivers.
As a result of this, when these early drivers were converted from
platform data to the new DSA OF bindings, there was no link information
translated into the first DT bindings.
https://lore.kernel.org/all/YtXFtTsf++AeDm1l@lunn.ch/
Later, phylink was adopted for shared ports as well, and today we have a
workaround in place, introduced by commit a20f997010c4 ("net: dsa: Don't
instantiate phylink for CPU/DSA ports unless needed"). There, DSA checks
for the presence of phy-handle/fixed-link/managed OF properties, and if
missing, phylink registration would be skipped. This is because phylink
is optional for some drivers (the shared ports already work without it),
but the process of starting to register a port with phylink is
irreversible: if phylink_create() fails to find the fwnode properties it
needs, it bails out and it leaves the ports inoperational (because
phylink expects ports to be initially down, so DSA necessarily takes
them down, and doesn't know how to put them back up again).
DSA being a common framework, new drivers opt into this workaround
willy-nilly, but the ideal behavior from the DSA core's side would have
been to not interfere with phylink's process of failing at all. This
isn't possible because of regression concerns with pre-phylink DT blobs,
but at least DSA should put a stop to the proliferation of more of such
cases that rely on the workaround to skip phylink registration, and
sanitize the environment that new drivers work in.
To that end, create a list of compatible strings for which the
workaround is preserved, and don't apply the workaround for any drivers
outside that list (this includes new drivers).
In some cases, we make the assumption that even existing drivers don't
rely on DSA's workaround, and we do this by looking at the device trees
in which they appear. We can't fully know what is the situation with
downstream DT blobs, but we can guess the overall trend by studying the
DT blobs that were submitted upstream. If there are upstream blobs that
have lacking descriptions, we take it as very likely that there are many
more downstream blobs that do so too. If all upstream blobs have
complete descriptions, we take that as a hint that the driver is a
candidate for enforcing strict DT bindings (considering that most
bindings are copy-pasted). If there are no upstream DT blobs, we take
the conservative route of allowing the workaround, unless the driver
maintainer instructs us otherwise.
The driver situation is as follows:
ar9331
~~~~~~
compatible strings:
- qca,ar9331-switch
1 occurrence in mainline device trees, part of SoC dtsi
(arch/mips/boot/dts/qca/ar9331.dtsi), description is not problematic.
Verdict: opt into strict DT bindings and out of workarounds.
b53
~~~
compatible strings:
- brcm,bcm5325
- brcm,bcm53115
- brcm,bcm53125
- brcm,bcm53128
- brcm,bcm5365
- brcm,bcm5389
- brcm,bcm5395
- brcm,bcm5397
- brcm,bcm5398
- brcm,bcm53010-srab
- brcm,bcm53011-srab
- brcm,bcm53012-srab
- brcm,bcm53018-srab
- brcm,bcm53019-srab
- brcm,bcm5301x-srab
- brcm,bcm11360-srab
- brcm,bcm58522-srab
- brcm,bcm58525-srab
- brcm,bcm58535-srab
- brcm,bcm58622-srab
- brcm,bcm58623-srab
- brcm,bcm58625-srab
- brcm,bcm88312-srab
- brcm,cygnus-srab
- brcm,nsp-srab
- brcm,omega-srab
- brcm,bcm3384-switch
- brcm,bcm6328-switch
- brcm,bcm6368-switch
- brcm,bcm63xx-switch
I've found at least these mainline DT blobs with problems:
arch/arm/boot/dts/bcm47094-linksys-panamera.dts
- lacks phy-mode
arch/arm/boot/dts/bcm47189-tenda-ac9.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47081-luxul-xap-1410.dts
arch/arm/boot/dts/bcm47081-luxul-xwr-1200.dts
arch/arm/boot/dts/bcm47081-buffalo-wzr-600dhp2.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47094-luxul-xbr-4500.dts
arch/arm/boot/dts/bcm4708-smartrg-sr400ac.dts
arch/arm/boot/dts/bcm4708-luxul-xap-1510.dts
arch/arm/boot/dts/bcm953012er.dts
arch/arm/boot/dts/bcm4708-netgear-r6250.dts
arch/arm/boot/dts/bcm4708-buffalo-wzr-1166dhp-common.dtsi
arch/arm/boot/dts/bcm4708-luxul-xwc-1000.dts
arch/arm/boot/dts/bcm47094-luxul-abr-4500.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm53016-meraki-mr32.dts
- lacks phy-mode
Verdict: opt into DSA workarounds.
bcm_sf2
~~~~~~~
compatible strings:
- brcm,bcm4908-switch
- brcm,bcm7445-switch-v4.0
- brcm,bcm7278-switch-v4.0
- brcm,bcm7278-switch-v4.8
A single occurrence in mainline
(arch/arm64/boot/dts/broadcom/bcm4908/bcm4908.dtsi), part of a SoC
dtsi, valid description. Florian Fainelli explains that most of the
bcm_sf2 device trees lack a full description for the internal IMP
ports.
Verdict: opt the BCM4908 into strict DT bindings, and opt the rest
into the workarounds. Note that even though BCM4908 has strict DT
bindings, it still does not register with phylink on the IMP port
due to it implementing ->adjust_link().
hellcreek
~~~~~~~~~
compatible strings:
- hirschmann,hellcreek-de1soc-r1
No occurrence in mainline device trees. Kurt Kanzenbach explains
that the downstream device trees lacked phy-mode and fixed link, and
needed work, but were fixed in the meantime.
Verdict: opt into strict DT bindings and out of workarounds.
lan9303
~~~~~~~
compatible strings:
- smsc,lan9303-mdio
- smsc,lan9303-i2c
1 occurrence in mainline device trees:
arch/arm/boot/dts/imx53-kp-hsc.dts
- no phy-mode, no fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
lantiq_gswip
~~~~~~~~~~~~
compatible strings:
- lantiq,xrx200-gswip
- lantiq,xrx300-gswip
- lantiq,xrx330-gswip
No occurrences in mainline device trees. Martin Blumenstingl
confirms that the downstream OpenWrt device trees lack a proper
fixed-link and need work, and that the incomplete description can
even be seen in the example from
Documentation/devicetree/bindings/net/dsa/lantiq-gswip.txt.
Verdict: opt out of strict DT bindings and into workarounds.
microchip ksz
~~~~~~~~~~~~~
compatible strings:
- microchip,ksz8765
- microchip,ksz8794
- microchip,ksz8795
- microchip,ksz8863
- microchip,ksz8873
- microchip,ksz9477
- microchip,ksz9897
- microchip,ksz9893
- microchip,ksz9563
- microchip,ksz8563
- microchip,ksz9567
- microchip,lan9370
- microchip,lan9371
- microchip,lan9372
- microchip,lan9373
- microchip,lan9374
5 occurrences in mainline device trees, all descriptions are valid.
But we had a snafu for the ksz8795 and ksz9477 drivers where the
phy-mode property would be expected to be located directly under the
'switch' node rather than under a port OF node. It was fixed by
commit edecfa98f602 ("net: dsa: microchip: look for phy-mode in port
nodes"). The driver still has compatibility with the old DT blobs.
The lan937x support was added later than the above snafu was fixed,
and even though it has support for the broken DT blobs by virtue of
sharing a common probing function, I'll take it that its DT blobs
are correct.
Verdict: opt lan937x into strict DT bindings, and the others out.
mt7530
~~~~~~
compatible strings
- mediatek,mt7621
- mediatek,mt7530
- mediatek,mt7531
Multiple occurrences in mainline device trees, one is part of an SoC
dtsi (arch/mips/boot/dts/ralink/mt7621.dtsi), all descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
mv88e6060
~~~~~~~~~
compatible string:
- marvell,mv88e6060
no occurrences in mainline, nobody knows anybody who uses it.
Verdict: opt out of strict DT bindings and into workarounds.
mv88e6xxx
~~~~~~~~~
compatible strings:
- marvell,mv88e6085
- marvell,mv88e6190
- marvell,mv88e6250
Device trees that have incomplete descriptions of CPU or DSA ports:
arch/arm64/boot/dts/freescale/imx8mq-zii-ultra.dtsi
- lacks phy-mode
arch/arm64/boot/dts/marvell/cn9130-crb.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-spu3.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-mv88f6281gtw-ge.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-spb4.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-cfu1.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-dev-rev-c.dts
- lacks phy-mode on CPU port, fixed-link on DSA ports
arch/arm/boot/dts/vf610-zii-dev-rev-b.dts
- lacks phy-mode on CPU port
arch/arm/boot/dts/armada-381-netgear-gs110emx.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-scu4-aib.dts
- lacks fixed-link on xgmii DSA ports and/or in-band-status on
2500base-x DSA ports, and phy-mode on CPU port
arch/arm/boot/dts/imx6qdl-gw5904.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-385-clearfog-gtr-l8.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-dtu.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-dir665.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-rd88f6281.dtsi
- lacks phy-mode
arch/arm/boot/dts/orion5x-netgear-wnr854t.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-388-clearfog.dts
- lacks phy-mode
arch/arm/boot/dts/armada-xp-linksys-mamba.dts
- lacks phy-mode
arch/arm/boot/dts/armada-385-linksys.dtsi
- lacks phy-mode
arch/arm/boot/dts/imx6q-b450v3.dts
arch/arm/boot/dts/imx6q-b850v3.dts
- has a phy-handle but not a phy-mode?
arch/arm/boot/dts/armada-370-rd.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-linksys-viper.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-rdu1.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-scu2-mezz.dts
- lacks phy-mode
arch/arm/boot/dts/imx6qdl-zii-rdu2.dtsi
- lacks phy-mode
arch/arm/boot/dts/armada-385-clearfog-gtr-s4.dts
- lacks phy-mode and fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
ocelot
~~~~~~
compatible strings:
- mscc,vsc9953-switch
- felix (arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi) is a PCI
device, has no compatible string
2 occurrences in mainline, both are part of SoC dtsi and complete.
Verdict: opt into strict DT bindings and out of workarounds.
qca8k
~~~~~
compatible strings:
- qca,qca8327
- qca,qca8328
- qca,qca8334
- qca,qca8337
5 occurrences in mainline device trees, none of the descriptions are
problematic.
Verdict: opt into strict DT bindings and out of workarounds.
realtek
~~~~~~~
compatible strings:
- realtek,rtl8366rb
- realtek,rtl8365mb
2 occurrences in mainline, both descriptions are fine, additionally
rtl8365mb.c has a comment "The device tree firmware should also
specify the link partner of the extension port - either via a
fixed-link or other phy-handle."
Verdict: opt into strict DT bindings and out of workarounds.
rzn1_a5psw
~~~~~~~~~~
compatible strings:
- renesas,rzn1-a5psw
One single occurrence, part of SoC dtsi
(arch/arm/boot/dts/r9a06g032.dtsi), description is fine.
Verdict: opt into strict DT bindings and out of workarounds.
sja1105
~~~~~~~
Driver already validates its port OF nodes in
sja1105_parse_ports_node().
Verdict: opt into strict DT bindings and out of workarounds.
vsc73xx
~~~~~~~
compatible strings:
- vitesse,vsc7385
- vitesse,vsc7388
- vitesse,vsc7395
- vitesse,vsc7398
2 occurrences in mainline device trees, both descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
xrs700x
~~~~~~~
compatible strings:
- arrow,xrs7003e
- arrow,xrs7003f
- arrow,xrs7004e
- arrow,xrs7004f
no occurrences in mainline, we don't know.
Verdict: opt out of strict DT bindings and into workarounds.
Because there is a pattern where newly added switches reuse existing
drivers more often than introducing new ones, I've opted for deciding
who gets to opt into the workaround based on an OF compatible match
table in the DSA core. The alternative would have been to add another
boolean property to struct dsa_switch, like configure_vlan_while_not_filtering.
But this avoids situations where sometimes driver maintainers obfuscate
what goes on by sharing a common probing function, and therefore making
new switches inherit old quirks.
Side note, we also warn about missing properties for drivers that rely
on the workaround. This isn't an indication that we'll break
compatibility with those DT blobs any time soon, but is rather done to
raise awareness about the change, for future DT blob authors.
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Frank Rowand <frowand.list@gmail.com>
Acked-by: Alvin Šipraga <alsi@bang-olufsen.dk> # realtek
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-18 19:55:00 +08:00
|
|
|
* invoked for them.
|
|
|
|
|
*
|
|
|
|
|
* If you are considering expanding this table for newly introduced switches,
|
|
|
|
|
* think again. It is OK to remove switches from this table if there aren't DT
|
|
|
|
|
* blobs in circulation which rely on defaulting the shared ports.
|
|
|
|
|
*/
|
|
|
|
|
static const char * const dsa_switches_apply_workarounds[] = {
|
|
|
|
|
#if IS_ENABLED(CONFIG_NET_DSA_XRS700X)
|
|
|
|
|
"arrow,xrs7003e",
|
|
|
|
|
"arrow,xrs7003f",
|
|
|
|
|
"arrow,xrs7004e",
|
|
|
|
|
"arrow,xrs7004f",
|
|
|
|
|
#endif
|
|
|
|
|
#if IS_ENABLED(CONFIG_B53)
|
|
|
|
|
"brcm,bcm5325",
|
|
|
|
|
"brcm,bcm53115",
|
|
|
|
|
"brcm,bcm53125",
|
|
|
|
|
"brcm,bcm53128",
|
|
|
|
|
"brcm,bcm5365",
|
|
|
|
|
"brcm,bcm5389",
|
|
|
|
|
"brcm,bcm5395",
|
|
|
|
|
"brcm,bcm5397",
|
|
|
|
|
"brcm,bcm5398",
|
|
|
|
|
"brcm,bcm53010-srab",
|
|
|
|
|
"brcm,bcm53011-srab",
|
|
|
|
|
"brcm,bcm53012-srab",
|
|
|
|
|
"brcm,bcm53018-srab",
|
|
|
|
|
"brcm,bcm53019-srab",
|
|
|
|
|
"brcm,bcm5301x-srab",
|
|
|
|
|
"brcm,bcm11360-srab",
|
|
|
|
|
"brcm,bcm58522-srab",
|
|
|
|
|
"brcm,bcm58525-srab",
|
|
|
|
|
"brcm,bcm58535-srab",
|
|
|
|
|
"brcm,bcm58622-srab",
|
|
|
|
|
"brcm,bcm58623-srab",
|
|
|
|
|
"brcm,bcm58625-srab",
|
|
|
|
|
"brcm,bcm88312-srab",
|
|
|
|
|
"brcm,cygnus-srab",
|
|
|
|
|
"brcm,nsp-srab",
|
|
|
|
|
"brcm,omega-srab",
|
|
|
|
|
"brcm,bcm3384-switch",
|
|
|
|
|
"brcm,bcm6328-switch",
|
|
|
|
|
"brcm,bcm6368-switch",
|
|
|
|
|
"brcm,bcm63xx-switch",
|
|
|
|
|
#endif
|
|
|
|
|
#if IS_ENABLED(CONFIG_NET_DSA_BCM_SF2)
|
|
|
|
|
"brcm,bcm7445-switch-v4.0",
|
|
|
|
|
"brcm,bcm7278-switch-v4.0",
|
|
|
|
|
"brcm,bcm7278-switch-v4.8",
|
|
|
|
|
#endif
|
|
|
|
|
#if IS_ENABLED(CONFIG_NET_DSA_LANTIQ_GSWIP)
|
|
|
|
|
"lantiq,xrx200-gswip",
|
|
|
|
|
"lantiq,xrx300-gswip",
|
|
|
|
|
"lantiq,xrx330-gswip",
|
|
|
|
|
#endif
|
|
|
|
|
#if IS_ENABLED(CONFIG_NET_DSA_MV88E6060)
|
|
|
|
|
"marvell,mv88e6060",
|
|
|
|
|
#endif
|
|
|
|
|
#if IS_ENABLED(CONFIG_NET_DSA_MV88E6XXX)
|
|
|
|
|
"marvell,mv88e6085",
|
|
|
|
|
"marvell,mv88e6190",
|
|
|
|
|
"marvell,mv88e6250",
|
|
|
|
|
#endif
|
|
|
|
|
#if IS_ENABLED(CONFIG_NET_DSA_MICROCHIP_KSZ_COMMON)
|
|
|
|
|
"microchip,ksz8765",
|
|
|
|
|
"microchip,ksz8794",
|
|
|
|
|
"microchip,ksz8795",
|
|
|
|
|
"microchip,ksz8863",
|
|
|
|
|
"microchip,ksz8873",
|
|
|
|
|
"microchip,ksz9477",
|
|
|
|
|
"microchip,ksz9897",
|
|
|
|
|
"microchip,ksz9893",
|
|
|
|
|
"microchip,ksz9563",
|
|
|
|
|
"microchip,ksz8563",
|
|
|
|
|
"microchip,ksz9567",
|
|
|
|
|
#endif
|
|
|
|
|
#if IS_ENABLED(CONFIG_NET_DSA_SMSC_LAN9303_MDIO)
|
|
|
|
|
"smsc,lan9303-mdio",
|
|
|
|
|
#endif
|
|
|
|
|
#if IS_ENABLED(CONFIG_NET_DSA_SMSC_LAN9303_I2C)
|
|
|
|
|
"smsc,lan9303-i2c",
|
|
|
|
|
#endif
|
|
|
|
|
NULL,
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
static void dsa_shared_port_validate_of(struct dsa_port *dp,
|
|
|
|
|
bool *missing_phy_mode,
|
|
|
|
|
bool *missing_link_description)
|
|
|
|
|
{
|
|
|
|
|
struct device_node *dn = dp->dn, *phy_np;
|
|
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
phy_interface_t mode;
|
|
|
|
|
|
|
|
|
|
*missing_phy_mode = false;
|
|
|
|
|
*missing_link_description = false;
|
|
|
|
|
|
|
|
|
|
if (of_get_phy_mode(dn, &mode)) {
|
|
|
|
|
*missing_phy_mode = true;
|
|
|
|
|
dev_err(ds->dev,
|
|
|
|
|
"OF node %pOF of %s port %d lacks the required \"phy-mode\" property\n",
|
|
|
|
|
dn, dsa_port_is_cpu(dp) ? "CPU" : "DSA", dp->index);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Note: of_phy_is_fixed_link() also returns true for
|
|
|
|
|
* managed = "in-band-status"
|
|
|
|
|
*/
|
|
|
|
|
if (of_phy_is_fixed_link(dn))
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
phy_np = of_parse_phandle(dn, "phy-handle", 0);
|
|
|
|
|
if (phy_np) {
|
|
|
|
|
of_node_put(phy_np);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
*missing_link_description = true;
|
|
|
|
|
|
|
|
|
|
dev_err(ds->dev,
|
|
|
|
|
"OF node %pOF of %s port %d lacks the required \"phy-handle\", \"fixed-link\" or \"managed\" properties\n",
|
|
|
|
|
dn, dsa_port_is_cpu(dp) ? "CPU" : "DSA", dp->index);
|
|
|
|
|
}
|
|
|
|
|
|
2022-08-18 19:54:59 +08:00
|
|
|
int dsa_shared_port_link_register_of(struct dsa_port *dp)
|
2017-10-26 22:50:07 +08:00
|
|
|
{
|
2019-05-29 01:38:16 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
net: dsa: make phylink-related OF properties mandatory on DSA and CPU ports
Early DSA drivers were kind of simplistic in that they assumed a fairly
narrow hardware layout. User ports would have integrated PHYs at an
internal MDIO address that is derivable from the port number, and shared
(DSA and CPU) ports would have an MII-style (serial or parallel)
connection to another MAC. Phylib and then phylink were used to drive
the internal PHYs, and this needed little to no description through the
platform data structures. Bringing up the shared ports at the maximum
supported link speed was the responsibility of the drivers.
As a result of this, when these early drivers were converted from
platform data to the new DSA OF bindings, there was no link information
translated into the first DT bindings.
https://lore.kernel.org/all/YtXFtTsf++AeDm1l@lunn.ch/
Later, phylink was adopted for shared ports as well, and today we have a
workaround in place, introduced by commit a20f997010c4 ("net: dsa: Don't
instantiate phylink for CPU/DSA ports unless needed"). There, DSA checks
for the presence of phy-handle/fixed-link/managed OF properties, and if
missing, phylink registration would be skipped. This is because phylink
is optional for some drivers (the shared ports already work without it),
but the process of starting to register a port with phylink is
irreversible: if phylink_create() fails to find the fwnode properties it
needs, it bails out and it leaves the ports inoperational (because
phylink expects ports to be initially down, so DSA necessarily takes
them down, and doesn't know how to put them back up again).
DSA being a common framework, new drivers opt into this workaround
willy-nilly, but the ideal behavior from the DSA core's side would have
been to not interfere with phylink's process of failing at all. This
isn't possible because of regression concerns with pre-phylink DT blobs,
but at least DSA should put a stop to the proliferation of more of such
cases that rely on the workaround to skip phylink registration, and
sanitize the environment that new drivers work in.
To that end, create a list of compatible strings for which the
workaround is preserved, and don't apply the workaround for any drivers
outside that list (this includes new drivers).
In some cases, we make the assumption that even existing drivers don't
rely on DSA's workaround, and we do this by looking at the device trees
in which they appear. We can't fully know what is the situation with
downstream DT blobs, but we can guess the overall trend by studying the
DT blobs that were submitted upstream. If there are upstream blobs that
have lacking descriptions, we take it as very likely that there are many
more downstream blobs that do so too. If all upstream blobs have
complete descriptions, we take that as a hint that the driver is a
candidate for enforcing strict DT bindings (considering that most
bindings are copy-pasted). If there are no upstream DT blobs, we take
the conservative route of allowing the workaround, unless the driver
maintainer instructs us otherwise.
The driver situation is as follows:
ar9331
~~~~~~
compatible strings:
- qca,ar9331-switch
1 occurrence in mainline device trees, part of SoC dtsi
(arch/mips/boot/dts/qca/ar9331.dtsi), description is not problematic.
Verdict: opt into strict DT bindings and out of workarounds.
b53
~~~
compatible strings:
- brcm,bcm5325
- brcm,bcm53115
- brcm,bcm53125
- brcm,bcm53128
- brcm,bcm5365
- brcm,bcm5389
- brcm,bcm5395
- brcm,bcm5397
- brcm,bcm5398
- brcm,bcm53010-srab
- brcm,bcm53011-srab
- brcm,bcm53012-srab
- brcm,bcm53018-srab
- brcm,bcm53019-srab
- brcm,bcm5301x-srab
- brcm,bcm11360-srab
- brcm,bcm58522-srab
- brcm,bcm58525-srab
- brcm,bcm58535-srab
- brcm,bcm58622-srab
- brcm,bcm58623-srab
- brcm,bcm58625-srab
- brcm,bcm88312-srab
- brcm,cygnus-srab
- brcm,nsp-srab
- brcm,omega-srab
- brcm,bcm3384-switch
- brcm,bcm6328-switch
- brcm,bcm6368-switch
- brcm,bcm63xx-switch
I've found at least these mainline DT blobs with problems:
arch/arm/boot/dts/bcm47094-linksys-panamera.dts
- lacks phy-mode
arch/arm/boot/dts/bcm47189-tenda-ac9.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47081-luxul-xap-1410.dts
arch/arm/boot/dts/bcm47081-luxul-xwr-1200.dts
arch/arm/boot/dts/bcm47081-buffalo-wzr-600dhp2.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47094-luxul-xbr-4500.dts
arch/arm/boot/dts/bcm4708-smartrg-sr400ac.dts
arch/arm/boot/dts/bcm4708-luxul-xap-1510.dts
arch/arm/boot/dts/bcm953012er.dts
arch/arm/boot/dts/bcm4708-netgear-r6250.dts
arch/arm/boot/dts/bcm4708-buffalo-wzr-1166dhp-common.dtsi
arch/arm/boot/dts/bcm4708-luxul-xwc-1000.dts
arch/arm/boot/dts/bcm47094-luxul-abr-4500.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm53016-meraki-mr32.dts
- lacks phy-mode
Verdict: opt into DSA workarounds.
bcm_sf2
~~~~~~~
compatible strings:
- brcm,bcm4908-switch
- brcm,bcm7445-switch-v4.0
- brcm,bcm7278-switch-v4.0
- brcm,bcm7278-switch-v4.8
A single occurrence in mainline
(arch/arm64/boot/dts/broadcom/bcm4908/bcm4908.dtsi), part of a SoC
dtsi, valid description. Florian Fainelli explains that most of the
bcm_sf2 device trees lack a full description for the internal IMP
ports.
Verdict: opt the BCM4908 into strict DT bindings, and opt the rest
into the workarounds. Note that even though BCM4908 has strict DT
bindings, it still does not register with phylink on the IMP port
due to it implementing ->adjust_link().
hellcreek
~~~~~~~~~
compatible strings:
- hirschmann,hellcreek-de1soc-r1
No occurrence in mainline device trees. Kurt Kanzenbach explains
that the downstream device trees lacked phy-mode and fixed link, and
needed work, but were fixed in the meantime.
Verdict: opt into strict DT bindings and out of workarounds.
lan9303
~~~~~~~
compatible strings:
- smsc,lan9303-mdio
- smsc,lan9303-i2c
1 occurrence in mainline device trees:
arch/arm/boot/dts/imx53-kp-hsc.dts
- no phy-mode, no fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
lantiq_gswip
~~~~~~~~~~~~
compatible strings:
- lantiq,xrx200-gswip
- lantiq,xrx300-gswip
- lantiq,xrx330-gswip
No occurrences in mainline device trees. Martin Blumenstingl
confirms that the downstream OpenWrt device trees lack a proper
fixed-link and need work, and that the incomplete description can
even be seen in the example from
Documentation/devicetree/bindings/net/dsa/lantiq-gswip.txt.
Verdict: opt out of strict DT bindings and into workarounds.
microchip ksz
~~~~~~~~~~~~~
compatible strings:
- microchip,ksz8765
- microchip,ksz8794
- microchip,ksz8795
- microchip,ksz8863
- microchip,ksz8873
- microchip,ksz9477
- microchip,ksz9897
- microchip,ksz9893
- microchip,ksz9563
- microchip,ksz8563
- microchip,ksz9567
- microchip,lan9370
- microchip,lan9371
- microchip,lan9372
- microchip,lan9373
- microchip,lan9374
5 occurrences in mainline device trees, all descriptions are valid.
But we had a snafu for the ksz8795 and ksz9477 drivers where the
phy-mode property would be expected to be located directly under the
'switch' node rather than under a port OF node. It was fixed by
commit edecfa98f602 ("net: dsa: microchip: look for phy-mode in port
nodes"). The driver still has compatibility with the old DT blobs.
The lan937x support was added later than the above snafu was fixed,
and even though it has support for the broken DT blobs by virtue of
sharing a common probing function, I'll take it that its DT blobs
are correct.
Verdict: opt lan937x into strict DT bindings, and the others out.
mt7530
~~~~~~
compatible strings
- mediatek,mt7621
- mediatek,mt7530
- mediatek,mt7531
Multiple occurrences in mainline device trees, one is part of an SoC
dtsi (arch/mips/boot/dts/ralink/mt7621.dtsi), all descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
mv88e6060
~~~~~~~~~
compatible string:
- marvell,mv88e6060
no occurrences in mainline, nobody knows anybody who uses it.
Verdict: opt out of strict DT bindings and into workarounds.
mv88e6xxx
~~~~~~~~~
compatible strings:
- marvell,mv88e6085
- marvell,mv88e6190
- marvell,mv88e6250
Device trees that have incomplete descriptions of CPU or DSA ports:
arch/arm64/boot/dts/freescale/imx8mq-zii-ultra.dtsi
- lacks phy-mode
arch/arm64/boot/dts/marvell/cn9130-crb.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-spu3.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-mv88f6281gtw-ge.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-spb4.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-cfu1.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-dev-rev-c.dts
- lacks phy-mode on CPU port, fixed-link on DSA ports
arch/arm/boot/dts/vf610-zii-dev-rev-b.dts
- lacks phy-mode on CPU port
arch/arm/boot/dts/armada-381-netgear-gs110emx.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-scu4-aib.dts
- lacks fixed-link on xgmii DSA ports and/or in-band-status on
2500base-x DSA ports, and phy-mode on CPU port
arch/arm/boot/dts/imx6qdl-gw5904.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-385-clearfog-gtr-l8.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-dtu.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-dir665.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-rd88f6281.dtsi
- lacks phy-mode
arch/arm/boot/dts/orion5x-netgear-wnr854t.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-388-clearfog.dts
- lacks phy-mode
arch/arm/boot/dts/armada-xp-linksys-mamba.dts
- lacks phy-mode
arch/arm/boot/dts/armada-385-linksys.dtsi
- lacks phy-mode
arch/arm/boot/dts/imx6q-b450v3.dts
arch/arm/boot/dts/imx6q-b850v3.dts
- has a phy-handle but not a phy-mode?
arch/arm/boot/dts/armada-370-rd.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-linksys-viper.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-rdu1.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-scu2-mezz.dts
- lacks phy-mode
arch/arm/boot/dts/imx6qdl-zii-rdu2.dtsi
- lacks phy-mode
arch/arm/boot/dts/armada-385-clearfog-gtr-s4.dts
- lacks phy-mode and fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
ocelot
~~~~~~
compatible strings:
- mscc,vsc9953-switch
- felix (arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi) is a PCI
device, has no compatible string
2 occurrences in mainline, both are part of SoC dtsi and complete.
Verdict: opt into strict DT bindings and out of workarounds.
qca8k
~~~~~
compatible strings:
- qca,qca8327
- qca,qca8328
- qca,qca8334
- qca,qca8337
5 occurrences in mainline device trees, none of the descriptions are
problematic.
Verdict: opt into strict DT bindings and out of workarounds.
realtek
~~~~~~~
compatible strings:
- realtek,rtl8366rb
- realtek,rtl8365mb
2 occurrences in mainline, both descriptions are fine, additionally
rtl8365mb.c has a comment "The device tree firmware should also
specify the link partner of the extension port - either via a
fixed-link or other phy-handle."
Verdict: opt into strict DT bindings and out of workarounds.
rzn1_a5psw
~~~~~~~~~~
compatible strings:
- renesas,rzn1-a5psw
One single occurrence, part of SoC dtsi
(arch/arm/boot/dts/r9a06g032.dtsi), description is fine.
Verdict: opt into strict DT bindings and out of workarounds.
sja1105
~~~~~~~
Driver already validates its port OF nodes in
sja1105_parse_ports_node().
Verdict: opt into strict DT bindings and out of workarounds.
vsc73xx
~~~~~~~
compatible strings:
- vitesse,vsc7385
- vitesse,vsc7388
- vitesse,vsc7395
- vitesse,vsc7398
2 occurrences in mainline device trees, both descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
xrs700x
~~~~~~~
compatible strings:
- arrow,xrs7003e
- arrow,xrs7003f
- arrow,xrs7004e
- arrow,xrs7004f
no occurrences in mainline, we don't know.
Verdict: opt out of strict DT bindings and into workarounds.
Because there is a pattern where newly added switches reuse existing
drivers more often than introducing new ones, I've opted for deciding
who gets to opt into the workaround based on an OF compatible match
table in the DSA core. The alternative would have been to add another
boolean property to struct dsa_switch, like configure_vlan_while_not_filtering.
But this avoids situations where sometimes driver maintainers obfuscate
what goes on by sharing a common probing function, and therefore making
new switches inherit old quirks.
Side note, we also warn about missing properties for drivers that rely
on the workaround. This isn't an indication that we'll break
compatibility with those DT blobs any time soon, but is rather done to
raise awareness about the change, for future DT blob authors.
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Frank Rowand <frowand.list@gmail.com>
Acked-by: Alvin Šipraga <alsi@bang-olufsen.dk> # realtek
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-18 19:55:00 +08:00
|
|
|
bool missing_link_description;
|
|
|
|
|
bool missing_phy_mode;
|
2020-04-14 08:34:39 +08:00
|
|
|
int port = dp->index;
|
2019-05-29 01:38:16 +08:00
|
|
|
|
net: dsa: make phylink-related OF properties mandatory on DSA and CPU ports
Early DSA drivers were kind of simplistic in that they assumed a fairly
narrow hardware layout. User ports would have integrated PHYs at an
internal MDIO address that is derivable from the port number, and shared
(DSA and CPU) ports would have an MII-style (serial or parallel)
connection to another MAC. Phylib and then phylink were used to drive
the internal PHYs, and this needed little to no description through the
platform data structures. Bringing up the shared ports at the maximum
supported link speed was the responsibility of the drivers.
As a result of this, when these early drivers were converted from
platform data to the new DSA OF bindings, there was no link information
translated into the first DT bindings.
https://lore.kernel.org/all/YtXFtTsf++AeDm1l@lunn.ch/
Later, phylink was adopted for shared ports as well, and today we have a
workaround in place, introduced by commit a20f997010c4 ("net: dsa: Don't
instantiate phylink for CPU/DSA ports unless needed"). There, DSA checks
for the presence of phy-handle/fixed-link/managed OF properties, and if
missing, phylink registration would be skipped. This is because phylink
is optional for some drivers (the shared ports already work without it),
but the process of starting to register a port with phylink is
irreversible: if phylink_create() fails to find the fwnode properties it
needs, it bails out and it leaves the ports inoperational (because
phylink expects ports to be initially down, so DSA necessarily takes
them down, and doesn't know how to put them back up again).
DSA being a common framework, new drivers opt into this workaround
willy-nilly, but the ideal behavior from the DSA core's side would have
been to not interfere with phylink's process of failing at all. This
isn't possible because of regression concerns with pre-phylink DT blobs,
but at least DSA should put a stop to the proliferation of more of such
cases that rely on the workaround to skip phylink registration, and
sanitize the environment that new drivers work in.
To that end, create a list of compatible strings for which the
workaround is preserved, and don't apply the workaround for any drivers
outside that list (this includes new drivers).
In some cases, we make the assumption that even existing drivers don't
rely on DSA's workaround, and we do this by looking at the device trees
in which they appear. We can't fully know what is the situation with
downstream DT blobs, but we can guess the overall trend by studying the
DT blobs that were submitted upstream. If there are upstream blobs that
have lacking descriptions, we take it as very likely that there are many
more downstream blobs that do so too. If all upstream blobs have
complete descriptions, we take that as a hint that the driver is a
candidate for enforcing strict DT bindings (considering that most
bindings are copy-pasted). If there are no upstream DT blobs, we take
the conservative route of allowing the workaround, unless the driver
maintainer instructs us otherwise.
The driver situation is as follows:
ar9331
~~~~~~
compatible strings:
- qca,ar9331-switch
1 occurrence in mainline device trees, part of SoC dtsi
(arch/mips/boot/dts/qca/ar9331.dtsi), description is not problematic.
Verdict: opt into strict DT bindings and out of workarounds.
b53
~~~
compatible strings:
- brcm,bcm5325
- brcm,bcm53115
- brcm,bcm53125
- brcm,bcm53128
- brcm,bcm5365
- brcm,bcm5389
- brcm,bcm5395
- brcm,bcm5397
- brcm,bcm5398
- brcm,bcm53010-srab
- brcm,bcm53011-srab
- brcm,bcm53012-srab
- brcm,bcm53018-srab
- brcm,bcm53019-srab
- brcm,bcm5301x-srab
- brcm,bcm11360-srab
- brcm,bcm58522-srab
- brcm,bcm58525-srab
- brcm,bcm58535-srab
- brcm,bcm58622-srab
- brcm,bcm58623-srab
- brcm,bcm58625-srab
- brcm,bcm88312-srab
- brcm,cygnus-srab
- brcm,nsp-srab
- brcm,omega-srab
- brcm,bcm3384-switch
- brcm,bcm6328-switch
- brcm,bcm6368-switch
- brcm,bcm63xx-switch
I've found at least these mainline DT blobs with problems:
arch/arm/boot/dts/bcm47094-linksys-panamera.dts
- lacks phy-mode
arch/arm/boot/dts/bcm47189-tenda-ac9.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47081-luxul-xap-1410.dts
arch/arm/boot/dts/bcm47081-luxul-xwr-1200.dts
arch/arm/boot/dts/bcm47081-buffalo-wzr-600dhp2.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47094-luxul-xbr-4500.dts
arch/arm/boot/dts/bcm4708-smartrg-sr400ac.dts
arch/arm/boot/dts/bcm4708-luxul-xap-1510.dts
arch/arm/boot/dts/bcm953012er.dts
arch/arm/boot/dts/bcm4708-netgear-r6250.dts
arch/arm/boot/dts/bcm4708-buffalo-wzr-1166dhp-common.dtsi
arch/arm/boot/dts/bcm4708-luxul-xwc-1000.dts
arch/arm/boot/dts/bcm47094-luxul-abr-4500.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm53016-meraki-mr32.dts
- lacks phy-mode
Verdict: opt into DSA workarounds.
bcm_sf2
~~~~~~~
compatible strings:
- brcm,bcm4908-switch
- brcm,bcm7445-switch-v4.0
- brcm,bcm7278-switch-v4.0
- brcm,bcm7278-switch-v4.8
A single occurrence in mainline
(arch/arm64/boot/dts/broadcom/bcm4908/bcm4908.dtsi), part of a SoC
dtsi, valid description. Florian Fainelli explains that most of the
bcm_sf2 device trees lack a full description for the internal IMP
ports.
Verdict: opt the BCM4908 into strict DT bindings, and opt the rest
into the workarounds. Note that even though BCM4908 has strict DT
bindings, it still does not register with phylink on the IMP port
due to it implementing ->adjust_link().
hellcreek
~~~~~~~~~
compatible strings:
- hirschmann,hellcreek-de1soc-r1
No occurrence in mainline device trees. Kurt Kanzenbach explains
that the downstream device trees lacked phy-mode and fixed link, and
needed work, but were fixed in the meantime.
Verdict: opt into strict DT bindings and out of workarounds.
lan9303
~~~~~~~
compatible strings:
- smsc,lan9303-mdio
- smsc,lan9303-i2c
1 occurrence in mainline device trees:
arch/arm/boot/dts/imx53-kp-hsc.dts
- no phy-mode, no fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
lantiq_gswip
~~~~~~~~~~~~
compatible strings:
- lantiq,xrx200-gswip
- lantiq,xrx300-gswip
- lantiq,xrx330-gswip
No occurrences in mainline device trees. Martin Blumenstingl
confirms that the downstream OpenWrt device trees lack a proper
fixed-link and need work, and that the incomplete description can
even be seen in the example from
Documentation/devicetree/bindings/net/dsa/lantiq-gswip.txt.
Verdict: opt out of strict DT bindings and into workarounds.
microchip ksz
~~~~~~~~~~~~~
compatible strings:
- microchip,ksz8765
- microchip,ksz8794
- microchip,ksz8795
- microchip,ksz8863
- microchip,ksz8873
- microchip,ksz9477
- microchip,ksz9897
- microchip,ksz9893
- microchip,ksz9563
- microchip,ksz8563
- microchip,ksz9567
- microchip,lan9370
- microchip,lan9371
- microchip,lan9372
- microchip,lan9373
- microchip,lan9374
5 occurrences in mainline device trees, all descriptions are valid.
But we had a snafu for the ksz8795 and ksz9477 drivers where the
phy-mode property would be expected to be located directly under the
'switch' node rather than under a port OF node. It was fixed by
commit edecfa98f602 ("net: dsa: microchip: look for phy-mode in port
nodes"). The driver still has compatibility with the old DT blobs.
The lan937x support was added later than the above snafu was fixed,
and even though it has support for the broken DT blobs by virtue of
sharing a common probing function, I'll take it that its DT blobs
are correct.
Verdict: opt lan937x into strict DT bindings, and the others out.
mt7530
~~~~~~
compatible strings
- mediatek,mt7621
- mediatek,mt7530
- mediatek,mt7531
Multiple occurrences in mainline device trees, one is part of an SoC
dtsi (arch/mips/boot/dts/ralink/mt7621.dtsi), all descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
mv88e6060
~~~~~~~~~
compatible string:
- marvell,mv88e6060
no occurrences in mainline, nobody knows anybody who uses it.
Verdict: opt out of strict DT bindings and into workarounds.
mv88e6xxx
~~~~~~~~~
compatible strings:
- marvell,mv88e6085
- marvell,mv88e6190
- marvell,mv88e6250
Device trees that have incomplete descriptions of CPU or DSA ports:
arch/arm64/boot/dts/freescale/imx8mq-zii-ultra.dtsi
- lacks phy-mode
arch/arm64/boot/dts/marvell/cn9130-crb.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-spu3.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-mv88f6281gtw-ge.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-spb4.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-cfu1.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-dev-rev-c.dts
- lacks phy-mode on CPU port, fixed-link on DSA ports
arch/arm/boot/dts/vf610-zii-dev-rev-b.dts
- lacks phy-mode on CPU port
arch/arm/boot/dts/armada-381-netgear-gs110emx.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-scu4-aib.dts
- lacks fixed-link on xgmii DSA ports and/or in-band-status on
2500base-x DSA ports, and phy-mode on CPU port
arch/arm/boot/dts/imx6qdl-gw5904.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-385-clearfog-gtr-l8.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-dtu.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-dir665.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-rd88f6281.dtsi
- lacks phy-mode
arch/arm/boot/dts/orion5x-netgear-wnr854t.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-388-clearfog.dts
- lacks phy-mode
arch/arm/boot/dts/armada-xp-linksys-mamba.dts
- lacks phy-mode
arch/arm/boot/dts/armada-385-linksys.dtsi
- lacks phy-mode
arch/arm/boot/dts/imx6q-b450v3.dts
arch/arm/boot/dts/imx6q-b850v3.dts
- has a phy-handle but not a phy-mode?
arch/arm/boot/dts/armada-370-rd.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-linksys-viper.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-rdu1.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-scu2-mezz.dts
- lacks phy-mode
arch/arm/boot/dts/imx6qdl-zii-rdu2.dtsi
- lacks phy-mode
arch/arm/boot/dts/armada-385-clearfog-gtr-s4.dts
- lacks phy-mode and fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
ocelot
~~~~~~
compatible strings:
- mscc,vsc9953-switch
- felix (arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi) is a PCI
device, has no compatible string
2 occurrences in mainline, both are part of SoC dtsi and complete.
Verdict: opt into strict DT bindings and out of workarounds.
qca8k
~~~~~
compatible strings:
- qca,qca8327
- qca,qca8328
- qca,qca8334
- qca,qca8337
5 occurrences in mainline device trees, none of the descriptions are
problematic.
Verdict: opt into strict DT bindings and out of workarounds.
realtek
~~~~~~~
compatible strings:
- realtek,rtl8366rb
- realtek,rtl8365mb
2 occurrences in mainline, both descriptions are fine, additionally
rtl8365mb.c has a comment "The device tree firmware should also
specify the link partner of the extension port - either via a
fixed-link or other phy-handle."
Verdict: opt into strict DT bindings and out of workarounds.
rzn1_a5psw
~~~~~~~~~~
compatible strings:
- renesas,rzn1-a5psw
One single occurrence, part of SoC dtsi
(arch/arm/boot/dts/r9a06g032.dtsi), description is fine.
Verdict: opt into strict DT bindings and out of workarounds.
sja1105
~~~~~~~
Driver already validates its port OF nodes in
sja1105_parse_ports_node().
Verdict: opt into strict DT bindings and out of workarounds.
vsc73xx
~~~~~~~
compatible strings:
- vitesse,vsc7385
- vitesse,vsc7388
- vitesse,vsc7395
- vitesse,vsc7398
2 occurrences in mainline device trees, both descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
xrs700x
~~~~~~~
compatible strings:
- arrow,xrs7003e
- arrow,xrs7003f
- arrow,xrs7004e
- arrow,xrs7004f
no occurrences in mainline, we don't know.
Verdict: opt out of strict DT bindings and into workarounds.
Because there is a pattern where newly added switches reuse existing
drivers more often than introducing new ones, I've opted for deciding
who gets to opt into the workaround based on an OF compatible match
table in the DSA core. The alternative would have been to add another
boolean property to struct dsa_switch, like configure_vlan_while_not_filtering.
But this avoids situations where sometimes driver maintainers obfuscate
what goes on by sharing a common probing function, and therefore making
new switches inherit old quirks.
Side note, we also warn about missing properties for drivers that rely
on the workaround. This isn't an indication that we'll break
compatibility with those DT blobs any time soon, but is rather done to
raise awareness about the change, for future DT blob authors.
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Frank Rowand <frowand.list@gmail.com>
Acked-by: Alvin Šipraga <alsi@bang-olufsen.dk> # realtek
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-18 19:55:00 +08:00
|
|
|
dsa_shared_port_validate_of(dp, &missing_phy_mode,
|
|
|
|
|
&missing_link_description);
|
|
|
|
|
|
|
|
|
|
if ((missing_phy_mode || missing_link_description) &&
|
|
|
|
|
!of_device_compatible_match(ds->dev->of_node,
|
|
|
|
|
dsa_switches_apply_workarounds))
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
2020-03-11 23:24:24 +08:00
|
|
|
if (!ds->ops->adjust_link) {
|
net: dsa: make phylink-related OF properties mandatory on DSA and CPU ports
Early DSA drivers were kind of simplistic in that they assumed a fairly
narrow hardware layout. User ports would have integrated PHYs at an
internal MDIO address that is derivable from the port number, and shared
(DSA and CPU) ports would have an MII-style (serial or parallel)
connection to another MAC. Phylib and then phylink were used to drive
the internal PHYs, and this needed little to no description through the
platform data structures. Bringing up the shared ports at the maximum
supported link speed was the responsibility of the drivers.
As a result of this, when these early drivers were converted from
platform data to the new DSA OF bindings, there was no link information
translated into the first DT bindings.
https://lore.kernel.org/all/YtXFtTsf++AeDm1l@lunn.ch/
Later, phylink was adopted for shared ports as well, and today we have a
workaround in place, introduced by commit a20f997010c4 ("net: dsa: Don't
instantiate phylink for CPU/DSA ports unless needed"). There, DSA checks
for the presence of phy-handle/fixed-link/managed OF properties, and if
missing, phylink registration would be skipped. This is because phylink
is optional for some drivers (the shared ports already work without it),
but the process of starting to register a port with phylink is
irreversible: if phylink_create() fails to find the fwnode properties it
needs, it bails out and it leaves the ports inoperational (because
phylink expects ports to be initially down, so DSA necessarily takes
them down, and doesn't know how to put them back up again).
DSA being a common framework, new drivers opt into this workaround
willy-nilly, but the ideal behavior from the DSA core's side would have
been to not interfere with phylink's process of failing at all. This
isn't possible because of regression concerns with pre-phylink DT blobs,
but at least DSA should put a stop to the proliferation of more of such
cases that rely on the workaround to skip phylink registration, and
sanitize the environment that new drivers work in.
To that end, create a list of compatible strings for which the
workaround is preserved, and don't apply the workaround for any drivers
outside that list (this includes new drivers).
In some cases, we make the assumption that even existing drivers don't
rely on DSA's workaround, and we do this by looking at the device trees
in which they appear. We can't fully know what is the situation with
downstream DT blobs, but we can guess the overall trend by studying the
DT blobs that were submitted upstream. If there are upstream blobs that
have lacking descriptions, we take it as very likely that there are many
more downstream blobs that do so too. If all upstream blobs have
complete descriptions, we take that as a hint that the driver is a
candidate for enforcing strict DT bindings (considering that most
bindings are copy-pasted). If there are no upstream DT blobs, we take
the conservative route of allowing the workaround, unless the driver
maintainer instructs us otherwise.
The driver situation is as follows:
ar9331
~~~~~~
compatible strings:
- qca,ar9331-switch
1 occurrence in mainline device trees, part of SoC dtsi
(arch/mips/boot/dts/qca/ar9331.dtsi), description is not problematic.
Verdict: opt into strict DT bindings and out of workarounds.
b53
~~~
compatible strings:
- brcm,bcm5325
- brcm,bcm53115
- brcm,bcm53125
- brcm,bcm53128
- brcm,bcm5365
- brcm,bcm5389
- brcm,bcm5395
- brcm,bcm5397
- brcm,bcm5398
- brcm,bcm53010-srab
- brcm,bcm53011-srab
- brcm,bcm53012-srab
- brcm,bcm53018-srab
- brcm,bcm53019-srab
- brcm,bcm5301x-srab
- brcm,bcm11360-srab
- brcm,bcm58522-srab
- brcm,bcm58525-srab
- brcm,bcm58535-srab
- brcm,bcm58622-srab
- brcm,bcm58623-srab
- brcm,bcm58625-srab
- brcm,bcm88312-srab
- brcm,cygnus-srab
- brcm,nsp-srab
- brcm,omega-srab
- brcm,bcm3384-switch
- brcm,bcm6328-switch
- brcm,bcm6368-switch
- brcm,bcm63xx-switch
I've found at least these mainline DT blobs with problems:
arch/arm/boot/dts/bcm47094-linksys-panamera.dts
- lacks phy-mode
arch/arm/boot/dts/bcm47189-tenda-ac9.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47081-luxul-xap-1410.dts
arch/arm/boot/dts/bcm47081-luxul-xwr-1200.dts
arch/arm/boot/dts/bcm47081-buffalo-wzr-600dhp2.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47094-luxul-xbr-4500.dts
arch/arm/boot/dts/bcm4708-smartrg-sr400ac.dts
arch/arm/boot/dts/bcm4708-luxul-xap-1510.dts
arch/arm/boot/dts/bcm953012er.dts
arch/arm/boot/dts/bcm4708-netgear-r6250.dts
arch/arm/boot/dts/bcm4708-buffalo-wzr-1166dhp-common.dtsi
arch/arm/boot/dts/bcm4708-luxul-xwc-1000.dts
arch/arm/boot/dts/bcm47094-luxul-abr-4500.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm53016-meraki-mr32.dts
- lacks phy-mode
Verdict: opt into DSA workarounds.
bcm_sf2
~~~~~~~
compatible strings:
- brcm,bcm4908-switch
- brcm,bcm7445-switch-v4.0
- brcm,bcm7278-switch-v4.0
- brcm,bcm7278-switch-v4.8
A single occurrence in mainline
(arch/arm64/boot/dts/broadcom/bcm4908/bcm4908.dtsi), part of a SoC
dtsi, valid description. Florian Fainelli explains that most of the
bcm_sf2 device trees lack a full description for the internal IMP
ports.
Verdict: opt the BCM4908 into strict DT bindings, and opt the rest
into the workarounds. Note that even though BCM4908 has strict DT
bindings, it still does not register with phylink on the IMP port
due to it implementing ->adjust_link().
hellcreek
~~~~~~~~~
compatible strings:
- hirschmann,hellcreek-de1soc-r1
No occurrence in mainline device trees. Kurt Kanzenbach explains
that the downstream device trees lacked phy-mode and fixed link, and
needed work, but were fixed in the meantime.
Verdict: opt into strict DT bindings and out of workarounds.
lan9303
~~~~~~~
compatible strings:
- smsc,lan9303-mdio
- smsc,lan9303-i2c
1 occurrence in mainline device trees:
arch/arm/boot/dts/imx53-kp-hsc.dts
- no phy-mode, no fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
lantiq_gswip
~~~~~~~~~~~~
compatible strings:
- lantiq,xrx200-gswip
- lantiq,xrx300-gswip
- lantiq,xrx330-gswip
No occurrences in mainline device trees. Martin Blumenstingl
confirms that the downstream OpenWrt device trees lack a proper
fixed-link and need work, and that the incomplete description can
even be seen in the example from
Documentation/devicetree/bindings/net/dsa/lantiq-gswip.txt.
Verdict: opt out of strict DT bindings and into workarounds.
microchip ksz
~~~~~~~~~~~~~
compatible strings:
- microchip,ksz8765
- microchip,ksz8794
- microchip,ksz8795
- microchip,ksz8863
- microchip,ksz8873
- microchip,ksz9477
- microchip,ksz9897
- microchip,ksz9893
- microchip,ksz9563
- microchip,ksz8563
- microchip,ksz9567
- microchip,lan9370
- microchip,lan9371
- microchip,lan9372
- microchip,lan9373
- microchip,lan9374
5 occurrences in mainline device trees, all descriptions are valid.
But we had a snafu for the ksz8795 and ksz9477 drivers where the
phy-mode property would be expected to be located directly under the
'switch' node rather than under a port OF node. It was fixed by
commit edecfa98f602 ("net: dsa: microchip: look for phy-mode in port
nodes"). The driver still has compatibility with the old DT blobs.
The lan937x support was added later than the above snafu was fixed,
and even though it has support for the broken DT blobs by virtue of
sharing a common probing function, I'll take it that its DT blobs
are correct.
Verdict: opt lan937x into strict DT bindings, and the others out.
mt7530
~~~~~~
compatible strings
- mediatek,mt7621
- mediatek,mt7530
- mediatek,mt7531
Multiple occurrences in mainline device trees, one is part of an SoC
dtsi (arch/mips/boot/dts/ralink/mt7621.dtsi), all descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
mv88e6060
~~~~~~~~~
compatible string:
- marvell,mv88e6060
no occurrences in mainline, nobody knows anybody who uses it.
Verdict: opt out of strict DT bindings and into workarounds.
mv88e6xxx
~~~~~~~~~
compatible strings:
- marvell,mv88e6085
- marvell,mv88e6190
- marvell,mv88e6250
Device trees that have incomplete descriptions of CPU or DSA ports:
arch/arm64/boot/dts/freescale/imx8mq-zii-ultra.dtsi
- lacks phy-mode
arch/arm64/boot/dts/marvell/cn9130-crb.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-spu3.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-mv88f6281gtw-ge.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-spb4.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-cfu1.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-dev-rev-c.dts
- lacks phy-mode on CPU port, fixed-link on DSA ports
arch/arm/boot/dts/vf610-zii-dev-rev-b.dts
- lacks phy-mode on CPU port
arch/arm/boot/dts/armada-381-netgear-gs110emx.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-scu4-aib.dts
- lacks fixed-link on xgmii DSA ports and/or in-band-status on
2500base-x DSA ports, and phy-mode on CPU port
arch/arm/boot/dts/imx6qdl-gw5904.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-385-clearfog-gtr-l8.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-dtu.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-dir665.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-rd88f6281.dtsi
- lacks phy-mode
arch/arm/boot/dts/orion5x-netgear-wnr854t.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-388-clearfog.dts
- lacks phy-mode
arch/arm/boot/dts/armada-xp-linksys-mamba.dts
- lacks phy-mode
arch/arm/boot/dts/armada-385-linksys.dtsi
- lacks phy-mode
arch/arm/boot/dts/imx6q-b450v3.dts
arch/arm/boot/dts/imx6q-b850v3.dts
- has a phy-handle but not a phy-mode?
arch/arm/boot/dts/armada-370-rd.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-linksys-viper.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-rdu1.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-scu2-mezz.dts
- lacks phy-mode
arch/arm/boot/dts/imx6qdl-zii-rdu2.dtsi
- lacks phy-mode
arch/arm/boot/dts/armada-385-clearfog-gtr-s4.dts
- lacks phy-mode and fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
ocelot
~~~~~~
compatible strings:
- mscc,vsc9953-switch
- felix (arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi) is a PCI
device, has no compatible string
2 occurrences in mainline, both are part of SoC dtsi and complete.
Verdict: opt into strict DT bindings and out of workarounds.
qca8k
~~~~~
compatible strings:
- qca,qca8327
- qca,qca8328
- qca,qca8334
- qca,qca8337
5 occurrences in mainline device trees, none of the descriptions are
problematic.
Verdict: opt into strict DT bindings and out of workarounds.
realtek
~~~~~~~
compatible strings:
- realtek,rtl8366rb
- realtek,rtl8365mb
2 occurrences in mainline, both descriptions are fine, additionally
rtl8365mb.c has a comment "The device tree firmware should also
specify the link partner of the extension port - either via a
fixed-link or other phy-handle."
Verdict: opt into strict DT bindings and out of workarounds.
rzn1_a5psw
~~~~~~~~~~
compatible strings:
- renesas,rzn1-a5psw
One single occurrence, part of SoC dtsi
(arch/arm/boot/dts/r9a06g032.dtsi), description is fine.
Verdict: opt into strict DT bindings and out of workarounds.
sja1105
~~~~~~~
Driver already validates its port OF nodes in
sja1105_parse_ports_node().
Verdict: opt into strict DT bindings and out of workarounds.
vsc73xx
~~~~~~~
compatible strings:
- vitesse,vsc7385
- vitesse,vsc7388
- vitesse,vsc7395
- vitesse,vsc7398
2 occurrences in mainline device trees, both descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
xrs700x
~~~~~~~
compatible strings:
- arrow,xrs7003e
- arrow,xrs7003f
- arrow,xrs7004e
- arrow,xrs7004f
no occurrences in mainline, we don't know.
Verdict: opt out of strict DT bindings and into workarounds.
Because there is a pattern where newly added switches reuse existing
drivers more often than introducing new ones, I've opted for deciding
who gets to opt into the workaround based on an OF compatible match
table in the DSA core. The alternative would have been to add another
boolean property to struct dsa_switch, like configure_vlan_while_not_filtering.
But this avoids situations where sometimes driver maintainers obfuscate
what goes on by sharing a common probing function, and therefore making
new switches inherit old quirks.
Side note, we also warn about missing properties for drivers that rely
on the workaround. This isn't an indication that we'll break
compatibility with those DT blobs any time soon, but is rather done to
raise awareness about the change, for future DT blob authors.
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Frank Rowand <frowand.list@gmail.com>
Acked-by: Alvin Šipraga <alsi@bang-olufsen.dk> # realtek
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-18 19:55:00 +08:00
|
|
|
if (missing_link_description) {
|
|
|
|
|
dev_warn(ds->dev,
|
|
|
|
|
"Skipping phylink registration for %s port %d\n",
|
|
|
|
|
dsa_port_is_cpu(dp) ? "CPU" : "DSA", dp->index);
|
|
|
|
|
} else {
|
2020-04-14 08:34:39 +08:00
|
|
|
if (ds->ops->phylink_mac_link_down)
|
|
|
|
|
ds->ops->phylink_mac_link_down(ds, port,
|
|
|
|
|
MLO_AN_FIXED, PHY_INTERFACE_MODE_NA);
|
net: dsa: make phylink-related OF properties mandatory on DSA and CPU ports
Early DSA drivers were kind of simplistic in that they assumed a fairly
narrow hardware layout. User ports would have integrated PHYs at an
internal MDIO address that is derivable from the port number, and shared
(DSA and CPU) ports would have an MII-style (serial or parallel)
connection to another MAC. Phylib and then phylink were used to drive
the internal PHYs, and this needed little to no description through the
platform data structures. Bringing up the shared ports at the maximum
supported link speed was the responsibility of the drivers.
As a result of this, when these early drivers were converted from
platform data to the new DSA OF bindings, there was no link information
translated into the first DT bindings.
https://lore.kernel.org/all/YtXFtTsf++AeDm1l@lunn.ch/
Later, phylink was adopted for shared ports as well, and today we have a
workaround in place, introduced by commit a20f997010c4 ("net: dsa: Don't
instantiate phylink for CPU/DSA ports unless needed"). There, DSA checks
for the presence of phy-handle/fixed-link/managed OF properties, and if
missing, phylink registration would be skipped. This is because phylink
is optional for some drivers (the shared ports already work without it),
but the process of starting to register a port with phylink is
irreversible: if phylink_create() fails to find the fwnode properties it
needs, it bails out and it leaves the ports inoperational (because
phylink expects ports to be initially down, so DSA necessarily takes
them down, and doesn't know how to put them back up again).
DSA being a common framework, new drivers opt into this workaround
willy-nilly, but the ideal behavior from the DSA core's side would have
been to not interfere with phylink's process of failing at all. This
isn't possible because of regression concerns with pre-phylink DT blobs,
but at least DSA should put a stop to the proliferation of more of such
cases that rely on the workaround to skip phylink registration, and
sanitize the environment that new drivers work in.
To that end, create a list of compatible strings for which the
workaround is preserved, and don't apply the workaround for any drivers
outside that list (this includes new drivers).
In some cases, we make the assumption that even existing drivers don't
rely on DSA's workaround, and we do this by looking at the device trees
in which they appear. We can't fully know what is the situation with
downstream DT blobs, but we can guess the overall trend by studying the
DT blobs that were submitted upstream. If there are upstream blobs that
have lacking descriptions, we take it as very likely that there are many
more downstream blobs that do so too. If all upstream blobs have
complete descriptions, we take that as a hint that the driver is a
candidate for enforcing strict DT bindings (considering that most
bindings are copy-pasted). If there are no upstream DT blobs, we take
the conservative route of allowing the workaround, unless the driver
maintainer instructs us otherwise.
The driver situation is as follows:
ar9331
~~~~~~
compatible strings:
- qca,ar9331-switch
1 occurrence in mainline device trees, part of SoC dtsi
(arch/mips/boot/dts/qca/ar9331.dtsi), description is not problematic.
Verdict: opt into strict DT bindings and out of workarounds.
b53
~~~
compatible strings:
- brcm,bcm5325
- brcm,bcm53115
- brcm,bcm53125
- brcm,bcm53128
- brcm,bcm5365
- brcm,bcm5389
- brcm,bcm5395
- brcm,bcm5397
- brcm,bcm5398
- brcm,bcm53010-srab
- brcm,bcm53011-srab
- brcm,bcm53012-srab
- brcm,bcm53018-srab
- brcm,bcm53019-srab
- brcm,bcm5301x-srab
- brcm,bcm11360-srab
- brcm,bcm58522-srab
- brcm,bcm58525-srab
- brcm,bcm58535-srab
- brcm,bcm58622-srab
- brcm,bcm58623-srab
- brcm,bcm58625-srab
- brcm,bcm88312-srab
- brcm,cygnus-srab
- brcm,nsp-srab
- brcm,omega-srab
- brcm,bcm3384-switch
- brcm,bcm6328-switch
- brcm,bcm6368-switch
- brcm,bcm63xx-switch
I've found at least these mainline DT blobs with problems:
arch/arm/boot/dts/bcm47094-linksys-panamera.dts
- lacks phy-mode
arch/arm/boot/dts/bcm47189-tenda-ac9.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47081-luxul-xap-1410.dts
arch/arm/boot/dts/bcm47081-luxul-xwr-1200.dts
arch/arm/boot/dts/bcm47081-buffalo-wzr-600dhp2.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm47094-luxul-xbr-4500.dts
arch/arm/boot/dts/bcm4708-smartrg-sr400ac.dts
arch/arm/boot/dts/bcm4708-luxul-xap-1510.dts
arch/arm/boot/dts/bcm953012er.dts
arch/arm/boot/dts/bcm4708-netgear-r6250.dts
arch/arm/boot/dts/bcm4708-buffalo-wzr-1166dhp-common.dtsi
arch/arm/boot/dts/bcm4708-luxul-xwc-1000.dts
arch/arm/boot/dts/bcm47094-luxul-abr-4500.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/bcm53016-meraki-mr32.dts
- lacks phy-mode
Verdict: opt into DSA workarounds.
bcm_sf2
~~~~~~~
compatible strings:
- brcm,bcm4908-switch
- brcm,bcm7445-switch-v4.0
- brcm,bcm7278-switch-v4.0
- brcm,bcm7278-switch-v4.8
A single occurrence in mainline
(arch/arm64/boot/dts/broadcom/bcm4908/bcm4908.dtsi), part of a SoC
dtsi, valid description. Florian Fainelli explains that most of the
bcm_sf2 device trees lack a full description for the internal IMP
ports.
Verdict: opt the BCM4908 into strict DT bindings, and opt the rest
into the workarounds. Note that even though BCM4908 has strict DT
bindings, it still does not register with phylink on the IMP port
due to it implementing ->adjust_link().
hellcreek
~~~~~~~~~
compatible strings:
- hirschmann,hellcreek-de1soc-r1
No occurrence in mainline device trees. Kurt Kanzenbach explains
that the downstream device trees lacked phy-mode and fixed link, and
needed work, but were fixed in the meantime.
Verdict: opt into strict DT bindings and out of workarounds.
lan9303
~~~~~~~
compatible strings:
- smsc,lan9303-mdio
- smsc,lan9303-i2c
1 occurrence in mainline device trees:
arch/arm/boot/dts/imx53-kp-hsc.dts
- no phy-mode, no fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
lantiq_gswip
~~~~~~~~~~~~
compatible strings:
- lantiq,xrx200-gswip
- lantiq,xrx300-gswip
- lantiq,xrx330-gswip
No occurrences in mainline device trees. Martin Blumenstingl
confirms that the downstream OpenWrt device trees lack a proper
fixed-link and need work, and that the incomplete description can
even be seen in the example from
Documentation/devicetree/bindings/net/dsa/lantiq-gswip.txt.
Verdict: opt out of strict DT bindings and into workarounds.
microchip ksz
~~~~~~~~~~~~~
compatible strings:
- microchip,ksz8765
- microchip,ksz8794
- microchip,ksz8795
- microchip,ksz8863
- microchip,ksz8873
- microchip,ksz9477
- microchip,ksz9897
- microchip,ksz9893
- microchip,ksz9563
- microchip,ksz8563
- microchip,ksz9567
- microchip,lan9370
- microchip,lan9371
- microchip,lan9372
- microchip,lan9373
- microchip,lan9374
5 occurrences in mainline device trees, all descriptions are valid.
But we had a snafu for the ksz8795 and ksz9477 drivers where the
phy-mode property would be expected to be located directly under the
'switch' node rather than under a port OF node. It was fixed by
commit edecfa98f602 ("net: dsa: microchip: look for phy-mode in port
nodes"). The driver still has compatibility with the old DT blobs.
The lan937x support was added later than the above snafu was fixed,
and even though it has support for the broken DT blobs by virtue of
sharing a common probing function, I'll take it that its DT blobs
are correct.
Verdict: opt lan937x into strict DT bindings, and the others out.
mt7530
~~~~~~
compatible strings
- mediatek,mt7621
- mediatek,mt7530
- mediatek,mt7531
Multiple occurrences in mainline device trees, one is part of an SoC
dtsi (arch/mips/boot/dts/ralink/mt7621.dtsi), all descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
mv88e6060
~~~~~~~~~
compatible string:
- marvell,mv88e6060
no occurrences in mainline, nobody knows anybody who uses it.
Verdict: opt out of strict DT bindings and into workarounds.
mv88e6xxx
~~~~~~~~~
compatible strings:
- marvell,mv88e6085
- marvell,mv88e6190
- marvell,mv88e6250
Device trees that have incomplete descriptions of CPU or DSA ports:
arch/arm64/boot/dts/freescale/imx8mq-zii-ultra.dtsi
- lacks phy-mode
arch/arm64/boot/dts/marvell/cn9130-crb.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-spu3.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-mv88f6281gtw-ge.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-spb4.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-cfu1.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-dev-rev-c.dts
- lacks phy-mode on CPU port, fixed-link on DSA ports
arch/arm/boot/dts/vf610-zii-dev-rev-b.dts
- lacks phy-mode on CPU port
arch/arm/boot/dts/armada-381-netgear-gs110emx.dts
- lacks phy-mode
arch/arm/boot/dts/vf610-zii-scu4-aib.dts
- lacks fixed-link on xgmii DSA ports and/or in-band-status on
2500base-x DSA ports, and phy-mode on CPU port
arch/arm/boot/dts/imx6qdl-gw5904.dtsi
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-385-clearfog-gtr-l8.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/vf610-zii-ssmb-dtu.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-dir665.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-rd88f6281.dtsi
- lacks phy-mode
arch/arm/boot/dts/orion5x-netgear-wnr854t.dts
- lacks phy-mode and fixed-link
arch/arm/boot/dts/armada-388-clearfog.dts
- lacks phy-mode
arch/arm/boot/dts/armada-xp-linksys-mamba.dts
- lacks phy-mode
arch/arm/boot/dts/armada-385-linksys.dtsi
- lacks phy-mode
arch/arm/boot/dts/imx6q-b450v3.dts
arch/arm/boot/dts/imx6q-b850v3.dts
- has a phy-handle but not a phy-mode?
arch/arm/boot/dts/armada-370-rd.dts
- lacks phy-mode
arch/arm/boot/dts/kirkwood-linksys-viper.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-rdu1.dts
- lacks phy-mode
arch/arm/boot/dts/imx51-zii-scu2-mezz.dts
- lacks phy-mode
arch/arm/boot/dts/imx6qdl-zii-rdu2.dtsi
- lacks phy-mode
arch/arm/boot/dts/armada-385-clearfog-gtr-s4.dts
- lacks phy-mode and fixed-link
Verdict: opt out of strict DT bindings and into workarounds.
ocelot
~~~~~~
compatible strings:
- mscc,vsc9953-switch
- felix (arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi) is a PCI
device, has no compatible string
2 occurrences in mainline, both are part of SoC dtsi and complete.
Verdict: opt into strict DT bindings and out of workarounds.
qca8k
~~~~~
compatible strings:
- qca,qca8327
- qca,qca8328
- qca,qca8334
- qca,qca8337
5 occurrences in mainline device trees, none of the descriptions are
problematic.
Verdict: opt into strict DT bindings and out of workarounds.
realtek
~~~~~~~
compatible strings:
- realtek,rtl8366rb
- realtek,rtl8365mb
2 occurrences in mainline, both descriptions are fine, additionally
rtl8365mb.c has a comment "The device tree firmware should also
specify the link partner of the extension port - either via a
fixed-link or other phy-handle."
Verdict: opt into strict DT bindings and out of workarounds.
rzn1_a5psw
~~~~~~~~~~
compatible strings:
- renesas,rzn1-a5psw
One single occurrence, part of SoC dtsi
(arch/arm/boot/dts/r9a06g032.dtsi), description is fine.
Verdict: opt into strict DT bindings and out of workarounds.
sja1105
~~~~~~~
Driver already validates its port OF nodes in
sja1105_parse_ports_node().
Verdict: opt into strict DT bindings and out of workarounds.
vsc73xx
~~~~~~~
compatible strings:
- vitesse,vsc7385
- vitesse,vsc7388
- vitesse,vsc7395
- vitesse,vsc7398
2 occurrences in mainline device trees, both descriptions are fine.
Verdict: opt into strict DT bindings and out of workarounds.
xrs700x
~~~~~~~
compatible strings:
- arrow,xrs7003e
- arrow,xrs7003f
- arrow,xrs7004e
- arrow,xrs7004f
no occurrences in mainline, we don't know.
Verdict: opt out of strict DT bindings and into workarounds.
Because there is a pattern where newly added switches reuse existing
drivers more often than introducing new ones, I've opted for deciding
who gets to opt into the workaround based on an OF compatible match
table in the DSA core. The alternative would have been to add another
boolean property to struct dsa_switch, like configure_vlan_while_not_filtering.
But this avoids situations where sometimes driver maintainers obfuscate
what goes on by sharing a common probing function, and therefore making
new switches inherit old quirks.
Side note, we also warn about missing properties for drivers that rely
on the workaround. This isn't an indication that we'll break
compatibility with those DT blobs any time soon, but is rather done to
raise awareness about the change, for future DT blob authors.
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Frank Rowand <frowand.list@gmail.com>
Acked-by: Alvin Šipraga <alsi@bang-olufsen.dk> # realtek
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-08-18 19:55:00 +08:00
|
|
|
|
2022-08-18 19:54:59 +08:00
|
|
|
return dsa_shared_port_phylink_register(dp);
|
2020-04-14 08:34:39 +08:00
|
|
|
}
|
2020-03-11 23:24:24 +08:00
|
|
|
return 0;
|
|
|
|
|
}
|
2019-05-29 01:38:16 +08:00
|
|
|
|
|
|
|
|
dev_warn(ds->dev,
|
|
|
|
|
"Using legacy PHYLIB callbacks. Please migrate to PHYLINK!\n");
|
|
|
|
|
|
2018-01-23 23:03:46 +08:00
|
|
|
if (of_phy_is_fixed_link(dp->dn))
|
2022-08-18 19:54:59 +08:00
|
|
|
return dsa_shared_port_fixed_link_register_of(dp);
|
2018-01-23 23:03:46 +08:00
|
|
|
else
|
2022-08-18 19:54:59 +08:00
|
|
|
return dsa_shared_port_setup_phy_of(dp, true);
|
2018-01-23 23:03:46 +08:00
|
|
|
}
|
2017-10-26 22:50:07 +08:00
|
|
|
|
2022-08-18 19:54:59 +08:00
|
|
|
void dsa_shared_port_link_unregister_of(struct dsa_port *dp)
|
2018-01-23 23:03:46 +08:00
|
|
|
{
|
2019-05-29 01:38:16 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
|
|
|
|
|
2020-03-11 23:24:24 +08:00
|
|
|
if (!ds->ops->adjust_link && dp->pl) {
|
2019-05-29 01:38:16 +08:00
|
|
|
rtnl_lock();
|
|
|
|
|
phylink_disconnect_phy(dp->pl);
|
|
|
|
|
rtnl_unlock();
|
net: dsa: don't leave dangling pointers in dp->pl when failing
There is a desire to simplify the dsa_port registration path with
devlink, and this involves reworking a bit how user ports which fail to
connect to their PHY (because it's missing) get reinitialized as UNUSED
devlink ports.
The desire is for the change to look something like this; basically
dsa_port_setup() has failed, we just change dp->type and call
dsa_port_setup() again.
-/* Destroy the current devlink port, and create a new one which has the UNUSED
- * flavour.
- */
-static int dsa_port_reinit_as_unused(struct dsa_port *dp)
+static int dsa_port_setup_as_unused(struct dsa_port *dp)
{
- dsa_port_devlink_teardown(dp);
dp->type = DSA_PORT_TYPE_UNUSED;
- return dsa_port_devlink_setup(dp);
+ return dsa_port_setup(dp);
}
For an UNUSED port, dsa_port_setup() mostly only calls dsa_port_devlink_setup()
anyway, so we could get away with calling just that. But if we call the
full blown dsa_port_setup(dp) (which will be needed to properly set
dp->setup = true), the callee will have the tendency to go through this
code block too, and call dsa_port_disable(dp):
switch (dp->type) {
case DSA_PORT_TYPE_UNUSED:
dsa_port_disable(dp);
break;
That is not very good, because dsa_port_disable() has this hidden inside
of it:
if (dp->pl)
phylink_stop(dp->pl);
Fact is, we are not prepared to handle a call to dsa_port_disable() with
a struct dsa_port that came from a previous (and failed) call to
dsa_port_setup(). We do not clean up dp->pl, and this will make the
second call to dsa_port_setup() call phylink_stop() on a dangling dp->pl
pointer.
Solve this by creating an API for phylink destruction which is symmetric
to the phylink creation, and never leave dp->pl set to anything except
NULL or a valid phylink structure.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-09-29 15:29:00 +08:00
|
|
|
dsa_port_phylink_destroy(dp);
|
2019-05-29 01:38:16 +08:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
2018-01-23 23:03:46 +08:00
|
|
|
if (of_phy_is_fixed_link(dp->dn))
|
|
|
|
|
of_phy_deregister_fixed_link(dp->dn);
|
|
|
|
|
else
|
2022-08-18 19:54:59 +08:00
|
|
|
dsa_shared_port_setup_phy_of(dp, false);
|
2017-10-26 22:50:07 +08:00
|
|
|
}
|
2018-04-26 03:12:52 +08:00
|
|
|
|
2023-09-22 21:31:04 +08:00
|
|
|
int dsa_port_hsr_join(struct dsa_port *dp, struct net_device *hsr,
|
|
|
|
|
struct netlink_ext_ack *extack)
|
2021-02-10 09:02:12 +08:00
|
|
|
{
|
2022-01-05 21:18:13 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
2021-02-10 09:02:12 +08:00
|
|
|
int err;
|
|
|
|
|
|
2022-01-05 21:18:13 +08:00
|
|
|
if (!ds->ops->port_hsr_join)
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
|
2021-02-10 09:02:12 +08:00
|
|
|
dp->hsr_dev = hsr;
|
|
|
|
|
|
2023-09-22 21:31:04 +08:00
|
|
|
err = ds->ops->port_hsr_join(ds, dp->index, hsr, extack);
|
2021-02-10 09:02:12 +08:00
|
|
|
if (err)
|
|
|
|
|
dp->hsr_dev = NULL;
|
|
|
|
|
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void dsa_port_hsr_leave(struct dsa_port *dp, struct net_device *hsr)
|
|
|
|
|
{
|
2022-01-05 21:18:13 +08:00
|
|
|
struct dsa_switch *ds = dp->ds;
|
2021-02-10 09:02:12 +08:00
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
dp->hsr_dev = NULL;
|
|
|
|
|
|
2022-01-05 21:18:13 +08:00
|
|
|
if (ds->ops->port_hsr_leave) {
|
|
|
|
|
err = ds->ops->port_hsr_leave(ds, dp->index, hsr);
|
|
|
|
|
if (err)
|
|
|
|
|
dev_err(dp->ds->dev,
|
|
|
|
|
"port %d failed to leave HSR %s: %pe\n",
|
|
|
|
|
dp->index, hsr->name, ERR_PTR(err));
|
|
|
|
|
}
|
2021-02-10 09:02:12 +08:00
|
|
|
}
|
net: dsa: tag_8021q: add proper cross-chip notifier support
The big problem which mandates cross-chip notifiers for tag_8021q is
this:
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
When the user runs:
ip link add br0 type bridge
ip link set sw0p0 master br0
ip link set sw2p0 master br0
It doesn't work.
This is because dsa_8021q_crosschip_bridge_join() assumes that "ds" and
"other_ds" are at most 1 hop away from each other, so it is sufficient
to add the RX VLAN of {ds, port} into {other_ds, other_port} and vice
versa and presto, the cross-chip link works. When there is another
switch in the middle, such as in this case switch 1 with its DSA links
sw1p3 and sw1p4, somebody needs to tell it about these VLANs too.
Which is exactly why the problem is quadratic: when a port joins a
bridge, for each port in the tree that's already in that same bridge we
notify a tag_8021q VLAN addition of that port's RX VLAN to the entire
tree. It is a very complicated web of VLANs.
It must be mentioned that currently we install tag_8021q VLANs on too
many ports (DSA links - to be precise, on all of them). For example,
when sw2p0 joins br0, and assuming sw1p0 was part of br0 too, we add the
RX VLAN of sw2p0 on the DSA links of switch 0 too, even though there
isn't any port of switch 0 that is a member of br0 (at least yet).
In theory we could notify only the switches which sit in between the
port joining the bridge and the port reacting to that bridge_join event.
But in practice that is impossible, because of the way 'link' properties
are described in the device tree. The DSA bindings require DT writers to
list out not only the real/physical DSA links, but in fact the entire
routing table, like for example switch 0 above will have:
sw0p3: port@3 {
link = <&sw1p4 &sw2p4>;
};
This was done because:
/* TODO: ideally DSA ports would have a single dp->link_dp member,
* and no dst->rtable nor this struct dsa_link would be needed,
* but this would require some more complex tree walking,
* so keep it stupid at the moment and list them all.
*/
but it is a perfect example of a situation where too much information is
actively detrimential, because we are now in the position where we
cannot distinguish a real DSA link from one that is put there to avoid
the 'complex tree walking'. And because DT is ABI, there is not much we
can change.
And because we do not know which DSA links are real and which ones
aren't, we can't really know if DSA switch A is in the data path between
switches B and C, in the general case.
So this is why tag_8021q RX VLANs are added on all DSA links, and
probably why it will never change.
On the other hand, at least the number of additions/deletions is well
balanced, and this means that once we implement reference counting at
the cross-chip notifier level a la fdb/mdb, there is absolutely zero
need for a struct dsa_8021q_crosschip_link, it's all self-managing.
In fact, with the tag_8021q notifiers emitted from the bridge join
notifiers, it becomes so generic that sja1105 does not need to do
anything anymore, we can just delete its implementation of the
.crosschip_bridge_{join,leave} methods.
Among other things we can simply delete is the home-grown implementation
of sja1105_notify_crosschip_switches(). The reason why that is wrong is
because it is not quadratic - it only covers remote switches to which we
have a cross-chip bridging link and that does not cover in-between
switches. This deletion is part of the same patch because sja1105 used
to poke deep inside the guts of the tag_8021q context in order to do
that. Because the cross-chip links went away, so needs the sja1105 code.
Last but not least, dsa_8021q_setup_port() is simplified (and also
renamed). Because our TAG_8021Q_VLAN_ADD notifier is designed to react
on the CPU port too, the four dsa_8021q_vid_apply() calls:
- 1 for RX VLAN on user port
- 1 for the user port's RX VLAN on the CPU port
- 1 for TX VLAN on user port
- 1 for the user port's TX VLAN on the CPU port
now get squashed into only 2 notifier calls via
dsa_port_tag_8021q_vlan_add.
And because the notifiers to add and to delete a tag_8021q VLAN are
distinct, now we finally break up the port setup and teardown into
separate functions instead of relying on a "bool enabled" flag which
tells us what to do. Arguably it should have been this way from the
get go.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-20 01:14:52 +08:00
|
|
|
|
net: dsa: tag_8021q: don't broadcast during setup/teardown
Currently, on my board with multiple sja1105 switches in disjoint trees
described in commit f66a6a69f97a ("net: dsa: permit cross-chip bridging
between all trees in the system"), rebooting the board triggers the
following benign warnings:
[ 12.345566] sja1105 spi2.0: port 0 failed to notify tag_8021q VLAN 1088 deletion: -ENOENT
[ 12.353804] sja1105 spi2.0: port 0 failed to notify tag_8021q VLAN 2112 deletion: -ENOENT
[ 12.362019] sja1105 spi2.0: port 1 failed to notify tag_8021q VLAN 1089 deletion: -ENOENT
[ 12.370246] sja1105 spi2.0: port 1 failed to notify tag_8021q VLAN 2113 deletion: -ENOENT
[ 12.378466] sja1105 spi2.0: port 2 failed to notify tag_8021q VLAN 1090 deletion: -ENOENT
[ 12.386683] sja1105 spi2.0: port 2 failed to notify tag_8021q VLAN 2114 deletion: -ENOENT
Basically switch 1 calls dsa_tag_8021q_unregister, and switch 1's TX and
RX VLANs cannot be found on switch 2's CPU port.
But why would switch 2 even attempt to delete switch 1's TX and RX
tag_8021q VLANs from its CPU port? Well, because we use dsa_broadcast,
and it is supposed that it had added those VLANs in the first place
(because in dsa_port_tag_8021q_vlan_match, all CPU ports match
regardless of their tree index or switch index).
The two trees probe asynchronously, and when switch 1 probed, it called
dsa_broadcast which did not notify the tree of switch 2, because that
didn't probe yet. But during unbind, switch 2's tree _is_ probed, so it
_is_ notified of the deletion.
Before jumping to introduce a synchronization mechanism between the
probing across disjoint switch trees, let's take a step back and see
whether we _need_ to do that in the first place.
The RX and TX VLANs of switch 1 would be needed on switch 2's CPU port
only if switch 1 and 2 were part of a cross-chip bridge. And
dsa_tag_8021q_bridge_join takes care precisely of that (but if probing
was synchronous, the bridge_join would just end up bumping the VLANs'
refcount, because they are already installed by the setup path).
Since by the time the ports are bridged, all DSA trees are already set
up, and we don't need the tag_8021q VLANs of one switch installed on the
other switches during probe time, the answer is that we don't need to
fix the synchronization issue.
So make the setup and teardown code paths call dsa_port_notify, which
notifies only the local tree, and the bridge code paths call
dsa_broadcast, which let the other trees know as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-11 21:46:06 +08:00
|
|
|
int dsa_port_tag_8021q_vlan_add(struct dsa_port *dp, u16 vid, bool broadcast)
|
net: dsa: tag_8021q: add proper cross-chip notifier support
The big problem which mandates cross-chip notifiers for tag_8021q is
this:
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
When the user runs:
ip link add br0 type bridge
ip link set sw0p0 master br0
ip link set sw2p0 master br0
It doesn't work.
This is because dsa_8021q_crosschip_bridge_join() assumes that "ds" and
"other_ds" are at most 1 hop away from each other, so it is sufficient
to add the RX VLAN of {ds, port} into {other_ds, other_port} and vice
versa and presto, the cross-chip link works. When there is another
switch in the middle, such as in this case switch 1 with its DSA links
sw1p3 and sw1p4, somebody needs to tell it about these VLANs too.
Which is exactly why the problem is quadratic: when a port joins a
bridge, for each port in the tree that's already in that same bridge we
notify a tag_8021q VLAN addition of that port's RX VLAN to the entire
tree. It is a very complicated web of VLANs.
It must be mentioned that currently we install tag_8021q VLANs on too
many ports (DSA links - to be precise, on all of them). For example,
when sw2p0 joins br0, and assuming sw1p0 was part of br0 too, we add the
RX VLAN of sw2p0 on the DSA links of switch 0 too, even though there
isn't any port of switch 0 that is a member of br0 (at least yet).
In theory we could notify only the switches which sit in between the
port joining the bridge and the port reacting to that bridge_join event.
But in practice that is impossible, because of the way 'link' properties
are described in the device tree. The DSA bindings require DT writers to
list out not only the real/physical DSA links, but in fact the entire
routing table, like for example switch 0 above will have:
sw0p3: port@3 {
link = <&sw1p4 &sw2p4>;
};
This was done because:
/* TODO: ideally DSA ports would have a single dp->link_dp member,
* and no dst->rtable nor this struct dsa_link would be needed,
* but this would require some more complex tree walking,
* so keep it stupid at the moment and list them all.
*/
but it is a perfect example of a situation where too much information is
actively detrimential, because we are now in the position where we
cannot distinguish a real DSA link from one that is put there to avoid
the 'complex tree walking'. And because DT is ABI, there is not much we
can change.
And because we do not know which DSA links are real and which ones
aren't, we can't really know if DSA switch A is in the data path between
switches B and C, in the general case.
So this is why tag_8021q RX VLANs are added on all DSA links, and
probably why it will never change.
On the other hand, at least the number of additions/deletions is well
balanced, and this means that once we implement reference counting at
the cross-chip notifier level a la fdb/mdb, there is absolutely zero
need for a struct dsa_8021q_crosschip_link, it's all self-managing.
In fact, with the tag_8021q notifiers emitted from the bridge join
notifiers, it becomes so generic that sja1105 does not need to do
anything anymore, we can just delete its implementation of the
.crosschip_bridge_{join,leave} methods.
Among other things we can simply delete is the home-grown implementation
of sja1105_notify_crosschip_switches(). The reason why that is wrong is
because it is not quadratic - it only covers remote switches to which we
have a cross-chip bridging link and that does not cover in-between
switches. This deletion is part of the same patch because sja1105 used
to poke deep inside the guts of the tag_8021q context in order to do
that. Because the cross-chip links went away, so needs the sja1105 code.
Last but not least, dsa_8021q_setup_port() is simplified (and also
renamed). Because our TAG_8021Q_VLAN_ADD notifier is designed to react
on the CPU port too, the four dsa_8021q_vid_apply() calls:
- 1 for RX VLAN on user port
- 1 for the user port's RX VLAN on the CPU port
- 1 for TX VLAN on user port
- 1 for the user port's TX VLAN on the CPU port
now get squashed into only 2 notifier calls via
dsa_port_tag_8021q_vlan_add.
And because the notifiers to add and to delete a tag_8021q VLAN are
distinct, now we finally break up the port setup and teardown into
separate functions instead of relying on a "bool enabled" flag which
tells us what to do. Arguably it should have been this way from the
get go.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-20 01:14:52 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_notifier_tag_8021q_vlan_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
net: dsa: tag_8021q: add proper cross-chip notifier support
The big problem which mandates cross-chip notifiers for tag_8021q is
this:
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
When the user runs:
ip link add br0 type bridge
ip link set sw0p0 master br0
ip link set sw2p0 master br0
It doesn't work.
This is because dsa_8021q_crosschip_bridge_join() assumes that "ds" and
"other_ds" are at most 1 hop away from each other, so it is sufficient
to add the RX VLAN of {ds, port} into {other_ds, other_port} and vice
versa and presto, the cross-chip link works. When there is another
switch in the middle, such as in this case switch 1 with its DSA links
sw1p3 and sw1p4, somebody needs to tell it about these VLANs too.
Which is exactly why the problem is quadratic: when a port joins a
bridge, for each port in the tree that's already in that same bridge we
notify a tag_8021q VLAN addition of that port's RX VLAN to the entire
tree. It is a very complicated web of VLANs.
It must be mentioned that currently we install tag_8021q VLANs on too
many ports (DSA links - to be precise, on all of them). For example,
when sw2p0 joins br0, and assuming sw1p0 was part of br0 too, we add the
RX VLAN of sw2p0 on the DSA links of switch 0 too, even though there
isn't any port of switch 0 that is a member of br0 (at least yet).
In theory we could notify only the switches which sit in between the
port joining the bridge and the port reacting to that bridge_join event.
But in practice that is impossible, because of the way 'link' properties
are described in the device tree. The DSA bindings require DT writers to
list out not only the real/physical DSA links, but in fact the entire
routing table, like for example switch 0 above will have:
sw0p3: port@3 {
link = <&sw1p4 &sw2p4>;
};
This was done because:
/* TODO: ideally DSA ports would have a single dp->link_dp member,
* and no dst->rtable nor this struct dsa_link would be needed,
* but this would require some more complex tree walking,
* so keep it stupid at the moment and list them all.
*/
but it is a perfect example of a situation where too much information is
actively detrimential, because we are now in the position where we
cannot distinguish a real DSA link from one that is put there to avoid
the 'complex tree walking'. And because DT is ABI, there is not much we
can change.
And because we do not know which DSA links are real and which ones
aren't, we can't really know if DSA switch A is in the data path between
switches B and C, in the general case.
So this is why tag_8021q RX VLANs are added on all DSA links, and
probably why it will never change.
On the other hand, at least the number of additions/deletions is well
balanced, and this means that once we implement reference counting at
the cross-chip notifier level a la fdb/mdb, there is absolutely zero
need for a struct dsa_8021q_crosschip_link, it's all self-managing.
In fact, with the tag_8021q notifiers emitted from the bridge join
notifiers, it becomes so generic that sja1105 does not need to do
anything anymore, we can just delete its implementation of the
.crosschip_bridge_{join,leave} methods.
Among other things we can simply delete is the home-grown implementation
of sja1105_notify_crosschip_switches(). The reason why that is wrong is
because it is not quadratic - it only covers remote switches to which we
have a cross-chip bridging link and that does not cover in-between
switches. This deletion is part of the same patch because sja1105 used
to poke deep inside the guts of the tag_8021q context in order to do
that. Because the cross-chip links went away, so needs the sja1105 code.
Last but not least, dsa_8021q_setup_port() is simplified (and also
renamed). Because our TAG_8021Q_VLAN_ADD notifier is designed to react
on the CPU port too, the four dsa_8021q_vid_apply() calls:
- 1 for RX VLAN on user port
- 1 for the user port's RX VLAN on the CPU port
- 1 for TX VLAN on user port
- 1 for the user port's TX VLAN on the CPU port
now get squashed into only 2 notifier calls via
dsa_port_tag_8021q_vlan_add.
And because the notifiers to add and to delete a tag_8021q VLAN are
distinct, now we finally break up the port setup and teardown into
separate functions instead of relying on a "bool enabled" flag which
tells us what to do. Arguably it should have been this way from the
get go.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-20 01:14:52 +08:00
|
|
|
.vid = vid,
|
|
|
|
|
};
|
|
|
|
|
|
net: dsa: tag_8021q: don't broadcast during setup/teardown
Currently, on my board with multiple sja1105 switches in disjoint trees
described in commit f66a6a69f97a ("net: dsa: permit cross-chip bridging
between all trees in the system"), rebooting the board triggers the
following benign warnings:
[ 12.345566] sja1105 spi2.0: port 0 failed to notify tag_8021q VLAN 1088 deletion: -ENOENT
[ 12.353804] sja1105 spi2.0: port 0 failed to notify tag_8021q VLAN 2112 deletion: -ENOENT
[ 12.362019] sja1105 spi2.0: port 1 failed to notify tag_8021q VLAN 1089 deletion: -ENOENT
[ 12.370246] sja1105 spi2.0: port 1 failed to notify tag_8021q VLAN 2113 deletion: -ENOENT
[ 12.378466] sja1105 spi2.0: port 2 failed to notify tag_8021q VLAN 1090 deletion: -ENOENT
[ 12.386683] sja1105 spi2.0: port 2 failed to notify tag_8021q VLAN 2114 deletion: -ENOENT
Basically switch 1 calls dsa_tag_8021q_unregister, and switch 1's TX and
RX VLANs cannot be found on switch 2's CPU port.
But why would switch 2 even attempt to delete switch 1's TX and RX
tag_8021q VLANs from its CPU port? Well, because we use dsa_broadcast,
and it is supposed that it had added those VLANs in the first place
(because in dsa_port_tag_8021q_vlan_match, all CPU ports match
regardless of their tree index or switch index).
The two trees probe asynchronously, and when switch 1 probed, it called
dsa_broadcast which did not notify the tree of switch 2, because that
didn't probe yet. But during unbind, switch 2's tree _is_ probed, so it
_is_ notified of the deletion.
Before jumping to introduce a synchronization mechanism between the
probing across disjoint switch trees, let's take a step back and see
whether we _need_ to do that in the first place.
The RX and TX VLANs of switch 1 would be needed on switch 2's CPU port
only if switch 1 and 2 were part of a cross-chip bridge. And
dsa_tag_8021q_bridge_join takes care precisely of that (but if probing
was synchronous, the bridge_join would just end up bumping the VLANs'
refcount, because they are already installed by the setup path).
Since by the time the ports are bridged, all DSA trees are already set
up, and we don't need the tag_8021q VLANs of one switch installed on the
other switches during probe time, the answer is that we don't need to
fix the synchronization issue.
So make the setup and teardown code paths call dsa_port_notify, which
notifies only the local tree, and the bridge code paths call
dsa_broadcast, which let the other trees know as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-11 21:46:06 +08:00
|
|
|
if (broadcast)
|
|
|
|
|
return dsa_broadcast(DSA_NOTIFIER_TAG_8021Q_VLAN_ADD, &info);
|
|
|
|
|
|
|
|
|
|
return dsa_port_notify(dp, DSA_NOTIFIER_TAG_8021Q_VLAN_ADD, &info);
|
net: dsa: tag_8021q: add proper cross-chip notifier support
The big problem which mandates cross-chip notifiers for tag_8021q is
this:
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
When the user runs:
ip link add br0 type bridge
ip link set sw0p0 master br0
ip link set sw2p0 master br0
It doesn't work.
This is because dsa_8021q_crosschip_bridge_join() assumes that "ds" and
"other_ds" are at most 1 hop away from each other, so it is sufficient
to add the RX VLAN of {ds, port} into {other_ds, other_port} and vice
versa and presto, the cross-chip link works. When there is another
switch in the middle, such as in this case switch 1 with its DSA links
sw1p3 and sw1p4, somebody needs to tell it about these VLANs too.
Which is exactly why the problem is quadratic: when a port joins a
bridge, for each port in the tree that's already in that same bridge we
notify a tag_8021q VLAN addition of that port's RX VLAN to the entire
tree. It is a very complicated web of VLANs.
It must be mentioned that currently we install tag_8021q VLANs on too
many ports (DSA links - to be precise, on all of them). For example,
when sw2p0 joins br0, and assuming sw1p0 was part of br0 too, we add the
RX VLAN of sw2p0 on the DSA links of switch 0 too, even though there
isn't any port of switch 0 that is a member of br0 (at least yet).
In theory we could notify only the switches which sit in between the
port joining the bridge and the port reacting to that bridge_join event.
But in practice that is impossible, because of the way 'link' properties
are described in the device tree. The DSA bindings require DT writers to
list out not only the real/physical DSA links, but in fact the entire
routing table, like for example switch 0 above will have:
sw0p3: port@3 {
link = <&sw1p4 &sw2p4>;
};
This was done because:
/* TODO: ideally DSA ports would have a single dp->link_dp member,
* and no dst->rtable nor this struct dsa_link would be needed,
* but this would require some more complex tree walking,
* so keep it stupid at the moment and list them all.
*/
but it is a perfect example of a situation where too much information is
actively detrimential, because we are now in the position where we
cannot distinguish a real DSA link from one that is put there to avoid
the 'complex tree walking'. And because DT is ABI, there is not much we
can change.
And because we do not know which DSA links are real and which ones
aren't, we can't really know if DSA switch A is in the data path between
switches B and C, in the general case.
So this is why tag_8021q RX VLANs are added on all DSA links, and
probably why it will never change.
On the other hand, at least the number of additions/deletions is well
balanced, and this means that once we implement reference counting at
the cross-chip notifier level a la fdb/mdb, there is absolutely zero
need for a struct dsa_8021q_crosschip_link, it's all self-managing.
In fact, with the tag_8021q notifiers emitted from the bridge join
notifiers, it becomes so generic that sja1105 does not need to do
anything anymore, we can just delete its implementation of the
.crosschip_bridge_{join,leave} methods.
Among other things we can simply delete is the home-grown implementation
of sja1105_notify_crosschip_switches(). The reason why that is wrong is
because it is not quadratic - it only covers remote switches to which we
have a cross-chip bridging link and that does not cover in-between
switches. This deletion is part of the same patch because sja1105 used
to poke deep inside the guts of the tag_8021q context in order to do
that. Because the cross-chip links went away, so needs the sja1105 code.
Last but not least, dsa_8021q_setup_port() is simplified (and also
renamed). Because our TAG_8021Q_VLAN_ADD notifier is designed to react
on the CPU port too, the four dsa_8021q_vid_apply() calls:
- 1 for RX VLAN on user port
- 1 for the user port's RX VLAN on the CPU port
- 1 for TX VLAN on user port
- 1 for the user port's TX VLAN on the CPU port
now get squashed into only 2 notifier calls via
dsa_port_tag_8021q_vlan_add.
And because the notifiers to add and to delete a tag_8021q VLAN are
distinct, now we finally break up the port setup and teardown into
separate functions instead of relying on a "bool enabled" flag which
tells us what to do. Arguably it should have been this way from the
get go.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-20 01:14:52 +08:00
|
|
|
}
|
|
|
|
|
|
net: dsa: tag_8021q: don't broadcast during setup/teardown
Currently, on my board with multiple sja1105 switches in disjoint trees
described in commit f66a6a69f97a ("net: dsa: permit cross-chip bridging
between all trees in the system"), rebooting the board triggers the
following benign warnings:
[ 12.345566] sja1105 spi2.0: port 0 failed to notify tag_8021q VLAN 1088 deletion: -ENOENT
[ 12.353804] sja1105 spi2.0: port 0 failed to notify tag_8021q VLAN 2112 deletion: -ENOENT
[ 12.362019] sja1105 spi2.0: port 1 failed to notify tag_8021q VLAN 1089 deletion: -ENOENT
[ 12.370246] sja1105 spi2.0: port 1 failed to notify tag_8021q VLAN 2113 deletion: -ENOENT
[ 12.378466] sja1105 spi2.0: port 2 failed to notify tag_8021q VLAN 1090 deletion: -ENOENT
[ 12.386683] sja1105 spi2.0: port 2 failed to notify tag_8021q VLAN 2114 deletion: -ENOENT
Basically switch 1 calls dsa_tag_8021q_unregister, and switch 1's TX and
RX VLANs cannot be found on switch 2's CPU port.
But why would switch 2 even attempt to delete switch 1's TX and RX
tag_8021q VLANs from its CPU port? Well, because we use dsa_broadcast,
and it is supposed that it had added those VLANs in the first place
(because in dsa_port_tag_8021q_vlan_match, all CPU ports match
regardless of their tree index or switch index).
The two trees probe asynchronously, and when switch 1 probed, it called
dsa_broadcast which did not notify the tree of switch 2, because that
didn't probe yet. But during unbind, switch 2's tree _is_ probed, so it
_is_ notified of the deletion.
Before jumping to introduce a synchronization mechanism between the
probing across disjoint switch trees, let's take a step back and see
whether we _need_ to do that in the first place.
The RX and TX VLANs of switch 1 would be needed on switch 2's CPU port
only if switch 1 and 2 were part of a cross-chip bridge. And
dsa_tag_8021q_bridge_join takes care precisely of that (but if probing
was synchronous, the bridge_join would just end up bumping the VLANs'
refcount, because they are already installed by the setup path).
Since by the time the ports are bridged, all DSA trees are already set
up, and we don't need the tag_8021q VLANs of one switch installed on the
other switches during probe time, the answer is that we don't need to
fix the synchronization issue.
So make the setup and teardown code paths call dsa_port_notify, which
notifies only the local tree, and the bridge code paths call
dsa_broadcast, which let the other trees know as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-11 21:46:06 +08:00
|
|
|
void dsa_port_tag_8021q_vlan_del(struct dsa_port *dp, u16 vid, bool broadcast)
|
net: dsa: tag_8021q: add proper cross-chip notifier support
The big problem which mandates cross-chip notifiers for tag_8021q is
this:
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
When the user runs:
ip link add br0 type bridge
ip link set sw0p0 master br0
ip link set sw2p0 master br0
It doesn't work.
This is because dsa_8021q_crosschip_bridge_join() assumes that "ds" and
"other_ds" are at most 1 hop away from each other, so it is sufficient
to add the RX VLAN of {ds, port} into {other_ds, other_port} and vice
versa and presto, the cross-chip link works. When there is another
switch in the middle, such as in this case switch 1 with its DSA links
sw1p3 and sw1p4, somebody needs to tell it about these VLANs too.
Which is exactly why the problem is quadratic: when a port joins a
bridge, for each port in the tree that's already in that same bridge we
notify a tag_8021q VLAN addition of that port's RX VLAN to the entire
tree. It is a very complicated web of VLANs.
It must be mentioned that currently we install tag_8021q VLANs on too
many ports (DSA links - to be precise, on all of them). For example,
when sw2p0 joins br0, and assuming sw1p0 was part of br0 too, we add the
RX VLAN of sw2p0 on the DSA links of switch 0 too, even though there
isn't any port of switch 0 that is a member of br0 (at least yet).
In theory we could notify only the switches which sit in between the
port joining the bridge and the port reacting to that bridge_join event.
But in practice that is impossible, because of the way 'link' properties
are described in the device tree. The DSA bindings require DT writers to
list out not only the real/physical DSA links, but in fact the entire
routing table, like for example switch 0 above will have:
sw0p3: port@3 {
link = <&sw1p4 &sw2p4>;
};
This was done because:
/* TODO: ideally DSA ports would have a single dp->link_dp member,
* and no dst->rtable nor this struct dsa_link would be needed,
* but this would require some more complex tree walking,
* so keep it stupid at the moment and list them all.
*/
but it is a perfect example of a situation where too much information is
actively detrimential, because we are now in the position where we
cannot distinguish a real DSA link from one that is put there to avoid
the 'complex tree walking'. And because DT is ABI, there is not much we
can change.
And because we do not know which DSA links are real and which ones
aren't, we can't really know if DSA switch A is in the data path between
switches B and C, in the general case.
So this is why tag_8021q RX VLANs are added on all DSA links, and
probably why it will never change.
On the other hand, at least the number of additions/deletions is well
balanced, and this means that once we implement reference counting at
the cross-chip notifier level a la fdb/mdb, there is absolutely zero
need for a struct dsa_8021q_crosschip_link, it's all self-managing.
In fact, with the tag_8021q notifiers emitted from the bridge join
notifiers, it becomes so generic that sja1105 does not need to do
anything anymore, we can just delete its implementation of the
.crosschip_bridge_{join,leave} methods.
Among other things we can simply delete is the home-grown implementation
of sja1105_notify_crosschip_switches(). The reason why that is wrong is
because it is not quadratic - it only covers remote switches to which we
have a cross-chip bridging link and that does not cover in-between
switches. This deletion is part of the same patch because sja1105 used
to poke deep inside the guts of the tag_8021q context in order to do
that. Because the cross-chip links went away, so needs the sja1105 code.
Last but not least, dsa_8021q_setup_port() is simplified (and also
renamed). Because our TAG_8021Q_VLAN_ADD notifier is designed to react
on the CPU port too, the four dsa_8021q_vid_apply() calls:
- 1 for RX VLAN on user port
- 1 for the user port's RX VLAN on the CPU port
- 1 for TX VLAN on user port
- 1 for the user port's TX VLAN on the CPU port
now get squashed into only 2 notifier calls via
dsa_port_tag_8021q_vlan_add.
And because the notifiers to add and to delete a tag_8021q VLAN are
distinct, now we finally break up the port setup and teardown into
separate functions instead of relying on a "bool enabled" flag which
tells us what to do. Arguably it should have been this way from the
get go.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-20 01:14:52 +08:00
|
|
|
{
|
|
|
|
|
struct dsa_notifier_tag_8021q_vlan_info info = {
|
net: dsa: make cross-chip notifiers more efficient for host events
To determine whether a given port should react to the port targeted by
the notifier, dsa_port_host_vlan_match() and dsa_port_host_address_match()
look at the positioning of the switch port currently executing the
notifier relative to the switch port for which the notifier was emitted.
To maintain stylistic compatibility with the other match functions from
switch.c, the host address and host VLAN match functions take the
notifier information about targeted port, switch and tree indices as
argument. However, these functions only use that information to retrieve
the struct dsa_port *targeted_dp, which is an invariant for the outer
loop that calls them. So it makes more sense to calculate the targeted
dp only once, and pass it to them as argument.
But furthermore, the targeted dp is actually known at the time the call
to dsa_port_notify() is made. It is just that we decide to only save the
indices of the port, switch and tree in the notifier structure, just to
retrace our steps and find the dp again using dsa_switch_find() and
dsa_to_port().
But both the above functions are relatively expensive, since they need
to iterate through lists. It appears more straightforward to make all
notifiers just pass the targeted dp inside their info structure, and
have the code that needs the indices to look at info->dp->index instead
of info->port, or info->dp->ds->index instead of info->sw_index, or
info->dp->ds->dst->index instead of info->tree_index.
For the sake of consistency, all cross-chip notifiers are converted to
pass the "dp" directly.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-04-15 23:46:22 +08:00
|
|
|
.dp = dp,
|
net: dsa: tag_8021q: add proper cross-chip notifier support
The big problem which mandates cross-chip notifiers for tag_8021q is
this:
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
When the user runs:
ip link add br0 type bridge
ip link set sw0p0 master br0
ip link set sw2p0 master br0
It doesn't work.
This is because dsa_8021q_crosschip_bridge_join() assumes that "ds" and
"other_ds" are at most 1 hop away from each other, so it is sufficient
to add the RX VLAN of {ds, port} into {other_ds, other_port} and vice
versa and presto, the cross-chip link works. When there is another
switch in the middle, such as in this case switch 1 with its DSA links
sw1p3 and sw1p4, somebody needs to tell it about these VLANs too.
Which is exactly why the problem is quadratic: when a port joins a
bridge, for each port in the tree that's already in that same bridge we
notify a tag_8021q VLAN addition of that port's RX VLAN to the entire
tree. It is a very complicated web of VLANs.
It must be mentioned that currently we install tag_8021q VLANs on too
many ports (DSA links - to be precise, on all of them). For example,
when sw2p0 joins br0, and assuming sw1p0 was part of br0 too, we add the
RX VLAN of sw2p0 on the DSA links of switch 0 too, even though there
isn't any port of switch 0 that is a member of br0 (at least yet).
In theory we could notify only the switches which sit in between the
port joining the bridge and the port reacting to that bridge_join event.
But in practice that is impossible, because of the way 'link' properties
are described in the device tree. The DSA bindings require DT writers to
list out not only the real/physical DSA links, but in fact the entire
routing table, like for example switch 0 above will have:
sw0p3: port@3 {
link = <&sw1p4 &sw2p4>;
};
This was done because:
/* TODO: ideally DSA ports would have a single dp->link_dp member,
* and no dst->rtable nor this struct dsa_link would be needed,
* but this would require some more complex tree walking,
* so keep it stupid at the moment and list them all.
*/
but it is a perfect example of a situation where too much information is
actively detrimential, because we are now in the position where we
cannot distinguish a real DSA link from one that is put there to avoid
the 'complex tree walking'. And because DT is ABI, there is not much we
can change.
And because we do not know which DSA links are real and which ones
aren't, we can't really know if DSA switch A is in the data path between
switches B and C, in the general case.
So this is why tag_8021q RX VLANs are added on all DSA links, and
probably why it will never change.
On the other hand, at least the number of additions/deletions is well
balanced, and this means that once we implement reference counting at
the cross-chip notifier level a la fdb/mdb, there is absolutely zero
need for a struct dsa_8021q_crosschip_link, it's all self-managing.
In fact, with the tag_8021q notifiers emitted from the bridge join
notifiers, it becomes so generic that sja1105 does not need to do
anything anymore, we can just delete its implementation of the
.crosschip_bridge_{join,leave} methods.
Among other things we can simply delete is the home-grown implementation
of sja1105_notify_crosschip_switches(). The reason why that is wrong is
because it is not quadratic - it only covers remote switches to which we
have a cross-chip bridging link and that does not cover in-between
switches. This deletion is part of the same patch because sja1105 used
to poke deep inside the guts of the tag_8021q context in order to do
that. Because the cross-chip links went away, so needs the sja1105 code.
Last but not least, dsa_8021q_setup_port() is simplified (and also
renamed). Because our TAG_8021Q_VLAN_ADD notifier is designed to react
on the CPU port too, the four dsa_8021q_vid_apply() calls:
- 1 for RX VLAN on user port
- 1 for the user port's RX VLAN on the CPU port
- 1 for TX VLAN on user port
- 1 for the user port's TX VLAN on the CPU port
now get squashed into only 2 notifier calls via
dsa_port_tag_8021q_vlan_add.
And because the notifiers to add and to delete a tag_8021q VLAN are
distinct, now we finally break up the port setup and teardown into
separate functions instead of relying on a "bool enabled" flag which
tells us what to do. Arguably it should have been this way from the
get go.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-20 01:14:52 +08:00
|
|
|
.vid = vid,
|
|
|
|
|
};
|
|
|
|
|
int err;
|
|
|
|
|
|
net: dsa: tag_8021q: don't broadcast during setup/teardown
Currently, on my board with multiple sja1105 switches in disjoint trees
described in commit f66a6a69f97a ("net: dsa: permit cross-chip bridging
between all trees in the system"), rebooting the board triggers the
following benign warnings:
[ 12.345566] sja1105 spi2.0: port 0 failed to notify tag_8021q VLAN 1088 deletion: -ENOENT
[ 12.353804] sja1105 spi2.0: port 0 failed to notify tag_8021q VLAN 2112 deletion: -ENOENT
[ 12.362019] sja1105 spi2.0: port 1 failed to notify tag_8021q VLAN 1089 deletion: -ENOENT
[ 12.370246] sja1105 spi2.0: port 1 failed to notify tag_8021q VLAN 2113 deletion: -ENOENT
[ 12.378466] sja1105 spi2.0: port 2 failed to notify tag_8021q VLAN 1090 deletion: -ENOENT
[ 12.386683] sja1105 spi2.0: port 2 failed to notify tag_8021q VLAN 2114 deletion: -ENOENT
Basically switch 1 calls dsa_tag_8021q_unregister, and switch 1's TX and
RX VLANs cannot be found on switch 2's CPU port.
But why would switch 2 even attempt to delete switch 1's TX and RX
tag_8021q VLANs from its CPU port? Well, because we use dsa_broadcast,
and it is supposed that it had added those VLANs in the first place
(because in dsa_port_tag_8021q_vlan_match, all CPU ports match
regardless of their tree index or switch index).
The two trees probe asynchronously, and when switch 1 probed, it called
dsa_broadcast which did not notify the tree of switch 2, because that
didn't probe yet. But during unbind, switch 2's tree _is_ probed, so it
_is_ notified of the deletion.
Before jumping to introduce a synchronization mechanism between the
probing across disjoint switch trees, let's take a step back and see
whether we _need_ to do that in the first place.
The RX and TX VLANs of switch 1 would be needed on switch 2's CPU port
only if switch 1 and 2 were part of a cross-chip bridge. And
dsa_tag_8021q_bridge_join takes care precisely of that (but if probing
was synchronous, the bridge_join would just end up bumping the VLANs'
refcount, because they are already installed by the setup path).
Since by the time the ports are bridged, all DSA trees are already set
up, and we don't need the tag_8021q VLANs of one switch installed on the
other switches during probe time, the answer is that we don't need to
fix the synchronization issue.
So make the setup and teardown code paths call dsa_port_notify, which
notifies only the local tree, and the bridge code paths call
dsa_broadcast, which let the other trees know as well.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-11 21:46:06 +08:00
|
|
|
if (broadcast)
|
|
|
|
|
err = dsa_broadcast(DSA_NOTIFIER_TAG_8021Q_VLAN_DEL, &info);
|
|
|
|
|
else
|
|
|
|
|
err = dsa_port_notify(dp, DSA_NOTIFIER_TAG_8021Q_VLAN_DEL, &info);
|
net: dsa: tag_8021q: add proper cross-chip notifier support
The big problem which mandates cross-chip notifiers for tag_8021q is
this:
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
When the user runs:
ip link add br0 type bridge
ip link set sw0p0 master br0
ip link set sw2p0 master br0
It doesn't work.
This is because dsa_8021q_crosschip_bridge_join() assumes that "ds" and
"other_ds" are at most 1 hop away from each other, so it is sufficient
to add the RX VLAN of {ds, port} into {other_ds, other_port} and vice
versa and presto, the cross-chip link works. When there is another
switch in the middle, such as in this case switch 1 with its DSA links
sw1p3 and sw1p4, somebody needs to tell it about these VLANs too.
Which is exactly why the problem is quadratic: when a port joins a
bridge, for each port in the tree that's already in that same bridge we
notify a tag_8021q VLAN addition of that port's RX VLAN to the entire
tree. It is a very complicated web of VLANs.
It must be mentioned that currently we install tag_8021q VLANs on too
many ports (DSA links - to be precise, on all of them). For example,
when sw2p0 joins br0, and assuming sw1p0 was part of br0 too, we add the
RX VLAN of sw2p0 on the DSA links of switch 0 too, even though there
isn't any port of switch 0 that is a member of br0 (at least yet).
In theory we could notify only the switches which sit in between the
port joining the bridge and the port reacting to that bridge_join event.
But in practice that is impossible, because of the way 'link' properties
are described in the device tree. The DSA bindings require DT writers to
list out not only the real/physical DSA links, but in fact the entire
routing table, like for example switch 0 above will have:
sw0p3: port@3 {
link = <&sw1p4 &sw2p4>;
};
This was done because:
/* TODO: ideally DSA ports would have a single dp->link_dp member,
* and no dst->rtable nor this struct dsa_link would be needed,
* but this would require some more complex tree walking,
* so keep it stupid at the moment and list them all.
*/
but it is a perfect example of a situation where too much information is
actively detrimential, because we are now in the position where we
cannot distinguish a real DSA link from one that is put there to avoid
the 'complex tree walking'. And because DT is ABI, there is not much we
can change.
And because we do not know which DSA links are real and which ones
aren't, we can't really know if DSA switch A is in the data path between
switches B and C, in the general case.
So this is why tag_8021q RX VLANs are added on all DSA links, and
probably why it will never change.
On the other hand, at least the number of additions/deletions is well
balanced, and this means that once we implement reference counting at
the cross-chip notifier level a la fdb/mdb, there is absolutely zero
need for a struct dsa_8021q_crosschip_link, it's all self-managing.
In fact, with the tag_8021q notifiers emitted from the bridge join
notifiers, it becomes so generic that sja1105 does not need to do
anything anymore, we can just delete its implementation of the
.crosschip_bridge_{join,leave} methods.
Among other things we can simply delete is the home-grown implementation
of sja1105_notify_crosschip_switches(). The reason why that is wrong is
because it is not quadratic - it only covers remote switches to which we
have a cross-chip bridging link and that does not cover in-between
switches. This deletion is part of the same patch because sja1105 used
to poke deep inside the guts of the tag_8021q context in order to do
that. Because the cross-chip links went away, so needs the sja1105 code.
Last but not least, dsa_8021q_setup_port() is simplified (and also
renamed). Because our TAG_8021Q_VLAN_ADD notifier is designed to react
on the CPU port too, the four dsa_8021q_vid_apply() calls:
- 1 for RX VLAN on user port
- 1 for the user port's RX VLAN on the CPU port
- 1 for TX VLAN on user port
- 1 for the user port's TX VLAN on the CPU port
now get squashed into only 2 notifier calls via
dsa_port_tag_8021q_vlan_add.
And because the notifiers to add and to delete a tag_8021q VLAN are
distinct, now we finally break up the port setup and teardown into
separate functions instead of relying on a "bool enabled" flag which
tells us what to do. Arguably it should have been this way from the
get go.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-20 01:14:52 +08:00
|
|
|
if (err)
|
2021-08-11 21:46:05 +08:00
|
|
|
dev_err(dp->ds->dev,
|
|
|
|
|
"port %d failed to notify tag_8021q VLAN %d deletion: %pe\n",
|
|
|
|
|
dp->index, vid, ERR_PTR(err));
|
net: dsa: tag_8021q: add proper cross-chip notifier support
The big problem which mandates cross-chip notifiers for tag_8021q is
this:
|
sw0p0 sw0p1 sw0p2 sw0p3 sw0p4
[ user ] [ user ] [ user ] [ dsa ] [ cpu ]
|
+---------+
|
sw1p0 sw1p1 sw1p2 sw1p3 sw1p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
|
+---------+
|
sw2p0 sw2p1 sw2p2 sw2p3 sw2p4
[ user ] [ user ] [ user ] [ dsa ] [ dsa ]
When the user runs:
ip link add br0 type bridge
ip link set sw0p0 master br0
ip link set sw2p0 master br0
It doesn't work.
This is because dsa_8021q_crosschip_bridge_join() assumes that "ds" and
"other_ds" are at most 1 hop away from each other, so it is sufficient
to add the RX VLAN of {ds, port} into {other_ds, other_port} and vice
versa and presto, the cross-chip link works. When there is another
switch in the middle, such as in this case switch 1 with its DSA links
sw1p3 and sw1p4, somebody needs to tell it about these VLANs too.
Which is exactly why the problem is quadratic: when a port joins a
bridge, for each port in the tree that's already in that same bridge we
notify a tag_8021q VLAN addition of that port's RX VLAN to the entire
tree. It is a very complicated web of VLANs.
It must be mentioned that currently we install tag_8021q VLANs on too
many ports (DSA links - to be precise, on all of them). For example,
when sw2p0 joins br0, and assuming sw1p0 was part of br0 too, we add the
RX VLAN of sw2p0 on the DSA links of switch 0 too, even though there
isn't any port of switch 0 that is a member of br0 (at least yet).
In theory we could notify only the switches which sit in between the
port joining the bridge and the port reacting to that bridge_join event.
But in practice that is impossible, because of the way 'link' properties
are described in the device tree. The DSA bindings require DT writers to
list out not only the real/physical DSA links, but in fact the entire
routing table, like for example switch 0 above will have:
sw0p3: port@3 {
link = <&sw1p4 &sw2p4>;
};
This was done because:
/* TODO: ideally DSA ports would have a single dp->link_dp member,
* and no dst->rtable nor this struct dsa_link would be needed,
* but this would require some more complex tree walking,
* so keep it stupid at the moment and list them all.
*/
but it is a perfect example of a situation where too much information is
actively detrimential, because we are now in the position where we
cannot distinguish a real DSA link from one that is put there to avoid
the 'complex tree walking'. And because DT is ABI, there is not much we
can change.
And because we do not know which DSA links are real and which ones
aren't, we can't really know if DSA switch A is in the data path between
switches B and C, in the general case.
So this is why tag_8021q RX VLANs are added on all DSA links, and
probably why it will never change.
On the other hand, at least the number of additions/deletions is well
balanced, and this means that once we implement reference counting at
the cross-chip notifier level a la fdb/mdb, there is absolutely zero
need for a struct dsa_8021q_crosschip_link, it's all self-managing.
In fact, with the tag_8021q notifiers emitted from the bridge join
notifiers, it becomes so generic that sja1105 does not need to do
anything anymore, we can just delete its implementation of the
.crosschip_bridge_{join,leave} methods.
Among other things we can simply delete is the home-grown implementation
of sja1105_notify_crosschip_switches(). The reason why that is wrong is
because it is not quadratic - it only covers remote switches to which we
have a cross-chip bridging link and that does not cover in-between
switches. This deletion is part of the same patch because sja1105 used
to poke deep inside the guts of the tag_8021q context in order to do
that. Because the cross-chip links went away, so needs the sja1105 code.
Last but not least, dsa_8021q_setup_port() is simplified (and also
renamed). Because our TAG_8021Q_VLAN_ADD notifier is designed to react
on the CPU port too, the four dsa_8021q_vid_apply() calls:
- 1 for RX VLAN on user port
- 1 for the user port's RX VLAN on the CPU port
- 1 for TX VLAN on user port
- 1 for the user port's TX VLAN on the CPU port
now get squashed into only 2 notifier calls via
dsa_port_tag_8021q_vlan_add.
And because the notifiers to add and to delete a tag_8021q VLAN are
distinct, now we finally break up the port setup and teardown into
separate functions instead of relying on a "bool enabled" flag which
tells us what to do. Arguably it should have been this way from the
get go.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-07-20 01:14:52 +08:00
|
|
|
}
|
