[PATCH net-next 0/2] net: dsa: b53: Remove unconditional CPU VLAN tagging

This patch series removes the unconditional egress tagging of VLANs
added to a b53 switch's CPU port. This had led to adding the
untag_bridge_pvid attribute in the DSA framework and is causing problems
with stacking of switches such as the configuration being used by
Matthew.

Since we use Broadcom tags for all of the supported switches nowadays
the original situation that led to doing this unconditional egress
tagging is no longer warranted.

Florian Fainelli (2):
  net: dsa: b53: Do not force tagging on CPU port VLANs
  net: dsa: Remove bridge PVID untagging

 drivers/net/dsa/b53/b53_common.c |  5 ++-
 include/net/dsa.h                |  8 -----
 net/dsa/dsa.c                    |  9 -----
 net/dsa/dsa_priv.h               | 59 --------------------------------
 4 files changed, 2 insertions(+), 79 deletions(-)

-- 
2.25.1

Commit ca8931948344 ("net: dsa: b53: Keep CPU port as tagged in all
VLANs") forced the CPU port to be always tagged in any VLAN membership.
This was necessary back then because we did not support Broadcom tags
for all configurations so the only way to differentiate tagged and
untagged traffic while DSA_TAG_PROTO_NONE was used was to force the CPU
port into being always tagged.

This is not necessary anymore since 8fab459e69ab ("net: dsa: b53: Enable
Broadcom tags for 531x5/539x families") and we can simply program what
we are being told now, regardless of the port being CPU or user-facing.

Reported-by: Matthew Hagan <redacted>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
 drivers/net/dsa/b53/b53_common.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c
index 3ca6b394dd5f..56e3b42ec28c 100644
--- a/drivers/net/dsa/b53/b53_common.c
+++ b/drivers/net/dsa/b53/b53_common.c

@@ -1477,7 +1477,7 @@ int b53_vlan_add(struct dsa_switch *ds, int port,
 		untagged = true;
 
 	vl->members |= BIT(port);
-	if (untagged && !dsa_is_cpu_port(ds, port))
+	if (untagged)
 		vl->untag |= BIT(port);
 	else
 		vl->untag &= ~BIT(port);

@@ -1514,7 +1514,7 @@ int b53_vlan_del(struct dsa_switch *ds, int port,
 	if (pvid == vlan->vid)
 		pvid = b53_default_pvid(dev);
 
-	if (untagged && !dsa_is_cpu_port(ds, port))
+	if (untagged)
 		vl->untag &= ~(BIT(port));
 
 	b53_set_vlan_entry(dev, vlan->vid, vl);

-- 
2.25.1

Now that the only user of the bridge PVID untagging code is no longer
forcing the PVID untagged VLAN configuration to be PVID egress tagged
for the CPU, we no longer need this support code to pop the VLAN tag
automatically.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
 drivers/net/dsa/b53/b53_common.c |  1 -
 include/net/dsa.h                |  8 -----
 net/dsa/dsa.c                    |  9 -----
 net/dsa/dsa_priv.h               | 59 --------------------------------
 4 files changed, 77 deletions(-)

diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c
index 56e3b42ec28c..dc43dadd6d31 100644
--- a/drivers/net/dsa/b53/b53_common.c
+++ b/drivers/net/dsa/b53/b53_common.c

@@ -2660,7 +2660,6 @@ struct b53_device *b53_switch_alloc(struct device *base,
 	dev->priv = priv;
 	dev->ops = ops;
 	ds->ops = &b53_switch_ops;
-	ds->untag_bridge_pvid = true;
 	dev->vlan_enabled = true;
 	/* Let DSA handle the case were multiple bridges span the same switch
 	 * device and different VLAN awareness settings are requested, which

diff --git a/include/net/dsa.h b/include/net/dsa.h
index e1a2610a0e06..216443820a7e 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h

@@ -355,14 +355,6 @@ struct dsa_switch {
 	 */
 	bool			configure_vlan_while_not_filtering;
 
-	/* If the switch driver always programs the CPU port as egress tagged
-	 * despite the VLAN configuration indicating otherwise, then setting
-	 * @untag_bridge_pvid will force the DSA receive path to pop the bridge's
-	 * default_pvid VLAN tagged frames to offer a consistent behavior
-	 * between a vlan_filtering=0 and vlan_filtering=1 bridge device.
-	 */
-	bool			untag_bridge_pvid;
-
 	/* Let DSA manage the FDB entries towards the CPU, based on the
 	 * software bridge database.
 	 */

diff --git a/net/dsa/dsa.c b/net/dsa/dsa.c
index 84cad1be9ce4..daac329b6e93 100644
--- a/net/dsa/dsa.c
+++ b/net/dsa/dsa.c

@@ -260,15 +260,6 @@ static int dsa_switch_rcv(struct sk_buff *skb, struct net_device *dev,
 
 	p = netdev_priv(skb->dev);
 
-	if (unlikely(cpu_dp->ds->untag_bridge_pvid)) {
-		nskb = dsa_untag_bridge_pvid(skb);
-		if (!nskb) {
-			kfree_skb(skb);
-			return 0;
-		}
-		skb = nskb;
-	}
-
 	dev_sw_netstats_rx_add(skb->dev, skb->len);
 
 	if (dsa_skb_defer_rx_timestamp(p, skb))

diff --git a/net/dsa/dsa_priv.h b/net/dsa/dsa_priv.h
index 92282de54230..08d915d951b0 100644
--- a/net/dsa/dsa_priv.h
+++ b/net/dsa/dsa_priv.h

@@ -290,65 +290,6 @@ dsa_slave_to_master(const struct net_device *dev)
 	return dp->cpu_dp->master;
 }
 
-/* If under a bridge with vlan_filtering=0, make sure to send pvid-tagged
- * frames as untagged, since the bridge will not untag them.
- */
-static inline struct sk_buff *dsa_untag_bridge_pvid(struct sk_buff *skb)
-{
-	struct dsa_port *dp = dsa_slave_to_port(skb->dev);
-	struct net_device *br = dp->bridge_dev;
-	struct net_device *dev = skb->dev;
-	struct net_device *upper_dev;
-	u16 vid, pvid, proto;
-	int err;
-
-	if (!br || br_vlan_enabled(br))
-		return skb;
-
-	err = br_vlan_get_proto(br, &proto);
-	if (err)
-		return skb;
-
-	/* Move VLAN tag from data to hwaccel */
-	if (!skb_vlan_tag_present(skb) && skb->protocol == htons(proto)) {
-		skb = skb_vlan_untag(skb);
-		if (!skb)
-			return NULL;
-	}
-
-	if (!skb_vlan_tag_present(skb))
-		return skb;
-
-	vid = skb_vlan_tag_get_id(skb);
-
-	/* We already run under an RCU read-side critical section since
-	 * we are called from netif_receive_skb_list_internal().
-	 */
-	err = br_vlan_get_pvid_rcu(dev, &pvid);
-	if (err)
-		return skb;
-
-	if (vid != pvid)
-		return skb;
-
-	/* The sad part about attempting to untag from DSA is that we
-	 * don't know, unless we check, if the skb will end up in
-	 * the bridge's data path - br_allowed_ingress() - or not.
-	 * For example, there might be an 8021q upper for the
-	 * default_pvid of the bridge, which will steal VLAN-tagged traffic
-	 * from the bridge's data path. This is a configuration that DSA
-	 * supports because vlan_filtering is 0. In that case, we should
-	 * definitely keep the tag, to make sure it keeps working.
-	 */
-	upper_dev = __vlan_find_dev_deep_rcu(br, htons(proto), vid);
-	if (upper_dev)
-		return skb;
-
-	__vlan_hwaccel_clear_tag(skb);
-
-	return skb;
-}
-
 /* switch.c */
 int dsa_switch_register_notifier(struct dsa_switch *ds);
 void dsa_switch_unregister_notifier(struct dsa_switch *ds);

-- 
2.25.1

On Mon, Jun 07, 2021 at 03:08:42PM -0700, Florian Fainelli wrote:

Don't you want to keep this functionality for BCM5325 / BCM5365 and
such, which still use DSA_TAG_PROTO_NONE?

On 6/7/2021 3:22 PM, Vladimir Oltean wrote:

On Mon, Jun 07, 2021 at 03:08:42PM -0700, Florian Fainelli wrote:

quoted

Commit ca8931948344 ("net: dsa: b53: Keep CPU port as tagged in all
VLANs") forced the CPU port to be always tagged in any VLAN membership.
This was necessary back then because we did not support Broadcom tags
for all configurations so the only way to differentiate tagged and
untagged traffic while DSA_TAG_PROTO_NONE was used was to force the CPU
port into being always tagged.

This is not necessary anymore since 8fab459e69ab ("net: dsa: b53: Enable
Broadcom tags for 531x5/539x families") and we can simply program what
we are being told now, regardless of the port being CPU or user-facing.

Reported-by: Matthew Hagan <redacted>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
 drivers/net/dsa/b53/b53_common.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c
index 3ca6b394dd5f..56e3b42ec28c 100644
--- a/drivers/net/dsa/b53/b53_common.c
+++ b/drivers/net/dsa/b53/b53_common.c

@@ -1477,7 +1477,7 @@ int b53_vlan_add(struct dsa_switch *ds, int port,
 		untagged = true;
 
 	vl->members |= BIT(port);
-	if (untagged && !dsa_is_cpu_port(ds, port))
+	if (untagged)
 		vl->untag |= BIT(port);
 	else
 		vl->untag &= ~BIT(port);

@@ -1514,7 +1514,7 @@ int b53_vlan_del(struct dsa_switch *ds, int port,
 	if (pvid == vlan->vid)
 		pvid = b53_default_pvid(dev);
 
-	if (untagged && !dsa_is_cpu_port(ds, port))
+	if (untagged)
 		vl->untag &= ~(BIT(port));
 
 	b53_set_vlan_entry(dev, vlan->vid, vl);

-- 
2.25.1

Don't you want to keep this functionality for BCM5325 / BCM5365 and
such, which still use DSA_TAG_PROTO_NONE?

Humm, in premise yes, however I am debating removing support for
5325/5365 entirely, nobody that I know of has even been trying to get
those devices to work with that driver.
-- 
Florian

On 6/7/2021 3:31 PM, Florian Fainelli wrote:

On 6/7/2021 3:22 PM, Vladimir Oltean wrote:

quoted

On Mon, Jun 07, 2021 at 03:08:42PM -0700, Florian Fainelli wrote:

quoted

Commit ca8931948344 ("net: dsa: b53: Keep CPU port as tagged in all
VLANs") forced the CPU port to be always tagged in any VLAN membership.
This was necessary back then because we did not support Broadcom tags
for all configurations so the only way to differentiate tagged and
untagged traffic while DSA_TAG_PROTO_NONE was used was to force the CPU
port into being always tagged.

This is not necessary anymore since 8fab459e69ab ("net: dsa: b53: Enable
Broadcom tags for 531x5/539x families") and we can simply program what
we are being told now, regardless of the port being CPU or user-facing.

Reported-by: Matthew Hagan <redacted>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
 drivers/net/dsa/b53/b53_common.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c
index 3ca6b394dd5f..56e3b42ec28c 100644
--- a/drivers/net/dsa/b53/b53_common.c
+++ b/drivers/net/dsa/b53/b53_common.c

@@ -1477,7 +1477,7 @@ int b53_vlan_add(struct dsa_switch *ds, int port,
 		untagged = true;
 
 	vl->members |= BIT(port);
-	if (untagged && !dsa_is_cpu_port(ds, port))
+	if (untagged)
 		vl->untag |= BIT(port);
 	else
 		vl->untag &= ~BIT(port);

@@ -1514,7 +1514,7 @@ int b53_vlan_del(struct dsa_switch *ds, int port,
 	if (pvid == vlan->vid)
 		pvid = b53_default_pvid(dev);
 
-	if (untagged && !dsa_is_cpu_port(ds, port))
+	if (untagged)
 		vl->untag &= ~(BIT(port));
 
 	b53_set_vlan_entry(dev, vlan->vid, vl);

-- 
2.25.1

Don't you want to keep this functionality for BCM5325 / BCM5365 and
such, which still use DSA_TAG_PROTO_NONE?

Humm, in premise yes, however I am debating removing support for
5325/5365 entirely, nobody that I know of has even been trying to get
those devices to work with that driver.

On second thought, we just need to have those devices return
DSA_TAG_PROTO_BRCM_LEGACY which is what they use, there does appear to
be a couple of users, including myself, I had not realized that the
device I was using used a 5325 (thought it was a 53125). V2 coming.
-- 
Florian