[PATCH net-next v7 00/12] net: vxlan: add skb drop reasons support

[PATCH net-next v7 00/12] net: vxlan: add skb drop reasons support

[Menglong Dong]

In this series, we add skb drop reasons support to VXLAN, and following
new skb drop reasons are introduced:

  SKB_DROP_REASON_VXLAN_INVALID_HDR
  SKB_DROP_REASON_VXLAN_VNI_NOT_FOUND
  SKB_DROP_REASON_VXLAN_ENTRY_EXISTS
  SKB_DROP_REASON_VXLAN_NO_REMOTE
  SKB_DROP_REASON_MAC_INVALID_SOURCE
  SKB_DROP_REASON_IP_TUNNEL_ECN
  SKB_DROP_REASON_TUNNEL_TXINFO
  SKB_DROP_REASON_LOCAL_MAC

We add some helper functions in this series, who will capture the drop
reasons from pskb_may_pull_reason and return them:

  pskb_network_may_pull_reason()
  pskb_inet_may_pull_reason()

And we also make the following functions return skb drop reasons:

  skb_vlan_inet_prepare()
  vxlan_remcsum()
  vxlan_snoop()
  vxlan_set_mac()

Changes since v6:
- fix some typos in the document for SKB_DROP_REASON_TUNNEL_TXINFO

Changes since v5:
- fix some typos in the document for SKB_DROP_REASON_TUNNEL_TXINFO

Changes since v4:
- make skb_vlan_inet_prepare() return drop reasons, instead of introduce
  a wrapper for it in the 3rd patch.
- modify the document for SKB_DROP_REASON_LOCAL_MAC and
  SKB_DROP_REASON_TUNNEL_TXINFO.

Changes since v3:
- rename SKB_DROP_REASON_VXLAN_INVALID_SMAC to
  SKB_DROP_REASON_MAC_INVALID_SOURCE in the 6th patch

Changes since v2:
- move all the drop reasons of VXLAN to the "core", instead of introducing
  the VXLAN drop reason subsystem
- add the 6th patch, which capture the drop reasons from vxlan_snoop()
- move the commits for vxlan_remcsum() and vxlan_set_mac() after
  vxlan_rcv() to update the call of them accordingly
- fix some format problems

Changes since v1:
- document all the drop reasons that we introduce
- rename the drop reasons to make them more descriptive, as Ido advised
- remove the 2nd patch, which introduce the SKB_DR_RESET
- add the 4th patch, which adds skb_vlan_inet_prepare_reason() helper
- introduce the 6th patch, which make vxlan_set_mac return drop reasons
- introduce the 10th patch, which uses VXLAN_DROP_NO_REMOTE as the drop
  reasons, as Ido advised

Menglong Dong (12):
  net: skb: add pskb_network_may_pull_reason() helper
  net: tunnel: add pskb_inet_may_pull_reason() helper
  net: tunnel: make skb_vlan_inet_prepare() return drop reasons
  net: vxlan: add skb drop reasons to vxlan_rcv()
  net: vxlan: make vxlan_remcsum() return drop reasons
  net: vxlan: make vxlan_snoop() return drop reasons
  net: vxlan: make vxlan_set_mac() return drop reasons
  net: vxlan: use kfree_skb_reason() in vxlan_xmit()
  net: vxlan: add drop reasons support to vxlan_xmit_one()
  net: vxlan: use kfree_skb_reason() in vxlan_mdb_xmit()
  net: vxlan: use kfree_skb_reason() in vxlan_encap_bypass()
  net: vxlan: use kfree_skb_reason() in encap_bypass_if_local()

 drivers/net/bareudp.c          |   4 +-
 drivers/net/geneve.c           |   4 +-
 drivers/net/vxlan/vxlan_core.c | 111 +++++++++++++++++++++------------
 drivers/net/vxlan/vxlan_mdb.c  |   2 +-
 include/linux/skbuff.h         |   8 ++-
 include/net/dropreason-core.h  |  40 ++++++++++++
 include/net/ip_tunnels.h       |  23 ++++---
 7 files changed, 139 insertions(+), 53 deletions(-)

-- 
2.39.5

[PATCH net-next v7 01/12] net: skb: add pskb_network_may_pull_reason() helper

[Menglong Dong]

Introduce the function pskb_network_may_pull_reason() and make
pskb_network_may_pull() a simple inline call to it. The drop reasons of
it just come from pskb_may_pull_reason.

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
---
 include/linux/skbuff.h | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 39f1d16f3628..48f1e0fa2a13 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -3130,9 +3130,15 @@ static inline int skb_inner_network_offset(const struct sk_buff *skb)
 	return skb_inner_network_header(skb) - skb->data;
 }
 
+static inline enum skb_drop_reason
+pskb_network_may_pull_reason(struct sk_buff *skb, unsigned int len)
+{
+	return pskb_may_pull_reason(skb, skb_network_offset(skb) + len);
+}
+
 static inline int pskb_network_may_pull(struct sk_buff *skb, unsigned int len)
 {
-	return pskb_may_pull(skb, skb_network_offset(skb) + len);
+	return pskb_network_may_pull_reason(skb, len) == SKB_NOT_DROPPED_YET;
 }
 
 /*
-- 
2.39.5

[PATCH net-next v7 02/12] net: tunnel: add pskb_inet_may_pull_reason() helper

[Menglong Dong]

Introduce the function pskb_inet_may_pull_reason() and make
pskb_inet_may_pull a simple inline call to it. The drop reasons of it just
come from pskb_may_pull_reason().

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
---
 include/net/ip_tunnels.h | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/include/net/ip_tunnels.h b/include/net/ip_tunnels.h
index 6194fbb564c6..7fc2f7bf837a 100644
--- a/include/net/ip_tunnels.h
+++ b/include/net/ip_tunnels.h
@@ -439,7 +439,8 @@ int ip_tunnel_encap_del_ops(const struct ip_tunnel_encap_ops *op,
 int ip_tunnel_encap_setup(struct ip_tunnel *t,
 			  struct ip_tunnel_encap *ipencap);
 
-static inline bool pskb_inet_may_pull(struct sk_buff *skb)
+static inline enum skb_drop_reason
+pskb_inet_may_pull_reason(struct sk_buff *skb)
 {
 	int nhlen;
 
@@ -456,7 +457,12 @@ static inline bool pskb_inet_may_pull(struct sk_buff *skb)
 		nhlen = 0;
 	}
 
-	return pskb_network_may_pull(skb, nhlen);
+	return pskb_network_may_pull_reason(skb, nhlen);
+}
+
+static inline bool pskb_inet_may_pull(struct sk_buff *skb)
+{
+	return pskb_inet_may_pull_reason(skb) == SKB_NOT_DROPPED_YET;
 }
 
 /* Variant of pskb_inet_may_pull().
-- 
2.39.5

[PATCH net-next v7 03/12] net: tunnel: make skb_vlan_inet_prepare() return drop reasons

[Menglong Dong]

Make skb_vlan_inet_prepare return the skb drop reasons, which is just
what pskb_may_pull_reason() returns. Meanwhile, adjust all the call of
it.

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
---
v5:
- make skb_vlan_inet_prepare() return drop reasons, instead of introduce
  a wrapper for it.
v3:
- fix some format problems,  as Alexander advised
---
 drivers/net/bareudp.c          |  4 ++--
 drivers/net/geneve.c           |  4 ++--
 drivers/net/vxlan/vxlan_core.c |  2 +-
 include/net/ip_tunnels.h       | 13 ++++++++-----
 4 files changed, 13 insertions(+), 10 deletions(-)

diff --git a/drivers/net/bareudp.c b/drivers/net/bareudp.c
index e057526448d7..fa2dd76ba3d9 100644
--- a/drivers/net/bareudp.c
+++ b/drivers/net/bareudp.c
@@ -317,7 +317,7 @@ static int bareudp_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 	__be32 saddr;
 	int err;
 
-	if (!skb_vlan_inet_prepare(skb, skb->protocol != htons(ETH_P_TEB)))
+	if (skb_vlan_inet_prepare(skb, skb->protocol != htons(ETH_P_TEB)))
 		return -EINVAL;
 
 	if (!sock)
@@ -387,7 +387,7 @@ static int bareudp6_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 	__be16 sport;
 	int err;
 
-	if (!skb_vlan_inet_prepare(skb, skb->protocol != htons(ETH_P_TEB)))
+	if (skb_vlan_inet_prepare(skb, skb->protocol != htons(ETH_P_TEB)))
 		return -EINVAL;
 
 	if (!sock)
diff --git a/drivers/net/geneve.c b/drivers/net/geneve.c
index 7f611c74eb62..2f29b1386b1c 100644
--- a/drivers/net/geneve.c
+++ b/drivers/net/geneve.c
@@ -827,7 +827,7 @@ static int geneve_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 	__be16 sport;
 	int err;
 
-	if (!skb_vlan_inet_prepare(skb, inner_proto_inherit))
+	if (skb_vlan_inet_prepare(skb, inner_proto_inherit))
 		return -EINVAL;
 
 	if (!gs4)
@@ -937,7 +937,7 @@ static int geneve6_xmit_skb(struct sk_buff *skb, struct net_device *dev,
 	__be16 sport;
 	int err;
 
-	if (!skb_vlan_inet_prepare(skb, inner_proto_inherit))
+	if (skb_vlan_inet_prepare(skb, inner_proto_inherit))
 		return -EINVAL;
 
 	if (!gs6)
diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index 53dcb9fffc04..0359c750d81e 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -2356,7 +2356,7 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 	__be32 vni = 0;
 
 	no_eth_encap = flags & VXLAN_F_GPE && skb->protocol != htons(ETH_P_TEB);
-	if (!skb_vlan_inet_prepare(skb, no_eth_encap))
+	if (skb_vlan_inet_prepare(skb, no_eth_encap))
 		goto drop;
 
 	old_iph = ip_hdr(skb);
diff --git a/include/net/ip_tunnels.h b/include/net/ip_tunnels.h
index 7fc2f7bf837a..4e4f9e24c9c1 100644
--- a/include/net/ip_tunnels.h
+++ b/include/net/ip_tunnels.h
@@ -467,11 +467,12 @@ static inline bool pskb_inet_may_pull(struct sk_buff *skb)
 
 /* Variant of pskb_inet_may_pull().
  */
-static inline bool skb_vlan_inet_prepare(struct sk_buff *skb,
-					 bool inner_proto_inherit)
+static inline enum skb_drop_reason
+skb_vlan_inet_prepare(struct sk_buff *skb, bool inner_proto_inherit)
 {
 	int nhlen = 0, maclen = inner_proto_inherit ? 0 : ETH_HLEN;
 	__be16 type = skb->protocol;
+	enum skb_drop_reason reason;
 
 	/* Essentially this is skb_protocol(skb, true)
 	 * And we get MAC len.
@@ -492,11 +493,13 @@ static inline bool skb_vlan_inet_prepare(struct sk_buff *skb,
 	/* For ETH_P_IPV6/ETH_P_IP we make sure to pull
 	 * a base network header in skb->head.
 	 */
-	if (!pskb_may_pull(skb, maclen + nhlen))
-		return false;
+	reason = pskb_may_pull_reason(skb, maclen + nhlen);
+	if (reason)
+		return reason;
 
 	skb_set_network_header(skb, maclen);
-	return true;
+
+	return SKB_NOT_DROPPED_YET;
 }
 
 static inline int ip_encap_hlen(struct ip_tunnel_encap *e)
-- 
2.39.5

Re: [PATCH net-next v7 03/12] net: tunnel: make skb_vlan_inet_prepare() return drop reasons

[Simon Horman]

On Wed, Oct 09, 2024 at 10:28:21AM +0800, Menglong Dong wrote:
> Make skb_vlan_inet_prepare return the skb drop reasons, which is just
> what pskb_may_pull_reason() returns. Meanwhile, adjust all the call of
> it.
> 
> Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> ---
> v5:
> - make skb_vlan_inet_prepare() return drop reasons, instead of introduce
>   a wrapper for it.
> v3:
> - fix some format problems,  as Alexander advised

Reviewed-by: Simon Horman <horms@kernel.org>

[PATCH net-next v7 04/12] net: vxlan: add skb drop reasons to vxlan_rcv()

[Menglong Dong]

Introduce skb drop reasons to the function vxlan_rcv(). Following new
drop reasons are added:

  SKB_DROP_REASON_VXLAN_INVALID_HDR
  SKB_DROP_REASON_VXLAN_VNI_NOT_FOUND
  SKB_DROP_REASON_IP_TUNNEL_ECN

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
---
v3:
- modify vxlan_set_mac() and vxlan_remcsum() after this patch
v2:
- rename the drop reasons, as Ido advised.
- document the drop reasons
---
 drivers/net/vxlan/vxlan_core.c | 26 ++++++++++++++++++++------
 include/net/dropreason-core.h  | 16 ++++++++++++++++
 2 files changed, 36 insertions(+), 6 deletions(-)

diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index 0359c750d81e..4997a2c09c14 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -1671,13 +1671,15 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 	struct vxlan_metadata _md;
 	struct vxlan_metadata *md = &_md;
 	__be16 protocol = htons(ETH_P_TEB);
+	enum skb_drop_reason reason;
 	bool raw_proto = false;
 	void *oiph;
 	__be32 vni = 0;
 	int nh;
 
 	/* Need UDP and VXLAN header to be present */
-	if (!pskb_may_pull(skb, VXLAN_HLEN))
+	reason = pskb_may_pull_reason(skb, VXLAN_HLEN);
+	if (reason)
 		goto drop;
 
 	unparsed = *vxlan_hdr(skb);
@@ -1686,6 +1688,7 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 		netdev_dbg(skb->dev, "invalid vxlan flags=%#x vni=%#x\n",
 			   ntohl(vxlan_hdr(skb)->vx_flags),
 			   ntohl(vxlan_hdr(skb)->vx_vni));
+		reason = SKB_DROP_REASON_VXLAN_INVALID_HDR;
 		/* Return non vxlan pkt */
 		goto drop;
 	}
@@ -1699,8 +1702,10 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 	vni = vxlan_vni(vxlan_hdr(skb)->vx_vni);
 
 	vxlan = vxlan_vs_find_vni(vs, skb->dev->ifindex, vni, &vninode);
-	if (!vxlan)
+	if (!vxlan) {
+		reason = SKB_DROP_REASON_VXLAN_VNI_NOT_FOUND;
 		goto drop;
+	}
 
 	/* For backwards compatibility, only allow reserved fields to be
 	 * used by VXLAN extensions if explicitly requested.
@@ -1713,8 +1718,10 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 	}
 
 	if (__iptunnel_pull_header(skb, VXLAN_HLEN, protocol, raw_proto,
-				   !net_eq(vxlan->net, dev_net(vxlan->dev))))
+				   !net_eq(vxlan->net, dev_net(vxlan->dev)))) {
+		reason = SKB_DROP_REASON_NOMEM;
 		goto drop;
+	}
 
 	if (vs->flags & VXLAN_F_REMCSUM_RX)
 		if (unlikely(!vxlan_remcsum(&unparsed, skb, vs->flags)))
@@ -1728,8 +1735,10 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 		tun_dst = udp_tun_rx_dst(skb, vxlan_get_sk_family(vs), flags,
 					 key32_to_tunnel_id(vni), sizeof(*md));
 
-		if (!tun_dst)
+		if (!tun_dst) {
+			reason = SKB_DROP_REASON_NOMEM;
 			goto drop;
+		}
 
 		md = ip_tunnel_info_opts(&tun_dst->u.tun_info);
 
@@ -1753,6 +1762,7 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 		 * is more robust and provides a little more security in
 		 * adding extensions to VXLAN.
 		 */
+		reason = SKB_DROP_REASON_VXLAN_INVALID_HDR;
 		goto drop;
 	}
 
@@ -1773,7 +1783,8 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 
 	skb_reset_network_header(skb);
 
-	if (!pskb_inet_may_pull(skb)) {
+	reason = pskb_inet_may_pull_reason(skb);
+	if (reason) {
 		DEV_STATS_INC(vxlan->dev, rx_length_errors);
 		DEV_STATS_INC(vxlan->dev, rx_errors);
 		vxlan_vnifilter_count(vxlan, vni, vninode,
@@ -1785,6 +1796,7 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 	oiph = skb->head + nh;
 
 	if (!vxlan_ecn_decapsulate(vs, oiph, skb)) {
+		reason = SKB_DROP_REASON_IP_TUNNEL_ECN;
 		DEV_STATS_INC(vxlan->dev, rx_frame_errors);
 		DEV_STATS_INC(vxlan->dev, rx_errors);
 		vxlan_vnifilter_count(vxlan, vni, vninode,
@@ -1799,6 +1811,7 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 		dev_core_stats_rx_dropped_inc(vxlan->dev);
 		vxlan_vnifilter_count(vxlan, vni, vninode,
 				      VXLAN_VNI_STATS_RX_DROPS, 0);
+		reason = SKB_DROP_REASON_DEV_READY;
 		goto drop;
 	}
 
@@ -1811,8 +1824,9 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 	return 0;
 
 drop:
+	reason = reason ?: SKB_DROP_REASON_NOT_SPECIFIED;
 	/* Consume bad packet */
-	kfree_skb(skb);
+	kfree_skb_reason(skb, reason);
 	return 0;
 }
 
diff --git a/include/net/dropreason-core.h b/include/net/dropreason-core.h
index 4748680e8c88..98259d2b3e92 100644
--- a/include/net/dropreason-core.h
+++ b/include/net/dropreason-core.h
@@ -92,6 +92,9 @@
 	FN(PACKET_SOCK_ERROR)		\
 	FN(TC_CHAIN_NOTFOUND)		\
 	FN(TC_RECLASSIFY_LOOP)		\
+	FN(VXLAN_INVALID_HDR)		\
+	FN(VXLAN_VNI_NOT_FOUND)		\
+	FN(IP_TUNNEL_ECN)		\
 	FNe(MAX)
 
 /**
@@ -418,6 +421,19 @@ enum skb_drop_reason {
 	 * iterations.
 	 */
 	SKB_DROP_REASON_TC_RECLASSIFY_LOOP,
+	/**
+	 * @SKB_DROP_REASON_VXLAN_INVALID_HDR: VXLAN header is invalid. E.g.:
+	 * 1) reserved fields are not zero
+	 * 2) "I" flag is not set
+	 */
+	SKB_DROP_REASON_VXLAN_INVALID_HDR,
+	/** @SKB_DROP_REASON_VXLAN_VNI_NOT_FOUND: no VXLAN device found for VNI */
+	SKB_DROP_REASON_VXLAN_VNI_NOT_FOUND,
+	/**
+	 * @SKB_DROP_REASON_IP_TUNNEL_ECN: skb is dropped according to
+	 * RFC 6040 4.2, see __INET_ECN_decapsulate() for detail.
+	 */
+	SKB_DROP_REASON_IP_TUNNEL_ECN,
 	/**
 	 * @SKB_DROP_REASON_MAX: the maximum of core drop reasons, which
 	 * shouldn't be used as a real 'reason' - only for tracing code gen
-- 
2.39.5

Re: [PATCH net-next v7 04/12] net: vxlan: add skb drop reasons to vxlan_rcv()

[Ido Schimmel]

On Wed, Oct 09, 2024 at 10:28:22AM +0800, Menglong Dong wrote:
> Introduce skb drop reasons to the function vxlan_rcv(). Following new
> drop reasons are added:
> 
>   SKB_DROP_REASON_VXLAN_INVALID_HDR
>   SKB_DROP_REASON_VXLAN_VNI_NOT_FOUND
>   SKB_DROP_REASON_IP_TUNNEL_ECN
> 
> Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> Reviewed-by: Simon Horman <horms@kernel.org>

Reviewed-by: Ido Schimmel <idosch@nvidia.com>

[PATCH net-next v7 05/12] net: vxlan: make vxlan_remcsum() return drop reasons

[Menglong Dong]

Make vxlan_remcsum() support skb drop reasons by changing the return
value type of it from bool to enum skb_drop_reason.

The only drop reason in vxlan_remcsum() comes from pskb_may_pull_reason(),
so we just return it.

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
---
v3:
- add a empty newline before return, as Alexander advised
- adjust the call of vxlan_remcsum()
---
 drivers/net/vxlan/vxlan_core.c | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index 4997a2c09c14..34b44755f663 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -1551,9 +1551,11 @@ static void vxlan_sock_release(struct vxlan_dev *vxlan)
 #endif
 }
 
-static bool vxlan_remcsum(struct vxlanhdr *unparsed,
-			  struct sk_buff *skb, u32 vxflags)
+static enum skb_drop_reason vxlan_remcsum(struct vxlanhdr *unparsed,
+					  struct sk_buff *skb,
+					  u32 vxflags)
 {
+	enum skb_drop_reason reason;
 	size_t start, offset;
 
 	if (!(unparsed->vx_flags & VXLAN_HF_RCO) || skb->remcsum_offload)
@@ -1562,15 +1564,17 @@ static bool vxlan_remcsum(struct vxlanhdr *unparsed,
 	start = vxlan_rco_start(unparsed->vx_vni);
 	offset = start + vxlan_rco_offset(unparsed->vx_vni);
 
-	if (!pskb_may_pull(skb, offset + sizeof(u16)))
-		return false;
+	reason = pskb_may_pull_reason(skb, offset + sizeof(u16));
+	if (reason)
+		return reason;
 
 	skb_remcsum_process(skb, (void *)(vxlan_hdr(skb) + 1), start, offset,
 			    !!(vxflags & VXLAN_F_REMCSUM_NOPARTIAL));
 out:
 	unparsed->vx_flags &= ~VXLAN_HF_RCO;
 	unparsed->vx_vni &= VXLAN_VNI_MASK;
-	return true;
+
+	return SKB_NOT_DROPPED_YET;
 }
 
 static void vxlan_parse_gbp_hdr(struct vxlanhdr *unparsed,
@@ -1723,9 +1727,11 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 		goto drop;
 	}
 
-	if (vs->flags & VXLAN_F_REMCSUM_RX)
-		if (unlikely(!vxlan_remcsum(&unparsed, skb, vs->flags)))
+	if (vs->flags & VXLAN_F_REMCSUM_RX) {
+		reason = vxlan_remcsum(&unparsed, skb, vs->flags);
+		if (unlikely(reason))
 			goto drop;
+	}
 
 	if (vxlan_collect_metadata(vs)) {
 		IP_TUNNEL_DECLARE_FLAGS(flags) = { };
-- 
2.39.5

Re: [PATCH net-next v7 05/12] net: vxlan: make vxlan_remcsum() return drop reasons

[Ido Schimmel]

On Wed, Oct 09, 2024 at 10:28:23AM +0800, Menglong Dong wrote:
> Make vxlan_remcsum() support skb drop reasons by changing the return
> value type of it from bool to enum skb_drop_reason.
> 
> The only drop reason in vxlan_remcsum() comes from pskb_may_pull_reason(),
> so we just return it.
> 
> Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> Reviewed-by: Simon Horman <horms@kernel.org>
> Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>

Reviewed-by: Ido Schimmel <idosch@nvidia.com>

[PATCH net-next v7 06/12] net: vxlan: make vxlan_snoop() return drop reasons

[Menglong Dong]

Change the return type of vxlan_snoop() from bool to enum
skb_drop_reason. In this commit, two drop reasons are introduced:

  SKB_DROP_REASON_MAC_INVALID_SOURCE
  SKB_DROP_REASON_VXLAN_ENTRY_EXISTS

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
---
v5:
- rename SKB_DROP_REASON_VXLAN_INVALID_SMAC to
  SKB_DROP_REASON_MAC_INVALID_SOURCE in the commit log
v4:
- rename SKB_DROP_REASON_VXLAN_INVALID_SMAC to
  SKB_DROP_REASON_MAC_INVALID_SOURCE
---
 drivers/net/vxlan/vxlan_core.c | 17 +++++++++--------
 include/net/dropreason-core.h  |  9 +++++++++
 2 files changed, 18 insertions(+), 8 deletions(-)

diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index 34b44755f663..1a81a3957327 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -1437,9 +1437,10 @@ static int vxlan_fdb_get(struct sk_buff *skb,
  * and Tunnel endpoint.
  * Return true if packet is bogus and should be dropped.
  */
-static bool vxlan_snoop(struct net_device *dev,
-			union vxlan_addr *src_ip, const u8 *src_mac,
-			u32 src_ifindex, __be32 vni)
+static enum skb_drop_reason vxlan_snoop(struct net_device *dev,
+					union vxlan_addr *src_ip,
+					const u8 *src_mac, u32 src_ifindex,
+					__be32 vni)
 {
 	struct vxlan_dev *vxlan = netdev_priv(dev);
 	struct vxlan_fdb *f;
@@ -1447,7 +1448,7 @@ static bool vxlan_snoop(struct net_device *dev,
 
 	/* Ignore packets from invalid src-address */
 	if (!is_valid_ether_addr(src_mac))
-		return true;
+		return SKB_DROP_REASON_MAC_INVALID_SOURCE;
 
 #if IS_ENABLED(CONFIG_IPV6)
 	if (src_ip->sa.sa_family == AF_INET6 &&
@@ -1461,15 +1462,15 @@ static bool vxlan_snoop(struct net_device *dev,
 
 		if (likely(vxlan_addr_equal(&rdst->remote_ip, src_ip) &&
 			   rdst->remote_ifindex == ifindex))
-			return false;
+			return SKB_NOT_DROPPED_YET;
 
 		/* Don't migrate static entries, drop packets */
 		if (f->state & (NUD_PERMANENT | NUD_NOARP))
-			return true;
+			return SKB_DROP_REASON_VXLAN_ENTRY_EXISTS;
 
 		/* Don't override an fdb with nexthop with a learnt entry */
 		if (rcu_access_pointer(f->nh))
-			return true;
+			return SKB_DROP_REASON_VXLAN_ENTRY_EXISTS;
 
 		if (net_ratelimit())
 			netdev_info(dev,
@@ -1497,7 +1498,7 @@ static bool vxlan_snoop(struct net_device *dev,
 		spin_unlock(&vxlan->hash_lock[hash_index]);
 	}
 
-	return false;
+	return SKB_NOT_DROPPED_YET;
 }
 
 static bool __vxlan_sock_release_prep(struct vxlan_sock *vs)
diff --git a/include/net/dropreason-core.h b/include/net/dropreason-core.h
index 98259d2b3e92..1cb8d7c953be 100644
--- a/include/net/dropreason-core.h
+++ b/include/net/dropreason-core.h
@@ -94,6 +94,8 @@
 	FN(TC_RECLASSIFY_LOOP)		\
 	FN(VXLAN_INVALID_HDR)		\
 	FN(VXLAN_VNI_NOT_FOUND)		\
+	FN(MAC_INVALID_SOURCE)		\
+	FN(VXLAN_ENTRY_EXISTS)		\
 	FN(IP_TUNNEL_ECN)		\
 	FNe(MAX)
 
@@ -429,6 +431,13 @@ enum skb_drop_reason {
 	SKB_DROP_REASON_VXLAN_INVALID_HDR,
 	/** @SKB_DROP_REASON_VXLAN_VNI_NOT_FOUND: no VXLAN device found for VNI */
 	SKB_DROP_REASON_VXLAN_VNI_NOT_FOUND,
+	/** @SKB_DROP_REASON_MAC_INVALID_SOURCE: source mac is invalid */
+	SKB_DROP_REASON_MAC_INVALID_SOURCE,
+	/**
+	 * @SKB_DROP_REASON_VXLAN_ENTRY_EXISTS: trying to migrate a static
+	 * entry or an entry pointing to a nexthop.
+	 */
+	SKB_DROP_REASON_VXLAN_ENTRY_EXISTS,
 	/**
 	 * @SKB_DROP_REASON_IP_TUNNEL_ECN: skb is dropped according to
 	 * RFC 6040 4.2, see __INET_ECN_decapsulate() for detail.
-- 
2.39.5

Re: [PATCH net-next v7 06/12] net: vxlan: make vxlan_snoop() return drop reasons

[Ido Schimmel]

On Wed, Oct 09, 2024 at 10:28:24AM +0800, Menglong Dong wrote:
> Change the return type of vxlan_snoop() from bool to enum
> skb_drop_reason. In this commit, two drop reasons are introduced:
> 
>   SKB_DROP_REASON_MAC_INVALID_SOURCE
>   SKB_DROP_REASON_VXLAN_ENTRY_EXISTS
> 
> Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> Reviewed-by: Simon Horman <horms@kernel.org>

Reviewed-by: Ido Schimmel <idosch@nvidia.com>

IMO the second reason is quite obscure and unlikely to be very useful,
but time will tell. The closest thing in the bridge driver is 802.1X /
MAB support (see "locked" and "mab" bridge link attributes in "man
bridge"), but I don't think it's close enough to allow us making this
reason more generic.

[...]

> diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
> index 34b44755f663..1a81a3957327 100644
> --- a/drivers/net/vxlan/vxlan_core.c
> +++ b/drivers/net/vxlan/vxlan_core.c
> @@ -1437,9 +1437,10 @@ static int vxlan_fdb_get(struct sk_buff *skb,
>   * and Tunnel endpoint.
>   * Return true if packet is bogus and should be dropped.

The last line is no longer correct so please remove it in a follow up
patch (unless you need another version).

>   */
> -static bool vxlan_snoop(struct net_device *dev,
> -			union vxlan_addr *src_ip, const u8 *src_mac,
> -			u32 src_ifindex, __be32 vni)
> +static enum skb_drop_reason vxlan_snoop(struct net_device *dev,
> +					union vxlan_addr *src_ip,
> +					const u8 *src_mac, u32 src_ifindex,
> +					__be32 vni)
>  {
>  	struct vxlan_dev *vxlan = netdev_priv(dev);
>  	struct vxlan_fdb *f;

Re: [PATCH net-next v7 06/12] net: vxlan: make vxlan_snoop() return drop reasons

[Menglong Dong]

On Sun, Oct 13, 2024 at 7:41 PM Ido Schimmel <idosch@nvidia.com> wrote:
>
> On Wed, Oct 09, 2024 at 10:28:24AM +0800, Menglong Dong wrote:
> > Change the return type of vxlan_snoop() from bool to enum
> > skb_drop_reason. In this commit, two drop reasons are introduced:
> >
> >   SKB_DROP_REASON_MAC_INVALID_SOURCE
> >   SKB_DROP_REASON_VXLAN_ENTRY_EXISTS
> >
> > Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> > Reviewed-by: Simon Horman <horms@kernel.org>
>
> Reviewed-by: Ido Schimmel <idosch@nvidia.com>
>
> IMO the second reason is quite obscure and unlikely to be very useful,
> but time will tell. The closest thing in the bridge driver is 802.1X /
> MAB support (see "locked" and "mab" bridge link attributes in "man
> bridge"), but I don't think it's close enough to allow us making this
> reason more generic.

Yeah, the concept of the second reason is a little obscure to
the users.

>
> [...]
>
> > diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
> > index 34b44755f663..1a81a3957327 100644
> > --- a/drivers/net/vxlan/vxlan_core.c
> > +++ b/drivers/net/vxlan/vxlan_core.c
> > @@ -1437,9 +1437,10 @@ static int vxlan_fdb_get(struct sk_buff *skb,
> >   * and Tunnel endpoint.
> >   * Return true if packet is bogus and should be dropped.
>
> The last line is no longer correct so please remove it in a follow up
> patch (unless you need another version).
>

Okay, I'll remove it in the next version.

Thanks!
Menglong Dong

> >   */
> > -static bool vxlan_snoop(struct net_device *dev,
> > -                     union vxlan_addr *src_ip, const u8 *src_mac,
> > -                     u32 src_ifindex, __be32 vni)
> > +static enum skb_drop_reason vxlan_snoop(struct net_device *dev,
> > +                                     union vxlan_addr *src_ip,
> > +                                     const u8 *src_mac, u32 src_ifindex,
> > +                                     __be32 vni)
> >  {
> >       struct vxlan_dev *vxlan = netdev_priv(dev);
> >       struct vxlan_fdb *f;

[PATCH net-next v7 07/12] net: vxlan: make vxlan_set_mac() return drop reasons

[Menglong Dong]

Change the return type of vxlan_set_mac() from bool to enum
skb_drop_reason. In this commit, the drop reason
"SKB_DROP_REASON_LOCAL_MAC" is introduced for the case that the source
mac of the packet is a local mac.

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
---
v5:
- modify the document of SKB_DROP_REASON_LOCAL_MAC
v3:
- adjust the call of vxlan_set_mac()
- add SKB_DROP_REASON_LOCAL_MAC
---
 drivers/net/vxlan/vxlan_core.c | 19 ++++++++++---------
 include/net/dropreason-core.h  |  6 ++++++
 2 files changed, 16 insertions(+), 9 deletions(-)

diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index 1a81a3957327..41191a28252a 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -1609,9 +1609,9 @@ static void vxlan_parse_gbp_hdr(struct vxlanhdr *unparsed,
 	unparsed->vx_flags &= ~VXLAN_GBP_USED_BITS;
 }
 
-static bool vxlan_set_mac(struct vxlan_dev *vxlan,
-			  struct vxlan_sock *vs,
-			  struct sk_buff *skb, __be32 vni)
+static enum skb_drop_reason vxlan_set_mac(struct vxlan_dev *vxlan,
+					  struct vxlan_sock *vs,
+					  struct sk_buff *skb, __be32 vni)
 {
 	union vxlan_addr saddr;
 	u32 ifindex = skb->dev->ifindex;
@@ -1622,7 +1622,7 @@ static bool vxlan_set_mac(struct vxlan_dev *vxlan,
 
 	/* Ignore packet loops (and multicast echo) */
 	if (ether_addr_equal(eth_hdr(skb)->h_source, vxlan->dev->dev_addr))
-		return false;
+		return SKB_DROP_REASON_LOCAL_MAC;
 
 	/* Get address from the outer IP header */
 	if (vxlan_get_sk_family(vs) == AF_INET) {
@@ -1635,11 +1635,11 @@ static bool vxlan_set_mac(struct vxlan_dev *vxlan,
 #endif
 	}
 
-	if ((vxlan->cfg.flags & VXLAN_F_LEARN) &&
-	    vxlan_snoop(skb->dev, &saddr, eth_hdr(skb)->h_source, ifindex, vni))
-		return false;
+	if (!(vxlan->cfg.flags & VXLAN_F_LEARN))
+		return SKB_NOT_DROPPED_YET;
 
-	return true;
+	return vxlan_snoop(skb->dev, &saddr, eth_hdr(skb)->h_source,
+			   ifindex, vni);
 }
 
 static bool vxlan_ecn_decapsulate(struct vxlan_sock *vs, void *oiph,
@@ -1774,7 +1774,8 @@ static int vxlan_rcv(struct sock *sk, struct sk_buff *skb)
 	}
 
 	if (!raw_proto) {
-		if (!vxlan_set_mac(vxlan, vs, skb, vni))
+		reason = vxlan_set_mac(vxlan, vs, skb, vni);
+		if (reason)
 			goto drop;
 	} else {
 		skb_reset_mac_header(skb);
diff --git a/include/net/dropreason-core.h b/include/net/dropreason-core.h
index 1cb8d7c953be..fbf92d442c1b 100644
--- a/include/net/dropreason-core.h
+++ b/include/net/dropreason-core.h
@@ -97,6 +97,7 @@
 	FN(MAC_INVALID_SOURCE)		\
 	FN(VXLAN_ENTRY_EXISTS)		\
 	FN(IP_TUNNEL_ECN)		\
+	FN(LOCAL_MAC)			\
 	FNe(MAX)
 
 /**
@@ -443,6 +444,11 @@ enum skb_drop_reason {
 	 * RFC 6040 4.2, see __INET_ECN_decapsulate() for detail.
 	 */
 	SKB_DROP_REASON_IP_TUNNEL_ECN,
+	/**
+	 * @SKB_DROP_REASON_LOCAL_MAC: the source MAC address is equal to
+	 * the MAC address of the local netdev.
+	 */
+	SKB_DROP_REASON_LOCAL_MAC,
 	/**
 	 * @SKB_DROP_REASON_MAX: the maximum of core drop reasons, which
 	 * shouldn't be used as a real 'reason' - only for tracing code gen
-- 
2.39.5

Re: [PATCH net-next v7 07/12] net: vxlan: make vxlan_set_mac() return drop reasons

[Ido Schimmel]

On Wed, Oct 09, 2024 at 10:28:25AM +0800, Menglong Dong wrote:
> Change the return type of vxlan_set_mac() from bool to enum
> skb_drop_reason. In this commit, the drop reason
> "SKB_DROP_REASON_LOCAL_MAC" is introduced for the case that the source
> mac of the packet is a local mac.
> 
> Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> Reviewed-by: Simon Horman <horms@kernel.org>

Reviewed-by: Ido Schimmel <idosch@nvidia.com>

[PATCH net-next v7 08/12] net: vxlan: use kfree_skb_reason() in vxlan_xmit()

[Menglong Dong]

Replace kfree_skb() with kfree_skb_reason() in vxlan_xmit(). Following
new skb drop reasons are introduced for vxlan:

/* no remote found for xmit */
SKB_DROP_REASON_VXLAN_NO_REMOTE
/* packet without necessary metadata reached a device which is
 * in "external" mode
 */
SKB_DROP_REASON_TUNNEL_TXINFO

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
---
v7:
- fix some typos in the document for SKB_DROP_REASON_TUNNEL_TXINFO
v6:
- fix some typos in the document for SKB_DROP_REASON_TUNNEL_TXINFO
v5:
- modify the document for SKB_DROP_REASON_TUNNEL_TXINFO
v2:
- move the drop reason "TXINFO" from vxlan to core
- rename VXLAN_DROP_REMOTE to VXLAN_DROP_NO_REMOTE
---
 drivers/net/vxlan/vxlan_core.c | 6 +++---
 include/net/dropreason-core.h  | 9 +++++++++
 2 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index 41191a28252a..b677ec901807 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -2730,7 +2730,7 @@ static netdev_tx_t vxlan_xmit(struct sk_buff *skb, struct net_device *dev)
 			if (info && info->mode & IP_TUNNEL_INFO_TX)
 				vxlan_xmit_one(skb, dev, vni, NULL, false);
 			else
-				kfree_skb(skb);
+				kfree_skb_reason(skb, SKB_DROP_REASON_TUNNEL_TXINFO);
 			return NETDEV_TX_OK;
 		}
 	}
@@ -2793,7 +2793,7 @@ static netdev_tx_t vxlan_xmit(struct sk_buff *skb, struct net_device *dev)
 			dev_core_stats_tx_dropped_inc(dev);
 			vxlan_vnifilter_count(vxlan, vni, NULL,
 					      VXLAN_VNI_STATS_TX_DROPS, 0);
-			kfree_skb(skb);
+			kfree_skb_reason(skb, SKB_DROP_REASON_VXLAN_NO_REMOTE);
 			return NETDEV_TX_OK;
 		}
 	}
@@ -2816,7 +2816,7 @@ static netdev_tx_t vxlan_xmit(struct sk_buff *skb, struct net_device *dev)
 		if (fdst)
 			vxlan_xmit_one(skb, dev, vni, fdst, did_rsc);
 		else
-			kfree_skb(skb);
+			kfree_skb_reason(skb, SKB_DROP_REASON_VXLAN_NO_REMOTE);
 	}
 
 	return NETDEV_TX_OK;
diff --git a/include/net/dropreason-core.h b/include/net/dropreason-core.h
index fbf92d442c1b..d59bb96c5a02 100644
--- a/include/net/dropreason-core.h
+++ b/include/net/dropreason-core.h
@@ -96,7 +96,9 @@
 	FN(VXLAN_VNI_NOT_FOUND)		\
 	FN(MAC_INVALID_SOURCE)		\
 	FN(VXLAN_ENTRY_EXISTS)		\
+	FN(VXLAN_NO_REMOTE)		\
 	FN(IP_TUNNEL_ECN)		\
+	FN(TUNNEL_TXINFO)		\
 	FN(LOCAL_MAC)			\
 	FNe(MAX)
 
@@ -439,11 +441,18 @@ enum skb_drop_reason {
 	 * entry or an entry pointing to a nexthop.
 	 */
 	SKB_DROP_REASON_VXLAN_ENTRY_EXISTS,
+	/** @SKB_DROP_REASON_VXLAN_NO_REMOTE: no remote found for xmit */
+	SKB_DROP_REASON_VXLAN_NO_REMOTE,
 	/**
 	 * @SKB_DROP_REASON_IP_TUNNEL_ECN: skb is dropped according to
 	 * RFC 6040 4.2, see __INET_ECN_decapsulate() for detail.
 	 */
 	SKB_DROP_REASON_IP_TUNNEL_ECN,
+	/**
+	 * @SKB_DROP_REASON_TUNNEL_TXINFO: packet without necessary metadata
+	 * reached a device which is in "external" mode.
+	 */
+	SKB_DROP_REASON_TUNNEL_TXINFO,
 	/**
 	 * @SKB_DROP_REASON_LOCAL_MAC: the source MAC address is equal to
 	 * the MAC address of the local netdev.
-- 
2.39.5

Re: [PATCH net-next v7 08/12] net: vxlan: use kfree_skb_reason() in vxlan_xmit()

[Ido Schimmel]

On Wed, Oct 09, 2024 at 10:28:26AM +0800, Menglong Dong wrote:
> Replace kfree_skb() with kfree_skb_reason() in vxlan_xmit(). Following
> new skb drop reasons are introduced for vxlan:
> 
> /* no remote found for xmit */
> SKB_DROP_REASON_VXLAN_NO_REMOTE
> /* packet without necessary metadata reached a device which is
>  * in "external" mode
>  */
> SKB_DROP_REASON_TUNNEL_TXINFO
> 
> Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> Reviewed-by: Simon Horman <horms@kernel.org>

Reviewed-by: Ido Schimmel <idosch@nvidia.com>

The first reason might be useful for the bridge driver as well when
there are no ports to forward the packet to (because of egress filtering
for example), but we can make it more generic if / when the bridge
driver is annotated.

Re: [PATCH net-next v7 08/12] net: vxlan: use kfree_skb_reason() in vxlan_xmit()

[Menglong Dong]

On Sun, Oct 13, 2024 at 8:43 PM Ido Schimmel <idosch@nvidia.com> wrote:
>
> On Wed, Oct 09, 2024 at 10:28:26AM +0800, Menglong Dong wrote:
> > Replace kfree_skb() with kfree_skb_reason() in vxlan_xmit(). Following
> > new skb drop reasons are introduced for vxlan:
> >
> > /* no remote found for xmit */
> > SKB_DROP_REASON_VXLAN_NO_REMOTE
> > /* packet without necessary metadata reached a device which is
> >  * in "external" mode
> >  */
> > SKB_DROP_REASON_TUNNEL_TXINFO
> >
> > Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> > Reviewed-by: Simon Horman <horms@kernel.org>
>
> Reviewed-by: Ido Schimmel <idosch@nvidia.com>
>
> The first reason might be useful for the bridge driver as well when
> there are no ports to forward the packet to (because of egress filtering
> for example), but we can make it more generic if / when the bridge
> driver is annotated.

You are right. As we already need a new version, so we can
do something for this patch too. As you said, maybe we can rename the
reason VXLAN_NO_REMOTE to NO_REMOTE for more generic
usage?

Re: [PATCH net-next v7 08/12] net: vxlan: use kfree_skb_reason() in vxlan_xmit()

[Ido Schimmel]

On Mon, Oct 14, 2024 at 08:35:57PM +0800, Menglong Dong wrote:
> On Sun, Oct 13, 2024 at 8:43 PM Ido Schimmel <idosch@nvidia.com> wrote:
> >
> > On Wed, Oct 09, 2024 at 10:28:26AM +0800, Menglong Dong wrote:
> > > Replace kfree_skb() with kfree_skb_reason() in vxlan_xmit(). Following
> > > new skb drop reasons are introduced for vxlan:
> > >
> > > /* no remote found for xmit */
> > > SKB_DROP_REASON_VXLAN_NO_REMOTE
> > > /* packet without necessary metadata reached a device which is
> > >  * in "external" mode
> > >  */
> > > SKB_DROP_REASON_TUNNEL_TXINFO
> > >
> > > Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> > > Reviewed-by: Simon Horman <horms@kernel.org>
> >
> > Reviewed-by: Ido Schimmel <idosch@nvidia.com>
> >
> > The first reason might be useful for the bridge driver as well when
> > there are no ports to forward the packet to (because of egress filtering
> > for example), but we can make it more generic if / when the bridge
> > driver is annotated.
> 
> You are right. As we already need a new version, so we can
> do something for this patch too. As you said, maybe we can rename the
> reason VXLAN_NO_REMOTE to NO_REMOTE for more generic
> usage?

"NO_REMOTE" is not really applicable to the bridge driver as there are
no remotes, but bridge ports. I'm fine with keeping it as is for now and
changing it later if / when needed.

Re: [PATCH net-next v7 08/12] net: vxlan: use kfree_skb_reason() in vxlan_xmit()

[Menglong Dong]

On Mon, Oct 14, 2024 at 11:47 PM Ido Schimmel <idosch@nvidia.com> wrote:
>
> On Mon, Oct 14, 2024 at 08:35:57PM +0800, Menglong Dong wrote:
> > On Sun, Oct 13, 2024 at 8:43 PM Ido Schimmel <idosch@nvidia.com> wrote:
> > >
> > > On Wed, Oct 09, 2024 at 10:28:26AM +0800, Menglong Dong wrote:
> > > > Replace kfree_skb() with kfree_skb_reason() in vxlan_xmit(). Following
> > > > new skb drop reasons are introduced for vxlan:
> > > >
> > > > /* no remote found for xmit */
> > > > SKB_DROP_REASON_VXLAN_NO_REMOTE
> > > > /* packet without necessary metadata reached a device which is
> > > >  * in "external" mode
> > > >  */
> > > > SKB_DROP_REASON_TUNNEL_TXINFO
> > > >
> > > > Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> > > > Reviewed-by: Simon Horman <horms@kernel.org>
> > >
> > > Reviewed-by: Ido Schimmel <idosch@nvidia.com>
> > >
> > > The first reason might be useful for the bridge driver as well when
> > > there are no ports to forward the packet to (because of egress filtering
> > > for example), but we can make it more generic if / when the bridge
> > > driver is annotated.
> >
> > You are right. As we already need a new version, so we can
> > do something for this patch too. As you said, maybe we can rename the
> > reason VXLAN_NO_REMOTE to NO_REMOTE for more generic
> > usage?
>
> "NO_REMOTE" is not really applicable to the bridge driver as there are
> no remotes, but bridge ports. I'm fine with keeping it as is for now and
> changing it later if / when needed.

Okay!

[PATCH net-next v7 09/12] net: vxlan: add drop reasons support to vxlan_xmit_one()

[Menglong Dong]

Replace kfree_skb/dev_kfree_skb with kfree_skb_reason in vxlan_xmit_one.
No drop reasons are introduced in this commit.

The only concern of mine is replacing dev_kfree_skb with
kfree_skb_reason. The dev_kfree_skb is equal to consume_skb, and I'm not
sure if we can change it to kfree_skb here. In my option, the skb is
"dropped" here, isn't it?

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
---
 drivers/net/vxlan/vxlan_core.c | 19 ++++++++++++++-----
 1 file changed, 14 insertions(+), 5 deletions(-)

diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index b677ec901807..508693fa4fd9 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -2374,13 +2374,16 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 	bool use_cache;
 	bool udp_sum = false;
 	bool xnet = !net_eq(vxlan->net, dev_net(vxlan->dev));
+	enum skb_drop_reason reason;
 	bool no_eth_encap;
 	__be32 vni = 0;
 
 	no_eth_encap = flags & VXLAN_F_GPE && skb->protocol != htons(ETH_P_TEB);
-	if (skb_vlan_inet_prepare(skb, no_eth_encap))
+	reason = skb_vlan_inet_prepare(skb, no_eth_encap);
+	if (reason)
 		goto drop;
 
+	reason = SKB_DROP_REASON_NOT_SPECIFIED;
 	old_iph = ip_hdr(skb);
 
 	info = skb_tunnel_info(skb);
@@ -2484,6 +2487,7 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 					   tos, use_cache ? dst_cache : NULL);
 		if (IS_ERR(rt)) {
 			err = PTR_ERR(rt);
+			reason = SKB_DROP_REASON_IP_OUTNOROUTES;
 			goto tx_error;
 		}
 
@@ -2535,8 +2539,10 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 		ttl = ttl ? : ip4_dst_hoplimit(&rt->dst);
 		err = vxlan_build_skb(skb, ndst, sizeof(struct iphdr),
 				      vni, md, flags, udp_sum);
-		if (err < 0)
+		if (err < 0) {
+			reason = SKB_DROP_REASON_NOMEM;
 			goto tx_error;
+		}
 
 		udp_tunnel_xmit_skb(rt, sock4->sock->sk, skb, saddr,
 				    pkey->u.ipv4.dst, tos, ttl, df,
@@ -2556,6 +2562,7 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 		if (IS_ERR(ndst)) {
 			err = PTR_ERR(ndst);
 			ndst = NULL;
+			reason = SKB_DROP_REASON_IP_OUTNOROUTES;
 			goto tx_error;
 		}
 
@@ -2596,8 +2603,10 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 		skb_scrub_packet(skb, xnet);
 		err = vxlan_build_skb(skb, ndst, sizeof(struct ipv6hdr),
 				      vni, md, flags, udp_sum);
-		if (err < 0)
+		if (err < 0) {
+			reason = SKB_DROP_REASON_NOMEM;
 			goto tx_error;
+		}
 
 		udp_tunnel6_xmit_skb(ndst, sock6->sock->sk, skb, dev,
 				     &saddr, &pkey->u.ipv6.dst, tos, ttl,
@@ -2612,7 +2621,7 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 drop:
 	dev_core_stats_tx_dropped_inc(dev);
 	vxlan_vnifilter_count(vxlan, vni, NULL, VXLAN_VNI_STATS_TX_DROPS, 0);
-	dev_kfree_skb(skb);
+	kfree_skb_reason(skb, reason);
 	return;
 
 tx_error:
@@ -2624,7 +2633,7 @@ void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 	dst_release(ndst);
 	DEV_STATS_INC(dev, tx_errors);
 	vxlan_vnifilter_count(vxlan, vni, NULL, VXLAN_VNI_STATS_TX_ERRORS, 0);
-	kfree_skb(skb);
+	kfree_skb_reason(skb, reason);
 }
 
 static void vxlan_xmit_nh(struct sk_buff *skb, struct net_device *dev,
-- 
2.39.5

Re: [PATCH net-next v7 09/12] net: vxlan: add drop reasons support to vxlan_xmit_one()

[Ido Schimmel]

On Wed, Oct 09, 2024 at 10:28:27AM +0800, Menglong Dong wrote:
> Replace kfree_skb/dev_kfree_skb with kfree_skb_reason in vxlan_xmit_one.
> No drop reasons are introduced in this commit.
> 
> The only concern of mine is replacing dev_kfree_skb with
> kfree_skb_reason. The dev_kfree_skb is equal to consume_skb, and I'm not
> sure if we can change it to kfree_skb here. In my option, the skb is
> "dropped" here, isn't it?
> 
> Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> Reviewed-by: Simon Horman <horms@kernel.org>

Reviewed-by: Ido Schimmel <idosch@nvidia.com>

[PATCH net-next v7 10/12] net: vxlan: use kfree_skb_reason() in vxlan_mdb_xmit()

[Menglong Dong]

Replace kfree_skb() with kfree_skb_reason() in vxlan_mdb_xmit. No drop
reasons are introduced in this commit.

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
---
 drivers/net/vxlan/vxlan_mdb.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/vxlan/vxlan_mdb.c b/drivers/net/vxlan/vxlan_mdb.c
index 60eb95a06d55..e1173ae13428 100644
--- a/drivers/net/vxlan/vxlan_mdb.c
+++ b/drivers/net/vxlan/vxlan_mdb.c
@@ -1712,7 +1712,7 @@ netdev_tx_t vxlan_mdb_xmit(struct vxlan_dev *vxlan,
 		vxlan_xmit_one(skb, vxlan->dev, src_vni,
 			       rcu_dereference(fremote->rd), false);
 	else
-		kfree_skb(skb);
+		kfree_skb_reason(skb, SKB_DROP_REASON_VXLAN_NO_REMOTE);
 
 	return NETDEV_TX_OK;
 }
-- 
2.39.5

Re: [PATCH net-next v7 10/12] net: vxlan: use kfree_skb_reason() in vxlan_mdb_xmit()

[Ido Schimmel]

On Wed, Oct 09, 2024 at 10:28:28AM +0800, Menglong Dong wrote:
> Replace kfree_skb() with kfree_skb_reason() in vxlan_mdb_xmit. No drop
> reasons are introduced in this commit.
> 
> Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> Reviewed-by: Simon Horman <horms@kernel.org>

Reviewed-by: Ido Schimmel <idosch@nvidia.com>

[PATCH net-next v7 11/12] net: vxlan: use kfree_skb_reason() in vxlan_encap_bypass()

[Menglong Dong]

Replace kfree_skb with kfree_skb_reason in vxlan_encap_bypass, and no new
skb drop reason is added in this commit.

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
---
 drivers/net/vxlan/vxlan_core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index 508693fa4fd9..da4de19d0331 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -2290,7 +2290,7 @@ static void vxlan_encap_bypass(struct sk_buff *skb, struct vxlan_dev *src_vxlan,
 	rcu_read_lock();
 	dev = skb->dev;
 	if (unlikely(!(dev->flags & IFF_UP))) {
-		kfree_skb(skb);
+		kfree_skb_reason(skb, SKB_DROP_REASON_DEV_READY);
 		goto drop;
 	}
 
-- 
2.39.5

Re: [PATCH net-next v7 11/12] net: vxlan: use kfree_skb_reason() in vxlan_encap_bypass()

[Ido Schimmel]

On Wed, Oct 09, 2024 at 10:28:29AM +0800, Menglong Dong wrote:
> Replace kfree_skb with kfree_skb_reason in vxlan_encap_bypass, and no new
> skb drop reason is added in this commit.
> 
> Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> Reviewed-by: Simon Horman <horms@kernel.org>

Reviewed-by: Ido Schimmel <idosch@nvidia.com>

[PATCH net-next v7 12/12] net: vxlan: use kfree_skb_reason() in encap_bypass_if_local()

[Menglong Dong]

Replace kfree_skb() with kfree_skb_reason() in encap_bypass_if_local, and
no new skb drop reason is added in this commit.

Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
---
 drivers/net/vxlan/vxlan_core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
index da4de19d0331..f7e94bb8e30e 100644
--- a/drivers/net/vxlan/vxlan_core.c
+++ b/drivers/net/vxlan/vxlan_core.c
@@ -2341,7 +2341,7 @@ static int encap_bypass_if_local(struct sk_buff *skb, struct net_device *dev,
 			DEV_STATS_INC(dev, tx_errors);
 			vxlan_vnifilter_count(vxlan, vni, NULL,
 					      VXLAN_VNI_STATS_TX_ERRORS, 0);
-			kfree_skb(skb);
+			kfree_skb_reason(skb, SKB_DROP_REASON_VXLAN_INVALID_HDR);
 
 			return -ENOENT;
 		}
-- 
2.39.5

Re: [PATCH net-next v7 12/12] net: vxlan: use kfree_skb_reason() in encap_bypass_if_local()

[Ido Schimmel]

On Wed, Oct 09, 2024 at 10:28:30AM +0800, Menglong Dong wrote:
> Replace kfree_skb() with kfree_skb_reason() in encap_bypass_if_local, and
> no new skb drop reason is added in this commit.
> 
> Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> Reviewed-by: Simon Horman <horms@kernel.org>
> Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
> ---
>  drivers/net/vxlan/vxlan_core.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
> index da4de19d0331..f7e94bb8e30e 100644
> --- a/drivers/net/vxlan/vxlan_core.c
> +++ b/drivers/net/vxlan/vxlan_core.c
> @@ -2341,7 +2341,7 @@ static int encap_bypass_if_local(struct sk_buff *skb, struct net_device *dev,
>  			DEV_STATS_INC(dev, tx_errors);
>  			vxlan_vnifilter_count(vxlan, vni, NULL,
>  					      VXLAN_VNI_STATS_TX_ERRORS, 0);
> -			kfree_skb(skb);
> +			kfree_skb_reason(skb, SKB_DROP_REASON_VXLAN_INVALID_HDR);

Shouldn't this be SKB_DROP_REASON_VXLAN_VNI_NOT_FOUND ?

>  
>  			return -ENOENT;
>  		}
> -- 
> 2.39.5
> 

Re: [PATCH net-next v7 12/12] net: vxlan: use kfree_skb_reason() in encap_bypass_if_local()

[Menglong Dong]

On Sun, Oct 13, 2024 at 9:24 PM Ido Schimmel <idosch@nvidia.com> wrote:
>
> On Wed, Oct 09, 2024 at 10:28:30AM +0800, Menglong Dong wrote:
> > Replace kfree_skb() with kfree_skb_reason() in encap_bypass_if_local, and
> > no new skb drop reason is added in this commit.
> >
> > Signed-off-by: Menglong Dong <dongml2@chinatelecom.cn>
> > Reviewed-by: Simon Horman <horms@kernel.org>
> > Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
> > ---
> >  drivers/net/vxlan/vxlan_core.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c
> > index da4de19d0331..f7e94bb8e30e 100644
> > --- a/drivers/net/vxlan/vxlan_core.c
> > +++ b/drivers/net/vxlan/vxlan_core.c
> > @@ -2341,7 +2341,7 @@ static int encap_bypass_if_local(struct sk_buff *skb, struct net_device *dev,
> >                       DEV_STATS_INC(dev, tx_errors);
> >                       vxlan_vnifilter_count(vxlan, vni, NULL,
> >                                             VXLAN_VNI_STATS_TX_ERRORS, 0);
> > -                     kfree_skb(skb);
> > +                     kfree_skb_reason(skb, SKB_DROP_REASON_VXLAN_INVALID_HDR);
>
> Shouldn't this be SKB_DROP_REASON_VXLAN_VNI_NOT_FOUND ?
>

Enn.....It should be SKB_DROP_REASON_VXLAN_VNI_NOT_FOUND. I even wonder
why I used SKB_DROP_REASON_VXLAN_INVALID_HDR here.

It seems that we need a new version, and I'll follow you comment
in the other patches of this series too.

Thanks!
Menglong Dong

> >
> >                       return -ENOENT;
> >               }
> > --
> > 2.39.5
> >

Re: [PATCH net-next v7 00/12] net: vxlan: add skb drop reasons support

[Jakub Kicinski]

On Wed,  9 Oct 2024 10:28:18 +0800 Menglong Dong wrote:
> In this series, we add skb drop reasons support to VXLAN, and following
> new skb drop reasons are introduced:

Looks like DaveM already applied this (silently?)
so please *do* follow up on Ido's reviews but as incremental patches
rather than v8

Re: [PATCH net-next v7 00/12] net: vxlan: add skb drop reasons support

[Menglong Dong]

On Tue, Oct 15, 2024 at 8:22 AM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Wed,  9 Oct 2024 10:28:18 +0800 Menglong Dong wrote:
> > In this series, we add skb drop reasons support to VXLAN, and following
> > new skb drop reasons are introduced:
>
> Looks like DaveM already applied this (silently?)

Yeah, the bot didn't notify me this time :/

> so please *do* follow up on Ido's reviews but as incremental patches
> rather than v8

Okay, I'll send new patches to fix the comment of vxlan_snoop(),
and use SKB_DROP_REASON_VXLAN_VNI_NOT_ FOUND in
encap_bypass_if_local().

Thanks!
Menglong Dong