Merge branch 'skb-mono-delivery-time'

Martin KaFai Lau says:

====================
Preserve mono delivery time (EDT) in skb->tstamp

skb->tstamp was first used as the (rcv) timestamp.
The major usage is to report it to the user (e.g. SO_TIMESTAMP).

Later, skb->tstamp is also set as the (future) delivery_time (e.g. EDT in TCP)
during egress and used by the qdisc (e.g. sch_fq) to make decision on when
the skb can be passed to the dev.

Currently, there is no way to tell skb->tstamp having the (rcv) timestamp
or the delivery_time, so it is always reset to 0 whenever forwarded
between egress and ingress.

While it makes sense to always clear the (rcv) timestamp in skb->tstamp
to avoid confusing sch_fq that expects the delivery_time, it is a
performance issue [0] to clear the delivery_time if the skb finally
egress to a fq@phy-dev.

This set is to keep the mono delivery time and make it available to
the final egress interface.  Please see individual patch for
the details.

[0] (slide 22): https://linuxplumbersconf.org/event/11/contributions/953/attachments/867/1658/LPC_2021_BPF_Datapath_Extensions.pdf

v6:
- Add kdoc and use non-UAPI type in patch 6 (Jakub)

v5:
netdev:
- Patch 3 in v4 is broken down into smaller patches 3, 4, and 5 in v5
- The mono_delivery_time bit clearing in __skb_tstamp_tx() is
  done in __net_timestamp() instead.  This is patch 4 in v5.
- Missed a skb_clear_delivery_time() for the 'skip_classify' case
  in dev.c in v4.  That is fixed in patch 5 in v5 for correctness.
  The skb_clear_delivery_time() will be moved to a later
  stage in Patch 10, so it was an intermediate error in v4.
- Added delivery time handling for nfnetlink_{log, queue}.c in patch 9 (Daniel)
- Added delivery time handling in the IPv6 IOAM hop-by-hop option which has
  an experimental IANA assigned value 49 in patch 8
- Added delivery time handling in nf_conntrack for the ipv6 defrag case
  in patch 7
- Removed unlikely() from testing skb->mono_delivery_time (Daniel)

bpf:
- Remove the skb->tstamp dance in ingress.  Depends on bpf insn
  rewrite to return 0 if skb->tstamp has delivery time in patch 11.
  It is to backward compatible with the existing tc-bpf@ingress in
  patch 11.
- bpf_set_delivery_time() will also allow dtime == 0 and
  dtime_type == BPF_SKB_DELIVERY_TIME_NONE as argument
  in patch 12.

v4:
netdev:
- Push the skb_clear_delivery_time() from
  ip_local_deliver() and ip6_input() to
  ip_local_deliver_finish() and ip6_input_finish()
  to accommodate the ipvs forward path.
  This is the notable change in v4 at the netdev side.

    - Patch 3/8 first does the skb_clear_delivery_time() after
      sch_handle_ingress() in dev.c and this will make the
      tc-bpf forward path work via the bpf_redirect_*() helper.

    - The next patch 4/8 (new in v4) will then postpone the
      skb_clear_delivery_time() from dev.c to
      the ip_local_deliver_finish() and ip6_input_finish() after
      taking care of the tstamp usage in the ip defrag case.
      This will make the kernel forward path also work, e.g.
      the ip[6]_forward().

- Fixed a case v3 which missed setting the skb->mono_delivery_time bit
  when sending TCP rst/ack in some cases (e.g. from a ctl_sk).
  That case happens at ip_send_unicast_reply() and
  tcp_v6_send_response().  It is fixed in patch 1/8 (and
  then patch 3/8) in v4.

bpf:
- Adding __sk_buff->delivery_time_type instead of adding
  __sk_buff->mono_delivery_time as in v3.  The tc-bpf can stay with
  one __sk_buff->tstamp instead of having two 'time' fields
  while one is 0 and another is not.
  tc-bpf can use the new __sk_buff->delivery_time_type to tell
  what is stored in __sk_buff->tstamp.
- bpf_skb_set_delivery_time() helper is added to set
  __sk_buff->tstamp from non mono delivery_time to
  mono delivery_time
- Most of the convert_ctx_access() bpf insn rewrite in v3
  is gone, so no new rewrite added for __sk_buff->tstamp.
  The only rewrite added is for reading the new
  __sk_buff->delivery_time_type.
- Added selftests, test_tc_dtime.c

v3:
- Feedback from v2 is using shinfo(skb)->tx_flags could be racy.
- Considered to reuse a few bits in skb->tstamp to represent
  different semantics, other than more code churns, it will break
  the bpf usecase which currently can write and then read back
  the skb->tstamp.
- Went back to v1 idea on adding a bit to skb and address the
  feedbacks on v1:
- Added one bit skb->mono_delivery_time to flag that
  the skb->tstamp has the mono delivery_time (EDT), instead
  of adding a bit to flag if the skb->tstamp has been forwarded or not.
- Instead of resetting the delivery_time back to the (rcv) timestamp
  during recvmsg syscall which may be too late and not useful,
  the delivery_time reset in v3 happens earlier once the stack
  knows that the skb will be delivered locally.
- Handled the tapping@ingress case by af_packet
- No need to change the (rcv) timestamp to mono clock base as in v1.
  The added one bit to flag skb->mono_delivery_time is enough
  to keep the EDT delivery_time during forward.
- Added logic to the bpf side to make the existing bpf
  running at ingress can still get the (rcv) timestamp
  when reading the __sk_buff->tstamp.  New __sk_buff->mono_delivery_time
  is also added.  Test is still needed to test this piece.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

38 files changed, 1174 insertions, 73 deletions

diff --git a/drivers/net/loopback.c b/drivers/net/loopback.c
index d05f86fe78c9..720394c0639b 100644
--- a/drivers/net/loopback.c
+++ b/drivers/net/loopback.c
@@ -74,7 +74,7 @@ static netdev_tx_t loopback_xmit(struct sk_buff *skb,
 	skb_tx_timestamp(skb);
 
 	/* do not fool net_timestamp_check() with various clock bases */
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 
 	skb_orphan(skb);
 
diff --git a/include/linux/filter.h b/include/linux/filter.h
index 1cb1af917617..9bf26307247f 100644
--- a/include/linux/filter.h
+++ b/include/linux/filter.h
@@ -572,7 +572,8 @@ struct bpf_prog {
 				has_callchain_buf:1, /* callchain buffer allocated? */
 				enforce_expected_attach_type:1, /* Enforce expected_attach_type checking at attach time */
 				call_get_stack:1, /* Do we call bpf_get_stack() or bpf_get_stackid() */
-				call_get_func_ip:1; /* Do we call get_func_ip() */
+				call_get_func_ip:1, /* Do we call get_func_ip() */
+				delivery_time_access:1; /* Accessed __sk_buff->delivery_time_type */
 	enum bpf_prog_type	type;		/* Type of BPF program */
 	enum bpf_attach_type	expected_attach_type; /* For some prog types */
 	u32			len;		/* Number of filter blocks */
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index d67941f78b92..5445860e1ba6 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -795,6 +795,10 @@ typedef unsigned char *sk_buff_data_t;
  *	@dst_pending_confirm: need to confirm neighbour
  *	@decrypted: Decrypted SKB
  *	@slow_gro: state present at GRO time, slower prepare step required
+ *	@mono_delivery_time: When set, skb->tstamp has the
+ *		delivery_time in mono clock base (i.e. EDT).  Otherwise, the
+ *		skb->tstamp has the (rcv) timestamp at ingress and
+ *		delivery_time at egress.
  *	@napi_id: id of the NAPI struct this skb came from
  *	@sender_cpu: (aka @napi_id) source CPU in XPS
  *	@secmark: security marking
@@ -937,8 +941,12 @@ struct sk_buff {
 	__u8			vlan_present:1;	/* See PKT_VLAN_PRESENT_BIT */
 	__u8			csum_complete_sw:1;
 	__u8			csum_level:2;
-	__u8			csum_not_inet:1;
 	__u8			dst_pending_confirm:1;
+	__u8			mono_delivery_time:1;
+#ifdef CONFIG_NET_CLS_ACT
+	__u8			tc_skip_classify:1;
+	__u8			tc_at_ingress:1;
+#endif
 #ifdef CONFIG_IPV6_NDISC_NODETYPE
 	__u8			ndisc_nodetype:2;
 #endif
@@ -950,10 +958,6 @@ struct sk_buff {
 	__u8			offload_fwd_mark:1;
 	__u8			offload_l3_fwd_mark:1;
 #endif
-#ifdef CONFIG_NET_CLS_ACT
-	__u8			tc_skip_classify:1;
-	__u8			tc_at_ingress:1;
-#endif
 	__u8			redirected:1;
 #ifdef CONFIG_NET_REDIRECT
 	__u8			from_ingress:1;
@@ -965,6 +969,7 @@ struct sk_buff {
 	__u8			decrypted:1;
 #endif
 	__u8			slow_gro:1;
+	__u8			csum_not_inet:1;
 
 #ifdef CONFIG_NET_SCHED
 	__u16			tc_index;	/* traffic control index */
@@ -1042,10 +1047,16 @@ struct sk_buff {
 /* if you move pkt_vlan_present around you also must adapt these constants */
 #ifdef __BIG_ENDIAN_BITFIELD
 #define PKT_VLAN_PRESENT_BIT	7
+#define TC_AT_INGRESS_MASK		(1 << 0)
+#define SKB_MONO_DELIVERY_TIME_MASK	(1 << 2)
 #else
 #define PKT_VLAN_PRESENT_BIT	0
+#define TC_AT_INGRESS_MASK		(1 << 7)
+#define SKB_MONO_DELIVERY_TIME_MASK	(1 << 5)
 #endif
 #define PKT_VLAN_PRESENT_OFFSET	offsetof(struct sk_buff, __pkt_vlan_present_offset)
+#define TC_AT_INGRESS_OFFSET offsetof(struct sk_buff, __pkt_vlan_present_offset)
+#define SKB_MONO_DELIVERY_TIME_OFFSET offsetof(struct sk_buff, __pkt_vlan_present_offset)
 
 #ifdef __KERNEL__
 /*
@@ -3976,6 +3987,7 @@ static inline void skb_get_new_timestampns(const struct sk_buff *skb,
 static inline void __net_timestamp(struct sk_buff *skb)
 {
 	skb->tstamp = ktime_get_real();
+	skb->mono_delivery_time = 0;
 }
 
 static inline ktime_t net_timedelta(ktime_t t)
@@ -3983,6 +3995,56 @@ static inline ktime_t net_timedelta(ktime_t t)
 	return ktime_sub(ktime_get_real(), t);
 }
 
+static inline void skb_set_delivery_time(struct sk_buff *skb, ktime_t kt,
+					 bool mono)
+{
+	skb->tstamp = kt;
+	skb->mono_delivery_time = kt && mono;
+}
+
+DECLARE_STATIC_KEY_FALSE(netstamp_needed_key);
+
+/* It is used in the ingress path to clear the delivery_time.
+ * If needed, set the skb->tstamp to the (rcv) timestamp.
+ */
+static inline void skb_clear_delivery_time(struct sk_buff *skb)
+{
+	if (skb->mono_delivery_time) {
+		skb->mono_delivery_time = 0;
+		if (static_branch_unlikely(&netstamp_needed_key))
+			skb->tstamp = ktime_get_real();
+		else
+			skb->tstamp = 0;
+	}
+}
+
+static inline void skb_clear_tstamp(struct sk_buff *skb)
+{
+	if (skb->mono_delivery_time)
+		return;
+
+	skb->tstamp = 0;
+}
+
+static inline ktime_t skb_tstamp(const struct sk_buff *skb)
+{
+	if (skb->mono_delivery_time)
+		return 0;
+
+	return skb->tstamp;
+}
+
+static inline ktime_t skb_tstamp_cond(const struct sk_buff *skb, bool cond)
+{
+	if (!skb->mono_delivery_time && skb->tstamp)
+		return skb->tstamp;
+
+	if (static_branch_unlikely(&netstamp_needed_key) || cond)
+		return ktime_get_real();
+
+	return 0;
+}
+
 static inline u8 skb_metadata_len(const struct sk_buff *skb)
 {
 	return skb_shinfo(skb)->meta_len;
@@ -4839,7 +4901,7 @@ static inline void skb_set_redirected(struct sk_buff *skb, bool from_ingress)
 #ifdef CONFIG_NET_REDIRECT
 	skb->from_ingress = from_ingress;
 	if (skb->from_ingress)
-		skb->tstamp = 0;
+		skb_clear_tstamp(skb);
 #endif
 }
 
diff --git a/include/net/inet_frag.h b/include/net/inet_frag.h
index 63540be0fc34..911ad930867d 100644
--- a/include/net/inet_frag.h
+++ b/include/net/inet_frag.h
@@ -70,6 +70,7 @@ struct frag_v6_compare_key {
  * @stamp: timestamp of the last received fragment
  * @len: total length of the original datagram
  * @meat: length of received fragments so far
+ * @mono_delivery_time: stamp has a mono delivery time (EDT)
  * @flags: fragment queue flags
  * @max_size: maximum received fragment size
  * @fqdir: pointer to struct fqdir
@@ -90,6 +91,7 @@ struct inet_frag_queue {
 	ktime_t			stamp;
 	int			len;
 	int			meat;
+	u8			mono_delivery_time;
 	__u8			flags;
 	u16			max_size;
 	struct fqdir		*fqdir;
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index afe3d0d7f5f2..4eebea830613 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -5086,6 +5086,37 @@ union bpf_attr {
  *	Return
  *		0 on success, or a negative error in case of failure. On error
  *		*dst* buffer is zeroed out.
+ *
+ * long bpf_skb_set_delivery_time(struct sk_buff *skb, u64 dtime, u32 dtime_type)
+ *	Description
+ *		Set a *dtime* (delivery time) to the __sk_buff->tstamp and also
+ *		change the __sk_buff->delivery_time_type to *dtime_type*.
+ *
+ *		When setting a delivery time (non zero *dtime*) to
+ *		__sk_buff->tstamp, only BPF_SKB_DELIVERY_TIME_MONO *dtime_type*
+ *		is supported.  It is the only delivery_time_type that will be
+ *		kept after bpf_redirect_*().
+ *
+ *		If there is no need to change the __sk_buff->delivery_time_type,
+ *		the delivery time can be directly written to __sk_buff->tstamp
+ *		instead.
+ *
+ *		*dtime* 0 and *dtime_type* BPF_SKB_DELIVERY_TIME_NONE
+ *		can be used to clear any delivery time stored in
+ *		__sk_buff->tstamp.
+ *
+ *		Only IPv4 and IPv6 skb->protocol are supported.
+ *
+ *		This function is most useful when it needs to set a
+ *		mono delivery time to __sk_buff->tstamp and then
+ *		bpf_redirect_*() to the egress of an iface.  For example,
+ *		changing the (rcv) timestamp in __sk_buff->tstamp at
+ *		ingress to a mono delivery time and then bpf_redirect_*()
+ *		to sch_fq@phy-dev.
+ *	Return
+ *		0 on success.
+ *		**-EINVAL** for invalid input
+ *		**-EOPNOTSUPP** for unsupported delivery_time_type and protocol
  */
 #define __BPF_FUNC_MAPPER(FN)		\
 	FN(unspec),			\
@@ -5280,6 +5311,7 @@ union bpf_attr {
 	FN(xdp_load_bytes),		\
 	FN(xdp_store_bytes),		\
 	FN(copy_from_user_task),	\
+	FN(skb_set_delivery_time),      \
 	/* */
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper
@@ -5469,6 +5501,12 @@ union {					\
 	__u64 :64;			\
 } __attribute__((aligned(8)))
 
+enum {
+	BPF_SKB_DELIVERY_TIME_NONE,
+	BPF_SKB_DELIVERY_TIME_UNSPEC,
+	BPF_SKB_DELIVERY_TIME_MONO,
+};
+
 /* user accessible mirror of in-kernel sk_buff.
  * new fields can only be added to the end of this structure
  */
@@ -5509,7 +5547,8 @@ struct __sk_buff {
 	__u32 gso_segs;
 	__bpf_md_ptr(struct bpf_sock *, sk);
 	__u32 gso_size;
-	__u32 :32;		/* Padding, future use. */
+	__u8  delivery_time_type;
+	__u32 :24;		/* Padding, future use. */
 	__u64 hwtstamp;
 };
 
diff --git a/net/bridge/br_forward.c b/net/bridge/br_forward.c
index ec646656dbf1..02bb620d3b8d 100644
--- a/net/bridge/br_forward.c
+++ b/net/bridge/br_forward.c
@@ -62,7 +62,7 @@ EXPORT_SYMBOL_GPL(br_dev_queue_push_xmit);
 
 int br_forward_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
 {
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 	return NF_HOOK(NFPROTO_BRIDGE, NF_BR_POST_ROUTING,
 		       net, sk, skb, NULL, skb->dev,
 		       br_dev_queue_push_xmit);
diff --git a/net/bridge/netfilter/nf_conntrack_bridge.c b/net/bridge/netfilter/nf_conntrack_bridge.c
index fdbed3158555..ebfb2a5c59e4 100644
--- a/net/bridge/netfilter/nf_conntrack_bridge.c
+++ b/net/bridge/netfilter/nf_conntrack_bridge.c
@@ -32,6 +32,7 @@ static int nf_br_ip_fragment(struct net *net, struct sock *sk,
 					   struct sk_buff *))
 {
 	int frag_max_size = BR_INPUT_SKB_CB(skb)->frag_max_size;
+	bool mono_delivery_time = skb->mono_delivery_time;
 	unsigned int hlen, ll_rs, mtu;
 	ktime_t tstamp = skb->tstamp;
 	struct ip_frag_state state;
@@ -81,7 +82,7 @@ static int nf_br_ip_fragment(struct net *net, struct sock *sk,
 			if (iter.frag)
 				ip_fraglist_prepare(skb, &iter);
 
-			skb->tstamp = tstamp;
+			skb_set_delivery_time(skb, tstamp, mono_delivery_time);
 			err = output(net, sk, data, skb);
 			if (err || !iter.frag)
 				break;
@@ -112,7 +113,7 @@ slow_path:
 			goto blackhole;
 		}
 
-		skb2->tstamp = tstamp;
+		skb_set_delivery_time(skb2, tstamp, mono_delivery_time);
 		err = output(net, sk, data, skb2);
 		if (err)
 			goto blackhole;
diff --git a/net/core/dev.c b/net/core/dev.c
index c9e54e5ad48d..5db2443c2371 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -2047,7 +2047,8 @@ void net_dec_egress_queue(void)
 EXPORT_SYMBOL_GPL(net_dec_egress_queue);
 #endif
 
-static DEFINE_STATIC_KEY_FALSE(netstamp_needed_key);
+DEFINE_STATIC_KEY_FALSE(netstamp_needed_key);
+EXPORT_SYMBOL(netstamp_needed_key);
 #ifdef CONFIG_JUMP_LABEL
 static atomic_t netstamp_needed_deferred;
 static atomic_t netstamp_wanted;
@@ -2108,14 +2109,15 @@ EXPORT_SYMBOL(net_disable_timestamp);
 static inline void net_timestamp_set(struct sk_buff *skb)
 {
 	skb->tstamp = 0;
+	skb->mono_delivery_time = 0;
 	if (static_branch_unlikely(&netstamp_needed_key))
-		__net_timestamp(skb);
+		skb->tstamp = ktime_get_real();
 }
 
 #define net_timestamp_check(COND, SKB)				\
 	if (static_branch_unlikely(&netstamp_needed_key)) {	\
 		if ((COND) && !(SKB)->tstamp)			\
-			__net_timestamp(SKB);			\
+			(SKB)->tstamp = ktime_get_real();	\
 	}							\
 
 bool is_skb_forwardable(const struct net_device *dev, const struct sk_buff *skb)
diff --git a/net/core/filter.c b/net/core/filter.c
index 65869fd510e8..88767f7da150 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -2107,7 +2107,7 @@ static inline int __bpf_tx_skb(struct net_device *dev, struct sk_buff *skb)
 	}
 
 	skb->dev = dev;
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 
 	dev_xmit_recursion_inc();
 	ret = dev_queue_xmit(skb);
@@ -2176,7 +2176,7 @@ static int bpf_out_neigh_v6(struct net *net, struct sk_buff *skb,
 	}
 
 	skb->dev = dev;
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 
 	if (unlikely(skb_headroom(skb) < hh_len && dev->header_ops)) {
 		skb = skb_expand_head(skb, hh_len);
@@ -2274,7 +2274,7 @@ static int bpf_out_neigh_v4(struct net *net, struct sk_buff *skb,
 	}
 
 	skb->dev = dev;
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 
 	if (unlikely(skb_headroom(skb) < hh_len && dev->header_ops)) {
 		skb = skb_expand_head(skb, hh_len);
@@ -7388,6 +7388,43 @@ static const struct bpf_func_proto bpf_sock_ops_reserve_hdr_opt_proto = {
 	.arg3_type	= ARG_ANYTHING,
 };
 
+BPF_CALL_3(bpf_skb_set_delivery_time, struct sk_buff *, skb,
+	   u64, dtime, u32, dtime_type)
+{
+	/* skb_clear_delivery_time() is done for inet protocol */
+	if (skb->protocol != htons(ETH_P_IP) &&
+	    skb->protocol != htons(ETH_P_IPV6))
+		return -EOPNOTSUPP;
+
+	switch (dtime_type) {
+	case BPF_SKB_DELIVERY_TIME_MONO:
+		if (!dtime)
+			return -EINVAL;
+		skb->tstamp = dtime;
+		skb->mono_delivery_time = 1;
+		break;
+	case BPF_SKB_DELIVERY_TIME_NONE:
+		if (dtime)
+			return -EINVAL;
+		skb->tstamp = 0;
+		skb->mono_delivery_time = 0;
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static const struct bpf_func_proto bpf_skb_set_delivery_time_proto = {
+	.func           = bpf_skb_set_delivery_time,
+	.gpl_only       = false,
+	.ret_type       = RET_INTEGER,
+	.arg1_type      = ARG_PTR_TO_CTX,
+	.arg2_type      = ARG_ANYTHING,
+	.arg3_type      = ARG_ANYTHING,
+};
+
 #endif /* CONFIG_INET */
 
 bool bpf_helper_changes_pkt_data(void *func)
@@ -7749,6 +7786,8 @@ tc_cls_act_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_tcp_gen_syncookie_proto;
 	case BPF_FUNC_sk_assign:
 		return &bpf_sk_assign_proto;
+	case BPF_FUNC_skb_set_delivery_time:
+		return &bpf_skb_set_delivery_time_proto;
 #endif
 	default:
 		return bpf_sk_base_func_proto(func_id);
@@ -8088,7 +8127,9 @@ static bool bpf_skb_is_valid_access(int off, int size, enum bpf_access_type type
 			return false;
 		info->reg_type = PTR_TO_SOCK_COMMON_OR_NULL;
 		break;
-	case offsetofend(struct __sk_buff, gso_size) ... offsetof(struct __sk_buff, hwtstamp) - 1:
+	case offsetof(struct __sk_buff, delivery_time_type):
+		return false;
+	case offsetofend(struct __sk_buff, delivery_time_type) ... offsetof(struct __sk_buff, hwtstamp) - 1:
 		/* Explicitly prohibit access to padding in __sk_buff. */
 		return false;
 	default:
@@ -8443,6 +8484,15 @@ static bool tc_cls_act_is_valid_access(int off, int size,
 		break;
 	case bpf_ctx_range_till(struct __sk_buff, family, local_port):
 		return false;
+	case offsetof(struct __sk_buff, delivery_time_type):
+		/* The convert_ctx_access() on reading and writing
+		 * __sk_buff->tstamp depends on whether the bpf prog
+		 * has used __sk_buff->delivery_time_type or not.
+		 * Thus, we need to set prog->delivery_time_access
+		 * earlier during is_valid_access() here.
+		 */
+		((struct bpf_prog *)prog)->delivery_time_access = 1;
+		return size == sizeof(__u8);
 	}
 
 	return bpf_skb_is_valid_access(off, size, type, prog, info);
@@ -8838,6 +8888,45 @@ static u32 flow_dissector_convert_ctx_access(enum bpf_access_type type,
 	return insn - insn_buf;
 }
 
+static struct bpf_insn *bpf_convert_dtime_type_read(const struct bpf_insn *si,
+						    struct bpf_insn *insn)
+{
+	__u8 value_reg = si->dst_reg;
+	__u8 skb_reg = si->src_reg;
+	__u8 tmp_reg = BPF_REG_AX;
+
+	*insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg,
+			      SKB_MONO_DELIVERY_TIME_OFFSET);
+	*insn++ = BPF_ALU32_IMM(BPF_AND, tmp_reg,
+				SKB_MONO_DELIVERY_TIME_MASK);
+	*insn++ = BPF_JMP32_IMM(BPF_JEQ, tmp_reg, 0, 2);
+	/* value_reg = BPF_SKB_DELIVERY_TIME_MONO */
+	*insn++ = BPF_MOV32_IMM(value_reg, BPF_SKB_DELIVERY_TIME_MONO);
+	*insn++ = BPF_JMP_A(IS_ENABLED(CONFIG_NET_CLS_ACT) ? 10 : 5);
+
+	*insn++ = BPF_LDX_MEM(BPF_DW, tmp_reg, skb_reg,
+			      offsetof(struct sk_buff, tstamp));
+	*insn++ = BPF_JMP_IMM(BPF_JNE, tmp_reg, 0, 2);
+	/* value_reg = BPF_SKB_DELIVERY_TIME_NONE */
+	*insn++ = BPF_MOV32_IMM(value_reg, BPF_SKB_DELIVERY_TIME_NONE);
+	*insn++ = BPF_JMP_A(IS_ENABLED(CONFIG_NET_CLS_ACT) ? 6 : 1);
+
+#ifdef CONFIG_NET_CLS_ACT
+	*insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg, TC_AT_INGRESS_OFFSET);
+	*insn++ = BPF_ALU32_IMM(BPF_AND, tmp_reg, TC_AT_INGRESS_MASK);
+	*insn++ = BPF_JMP32_IMM(BPF_JEQ, tmp_reg, 0, 2);
+	/* At ingress, value_reg = 0 */
+	*insn++ = BPF_MOV32_IMM(value_reg, 0);
+	*insn++ = BPF_JMP_A(1);
+#endif
+
+	/* value_reg = BPF_SKB_DELIVERYT_TIME_UNSPEC */
+	*insn++ = BPF_MOV32_IMM(value_reg, BPF_SKB_DELIVERY_TIME_UNSPEC);
+
+	/* 15 insns with CONFIG_NET_CLS_ACT */
+	return insn;
+}
+
 static struct bpf_insn *bpf_convert_shinfo_access(const struct bpf_insn *si,
 						  struct bpf_insn *insn)
 {
@@ -8859,6 +8948,71 @@ static struct bpf_insn *bpf_convert_shinfo_access(const struct bpf_insn *si,
 	return insn;
 }
 
+static struct bpf_insn *bpf_convert_tstamp_read(const struct bpf_prog *prog,
+						const struct bpf_insn *si,
+						struct bpf_insn *insn)
+{
+	__u8 value_reg = si->dst_reg;
+	__u8 skb_reg = si->src_reg;
+
+#ifdef CONFIG_NET_CLS_ACT
+	if (!prog->delivery_time_access) {
+		__u8 tmp_reg = BPF_REG_AX;
+
+		*insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg, TC_AT_INGRESS_OFFSET);
+		*insn++ = BPF_ALU32_IMM(BPF_AND, tmp_reg, TC_AT_INGRESS_MASK);
+		*insn++ = BPF_JMP32_IMM(BPF_JEQ, tmp_reg, 0, 5);
+		/* @ingress, read __sk_buff->tstamp as the (rcv) timestamp,
+		 * so check the skb->mono_delivery_time.
+		 */
+		*insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg,
+				      SKB_MONO_DELIVERY_TIME_OFFSET);
+		*insn++ = BPF_ALU32_IMM(BPF_AND, tmp_reg,
+					SKB_MONO_DELIVERY_TIME_MASK);
+		*insn++ = BPF_JMP32_IMM(BPF_JEQ, tmp_reg, 0, 2);
+		/* skb->mono_delivery_time is set, read 0 as the (rcv) timestamp. */
+		*insn++ = BPF_MOV64_IMM(value_reg, 0);
+		*insn++ = BPF_JMP_A(1);
+	}
+#endif
+
+	*insn++ = BPF_LDX_MEM(BPF_DW, value_reg, skb_reg,
+			      offsetof(struct sk_buff, tstamp));
+	return insn;
+}
+
+static struct bpf_insn *bpf_convert_tstamp_write(const struct bpf_prog *prog,
+						 const struct bpf_insn *si,
+						 struct bpf_insn *insn)
+{
+	__u8 value_reg = si->src_reg;
+	__u8 skb_reg = si->dst_reg;
+
+#ifdef CONFIG_NET_CLS_ACT
+	if (!prog->delivery_time_access) {
+		__u8 tmp_reg = BPF_REG_AX;
+
+		*insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg, TC_AT_INGRESS_OFFSET);
+		*insn++ = BPF_ALU32_IMM(BPF_AND, tmp_reg, TC_AT_INGRESS_MASK);
+		*insn++ = BPF_JMP32_IMM(BPF_JEQ, tmp_reg, 0, 3);
+		/* Writing __sk_buff->tstamp at ingress as the (rcv) timestamp.
+		 * Clear the skb->mono_delivery_time.
+		 */
+		*insn++ = BPF_LDX_MEM(BPF_B, tmp_reg, skb_reg,
+				      SKB_MONO_DELIVERY_TIME_OFFSET);
+		*insn++ = BPF_ALU32_IMM(BPF_AND, tmp_reg,
+					~SKB_MONO_DELIVERY_TIME_MASK);
+		*insn++ = BPF_STX_MEM(BPF_B, skb_reg, tmp_reg,
+				      SKB_MONO_DELIVERY_TIME_OFFSET);
+	}
+#endif
+
+	/* skb->tstamp = tstamp */
+	*insn++ = BPF_STX_MEM(BPF_DW, skb_reg, value_reg,
+			      offsetof(struct sk_buff, tstamp));
+	return insn;
+}
+
 static u32 bpf_convert_ctx_access(enum bpf_access_type type,
 				  const struct bpf_insn *si,
 				  struct bpf_insn *insn_buf,
@@ -9167,17 +9321,13 @@ static u32 bpf_convert_ctx_access(enum bpf_access_type type,
 		BUILD_BUG_ON(sizeof_field(struct sk_buff, tstamp) != 8);
 
 		if (type == BPF_WRITE)
-			*insn++ = BPF_STX_MEM(BPF_DW,
-					      si->dst_reg, si->src_reg,
-					      bpf_target_off(struct sk_buff,
-							     tstamp, 8,
-							     target_size));
+			insn = bpf_convert_tstamp_write(prog, si, insn);
 		else
-			*insn++ = BPF_LDX_MEM(BPF_DW,
-					      si->dst_reg, si->src_reg,
-					      bpf_target_off(struct sk_buff,
-							     tstamp, 8,
-							     target_size));
+			insn = bpf_convert_tstamp_read(prog, si, insn);
+		break;
+
+	case offsetof(struct __sk_buff, delivery_time_type):
+		insn = bpf_convert_dtime_type_read(si, insn);
 		break;
 
 	case offsetof(struct __sk_buff, gso_segs):
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index b32c5d782fe1..e5082836295b 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -4851,7 +4851,7 @@ void __skb_tstamp_tx(struct sk_buff *orig_skb,
 	if (hwtstamps)
 		*skb_hwtstamps(skb) = *hwtstamps;
 	else
-		skb->tstamp = ktime_get_real();
+		__net_timestamp(skb);
 
 	__skb_complete_tx_timestamp(skb, sk, tstype, opt_stats);
 }
@@ -5381,7 +5381,7 @@ void skb_scrub_packet(struct sk_buff *skb, bool xnet)
 
 	ipvs_reset(skb);
 	skb->mark = 0;
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 }
 EXPORT_SYMBOL_GPL(skb_scrub_packet);
 
diff --git a/net/ieee802154/6lowpan/reassembly.c b/net/ieee802154/6lowpan/reassembly.c
index be6f06adefe0..a91283d1e5bf 100644
--- a/net/ieee802154/6lowpan/reassembly.c
+++ b/net/ieee802154/6lowpan/reassembly.c
@@ -130,6 +130,7 @@ static int lowpan_frag_queue(struct lowpan_frag_queue *fq,
 		goto err;
 
 	fq->q.stamp = skb->tstamp;
+	fq->q.mono_delivery_time = skb->mono_delivery_time;
 	if (frag_type == LOWPAN_DISPATCH_FRAG1)
 		fq->q.flags |= INET_FRAG_FIRST_IN;
 
diff --git a/net/ipv4/inet_fragment.c b/net/ipv4/inet_fragment.c
index 341096807100..63948f6aeca0 100644
--- a/net/ipv4/inet_fragment.c
+++ b/net/ipv4/inet_fragment.c
@@ -572,6 +572,7 @@ void inet_frag_reasm_finish(struct inet_frag_queue *q, struct sk_buff *head,
 	skb_mark_not_on_list(head);
 	head->prev = NULL;
 	head->tstamp = q->stamp;
+	head->mono_delivery_time = q->mono_delivery_time;
 }
 EXPORT_SYMBOL(inet_frag_reasm_finish);
 
diff --git a/net/ipv4/ip_forward.c b/net/ipv4/ip_forward.c
index 00ec819f949b..92ba3350274b 100644
--- a/net/ipv4/ip_forward.c
+++ b/net/ipv4/ip_forward.c
@@ -79,7 +79,7 @@ static int ip_forward_finish(struct net *net, struct sock *sk, struct sk_buff *s
 	if (unlikely(opt->optlen))
 		ip_forward_options(skb);
 
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 	return dst_output(net, sk, skb);
 }
 
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index fad803d2d711..fb153569889e 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -349,6 +349,7 @@ static int ip_frag_queue(struct ipq *qp, struct sk_buff *skb)
 		qp->iif = dev->ifindex;
 
 	qp->q.stamp = skb->tstamp;
+	qp->q.mono_delivery_time = skb->mono_delivery_time;
 	qp->q.meat += skb->len;
 	qp->ecn |= ecn;
 	add_frag_mem_limit(qp->q.fqdir, skb->truesize);
diff --git a/net/ipv4/ip_input.c b/net/ipv4/ip_input.c
index d94f9f7e60c3..95f7bb052784 100644
--- a/net/ipv4/ip_input.c
+++ b/net/ipv4/ip_input.c
@@ -226,6 +226,7 @@ resubmit:
 
 static int ip_local_deliver_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
 {
+	skb_clear_delivery_time(skb);
 	__skb_pull(skb, skb_network_header_len(skb));
 
 	rcu_read_lock();
diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
index 6df3545c891d..00b4bf26fd93 100644
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -761,6 +761,7 @@ int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
 {
 	struct iphdr *iph;
 	struct sk_buff *skb2;
+	bool mono_delivery_time = skb->mono_delivery_time;
 	struct rtable *rt = skb_rtable(skb);
 	unsigned int mtu, hlen, ll_rs;
 	struct ip_fraglist_iter iter;
@@ -852,7 +853,7 @@ int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
 				}
 			}
 
-			skb->tstamp = tstamp;
+			skb_set_delivery_time(skb, tstamp, mono_delivery_time);
 			err = output(net, sk, skb);
 
 			if (!err)
@@ -908,7 +909,7 @@ slow_path:
 		/*
 		 *	Put this fragment into the sending queue.
 		 */
-		skb2->tstamp = tstamp;
+		skb_set_delivery_time(skb2, tstamp, mono_delivery_time);
 		err = output(net, sk, skb2);
 		if (err)
 			goto fail;
@@ -1727,6 +1728,7 @@ void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
 			  arg->csumoffset) = csum_fold(csum_add(nskb->csum,
 								arg->csum));
 		nskb->ip_summed = CHECKSUM_NONE;
+		nskb->mono_delivery_time = !!transmit_time;
 		ip_push_pending_frames(sk, &fl4);
 	}
 out:
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index e76bf1e9251e..2319531267c6 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -1253,7 +1253,7 @@ static int __tcp_transmit_skb(struct sock *sk, struct sk_buff *skb,
 	tp = tcp_sk(sk);
 	prior_wstamp = tp->tcp_wstamp_ns;
 	tp->tcp_wstamp_ns = max(tp->tcp_wstamp_ns, tp->tcp_clock_cache);
-	skb->skb_mstamp_ns = tp->tcp_wstamp_ns;
+	skb_set_delivery_time(skb, tp->tcp_wstamp_ns, true);
 	if (clone_it) {
 		oskb = skb;
 
@@ -1589,7 +1589,7 @@ int tcp_fragment(struct sock *sk, enum tcp_queue tcp_queue,
 
 	skb_split(skb, buff, len);
 
-	buff->tstamp = skb->tstamp;
+	skb_set_delivery_time(buff, skb->tstamp, true);
 	tcp_fragment_tstamp(skb, buff);
 
 	old_factor = tcp_skb_pcount(skb);
@@ -2616,7 +2616,8 @@ static bool tcp_write_xmit(struct sock *sk, unsigned int mss_now, int nonagle,
 
 		if (unlikely(tp->repair) && tp->repair_queue == TCP_SEND_QUEUE) {
 			/* "skb_mstamp_ns" is used as a start point for the retransmit timer */
-			skb->skb_mstamp_ns = tp->tcp_wstamp_ns = tp->tcp_clock_cache;
+			tp->tcp_wstamp_ns = tp->tcp_clock_cache;
+			skb_set_delivery_time(skb, tp->tcp_wstamp_ns, true);
 			list_move_tail(&skb->tcp_tsorted_anchor, &tp->tsorted_sent_queue);
 			tcp_init_tso_segs(skb, mss_now);
 			goto repair; /* Skip network transmission */
@@ -3541,11 +3542,12 @@ struct sk_buff *tcp_make_synack(const struct sock *sk, struct dst_entry *dst,
 	now = tcp_clock_ns();
 #ifdef CONFIG_SYN_COOKIES
 	if (unlikely(synack_type == TCP_SYNACK_COOKIE && ireq->tstamp_ok))
-		skb->skb_mstamp_ns = cookie_init_timestamp(req, now);
+		skb_set_delivery_time(skb, cookie_init_timestamp(req, now),
+				      true);
 	else
 #endif
 	{
-		skb->skb_mstamp_ns = now;
+		skb_set_delivery_time(skb, now, true);
 		if (!tcp_rsk(req)->snt_synack) /* Timestamp first SYNACK */
 			tcp_rsk(req)->snt_synack = tcp_skb_timestamp_us(skb);
 	}
@@ -3594,7 +3596,7 @@ struct sk_buff *tcp_make_synack(const struct sock *sk, struct dst_entry *dst,
 	bpf_skops_write_hdr_opt((struct sock *)sk, skb, req, syn_skb,
 				synack_type, &opts);
 
-	skb->skb_mstamp_ns = now;
+	skb_set_delivery_time(skb, now, true);
 	tcp_add_tx_delay(skb, tp);
 
 	return skb;
@@ -3771,7 +3773,7 @@ static int tcp_send_syn_data(struct sock *sk, struct sk_buff *syn)
 
 	err = tcp_transmit_skb(sk, syn_data, 1, sk->sk_allocation);
 
-	syn->skb_mstamp_ns = syn_data->skb_mstamp_ns;
+	skb_set_delivery_time(syn, syn_data->skb_mstamp_ns, true);
 
 	/* Now full SYN+DATA was cloned and sent (or not),
 	 * remove the SYN from the original skb (syn_data)
diff --git a/net/ipv6/ioam6.c b/net/ipv6/ioam6.c
index e159eb4328a8..1098131ed90c 100644
--- a/net/ipv6/ioam6.c
+++ b/net/ipv6/ioam6.c
@@ -635,7 +635,8 @@ static void __ioam6_fill_trace_data(struct sk_buff *skb,
 				    struct ioam6_schema *sc,
 				    u8 sclen, bool is_input)
 {
-	struct __kernel_sock_timeval ts;
+	struct timespec64 ts;
+	ktime_t tstamp;
 	u64 raw64;
 	u32 raw32;
 	u16 raw16;
@@ -680,10 +681,9 @@ static void __ioam6_fill_trace_data(struct sk_buff *skb,
 		if (!skb->dev) {
 			*(__be32 *)data = cpu_to_be32(IOAM6_U32_UNAVAILABLE);
 		} else {
-			if (!skb->tstamp)
-				__net_timestamp(skb);
+			tstamp = skb_tstamp_cond(skb, true);
+			ts = ktime_to_timespec64(tstamp);
 
-			skb_get_new_timestamp(skb, &ts);
 			*(__be32 *)data = cpu_to_be32((u32)ts.tv_sec);
 		}
 		data += sizeof(__be32);
@@ -694,13 +694,12 @@ static void __ioam6_fill_trace_data(struct sk_buff *skb,
 		if (!skb->dev) {
 			*(__be32 *)data = cpu_to_be32(IOAM6_U32_UNAVAILABLE);
 		} else {
-			if (!skb->tstamp)
-				__net_timestamp(skb);
+			if (!trace->type.bit2) {
+				tstamp = skb_tstamp_cond(skb, true);
+				ts = ktime_to_timespec64(tstamp);
+			}
 
-			if (!trace->type.bit2)
-				skb_get_new_timestamp(skb, &ts);
-
-			*(__be32 *)data = cpu_to_be32((u32)ts.tv_usec);
+			*(__be32 *)data = cpu_to_be32((u32)(ts.tv_nsec / NSEC_PER_USEC));
 		}
 		data += sizeof(__be32);
 	}
diff --git a/net/ipv6/ip6_input.c b/net/ipv6/ip6_input.c
index d4b1e2c5aa76..5b5ea35635f9 100644
--- a/net/ipv6/ip6_input.c
+++ b/net/ipv6/ip6_input.c
@@ -459,6 +459,7 @@ discard:
 
 static int ip6_input_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
 {
+	skb_clear_delivery_time(skb);
 	rcu_read_lock();
 	ip6_protocol_deliver_rcu(net, skb, 0, false);
 	rcu_read_unlock();
diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index c5edc86b18bd..50db9b20d746 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -440,7 +440,7 @@ static inline int ip6_forward_finish(struct net *net, struct sock *sk,
 	}
 #endif
 
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 	return dst_output(net, sk, skb);
 }
 
@@ -813,6 +813,7 @@ int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
 	struct rt6_info *rt = (struct rt6_info *)skb_dst(skb);
 	struct ipv6_pinfo *np = skb->sk && !dev_recursion_level() ?
 				inet6_sk(skb->sk) : NULL;
+	bool mono_delivery_time = skb->mono_delivery_time;
 	struct ip6_frag_state state;
 	unsigned int mtu, hlen, nexthdr_offset;
 	ktime_t tstamp = skb->tstamp;
@@ -903,7 +904,7 @@ int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
 			if (iter.frag)
 				ip6_fraglist_prepare(skb, &iter);
 
-			skb->tstamp = tstamp;
+			skb_set_delivery_time(skb, tstamp, mono_delivery_time);
 			err = output(net, sk, skb);
 			if (!err)
 				IP6_INC_STATS(net, ip6_dst_idev(&rt->dst),
@@ -962,7 +963,7 @@ slow_path:
 		/*
 		 *	Put this fragment into the sending queue.
 		 */
-		frag->tstamp = tstamp;
+		skb_set_delivery_time(frag, tstamp, mono_delivery_time);
 		err = output(net, sk, frag);
 		if (err)
 			goto fail;
diff --git a/net/ipv6/netfilter.c b/net/ipv6/netfilter.c
index 6ab710b5a1a8..1da332450d98 100644
--- a/net/ipv6/netfilter.c
+++ b/net/ipv6/netfilter.c
@@ -121,6 +121,7 @@ int br_ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
 				  struct sk_buff *))
 {
 	int frag_max_size = BR_INPUT_SKB_CB(skb)->frag_max_size;
+	bool mono_delivery_time = skb->mono_delivery_time;
 	ktime_t tstamp = skb->tstamp;
 	struct ip6_frag_state state;
 	u8 *prevhdr, nexthdr = 0;
@@ -186,7 +187,7 @@ int br_ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
 			if (iter.frag)
 				ip6_fraglist_prepare(skb, &iter);
 
-			skb->tstamp = tstamp;
+			skb_set_delivery_time(skb, tstamp, mono_delivery_time);
 			err = output(net, sk, data, skb);
 			if (err || !iter.frag)
 				break;
@@ -219,7 +220,7 @@ slow_path:
 			goto blackhole;
 		}
 
-		skb2->tstamp = tstamp;
+		skb_set_delivery_time(skb2, tstamp, mono_delivery_time);
 		err = output(net, sk, data, skb2);
 		if (err)
 			goto blackhole;
diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c b/net/ipv6/netfilter/nf_conntrack_reasm.c
index 5c47be29b9ee..7dd3629dd19e 100644
--- a/net/ipv6/netfilter/nf_conntrack_reasm.c
+++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
@@ -264,6 +264,7 @@ static int nf_ct_frag6_queue(struct frag_queue *fq, struct sk_buff *skb,
 		fq->iif = dev->ifindex;
 
 	fq->q.stamp = skb->tstamp;
+	fq->q.mono_delivery_time = skb->mono_delivery_time;
 	fq->q.meat += skb->len;
 	fq->ecn |= ecn;
 	if (payload_len > fq->q.max_size)
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 28e44782c94d..ff866f2a879e 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -194,6 +194,7 @@ static int ip6_frag_queue(struct frag_queue *fq, struct sk_buff *skb,
 		fq->iif = dev->ifindex;
 
 	fq->q.stamp = skb->tstamp;
+	fq->q.mono_delivery_time = skb->mono_delivery_time;
 	fq->q.meat += skb->len;
 	fq->ecn |= ecn;
 	add_frag_mem_limit(fq->q.fqdir, skb->truesize);
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index e98af869ff3a..cb2bb7d2e907 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -940,7 +940,7 @@ static void tcp_v6_send_response(const struct sock *sk, struct sk_buff *skb, u32
 		} else {
 			mark = sk->sk_mark;
 		}
-		buff->tstamp = tcp_transmit_time(sk);
+		skb_set_delivery_time(buff, tcp_transmit_time(sk), true);
 	}
 	fl6.flowi6_mark = IP6_REPLY_MARK(net, skb->mark) ?: mark;
 	fl6.fl6_dport = t1->dest;
diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
index d2e5a8f644b8..029171379884 100644
--- a/net/netfilter/ipvs/ip_vs_xmit.c
+++ b/net/netfilter/ipvs/ip_vs_xmit.c
@@ -610,7 +610,7 @@ static inline int ip_vs_tunnel_xmit_prepare(struct sk_buff *skb,
 		nf_reset_ct(skb);
 		skb_forward_csum(skb);
 		if (skb->dev)
-			skb->tstamp = 0;
+			skb_clear_tstamp(skb);
 	}
 	return ret;
 }
@@ -652,7 +652,7 @@ static inline int ip_vs_nat_send_or_cont(int pf, struct sk_buff *skb,
 	if (!local) {
 		skb_forward_csum(skb);
 		if (skb->dev)
-			skb->tstamp = 0;
+			skb_clear_tstamp(skb);
 		NF_HOOK(pf, NF_INET_LOCAL_OUT, cp->ipvs->net, NULL, skb,
 			NULL, skb_dst(skb)->dev, dst_output);
 	} else
@@ -674,7 +674,7 @@ static inline int ip_vs_send_or_cont(int pf, struct sk_buff *skb,
 		ip_vs_drop_early_demux_sk(skb);
 		skb_forward_csum(skb);
 		if (skb->dev)
-			skb->tstamp = 0;
+			skb_clear_tstamp(skb);
 		NF_HOOK(pf, NF_INET_LOCAL_OUT, cp->ipvs->net, NULL, skb,
 			NULL, skb_dst(skb)->dev, dst_output);
 	} else
diff --git a/net/netfilter/nf_dup_netdev.c b/net/netfilter/nf_dup_netdev.c
index a579e59ee5c5..7873bd1389c3 100644
--- a/net/netfilter/nf_dup_netdev.c
+++ b/net/netfilter/nf_dup_netdev.c
@@ -19,7 +19,7 @@ static void nf_do_netdev_egress(struct sk_buff *skb, struct net_device *dev)
 		skb_push(skb, skb->mac_len);
 
 	skb->dev = dev;
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 	dev_queue_xmit(skb);
 }
 
diff --git a/net/netfilter/nf_flow_table_ip.c b/net/netfilter/nf_flow_table_ip.c
index 889cf88d3dba..f1d387129f02 100644
--- a/net/netfilter/nf_flow_table_ip.c
+++ b/net/netfilter/nf_flow_table_ip.c
@@ -376,7 +376,7 @@ nf_flow_offload_ip_hook(void *priv, struct sk_buff *skb,
 	nf_flow_nat_ip(flow, skb, thoff, dir, iph);
 
 	ip_decrease_ttl(iph);
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 
 	if (flow_table->flags & NF_FLOWTABLE_COUNTER)
 		nf_ct_acct_update(flow->ct, tuplehash->tuple.dir, skb->len);
@@ -611,7 +611,7 @@ nf_flow_offload_ipv6_hook(void *priv, struct sk_buff *skb,
 	nf_flow_nat_ipv6(flow, skb, dir, ip6h);
 
 	ip6h->hop_limit--;
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 
 	if (flow_table->flags & NF_FLOWTABLE_COUNTER)
 		nf_ct_acct_update(flow->ct, tuplehash->tuple.dir, skb->len);
diff --git a/net/netfilter/nfnetlink_log.c b/net/netfilter/nfnetlink_log.c
index ae9c0756bba5..d97eb280cb2e 100644
--- a/net/netfilter/nfnetlink_log.c
+++ b/net/netfilter/nfnetlink_log.c
@@ -460,6 +460,7 @@ __build_packet_message(struct nfnl_log_net *log,
 	sk_buff_data_t old_tail = inst->skb->tail;
 	struct sock *sk;
 	const unsigned char *hwhdrp;
+	ktime_t tstamp;
 
 	nlh = nfnl_msg_put(inst->skb, 0, 0,
 			   nfnl_msg_type(NFNL_SUBSYS_ULOG, NFULNL_MSG_PACKET),
@@ -588,9 +589,10 @@ __build_packet_message(struct nfnl_log_net *log,
 			goto nla_put_failure;
 	}
 
-	if (hooknum <= NF_INET_FORWARD && skb->tstamp) {
+	tstamp = skb_tstamp_cond(skb, false);
+	if (hooknum <= NF_INET_FORWARD && tstamp) {
 		struct nfulnl_msg_packet_timestamp ts;
-		struct timespec64 kts = ktime_to_timespec64(skb->tstamp);
+		struct timespec64 kts = ktime_to_timespec64(tstamp);
 		ts.sec = cpu_to_be64(kts.tv_sec);
 		ts.usec = cpu_to_be64(kts.tv_nsec / NSEC_PER_USEC);
 
diff --git a/net/netfilter/nfnetlink_queue.c b/net/netfilter/nfnetlink_queue.c
index 8c15978d9258..db9b5357f2ca 100644
--- a/net/netfilter/nfnetlink_queue.c
+++ b/net/netfilter/nfnetlink_queue.c
@@ -392,6 +392,7 @@ nfqnl_build_packet_message(struct net *net, struct nfqnl_instance *queue,
 	bool csum_verify;
 	char *secdata = NULL;
 	u32 seclen = 0;
+	ktime_t tstamp;
 
 	size = nlmsg_total_size(sizeof(struct nfgenmsg))
 		+ nla_total_size(sizeof(struct nfqnl_msg_packet_hdr))
@@ -407,7 +408,8 @@ nfqnl_build_packet_message(struct net *net, struct nfqnl_instance *queue,
 		+ nla_total_size(sizeof(u_int32_t))	/* skbinfo */
 		+ nla_total_size(sizeof(u_int32_t));	/* cap_len */
 
-	if (entskb->tstamp)
+	tstamp = skb_tstamp_cond(entskb, false);
+	if (tstamp)
 		size += nla_total_size(sizeof(struct nfqnl_msg_packet_timestamp));
 
 	size += nfqnl_get_bridge_size(entry);
@@ -582,9 +584,9 @@ nfqnl_build_packet_message(struct net *net, struct nfqnl_instance *queue,
 	if (nfqnl_put_bridge(entry, skb) < 0)
 		goto nla_put_failure;
 
-	if (entry->state.hook <= NF_INET_FORWARD && entskb->tstamp) {
+	if (entry->state.hook <= NF_INET_FORWARD && tstamp) {
 		struct nfqnl_msg_packet_timestamp ts;
-		struct timespec64 kts = ktime_to_timespec64(entskb->tstamp);
+		struct timespec64 kts = ktime_to_timespec64(tstamp);
 
 		ts.sec = cpu_to_be64(kts.tv_sec);
 		ts.usec = cpu_to_be64(kts.tv_nsec / NSEC_PER_USEC);
diff --git a/net/netfilter/nft_fwd_netdev.c b/net/netfilter/nft_fwd_netdev.c
index 619e394a91de..08e7a289738e 100644
--- a/net/netfilter/nft_fwd_netdev.c
+++ b/net/netfilter/nft_fwd_netdev.c
@@ -145,7 +145,7 @@ static void nft_fwd_neigh_eval(const struct nft_expr *expr,
 		return;
 
 	skb->dev = dev;
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 	neigh_xmit(neigh_table, dev, addr, skb);
 out:
 	regs->verdict.code = verdict;
diff --git a/net/openvswitch/vport.c b/net/openvswitch/vport.c
index cf2ce5812489..82a74f998966 100644
--- a/net/openvswitch/vport.c
+++ b/net/openvswitch/vport.c
@@ -507,7 +507,7 @@ void ovs_vport_send(struct vport *vport, struct sk_buff *skb, u8 mac_proto)
 	}
 
 	skb->dev = vport->dev;
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 	vport->ops->send(skb);
 	return;
 
diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index ab87f22cc7ec..1b93ce1a5600 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -460,7 +460,7 @@ static __u32 tpacket_get_timestamp(struct sk_buff *skb, struct timespec64 *ts,
 		return TP_STATUS_TS_RAW_HARDWARE;
 
 	if ((flags & SOF_TIMESTAMPING_SOFTWARE) &&
-	    ktime_to_timespec64_cond(skb->tstamp, ts))
+	    ktime_to_timespec64_cond(skb_tstamp(skb), ts))
 		return TP_STATUS_TS_SOFTWARE;
 
 	return 0;
@@ -2199,6 +2199,7 @@ static int packet_rcv(struct sk_buff *skb, struct net_device *dev,
 	spin_lock(&sk->sk_receive_queue.lock);
 	po->stats.stats1.tp_packets++;
 	sock_skb_set_dropcount(sk, skb);
+	skb_clear_delivery_time(skb);
 	__skb_queue_tail(&sk->sk_receive_queue, skb);
 	spin_unlock(&sk->sk_receive_queue.lock);
 	sk->sk_data_ready(sk);
@@ -2377,6 +2378,7 @@ static int tpacket_rcv(struct sk_buff *skb, struct net_device *dev,
 	po->stats.stats1.tp_packets++;
 	if (copy_skb) {
 		status |= TP_STATUS_COPY;
+		skb_clear_delivery_time(copy_skb);
 		__skb_queue_tail(&sk->sk_receive_queue, copy_skb);
 	}
 	spin_unlock(&sk->sk_receive_queue.lock);
diff --git a/net/sched/act_bpf.c b/net/sched/act_bpf.c
index a77d8908e737..fea2d78b9ddc 100644
--- a/net/sched/act_bpf.c
+++ b/net/sched/act_bpf.c
@@ -53,6 +53,8 @@ static int tcf_bpf_act(struct sk_buff *skb, const struct tc_action *act,
 		bpf_compute_data_pointers(skb);
 		filter_res = bpf_prog_run(filter, skb);
 	}
+	if (unlikely(!skb->tstamp && skb->mono_delivery_time))
+		skb->mono_delivery_time = 0;
 	if (skb_sk_is_prefetched(skb) && filter_res != TC_ACT_OK)
 		skb_orphan(skb);
 
diff --git a/net/sched/cls_bpf.c b/net/sched/cls_bpf.c
index df19a847829e..c85b85a192bf 100644
--- a/net/sched/cls_bpf.c
+++ b/net/sched/cls_bpf.c
@@ -102,6 +102,8 @@ static int cls_bpf_classify(struct sk_buff *skb, const struct tcf_proto *tp,
 			bpf_compute_data_pointers(skb);
 			filter_res = bpf_prog_run(prog->filter, skb);
 		}
+		if (unlikely(!skb->tstamp && skb->mono_delivery_time))
+			skb->mono_delivery_time = 0;
 
 		if (prog->exts_integrated) {
 			res->class   = 0;
diff --git a/net/xfrm/xfrm_interface.c b/net/xfrm/xfrm_interface.c
index 57448fc519fc..4991e99ced9a 100644
--- a/net/xfrm/xfrm_interface.c
+++ b/net/xfrm/xfrm_interface.c
@@ -190,7 +190,7 @@ static void xfrmi_dev_uninit(struct net_device *dev)
 
 static void xfrmi_scrub_packet(struct sk_buff *skb, bool xnet)
 {
-	skb->tstamp = 0;
+	skb_clear_tstamp(skb);
 	skb->pkt_type = PACKET_HOST;
 	skb->skb_iif = 0;
 	skb->ignore_df = 0;
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index afe3d0d7f5f2..4eebea830613 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -5086,6 +5086,37 @@ union bpf_attr {
  *	Return
  *		0 on success, or a negative error in case of failure. On error
  *		*dst* buffer is zeroed out.
+ *
+ * long bpf_skb_set_delivery_time(struct sk_buff *skb, u64 dtime, u32 dtime_type)
+ *	Description
+ *		Set a *dtime* (delivery time) to the __sk_buff->tstamp and also
+ *		change the __sk_buff->delivery_time_type to *dtime_type*.
+ *
+ *		When setting a delivery time (non zero *dtime*) to
+ *		__sk_buff->tstamp, only BPF_SKB_DELIVERY_TIME_MONO *dtime_type*
+ *		is supported.  It is the only delivery_time_type that will be
+ *		kept after bpf_redirect_*().
+ *
+ *		If there is no need to change the __sk_buff->delivery_time_type,
+ *		the delivery time can be directly written to __sk_buff->tstamp
+ *		instead.
+ *
+ *		*dtime* 0 and *dtime_type* BPF_SKB_DELIVERY_TIME_NONE
+ *		can be used to clear any delivery time stored in
+ *		__sk_buff->tstamp.
+ *
+ *		Only IPv4 and IPv6 skb->protocol are supported.
+ *
+ *		This function is most useful when it needs to set a
+ *		mono delivery time to __sk_buff->tstamp and then
+ *		bpf_redirect_*() to the egress of an iface.  For example,
+ *		changing the (rcv) timestamp in __sk_buff->tstamp at
+ *		ingress to a mono delivery time and then bpf_redirect_*()
+ *		to sch_fq@phy-dev.
+ *	Return
+ *		0 on success.
+ *		**-EINVAL** for invalid input
+ *		**-EOPNOTSUPP** for unsupported delivery_time_type and protocol
  */
 #define __BPF_FUNC_MAPPER(FN)		\
 	FN(unspec),			\
@@ -5280,6 +5311,7 @@ union bpf_attr {
 	FN(xdp_load_bytes),		\
 	FN(xdp_store_bytes),		\
 	FN(copy_from_user_task),	\
+	FN(skb_set_delivery_time),      \
 	/* */
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper
@@ -5469,6 +5501,12 @@ union {					\
 	__u64 :64;			\
 } __attribute__((aligned(8)))
 
+enum {
+	BPF_SKB_DELIVERY_TIME_NONE,
+	BPF_SKB_DELIVERY_TIME_UNSPEC,
+	BPF_SKB_DELIVERY_TIME_MONO,
+};
+
 /* user accessible mirror of in-kernel sk_buff.
  * new fields can only be added to the end of this structure
  */
@@ -5509,7 +5547,8 @@ struct __sk_buff {
 	__u32 gso_segs;
 	__bpf_md_ptr(struct bpf_sock *, sk);
 	__u32 gso_size;
-	__u32 :32;		/* Padding, future use. */
+	__u8  delivery_time_type;
+	__u32 :24;		/* Padding, future use. */
 	__u64 hwtstamp;
 };
 
diff --git a/tools/testing/selftests/bpf/prog_tests/tc_redirect.c b/tools/testing/selftests/bpf/prog_tests/tc_redirect.c
index 647b0a833628..2b255e28ed26 100644
--- a/tools/testing/selftests/bpf/prog_tests/tc_redirect.c
+++ b/tools/testing/selftests/bpf/prog_tests/tc_redirect.c
@@ -17,6 +17,8 @@
 #include <linux/if_tun.h>
 #include <linux/limits.h>
 #include <linux/sysctl.h>
+#include <linux/time_types.h>
+#include <linux/net_tstamp.h>
 #include <sched.h>
 #include <stdbool.h>
 #include <stdio.h>
@@ -29,6 +31,11 @@
 #include "test_tc_neigh_fib.skel.h"
 #include "test_tc_neigh.skel.h"
 #include "test_tc_peer.skel.h"
+#include "test_tc_dtime.skel.h"
+
+#ifndef TCP_TX_DELAY
+#define TCP_TX_DELAY 37
+#endif
 
 #define NS_SRC "ns_src"
 #define NS_FWD "ns_fwd"
@@ -61,6 +68,7 @@
 #define CHK_PROG_PIN_FILE "/sys/fs/bpf/test_tc_chk"
 
 #define TIMEOUT_MILLIS 10000
+#define NSEC_PER_SEC 1000000000ULL
 
 #define log_err(MSG, ...) \
 	fprintf(stderr, "(%s:%d: errno: %s) " MSG "\n", \
@@ -440,6 +448,431 @@ static int set_forwarding(bool enable)
 	return 0;
 }
 
+static void rcv_tstamp(int fd, const char *expected, size_t s)
+{
+	struct __kernel_timespec pkt_ts = {};
+	char ctl[CMSG_SPACE(sizeof(pkt_ts))];
+	struct timespec now_ts;
+	struct msghdr msg = {};
+	__u64 now_ns, pkt_ns;
+	struct cmsghdr *cmsg;
+	struct iovec iov;
+	char data[32];
+	int ret;
+
+	iov.iov_base = data;
+	iov.iov_len = sizeof(data);
+	msg.msg_iov = &iov;
+	msg.msg_iovlen = 1;
+	msg.msg_control = &ctl;
+	msg.msg_controllen = sizeof(ctl);
+
+	ret = recvmsg(fd, &msg, 0);
+	if (!ASSERT_EQ(ret, s, "recvmsg"))
+		return;
+	ASSERT_STRNEQ(data, expected, s, "expected rcv data");
+
+	cmsg = CMSG_FIRSTHDR(&msg);
+	if (cmsg && cmsg->cmsg_level == SOL_SOCKET &&
+	    cmsg->cmsg_type == SO_TIMESTAMPNS_NEW)
+		memcpy(&pkt_ts, CMSG_DATA(cmsg), sizeof(pkt_ts));
+
+	pkt_ns = pkt_ts.tv_sec * NSEC_PER_SEC + pkt_ts.tv_nsec;
+	ASSERT_NEQ(pkt_ns, 0, "pkt rcv tstamp");
+
+	ret = clock_gettime(CLOCK_REALTIME, &now_ts);
+	ASSERT_OK(ret, "clock_gettime");
+	now_ns = now_ts.tv_sec * NSEC_PER_SEC + now_ts.tv_nsec;
+
+	if (ASSERT_GE(now_ns, pkt_ns, "check rcv tstamp"))
+		ASSERT_LT(now_ns - pkt_ns, 5 * NSEC_PER_SEC,
+			  "check rcv tstamp");
+}
+
+static void snd_tstamp(int fd, char *b, size_t s)
+{
+	struct sock_txtime opt = { .clockid = CLOCK_TAI };
+	char ctl[CMSG_SPACE(sizeof(__u64))];
+	struct timespec now_ts;
+	struct msghdr msg = {};
+	struct cmsghdr *cmsg;
+	struct iovec iov;
+	__u64 now_ns;
+	int ret;
+
+	ret = clock_gettime(CLOCK_TAI, &now_ts);
+	ASSERT_OK(ret, "clock_get_time(CLOCK_TAI)");
+	now_ns = now_ts.tv_sec * NSEC_PER_SEC + now_ts.tv_nsec;
+
+	iov.iov_base = b;
+	iov.iov_len = s;
+	msg.msg_iov = &iov;
+	msg.msg_iovlen = 1;
+	msg.msg_control = &ctl;
+	msg.msg_controllen = sizeof(ctl);
+
+	cmsg = CMSG_FIRSTHDR(&msg);
+	cmsg->cmsg_level = SOL_SOCKET;
+	cmsg->cmsg_type = SCM_TXTIME;
+	cmsg->cmsg_len = CMSG_LEN(sizeof(now_ns));
+	*(__u64 *)CMSG_DATA(cmsg) = now_ns;
+
+	ret = setsockopt(fd, SOL_SOCKET, SO_TXTIME, &opt, sizeof(opt));
+	ASSERT_OK(ret, "setsockopt(SO_TXTIME)");
+
+	ret = sendmsg(fd, &msg, 0);
+	ASSERT_EQ(ret, s, "sendmsg");
+}
+
+static void test_inet_dtime(int family, int type, const char *addr, __u16 port)
+{
+	int opt = 1, accept_fd = -1, client_fd = -1, listen_fd, err;
+	char buf[] = "testing testing";
+	struct nstoken *nstoken;
+
+	nstoken = open_netns(NS_DST);
+	if (!ASSERT_OK_PTR(nstoken, "setns dst"))
+		return;
+	listen_fd = start_server(family, type, addr, port, 0);
+	close_netns(nstoken);
+
+	if (!ASSERT_GE(listen_fd, 0, "listen"))
+		return;
+
+	/* Ensure the kernel puts the (rcv) timestamp for all skb */
+	err = setsockopt(listen_fd, SOL_SOCKET, SO_TIMESTAMPNS_NEW,
+			 &opt, sizeof(opt));
+	if (!ASSERT_OK(err, "setsockopt(SO_TIMESTAMPNS_NEW)"))
+		goto done;
+
+	if (type == SOCK_STREAM) {
+		/* Ensure the kernel set EDT when sending out rst/ack
+		 * from the kernel's ctl_sk.
+		 */
+		err = setsockopt(listen_fd, SOL_TCP, TCP_TX_DELAY, &opt,
+				 sizeof(opt));
+		if (!ASSERT_OK(err, "setsockopt(TCP_TX_DELAY)"))
+			goto done;
+	}
+
+	nstoken = open_netns(NS_SRC);
+	if (!ASSERT_OK_PTR(nstoken, "setns src"))
+		goto done;
+	client_fd = connect_to_fd(listen_fd, TIMEOUT_MILLIS);
+	close_netns(nstoken);
+
+	if (!ASSERT_GE(client_fd, 0, "connect_to_fd"))
+		goto done;
+
+	if (type == SOCK_STREAM) {
+		int n;
+
+		accept_fd = accept(listen_fd, NULL, NULL);
+		if (!ASSERT_GE(accept_fd, 0, "accept"))
+			goto done;
+
+		n = write(client_fd, buf, sizeof(buf));
+		if (!ASSERT_EQ(n, sizeof(buf), "send to server"))
+			goto done;
+		rcv_tstamp(accept_fd, buf, sizeof(buf));
+	} else {
+		snd_tstamp(client_fd, buf, sizeof(buf));
+		rcv_tstamp(listen_fd, buf, sizeof(buf));
+	}
+
+done:
+	close(listen_fd);
+	if (accept_fd != -1)
+		close(accept_fd);
+	if (client_fd != -1)
+		close(client_fd);
+}
+
+static int netns_load_dtime_bpf(struct test_tc_dtime *skel)
+{
+	struct nstoken *nstoken;
+
+#define PIN_FNAME(__file) "/sys/fs/bpf/" #__file
+#define PIN(__prog) ({							\
+		int err = bpf_program__pin(skel->progs.__prog, PIN_FNAME(__prog)); \
+		if (!ASSERT_OK(err, "pin " #__prog))		\
+			goto fail;					\
+		})
+
+	/* setup ns_src tc progs */
+	nstoken = open_netns(NS_SRC);
+	if (!ASSERT_OK_PTR(nstoken, "setns " NS_SRC))
+		return -1;
+	PIN(egress_host);
+	PIN(ingress_host);
+	SYS("tc qdisc add dev veth_src clsact");
+	SYS("tc filter add dev veth_src ingress bpf da object-pinned "
+	    PIN_FNAME(ingress_host));
+	SYS("tc filter add dev veth_src egress bpf da object-pinned "
+	    PIN_FNAME(egress_host));
+	close_netns(nstoken);
+
+	/* setup ns_dst tc progs */
+	nstoken = open_netns(NS_DST);
+	if (!ASSERT_OK_PTR(nstoken, "setns " NS_DST))
+		return -1;
+	PIN(egress_host);
+	PIN(ingress_host);
+	SYS("tc qdisc add dev veth_dst clsact");
+	SYS("tc filter add dev veth_dst ingress bpf da object-pinned "
+	    PIN_FNAME(ingress_host));
+	SYS("tc filter add dev veth_dst egress bpf da object-pinned "
+	    PIN_FNAME(egress_host));
+	close_netns(nstoken);
+
+	/* setup ns_fwd tc progs */
+	nstoken = open_netns(NS_FWD);
+	if (!ASSERT_OK_PTR(nstoken, "setns " NS_FWD))
+		return -1;
+	PIN(ingress_fwdns_prio100);
+	PIN(egress_fwdns_prio100);
+	PIN(ingress_fwdns_prio101);
+	PIN(egress_fwdns_prio101);
+	SYS("tc qdisc add dev veth_dst_fwd clsact");
+	SYS("tc filter add dev veth_dst_fwd ingress prio 100 bpf da object-pinned "
+	    PIN_FNAME(ingress_fwdns_prio100));
+	SYS("tc filter add dev veth_dst_fwd ingress prio 101 bpf da object-pinned "
+	    PIN_FNAME(ingress_fwdns_prio101));
+	SYS("tc filter add dev veth_dst_fwd egress prio 100 bpf da object-pinned "
+	    PIN_FNAME(egress_fwdns_prio100));
+	SYS("tc filter add dev veth_dst_fwd egress prio 101 bpf da object-pinned "
+	    PIN_FNAME(egress_fwdns_prio101));
+	SYS("tc qdisc add dev veth_src_fwd clsact");
+	SYS("tc filter add dev veth_src_fwd ingress prio 100 bpf da object-pinned "
+	    PIN_FNAME(ingress_fwdns_prio100));
+	SYS("tc filter add dev veth_src_fwd ingress prio 101 bpf da object-pinned "
+	    PIN_FNAME(ingress_fwdns_prio101));
+	SYS("tc filter add dev veth_src_fwd egress prio 100 bpf da object-pinned "
+	    PIN_FNAME(egress_fwdns_prio100));
+	SYS("tc filter add dev veth_src_fwd egress prio 101 bpf da object-pinned "
+	    PIN_FNAME(egress_fwdns_prio101));
+	close_netns(nstoken);
+
+#undef PIN
+
+	return 0;
+
+fail:
+	close_netns(nstoken);
+	return -1;
+}
+
+enum {
+	INGRESS_FWDNS_P100,
+	INGRESS_FWDNS_P101,
+	EGRESS_FWDNS_P100,
+	EGRESS_FWDNS_P101,
+	INGRESS_ENDHOST,
+	EGRESS_ENDHOST,
+	SET_DTIME,
+	__MAX_CNT,
+};
+
+const char *cnt_names[] = {
+	"ingress_fwdns_p100",
+	"ingress_fwdns_p101",
+	"egress_fwdns_p100",
+	"egress_fwdns_p101",
+	"ingress_endhost",
+	"egress_endhost",
+	"set_dtime",
+};
+
+enum {
+	TCP_IP6_CLEAR_DTIME,
+	TCP_IP4,
+	TCP_IP6,
+	UDP_IP4,
+	UDP_IP6,
+	TCP_IP4_RT_FWD,
+	TCP_IP6_RT_FWD,
+	UDP_IP4_RT_FWD,
+	UDP_IP6_RT_FWD,
+	UKN_TEST,
+	__NR_TESTS,
+};
+
+const char *test_names[] = {
+	"tcp ip6 clear dtime",
+	"tcp ip4",
+	"tcp ip6",
+	"udp ip4",
+	"udp ip6",
+	"tcp ip4 rt fwd",
+	"tcp ip6 rt fwd",
+	"udp ip4 rt fwd",
+	"udp ip6 rt fwd",
+};
+
+static const char *dtime_cnt_str(int test, int cnt)
+{
+	static char name[64];
+
+	snprintf(name, sizeof(name), "%s %s", test_names[test], cnt_names[cnt]);
+
+	return name;
+}
+
+static const char *dtime_err_str(int test, int cnt)
+{
+	static char name[64];
+
+	snprintf(name, sizeof(name), "%s %s errs", test_names[test],
+		 cnt_names[cnt]);
+
+	return name;
+}
+
+static void test_tcp_clear_dtime(struct test_tc_dtime *skel)
+{
+	int i, t = TCP_IP6_CLEAR_DTIME;
+	__u32 *dtimes = skel->bss->dtimes[t];
+	__u32 *errs = skel->bss->errs[t];
+
+	skel->bss->test = t;
+	test_inet_dtime(AF_INET6, SOCK_STREAM, IP6_DST, 0);
+
+	ASSERT_EQ(dtimes[INGRESS_FWDNS_P100], 0,
+		  dtime_cnt_str(t, INGRESS_FWDNS_P100));
+	ASSERT_EQ(dtimes[INGRESS_FWDNS_P101], 0,
+		  dtime_cnt_str(t, INGRESS_FWDNS_P101));
+	ASSERT_GT(dtimes[EGRESS_FWDNS_P100], 0,
+		  dtime_cnt_str(t, EGRESS_FWDNS_P100));
+	ASSERT_EQ(dtimes[EGRESS_FWDNS_P101], 0,
+		  dtime_cnt_str(t, EGRESS_FWDNS_P101));
+	ASSERT_GT(dtimes[EGRESS_ENDHOST], 0,
+		  dtime_cnt_str(t, EGRESS_ENDHOST));
+	ASSERT_GT(dtimes[INGRESS_ENDHOST], 0,
+		  dtime_cnt_str(t, INGRESS_ENDHOST));
+
+	for (i = INGRESS_FWDNS_P100; i < __MAX_CNT; i++)
+		ASSERT_EQ(errs[i], 0, dtime_err_str(t, i));
+}
+
+static void test_tcp_dtime(struct test_tc_dtime *skel, int family, bool bpf_fwd)
+{
+	__u32 *dtimes, *errs;
+	const char *addr;
+	int i, t;
+
+	if (family == AF_INET) {
+		t = bpf_fwd ? TCP_IP4 : TCP_IP4_RT_FWD;
+		addr = IP4_DST;
+	} else {
+		t = bpf_fwd ? TCP_IP6 : TCP_IP6_RT_FWD;
+		addr = IP6_DST;
+	}
+
+	dtimes = skel->bss->dtimes[t];
+	errs = skel->bss->errs[t];
+
+	skel->bss->test = t;
+	test_inet_dtime(family, SOCK_STREAM, addr, 0);
+
+	/* fwdns_prio100 prog does not read delivery_time_type, so
+	 * kernel puts the (rcv) timetamp in __sk_buff->tstamp
+	 */
+	ASSERT_EQ(dtimes[INGRESS_FWDNS_P100], 0,
+		  dtime_cnt_str(t, INGRESS_FWDNS_P100));
+	for (i = INGRESS_FWDNS_P101; i < SET_DTIME; i++)
+		ASSERT_GT(dtimes[i], 0, dtime_cnt_str(t, i));
+
+	for (i = INGRESS_FWDNS_P100; i < __MAX_CNT; i++)
+		ASSERT_EQ(errs[i], 0, dtime_err_str(t, i));
+}
+
+static void test_udp_dtime(struct test_tc_dtime *skel, int family, bool bpf_fwd)
+{
+	__u32 *dtimes, *errs;
+	const char *addr;
+	int i, t;
+
+	if (family == AF_INET) {
+		t = bpf_fwd ? UDP_IP4 : UDP_IP4_RT_FWD;
+		addr = IP4_DST;
+	} else {
+		t = bpf_fwd ? UDP_IP6 : UDP_IP6_RT_FWD;
+		addr = IP6_DST;
+	}
+
+	dtimes = skel->bss->dtimes[t];
+	errs = skel->bss->errs[t];
+
+	skel->bss->test = t;
+	test_inet_dtime(family, SOCK_DGRAM, addr, 0);
+
+	ASSERT_EQ(dtimes[INGRESS_FWDNS_P100], 0,
+		  dtime_cnt_str(t, INGRESS_FWDNS_P100));
+	/* non mono delivery time is not forwarded */
+	ASSERT_EQ(dtimes[INGRESS_FWDNS_P101], 0,
+		  dtime_cnt_str(t, INGRESS_FWDNS_P100));
+	for (i = EGRESS_FWDNS_P100; i < SET_DTIME; i++)
+		ASSERT_GT(dtimes[i], 0, dtime_cnt_str(t, i));
+
+	for (i = INGRESS_FWDNS_P100; i < __MAX_CNT; i++)
+		ASSERT_EQ(errs[i], 0, dtime_err_str(t, i));
+}
+
+static void test_tc_redirect_dtime(struct netns_setup_result *setup_result)
+{
+	struct test_tc_dtime *skel;
+	struct nstoken *nstoken;
+	int err;
+
+	skel = test_tc_dtime__open();
+	if (!ASSERT_OK_PTR(skel, "test_tc_dtime__open"))
+		return;
+
+	skel->rodata->IFINDEX_SRC = setup_result->ifindex_veth_src_fwd;
+	skel->rodata->IFINDEX_DST = setup_result->ifindex_veth_dst_fwd;
+
+	err = test_tc_dtime__load(skel);
+	if (!ASSERT_OK(err, "test_tc_dtime__load"))
+		goto done;
+
+	if (netns_load_dtime_bpf(skel))
+		goto done;
+
+	nstoken = open_netns(NS_FWD);
+	if (!ASSERT_OK_PTR(nstoken, "setns fwd"))
+		goto done;
+	err = set_forwarding(false);
+	close_netns(nstoken);
+	if (!ASSERT_OK(err, "disable forwarding"))
+		goto done;
+
+	test_tcp_clear_dtime(skel);
+
+	test_tcp_dtime(skel, AF_INET, true);
+	test_tcp_dtime(skel, AF_INET6, true);
+	test_udp_dtime(skel, AF_INET, true);
+	test_udp_dtime(skel, AF_INET6, true);
+
+	/* Test the kernel ip[6]_forward path instead
+	 * of bpf_redirect_neigh().
+	 */
+	nstoken = open_netns(NS_FWD);
+	if (!ASSERT_OK_PTR(nstoken, "setns fwd"))
+		goto done;
+	err = set_forwarding(true);
+	close_netns(nstoken);
+	if (!ASSERT_OK(err, "enable forwarding"))
+		goto done;
+
+	test_tcp_dtime(skel, AF_INET, false);
+	test_tcp_dtime(skel, AF_INET6, false);
+	test_udp_dtime(skel, AF_INET, false);
+	test_udp_dtime(skel, AF_INET6, false);
+
+done:
+	test_tc_dtime__destroy(skel);
+}
+
 static void test_tc_redirect_neigh_fib(struct netns_setup_result *setup_result)
 {
 	struct nstoken *nstoken = NULL;
@@ -787,6 +1220,7 @@ static void *test_tc_redirect_run_tests(void *arg)
 	RUN_TEST(tc_redirect_peer_l3);
 	RUN_TEST(tc_redirect_neigh);
 	RUN_TEST(tc_redirect_neigh_fib);
+	RUN_TEST(tc_redirect_dtime);
 	return NULL;
 }
 
diff --git a/tools/testing/selftests/bpf/progs/test_tc_dtime.c b/tools/testing/selftests/bpf/progs/test_tc_dtime.c
new file mode 100644
index 000000000000..9d9e8e17b8a0
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/test_tc_dtime.c
@@ -0,0 +1,349 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2022 Meta
+
+#include <stddef.h>
+#include <stdint.h>
+#include <stdbool.h>
+#include <linux/bpf.h>
+#include <linux/stddef.h>
+#include <linux/pkt_cls.h>
+#include <linux/if_ether.h>
+#include <linux/in.h>
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_endian.h>
+#include <sys/socket.h>
+
+/* veth_src --- veth_src_fwd --- veth_det_fwd --- veth_dst
+ *           |                                 |
+ *  ns_src   |              ns_fwd             |   ns_dst
+ *
+ * ns_src and ns_dst: ENDHOST namespace
+ *            ns_fwd: Fowarding namespace
+ */
+
+#define ctx_ptr(field)		(void *)(long)(field)
+
+#define ip4_src			__bpf_htonl(0xac100164) /* 172.16.1.100 */
+#define ip4_dst			__bpf_htonl(0xac100264) /* 172.16.2.100 */
+
+#define ip6_src			{ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \
+				  0x00, 0x01, 0xde, 0xad, 0xbe, 0xef, 0xca, 0xfe }
+#define ip6_dst			{ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, \
+				  0x00, 0x02, 0xde, 0xad, 0xbe, 0xef, 0xca, 0xfe }
+
+#define v6_equal(a, b)		(a.s6_addr32[0] == b.s6_addr32[0] && \
+				 a.s6_addr32[1] == b.s6_addr32[1] && \
+				 a.s6_addr32[2] == b.s6_addr32[2] && \
+				 a.s6_addr32[3] == b.s6_addr32[3])
+
+volatile const __u32 IFINDEX_SRC;
+volatile const __u32 IFINDEX_DST;
+
+#define EGRESS_ENDHOST_MAGIC	0x0b9fbeef
+#define INGRESS_FWDNS_MAGIC	0x1b9fbeef
+#define EGRESS_FWDNS_MAGIC	0x2b9fbeef
+
+enum {
+	INGRESS_FWDNS_P100,
+	INGRESS_FWDNS_P101,
+	EGRESS_FWDNS_P100,
+	EGRESS_FWDNS_P101,
+	INGRESS_ENDHOST,
+	EGRESS_ENDHOST,
+	SET_DTIME,
+	__MAX_CNT,
+};
+
+enum {
+	TCP_IP6_CLEAR_DTIME,
+	TCP_IP4,
+	TCP_IP6,
+	UDP_IP4,
+	UDP_IP6,
+	TCP_IP4_RT_FWD,
+	TCP_IP6_RT_FWD,
+	UDP_IP4_RT_FWD,
+	UDP_IP6_RT_FWD,
+	UKN_TEST,
+	__NR_TESTS,
+};
+
+enum {
+	SRC_NS = 1,
+	DST_NS,
+};
+
+__u32 dtimes[__NR_TESTS][__MAX_CNT] = {};
+__u32 errs[__NR_TESTS][__MAX_CNT] = {};
+__u32 test = 0;
+
+static void inc_dtimes(__u32 idx)
+{
+	if (test < __NR_TESTS)
+		dtimes[test][idx]++;
+	else
+		dtimes[UKN_TEST][idx]++;
+}
+
+static void inc_errs(__u32 idx)
+{
+	if (test < __NR_TESTS)
+		errs[test][idx]++;
+	else
+		errs[UKN_TEST][idx]++;
+}
+
+static int skb_proto(int type)
+{
+	return type & 0xff;
+}
+
+static int skb_ns(int type)
+{
+	return (type >> 8) & 0xff;
+}
+
+static bool fwdns_clear_dtime(void)
+{
+	return test == TCP_IP6_CLEAR_DTIME;
+}
+
+static bool bpf_fwd(void)
+{
+	return test < TCP_IP4_RT_FWD;
+}
+
+/* -1: parse error: TC_ACT_SHOT
+ *  0: not testing traffic: TC_ACT_OK
+ * >0: first byte is the inet_proto, second byte has the netns
+ *     of the sender
+ */
+static int skb_get_type(struct __sk_buff *skb)
+{
+	void *data_end = ctx_ptr(skb->data_end);
+	void *data = ctx_ptr(skb->data);
+	__u8 inet_proto = 0, ns = 0;
+	struct ipv6hdr *ip6h;
+	struct iphdr *iph;
+
+	switch (skb->protocol) {
+	case __bpf_htons(ETH_P_IP):
+		iph = data + sizeof(struct ethhdr);
+		if (iph + 1 > data_end)
+			return -1;
+		if (iph->saddr == ip4_src)
+			ns = SRC_NS;
+		else if (iph->saddr == ip4_dst)
+			ns = DST_NS;
+		inet_proto = iph->protocol;
+		break;
+	case __bpf_htons(ETH_P_IPV6):
+		ip6h = data + sizeof(struct ethhdr);
+		if (ip6h + 1 > data_end)
+			return -1;
+		if (v6_equal(ip6h->saddr, (struct in6_addr)ip6_src))
+			ns = SRC_NS;
+		else if (v6_equal(ip6h->saddr, (struct in6_addr)ip6_dst))
+			ns = DST_NS;
+		inet_proto = ip6h->nexthdr;
+		break;
+	default:
+		return 0;
+	}
+
+	if ((inet_proto != IPPROTO_TCP && inet_proto != IPPROTO_UDP) || !ns)
+		return 0;
+
+	return (ns << 8 | inet_proto);
+}
+
+/* format: direction@iface@netns
+ * egress@veth_(src|dst)@ns_(src|dst)
+ */
+SEC("tc")
+int egress_host(struct __sk_buff *skb)
+{
+	int skb_type;
+
+	skb_type = skb_get_type(skb);
+	if (skb_type == -1)
+		return TC_ACT_SHOT;
+	if (!skb_type)
+		return TC_ACT_OK;
+
+	if (skb_proto(skb_type) == IPPROTO_TCP) {
+		if (skb->delivery_time_type == BPF_SKB_DELIVERY_TIME_MONO &&
+		    skb->tstamp)
+			inc_dtimes(EGRESS_ENDHOST);
+		else
+			inc_errs(EGRESS_ENDHOST);
+	} else {
+		if (skb->delivery_time_type == BPF_SKB_DELIVERY_TIME_UNSPEC &&
+		    skb->tstamp)
+			inc_dtimes(EGRESS_ENDHOST);
+		else
+			inc_errs(EGRESS_ENDHOST);
+	}
+
+	skb->tstamp = EGRESS_ENDHOST_MAGIC;
+
+	return TC_ACT_OK;
+}
+
+/* ingress@veth_(src|dst)@ns_(src|dst) */
+SEC("tc")
+int ingress_host(struct __sk_buff *skb)
+{
+	int skb_type;
+
+	skb_type = skb_get_type(skb);
+	if (skb_type == -1)
+		return TC_ACT_SHOT;
+	if (!skb_type)
+		return TC_ACT_OK;
+
+	if (skb->delivery_time_type == BPF_SKB_DELIVERY_TIME_MONO &&
+	    skb->tstamp == EGRESS_FWDNS_MAGIC)
+		inc_dtimes(INGRESS_ENDHOST);
+	else
+		inc_errs(INGRESS_ENDHOST);
+
+	return TC_ACT_OK;
+}
+
+/* ingress@veth_(src|dst)_fwd@ns_fwd priority 100 */
+SEC("tc")
+int ingress_fwdns_prio100(struct __sk_buff *skb)
+{
+	int skb_type;
+
+	skb_type = skb_get_type(skb);
+	if (skb_type == -1)
+		return TC_ACT_SHOT;
+	if (!skb_type)
+		return TC_ACT_OK;
+
+	/* delivery_time is only available to the ingress
+	 * if the tc-bpf checks the skb->delivery_time_type.
+	 */
+	if (skb->tstamp == EGRESS_ENDHOST_MAGIC)
+		inc_errs(INGRESS_FWDNS_P100);
+
+	if (fwdns_clear_dtime())
+		skb->tstamp = 0;
+
+	return TC_ACT_UNSPEC;
+}
+
+/* egress@veth_(src|dst)_fwd@ns_fwd priority 100 */
+SEC("tc")
+int egress_fwdns_prio100(struct __sk_buff *skb)
+{
+	int skb_type;
+
+	skb_type = skb_get_type(skb);
+	if (skb_type == -1)
+		return TC_ACT_SHOT;
+	if (!skb_type)
+		return TC_ACT_OK;
+
+	/* delivery_time is always available to egress even
+	 * the tc-bpf did not use the delivery_time_type.
+	 */
+	if (skb->tstamp == INGRESS_FWDNS_MAGIC)
+		inc_dtimes(EGRESS_FWDNS_P100);
+	else
+		inc_errs(EGRESS_FWDNS_P100);
+
+	if (fwdns_clear_dtime())
+		skb->tstamp = 0;
+
+	return TC_ACT_UNSPEC;
+}
+
+/* ingress@veth_(src|dst)_fwd@ns_fwd priority 101 */
+SEC("tc")
+int ingress_fwdns_prio101(struct __sk_buff *skb)
+{
+	__u64 expected_dtime = EGRESS_ENDHOST_MAGIC;
+	int skb_type;
+
+	skb_type = skb_get_type(skb);
+	if (skb_type == -1 || !skb_type)
+		/* Should have handled in prio100 */
+		return TC_ACT_SHOT;
+
+	if (skb_proto(skb_type) == IPPROTO_UDP)
+		expected_dtime = 0;
+
+	if (skb->delivery_time_type) {
+		if (fwdns_clear_dtime() ||
+		    skb->delivery_time_type != BPF_SKB_DELIVERY_TIME_MONO ||
+		    skb->tstamp != expected_dtime)
+			inc_errs(INGRESS_FWDNS_P101);
+		else
+			inc_dtimes(INGRESS_FWDNS_P101);
+	} else {
+		if (!fwdns_clear_dtime() && expected_dtime)
+			inc_errs(INGRESS_FWDNS_P101);
+	}
+
+	if (skb->delivery_time_type == BPF_SKB_DELIVERY_TIME_MONO) {
+		skb->tstamp = INGRESS_FWDNS_MAGIC;
+	} else {
+		if (bpf_skb_set_delivery_time(skb, INGRESS_FWDNS_MAGIC,
+					      BPF_SKB_DELIVERY_TIME_MONO))
+			inc_errs(SET_DTIME);
+		if (!bpf_skb_set_delivery_time(skb, INGRESS_FWDNS_MAGIC,
+					       BPF_SKB_DELIVERY_TIME_UNSPEC))
+			inc_errs(SET_DTIME);
+	}
+
+	if (skb_ns(skb_type) == SRC_NS)
+		return bpf_fwd() ?
+			bpf_redirect_neigh(IFINDEX_DST, NULL, 0, 0) : TC_ACT_OK;
+	else
+		return bpf_fwd() ?
+			bpf_redirect_neigh(IFINDEX_SRC, NULL, 0, 0) : TC_ACT_OK;
+}
+
+/* egress@veth_(src|dst)_fwd@ns_fwd priority 101 */
+SEC("tc")
+int egress_fwdns_prio101(struct __sk_buff *skb)
+{
+	int skb_type;
+
+	skb_type = skb_get_type(skb);
+	if (skb_type == -1 || !skb_type)
+		/* Should have handled in prio100 */
+		return TC_ACT_SHOT;
+
+	if (skb->delivery_time_type) {
+		if (fwdns_clear_dtime() ||
+		    skb->delivery_time_type != BPF_SKB_DELIVERY_TIME_MONO ||
+		    skb->tstamp != INGRESS_FWDNS_MAGIC)
+			inc_errs(EGRESS_FWDNS_P101);
+		else
+			inc_dtimes(EGRESS_FWDNS_P101);
+	} else {
+		if (!fwdns_clear_dtime())
+			inc_errs(EGRESS_FWDNS_P101);
+	}
+
+	if (skb->delivery_time_type == BPF_SKB_DELIVERY_TIME_MONO) {
+		skb->tstamp = EGRESS_FWDNS_MAGIC;
+	} else {
+		if (bpf_skb_set_delivery_time(skb, EGRESS_FWDNS_MAGIC,
+					      BPF_SKB_DELIVERY_TIME_MONO))
+			inc_errs(SET_DTIME);
+		if (!bpf_skb_set_delivery_time(skb, EGRESS_FWDNS_MAGIC,
+					       BPF_SKB_DELIVERY_TIME_UNSPEC))
+			inc_errs(SET_DTIME);
+	}
+
+	return TC_ACT_OK;
+}
+
+char __license[] SEC("license") = "GPL";