selftests/bpf: Add bpf_burst scheduler & test

This patch implements the burst BPF MPTCP scheduler, named bpf_burst,
which is the default scheduler in protocol.c. bpf_burst_get_send() uses
the same logic as mptcp_subflow_get_send() and bpf_burst_get_retrans
uses the same logic as mptcp_subflow_get_retrans().

Using MPTCP_SCHED_TEST macro to add a new test for this bpf_burst
scheduler, the arguments "1 1" means data has been sent on both net
devices. Run this test by RUN_MPTCP_TEST macro.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
Reviewed-by: Mat Martineau <martineau@kernel.org>
Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>

3 files changed, 209 insertions, 0 deletions

diff --git a/tools/testing/selftests/bpf/bpf_tcp_helpers.h b/tools/testing/selftests/bpf/bpf_tcp_helpers.h
index cd4b6e6f14f9da..2c71226b3631d4 100644
--- a/tools/testing/selftests/bpf/bpf_tcp_helpers.h
+++ b/tools/testing/selftests/bpf/bpf_tcp_helpers.h
@@ -36,6 +36,8 @@ enum sk_pacing {
 struct sock {
 	struct sock_common	__sk_common;
 #define sk_state		__sk_common.skc_state
+	int			sk_sndbuf;
+	int			sk_wmem_queued;
 	unsigned long		sk_pacing_rate;
 	__u32			sk_pacing_status; /* see enum sk_pacing */
 } __attribute__((preserve_access_index));
@@ -85,6 +87,7 @@ struct tcp_sock {
 	__u32	lsndtime;
 	__u32	prior_cwnd;
 	__u64	tcp_mstamp;	/* most recent packet received/sent */
+	__u32	write_seq;	/* Tail(+1) of data held in tcp send buffer */
 	bool	is_mptcp;
 } __attribute__((preserve_access_index));
 
@@ -234,7 +237,9 @@ extern void tcp_cong_avoid_ai(struct tcp_sock *tp, __u32 w, __u32 acked) __ksym;
 #define MPTCP_SUBFLOWS_MAX	8
 
 struct mptcp_subflow_context {
+	unsigned long avg_pacing_rate;
 	__u32	backup : 1;
+	__u8	stale_count;
 	struct	sock *tcp_sock;	    /* tcp sk backpointer */
 } __attribute__((preserve_access_index));
 
@@ -257,6 +262,8 @@ struct mptcp_sched_ops {
 struct mptcp_sock {
 	struct inet_connection_sock	sk;
 
+	__u64		snd_nxt;
+	int		snd_burst;
 	__u32		token;
 	struct sock	*first;
 	char		ca_name[TCP_CA_NAME_MAX];
diff --git a/tools/testing/selftests/bpf/prog_tests/mptcp.c b/tools/testing/selftests/bpf/prog_tests/mptcp.c
index 3964dda7900f65..d3cc3541dd4ffe 100644
--- a/tools/testing/selftests/bpf/prog_tests/mptcp.c
+++ b/tools/testing/selftests/bpf/prog_tests/mptcp.c
@@ -13,6 +13,7 @@
 #include "mptcp_bpf_bkup.skel.h"
 #include "mptcp_bpf_rr.skel.h"
 #include "mptcp_bpf_red.skel.h"
+#include "mptcp_bpf_burst.skel.h"
 
 #define NS_TEST "mptcp_ns"
 #define WITH_DATA	true
@@ -545,6 +546,7 @@ MPTCP_SCHED_TEST(first, WITH_DATA, WITHOUT_DATA);
 MPTCP_SCHED_TEST(bkup, WITH_DATA, WITHOUT_DATA);
 MPTCP_SCHED_TEST(rr, WITH_DATA, WITH_DATA);
 MPTCP_SCHED_TEST(red, WITH_DATA, WITH_DATA);
+MPTCP_SCHED_TEST(burst, WITH_DATA, WITH_DATA);
 
 #define RUN_MPTCP_TEST(suffix)					\
 do {								\
@@ -561,4 +563,5 @@ void test_mptcp(void)
 	RUN_MPTCP_TEST(bkup);
 	RUN_MPTCP_TEST(rr);
 	RUN_MPTCP_TEST(red);
+	RUN_MPTCP_TEST(burst);
 }
diff --git a/tools/testing/selftests/bpf/progs/mptcp_bpf_burst.c b/tools/testing/selftests/bpf/progs/mptcp_bpf_burst.c
new file mode 100644
index 00000000000000..b3c8115648667f
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/mptcp_bpf_burst.c
@@ -0,0 +1,199 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2023, SUSE. */
+
+#include <linux/bpf.h>
+#include <limits.h>
+#include "bpf_tcp_helpers.h"
+
+char _license[] SEC("license") = "GPL";
+
+#define MPTCP_SEND_BURST_SIZE	65428
+
+struct subflow_send_info {
+	__u8 subflow_id;
+	__u64 linger_time;
+};
+
+extern bool mptcp_subflow_active(struct mptcp_subflow_context *subflow) __ksym;
+extern void mptcp_set_timeout(struct sock *sk) __ksym;
+extern __u64 mptcp_wnd_end(const struct mptcp_sock *msk) __ksym;
+extern bool tcp_stream_memory_free(const struct sock *sk, int wake) __ksym;
+extern bool bpf_mptcp_subflow_queues_empty(struct sock *sk) __ksym;
+extern void mptcp_pm_subflow_chk_stale(const struct mptcp_sock *msk, struct sock *ssk) __ksym;
+
+#define SSK_MODE_ACTIVE	0
+#define SSK_MODE_BACKUP	1
+#define SSK_MODE_MAX	2
+
+static __always_inline __u64 div_u64(__u64 dividend, __u32 divisor)
+{
+	return dividend / divisor;
+}
+
+static __always_inline bool tcp_write_queue_empty(struct sock *sk)
+{
+	const struct tcp_sock *tp = bpf_skc_to_tcp_sock(sk);
+
+	return tp ? tp->write_seq == tp->snd_nxt : true;
+}
+
+static __always_inline bool tcp_rtx_and_write_queues_empty(struct sock *sk)
+{
+	return bpf_mptcp_subflow_queues_empty(sk) && tcp_write_queue_empty(sk);
+}
+
+static __always_inline bool __sk_stream_memory_free(const struct sock *sk, int wake)
+{
+	if (sk->sk_wmem_queued >= sk->sk_sndbuf)
+		return false;
+
+	return tcp_stream_memory_free(sk, wake);
+}
+
+static __always_inline bool sk_stream_memory_free(const struct sock *sk)
+{
+	return __sk_stream_memory_free(sk, 0);
+}
+
+SEC("struct_ops/mptcp_sched_burst_init")
+void BPF_PROG(mptcp_sched_burst_init, struct mptcp_sock *msk)
+{
+}
+
+SEC("struct_ops/mptcp_sched_burst_release")
+void BPF_PROG(mptcp_sched_burst_release, struct mptcp_sock *msk)
+{
+}
+
+static int bpf_burst_get_send(struct mptcp_sock *msk,
+			      struct mptcp_sched_data *data)
+{
+	struct subflow_send_info send_info[SSK_MODE_MAX];
+	struct mptcp_subflow_context *subflow;
+	struct sock *sk = (struct sock *)msk;
+	__u32 pace, burst, wmem;
+	__u64 linger_time;
+	struct sock *ssk;
+	int i;
+
+	/* pick the subflow with the lower wmem/wspace ratio */
+	for (i = 0; i < SSK_MODE_MAX; ++i) {
+		send_info[i].subflow_id = MPTCP_SUBFLOWS_MAX;
+		send_info[i].linger_time = -1;
+	}
+
+	for (i = 0; i < data->subflows && i < MPTCP_SUBFLOWS_MAX; i++) {
+		subflow = bpf_mptcp_subflow_ctx_by_pos(data, i);
+		if (!subflow)
+			break;
+
+		ssk = mptcp_subflow_tcp_sock(subflow);
+		if (!mptcp_subflow_active(subflow))
+			continue;
+
+		pace = subflow->avg_pacing_rate;
+		if (!pace) {
+			/* init pacing rate from socket */
+			subflow->avg_pacing_rate = ssk->sk_pacing_rate;
+			pace = subflow->avg_pacing_rate;
+			if (!pace)
+				continue;
+		}
+
+		linger_time = div_u64((__u64)ssk->sk_wmem_queued << 32, pace);
+		if (linger_time < send_info[subflow->backup].linger_time) {
+			send_info[subflow->backup].subflow_id = i;
+			send_info[subflow->backup].linger_time = linger_time;
+		}
+	}
+	mptcp_set_timeout(sk);
+
+	/* pick the best backup if no other subflow is active */
+	if (send_info[SSK_MODE_ACTIVE].subflow_id == MPTCP_SUBFLOWS_MAX)
+		send_info[SSK_MODE_ACTIVE].subflow_id = send_info[SSK_MODE_BACKUP].subflow_id;
+
+	subflow = bpf_mptcp_subflow_ctx_by_pos(data, send_info[SSK_MODE_ACTIVE].subflow_id);
+	if (!subflow)
+		return -1;
+	ssk = mptcp_subflow_tcp_sock(subflow);
+	if (!ssk || !sk_stream_memory_free(ssk))
+		return -1;
+
+	burst = min(MPTCP_SEND_BURST_SIZE, mptcp_wnd_end(msk) - msk->snd_nxt);
+	wmem = ssk->sk_wmem_queued;
+	if (!burst)
+		goto out;
+
+	subflow->avg_pacing_rate = div_u64((__u64)subflow->avg_pacing_rate * wmem +
+					   ssk->sk_pacing_rate * burst,
+					   burst + wmem);
+	msk->snd_burst = burst;
+
+out:
+	mptcp_subflow_set_scheduled(subflow, true);
+	return 0;
+}
+
+static int bpf_burst_get_retrans(struct mptcp_sock *msk,
+				 struct mptcp_sched_data *data)
+{
+	int backup = MPTCP_SUBFLOWS_MAX, pick = MPTCP_SUBFLOWS_MAX, subflow_id;
+	struct mptcp_subflow_context *subflow;
+	int min_stale_count = INT_MAX;
+	struct sock *ssk;
+
+	for (int i = 0; i < data->subflows && i < MPTCP_SUBFLOWS_MAX; i++) {
+		subflow = bpf_mptcp_subflow_ctx_by_pos(data, i);
+		if (!subflow)
+			break;
+
+		if (!mptcp_subflow_active(subflow))
+			continue;
+
+		ssk = mptcp_subflow_tcp_sock(subflow);
+		/* still data outstanding at TCP level? skip this */
+		if (!tcp_rtx_and_write_queues_empty(ssk)) {
+			mptcp_pm_subflow_chk_stale(msk, ssk);
+			min_stale_count = min(min_stale_count, subflow->stale_count);
+			continue;
+		}
+
+		if (subflow->backup) {
+			if (backup == MPTCP_SUBFLOWS_MAX)
+				backup = i;
+			continue;
+		}
+
+		if (pick == MPTCP_SUBFLOWS_MAX)
+			pick = i;
+	}
+
+	if (pick < MPTCP_SUBFLOWS_MAX) {
+		subflow_id = pick;
+		goto out;
+	}
+	subflow_id = min_stale_count > 1 ? backup : MPTCP_SUBFLOWS_MAX;
+
+out:
+	subflow = bpf_mptcp_subflow_ctx_by_pos(data, subflow_id);
+	if (!subflow)
+		return -1;
+	mptcp_subflow_set_scheduled(subflow, true);
+	return 0;
+}
+
+int BPF_STRUCT_OPS(bpf_burst_get_subflow, struct mptcp_sock *msk,
+		   struct mptcp_sched_data *data)
+{
+	if (data->reinject)
+		return bpf_burst_get_retrans(msk, data);
+	return bpf_burst_get_send(msk, data);
+}
+
+SEC(".struct_ops")
+struct mptcp_sched_ops burst = {
+	.init		= (void *)mptcp_sched_burst_init,
+	.release	= (void *)mptcp_sched_burst_release,
+	.get_subflow	= (void *)bpf_burst_get_subflow,
+	.name		= "bpf_burst",
+};