net-tcp_bbr: v2: Remove BBRv2 tcp congestion control patchset

This reverts commit 187eba0...7c99a90

Signed-off-by: Alexandre Frade <kernel@xanmod.org>

include/linux/tcp.h

@@ -255,8 +255,7 @@ struct tcp_sock {
 	u8	compressed_ack;
 	u8	dup_ack_counter:2,
 		tlp_retrans:1,	/* TLP is a retransmission */
-		fast_ack_mode:2, /* which fast ack mode ? */
-		unused:3;
+		unused:5;
 	u32	chrono_start;	/* Start time in jiffies of a TCP chrono */
 	u32	chrono_stat[3];	/* Time in jiffies for chrono_stat stats */
 	u8	chrono_type:2,	/* current chronograph type */

include/net/inet_connection_sock.h

@@ -135,8 +135,7 @@ struct inet_connection_sock {
 	u32			  icsk_probes_tstamp;
 	u32			  icsk_user_timeout;
 
-/* XXX inflated by temporary internal debugging info */
-	u64			  icsk_ca_priv[216 / sizeof(u64)];
+	u64			  icsk_ca_priv[104 / sizeof(u64)];
 #define ICSK_CA_PRIV_SIZE	  sizeof_field(struct inet_connection_sock, icsk_ca_priv)
 };
 

include/net/tcp.h

@@ -375,7 +375,6 @@ static inline void tcp_dec_quickack_mode(struct sock *sk,
 #define	TCP_ECN_QUEUE_CWR	2
 #define	TCP_ECN_DEMAND_CWR	4
 #define	TCP_ECN_SEEN		8
-#define	TCP_ECN_ECT_PERMANENT	16
 
 enum tcp_tw_status {
 	TCP_TW_SUCCESS = 0,
@@ -825,11 +824,6 @@ static inline u32 tcp_stamp_us_delta(u64 t1, u64 t0)
 	return max_t(s64, t1 - t0, 0);
 }
 
-static inline u32 tcp_stamp32_us_delta(u32 t1, u32 t0)
-{
-	return max_t(s32, t1 - t0, 0);
-}
-
 static inline u32 tcp_skb_timestamp(const struct sk_buff *skb)
 {
 	return tcp_ns_to_ts(skb->skb_mstamp_ns);
@@ -905,14 +899,9 @@ struct tcp_skb_cb {
 			/* pkts S/ACKed so far upon tx of skb, incl retrans: */
 			__u32 delivered;
 			/* start of send pipeline phase */
-			u32 first_tx_mstamp;
+			u64 first_tx_mstamp;
 			/* when we reached the "delivered" count */
-			u32 delivered_mstamp;
-#define TCPCB_IN_FLIGHT_BITS 20
-#define TCPCB_IN_FLIGHT_MAX ((1U << TCPCB_IN_FLIGHT_BITS) - 1)
-			u32 in_flight:20,   /* packets in flight at transmit */
-			    unused2:12;
-			u32 lost;	/* packets lost so far upon tx of skb */
+			u64 delivered_mstamp;
 		} tx;   /* only used for outgoing skbs */
 		union {
 			struct inet_skb_parm	h4;
@@ -1038,11 +1027,7 @@ enum tcp_ca_ack_event_flags {
 #define TCP_CONG_NON_RESTRICTED 0x1
 /* Requires ECN/ECT set on all packets */
 #define TCP_CONG_NEEDS_ECN	0x2
-/* Wants notification of CE events (CA_EVENT_ECN_IS_CE, CA_EVENT_ECN_NO_CE). */
-#define TCP_CONG_WANTS_CE_EVENTS	0x4
-#define TCP_CONG_MASK	(TCP_CONG_NON_RESTRICTED | \
-			 TCP_CONG_NEEDS_ECN | \
-			 TCP_CONG_WANTS_CE_EVENTS)
+#define TCP_CONG_MASK	(TCP_CONG_NON_RESTRICTED | TCP_CONG_NEEDS_ECN)
 
 union tcp_cc_info;
 
@@ -1062,11 +1047,8 @@ struct ack_sample {
  */
 struct rate_sample {
 	u64  prior_mstamp; /* starting timestamp for interval */
-	u32  prior_lost;	/* tp->lost at "prior_mstamp" */
 	u32  prior_delivered;	/* tp->delivered at "prior_mstamp" */
 	u32  prior_delivered_ce;/* tp->delivered_ce at "prior_mstamp" */
-	u32 tx_in_flight;	/* packets in flight at starting timestamp */
-	s32  lost;		/* number of packets lost over interval */
 	s32  delivered;		/* number of packets delivered over interval */
 	s32  delivered_ce;	/* number of packets delivered w/ CE marks*/
 	long interval_us;	/* time for tp->delivered to incr "delivered" */
@@ -1080,7 +1062,6 @@ struct rate_sample {
 	bool is_app_limited;	/* is sample from packet with bubble in pipe? */
 	bool is_retrans;	/* is sample from retransmission? */
 	bool is_ack_delayed;	/* is this (likely) a delayed ACK? */
-	bool is_ece;		/* did this ACK have ECN marked? */
 };
 
 struct tcp_congestion_ops {
@@ -1104,11 +1085,8 @@ struct tcp_congestion_ops {
 	/* hook for packet ack accounting (optional) */
 	void (*pkts_acked)(struct sock *sk, const struct ack_sample *sample);
 
-	/* pick target number of segments per TSO/GSO skb (optional): */
-	u32 (*tso_segs)(struct sock *sk, unsigned int mss_now);
-
-	/* react to a specific lost skb (optional) */
-	void (*skb_marked_lost)(struct sock *sk, const struct sk_buff *skb);
+	/* override sysctl_tcp_min_tso_segs */
+	u32 (*min_tso_segs)(struct sock *sk);
 
 	/* call when packets are delivered to update cwnd and pacing rate,
 	 * after all the ca_state processing. (optional)
@@ -1174,14 +1152,6 @@ static inline char *tcp_ca_get_name_by_key(u32 key, char *buffer)
 }
 #endif
 
-static inline bool tcp_ca_wants_ce_events(const struct sock *sk)
-{
-	const struct inet_connection_sock *icsk = inet_csk(sk);
-
-	return icsk->icsk_ca_ops->flags & (TCP_CONG_NEEDS_ECN |
-					   TCP_CONG_WANTS_CE_EVENTS);
-}
-
 static inline bool tcp_ca_needs_ecn(const struct sock *sk)
 {
 	const struct inet_connection_sock *icsk = inet_csk(sk);
@@ -1201,7 +1171,6 @@ static inline void tcp_ca_event(struct sock *sk, const enum tcp_ca_event event)
 void tcp_set_ca_state(struct sock *sk, const u8 ca_state);
 
 /* From tcp_rate.c */
-void tcp_set_tx_in_flight(struct sock *sk, struct sk_buff *skb);
 void tcp_rate_skb_sent(struct sock *sk, struct sk_buff *skb);
 void tcp_rate_skb_delivered(struct sock *sk, struct sk_buff *skb,
 			    struct rate_sample *rs);

include/uapi/linux/inet_diag.h

@@ -231,42 +231,9 @@ struct tcp_bbr_info {
 	__u32	bbr_cwnd_gain;		/* cwnd gain shifted left 8 bits */
 };
 
-/* Phase as reported in netlink/ss stats. */
-enum tcp_bbr2_phase {
-	BBR2_PHASE_INVALID		= 0,
-	BBR2_PHASE_STARTUP		= 1,
-	BBR2_PHASE_DRAIN		= 2,
-	BBR2_PHASE_PROBE_RTT		= 3,
-	BBR2_PHASE_PROBE_BW_UP		= 4,
-	BBR2_PHASE_PROBE_BW_DOWN	= 5,
-	BBR2_PHASE_PROBE_BW_CRUISE	= 6,
-	BBR2_PHASE_PROBE_BW_REFILL	= 7
-};
-
-struct tcp_bbr2_info {
-	/* u64 bw: bandwidth (app throughput) estimate in Byte per sec: */
-	__u32	bbr_bw_lsb;		/* lower 32 bits of bw */
-	__u32	bbr_bw_msb;		/* upper 32 bits of bw */
-	__u32	bbr_min_rtt;		/* min-filtered RTT in uSec */
-	__u32	bbr_pacing_gain;	/* pacing gain shifted left 8 bits */
-	__u32	bbr_cwnd_gain;		/* cwnd gain shifted left 8 bits */
-	__u32	bbr_bw_hi_lsb;		/* lower 32 bits of bw_hi */
-	__u32	bbr_bw_hi_msb;		/* upper 32 bits of bw_hi */
-	__u32	bbr_bw_lo_lsb;		/* lower 32 bits of bw_lo */
-	__u32	bbr_bw_lo_msb;		/* upper 32 bits of bw_lo */
-	__u8	bbr_mode;		/* current bbr_mode in state machine */
-	__u8	bbr_phase;		/* current state machine phase */
-	__u8	unused1;		/* alignment padding; not used yet */
-	__u8	bbr_version;		/* MUST be at this offset in struct */
-	__u32	bbr_inflight_lo;	/* lower/short-term data volume bound */
-	__u32	bbr_inflight_hi;	/* higher/long-term data volume bound */
-	__u32	bbr_extra_acked;	/* max excess packets ACKed in epoch */
-};
-
 union tcp_cc_info {
 	struct tcpvegas_info	vegas;
 	struct tcp_dctcp_info	dctcp;
 	struct tcp_bbr_info	bbr;
-	struct tcp_bbr2_info	bbr2;
 };
 #endif /* _UAPI_INET_DIAG_H_ */

net/ipv4/Kconfig

@@ -678,24 +678,6 @@ config TCP_CONG_BBR
 	  AQM schemes that do not provide a delay signal. It requires the fq
 	  ("Fair Queue") pacing packet scheduler.
 
-config TCP_CONG_BBR2
-	tristate "BBR2 TCP"
-	default n
-	help
-
-	BBR2 TCP congestion control is a model-based congestion control
-	algorithm that aims to maximize network utilization, keep queues and
-	retransmit rates low, and to be able to coexist with Reno/CUBIC in
-	common scenarios. It builds an explicit model of the network path.  It
-	tolerates a targeted degree of random packet loss and delay that are
-	unrelated to congestion. It can operate over LAN, WAN, cellular, wifi,
-	or cable modem links, and can use DCTCP-L4S-style ECN signals.  It can
-	coexist with flows that use loss-based congestion control, and can
-	operate with shallow buffers, deep buffers, bufferbloat, policers, or
-	AQM schemes that do not provide a delay signal. It requires pacing,
-	using either TCP internal pacing or the fq ("Fair Queue") pacing packet
-	scheduler.
-
 choice
 	prompt "Default TCP congestion control"
 	default DEFAULT_CUBIC
@@ -733,9 +715,6 @@ choice
 	config DEFAULT_BBR
 		bool "BBR" if TCP_CONG_BBR=y
 
-	config DEFAULT_BBR2
-		bool "BBR2" if TCP_CONG_BBR2=y
-
 	config DEFAULT_RENO
 		bool "Reno"
 endchoice
@@ -760,7 +739,6 @@ config DEFAULT_TCP_CONG
 	default "dctcp" if DEFAULT_DCTCP
 	default "cdg" if DEFAULT_CDG
 	default "bbr" if DEFAULT_BBR
-	default "bbr2" if DEFAULT_BBR2
 	default "cubic"
 
 config TCP_MD5SIG

net/ipv4/Makefile

@@ -47,7 +47,6 @@ obj-$(CONFIG_INET_TCP_DIAG) += tcp_diag.o
 obj-$(CONFIG_INET_UDP_DIAG) += udp_diag.o
 obj-$(CONFIG_INET_RAW_DIAG) += raw_diag.o
 obj-$(CONFIG_TCP_CONG_BBR) += tcp_bbr.o
-obj-$(CONFIG_TCP_CONG_BBR2) += tcp_bbr2.o
 obj-$(CONFIG_TCP_CONG_BIC) += tcp_bic.o
 obj-$(CONFIG_TCP_CONG_CDG) += tcp_cdg.o
 obj-$(CONFIG_TCP_CONG_CUBIC) += tcp_cubic.o

net/ipv4/tcp.c

@@ -3192,7 +3192,6 @@ int tcp_disconnect(struct sock *sk, int flags)
 	tp->rx_opt.dsack = 0;
 	tp->rx_opt.num_sacks = 0;
 	tp->rcv_ooopack = 0;
-	tp->fast_ack_mode = 0;
 
 
 	/* Clean up fastopen related fields */

net/ipv4/tcp_bbr.c

@@ -294,40 +294,26 @@ static void bbr_set_pacing_rate(struct sock *sk, u32 bw, int gain)
 		sk->sk_pacing_rate = rate;
 }
 
+/* override sysctl_tcp_min_tso_segs */
 __bpf_kfunc static u32 bbr_min_tso_segs(struct sock *sk)
 {
 	return sk->sk_pacing_rate < (bbr_min_tso_rate >> 3) ? 1 : 2;
 }
 
-/* Return the number of segments BBR would like in a TSO/GSO skb, given
- * a particular max gso size as a constraint.
- */
-static u32 bbr_tso_segs_generic(struct sock *sk, unsigned int mss_now,
-				u32 gso_max_size)
-{
-	u32 segs;
-	u64 bytes;
-
-	/* Budget a TSO/GSO burst size allowance based on bw (pacing_rate). */
-	bytes = sk->sk_pacing_rate >> sk->sk_pacing_shift;
-
-	bytes = min_t(u32, bytes, gso_max_size - 1 - MAX_TCP_HEADER);
-	segs = max_t(u32, div_u64(bytes, mss_now), bbr_min_tso_segs(sk));
-	return segs;
-}
-
-/* Custom tcp_tso_autosize() for BBR, used at transmit time to cap skb size. */
-static u32  bbr_tso_segs(struct sock *sk, unsigned int mss_now)
-{
-	return bbr_tso_segs_generic(sk, mss_now, sk->sk_gso_max_size);
-}
-
-/* Like bbr_tso_segs(), using mss_cache, ignoring driver's sk_gso_max_size. */
 static u32 bbr_tso_segs_goal(struct sock *sk)
 {
 	struct tcp_sock *tp = tcp_sk(sk);
+	u32 segs, bytes;
+
+	/* Sort of tcp_tso_autosize() but ignoring
+	 * driver provided sk_gso_max_size.
+	 */
+	bytes = min_t(unsigned long,
+		      sk->sk_pacing_rate >> READ_ONCE(sk->sk_pacing_shift),
+		      GSO_LEGACY_MAX_SIZE - 1 - MAX_TCP_HEADER);
+	segs = max_t(u32, bytes / tp->mss_cache, bbr_min_tso_segs(sk));
 
-	return  bbr_tso_segs_generic(sk, tp->mss_cache, GSO_LEGACY_MAX_SIZE);
+	return min(segs, 0x7FU);
 }
 
 /* Save "last known good" cwnd so we can restore it after losses or PROBE_RTT */
@@ -1163,7 +1149,7 @@ static struct tcp_congestion_ops tcp_bbr_cong_ops __read_mostly = {
 	.undo_cwnd	= bbr_undo_cwnd,
 	.cwnd_event	= bbr_cwnd_event,
 	.ssthresh	= bbr_ssthresh,
-	.tso_segs	= bbr_tso_segs,
+	.min_tso_segs	= bbr_min_tso_segs,
 	.get_info	= bbr_get_info,
 	.set_state	= bbr_set_state,
 };