Commit b1394967 authored by David S. Miller's avatar David S. Miller
Browse files

Merge branch 'tcp-second-round-for-EDT-conversion' 



Eric Dumazet says:

====================
tcp: second round for EDT conversion

First round of EDT patches left TCP stack in a non optimal state.

- High speed flows suffered from loss of performance, addressed
  by the first patch of this series.

- Second patch brings pacing to the current state of networking,
  since we now reach ~100 Gbit on a single TCP flow.

- Third patch implements a mitigation for scheduling delays,
  like the one we did in sch_fq in the past.

- Fourth patch removes one special case in sch_fq for ACK packets.

- Fifth patch removes a serious perfomance cost for TCP internal
  pacing. We should setup the high resolution timer only if
  really needed.

- Sixth patch fixes a typo in BBR.

- Last patch is one minor change in cdg congestion control.

Neal Cardwell also has a patch series fixing BBR after
EDT adoption.
====================

Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parents 1a3aea25 825e1c52

include/linux/tcp.h

+1 −0
Original line number Diff line number Diff line
@@ -249,6 +249,7 @@ struct tcp_sock { 
	u32	tlp_high_seq;	/* snd_nxt at the time of TLP retransmit. */



	u64	tcp_wstamp_ns;	/* departure time for next sent data packet */

	u64	tcp_clock_cache; /* cache last tcp_clock_ns() (see tcp_mstamp_refresh()) */



/* RTT measurement */

	u64	tcp_mstamp;	/* most recent packet received/sent */

include/net/sock.h

+2 −2
Original line number Diff line number Diff line
@@ -422,8 +422,8 @@ struct sock { 
	struct timer_list	sk_timer;

	__u32			sk_priority;

	__u32			sk_mark;

	u32			sk_pacing_rate; /* bytes per second */

	u32			sk_max_pacing_rate;

	unsigned long		sk_pacing_rate; /* bytes per second */

	unsigned long		sk_max_pacing_rate;

	struct page_frag	sk_frag;

	netdev_features_t	sk_route_caps;

	netdev_features_t	sk_route_nocaps;

net/core/filter.c

+2 −2
Original line number Diff line number Diff line
@@ -3927,8 +3927,8 @@ BPF_CALL_5(bpf_setsockopt, struct bpf_sock_ops_kern *, bpf_sock, 
			sk->sk_userlocks |= SOCK_SNDBUF_LOCK;

			sk->sk_sndbuf = max_t(int, val * 2, SOCK_MIN_SNDBUF);

			break;

		case SO_MAX_PACING_RATE:

			sk->sk_max_pacing_rate = val;

		case SO_MAX_PACING_RATE: /* 32bit version */

			sk->sk_max_pacing_rate = (val == ~0U) ? ~0UL : val;

			sk->sk_pacing_rate = min(sk->sk_pacing_rate,

						 sk->sk_max_pacing_rate);

			break;

net/core/sock.c

+5 −4
Original line number Diff line number Diff line
@@ -998,7 +998,7 @@ set_rcvbuf: 
			cmpxchg(&sk->sk_pacing_status,

				SK_PACING_NONE,

				SK_PACING_NEEDED);

		sk->sk_max_pacing_rate = val;

		sk->sk_max_pacing_rate = (val == ~0U) ? ~0UL : val;

		sk->sk_pacing_rate = min(sk->sk_pacing_rate,

					 sk->sk_max_pacing_rate);

		break;
@@ -1336,7 +1336,8 @@ int sock_getsockopt(struct socket *sock, int level, int optname, 
#endif



	case SO_MAX_PACING_RATE:

		v.val = sk->sk_max_pacing_rate;

		/* 32bit version */

		v.val = min_t(unsigned long, sk->sk_max_pacing_rate, ~0U);

		break;



	case SO_INCOMING_CPU:
@@ -2810,8 +2811,8 @@ void sock_init_data(struct socket *sock, struct sock *sk) 
	sk->sk_ll_usec		=	sysctl_net_busy_read;

#endif



	sk->sk_max_pacing_rate = ~0U;

	sk->sk_pacing_rate = ~0U;

	sk->sk_max_pacing_rate = ~0UL;

	sk->sk_pacing_rate = ~0UL;

	sk->sk_pacing_shift = 10;

	sk->sk_incoming_cpu = -1;

net/ipv4/tcp.c

+5 −5
Original line number Diff line number Diff line
@@ -3111,10 +3111,10 @@ void tcp_get_info(struct sock *sk, struct tcp_info *info) 
{

	const struct tcp_sock *tp = tcp_sk(sk); /* iff sk_type == SOCK_STREAM */

	const struct inet_connection_sock *icsk = inet_csk(sk);

	unsigned long rate;

	u32 now;

	u64 rate64;

	bool slow;

	u32 rate;



	memset(info, 0, sizeof(*info));

	if (sk->sk_type != SOCK_STREAM)
@@ -3124,11 +3124,11 @@ void tcp_get_info(struct sock *sk, struct tcp_info *info) 


	/* Report meaningful fields for all TCP states, including listeners */

	rate = READ_ONCE(sk->sk_pacing_rate);

	rate64 = rate != ~0U ? rate : ~0ULL;

	rate64 = (rate != ~0UL) ? rate : ~0ULL;

	info->tcpi_pacing_rate = rate64;



	rate = READ_ONCE(sk->sk_max_pacing_rate);

	rate64 = rate != ~0U ? rate : ~0ULL;

	rate64 = (rate != ~0UL) ? rate : ~0ULL;

	info->tcpi_max_pacing_rate = rate64;



	info->tcpi_reordering = tp->reordering;
@@ -3254,8 +3254,8 @@ struct sk_buff *tcp_get_timestamping_opt_stats(const struct sock *sk) 
	const struct tcp_sock *tp = tcp_sk(sk);

	struct sk_buff *stats;

	struct tcp_info info;

	unsigned long rate;

	u64 rate64;

	u32 rate;



	stats = alloc_skb(tcp_opt_stats_get_size(), GFP_ATOMIC);

	if (!stats)
@@ -3274,7 +3274,7 @@ struct sk_buff *tcp_get_timestamping_opt_stats(const struct sock *sk) 
			  tp->total_retrans, TCP_NLA_PAD);



	rate = READ_ONCE(sk->sk_pacing_rate);

	rate64 = rate != ~0U ? rate : ~0ULL;

	rate64 = (rate != ~0UL) ? rate : ~0ULL;

	nla_put_u64_64bit(stats, TCP_NLA_PACING_RATE, rate64, TCP_NLA_PAD);



	rate64 = tcp_compute_delivery_rate(tp);

CRA Git | Maintained and supported by SUSTech CRA and CCSE