Orange Pi5 kernel

^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   1) // SPDX-License-Identifier: GPL-2.0-only
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   2) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   3)  * TCP NV: TCP with Congestion Avoidance
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   4)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   5)  * TCP-NV is a successor of TCP-Vegas that has been developed to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   6)  * deal with the issues that occur in modern networks.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   7)  * Like TCP-Vegas, TCP-NV supports true congestion avoidance,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   8)  * the ability to detect congestion before packet losses occur.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300   9)  * When congestion (queue buildup) starts to occur, TCP-NV
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  10)  * predicts what the cwnd size should be for the current
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  11)  * throughput and it reduces the cwnd proportionally to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  12)  * the difference between the current cwnd and the predicted cwnd.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  13)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  14)  * NV is only recommeneded for traffic within a data center, and when
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  15)  * all the flows are NV (at least those within the data center). This
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  16)  * is due to the inherent unfairness between flows using losses to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  17)  * detect congestion (congestion control) and those that use queue
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  18)  * buildup to detect congestion (congestion avoidance).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  19)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  20)  * Note: High NIC coalescence values may lower the performance of NV
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  21)  * due to the increased noise in RTT values. In particular, we have
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  22)  * seen issues with rx-frames values greater than 8.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  23)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  24)  * TODO:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  25)  * 1) Add mechanism to deal with reverse congestion.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  26)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  27) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  28) #include <linux/mm.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  29) #include <linux/module.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  30) #include <linux/math64.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  31) #include <net/tcp.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  32) #include <linux/inet_diag.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  33) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  34) /* TCP NV parameters
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  35)  *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  36)  * nv_pad		Max number of queued packets allowed in network
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  37)  * nv_pad_buffer	Do not grow cwnd if this closed to nv_pad
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  38)  * nv_reset_period	How often (in) seconds)to reset min_rtt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  39)  * nv_min_cwnd		Don't decrease cwnd below this if there are no losses
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  40)  * nv_cong_dec_mult	Decrease cwnd by X% (30%) of congestion when detected
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  41)  * nv_ssthresh_factor	On congestion set ssthresh to this * <desired cwnd> / 8
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  42)  * nv_rtt_factor	RTT averaging factor
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  43)  * nv_loss_dec_factor	Decrease cwnd to this (80%) when losses occur
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  44)  * nv_dec_eval_min_calls	Wait this many RTT measurements before dec cwnd
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  45)  * nv_inc_eval_min_calls	Wait this many RTT measurements before inc cwnd
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  46)  * nv_ssthresh_eval_min_calls	Wait this many RTT measurements before stopping
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  47)  *				slow-start due to congestion
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  48)  * nv_stop_rtt_cnt	Only grow cwnd for this many RTTs after non-congestion
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  49)  * nv_rtt_min_cnt	Wait these many RTTs before making congesion decision
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  50)  * nv_cwnd_growth_rate_neg
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  51)  * nv_cwnd_growth_rate_pos
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  52)  *	How quickly to double growth rate (not rate) of cwnd when not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  53)  *	congested. One value (nv_cwnd_growth_rate_neg) for when
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  54)  *	rate < 1 pkt/RTT (after losses). The other (nv_cwnd_growth_rate_pos)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  55)  *	otherwise.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  56)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  57) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  58) static int nv_pad __read_mostly = 10;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  59) static int nv_pad_buffer __read_mostly = 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  60) static int nv_reset_period __read_mostly = 5; /* in seconds */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  61) static int nv_min_cwnd __read_mostly = 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  62) static int nv_cong_dec_mult __read_mostly = 30 * 128 / 100; /* = 30% */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  63) static int nv_ssthresh_factor __read_mostly = 8; /* = 1 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  64) static int nv_rtt_factor __read_mostly = 128; /* = 1/2*old + 1/2*new */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  65) static int nv_loss_dec_factor __read_mostly = 819; /* => 80% */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  66) static int nv_cwnd_growth_rate_neg __read_mostly = 8;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  67) static int nv_cwnd_growth_rate_pos __read_mostly; /* 0 => fixed like Reno */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  68) static int nv_dec_eval_min_calls __read_mostly = 60;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  69) static int nv_inc_eval_min_calls __read_mostly = 20;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  70) static int nv_ssthresh_eval_min_calls __read_mostly = 30;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  71) static int nv_stop_rtt_cnt __read_mostly = 10;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  72) static int nv_rtt_min_cnt __read_mostly = 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  73) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  74) module_param(nv_pad, int, 0644);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  75) MODULE_PARM_DESC(nv_pad, "max queued packets allowed in network");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  76) module_param(nv_reset_period, int, 0644);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  77) MODULE_PARM_DESC(nv_reset_period, "nv_min_rtt reset period (secs)");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  78) module_param(nv_min_cwnd, int, 0644);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  79) MODULE_PARM_DESC(nv_min_cwnd, "NV will not decrease cwnd below this value"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  80) 		 " without losses");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  81) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  82) /* TCP NV Parameters */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  83) struct tcpnv {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  84) 	unsigned long nv_min_rtt_reset_jiffies;  /* when to switch to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  85) 						  * nv_min_rtt_new */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  86) 	s8  cwnd_growth_factor;	/* Current cwnd growth factor,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  87) 				 * < 0 => less than 1 packet/RTT */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  88) 	u8  available8;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  89) 	u16 available16;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  90) 	u8  nv_allow_cwnd_growth:1, /* whether cwnd can grow */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  91) 		nv_reset:1,	    /* whether to reset values */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  92) 		nv_catchup:1;	    /* whether we are growing because
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  93) 				     * of temporary cwnd decrease */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  94) 	u8  nv_eval_call_cnt;	/* call count since last eval */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  95) 	u8  nv_min_cwnd;	/* nv won't make a ca decision if cwnd is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  96) 				 * smaller than this. It may grow to handle
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  97) 				 * TSO, LRO and interrupt coalescence because
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  98) 				 * with these a small cwnd cannot saturate
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300  99) 				 * the link. Note that this is different from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) 				 * the file local nv_min_cwnd */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) 	u8  nv_rtt_cnt;		/* RTTs without making ca decision */;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) 	u32 nv_last_rtt;	/* last rtt */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) 	u32 nv_min_rtt;		/* active min rtt. Used to determine slope */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) 	u32 nv_min_rtt_new;	/* min rtt for future use */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105) 	u32 nv_base_rtt;        /* If non-zero it represents the threshold for
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) 				 * congestion */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) 	u32 nv_lower_bound_rtt; /* Used in conjunction with nv_base_rtt. It is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) 				 * set to 80% of nv_base_rtt. It helps reduce
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109) 				 * unfairness between flows */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) 	u32 nv_rtt_max_rate;	/* max rate seen during current RTT */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) 	u32 nv_rtt_start_seq;	/* current RTT ends when packet arrives
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112) 				 * acking beyond nv_rtt_start_seq */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113) 	u32 nv_last_snd_una;	/* Previous value of tp->snd_una. It is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) 				 * used to determine bytes acked since last
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) 				 * call to bictcp_acked */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) 	u32 nv_no_cong_cnt;	/* Consecutive no congestion decisions */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119) #define NV_INIT_RTT	  U32_MAX
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) #define NV_MIN_CWND	  4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) #define NV_MIN_CWND_GROW  2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122) #define NV_TSO_CWND_BOUND 80
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) static inline void tcpnv_reset(struct tcpnv *ca, struct sock *sk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) 	struct tcp_sock *tp = tcp_sk(sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) 	ca->nv_reset = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) 	ca->nv_no_cong_cnt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) 	ca->nv_rtt_cnt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131) 	ca->nv_last_rtt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132) 	ca->nv_rtt_max_rate = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133) 	ca->nv_rtt_start_seq = tp->snd_una;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134) 	ca->nv_eval_call_cnt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) 	ca->nv_last_snd_una = tp->snd_una;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) static void tcpnv_init(struct sock *sk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140) 	struct tcpnv *ca = inet_csk_ca(sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141) 	int base_rtt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) 	tcpnv_reset(ca, sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145) 	/* See if base_rtt is available from socket_ops bpf program.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146) 	 * It is meant to be used in environments, such as communication
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) 	 * within a datacenter, where we have reasonable estimates of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) 	 * RTTs
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149) 	 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150) 	base_rtt = tcp_call_bpf(sk, BPF_SOCK_OPS_BASE_RTT, 0, NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) 	if (base_rtt > 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) 		ca->nv_base_rtt = base_rtt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153) 		ca->nv_lower_bound_rtt = (base_rtt * 205) >> 8; /* 80% */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) 	} else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) 		ca->nv_base_rtt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156) 		ca->nv_lower_bound_rtt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) 	ca->nv_allow_cwnd_growth = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160) 	ca->nv_min_rtt_reset_jiffies = jiffies + 2 * HZ;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161) 	ca->nv_min_rtt = NV_INIT_RTT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162) 	ca->nv_min_rtt_new = NV_INIT_RTT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) 	ca->nv_min_cwnd = NV_MIN_CWND;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164) 	ca->nv_catchup = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165) 	ca->cwnd_growth_factor = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) /* If provided, apply upper (base_rtt) and lower (lower_bound_rtt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169)  * bounds to RTT.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) inline u32 nv_get_bounded_rtt(struct tcpnv *ca, u32 val)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173) 	if (ca->nv_lower_bound_rtt > 0 && val < ca->nv_lower_bound_rtt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) 		return ca->nv_lower_bound_rtt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175) 	else if (ca->nv_base_rtt > 0 && val > ca->nv_base_rtt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176) 		return ca->nv_base_rtt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) 	else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) 		return val;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181) static void tcpnv_cong_avoid(struct sock *sk, u32 ack, u32 acked)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183) 	struct tcp_sock *tp = tcp_sk(sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) 	struct tcpnv *ca = inet_csk_ca(sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) 	u32 cnt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187) 	if (!tcp_is_cwnd_limited(sk))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188) 		return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190) 	/* Only grow cwnd if NV has not detected congestion */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) 	if (!ca->nv_allow_cwnd_growth)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) 		return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) 	if (tcp_in_slow_start(tp)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195) 		acked = tcp_slow_start(tp, acked);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) 		if (!acked)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) 			return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) 	if (ca->cwnd_growth_factor < 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201) 		cnt = tp->snd_cwnd << -ca->cwnd_growth_factor;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202) 		tcp_cong_avoid_ai(tp, cnt, acked);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203) 	} else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204) 		cnt = max(4U, tp->snd_cwnd >> ca->cwnd_growth_factor);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205) 		tcp_cong_avoid_ai(tp, cnt, acked);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) static u32 tcpnv_recalc_ssthresh(struct sock *sk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211) 	const struct tcp_sock *tp = tcp_sk(sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213) 	return max((tp->snd_cwnd * nv_loss_dec_factor) >> 10, 2U);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216) static void tcpnv_state(struct sock *sk, u8 new_state)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218) 	struct tcpnv *ca = inet_csk_ca(sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220) 	if (new_state == TCP_CA_Open && ca->nv_reset) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221) 		tcpnv_reset(ca, sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222) 	} else if (new_state == TCP_CA_Loss || new_state == TCP_CA_CWR ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223) 		new_state == TCP_CA_Recovery) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) 		ca->nv_reset = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225) 		ca->nv_allow_cwnd_growth = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226) 		if (new_state == TCP_CA_Loss) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227) 			/* Reset cwnd growth factor to Reno value */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228) 			if (ca->cwnd_growth_factor > 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229) 				ca->cwnd_growth_factor = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) 			/* Decrease growth rate if allowed */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231) 			if (nv_cwnd_growth_rate_neg > 0 &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) 			    ca->cwnd_growth_factor > -8)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) 				ca->cwnd_growth_factor--;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) 		}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) /* Do congestion avoidance calculations for TCP-NV
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239)  */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240) static void tcpnv_acked(struct sock *sk, const struct ack_sample *sample)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242) 	const struct inet_connection_sock *icsk = inet_csk(sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243) 	struct tcp_sock *tp = tcp_sk(sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244) 	struct tcpnv *ca = inet_csk_ca(sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245) 	unsigned long now = jiffies;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) 	u64 rate64;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) 	u32 rate, max_win, cwnd_by_slope;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248) 	u32 avg_rtt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249) 	u32 bytes_acked = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 251) 	/* Some calls are for duplicates without timetamps */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 252) 	if (sample->rtt_us < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 253) 		return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255) 	/* If not in TCP_CA_Open or TCP_CA_Disorder states, skip. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) 	if (icsk->icsk_ca_state != TCP_CA_Open &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) 	    icsk->icsk_ca_state != TCP_CA_Disorder)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258) 		return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260) 	/* Stop cwnd growth if we were in catch up mode */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261) 	if (ca->nv_catchup && tp->snd_cwnd >= nv_min_cwnd) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262) 		ca->nv_catchup = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263) 		ca->nv_allow_cwnd_growth = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266) 	bytes_acked = tp->snd_una - ca->nv_last_snd_una;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267) 	ca->nv_last_snd_una = tp->snd_una;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269) 	if (sample->in_flight == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) 		return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272) 	/* Calculate moving average of RTT */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273) 	if (nv_rtt_factor > 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) 		if (ca->nv_last_rtt > 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275) 			avg_rtt = (((u64)sample->rtt_us) * nv_rtt_factor +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) 				   ((u64)ca->nv_last_rtt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) 				   * (256 - nv_rtt_factor)) >> 8;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278) 		} else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) 			avg_rtt = sample->rtt_us;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280) 			ca->nv_min_rtt = avg_rtt << 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281) 		}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) 		ca->nv_last_rtt = avg_rtt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283) 	} else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284) 		avg_rtt = sample->rtt_us;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287) 	/* rate in 100's bits per second */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288) 	rate64 = ((u64)sample->in_flight) * 80000;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) 	do_div(rate64, avg_rtt ?: 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290) 	rate = (u32)rate64;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292) 	/* Remember the maximum rate seen during this RTT
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) 	 * Note: It may be more than one RTT. This function should be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294) 	 *       called at least nv_dec_eval_min_calls times.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295) 	 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296) 	if (ca->nv_rtt_max_rate < rate)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 297) 		ca->nv_rtt_max_rate = rate;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 298) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 299) 	/* We have valid information, increment counter */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 300) 	if (ca->nv_eval_call_cnt < 255)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 301) 		ca->nv_eval_call_cnt++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 302) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 303) 	/* Apply bounds to rtt. Only used to update min_rtt */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 304) 	avg_rtt = nv_get_bounded_rtt(ca, avg_rtt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 305) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 306) 	/* update min rtt if necessary */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 307) 	if (avg_rtt < ca->nv_min_rtt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 308) 		ca->nv_min_rtt = avg_rtt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 309) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 310) 	/* update future min_rtt if necessary */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 311) 	if (avg_rtt < ca->nv_min_rtt_new)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 312) 		ca->nv_min_rtt_new = avg_rtt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 313) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 314) 	/* nv_min_rtt is updated with the minimum (possibley averaged) rtt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 315) 	 * seen in the last sysctl_tcp_nv_reset_period seconds (i.e. a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 316) 	 * warm reset). This new nv_min_rtt will be continued to be updated
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 317) 	 * and be used for another sysctl_tcp_nv_reset_period seconds,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 318) 	 * when it will be updated again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 319) 	 * In practice we introduce some randomness, so the actual period used
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 320) 	 * is chosen randomly from the range:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 321) 	 *   [sysctl_tcp_nv_reset_period*3/4, sysctl_tcp_nv_reset_period*5/4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 322) 	 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 323) 	if (time_after_eq(now, ca->nv_min_rtt_reset_jiffies)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 324) 		unsigned char rand;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 325) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 326) 		ca->nv_min_rtt = ca->nv_min_rtt_new;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 327) 		ca->nv_min_rtt_new = NV_INIT_RTT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 328) 		get_random_bytes(&rand, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 329) 		ca->nv_min_rtt_reset_jiffies =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 330) 			now + ((nv_reset_period * (384 + rand) * HZ) >> 9);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 331) 		/* Every so often we decrease ca->nv_min_cwnd in case previous
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 332) 		 *  value is no longer accurate.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 333) 		 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 334) 		ca->nv_min_cwnd = max(ca->nv_min_cwnd / 2, NV_MIN_CWND);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 335) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 336) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 337) 	/* Once per RTT check if we need to do congestion avoidance */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 338) 	if (before(ca->nv_rtt_start_seq, tp->snd_una)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 339) 		ca->nv_rtt_start_seq = tp->snd_nxt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 340) 		if (ca->nv_rtt_cnt < 0xff)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 341) 			/* Increase counter for RTTs without CA decision */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 342) 			ca->nv_rtt_cnt++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 343) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 344) 		/* If this function is only called once within an RTT
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 345) 		 * the cwnd is probably too small (in some cases due to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 346) 		 * tso, lro or interrupt coalescence), so we increase
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 347) 		 * ca->nv_min_cwnd.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 348) 		 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 349) 		if (ca->nv_eval_call_cnt == 1 &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 350) 		    bytes_acked >= (ca->nv_min_cwnd - 1) * tp->mss_cache &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 351) 		    ca->nv_min_cwnd < (NV_TSO_CWND_BOUND + 1)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 352) 			ca->nv_min_cwnd = min(ca->nv_min_cwnd
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 353) 					      + NV_MIN_CWND_GROW,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 354) 					      NV_TSO_CWND_BOUND + 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 355) 			ca->nv_rtt_start_seq = tp->snd_nxt +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 356) 				ca->nv_min_cwnd * tp->mss_cache;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 357) 			ca->nv_eval_call_cnt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 358) 			ca->nv_allow_cwnd_growth = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 359) 			return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 360) 		}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 361) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 362) 		/* Find the ideal cwnd for current rate from slope
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 363) 		 * slope = 80000.0 * mss / nv_min_rtt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 364) 		 * cwnd_by_slope = nv_rtt_max_rate / slope
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 365) 		 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 366) 		cwnd_by_slope = (u32)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 367) 			div64_u64(((u64)ca->nv_rtt_max_rate) * ca->nv_min_rtt,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 368) 				  80000ULL * tp->mss_cache);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 369) 		max_win = cwnd_by_slope + nv_pad;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 370) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 371) 		/* If cwnd > max_win, decrease cwnd
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 372) 		 * if cwnd < max_win, grow cwnd
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 373) 		 * else leave the same
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 374) 		 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 375) 		if (tp->snd_cwnd > max_win) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 376) 			/* there is congestion, check that it is ok
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 377) 			 * to make a CA decision
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 378) 			 * 1. We should have at least nv_dec_eval_min_calls
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 379) 			 *    data points before making a CA  decision
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 380) 			 * 2. We only make a congesion decision after
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 381) 			 *    nv_rtt_min_cnt RTTs
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 382) 			 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 383) 			if (ca->nv_rtt_cnt < nv_rtt_min_cnt) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 384) 				return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 385) 			} else if (tp->snd_ssthresh == TCP_INFINITE_SSTHRESH) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 386) 				if (ca->nv_eval_call_cnt <
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 387) 				    nv_ssthresh_eval_min_calls)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 388) 					return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 389) 				/* otherwise we will decrease cwnd */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 390) 			} else if (ca->nv_eval_call_cnt <
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 391) 				   nv_dec_eval_min_calls) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 392) 				if (ca->nv_allow_cwnd_growth &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 393) 				    ca->nv_rtt_cnt > nv_stop_rtt_cnt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 394) 					ca->nv_allow_cwnd_growth = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 395) 				return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 396) 			}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 397) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 398) 			/* We have enough data to determine we are congested */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 399) 			ca->nv_allow_cwnd_growth = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 400) 			tp->snd_ssthresh =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 401) 				(nv_ssthresh_factor * max_win) >> 3;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 402) 			if (tp->snd_cwnd - max_win > 2) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 403) 				/* gap > 2, we do exponential cwnd decrease */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 404) 				int dec;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 405) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 406) 				dec = max(2U, ((tp->snd_cwnd - max_win) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 407) 					       nv_cong_dec_mult) >> 7);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 408) 				tp->snd_cwnd -= dec;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 409) 			} else if (nv_cong_dec_mult > 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 410) 				tp->snd_cwnd = max_win;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 411) 			}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 412) 			if (ca->cwnd_growth_factor > 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 413) 				ca->cwnd_growth_factor = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 414) 			ca->nv_no_cong_cnt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 415) 		} else if (tp->snd_cwnd <= max_win - nv_pad_buffer) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 416) 			/* There is no congestion, grow cwnd if allowed*/
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 417) 			if (ca->nv_eval_call_cnt < nv_inc_eval_min_calls)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 418) 				return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 419) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 420) 			ca->nv_allow_cwnd_growth = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 421) 			ca->nv_no_cong_cnt++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 422) 			if (ca->cwnd_growth_factor < 0 &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 423) 			    nv_cwnd_growth_rate_neg > 0 &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 424) 			    ca->nv_no_cong_cnt > nv_cwnd_growth_rate_neg) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 425) 				ca->cwnd_growth_factor++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 426) 				ca->nv_no_cong_cnt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 427) 			} else if (ca->cwnd_growth_factor >= 0 &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 428) 				   nv_cwnd_growth_rate_pos > 0 &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 429) 				   ca->nv_no_cong_cnt >
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 430) 				   nv_cwnd_growth_rate_pos) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 431) 				ca->cwnd_growth_factor++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 432) 				ca->nv_no_cong_cnt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 433) 			}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 434) 		} else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 435) 			/* cwnd is in-between, so do nothing */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 436) 			return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 437) 		}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 438) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 439) 		/* update state */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 440) 		ca->nv_eval_call_cnt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 441) 		ca->nv_rtt_cnt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 442) 		ca->nv_rtt_max_rate = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 443) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 444) 		/* Don't want to make cwnd < nv_min_cwnd
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 445) 		 * (it wasn't before, if it is now is because nv
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 446) 		 *  decreased it).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 447) 		 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 448) 		if (tp->snd_cwnd < nv_min_cwnd)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 449) 			tp->snd_cwnd = nv_min_cwnd;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 450) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 451) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 452) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 453) /* Extract info for Tcp socket info provided via netlink */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 454) static size_t tcpnv_get_info(struct sock *sk, u32 ext, int *attr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 455) 			     union tcp_cc_info *info)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 456) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 457) 	const struct tcpnv *ca = inet_csk_ca(sk);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 458) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 459) 	if (ext & (1 << (INET_DIAG_VEGASINFO - 1))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 460) 		info->vegas.tcpv_enabled = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 461) 		info->vegas.tcpv_rttcnt = ca->nv_rtt_cnt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 462) 		info->vegas.tcpv_rtt = ca->nv_last_rtt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 463) 		info->vegas.tcpv_minrtt = ca->nv_min_rtt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 464) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 465) 		*attr = INET_DIAG_VEGASINFO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 466) 		return sizeof(struct tcpvegas_info);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 467) 	}
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 468) 	return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 469) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 470) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 471) static struct tcp_congestion_ops tcpnv __read_mostly = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 472) 	.init		= tcpnv_init,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 473) 	.ssthresh	= tcpnv_recalc_ssthresh,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 474) 	.cong_avoid	= tcpnv_cong_avoid,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 475) 	.set_state	= tcpnv_state,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 476) 	.undo_cwnd	= tcp_reno_undo_cwnd,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 477) 	.pkts_acked     = tcpnv_acked,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 478) 	.get_info	= tcpnv_get_info,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 479) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 480) 	.owner		= THIS_MODULE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 481) 	.name		= "nv",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 482) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 483) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 484) static int __init tcpnv_register(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 485) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 486) 	BUILD_BUG_ON(sizeof(struct tcpnv) > ICSK_CA_PRIV_SIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 487) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 488) 	return tcp_register_congestion_control(&tcpnv);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 489) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 490) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 491) static void __exit tcpnv_unregister(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 492) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 493) 	tcp_unregister_congestion_control(&tcpnv);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 494) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 495) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 496) module_init(tcpnv_register);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 497) module_exit(tcpnv_unregister);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 498) 
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 499) MODULE_AUTHOR("Lawrence Brakmo");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 500) MODULE_LICENSE("GPL");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 501) MODULE_DESCRIPTION("TCP NV");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 502) MODULE_VERSION("1.0");