1/*- 2 * 3 * SPDX-License-Identifier: BSD-3-Clause 4 * 5 * Copyright (c) 2018-2020 6 * Netflix Inc. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 * __FBSDID("$FreeBSD$"); 29 * 30 */ 31/** 32 * Author: Randall Stewart <rrs@netflix.com> 33 */ 34#ifndef __tcp_ratelimit_h__ 35#define __tcp_ratelimit_h__ 36 37struct m_snd_tag; 38 39/* Flags on an individual rate */ 40#define HDWRPACE_INITED 0x0001 41#define HDWRPACE_TAGPRESENT 0x0002 42#define HDWRPACE_IFPDEPARTED 0x0004 43struct tcp_hwrate_limit_table { 44 const struct tcp_rate_set *ptbl; /* Pointer to parent table */ 45 struct m_snd_tag *tag; /* Send tag if needed (chelsio) */ 46 long rate; /* Rate we get in Bytes per second (Bps) */ 47 long using; /* How many flows are using this hdwr rate. */ 48 long rs_num_enobufs; 49 uint32_t time_between; /* Time-Gap between packets at this rate */ 50 uint32_t flags; 51}; 52 53/* Rateset flags */ 54#define RS_IS_DEFF 0x0001 /* Its a lagg, do a double lookup */ 55#define RS_IS_INTF 0x0002 /* Its a plain interface */ 56#define RS_NO_PRE 0x0004 /* The interfacd has set rates */ 57#define RS_INT_TBL 0x0010 /* 58 * The table is the internal version 59 * which has special setup requirements. 60 */ 61#define RS_IS_DEAD 0x0020 /* The RS is dead list */ 62#define RS_FUNERAL_SCHD 0x0040 /* Is a epoch call scheduled to bury this guy?*/ 63#define RS_INTF_NO_SUP 0x0100 /* The interface does not support the ratelimiting */ 64 65struct tcp_rate_set { 66 struct sysctl_ctx_list sysctl_ctx; 67 CK_LIST_ENTRY(tcp_rate_set) next; 68 struct ifnet *rs_ifp; 69 struct tcp_hwrate_limit_table *rs_rlt; 70 uint64_t rs_flows_using; 71 uint64_t rs_flow_limit; 72 uint32_t rs_if_dunit; 73 int rs_rate_cnt; 74 int rs_min_seg; 75 int rs_highest_valid; 76 int rs_lowest_valid; 77 int rs_disable; 78 int rs_flags; 79 struct epoch_context rs_epoch_ctx; 80}; 81 82CK_LIST_HEAD(head_tcp_rate_set, tcp_rate_set); 83 84/* Request flags */ 85#define RS_PACING_EXACT_MATCH 0x0001 /* Need an exact match for rate */ 86#define RS_PACING_GT 0x0002 /* Greater than requested */ 87#define RS_PACING_GEQ 0x0004 /* Greater than or equal too */ 88#define RS_PACING_LT 0x0008 /* Less than requested rate */ 89#define RS_PACING_SUB_OK 0x0010 /* If a rate can't be found get the 90 * next best rate (highest or lowest). */ 91#ifdef _KERNEL 92#ifndef ETHERNET_SEGMENT_SIZE 93#define ETHERNET_SEGMENT_SIZE 1514 94#endif 95#ifdef RATELIMIT 96#define DETAILED_RATELIMIT_SYSCTL 1 /* 97 * Undefine this if you don't want 98 * detailed rates to appear in 99 * net.inet.tcp.rl. 100 * With the defintion each rate 101 * shows up in your sysctl tree 102 * this can be big. 103 */ 104uint64_t inline 105tcp_hw_highest_rate(const struct tcp_hwrate_limit_table *rle) 106{ 107 return (rle->ptbl->rs_rlt[rle->ptbl->rs_highest_valid].rate); 108} 109 110uint64_t 111tcp_hw_highest_rate_ifp(struct ifnet *ifp, struct inpcb *inp); 112 113const struct tcp_hwrate_limit_table * 114tcp_set_pacing_rate(struct tcpcb *tp, struct ifnet *ifp, 115 uint64_t bytes_per_sec, int flags, int *error, uint64_t *lower_rate); 116 117const struct tcp_hwrate_limit_table * 118tcp_chg_pacing_rate(const struct tcp_hwrate_limit_table *crte, 119 struct tcpcb *tp, struct ifnet *ifp, 120 uint64_t bytes_per_sec, int flags, int *error, uint64_t *lower_rate); 121void 122tcp_rel_pacing_rate(const struct tcp_hwrate_limit_table *crte, 123 struct tcpcb *tp); 124#else 125static inline const struct tcp_hwrate_limit_table * 126tcp_set_pacing_rate(struct tcpcb *tp, struct ifnet *ifp, 127 uint64_t bytes_per_sec, int flags, int *error, uint64_t *lower_rate) 128{ 129 if (error) 130 *error = EOPNOTSUPP; 131 return (NULL); 132} 133 134static inline const struct tcp_hwrate_limit_table * 135tcp_chg_pacing_rate(const struct tcp_hwrate_limit_table *crte, 136 struct tcpcb *tp, struct ifnet *ifp, 137 uint64_t bytes_per_sec, int flags, int *error, uint64_t *lower_rate) 138{ 139 if (error) 140 *error = EOPNOTSUPP; 141 return (NULL); 142} 143 144static inline void 145tcp_rel_pacing_rate(const struct tcp_hwrate_limit_table *crte, 146 struct tcpcb *tp) 147{ 148 return; 149} 150 151static uint64_t inline 152tcp_hw_highest_rate(const struct tcp_hwrate_limit_table *rle) 153{ 154 return (0); 155} 156 157static uint64_t inline 158tcp_hw_highest_rate_ifp(struct ifnet *ifp, struct inpcb *inp) 159{ 160 return (0); 161} 162 163 164#endif 165/* 166 * Given a b/w and a segsiz, and optional hardware 167 * rate limit, return the ideal size to burst 168 * out at once. Note the parameter can_use_1mss 169 * dictates if the transport will tolerate a 1mss 170 * limit, if not it will bottom out at 2mss (think 171 * delayed ack). 172 */ 173uint32_t 174tcp_get_pacing_burst_size(struct tcpcb *tp, uint64_t bw, uint32_t segsiz, int can_use_1mss, 175 const struct tcp_hwrate_limit_table *te, int *err); 176 177 178void 179tcp_rl_log_enobuf(const struct tcp_hwrate_limit_table *rte); 180 181#endif 182#endif 183