1// SPDX-License-Identifier: GPL-2.0-only
2/*
3 * CAIA Delay-Gradient (CDG) congestion control
4 *
5 * This implementation is based on the paper:
6 * D.A. Hayes and G. Armitage. "Revisiting TCP congestion control using
7 * delay gradients." In IFIP Networking, pages 328-341. Springer, 2011.
8 *
9 * Scavenger traffic (Less-than-Best-Effort) should disable coexistence
10 * heuristics using parameters use_shadow=0 and use_ineff=0.
11 *
12 * Parameters window, backoff_beta, and backoff_factor are crucial for
13 * throughput and delay. Future work is needed to determine better defaults,
14 * and to provide guidelines for use in different environments/contexts.
15 *
16 * Except for window, knobs are configured via /sys/module/tcp_cdg/parameters/.
17 * Parameter window is only configurable when loading tcp_cdg as a module.
18 *
19 * Notable differences from paper/FreeBSD:
20 * o Using Hybrid Slow start and Proportional Rate Reduction.
21 * o Add toggle for shadow window mechanism. Suggested by David Hayes.
22 * o Add toggle for non-congestion loss tolerance.
23 * o Scaling parameter G is changed to a backoff factor;
24 * conversion is given by: backoff_factor = 1000/(G * window).
25 * o Limit shadow window to 2 * cwnd, or to cwnd when application limited.
26 * o More accurate e^-x.
27 */
28#include <linux/kernel.h>
29#include <linux/random.h>
30#include <linux/module.h>
31#include <linux/sched/clock.h>
32
33#include <net/tcp.h>
34
35#define HYSTART_ACK_TRAIN 1
36#define HYSTART_DELAY 2
37
38static int window __read_mostly = 8;
39static unsigned int backoff_beta __read_mostly = 0.7071 * 1024; /* sqrt 0.5 */
40static unsigned int backoff_factor __read_mostly = 42;
41static unsigned int hystart_detect __read_mostly = 3;
42static unsigned int use_ineff __read_mostly = 5;
43static bool use_shadow __read_mostly = true;
44static bool use_tolerance __read_mostly;
45
46module_param(window, int, 0444);
47MODULE_PARM_DESC(window, "gradient window size (power of two <= 256)");
48module_param(backoff_beta, uint, 0644);
49MODULE_PARM_DESC(backoff_beta, "backoff beta (0-1024)");
50module_param(backoff_factor, uint, 0644);
51MODULE_PARM_DESC(backoff_factor, "backoff probability scale factor");
52module_param(hystart_detect, uint, 0644);
53MODULE_PARM_DESC(hystart_detect, "use Hybrid Slow start "
54 "(0: disabled, 1: ACK train, 2: delay threshold, 3: both)");
55module_param(use_ineff, uint, 0644);
56MODULE_PARM_DESC(use_ineff, "use ineffectual backoff detection (threshold)");
57module_param(use_shadow, bool, 0644);
58MODULE_PARM_DESC(use_shadow, "use shadow window heuristic");
59module_param(use_tolerance, bool, 0644);
60MODULE_PARM_DESC(use_tolerance, "use loss tolerance heuristic");
61
62struct cdg_minmax {
63 union {
64 struct {
65 s32 min;
66 s32 max;
67 };
68 u64 v64;
69 };
70};
71
72enum cdg_state {
73 CDG_UNKNOWN = 0,
74 CDG_NONFULL = 1,
75 CDG_FULL = 2,
76 CDG_BACKOFF = 3,
77};
78
79struct cdg {
80 struct cdg_minmax rtt;
81 struct cdg_minmax rtt_prev;
82 struct cdg_minmax *gradients;
83 struct cdg_minmax gsum;
84 bool gfilled;
85 u8 tail;
86 u8 state;
87 u8 delack;
88 u32 rtt_seq;
89 u32 shadow_wnd;
90 u16 backoff_cnt;
91 u16 sample_cnt;
92 s32 delay_min;
93 u32 last_ack;
94 u32 round_start;
95};
96
97/**
98 * nexp_u32 - negative base-e exponential
99 * @ux: x in units of micro
100 *
101 * Returns exp(ux * -1e-6) * U32_MAX.
102 */
103static u32 __pure nexp_u32(u32 ux)
104{
105 static const u16 v[] = {
106 /* exp(-x)*65536-1 for x = 0, 0.000256, 0.000512, ... */
107 65535,
108 65518, 65501, 65468, 65401, 65267, 65001, 64470, 63422,
109 61378, 57484, 50423, 38795, 22965, 8047, 987, 14,
110 };
111 u32 msb = ux >> 8;
112 u32 res;
113 int i;
114
115 /* Cut off when ux >= 2^24 (actual result is <= 222/U32_MAX). */
116 if (msb > U16_MAX)
117 return 0;
118
119 /* Scale first eight bits linearly: */
120 res = U32_MAX - (ux & 0xff) * (U32_MAX / 1000000);
121
122 /* Obtain e^(x + y + ...) by computing e^x * e^y * ...: */
123 for (i = 1; msb; i++, msb >>= 1) {
124 u32 y = v[i & -(msb & 1)] + U32_C(1);
125
126 res = ((u64)res * y) >> 16;
127 }
128
129 return res;
130}
131
132/* Based on the HyStart algorithm (by Ha et al.) that is implemented in
133 * tcp_cubic. Differences/experimental changes:
134 * o Using Hayes' delayed ACK filter.
135 * o Using a usec clock for the ACK train.
136 * o Reset ACK train when application limited.
137 * o Invoked at any cwnd (i.e. also when cwnd < 16).
138 * o Invoked only when cwnd < ssthresh (i.e. not when cwnd == ssthresh).
139 */
140static void tcp_cdg_hystart_update(struct sock *sk)
141{
142 struct cdg *ca = inet_csk_ca(sk);
143 struct tcp_sock *tp = tcp_sk(sk);
144
145 ca->delay_min = min_not_zero(ca->delay_min, ca->rtt.min);
146 if (ca->delay_min == 0)
147 return;
148
149 if (hystart_detect & HYSTART_ACK_TRAIN) {
150 u32 now_us = tp->tcp_mstamp;
151
152 if (ca->last_ack == 0 || !tcp_is_cwnd_limited(sk)) {
153 ca->last_ack = now_us;
154 ca->round_start = now_us;
155 } else if (before(seq1: now_us, seq2: ca->last_ack + 3000)) {
156 u32 base_owd = max(ca->delay_min / 2U, 125U);
157
158 ca->last_ack = now_us;
159 if (after(now_us, ca->round_start + base_owd)) {
160 NET_INC_STATS(sock_net(sk),
161 LINUX_MIB_TCPHYSTARTTRAINDETECT);
162 NET_ADD_STATS(sock_net(sk),
163 LINUX_MIB_TCPHYSTARTTRAINCWND,
164 tcp_snd_cwnd(tp));
165 tp->snd_ssthresh = tcp_snd_cwnd(tp);
166 return;
167 }
168 }
169 }
170
171 if (hystart_detect & HYSTART_DELAY) {
172 if (ca->sample_cnt < 8) {
173 ca->sample_cnt++;
174 } else {
175 s32 thresh = max(ca->delay_min + ca->delay_min / 8U,
176 125U);
177
178 if (ca->rtt.min > thresh) {
179 NET_INC_STATS(sock_net(sk),
180 LINUX_MIB_TCPHYSTARTDELAYDETECT);
181 NET_ADD_STATS(sock_net(sk),
182 LINUX_MIB_TCPHYSTARTDELAYCWND,
183 tcp_snd_cwnd(tp));
184 tp->snd_ssthresh = tcp_snd_cwnd(tp);
185 }
186 }
187 }
188}
189
190static s32 tcp_cdg_grad(struct cdg *ca)
191{
192 s32 gmin = ca->rtt.min - ca->rtt_prev.min;
193 s32 gmax = ca->rtt.max - ca->rtt_prev.max;
194 s32 grad;
195
196 if (ca->gradients) {
197 ca->gsum.min += gmin - ca->gradients[ca->tail].min;
198 ca->gsum.max += gmax - ca->gradients[ca->tail].max;
199 ca->gradients[ca->tail].min = gmin;
200 ca->gradients[ca->tail].max = gmax;
201 ca->tail = (ca->tail + 1) & (window - 1);
202 gmin = ca->gsum.min;
203 gmax = ca->gsum.max;
204 }
205
206 /* We keep sums to ignore gradients during cwnd reductions;
207 * the paper's smoothed gradients otherwise simplify to:
208 * (rtt_latest - rtt_oldest) / window.
209 *
210 * We also drop division by window here.
211 */
212 grad = gmin > 0 ? gmin : gmax;
213
214 /* Extrapolate missing values in gradient window: */
215 if (!ca->gfilled) {
216 if (!ca->gradients && window > 1)
217 grad *= window; /* Memory allocation failed. */
218 else if (ca->tail == 0)
219 ca->gfilled = true;
220 else
221 grad = (grad * window) / (int)ca->tail;
222 }
223
224 /* Backoff was effectual: */
225 if (gmin <= -32 || gmax <= -32)
226 ca->backoff_cnt = 0;
227
228 if (use_tolerance) {
229 /* Reduce small variations to zero: */
230 gmin = DIV_ROUND_CLOSEST(gmin, 64);
231 gmax = DIV_ROUND_CLOSEST(gmax, 64);
232
233 if (gmin > 0 && gmax <= 0)
234 ca->state = CDG_FULL;
235 else if ((gmin > 0 && gmax > 0) || gmax < 0)
236 ca->state = CDG_NONFULL;
237 }
238 return grad;
239}
240
241static bool tcp_cdg_backoff(struct sock *sk, u32 grad)
242{
243 struct cdg *ca = inet_csk_ca(sk);
244 struct tcp_sock *tp = tcp_sk(sk);
245
246 if (get_random_u32() <= nexp_u32(ux: grad * backoff_factor))
247 return false;
248
249 if (use_ineff) {
250 ca->backoff_cnt++;
251 if (ca->backoff_cnt > use_ineff)
252 return false;
253 }
254
255 ca->shadow_wnd = max(ca->shadow_wnd, tcp_snd_cwnd(tp));
256 ca->state = CDG_BACKOFF;
257 tcp_enter_cwr(sk);
258 return true;
259}
260
261/* Not called in CWR or Recovery state. */
262static void tcp_cdg_cong_avoid(struct sock *sk, u32 ack, u32 acked)
263{
264 struct cdg *ca = inet_csk_ca(sk);
265 struct tcp_sock *tp = tcp_sk(sk);
266 u32 prior_snd_cwnd;
267 u32 incr;
268
269 if (tcp_in_slow_start(tp) && hystart_detect)
270 tcp_cdg_hystart_update(sk);
271
272 if (after(ack, ca->rtt_seq) && ca->rtt.v64) {
273 s32 grad = 0;
274
275 if (ca->rtt_prev.v64)
276 grad = tcp_cdg_grad(ca);
277 ca->rtt_seq = tp->snd_nxt;
278 ca->rtt_prev = ca->rtt;
279 ca->rtt.v64 = 0;
280 ca->last_ack = 0;
281 ca->sample_cnt = 0;
282
283 if (grad > 0 && tcp_cdg_backoff(sk, grad))
284 return;
285 }
286
287 if (!tcp_is_cwnd_limited(sk)) {
288 ca->shadow_wnd = min(ca->shadow_wnd, tcp_snd_cwnd(tp));
289 return;
290 }
291
292 prior_snd_cwnd = tcp_snd_cwnd(tp);
293 tcp_reno_cong_avoid(sk, ack, acked);
294
295 incr = tcp_snd_cwnd(tp) - prior_snd_cwnd;
296 ca->shadow_wnd = max(ca->shadow_wnd, ca->shadow_wnd + incr);
297}
298
299static void tcp_cdg_acked(struct sock *sk, const struct ack_sample *sample)
300{
301 struct cdg *ca = inet_csk_ca(sk);
302 struct tcp_sock *tp = tcp_sk(sk);
303
304 if (sample->rtt_us <= 0)
305 return;
306
307 /* A heuristic for filtering delayed ACKs, adapted from:
308 * D.A. Hayes. "Timing enhancements to the FreeBSD kernel to support
309 * delay and rate based TCP mechanisms." TR 100219A. CAIA, 2010.
310 */
311 if (tp->sacked_out == 0) {
312 if (sample->pkts_acked == 1 && ca->delack) {
313 /* A delayed ACK is only used for the minimum if it is
314 * provenly lower than an existing non-zero minimum.
315 */
316 ca->rtt.min = min(ca->rtt.min, sample->rtt_us);
317 ca->delack--;
318 return;
319 } else if (sample->pkts_acked > 1 && ca->delack < 5) {
320 ca->delack++;
321 }
322 }
323
324 ca->rtt.min = min_not_zero(ca->rtt.min, sample->rtt_us);
325 ca->rtt.max = max(ca->rtt.max, sample->rtt_us);
326}
327
328static u32 tcp_cdg_ssthresh(struct sock *sk)
329{
330 struct cdg *ca = inet_csk_ca(sk);
331 struct tcp_sock *tp = tcp_sk(sk);
332
333 if (ca->state == CDG_BACKOFF)
334 return max(2U, (tcp_snd_cwnd(tp) * min(1024U, backoff_beta)) >> 10);
335
336 if (ca->state == CDG_NONFULL && use_tolerance)
337 return tcp_snd_cwnd(tp);
338
339 ca->shadow_wnd = min(ca->shadow_wnd >> 1, tcp_snd_cwnd(tp));
340 if (use_shadow)
341 return max3(2U, ca->shadow_wnd, tcp_snd_cwnd(tp) >> 1);
342 return max(2U, tcp_snd_cwnd(tp) >> 1);
343}
344
345static void tcp_cdg_cwnd_event(struct sock *sk, const enum tcp_ca_event ev)
346{
347 struct cdg *ca = inet_csk_ca(sk);
348 struct tcp_sock *tp = tcp_sk(sk);
349 struct cdg_minmax *gradients;
350
351 switch (ev) {
352 case CA_EVENT_CWND_RESTART:
353 gradients = ca->gradients;
354 if (gradients)
355 memset(gradients, 0, window * sizeof(gradients[0]));
356 memset(ca, 0, sizeof(*ca));
357
358 ca->gradients = gradients;
359 ca->rtt_seq = tp->snd_nxt;
360 ca->shadow_wnd = tcp_snd_cwnd(tp);
361 break;
362 case CA_EVENT_COMPLETE_CWR:
363 ca->state = CDG_UNKNOWN;
364 ca->rtt_seq = tp->snd_nxt;
365 ca->rtt_prev = ca->rtt;
366 ca->rtt.v64 = 0;
367 break;
368 default:
369 break;
370 }
371}
372
373static void tcp_cdg_init(struct sock *sk)
374{
375 struct cdg *ca = inet_csk_ca(sk);
376 struct tcp_sock *tp = tcp_sk(sk);
377
378 ca->gradients = NULL;
379 /* We silently fall back to window = 1 if allocation fails. */
380 if (window > 1)
381 ca->gradients = kcalloc(n: window, size: sizeof(ca->gradients[0]),
382 GFP_NOWAIT | __GFP_NOWARN);
383 ca->rtt_seq = tp->snd_nxt;
384 ca->shadow_wnd = tcp_snd_cwnd(tp);
385}
386
387static void tcp_cdg_release(struct sock *sk)
388{
389 struct cdg *ca = inet_csk_ca(sk);
390
391 kfree(objp: ca->gradients);
392 ca->gradients = NULL;
393}
394
395static struct tcp_congestion_ops tcp_cdg __read_mostly = {
396 .cong_avoid = tcp_cdg_cong_avoid,
397 .cwnd_event = tcp_cdg_cwnd_event,
398 .pkts_acked = tcp_cdg_acked,
399 .undo_cwnd = tcp_reno_undo_cwnd,
400 .ssthresh = tcp_cdg_ssthresh,
401 .release = tcp_cdg_release,
402 .init = tcp_cdg_init,
403 .owner = THIS_MODULE,
404 .name = "cdg",
405};
406
407static int __init tcp_cdg_register(void)
408{
409 if (backoff_beta > 1024 || window < 1 || window > 256)
410 return -ERANGE;
411 if (!is_power_of_2(n: window))
412 return -EINVAL;
413
414 BUILD_BUG_ON(sizeof(struct cdg) > ICSK_CA_PRIV_SIZE);
415 tcp_register_congestion_control(type: &tcp_cdg);
416 return 0;
417}
418
419static void __exit tcp_cdg_unregister(void)
420{
421 tcp_unregister_congestion_control(type: &tcp_cdg);
422}
423
424module_init(tcp_cdg_register);
425module_exit(tcp_cdg_unregister);
426MODULE_AUTHOR("Kenneth Klette Jonassen");
427MODULE_LICENSE("GPL");
428MODULE_DESCRIPTION("TCP CDG");
429

source code of linux/net/ipv4/tcp_cdg.c