1 /*- 2 * Copyright (C) 1998-2003 3 * Sony Computer Science Laboratories Inc. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 14 * THIS SOFTWARE IS PROVIDED BY SONY CSL AND CONTRIBUTORS ``AS IS'' AND 15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 17 * ARE DISCLAIMED. IN NO EVENT SHALL SONY CSL OR CONTRIBUTORS BE LIABLE 18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 24 * SUCH DAMAGE. 25 */ 26 /*- 27 * Copyright (c) 1990-1994 Regents of the University of California. 28 * All rights reserved. 29 * 30 * Redistribution and use in source and binary forms, with or without 31 * modification, are permitted provided that the following conditions 32 * are met: 33 * 1. Redistributions of source code must retain the above copyright 34 * notice, this list of conditions and the following disclaimer. 35 * 2. Redistributions in binary form must reproduce the above copyright 36 * notice, this list of conditions and the following disclaimer in the 37 * documentation and/or other materials provided with the distribution. 38 * 3. All advertising materials mentioning features or use of this software 39 * must display the following acknowledgement: 40 * This product includes software developed by the Computer Systems 41 * Engineering Group at Lawrence Berkeley Laboratory. 42 * 4. Neither the name of the University nor of the Laboratory may be used 43 * to endorse or promote products derived from this software without 44 * specific prior written permission. 45 * 46 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 47 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 48 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 49 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 50 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 51 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 52 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 53 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 54 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 55 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 56 * SUCH DAMAGE. 57 * 58 * $KAME: altq_rio.c,v 1.17 2003/07/10 12:07:49 kjc Exp $ 59 */ 60 61 #include "opt_altq.h" 62 #include "opt_inet.h" 63 #include "opt_inet6.h" 64 #ifdef ALTQ_RIO /* rio is enabled by ALTQ_RIO option in opt_altq.h */ 65 66 #include <sys/param.h> 67 #include <sys/malloc.h> 68 #include <sys/mbuf.h> 69 #include <sys/socket.h> 70 #include <sys/systm.h> 71 #include <sys/errno.h> 72 #if 1 /* ALTQ3_COMPAT */ 73 #include <sys/proc.h> 74 #include <sys/sockio.h> 75 #include <sys/kernel.h> 76 #endif 77 78 #include <net/if.h> 79 #include <net/if_var.h> 80 81 #include <netinet/in.h> 82 #include <netinet/in_systm.h> 83 #include <netinet/ip.h> 84 #ifdef INET6 85 #include <netinet/ip6.h> 86 #endif 87 88 #include <netpfil/pf/pf.h> 89 #include <netpfil/pf/pf_altq.h> 90 #include <net/altq/altq.h> 91 #include <net/altq/altq_cdnr.h> 92 #include <net/altq/altq_red.h> 93 #include <net/altq/altq_rio.h> 94 95 /* 96 * RIO: RED with IN/OUT bit 97 * described in 98 * "Explicit Allocation of Best Effort Packet Delivery Service" 99 * David D. Clark and Wenjia Fang, MIT Lab for Computer Science 100 * http://diffserv.lcs.mit.edu/Papers/exp-alloc-ddc-wf.{ps,pdf} 101 * 102 * this implementation is extended to support more than 2 drop precedence 103 * values as described in RFC2597 (Assured Forwarding PHB Group). 104 * 105 */ 106 /* 107 * AF DS (differentiated service) codepoints. 108 * (classes can be mapped to CBQ or H-FSC classes.) 109 * 110 * 0 1 2 3 4 5 6 7 111 * +---+---+---+---+---+---+---+---+ 112 * | CLASS |DropPre| 0 | CU | 113 * +---+---+---+---+---+---+---+---+ 114 * 115 * class 1: 001 116 * class 2: 010 117 * class 3: 011 118 * class 4: 100 119 * 120 * low drop prec: 01 121 * medium drop prec: 10 122 * high drop prec: 01 123 */ 124 125 /* normal red parameters */ 126 #define W_WEIGHT 512 /* inverse of weight of EWMA (511/512) */ 127 /* q_weight = 0.00195 */ 128 129 /* red parameters for a slow link */ 130 #define W_WEIGHT_1 128 /* inverse of weight of EWMA (127/128) */ 131 /* q_weight = 0.0078125 */ 132 133 /* red parameters for a very slow link (e.g., dialup) */ 134 #define W_WEIGHT_2 64 /* inverse of weight of EWMA (63/64) */ 135 /* q_weight = 0.015625 */ 136 137 /* fixed-point uses 12-bit decimal places */ 138 #define FP_SHIFT 12 /* fixed-point shift */ 139 140 /* red parameters for drop probability */ 141 #define INV_P_MAX 10 /* inverse of max drop probability */ 142 #define TH_MIN 5 /* min threshold */ 143 #define TH_MAX 15 /* max threshold */ 144 145 #define RIO_LIMIT 60 /* default max queue length */ 146 #define RIO_STATS /* collect statistics */ 147 148 #define TV_DELTA(a, b, delta) { \ 149 int xxs; \ 150 \ 151 delta = (a)->tv_usec - (b)->tv_usec; \ 152 if ((xxs = (a)->tv_sec - (b)->tv_sec) != 0) { \ 153 if (xxs < 0) { \ 154 delta = 60000000; \ 155 } else if (xxs > 4) { \ 156 if (xxs > 60) \ 157 delta = 60000000; \ 158 else \ 159 delta += xxs * 1000000; \ 160 } else while (xxs > 0) { \ 161 delta += 1000000; \ 162 xxs--; \ 163 } \ 164 } \ 165 } 166 167 /* default rio parameter values */ 168 static struct redparams default_rio_params[RIO_NDROPPREC] = { 169 /* th_min, th_max, inv_pmax */ 170 { TH_MAX * 2 + TH_MIN, TH_MAX * 3, INV_P_MAX }, /* low drop precedence */ 171 { TH_MAX + TH_MIN, TH_MAX * 2, INV_P_MAX }, /* medium drop precedence */ 172 { TH_MIN, TH_MAX, INV_P_MAX } /* high drop precedence */ 173 }; 174 175 /* internal function prototypes */ 176 static int dscp2index(u_int8_t); 177 178 rio_t * 179 rio_alloc(int weight, struct redparams *params, int flags, int pkttime) 180 { 181 rio_t *rp; 182 int w, i; 183 int npkts_per_sec; 184 185 rp = malloc(sizeof(rio_t), M_DEVBUF, M_NOWAIT | M_ZERO); 186 if (rp == NULL) 187 return (NULL); 188 189 rp->rio_flags = flags; 190 if (pkttime == 0) 191 /* default packet time: 1000 bytes / 10Mbps * 8 * 1000000 */ 192 rp->rio_pkttime = 800; 193 else 194 rp->rio_pkttime = pkttime; 195 196 if (weight != 0) 197 rp->rio_weight = weight; 198 else { 199 /* use default */ 200 rp->rio_weight = W_WEIGHT; 201 202 /* when the link is very slow, adjust red parameters */ 203 npkts_per_sec = 1000000 / rp->rio_pkttime; 204 if (npkts_per_sec < 50) { 205 /* up to about 400Kbps */ 206 rp->rio_weight = W_WEIGHT_2; 207 } else if (npkts_per_sec < 300) { 208 /* up to about 2.4Mbps */ 209 rp->rio_weight = W_WEIGHT_1; 210 } 211 } 212 213 /* calculate wshift. weight must be power of 2 */ 214 w = rp->rio_weight; 215 for (i = 0; w > 1; i++) 216 w = w >> 1; 217 rp->rio_wshift = i; 218 w = 1 << rp->rio_wshift; 219 if (w != rp->rio_weight) { 220 printf("invalid weight value %d for red! use %d\n", 221 rp->rio_weight, w); 222 rp->rio_weight = w; 223 } 224 225 /* allocate weight table */ 226 rp->rio_wtab = wtab_alloc(rp->rio_weight); 227 228 for (i = 0; i < RIO_NDROPPREC; i++) { 229 struct dropprec_state *prec = &rp->rio_precstate[i]; 230 231 prec->avg = 0; 232 prec->idle = 1; 233 234 if (params == NULL || params[i].inv_pmax == 0) 235 prec->inv_pmax = default_rio_params[i].inv_pmax; 236 else 237 prec->inv_pmax = params[i].inv_pmax; 238 if (params == NULL || params[i].th_min == 0) 239 prec->th_min = default_rio_params[i].th_min; 240 else 241 prec->th_min = params[i].th_min; 242 if (params == NULL || params[i].th_max == 0) 243 prec->th_max = default_rio_params[i].th_max; 244 else 245 prec->th_max = params[i].th_max; 246 247 /* 248 * th_min_s and th_max_s are scaled versions of th_min 249 * and th_max to be compared with avg. 250 */ 251 prec->th_min_s = prec->th_min << (rp->rio_wshift + FP_SHIFT); 252 prec->th_max_s = prec->th_max << (rp->rio_wshift + FP_SHIFT); 253 254 /* 255 * precompute probability denominator 256 * probd = (2 * (TH_MAX-TH_MIN) / pmax) in fixed-point 257 */ 258 prec->probd = (2 * (prec->th_max - prec->th_min) 259 * prec->inv_pmax) << FP_SHIFT; 260 261 microtime(&prec->last); 262 } 263 264 return (rp); 265 } 266 267 void 268 rio_destroy(rio_t *rp) 269 { 270 wtab_destroy(rp->rio_wtab); 271 free(rp, M_DEVBUF); 272 } 273 274 void 275 rio_getstats(rio_t *rp, struct redstats *sp) 276 { 277 int i; 278 279 for (i = 0; i < RIO_NDROPPREC; i++) { 280 bcopy(&rp->q_stats[i], sp, sizeof(struct redstats)); 281 sp->q_avg = rp->rio_precstate[i].avg >> rp->rio_wshift; 282 sp++; 283 } 284 } 285 286 #if (RIO_NDROPPREC == 3) 287 /* 288 * internally, a drop precedence value is converted to an index 289 * starting from 0. 290 */ 291 static int 292 dscp2index(u_int8_t dscp) 293 { 294 int dpindex = dscp & AF_DROPPRECMASK; 295 296 if (dpindex == 0) 297 return (0); 298 return ((dpindex >> 3) - 1); 299 } 300 #endif 301 302 #if 1 303 /* 304 * kludge: when a packet is dequeued, we need to know its drop precedence 305 * in order to keep the queue length of each drop precedence. 306 * use m_pkthdr.rcvif to pass this info. 307 */ 308 #define RIOM_SET_PRECINDEX(m, idx) \ 309 do { (m)->m_pkthdr.rcvif = (void *)((long)(idx)); } while (0) 310 #define RIOM_GET_PRECINDEX(m) \ 311 ({ long idx; idx = (long)((m)->m_pkthdr.rcvif); \ 312 (m)->m_pkthdr.rcvif = NULL; idx; }) 313 #endif 314 315 int 316 rio_addq(rio_t *rp, class_queue_t *q, struct mbuf *m, 317 struct altq_pktattr *pktattr) 318 { 319 int avg, droptype; 320 u_int8_t dsfield, odsfield; 321 int dpindex, i, n, t; 322 struct timeval now; 323 struct dropprec_state *prec; 324 325 dsfield = odsfield = read_dsfield(m, pktattr); 326 dpindex = dscp2index(dsfield); 327 328 /* 329 * update avg of the precedence states whose drop precedence 330 * is larger than or equal to the drop precedence of the packet 331 */ 332 now.tv_sec = 0; 333 for (i = dpindex; i < RIO_NDROPPREC; i++) { 334 prec = &rp->rio_precstate[i]; 335 avg = prec->avg; 336 if (prec->idle) { 337 prec->idle = 0; 338 if (now.tv_sec == 0) 339 microtime(&now); 340 t = (now.tv_sec - prec->last.tv_sec); 341 if (t > 60) 342 avg = 0; 343 else { 344 t = t * 1000000 + 345 (now.tv_usec - prec->last.tv_usec); 346 n = t / rp->rio_pkttime; 347 /* calculate (avg = (1 - Wq)^n * avg) */ 348 if (n > 0) 349 avg = (avg >> FP_SHIFT) * 350 pow_w(rp->rio_wtab, n); 351 } 352 } 353 354 /* run estimator. (avg is scaled by WEIGHT in fixed-point) */ 355 avg += (prec->qlen << FP_SHIFT) - (avg >> rp->rio_wshift); 356 prec->avg = avg; /* save the new value */ 357 /* 358 * count keeps a tally of arriving traffic that has not 359 * been dropped. 360 */ 361 prec->count++; 362 } 363 364 prec = &rp->rio_precstate[dpindex]; 365 avg = prec->avg; 366 367 /* see if we drop early */ 368 droptype = DTYPE_NODROP; 369 if (avg >= prec->th_min_s && prec->qlen > 1) { 370 if (avg >= prec->th_max_s) { 371 /* avg >= th_max: forced drop */ 372 droptype = DTYPE_FORCED; 373 } else if (prec->old == 0) { 374 /* first exceeds th_min */ 375 prec->count = 1; 376 prec->old = 1; 377 } else if (drop_early((avg - prec->th_min_s) >> rp->rio_wshift, 378 prec->probd, prec->count)) { 379 /* unforced drop by red */ 380 droptype = DTYPE_EARLY; 381 } 382 } else { 383 /* avg < th_min */ 384 prec->old = 0; 385 } 386 387 /* 388 * if the queue length hits the hard limit, it's a forced drop. 389 */ 390 if (droptype == DTYPE_NODROP && qlen(q) >= qlimit(q)) 391 droptype = DTYPE_FORCED; 392 393 if (droptype != DTYPE_NODROP) { 394 /* always drop incoming packet (as opposed to randomdrop) */ 395 for (i = dpindex; i < RIO_NDROPPREC; i++) 396 rp->rio_precstate[i].count = 0; 397 #ifdef RIO_STATS 398 if (droptype == DTYPE_EARLY) 399 rp->q_stats[dpindex].drop_unforced++; 400 else 401 rp->q_stats[dpindex].drop_forced++; 402 PKTCNTR_ADD(&rp->q_stats[dpindex].drop_cnt, m_pktlen(m)); 403 #endif 404 m_freem(m); 405 return (-1); 406 } 407 408 for (i = dpindex; i < RIO_NDROPPREC; i++) 409 rp->rio_precstate[i].qlen++; 410 411 /* save drop precedence index in mbuf hdr */ 412 RIOM_SET_PRECINDEX(m, dpindex); 413 414 if (rp->rio_flags & RIOF_CLEARDSCP) 415 dsfield &= ~DSCP_MASK; 416 417 if (dsfield != odsfield) 418 write_dsfield(m, pktattr, dsfield); 419 420 _addq(q, m); 421 422 #ifdef RIO_STATS 423 PKTCNTR_ADD(&rp->q_stats[dpindex].xmit_cnt, m_pktlen(m)); 424 #endif 425 return (0); 426 } 427 428 struct mbuf * 429 rio_getq(rio_t *rp, class_queue_t *q) 430 { 431 struct mbuf *m; 432 int dpindex, i; 433 434 if ((m = _getq(q)) == NULL) 435 return NULL; 436 437 dpindex = RIOM_GET_PRECINDEX(m); 438 for (i = dpindex; i < RIO_NDROPPREC; i++) { 439 if (--rp->rio_precstate[i].qlen == 0) { 440 if (rp->rio_precstate[i].idle == 0) { 441 rp->rio_precstate[i].idle = 1; 442 microtime(&rp->rio_precstate[i].last); 443 } 444 } 445 } 446 return (m); 447 } 448 449 #endif /* ALTQ_RIO */ 450