1 /* 2 * Copyright (c) 2007 The University of Aberdeen, Scotland, UK 3 * Copyright (c) 2005-7 The University of Waikato, Hamilton, New Zealand. 4 * 5 * An implementation of the DCCP protocol 6 * 7 * This code has been developed by the University of Waikato WAND 8 * research group. For further information please see http://www.wand.net.nz/ 9 * or e-mail Ian McDonald - ian.mcdonald@jandi.co.nz 10 * 11 * This code also uses code from Lulea University, rereleased as GPL by its 12 * authors: 13 * Copyright (c) 2003 Nils-Erik Mattsson, Joacim Haggmark, Magnus Erixzon 14 * 15 * Changes to meet Linux coding standards, to make it meet latest ccid3 draft 16 * and to make it work as a loadable module in the DCCP stack written by 17 * Arnaldo Carvalho de Melo <acme@conectiva.com.br>. 18 * 19 * Copyright (c) 2005 Arnaldo Carvalho de Melo <acme@conectiva.com.br> 20 * 21 * This program is free software; you can redistribute it and/or modify 22 * it under the terms of the GNU General Public License as published by 23 * the Free Software Foundation; either version 2 of the License, or 24 * (at your option) any later version. 25 * 26 * This program is distributed in the hope that it will be useful, 27 * but WITHOUT ANY WARRANTY; without even the implied warranty of 28 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 29 * GNU General Public License for more details. 30 * 31 * You should have received a copy of the GNU General Public License 32 * along with this program; if not, write to the Free Software 33 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. 34 */ 35 36 #include <linux/string.h> 37 #include <linux/slab.h> 38 #include "packet_history.h" 39 #include "../../dccp.h" 40 41 /* 42 * Transmitter History Routines 43 */ 44 static struct kmem_cache *tfrc_tx_hist_slab; 45 46 int __init tfrc_tx_packet_history_init(void) 47 { 48 tfrc_tx_hist_slab = kmem_cache_create("tfrc_tx_hist", 49 sizeof(struct tfrc_tx_hist_entry), 50 0, SLAB_HWCACHE_ALIGN, NULL); 51 return tfrc_tx_hist_slab == NULL ? -ENOBUFS : 0; 52 } 53 54 void tfrc_tx_packet_history_exit(void) 55 { 56 if (tfrc_tx_hist_slab != NULL) { 57 kmem_cache_destroy(tfrc_tx_hist_slab); 58 tfrc_tx_hist_slab = NULL; 59 } 60 } 61 62 int tfrc_tx_hist_add(struct tfrc_tx_hist_entry **headp, u64 seqno) 63 { 64 struct tfrc_tx_hist_entry *entry = kmem_cache_alloc(tfrc_tx_hist_slab, gfp_any()); 65 66 if (entry == NULL) 67 return -ENOBUFS; 68 entry->seqno = seqno; 69 entry->stamp = ktime_get_real(); 70 entry->next = *headp; 71 *headp = entry; 72 return 0; 73 } 74 75 void tfrc_tx_hist_purge(struct tfrc_tx_hist_entry **headp) 76 { 77 struct tfrc_tx_hist_entry *head = *headp; 78 79 while (head != NULL) { 80 struct tfrc_tx_hist_entry *next = head->next; 81 82 kmem_cache_free(tfrc_tx_hist_slab, head); 83 head = next; 84 } 85 86 *headp = NULL; 87 } 88 89 /* 90 * Receiver History Routines 91 */ 92 static struct kmem_cache *tfrc_rx_hist_slab; 93 94 int __init tfrc_rx_packet_history_init(void) 95 { 96 tfrc_rx_hist_slab = kmem_cache_create("tfrc_rxh_cache", 97 sizeof(struct tfrc_rx_hist_entry), 98 0, SLAB_HWCACHE_ALIGN, NULL); 99 return tfrc_rx_hist_slab == NULL ? -ENOBUFS : 0; 100 } 101 102 void tfrc_rx_packet_history_exit(void) 103 { 104 if (tfrc_rx_hist_slab != NULL) { 105 kmem_cache_destroy(tfrc_rx_hist_slab); 106 tfrc_rx_hist_slab = NULL; 107 } 108 } 109 110 static inline void tfrc_rx_hist_entry_from_skb(struct tfrc_rx_hist_entry *entry, 111 const struct sk_buff *skb, 112 const u64 ndp) 113 { 114 const struct dccp_hdr *dh = dccp_hdr(skb); 115 116 entry->tfrchrx_seqno = DCCP_SKB_CB(skb)->dccpd_seq; 117 entry->tfrchrx_ccval = dh->dccph_ccval; 118 entry->tfrchrx_type = dh->dccph_type; 119 entry->tfrchrx_ndp = ndp; 120 entry->tfrchrx_tstamp = ktime_get_real(); 121 } 122 123 void tfrc_rx_hist_add_packet(struct tfrc_rx_hist *h, 124 const struct sk_buff *skb, 125 const u64 ndp) 126 { 127 struct tfrc_rx_hist_entry *entry = tfrc_rx_hist_last_rcv(h); 128 129 tfrc_rx_hist_entry_from_skb(entry, skb, ndp); 130 } 131 132 /* has the packet contained in skb been seen before? */ 133 int tfrc_rx_hist_duplicate(struct tfrc_rx_hist *h, struct sk_buff *skb) 134 { 135 const u64 seq = DCCP_SKB_CB(skb)->dccpd_seq; 136 int i; 137 138 if (dccp_delta_seqno(tfrc_rx_hist_loss_prev(h)->tfrchrx_seqno, seq) <= 0) 139 return 1; 140 141 for (i = 1; i <= h->loss_count; i++) 142 if (tfrc_rx_hist_entry(h, i)->tfrchrx_seqno == seq) 143 return 1; 144 145 return 0; 146 } 147 148 static void tfrc_rx_hist_swap(struct tfrc_rx_hist *h, const u8 a, const u8 b) 149 { 150 const u8 idx_a = tfrc_rx_hist_index(h, a), 151 idx_b = tfrc_rx_hist_index(h, b); 152 struct tfrc_rx_hist_entry *tmp = h->ring[idx_a]; 153 154 h->ring[idx_a] = h->ring[idx_b]; 155 h->ring[idx_b] = tmp; 156 } 157 158 /* 159 * Private helper functions for loss detection. 160 * 161 * In the descriptions, `Si' refers to the sequence number of entry number i, 162 * whose NDP count is `Ni' (lower case is used for variables). 163 * Note: All __xxx_loss functions expect that a test against duplicates has been 164 * performed already: the seqno of the skb must not be less than the seqno 165 * of loss_prev; and it must not equal that of any valid history entry. 166 */ 167 static void __do_track_loss(struct tfrc_rx_hist *h, struct sk_buff *skb, u64 n1) 168 { 169 u64 s0 = tfrc_rx_hist_loss_prev(h)->tfrchrx_seqno, 170 s1 = DCCP_SKB_CB(skb)->dccpd_seq; 171 172 if (!dccp_loss_free(s0, s1, n1)) { /* gap between S0 and S1 */ 173 h->loss_count = 1; 174 tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 1), skb, n1); 175 } 176 } 177 178 static void __one_after_loss(struct tfrc_rx_hist *h, struct sk_buff *skb, u32 n2) 179 { 180 u64 s0 = tfrc_rx_hist_loss_prev(h)->tfrchrx_seqno, 181 s1 = tfrc_rx_hist_entry(h, 1)->tfrchrx_seqno, 182 s2 = DCCP_SKB_CB(skb)->dccpd_seq; 183 184 if (likely(dccp_delta_seqno(s1, s2) > 0)) { /* S1 < S2 */ 185 h->loss_count = 2; 186 tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 2), skb, n2); 187 return; 188 } 189 190 /* S0 < S2 < S1 */ 191 192 if (dccp_loss_free(s0, s2, n2)) { 193 u64 n1 = tfrc_rx_hist_entry(h, 1)->tfrchrx_ndp; 194 195 if (dccp_loss_free(s2, s1, n1)) { 196 /* hole is filled: S0, S2, and S1 are consecutive */ 197 h->loss_count = 0; 198 h->loss_start = tfrc_rx_hist_index(h, 1); 199 } else 200 /* gap between S2 and S1: just update loss_prev */ 201 tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_loss_prev(h), skb, n2); 202 203 } else { /* gap between S0 and S2 */ 204 /* 205 * Reorder history to insert S2 between S0 and S1 206 */ 207 tfrc_rx_hist_swap(h, 0, 3); 208 h->loss_start = tfrc_rx_hist_index(h, 3); 209 tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 1), skb, n2); 210 h->loss_count = 2; 211 } 212 } 213 214 /* return 1 if a new loss event has been identified */ 215 static int __two_after_loss(struct tfrc_rx_hist *h, struct sk_buff *skb, u32 n3) 216 { 217 u64 s0 = tfrc_rx_hist_loss_prev(h)->tfrchrx_seqno, 218 s1 = tfrc_rx_hist_entry(h, 1)->tfrchrx_seqno, 219 s2 = tfrc_rx_hist_entry(h, 2)->tfrchrx_seqno, 220 s3 = DCCP_SKB_CB(skb)->dccpd_seq; 221 222 if (likely(dccp_delta_seqno(s2, s3) > 0)) { /* S2 < S3 */ 223 h->loss_count = 3; 224 tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 3), skb, n3); 225 return 1; 226 } 227 228 /* S3 < S2 */ 229 230 if (dccp_delta_seqno(s1, s3) > 0) { /* S1 < S3 < S2 */ 231 /* 232 * Reorder history to insert S3 between S1 and S2 233 */ 234 tfrc_rx_hist_swap(h, 2, 3); 235 tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 2), skb, n3); 236 h->loss_count = 3; 237 return 1; 238 } 239 240 /* S0 < S3 < S1 */ 241 242 if (dccp_loss_free(s0, s3, n3)) { 243 u64 n1 = tfrc_rx_hist_entry(h, 1)->tfrchrx_ndp; 244 245 if (dccp_loss_free(s3, s1, n1)) { 246 /* hole between S0 and S1 filled by S3 */ 247 u64 n2 = tfrc_rx_hist_entry(h, 2)->tfrchrx_ndp; 248 249 if (dccp_loss_free(s1, s2, n2)) { 250 /* entire hole filled by S0, S3, S1, S2 */ 251 h->loss_start = tfrc_rx_hist_index(h, 2); 252 h->loss_count = 0; 253 } else { 254 /* gap remains between S1 and S2 */ 255 h->loss_start = tfrc_rx_hist_index(h, 1); 256 h->loss_count = 1; 257 } 258 259 } else /* gap exists between S3 and S1, loss_count stays at 2 */ 260 tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_loss_prev(h), skb, n3); 261 262 return 0; 263 } 264 265 /* 266 * The remaining case: S0 < S3 < S1 < S2; gap between S0 and S3 267 * Reorder history to insert S3 between S0 and S1. 268 */ 269 tfrc_rx_hist_swap(h, 0, 3); 270 h->loss_start = tfrc_rx_hist_index(h, 3); 271 tfrc_rx_hist_entry_from_skb(tfrc_rx_hist_entry(h, 1), skb, n3); 272 h->loss_count = 3; 273 274 return 1; 275 } 276 277 /* recycle RX history records to continue loss detection if necessary */ 278 static void __three_after_loss(struct tfrc_rx_hist *h) 279 { 280 /* 281 * At this stage we know already that there is a gap between S0 and S1 282 * (since S0 was the highest sequence number received before detecting 283 * the loss). To recycle the loss record, it is thus only necessary to 284 * check for other possible gaps between S1/S2 and between S2/S3. 285 */ 286 u64 s1 = tfrc_rx_hist_entry(h, 1)->tfrchrx_seqno, 287 s2 = tfrc_rx_hist_entry(h, 2)->tfrchrx_seqno, 288 s3 = tfrc_rx_hist_entry(h, 3)->tfrchrx_seqno; 289 u64 n2 = tfrc_rx_hist_entry(h, 2)->tfrchrx_ndp, 290 n3 = tfrc_rx_hist_entry(h, 3)->tfrchrx_ndp; 291 292 if (dccp_loss_free(s1, s2, n2)) { 293 294 if (dccp_loss_free(s2, s3, n3)) { 295 /* no gap between S2 and S3: entire hole is filled */ 296 h->loss_start = tfrc_rx_hist_index(h, 3); 297 h->loss_count = 0; 298 } else { 299 /* gap between S2 and S3 */ 300 h->loss_start = tfrc_rx_hist_index(h, 2); 301 h->loss_count = 1; 302 } 303 304 } else { /* gap between S1 and S2 */ 305 h->loss_start = tfrc_rx_hist_index(h, 1); 306 h->loss_count = 2; 307 } 308 } 309 310 /** 311 * tfrc_rx_handle_loss - Loss detection and further processing 312 * @h: The non-empty RX history object 313 * @lh: Loss Intervals database to update 314 * @skb: Currently received packet 315 * @ndp: The NDP count belonging to @skb 316 * @calc_first_li: Caller-dependent computation of first loss interval in @lh 317 * @sk: Used by @calc_first_li (see tfrc_lh_interval_add) 318 * 319 * Chooses action according to pending loss, updates LI database when a new 320 * loss was detected, and does required post-processing. Returns 1 when caller 321 * should send feedback, 0 otherwise. 322 * Since it also takes care of reordering during loss detection and updates the 323 * records accordingly, the caller should not perform any more RX history 324 * operations when loss_count is greater than 0 after calling this function. 325 */ 326 int tfrc_rx_handle_loss(struct tfrc_rx_hist *h, 327 struct tfrc_loss_hist *lh, 328 struct sk_buff *skb, const u64 ndp, 329 u32 (*calc_first_li)(struct sock *), struct sock *sk) 330 { 331 int is_new_loss = 0; 332 333 if (h->loss_count == 0) { 334 __do_track_loss(h, skb, ndp); 335 } else if (h->loss_count == 1) { 336 __one_after_loss(h, skb, ndp); 337 } else if (h->loss_count != 2) { 338 DCCP_BUG("invalid loss_count %d", h->loss_count); 339 } else if (__two_after_loss(h, skb, ndp)) { 340 /* 341 * Update Loss Interval database and recycle RX records 342 */ 343 is_new_loss = tfrc_lh_interval_add(lh, h, calc_first_li, sk); 344 __three_after_loss(h); 345 } 346 return is_new_loss; 347 } 348 349 int tfrc_rx_hist_alloc(struct tfrc_rx_hist *h) 350 { 351 int i; 352 353 for (i = 0; i <= TFRC_NDUPACK; i++) { 354 h->ring[i] = kmem_cache_alloc(tfrc_rx_hist_slab, GFP_ATOMIC); 355 if (h->ring[i] == NULL) 356 goto out_free; 357 } 358 359 h->loss_count = h->loss_start = 0; 360 return 0; 361 362 out_free: 363 while (i-- != 0) { 364 kmem_cache_free(tfrc_rx_hist_slab, h->ring[i]); 365 h->ring[i] = NULL; 366 } 367 return -ENOBUFS; 368 } 369 370 void tfrc_rx_hist_purge(struct tfrc_rx_hist *h) 371 { 372 int i; 373 374 for (i = 0; i <= TFRC_NDUPACK; ++i) 375 if (h->ring[i] != NULL) { 376 kmem_cache_free(tfrc_rx_hist_slab, h->ring[i]); 377 h->ring[i] = NULL; 378 } 379 } 380 381 /** 382 * tfrc_rx_hist_rtt_last_s - reference entry to compute RTT samples against 383 */ 384 static inline struct tfrc_rx_hist_entry * 385 tfrc_rx_hist_rtt_last_s(const struct tfrc_rx_hist *h) 386 { 387 return h->ring[0]; 388 } 389 390 /** 391 * tfrc_rx_hist_rtt_prev_s - previously suitable (wrt rtt_last_s) RTT-sampling entry 392 */ 393 static inline struct tfrc_rx_hist_entry * 394 tfrc_rx_hist_rtt_prev_s(const struct tfrc_rx_hist *h) 395 { 396 return h->ring[h->rtt_sample_prev]; 397 } 398 399 /** 400 * tfrc_rx_hist_sample_rtt - Sample RTT from timestamp / CCVal 401 * Based on ideas presented in RFC 4342, 8.1. Returns 0 if it was not able 402 * to compute a sample with given data - calling function should check this. 403 */ 404 u32 tfrc_rx_hist_sample_rtt(struct tfrc_rx_hist *h, const struct sk_buff *skb) 405 { 406 u32 sample = 0, 407 delta_v = SUB16(dccp_hdr(skb)->dccph_ccval, 408 tfrc_rx_hist_rtt_last_s(h)->tfrchrx_ccval); 409 410 if (delta_v < 1 || delta_v > 4) { /* unsuitable CCVal delta */ 411 if (h->rtt_sample_prev == 2) { /* previous candidate stored */ 412 sample = SUB16(tfrc_rx_hist_rtt_prev_s(h)->tfrchrx_ccval, 413 tfrc_rx_hist_rtt_last_s(h)->tfrchrx_ccval); 414 if (sample) 415 sample = 4 / sample * 416 ktime_us_delta(tfrc_rx_hist_rtt_prev_s(h)->tfrchrx_tstamp, 417 tfrc_rx_hist_rtt_last_s(h)->tfrchrx_tstamp); 418 else /* 419 * FIXME: This condition is in principle not 420 * possible but occurs when CCID is used for 421 * two-way data traffic. I have tried to trace 422 * it, but the cause does not seem to be here. 423 */ 424 DCCP_BUG("please report to dccp@vger.kernel.org" 425 " => prev = %u, last = %u", 426 tfrc_rx_hist_rtt_prev_s(h)->tfrchrx_ccval, 427 tfrc_rx_hist_rtt_last_s(h)->tfrchrx_ccval); 428 } else if (delta_v < 1) { 429 h->rtt_sample_prev = 1; 430 goto keep_ref_for_next_time; 431 } 432 433 } else if (delta_v == 4) /* optimal match */ 434 sample = ktime_to_us(net_timedelta(tfrc_rx_hist_rtt_last_s(h)->tfrchrx_tstamp)); 435 else { /* suboptimal match */ 436 h->rtt_sample_prev = 2; 437 goto keep_ref_for_next_time; 438 } 439 440 if (unlikely(sample > DCCP_SANE_RTT_MAX)) { 441 DCCP_WARN("RTT sample %u too large, using max\n", sample); 442 sample = DCCP_SANE_RTT_MAX; 443 } 444 445 h->rtt_sample_prev = 0; /* use current entry as next reference */ 446 keep_ref_for_next_time: 447 448 return sample; 449 } 450