src/addrman.h

   1 // Copyright (c) 2012 Pieter Wuille
   2 // Distributed under the MIT/X11 software license, see the accompanying
   3 // file COPYING or http://www.opensource.org/licenses/mit-license.php.
   4 #ifndef _BITCOIN_ADDRMAN
   5 #define _BITCOIN_ADDRMAN 1
   6
   7 #include "netbase.h"
   8 #include "protocol.h"
   9 #include "util.h"
  10 #include "sync.h"
  11
  12
  13 #include <map>
  14 #include <vector>
  15
  16 #include <openssl/rand.h>
  17
  18
  19 /** Extended statistics about a CAddress */
  20 class CAddrInfo : public CAddress
  21 {
  22 private:
  23     // where knowledge about this address first came from
  24     CNetAddr source;
  25
  26     // last successful connection by us
  27     int64_t nLastSuccess;
  28
  29     // last try whatsoever by us:
  30     // int64_t CAddress::nLastTry
  31
  32     // connection attempts since last successful attempt
  33     int nAttempts;
  34
  35     // reference count in new sets (memory only)
  36     int nRefCount;
  37
  38     // in tried set? (memory only)
  39     bool fInTried;
  40
  41     // position in vRandom
  42     int nRandomPos;
  43
  44     friend class CAddrMan;
  45
  46 public:
  47
  48     IMPLEMENT_SERIALIZE(
  49         CAddress* pthis = (CAddress*)(this);
  50         READWRITE(*pthis);
  51         READWRITE(source);
  52         READWRITE(nLastSuccess);
  53         READWRITE(nAttempts);
  54     )
  55
  56     void Init()
  57     {
  58         nLastSuccess = 0;
  59         nLastTry = 0;
  60         nAttempts = 0;
  61         nRefCount = 0;
  62         fInTried = false;
  63         nRandomPos = -1;
  64     }
  65
  66     CAddrInfo(const CAddress &addrIn, const CNetAddr &addrSource) : CAddress(addrIn), source(addrSource)
  67     {
  68         Init();
  69     }
  70
  71     CAddrInfo() : CAddress(), source()
  72     {
  73         Init();
  74     }
  75
  76     // Calculate in which "tried" bucket this entry belongs
  77     int GetTriedBucket(const std::vector<unsigned char> &nKey) const;
  78
  79     // Calculate in which "new" bucket this entry belongs, given a certain source
  80     int GetNewBucket(const std::vector<unsigned char> &nKey, const CNetAddr& src) const;
  81
  82     // Calculate in which "new" bucket this entry belongs, using its default source
  83     int GetNewBucket(const std::vector<unsigned char> &nKey) const
  84     {
  85         return GetNewBucket(nKey, source);
  86     }
  87
  88     // Determine whether the statistics about this entry are bad enough so that it can just be deleted
  89     bool IsTerrible(int64_t nNow = GetAdjustedTime()) const;
  90
  91     // Calculate the relative chance this entry should be given when selecting nodes to connect to
  92     double GetChance(int64_t nNow = GetAdjustedTime()) const;
  93
  94 };
  95
  96 // Stochastic address manager
  97 //
  98 // Design goals:
  99 //  * Only keep a limited number of addresses around, so that addr.dat and memory requirements do not grow without bound.
 100 //  * Keep the address tables in-memory, and asynchronously dump the entire to able in addr.dat.
 101 //  * Make sure no (localized) attacker can fill the entire table with his nodes/addresses.
 102 //
 103 // To that end:
 104 //  * Addresses are organized into buckets.
 105 //    * Address that have not yet been tried go into 256 "new" buckets.
 106 //      * Based on the address range (/16 for IPv4) of source of the information, 32 buckets are selected at random
 107 //      * The actual bucket is chosen from one of these, based on the range the address itself is located.
 108 //      * One single address can occur in up to 4 different buckets, to increase selection chances for addresses that
 109 //        are seen frequently. The chance for increasing this multiplicity decreases exponentially.
 110 //      * When adding a new address to a full bucket, a randomly chosen entry (with a bias favoring less recently seen
 111 //        ones) is removed from it first.
 112 //    * Addresses of nodes that are known to be accessible go into 64 "tried" buckets.
 113 //      * Each address range selects at random 4 of these buckets.
 114 //      * The actual bucket is chosen from one of these, based on the full address.
 115 //      * When adding a new good address to a full bucket, a randomly chosen entry (with a bias favoring less recently
 116 //        tried ones) is evicted from it, back to the "new" buckets.
 117 //    * Bucket selection is based on cryptographic hashing, using a randomly-generated 256-bit key, which should not
 118 //      be observable by adversaries.
 119 //    * Several indexes are kept for high performance. Defining DEBUG_ADDRMAN will introduce frequent (and expensive)
 120 //      consistency checks for the entire data structure.
 121
 122 // total number of buckets for tried addresses
 123 #define ADDRMAN_TRIED_BUCKET_COUNT 64
 124
 125 // maximum allowed number of entries in buckets for tried addresses
 126 #define ADDRMAN_TRIED_BUCKET_SIZE 64
 127
 128 // total number of buckets for new addresses
 129 #define ADDRMAN_NEW_BUCKET_COUNT 256
 130
 131 // maximum allowed number of entries in buckets for new addresses
 132 #define ADDRMAN_NEW_BUCKET_SIZE 64
 133
 134 // over how many buckets entries with tried addresses from a single group (/16 for IPv4) are spread
 135 #define ADDRMAN_TRIED_BUCKETS_PER_GROUP 4
 136
 137 // over how many buckets entries with new addresses originating from a single group are spread
 138 #define ADDRMAN_NEW_BUCKETS_PER_SOURCE_GROUP 32
 139
 140 // in how many buckets for entries with new addresses a single address may occur
 141 #define ADDRMAN_NEW_BUCKETS_PER_ADDRESS 4
 142
 143 // how many entries in a bucket with tried addresses are inspected, when selecting one to replace
 144 #define ADDRMAN_TRIED_ENTRIES_INSPECT_ON_EVICT 4
 145
 146 // how old addresses can maximally be
 147 #define ADDRMAN_HORIZON_DAYS 30
 148
 149 // after how many failed attempts we give up on a new node
 150 #define ADDRMAN_RETRIES 3
 151
 152 // how many successive failures are allowed ...
 153 #define ADDRMAN_MAX_FAILURES 10
 154
 155 // ... in at least this many days
 156 #define ADDRMAN_MIN_FAIL_DAYS 7
 157
 158 // the maximum percentage of nodes to return in a getaddr call
 159 #define ADDRMAN_GETADDR_MAX_PCT 23
 160
 161 // the maximum number of nodes to return in a getaddr call
 162 #define ADDRMAN_GETADDR_MAX 2500
 163
 164 /** Stochastical (IP) address manager */
 165 class CAddrMan
 166 {
 167 private:
 168     // critical section to protect the inner data structures
 169     mutable CCriticalSection cs;
 170
 171     // secret key to randomize bucket select with
 172     std::vector<unsigned char> nKey;
 173
 174     // last used nId
 175     int nIdCount;
 176
 177     // table with information about all nIds
 178     std::map<int, CAddrInfo> mapInfo;
 179
 180     // find an nId based on its network address
 181     std::map<CNetAddr, int> mapAddr;
 182
 183     // randomly-ordered vector of all nIds
 184     std::vector<int> vRandom;
 185
 186     // number of "tried" entries
 187     int nTried;
 188
 189     // list of "tried" buckets
 190     std::vector<std::vector<int> > vvTried;
 191
 192     // number of (unique) "new" entries
 193     int nNew;
 194
 195     // list of "new" buckets
 196     std::vector<std::set<int> > vvNew;
 197
 198 protected:
 199
 200     // Find an entry.
 201     CAddrInfo* Find(const CNetAddr& addr, int *pnId = NULL);
 202
 203     // find an entry, creating it if necessary.
 204     // nTime and nServices of found node is updated, if necessary.
 205     CAddrInfo* Create(const CAddress &addr, const CNetAddr &addrSource, int *pnId = NULL);
 206
 207     // Swap two elements in vRandom.
 208     void SwapRandom(unsigned int nRandomPos1, unsigned int nRandomPos2);
 209
 210     // Return position in given bucket to replace.
 211     int SelectTried(int nKBucket);
 212
 213     // Remove an element from a "new" bucket.
 214     // This is the only place where actual deletes occur.
 215     // They are never deleted while in the "tried" table, only possibly evicted back to the "new" table.
 216     int ShrinkNew(int nUBucket);
 217
 218     // Move an entry from the "new" table(s) to the "tried" table
 219     // @pre vvUnkown[nOrigin].count(nId) != 0
 220     void MakeTried(CAddrInfo& info, int nId, int nOrigin);
 221
 222     // Mark an entry "good", possibly moving it from "new" to "tried".
 223     void Good_(const CService &addr, int64_t nTime);
 224
 225     // Add an entry to the "new" table.
 226     bool Add_(const CAddress &addr, const CNetAddr& source, int64_t nTimePenalty);
 227
 228     // Mark an entry as attempted to connect.
 229     void Attempt_(const CService &addr, int64_t nTime);
 230
 231     // Select an address to connect to.
 232     // nUnkBias determines how much to favor new addresses over tried ones (min=0, max=100)
 233     CAddress Select_(int nUnkBias);
 234
 235 #ifdef DEBUG_ADDRMAN
 236     // Perform consistency check. Returns an error code or zero.
 237     int Check_();
 238 #endif
 239
 240     // Select several addresses at once.
 241     void GetAddr_(std::vector<CAddress> &vAddr);
 242
 243     // Mark an entry as currently-connected-to.
 244     void Connected_(const CService &addr, int64_t nTime);
 245
 246 public:
 247
 248 #ifndef _MSC_VER
 249     IMPLEMENT_SERIALIZE
 250     (({
 251         // serialized format:
 252         // * version byte (currently 0)
 253         // * nKey
 254         // * nNew
 255         // * nTried
 256         // * number of "new" buckets
 257         // * all nNew addrinfos in vvNew
 258         // * all nTried addrinfos in vvTried
 259         // * for each bucket:
 260         //   * number of elements
 261         //   * for each element: index
 262         //
 263         // Notice that vvTried, mapAddr and vVector are never encoded explicitly;
 264         // they are instead reconstructed from the other information.
 265         //
 266         // vvNew is serialized, but only used if ADDRMAN_UNKOWN_BUCKET_COUNT didn't change,
 267         // otherwise it is reconstructed as well.
 268         //
 269         // This format is more complex, but significantly smaller (at most 1.5 MiB), and supports
 270         // changes to the ADDRMAN_ parameters without breaking the on-disk structure.
 271         {
 272             LOCK(cs);
 273             unsigned char nVersion = 0;
 274             READWRITE(nVersion);
 275             READWRITE(nKey);
 276             READWRITE(nNew);
 277             READWRITE(nTried);
 278
 279             CAddrMan *am = const_cast<CAddrMan*>(this);
 280             if (fWrite)
 281             {
 282                 int nUBuckets = ADDRMAN_NEW_BUCKET_COUNT;
 283                 READWRITE(nUBuckets);
 284                 std::map<int, int> mapUnkIds;
 285                 int nIds = 0;
 286                 for (std::map<int, CAddrInfo>::iterator it = am->mapInfo.begin(); it != am->mapInfo.end(); it++)
 287                 {
 288                     if (nIds == nNew) break; // this means nNew was wrong, oh ow
 289                     mapUnkIds[(*it).first] = nIds;
 290                     CAddrInfo &info = (*it).second;
 291                     if (info.nRefCount)
 292                     {
 293                         READWRITE(info);
 294                         nIds++;
 295                     }
 296                 }
 297                 nIds = 0;
 298                 for (std::map<int, CAddrInfo>::iterator it = am->mapInfo.begin(); it != am->mapInfo.end(); it++)
 299                 {
 300                     if (nIds == nTried) break; // this means nTried was wrong, oh ow
 301                     CAddrInfo &info = (*it).second;
 302                     if (info.fInTried)
 303                     {
 304                         READWRITE(info);
 305                         nIds++;
 306                     }
 307                 }
 308                 for (std::vector<std::set<int> >::iterator it = am->vvNew.begin(); it != am->vvNew.end(); it++)
 309                 {
 310                     const std::set<int> &vNew = (*it);
 311                     int nSize = vNew.size();
 312                     READWRITE(nSize);
 313                     for (std::set<int>::iterator it2 = vNew.begin(); it2 != vNew.end(); it2++)
 314                     {
 315                         int nIndex = mapUnkIds[*it2];
 316                         READWRITE(nIndex);
 317                     }
 318                 }
 319             } else {
 320                 int nUBuckets = 0;
 321                 READWRITE(nUBuckets);
 322                 am->nIdCount = 0;
 323                 am->mapInfo.clear();
 324                 am->mapAddr.clear();
 325                 am->vRandom.clear();
 326                 am->vvTried = std::vector<std::vector<int> >(ADDRMAN_TRIED_BUCKET_COUNT, std::vector<int>(0));
 327                 am->vvNew = std::vector<std::set<int> >(ADDRMAN_NEW_BUCKET_COUNT, std::set<int>());
 328                 for (int n = 0; n < am->nNew; n++)
 329                 {
 330                     CAddrInfo &info = am->mapInfo[n];
 331                     READWRITE(info);
 332                     am->mapAddr[info] = n;
 333                     info.nRandomPos = vRandom.size();
 334                     am->vRandom.push_back(n);
 335                     if (nUBuckets != ADDRMAN_NEW_BUCKET_COUNT)
 336                     {
 337                         am->vvNew[info.GetNewBucket(am->nKey)].insert(n);
 338                         info.nRefCount++;
 339                     }
 340                 }
 341                 am->nIdCount = am->nNew;
 342                 int nLost = 0;
 343                 for (int n = 0; n < am->nTried; n++)
 344                 {
 345                     CAddrInfo info;
 346                     READWRITE(info);
 347                     std::vector<int> &vTried = am->vvTried[info.GetTriedBucket(am->nKey)];
 348                     if (vTried.size() < ADDRMAN_TRIED_BUCKET_SIZE)
 349                     {
 350                         info.nRandomPos = vRandom.size();
 351                         info.fInTried = true;
 352                         am->vRandom.push_back(am->nIdCount);
 353                         am->mapInfo[am->nIdCount] = info;
 354                         am->mapAddr[info] = am->nIdCount;
 355                         vTried.push_back(am->nIdCount);
 356                         am->nIdCount++;
 357                     } else {
 358                         nLost++;
 359                     }
 360                 }
 361                 am->nTried -= nLost;
 362                 for (int b = 0; b < nUBuckets; b++)
 363                 {
 364                     std::set<int> &vNew = am->vvNew[b];
 365                     int nSize = 0;
 366                     READWRITE(nSize);
 367                     for (int n = 0; n < nSize; n++)
 368                     {
 369                         int nIndex = 0;
 370                         READWRITE(nIndex);
 371                         CAddrInfo &info = am->mapInfo[nIndex];
 372                         if (nUBuckets == ADDRMAN_NEW_BUCKET_COUNT && info.nRefCount < ADDRMAN_NEW_BUCKETS_PER_ADDRESS)
 373                         {
 374                             info.nRefCount++;
 375                             vNew.insert(nIndex);
 376                         }
 377                     }
 378                 }
 379             }
 380         }
 381     });)
 382 #else
 383     IMPLEMENT_SERIALIZE
 384             (
 385             LOCK(cs);
 386                 unsigned char
 387                     nVersion = 0;
 388
 389                 READWRITE(nVersion);
 390                 READWRITE(nKey);
 391                 READWRITE(nNew);
 392                 READWRITE(nTried);
 393
 394                 CAddrMan
 395                     *am = const_cast<CAddrMan*>(this);
 396
 397                 if (fWrite)
 398                 {
 399                     int
 400                         nUBuckets = ADDRMAN_NEW_BUCKET_COUNT;
 401
 402                     READWRITE(nUBuckets);
 403 /************
 404                     std::map<int, int>
 405                         mapUnkIds;
 406 ************/
 407                     int
 408                         nIds = 0;
 409                     for (std::map<int, CAddrInfo>::iterator it = am->mapInfo.begin(); it != am->mapInfo.end(); it++)
 410                     {
 411                         if (nIds == nNew)
 412                             break; // this means nNew was wrong, oh ow
 413                         mapUnkIds[(*it).first] = nIds;
 414
 415                         CAddrInfo
 416                             &info = (*it).second;
 417
 418                         if (info.nRefCount)
 419                         {
 420                             READWRITE(info);
 421                             nIds++;
 422                         }
 423                     }
 424                     nIds = 0;
 425                     for (std::map<int, CAddrInfo>::iterator it = am->mapInfo.begin(); it != am->mapInfo.end(); it++)
 426                     {
 427                         if (nIds == nTried)
 428                             break; /* this means nTried was wrong, oh ow */
 429
 430                         CAddrInfo
 431                             &info = (*it).second;
 432
 433                         if (info.fInTried)
 434                         {
 435                             READWRITE(info);
 436                             nIds++;
 437                         }
 438                     }
 439                     for (
 440                          std::vector<std::set<int> >::iterator it = am->vvNew.begin();
 441                          it != am->vvNew.end();
 442                          it++
 443                         )
 444                     {
 445                         std::set<int>
 446                             &vNew = (*it);
 447
 448                         int
 449                             nSize = int( vNew.size() );
 450
 451                         READWRITE(nSize);
 452                         for (std::set<int>::iterator it2 = vNew.begin(); it2 != vNew.end(); it2++)
 453                         {
 454                         int
 455                             nIndex = mapUnkIds[*it2];
 456                         READWRITE(nIndex);
 457                         }
 458                     }
 459                 }
 460                 else
 461                 {
 462                     int
 463                         nUBuckets = 0;
 464
 465                     READWRITE(nUBuckets);
 466                     am->nIdCount = 0;
 467                     am->mapInfo.clear();
 468                     am->mapAddr.clear();
 469                     am->vRandom.clear();
 470                     am->vvTried =
 471                         std::vector<std::vector<int> >(
 472                                                         ADDRMAN_TRIED_BUCKET_COUNT,
 473                                                         std::vector<int>(0)
 474                                                       );
 475                     am->vvNew =
 476                         std::vector<std::set<int> >(
 477                                                     ADDRMAN_NEW_BUCKET_COUNT,
 478                                                     std::set<int>()
 479                                                    );
 480                     for (int n = 0; n < am->nNew; n++)
 481                     {
 482                         CAddrInfo
 483                             &info = am->mapInfo[n];
 484
 485                         READWRITE(info);
 486                         am->mapAddr[info] = n;
 487                         info.nRandomPos = int( vRandom.size() );
 488                         am->vRandom.push_back(n);
 489                         if (nUBuckets != ADDRMAN_NEW_BUCKET_COUNT)
 490                         {
 491                             am->vvNew[info.GetNewBucket(am->nKey)].insert(n);
 492                             info.nRefCount++;
 493                         }
 494                     }
 495                     am->nIdCount = am->nNew;
 496
 497                     int
 498                         nLost = 0;
 499
 500                     for (int n = 0; n < am->nTried; n++)
 501                     {
 502                         CAddrInfo
 503                             info;
 504
 505                         READWRITE(info);
 506
 507                         std::vector<int>
 508                             &vTried = am->vvTried[info.GetTriedBucket(am->nKey)];
 509
 510                         if (vTried.size() < ADDRMAN_TRIED_BUCKET_SIZE)
 511                         {
 512                             info.nRandomPos = int( vRandom.size() );
 513                             info.fInTried = true;
 514                             am->vRandom.push_back(am->nIdCount);
 515                             am->mapInfo[am->nIdCount] = info;
 516                             am->mapAddr[info] = am->nIdCount;
 517                             vTried.push_back(am->nIdCount);
 518                             am->nIdCount++;
 519                         }
 520                         else
 521                         {
 522                             nLost++;
 523                         }
 524                     }
 525                     am->nTried -= nLost;
 526                     for (int b = 0; b < nUBuckets; b++)
 527                     {
 528                         std::set<int>
 529                             &vNew = am->vvNew[b];
 530
 531                         int
 532                             nSize = 0;
 533
 534                         READWRITE(nSize);
 535                         for (int n = 0; n < nSize; n++)
 536                         {
 537                             int
 538                                 nIndex = 0;
 539
 540                             READWRITE(nIndex);
 541
 542                             CAddrInfo
 543                                 &info = am->mapInfo[nIndex];
 544
 545                             if (
 546                                 (nUBuckets == ADDRMAN_NEW_BUCKET_COUNT) &&
 547                                 (info.nRefCount < ADDRMAN_NEW_BUCKETS_PER_ADDRESS)
 548                                )
 549                             {
 550                                 info.nRefCount++;
 551                                 vNew.insert(nIndex);
 552                             }
 553                         }
 554                     }
 555                 }
 556             )
 557
 558 #endif  //_MSC_VER
 559     CAddrMan() : vRandom(0), vvTried(ADDRMAN_TRIED_BUCKET_COUNT, std::vector<int>(0)), vvNew(ADDRMAN_NEW_BUCKET_COUNT, std::set<int>())
 560     {
 561          nKey.resize(32);
 562          RAND_bytes(&nKey[0], 32);
 563
 564          nIdCount = 0;
 565          nTried = 0;
 566          nNew = 0;
 567     }
 568
 569     // Return the number of (unique) addresses in all tables.
 570     int size()
 571     {
 572         return vRandom.size();
 573     }
 574
 575     // Consistency check
 576     void Check()
 577     {
 578 #ifdef DEBUG_ADDRMAN
 579         {
 580             LOCK(cs);
 581             int err;
 582             if ((err=Check_()))
 583                 printf("ADDRMAN CONSISTENCY CHECK FAILED!!! err=%i\n", err);
 584         }
 585 #endif
 586     }
 587
 588     // Add a single address.
 589     bool Add(const CAddress &addr, const CNetAddr& source, int64_t nTimePenalty = 0)
 590     {
 591         bool fRet = false;
 592         {
 593             LOCK(cs);
 594             Check();
 595             fRet |= Add_(addr, source, nTimePenalty);
 596             Check();
 597         }
 598         if (fRet)
 599             printf("Added %s from %s: %i tried, %i new\n", addr.ToStringIPPort().c_str(), source.ToString().c_str(), nTried, nNew);
 600         return fRet;
 601     }
 602
 603     // Add multiple addresses.
 604     bool Add(const std::vector<CAddress> &vAddr, const CNetAddr& source, int64_t nTimePenalty = 0)
 605     {
 606         int nAdd = 0;
 607         {
 608             LOCK(cs);
 609             Check();
 610             for (std::vector<CAddress>::const_iterator it = vAddr.begin(); it != vAddr.end(); it++)
 611                 nAdd += Add_(*it, source, nTimePenalty) ? 1 : 0;
 612             Check();
 613         }
 614         if (nAdd)
 615             printf("Added %i addresses from %s: %i tried, %i new\n", nAdd, source.ToString().c_str(), nTried, nNew);
 616         return nAdd > 0;
 617     }
 618
 619     // Mark an entry as accessible.
 620     void Good(const CService &addr, int64_t nTime = GetAdjustedTime())
 621     {
 622         {
 623             LOCK(cs);
 624             Check();
 625             Good_(addr, nTime);
 626             Check();
 627         }
 628     }
 629
 630     // Mark an entry as connection attempted to.
 631     void Attempt(const CService &addr, int64_t nTime = GetAdjustedTime())
 632     {
 633         {
 634             LOCK(cs);
 635             Check();
 636             Attempt_(addr, nTime);
 637             Check();
 638         }
 639     }
 640
 641     // Choose an address to connect to.
 642     // nUnkBias determines how much "new" entries are favored over "tried" ones (0-100).
 643     CAddress Select(int nUnkBias = 50)
 644     {
 645         CAddress addrRet;
 646         {
 647             LOCK(cs);
 648             Check();
 649             addrRet = Select_(nUnkBias);
 650             Check();
 651         }
 652         return addrRet;
 653     }
 654
 655     // Return a bunch of addresses, selected at random.
 656     std::vector<CAddress> GetAddr()
 657     {
 658         Check();
 659         std::vector<CAddress> vAddr;
 660         {
 661             LOCK(cs);
 662             GetAddr_(vAddr);
 663         }
 664         Check();
 665         return vAddr;
 666     }
 667
 668     // Mark an entry as currently-connected-to.
 669     void Connected(const CService &addr, int64_t nTime = GetAdjustedTime())
 670     {
 671         {
 672             LOCK(cs);
 673             Check();
 674             Connected_(addr, nTime);
 675             Check();
 676         }
 677     }
 678 };
 679
 680 #endif