test/fuzzinvariants.c

   1 /*
   2 ** 2022-06-14
   3 **
   4 ** The author disclaims copyright to this source code.  In place of
   5 ** a legal notice, here is a blessing:
   6 **
   7 **    May you do good and not evil.
   8 **    May you find forgiveness for yourself and forgive others.
   9 **    May you share freely, never taking more than you give.
  10 **
  11 *************************************************************************
  12 **
  13 ** This library is used by fuzzcheck to test query invariants.
  14 **
  15 ** An sqlite3_stmt is passed in that has just returned SQLITE_ROW.  This
  16 ** routine does:
  17 **
  18 **     *   Record the output of the current row
  19 **     *   Construct an alternative query that should return the same row
  20 **     *   Run the alternative query and verify that it does in fact return
  21 **         the same row
  22 **
  23 */
  24 #include "sqlite3.h"
  25 #include <stdio.h>
  26 #include <stdlib.h>
  27 #include <string.h>
  28 #include <ctype.h>
  29
  30 /* Forward references */
  31 static char *fuzz_invariant_sql(sqlite3_stmt*, int);
  32 static int sameValue(sqlite3_stmt*,int,sqlite3_stmt*,int,sqlite3_stmt*);
  33 static void reportInvariantFailed(sqlite3_stmt*,sqlite3_stmt*,int);
  34
  35 /*
  36 ** Do an invariant check on pStmt.  iCnt determines which invariant check to
  37 ** perform.  The first check is iCnt==0.
  38 **
  39 ** *pbCorrupt is a flag that, if true, indicates that the database file
  40 ** is known to be corrupt.  A value of non-zero means "yes, the database
  41 ** is corrupt".  A zero value means "we do not know whether or not the
  42 ** database is corrupt".  The value might be set prior to entry, or this
  43 ** routine might set the value.
  44 **
  45 ** Return values:
  46 **
  47 **     SQLITE_OK          This check was successful.
  48 **
  49 **     SQLITE_DONE        iCnt is out of range.  The caller typically sets
  50 **                        up a loop on iCnt starting with zero, and increments
  51 **                        iCnt until this code is returned.
  52 **
  53 **     SQLITE_CORRUPT     The invariant failed, but the underlying database
  54 **                        file is indicating that it is corrupt, which might
  55 **                        be the cause of the malfunction.  The *pCorrupt
  56 **                        value will also be set.
  57 **
  58 **     SQLITE_INTERNAL    The invariant failed, and the database file is not
  59 **                        corrupt.  (This never happens because this function
  60 **                        will call abort() following an invariant failure.)
  61 **
  62 **     (other)            Some other kind of error occurred.
  63 */
  64 int fuzz_invariant(
  65   sqlite3 *db,            /* The database connection */
  66   sqlite3_stmt *pStmt,    /* Test statement stopped on an SQLITE_ROW */
  67   int iCnt,               /* Invariant sequence number, starting at 0 */
  68   int iRow,               /* Current row number */
  69   int nRow,               /* Number of output rows from pStmt */
  70   int *pbCorrupt,         /* IN/OUT: Flag indicating a corrupt database file */
  71   int eVerbosity          /* How much debugging output */
  72 ){
  73   char *zTest;
  74   sqlite3_stmt *pTestStmt = 0;
  75   int rc;
  76   int i;
  77   int nCol;
  78   int nParam;
  79
  80   if( *pbCorrupt ) return SQLITE_DONE;
  81   nParam = sqlite3_bind_parameter_count(pStmt);
  82   if( nParam>100 ) return SQLITE_DONE;
  83   zTest = fuzz_invariant_sql(pStmt, iCnt);
  84   if( zTest==0 ) return SQLITE_DONE;
  85   rc = sqlite3_prepare_v2(db, zTest, -1, &pTestStmt, 0);
  86   if( rc ){
  87     if( eVerbosity ){
  88       printf("invariant compile failed: %s\n%s\n",
  89              sqlite3_errmsg(db), zTest);
  90     }
  91     sqlite3_free(zTest);
  92     sqlite3_finalize(pTestStmt);
  93     return rc;
  94   }
  95   sqlite3_free(zTest);
  96   nCol = sqlite3_column_count(pStmt);
  97   for(i=0; i<nCol; i++){
  98     rc = sqlite3_bind_value(pTestStmt,i+1+nParam,sqlite3_column_value(pStmt,i));
  99     if( rc!=SQLITE_OK && rc!=SQLITE_RANGE ){
 100       sqlite3_finalize(pTestStmt);
 101       return rc;
 102     }
 103   }
 104   if( eVerbosity>=2 ){
 105     char *zSql = sqlite3_expanded_sql(pTestStmt);
 106     printf("invariant-sql #%d:\n%s\n", iCnt, zSql);
 107     sqlite3_free(zSql);
 108   }
 109   while( (rc = sqlite3_step(pTestStmt))==SQLITE_ROW ){
 110     for(i=0; i<nCol; i++){
 111       if( !sameValue(pStmt, i, pTestStmt, i, 0) ) break;
 112     }
 113     if( i>=nCol ) break;
 114   }
 115   if( rc==SQLITE_DONE ){
 116     /* No matching output row found */
 117     sqlite3_stmt *pCk = 0;
 118
 119     /* This is not a fault if the database file is corrupt, because anything
 120     ** can happen with a corrupt database file */
 121     rc = sqlite3_prepare_v2(db, "PRAGMA integrity_check", -1, &pCk, 0);
 122     if( rc ){
 123       sqlite3_finalize(pCk);
 124       sqlite3_finalize(pTestStmt);
 125       return rc;
 126     }
 127     rc = sqlite3_step(pCk);
 128     if( rc!=SQLITE_ROW
 129      || sqlite3_column_text(pCk, 0)==0
 130      || strcmp((const char*)sqlite3_column_text(pCk,0),"ok")!=0
 131     ){
 132       *pbCorrupt = 1;
 133       sqlite3_finalize(pCk);
 134       sqlite3_finalize(pTestStmt);
 135       return SQLITE_CORRUPT;
 136     }
 137     sqlite3_finalize(pCk);
 138
 139     if( sqlite3_strlike("%group%by%",sqlite3_sql(pStmt),0)==0 ){
 140       /*
 141       ** If there is a GROUP BY clause, it might not cover every term in the
 142       ** output.  And then non-covered terms can take on a value from any
 143       ** row in the result set.  This can cause differing answers.
 144       */
 145       goto not_a_fault;
 146     }
 147
 148     if( sqlite3_strlike("%limit%)%order%by%", sqlite3_sql(pTestStmt),0)==0 ){
 149       /* crash-89bd6a6f8c6166e9a4c5f47b3e70b225f69b76c6
 150       ** Original statement is:
 151       **
 152       **    SELECT a,b,c* FROM t1 LIMIT 1%5<4
 153       **
 154       ** When running:
 155       **
 156       **    SELECT * FROM (...) ORDER BY 1
 157       **
 158       ** A different subset of the rows come out
 159       */
 160       goto not_a_fault;
 161     }
 162
 163     /* The original sameValue() comparison assumed a collating sequence
 164     ** of "binary".  It can sometimes get an incorrect result for different
 165     ** collating sequences.  So rerun the test with no assumptions about
 166     ** collations.
 167     */
 168     rc = sqlite3_prepare_v2(db,
 169        "SELECT ?1=?2 OR ?1=?2 COLLATE nocase OR ?1=?2 COLLATE rtrim",
 170        -1, &pCk, 0);
 171     if( rc==SQLITE_OK ){
 172       sqlite3_reset(pTestStmt);
 173       while( (rc = sqlite3_step(pTestStmt))==SQLITE_ROW ){
 174         for(i=0; i<nCol; i++){
 175           if( !sameValue(pStmt, i, pTestStmt, i, pCk) ) break;
 176         }
 177         if( i>=nCol ){
 178           sqlite3_finalize(pCk);
 179           goto not_a_fault;
 180         }
 181       }
 182     }
 183     sqlite3_finalize(pCk);
 184
 185     /* Invariants do not necessarily work if there are virtual tables
 186     ** involved in the query */
 187     rc = sqlite3_prepare_v2(db,
 188             "SELECT 1 FROM bytecode(?1) WHERE opcode='VOpen'", -1, &pCk, 0);
 189     if( rc==SQLITE_OK ){
 190       sqlite3_bind_pointer(pCk, 1, pStmt, "stmt-pointer", 0);
 191       rc = sqlite3_step(pCk);
 192     }
 193     sqlite3_finalize(pCk);
 194     if( rc==SQLITE_DONE ){
 195       reportInvariantFailed(pStmt, pTestStmt, iRow);
 196       return SQLITE_INTERNAL;
 197     }else if( eVerbosity>0 ){
 198       printf("invariant-error ignored due to the use of virtual tables\n");
 199     }
 200   }
 201 not_a_fault:
 202   sqlite3_finalize(pTestStmt);
 203   return SQLITE_OK;
 204 }
 205
 206
 207 /*
 208 ** Generate SQL used to test a statement invariant.
 209 **
 210 ** Return 0 if the iCnt is out of range.
 211 **
 212 ** iCnt meanings:
 213 **
 214 **   0     SELECT * FROM (<query>)
 215 **   1     SELECT DISTINCT * FROM (<query>)
 216 **   2     SELECT * FROM (<query>) WHERE ORDER BY 1
 217 **   3     SELECT DISTINCT * FROM (<query>) ORDER BY 1
 218 **   4     SELECT * FROM (<query>) WHERE <all-columns>=<all-values>
 219 **   5     SELECT DISTINCT * FROM (<query>) WHERE <all-columns=<all-values
 220 **   6     SELECT * FROM (<query>) WHERE <all-column>=<all-value> ORDER BY 1
 221 **   7     SELECT DISTINCT * FROM (<query>) WHERE <all-column>=<all-value>
 222 **                           ORDER BY 1
 223 **   N+0   SELECT * FROM (<query>) WHERE <nth-column>=<value>
 224 **   N+1   SELECT DISTINCT * FROM (<query>) WHERE <Nth-column>=<value>
 225 **   N+2   SELECT * FROM (<query>) WHERE <Nth-column>=<value> ORDER BY 1
 226 **   N+3   SELECT DISTINCT * FROM (<query>) WHERE <Nth-column>=<value>
 227 **                           ORDER BY N
 228 **
 229 */
 230 static char *fuzz_invariant_sql(sqlite3_stmt *pStmt, int iCnt){
 231   const char *zIn;
 232   size_t nIn;
 233   const char *zAnd = "WHERE";
 234   int i, j;
 235   sqlite3_str *pTest;
 236   sqlite3_stmt *pBase = 0;
 237   sqlite3 *db = sqlite3_db_handle(pStmt);
 238   int rc;
 239   int nCol = sqlite3_column_count(pStmt);
 240   int mxCnt;
 241   int bDistinct = 0;
 242   int bOrderBy = 0;
 243   int nParam = sqlite3_bind_parameter_count(pStmt);
 244
 245   switch( iCnt % 4 ){
 246     case 1:  bDistinct = 1;              break;
 247     case 2:  bOrderBy = 1;               break;
 248     case 3:  bDistinct = bOrderBy = 1;   break;
 249   }
 250   iCnt /= 4;
 251   mxCnt = nCol;
 252   if( iCnt<0 || iCnt>mxCnt ) return 0;
 253   zIn = sqlite3_sql(pStmt);
 254   if( zIn==0 ) return 0;
 255   nIn = strlen(zIn);
 256   while( nIn>0 && (isspace(zIn[nIn-1]) || zIn[nIn-1]==';') ) nIn--;
 257   if( strchr(zIn, '?') ) return 0;
 258   pTest = sqlite3_str_new(0);
 259   sqlite3_str_appendf(pTest, "SELECT %s* FROM (",
 260                       bDistinct ? "DISTINCT " : "");
 261   sqlite3_str_append(pTest, zIn, (int)nIn);
 262   sqlite3_str_append(pTest, ")", 1);
 263   rc = sqlite3_prepare_v2(db, sqlite3_str_value(pTest), -1, &pBase, 0);
 264   if( rc ){
 265     sqlite3_finalize(pBase);
 266     pBase = pStmt;
 267   }
 268   for(i=0; i<sqlite3_column_count(pStmt); i++){
 269     const char *zColName = sqlite3_column_name(pBase,i);
 270     const char *zSuffix = zColName ? strrchr(zColName, ':') : 0;
 271     if( zSuffix
 272      && isdigit(zSuffix[1])
 273      && (zSuffix[1]>'3' || isdigit(zSuffix[2]))
 274     ){
 275       /* This is a randomized column name and so cannot be used in the
 276       ** WHERE clause. */
 277       continue;
 278     }
 279     for(j=0; j<i; j++){
 280       const char *zPrior = sqlite3_column_name(pBase, j);
 281       if( sqlite3_stricmp(zPrior, zColName)==0 ) break;
 282     }
 283     if( j<i ){
 284       /* Duplicate column name */
 285       continue;
 286     }
 287     if( iCnt==0 ) continue;
 288     if( iCnt>1 && i+2!=iCnt ) continue;
 289     if( zColName==0 ) continue;
 290     if( sqlite3_column_type(pStmt, i)==SQLITE_NULL ){
 291       sqlite3_str_appendf(pTest, " %s \"%w\" ISNULL", zAnd, zColName);
 292     }else{
 293       sqlite3_str_appendf(pTest, " %s \"%w\"=?%d", zAnd, zColName,
 294                           i+1+nParam);
 295     }
 296     zAnd = "AND";
 297   }
 298   if( pBase!=pStmt ) sqlite3_finalize(pBase);
 299   if( bOrderBy ){
 300     sqlite3_str_appendf(pTest, " ORDER BY %d", iCnt>2 ? iCnt-1 : 1);
 301   }
 302   return sqlite3_str_finish(pTest);
 303 }
 304
 305 /*
 306 ** Return true if and only if v1 and is the same as v2.
 307 */
 308 static int sameValue(
 309   sqlite3_stmt *pS1, int i1,       /* Value to text on the left */
 310   sqlite3_stmt *pS2, int i2,       /* Value to test on the right */
 311   sqlite3_stmt *pTestCompare       /* COLLATE comparison statement or NULL */
 312 ){
 313   int x = 1;
 314   int t1 = sqlite3_column_type(pS1,i1);
 315   int t2 = sqlite3_column_type(pS2,i2);
 316   if( t1!=t2 ){
 317     if( (t1==SQLITE_INTEGER && t2==SQLITE_FLOAT)
 318      || (t1==SQLITE_FLOAT && t2==SQLITE_INTEGER)
 319     ){
 320       /* Comparison of numerics is ok */
 321     }else{
 322       return 0;
 323     }
 324   }
 325   switch( sqlite3_column_type(pS1,i1) ){
 326     case SQLITE_INTEGER: {
 327       x =  sqlite3_column_int64(pS1,i1)==sqlite3_column_int64(pS2,i2);
 328       break;
 329     }
 330     case SQLITE_FLOAT: {
 331       x = sqlite3_column_double(pS1,i1)==sqlite3_column_double(pS2,i2);
 332       break;
 333     }
 334     case SQLITE_TEXT: {
 335       int e1 = sqlite3_value_encoding(sqlite3_column_value(pS1,i1));
 336       int e2 = sqlite3_value_encoding(sqlite3_column_value(pS2,i2));
 337       if( e1!=e2 ){
 338         const char *z1 = (const char*)sqlite3_column_text(pS1,i1);
 339         const char *z2 = (const char*)sqlite3_column_text(pS2,i2);
 340         x = ((z1==0 && z2==0) || (z1!=0 && z2!=0 && strcmp(z1,z1)==0));
 341         printf("Encodings differ.  %d on left and %d on right\n", e1, e2);
 342         abort();
 343       }
 344       if( pTestCompare ){
 345         sqlite3_bind_value(pTestCompare, 1, sqlite3_column_value(pS1,i1));
 346         sqlite3_bind_value(pTestCompare, 2, sqlite3_column_value(pS2,i2));
 347         x = sqlite3_step(pTestCompare)==SQLITE_ROW
 348                       && sqlite3_column_int(pTestCompare,0)!=0;
 349         sqlite3_reset(pTestCompare);
 350         break;
 351       }
 352       if( e1!=SQLITE_UTF8 ){
 353         int len1 = sqlite3_column_bytes16(pS1,i1);
 354         const unsigned char *b1 = sqlite3_column_blob(pS1,i1);
 355         int len2 = sqlite3_column_bytes16(pS2,i2);
 356         const unsigned char *b2 = sqlite3_column_blob(pS2,i2);
 357         if( len1!=len2 ){
 358           x = 0;
 359         }else if( len1==0 ){
 360           x = 1;
 361         }else{
 362           x = (b1!=0 && b2!=0 && memcmp(b1,b2,len1)==0);
 363         }
 364         break;
 365       }
 366       /* Fall through into the SQLITE_BLOB case */
 367     }
 368     case SQLITE_BLOB: {
 369       int len1 = sqlite3_column_bytes(pS1,i1);
 370       const unsigned char *b1 = sqlite3_column_blob(pS1,i1);
 371       int len2 = sqlite3_column_bytes(pS2,i2);
 372       const unsigned char *b2 = sqlite3_column_blob(pS2,i2);
 373       if( len1!=len2 ){
 374         x = 0;
 375       }else if( len1==0 ){
 376         x = 1;
 377       }else{
 378         x = (b1!=0 && b2!=0 && memcmp(b1,b2,len1)==0);
 379       }
 380       break;
 381     }
 382   }
 383   return x;
 384 }
 385
 386 /*
 387 ** Print binary data as hex
 388 */
 389 static void printHex(const unsigned char *a, int n, int mx){
 390   int j;
 391   for(j=0; j<mx && j<n; j++){
 392     printf("%02x", a[j]);
 393   }
 394   if( j<n ) printf("...");
 395 }
 396
 397 /*
 398 ** Print a single row from the prepared statement
 399 */
 400 static void printRow(sqlite3_stmt *pStmt, int iRow){
 401   int i, n, nCol;
 402   unsigned const char *data;
 403   nCol = sqlite3_column_count(pStmt);
 404   for(i=0; i<nCol; i++){
 405     printf("row%d.col%d = ", iRow, i);
 406     switch( sqlite3_column_type(pStmt, i) ){
 407       case SQLITE_NULL: {
 408         printf("NULL\n");
 409         break;
 410       }
 411       case SQLITE_INTEGER: {
 412         printf("(integer) %lld\n", sqlite3_column_int64(pStmt, i));
 413         break;
 414       }
 415       case SQLITE_FLOAT: {
 416         printf("(float) %f\n", sqlite3_column_double(pStmt, i));
 417         break;
 418       }
 419       case SQLITE_TEXT: {
 420         switch( sqlite3_value_encoding(sqlite3_column_value(pStmt,i)) ){
 421           case SQLITE_UTF8: {
 422             printf("(utf8) x'");
 423             n = sqlite3_column_bytes(pStmt, i);
 424             data = sqlite3_column_blob(pStmt, i);
 425             printHex(data, n, 35);
 426             printf("'\n");
 427             break;
 428           }
 429           case SQLITE_UTF16BE: {
 430             printf("(utf16be) x'");
 431             n = sqlite3_column_bytes16(pStmt, i);
 432             data = sqlite3_column_blob(pStmt, i);
 433             printHex(data, n, 35);
 434             printf("'\n");
 435             break;
 436           }
 437           case SQLITE_UTF16LE: {
 438             printf("(utf16le) x'");
 439             n = sqlite3_column_bytes16(pStmt, i);
 440             data = sqlite3_column_blob(pStmt, i);
 441             printHex(data, n, 35);
 442             printf("'\n");
 443             break;
 444           }
 445           default: {
 446             printf("Illegal return from sqlite3_value_encoding(): %d\n",
 447                 sqlite3_value_encoding(sqlite3_column_value(pStmt,i)));
 448             abort();
 449           }
 450         }
 451         break;
 452       }
 453       case SQLITE_BLOB: {
 454         n = sqlite3_column_bytes(pStmt, i);
 455         data = sqlite3_column_blob(pStmt, i);
 456         printf("(blob %d bytes) x'", n);
 457         printHex(data, n, 35);
 458         printf("'\n");
 459         break;
 460       }
 461     }
 462   }
 463 }
 464
 465 /*
 466 ** Report a failure of the invariant:  The current output row of pOrig
 467 ** does not appear in any row of the output from pTest.
 468 */
 469 static void reportInvariantFailed(
 470   sqlite3_stmt *pOrig,   /* The original query */
 471   sqlite3_stmt *pTest,   /* The alternative test query with a missing row */
 472   int iRow               /* Row number in pOrig */
 473 ){
 474   int iTestRow = 0;
 475   printf("Invariant check failed on row %d.\n", iRow);
 476   printf("Original query --------------------------------------------------\n");
 477   printf("%s\n", sqlite3_expanded_sql(pOrig));
 478   printf("Alternative query -----------------------------------------------\n");
 479   printf("%s\n", sqlite3_expanded_sql(pTest));
 480   printf("Result row that is missing from the alternative -----------------\n");
 481   printRow(pOrig, iRow);
 482   printf("Complete results from the alternative query ---------------------\n");
 483   sqlite3_reset(pTest);
 484   while( sqlite3_step(pTest)==SQLITE_ROW ){
 485     iTestRow++;
 486     printRow(pTest, iTestRow);
 487   }
 488   sqlite3_finalize(pTest);
 489   abort();
 490 }