granicus.if.org Git - postgresql/blob - src/backend/optimizer/path/allpaths.c

   1 /*-------------------------------------------------------------------------
   2  *
   3  * allpaths.c
   4  *        Routines to find possible search paths for processing a query
   5  *
   6  * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
   7  * Portions Copyright (c) 1994, Regents of the University of California
   8  *
   9  *
  10  * IDENTIFICATION
  11  *        $PostgreSQL: pgsql/src/backend/optimizer/path/allpaths.c,v 1.117 2004/06/01 03:02:51 tgl Exp $
  12  *
  13  *-------------------------------------------------------------------------
  14  */
  15
  16 #include "postgres.h"
  17
  18 #ifdef OPTIMIZER_DEBUG
  19 #include "nodes/print.h"
  20 #endif
  21 #include "optimizer/clauses.h"
  22 #include "optimizer/cost.h"
  23 #include "optimizer/geqo.h"
  24 #include "optimizer/pathnode.h"
  25 #include "optimizer/paths.h"
  26 #include "optimizer/plancat.h"
  27 #include "optimizer/planner.h"
  28 #include "optimizer/prep.h"
  29 #include "optimizer/var.h"
  30 #include "parser/parsetree.h"
  31 #include "parser/parse_clause.h"
  32 #include "rewrite/rewriteManip.h"
  33
  34
  35 /* These parameters are set by GUC */
  36 bool            enable_geqo = false;    /* just in case GUC doesn't set it */
  37 int                     geqo_threshold;
  38
  39
  40 static void set_base_rel_pathlists(Query *root);
  41 static void set_plain_rel_pathlist(Query *root, RelOptInfo *rel,
  42                                            RangeTblEntry *rte);
  43 static void set_inherited_rel_pathlist(Query *root, RelOptInfo *rel,
  44                                                    Index rti, RangeTblEntry *rte,
  45                                                    List *inheritlist);
  46 static void set_subquery_pathlist(Query *root, RelOptInfo *rel,
  47                                           Index rti, RangeTblEntry *rte);
  48 static void set_function_pathlist(Query *root, RelOptInfo *rel,
  49                                           RangeTblEntry *rte);
  50 static RelOptInfo *make_one_rel_by_joins(Query *root, int levels_needed,
  51                                           List *initial_rels);
  52 static bool subquery_is_pushdown_safe(Query *subquery, Query *topquery,
  53                                                   bool *differentTypes);
  54 static bool recurse_pushdown_safe(Node *setOp, Query *topquery,
  55                                           bool *differentTypes);
  56 static void compare_tlist_datatypes(List *tlist, List *colTypes,
  57                                                 bool *differentTypes);
  58 static bool qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
  59                                           bool *differentTypes);
  60 static void subquery_push_qual(Query *subquery,
  61                                                            RangeTblEntry *rte, Index rti, Node *qual);
  62 static void recurse_push_qual(Node *setOp, Query *topquery,
  63                                                           RangeTblEntry *rte, Index rti, Node *qual);
  64
  65
  66 /*
  67  * make_one_rel
  68  *        Finds all possible access paths for executing a query, returning a
  69  *        single rel that represents the join of all base rels in the query.
  70  */
  71 RelOptInfo *
  72 make_one_rel(Query *root)
  73 {
  74         RelOptInfo *rel;
  75
  76         /*
  77          * Generate access paths for the base rels.
  78          */
  79         set_base_rel_pathlists(root);
  80
  81         /*
  82          * Generate access paths for the entire join tree.
  83          */
  84         Assert(root->jointree != NULL && IsA(root->jointree, FromExpr));
  85
  86         rel = make_fromexpr_rel(root, root->jointree);
  87
  88         /*
  89          * The result should join all the query's base rels.
  90          */
  91         Assert(bms_num_members(rel->relids) == list_length(root->base_rel_list));
  92
  93         return rel;
  94 }
  95
  96 /*
  97  * set_base_rel_pathlists
  98  *        Finds all paths available for scanning each base-relation entry.
  99  *        Sequential scan and any available indices are considered.
 100  *        Each useful path is attached to its relation's 'pathlist' field.
 101  */
 102 static void
 103 set_base_rel_pathlists(Query *root)
 104 {
 105         ListCell           *l;
 106
 107         foreach(l, root->base_rel_list)
 108         {
 109                 RelOptInfo *rel = (RelOptInfo *) lfirst(l);
 110                 Index           rti = rel->relid;
 111                 RangeTblEntry *rte;
 112                 List       *inheritlist;
 113
 114                 Assert(rti > 0);                /* better be base rel */
 115                 rte = rt_fetch(rti, root->rtable);
 116
 117                 if (rel->rtekind == RTE_SUBQUERY)
 118                 {
 119                         /* Subquery --- generate a separate plan for it */
 120                         set_subquery_pathlist(root, rel, rti, rte);
 121                 }
 122                 else if (rel->rtekind == RTE_FUNCTION)
 123                 {
 124                         /* RangeFunction --- generate a separate plan for it */
 125                         set_function_pathlist(root, rel, rte);
 126                 }
 127                 else if ((inheritlist = expand_inherited_rtentry(root, rti, true))
 128                                  != NIL)
 129                 {
 130                         /* Relation is root of an inheritance tree, process specially */
 131                         set_inherited_rel_pathlist(root, rel, rti, rte, inheritlist);
 132                 }
 133                 else
 134                 {
 135                         /* Plain relation */
 136                         set_plain_rel_pathlist(root, rel, rte);
 137                 }
 138
 139 #ifdef OPTIMIZER_DEBUG
 140                 debug_print_rel(root, rel);
 141 #endif
 142         }
 143 }
 144
 145 /*
 146  * set_plain_rel_pathlist
 147  *        Build access paths for a plain relation (no subquery, no inheritance)
 148  */
 149 static void
 150 set_plain_rel_pathlist(Query *root, RelOptInfo *rel, RangeTblEntry *rte)
 151 {
 152         /* Mark rel with estimated output rows, width, etc */
 153         set_baserel_size_estimates(root, rel);
 154
 155         /* Test any partial indexes of rel for applicability */
 156         check_partial_indexes(root, rel);
 157
 158         /*
 159          * Check to see if we can extract any restriction conditions from
 160          * join quals that are OR-of-AND structures.  If so, add them to the
 161          * rel's restriction list, and recompute the size estimates.
 162          */
 163         if (create_or_index_quals(root, rel))
 164                 set_baserel_size_estimates(root, rel);
 165
 166         /*
 167          * Generate paths and add them to the rel's pathlist.
 168          *
 169          * Note: add_path() will discard any paths that are dominated by another
 170          * available path, keeping only those paths that are superior along at
 171          * least one dimension of cost or sortedness.
 172          */
 173
 174         /* Consider sequential scan */
 175         add_path(rel, create_seqscan_path(root, rel));
 176
 177         /* Consider TID scans */
 178         create_tidscan_paths(root, rel);
 179
 180         /* Consider index paths for both simple and OR index clauses */
 181         create_index_paths(root, rel);
 182         create_or_index_paths(root, rel);
 183
 184         /* Now find the cheapest of the paths for this rel */
 185         set_cheapest(rel);
 186 }
 187
 188 /*
 189  * set_inherited_rel_pathlist
 190  *        Build access paths for a inheritance tree rooted at rel
 191  *
 192  * inheritlist is a list of RT indexes of all tables in the inheritance tree,
 193  * including a duplicate of the parent itself.  Note we will not come here
 194  * unless there's at least one child in addition to the parent.
 195  *
 196  * NOTE: the passed-in rel and RTE will henceforth represent the appended
 197  * result of the whole inheritance tree.  The members of inheritlist represent
 198  * the individual tables --- in particular, the inheritlist member that is a
 199  * duplicate of the parent RTE represents the parent table alone.
 200  * We will generate plans to scan the individual tables that refer to
 201  * the inheritlist RTEs, whereas Vars elsewhere in the plan tree that
 202  * refer to the original RTE are taken to refer to the append output.
 203  * In particular, this means we have separate RelOptInfos for the parent
 204  * table and for the append output, which is a good thing because they're
 205  * not the same size.
 206  */
 207 static void
 208 set_inherited_rel_pathlist(Query *root, RelOptInfo *rel,
 209                                                    Index rti, RangeTblEntry *rte,
 210                                                    List *inheritlist)
 211 {
 212         int                     parentRTindex = rti;
 213         Oid                     parentOID = rte->relid;
 214         List       *subpaths = NIL;
 215         ListCell   *il;
 216
 217         /*
 218          * XXX for now, can't handle inherited expansion of FOR UPDATE; can we
 219          * do better?
 220          */
 221         if (list_member_int(root->rowMarks, parentRTindex))
 222                 ereport(ERROR,
 223                                 (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
 224                                  errmsg("SELECT FOR UPDATE is not supported for inheritance queries")));
 225
 226         /*
 227          * The executor will check the parent table's access permissions when
 228          * it examines the parent's inheritlist entry.  There's no need to
 229          * check twice, so turn off access check bits in the original RTE.
 230          */
 231         rte->requiredPerms = 0;
 232
 233         /*
 234          * Initialize to compute size estimates for whole inheritance tree
 235          */
 236         rel->rows = 0;
 237         rel->width = 0;
 238
 239         /*
 240          * Generate access paths for each table in the tree (parent AND
 241          * children), and pick the cheapest path for each table.
 242          */
 243         foreach(il, inheritlist)
 244         {
 245                 int                     childRTindex = lfirst_int(il);
 246                 RangeTblEntry *childrte;
 247                 Oid                     childOID;
 248                 RelOptInfo *childrel;
 249                 ListCell   *parentvars;
 250                 ListCell   *childvars;
 251
 252                 childrte = rt_fetch(childRTindex, root->rtable);
 253                 childOID = childrte->relid;
 254
 255                 /*
 256                  * Make a RelOptInfo for the child so we can do planning.  Do NOT
 257                  * attach the RelOptInfo to the query's base_rel_list, however,
 258                  * since the child is not part of the main join tree.  Instead,
 259                  * the child RelOptInfo is added to other_rel_list.
 260                  */
 261                 childrel = build_other_rel(root, childRTindex);
 262
 263                 /*
 264                  * Copy the parent's targetlist and restriction quals to the
 265                  * child, with attribute-number adjustment as needed.  We don't
 266                  * bother to copy the join quals, since we can't do any joining of
 267                  * the individual tables.  Also, we just zap attr_needed rather
 268                  * than trying to adjust it; it won't be looked at in the child.
 269                  */
 270                 childrel->reltargetlist = (List *)
 271                         adjust_inherited_attrs((Node *) rel->reltargetlist,
 272                                                                    parentRTindex,
 273                                                                    parentOID,
 274                                                                    childRTindex,
 275                                                                    childOID);
 276                 childrel->attr_needed = NULL;
 277                 childrel->baserestrictinfo = (List *)
 278                         adjust_inherited_attrs((Node *) rel->baserestrictinfo,
 279                                                                    parentRTindex,
 280                                                                    parentOID,
 281                                                                    childRTindex,
 282                                                                    childOID);
 283
 284                 /*
 285                  * Now compute child access paths, and save the cheapest.
 286                  */
 287                 set_plain_rel_pathlist(root, childrel, childrte);
 288
 289                 subpaths = lappend(subpaths, childrel->cheapest_total_path);
 290
 291                 /*
 292                  * Propagate size information from the child back to the parent.
 293                  * For simplicity, we use the largest widths from any child as the
 294                  * parent estimates.
 295                  */
 296                 rel->rows += childrel->rows;
 297                 if (childrel->width > rel->width)
 298                         rel->width = childrel->width;
 299
 300                 forboth(parentvars, rel->reltargetlist,
 301                                 childvars, childrel->reltargetlist)
 302                 {
 303                         Var                *parentvar = (Var *) lfirst(parentvars);
 304                         Var                *childvar = (Var *) lfirst(childvars);
 305                         int                     parentndx = parentvar->varattno - rel->min_attr;
 306                         int                     childndx = childvar->varattno - childrel->min_attr;
 307
 308                         if (childrel->attr_widths[childndx] > rel->attr_widths[parentndx])
 309                                 rel->attr_widths[parentndx] = childrel->attr_widths[childndx];
 310                 }
 311         }
 312
 313         /*
 314          * Finally, build Append path and install it as the only access path
 315          * for the parent rel.
 316          */
 317         add_path(rel, (Path *) create_append_path(rel, subpaths));
 318
 319         /* Select cheapest path (pretty easy in this case...) */
 320         set_cheapest(rel);
 321 }
 322
 323 /*
 324  * set_subquery_pathlist
 325  *              Build the (single) access path for a subquery RTE
 326  */
 327 static void
 328 set_subquery_pathlist(Query *root, RelOptInfo *rel,
 329                                           Index rti, RangeTblEntry *rte)
 330 {
 331         Query      *subquery = rte->subquery;
 332         bool       *differentTypes;
 333         List       *pathkeys;
 334
 335         /* We need a workspace for keeping track of set-op type coercions */
 336         differentTypes = (bool *)
 337                 palloc0((list_length(subquery->targetList) + 1) * sizeof(bool));
 338
 339         /*
 340          * If there are any restriction clauses that have been attached to the
 341          * subquery relation, consider pushing them down to become HAVING
 342          * quals of the subquery itself.  (Not WHERE clauses, since they may
 343          * refer to subquery outputs that are aggregate results.  But
 344          * planner.c will transfer them into the subquery's WHERE if they do
 345          * not.)  This transformation is useful because it may allow us to
 346          * generate a better plan for the subquery than evaluating all the
 347          * subquery output rows and then filtering them.
 348          *
 349          * There are several cases where we cannot push down clauses.
 350          * Restrictions involving the subquery are checked by
 351          * subquery_is_pushdown_safe().  Restrictions on individual clauses
 352          * are checked by qual_is_pushdown_safe().
 353          *
 354          * Non-pushed-down clauses will get evaluated as qpquals of the
 355          * SubqueryScan node.
 356          *
 357          * XXX Are there any cases where we want to make a policy decision not to
 358          * push down a pushable qual, because it'd result in a worse plan?
 359          */
 360         if (rel->baserestrictinfo != NIL &&
 361                 subquery_is_pushdown_safe(subquery, subquery, differentTypes))
 362         {
 363                 /* OK to consider pushing down individual quals */
 364                 List       *upperrestrictlist = NIL;
 365                 ListCell   *l;
 366
 367                 foreach(l, rel->baserestrictinfo)
 368                 {
 369                         RestrictInfo *rinfo = (RestrictInfo *) lfirst(l);
 370                         Node       *clause = (Node *) rinfo->clause;
 371
 372                         if (qual_is_pushdown_safe(subquery, rti, clause, differentTypes))
 373                         {
 374                                 /* Push it down */
 375                                 subquery_push_qual(subquery, rte, rti, clause);
 376                         }
 377                         else
 378                         {
 379                                 /* Keep it in the upper query */
 380                                 upperrestrictlist = lappend(upperrestrictlist, rinfo);
 381                         }
 382                 }
 383                 rel->baserestrictinfo = upperrestrictlist;
 384         }
 385
 386         pfree(differentTypes);
 387
 388         /* Generate the plan for the subquery */
 389         rel->subplan = subquery_planner(subquery, 0.0 /* default case */ );
 390
 391         /* Copy number of output rows from subplan */
 392         rel->tuples = rel->subplan->plan_rows;
 393
 394         /* Mark rel with estimated output rows, width, etc */
 395         set_baserel_size_estimates(root, rel);
 396
 397         /* Convert subquery pathkeys to outer representation */
 398         pathkeys = build_subquery_pathkeys(root, rel, subquery);
 399
 400         /* Generate appropriate path */
 401         add_path(rel, create_subqueryscan_path(rel, pathkeys));
 402
 403         /* Select cheapest path (pretty easy in this case...) */
 404         set_cheapest(rel);
 405 }
 406
 407 /*
 408  * set_function_pathlist
 409  *              Build the (single) access path for a function RTE
 410  */
 411 static void
 412 set_function_pathlist(Query *root, RelOptInfo *rel, RangeTblEntry *rte)
 413 {
 414         /* Mark rel with estimated output rows, width, etc */
 415         set_function_size_estimates(root, rel);
 416
 417         /* Generate appropriate path */
 418         add_path(rel, create_functionscan_path(root, rel));
 419
 420         /* Select cheapest path (pretty easy in this case...) */
 421         set_cheapest(rel);
 422 }
 423
 424 /*
 425  * make_fromexpr_rel
 426  *        Build access paths for a FromExpr jointree node.
 427  */
 428 RelOptInfo *
 429 make_fromexpr_rel(Query *root, FromExpr *from)
 430 {
 431         int                     levels_needed;
 432         List       *initial_rels = NIL;
 433         ListCell   *jt;
 434
 435         /*
 436          * Count the number of child jointree nodes.  This is the depth of the
 437          * dynamic-programming algorithm we must employ to consider all ways
 438          * of joining the child nodes.
 439          */
 440         levels_needed = list_length(from->fromlist);
 441
 442         if (levels_needed <= 0)
 443                 return NULL;                    /* nothing to do? */
 444
 445         /*
 446          * Construct a list of rels corresponding to the child jointree nodes.
 447          * This may contain both base rels and rels constructed according to
 448          * explicit JOIN directives.
 449          */
 450         foreach(jt, from->fromlist)
 451         {
 452                 Node       *jtnode = (Node *) lfirst(jt);
 453
 454                 initial_rels = lappend(initial_rels,
 455                                                            make_jointree_rel(root, jtnode));
 456         }
 457
 458         if (levels_needed == 1)
 459         {
 460                 /*
 461                  * Single jointree node, so we're done.
 462                  */
 463                 return (RelOptInfo *) linitial(initial_rels);
 464         }
 465         else
 466         {
 467                 /*
 468                  * Consider the different orders in which we could join the rels,
 469                  * using either GEQO or regular optimizer.
 470                  */
 471                 if (enable_geqo && levels_needed >= geqo_threshold)
 472                         return geqo(root, levels_needed, initial_rels);
 473                 else
 474                         return make_one_rel_by_joins(root, levels_needed, initial_rels);
 475         }
 476 }
 477
 478 /*
 479  * make_one_rel_by_joins
 480  *        Find all possible joinpaths for a query by successively finding ways
 481  *        to join component relations into join relations.
 482  *
 483  * 'levels_needed' is the number of iterations needed, ie, the number of
 484  *              independent jointree items in the query.  This is > 1.
 485  *
 486  * 'initial_rels' is a list of RelOptInfo nodes for each independent
 487  *              jointree item.  These are the components to be joined together.
 488  *
 489  * Returns the final level of join relations, i.e., the relation that is
 490  * the result of joining all the original relations together.
 491  */
 492 static RelOptInfo *
 493 make_one_rel_by_joins(Query *root, int levels_needed, List *initial_rels)
 494 {
 495         List      **joinitems;
 496         int                     lev;
 497         RelOptInfo *rel;
 498
 499         /*
 500          * We employ a simple "dynamic programming" algorithm: we first find
 501          * all ways to build joins of two jointree items, then all ways to
 502          * build joins of three items (from two-item joins and single items),
 503          * then four-item joins, and so on until we have considered all ways
 504          * to join all the items into one rel.
 505          *
 506          * joinitems[j] is a list of all the j-item rels.  Initially we set
 507          * joinitems[1] to represent all the single-jointree-item relations.
 508          */
 509         joinitems = (List **) palloc0((levels_needed + 1) * sizeof(List *));
 510
 511         joinitems[1] = initial_rels;
 512
 513         for (lev = 2; lev <= levels_needed; lev++)
 514         {
 515                 ListCell   *x;
 516
 517                 /*
 518                  * Determine all possible pairs of relations to be joined at this
 519                  * level, and build paths for making each one from every available
 520                  * pair of lower-level relations.
 521                  */
 522                 joinitems[lev] = make_rels_by_joins(root, lev, joinitems);
 523
 524                 /*
 525                  * Do cleanup work on each just-processed rel.
 526                  */
 527                 foreach(x, joinitems[lev])
 528                 {
 529                         rel = (RelOptInfo *) lfirst(x);
 530
 531                         /* Find and save the cheapest paths for this rel */
 532                         set_cheapest(rel);
 533
 534 #ifdef OPTIMIZER_DEBUG
 535                         debug_print_rel(root, rel);
 536 #endif
 537                 }
 538         }
 539
 540         /*
 541          * We should have a single rel at the final level.
 542          */
 543         if (joinitems[levels_needed] == NIL)
 544                 elog(ERROR, "failed to build any %d-way joins", levels_needed);
 545         Assert(list_length(joinitems[levels_needed]) == 1);
 546
 547         rel = (RelOptInfo *) linitial(joinitems[levels_needed]);
 548
 549         return rel;
 550 }
 551
 552 /*****************************************************************************
 553  *                      PUSHING QUALS DOWN INTO SUBQUERIES
 554  *****************************************************************************/
 555
 556 /*
 557  * subquery_is_pushdown_safe - is a subquery safe for pushing down quals?
 558  *
 559  * subquery is the particular component query being checked.  topquery
 560  * is the top component of a set-operations tree (the same Query if no
 561  * set-op is involved).
 562  *
 563  * Conditions checked here:
 564  *
 565  * 1. If the subquery has a LIMIT clause, we must not push down any quals,
 566  * since that could change the set of rows returned.
 567  *
 568  * 2. If the subquery contains EXCEPT or EXCEPT ALL set ops we cannot push
 569  * quals into it, because that would change the results.
 570  *
 571  * 3. For subqueries using UNION/UNION ALL/INTERSECT/INTERSECT ALL, we can
 572  * push quals into each component query, but the quals can only reference
 573  * subquery columns that suffer no type coercions in the set operation.
 574  * Otherwise there are possible semantic gotchas.  So, we check the
 575  * component queries to see if any of them have different output types;
 576  * differentTypes[k] is set true if column k has different type in any
 577  * component.
 578  */
 579 static bool
 580 subquery_is_pushdown_safe(Query *subquery, Query *topquery,
 581                                                   bool *differentTypes)
 582 {
 583         SetOperationStmt *topop;
 584
 585         /* Check point 1 */
 586         if (subquery->limitOffset != NULL || subquery->limitCount != NULL)
 587                 return false;
 588
 589         /* Are we at top level, or looking at a setop component? */
 590         if (subquery == topquery)
 591         {
 592                 /* Top level, so check any component queries */
 593                 if (subquery->setOperations != NULL)
 594                         if (!recurse_pushdown_safe(subquery->setOperations, topquery,
 595                                                                            differentTypes))
 596                                 return false;
 597         }
 598         else
 599         {
 600                 /* Setop component must not have more components (too weird) */
 601                 if (subquery->setOperations != NULL)
 602                         return false;
 603                 /* Check whether setop component output types match top level */
 604                 topop = (SetOperationStmt *) topquery->setOperations;
 605                 Assert(topop && IsA(topop, SetOperationStmt));
 606                 compare_tlist_datatypes(subquery->targetList,
 607                                                                 topop->colTypes,
 608                                                                 differentTypes);
 609         }
 610         return true;
 611 }
 612
 613 /*
 614  * Helper routine to recurse through setOperations tree
 615  */
 616 static bool
 617 recurse_pushdown_safe(Node *setOp, Query *topquery,
 618                                           bool *differentTypes)
 619 {
 620         if (IsA(setOp, RangeTblRef))
 621         {
 622                 RangeTblRef *rtr = (RangeTblRef *) setOp;
 623                 RangeTblEntry *rte = rt_fetch(rtr->rtindex, topquery->rtable);
 624                 Query      *subquery = rte->subquery;
 625
 626                 Assert(subquery != NULL);
 627                 return subquery_is_pushdown_safe(subquery, topquery, differentTypes);
 628         }
 629         else if (IsA(setOp, SetOperationStmt))
 630         {
 631                 SetOperationStmt *op = (SetOperationStmt *) setOp;
 632
 633                 /* EXCEPT is no good */
 634                 if (op->op == SETOP_EXCEPT)
 635                         return false;
 636                 /* Else recurse */
 637                 if (!recurse_pushdown_safe(op->larg, topquery, differentTypes))
 638                         return false;
 639                 if (!recurse_pushdown_safe(op->rarg, topquery, differentTypes))
 640                         return false;
 641         }
 642         else
 643         {
 644                 elog(ERROR, "unrecognized node type: %d",
 645                          (int) nodeTag(setOp));
 646         }
 647         return true;
 648 }
 649
 650 /*
 651  * Compare tlist's datatypes against the list of set-operation result types.
 652  * For any items that are different, mark the appropriate element of
 653  * differentTypes[] to show that this column will have type conversions.
 654  */
 655 static void
 656 compare_tlist_datatypes(List *tlist, List *colTypes,
 657                                                 bool *differentTypes)
 658 {
 659         ListCell   *l;
 660         ListCell   *colType = list_head(colTypes);
 661
 662         foreach(l, tlist)
 663         {
 664                 TargetEntry *tle = (TargetEntry *) lfirst(l);
 665
 666                 if (tle->resdom->resjunk)
 667                         continue;                       /* ignore resjunk columns */
 668                 if (colType == NULL)
 669                         elog(ERROR, "wrong number of tlist entries");
 670                 if (tle->resdom->restype != lfirst_oid(colType))
 671                         differentTypes[tle->resdom->resno] = true;
 672                 colType = lnext(colType);
 673         }
 674         if (colType != NULL)
 675                 elog(ERROR, "wrong number of tlist entries");
 676 }
 677
 678 /*
 679  * qual_is_pushdown_safe - is a particular qual safe to push down?
 680  *
 681  * qual is a restriction clause applying to the given subquery (whose RTE
 682  * has index rti in the parent query).
 683  *
 684  * Conditions checked here:
 685  *
 686  * 1. The qual must not contain any subselects (mainly because I'm not sure
 687  * it will work correctly: sublinks will already have been transformed into
 688  * subplans in the qual, but not in the subquery).
 689  *
 690  * 2. The qual must not refer to any subquery output columns that were
 691  * found to have inconsistent types across a set operation tree by
 692  * subquery_is_pushdown_safe().
 693  *
 694  * 3. If the subquery uses DISTINCT ON, we must not push down any quals that
 695  * refer to non-DISTINCT output columns, because that could change the set
 696  * of rows returned.  This condition is vacuous for DISTINCT, because then
 697  * there are no non-DISTINCT output columns, but unfortunately it's fairly
 698  * expensive to tell the difference between DISTINCT and DISTINCT ON in the
 699  * parsetree representation.  It's cheaper to just make sure all the Vars
 700  * in the qual refer to DISTINCT columns.
 701  *
 702  * 4. We must not push down any quals that refer to subselect outputs that
 703  * return sets, else we'd introduce functions-returning-sets into the
 704  * subquery's WHERE/HAVING quals.
 705  */
 706 static bool
 707 qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
 708                                           bool *differentTypes)
 709 {
 710         bool            safe = true;
 711         List       *vars;
 712         ListCell   *vl;
 713         Bitmapset  *tested = NULL;
 714
 715         /* Refuse subselects (point 1) */
 716         if (contain_subplans(qual))
 717                 return false;
 718
 719         /*
 720          * Examine all Vars used in clause; since it's a restriction clause,
 721          * all such Vars must refer to subselect output columns.
 722          */
 723         vars = pull_var_clause(qual, false);
 724         foreach(vl, vars)
 725         {
 726                 Var                *var = (Var *) lfirst(vl);
 727                 TargetEntry *tle;
 728
 729                 Assert(var->varno == rti);
 730
 731                 /*
 732                  * We use a bitmapset to avoid testing the same attno more than
 733                  * once.  (NB: this only works because subquery outputs can't have
 734                  * negative attnos.)
 735                  */
 736                 if (bms_is_member(var->varattno, tested))
 737                         continue;
 738                 tested = bms_add_member(tested, var->varattno);
 739
 740                 /* Check point 2 */
 741                 if (differentTypes[var->varattno])
 742                 {
 743                         safe = false;
 744                         break;
 745                 }
 746
 747                 /* Must find the tlist element referenced by the Var */
 748                 tle = get_tle_by_resno(subquery->targetList, var->varattno);
 749                 Assert(tle != NULL);
 750                 Assert(!tle->resdom->resjunk);
 751
 752                 /* If subquery uses DISTINCT or DISTINCT ON, check point 3 */
 753                 if (subquery->distinctClause != NIL &&
 754                         !targetIsInSortList(tle, subquery->distinctClause))
 755                 {
 756                         /* non-DISTINCT column, so fail */
 757                         safe = false;
 758                         break;
 759                 }
 760
 761                 /* Refuse functions returning sets (point 4) */
 762                 if (expression_returns_set((Node *) tle->expr))
 763                 {
 764                         safe = false;
 765                         break;
 766                 }
 767         }
 768
 769         list_free(vars);
 770         bms_free(tested);
 771
 772         return safe;
 773 }
 774
 775 /*
 776  * subquery_push_qual - push down a qual that we have determined is safe
 777  */
 778 static void
 779 subquery_push_qual(Query *subquery, RangeTblEntry *rte, Index rti, Node *qual)
 780 {
 781         if (subquery->setOperations != NULL)
 782         {
 783                 /* Recurse to push it separately to each component query */
 784                 recurse_push_qual(subquery->setOperations, subquery, rte, rti, qual);
 785         }
 786         else
 787         {
 788                 /*
 789                  * We need to replace Vars in the qual (which must refer to
 790                  * outputs of the subquery) with copies of the subquery's
 791                  * targetlist expressions.      Note that at this point, any uplevel
 792                  * Vars in the qual should have been replaced with Params, so they
 793                  * need no work.
 794                  *
 795                  * This step also ensures that when we are pushing into a setop tree,
 796                  * each component query gets its own copy of the qual.
 797                  */
 798                 qual = ResolveNew(qual, rti, 0, rte,
 799                                                   subquery->targetList,
 800                                                   CMD_SELECT, 0);
 801                 subquery->havingQual = make_and_qual(subquery->havingQual,
 802                                                                                          qual);
 803
 804                 /*
 805                  * We need not change the subquery's hasAggs or hasSublinks flags,
 806                  * since we can't be pushing down any aggregates that weren't
 807                  * there before, and we don't push down subselects at all.
 808                  */
 809         }
 810 }
 811
 812 /*
 813  * Helper routine to recurse through setOperations tree
 814  */
 815 static void
 816 recurse_push_qual(Node *setOp, Query *topquery,
 817                                   RangeTblEntry *rte, Index rti, Node *qual)
 818 {
 819         if (IsA(setOp, RangeTblRef))
 820         {
 821                 RangeTblRef *rtr = (RangeTblRef *) setOp;
 822                 RangeTblEntry *subrte = rt_fetch(rtr->rtindex, topquery->rtable);
 823                 Query      *subquery = subrte->subquery;
 824
 825                 Assert(subquery != NULL);
 826                 subquery_push_qual(subquery, rte, rti, qual);
 827         }
 828         else if (IsA(setOp, SetOperationStmt))
 829         {
 830                 SetOperationStmt *op = (SetOperationStmt *) setOp;
 831
 832                 recurse_push_qual(op->larg, topquery, rte, rti, qual);
 833                 recurse_push_qual(op->rarg, topquery, rte, rti, qual);
 834         }
 835         else
 836         {
 837                 elog(ERROR, "unrecognized node type: %d",
 838                          (int) nodeTag(setOp));
 839         }
 840 }
 841
 842 /*****************************************************************************
 843  *                      DEBUG SUPPORT
 844  *****************************************************************************/
 845
 846 #ifdef OPTIMIZER_DEBUG
 847
 848 static void
 849 print_relids(Relids relids)
 850 {
 851         Relids          tmprelids;
 852         int                     x;
 853         bool            first = true;
 854
 855         tmprelids = bms_copy(relids);
 856         while ((x = bms_first_member(tmprelids)) >= 0)
 857         {
 858                 if (!first)
 859                         printf(" ");
 860                 printf("%d", x);
 861                 first = false;
 862         }
 863         bms_free(tmprelids);
 864 }
 865
 866 static void
 867 print_restrictclauses(Query *root, List *clauses)
 868 {
 869         ListCell   *l;
 870
 871         foreach(l, clauses)
 872         {
 873                 RestrictInfo *c = lfirst(l);
 874
 875                 print_expr((Node *) c->clause, root->rtable);
 876                 if (lnext(l))
 877                         printf(", ");
 878         }
 879 }
 880
 881 static void
 882 print_path(Query *root, Path *path, int indent)
 883 {
 884         const char *ptype;
 885         bool            join = false;
 886         Path       *subpath = NULL;
 887         int                     i;
 888
 889         switch (nodeTag(path))
 890         {
 891                 case T_Path:
 892                         ptype = "SeqScan";
 893                         break;
 894                 case T_IndexPath:
 895                         ptype = "IdxScan";
 896                         break;
 897                 case T_TidPath:
 898                         ptype = "TidScan";
 899                         break;
 900                 case T_AppendPath:
 901                         ptype = "Append";
 902                         break;
 903                 case T_ResultPath:
 904                         ptype = "Result";
 905                         subpath = ((ResultPath *) path)->subpath;
 906                         break;
 907                 case T_MaterialPath:
 908                         ptype = "Material";
 909                         subpath = ((MaterialPath *) path)->subpath;
 910                         break;
 911                 case T_UniquePath:
 912                         ptype = "Unique";
 913                         subpath = ((UniquePath *) path)->subpath;
 914                         break;
 915                 case T_NestPath:
 916                         ptype = "NestLoop";
 917                         join = true;
 918                         break;
 919                 case T_MergePath:
 920                         ptype = "MergeJoin";
 921                         join = true;
 922                         break;
 923                 case T_HashPath:
 924                         ptype = "HashJoin";
 925                         join = true;
 926                         break;
 927                 default:
 928                         ptype = "???Path";
 929                         break;
 930         }
 931
 932         for (i = 0; i < indent; i++)
 933                 printf("\t");
 934         printf("%s", ptype);
 935
 936         if (path->parent)
 937         {
 938                 printf("(");
 939                 print_relids(path->parent->relids);
 940                 printf(") rows=%.0f", path->parent->rows);
 941         }
 942         printf(" cost=%.2f..%.2f\n", path->startup_cost, path->total_cost);
 943
 944         if (path->pathkeys)
 945         {
 946                 for (i = 0; i < indent; i++)
 947                         printf("\t");
 948                 printf("  pathkeys: ");
 949                 print_pathkeys(path->pathkeys, root->rtable);
 950         }
 951
 952         if (join)
 953         {
 954                 JoinPath   *jp = (JoinPath *) path;
 955
 956                 for (i = 0; i < indent; i++)
 957                         printf("\t");
 958                 printf("  clauses: ");
 959                 print_restrictclauses(root, jp->joinrestrictinfo);
 960                 printf("\n");
 961
 962                 if (IsA(path, MergePath))
 963                 {
 964                         MergePath  *mp = (MergePath *) path;
 965
 966                         if (mp->outersortkeys || mp->innersortkeys)
 967                         {
 968                                 for (i = 0; i < indent; i++)
 969                                         printf("\t");
 970                                 printf("  sortouter=%d sortinner=%d\n",
 971                                            ((mp->outersortkeys) ? 1 : 0),
 972                                            ((mp->innersortkeys) ? 1 : 0));
 973                         }
 974                 }
 975
 976                 print_path(root, jp->outerjoinpath, indent + 1);
 977                 print_path(root, jp->innerjoinpath, indent + 1);
 978         }
 979
 980         if (subpath)
 981                 print_path(root, subpath, indent + 1);
 982 }
 983
 984 void
 985 debug_print_rel(Query *root, RelOptInfo *rel)
 986 {
 987         ListCell   *l;
 988
 989         printf("RELOPTINFO (");
 990         print_relids(rel->relids);
 991         printf("): rows=%.0f width=%d\n", rel->rows, rel->width);
 992
 993         if (rel->baserestrictinfo)
 994         {
 995                 printf("\tbaserestrictinfo: ");
 996                 print_restrictclauses(root, rel->baserestrictinfo);
 997                 printf("\n");
 998         }
 999
1000         foreach(l, rel->joininfo)
1001         {
1002                 JoinInfo   *j = (JoinInfo *) lfirst(l);
1003
1004                 printf("\tjoininfo (");
1005                 print_relids(j->unjoined_relids);
1006                 printf("): ");
1007                 print_restrictclauses(root, j->jinfo_restrictinfo);
1008                 printf("\n");
1009         }
1010
1011         printf("\tpath list:\n");
1012         foreach(l, rel->pathlist)
1013                 print_path(root, lfirst(l), 1);
1014         printf("\n\tcheapest startup path:\n");
1015         print_path(root, rel->cheapest_startup_path, 1);
1016         printf("\n\tcheapest total path:\n");
1017         print_path(root, rel->cheapest_total_path, 1);
1018         printf("\n");
1019         fflush(stdout);
1020 }
1021
1022 #endif   /* OPTIMIZER_DEBUG */