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) 1996-2007, 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 1.163 2007/04/21 21:01:44 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/parse_clause.h"
  31 #include "parser/parse_expr.h"
  32 #include "parser/parsetree.h"
  33 #include "rewrite/rewriteManip.h"
  34
  35
  36 /* These parameters are set by GUC */
  37 bool            enable_geqo = false;    /* just in case GUC doesn't set it */
  38 int                     geqo_threshold;
  39
  40
  41 static void set_base_rel_pathlists(PlannerInfo *root);
  42 static void set_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
  43                                                          Index rti, RangeTblEntry *rte);
  44 static void set_plain_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
  45                                            RangeTblEntry *rte);
  46 static void set_append_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
  47                                                 Index rti, RangeTblEntry *rte);
  48 static void set_subquery_pathlist(PlannerInfo *root, RelOptInfo *rel,
  49                                           Index rti, RangeTblEntry *rte);
  50 static void set_function_pathlist(PlannerInfo *root, RelOptInfo *rel,
  51                                           RangeTblEntry *rte);
  52 static void set_values_pathlist(PlannerInfo *root, RelOptInfo *rel,
  53                                         RangeTblEntry *rte);
  54 static RelOptInfo *make_rel_from_joinlist(PlannerInfo *root, List *joinlist);
  55 static RelOptInfo *make_one_rel_by_joins(PlannerInfo *root, int levels_needed,
  56                                           List *initial_rels);
  57 static bool subquery_is_pushdown_safe(Query *subquery, Query *topquery,
  58                                                   bool *differentTypes);
  59 static bool recurse_pushdown_safe(Node *setOp, Query *topquery,
  60                                           bool *differentTypes);
  61 static void compare_tlist_datatypes(List *tlist, List *colTypes,
  62                                                 bool *differentTypes);
  63 static bool qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
  64                                           bool *differentTypes);
  65 static void subquery_push_qual(Query *subquery,
  66                                    RangeTblEntry *rte, Index rti, Node *qual);
  67 static void recurse_push_qual(Node *setOp, Query *topquery,
  68                                   RangeTblEntry *rte, Index rti, Node *qual);
  69
  70
  71 /*
  72  * make_one_rel
  73  *        Finds all possible access paths for executing a query, returning a
  74  *        single rel that represents the join of all base rels in the query.
  75  */
  76 RelOptInfo *
  77 make_one_rel(PlannerInfo *root, List *joinlist)
  78 {
  79         RelOptInfo *rel;
  80
  81         /*
  82          * Generate access paths for the base rels.
  83          */
  84         set_base_rel_pathlists(root);
  85
  86         /*
  87          * Generate access paths for the entire join tree.
  88          */
  89         rel = make_rel_from_joinlist(root, joinlist);
  90
  91         /*
  92          * The result should join all and only the query's base rels.
  93          */
  94 #ifdef USE_ASSERT_CHECKING
  95         {
  96                 int                     num_base_rels = 0;
  97                 Index           rti;
  98
  99                 for (rti = 1; rti < root->simple_rel_array_size; rti++)
 100                 {
 101                         RelOptInfo *brel = root->simple_rel_array[rti];
 102
 103                         if (brel == NULL)
 104                                 continue;
 105
 106                         Assert(brel->relid == rti); /* sanity check on array */
 107
 108                         /* ignore RTEs that are "other rels" */
 109                         if (brel->reloptkind != RELOPT_BASEREL)
 110                                 continue;
 111
 112                         Assert(bms_is_member(rti, rel->relids));
 113                         num_base_rels++;
 114                 }
 115
 116                 Assert(bms_num_members(rel->relids) == num_base_rels);
 117         }
 118 #endif
 119
 120         return rel;
 121 }
 122
 123 /*
 124  * set_base_rel_pathlists
 125  *        Finds all paths available for scanning each base-relation entry.
 126  *        Sequential scan and any available indices are considered.
 127  *        Each useful path is attached to its relation's 'pathlist' field.
 128  */
 129 static void
 130 set_base_rel_pathlists(PlannerInfo *root)
 131 {
 132         Index           rti;
 133
 134         for (rti = 1; rti < root->simple_rel_array_size; rti++)
 135         {
 136                 RelOptInfo *rel = root->simple_rel_array[rti];
 137
 138                 /* there may be empty slots corresponding to non-baserel RTEs */
 139                 if (rel == NULL)
 140                         continue;
 141
 142                 Assert(rel->relid == rti);              /* sanity check on array */
 143
 144                 /* ignore RTEs that are "other rels" */
 145                 if (rel->reloptkind != RELOPT_BASEREL)
 146                         continue;
 147
 148                 set_rel_pathlist(root, rel, rti, root->simple_rte_array[rti]);
 149         }
 150 }
 151
 152 /*
 153  * set_rel_pathlist
 154  *        Build access paths for a base relation
 155  */
 156 static void
 157 set_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
 158                                  Index rti, RangeTblEntry *rte)
 159 {
 160         if (rte->inh)
 161         {
 162                 /* It's an "append relation", process accordingly */
 163                 set_append_rel_pathlist(root, rel, rti, rte);
 164         }
 165         else if (rel->rtekind == RTE_SUBQUERY)
 166         {
 167                 /* Subquery --- generate a separate plan for it */
 168                 set_subquery_pathlist(root, rel, rti, rte);
 169         }
 170         else if (rel->rtekind == RTE_FUNCTION)
 171         {
 172                 /* RangeFunction --- generate a separate plan for it */
 173                 set_function_pathlist(root, rel, rte);
 174         }
 175         else if (rel->rtekind == RTE_VALUES)
 176         {
 177                 /* Values list --- generate a separate plan for it */
 178                 set_values_pathlist(root, rel, rte);
 179         }
 180         else
 181         {
 182                 /* Plain relation */
 183                 Assert(rel->rtekind == RTE_RELATION);
 184                 set_plain_rel_pathlist(root, rel, rte);
 185         }
 186
 187 #ifdef OPTIMIZER_DEBUG
 188         debug_print_rel(root, rel);
 189 #endif
 190 }
 191
 192 /*
 193  * set_plain_rel_pathlist
 194  *        Build access paths for a plain relation (no subquery, no inheritance)
 195  */
 196 static void
 197 set_plain_rel_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 198 {
 199         /*
 200          * If we can prove we don't need to scan the rel via constraint exclusion,
 201          * set up a single dummy path for it.  (Rather than inventing a special
 202          * "dummy" path type, we represent this as an AppendPath with no members.)
 203          * We only need to check for regular baserels; if it's an otherrel, CE
 204          * was already checked in set_append_rel_pathlist().
 205          */
 206         if (rel->reloptkind == RELOPT_BASEREL &&
 207                 relation_excluded_by_constraints(rel, rte))
 208         {
 209                 /* Set dummy size estimates --- we leave attr_widths[] as zeroes */
 210                 rel->rows = 0;
 211                 rel->width = 0;
 212
 213                 add_path(rel, (Path *) create_append_path(rel, NIL));
 214
 215                 /* Select cheapest path (pretty easy in this case...) */
 216                 set_cheapest(rel);
 217
 218                 return;
 219         }
 220
 221         /* Mark rel with estimated output rows, width, etc */
 222         set_baserel_size_estimates(root, rel);
 223
 224         /* Test any partial indexes of rel for applicability */
 225         check_partial_indexes(root, rel);
 226
 227         /*
 228          * Check to see if we can extract any restriction conditions from join
 229          * quals that are OR-of-AND structures.  If so, add them to the rel's
 230          * restriction list, and recompute the size estimates.
 231          */
 232         if (create_or_index_quals(root, rel))
 233                 set_baserel_size_estimates(root, rel);
 234
 235         /*
 236          * Generate paths and add them to the rel's pathlist.
 237          *
 238          * Note: add_path() will discard any paths that are dominated by another
 239          * available path, keeping only those paths that are superior along at
 240          * least one dimension of cost or sortedness.
 241          */
 242
 243         /* Consider sequential scan */
 244         add_path(rel, create_seqscan_path(root, rel));
 245
 246         /* Consider index scans */
 247         create_index_paths(root, rel);
 248
 249         /* Consider TID scans */
 250         create_tidscan_paths(root, rel);
 251
 252         /* Now find the cheapest of the paths for this rel */
 253         set_cheapest(rel);
 254 }
 255
 256 /*
 257  * set_append_rel_pathlist
 258  *        Build access paths for an "append relation"
 259  *
 260  * The passed-in rel and RTE represent the entire append relation.      The
 261  * relation's contents are computed by appending together the output of
 262  * the individual member relations.  Note that in the inheritance case,
 263  * the first member relation is actually the same table as is mentioned in
 264  * the parent RTE ... but it has a different RTE and RelOptInfo.  This is
 265  * a good thing because their outputs are not the same size.
 266  */
 267 static void
 268 set_append_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
 269                                                 Index rti, RangeTblEntry *rte)
 270 {
 271         int                     parentRTindex = rti;
 272         List       *subpaths = NIL;
 273         ListCell   *l;
 274
 275         /*
 276          * XXX for now, can't handle inherited expansion of FOR UPDATE/SHARE; can
 277          * we do better?  (This will take some redesign because the executor
 278          * currently supposes that every rowMark relation is involved in every row
 279          * returned by the query.)
 280          */
 281         if (get_rowmark(root->parse, parentRTindex))
 282                 ereport(ERROR,
 283                                 (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
 284                                  errmsg("SELECT FOR UPDATE/SHARE is not supported for inheritance queries")));
 285
 286         /*
 287          * Initialize to compute size estimates for whole append relation
 288          */
 289         rel->rows = 0;
 290         rel->width = 0;
 291
 292         /*
 293          * Generate access paths for each member relation, and pick the cheapest
 294          * path for each one.
 295          */
 296         foreach(l, root->append_rel_list)
 297         {
 298                 AppendRelInfo *appinfo = (AppendRelInfo *) lfirst(l);
 299                 int                     childRTindex;
 300                 RangeTblEntry *childRTE;
 301                 RelOptInfo *childrel;
 302                 Path       *childpath;
 303                 ListCell   *parentvars;
 304                 ListCell   *childvars;
 305
 306                 /* append_rel_list contains all append rels; ignore others */
 307                 if (appinfo->parent_relid != parentRTindex)
 308                         continue;
 309
 310                 childRTindex = appinfo->child_relid;
 311                 childRTE = root->simple_rte_array[childRTindex];
 312
 313                 /*
 314                  * The child rel's RelOptInfo was already created during
 315                  * add_base_rels_to_query.
 316                  */
 317                 childrel = find_base_rel(root, childRTindex);
 318                 Assert(childrel->reloptkind == RELOPT_OTHER_MEMBER_REL);
 319
 320                 /*
 321                  * We have to copy the parent's targetlist and quals to the child,
 322                  * with appropriate substitution of variables.  However, only the
 323                  * baserestrictinfo quals are needed before we can check for
 324                  * constraint exclusion; so do that first and then check to see
 325                  * if we can disregard this child.
 326                  */
 327                 childrel->baserestrictinfo = (List *)
 328                         adjust_appendrel_attrs((Node *) rel->baserestrictinfo,
 329                                                                    appinfo);
 330
 331                 if (relation_excluded_by_constraints(childrel, childRTE))
 332                 {
 333                         /* this child need not be scanned, so just disregard it */
 334                         continue;
 335                 }
 336
 337                 /* CE failed, so finish copying targetlist and join quals */
 338                 childrel->joininfo = (List *)
 339                         adjust_appendrel_attrs((Node *) rel->joininfo,
 340                                                                    appinfo);
 341                 childrel->reltargetlist = (List *)
 342                         adjust_appendrel_attrs((Node *) rel->reltargetlist,
 343                                                                    appinfo);
 344
 345                 /*
 346                  * We have to make child entries in the EquivalenceClass data
 347                  * structures as well.
 348                  */
 349                 if (rel->has_eclass_joins)
 350                 {
 351                         add_child_rel_equivalences(root, appinfo, rel, childrel);
 352                         childrel->has_eclass_joins = true;
 353                 }
 354
 355                 /*
 356                  * Copy the parent's attr_needed data as well, with appropriate
 357                  * adjustment of relids and attribute numbers.
 358                  */
 359                 pfree(childrel->attr_needed);
 360                 childrel->attr_needed =
 361                         adjust_appendrel_attr_needed(rel, appinfo,
 362                                                                                  childrel->min_attr,
 363                                                                                  childrel->max_attr);
 364
 365                 /*
 366                  * Compute the child's access paths, and add the cheapest one to the
 367                  * Append path we are constructing for the parent.
 368                  *
 369                  * It's possible that the child is itself an appendrel, in which case
 370                  * we can "cut out the middleman" and just add its child paths to our
 371                  * own list.  (We don't try to do this earlier because we need to
 372                  * apply both levels of transformation to the quals.)
 373                  */
 374                 set_rel_pathlist(root, childrel, childRTindex, childRTE);
 375
 376                 childpath = childrel->cheapest_total_path;
 377                 if (IsA(childpath, AppendPath))
 378                         subpaths = list_concat(subpaths,
 379                                                                    ((AppendPath *) childpath)->subpaths);
 380                 else
 381                         subpaths = lappend(subpaths, childpath);
 382
 383                 /*
 384                  * Propagate size information from the child back to the parent. For
 385                  * simplicity, we use the largest widths from any child as the parent
 386                  * estimates.  (If you want to change this, beware of child
 387                  * attr_widths[] entries that haven't been set and are still 0.)
 388                  */
 389                 rel->rows += childrel->rows;
 390                 if (childrel->width > rel->width)
 391                         rel->width = childrel->width;
 392
 393                 forboth(parentvars, rel->reltargetlist,
 394                                 childvars, childrel->reltargetlist)
 395                 {
 396                         Var                *parentvar = (Var *) lfirst(parentvars);
 397                         Var                *childvar = (Var *) lfirst(childvars);
 398
 399                         if (IsA(parentvar, Var) &&
 400                                 IsA(childvar, Var))
 401                         {
 402                                 int                     pndx = parentvar->varattno - rel->min_attr;
 403                                 int                     cndx = childvar->varattno - childrel->min_attr;
 404
 405                                 if (childrel->attr_widths[cndx] > rel->attr_widths[pndx])
 406                                         rel->attr_widths[pndx] = childrel->attr_widths[cndx];
 407                         }
 408                 }
 409         }
 410
 411         /*
 412          * Set "raw tuples" count equal to "rows" for the appendrel; needed
 413          * because some places assume rel->tuples is valid for any baserel.
 414          */
 415         rel->tuples = rel->rows;
 416
 417         /*
 418          * Finally, build Append path and install it as the only access path for
 419          * the parent rel.      (Note: this is correct even if we have zero or one
 420          * live subpath due to constraint exclusion.)
 421          */
 422         add_path(rel, (Path *) create_append_path(rel, subpaths));
 423
 424         /* Select cheapest path (pretty easy in this case...) */
 425         set_cheapest(rel);
 426 }
 427
 428 /* quick-and-dirty test to see if any joining is needed */
 429 static bool
 430 has_multiple_baserels(PlannerInfo *root)
 431 {
 432         int                     num_base_rels = 0;
 433         Index           rti;
 434
 435         for (rti = 1; rti < root->simple_rel_array_size; rti++)
 436         {
 437                 RelOptInfo *brel = root->simple_rel_array[rti];
 438
 439                 if (brel == NULL)
 440                         continue;
 441
 442                 /* ignore RTEs that are "other rels" */
 443                 if (brel->reloptkind == RELOPT_BASEREL)
 444                         if (++num_base_rels > 1)
 445                                 return true;
 446         }
 447         return false;
 448 }
 449
 450 /*
 451  * set_subquery_pathlist
 452  *              Build the (single) access path for a subquery RTE
 453  */
 454 static void
 455 set_subquery_pathlist(PlannerInfo *root, RelOptInfo *rel,
 456                                           Index rti, RangeTblEntry *rte)
 457 {
 458         Query      *parse = root->parse;
 459         Query      *subquery = rte->subquery;
 460         bool       *differentTypes;
 461         double          tuple_fraction;
 462         PlannerInfo *subroot;
 463         List       *pathkeys;
 464
 465         /* We need a workspace for keeping track of set-op type coercions */
 466         differentTypes = (bool *)
 467                 palloc0((list_length(subquery->targetList) + 1) * sizeof(bool));
 468
 469         /*
 470          * If there are any restriction clauses that have been attached to the
 471          * subquery relation, consider pushing them down to become WHERE or HAVING
 472          * quals of the subquery itself.  This transformation is useful because it
 473          * may allow us to generate a better plan for the subquery than evaluating
 474          * all the subquery output rows and then filtering them.
 475          *
 476          * There are several cases where we cannot push down clauses. Restrictions
 477          * involving the subquery are checked by subquery_is_pushdown_safe().
 478          * Restrictions on individual clauses are checked by
 479          * qual_is_pushdown_safe().  Also, we don't want to push down
 480          * pseudoconstant clauses; better to have the gating node above the
 481          * subquery.
 482          *
 483          * Non-pushed-down clauses will get evaluated as qpquals of the
 484          * SubqueryScan node.
 485          *
 486          * XXX Are there any cases where we want to make a policy decision not to
 487          * push down a pushable qual, because it'd result in a worse plan?
 488          */
 489         if (rel->baserestrictinfo != NIL &&
 490                 subquery_is_pushdown_safe(subquery, subquery, differentTypes))
 491         {
 492                 /* OK to consider pushing down individual quals */
 493                 List       *upperrestrictlist = NIL;
 494                 ListCell   *l;
 495
 496                 foreach(l, rel->baserestrictinfo)
 497                 {
 498                         RestrictInfo *rinfo = (RestrictInfo *) lfirst(l);
 499                         Node       *clause = (Node *) rinfo->clause;
 500
 501                         if (!rinfo->pseudoconstant &&
 502                                 qual_is_pushdown_safe(subquery, rti, clause, differentTypes))
 503                         {
 504                                 /* Push it down */
 505                                 subquery_push_qual(subquery, rte, rti, clause);
 506                         }
 507                         else
 508                         {
 509                                 /* Keep it in the upper query */
 510                                 upperrestrictlist = lappend(upperrestrictlist, rinfo);
 511                         }
 512                 }
 513                 rel->baserestrictinfo = upperrestrictlist;
 514         }
 515
 516         pfree(differentTypes);
 517
 518         /*
 519          * We can safely pass the outer tuple_fraction down to the subquery if the
 520          * outer level has no joining, aggregation, or sorting to do. Otherwise
 521          * we'd better tell the subquery to plan for full retrieval. (XXX This
 522          * could probably be made more intelligent ...)
 523          */
 524         if (parse->hasAggs ||
 525                 parse->groupClause ||
 526                 parse->havingQual ||
 527                 parse->distinctClause ||
 528                 parse->sortClause ||
 529                 has_multiple_baserels(root))
 530                 tuple_fraction = 0.0;   /* default case */
 531         else
 532                 tuple_fraction = root->tuple_fraction;
 533
 534         /* Generate the plan for the subquery */
 535         rel->subplan = subquery_planner(root->glob, subquery,
 536                                                                         root->query_level + 1,
 537                                                                         tuple_fraction,
 538                                                                         &subroot);
 539         rel->subrtable = subroot->parse->rtable;
 540
 541         /* Copy number of output rows from subplan */
 542         rel->tuples = rel->subplan->plan_rows;
 543
 544         /* Mark rel with estimated output rows, width, etc */
 545         set_baserel_size_estimates(root, rel);
 546
 547         /* Convert subquery pathkeys to outer representation */
 548         pathkeys = convert_subquery_pathkeys(root, rel, subroot->query_pathkeys);
 549
 550         /* Generate appropriate path */
 551         add_path(rel, create_subqueryscan_path(rel, pathkeys));
 552
 553         /* Select cheapest path (pretty easy in this case...) */
 554         set_cheapest(rel);
 555 }
 556
 557 /*
 558  * set_function_pathlist
 559  *              Build the (single) access path for a function RTE
 560  */
 561 static void
 562 set_function_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 563 {
 564         /* Mark rel with estimated output rows, width, etc */
 565         set_function_size_estimates(root, rel);
 566
 567         /* Generate appropriate path */
 568         add_path(rel, create_functionscan_path(root, rel));
 569
 570         /* Select cheapest path (pretty easy in this case...) */
 571         set_cheapest(rel);
 572 }
 573
 574 /*
 575  * set_values_pathlist
 576  *              Build the (single) access path for a VALUES RTE
 577  */
 578 static void
 579 set_values_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 580 {
 581         /* Mark rel with estimated output rows, width, etc */
 582         set_values_size_estimates(root, rel);
 583
 584         /* Generate appropriate path */
 585         add_path(rel, create_valuesscan_path(root, rel));
 586
 587         /* Select cheapest path (pretty easy in this case...) */
 588         set_cheapest(rel);
 589 }
 590
 591 /*
 592  * make_rel_from_joinlist
 593  *        Build access paths using a "joinlist" to guide the join path search.
 594  *
 595  * See comments for deconstruct_jointree() for definition of the joinlist
 596  * data structure.
 597  */
 598 static RelOptInfo *
 599 make_rel_from_joinlist(PlannerInfo *root, List *joinlist)
 600 {
 601         int                     levels_needed;
 602         List       *initial_rels;
 603         ListCell   *jl;
 604
 605         /*
 606          * Count the number of child joinlist nodes.  This is the depth of the
 607          * dynamic-programming algorithm we must employ to consider all ways of
 608          * joining the child nodes.
 609          */
 610         levels_needed = list_length(joinlist);
 611
 612         if (levels_needed <= 0)
 613                 return NULL;                    /* nothing to do? */
 614
 615         /*
 616          * Construct a list of rels corresponding to the child joinlist nodes.
 617          * This may contain both base rels and rels constructed according to
 618          * sub-joinlists.
 619          */
 620         initial_rels = NIL;
 621         foreach(jl, joinlist)
 622         {
 623                 Node       *jlnode = (Node *) lfirst(jl);
 624                 RelOptInfo *thisrel;
 625
 626                 if (IsA(jlnode, RangeTblRef))
 627                 {
 628                         int                     varno = ((RangeTblRef *) jlnode)->rtindex;
 629
 630                         thisrel = find_base_rel(root, varno);
 631                 }
 632                 else if (IsA(jlnode, List))
 633                 {
 634                         /* Recurse to handle subproblem */
 635                         thisrel = make_rel_from_joinlist(root, (List *) jlnode);
 636                 }
 637                 else
 638                 {
 639                         elog(ERROR, "unrecognized joinlist node type: %d",
 640                                  (int) nodeTag(jlnode));
 641                         thisrel = NULL;         /* keep compiler quiet */
 642                 }
 643
 644                 initial_rels = lappend(initial_rels, thisrel);
 645         }
 646
 647         if (levels_needed == 1)
 648         {
 649                 /*
 650                  * Single joinlist node, so we're done.
 651                  */
 652                 return (RelOptInfo *) linitial(initial_rels);
 653         }
 654         else
 655         {
 656                 /*
 657                  * Consider the different orders in which we could join the rels,
 658                  * using either GEQO or regular optimizer.
 659                  */
 660                 if (enable_geqo && levels_needed >= geqo_threshold)
 661                         return geqo(root, levels_needed, initial_rels);
 662                 else
 663                         return make_one_rel_by_joins(root, levels_needed, initial_rels);
 664         }
 665 }
 666
 667 /*
 668  * make_one_rel_by_joins
 669  *        Find all possible joinpaths for a query by successively finding ways
 670  *        to join component relations into join relations.
 671  *
 672  * 'levels_needed' is the number of iterations needed, ie, the number of
 673  *              independent jointree items in the query.  This is > 1.
 674  *
 675  * 'initial_rels' is a list of RelOptInfo nodes for each independent
 676  *              jointree item.  These are the components to be joined together.
 677  *
 678  * Returns the final level of join relations, i.e., the relation that is
 679  * the result of joining all the original relations together.
 680  */
 681 static RelOptInfo *
 682 make_one_rel_by_joins(PlannerInfo *root, int levels_needed, List *initial_rels)
 683 {
 684         List      **joinitems;
 685         int                     lev;
 686         RelOptInfo *rel;
 687
 688         /*
 689          * We employ a simple "dynamic programming" algorithm: we first find all
 690          * ways to build joins of two jointree items, then all ways to build joins
 691          * of three items (from two-item joins and single items), then four-item
 692          * joins, and so on until we have considered all ways to join all the
 693          * items into one rel.
 694          *
 695          * joinitems[j] is a list of all the j-item rels.  Initially we set
 696          * joinitems[1] to represent all the single-jointree-item relations.
 697          */
 698         joinitems = (List **) palloc0((levels_needed + 1) * sizeof(List *));
 699
 700         joinitems[1] = initial_rels;
 701
 702         for (lev = 2; lev <= levels_needed; lev++)
 703         {
 704                 ListCell   *x;
 705
 706                 /*
 707                  * Determine all possible pairs of relations to be joined at this
 708                  * level, and build paths for making each one from every available
 709                  * pair of lower-level relations.
 710                  */
 711                 joinitems[lev] = make_rels_by_joins(root, lev, joinitems);
 712
 713                 /*
 714                  * Do cleanup work on each just-processed rel.
 715                  */
 716                 foreach(x, joinitems[lev])
 717                 {
 718                         rel = (RelOptInfo *) lfirst(x);
 719
 720                         /* Find and save the cheapest paths for this rel */
 721                         set_cheapest(rel);
 722
 723 #ifdef OPTIMIZER_DEBUG
 724                         debug_print_rel(root, rel);
 725 #endif
 726                 }
 727         }
 728
 729         /*
 730          * We should have a single rel at the final level.
 731          */
 732         if (joinitems[levels_needed] == NIL)
 733                 elog(ERROR, "failed to build any %d-way joins", levels_needed);
 734         Assert(list_length(joinitems[levels_needed]) == 1);
 735
 736         rel = (RelOptInfo *) linitial(joinitems[levels_needed]);
 737
 738         return rel;
 739 }
 740
 741 /*****************************************************************************
 742  *                      PUSHING QUALS DOWN INTO SUBQUERIES
 743  *****************************************************************************/
 744
 745 /*
 746  * subquery_is_pushdown_safe - is a subquery safe for pushing down quals?
 747  *
 748  * subquery is the particular component query being checked.  topquery
 749  * is the top component of a set-operations tree (the same Query if no
 750  * set-op is involved).
 751  *
 752  * Conditions checked here:
 753  *
 754  * 1. If the subquery has a LIMIT clause, we must not push down any quals,
 755  * since that could change the set of rows returned.
 756  *
 757  * 2. If the subquery contains EXCEPT or EXCEPT ALL set ops we cannot push
 758  * quals into it, because that would change the results.
 759  *
 760  * 3. For subqueries using UNION/UNION ALL/INTERSECT/INTERSECT ALL, we can
 761  * push quals into each component query, but the quals can only reference
 762  * subquery columns that suffer no type coercions in the set operation.
 763  * Otherwise there are possible semantic gotchas.  So, we check the
 764  * component queries to see if any of them have different output types;
 765  * differentTypes[k] is set true if column k has different type in any
 766  * component.
 767  */
 768 static bool
 769 subquery_is_pushdown_safe(Query *subquery, Query *topquery,
 770                                                   bool *differentTypes)
 771 {
 772         SetOperationStmt *topop;
 773
 774         /* Check point 1 */
 775         if (subquery->limitOffset != NULL || subquery->limitCount != NULL)
 776                 return false;
 777
 778         /* Are we at top level, or looking at a setop component? */
 779         if (subquery == topquery)
 780         {
 781                 /* Top level, so check any component queries */
 782                 if (subquery->setOperations != NULL)
 783                         if (!recurse_pushdown_safe(subquery->setOperations, topquery,
 784                                                                            differentTypes))
 785                                 return false;
 786         }
 787         else
 788         {
 789                 /* Setop component must not have more components (too weird) */
 790                 if (subquery->setOperations != NULL)
 791                         return false;
 792                 /* Check whether setop component output types match top level */
 793                 topop = (SetOperationStmt *) topquery->setOperations;
 794                 Assert(topop && IsA(topop, SetOperationStmt));
 795                 compare_tlist_datatypes(subquery->targetList,
 796                                                                 topop->colTypes,
 797                                                                 differentTypes);
 798         }
 799         return true;
 800 }
 801
 802 /*
 803  * Helper routine to recurse through setOperations tree
 804  */
 805 static bool
 806 recurse_pushdown_safe(Node *setOp, Query *topquery,
 807                                           bool *differentTypes)
 808 {
 809         if (IsA(setOp, RangeTblRef))
 810         {
 811                 RangeTblRef *rtr = (RangeTblRef *) setOp;
 812                 RangeTblEntry *rte = rt_fetch(rtr->rtindex, topquery->rtable);
 813                 Query      *subquery = rte->subquery;
 814
 815                 Assert(subquery != NULL);
 816                 return subquery_is_pushdown_safe(subquery, topquery, differentTypes);
 817         }
 818         else if (IsA(setOp, SetOperationStmt))
 819         {
 820                 SetOperationStmt *op = (SetOperationStmt *) setOp;
 821
 822                 /* EXCEPT is no good */
 823                 if (op->op == SETOP_EXCEPT)
 824                         return false;
 825                 /* Else recurse */
 826                 if (!recurse_pushdown_safe(op->larg, topquery, differentTypes))
 827                         return false;
 828                 if (!recurse_pushdown_safe(op->rarg, topquery, differentTypes))
 829                         return false;
 830         }
 831         else
 832         {
 833                 elog(ERROR, "unrecognized node type: %d",
 834                          (int) nodeTag(setOp));
 835         }
 836         return true;
 837 }
 838
 839 /*
 840  * Compare tlist's datatypes against the list of set-operation result types.
 841  * For any items that are different, mark the appropriate element of
 842  * differentTypes[] to show that this column will have type conversions.
 843  *
 844  * We don't have to care about typmods here: the only allowed difference
 845  * between set-op input and output typmods is input is a specific typmod
 846  * and output is -1, and that does not require a coercion.
 847  */
 848 static void
 849 compare_tlist_datatypes(List *tlist, List *colTypes,
 850                                                 bool *differentTypes)
 851 {
 852         ListCell   *l;
 853         ListCell   *colType = list_head(colTypes);
 854
 855         foreach(l, tlist)
 856         {
 857                 TargetEntry *tle = (TargetEntry *) lfirst(l);
 858
 859                 if (tle->resjunk)
 860                         continue;                       /* ignore resjunk columns */
 861                 if (colType == NULL)
 862                         elog(ERROR, "wrong number of tlist entries");
 863                 if (exprType((Node *) tle->expr) != lfirst_oid(colType))
 864                         differentTypes[tle->resno] = true;
 865                 colType = lnext(colType);
 866         }
 867         if (colType != NULL)
 868                 elog(ERROR, "wrong number of tlist entries");
 869 }
 870
 871 /*
 872  * qual_is_pushdown_safe - is a particular qual safe to push down?
 873  *
 874  * qual is a restriction clause applying to the given subquery (whose RTE
 875  * has index rti in the parent query).
 876  *
 877  * Conditions checked here:
 878  *
 879  * 1. The qual must not contain any subselects (mainly because I'm not sure
 880  * it will work correctly: sublinks will already have been transformed into
 881  * subplans in the qual, but not in the subquery).
 882  *
 883  * 2. The qual must not refer to the whole-row output of the subquery
 884  * (since there is no easy way to name that within the subquery itself).
 885  *
 886  * 3. The qual must not refer to any subquery output columns that were
 887  * found to have inconsistent types across a set operation tree by
 888  * subquery_is_pushdown_safe().
 889  *
 890  * 4. If the subquery uses DISTINCT ON, we must not push down any quals that
 891  * refer to non-DISTINCT output columns, because that could change the set
 892  * of rows returned.  This condition is vacuous for DISTINCT, because then
 893  * there are no non-DISTINCT output columns, but unfortunately it's fairly
 894  * expensive to tell the difference between DISTINCT and DISTINCT ON in the
 895  * parsetree representation.  It's cheaper to just make sure all the Vars
 896  * in the qual refer to DISTINCT columns.
 897  *
 898  * 5. We must not push down any quals that refer to subselect outputs that
 899  * return sets, else we'd introduce functions-returning-sets into the
 900  * subquery's WHERE/HAVING quals.
 901  *
 902  * 6. We must not push down any quals that refer to subselect outputs that
 903  * contain volatile functions, for fear of introducing strange results due
 904  * to multiple evaluation of a volatile function.
 905  */
 906 static bool
 907 qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
 908                                           bool *differentTypes)
 909 {
 910         bool            safe = true;
 911         List       *vars;
 912         ListCell   *vl;
 913         Bitmapset  *tested = NULL;
 914
 915         /* Refuse subselects (point 1) */
 916         if (contain_subplans(qual))
 917                 return false;
 918
 919         /*
 920          * Examine all Vars used in clause; since it's a restriction clause, all
 921          * such Vars must refer to subselect output columns.
 922          */
 923         vars = pull_var_clause(qual, false);
 924         foreach(vl, vars)
 925         {
 926                 Var                *var = (Var *) lfirst(vl);
 927                 TargetEntry *tle;
 928
 929                 Assert(var->varno == rti);
 930
 931                 /* Check point 2 */
 932                 if (var->varattno == 0)
 933                 {
 934                         safe = false;
 935                         break;
 936                 }
 937
 938                 /*
 939                  * We use a bitmapset to avoid testing the same attno more than once.
 940                  * (NB: this only works because subquery outputs can't have negative
 941                  * attnos.)
 942                  */
 943                 if (bms_is_member(var->varattno, tested))
 944                         continue;
 945                 tested = bms_add_member(tested, var->varattno);
 946
 947                 /* Check point 3 */
 948                 if (differentTypes[var->varattno])
 949                 {
 950                         safe = false;
 951                         break;
 952                 }
 953
 954                 /* Must find the tlist element referenced by the Var */
 955                 tle = get_tle_by_resno(subquery->targetList, var->varattno);
 956                 Assert(tle != NULL);
 957                 Assert(!tle->resjunk);
 958
 959                 /* If subquery uses DISTINCT or DISTINCT ON, check point 4 */
 960                 if (subquery->distinctClause != NIL &&
 961                         !targetIsInSortList(tle, InvalidOid, subquery->distinctClause))
 962                 {
 963                         /* non-DISTINCT column, so fail */
 964                         safe = false;
 965                         break;
 966                 }
 967
 968                 /* Refuse functions returning sets (point 5) */
 969                 if (expression_returns_set((Node *) tle->expr))
 970                 {
 971                         safe = false;
 972                         break;
 973                 }
 974
 975                 /* Refuse volatile functions (point 6) */
 976                 if (contain_volatile_functions((Node *) tle->expr))
 977                 {
 978                         safe = false;
 979                         break;
 980                 }
 981         }
 982
 983         list_free(vars);
 984         bms_free(tested);
 985
 986         return safe;
 987 }
 988
 989 /*
 990  * subquery_push_qual - push down a qual that we have determined is safe
 991  */
 992 static void
 993 subquery_push_qual(Query *subquery, RangeTblEntry *rte, Index rti, Node *qual)
 994 {
 995         if (subquery->setOperations != NULL)
 996         {
 997                 /* Recurse to push it separately to each component query */
 998                 recurse_push_qual(subquery->setOperations, subquery,
 999                                                   rte, rti, qual);
1000         }
1001         else
1002         {
1003                 /*
1004                  * We need to replace Vars in the qual (which must refer to outputs of
1005                  * the subquery) with copies of the subquery's targetlist expressions.
1006                  * Note that at this point, any uplevel Vars in the qual should have
1007                  * been replaced with Params, so they need no work.
1008                  *
1009                  * This step also ensures that when we are pushing into a setop tree,
1010                  * each component query gets its own copy of the qual.
1011                  */
1012                 qual = ResolveNew(qual, rti, 0, rte,
1013                                                   subquery->targetList,
1014                                                   CMD_SELECT, 0);
1015
1016                 /*
1017                  * Now attach the qual to the proper place: normally WHERE, but if the
1018                  * subquery uses grouping or aggregation, put it in HAVING (since the
1019                  * qual really refers to the group-result rows).
1020                  */
1021                 if (subquery->hasAggs || subquery->groupClause || subquery->havingQual)
1022                         subquery->havingQual = make_and_qual(subquery->havingQual, qual);
1023                 else
1024                         subquery->jointree->quals =
1025                                 make_and_qual(subquery->jointree->quals, qual);
1026
1027                 /*
1028                  * We need not change the subquery's hasAggs or hasSublinks flags,
1029                  * since we can't be pushing down any aggregates that weren't there
1030                  * before, and we don't push down subselects at all.
1031                  */
1032         }
1033 }
1034
1035 /*
1036  * Helper routine to recurse through setOperations tree
1037  */
1038 static void
1039 recurse_push_qual(Node *setOp, Query *topquery,
1040                                   RangeTblEntry *rte, Index rti, Node *qual)
1041 {
1042         if (IsA(setOp, RangeTblRef))
1043         {
1044                 RangeTblRef *rtr = (RangeTblRef *) setOp;
1045                 RangeTblEntry *subrte = rt_fetch(rtr->rtindex, topquery->rtable);
1046                 Query      *subquery = subrte->subquery;
1047
1048                 Assert(subquery != NULL);
1049                 subquery_push_qual(subquery, rte, rti, qual);
1050         }
1051         else if (IsA(setOp, SetOperationStmt))
1052         {
1053                 SetOperationStmt *op = (SetOperationStmt *) setOp;
1054
1055                 recurse_push_qual(op->larg, topquery, rte, rti, qual);
1056                 recurse_push_qual(op->rarg, topquery, rte, rti, qual);
1057         }
1058         else
1059         {
1060                 elog(ERROR, "unrecognized node type: %d",
1061                          (int) nodeTag(setOp));
1062         }
1063 }
1064
1065 /*****************************************************************************
1066  *                      DEBUG SUPPORT
1067  *****************************************************************************/
1068
1069 #ifdef OPTIMIZER_DEBUG
1070
1071 static void
1072 print_relids(Relids relids)
1073 {
1074         Relids          tmprelids;
1075         int                     x;
1076         bool            first = true;
1077
1078         tmprelids = bms_copy(relids);
1079         while ((x = bms_first_member(tmprelids)) >= 0)
1080         {
1081                 if (!first)
1082                         printf(" ");
1083                 printf("%d", x);
1084                 first = false;
1085         }
1086         bms_free(tmprelids);
1087 }
1088
1089 static void
1090 print_restrictclauses(PlannerInfo *root, List *clauses)
1091 {
1092         ListCell   *l;
1093
1094         foreach(l, clauses)
1095         {
1096                 RestrictInfo *c = lfirst(l);
1097
1098                 print_expr((Node *) c->clause, root->parse->rtable);
1099                 if (lnext(l))
1100                         printf(", ");
1101         }
1102 }
1103
1104 static void
1105 print_path(PlannerInfo *root, Path *path, int indent)
1106 {
1107         const char *ptype;
1108         bool            join = false;
1109         Path       *subpath = NULL;
1110         int                     i;
1111
1112         switch (nodeTag(path))
1113         {
1114                 case T_Path:
1115                         ptype = "SeqScan";
1116                         break;
1117                 case T_IndexPath:
1118                         ptype = "IdxScan";
1119                         break;
1120                 case T_BitmapHeapPath:
1121                         ptype = "BitmapHeapScan";
1122                         break;
1123                 case T_BitmapAndPath:
1124                         ptype = "BitmapAndPath";
1125                         break;
1126                 case T_BitmapOrPath:
1127                         ptype = "BitmapOrPath";
1128                         break;
1129                 case T_TidPath:
1130                         ptype = "TidScan";
1131                         break;
1132                 case T_AppendPath:
1133                         ptype = "Append";
1134                         break;
1135                 case T_ResultPath:
1136                         ptype = "Result";
1137                         break;
1138                 case T_MaterialPath:
1139                         ptype = "Material";
1140                         subpath = ((MaterialPath *) path)->subpath;
1141                         break;
1142                 case T_UniquePath:
1143                         ptype = "Unique";
1144                         subpath = ((UniquePath *) path)->subpath;
1145                         break;
1146                 case T_NestPath:
1147                         ptype = "NestLoop";
1148                         join = true;
1149                         break;
1150                 case T_MergePath:
1151                         ptype = "MergeJoin";
1152                         join = true;
1153                         break;
1154                 case T_HashPath:
1155                         ptype = "HashJoin";
1156                         join = true;
1157                         break;
1158                 default:
1159                         ptype = "???Path";
1160                         break;
1161         }
1162
1163         for (i = 0; i < indent; i++)
1164                 printf("\t");
1165         printf("%s", ptype);
1166
1167         if (path->parent)
1168         {
1169                 printf("(");
1170                 print_relids(path->parent->relids);
1171                 printf(") rows=%.0f", path->parent->rows);
1172         }
1173         printf(" cost=%.2f..%.2f\n", path->startup_cost, path->total_cost);
1174
1175         if (path->pathkeys)
1176         {
1177                 for (i = 0; i < indent; i++)
1178                         printf("\t");
1179                 printf("  pathkeys: ");
1180                 print_pathkeys(path->pathkeys, root->parse->rtable);
1181         }
1182
1183         if (join)
1184         {
1185                 JoinPath   *jp = (JoinPath *) path;
1186
1187                 for (i = 0; i < indent; i++)
1188                         printf("\t");
1189                 printf("  clauses: ");
1190                 print_restrictclauses(root, jp->joinrestrictinfo);
1191                 printf("\n");
1192
1193                 if (IsA(path, MergePath))
1194                 {
1195                         MergePath  *mp = (MergePath *) path;
1196
1197                         if (mp->outersortkeys || mp->innersortkeys)
1198                         {
1199                                 for (i = 0; i < indent; i++)
1200                                         printf("\t");
1201                                 printf("  sortouter=%d sortinner=%d\n",
1202                                            ((mp->outersortkeys) ? 1 : 0),
1203                                            ((mp->innersortkeys) ? 1 : 0));
1204                         }
1205                 }
1206
1207                 print_path(root, jp->outerjoinpath, indent + 1);
1208                 print_path(root, jp->innerjoinpath, indent + 1);
1209         }
1210
1211         if (subpath)
1212                 print_path(root, subpath, indent + 1);
1213 }
1214
1215 void
1216 debug_print_rel(PlannerInfo *root, RelOptInfo *rel)
1217 {
1218         ListCell   *l;
1219
1220         printf("RELOPTINFO (");
1221         print_relids(rel->relids);
1222         printf("): rows=%.0f width=%d\n", rel->rows, rel->width);
1223
1224         if (rel->baserestrictinfo)
1225         {
1226                 printf("\tbaserestrictinfo: ");
1227                 print_restrictclauses(root, rel->baserestrictinfo);
1228                 printf("\n");
1229         }
1230
1231         if (rel->joininfo)
1232         {
1233                 printf("\tjoininfo: ");
1234                 print_restrictclauses(root, rel->joininfo);
1235                 printf("\n");
1236         }
1237
1238         printf("\tpath list:\n");
1239         foreach(l, rel->pathlist)
1240                 print_path(root, lfirst(l), 1);
1241         printf("\n\tcheapest startup path:\n");
1242         print_path(root, rel->cheapest_startup_path, 1);
1243         printf("\n\tcheapest total path:\n");
1244         print_path(root, rel->cheapest_total_path, 1);
1245         printf("\n");
1246         fflush(stdout);
1247 }
1248
1249 #endif   /* OPTIMIZER_DEBUG */