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 1.164 2007/05/26 18:23:01 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_dummy_rel_pathlist(RelOptInfo *rel);
  49 static void set_subquery_pathlist(PlannerInfo *root, RelOptInfo *rel,
  50                                           Index rti, RangeTblEntry *rte);
  51 static void set_function_pathlist(PlannerInfo *root, RelOptInfo *rel,
  52                                           RangeTblEntry *rte);
  53 static void set_values_pathlist(PlannerInfo *root, RelOptInfo *rel,
  54                                         RangeTblEntry *rte);
  55 static RelOptInfo *make_rel_from_joinlist(PlannerInfo *root, List *joinlist);
  56 static RelOptInfo *make_one_rel_by_joins(PlannerInfo *root, int levels_needed,
  57                                           List *initial_rels);
  58 static bool subquery_is_pushdown_safe(Query *subquery, Query *topquery,
  59                                                   bool *differentTypes);
  60 static bool recurse_pushdown_safe(Node *setOp, Query *topquery,
  61                                           bool *differentTypes);
  62 static void compare_tlist_datatypes(List *tlist, List *colTypes,
  63                                                 bool *differentTypes);
  64 static bool qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
  65                                           bool *differentTypes);
  66 static void subquery_push_qual(Query *subquery,
  67                                    RangeTblEntry *rte, Index rti, Node *qual);
  68 static void recurse_push_qual(Node *setOp, Query *topquery,
  69                                   RangeTblEntry *rte, Index rti, Node *qual);
  70
  71
  72 /*
  73  * make_one_rel
  74  *        Finds all possible access paths for executing a query, returning a
  75  *        single rel that represents the join of all base rels in the query.
  76  */
  77 RelOptInfo *
  78 make_one_rel(PlannerInfo *root, List *joinlist)
  79 {
  80         RelOptInfo *rel;
  81
  82         /*
  83          * Generate access paths for the base rels.
  84          */
  85         set_base_rel_pathlists(root);
  86
  87         /*
  88          * Generate access paths for the entire join tree.
  89          */
  90         rel = make_rel_from_joinlist(root, joinlist);
  91
  92         /*
  93          * The result should join all and only the query's base rels.
  94          */
  95 #ifdef USE_ASSERT_CHECKING
  96         {
  97                 int                     num_base_rels = 0;
  98                 Index           rti;
  99
 100                 for (rti = 1; rti < root->simple_rel_array_size; rti++)
 101                 {
 102                         RelOptInfo *brel = root->simple_rel_array[rti];
 103
 104                         if (brel == NULL)
 105                                 continue;
 106
 107                         Assert(brel->relid == rti); /* sanity check on array */
 108
 109                         /* ignore RTEs that are "other rels" */
 110                         if (brel->reloptkind != RELOPT_BASEREL)
 111                                 continue;
 112
 113                         Assert(bms_is_member(rti, rel->relids));
 114                         num_base_rels++;
 115                 }
 116
 117                 Assert(bms_num_members(rel->relids) == num_base_rels);
 118         }
 119 #endif
 120
 121         return rel;
 122 }
 123
 124 /*
 125  * set_base_rel_pathlists
 126  *        Finds all paths available for scanning each base-relation entry.
 127  *        Sequential scan and any available indices are considered.
 128  *        Each useful path is attached to its relation's 'pathlist' field.
 129  */
 130 static void
 131 set_base_rel_pathlists(PlannerInfo *root)
 132 {
 133         Index           rti;
 134
 135         for (rti = 1; rti < root->simple_rel_array_size; rti++)
 136         {
 137                 RelOptInfo *rel = root->simple_rel_array[rti];
 138
 139                 /* there may be empty slots corresponding to non-baserel RTEs */
 140                 if (rel == NULL)
 141                         continue;
 142
 143                 Assert(rel->relid == rti);              /* sanity check on array */
 144
 145                 /* ignore RTEs that are "other rels" */
 146                 if (rel->reloptkind != RELOPT_BASEREL)
 147                         continue;
 148
 149                 set_rel_pathlist(root, rel, rti, root->simple_rte_array[rti]);
 150         }
 151 }
 152
 153 /*
 154  * set_rel_pathlist
 155  *        Build access paths for a base relation
 156  */
 157 static void
 158 set_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
 159                                  Index rti, RangeTblEntry *rte)
 160 {
 161         if (rte->inh)
 162         {
 163                 /* It's an "append relation", process accordingly */
 164                 set_append_rel_pathlist(root, rel, rti, rte);
 165         }
 166         else if (rel->rtekind == RTE_SUBQUERY)
 167         {
 168                 /* Subquery --- generate a separate plan for it */
 169                 set_subquery_pathlist(root, rel, rti, rte);
 170         }
 171         else if (rel->rtekind == RTE_FUNCTION)
 172         {
 173                 /* RangeFunction --- generate a separate plan for it */
 174                 set_function_pathlist(root, rel, rte);
 175         }
 176         else if (rel->rtekind == RTE_VALUES)
 177         {
 178                 /* Values list --- generate a separate plan for it */
 179                 set_values_pathlist(root, rel, rte);
 180         }
 181         else
 182         {
 183                 /* Plain relation */
 184                 Assert(rel->rtekind == RTE_RELATION);
 185                 set_plain_rel_pathlist(root, rel, rte);
 186         }
 187
 188 #ifdef OPTIMIZER_DEBUG
 189         debug_print_rel(root, rel);
 190 #endif
 191 }
 192
 193 /*
 194  * set_plain_rel_pathlist
 195  *        Build access paths for a plain relation (no subquery, no inheritance)
 196  */
 197 static void
 198 set_plain_rel_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 199 {
 200         /*
 201          * If we can prove we don't need to scan the rel via constraint exclusion,
 202          * set up a single dummy path for it.  We only need to check for regular
 203          * baserels; if it's an otherrel, CE was already checked in
 204          * set_append_rel_pathlist().
 205          */
 206         if (rel->reloptkind == RELOPT_BASEREL &&
 207                 relation_excluded_by_constraints(rel, rte))
 208         {
 209                 set_dummy_rel_pathlist(rel);
 210                 return;
 211         }
 212
 213         /* Mark rel with estimated output rows, width, etc */
 214         set_baserel_size_estimates(root, rel);
 215
 216         /* Test any partial indexes of rel for applicability */
 217         check_partial_indexes(root, rel);
 218
 219         /*
 220          * Check to see if we can extract any restriction conditions from join
 221          * quals that are OR-of-AND structures.  If so, add them to the rel's
 222          * restriction list, and recompute the size estimates.
 223          */
 224         if (create_or_index_quals(root, rel))
 225                 set_baserel_size_estimates(root, rel);
 226
 227         /*
 228          * Generate paths and add them to the rel's pathlist.
 229          *
 230          * Note: add_path() will discard any paths that are dominated by another
 231          * available path, keeping only those paths that are superior along at
 232          * least one dimension of cost or sortedness.
 233          */
 234
 235         /* Consider sequential scan */
 236         add_path(rel, create_seqscan_path(root, rel));
 237
 238         /* Consider index scans */
 239         create_index_paths(root, rel);
 240
 241         /* Consider TID scans */
 242         create_tidscan_paths(root, rel);
 243
 244         /* Now find the cheapest of the paths for this rel */
 245         set_cheapest(rel);
 246 }
 247
 248 /*
 249  * set_append_rel_pathlist
 250  *        Build access paths for an "append relation"
 251  *
 252  * The passed-in rel and RTE represent the entire append relation.      The
 253  * relation's contents are computed by appending together the output of
 254  * the individual member relations.  Note that in the inheritance case,
 255  * the first member relation is actually the same table as is mentioned in
 256  * the parent RTE ... but it has a different RTE and RelOptInfo.  This is
 257  * a good thing because their outputs are not the same size.
 258  */
 259 static void
 260 set_append_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
 261                                                 Index rti, RangeTblEntry *rte)
 262 {
 263         int                     parentRTindex = rti;
 264         List       *subpaths = NIL;
 265         ListCell   *l;
 266
 267         /*
 268          * XXX for now, can't handle inherited expansion of FOR UPDATE/SHARE; can
 269          * we do better?  (This will take some redesign because the executor
 270          * currently supposes that every rowMark relation is involved in every row
 271          * returned by the query.)
 272          */
 273         if (get_rowmark(root->parse, parentRTindex))
 274                 ereport(ERROR,
 275                                 (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
 276                                  errmsg("SELECT FOR UPDATE/SHARE is not supported for inheritance queries")));
 277
 278         /*
 279          * Initialize to compute size estimates for whole append relation
 280          */
 281         rel->rows = 0;
 282         rel->width = 0;
 283
 284         /*
 285          * Generate access paths for each member relation, and pick the cheapest
 286          * path for each one.
 287          */
 288         foreach(l, root->append_rel_list)
 289         {
 290                 AppendRelInfo *appinfo = (AppendRelInfo *) lfirst(l);
 291                 int                     childRTindex;
 292                 RangeTblEntry *childRTE;
 293                 RelOptInfo *childrel;
 294                 Path       *childpath;
 295                 ListCell   *parentvars;
 296                 ListCell   *childvars;
 297
 298                 /* append_rel_list contains all append rels; ignore others */
 299                 if (appinfo->parent_relid != parentRTindex)
 300                         continue;
 301
 302                 childRTindex = appinfo->child_relid;
 303                 childRTE = root->simple_rte_array[childRTindex];
 304
 305                 /*
 306                  * The child rel's RelOptInfo was already created during
 307                  * add_base_rels_to_query.
 308                  */
 309                 childrel = find_base_rel(root, childRTindex);
 310                 Assert(childrel->reloptkind == RELOPT_OTHER_MEMBER_REL);
 311
 312                 /*
 313                  * We have to copy the parent's targetlist and quals to the child,
 314                  * with appropriate substitution of variables.  However, only the
 315                  * baserestrictinfo quals are needed before we can check for
 316                  * constraint exclusion; so do that first and then check to see
 317                  * if we can disregard this child.
 318                  */
 319                 childrel->baserestrictinfo = (List *)
 320                         adjust_appendrel_attrs((Node *) rel->baserestrictinfo,
 321                                                                    appinfo);
 322
 323                 if (relation_excluded_by_constraints(childrel, childRTE))
 324                 {
 325                         /*
 326                          * This child need not be scanned, so we can omit it from the
 327                          * appendrel.  Mark it with a dummy cheapest-path though, in
 328                          * case best_appendrel_indexscan() looks at it later.
 329                          */
 330                         set_dummy_rel_pathlist(childrel);
 331                         continue;
 332                 }
 333
 334                 /* CE failed, so finish copying targetlist and join quals */
 335                 childrel->joininfo = (List *)
 336                         adjust_appendrel_attrs((Node *) rel->joininfo,
 337                                                                    appinfo);
 338                 childrel->reltargetlist = (List *)
 339                         adjust_appendrel_attrs((Node *) rel->reltargetlist,
 340                                                                    appinfo);
 341
 342                 /*
 343                  * We have to make child entries in the EquivalenceClass data
 344                  * structures as well.
 345                  */
 346                 if (rel->has_eclass_joins)
 347                 {
 348                         add_child_rel_equivalences(root, appinfo, rel, childrel);
 349                         childrel->has_eclass_joins = true;
 350                 }
 351
 352                 /*
 353                  * Copy the parent's attr_needed data as well, with appropriate
 354                  * adjustment of relids and attribute numbers.
 355                  */
 356                 pfree(childrel->attr_needed);
 357                 childrel->attr_needed =
 358                         adjust_appendrel_attr_needed(rel, appinfo,
 359                                                                                  childrel->min_attr,
 360                                                                                  childrel->max_attr);
 361
 362                 /*
 363                  * Compute the child's access paths, and add the cheapest one to the
 364                  * Append path we are constructing for the parent.
 365                  *
 366                  * It's possible that the child is itself an appendrel, in which case
 367                  * we can "cut out the middleman" and just add its child paths to our
 368                  * own list.  (We don't try to do this earlier because we need to
 369                  * apply both levels of transformation to the quals.)
 370                  */
 371                 set_rel_pathlist(root, childrel, childRTindex, childRTE);
 372
 373                 childpath = childrel->cheapest_total_path;
 374                 if (IsA(childpath, AppendPath))
 375                         subpaths = list_concat(subpaths,
 376                                                                    ((AppendPath *) childpath)->subpaths);
 377                 else
 378                         subpaths = lappend(subpaths, childpath);
 379
 380                 /*
 381                  * Propagate size information from the child back to the parent. For
 382                  * simplicity, we use the largest widths from any child as the parent
 383                  * estimates.  (If you want to change this, beware of child
 384                  * attr_widths[] entries that haven't been set and are still 0.)
 385                  */
 386                 rel->rows += childrel->rows;
 387                 if (childrel->width > rel->width)
 388                         rel->width = childrel->width;
 389
 390                 forboth(parentvars, rel->reltargetlist,
 391                                 childvars, childrel->reltargetlist)
 392                 {
 393                         Var                *parentvar = (Var *) lfirst(parentvars);
 394                         Var                *childvar = (Var *) lfirst(childvars);
 395
 396                         if (IsA(parentvar, Var) &&
 397                                 IsA(childvar, Var))
 398                         {
 399                                 int                     pndx = parentvar->varattno - rel->min_attr;
 400                                 int                     cndx = childvar->varattno - childrel->min_attr;
 401
 402                                 if (childrel->attr_widths[cndx] > rel->attr_widths[pndx])
 403                                         rel->attr_widths[pndx] = childrel->attr_widths[cndx];
 404                         }
 405                 }
 406         }
 407
 408         /*
 409          * Set "raw tuples" count equal to "rows" for the appendrel; needed
 410          * because some places assume rel->tuples is valid for any baserel.
 411          */
 412         rel->tuples = rel->rows;
 413
 414         /*
 415          * Finally, build Append path and install it as the only access path for
 416          * the parent rel.      (Note: this is correct even if we have zero or one
 417          * live subpath due to constraint exclusion.)
 418          */
 419         add_path(rel, (Path *) create_append_path(rel, subpaths));
 420
 421         /* Select cheapest path (pretty easy in this case...) */
 422         set_cheapest(rel);
 423 }
 424
 425 /*
 426  * set_dummy_rel_pathlist
 427  *        Build a dummy path for a relation that's been excluded by constraints
 428  *
 429  * Rather than inventing a special "dummy" path type, we represent this as an
 430  * AppendPath with no members.
 431  */
 432 static void
 433 set_dummy_rel_pathlist(RelOptInfo *rel)
 434 {
 435         /* Set dummy size estimates --- we leave attr_widths[] as zeroes */
 436         rel->rows = 0;
 437         rel->width = 0;
 438
 439         add_path(rel, (Path *) create_append_path(rel, NIL));
 440
 441         /* Select cheapest path (pretty easy in this case...) */
 442         set_cheapest(rel);
 443 }
 444
 445 /* quick-and-dirty test to see if any joining is needed */
 446 static bool
 447 has_multiple_baserels(PlannerInfo *root)
 448 {
 449         int                     num_base_rels = 0;
 450         Index           rti;
 451
 452         for (rti = 1; rti < root->simple_rel_array_size; rti++)
 453         {
 454                 RelOptInfo *brel = root->simple_rel_array[rti];
 455
 456                 if (brel == NULL)
 457                         continue;
 458
 459                 /* ignore RTEs that are "other rels" */
 460                 if (brel->reloptkind == RELOPT_BASEREL)
 461                         if (++num_base_rels > 1)
 462                                 return true;
 463         }
 464         return false;
 465 }
 466
 467 /*
 468  * set_subquery_pathlist
 469  *              Build the (single) access path for a subquery RTE
 470  */
 471 static void
 472 set_subquery_pathlist(PlannerInfo *root, RelOptInfo *rel,
 473                                           Index rti, RangeTblEntry *rte)
 474 {
 475         Query      *parse = root->parse;
 476         Query      *subquery = rte->subquery;
 477         bool       *differentTypes;
 478         double          tuple_fraction;
 479         PlannerInfo *subroot;
 480         List       *pathkeys;
 481
 482         /* We need a workspace for keeping track of set-op type coercions */
 483         differentTypes = (bool *)
 484                 palloc0((list_length(subquery->targetList) + 1) * sizeof(bool));
 485
 486         /*
 487          * If there are any restriction clauses that have been attached to the
 488          * subquery relation, consider pushing them down to become WHERE or HAVING
 489          * quals of the subquery itself.  This transformation is useful because it
 490          * may allow us to generate a better plan for the subquery than evaluating
 491          * all the subquery output rows and then filtering them.
 492          *
 493          * There are several cases where we cannot push down clauses. Restrictions
 494          * involving the subquery are checked by subquery_is_pushdown_safe().
 495          * Restrictions on individual clauses are checked by
 496          * qual_is_pushdown_safe().  Also, we don't want to push down
 497          * pseudoconstant clauses; better to have the gating node above the
 498          * subquery.
 499          *
 500          * Non-pushed-down clauses will get evaluated as qpquals of the
 501          * SubqueryScan node.
 502          *
 503          * XXX Are there any cases where we want to make a policy decision not to
 504          * push down a pushable qual, because it'd result in a worse plan?
 505          */
 506         if (rel->baserestrictinfo != NIL &&
 507                 subquery_is_pushdown_safe(subquery, subquery, differentTypes))
 508         {
 509                 /* OK to consider pushing down individual quals */
 510                 List       *upperrestrictlist = NIL;
 511                 ListCell   *l;
 512
 513                 foreach(l, rel->baserestrictinfo)
 514                 {
 515                         RestrictInfo *rinfo = (RestrictInfo *) lfirst(l);
 516                         Node       *clause = (Node *) rinfo->clause;
 517
 518                         if (!rinfo->pseudoconstant &&
 519                                 qual_is_pushdown_safe(subquery, rti, clause, differentTypes))
 520                         {
 521                                 /* Push it down */
 522                                 subquery_push_qual(subquery, rte, rti, clause);
 523                         }
 524                         else
 525                         {
 526                                 /* Keep it in the upper query */
 527                                 upperrestrictlist = lappend(upperrestrictlist, rinfo);
 528                         }
 529                 }
 530                 rel->baserestrictinfo = upperrestrictlist;
 531         }
 532
 533         pfree(differentTypes);
 534
 535         /*
 536          * We can safely pass the outer tuple_fraction down to the subquery if the
 537          * outer level has no joining, aggregation, or sorting to do. Otherwise
 538          * we'd better tell the subquery to plan for full retrieval. (XXX This
 539          * could probably be made more intelligent ...)
 540          */
 541         if (parse->hasAggs ||
 542                 parse->groupClause ||
 543                 parse->havingQual ||
 544                 parse->distinctClause ||
 545                 parse->sortClause ||
 546                 has_multiple_baserels(root))
 547                 tuple_fraction = 0.0;   /* default case */
 548         else
 549                 tuple_fraction = root->tuple_fraction;
 550
 551         /* Generate the plan for the subquery */
 552         rel->subplan = subquery_planner(root->glob, subquery,
 553                                                                         root->query_level + 1,
 554                                                                         tuple_fraction,
 555                                                                         &subroot);
 556         rel->subrtable = subroot->parse->rtable;
 557
 558         /* Copy number of output rows from subplan */
 559         rel->tuples = rel->subplan->plan_rows;
 560
 561         /* Mark rel with estimated output rows, width, etc */
 562         set_baserel_size_estimates(root, rel);
 563
 564         /* Convert subquery pathkeys to outer representation */
 565         pathkeys = convert_subquery_pathkeys(root, rel, subroot->query_pathkeys);
 566
 567         /* Generate appropriate path */
 568         add_path(rel, create_subqueryscan_path(rel, pathkeys));
 569
 570         /* Select cheapest path (pretty easy in this case...) */
 571         set_cheapest(rel);
 572 }
 573
 574 /*
 575  * set_function_pathlist
 576  *              Build the (single) access path for a function RTE
 577  */
 578 static void
 579 set_function_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 580 {
 581         /* Mark rel with estimated output rows, width, etc */
 582         set_function_size_estimates(root, rel);
 583
 584         /* Generate appropriate path */
 585         add_path(rel, create_functionscan_path(root, rel));
 586
 587         /* Select cheapest path (pretty easy in this case...) */
 588         set_cheapest(rel);
 589 }
 590
 591 /*
 592  * set_values_pathlist
 593  *              Build the (single) access path for a VALUES RTE
 594  */
 595 static void
 596 set_values_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 597 {
 598         /* Mark rel with estimated output rows, width, etc */
 599         set_values_size_estimates(root, rel);
 600
 601         /* Generate appropriate path */
 602         add_path(rel, create_valuesscan_path(root, rel));
 603
 604         /* Select cheapest path (pretty easy in this case...) */
 605         set_cheapest(rel);
 606 }
 607
 608 /*
 609  * make_rel_from_joinlist
 610  *        Build access paths using a "joinlist" to guide the join path search.
 611  *
 612  * See comments for deconstruct_jointree() for definition of the joinlist
 613  * data structure.
 614  */
 615 static RelOptInfo *
 616 make_rel_from_joinlist(PlannerInfo *root, List *joinlist)
 617 {
 618         int                     levels_needed;
 619         List       *initial_rels;
 620         ListCell   *jl;
 621
 622         /*
 623          * Count the number of child joinlist nodes.  This is the depth of the
 624          * dynamic-programming algorithm we must employ to consider all ways of
 625          * joining the child nodes.
 626          */
 627         levels_needed = list_length(joinlist);
 628
 629         if (levels_needed <= 0)
 630                 return NULL;                    /* nothing to do? */
 631
 632         /*
 633          * Construct a list of rels corresponding to the child joinlist nodes.
 634          * This may contain both base rels and rels constructed according to
 635          * sub-joinlists.
 636          */
 637         initial_rels = NIL;
 638         foreach(jl, joinlist)
 639         {
 640                 Node       *jlnode = (Node *) lfirst(jl);
 641                 RelOptInfo *thisrel;
 642
 643                 if (IsA(jlnode, RangeTblRef))
 644                 {
 645                         int                     varno = ((RangeTblRef *) jlnode)->rtindex;
 646
 647                         thisrel = find_base_rel(root, varno);
 648                 }
 649                 else if (IsA(jlnode, List))
 650                 {
 651                         /* Recurse to handle subproblem */
 652                         thisrel = make_rel_from_joinlist(root, (List *) jlnode);
 653                 }
 654                 else
 655                 {
 656                         elog(ERROR, "unrecognized joinlist node type: %d",
 657                                  (int) nodeTag(jlnode));
 658                         thisrel = NULL;         /* keep compiler quiet */
 659                 }
 660
 661                 initial_rels = lappend(initial_rels, thisrel);
 662         }
 663
 664         if (levels_needed == 1)
 665         {
 666                 /*
 667                  * Single joinlist node, so we're done.
 668                  */
 669                 return (RelOptInfo *) linitial(initial_rels);
 670         }
 671         else
 672         {
 673                 /*
 674                  * Consider the different orders in which we could join the rels,
 675                  * using either GEQO or regular optimizer.
 676                  */
 677                 if (enable_geqo && levels_needed >= geqo_threshold)
 678                         return geqo(root, levels_needed, initial_rels);
 679                 else
 680                         return make_one_rel_by_joins(root, levels_needed, initial_rels);
 681         }
 682 }
 683
 684 /*
 685  * make_one_rel_by_joins
 686  *        Find all possible joinpaths for a query by successively finding ways
 687  *        to join component relations into join relations.
 688  *
 689  * 'levels_needed' is the number of iterations needed, ie, the number of
 690  *              independent jointree items in the query.  This is > 1.
 691  *
 692  * 'initial_rels' is a list of RelOptInfo nodes for each independent
 693  *              jointree item.  These are the components to be joined together.
 694  *
 695  * Returns the final level of join relations, i.e., the relation that is
 696  * the result of joining all the original relations together.
 697  */
 698 static RelOptInfo *
 699 make_one_rel_by_joins(PlannerInfo *root, int levels_needed, List *initial_rels)
 700 {
 701         List      **joinitems;
 702         int                     lev;
 703         RelOptInfo *rel;
 704
 705         /*
 706          * We employ a simple "dynamic programming" algorithm: we first find all
 707          * ways to build joins of two jointree items, then all ways to build joins
 708          * of three items (from two-item joins and single items), then four-item
 709          * joins, and so on until we have considered all ways to join all the
 710          * items into one rel.
 711          *
 712          * joinitems[j] is a list of all the j-item rels.  Initially we set
 713          * joinitems[1] to represent all the single-jointree-item relations.
 714          */
 715         joinitems = (List **) palloc0((levels_needed + 1) * sizeof(List *));
 716
 717         joinitems[1] = initial_rels;
 718
 719         for (lev = 2; lev <= levels_needed; lev++)
 720         {
 721                 ListCell   *x;
 722
 723                 /*
 724                  * Determine all possible pairs of relations to be joined at this
 725                  * level, and build paths for making each one from every available
 726                  * pair of lower-level relations.
 727                  */
 728                 joinitems[lev] = make_rels_by_joins(root, lev, joinitems);
 729
 730                 /*
 731                  * Do cleanup work on each just-processed rel.
 732                  */
 733                 foreach(x, joinitems[lev])
 734                 {
 735                         rel = (RelOptInfo *) lfirst(x);
 736
 737                         /* Find and save the cheapest paths for this rel */
 738                         set_cheapest(rel);
 739
 740 #ifdef OPTIMIZER_DEBUG
 741                         debug_print_rel(root, rel);
 742 #endif
 743                 }
 744         }
 745
 746         /*
 747          * We should have a single rel at the final level.
 748          */
 749         if (joinitems[levels_needed] == NIL)
 750                 elog(ERROR, "failed to build any %d-way joins", levels_needed);
 751         Assert(list_length(joinitems[levels_needed]) == 1);
 752
 753         rel = (RelOptInfo *) linitial(joinitems[levels_needed]);
 754
 755         return rel;
 756 }
 757
 758 /*****************************************************************************
 759  *                      PUSHING QUALS DOWN INTO SUBQUERIES
 760  *****************************************************************************/
 761
 762 /*
 763  * subquery_is_pushdown_safe - is a subquery safe for pushing down quals?
 764  *
 765  * subquery is the particular component query being checked.  topquery
 766  * is the top component of a set-operations tree (the same Query if no
 767  * set-op is involved).
 768  *
 769  * Conditions checked here:
 770  *
 771  * 1. If the subquery has a LIMIT clause, we must not push down any quals,
 772  * since that could change the set of rows returned.
 773  *
 774  * 2. If the subquery contains EXCEPT or EXCEPT ALL set ops we cannot push
 775  * quals into it, because that would change the results.
 776  *
 777  * 3. For subqueries using UNION/UNION ALL/INTERSECT/INTERSECT ALL, we can
 778  * push quals into each component query, but the quals can only reference
 779  * subquery columns that suffer no type coercions in the set operation.
 780  * Otherwise there are possible semantic gotchas.  So, we check the
 781  * component queries to see if any of them have different output types;
 782  * differentTypes[k] is set true if column k has different type in any
 783  * component.
 784  */
 785 static bool
 786 subquery_is_pushdown_safe(Query *subquery, Query *topquery,
 787                                                   bool *differentTypes)
 788 {
 789         SetOperationStmt *topop;
 790
 791         /* Check point 1 */
 792         if (subquery->limitOffset != NULL || subquery->limitCount != NULL)
 793                 return false;
 794
 795         /* Are we at top level, or looking at a setop component? */
 796         if (subquery == topquery)
 797         {
 798                 /* Top level, so check any component queries */
 799                 if (subquery->setOperations != NULL)
 800                         if (!recurse_pushdown_safe(subquery->setOperations, topquery,
 801                                                                            differentTypes))
 802                                 return false;
 803         }
 804         else
 805         {
 806                 /* Setop component must not have more components (too weird) */
 807                 if (subquery->setOperations != NULL)
 808                         return false;
 809                 /* Check whether setop component output types match top level */
 810                 topop = (SetOperationStmt *) topquery->setOperations;
 811                 Assert(topop && IsA(topop, SetOperationStmt));
 812                 compare_tlist_datatypes(subquery->targetList,
 813                                                                 topop->colTypes,
 814                                                                 differentTypes);
 815         }
 816         return true;
 817 }
 818
 819 /*
 820  * Helper routine to recurse through setOperations tree
 821  */
 822 static bool
 823 recurse_pushdown_safe(Node *setOp, Query *topquery,
 824                                           bool *differentTypes)
 825 {
 826         if (IsA(setOp, RangeTblRef))
 827         {
 828                 RangeTblRef *rtr = (RangeTblRef *) setOp;
 829                 RangeTblEntry *rte = rt_fetch(rtr->rtindex, topquery->rtable);
 830                 Query      *subquery = rte->subquery;
 831
 832                 Assert(subquery != NULL);
 833                 return subquery_is_pushdown_safe(subquery, topquery, differentTypes);
 834         }
 835         else if (IsA(setOp, SetOperationStmt))
 836         {
 837                 SetOperationStmt *op = (SetOperationStmt *) setOp;
 838
 839                 /* EXCEPT is no good */
 840                 if (op->op == SETOP_EXCEPT)
 841                         return false;
 842                 /* Else recurse */
 843                 if (!recurse_pushdown_safe(op->larg, topquery, differentTypes))
 844                         return false;
 845                 if (!recurse_pushdown_safe(op->rarg, topquery, differentTypes))
 846                         return false;
 847         }
 848         else
 849         {
 850                 elog(ERROR, "unrecognized node type: %d",
 851                          (int) nodeTag(setOp));
 852         }
 853         return true;
 854 }
 855
 856 /*
 857  * Compare tlist's datatypes against the list of set-operation result types.
 858  * For any items that are different, mark the appropriate element of
 859  * differentTypes[] to show that this column will have type conversions.
 860  *
 861  * We don't have to care about typmods here: the only allowed difference
 862  * between set-op input and output typmods is input is a specific typmod
 863  * and output is -1, and that does not require a coercion.
 864  */
 865 static void
 866 compare_tlist_datatypes(List *tlist, List *colTypes,
 867                                                 bool *differentTypes)
 868 {
 869         ListCell   *l;
 870         ListCell   *colType = list_head(colTypes);
 871
 872         foreach(l, tlist)
 873         {
 874                 TargetEntry *tle = (TargetEntry *) lfirst(l);
 875
 876                 if (tle->resjunk)
 877                         continue;                       /* ignore resjunk columns */
 878                 if (colType == NULL)
 879                         elog(ERROR, "wrong number of tlist entries");
 880                 if (exprType((Node *) tle->expr) != lfirst_oid(colType))
 881                         differentTypes[tle->resno] = true;
 882                 colType = lnext(colType);
 883         }
 884         if (colType != NULL)
 885                 elog(ERROR, "wrong number of tlist entries");
 886 }
 887
 888 /*
 889  * qual_is_pushdown_safe - is a particular qual safe to push down?
 890  *
 891  * qual is a restriction clause applying to the given subquery (whose RTE
 892  * has index rti in the parent query).
 893  *
 894  * Conditions checked here:
 895  *
 896  * 1. The qual must not contain any subselects (mainly because I'm not sure
 897  * it will work correctly: sublinks will already have been transformed into
 898  * subplans in the qual, but not in the subquery).
 899  *
 900  * 2. The qual must not refer to the whole-row output of the subquery
 901  * (since there is no easy way to name that within the subquery itself).
 902  *
 903  * 3. The qual must not refer to any subquery output columns that were
 904  * found to have inconsistent types across a set operation tree by
 905  * subquery_is_pushdown_safe().
 906  *
 907  * 4. If the subquery uses DISTINCT ON, we must not push down any quals that
 908  * refer to non-DISTINCT output columns, because that could change the set
 909  * of rows returned.  This condition is vacuous for DISTINCT, because then
 910  * there are no non-DISTINCT output columns, but unfortunately it's fairly
 911  * expensive to tell the difference between DISTINCT and DISTINCT ON in the
 912  * parsetree representation.  It's cheaper to just make sure all the Vars
 913  * in the qual refer to DISTINCT columns.
 914  *
 915  * 5. We must not push down any quals that refer to subselect outputs that
 916  * return sets, else we'd introduce functions-returning-sets into the
 917  * subquery's WHERE/HAVING quals.
 918  *
 919  * 6. We must not push down any quals that refer to subselect outputs that
 920  * contain volatile functions, for fear of introducing strange results due
 921  * to multiple evaluation of a volatile function.
 922  */
 923 static bool
 924 qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
 925                                           bool *differentTypes)
 926 {
 927         bool            safe = true;
 928         List       *vars;
 929         ListCell   *vl;
 930         Bitmapset  *tested = NULL;
 931
 932         /* Refuse subselects (point 1) */
 933         if (contain_subplans(qual))
 934                 return false;
 935
 936         /*
 937          * Examine all Vars used in clause; since it's a restriction clause, all
 938          * such Vars must refer to subselect output columns.
 939          */
 940         vars = pull_var_clause(qual, false);
 941         foreach(vl, vars)
 942         {
 943                 Var                *var = (Var *) lfirst(vl);
 944                 TargetEntry *tle;
 945
 946                 Assert(var->varno == rti);
 947
 948                 /* Check point 2 */
 949                 if (var->varattno == 0)
 950                 {
 951                         safe = false;
 952                         break;
 953                 }
 954
 955                 /*
 956                  * We use a bitmapset to avoid testing the same attno more than once.
 957                  * (NB: this only works because subquery outputs can't have negative
 958                  * attnos.)
 959                  */
 960                 if (bms_is_member(var->varattno, tested))
 961                         continue;
 962                 tested = bms_add_member(tested, var->varattno);
 963
 964                 /* Check point 3 */
 965                 if (differentTypes[var->varattno])
 966                 {
 967                         safe = false;
 968                         break;
 969                 }
 970
 971                 /* Must find the tlist element referenced by the Var */
 972                 tle = get_tle_by_resno(subquery->targetList, var->varattno);
 973                 Assert(tle != NULL);
 974                 Assert(!tle->resjunk);
 975
 976                 /* If subquery uses DISTINCT or DISTINCT ON, check point 4 */
 977                 if (subquery->distinctClause != NIL &&
 978                         !targetIsInSortList(tle, InvalidOid, subquery->distinctClause))
 979                 {
 980                         /* non-DISTINCT column, so fail */
 981                         safe = false;
 982                         break;
 983                 }
 984
 985                 /* Refuse functions returning sets (point 5) */
 986                 if (expression_returns_set((Node *) tle->expr))
 987                 {
 988                         safe = false;
 989                         break;
 990                 }
 991
 992                 /* Refuse volatile functions (point 6) */
 993                 if (contain_volatile_functions((Node *) tle->expr))
 994                 {
 995                         safe = false;
 996                         break;
 997                 }
 998         }
 999
1000         list_free(vars);
1001         bms_free(tested);
1002
1003         return safe;
1004 }
1005
1006 /*
1007  * subquery_push_qual - push down a qual that we have determined is safe
1008  */
1009 static void
1010 subquery_push_qual(Query *subquery, RangeTblEntry *rte, Index rti, Node *qual)
1011 {
1012         if (subquery->setOperations != NULL)
1013         {
1014                 /* Recurse to push it separately to each component query */
1015                 recurse_push_qual(subquery->setOperations, subquery,
1016                                                   rte, rti, qual);
1017         }
1018         else
1019         {
1020                 /*
1021                  * We need to replace Vars in the qual (which must refer to outputs of
1022                  * the subquery) with copies of the subquery's targetlist expressions.
1023                  * Note that at this point, any uplevel Vars in the qual should have
1024                  * been replaced with Params, so they need no work.
1025                  *
1026                  * This step also ensures that when we are pushing into a setop tree,
1027                  * each component query gets its own copy of the qual.
1028                  */
1029                 qual = ResolveNew(qual, rti, 0, rte,
1030                                                   subquery->targetList,
1031                                                   CMD_SELECT, 0);
1032
1033                 /*
1034                  * Now attach the qual to the proper place: normally WHERE, but if the
1035                  * subquery uses grouping or aggregation, put it in HAVING (since the
1036                  * qual really refers to the group-result rows).
1037                  */
1038                 if (subquery->hasAggs || subquery->groupClause || subquery->havingQual)
1039                         subquery->havingQual = make_and_qual(subquery->havingQual, qual);
1040                 else
1041                         subquery->jointree->quals =
1042                                 make_and_qual(subquery->jointree->quals, qual);
1043
1044                 /*
1045                  * We need not change the subquery's hasAggs or hasSublinks flags,
1046                  * since we can't be pushing down any aggregates that weren't there
1047                  * before, and we don't push down subselects at all.
1048                  */
1049         }
1050 }
1051
1052 /*
1053  * Helper routine to recurse through setOperations tree
1054  */
1055 static void
1056 recurse_push_qual(Node *setOp, Query *topquery,
1057                                   RangeTblEntry *rte, Index rti, Node *qual)
1058 {
1059         if (IsA(setOp, RangeTblRef))
1060         {
1061                 RangeTblRef *rtr = (RangeTblRef *) setOp;
1062                 RangeTblEntry *subrte = rt_fetch(rtr->rtindex, topquery->rtable);
1063                 Query      *subquery = subrte->subquery;
1064
1065                 Assert(subquery != NULL);
1066                 subquery_push_qual(subquery, rte, rti, qual);
1067         }
1068         else if (IsA(setOp, SetOperationStmt))
1069         {
1070                 SetOperationStmt *op = (SetOperationStmt *) setOp;
1071
1072                 recurse_push_qual(op->larg, topquery, rte, rti, qual);
1073                 recurse_push_qual(op->rarg, topquery, rte, rti, qual);
1074         }
1075         else
1076         {
1077                 elog(ERROR, "unrecognized node type: %d",
1078                          (int) nodeTag(setOp));
1079         }
1080 }
1081
1082 /*****************************************************************************
1083  *                      DEBUG SUPPORT
1084  *****************************************************************************/
1085
1086 #ifdef OPTIMIZER_DEBUG
1087
1088 static void
1089 print_relids(Relids relids)
1090 {
1091         Relids          tmprelids;
1092         int                     x;
1093         bool            first = true;
1094
1095         tmprelids = bms_copy(relids);
1096         while ((x = bms_first_member(tmprelids)) >= 0)
1097         {
1098                 if (!first)
1099                         printf(" ");
1100                 printf("%d", x);
1101                 first = false;
1102         }
1103         bms_free(tmprelids);
1104 }
1105
1106 static void
1107 print_restrictclauses(PlannerInfo *root, List *clauses)
1108 {
1109         ListCell   *l;
1110
1111         foreach(l, clauses)
1112         {
1113                 RestrictInfo *c = lfirst(l);
1114
1115                 print_expr((Node *) c->clause, root->parse->rtable);
1116                 if (lnext(l))
1117                         printf(", ");
1118         }
1119 }
1120
1121 static void
1122 print_path(PlannerInfo *root, Path *path, int indent)
1123 {
1124         const char *ptype;
1125         bool            join = false;
1126         Path       *subpath = NULL;
1127         int                     i;
1128
1129         switch (nodeTag(path))
1130         {
1131                 case T_Path:
1132                         ptype = "SeqScan";
1133                         break;
1134                 case T_IndexPath:
1135                         ptype = "IdxScan";
1136                         break;
1137                 case T_BitmapHeapPath:
1138                         ptype = "BitmapHeapScan";
1139                         break;
1140                 case T_BitmapAndPath:
1141                         ptype = "BitmapAndPath";
1142                         break;
1143                 case T_BitmapOrPath:
1144                         ptype = "BitmapOrPath";
1145                         break;
1146                 case T_TidPath:
1147                         ptype = "TidScan";
1148                         break;
1149                 case T_AppendPath:
1150                         ptype = "Append";
1151                         break;
1152                 case T_ResultPath:
1153                         ptype = "Result";
1154                         break;
1155                 case T_MaterialPath:
1156                         ptype = "Material";
1157                         subpath = ((MaterialPath *) path)->subpath;
1158                         break;
1159                 case T_UniquePath:
1160                         ptype = "Unique";
1161                         subpath = ((UniquePath *) path)->subpath;
1162                         break;
1163                 case T_NestPath:
1164                         ptype = "NestLoop";
1165                         join = true;
1166                         break;
1167                 case T_MergePath:
1168                         ptype = "MergeJoin";
1169                         join = true;
1170                         break;
1171                 case T_HashPath:
1172                         ptype = "HashJoin";
1173                         join = true;
1174                         break;
1175                 default:
1176                         ptype = "???Path";
1177                         break;
1178         }
1179
1180         for (i = 0; i < indent; i++)
1181                 printf("\t");
1182         printf("%s", ptype);
1183
1184         if (path->parent)
1185         {
1186                 printf("(");
1187                 print_relids(path->parent->relids);
1188                 printf(") rows=%.0f", path->parent->rows);
1189         }
1190         printf(" cost=%.2f..%.2f\n", path->startup_cost, path->total_cost);
1191
1192         if (path->pathkeys)
1193         {
1194                 for (i = 0; i < indent; i++)
1195                         printf("\t");
1196                 printf("  pathkeys: ");
1197                 print_pathkeys(path->pathkeys, root->parse->rtable);
1198         }
1199
1200         if (join)
1201         {
1202                 JoinPath   *jp = (JoinPath *) path;
1203
1204                 for (i = 0; i < indent; i++)
1205                         printf("\t");
1206                 printf("  clauses: ");
1207                 print_restrictclauses(root, jp->joinrestrictinfo);
1208                 printf("\n");
1209
1210                 if (IsA(path, MergePath))
1211                 {
1212                         MergePath  *mp = (MergePath *) path;
1213
1214                         if (mp->outersortkeys || mp->innersortkeys)
1215                         {
1216                                 for (i = 0; i < indent; i++)
1217                                         printf("\t");
1218                                 printf("  sortouter=%d sortinner=%d\n",
1219                                            ((mp->outersortkeys) ? 1 : 0),
1220                                            ((mp->innersortkeys) ? 1 : 0));
1221                         }
1222                 }
1223
1224                 print_path(root, jp->outerjoinpath, indent + 1);
1225                 print_path(root, jp->innerjoinpath, indent + 1);
1226         }
1227
1228         if (subpath)
1229                 print_path(root, subpath, indent + 1);
1230 }
1231
1232 void
1233 debug_print_rel(PlannerInfo *root, RelOptInfo *rel)
1234 {
1235         ListCell   *l;
1236
1237         printf("RELOPTINFO (");
1238         print_relids(rel->relids);
1239         printf("): rows=%.0f width=%d\n", rel->rows, rel->width);
1240
1241         if (rel->baserestrictinfo)
1242         {
1243                 printf("\tbaserestrictinfo: ");
1244                 print_restrictclauses(root, rel->baserestrictinfo);
1245                 printf("\n");
1246         }
1247
1248         if (rel->joininfo)
1249         {
1250                 printf("\tjoininfo: ");
1251                 print_restrictclauses(root, rel->joininfo);
1252                 printf("\n");
1253         }
1254
1255         printf("\tpath list:\n");
1256         foreach(l, rel->pathlist)
1257                 print_path(root, lfirst(l), 1);
1258         printf("\n\tcheapest startup path:\n");
1259         print_path(root, rel->cheapest_startup_path, 1);
1260         printf("\n\tcheapest total path:\n");
1261         print_path(root, rel->cheapest_total_path, 1);
1262         printf("\n");
1263         fflush(stdout);
1264 }
1265
1266 #endif   /* OPTIMIZER_DEBUG */