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) 1996-2008, 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 1.167 2008/01/01 19:45:50 momjian 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 /* Hook for plugins to replace standard_join_search() */
  41 join_search_hook_type join_search_hook = NULL;
  42
  43
  44 static void set_base_rel_pathlists(PlannerInfo *root);
  45 static void set_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
  46                                  Index rti, RangeTblEntry *rte);
  47 static void set_plain_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
  48                                            RangeTblEntry *rte);
  49 static void set_append_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
  50                                                 Index rti, RangeTblEntry *rte);
  51 static void set_dummy_rel_pathlist(RelOptInfo *rel);
  52 static void set_subquery_pathlist(PlannerInfo *root, RelOptInfo *rel,
  53                                           Index rti, RangeTblEntry *rte);
  54 static void set_function_pathlist(PlannerInfo *root, RelOptInfo *rel,
  55                                           RangeTblEntry *rte);
  56 static void set_values_pathlist(PlannerInfo *root, RelOptInfo *rel,
  57                                         RangeTblEntry *rte);
  58 static RelOptInfo *make_rel_from_joinlist(PlannerInfo *root, List *joinlist);
  59 static bool subquery_is_pushdown_safe(Query *subquery, Query *topquery,
  60                                                   bool *differentTypes);
  61 static bool recurse_pushdown_safe(Node *setOp, Query *topquery,
  62                                           bool *differentTypes);
  63 static void compare_tlist_datatypes(List *tlist, List *colTypes,
  64                                                 bool *differentTypes);
  65 static bool qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
  66                                           bool *differentTypes);
  67 static void subquery_push_qual(Query *subquery,
  68                                    RangeTblEntry *rte, Index rti, Node *qual);
  69 static void recurse_push_qual(Node *setOp, Query *topquery,
  70                                   RangeTblEntry *rte, Index rti, Node *qual);
  71
  72
  73 /*
  74  * make_one_rel
  75  *        Finds all possible access paths for executing a query, returning a
  76  *        single rel that represents the join of all base rels in the query.
  77  */
  78 RelOptInfo *
  79 make_one_rel(PlannerInfo *root, List *joinlist)
  80 {
  81         RelOptInfo *rel;
  82
  83         /*
  84          * Generate access paths for the base rels.
  85          */
  86         set_base_rel_pathlists(root);
  87
  88         /*
  89          * Generate access paths for the entire join tree.
  90          */
  91         rel = make_rel_from_joinlist(root, joinlist);
  92
  93         /*
  94          * The result should join all and only the query's base rels.
  95          */
  96 #ifdef USE_ASSERT_CHECKING
  97         {
  98                 int                     num_base_rels = 0;
  99                 Index           rti;
 100
 101                 for (rti = 1; rti < root->simple_rel_array_size; rti++)
 102                 {
 103                         RelOptInfo *brel = root->simple_rel_array[rti];
 104
 105                         if (brel == NULL)
 106                                 continue;
 107
 108                         Assert(brel->relid == rti); /* sanity check on array */
 109
 110                         /* ignore RTEs that are "other rels" */
 111                         if (brel->reloptkind != RELOPT_BASEREL)
 112                                 continue;
 113
 114                         Assert(bms_is_member(rti, rel->relids));
 115                         num_base_rels++;
 116                 }
 117
 118                 Assert(bms_num_members(rel->relids) == num_base_rels);
 119         }
 120 #endif
 121
 122         return rel;
 123 }
 124
 125 /*
 126  * set_base_rel_pathlists
 127  *        Finds all paths available for scanning each base-relation entry.
 128  *        Sequential scan and any available indices are considered.
 129  *        Each useful path is attached to its relation's 'pathlist' field.
 130  */
 131 static void
 132 set_base_rel_pathlists(PlannerInfo *root)
 133 {
 134         Index           rti;
 135
 136         for (rti = 1; rti < root->simple_rel_array_size; rti++)
 137         {
 138                 RelOptInfo *rel = root->simple_rel_array[rti];
 139
 140                 /* there may be empty slots corresponding to non-baserel RTEs */
 141                 if (rel == NULL)
 142                         continue;
 143
 144                 Assert(rel->relid == rti);              /* sanity check on array */
 145
 146                 /* ignore RTEs that are "other rels" */
 147                 if (rel->reloptkind != RELOPT_BASEREL)
 148                         continue;
 149
 150                 set_rel_pathlist(root, rel, rti, root->simple_rte_array[rti]);
 151         }
 152 }
 153
 154 /*
 155  * set_rel_pathlist
 156  *        Build access paths for a base relation
 157  */
 158 static void
 159 set_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
 160                                  Index rti, RangeTblEntry *rte)
 161 {
 162         if (rte->inh)
 163         {
 164                 /* It's an "append relation", process accordingly */
 165                 set_append_rel_pathlist(root, rel, rti, rte);
 166         }
 167         else if (rel->rtekind == RTE_SUBQUERY)
 168         {
 169                 /* Subquery --- generate a separate plan for it */
 170                 set_subquery_pathlist(root, rel, rti, rte);
 171         }
 172         else if (rel->rtekind == RTE_FUNCTION)
 173         {
 174                 /* RangeFunction --- generate a separate plan for it */
 175                 set_function_pathlist(root, rel, rte);
 176         }
 177         else if (rel->rtekind == RTE_VALUES)
 178         {
 179                 /* Values list --- generate a separate plan for it */
 180                 set_values_pathlist(root, rel, rte);
 181         }
 182         else
 183         {
 184                 /* Plain relation */
 185                 Assert(rel->rtekind == RTE_RELATION);
 186                 set_plain_rel_pathlist(root, rel, rte);
 187         }
 188
 189 #ifdef OPTIMIZER_DEBUG
 190         debug_print_rel(root, rel);
 191 #endif
 192 }
 193
 194 /*
 195  * set_plain_rel_pathlist
 196  *        Build access paths for a plain relation (no subquery, no inheritance)
 197  */
 198 static void
 199 set_plain_rel_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 200 {
 201         /*
 202          * If we can prove we don't need to scan the rel via constraint exclusion,
 203          * set up a single dummy path for it.  We only need to check for regular
 204          * baserels; if it's an otherrel, CE was already checked in
 205          * set_append_rel_pathlist().
 206          */
 207         if (rel->reloptkind == RELOPT_BASEREL &&
 208                 relation_excluded_by_constraints(rel, rte))
 209         {
 210                 set_dummy_rel_pathlist(rel);
 211                 return;
 212         }
 213
 214         /* Mark rel with estimated output rows, width, etc */
 215         set_baserel_size_estimates(root, rel);
 216
 217         /* Test any partial indexes of rel for applicability */
 218         check_partial_indexes(root, rel);
 219
 220         /*
 221          * Check to see if we can extract any restriction conditions from join
 222          * quals that are OR-of-AND structures.  If so, add them to the rel's
 223          * restriction list, and recompute the size estimates.
 224          */
 225         if (create_or_index_quals(root, rel))
 226                 set_baserel_size_estimates(root, rel);
 227
 228         /*
 229          * Generate paths and add them to the rel's pathlist.
 230          *
 231          * Note: add_path() will discard any paths that are dominated by another
 232          * available path, keeping only those paths that are superior along at
 233          * least one dimension of cost or sortedness.
 234          */
 235
 236         /* Consider sequential scan */
 237         add_path(rel, create_seqscan_path(root, rel));
 238
 239         /* Consider index scans */
 240         create_index_paths(root, rel);
 241
 242         /* Consider TID scans */
 243         create_tidscan_paths(root, rel);
 244
 245         /* Now find the cheapest of the paths for this rel */
 246         set_cheapest(rel);
 247 }
 248
 249 /*
 250  * set_append_rel_pathlist
 251  *        Build access paths for an "append relation"
 252  *
 253  * The passed-in rel and RTE represent the entire append relation.      The
 254  * relation's contents are computed by appending together the output of
 255  * the individual member relations.  Note that in the inheritance case,
 256  * the first member relation is actually the same table as is mentioned in
 257  * the parent RTE ... but it has a different RTE and RelOptInfo.  This is
 258  * a good thing because their outputs are not the same size.
 259  */
 260 static void
 261 set_append_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
 262                                                 Index rti, RangeTblEntry *rte)
 263 {
 264         int                     parentRTindex = rti;
 265         List       *subpaths = NIL;
 266         ListCell   *l;
 267
 268         /*
 269          * XXX for now, can't handle inherited expansion of FOR UPDATE/SHARE; can
 270          * we do better?  (This will take some redesign because the executor
 271          * currently supposes that every rowMark relation is involved in every row
 272          * returned by the query.)
 273          */
 274         if (get_rowmark(root->parse, parentRTindex))
 275                 ereport(ERROR,
 276                                 (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
 277                                  errmsg("SELECT FOR UPDATE/SHARE is not supported for inheritance queries")));
 278
 279         /*
 280          * Initialize to compute size estimates for whole append relation
 281          */
 282         rel->rows = 0;
 283         rel->width = 0;
 284
 285         /*
 286          * Generate access paths for each member relation, and pick the cheapest
 287          * path for each one.
 288          */
 289         foreach(l, root->append_rel_list)
 290         {
 291                 AppendRelInfo *appinfo = (AppendRelInfo *) lfirst(l);
 292                 int                     childRTindex;
 293                 RangeTblEntry *childRTE;
 294                 RelOptInfo *childrel;
 295                 Path       *childpath;
 296                 ListCell   *parentvars;
 297                 ListCell   *childvars;
 298
 299                 /* append_rel_list contains all append rels; ignore others */
 300                 if (appinfo->parent_relid != parentRTindex)
 301                         continue;
 302
 303                 childRTindex = appinfo->child_relid;
 304                 childRTE = root->simple_rte_array[childRTindex];
 305
 306                 /*
 307                  * The child rel's RelOptInfo was already created during
 308                  * add_base_rels_to_query.
 309                  */
 310                 childrel = find_base_rel(root, childRTindex);
 311                 Assert(childrel->reloptkind == RELOPT_OTHER_MEMBER_REL);
 312
 313                 /*
 314                  * We have to copy the parent's targetlist and quals to the child,
 315                  * with appropriate substitution of variables.  However, only the
 316                  * baserestrictinfo quals are needed before we can check for
 317                  * constraint exclusion; so do that first and then check to see if we
 318                  * can disregard this child.
 319                  */
 320                 childrel->baserestrictinfo = (List *)
 321                         adjust_appendrel_attrs((Node *) rel->baserestrictinfo,
 322                                                                    appinfo);
 323
 324                 if (relation_excluded_by_constraints(childrel, childRTE))
 325                 {
 326                         /*
 327                          * This child need not be scanned, so we can omit it from the
 328                          * appendrel.  Mark it with a dummy cheapest-path though, in case
 329                          * best_appendrel_indexscan() looks at it later.
 330                          */
 331                         set_dummy_rel_pathlist(childrel);
 332                         continue;
 333                 }
 334
 335                 /* CE failed, so finish copying targetlist and join quals */
 336                 childrel->joininfo = (List *)
 337                         adjust_appendrel_attrs((Node *) rel->joininfo,
 338                                                                    appinfo);
 339                 childrel->reltargetlist = (List *)
 340                         adjust_appendrel_attrs((Node *) rel->reltargetlist,
 341                                                                    appinfo);
 342
 343                 /*
 344                  * We have to make child entries in the EquivalenceClass data
 345                  * structures as well.
 346                  */
 347                 if (rel->has_eclass_joins)
 348                 {
 349                         add_child_rel_equivalences(root, appinfo, rel, childrel);
 350                         childrel->has_eclass_joins = true;
 351                 }
 352
 353                 /*
 354                  * Copy the parent's attr_needed data as well, with appropriate
 355                  * adjustment of relids and attribute numbers.
 356                  */
 357                 pfree(childrel->attr_needed);
 358                 childrel->attr_needed =
 359                         adjust_appendrel_attr_needed(rel, appinfo,
 360                                                                                  childrel->min_attr,
 361                                                                                  childrel->max_attr);
 362
 363                 /*
 364                  * Compute the child's access paths, and add the cheapest one to the
 365                  * Append path we are constructing for the parent.
 366                  *
 367                  * It's possible that the child is itself an appendrel, in which case
 368                  * we can "cut out the middleman" and just add its child paths to our
 369                  * own list.  (We don't try to do this earlier because we need to
 370                  * apply both levels of transformation to the quals.)
 371                  */
 372                 set_rel_pathlist(root, childrel, childRTindex, childRTE);
 373
 374                 childpath = childrel->cheapest_total_path;
 375                 if (IsA(childpath, AppendPath))
 376                         subpaths = list_concat(subpaths,
 377                                                                    ((AppendPath *) childpath)->subpaths);
 378                 else
 379                         subpaths = lappend(subpaths, childpath);
 380
 381                 /*
 382                  * Propagate size information from the child back to the parent. For
 383                  * simplicity, we use the largest widths from any child as the parent
 384                  * estimates.  (If you want to change this, beware of child
 385                  * attr_widths[] entries that haven't been set and are still 0.)
 386                  */
 387                 rel->rows += childrel->rows;
 388                 if (childrel->width > rel->width)
 389                         rel->width = childrel->width;
 390
 391                 forboth(parentvars, rel->reltargetlist,
 392                                 childvars, childrel->reltargetlist)
 393                 {
 394                         Var                *parentvar = (Var *) lfirst(parentvars);
 395                         Var                *childvar = (Var *) lfirst(childvars);
 396
 397                         if (IsA(parentvar, Var) &&
 398                                 IsA(childvar, Var))
 399                         {
 400                                 int                     pndx = parentvar->varattno - rel->min_attr;
 401                                 int                     cndx = childvar->varattno - childrel->min_attr;
 402
 403                                 if (childrel->attr_widths[cndx] > rel->attr_widths[pndx])
 404                                         rel->attr_widths[pndx] = childrel->attr_widths[cndx];
 405                         }
 406                 }
 407         }
 408
 409         /*
 410          * Set "raw tuples" count equal to "rows" for the appendrel; needed
 411          * because some places assume rel->tuples is valid for any baserel.
 412          */
 413         rel->tuples = rel->rows;
 414
 415         /*
 416          * Finally, build Append path and install it as the only access path for
 417          * the parent rel.      (Note: this is correct even if we have zero or one
 418          * live subpath due to constraint exclusion.)
 419          */
 420         add_path(rel, (Path *) create_append_path(rel, subpaths));
 421
 422         /* Select cheapest path (pretty easy in this case...) */
 423         set_cheapest(rel);
 424 }
 425
 426 /*
 427  * set_dummy_rel_pathlist
 428  *        Build a dummy path for a relation that's been excluded by constraints
 429  *
 430  * Rather than inventing a special "dummy" path type, we represent this as an
 431  * AppendPath with no members.
 432  */
 433 static void
 434 set_dummy_rel_pathlist(RelOptInfo *rel)
 435 {
 436         /* Set dummy size estimates --- we leave attr_widths[] as zeroes */
 437         rel->rows = 0;
 438         rel->width = 0;
 439
 440         add_path(rel, (Path *) create_append_path(rel, NIL));
 441
 442         /* Select cheapest path (pretty easy in this case...) */
 443         set_cheapest(rel);
 444 }
 445
 446 /* quick-and-dirty test to see if any joining is needed */
 447 static bool
 448 has_multiple_baserels(PlannerInfo *root)
 449 {
 450         int                     num_base_rels = 0;
 451         Index           rti;
 452
 453         for (rti = 1; rti < root->simple_rel_array_size; rti++)
 454         {
 455                 RelOptInfo *brel = root->simple_rel_array[rti];
 456
 457                 if (brel == NULL)
 458                         continue;
 459
 460                 /* ignore RTEs that are "other rels" */
 461                 if (brel->reloptkind == RELOPT_BASEREL)
 462                         if (++num_base_rels > 1)
 463                                 return true;
 464         }
 465         return false;
 466 }
 467
 468 /*
 469  * set_subquery_pathlist
 470  *              Build the (single) access path for a subquery RTE
 471  */
 472 static void
 473 set_subquery_pathlist(PlannerInfo *root, RelOptInfo *rel,
 474                                           Index rti, RangeTblEntry *rte)
 475 {
 476         Query      *parse = root->parse;
 477         Query      *subquery = rte->subquery;
 478         bool       *differentTypes;
 479         double          tuple_fraction;
 480         PlannerInfo *subroot;
 481         List       *pathkeys;
 482
 483         /* We need a workspace for keeping track of set-op type coercions */
 484         differentTypes = (bool *)
 485                 palloc0((list_length(subquery->targetList) + 1) * sizeof(bool));
 486
 487         /*
 488          * If there are any restriction clauses that have been attached to the
 489          * subquery relation, consider pushing them down to become WHERE or HAVING
 490          * quals of the subquery itself.  This transformation is useful because it
 491          * may allow us to generate a better plan for the subquery than evaluating
 492          * all the subquery output rows and then filtering them.
 493          *
 494          * There are several cases where we cannot push down clauses. Restrictions
 495          * involving the subquery are checked by subquery_is_pushdown_safe().
 496          * Restrictions on individual clauses are checked by
 497          * qual_is_pushdown_safe().  Also, we don't want to push down
 498          * pseudoconstant clauses; better to have the gating node above the
 499          * subquery.
 500          *
 501          * Non-pushed-down clauses will get evaluated as qpquals of the
 502          * SubqueryScan node.
 503          *
 504          * XXX Are there any cases where we want to make a policy decision not to
 505          * push down a pushable qual, because it'd result in a worse plan?
 506          */
 507         if (rel->baserestrictinfo != NIL &&
 508                 subquery_is_pushdown_safe(subquery, subquery, differentTypes))
 509         {
 510                 /* OK to consider pushing down individual quals */
 511                 List       *upperrestrictlist = NIL;
 512                 ListCell   *l;
 513
 514                 foreach(l, rel->baserestrictinfo)
 515                 {
 516                         RestrictInfo *rinfo = (RestrictInfo *) lfirst(l);
 517                         Node       *clause = (Node *) rinfo->clause;
 518
 519                         if (!rinfo->pseudoconstant &&
 520                                 qual_is_pushdown_safe(subquery, rti, clause, differentTypes))
 521                         {
 522                                 /* Push it down */
 523                                 subquery_push_qual(subquery, rte, rti, clause);
 524                         }
 525                         else
 526                         {
 527                                 /* Keep it in the upper query */
 528                                 upperrestrictlist = lappend(upperrestrictlist, rinfo);
 529                         }
 530                 }
 531                 rel->baserestrictinfo = upperrestrictlist;
 532         }
 533
 534         pfree(differentTypes);
 535
 536         /*
 537          * We can safely pass the outer tuple_fraction down to the subquery if the
 538          * outer level has no joining, aggregation, or sorting to do. Otherwise
 539          * we'd better tell the subquery to plan for full retrieval. (XXX This
 540          * could probably be made more intelligent ...)
 541          */
 542         if (parse->hasAggs ||
 543                 parse->groupClause ||
 544                 parse->havingQual ||
 545                 parse->distinctClause ||
 546                 parse->sortClause ||
 547                 has_multiple_baserels(root))
 548                 tuple_fraction = 0.0;   /* default case */
 549         else
 550                 tuple_fraction = root->tuple_fraction;
 551
 552         /* Generate the plan for the subquery */
 553         rel->subplan = subquery_planner(root->glob, subquery,
 554                                                                         root->query_level + 1,
 555                                                                         tuple_fraction,
 556                                                                         &subroot);
 557         rel->subrtable = subroot->parse->rtable;
 558
 559         /* Copy number of output rows from subplan */
 560         rel->tuples = rel->subplan->plan_rows;
 561
 562         /* Mark rel with estimated output rows, width, etc */
 563         set_baserel_size_estimates(root, rel);
 564
 565         /* Convert subquery pathkeys to outer representation */
 566         pathkeys = convert_subquery_pathkeys(root, rel, subroot->query_pathkeys);
 567
 568         /* Generate appropriate path */
 569         add_path(rel, create_subqueryscan_path(rel, pathkeys));
 570
 571         /* Select cheapest path (pretty easy in this case...) */
 572         set_cheapest(rel);
 573 }
 574
 575 /*
 576  * set_function_pathlist
 577  *              Build the (single) access path for a function RTE
 578  */
 579 static void
 580 set_function_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 581 {
 582         /* Mark rel with estimated output rows, width, etc */
 583         set_function_size_estimates(root, rel);
 584
 585         /* Generate appropriate path */
 586         add_path(rel, create_functionscan_path(root, rel));
 587
 588         /* Select cheapest path (pretty easy in this case...) */
 589         set_cheapest(rel);
 590 }
 591
 592 /*
 593  * set_values_pathlist
 594  *              Build the (single) access path for a VALUES RTE
 595  */
 596 static void
 597 set_values_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 598 {
 599         /* Mark rel with estimated output rows, width, etc */
 600         set_values_size_estimates(root, rel);
 601
 602         /* Generate appropriate path */
 603         add_path(rel, create_valuesscan_path(root, rel));
 604
 605         /* Select cheapest path (pretty easy in this case...) */
 606         set_cheapest(rel);
 607 }
 608
 609 /*
 610  * make_rel_from_joinlist
 611  *        Build access paths using a "joinlist" to guide the join path search.
 612  *
 613  * See comments for deconstruct_jointree() for definition of the joinlist
 614  * data structure.
 615  */
 616 static RelOptInfo *
 617 make_rel_from_joinlist(PlannerInfo *root, List *joinlist)
 618 {
 619         int                     levels_needed;
 620         List       *initial_rels;
 621         ListCell   *jl;
 622
 623         /*
 624          * Count the number of child joinlist nodes.  This is the depth of the
 625          * dynamic-programming algorithm we must employ to consider all ways of
 626          * joining the child nodes.
 627          */
 628         levels_needed = list_length(joinlist);
 629
 630         if (levels_needed <= 0)
 631                 return NULL;                    /* nothing to do? */
 632
 633         /*
 634          * Construct a list of rels corresponding to the child joinlist nodes.
 635          * This may contain both base rels and rels constructed according to
 636          * sub-joinlists.
 637          */
 638         initial_rels = NIL;
 639         foreach(jl, joinlist)
 640         {
 641                 Node       *jlnode = (Node *) lfirst(jl);
 642                 RelOptInfo *thisrel;
 643
 644                 if (IsA(jlnode, RangeTblRef))
 645                 {
 646                         int                     varno = ((RangeTblRef *) jlnode)->rtindex;
 647
 648                         thisrel = find_base_rel(root, varno);
 649                 }
 650                 else if (IsA(jlnode, List))
 651                 {
 652                         /* Recurse to handle subproblem */
 653                         thisrel = make_rel_from_joinlist(root, (List *) jlnode);
 654                 }
 655                 else
 656                 {
 657                         elog(ERROR, "unrecognized joinlist node type: %d",
 658                                  (int) nodeTag(jlnode));
 659                         thisrel = NULL;         /* keep compiler quiet */
 660                 }
 661
 662                 initial_rels = lappend(initial_rels, thisrel);
 663         }
 664
 665         if (levels_needed == 1)
 666         {
 667                 /*
 668                  * Single joinlist node, so we're done.
 669                  */
 670                 return (RelOptInfo *) linitial(initial_rels);
 671         }
 672         else
 673         {
 674                 /*
 675                  * Consider the different orders in which we could join the rels,
 676                  * using a plugin, GEQO, or the regular join search code.
 677                  */
 678                 if (join_search_hook)
 679                         return (*join_search_hook) (root, levels_needed, initial_rels);
 680                 else if (enable_geqo && levels_needed >= geqo_threshold)
 681                         return geqo(root, levels_needed, initial_rels);
 682                 else
 683                         return standard_join_search(root, levels_needed, initial_rels);
 684         }
 685 }
 686
 687 /*
 688  * standard_join_search
 689  *        Find possible joinpaths for a query by successively finding ways
 690  *        to join component relations into join relations.
 691  *
 692  * 'levels_needed' is the number of iterations needed, ie, the number of
 693  *              independent jointree items in the query.  This is > 1.
 694  *
 695  * 'initial_rels' is a list of RelOptInfo nodes for each independent
 696  *              jointree item.  These are the components to be joined together.
 697  *              Note that levels_needed == list_length(initial_rels).
 698  *
 699  * Returns the final level of join relations, i.e., the relation that is
 700  * the result of joining all the original relations together.
 701  * At least one implementation path must be provided for this relation and
 702  * all required sub-relations.
 703  *
 704  * To support loadable plugins that modify planner behavior by changing the
 705  * join searching algorithm, we provide a hook variable that lets a plugin
 706  * replace or supplement this function.  Any such hook must return the same
 707  * final join relation as the standard code would, but it might have a
 708  * different set of implementation paths attached, and only the sub-joinrels
 709  * needed for these paths need have been instantiated.
 710  *
 711  * Note to plugin authors: the functions invoked during standard_join_search()
 712  * modify root->join_rel_list and root->join_rel_hash.  If you want to do more
 713  * than one join-order search, you'll probably need to save and restore the
 714  * original states of those data structures.  See geqo_eval() for an example.
 715  */
 716 RelOptInfo *
 717 standard_join_search(PlannerInfo *root, int levels_needed, List *initial_rels)
 718 {
 719         List      **joinitems;
 720         int                     lev;
 721         RelOptInfo *rel;
 722
 723         /*
 724          * We employ a simple "dynamic programming" algorithm: we first find all
 725          * ways to build joins of two jointree items, then all ways to build joins
 726          * of three items (from two-item joins and single items), then four-item
 727          * joins, and so on until we have considered all ways to join all the
 728          * items into one rel.
 729          *
 730          * joinitems[j] is a list of all the j-item rels.  Initially we set
 731          * joinitems[1] to represent all the single-jointree-item relations.
 732          */
 733         joinitems = (List **) palloc0((levels_needed + 1) * sizeof(List *));
 734
 735         joinitems[1] = initial_rels;
 736
 737         for (lev = 2; lev <= levels_needed; lev++)
 738         {
 739                 ListCell   *x;
 740
 741                 /*
 742                  * Determine all possible pairs of relations to be joined at this
 743                  * level, and build paths for making each one from every available
 744                  * pair of lower-level relations.
 745                  */
 746                 joinitems[lev] = join_search_one_level(root, lev, joinitems);
 747
 748                 /*
 749                  * Do cleanup work on each just-processed rel.
 750                  */
 751                 foreach(x, joinitems[lev])
 752                 {
 753                         rel = (RelOptInfo *) lfirst(x);
 754
 755                         /* Find and save the cheapest paths for this rel */
 756                         set_cheapest(rel);
 757
 758 #ifdef OPTIMIZER_DEBUG
 759                         debug_print_rel(root, rel);
 760 #endif
 761                 }
 762         }
 763
 764         /*
 765          * We should have a single rel at the final level.
 766          */
 767         if (joinitems[levels_needed] == NIL)
 768                 elog(ERROR, "failed to build any %d-way joins", levels_needed);
 769         Assert(list_length(joinitems[levels_needed]) == 1);
 770
 771         rel = (RelOptInfo *) linitial(joinitems[levels_needed]);
 772
 773         return rel;
 774 }
 775
 776 /*****************************************************************************
 777  *                      PUSHING QUALS DOWN INTO SUBQUERIES
 778  *****************************************************************************/
 779
 780 /*
 781  * subquery_is_pushdown_safe - is a subquery safe for pushing down quals?
 782  *
 783  * subquery is the particular component query being checked.  topquery
 784  * is the top component of a set-operations tree (the same Query if no
 785  * set-op is involved).
 786  *
 787  * Conditions checked here:
 788  *
 789  * 1. If the subquery has a LIMIT clause, we must not push down any quals,
 790  * since that could change the set of rows returned.
 791  *
 792  * 2. If the subquery contains EXCEPT or EXCEPT ALL set ops we cannot push
 793  * quals into it, because that would change the results.
 794  *
 795  * 3. For subqueries using UNION/UNION ALL/INTERSECT/INTERSECT ALL, we can
 796  * push quals into each component query, but the quals can only reference
 797  * subquery columns that suffer no type coercions in the set operation.
 798  * Otherwise there are possible semantic gotchas.  So, we check the
 799  * component queries to see if any of them have different output types;
 800  * differentTypes[k] is set true if column k has different type in any
 801  * component.
 802  */
 803 static bool
 804 subquery_is_pushdown_safe(Query *subquery, Query *topquery,
 805                                                   bool *differentTypes)
 806 {
 807         SetOperationStmt *topop;
 808
 809         /* Check point 1 */
 810         if (subquery->limitOffset != NULL || subquery->limitCount != NULL)
 811                 return false;
 812
 813         /* Are we at top level, or looking at a setop component? */
 814         if (subquery == topquery)
 815         {
 816                 /* Top level, so check any component queries */
 817                 if (subquery->setOperations != NULL)
 818                         if (!recurse_pushdown_safe(subquery->setOperations, topquery,
 819                                                                            differentTypes))
 820                                 return false;
 821         }
 822         else
 823         {
 824                 /* Setop component must not have more components (too weird) */
 825                 if (subquery->setOperations != NULL)
 826                         return false;
 827                 /* Check whether setop component output types match top level */
 828                 topop = (SetOperationStmt *) topquery->setOperations;
 829                 Assert(topop && IsA(topop, SetOperationStmt));
 830                 compare_tlist_datatypes(subquery->targetList,
 831                                                                 topop->colTypes,
 832                                                                 differentTypes);
 833         }
 834         return true;
 835 }
 836
 837 /*
 838  * Helper routine to recurse through setOperations tree
 839  */
 840 static bool
 841 recurse_pushdown_safe(Node *setOp, Query *topquery,
 842                                           bool *differentTypes)
 843 {
 844         if (IsA(setOp, RangeTblRef))
 845         {
 846                 RangeTblRef *rtr = (RangeTblRef *) setOp;
 847                 RangeTblEntry *rte = rt_fetch(rtr->rtindex, topquery->rtable);
 848                 Query      *subquery = rte->subquery;
 849
 850                 Assert(subquery != NULL);
 851                 return subquery_is_pushdown_safe(subquery, topquery, differentTypes);
 852         }
 853         else if (IsA(setOp, SetOperationStmt))
 854         {
 855                 SetOperationStmt *op = (SetOperationStmt *) setOp;
 856
 857                 /* EXCEPT is no good */
 858                 if (op->op == SETOP_EXCEPT)
 859                         return false;
 860                 /* Else recurse */
 861                 if (!recurse_pushdown_safe(op->larg, topquery, differentTypes))
 862                         return false;
 863                 if (!recurse_pushdown_safe(op->rarg, topquery, differentTypes))
 864                         return false;
 865         }
 866         else
 867         {
 868                 elog(ERROR, "unrecognized node type: %d",
 869                          (int) nodeTag(setOp));
 870         }
 871         return true;
 872 }
 873
 874 /*
 875  * Compare tlist's datatypes against the list of set-operation result types.
 876  * For any items that are different, mark the appropriate element of
 877  * differentTypes[] to show that this column will have type conversions.
 878  *
 879  * We don't have to care about typmods here: the only allowed difference
 880  * between set-op input and output typmods is input is a specific typmod
 881  * and output is -1, and that does not require a coercion.
 882  */
 883 static void
 884 compare_tlist_datatypes(List *tlist, List *colTypes,
 885                                                 bool *differentTypes)
 886 {
 887         ListCell   *l;
 888         ListCell   *colType = list_head(colTypes);
 889
 890         foreach(l, tlist)
 891         {
 892                 TargetEntry *tle = (TargetEntry *) lfirst(l);
 893
 894                 if (tle->resjunk)
 895                         continue;                       /* ignore resjunk columns */
 896                 if (colType == NULL)
 897                         elog(ERROR, "wrong number of tlist entries");
 898                 if (exprType((Node *) tle->expr) != lfirst_oid(colType))
 899                         differentTypes[tle->resno] = true;
 900                 colType = lnext(colType);
 901         }
 902         if (colType != NULL)
 903                 elog(ERROR, "wrong number of tlist entries");
 904 }
 905
 906 /*
 907  * qual_is_pushdown_safe - is a particular qual safe to push down?
 908  *
 909  * qual is a restriction clause applying to the given subquery (whose RTE
 910  * has index rti in the parent query).
 911  *
 912  * Conditions checked here:
 913  *
 914  * 1. The qual must not contain any subselects (mainly because I'm not sure
 915  * it will work correctly: sublinks will already have been transformed into
 916  * subplans in the qual, but not in the subquery).
 917  *
 918  * 2. The qual must not refer to the whole-row output of the subquery
 919  * (since there is no easy way to name that within the subquery itself).
 920  *
 921  * 3. The qual must not refer to any subquery output columns that were
 922  * found to have inconsistent types across a set operation tree by
 923  * subquery_is_pushdown_safe().
 924  *
 925  * 4. If the subquery uses DISTINCT ON, we must not push down any quals that
 926  * refer to non-DISTINCT output columns, because that could change the set
 927  * of rows returned.  This condition is vacuous for DISTINCT, because then
 928  * there are no non-DISTINCT output columns, but unfortunately it's fairly
 929  * expensive to tell the difference between DISTINCT and DISTINCT ON in the
 930  * parsetree representation.  It's cheaper to just make sure all the Vars
 931  * in the qual refer to DISTINCT columns.
 932  *
 933  * 5. We must not push down any quals that refer to subselect outputs that
 934  * return sets, else we'd introduce functions-returning-sets into the
 935  * subquery's WHERE/HAVING quals.
 936  *
 937  * 6. We must not push down any quals that refer to subselect outputs that
 938  * contain volatile functions, for fear of introducing strange results due
 939  * to multiple evaluation of a volatile function.
 940  */
 941 static bool
 942 qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
 943                                           bool *differentTypes)
 944 {
 945         bool            safe = true;
 946         List       *vars;
 947         ListCell   *vl;
 948         Bitmapset  *tested = NULL;
 949
 950         /* Refuse subselects (point 1) */
 951         if (contain_subplans(qual))
 952                 return false;
 953
 954         /*
 955          * Examine all Vars used in clause; since it's a restriction clause, all
 956          * such Vars must refer to subselect output columns.
 957          */
 958         vars = pull_var_clause(qual, false);
 959         foreach(vl, vars)
 960         {
 961                 Var                *var = (Var *) lfirst(vl);
 962                 TargetEntry *tle;
 963
 964                 Assert(var->varno == rti);
 965
 966                 /* Check point 2 */
 967                 if (var->varattno == 0)
 968                 {
 969                         safe = false;
 970                         break;
 971                 }
 972
 973                 /*
 974                  * We use a bitmapset to avoid testing the same attno more than once.
 975                  * (NB: this only works because subquery outputs can't have negative
 976                  * attnos.)
 977                  */
 978                 if (bms_is_member(var->varattno, tested))
 979                         continue;
 980                 tested = bms_add_member(tested, var->varattno);
 981
 982                 /* Check point 3 */
 983                 if (differentTypes[var->varattno])
 984                 {
 985                         safe = false;
 986                         break;
 987                 }
 988
 989                 /* Must find the tlist element referenced by the Var */
 990                 tle = get_tle_by_resno(subquery->targetList, var->varattno);
 991                 Assert(tle != NULL);
 992                 Assert(!tle->resjunk);
 993
 994                 /* If subquery uses DISTINCT or DISTINCT ON, check point 4 */
 995                 if (subquery->distinctClause != NIL &&
 996                         !targetIsInSortList(tle, InvalidOid, subquery->distinctClause))
 997                 {
 998                         /* non-DISTINCT column, so fail */
 999                         safe = false;
1000                         break;
1001                 }
1002
1003                 /* Refuse functions returning sets (point 5) */
1004                 if (expression_returns_set((Node *) tle->expr))
1005                 {
1006                         safe = false;
1007                         break;
1008                 }
1009
1010                 /* Refuse volatile functions (point 6) */
1011                 if (contain_volatile_functions((Node *) tle->expr))
1012                 {
1013                         safe = false;
1014                         break;
1015                 }
1016         }
1017
1018         list_free(vars);
1019         bms_free(tested);
1020
1021         return safe;
1022 }
1023
1024 /*
1025  * subquery_push_qual - push down a qual that we have determined is safe
1026  */
1027 static void
1028 subquery_push_qual(Query *subquery, RangeTblEntry *rte, Index rti, Node *qual)
1029 {
1030         if (subquery->setOperations != NULL)
1031         {
1032                 /* Recurse to push it separately to each component query */
1033                 recurse_push_qual(subquery->setOperations, subquery,
1034                                                   rte, rti, qual);
1035         }
1036         else
1037         {
1038                 /*
1039                  * We need to replace Vars in the qual (which must refer to outputs of
1040                  * the subquery) with copies of the subquery's targetlist expressions.
1041                  * Note that at this point, any uplevel Vars in the qual should have
1042                  * been replaced with Params, so they need no work.
1043                  *
1044                  * This step also ensures that when we are pushing into a setop tree,
1045                  * each component query gets its own copy of the qual.
1046                  */
1047                 qual = ResolveNew(qual, rti, 0, rte,
1048                                                   subquery->targetList,
1049                                                   CMD_SELECT, 0);
1050
1051                 /*
1052                  * Now attach the qual to the proper place: normally WHERE, but if the
1053                  * subquery uses grouping or aggregation, put it in HAVING (since the
1054                  * qual really refers to the group-result rows).
1055                  */
1056                 if (subquery->hasAggs || subquery->groupClause || subquery->havingQual)
1057                         subquery->havingQual = make_and_qual(subquery->havingQual, qual);
1058                 else
1059                         subquery->jointree->quals =
1060                                 make_and_qual(subquery->jointree->quals, qual);
1061
1062                 /*
1063                  * We need not change the subquery's hasAggs or hasSublinks flags,
1064                  * since we can't be pushing down any aggregates that weren't there
1065                  * before, and we don't push down subselects at all.
1066                  */
1067         }
1068 }
1069
1070 /*
1071  * Helper routine to recurse through setOperations tree
1072  */
1073 static void
1074 recurse_push_qual(Node *setOp, Query *topquery,
1075                                   RangeTblEntry *rte, Index rti, Node *qual)
1076 {
1077         if (IsA(setOp, RangeTblRef))
1078         {
1079                 RangeTblRef *rtr = (RangeTblRef *) setOp;
1080                 RangeTblEntry *subrte = rt_fetch(rtr->rtindex, topquery->rtable);
1081                 Query      *subquery = subrte->subquery;
1082
1083                 Assert(subquery != NULL);
1084                 subquery_push_qual(subquery, rte, rti, qual);
1085         }
1086         else if (IsA(setOp, SetOperationStmt))
1087         {
1088                 SetOperationStmt *op = (SetOperationStmt *) setOp;
1089
1090                 recurse_push_qual(op->larg, topquery, rte, rti, qual);
1091                 recurse_push_qual(op->rarg, topquery, rte, rti, qual);
1092         }
1093         else
1094         {
1095                 elog(ERROR, "unrecognized node type: %d",
1096                          (int) nodeTag(setOp));
1097         }
1098 }
1099
1100 /*****************************************************************************
1101  *                      DEBUG SUPPORT
1102  *****************************************************************************/
1103
1104 #ifdef OPTIMIZER_DEBUG
1105
1106 static void
1107 print_relids(Relids relids)
1108 {
1109         Relids          tmprelids;
1110         int                     x;
1111         bool            first = true;
1112
1113         tmprelids = bms_copy(relids);
1114         while ((x = bms_first_member(tmprelids)) >= 0)
1115         {
1116                 if (!first)
1117                         printf(" ");
1118                 printf("%d", x);
1119                 first = false;
1120         }
1121         bms_free(tmprelids);
1122 }
1123
1124 static void
1125 print_restrictclauses(PlannerInfo *root, List *clauses)
1126 {
1127         ListCell   *l;
1128
1129         foreach(l, clauses)
1130         {
1131                 RestrictInfo *c = lfirst(l);
1132
1133                 print_expr((Node *) c->clause, root->parse->rtable);
1134                 if (lnext(l))
1135                         printf(", ");
1136         }
1137 }
1138
1139 static void
1140 print_path(PlannerInfo *root, Path *path, int indent)
1141 {
1142         const char *ptype;
1143         bool            join = false;
1144         Path       *subpath = NULL;
1145         int                     i;
1146
1147         switch (nodeTag(path))
1148         {
1149                 case T_Path:
1150                         ptype = "SeqScan";
1151                         break;
1152                 case T_IndexPath:
1153                         ptype = "IdxScan";
1154                         break;
1155                 case T_BitmapHeapPath:
1156                         ptype = "BitmapHeapScan";
1157                         break;
1158                 case T_BitmapAndPath:
1159                         ptype = "BitmapAndPath";
1160                         break;
1161                 case T_BitmapOrPath:
1162                         ptype = "BitmapOrPath";
1163                         break;
1164                 case T_TidPath:
1165                         ptype = "TidScan";
1166                         break;
1167                 case T_AppendPath:
1168                         ptype = "Append";
1169                         break;
1170                 case T_ResultPath:
1171                         ptype = "Result";
1172                         break;
1173                 case T_MaterialPath:
1174                         ptype = "Material";
1175                         subpath = ((MaterialPath *) path)->subpath;
1176                         break;
1177                 case T_UniquePath:
1178                         ptype = "Unique";
1179                         subpath = ((UniquePath *) path)->subpath;
1180                         break;
1181                 case T_NestPath:
1182                         ptype = "NestLoop";
1183                         join = true;
1184                         break;
1185                 case T_MergePath:
1186                         ptype = "MergeJoin";
1187                         join = true;
1188                         break;
1189                 case T_HashPath:
1190                         ptype = "HashJoin";
1191                         join = true;
1192                         break;
1193                 default:
1194                         ptype = "???Path";
1195                         break;
1196         }
1197
1198         for (i = 0; i < indent; i++)
1199                 printf("\t");
1200         printf("%s", ptype);
1201
1202         if (path->parent)
1203         {
1204                 printf("(");
1205                 print_relids(path->parent->relids);
1206                 printf(") rows=%.0f", path->parent->rows);
1207         }
1208         printf(" cost=%.2f..%.2f\n", path->startup_cost, path->total_cost);
1209
1210         if (path->pathkeys)
1211         {
1212                 for (i = 0; i < indent; i++)
1213                         printf("\t");
1214                 printf("  pathkeys: ");
1215                 print_pathkeys(path->pathkeys, root->parse->rtable);
1216         }
1217
1218         if (join)
1219         {
1220                 JoinPath   *jp = (JoinPath *) path;
1221
1222                 for (i = 0; i < indent; i++)
1223                         printf("\t");
1224                 printf("  clauses: ");
1225                 print_restrictclauses(root, jp->joinrestrictinfo);
1226                 printf("\n");
1227
1228                 if (IsA(path, MergePath))
1229                 {
1230                         MergePath  *mp = (MergePath *) path;
1231
1232                         if (mp->outersortkeys || mp->innersortkeys)
1233                         {
1234                                 for (i = 0; i < indent; i++)
1235                                         printf("\t");
1236                                 printf("  sortouter=%d sortinner=%d\n",
1237                                            ((mp->outersortkeys) ? 1 : 0),
1238                                            ((mp->innersortkeys) ? 1 : 0));
1239                         }
1240                 }
1241
1242                 print_path(root, jp->outerjoinpath, indent + 1);
1243                 print_path(root, jp->innerjoinpath, indent + 1);
1244         }
1245
1246         if (subpath)
1247                 print_path(root, subpath, indent + 1);
1248 }
1249
1250 void
1251 debug_print_rel(PlannerInfo *root, RelOptInfo *rel)
1252 {
1253         ListCell   *l;
1254
1255         printf("RELOPTINFO (");
1256         print_relids(rel->relids);
1257         printf("): rows=%.0f width=%d\n", rel->rows, rel->width);
1258
1259         if (rel->baserestrictinfo)
1260         {
1261                 printf("\tbaserestrictinfo: ");
1262                 print_restrictclauses(root, rel->baserestrictinfo);
1263                 printf("\n");
1264         }
1265
1266         if (rel->joininfo)
1267         {
1268                 printf("\tjoininfo: ");
1269                 print_restrictclauses(root, rel->joininfo);
1270                 printf("\n");
1271         }
1272
1273         printf("\tpath list:\n");
1274         foreach(l, rel->pathlist)
1275                 print_path(root, lfirst(l), 1);
1276         printf("\n\tcheapest startup path:\n");
1277         print_path(root, rel->cheapest_startup_path, 1);
1278         printf("\n\tcheapest total path:\n");
1279         print_path(root, rel->cheapest_total_path, 1);
1280         printf("\n");
1281         fflush(stdout);
1282 }
1283
1284 #endif   /* OPTIMIZER_DEBUG */