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 1.159 2007/02/19 07:03:28 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, Index rti);
  43 static void set_plain_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
  44                                            RangeTblEntry *rte);
  45 static void set_append_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
  46                                                 Index rti, RangeTblEntry *rte);
  47 static void set_subquery_pathlist(PlannerInfo *root, RelOptInfo *rel,
  48                                           Index rti, RangeTblEntry *rte);
  49 static void set_function_pathlist(PlannerInfo *root, RelOptInfo *rel,
  50                                           RangeTblEntry *rte);
  51 static void set_values_pathlist(PlannerInfo *root, RelOptInfo *rel,
  52                                         RangeTblEntry *rte);
  53 static RelOptInfo *make_rel_from_joinlist(PlannerInfo *root, List *joinlist);
  54 static RelOptInfo *make_one_rel_by_joins(PlannerInfo *root, int levels_needed,
  55                                           List *initial_rels);
  56 static bool subquery_is_pushdown_safe(Query *subquery, Query *topquery,
  57                                                   bool *differentTypes);
  58 static bool recurse_pushdown_safe(Node *setOp, Query *topquery,
  59                                           bool *differentTypes);
  60 static void compare_tlist_datatypes(List *tlist, List *colTypes,
  61                                                 bool *differentTypes);
  62 static bool qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
  63                                           bool *differentTypes);
  64 static void subquery_push_qual(Query *subquery,
  65                                    RangeTblEntry *rte, Index rti, Node *qual);
  66 static void recurse_push_qual(Node *setOp, Query *topquery,
  67                                   RangeTblEntry *rte, Index rti, Node *qual);
  68
  69
  70 /*
  71  * make_one_rel
  72  *        Finds all possible access paths for executing a query, returning a
  73  *        single rel that represents the join of all base rels in the query.
  74  */
  75 RelOptInfo *
  76 make_one_rel(PlannerInfo *root, List *joinlist)
  77 {
  78         RelOptInfo *rel;
  79
  80         /*
  81          * Generate access paths for the base rels.
  82          */
  83         set_base_rel_pathlists(root);
  84
  85         /*
  86          * Generate access paths for the entire join tree.
  87          */
  88         rel = make_rel_from_joinlist(root, joinlist);
  89
  90         /*
  91          * The result should join all and only the query's base rels.
  92          */
  93 #ifdef USE_ASSERT_CHECKING
  94         {
  95                 int                     num_base_rels = 0;
  96                 Index           rti;
  97
  98                 for (rti = 1; rti < root->simple_rel_array_size; rti++)
  99                 {
 100                         RelOptInfo *brel = root->simple_rel_array[rti];
 101
 102                         if (brel == NULL)
 103                                 continue;
 104
 105                         Assert(brel->relid == rti); /* sanity check on array */
 106
 107                         /* ignore RTEs that are "other rels" */
 108                         if (brel->reloptkind != RELOPT_BASEREL)
 109                                 continue;
 110
 111                         Assert(bms_is_member(rti, rel->relids));
 112                         num_base_rels++;
 113                 }
 114
 115                 Assert(bms_num_members(rel->relids) == num_base_rels);
 116         }
 117 #endif
 118
 119         return rel;
 120 }
 121
 122 /*
 123  * set_base_rel_pathlists
 124  *        Finds all paths available for scanning each base-relation entry.
 125  *        Sequential scan and any available indices are considered.
 126  *        Each useful path is attached to its relation's 'pathlist' field.
 127  */
 128 static void
 129 set_base_rel_pathlists(PlannerInfo *root)
 130 {
 131         Index           rti;
 132
 133         for (rti = 1; rti < root->simple_rel_array_size; rti++)
 134         {
 135                 RelOptInfo *rel = root->simple_rel_array[rti];
 136
 137                 /* there may be empty slots corresponding to non-baserel RTEs */
 138                 if (rel == NULL)
 139                         continue;
 140
 141                 Assert(rel->relid == rti);              /* sanity check on array */
 142
 143                 /* ignore RTEs that are "other rels" */
 144                 if (rel->reloptkind != RELOPT_BASEREL)
 145                         continue;
 146
 147                 set_rel_pathlist(root, rel, rti);
 148         }
 149 }
 150
 151 /*
 152  * set_rel_pathlist
 153  *        Build access paths for a base relation
 154  */
 155 static void
 156 set_rel_pathlist(PlannerInfo *root, RelOptInfo *rel, Index rti)
 157 {
 158         RangeTblEntry *rte = rt_fetch(rti, root->parse->rtable);
 159
 160         if (rte->inh)
 161         {
 162                 /* It's an "append relation", process accordingly */
 163                 set_append_rel_pathlist(root, rel, rti, rte);
 164         }
 165         else if (rel->rtekind == RTE_SUBQUERY)
 166         {
 167                 /* Subquery --- generate a separate plan for it */
 168                 set_subquery_pathlist(root, rel, rti, rte);
 169         }
 170         else if (rel->rtekind == RTE_FUNCTION)
 171         {
 172                 /* RangeFunction --- generate a separate plan for it */
 173                 set_function_pathlist(root, rel, rte);
 174         }
 175         else if (rel->rtekind == RTE_VALUES)
 176         {
 177                 /* Values list --- generate a separate plan for it */
 178                 set_values_pathlist(root, rel, rte);
 179         }
 180         else
 181         {
 182                 /* Plain relation */
 183                 Assert(rel->rtekind == RTE_RELATION);
 184                 set_plain_rel_pathlist(root, rel, rte);
 185         }
 186
 187 #ifdef OPTIMIZER_DEBUG
 188         debug_print_rel(root, rel);
 189 #endif
 190 }
 191
 192 /*
 193  * set_plain_rel_pathlist
 194  *        Build access paths for a plain relation (no subquery, no inheritance)
 195  */
 196 static void
 197 set_plain_rel_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 198 {
 199         /* Mark rel with estimated output rows, width, etc */
 200         set_baserel_size_estimates(root, rel);
 201
 202         /* Test any partial indexes of rel for applicability */
 203         check_partial_indexes(root, rel);
 204
 205         /*
 206          * Check to see if we can extract any restriction conditions from join
 207          * quals that are OR-of-AND structures.  If so, add them to the rel's
 208          * restriction list, and recompute the size estimates.
 209          */
 210         if (create_or_index_quals(root, rel))
 211                 set_baserel_size_estimates(root, rel);
 212
 213         /*
 214          * If we can prove we don't need to scan the rel via constraint exclusion,
 215          * set up a single dummy path for it.  (Rather than inventing a special
 216          * "dummy" path type, we represent this as an AppendPath with no members.)
 217          */
 218         if (relation_excluded_by_constraints(rel, rte))
 219         {
 220                 /* Reset output-rows estimate to 0 */
 221                 rel->rows = 0;
 222
 223                 add_path(rel, (Path *) create_append_path(rel, NIL));
 224
 225                 /* Select cheapest path (pretty easy in this case...) */
 226                 set_cheapest(rel);
 227
 228                 return;
 229         }
 230
 231         /*
 232          * Generate paths and add them to the rel's pathlist.
 233          *
 234          * Note: add_path() will discard any paths that are dominated by another
 235          * available path, keeping only those paths that are superior along at
 236          * least one dimension of cost or sortedness.
 237          */
 238
 239         /* Consider sequential scan */
 240         add_path(rel, create_seqscan_path(root, rel));
 241
 242         /* Consider index scans */
 243         create_index_paths(root, rel);
 244
 245         /* Consider TID scans */
 246         create_tidscan_paths(root, rel);
 247
 248         /* Now find the cheapest of the paths for this rel */
 249         set_cheapest(rel);
 250 }
 251
 252 /*
 253  * set_append_rel_pathlist
 254  *        Build access paths for an "append relation"
 255  *
 256  * The passed-in rel and RTE represent the entire append relation.      The
 257  * relation's contents are computed by appending together the output of
 258  * the individual member relations.  Note that in the inheritance case,
 259  * the first member relation is actually the same table as is mentioned in
 260  * the parent RTE ... but it has a different RTE and RelOptInfo.  This is
 261  * a good thing because their outputs are not the same size.
 262  */
 263 static void
 264 set_append_rel_pathlist(PlannerInfo *root, RelOptInfo *rel,
 265                                                 Index rti, RangeTblEntry *rte)
 266 {
 267         int                     parentRTindex = rti;
 268         List       *subpaths = NIL;
 269         ListCell   *l;
 270
 271         /*
 272          * XXX for now, can't handle inherited expansion of FOR UPDATE/SHARE; can
 273          * we do better?  (This will take some redesign because the executor
 274          * currently supposes that every rowMark relation is involved in every row
 275          * returned by the query.)
 276          */
 277         if (get_rowmark(root->parse, parentRTindex))
 278                 ereport(ERROR,
 279                                 (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
 280                                  errmsg("SELECT FOR UPDATE/SHARE is not supported for inheritance queries")));
 281
 282         /*
 283          * Initialize to compute size estimates for whole append relation
 284          */
 285         rel->rows = 0;
 286         rel->width = 0;
 287
 288         /*
 289          * Generate access paths for each member relation, and pick the cheapest
 290          * path for each one.
 291          */
 292         foreach(l, root->append_rel_list)
 293         {
 294                 AppendRelInfo *appinfo = (AppendRelInfo *) lfirst(l);
 295                 int                     childRTindex;
 296                 RelOptInfo *childrel;
 297                 Path       *childpath;
 298                 ListCell   *parentvars;
 299                 ListCell   *childvars;
 300
 301                 /* append_rel_list contains all append rels; ignore others */
 302                 if (appinfo->parent_relid != parentRTindex)
 303                         continue;
 304
 305                 childRTindex = appinfo->child_relid;
 306
 307                 /*
 308                  * The child rel's RelOptInfo was already created during
 309                  * add_base_rels_to_query.
 310                  */
 311                 childrel = find_base_rel(root, childRTindex);
 312                 Assert(childrel->reloptkind == RELOPT_OTHER_MEMBER_REL);
 313
 314                 /*
 315                  * Copy the parent's targetlist and quals to the child, with
 316                  * appropriate substitution of variables.
 317                  */
 318                 childrel->reltargetlist = (List *)
 319                         adjust_appendrel_attrs((Node *) rel->reltargetlist,
 320                                                                    appinfo);
 321                 childrel->baserestrictinfo = (List *)
 322                         adjust_appendrel_attrs((Node *) rel->baserestrictinfo,
 323                                                                    appinfo);
 324                 childrel->joininfo = (List *)
 325                         adjust_appendrel_attrs((Node *) rel->joininfo,
 326                                                                    appinfo);
 327
 328                 /*
 329                  * We have to make child entries in the EquivalenceClass data
 330                  * structures as well.
 331                  */
 332                 if (rel->has_eclass_joins)
 333                 {
 334                         add_child_rel_equivalences(root, appinfo, rel, childrel);
 335                         childrel->has_eclass_joins = true;
 336                 }
 337
 338                 /*
 339                  * Copy the parent's attr_needed data as well, with appropriate
 340                  * adjustment of relids and attribute numbers.
 341                  */
 342                 pfree(childrel->attr_needed);
 343                 childrel->attr_needed =
 344                         adjust_appendrel_attr_needed(rel, appinfo,
 345                                                                                  childrel->min_attr,
 346                                                                                  childrel->max_attr);
 347
 348                 /*
 349                  * Compute the child's access paths, and add the cheapest one to the
 350                  * Append path we are constructing for the parent.
 351                  *
 352                  * It's possible that the child is itself an appendrel, in which case
 353                  * we can "cut out the middleman" and just add its child paths to our
 354                  * own list.  (We don't try to do this earlier because we need to
 355                  * apply both levels of transformation to the quals.) This test also
 356                  * handles the case where the child rel need not be scanned because of
 357                  * constraint exclusion: it'll have an Append path with no subpaths,
 358                  * and will vanish from our list.
 359                  */
 360                 set_rel_pathlist(root, childrel, childRTindex);
 361
 362                 childpath = childrel->cheapest_total_path;
 363                 if (IsA(childpath, AppendPath))
 364                         subpaths = list_concat(subpaths,
 365                                                                    ((AppendPath *) childpath)->subpaths);
 366                 else
 367                         subpaths = lappend(subpaths, childpath);
 368
 369                 /*
 370                  * Propagate size information from the child back to the parent. For
 371                  * simplicity, we use the largest widths from any child as the parent
 372                  * estimates.
 373                  */
 374                 rel->rows += childrel->rows;
 375                 if (childrel->width > rel->width)
 376                         rel->width = childrel->width;
 377
 378                 forboth(parentvars, rel->reltargetlist,
 379                                 childvars, childrel->reltargetlist)
 380                 {
 381                         Var                *parentvar = (Var *) lfirst(parentvars);
 382                         Var                *childvar = (Var *) lfirst(childvars);
 383
 384                         if (IsA(parentvar, Var) &&
 385                                 IsA(childvar, Var))
 386                         {
 387                                 int                     pndx = parentvar->varattno - rel->min_attr;
 388                                 int                     cndx = childvar->varattno - childrel->min_attr;
 389
 390                                 if (childrel->attr_widths[cndx] > rel->attr_widths[pndx])
 391                                         rel->attr_widths[pndx] = childrel->attr_widths[cndx];
 392                         }
 393                 }
 394         }
 395
 396         /*
 397          * Set "raw tuples" count equal to "rows" for the appendrel; needed
 398          * because some places assume rel->tuples is valid for any baserel.
 399          */
 400         rel->tuples = rel->rows;
 401
 402         /*
 403          * Finally, build Append path and install it as the only access path for
 404          * the parent rel.      (Note: this is correct even if we have zero or one
 405          * live subpath due to constraint exclusion.)
 406          */
 407         add_path(rel, (Path *) create_append_path(rel, subpaths));
 408
 409         /* Select cheapest path (pretty easy in this case...) */
 410         set_cheapest(rel);
 411 }
 412
 413 /* quick-and-dirty test to see if any joining is needed */
 414 static bool
 415 has_multiple_baserels(PlannerInfo *root)
 416 {
 417         int                     num_base_rels = 0;
 418         Index           rti;
 419
 420         for (rti = 1; rti < root->simple_rel_array_size; rti++)
 421         {
 422                 RelOptInfo *brel = root->simple_rel_array[rti];
 423
 424                 if (brel == NULL)
 425                         continue;
 426
 427                 /* ignore RTEs that are "other rels" */
 428                 if (brel->reloptkind == RELOPT_BASEREL)
 429                         if (++num_base_rels > 1)
 430                                 return true;
 431         }
 432         return false;
 433 }
 434
 435 /*
 436  * set_subquery_pathlist
 437  *              Build the (single) access path for a subquery RTE
 438  */
 439 static void
 440 set_subquery_pathlist(PlannerInfo *root, RelOptInfo *rel,
 441                                           Index rti, RangeTblEntry *rte)
 442 {
 443         Query      *parse = root->parse;
 444         Query      *subquery = rte->subquery;
 445         bool       *differentTypes;
 446         double          tuple_fraction;
 447         List       *pathkeys;
 448         List       *subquery_pathkeys;
 449
 450         /* We need a workspace for keeping track of set-op type coercions */
 451         differentTypes = (bool *)
 452                 palloc0((list_length(subquery->targetList) + 1) * sizeof(bool));
 453
 454         /*
 455          * If there are any restriction clauses that have been attached to the
 456          * subquery relation, consider pushing them down to become WHERE or HAVING
 457          * quals of the subquery itself.  This transformation is useful because it
 458          * may allow us to generate a better plan for the subquery than evaluating
 459          * all the subquery output rows and then filtering them.
 460          *
 461          * There are several cases where we cannot push down clauses. Restrictions
 462          * involving the subquery are checked by subquery_is_pushdown_safe().
 463          * Restrictions on individual clauses are checked by
 464          * qual_is_pushdown_safe().  Also, we don't want to push down
 465          * pseudoconstant clauses; better to have the gating node above the
 466          * subquery.
 467          *
 468          * Non-pushed-down clauses will get evaluated as qpquals of the
 469          * SubqueryScan node.
 470          *
 471          * XXX Are there any cases where we want to make a policy decision not to
 472          * push down a pushable qual, because it'd result in a worse plan?
 473          */
 474         if (rel->baserestrictinfo != NIL &&
 475                 subquery_is_pushdown_safe(subquery, subquery, differentTypes))
 476         {
 477                 /* OK to consider pushing down individual quals */
 478                 List       *upperrestrictlist = NIL;
 479                 ListCell   *l;
 480
 481                 foreach(l, rel->baserestrictinfo)
 482                 {
 483                         RestrictInfo *rinfo = (RestrictInfo *) lfirst(l);
 484                         Node       *clause = (Node *) rinfo->clause;
 485
 486                         if (!rinfo->pseudoconstant &&
 487                                 qual_is_pushdown_safe(subquery, rti, clause, differentTypes))
 488                         {
 489                                 /* Push it down */
 490                                 subquery_push_qual(subquery, rte, rti, clause);
 491                         }
 492                         else
 493                         {
 494                                 /* Keep it in the upper query */
 495                                 upperrestrictlist = lappend(upperrestrictlist, rinfo);
 496                         }
 497                 }
 498                 rel->baserestrictinfo = upperrestrictlist;
 499         }
 500
 501         pfree(differentTypes);
 502
 503         /*
 504          * We can safely pass the outer tuple_fraction down to the subquery if the
 505          * outer level has no joining, aggregation, or sorting to do. Otherwise
 506          * we'd better tell the subquery to plan for full retrieval. (XXX This
 507          * could probably be made more intelligent ...)
 508          */
 509         if (parse->hasAggs ||
 510                 parse->groupClause ||
 511                 parse->havingQual ||
 512                 parse->distinctClause ||
 513                 parse->sortClause ||
 514                 has_multiple_baserels(root))
 515                 tuple_fraction = 0.0;   /* default case */
 516         else
 517                 tuple_fraction = root->tuple_fraction;
 518
 519         /* Generate the plan for the subquery */
 520         rel->subplan = subquery_planner(root->glob, subquery,
 521                                                                         root->query_level + 1,
 522                                                                         tuple_fraction,
 523                                                                         &subquery_pathkeys);
 524
 525         /* Copy number of output rows from subplan */
 526         rel->tuples = rel->subplan->plan_rows;
 527
 528         /* Mark rel with estimated output rows, width, etc */
 529         set_baserel_size_estimates(root, rel);
 530
 531         /* Convert subquery pathkeys to outer representation */
 532         pathkeys = convert_subquery_pathkeys(root, rel, subquery_pathkeys);
 533
 534         /* Generate appropriate path */
 535         add_path(rel, create_subqueryscan_path(rel, pathkeys));
 536
 537         /* Select cheapest path (pretty easy in this case...) */
 538         set_cheapest(rel);
 539 }
 540
 541 /*
 542  * set_function_pathlist
 543  *              Build the (single) access path for a function RTE
 544  */
 545 static void
 546 set_function_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 547 {
 548         /* Mark rel with estimated output rows, width, etc */
 549         set_function_size_estimates(root, rel);
 550
 551         /* Generate appropriate path */
 552         add_path(rel, create_functionscan_path(root, rel));
 553
 554         /* Select cheapest path (pretty easy in this case...) */
 555         set_cheapest(rel);
 556 }
 557
 558 /*
 559  * set_values_pathlist
 560  *              Build the (single) access path for a VALUES RTE
 561  */
 562 static void
 563 set_values_pathlist(PlannerInfo *root, RelOptInfo *rel, RangeTblEntry *rte)
 564 {
 565         /* Mark rel with estimated output rows, width, etc */
 566         set_values_size_estimates(root, rel);
 567
 568         /* Generate appropriate path */
 569         add_path(rel, create_valuesscan_path(root, rel));
 570
 571         /* Select cheapest path (pretty easy in this case...) */
 572         set_cheapest(rel);
 573 }
 574
 575 /*
 576  * make_rel_from_joinlist
 577  *        Build access paths using a "joinlist" to guide the join path search.
 578  *
 579  * See comments for deconstruct_jointree() for definition of the joinlist
 580  * data structure.
 581  */
 582 static RelOptInfo *
 583 make_rel_from_joinlist(PlannerInfo *root, List *joinlist)
 584 {
 585         int                     levels_needed;
 586         List       *initial_rels;
 587         ListCell   *jl;
 588
 589         /*
 590          * Count the number of child joinlist nodes.  This is the depth of the
 591          * dynamic-programming algorithm we must employ to consider all ways of
 592          * joining the child nodes.
 593          */
 594         levels_needed = list_length(joinlist);
 595
 596         if (levels_needed <= 0)
 597                 return NULL;                    /* nothing to do? */
 598
 599         /*
 600          * Construct a list of rels corresponding to the child joinlist nodes.
 601          * This may contain both base rels and rels constructed according to
 602          * sub-joinlists.
 603          */
 604         initial_rels = NIL;
 605         foreach(jl, joinlist)
 606         {
 607                 Node       *jlnode = (Node *) lfirst(jl);
 608                 RelOptInfo *thisrel;
 609
 610                 if (IsA(jlnode, RangeTblRef))
 611                 {
 612                         int                     varno = ((RangeTblRef *) jlnode)->rtindex;
 613
 614                         thisrel = find_base_rel(root, varno);
 615                 }
 616                 else if (IsA(jlnode, List))
 617                 {
 618                         /* Recurse to handle subproblem */
 619                         thisrel = make_rel_from_joinlist(root, (List *) jlnode);
 620                 }
 621                 else
 622                 {
 623                         elog(ERROR, "unrecognized joinlist node type: %d",
 624                                  (int) nodeTag(jlnode));
 625                         thisrel = NULL;         /* keep compiler quiet */
 626                 }
 627
 628                 initial_rels = lappend(initial_rels, thisrel);
 629         }
 630
 631         if (levels_needed == 1)
 632         {
 633                 /*
 634                  * Single joinlist node, so we're done.
 635                  */
 636                 return (RelOptInfo *) linitial(initial_rels);
 637         }
 638         else
 639         {
 640                 /*
 641                  * Consider the different orders in which we could join the rels,
 642                  * using either GEQO or regular optimizer.
 643                  */
 644                 if (enable_geqo && levels_needed >= geqo_threshold)
 645                         return geqo(root, levels_needed, initial_rels);
 646                 else
 647                         return make_one_rel_by_joins(root, levels_needed, initial_rels);
 648         }
 649 }
 650
 651 /*
 652  * make_one_rel_by_joins
 653  *        Find all possible joinpaths for a query by successively finding ways
 654  *        to join component relations into join relations.
 655  *
 656  * 'levels_needed' is the number of iterations needed, ie, the number of
 657  *              independent jointree items in the query.  This is > 1.
 658  *
 659  * 'initial_rels' is a list of RelOptInfo nodes for each independent
 660  *              jointree item.  These are the components to be joined together.
 661  *
 662  * Returns the final level of join relations, i.e., the relation that is
 663  * the result of joining all the original relations together.
 664  */
 665 static RelOptInfo *
 666 make_one_rel_by_joins(PlannerInfo *root, int levels_needed, List *initial_rels)
 667 {
 668         List      **joinitems;
 669         int                     lev;
 670         RelOptInfo *rel;
 671
 672         /*
 673          * We employ a simple "dynamic programming" algorithm: we first find all
 674          * ways to build joins of two jointree items, then all ways to build joins
 675          * of three items (from two-item joins and single items), then four-item
 676          * joins, and so on until we have considered all ways to join all the
 677          * items into one rel.
 678          *
 679          * joinitems[j] is a list of all the j-item rels.  Initially we set
 680          * joinitems[1] to represent all the single-jointree-item relations.
 681          */
 682         joinitems = (List **) palloc0((levels_needed + 1) * sizeof(List *));
 683
 684         joinitems[1] = initial_rels;
 685
 686         for (lev = 2; lev <= levels_needed; lev++)
 687         {
 688                 ListCell   *x;
 689
 690                 /*
 691                  * Determine all possible pairs of relations to be joined at this
 692                  * level, and build paths for making each one from every available
 693                  * pair of lower-level relations.
 694                  */
 695                 joinitems[lev] = make_rels_by_joins(root, lev, joinitems);
 696
 697                 /*
 698                  * Do cleanup work on each just-processed rel.
 699                  */
 700                 foreach(x, joinitems[lev])
 701                 {
 702                         rel = (RelOptInfo *) lfirst(x);
 703
 704                         /* Find and save the cheapest paths for this rel */
 705                         set_cheapest(rel);
 706
 707 #ifdef OPTIMIZER_DEBUG
 708                         debug_print_rel(root, rel);
 709 #endif
 710                 }
 711         }
 712
 713         /*
 714          * We should have a single rel at the final level.
 715          */
 716         if (joinitems[levels_needed] == NIL)
 717                 elog(ERROR, "failed to build any %d-way joins", levels_needed);
 718         Assert(list_length(joinitems[levels_needed]) == 1);
 719
 720         rel = (RelOptInfo *) linitial(joinitems[levels_needed]);
 721
 722         return rel;
 723 }
 724
 725 /*****************************************************************************
 726  *                      PUSHING QUALS DOWN INTO SUBQUERIES
 727  *****************************************************************************/
 728
 729 /*
 730  * subquery_is_pushdown_safe - is a subquery safe for pushing down quals?
 731  *
 732  * subquery is the particular component query being checked.  topquery
 733  * is the top component of a set-operations tree (the same Query if no
 734  * set-op is involved).
 735  *
 736  * Conditions checked here:
 737  *
 738  * 1. If the subquery has a LIMIT clause, we must not push down any quals,
 739  * since that could change the set of rows returned.
 740  *
 741  * 2. If the subquery contains EXCEPT or EXCEPT ALL set ops we cannot push
 742  * quals into it, because that would change the results.
 743  *
 744  * 3. For subqueries using UNION/UNION ALL/INTERSECT/INTERSECT ALL, we can
 745  * push quals into each component query, but the quals can only reference
 746  * subquery columns that suffer no type coercions in the set operation.
 747  * Otherwise there are possible semantic gotchas.  So, we check the
 748  * component queries to see if any of them have different output types;
 749  * differentTypes[k] is set true if column k has different type in any
 750  * component.
 751  */
 752 static bool
 753 subquery_is_pushdown_safe(Query *subquery, Query *topquery,
 754                                                   bool *differentTypes)
 755 {
 756         SetOperationStmt *topop;
 757
 758         /* Check point 1 */
 759         if (subquery->limitOffset != NULL || subquery->limitCount != NULL)
 760                 return false;
 761
 762         /* Are we at top level, or looking at a setop component? */
 763         if (subquery == topquery)
 764         {
 765                 /* Top level, so check any component queries */
 766                 if (subquery->setOperations != NULL)
 767                         if (!recurse_pushdown_safe(subquery->setOperations, topquery,
 768                                                                            differentTypes))
 769                                 return false;
 770         }
 771         else
 772         {
 773                 /* Setop component must not have more components (too weird) */
 774                 if (subquery->setOperations != NULL)
 775                         return false;
 776                 /* Check whether setop component output types match top level */
 777                 topop = (SetOperationStmt *) topquery->setOperations;
 778                 Assert(topop && IsA(topop, SetOperationStmt));
 779                 compare_tlist_datatypes(subquery->targetList,
 780                                                                 topop->colTypes,
 781                                                                 differentTypes);
 782         }
 783         return true;
 784 }
 785
 786 /*
 787  * Helper routine to recurse through setOperations tree
 788  */
 789 static bool
 790 recurse_pushdown_safe(Node *setOp, Query *topquery,
 791                                           bool *differentTypes)
 792 {
 793         if (IsA(setOp, RangeTblRef))
 794         {
 795                 RangeTblRef *rtr = (RangeTblRef *) setOp;
 796                 RangeTblEntry *rte = rt_fetch(rtr->rtindex, topquery->rtable);
 797                 Query      *subquery = rte->subquery;
 798
 799                 Assert(subquery != NULL);
 800                 return subquery_is_pushdown_safe(subquery, topquery, differentTypes);
 801         }
 802         else if (IsA(setOp, SetOperationStmt))
 803         {
 804                 SetOperationStmt *op = (SetOperationStmt *) setOp;
 805
 806                 /* EXCEPT is no good */
 807                 if (op->op == SETOP_EXCEPT)
 808                         return false;
 809                 /* Else recurse */
 810                 if (!recurse_pushdown_safe(op->larg, topquery, differentTypes))
 811                         return false;
 812                 if (!recurse_pushdown_safe(op->rarg, topquery, differentTypes))
 813                         return false;
 814         }
 815         else
 816         {
 817                 elog(ERROR, "unrecognized node type: %d",
 818                          (int) nodeTag(setOp));
 819         }
 820         return true;
 821 }
 822
 823 /*
 824  * Compare tlist's datatypes against the list of set-operation result types.
 825  * For any items that are different, mark the appropriate element of
 826  * differentTypes[] to show that this column will have type conversions.
 827  *
 828  * We don't have to care about typmods here: the only allowed difference
 829  * between set-op input and output typmods is input is a specific typmod
 830  * and output is -1, and that does not require a coercion.
 831  */
 832 static void
 833 compare_tlist_datatypes(List *tlist, List *colTypes,
 834                                                 bool *differentTypes)
 835 {
 836         ListCell   *l;
 837         ListCell   *colType = list_head(colTypes);
 838
 839         foreach(l, tlist)
 840         {
 841                 TargetEntry *tle = (TargetEntry *) lfirst(l);
 842
 843                 if (tle->resjunk)
 844                         continue;                       /* ignore resjunk columns */
 845                 if (colType == NULL)
 846                         elog(ERROR, "wrong number of tlist entries");
 847                 if (exprType((Node *) tle->expr) != lfirst_oid(colType))
 848                         differentTypes[tle->resno] = true;
 849                 colType = lnext(colType);
 850         }
 851         if (colType != NULL)
 852                 elog(ERROR, "wrong number of tlist entries");
 853 }
 854
 855 /*
 856  * qual_is_pushdown_safe - is a particular qual safe to push down?
 857  *
 858  * qual is a restriction clause applying to the given subquery (whose RTE
 859  * has index rti in the parent query).
 860  *
 861  * Conditions checked here:
 862  *
 863  * 1. The qual must not contain any subselects (mainly because I'm not sure
 864  * it will work correctly: sublinks will already have been transformed into
 865  * subplans in the qual, but not in the subquery).
 866  *
 867  * 2. The qual must not refer to the whole-row output of the subquery
 868  * (since there is no easy way to name that within the subquery itself).
 869  *
 870  * 3. The qual must not refer to any subquery output columns that were
 871  * found to have inconsistent types across a set operation tree by
 872  * subquery_is_pushdown_safe().
 873  *
 874  * 4. If the subquery uses DISTINCT ON, we must not push down any quals that
 875  * refer to non-DISTINCT output columns, because that could change the set
 876  * of rows returned.  This condition is vacuous for DISTINCT, because then
 877  * there are no non-DISTINCT output columns, but unfortunately it's fairly
 878  * expensive to tell the difference between DISTINCT and DISTINCT ON in the
 879  * parsetree representation.  It's cheaper to just make sure all the Vars
 880  * in the qual refer to DISTINCT columns.
 881  *
 882  * 5. We must not push down any quals that refer to subselect outputs that
 883  * return sets, else we'd introduce functions-returning-sets into the
 884  * subquery's WHERE/HAVING quals.
 885  *
 886  * 6. We must not push down any quals that refer to subselect outputs that
 887  * contain volatile functions, for fear of introducing strange results due
 888  * to multiple evaluation of a volatile function.
 889  */
 890 static bool
 891 qual_is_pushdown_safe(Query *subquery, Index rti, Node *qual,
 892                                           bool *differentTypes)
 893 {
 894         bool            safe = true;
 895         List       *vars;
 896         ListCell   *vl;
 897         Bitmapset  *tested = NULL;
 898
 899         /* Refuse subselects (point 1) */
 900         if (contain_subplans(qual))
 901                 return false;
 902
 903         /*
 904          * Examine all Vars used in clause; since it's a restriction clause, all
 905          * such Vars must refer to subselect output columns.
 906          */
 907         vars = pull_var_clause(qual, false);
 908         foreach(vl, vars)
 909         {
 910                 Var                *var = (Var *) lfirst(vl);
 911                 TargetEntry *tle;
 912
 913                 Assert(var->varno == rti);
 914
 915                 /* Check point 2 */
 916                 if (var->varattno == 0)
 917                 {
 918                         safe = false;
 919                         break;
 920                 }
 921
 922                 /*
 923                  * We use a bitmapset to avoid testing the same attno more than once.
 924                  * (NB: this only works because subquery outputs can't have negative
 925                  * attnos.)
 926                  */
 927                 if (bms_is_member(var->varattno, tested))
 928                         continue;
 929                 tested = bms_add_member(tested, var->varattno);
 930
 931                 /* Check point 3 */
 932                 if (differentTypes[var->varattno])
 933                 {
 934                         safe = false;
 935                         break;
 936                 }
 937
 938                 /* Must find the tlist element referenced by the Var */
 939                 tle = get_tle_by_resno(subquery->targetList, var->varattno);
 940                 Assert(tle != NULL);
 941                 Assert(!tle->resjunk);
 942
 943                 /* If subquery uses DISTINCT or DISTINCT ON, check point 4 */
 944                 if (subquery->distinctClause != NIL &&
 945                         !targetIsInSortList(tle, InvalidOid, subquery->distinctClause))
 946                 {
 947                         /* non-DISTINCT column, so fail */
 948                         safe = false;
 949                         break;
 950                 }
 951
 952                 /* Refuse functions returning sets (point 5) */
 953                 if (expression_returns_set((Node *) tle->expr))
 954                 {
 955                         safe = false;
 956                         break;
 957                 }
 958
 959                 /* Refuse volatile functions (point 6) */
 960                 if (contain_volatile_functions((Node *) tle->expr))
 961                 {
 962                         safe = false;
 963                         break;
 964                 }
 965         }
 966
 967         list_free(vars);
 968         bms_free(tested);
 969
 970         return safe;
 971 }
 972
 973 /*
 974  * subquery_push_qual - push down a qual that we have determined is safe
 975  */
 976 static void
 977 subquery_push_qual(Query *subquery, RangeTblEntry *rte, Index rti, Node *qual)
 978 {
 979         if (subquery->setOperations != NULL)
 980         {
 981                 /* Recurse to push it separately to each component query */
 982                 recurse_push_qual(subquery->setOperations, subquery,
 983                                                   rte, rti, qual);
 984         }
 985         else
 986         {
 987                 /*
 988                  * We need to replace Vars in the qual (which must refer to outputs of
 989                  * the subquery) with copies of the subquery's targetlist expressions.
 990                  * Note that at this point, any uplevel Vars in the qual should have
 991                  * been replaced with Params, so they need no work.
 992                  *
 993                  * This step also ensures that when we are pushing into a setop tree,
 994                  * each component query gets its own copy of the qual.
 995                  */
 996                 qual = ResolveNew(qual, rti, 0, rte,
 997                                                   subquery->targetList,
 998                                                   CMD_SELECT, 0);
 999
1000                 /*
1001                  * Now attach the qual to the proper place: normally WHERE, but if the
1002                  * subquery uses grouping or aggregation, put it in HAVING (since the
1003                  * qual really refers to the group-result rows).
1004                  */
1005                 if (subquery->hasAggs || subquery->groupClause || subquery->havingQual)
1006                         subquery->havingQual = make_and_qual(subquery->havingQual, qual);
1007                 else
1008                         subquery->jointree->quals =
1009                                 make_and_qual(subquery->jointree->quals, qual);
1010
1011                 /*
1012                  * We need not change the subquery's hasAggs or hasSublinks flags,
1013                  * since we can't be pushing down any aggregates that weren't there
1014                  * before, and we don't push down subselects at all.
1015                  */
1016         }
1017 }
1018
1019 /*
1020  * Helper routine to recurse through setOperations tree
1021  */
1022 static void
1023 recurse_push_qual(Node *setOp, Query *topquery,
1024                                   RangeTblEntry *rte, Index rti, Node *qual)
1025 {
1026         if (IsA(setOp, RangeTblRef))
1027         {
1028                 RangeTblRef *rtr = (RangeTblRef *) setOp;
1029                 RangeTblEntry *subrte = rt_fetch(rtr->rtindex, topquery->rtable);
1030                 Query      *subquery = subrte->subquery;
1031
1032                 Assert(subquery != NULL);
1033                 subquery_push_qual(subquery, rte, rti, qual);
1034         }
1035         else if (IsA(setOp, SetOperationStmt))
1036         {
1037                 SetOperationStmt *op = (SetOperationStmt *) setOp;
1038
1039                 recurse_push_qual(op->larg, topquery, rte, rti, qual);
1040                 recurse_push_qual(op->rarg, topquery, rte, rti, qual);
1041         }
1042         else
1043         {
1044                 elog(ERROR, "unrecognized node type: %d",
1045                          (int) nodeTag(setOp));
1046         }
1047 }
1048
1049 /*****************************************************************************
1050  *                      DEBUG SUPPORT
1051  *****************************************************************************/
1052
1053 #ifdef OPTIMIZER_DEBUG
1054
1055 static void
1056 print_relids(Relids relids)
1057 {
1058         Relids          tmprelids;
1059         int                     x;
1060         bool            first = true;
1061
1062         tmprelids = bms_copy(relids);
1063         while ((x = bms_first_member(tmprelids)) >= 0)
1064         {
1065                 if (!first)
1066                         printf(" ");
1067                 printf("%d", x);
1068                 first = false;
1069         }
1070         bms_free(tmprelids);
1071 }
1072
1073 static void
1074 print_restrictclauses(PlannerInfo *root, List *clauses)
1075 {
1076         ListCell   *l;
1077
1078         foreach(l, clauses)
1079         {
1080                 RestrictInfo *c = lfirst(l);
1081
1082                 print_expr((Node *) c->clause, root->parse->rtable);
1083                 if (lnext(l))
1084                         printf(", ");
1085         }
1086 }
1087
1088 static void
1089 print_path(PlannerInfo *root, Path *path, int indent)
1090 {
1091         const char *ptype;
1092         bool            join = false;
1093         Path       *subpath = NULL;
1094         int                     i;
1095
1096         switch (nodeTag(path))
1097         {
1098                 case T_Path:
1099                         ptype = "SeqScan";
1100                         break;
1101                 case T_IndexPath:
1102                         ptype = "IdxScan";
1103                         break;
1104                 case T_BitmapHeapPath:
1105                         ptype = "BitmapHeapScan";
1106                         break;
1107                 case T_BitmapAndPath:
1108                         ptype = "BitmapAndPath";
1109                         break;
1110                 case T_BitmapOrPath:
1111                         ptype = "BitmapOrPath";
1112                         break;
1113                 case T_TidPath:
1114                         ptype = "TidScan";
1115                         break;
1116                 case T_AppendPath:
1117                         ptype = "Append";
1118                         break;
1119                 case T_ResultPath:
1120                         ptype = "Result";
1121                         break;
1122                 case T_MaterialPath:
1123                         ptype = "Material";
1124                         subpath = ((MaterialPath *) path)->subpath;
1125                         break;
1126                 case T_UniquePath:
1127                         ptype = "Unique";
1128                         subpath = ((UniquePath *) path)->subpath;
1129                         break;
1130                 case T_NestPath:
1131                         ptype = "NestLoop";
1132                         join = true;
1133                         break;
1134                 case T_MergePath:
1135                         ptype = "MergeJoin";
1136                         join = true;
1137                         break;
1138                 case T_HashPath:
1139                         ptype = "HashJoin";
1140                         join = true;
1141                         break;
1142                 default:
1143                         ptype = "???Path";
1144                         break;
1145         }
1146
1147         for (i = 0; i < indent; i++)
1148                 printf("\t");
1149         printf("%s", ptype);
1150
1151         if (path->parent)
1152         {
1153                 printf("(");
1154                 print_relids(path->parent->relids);
1155                 printf(") rows=%.0f", path->parent->rows);
1156         }
1157         printf(" cost=%.2f..%.2f\n", path->startup_cost, path->total_cost);
1158
1159         if (path->pathkeys)
1160         {
1161                 for (i = 0; i < indent; i++)
1162                         printf("\t");
1163                 printf("  pathkeys: ");
1164                 print_pathkeys(path->pathkeys, root->parse->rtable);
1165         }
1166
1167         if (join)
1168         {
1169                 JoinPath   *jp = (JoinPath *) path;
1170
1171                 for (i = 0; i < indent; i++)
1172                         printf("\t");
1173                 printf("  clauses: ");
1174                 print_restrictclauses(root, jp->joinrestrictinfo);
1175                 printf("\n");
1176
1177                 if (IsA(path, MergePath))
1178                 {
1179                         MergePath  *mp = (MergePath *) path;
1180
1181                         if (mp->outersortkeys || mp->innersortkeys)
1182                         {
1183                                 for (i = 0; i < indent; i++)
1184                                         printf("\t");
1185                                 printf("  sortouter=%d sortinner=%d\n",
1186                                            ((mp->outersortkeys) ? 1 : 0),
1187                                            ((mp->innersortkeys) ? 1 : 0));
1188                         }
1189                 }
1190
1191                 print_path(root, jp->outerjoinpath, indent + 1);
1192                 print_path(root, jp->innerjoinpath, indent + 1);
1193         }
1194
1195         if (subpath)
1196                 print_path(root, subpath, indent + 1);
1197 }
1198
1199 void
1200 debug_print_rel(PlannerInfo *root, RelOptInfo *rel)
1201 {
1202         ListCell   *l;
1203
1204         printf("RELOPTINFO (");
1205         print_relids(rel->relids);
1206         printf("): rows=%.0f width=%d\n", rel->rows, rel->width);
1207
1208         if (rel->baserestrictinfo)
1209         {
1210                 printf("\tbaserestrictinfo: ");
1211                 print_restrictclauses(root, rel->baserestrictinfo);
1212                 printf("\n");
1213         }
1214
1215         if (rel->joininfo)
1216         {
1217                 printf("\tjoininfo: ");
1218                 print_restrictclauses(root, rel->joininfo);
1219                 printf("\n");
1220         }
1221
1222         printf("\tpath list:\n");
1223         foreach(l, rel->pathlist)
1224                 print_path(root, lfirst(l), 1);
1225         printf("\n\tcheapest startup path:\n");
1226         print_path(root, rel->cheapest_startup_path, 1);
1227         printf("\n\tcheapest total path:\n");
1228         print_path(root, rel->cheapest_total_path, 1);
1229         printf("\n");
1230         fflush(stdout);
1231 }
1232
1233 #endif   /* OPTIMIZER_DEBUG */