Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * pathnode.c
4 : * Routines to manipulate pathlists and create path nodes
5 : *
6 : * Portions Copyright (c) 1996-2025, PostgreSQL Global Development Group
7 : * Portions Copyright (c) 1994, Regents of the University of California
8 : *
9 : *
10 : * IDENTIFICATION
11 : * src/backend/optimizer/util/pathnode.c
12 : *
13 : *-------------------------------------------------------------------------
14 : */
15 : #include "postgres.h"
16 :
17 : #include <math.h>
18 :
19 : #include "foreign/fdwapi.h"
20 : #include "miscadmin.h"
21 : #include "nodes/extensible.h"
22 : #include "optimizer/appendinfo.h"
23 : #include "optimizer/clauses.h"
24 : #include "optimizer/cost.h"
25 : #include "optimizer/optimizer.h"
26 : #include "optimizer/pathnode.h"
27 : #include "optimizer/paths.h"
28 : #include "optimizer/planmain.h"
29 : #include "optimizer/tlist.h"
30 : #include "parser/parsetree.h"
31 : #include "utils/memutils.h"
32 : #include "utils/selfuncs.h"
33 :
34 : typedef enum
35 : {
36 : COSTS_EQUAL, /* path costs are fuzzily equal */
37 : COSTS_BETTER1, /* first path is cheaper than second */
38 : COSTS_BETTER2, /* second path is cheaper than first */
39 : COSTS_DIFFERENT, /* neither path dominates the other on cost */
40 : } PathCostComparison;
41 :
42 : /*
43 : * STD_FUZZ_FACTOR is the normal fuzz factor for compare_path_costs_fuzzily.
44 : * XXX is it worth making this user-controllable? It provides a tradeoff
45 : * between planner runtime and the accuracy of path cost comparisons.
46 : */
47 : #define STD_FUZZ_FACTOR 1.01
48 :
49 : static List *translate_sub_tlist(List *tlist, int relid);
50 : static int append_total_cost_compare(const ListCell *a, const ListCell *b);
51 : static int append_startup_cost_compare(const ListCell *a, const ListCell *b);
52 : static List *reparameterize_pathlist_by_child(PlannerInfo *root,
53 : List *pathlist,
54 : RelOptInfo *child_rel);
55 : static bool pathlist_is_reparameterizable_by_child(List *pathlist,
56 : RelOptInfo *child_rel);
57 :
58 :
59 : /*****************************************************************************
60 : * MISC. PATH UTILITIES
61 : *****************************************************************************/
62 :
63 : /*
64 : * compare_path_costs
65 : * Return -1, 0, or +1 according as path1 is cheaper, the same cost,
66 : * or more expensive than path2 for the specified criterion.
67 : */
68 : int
69 1052828 : compare_path_costs(Path *path1, Path *path2, CostSelector criterion)
70 : {
71 : /* Number of disabled nodes, if different, trumps all else. */
72 1052828 : if (unlikely(path1->disabled_nodes != path2->disabled_nodes))
73 : {
74 2604 : if (path1->disabled_nodes < path2->disabled_nodes)
75 2604 : return -1;
76 : else
77 0 : return +1;
78 : }
79 :
80 1050224 : if (criterion == STARTUP_COST)
81 : {
82 532798 : if (path1->startup_cost < path2->startup_cost)
83 318934 : return -1;
84 213864 : if (path1->startup_cost > path2->startup_cost)
85 104748 : return +1;
86 :
87 : /*
88 : * If paths have the same startup cost (not at all unlikely), order
89 : * them by total cost.
90 : */
91 109116 : if (path1->total_cost < path2->total_cost)
92 55678 : return -1;
93 53438 : if (path1->total_cost > path2->total_cost)
94 4490 : return +1;
95 : }
96 : else
97 : {
98 517426 : if (path1->total_cost < path2->total_cost)
99 485016 : return -1;
100 32410 : if (path1->total_cost > path2->total_cost)
101 3882 : return +1;
102 :
103 : /*
104 : * If paths have the same total cost, order them by startup cost.
105 : */
106 28528 : if (path1->startup_cost < path2->startup_cost)
107 3240 : return -1;
108 25288 : if (path1->startup_cost > path2->startup_cost)
109 32 : return +1;
110 : }
111 74204 : return 0;
112 : }
113 :
114 : /*
115 : * compare_fractional_path_costs
116 : * Return -1, 0, or +1 according as path1 is cheaper, the same cost,
117 : * or more expensive than path2 for fetching the specified fraction
118 : * of the total tuples.
119 : *
120 : * If fraction is <= 0 or > 1, we interpret it as 1, ie, we select the
121 : * path with the cheaper total_cost.
122 : */
123 : int
124 5920 : compare_fractional_path_costs(Path *path1, Path *path2,
125 : double fraction)
126 : {
127 : Cost cost1,
128 : cost2;
129 :
130 : /* Number of disabled nodes, if different, trumps all else. */
131 5920 : if (unlikely(path1->disabled_nodes != path2->disabled_nodes))
132 : {
133 36 : if (path1->disabled_nodes < path2->disabled_nodes)
134 36 : return -1;
135 : else
136 0 : return +1;
137 : }
138 :
139 5884 : if (fraction <= 0.0 || fraction >= 1.0)
140 1806 : return compare_path_costs(path1, path2, TOTAL_COST);
141 4078 : cost1 = path1->startup_cost +
142 4078 : fraction * (path1->total_cost - path1->startup_cost);
143 4078 : cost2 = path2->startup_cost +
144 4078 : fraction * (path2->total_cost - path2->startup_cost);
145 4078 : if (cost1 < cost2)
146 3400 : return -1;
147 678 : if (cost1 > cost2)
148 678 : return +1;
149 0 : return 0;
150 : }
151 :
152 : /*
153 : * compare_path_costs_fuzzily
154 : * Compare the costs of two paths to see if either can be said to
155 : * dominate the other.
156 : *
157 : * We use fuzzy comparisons so that add_path() can avoid keeping both of
158 : * a pair of paths that really have insignificantly different cost.
159 : *
160 : * The fuzz_factor argument must be 1.0 plus delta, where delta is the
161 : * fraction of the smaller cost that is considered to be a significant
162 : * difference. For example, fuzz_factor = 1.01 makes the fuzziness limit
163 : * be 1% of the smaller cost.
164 : *
165 : * The two paths are said to have "equal" costs if both startup and total
166 : * costs are fuzzily the same. Path1 is said to be better than path2 if
167 : * it has fuzzily better startup cost and fuzzily no worse total cost,
168 : * or if it has fuzzily better total cost and fuzzily no worse startup cost.
169 : * Path2 is better than path1 if the reverse holds. Finally, if one path
170 : * is fuzzily better than the other on startup cost and fuzzily worse on
171 : * total cost, we just say that their costs are "different", since neither
172 : * dominates the other across the whole performance spectrum.
173 : *
174 : * This function also enforces a policy rule that paths for which the relevant
175 : * one of parent->consider_startup and parent->consider_param_startup is false
176 : * cannot survive comparisons solely on the grounds of good startup cost, so
177 : * we never return COSTS_DIFFERENT when that is true for the total-cost loser.
178 : * (But if total costs are fuzzily equal, we compare startup costs anyway,
179 : * in hopes of eliminating one path or the other.)
180 : */
181 : static PathCostComparison
182 4336582 : compare_path_costs_fuzzily(Path *path1, Path *path2, double fuzz_factor)
183 : {
184 : #define CONSIDER_PATH_STARTUP_COST(p) \
185 : ((p)->param_info == NULL ? (p)->parent->consider_startup : (p)->parent->consider_param_startup)
186 :
187 : /* Number of disabled nodes, if different, trumps all else. */
188 4336582 : if (unlikely(path1->disabled_nodes != path2->disabled_nodes))
189 : {
190 30008 : if (path1->disabled_nodes < path2->disabled_nodes)
191 16246 : return COSTS_BETTER1;
192 : else
193 13762 : return COSTS_BETTER2;
194 : }
195 :
196 : /*
197 : * Check total cost first since it's more likely to be different; many
198 : * paths have zero startup cost.
199 : */
200 4306574 : if (path1->total_cost > path2->total_cost * fuzz_factor)
201 : {
202 : /* path1 fuzzily worse on total cost */
203 2273328 : if (CONSIDER_PATH_STARTUP_COST(path1) &&
204 125122 : path2->startup_cost > path1->startup_cost * fuzz_factor)
205 : {
206 : /* ... but path2 fuzzily worse on startup, so DIFFERENT */
207 84326 : return COSTS_DIFFERENT;
208 : }
209 : /* else path2 dominates */
210 2189002 : return COSTS_BETTER2;
211 : }
212 2033246 : if (path2->total_cost > path1->total_cost * fuzz_factor)
213 : {
214 : /* path2 fuzzily worse on total cost */
215 1026050 : if (CONSIDER_PATH_STARTUP_COST(path2) &&
216 54222 : path1->startup_cost > path2->startup_cost * fuzz_factor)
217 : {
218 : /* ... but path1 fuzzily worse on startup, so DIFFERENT */
219 35408 : return COSTS_DIFFERENT;
220 : }
221 : /* else path1 dominates */
222 990642 : return COSTS_BETTER1;
223 : }
224 : /* fuzzily the same on total cost ... */
225 1007196 : if (path1->startup_cost > path2->startup_cost * fuzz_factor)
226 : {
227 : /* ... but path1 fuzzily worse on startup, so path2 wins */
228 395034 : return COSTS_BETTER2;
229 : }
230 612162 : if (path2->startup_cost > path1->startup_cost * fuzz_factor)
231 : {
232 : /* ... but path2 fuzzily worse on startup, so path1 wins */
233 68908 : return COSTS_BETTER1;
234 : }
235 : /* fuzzily the same on both costs */
236 543254 : return COSTS_EQUAL;
237 :
238 : #undef CONSIDER_PATH_STARTUP_COST
239 : }
240 :
241 : /*
242 : * set_cheapest
243 : * Find the minimum-cost paths from among a relation's paths,
244 : * and save them in the rel's cheapest-path fields.
245 : *
246 : * cheapest_total_path is normally the cheapest-total-cost unparameterized
247 : * path; but if there are no unparameterized paths, we assign it to be the
248 : * best (cheapest least-parameterized) parameterized path. However, only
249 : * unparameterized paths are considered candidates for cheapest_startup_path,
250 : * so that will be NULL if there are no unparameterized paths.
251 : *
252 : * The cheapest_parameterized_paths list collects all parameterized paths
253 : * that have survived the add_path() tournament for this relation. (Since
254 : * add_path ignores pathkeys for a parameterized path, these will be paths
255 : * that have best cost or best row count for their parameterization. We
256 : * may also have both a parallel-safe and a non-parallel-safe path in some
257 : * cases for the same parameterization in some cases, but this should be
258 : * relatively rare since, most typically, all paths for the same relation
259 : * will be parallel-safe or none of them will.)
260 : *
261 : * cheapest_parameterized_paths always includes the cheapest-total
262 : * unparameterized path, too, if there is one; the users of that list find
263 : * it more convenient if that's included.
264 : *
265 : * This is normally called only after we've finished constructing the path
266 : * list for the rel node.
267 : */
268 : void
269 2075020 : set_cheapest(RelOptInfo *parent_rel)
270 : {
271 : Path *cheapest_startup_path;
272 : Path *cheapest_total_path;
273 : Path *best_param_path;
274 : List *parameterized_paths;
275 : ListCell *p;
276 :
277 : Assert(IsA(parent_rel, RelOptInfo));
278 :
279 2075020 : if (parent_rel->pathlist == NIL)
280 0 : elog(ERROR, "could not devise a query plan for the given query");
281 :
282 2075020 : cheapest_startup_path = cheapest_total_path = best_param_path = NULL;
283 2075020 : parameterized_paths = NIL;
284 :
285 4688934 : foreach(p, parent_rel->pathlist)
286 : {
287 2613914 : Path *path = (Path *) lfirst(p);
288 : int cmp;
289 :
290 2613914 : if (path->param_info)
291 : {
292 : /* Parameterized path, so add it to parameterized_paths */
293 134390 : parameterized_paths = lappend(parameterized_paths, path);
294 :
295 : /*
296 : * If we have an unparameterized cheapest-total, we no longer care
297 : * about finding the best parameterized path, so move on.
298 : */
299 134390 : if (cheapest_total_path)
300 26100 : continue;
301 :
302 : /*
303 : * Otherwise, track the best parameterized path, which is the one
304 : * with least total cost among those of the minimum
305 : * parameterization.
306 : */
307 108290 : if (best_param_path == NULL)
308 98166 : best_param_path = path;
309 : else
310 : {
311 10124 : switch (bms_subset_compare(PATH_REQ_OUTER(path),
312 10124 : PATH_REQ_OUTER(best_param_path)))
313 : {
314 54 : case BMS_EQUAL:
315 : /* keep the cheaper one */
316 54 : if (compare_path_costs(path, best_param_path,
317 : TOTAL_COST) < 0)
318 0 : best_param_path = path;
319 54 : break;
320 350 : case BMS_SUBSET1:
321 : /* new path is less-parameterized */
322 350 : best_param_path = path;
323 350 : break;
324 0 : case BMS_SUBSET2:
325 : /* old path is less-parameterized, keep it */
326 0 : break;
327 9720 : case BMS_DIFFERENT:
328 :
329 : /*
330 : * This means that neither path has the least possible
331 : * parameterization for the rel. We'll sit on the old
332 : * path until something better comes along.
333 : */
334 9720 : break;
335 : }
336 108290 : }
337 : }
338 : else
339 : {
340 : /* Unparameterized path, so consider it for cheapest slots */
341 2479524 : if (cheapest_total_path == NULL)
342 : {
343 2063696 : cheapest_startup_path = cheapest_total_path = path;
344 2063696 : continue;
345 : }
346 :
347 : /*
348 : * If we find two paths of identical costs, try to keep the
349 : * better-sorted one. The paths might have unrelated sort
350 : * orderings, in which case we can only guess which might be
351 : * better to keep, but if one is superior then we definitely
352 : * should keep that one.
353 : */
354 415828 : cmp = compare_path_costs(cheapest_startup_path, path, STARTUP_COST);
355 415828 : if (cmp > 0 ||
356 440 : (cmp == 0 &&
357 440 : compare_pathkeys(cheapest_startup_path->pathkeys,
358 : path->pathkeys) == PATHKEYS_BETTER2))
359 81474 : cheapest_startup_path = path;
360 :
361 415828 : cmp = compare_path_costs(cheapest_total_path, path, TOTAL_COST);
362 415828 : if (cmp > 0 ||
363 48 : (cmp == 0 &&
364 48 : compare_pathkeys(cheapest_total_path->pathkeys,
365 : path->pathkeys) == PATHKEYS_BETTER2))
366 0 : cheapest_total_path = path;
367 : }
368 : }
369 :
370 : /* Add cheapest unparameterized path, if any, to parameterized_paths */
371 2075020 : if (cheapest_total_path)
372 2063696 : parameterized_paths = lcons(cheapest_total_path, parameterized_paths);
373 :
374 : /*
375 : * If there is no unparameterized path, use the best parameterized path as
376 : * cheapest_total_path (but not as cheapest_startup_path).
377 : */
378 2075020 : if (cheapest_total_path == NULL)
379 11324 : cheapest_total_path = best_param_path;
380 : Assert(cheapest_total_path != NULL);
381 :
382 2075020 : parent_rel->cheapest_startup_path = cheapest_startup_path;
383 2075020 : parent_rel->cheapest_total_path = cheapest_total_path;
384 2075020 : parent_rel->cheapest_unique_path = NULL; /* computed only if needed */
385 2075020 : parent_rel->cheapest_parameterized_paths = parameterized_paths;
386 2075020 : }
387 :
388 : /*
389 : * add_path
390 : * Consider a potential implementation path for the specified parent rel,
391 : * and add it to the rel's pathlist if it is worthy of consideration.
392 : *
393 : * A path is worthy if it has a better sort order (better pathkeys) or
394 : * cheaper cost (as defined below), or generates fewer rows, than any
395 : * existing path that has the same or superset parameterization rels. We
396 : * also consider parallel-safe paths more worthy than others.
397 : *
398 : * Cheaper cost can mean either a cheaper total cost or a cheaper startup
399 : * cost; if one path is cheaper in one of these aspects and another is
400 : * cheaper in the other, we keep both. However, when some path type is
401 : * disabled (e.g. due to enable_seqscan=false), the number of times that
402 : * a disabled path type is used is considered to be a higher-order
403 : * component of the cost. Hence, if path A uses no disabled path type,
404 : * and path B uses 1 or more disabled path types, A is cheaper, no matter
405 : * what we estimate for the startup and total costs. The startup and total
406 : * cost essentially act as a tiebreak when comparing paths that use equal
407 : * numbers of disabled path nodes; but in practice this tiebreak is almost
408 : * always used, since normally no path types are disabled.
409 : *
410 : * In addition to possibly adding new_path, we also remove from the rel's
411 : * pathlist any old paths that are dominated by new_path --- that is,
412 : * new_path is cheaper, at least as well ordered, generates no more rows,
413 : * requires no outer rels not required by the old path, and is no less
414 : * parallel-safe.
415 : *
416 : * In most cases, a path with a superset parameterization will generate
417 : * fewer rows (since it has more join clauses to apply), so that those two
418 : * figures of merit move in opposite directions; this means that a path of
419 : * one parameterization can seldom dominate a path of another. But such
420 : * cases do arise, so we make the full set of checks anyway.
421 : *
422 : * There are two policy decisions embedded in this function, along with
423 : * its sibling add_path_precheck. First, we treat all parameterized paths
424 : * as having NIL pathkeys, so that they cannot win comparisons on the
425 : * basis of sort order. This is to reduce the number of parameterized
426 : * paths that are kept; see discussion in src/backend/optimizer/README.
427 : *
428 : * Second, we only consider cheap startup cost to be interesting if
429 : * parent_rel->consider_startup is true for an unparameterized path, or
430 : * parent_rel->consider_param_startup is true for a parameterized one.
431 : * Again, this allows discarding useless paths sooner.
432 : *
433 : * The pathlist is kept sorted by disabled_nodes and then by total_cost,
434 : * with cheaper paths at the front. Within this routine, that's simply a
435 : * speed hack: doing it that way makes it more likely that we will reject
436 : * an inferior path after a few comparisons, rather than many comparisons.
437 : * However, add_path_precheck relies on this ordering to exit early
438 : * when possible.
439 : *
440 : * NOTE: discarded Path objects are immediately pfree'd to reduce planner
441 : * memory consumption. We dare not try to free the substructure of a Path,
442 : * since much of it may be shared with other Paths or the query tree itself;
443 : * but just recycling discarded Path nodes is a very useful savings in
444 : * a large join tree. We can recycle the List nodes of pathlist, too.
445 : *
446 : * As noted in optimizer/README, deleting a previously-accepted Path is
447 : * safe because we know that Paths of this rel cannot yet be referenced
448 : * from any other rel, such as a higher-level join. However, in some cases
449 : * it is possible that a Path is referenced by another Path for its own
450 : * rel; we must not delete such a Path, even if it is dominated by the new
451 : * Path. Currently this occurs only for IndexPath objects, which may be
452 : * referenced as children of BitmapHeapPaths as well as being paths in
453 : * their own right. Hence, we don't pfree IndexPaths when rejecting them.
454 : *
455 : * 'parent_rel' is the relation entry to which the path corresponds.
456 : * 'new_path' is a potential path for parent_rel.
457 : *
458 : * Returns nothing, but modifies parent_rel->pathlist.
459 : */
460 : void
461 4374574 : add_path(RelOptInfo *parent_rel, Path *new_path)
462 : {
463 4374574 : bool accept_new = true; /* unless we find a superior old path */
464 4374574 : int insert_at = 0; /* where to insert new item */
465 : List *new_path_pathkeys;
466 : ListCell *p1;
467 :
468 : /*
469 : * This is a convenient place to check for query cancel --- no part of the
470 : * planner goes very long without calling add_path().
471 : */
472 4374574 : CHECK_FOR_INTERRUPTS();
473 :
474 : /* Pretend parameterized paths have no pathkeys, per comment above */
475 4374574 : new_path_pathkeys = new_path->param_info ? NIL : new_path->pathkeys;
476 :
477 : /*
478 : * Loop to check proposed new path against old paths. Note it is possible
479 : * for more than one old path to be tossed out because new_path dominates
480 : * it.
481 : */
482 6714024 : foreach(p1, parent_rel->pathlist)
483 : {
484 3985216 : Path *old_path = (Path *) lfirst(p1);
485 3985216 : bool remove_old = false; /* unless new proves superior */
486 : PathCostComparison costcmp;
487 : PathKeysComparison keyscmp;
488 : BMS_Comparison outercmp;
489 :
490 : /*
491 : * Do a fuzzy cost comparison with standard fuzziness limit.
492 : */
493 3985216 : costcmp = compare_path_costs_fuzzily(new_path, old_path,
494 : STD_FUZZ_FACTOR);
495 :
496 : /*
497 : * If the two paths compare differently for startup and total cost,
498 : * then we want to keep both, and we can skip comparing pathkeys and
499 : * required_outer rels. If they compare the same, proceed with the
500 : * other comparisons. Row count is checked last. (We make the tests
501 : * in this order because the cost comparison is most likely to turn
502 : * out "different", and the pathkeys comparison next most likely. As
503 : * explained above, row count very seldom makes a difference, so even
504 : * though it's cheap to compare there's not much point in checking it
505 : * earlier.)
506 : */
507 3985216 : if (costcmp != COSTS_DIFFERENT)
508 : {
509 : /* Similarly check to see if either dominates on pathkeys */
510 : List *old_path_pathkeys;
511 :
512 3865512 : old_path_pathkeys = old_path->param_info ? NIL : old_path->pathkeys;
513 3865512 : keyscmp = compare_pathkeys(new_path_pathkeys,
514 : old_path_pathkeys);
515 3865512 : if (keyscmp != PATHKEYS_DIFFERENT)
516 : {
517 3663644 : switch (costcmp)
518 : {
519 379762 : case COSTS_EQUAL:
520 379762 : outercmp = bms_subset_compare(PATH_REQ_OUTER(new_path),
521 379762 : PATH_REQ_OUTER(old_path));
522 379762 : if (keyscmp == PATHKEYS_BETTER1)
523 : {
524 6128 : if ((outercmp == BMS_EQUAL ||
525 6128 : outercmp == BMS_SUBSET1) &&
526 6128 : new_path->rows <= old_path->rows &&
527 6120 : new_path->parallel_safe >= old_path->parallel_safe)
528 6120 : remove_old = true; /* new dominates old */
529 : }
530 373634 : else if (keyscmp == PATHKEYS_BETTER2)
531 : {
532 17070 : if ((outercmp == BMS_EQUAL ||
533 17070 : outercmp == BMS_SUBSET2) &&
534 17070 : new_path->rows >= old_path->rows &&
535 17070 : new_path->parallel_safe <= old_path->parallel_safe)
536 17070 : accept_new = false; /* old dominates new */
537 : }
538 : else /* keyscmp == PATHKEYS_EQUAL */
539 : {
540 356564 : if (outercmp == BMS_EQUAL)
541 : {
542 : /*
543 : * Same pathkeys and outer rels, and fuzzily
544 : * the same cost, so keep just one; to decide
545 : * which, first check parallel-safety, then
546 : * rows, then do a fuzzy cost comparison with
547 : * very small fuzz limit. (We used to do an
548 : * exact cost comparison, but that results in
549 : * annoying platform-specific plan variations
550 : * due to roundoff in the cost estimates.) If
551 : * things are still tied, arbitrarily keep
552 : * only the old path. Notice that we will
553 : * keep only the old path even if the
554 : * less-fuzzy comparison decides the startup
555 : * and total costs compare differently.
556 : */
557 351666 : if (new_path->parallel_safe >
558 351666 : old_path->parallel_safe)
559 42 : remove_old = true; /* new dominates old */
560 351624 : else if (new_path->parallel_safe <
561 351624 : old_path->parallel_safe)
562 54 : accept_new = false; /* old dominates new */
563 351570 : else if (new_path->rows < old_path->rows)
564 0 : remove_old = true; /* new dominates old */
565 351570 : else if (new_path->rows > old_path->rows)
566 204 : accept_new = false; /* old dominates new */
567 351366 : else if (compare_path_costs_fuzzily(new_path,
568 : old_path,
569 : 1.0000000001) == COSTS_BETTER1)
570 16256 : remove_old = true; /* new dominates old */
571 : else
572 335110 : accept_new = false; /* old equals or
573 : * dominates new */
574 : }
575 4898 : else if (outercmp == BMS_SUBSET1 &&
576 1264 : new_path->rows <= old_path->rows &&
577 1248 : new_path->parallel_safe >= old_path->parallel_safe)
578 1248 : remove_old = true; /* new dominates old */
579 3650 : else if (outercmp == BMS_SUBSET2 &&
580 3006 : new_path->rows >= old_path->rows &&
581 2972 : new_path->parallel_safe <= old_path->parallel_safe)
582 2972 : accept_new = false; /* old dominates new */
583 : /* else different parameterizations, keep both */
584 : }
585 379762 : break;
586 1031050 : case COSTS_BETTER1:
587 1031050 : if (keyscmp != PATHKEYS_BETTER2)
588 : {
589 715136 : outercmp = bms_subset_compare(PATH_REQ_OUTER(new_path),
590 715136 : PATH_REQ_OUTER(old_path));
591 715136 : if ((outercmp == BMS_EQUAL ||
592 609572 : outercmp == BMS_SUBSET1) &&
593 609572 : new_path->rows <= old_path->rows &&
594 605760 : new_path->parallel_safe >= old_path->parallel_safe)
595 603264 : remove_old = true; /* new dominates old */
596 : }
597 1031050 : break;
598 2252832 : case COSTS_BETTER2:
599 2252832 : if (keyscmp != PATHKEYS_BETTER1)
600 : {
601 1435804 : outercmp = bms_subset_compare(PATH_REQ_OUTER(new_path),
602 1435804 : PATH_REQ_OUTER(old_path));
603 1435804 : if ((outercmp == BMS_EQUAL ||
604 1349600 : outercmp == BMS_SUBSET2) &&
605 1349600 : new_path->rows >= old_path->rows &&
606 1292076 : new_path->parallel_safe <= old_path->parallel_safe)
607 1290356 : accept_new = false; /* old dominates new */
608 : }
609 2252832 : break;
610 0 : case COSTS_DIFFERENT:
611 :
612 : /*
613 : * can't get here, but keep this case to keep compiler
614 : * quiet
615 : */
616 0 : break;
617 : }
618 321572 : }
619 : }
620 :
621 : /*
622 : * Remove current element from pathlist if dominated by new.
623 : */
624 3985216 : if (remove_old)
625 : {
626 626930 : parent_rel->pathlist = foreach_delete_current(parent_rel->pathlist,
627 : p1);
628 :
629 : /*
630 : * Delete the data pointed-to by the deleted cell, if possible
631 : */
632 626930 : if (!IsA(old_path, IndexPath))
633 609728 : pfree(old_path);
634 : }
635 : else
636 : {
637 : /*
638 : * new belongs after this old path if it has more disabled nodes
639 : * or if it has the same number of nodes but a greater total cost
640 : */
641 3358286 : if (new_path->disabled_nodes > old_path->disabled_nodes ||
642 3344524 : (new_path->disabled_nodes == old_path->disabled_nodes &&
643 3343660 : new_path->total_cost >= old_path->total_cost))
644 2813310 : insert_at = foreach_current_index(p1) + 1;
645 : }
646 :
647 : /*
648 : * If we found an old path that dominates new_path, we can quit
649 : * scanning the pathlist; we will not add new_path, and we assume
650 : * new_path cannot dominate any other elements of the pathlist.
651 : */
652 3985216 : if (!accept_new)
653 1645766 : break;
654 : }
655 :
656 4374574 : if (accept_new)
657 : {
658 : /* Accept the new path: insert it at proper place in pathlist */
659 2728808 : parent_rel->pathlist =
660 2728808 : list_insert_nth(parent_rel->pathlist, insert_at, new_path);
661 : }
662 : else
663 : {
664 : /* Reject and recycle the new path */
665 1645766 : if (!IsA(new_path, IndexPath))
666 1542880 : pfree(new_path);
667 : }
668 4374574 : }
669 :
670 : /*
671 : * add_path_precheck
672 : * Check whether a proposed new path could possibly get accepted.
673 : * We assume we know the path's pathkeys and parameterization accurately,
674 : * and have lower bounds for its costs.
675 : *
676 : * Note that we do not know the path's rowcount, since getting an estimate for
677 : * that is too expensive to do before prechecking. We assume here that paths
678 : * of a superset parameterization will generate fewer rows; if that holds,
679 : * then paths with different parameterizations cannot dominate each other
680 : * and so we can simply ignore existing paths of another parameterization.
681 : * (In the infrequent cases where that rule of thumb fails, add_path will
682 : * get rid of the inferior path.)
683 : *
684 : * At the time this is called, we haven't actually built a Path structure,
685 : * so the required information has to be passed piecemeal.
686 : */
687 : bool
688 4812308 : add_path_precheck(RelOptInfo *parent_rel, int disabled_nodes,
689 : Cost startup_cost, Cost total_cost,
690 : List *pathkeys, Relids required_outer)
691 : {
692 : List *new_path_pathkeys;
693 : bool consider_startup;
694 : ListCell *p1;
695 :
696 : /* Pretend parameterized paths have no pathkeys, per add_path policy */
697 4812308 : new_path_pathkeys = required_outer ? NIL : pathkeys;
698 :
699 : /* Decide whether new path's startup cost is interesting */
700 4812308 : consider_startup = required_outer ? parent_rel->consider_param_startup : parent_rel->consider_startup;
701 :
702 6293562 : foreach(p1, parent_rel->pathlist)
703 : {
704 5985172 : Path *old_path = (Path *) lfirst(p1);
705 : PathKeysComparison keyscmp;
706 :
707 : /*
708 : * Since the pathlist is sorted by disabled_nodes and then by
709 : * total_cost, we can stop looking once we reach a path with more
710 : * disabled nodes, or the same number of disabled nodes plus a
711 : * total_cost larger than the new path's.
712 : */
713 5985172 : if (unlikely(old_path->disabled_nodes != disabled_nodes))
714 : {
715 11892 : if (disabled_nodes < old_path->disabled_nodes)
716 318 : break;
717 : }
718 5973280 : else if (total_cost <= old_path->total_cost * STD_FUZZ_FACTOR)
719 1705978 : break;
720 :
721 : /*
722 : * We are looking for an old_path with the same parameterization (and
723 : * by assumption the same rowcount) that dominates the new path on
724 : * pathkeys as well as both cost metrics. If we find one, we can
725 : * reject the new path.
726 : *
727 : * Cost comparisons here should match compare_path_costs_fuzzily.
728 : */
729 : /* new path can win on startup cost only if consider_startup */
730 4278876 : if (startup_cost > old_path->startup_cost * STD_FUZZ_FACTOR ||
731 2015520 : !consider_startup)
732 : {
733 : /* new path loses on cost, so check pathkeys... */
734 : List *old_path_pathkeys;
735 :
736 4186206 : old_path_pathkeys = old_path->param_info ? NIL : old_path->pathkeys;
737 4186206 : keyscmp = compare_pathkeys(new_path_pathkeys,
738 : old_path_pathkeys);
739 4186206 : if (keyscmp == PATHKEYS_EQUAL ||
740 : keyscmp == PATHKEYS_BETTER2)
741 : {
742 : /* new path does not win on pathkeys... */
743 2851066 : if (bms_equal(required_outer, PATH_REQ_OUTER(old_path)))
744 : {
745 : /* Found an old path that dominates the new one */
746 2797622 : return false;
747 : }
748 : }
749 : }
750 : }
751 :
752 2014686 : return true;
753 : }
754 :
755 : /*
756 : * add_partial_path
757 : * Like add_path, our goal here is to consider whether a path is worthy
758 : * of being kept around, but the considerations here are a bit different.
759 : * A partial path is one which can be executed in any number of workers in
760 : * parallel such that each worker will generate a subset of the path's
761 : * overall result.
762 : *
763 : * As in add_path, the partial_pathlist is kept sorted with the cheapest
764 : * total path in front. This is depended on by multiple places, which
765 : * just take the front entry as the cheapest path without searching.
766 : *
767 : * We don't generate parameterized partial paths for several reasons. Most
768 : * importantly, they're not safe to execute, because there's nothing to
769 : * make sure that a parallel scan within the parameterized portion of the
770 : * plan is running with the same value in every worker at the same time.
771 : * Fortunately, it seems unlikely to be worthwhile anyway, because having
772 : * each worker scan the entire outer relation and a subset of the inner
773 : * relation will generally be a terrible plan. The inner (parameterized)
774 : * side of the plan will be small anyway. There could be rare cases where
775 : * this wins big - e.g. if join order constraints put a 1-row relation on
776 : * the outer side of the topmost join with a parameterized plan on the inner
777 : * side - but we'll have to be content not to handle such cases until
778 : * somebody builds an executor infrastructure that can cope with them.
779 : *
780 : * Because we don't consider parameterized paths here, we also don't
781 : * need to consider the row counts as a measure of quality: every path will
782 : * produce the same number of rows. Neither do we need to consider startup
783 : * costs: parallelism is only used for plans that will be run to completion.
784 : * Therefore, this routine is much simpler than add_path: it needs to
785 : * consider only disabled nodes, pathkeys and total cost.
786 : *
787 : * As with add_path, we pfree paths that are found to be dominated by
788 : * another partial path; this requires that there be no other references to
789 : * such paths yet. Hence, GatherPaths must not be created for a rel until
790 : * we're done creating all partial paths for it. Unlike add_path, we don't
791 : * take an exception for IndexPaths as partial index paths won't be
792 : * referenced by partial BitmapHeapPaths.
793 : */
794 : void
795 102982 : add_partial_path(RelOptInfo *parent_rel, Path *new_path)
796 : {
797 102982 : bool accept_new = true; /* unless we find a superior old path */
798 102982 : int insert_at = 0; /* where to insert new item */
799 : ListCell *p1;
800 :
801 : /* Check for query cancel. */
802 102982 : CHECK_FOR_INTERRUPTS();
803 :
804 : /* Path to be added must be parallel safe. */
805 : Assert(new_path->parallel_safe);
806 :
807 : /* Relation should be OK for parallelism, too. */
808 : Assert(parent_rel->consider_parallel);
809 :
810 : /*
811 : * As in add_path, throw out any paths which are dominated by the new
812 : * path, but throw out the new path if some existing path dominates it.
813 : */
814 137428 : foreach(p1, parent_rel->partial_pathlist)
815 : {
816 53706 : Path *old_path = (Path *) lfirst(p1);
817 53706 : bool remove_old = false; /* unless new proves superior */
818 : PathKeysComparison keyscmp;
819 :
820 : /* Compare pathkeys. */
821 53706 : keyscmp = compare_pathkeys(new_path->pathkeys, old_path->pathkeys);
822 :
823 : /* Unless pathkeys are incompatible, keep just one of the two paths. */
824 53706 : if (keyscmp != PATHKEYS_DIFFERENT)
825 : {
826 53496 : if (unlikely(new_path->disabled_nodes != old_path->disabled_nodes))
827 : {
828 1460 : if (new_path->disabled_nodes > old_path->disabled_nodes)
829 956 : accept_new = false;
830 : else
831 504 : remove_old = true;
832 : }
833 52036 : else if (new_path->total_cost > old_path->total_cost
834 52036 : * STD_FUZZ_FACTOR)
835 : {
836 : /* New path costs more; keep it only if pathkeys are better. */
837 18256 : if (keyscmp != PATHKEYS_BETTER1)
838 9550 : accept_new = false;
839 : }
840 33780 : else if (old_path->total_cost > new_path->total_cost
841 33780 : * STD_FUZZ_FACTOR)
842 : {
843 : /* Old path costs more; keep it only if pathkeys are better. */
844 24558 : if (keyscmp != PATHKEYS_BETTER2)
845 12682 : remove_old = true;
846 : }
847 9222 : else if (keyscmp == PATHKEYS_BETTER1)
848 : {
849 : /* Costs are about the same, new path has better pathkeys. */
850 0 : remove_old = true;
851 : }
852 9222 : else if (keyscmp == PATHKEYS_BETTER2)
853 : {
854 : /* Costs are about the same, old path has better pathkeys. */
855 1740 : accept_new = false;
856 : }
857 7482 : else if (old_path->total_cost > new_path->total_cost * 1.0000000001)
858 : {
859 : /* Pathkeys are the same, and the old path costs more. */
860 468 : remove_old = true;
861 : }
862 : else
863 : {
864 : /*
865 : * Pathkeys are the same, and new path isn't materially
866 : * cheaper.
867 : */
868 7014 : accept_new = false;
869 : }
870 : }
871 :
872 : /*
873 : * Remove current element from partial_pathlist if dominated by new.
874 : */
875 53706 : if (remove_old)
876 : {
877 13654 : parent_rel->partial_pathlist =
878 13654 : foreach_delete_current(parent_rel->partial_pathlist, p1);
879 13654 : pfree(old_path);
880 : }
881 : else
882 : {
883 : /* new belongs after this old path if it has cost >= old's */
884 40052 : if (new_path->total_cost >= old_path->total_cost)
885 27190 : insert_at = foreach_current_index(p1) + 1;
886 : }
887 :
888 : /*
889 : * If we found an old path that dominates new_path, we can quit
890 : * scanning the partial_pathlist; we will not add new_path, and we
891 : * assume new_path cannot dominate any later path.
892 : */
893 53706 : if (!accept_new)
894 19260 : break;
895 : }
896 :
897 102982 : if (accept_new)
898 : {
899 : /* Accept the new path: insert it at proper place */
900 83722 : parent_rel->partial_pathlist =
901 83722 : list_insert_nth(parent_rel->partial_pathlist, insert_at, new_path);
902 : }
903 : else
904 : {
905 : /* Reject and recycle the new path */
906 19260 : pfree(new_path);
907 : }
908 102982 : }
909 :
910 : /*
911 : * add_partial_path_precheck
912 : * Check whether a proposed new partial path could possibly get accepted.
913 : *
914 : * Unlike add_path_precheck, we can ignore startup cost and parameterization,
915 : * since they don't matter for partial paths (see add_partial_path). But
916 : * we do want to make sure we don't add a partial path if there's already
917 : * a complete path that dominates it, since in that case the proposed path
918 : * is surely a loser.
919 : */
920 : bool
921 83920 : add_partial_path_precheck(RelOptInfo *parent_rel, int disabled_nodes,
922 : Cost total_cost, List *pathkeys)
923 : {
924 : ListCell *p1;
925 :
926 : /*
927 : * Our goal here is twofold. First, we want to find out whether this path
928 : * is clearly inferior to some existing partial path. If so, we want to
929 : * reject it immediately. Second, we want to find out whether this path
930 : * is clearly superior to some existing partial path -- at least, modulo
931 : * final cost computations. If so, we definitely want to consider it.
932 : *
933 : * Unlike add_path(), we always compare pathkeys here. This is because we
934 : * expect partial_pathlist to be very short, and getting a definitive
935 : * answer at this stage avoids the need to call add_path_precheck.
936 : */
937 114300 : foreach(p1, parent_rel->partial_pathlist)
938 : {
939 92718 : Path *old_path = (Path *) lfirst(p1);
940 : PathKeysComparison keyscmp;
941 :
942 92718 : keyscmp = compare_pathkeys(pathkeys, old_path->pathkeys);
943 92718 : if (keyscmp != PATHKEYS_DIFFERENT)
944 : {
945 92526 : if (total_cost > old_path->total_cost * STD_FUZZ_FACTOR &&
946 : keyscmp != PATHKEYS_BETTER1)
947 62338 : return false;
948 44806 : if (old_path->total_cost > total_cost * STD_FUZZ_FACTOR &&
949 : keyscmp != PATHKEYS_BETTER2)
950 14618 : return true;
951 : }
952 : }
953 :
954 : /*
955 : * This path is neither clearly inferior to an existing partial path nor
956 : * clearly good enough that it might replace one. Compare it to
957 : * non-parallel plans. If it loses even before accounting for the cost of
958 : * the Gather node, we should definitely reject it.
959 : *
960 : * Note that we pass the total_cost to add_path_precheck twice. This is
961 : * because it's never advantageous to consider the startup cost of a
962 : * partial path; the resulting plans, if run in parallel, will be run to
963 : * completion.
964 : */
965 21582 : if (!add_path_precheck(parent_rel, disabled_nodes, total_cost, total_cost,
966 : pathkeys, NULL))
967 2518 : return false;
968 :
969 19064 : return true;
970 : }
971 :
972 :
973 : /*****************************************************************************
974 : * PATH NODE CREATION ROUTINES
975 : *****************************************************************************/
976 :
977 : /*
978 : * create_seqscan_path
979 : * Creates a path corresponding to a sequential scan, returning the
980 : * pathnode.
981 : */
982 : Path *
983 427202 : create_seqscan_path(PlannerInfo *root, RelOptInfo *rel,
984 : Relids required_outer, int parallel_workers)
985 : {
986 427202 : Path *pathnode = makeNode(Path);
987 :
988 427202 : pathnode->pathtype = T_SeqScan;
989 427202 : pathnode->parent = rel;
990 427202 : pathnode->pathtarget = rel->reltarget;
991 427202 : pathnode->param_info = get_baserel_parampathinfo(root, rel,
992 : required_outer);
993 427202 : pathnode->parallel_aware = (parallel_workers > 0);
994 427202 : pathnode->parallel_safe = rel->consider_parallel;
995 427202 : pathnode->parallel_workers = parallel_workers;
996 427202 : pathnode->pathkeys = NIL; /* seqscan has unordered result */
997 :
998 427202 : cost_seqscan(pathnode, root, rel, pathnode->param_info);
999 :
1000 427202 : return pathnode;
1001 : }
1002 :
1003 : /*
1004 : * create_samplescan_path
1005 : * Creates a path node for a sampled table scan.
1006 : */
1007 : Path *
1008 306 : create_samplescan_path(PlannerInfo *root, RelOptInfo *rel, Relids required_outer)
1009 : {
1010 306 : Path *pathnode = makeNode(Path);
1011 :
1012 306 : pathnode->pathtype = T_SampleScan;
1013 306 : pathnode->parent = rel;
1014 306 : pathnode->pathtarget = rel->reltarget;
1015 306 : pathnode->param_info = get_baserel_parampathinfo(root, rel,
1016 : required_outer);
1017 306 : pathnode->parallel_aware = false;
1018 306 : pathnode->parallel_safe = rel->consider_parallel;
1019 306 : pathnode->parallel_workers = 0;
1020 306 : pathnode->pathkeys = NIL; /* samplescan has unordered result */
1021 :
1022 306 : cost_samplescan(pathnode, root, rel, pathnode->param_info);
1023 :
1024 306 : return pathnode;
1025 : }
1026 :
1027 : /*
1028 : * create_index_path
1029 : * Creates a path node for an index scan.
1030 : *
1031 : * 'index' is a usable index.
1032 : * 'indexclauses' is a list of IndexClause nodes representing clauses
1033 : * to be enforced as qual conditions in the scan.
1034 : * 'indexorderbys' is a list of bare expressions (no RestrictInfos)
1035 : * to be used as index ordering operators in the scan.
1036 : * 'indexorderbycols' is an integer list of index column numbers (zero based)
1037 : * the ordering operators can be used with.
1038 : * 'pathkeys' describes the ordering of the path.
1039 : * 'indexscandir' is either ForwardScanDirection or BackwardScanDirection.
1040 : * 'indexonly' is true if an index-only scan is wanted.
1041 : * 'required_outer' is the set of outer relids for a parameterized path.
1042 : * 'loop_count' is the number of repetitions of the indexscan to factor into
1043 : * estimates of caching behavior.
1044 : * 'partial_path' is true if constructing a parallel index scan path.
1045 : *
1046 : * Returns the new path node.
1047 : */
1048 : IndexPath *
1049 786312 : create_index_path(PlannerInfo *root,
1050 : IndexOptInfo *index,
1051 : List *indexclauses,
1052 : List *indexorderbys,
1053 : List *indexorderbycols,
1054 : List *pathkeys,
1055 : ScanDirection indexscandir,
1056 : bool indexonly,
1057 : Relids required_outer,
1058 : double loop_count,
1059 : bool partial_path)
1060 : {
1061 786312 : IndexPath *pathnode = makeNode(IndexPath);
1062 786312 : RelOptInfo *rel = index->rel;
1063 :
1064 786312 : pathnode->path.pathtype = indexonly ? T_IndexOnlyScan : T_IndexScan;
1065 786312 : pathnode->path.parent = rel;
1066 786312 : pathnode->path.pathtarget = rel->reltarget;
1067 786312 : pathnode->path.param_info = get_baserel_parampathinfo(root, rel,
1068 : required_outer);
1069 786312 : pathnode->path.parallel_aware = false;
1070 786312 : pathnode->path.parallel_safe = rel->consider_parallel;
1071 786312 : pathnode->path.parallel_workers = 0;
1072 786312 : pathnode->path.pathkeys = pathkeys;
1073 :
1074 786312 : pathnode->indexinfo = index;
1075 786312 : pathnode->indexclauses = indexclauses;
1076 786312 : pathnode->indexorderbys = indexorderbys;
1077 786312 : pathnode->indexorderbycols = indexorderbycols;
1078 786312 : pathnode->indexscandir = indexscandir;
1079 :
1080 786312 : cost_index(pathnode, root, loop_count, partial_path);
1081 :
1082 786312 : return pathnode;
1083 : }
1084 :
1085 : /*
1086 : * create_bitmap_heap_path
1087 : * Creates a path node for a bitmap scan.
1088 : *
1089 : * 'bitmapqual' is a tree of IndexPath, BitmapAndPath, and BitmapOrPath nodes.
1090 : * 'required_outer' is the set of outer relids for a parameterized path.
1091 : * 'loop_count' is the number of repetitions of the indexscan to factor into
1092 : * estimates of caching behavior.
1093 : *
1094 : * loop_count should match the value used when creating the component
1095 : * IndexPaths.
1096 : */
1097 : BitmapHeapPath *
1098 343306 : create_bitmap_heap_path(PlannerInfo *root,
1099 : RelOptInfo *rel,
1100 : Path *bitmapqual,
1101 : Relids required_outer,
1102 : double loop_count,
1103 : int parallel_degree)
1104 : {
1105 343306 : BitmapHeapPath *pathnode = makeNode(BitmapHeapPath);
1106 :
1107 343306 : pathnode->path.pathtype = T_BitmapHeapScan;
1108 343306 : pathnode->path.parent = rel;
1109 343306 : pathnode->path.pathtarget = rel->reltarget;
1110 343306 : pathnode->path.param_info = get_baserel_parampathinfo(root, rel,
1111 : required_outer);
1112 343306 : pathnode->path.parallel_aware = (parallel_degree > 0);
1113 343306 : pathnode->path.parallel_safe = rel->consider_parallel;
1114 343306 : pathnode->path.parallel_workers = parallel_degree;
1115 343306 : pathnode->path.pathkeys = NIL; /* always unordered */
1116 :
1117 343306 : pathnode->bitmapqual = bitmapqual;
1118 :
1119 343306 : cost_bitmap_heap_scan(&pathnode->path, root, rel,
1120 : pathnode->path.param_info,
1121 : bitmapqual, loop_count);
1122 :
1123 343306 : return pathnode;
1124 : }
1125 :
1126 : /*
1127 : * create_bitmap_and_path
1128 : * Creates a path node representing a BitmapAnd.
1129 : */
1130 : BitmapAndPath *
1131 52002 : create_bitmap_and_path(PlannerInfo *root,
1132 : RelOptInfo *rel,
1133 : List *bitmapquals)
1134 : {
1135 52002 : BitmapAndPath *pathnode = makeNode(BitmapAndPath);
1136 52002 : Relids required_outer = NULL;
1137 : ListCell *lc;
1138 :
1139 52002 : pathnode->path.pathtype = T_BitmapAnd;
1140 52002 : pathnode->path.parent = rel;
1141 52002 : pathnode->path.pathtarget = rel->reltarget;
1142 :
1143 : /*
1144 : * Identify the required outer rels as the union of what the child paths
1145 : * depend on. (Alternatively, we could insist that the caller pass this
1146 : * in, but it's more convenient and reliable to compute it here.)
1147 : */
1148 156006 : foreach(lc, bitmapquals)
1149 : {
1150 104004 : Path *bitmapqual = (Path *) lfirst(lc);
1151 :
1152 104004 : required_outer = bms_add_members(required_outer,
1153 104004 : PATH_REQ_OUTER(bitmapqual));
1154 : }
1155 52002 : pathnode->path.param_info = get_baserel_parampathinfo(root, rel,
1156 : required_outer);
1157 :
1158 : /*
1159 : * Currently, a BitmapHeapPath, BitmapAndPath, or BitmapOrPath will be
1160 : * parallel-safe if and only if rel->consider_parallel is set. So, we can
1161 : * set the flag for this path based only on the relation-level flag,
1162 : * without actually iterating over the list of children.
1163 : */
1164 52002 : pathnode->path.parallel_aware = false;
1165 52002 : pathnode->path.parallel_safe = rel->consider_parallel;
1166 52002 : pathnode->path.parallel_workers = 0;
1167 :
1168 52002 : pathnode->path.pathkeys = NIL; /* always unordered */
1169 :
1170 52002 : pathnode->bitmapquals = bitmapquals;
1171 :
1172 : /* this sets bitmapselectivity as well as the regular cost fields: */
1173 52002 : cost_bitmap_and_node(pathnode, root);
1174 :
1175 52002 : return pathnode;
1176 : }
1177 :
1178 : /*
1179 : * create_bitmap_or_path
1180 : * Creates a path node representing a BitmapOr.
1181 : */
1182 : BitmapOrPath *
1183 976 : create_bitmap_or_path(PlannerInfo *root,
1184 : RelOptInfo *rel,
1185 : List *bitmapquals)
1186 : {
1187 976 : BitmapOrPath *pathnode = makeNode(BitmapOrPath);
1188 976 : Relids required_outer = NULL;
1189 : ListCell *lc;
1190 :
1191 976 : pathnode->path.pathtype = T_BitmapOr;
1192 976 : pathnode->path.parent = rel;
1193 976 : pathnode->path.pathtarget = rel->reltarget;
1194 :
1195 : /*
1196 : * Identify the required outer rels as the union of what the child paths
1197 : * depend on. (Alternatively, we could insist that the caller pass this
1198 : * in, but it's more convenient and reliable to compute it here.)
1199 : */
1200 2736 : foreach(lc, bitmapquals)
1201 : {
1202 1760 : Path *bitmapqual = (Path *) lfirst(lc);
1203 :
1204 1760 : required_outer = bms_add_members(required_outer,
1205 1760 : PATH_REQ_OUTER(bitmapqual));
1206 : }
1207 976 : pathnode->path.param_info = get_baserel_parampathinfo(root, rel,
1208 : required_outer);
1209 :
1210 : /*
1211 : * Currently, a BitmapHeapPath, BitmapAndPath, or BitmapOrPath will be
1212 : * parallel-safe if and only if rel->consider_parallel is set. So, we can
1213 : * set the flag for this path based only on the relation-level flag,
1214 : * without actually iterating over the list of children.
1215 : */
1216 976 : pathnode->path.parallel_aware = false;
1217 976 : pathnode->path.parallel_safe = rel->consider_parallel;
1218 976 : pathnode->path.parallel_workers = 0;
1219 :
1220 976 : pathnode->path.pathkeys = NIL; /* always unordered */
1221 :
1222 976 : pathnode->bitmapquals = bitmapquals;
1223 :
1224 : /* this sets bitmapselectivity as well as the regular cost fields: */
1225 976 : cost_bitmap_or_node(pathnode, root);
1226 :
1227 976 : return pathnode;
1228 : }
1229 :
1230 : /*
1231 : * create_tidscan_path
1232 : * Creates a path corresponding to a scan by TID, returning the pathnode.
1233 : */
1234 : TidPath *
1235 852 : create_tidscan_path(PlannerInfo *root, RelOptInfo *rel, List *tidquals,
1236 : Relids required_outer)
1237 : {
1238 852 : TidPath *pathnode = makeNode(TidPath);
1239 :
1240 852 : pathnode->path.pathtype = T_TidScan;
1241 852 : pathnode->path.parent = rel;
1242 852 : pathnode->path.pathtarget = rel->reltarget;
1243 852 : pathnode->path.param_info = get_baserel_parampathinfo(root, rel,
1244 : required_outer);
1245 852 : pathnode->path.parallel_aware = false;
1246 852 : pathnode->path.parallel_safe = rel->consider_parallel;
1247 852 : pathnode->path.parallel_workers = 0;
1248 852 : pathnode->path.pathkeys = NIL; /* always unordered */
1249 :
1250 852 : pathnode->tidquals = tidquals;
1251 :
1252 852 : cost_tidscan(&pathnode->path, root, rel, tidquals,
1253 : pathnode->path.param_info);
1254 :
1255 852 : return pathnode;
1256 : }
1257 :
1258 : /*
1259 : * create_tidrangescan_path
1260 : * Creates a path corresponding to a scan by a range of TIDs, returning
1261 : * the pathnode.
1262 : */
1263 : TidRangePath *
1264 1940 : create_tidrangescan_path(PlannerInfo *root, RelOptInfo *rel,
1265 : List *tidrangequals, Relids required_outer)
1266 : {
1267 1940 : TidRangePath *pathnode = makeNode(TidRangePath);
1268 :
1269 1940 : pathnode->path.pathtype = T_TidRangeScan;
1270 1940 : pathnode->path.parent = rel;
1271 1940 : pathnode->path.pathtarget = rel->reltarget;
1272 1940 : pathnode->path.param_info = get_baserel_parampathinfo(root, rel,
1273 : required_outer);
1274 1940 : pathnode->path.parallel_aware = false;
1275 1940 : pathnode->path.parallel_safe = rel->consider_parallel;
1276 1940 : pathnode->path.parallel_workers = 0;
1277 1940 : pathnode->path.pathkeys = NIL; /* always unordered */
1278 :
1279 1940 : pathnode->tidrangequals = tidrangequals;
1280 :
1281 1940 : cost_tidrangescan(&pathnode->path, root, rel, tidrangequals,
1282 : pathnode->path.param_info);
1283 :
1284 1940 : return pathnode;
1285 : }
1286 :
1287 : /*
1288 : * create_append_path
1289 : * Creates a path corresponding to an Append plan, returning the
1290 : * pathnode.
1291 : *
1292 : * Note that we must handle subpaths = NIL, representing a dummy access path.
1293 : * Also, there are callers that pass root = NULL.
1294 : *
1295 : * 'rows', when passed as a non-negative number, will be used to overwrite the
1296 : * returned path's row estimate. Otherwise, the row estimate is calculated
1297 : * by totalling the row estimates from the 'subpaths' list.
1298 : */
1299 : AppendPath *
1300 76092 : create_append_path(PlannerInfo *root,
1301 : RelOptInfo *rel,
1302 : List *subpaths, List *partial_subpaths,
1303 : List *pathkeys, Relids required_outer,
1304 : int parallel_workers, bool parallel_aware,
1305 : double rows)
1306 : {
1307 76092 : AppendPath *pathnode = makeNode(AppendPath);
1308 : ListCell *l;
1309 :
1310 : Assert(!parallel_aware || parallel_workers > 0);
1311 :
1312 76092 : pathnode->path.pathtype = T_Append;
1313 76092 : pathnode->path.parent = rel;
1314 76092 : pathnode->path.pathtarget = rel->reltarget;
1315 :
1316 : /*
1317 : * If this is for a baserel (not a join or non-leaf partition), we prefer
1318 : * to apply get_baserel_parampathinfo to construct a full ParamPathInfo
1319 : * for the path. This supports building a Memoize path atop this path,
1320 : * and if this is a partitioned table the info may be useful for run-time
1321 : * pruning (cf make_partition_pruneinfo()).
1322 : *
1323 : * However, if we don't have "root" then that won't work and we fall back
1324 : * on the simpler get_appendrel_parampathinfo. There's no point in doing
1325 : * the more expensive thing for a dummy path, either.
1326 : */
1327 76092 : if (rel->reloptkind == RELOPT_BASEREL && root && subpaths != NIL)
1328 37786 : pathnode->path.param_info = get_baserel_parampathinfo(root,
1329 : rel,
1330 : required_outer);
1331 : else
1332 38306 : pathnode->path.param_info = get_appendrel_parampathinfo(rel,
1333 : required_outer);
1334 :
1335 76092 : pathnode->path.parallel_aware = parallel_aware;
1336 76092 : pathnode->path.parallel_safe = rel->consider_parallel;
1337 76092 : pathnode->path.parallel_workers = parallel_workers;
1338 76092 : pathnode->path.pathkeys = pathkeys;
1339 :
1340 : /*
1341 : * For parallel append, non-partial paths are sorted by descending total
1342 : * costs. That way, the total time to finish all non-partial paths is
1343 : * minimized. Also, the partial paths are sorted by descending startup
1344 : * costs. There may be some paths that require to do startup work by a
1345 : * single worker. In such case, it's better for workers to choose the
1346 : * expensive ones first, whereas the leader should choose the cheapest
1347 : * startup plan.
1348 : */
1349 76092 : if (pathnode->path.parallel_aware)
1350 : {
1351 : /*
1352 : * We mustn't fiddle with the order of subpaths when the Append has
1353 : * pathkeys. The order they're listed in is critical to keeping the
1354 : * pathkeys valid.
1355 : */
1356 : Assert(pathkeys == NIL);
1357 :
1358 25718 : list_sort(subpaths, append_total_cost_compare);
1359 25718 : list_sort(partial_subpaths, append_startup_cost_compare);
1360 : }
1361 76092 : pathnode->first_partial_path = list_length(subpaths);
1362 76092 : pathnode->subpaths = list_concat(subpaths, partial_subpaths);
1363 :
1364 : /*
1365 : * Apply query-wide LIMIT if known and path is for sole base relation.
1366 : * (Handling this at this low level is a bit klugy.)
1367 : */
1368 76092 : if (root != NULL && bms_equal(rel->relids, root->all_query_rels))
1369 39078 : pathnode->limit_tuples = root->limit_tuples;
1370 : else
1371 37014 : pathnode->limit_tuples = -1.0;
1372 :
1373 249706 : foreach(l, pathnode->subpaths)
1374 : {
1375 173614 : Path *subpath = (Path *) lfirst(l);
1376 :
1377 308366 : pathnode->path.parallel_safe = pathnode->path.parallel_safe &&
1378 134752 : subpath->parallel_safe;
1379 :
1380 : /* All child paths must have same parameterization */
1381 : Assert(bms_equal(PATH_REQ_OUTER(subpath), required_outer));
1382 : }
1383 :
1384 : Assert(!parallel_aware || pathnode->path.parallel_safe);
1385 :
1386 : /*
1387 : * If there's exactly one child path then the output of the Append is
1388 : * necessarily ordered the same as the child's, so we can inherit the
1389 : * child's pathkeys if any, overriding whatever the caller might've said.
1390 : * Furthermore, if the child's parallel awareness matches the Append's,
1391 : * then the Append is a no-op and will be discarded later (in setrefs.c).
1392 : * Then we can inherit the child's size and cost too, effectively charging
1393 : * zero for the Append. Otherwise, we must do the normal costsize
1394 : * calculation.
1395 : */
1396 76092 : if (list_length(pathnode->subpaths) == 1)
1397 : {
1398 22248 : Path *child = (Path *) linitial(pathnode->subpaths);
1399 :
1400 22248 : if (child->parallel_aware == parallel_aware)
1401 : {
1402 21810 : pathnode->path.rows = child->rows;
1403 21810 : pathnode->path.startup_cost = child->startup_cost;
1404 21810 : pathnode->path.total_cost = child->total_cost;
1405 : }
1406 : else
1407 438 : cost_append(pathnode);
1408 : /* Must do this last, else cost_append complains */
1409 22248 : pathnode->path.pathkeys = child->pathkeys;
1410 : }
1411 : else
1412 53844 : cost_append(pathnode);
1413 :
1414 : /* If the caller provided a row estimate, override the computed value. */
1415 76092 : if (rows >= 0)
1416 576 : pathnode->path.rows = rows;
1417 :
1418 76092 : return pathnode;
1419 : }
1420 :
1421 : /*
1422 : * append_total_cost_compare
1423 : * list_sort comparator for sorting append child paths
1424 : * by total_cost descending
1425 : *
1426 : * For equal total costs, we fall back to comparing startup costs; if those
1427 : * are equal too, break ties using bms_compare on the paths' relids.
1428 : * (This is to avoid getting unpredictable results from list_sort.)
1429 : */
1430 : static int
1431 4596 : append_total_cost_compare(const ListCell *a, const ListCell *b)
1432 : {
1433 4596 : Path *path1 = (Path *) lfirst(a);
1434 4596 : Path *path2 = (Path *) lfirst(b);
1435 : int cmp;
1436 :
1437 4596 : cmp = compare_path_costs(path1, path2, TOTAL_COST);
1438 4596 : if (cmp != 0)
1439 4320 : return -cmp;
1440 276 : return bms_compare(path1->parent->relids, path2->parent->relids);
1441 : }
1442 :
1443 : /*
1444 : * append_startup_cost_compare
1445 : * list_sort comparator for sorting append child paths
1446 : * by startup_cost descending
1447 : *
1448 : * For equal startup costs, we fall back to comparing total costs; if those
1449 : * are equal too, break ties using bms_compare on the paths' relids.
1450 : * (This is to avoid getting unpredictable results from list_sort.)
1451 : */
1452 : static int
1453 34076 : append_startup_cost_compare(const ListCell *a, const ListCell *b)
1454 : {
1455 34076 : Path *path1 = (Path *) lfirst(a);
1456 34076 : Path *path2 = (Path *) lfirst(b);
1457 : int cmp;
1458 :
1459 34076 : cmp = compare_path_costs(path1, path2, STARTUP_COST);
1460 34076 : if (cmp != 0)
1461 13422 : return -cmp;
1462 20654 : return bms_compare(path1->parent->relids, path2->parent->relids);
1463 : }
1464 :
1465 : /*
1466 : * create_merge_append_path
1467 : * Creates a path corresponding to a MergeAppend plan, returning the
1468 : * pathnode.
1469 : */
1470 : MergeAppendPath *
1471 4278 : create_merge_append_path(PlannerInfo *root,
1472 : RelOptInfo *rel,
1473 : List *subpaths,
1474 : List *pathkeys,
1475 : Relids required_outer)
1476 : {
1477 4278 : MergeAppendPath *pathnode = makeNode(MergeAppendPath);
1478 : int input_disabled_nodes;
1479 : Cost input_startup_cost;
1480 : Cost input_total_cost;
1481 : ListCell *l;
1482 :
1483 : /*
1484 : * We don't currently support parameterized MergeAppend paths, as
1485 : * explained in the comments for generate_orderedappend_paths.
1486 : */
1487 : Assert(bms_is_empty(rel->lateral_relids) && bms_is_empty(required_outer));
1488 :
1489 4278 : pathnode->path.pathtype = T_MergeAppend;
1490 4278 : pathnode->path.parent = rel;
1491 4278 : pathnode->path.pathtarget = rel->reltarget;
1492 4278 : pathnode->path.param_info = NULL;
1493 4278 : pathnode->path.parallel_aware = false;
1494 4278 : pathnode->path.parallel_safe = rel->consider_parallel;
1495 4278 : pathnode->path.parallel_workers = 0;
1496 4278 : pathnode->path.pathkeys = pathkeys;
1497 4278 : pathnode->subpaths = subpaths;
1498 :
1499 : /*
1500 : * Apply query-wide LIMIT if known and path is for sole base relation.
1501 : * (Handling this at this low level is a bit klugy.)
1502 : */
1503 4278 : if (bms_equal(rel->relids, root->all_query_rels))
1504 2142 : pathnode->limit_tuples = root->limit_tuples;
1505 : else
1506 2136 : pathnode->limit_tuples = -1.0;
1507 :
1508 : /*
1509 : * Add up the sizes and costs of the input paths.
1510 : */
1511 4278 : pathnode->path.rows = 0;
1512 4278 : input_disabled_nodes = 0;
1513 4278 : input_startup_cost = 0;
1514 4278 : input_total_cost = 0;
1515 15990 : foreach(l, subpaths)
1516 : {
1517 11712 : Path *subpath = (Path *) lfirst(l);
1518 :
1519 : /* All child paths should be unparameterized */
1520 : Assert(bms_is_empty(PATH_REQ_OUTER(subpath)));
1521 :
1522 11712 : pathnode->path.rows += subpath->rows;
1523 20710 : pathnode->path.parallel_safe = pathnode->path.parallel_safe &&
1524 8998 : subpath->parallel_safe;
1525 :
1526 11712 : if (pathkeys_contained_in(pathkeys, subpath->pathkeys))
1527 : {
1528 : /* Subpath is adequately ordered, we won't need to sort it */
1529 11378 : input_disabled_nodes += subpath->disabled_nodes;
1530 11378 : input_startup_cost += subpath->startup_cost;
1531 11378 : input_total_cost += subpath->total_cost;
1532 : }
1533 : else
1534 : {
1535 : /* We'll need to insert a Sort node, so include cost for that */
1536 : Path sort_path; /* dummy for result of cost_sort */
1537 :
1538 334 : cost_sort(&sort_path,
1539 : root,
1540 : pathkeys,
1541 : subpath->disabled_nodes,
1542 : subpath->total_cost,
1543 : subpath->rows,
1544 334 : subpath->pathtarget->width,
1545 : 0.0,
1546 : work_mem,
1547 : pathnode->limit_tuples);
1548 334 : input_disabled_nodes += sort_path.disabled_nodes;
1549 334 : input_startup_cost += sort_path.startup_cost;
1550 334 : input_total_cost += sort_path.total_cost;
1551 : }
1552 : }
1553 :
1554 : /*
1555 : * Now we can compute total costs of the MergeAppend. If there's exactly
1556 : * one child path and its parallel awareness matches that of the
1557 : * MergeAppend, then the MergeAppend is a no-op and will be discarded
1558 : * later (in setrefs.c); otherwise we do the normal cost calculation.
1559 : */
1560 4278 : if (list_length(subpaths) == 1 &&
1561 112 : ((Path *) linitial(subpaths))->parallel_aware ==
1562 112 : pathnode->path.parallel_aware)
1563 : {
1564 112 : pathnode->path.disabled_nodes = input_disabled_nodes;
1565 112 : pathnode->path.startup_cost = input_startup_cost;
1566 112 : pathnode->path.total_cost = input_total_cost;
1567 : }
1568 : else
1569 4166 : cost_merge_append(&pathnode->path, root,
1570 : pathkeys, list_length(subpaths),
1571 : input_disabled_nodes,
1572 : input_startup_cost, input_total_cost,
1573 : pathnode->path.rows);
1574 :
1575 4278 : return pathnode;
1576 : }
1577 :
1578 : /*
1579 : * create_group_result_path
1580 : * Creates a path representing a Result-and-nothing-else plan.
1581 : *
1582 : * This is only used for degenerate grouping cases, in which we know we
1583 : * need to produce one result row, possibly filtered by a HAVING qual.
1584 : */
1585 : GroupResultPath *
1586 196000 : create_group_result_path(PlannerInfo *root, RelOptInfo *rel,
1587 : PathTarget *target, List *havingqual)
1588 : {
1589 196000 : GroupResultPath *pathnode = makeNode(GroupResultPath);
1590 :
1591 196000 : pathnode->path.pathtype = T_Result;
1592 196000 : pathnode->path.parent = rel;
1593 196000 : pathnode->path.pathtarget = target;
1594 196000 : pathnode->path.param_info = NULL; /* there are no other rels... */
1595 196000 : pathnode->path.parallel_aware = false;
1596 196000 : pathnode->path.parallel_safe = rel->consider_parallel;
1597 196000 : pathnode->path.parallel_workers = 0;
1598 196000 : pathnode->path.pathkeys = NIL;
1599 196000 : pathnode->quals = havingqual;
1600 :
1601 : /*
1602 : * We can't quite use cost_resultscan() because the quals we want to
1603 : * account for are not baserestrict quals of the rel. Might as well just
1604 : * hack it here.
1605 : */
1606 196000 : pathnode->path.rows = 1;
1607 196000 : pathnode->path.startup_cost = target->cost.startup;
1608 196000 : pathnode->path.total_cost = target->cost.startup +
1609 196000 : cpu_tuple_cost + target->cost.per_tuple;
1610 :
1611 : /*
1612 : * Add cost of qual, if any --- but we ignore its selectivity, since our
1613 : * rowcount estimate should be 1 no matter what the qual is.
1614 : */
1615 196000 : if (havingqual)
1616 : {
1617 : QualCost qual_cost;
1618 :
1619 602 : cost_qual_eval(&qual_cost, havingqual, root);
1620 : /* havingqual is evaluated once at startup */
1621 602 : pathnode->path.startup_cost += qual_cost.startup + qual_cost.per_tuple;
1622 602 : pathnode->path.total_cost += qual_cost.startup + qual_cost.per_tuple;
1623 : }
1624 :
1625 196000 : return pathnode;
1626 : }
1627 :
1628 : /*
1629 : * create_material_path
1630 : * Creates a path corresponding to a Material plan, returning the
1631 : * pathnode.
1632 : */
1633 : MaterialPath *
1634 532818 : create_material_path(RelOptInfo *rel, Path *subpath)
1635 : {
1636 532818 : MaterialPath *pathnode = makeNode(MaterialPath);
1637 :
1638 : Assert(subpath->parent == rel);
1639 :
1640 532818 : pathnode->path.pathtype = T_Material;
1641 532818 : pathnode->path.parent = rel;
1642 532818 : pathnode->path.pathtarget = rel->reltarget;
1643 532818 : pathnode->path.param_info = subpath->param_info;
1644 532818 : pathnode->path.parallel_aware = false;
1645 1012404 : pathnode->path.parallel_safe = rel->consider_parallel &&
1646 479586 : subpath->parallel_safe;
1647 532818 : pathnode->path.parallel_workers = subpath->parallel_workers;
1648 532818 : pathnode->path.pathkeys = subpath->pathkeys;
1649 :
1650 532818 : pathnode->subpath = subpath;
1651 :
1652 532818 : cost_material(&pathnode->path,
1653 : subpath->disabled_nodes,
1654 : subpath->startup_cost,
1655 : subpath->total_cost,
1656 : subpath->rows,
1657 532818 : subpath->pathtarget->width);
1658 :
1659 532818 : return pathnode;
1660 : }
1661 :
1662 : /*
1663 : * create_memoize_path
1664 : * Creates a path corresponding to a Memoize plan, returning the pathnode.
1665 : */
1666 : MemoizePath *
1667 313198 : create_memoize_path(PlannerInfo *root, RelOptInfo *rel, Path *subpath,
1668 : List *param_exprs, List *hash_operators,
1669 : bool singlerow, bool binary_mode, double calls)
1670 : {
1671 313198 : MemoizePath *pathnode = makeNode(MemoizePath);
1672 :
1673 : Assert(subpath->parent == rel);
1674 :
1675 313198 : pathnode->path.pathtype = T_Memoize;
1676 313198 : pathnode->path.parent = rel;
1677 313198 : pathnode->path.pathtarget = rel->reltarget;
1678 313198 : pathnode->path.param_info = subpath->param_info;
1679 313198 : pathnode->path.parallel_aware = false;
1680 615182 : pathnode->path.parallel_safe = rel->consider_parallel &&
1681 301984 : subpath->parallel_safe;
1682 313198 : pathnode->path.parallel_workers = subpath->parallel_workers;
1683 313198 : pathnode->path.pathkeys = subpath->pathkeys;
1684 :
1685 313198 : pathnode->subpath = subpath;
1686 313198 : pathnode->hash_operators = hash_operators;
1687 313198 : pathnode->param_exprs = param_exprs;
1688 313198 : pathnode->singlerow = singlerow;
1689 313198 : pathnode->binary_mode = binary_mode;
1690 313198 : pathnode->calls = clamp_row_est(calls);
1691 :
1692 : /*
1693 : * For now we set est_entries to 0. cost_memoize_rescan() does all the
1694 : * hard work to determine how many cache entries there are likely to be,
1695 : * so it seems best to leave it up to that function to fill this field in.
1696 : * If left at 0, the executor will make a guess at a good value.
1697 : */
1698 313198 : pathnode->est_entries = 0;
1699 :
1700 : /* we should not generate this path type when enable_memoize=false */
1701 : Assert(enable_memoize);
1702 313198 : pathnode->path.disabled_nodes = subpath->disabled_nodes;
1703 :
1704 : /*
1705 : * Add a small additional charge for caching the first entry. All the
1706 : * harder calculations for rescans are performed in cost_memoize_rescan().
1707 : */
1708 313198 : pathnode->path.startup_cost = subpath->startup_cost + cpu_tuple_cost;
1709 313198 : pathnode->path.total_cost = subpath->total_cost + cpu_tuple_cost;
1710 313198 : pathnode->path.rows = subpath->rows;
1711 :
1712 313198 : return pathnode;
1713 : }
1714 :
1715 : /*
1716 : * create_unique_path
1717 : * Creates a path representing elimination of distinct rows from the
1718 : * input data. Distinct-ness is defined according to the needs of the
1719 : * semijoin represented by sjinfo. If it is not possible to identify
1720 : * how to make the data unique, NULL is returned.
1721 : *
1722 : * If used at all, this is likely to be called repeatedly on the same rel;
1723 : * and the input subpath should always be the same (the cheapest_total path
1724 : * for the rel). So we cache the result.
1725 : */
1726 : UniquePath *
1727 47228 : create_unique_path(PlannerInfo *root, RelOptInfo *rel, Path *subpath,
1728 : SpecialJoinInfo *sjinfo)
1729 : {
1730 : UniquePath *pathnode;
1731 : Path sort_path; /* dummy for result of cost_sort */
1732 : Path agg_path; /* dummy for result of cost_agg */
1733 : MemoryContext oldcontext;
1734 : int numCols;
1735 :
1736 : /* Caller made a mistake if subpath isn't cheapest_total ... */
1737 : Assert(subpath == rel->cheapest_total_path);
1738 : Assert(subpath->parent == rel);
1739 : /* ... or if SpecialJoinInfo is the wrong one */
1740 : Assert(sjinfo->jointype == JOIN_SEMI);
1741 : Assert(bms_equal(rel->relids, sjinfo->syn_righthand));
1742 :
1743 : /* If result already cached, return it */
1744 47228 : if (rel->cheapest_unique_path)
1745 40776 : return (UniquePath *) rel->cheapest_unique_path;
1746 :
1747 : /* If it's not possible to unique-ify, return NULL */
1748 6452 : if (!(sjinfo->semi_can_btree || sjinfo->semi_can_hash))
1749 120 : return NULL;
1750 :
1751 : /*
1752 : * When called during GEQO join planning, we are in a short-lived memory
1753 : * context. We must make sure that the path and any subsidiary data
1754 : * structures created for a baserel survive the GEQO cycle, else the
1755 : * baserel is trashed for future GEQO cycles. On the other hand, when we
1756 : * are creating those for a joinrel during GEQO, we don't want them to
1757 : * clutter the main planning context. Upshot is that the best solution is
1758 : * to explicitly allocate memory in the same context the given RelOptInfo
1759 : * is in.
1760 : */
1761 6332 : oldcontext = MemoryContextSwitchTo(GetMemoryChunkContext(rel));
1762 :
1763 6332 : pathnode = makeNode(UniquePath);
1764 :
1765 6332 : pathnode->path.pathtype = T_Unique;
1766 6332 : pathnode->path.parent = rel;
1767 6332 : pathnode->path.pathtarget = rel->reltarget;
1768 6332 : pathnode->path.param_info = subpath->param_info;
1769 6332 : pathnode->path.parallel_aware = false;
1770 12024 : pathnode->path.parallel_safe = rel->consider_parallel &&
1771 5692 : subpath->parallel_safe;
1772 6332 : pathnode->path.parallel_workers = subpath->parallel_workers;
1773 :
1774 : /*
1775 : * Assume the output is unsorted, since we don't necessarily have pathkeys
1776 : * to represent it. (This might get overridden below.)
1777 : */
1778 6332 : pathnode->path.pathkeys = NIL;
1779 :
1780 6332 : pathnode->subpath = subpath;
1781 :
1782 : /*
1783 : * Under GEQO and when planning child joins, the sjinfo might be
1784 : * short-lived, so we'd better make copies of data structures we extract
1785 : * from it.
1786 : */
1787 6332 : pathnode->in_operators = copyObject(sjinfo->semi_operators);
1788 6332 : pathnode->uniq_exprs = copyObject(sjinfo->semi_rhs_exprs);
1789 :
1790 : /*
1791 : * If the input is a relation and it has a unique index that proves the
1792 : * semi_rhs_exprs are unique, then we don't need to do anything. Note
1793 : * that relation_has_unique_index_for automatically considers restriction
1794 : * clauses for the rel, as well.
1795 : */
1796 7268 : if (rel->rtekind == RTE_RELATION && sjinfo->semi_can_btree &&
1797 936 : relation_has_unique_index_for(root, rel, NIL,
1798 : sjinfo->semi_rhs_exprs,
1799 : sjinfo->semi_operators))
1800 : {
1801 0 : pathnode->umethod = UNIQUE_PATH_NOOP;
1802 0 : pathnode->path.rows = rel->rows;
1803 0 : pathnode->path.disabled_nodes = subpath->disabled_nodes;
1804 0 : pathnode->path.startup_cost = subpath->startup_cost;
1805 0 : pathnode->path.total_cost = subpath->total_cost;
1806 0 : pathnode->path.pathkeys = subpath->pathkeys;
1807 :
1808 0 : rel->cheapest_unique_path = (Path *) pathnode;
1809 :
1810 0 : MemoryContextSwitchTo(oldcontext);
1811 :
1812 0 : return pathnode;
1813 : }
1814 :
1815 : /*
1816 : * If the input is a subquery whose output must be unique already, then we
1817 : * don't need to do anything. The test for uniqueness has to consider
1818 : * exactly which columns we are extracting; for example "SELECT DISTINCT
1819 : * x,y" doesn't guarantee that x alone is distinct. So we cannot check for
1820 : * this optimization unless semi_rhs_exprs consists only of simple Vars
1821 : * referencing subquery outputs. (Possibly we could do something with
1822 : * expressions in the subquery outputs, too, but for now keep it simple.)
1823 : */
1824 6332 : if (rel->rtekind == RTE_SUBQUERY)
1825 : {
1826 3302 : RangeTblEntry *rte = planner_rt_fetch(rel->relid, root);
1827 :
1828 3302 : if (query_supports_distinctness(rte->subquery))
1829 : {
1830 : List *sub_tlist_colnos;
1831 :
1832 3248 : sub_tlist_colnos = translate_sub_tlist(sjinfo->semi_rhs_exprs,
1833 3248 : rel->relid);
1834 :
1835 3444 : if (sub_tlist_colnos &&
1836 196 : query_is_distinct_for(rte->subquery,
1837 : sub_tlist_colnos,
1838 : sjinfo->semi_operators))
1839 : {
1840 0 : pathnode->umethod = UNIQUE_PATH_NOOP;
1841 0 : pathnode->path.rows = rel->rows;
1842 0 : pathnode->path.disabled_nodes = subpath->disabled_nodes;
1843 0 : pathnode->path.startup_cost = subpath->startup_cost;
1844 0 : pathnode->path.total_cost = subpath->total_cost;
1845 0 : pathnode->path.pathkeys = subpath->pathkeys;
1846 :
1847 0 : rel->cheapest_unique_path = (Path *) pathnode;
1848 :
1849 0 : MemoryContextSwitchTo(oldcontext);
1850 :
1851 0 : return pathnode;
1852 : }
1853 : }
1854 : }
1855 :
1856 : /* Estimate number of output rows */
1857 6332 : pathnode->path.rows = estimate_num_groups(root,
1858 : sjinfo->semi_rhs_exprs,
1859 : rel->rows,
1860 : NULL,
1861 : NULL);
1862 6332 : numCols = list_length(sjinfo->semi_rhs_exprs);
1863 :
1864 6332 : if (sjinfo->semi_can_btree)
1865 : {
1866 : /*
1867 : * Estimate cost for sort+unique implementation
1868 : */
1869 6332 : cost_sort(&sort_path, root, NIL,
1870 : subpath->disabled_nodes,
1871 : subpath->total_cost,
1872 : rel->rows,
1873 6332 : subpath->pathtarget->width,
1874 : 0.0,
1875 : work_mem,
1876 : -1.0);
1877 :
1878 : /*
1879 : * Charge one cpu_operator_cost per comparison per input tuple. We
1880 : * assume all columns get compared at most of the tuples. (XXX
1881 : * probably this is an overestimate.) This should agree with
1882 : * create_upper_unique_path.
1883 : */
1884 6332 : sort_path.total_cost += cpu_operator_cost * rel->rows * numCols;
1885 : }
1886 :
1887 6332 : if (sjinfo->semi_can_hash)
1888 : {
1889 : /*
1890 : * Estimate the overhead per hashtable entry at 64 bytes (same as in
1891 : * planner.c).
1892 : */
1893 6332 : int hashentrysize = subpath->pathtarget->width + 64;
1894 :
1895 6332 : if (hashentrysize * pathnode->path.rows > get_hash_memory_limit())
1896 : {
1897 : /*
1898 : * We should not try to hash. Hack the SpecialJoinInfo to
1899 : * remember this, in case we come through here again.
1900 : */
1901 0 : sjinfo->semi_can_hash = false;
1902 : }
1903 : else
1904 6332 : cost_agg(&agg_path, root,
1905 : AGG_HASHED, NULL,
1906 : numCols, pathnode->path.rows,
1907 : NIL,
1908 : subpath->disabled_nodes,
1909 : subpath->startup_cost,
1910 : subpath->total_cost,
1911 : rel->rows,
1912 6332 : subpath->pathtarget->width);
1913 : }
1914 :
1915 6332 : if (sjinfo->semi_can_btree && sjinfo->semi_can_hash)
1916 : {
1917 6332 : if (agg_path.disabled_nodes < sort_path.disabled_nodes ||
1918 6326 : (agg_path.disabled_nodes == sort_path.disabled_nodes &&
1919 6326 : agg_path.total_cost < sort_path.total_cost))
1920 6074 : pathnode->umethod = UNIQUE_PATH_HASH;
1921 : else
1922 258 : pathnode->umethod = UNIQUE_PATH_SORT;
1923 : }
1924 0 : else if (sjinfo->semi_can_btree)
1925 0 : pathnode->umethod = UNIQUE_PATH_SORT;
1926 0 : else if (sjinfo->semi_can_hash)
1927 0 : pathnode->umethod = UNIQUE_PATH_HASH;
1928 : else
1929 : {
1930 : /* we can get here only if we abandoned hashing above */
1931 0 : MemoryContextSwitchTo(oldcontext);
1932 0 : return NULL;
1933 : }
1934 :
1935 6332 : if (pathnode->umethod == UNIQUE_PATH_HASH)
1936 : {
1937 6074 : pathnode->path.disabled_nodes = agg_path.disabled_nodes;
1938 6074 : pathnode->path.startup_cost = agg_path.startup_cost;
1939 6074 : pathnode->path.total_cost = agg_path.total_cost;
1940 : }
1941 : else
1942 : {
1943 258 : pathnode->path.disabled_nodes = sort_path.disabled_nodes;
1944 258 : pathnode->path.startup_cost = sort_path.startup_cost;
1945 258 : pathnode->path.total_cost = sort_path.total_cost;
1946 : }
1947 :
1948 6332 : rel->cheapest_unique_path = (Path *) pathnode;
1949 :
1950 6332 : MemoryContextSwitchTo(oldcontext);
1951 :
1952 6332 : return pathnode;
1953 : }
1954 :
1955 : /*
1956 : * create_gather_merge_path
1957 : *
1958 : * Creates a path corresponding to a gather merge scan, returning
1959 : * the pathnode.
1960 : */
1961 : GatherMergePath *
1962 10190 : create_gather_merge_path(PlannerInfo *root, RelOptInfo *rel, Path *subpath,
1963 : PathTarget *target, List *pathkeys,
1964 : Relids required_outer, double *rows)
1965 : {
1966 10190 : GatherMergePath *pathnode = makeNode(GatherMergePath);
1967 10190 : int input_disabled_nodes = 0;
1968 10190 : Cost input_startup_cost = 0;
1969 10190 : Cost input_total_cost = 0;
1970 :
1971 : Assert(subpath->parallel_safe);
1972 : Assert(pathkeys);
1973 :
1974 : /*
1975 : * The subpath should guarantee that it is adequately ordered either by
1976 : * adding an explicit sort node or by using presorted input. We cannot
1977 : * add an explicit Sort node for the subpath in createplan.c on additional
1978 : * pathkeys, because we can't guarantee the sort would be safe. For
1979 : * example, expressions may be volatile or otherwise parallel unsafe.
1980 : */
1981 10190 : if (!pathkeys_contained_in(pathkeys, subpath->pathkeys))
1982 0 : elog(ERROR, "gather merge input not sufficiently sorted");
1983 :
1984 10190 : pathnode->path.pathtype = T_GatherMerge;
1985 10190 : pathnode->path.parent = rel;
1986 10190 : pathnode->path.param_info = get_baserel_parampathinfo(root, rel,
1987 : required_outer);
1988 10190 : pathnode->path.parallel_aware = false;
1989 :
1990 10190 : pathnode->subpath = subpath;
1991 10190 : pathnode->num_workers = subpath->parallel_workers;
1992 10190 : pathnode->path.pathkeys = pathkeys;
1993 10190 : pathnode->path.pathtarget = target ? target : rel->reltarget;
1994 :
1995 10190 : input_disabled_nodes += subpath->disabled_nodes;
1996 10190 : input_startup_cost += subpath->startup_cost;
1997 10190 : input_total_cost += subpath->total_cost;
1998 :
1999 10190 : cost_gather_merge(pathnode, root, rel, pathnode->path.param_info,
2000 : input_disabled_nodes, input_startup_cost,
2001 : input_total_cost, rows);
2002 :
2003 10190 : return pathnode;
2004 : }
2005 :
2006 : /*
2007 : * translate_sub_tlist - get subquery column numbers represented by tlist
2008 : *
2009 : * The given targetlist usually contains only Vars referencing the given relid.
2010 : * Extract their varattnos (ie, the column numbers of the subquery) and return
2011 : * as an integer List.
2012 : *
2013 : * If any of the tlist items is not a simple Var, we cannot determine whether
2014 : * the subquery's uniqueness condition (if any) matches ours, so punt and
2015 : * return NIL.
2016 : */
2017 : static List *
2018 3248 : translate_sub_tlist(List *tlist, int relid)
2019 : {
2020 3248 : List *result = NIL;
2021 : ListCell *l;
2022 :
2023 3444 : foreach(l, tlist)
2024 : {
2025 3248 : Var *var = (Var *) lfirst(l);
2026 :
2027 3248 : if (!var || !IsA(var, Var) ||
2028 196 : var->varno != relid)
2029 3052 : return NIL; /* punt */
2030 :
2031 196 : result = lappend_int(result, var->varattno);
2032 : }
2033 196 : return result;
2034 : }
2035 :
2036 : /*
2037 : * create_gather_path
2038 : * Creates a path corresponding to a gather scan, returning the
2039 : * pathnode.
2040 : *
2041 : * 'rows' may optionally be set to override row estimates from other sources.
2042 : */
2043 : GatherPath *
2044 19168 : create_gather_path(PlannerInfo *root, RelOptInfo *rel, Path *subpath,
2045 : PathTarget *target, Relids required_outer, double *rows)
2046 : {
2047 19168 : GatherPath *pathnode = makeNode(GatherPath);
2048 :
2049 : Assert(subpath->parallel_safe);
2050 :
2051 19168 : pathnode->path.pathtype = T_Gather;
2052 19168 : pathnode->path.parent = rel;
2053 19168 : pathnode->path.pathtarget = target;
2054 19168 : pathnode->path.param_info = get_baserel_parampathinfo(root, rel,
2055 : required_outer);
2056 19168 : pathnode->path.parallel_aware = false;
2057 19168 : pathnode->path.parallel_safe = false;
2058 19168 : pathnode->path.parallel_workers = 0;
2059 19168 : pathnode->path.pathkeys = NIL; /* Gather has unordered result */
2060 :
2061 19168 : pathnode->subpath = subpath;
2062 19168 : pathnode->num_workers = subpath->parallel_workers;
2063 19168 : pathnode->single_copy = false;
2064 :
2065 19168 : if (pathnode->num_workers == 0)
2066 : {
2067 0 : pathnode->path.pathkeys = subpath->pathkeys;
2068 0 : pathnode->num_workers = 1;
2069 0 : pathnode->single_copy = true;
2070 : }
2071 :
2072 19168 : cost_gather(pathnode, root, rel, pathnode->path.param_info, rows);
2073 :
2074 19168 : return pathnode;
2075 : }
2076 :
2077 : /*
2078 : * create_subqueryscan_path
2079 : * Creates a path corresponding to a scan of a subquery,
2080 : * returning the pathnode.
2081 : *
2082 : * Caller must pass trivial_pathtarget = true if it believes rel->reltarget to
2083 : * be trivial, ie just a fetch of all the subquery output columns in order.
2084 : * While we could determine that here, the caller can usually do it more
2085 : * efficiently (or at least amortize it over multiple calls).
2086 : */
2087 : SubqueryScanPath *
2088 48930 : create_subqueryscan_path(PlannerInfo *root, RelOptInfo *rel, Path *subpath,
2089 : bool trivial_pathtarget,
2090 : List *pathkeys, Relids required_outer)
2091 : {
2092 48930 : SubqueryScanPath *pathnode = makeNode(SubqueryScanPath);
2093 :
2094 48930 : pathnode->path.pathtype = T_SubqueryScan;
2095 48930 : pathnode->path.parent = rel;
2096 48930 : pathnode->path.pathtarget = rel->reltarget;
2097 48930 : pathnode->path.param_info = get_baserel_parampathinfo(root, rel,
2098 : required_outer);
2099 48930 : pathnode->path.parallel_aware = false;
2100 82550 : pathnode->path.parallel_safe = rel->consider_parallel &&
2101 33620 : subpath->parallel_safe;
2102 48930 : pathnode->path.parallel_workers = subpath->parallel_workers;
2103 48930 : pathnode->path.pathkeys = pathkeys;
2104 48930 : pathnode->subpath = subpath;
2105 :
2106 48930 : cost_subqueryscan(pathnode, root, rel, pathnode->path.param_info,
2107 : trivial_pathtarget);
2108 :
2109 48930 : return pathnode;
2110 : }
2111 :
2112 : /*
2113 : * create_functionscan_path
2114 : * Creates a path corresponding to a sequential scan of a function,
2115 : * returning the pathnode.
2116 : */
2117 : Path *
2118 54584 : create_functionscan_path(PlannerInfo *root, RelOptInfo *rel,
2119 : List *pathkeys, Relids required_outer)
2120 : {
2121 54584 : Path *pathnode = makeNode(Path);
2122 :
2123 54584 : pathnode->pathtype = T_FunctionScan;
2124 54584 : pathnode->parent = rel;
2125 54584 : pathnode->pathtarget = rel->reltarget;
2126 54584 : pathnode->param_info = get_baserel_parampathinfo(root, rel,
2127 : required_outer);
2128 54584 : pathnode->parallel_aware = false;
2129 54584 : pathnode->parallel_safe = rel->consider_parallel;
2130 54584 : pathnode->parallel_workers = 0;
2131 54584 : pathnode->pathkeys = pathkeys;
2132 :
2133 54584 : cost_functionscan(pathnode, root, rel, pathnode->param_info);
2134 :
2135 54584 : return pathnode;
2136 : }
2137 :
2138 : /*
2139 : * create_tablefuncscan_path
2140 : * Creates a path corresponding to a sequential scan of a table function,
2141 : * returning the pathnode.
2142 : */
2143 : Path *
2144 626 : create_tablefuncscan_path(PlannerInfo *root, RelOptInfo *rel,
2145 : Relids required_outer)
2146 : {
2147 626 : Path *pathnode = makeNode(Path);
2148 :
2149 626 : pathnode->pathtype = T_TableFuncScan;
2150 626 : pathnode->parent = rel;
2151 626 : pathnode->pathtarget = rel->reltarget;
2152 626 : pathnode->param_info = get_baserel_parampathinfo(root, rel,
2153 : required_outer);
2154 626 : pathnode->parallel_aware = false;
2155 626 : pathnode->parallel_safe = rel->consider_parallel;
2156 626 : pathnode->parallel_workers = 0;
2157 626 : pathnode->pathkeys = NIL; /* result is always unordered */
2158 :
2159 626 : cost_tablefuncscan(pathnode, root, rel, pathnode->param_info);
2160 :
2161 626 : return pathnode;
2162 : }
2163 :
2164 : /*
2165 : * create_valuesscan_path
2166 : * Creates a path corresponding to a scan of a VALUES list,
2167 : * returning the pathnode.
2168 : */
2169 : Path *
2170 8244 : create_valuesscan_path(PlannerInfo *root, RelOptInfo *rel,
2171 : Relids required_outer)
2172 : {
2173 8244 : Path *pathnode = makeNode(Path);
2174 :
2175 8244 : pathnode->pathtype = T_ValuesScan;
2176 8244 : pathnode->parent = rel;
2177 8244 : pathnode->pathtarget = rel->reltarget;
2178 8244 : pathnode->param_info = get_baserel_parampathinfo(root, rel,
2179 : required_outer);
2180 8244 : pathnode->parallel_aware = false;
2181 8244 : pathnode->parallel_safe = rel->consider_parallel;
2182 8244 : pathnode->parallel_workers = 0;
2183 8244 : pathnode->pathkeys = NIL; /* result is always unordered */
2184 :
2185 8244 : cost_valuesscan(pathnode, root, rel, pathnode->param_info);
2186 :
2187 8244 : return pathnode;
2188 : }
2189 :
2190 : /*
2191 : * create_ctescan_path
2192 : * Creates a path corresponding to a scan of a non-self-reference CTE,
2193 : * returning the pathnode.
2194 : */
2195 : Path *
2196 4092 : create_ctescan_path(PlannerInfo *root, RelOptInfo *rel,
2197 : List *pathkeys, Relids required_outer)
2198 : {
2199 4092 : Path *pathnode = makeNode(Path);
2200 :
2201 4092 : pathnode->pathtype = T_CteScan;
2202 4092 : pathnode->parent = rel;
2203 4092 : pathnode->pathtarget = rel->reltarget;
2204 4092 : pathnode->param_info = get_baserel_parampathinfo(root, rel,
2205 : required_outer);
2206 4092 : pathnode->parallel_aware = false;
2207 4092 : pathnode->parallel_safe = rel->consider_parallel;
2208 4092 : pathnode->parallel_workers = 0;
2209 4092 : pathnode->pathkeys = pathkeys;
2210 :
2211 4092 : cost_ctescan(pathnode, root, rel, pathnode->param_info);
2212 :
2213 4092 : return pathnode;
2214 : }
2215 :
2216 : /*
2217 : * create_namedtuplestorescan_path
2218 : * Creates a path corresponding to a scan of a named tuplestore, returning
2219 : * the pathnode.
2220 : */
2221 : Path *
2222 470 : create_namedtuplestorescan_path(PlannerInfo *root, RelOptInfo *rel,
2223 : Relids required_outer)
2224 : {
2225 470 : Path *pathnode = makeNode(Path);
2226 :
2227 470 : pathnode->pathtype = T_NamedTuplestoreScan;
2228 470 : pathnode->parent = rel;
2229 470 : pathnode->pathtarget = rel->reltarget;
2230 470 : pathnode->param_info = get_baserel_parampathinfo(root, rel,
2231 : required_outer);
2232 470 : pathnode->parallel_aware = false;
2233 470 : pathnode->parallel_safe = rel->consider_parallel;
2234 470 : pathnode->parallel_workers = 0;
2235 470 : pathnode->pathkeys = NIL; /* result is always unordered */
2236 :
2237 470 : cost_namedtuplestorescan(pathnode, root, rel, pathnode->param_info);
2238 :
2239 470 : return pathnode;
2240 : }
2241 :
2242 : /*
2243 : * create_resultscan_path
2244 : * Creates a path corresponding to a scan of an RTE_RESULT relation,
2245 : * returning the pathnode.
2246 : */
2247 : Path *
2248 4244 : create_resultscan_path(PlannerInfo *root, RelOptInfo *rel,
2249 : Relids required_outer)
2250 : {
2251 4244 : Path *pathnode = makeNode(Path);
2252 :
2253 4244 : pathnode->pathtype = T_Result;
2254 4244 : pathnode->parent = rel;
2255 4244 : pathnode->pathtarget = rel->reltarget;
2256 4244 : pathnode->param_info = get_baserel_parampathinfo(root, rel,
2257 : required_outer);
2258 4244 : pathnode->parallel_aware = false;
2259 4244 : pathnode->parallel_safe = rel->consider_parallel;
2260 4244 : pathnode->parallel_workers = 0;
2261 4244 : pathnode->pathkeys = NIL; /* result is always unordered */
2262 :
2263 4244 : cost_resultscan(pathnode, root, rel, pathnode->param_info);
2264 :
2265 4244 : return pathnode;
2266 : }
2267 :
2268 : /*
2269 : * create_worktablescan_path
2270 : * Creates a path corresponding to a scan of a self-reference CTE,
2271 : * returning the pathnode.
2272 : */
2273 : Path *
2274 1012 : create_worktablescan_path(PlannerInfo *root, RelOptInfo *rel,
2275 : Relids required_outer)
2276 : {
2277 1012 : Path *pathnode = makeNode(Path);
2278 :
2279 1012 : pathnode->pathtype = T_WorkTableScan;
2280 1012 : pathnode->parent = rel;
2281 1012 : pathnode->pathtarget = rel->reltarget;
2282 1012 : pathnode->param_info = get_baserel_parampathinfo(root, rel,
2283 : required_outer);
2284 1012 : pathnode->parallel_aware = false;
2285 1012 : pathnode->parallel_safe = rel->consider_parallel;
2286 1012 : pathnode->parallel_workers = 0;
2287 1012 : pathnode->pathkeys = NIL; /* result is always unordered */
2288 :
2289 : /* Cost is the same as for a regular CTE scan */
2290 1012 : cost_ctescan(pathnode, root, rel, pathnode->param_info);
2291 :
2292 1012 : return pathnode;
2293 : }
2294 :
2295 : /*
2296 : * create_foreignscan_path
2297 : * Creates a path corresponding to a scan of a foreign base table,
2298 : * returning the pathnode.
2299 : *
2300 : * This function is never called from core Postgres; rather, it's expected
2301 : * to be called by the GetForeignPaths function of a foreign data wrapper.
2302 : * We make the FDW supply all fields of the path, since we do not have any way
2303 : * to calculate them in core. However, there is a usually-sane default for
2304 : * the pathtarget (rel->reltarget), so we let a NULL for "target" select that.
2305 : */
2306 : ForeignPath *
2307 3648 : create_foreignscan_path(PlannerInfo *root, RelOptInfo *rel,
2308 : PathTarget *target,
2309 : double rows, int disabled_nodes,
2310 : Cost startup_cost, Cost total_cost,
2311 : List *pathkeys,
2312 : Relids required_outer,
2313 : Path *fdw_outerpath,
2314 : List *fdw_restrictinfo,
2315 : List *fdw_private)
2316 : {
2317 3648 : ForeignPath *pathnode = makeNode(ForeignPath);
2318 :
2319 : /* Historically some FDWs were confused about when to use this */
2320 : Assert(IS_SIMPLE_REL(rel));
2321 :
2322 3648 : pathnode->path.pathtype = T_ForeignScan;
2323 3648 : pathnode->path.parent = rel;
2324 3648 : pathnode->path.pathtarget = target ? target : rel->reltarget;
2325 3648 : pathnode->path.param_info = get_baserel_parampathinfo(root, rel,
2326 : required_outer);
2327 3648 : pathnode->path.parallel_aware = false;
2328 3648 : pathnode->path.parallel_safe = rel->consider_parallel;
2329 3648 : pathnode->path.parallel_workers = 0;
2330 3648 : pathnode->path.rows = rows;
2331 3648 : pathnode->path.disabled_nodes = disabled_nodes;
2332 3648 : pathnode->path.startup_cost = startup_cost;
2333 3648 : pathnode->path.total_cost = total_cost;
2334 3648 : pathnode->path.pathkeys = pathkeys;
2335 :
2336 3648 : pathnode->fdw_outerpath = fdw_outerpath;
2337 3648 : pathnode->fdw_restrictinfo = fdw_restrictinfo;
2338 3648 : pathnode->fdw_private = fdw_private;
2339 :
2340 3648 : return pathnode;
2341 : }
2342 :
2343 : /*
2344 : * create_foreign_join_path
2345 : * Creates a path corresponding to a scan of a foreign join,
2346 : * returning the pathnode.
2347 : *
2348 : * This function is never called from core Postgres; rather, it's expected
2349 : * to be called by the GetForeignJoinPaths function of a foreign data wrapper.
2350 : * We make the FDW supply all fields of the path, since we do not have any way
2351 : * to calculate them in core. However, there is a usually-sane default for
2352 : * the pathtarget (rel->reltarget), so we let a NULL for "target" select that.
2353 : */
2354 : ForeignPath *
2355 1200 : create_foreign_join_path(PlannerInfo *root, RelOptInfo *rel,
2356 : PathTarget *target,
2357 : double rows, int disabled_nodes,
2358 : Cost startup_cost, Cost total_cost,
2359 : List *pathkeys,
2360 : Relids required_outer,
2361 : Path *fdw_outerpath,
2362 : List *fdw_restrictinfo,
2363 : List *fdw_private)
2364 : {
2365 1200 : ForeignPath *pathnode = makeNode(ForeignPath);
2366 :
2367 : /*
2368 : * We should use get_joinrel_parampathinfo to handle parameterized paths,
2369 : * but the API of this function doesn't support it, and existing
2370 : * extensions aren't yet trying to build such paths anyway. For the
2371 : * moment just throw an error if someone tries it; eventually we should
2372 : * revisit this.
2373 : */
2374 1200 : if (!bms_is_empty(required_outer) || !bms_is_empty(rel->lateral_relids))
2375 0 : elog(ERROR, "parameterized foreign joins are not supported yet");
2376 :
2377 1200 : pathnode->path.pathtype = T_ForeignScan;
2378 1200 : pathnode->path.parent = rel;
2379 1200 : pathnode->path.pathtarget = target ? target : rel->reltarget;
2380 1200 : pathnode->path.param_info = NULL; /* XXX see above */
2381 1200 : pathnode->path.parallel_aware = false;
2382 1200 : pathnode->path.parallel_safe = rel->consider_parallel;
2383 1200 : pathnode->path.parallel_workers = 0;
2384 1200 : pathnode->path.rows = rows;
2385 1200 : pathnode->path.disabled_nodes = disabled_nodes;
2386 1200 : pathnode->path.startup_cost = startup_cost;
2387 1200 : pathnode->path.total_cost = total_cost;
2388 1200 : pathnode->path.pathkeys = pathkeys;
2389 :
2390 1200 : pathnode->fdw_outerpath = fdw_outerpath;
2391 1200 : pathnode->fdw_restrictinfo = fdw_restrictinfo;
2392 1200 : pathnode->fdw_private = fdw_private;
2393 :
2394 1200 : return pathnode;
2395 : }
2396 :
2397 : /*
2398 : * create_foreign_upper_path
2399 : * Creates a path corresponding to an upper relation that's computed
2400 : * directly by an FDW, returning the pathnode.
2401 : *
2402 : * This function is never called from core Postgres; rather, it's expected to
2403 : * be called by the GetForeignUpperPaths function of a foreign data wrapper.
2404 : * We make the FDW supply all fields of the path, since we do not have any way
2405 : * to calculate them in core. However, there is a usually-sane default for
2406 : * the pathtarget (rel->reltarget), so we let a NULL for "target" select that.
2407 : */
2408 : ForeignPath *
2409 586 : create_foreign_upper_path(PlannerInfo *root, RelOptInfo *rel,
2410 : PathTarget *target,
2411 : double rows, int disabled_nodes,
2412 : Cost startup_cost, Cost total_cost,
2413 : List *pathkeys,
2414 : Path *fdw_outerpath,
2415 : List *fdw_restrictinfo,
2416 : List *fdw_private)
2417 : {
2418 586 : ForeignPath *pathnode = makeNode(ForeignPath);
2419 :
2420 : /*
2421 : * Upper relations should never have any lateral references, since joining
2422 : * is complete.
2423 : */
2424 : Assert(bms_is_empty(rel->lateral_relids));
2425 :
2426 586 : pathnode->path.pathtype = T_ForeignScan;
2427 586 : pathnode->path.parent = rel;
2428 586 : pathnode->path.pathtarget = target ? target : rel->reltarget;
2429 586 : pathnode->path.param_info = NULL;
2430 586 : pathnode->path.parallel_aware = false;
2431 586 : pathnode->path.parallel_safe = rel->consider_parallel;
2432 586 : pathnode->path.parallel_workers = 0;
2433 586 : pathnode->path.rows = rows;
2434 586 : pathnode->path.disabled_nodes = disabled_nodes;
2435 586 : pathnode->path.startup_cost = startup_cost;
2436 586 : pathnode->path.total_cost = total_cost;
2437 586 : pathnode->path.pathkeys = pathkeys;
2438 :
2439 586 : pathnode->fdw_outerpath = fdw_outerpath;
2440 586 : pathnode->fdw_restrictinfo = fdw_restrictinfo;
2441 586 : pathnode->fdw_private = fdw_private;
2442 :
2443 586 : return pathnode;
2444 : }
2445 :
2446 : /*
2447 : * calc_nestloop_required_outer
2448 : * Compute the required_outer set for a nestloop join path
2449 : *
2450 : * Note: when considering a child join, the inputs nonetheless use top-level
2451 : * parent relids
2452 : *
2453 : * Note: result must not share storage with either input
2454 : */
2455 : Relids
2456 3170834 : calc_nestloop_required_outer(Relids outerrelids,
2457 : Relids outer_paramrels,
2458 : Relids innerrelids,
2459 : Relids inner_paramrels)
2460 : {
2461 : Relids required_outer;
2462 :
2463 : /* inner_path can require rels from outer path, but not vice versa */
2464 : Assert(!bms_overlap(outer_paramrels, innerrelids));
2465 : /* easy case if inner path is not parameterized */
2466 3170834 : if (!inner_paramrels)
2467 2141958 : return bms_copy(outer_paramrels);
2468 : /* else, form the union ... */
2469 1028876 : required_outer = bms_union(outer_paramrels, inner_paramrels);
2470 : /* ... and remove any mention of now-satisfied outer rels */
2471 1028876 : required_outer = bms_del_members(required_outer,
2472 : outerrelids);
2473 1028876 : return required_outer;
2474 : }
2475 :
2476 : /*
2477 : * calc_non_nestloop_required_outer
2478 : * Compute the required_outer set for a merge or hash join path
2479 : *
2480 : * Note: result must not share storage with either input
2481 : */
2482 : Relids
2483 2097212 : calc_non_nestloop_required_outer(Path *outer_path, Path *inner_path)
2484 : {
2485 2097212 : Relids outer_paramrels = PATH_REQ_OUTER(outer_path);
2486 2097212 : Relids inner_paramrels = PATH_REQ_OUTER(inner_path);
2487 : Relids innerrelids PG_USED_FOR_ASSERTS_ONLY;
2488 : Relids outerrelids PG_USED_FOR_ASSERTS_ONLY;
2489 : Relids required_outer;
2490 :
2491 : /*
2492 : * Any parameterization of the input paths refers to topmost parents of
2493 : * the relevant relations, because reparameterize_path_by_child() hasn't
2494 : * been called yet. So we must consider topmost parents of the relations
2495 : * being joined, too, while checking for disallowed parameterization
2496 : * cases.
2497 : */
2498 2097212 : if (inner_path->parent->top_parent_relids)
2499 37526 : innerrelids = inner_path->parent->top_parent_relids;
2500 : else
2501 2059686 : innerrelids = inner_path->parent->relids;
2502 :
2503 2097212 : if (outer_path->parent->top_parent_relids)
2504 37526 : outerrelids = outer_path->parent->top_parent_relids;
2505 : else
2506 2059686 : outerrelids = outer_path->parent->relids;
2507 :
2508 : /* neither path can require rels from the other */
2509 : Assert(!bms_overlap(outer_paramrels, innerrelids));
2510 : Assert(!bms_overlap(inner_paramrels, outerrelids));
2511 : /* form the union ... */
2512 2097212 : required_outer = bms_union(outer_paramrels, inner_paramrels);
2513 : /* we do not need an explicit test for empty; bms_union gets it right */
2514 2097212 : return required_outer;
2515 : }
2516 :
2517 : /*
2518 : * create_nestloop_path
2519 : * Creates a pathnode corresponding to a nestloop join between two
2520 : * relations.
2521 : *
2522 : * 'joinrel' is the join relation.
2523 : * 'jointype' is the type of join required
2524 : * 'workspace' is the result from initial_cost_nestloop
2525 : * 'extra' contains various information about the join
2526 : * 'outer_path' is the outer path
2527 : * 'inner_path' is the inner path
2528 : * 'restrict_clauses' are the RestrictInfo nodes to apply at the join
2529 : * 'pathkeys' are the path keys of the new join path
2530 : * 'required_outer' is the set of required outer rels
2531 : *
2532 : * Returns the resulting path node.
2533 : */
2534 : NestPath *
2535 1400428 : create_nestloop_path(PlannerInfo *root,
2536 : RelOptInfo *joinrel,
2537 : JoinType jointype,
2538 : JoinCostWorkspace *workspace,
2539 : JoinPathExtraData *extra,
2540 : Path *outer_path,
2541 : Path *inner_path,
2542 : List *restrict_clauses,
2543 : List *pathkeys,
2544 : Relids required_outer)
2545 : {
2546 1400428 : NestPath *pathnode = makeNode(NestPath);
2547 1400428 : Relids inner_req_outer = PATH_REQ_OUTER(inner_path);
2548 : Relids outerrelids;
2549 :
2550 : /*
2551 : * Paths are parameterized by top-level parents, so run parameterization
2552 : * tests on the parent relids.
2553 : */
2554 1400428 : if (outer_path->parent->top_parent_relids)
2555 18748 : outerrelids = outer_path->parent->top_parent_relids;
2556 : else
2557 1381680 : outerrelids = outer_path->parent->relids;
2558 :
2559 : /*
2560 : * If the inner path is parameterized by the outer, we must drop any
2561 : * restrict_clauses that are due to be moved into the inner path. We have
2562 : * to do this now, rather than postpone the work till createplan time,
2563 : * because the restrict_clauses list can affect the size and cost
2564 : * estimates for this path. We detect such clauses by checking for serial
2565 : * number match to clauses already enforced in the inner path.
2566 : */
2567 1400428 : if (bms_overlap(inner_req_outer, outerrelids))
2568 : {
2569 387606 : Bitmapset *enforced_serials = get_param_path_clause_serials(inner_path);
2570 387606 : List *jclauses = NIL;
2571 : ListCell *lc;
2572 :
2573 860446 : foreach(lc, restrict_clauses)
2574 : {
2575 472840 : RestrictInfo *rinfo = (RestrictInfo *) lfirst(lc);
2576 :
2577 472840 : if (!bms_is_member(rinfo->rinfo_serial, enforced_serials))
2578 64390 : jclauses = lappend(jclauses, rinfo);
2579 : }
2580 387606 : restrict_clauses = jclauses;
2581 : }
2582 :
2583 1400428 : pathnode->jpath.path.pathtype = T_NestLoop;
2584 1400428 : pathnode->jpath.path.parent = joinrel;
2585 1400428 : pathnode->jpath.path.pathtarget = joinrel->reltarget;
2586 1400428 : pathnode->jpath.path.param_info =
2587 1400428 : get_joinrel_parampathinfo(root,
2588 : joinrel,
2589 : outer_path,
2590 : inner_path,
2591 : extra->sjinfo,
2592 : required_outer,
2593 : &restrict_clauses);
2594 1400428 : pathnode->jpath.path.parallel_aware = false;
2595 4082004 : pathnode->jpath.path.parallel_safe = joinrel->consider_parallel &&
2596 1400428 : outer_path->parallel_safe && inner_path->parallel_safe;
2597 : /* This is a foolish way to estimate parallel_workers, but for now... */
2598 1400428 : pathnode->jpath.path.parallel_workers = outer_path->parallel_workers;
2599 1400428 : pathnode->jpath.path.pathkeys = pathkeys;
2600 1400428 : pathnode->jpath.jointype = jointype;
2601 1400428 : pathnode->jpath.inner_unique = extra->inner_unique;
2602 1400428 : pathnode->jpath.outerjoinpath = outer_path;
2603 1400428 : pathnode->jpath.innerjoinpath = inner_path;
2604 1400428 : pathnode->jpath.joinrestrictinfo = restrict_clauses;
2605 :
2606 1400428 : final_cost_nestloop(root, pathnode, workspace, extra);
2607 :
2608 1400428 : return pathnode;
2609 : }
2610 :
2611 : /*
2612 : * create_mergejoin_path
2613 : * Creates a pathnode corresponding to a mergejoin join between
2614 : * two relations
2615 : *
2616 : * 'joinrel' is the join relation
2617 : * 'jointype' is the type of join required
2618 : * 'workspace' is the result from initial_cost_mergejoin
2619 : * 'extra' contains various information about the join
2620 : * 'outer_path' is the outer path
2621 : * 'inner_path' is the inner path
2622 : * 'restrict_clauses' are the RestrictInfo nodes to apply at the join
2623 : * 'pathkeys' are the path keys of the new join path
2624 : * 'required_outer' is the set of required outer rels
2625 : * 'mergeclauses' are the RestrictInfo nodes to use as merge clauses
2626 : * (this should be a subset of the restrict_clauses list)
2627 : * 'outersortkeys' are the sort varkeys for the outer relation
2628 : * 'innersortkeys' are the sort varkeys for the inner relation
2629 : * 'outer_presorted_keys' is the number of presorted keys of the outer path
2630 : */
2631 : MergePath *
2632 331908 : create_mergejoin_path(PlannerInfo *root,
2633 : RelOptInfo *joinrel,
2634 : JoinType jointype,
2635 : JoinCostWorkspace *workspace,
2636 : JoinPathExtraData *extra,
2637 : Path *outer_path,
2638 : Path *inner_path,
2639 : List *restrict_clauses,
2640 : List *pathkeys,
2641 : Relids required_outer,
2642 : List *mergeclauses,
2643 : List *outersortkeys,
2644 : List *innersortkeys,
2645 : int outer_presorted_keys)
2646 : {
2647 331908 : MergePath *pathnode = makeNode(MergePath);
2648 :
2649 331908 : pathnode->jpath.path.pathtype = T_MergeJoin;
2650 331908 : pathnode->jpath.path.parent = joinrel;
2651 331908 : pathnode->jpath.path.pathtarget = joinrel->reltarget;
2652 331908 : pathnode->jpath.path.param_info =
2653 331908 : get_joinrel_parampathinfo(root,
2654 : joinrel,
2655 : outer_path,
2656 : inner_path,
2657 : extra->sjinfo,
2658 : required_outer,
2659 : &restrict_clauses);
2660 331908 : pathnode->jpath.path.parallel_aware = false;
2661 960100 : pathnode->jpath.path.parallel_safe = joinrel->consider_parallel &&
2662 331908 : outer_path->parallel_safe && inner_path->parallel_safe;
2663 : /* This is a foolish way to estimate parallel_workers, but for now... */
2664 331908 : pathnode->jpath.path.parallel_workers = outer_path->parallel_workers;
2665 331908 : pathnode->jpath.path.pathkeys = pathkeys;
2666 331908 : pathnode->jpath.jointype = jointype;
2667 331908 : pathnode->jpath.inner_unique = extra->inner_unique;
2668 331908 : pathnode->jpath.outerjoinpath = outer_path;
2669 331908 : pathnode->jpath.innerjoinpath = inner_path;
2670 331908 : pathnode->jpath.joinrestrictinfo = restrict_clauses;
2671 331908 : pathnode->path_mergeclauses = mergeclauses;
2672 331908 : pathnode->outersortkeys = outersortkeys;
2673 331908 : pathnode->innersortkeys = innersortkeys;
2674 331908 : pathnode->outer_presorted_keys = outer_presorted_keys;
2675 : /* pathnode->skip_mark_restore will be set by final_cost_mergejoin */
2676 : /* pathnode->materialize_inner will be set by final_cost_mergejoin */
2677 :
2678 331908 : final_cost_mergejoin(root, pathnode, workspace, extra);
2679 :
2680 331908 : return pathnode;
2681 : }
2682 :
2683 : /*
2684 : * create_hashjoin_path
2685 : * Creates a pathnode corresponding to a hash join between two relations.
2686 : *
2687 : * 'joinrel' is the join relation
2688 : * 'jointype' is the type of join required
2689 : * 'workspace' is the result from initial_cost_hashjoin
2690 : * 'extra' contains various information about the join
2691 : * 'outer_path' is the cheapest outer path
2692 : * 'inner_path' is the cheapest inner path
2693 : * 'parallel_hash' to select Parallel Hash of inner path (shared hash table)
2694 : * 'restrict_clauses' are the RestrictInfo nodes to apply at the join
2695 : * 'required_outer' is the set of required outer rels
2696 : * 'hashclauses' are the RestrictInfo nodes to use as hash clauses
2697 : * (this should be a subset of the restrict_clauses list)
2698 : */
2699 : HashPath *
2700 296968 : create_hashjoin_path(PlannerInfo *root,
2701 : RelOptInfo *joinrel,
2702 : JoinType jointype,
2703 : JoinCostWorkspace *workspace,
2704 : JoinPathExtraData *extra,
2705 : Path *outer_path,
2706 : Path *inner_path,
2707 : bool parallel_hash,
2708 : List *restrict_clauses,
2709 : Relids required_outer,
2710 : List *hashclauses)
2711 : {
2712 296968 : HashPath *pathnode = makeNode(HashPath);
2713 :
2714 296968 : pathnode->jpath.path.pathtype = T_HashJoin;
2715 296968 : pathnode->jpath.path.parent = joinrel;
2716 296968 : pathnode->jpath.path.pathtarget = joinrel->reltarget;
2717 296968 : pathnode->jpath.path.param_info =
2718 296968 : get_joinrel_parampathinfo(root,
2719 : joinrel,
2720 : outer_path,
2721 : inner_path,
2722 : extra->sjinfo,
2723 : required_outer,
2724 : &restrict_clauses);
2725 296968 : pathnode->jpath.path.parallel_aware =
2726 296968 : joinrel->consider_parallel && parallel_hash;
2727 856184 : pathnode->jpath.path.parallel_safe = joinrel->consider_parallel &&
2728 296968 : outer_path->parallel_safe && inner_path->parallel_safe;
2729 : /* This is a foolish way to estimate parallel_workers, but for now... */
2730 296968 : pathnode->jpath.path.parallel_workers = outer_path->parallel_workers;
2731 :
2732 : /*
2733 : * A hashjoin never has pathkeys, since its output ordering is
2734 : * unpredictable due to possible batching. XXX If the inner relation is
2735 : * small enough, we could instruct the executor that it must not batch,
2736 : * and then we could assume that the output inherits the outer relation's
2737 : * ordering, which might save a sort step. However there is considerable
2738 : * downside if our estimate of the inner relation size is badly off. For
2739 : * the moment we don't risk it. (Note also that if we wanted to take this
2740 : * seriously, joinpath.c would have to consider many more paths for the
2741 : * outer rel than it does now.)
2742 : */
2743 296968 : pathnode->jpath.path.pathkeys = NIL;
2744 296968 : pathnode->jpath.jointype = jointype;
2745 296968 : pathnode->jpath.inner_unique = extra->inner_unique;
2746 296968 : pathnode->jpath.outerjoinpath = outer_path;
2747 296968 : pathnode->jpath.innerjoinpath = inner_path;
2748 296968 : pathnode->jpath.joinrestrictinfo = restrict_clauses;
2749 296968 : pathnode->path_hashclauses = hashclauses;
2750 : /* final_cost_hashjoin will fill in pathnode->num_batches */
2751 :
2752 296968 : final_cost_hashjoin(root, pathnode, workspace, extra);
2753 :
2754 296968 : return pathnode;
2755 : }
2756 :
2757 : /*
2758 : * create_projection_path
2759 : * Creates a pathnode that represents performing a projection.
2760 : *
2761 : * 'rel' is the parent relation associated with the result
2762 : * 'subpath' is the path representing the source of data
2763 : * 'target' is the PathTarget to be computed
2764 : */
2765 : ProjectionPath *
2766 384656 : create_projection_path(PlannerInfo *root,
2767 : RelOptInfo *rel,
2768 : Path *subpath,
2769 : PathTarget *target)
2770 : {
2771 384656 : ProjectionPath *pathnode = makeNode(ProjectionPath);
2772 : PathTarget *oldtarget;
2773 :
2774 : /*
2775 : * We mustn't put a ProjectionPath directly above another; it's useless
2776 : * and will confuse create_projection_plan. Rather than making sure all
2777 : * callers handle that, let's implement it here, by stripping off any
2778 : * ProjectionPath in what we're given. Given this rule, there won't be
2779 : * more than one.
2780 : */
2781 384656 : if (IsA(subpath, ProjectionPath))
2782 : {
2783 12 : ProjectionPath *subpp = (ProjectionPath *) subpath;
2784 :
2785 : Assert(subpp->path.parent == rel);
2786 12 : subpath = subpp->subpath;
2787 : Assert(!IsA(subpath, ProjectionPath));
2788 : }
2789 :
2790 384656 : pathnode->path.pathtype = T_Result;
2791 384656 : pathnode->path.parent = rel;
2792 384656 : pathnode->path.pathtarget = target;
2793 : /* For now, assume we are above any joins, so no parameterization */
2794 384656 : pathnode->path.param_info = NULL;
2795 384656 : pathnode->path.parallel_aware = false;
2796 876120 : pathnode->path.parallel_safe = rel->consider_parallel &&
2797 491072 : subpath->parallel_safe &&
2798 106416 : is_parallel_safe(root, (Node *) target->exprs);
2799 384656 : pathnode->path.parallel_workers = subpath->parallel_workers;
2800 : /* Projection does not change the sort order */
2801 384656 : pathnode->path.pathkeys = subpath->pathkeys;
2802 :
2803 384656 : pathnode->subpath = subpath;
2804 :
2805 : /*
2806 : * We might not need a separate Result node. If the input plan node type
2807 : * can project, we can just tell it to project something else. Or, if it
2808 : * can't project but the desired target has the same expression list as
2809 : * what the input will produce anyway, we can still give it the desired
2810 : * tlist (possibly changing its ressortgroupref labels, but nothing else).
2811 : * Note: in the latter case, create_projection_plan has to recheck our
2812 : * conclusion; see comments therein.
2813 : */
2814 384656 : oldtarget = subpath->pathtarget;
2815 386958 : if (is_projection_capable_path(subpath) ||
2816 2302 : equal(oldtarget->exprs, target->exprs))
2817 : {
2818 : /* No separate Result node needed */
2819 382480 : pathnode->dummypp = true;
2820 :
2821 : /*
2822 : * Set cost of plan as subpath's cost, adjusted for tlist replacement.
2823 : */
2824 382480 : pathnode->path.rows = subpath->rows;
2825 382480 : pathnode->path.disabled_nodes = subpath->disabled_nodes;
2826 382480 : pathnode->path.startup_cost = subpath->startup_cost +
2827 382480 : (target->cost.startup - oldtarget->cost.startup);
2828 382480 : pathnode->path.total_cost = subpath->total_cost +
2829 382480 : (target->cost.startup - oldtarget->cost.startup) +
2830 382480 : (target->cost.per_tuple - oldtarget->cost.per_tuple) * subpath->rows;
2831 : }
2832 : else
2833 : {
2834 : /* We really do need the Result node */
2835 2176 : pathnode->dummypp = false;
2836 :
2837 : /*
2838 : * The Result node's cost is cpu_tuple_cost per row, plus the cost of
2839 : * evaluating the tlist. There is no qual to worry about.
2840 : */
2841 2176 : pathnode->path.rows = subpath->rows;
2842 2176 : pathnode->path.disabled_nodes = subpath->disabled_nodes;
2843 2176 : pathnode->path.startup_cost = subpath->startup_cost +
2844 2176 : target->cost.startup;
2845 2176 : pathnode->path.total_cost = subpath->total_cost +
2846 2176 : target->cost.startup +
2847 2176 : (cpu_tuple_cost + target->cost.per_tuple) * subpath->rows;
2848 : }
2849 :
2850 384656 : return pathnode;
2851 : }
2852 :
2853 : /*
2854 : * apply_projection_to_path
2855 : * Add a projection step, or just apply the target directly to given path.
2856 : *
2857 : * This has the same net effect as create_projection_path(), except that if
2858 : * a separate Result plan node isn't needed, we just replace the given path's
2859 : * pathtarget with the desired one. This must be used only when the caller
2860 : * knows that the given path isn't referenced elsewhere and so can be modified
2861 : * in-place.
2862 : *
2863 : * If the input path is a GatherPath or GatherMergePath, we try to push the
2864 : * new target down to its input as well; this is a yet more invasive
2865 : * modification of the input path, which create_projection_path() can't do.
2866 : *
2867 : * Note that we mustn't change the source path's parent link; so when it is
2868 : * add_path'd to "rel" things will be a bit inconsistent. So far that has
2869 : * not caused any trouble.
2870 : *
2871 : * 'rel' is the parent relation associated with the result
2872 : * 'path' is the path representing the source of data
2873 : * 'target' is the PathTarget to be computed
2874 : */
2875 : Path *
2876 13916 : apply_projection_to_path(PlannerInfo *root,
2877 : RelOptInfo *rel,
2878 : Path *path,
2879 : PathTarget *target)
2880 : {
2881 : QualCost oldcost;
2882 :
2883 : /*
2884 : * If given path can't project, we might need a Result node, so make a
2885 : * separate ProjectionPath.
2886 : */
2887 13916 : if (!is_projection_capable_path(path))
2888 1462 : return (Path *) create_projection_path(root, rel, path, target);
2889 :
2890 : /*
2891 : * We can just jam the desired tlist into the existing path, being sure to
2892 : * update its cost estimates appropriately.
2893 : */
2894 12454 : oldcost = path->pathtarget->cost;
2895 12454 : path->pathtarget = target;
2896 :
2897 12454 : path->startup_cost += target->cost.startup - oldcost.startup;
2898 12454 : path->total_cost += target->cost.startup - oldcost.startup +
2899 12454 : (target->cost.per_tuple - oldcost.per_tuple) * path->rows;
2900 :
2901 : /*
2902 : * If the path happens to be a Gather or GatherMerge path, we'd like to
2903 : * arrange for the subpath to return the required target list so that
2904 : * workers can help project. But if there is something that is not
2905 : * parallel-safe in the target expressions, then we can't.
2906 : */
2907 12478 : if ((IsA(path, GatherPath) || IsA(path, GatherMergePath)) &&
2908 24 : is_parallel_safe(root, (Node *) target->exprs))
2909 : {
2910 : /*
2911 : * We always use create_projection_path here, even if the subpath is
2912 : * projection-capable, so as to avoid modifying the subpath in place.
2913 : * It seems unlikely at present that there could be any other
2914 : * references to the subpath, but better safe than sorry.
2915 : *
2916 : * Note that we don't change the parallel path's cost estimates; it
2917 : * might be appropriate to do so, to reflect the fact that the bulk of
2918 : * the target evaluation will happen in workers.
2919 : */
2920 24 : if (IsA(path, GatherPath))
2921 : {
2922 0 : GatherPath *gpath = (GatherPath *) path;
2923 :
2924 0 : gpath->subpath = (Path *)
2925 0 : create_projection_path(root,
2926 0 : gpath->subpath->parent,
2927 : gpath->subpath,
2928 : target);
2929 : }
2930 : else
2931 : {
2932 24 : GatherMergePath *gmpath = (GatherMergePath *) path;
2933 :
2934 24 : gmpath->subpath = (Path *)
2935 24 : create_projection_path(root,
2936 24 : gmpath->subpath->parent,
2937 : gmpath->subpath,
2938 : target);
2939 : }
2940 : }
2941 12430 : else if (path->parallel_safe &&
2942 4978 : !is_parallel_safe(root, (Node *) target->exprs))
2943 : {
2944 : /*
2945 : * We're inserting a parallel-restricted target list into a path
2946 : * currently marked parallel-safe, so we have to mark it as no longer
2947 : * safe.
2948 : */
2949 12 : path->parallel_safe = false;
2950 : }
2951 :
2952 12454 : return path;
2953 : }
2954 :
2955 : /*
2956 : * create_set_projection_path
2957 : * Creates a pathnode that represents performing a projection that
2958 : * includes set-returning functions.
2959 : *
2960 : * 'rel' is the parent relation associated with the result
2961 : * 'subpath' is the path representing the source of data
2962 : * 'target' is the PathTarget to be computed
2963 : */
2964 : ProjectSetPath *
2965 12110 : create_set_projection_path(PlannerInfo *root,
2966 : RelOptInfo *rel,
2967 : Path *subpath,
2968 : PathTarget *target)
2969 : {
2970 12110 : ProjectSetPath *pathnode = makeNode(ProjectSetPath);
2971 : double tlist_rows;
2972 : ListCell *lc;
2973 :
2974 12110 : pathnode->path.pathtype = T_ProjectSet;
2975 12110 : pathnode->path.parent = rel;
2976 12110 : pathnode->path.pathtarget = target;
2977 : /* For now, assume we are above any joins, so no parameterization */
2978 12110 : pathnode->path.param_info = NULL;
2979 12110 : pathnode->path.parallel_aware = false;
2980 29030 : pathnode->path.parallel_safe = rel->consider_parallel &&
2981 16884 : subpath->parallel_safe &&
2982 4774 : is_parallel_safe(root, (Node *) target->exprs);
2983 12110 : pathnode->path.parallel_workers = subpath->parallel_workers;
2984 : /* Projection does not change the sort order XXX? */
2985 12110 : pathnode->path.pathkeys = subpath->pathkeys;
2986 :
2987 12110 : pathnode->subpath = subpath;
2988 :
2989 : /*
2990 : * Estimate number of rows produced by SRFs for each row of input; if
2991 : * there's more than one in this node, use the maximum.
2992 : */
2993 12110 : tlist_rows = 1;
2994 26300 : foreach(lc, target->exprs)
2995 : {
2996 14190 : Node *node = (Node *) lfirst(lc);
2997 : double itemrows;
2998 :
2999 14190 : itemrows = expression_returns_set_rows(root, node);
3000 14190 : if (tlist_rows < itemrows)
3001 11780 : tlist_rows = itemrows;
3002 : }
3003 :
3004 : /*
3005 : * In addition to the cost of evaluating the tlist, charge cpu_tuple_cost
3006 : * per input row, and half of cpu_tuple_cost for each added output row.
3007 : * This is slightly bizarre maybe, but it's what 9.6 did; we may revisit
3008 : * this estimate later.
3009 : */
3010 12110 : pathnode->path.disabled_nodes = subpath->disabled_nodes;
3011 12110 : pathnode->path.rows = subpath->rows * tlist_rows;
3012 12110 : pathnode->path.startup_cost = subpath->startup_cost +
3013 12110 : target->cost.startup;
3014 12110 : pathnode->path.total_cost = subpath->total_cost +
3015 12110 : target->cost.startup +
3016 12110 : (cpu_tuple_cost + target->cost.per_tuple) * subpath->rows +
3017 12110 : (pathnode->path.rows - subpath->rows) * cpu_tuple_cost / 2;
3018 :
3019 12110 : return pathnode;
3020 : }
3021 :
3022 : /*
3023 : * create_incremental_sort_path
3024 : * Creates a pathnode that represents performing an incremental sort.
3025 : *
3026 : * 'rel' is the parent relation associated with the result
3027 : * 'subpath' is the path representing the source of data
3028 : * 'pathkeys' represents the desired sort order
3029 : * 'presorted_keys' is the number of keys by which the input path is
3030 : * already sorted
3031 : * 'limit_tuples' is the estimated bound on the number of output tuples,
3032 : * or -1 if no LIMIT or couldn't estimate
3033 : */
3034 : IncrementalSortPath *
3035 9958 : create_incremental_sort_path(PlannerInfo *root,
3036 : RelOptInfo *rel,
3037 : Path *subpath,
3038 : List *pathkeys,
3039 : int presorted_keys,
3040 : double limit_tuples)
3041 : {
3042 9958 : IncrementalSortPath *sort = makeNode(IncrementalSortPath);
3043 9958 : SortPath *pathnode = &sort->spath;
3044 :
3045 9958 : pathnode->path.pathtype = T_IncrementalSort;
3046 9958 : pathnode->path.parent = rel;
3047 : /* Sort doesn't project, so use source path's pathtarget */
3048 9958 : pathnode->path.pathtarget = subpath->pathtarget;
3049 : /* For now, assume we are above any joins, so no parameterization */
3050 9958 : pathnode->path.param_info = NULL;
3051 9958 : pathnode->path.parallel_aware = false;
3052 14978 : pathnode->path.parallel_safe = rel->consider_parallel &&
3053 5020 : subpath->parallel_safe;
3054 9958 : pathnode->path.parallel_workers = subpath->parallel_workers;
3055 9958 : pathnode->path.pathkeys = pathkeys;
3056 :
3057 9958 : pathnode->subpath = subpath;
3058 :
3059 9958 : cost_incremental_sort(&pathnode->path,
3060 : root, pathkeys, presorted_keys,
3061 : subpath->disabled_nodes,
3062 : subpath->startup_cost,
3063 : subpath->total_cost,
3064 : subpath->rows,
3065 9958 : subpath->pathtarget->width,
3066 : 0.0, /* XXX comparison_cost shouldn't be 0? */
3067 : work_mem, limit_tuples);
3068 :
3069 9958 : sort->nPresortedCols = presorted_keys;
3070 :
3071 9958 : return sort;
3072 : }
3073 :
3074 : /*
3075 : * create_sort_path
3076 : * Creates a pathnode that represents performing an explicit sort.
3077 : *
3078 : * 'rel' is the parent relation associated with the result
3079 : * 'subpath' is the path representing the source of data
3080 : * 'pathkeys' represents the desired sort order
3081 : * 'limit_tuples' is the estimated bound on the number of output tuples,
3082 : * or -1 if no LIMIT or couldn't estimate
3083 : */
3084 : SortPath *
3085 102146 : create_sort_path(PlannerInfo *root,
3086 : RelOptInfo *rel,
3087 : Path *subpath,
3088 : List *pathkeys,
3089 : double limit_tuples)
3090 : {
3091 102146 : SortPath *pathnode = makeNode(SortPath);
3092 :
3093 102146 : pathnode->path.pathtype = T_Sort;
3094 102146 : pathnode->path.parent = rel;
3095 : /* Sort doesn't project, so use source path's pathtarget */
3096 102146 : pathnode->path.pathtarget = subpath->pathtarget;
3097 : /* For now, assume we are above any joins, so no parameterization */
3098 102146 : pathnode->path.param_info = NULL;
3099 102146 : pathnode->path.parallel_aware = false;
3100 174712 : pathnode->path.parallel_safe = rel->consider_parallel &&
3101 72566 : subpath->parallel_safe;
3102 102146 : pathnode->path.parallel_workers = subpath->parallel_workers;
3103 102146 : pathnode->path.pathkeys = pathkeys;
3104 :
3105 102146 : pathnode->subpath = subpath;
3106 :
3107 102146 : cost_sort(&pathnode->path, root, pathkeys,
3108 : subpath->disabled_nodes,
3109 : subpath->total_cost,
3110 : subpath->rows,
3111 102146 : subpath->pathtarget->width,
3112 : 0.0, /* XXX comparison_cost shouldn't be 0? */
3113 : work_mem, limit_tuples);
3114 :
3115 102146 : return pathnode;
3116 : }
3117 :
3118 : /*
3119 : * create_group_path
3120 : * Creates a pathnode that represents performing grouping of presorted input
3121 : *
3122 : * 'rel' is the parent relation associated with the result
3123 : * 'subpath' is the path representing the source of data
3124 : * 'target' is the PathTarget to be computed
3125 : * 'groupClause' is a list of SortGroupClause's representing the grouping
3126 : * 'qual' is the HAVING quals if any
3127 : * 'numGroups' is the estimated number of groups
3128 : */
3129 : GroupPath *
3130 1214 : create_group_path(PlannerInfo *root,
3131 : RelOptInfo *rel,
3132 : Path *subpath,
3133 : List *groupClause,
3134 : List *qual,
3135 : double numGroups)
3136 : {
3137 1214 : GroupPath *pathnode = makeNode(GroupPath);
3138 1214 : PathTarget *target = rel->reltarget;
3139 :
3140 1214 : pathnode->path.pathtype = T_Group;
3141 1214 : pathnode->path.parent = rel;
3142 1214 : pathnode->path.pathtarget = target;
3143 : /* For now, assume we are above any joins, so no parameterization */
3144 1214 : pathnode->path.param_info = NULL;
3145 1214 : pathnode->path.parallel_aware = false;
3146 1958 : pathnode->path.parallel_safe = rel->consider_parallel &&
3147 744 : subpath->parallel_safe;
3148 1214 : pathnode->path.parallel_workers = subpath->parallel_workers;
3149 : /* Group doesn't change sort ordering */
3150 1214 : pathnode->path.pathkeys = subpath->pathkeys;
3151 :
3152 1214 : pathnode->subpath = subpath;
3153 :
3154 1214 : pathnode->groupClause = groupClause;
3155 1214 : pathnode->qual = qual;
3156 :
3157 1214 : cost_group(&pathnode->path, root,
3158 : list_length(groupClause),
3159 : numGroups,
3160 : qual,
3161 : subpath->disabled_nodes,
3162 : subpath->startup_cost, subpath->total_cost,
3163 : subpath->rows);
3164 :
3165 : /* add tlist eval cost for each output row */
3166 1214 : pathnode->path.startup_cost += target->cost.startup;
3167 1214 : pathnode->path.total_cost += target->cost.startup +
3168 1214 : target->cost.per_tuple * pathnode->path.rows;
3169 :
3170 1214 : return pathnode;
3171 : }
3172 :
3173 : /*
3174 : * create_upper_unique_path
3175 : * Creates a pathnode that represents performing an explicit Unique step
3176 : * on presorted input.
3177 : *
3178 : * This produces a Unique plan node, but the use-case is so different from
3179 : * create_unique_path that it doesn't seem worth trying to merge the two.
3180 : *
3181 : * 'rel' is the parent relation associated with the result
3182 : * 'subpath' is the path representing the source of data
3183 : * 'numCols' is the number of grouping columns
3184 : * 'numGroups' is the estimated number of groups
3185 : *
3186 : * The input path must be sorted on the grouping columns, plus possibly
3187 : * additional columns; so the first numCols pathkeys are the grouping columns
3188 : */
3189 : UpperUniquePath *
3190 8284 : create_upper_unique_path(PlannerInfo *root,
3191 : RelOptInfo *rel,
3192 : Path *subpath,
3193 : int numCols,
3194 : double numGroups)
3195 : {
3196 8284 : UpperUniquePath *pathnode = makeNode(UpperUniquePath);
3197 :
3198 8284 : pathnode->path.pathtype = T_Unique;
3199 8284 : pathnode->path.parent = rel;
3200 : /* Unique doesn't project, so use source path's pathtarget */
3201 8284 : pathnode->path.pathtarget = subpath->pathtarget;
3202 : /* For now, assume we are above any joins, so no parameterization */
3203 8284 : pathnode->path.param_info = NULL;
3204 8284 : pathnode->path.parallel_aware = false;
3205 13088 : pathnode->path.parallel_safe = rel->consider_parallel &&
3206 4804 : subpath->parallel_safe;
3207 8284 : pathnode->path.parallel_workers = subpath->parallel_workers;
3208 : /* Unique doesn't change the input ordering */
3209 8284 : pathnode->path.pathkeys = subpath->pathkeys;
3210 :
3211 8284 : pathnode->subpath = subpath;
3212 8284 : pathnode->numkeys = numCols;
3213 :
3214 : /*
3215 : * Charge one cpu_operator_cost per comparison per input tuple. We assume
3216 : * all columns get compared at most of the tuples. (XXX probably this is
3217 : * an overestimate.)
3218 : */
3219 8284 : pathnode->path.disabled_nodes = subpath->disabled_nodes;
3220 8284 : pathnode->path.startup_cost = subpath->startup_cost;
3221 8284 : pathnode->path.total_cost = subpath->total_cost +
3222 8284 : cpu_operator_cost * subpath->rows * numCols;
3223 8284 : pathnode->path.rows = numGroups;
3224 :
3225 8284 : return pathnode;
3226 : }
3227 :
3228 : /*
3229 : * create_agg_path
3230 : * Creates a pathnode that represents performing aggregation/grouping
3231 : *
3232 : * 'rel' is the parent relation associated with the result
3233 : * 'subpath' is the path representing the source of data
3234 : * 'target' is the PathTarget to be computed
3235 : * 'aggstrategy' is the Agg node's basic implementation strategy
3236 : * 'aggsplit' is the Agg node's aggregate-splitting mode
3237 : * 'groupClause' is a list of SortGroupClause's representing the grouping
3238 : * 'qual' is the HAVING quals if any
3239 : * 'aggcosts' contains cost info about the aggregate functions to be computed
3240 : * 'numGroups' is the estimated number of groups (1 if not grouping)
3241 : */
3242 : AggPath *
3243 57736 : create_agg_path(PlannerInfo *root,
3244 : RelOptInfo *rel,
3245 : Path *subpath,
3246 : PathTarget *target,
3247 : AggStrategy aggstrategy,
3248 : AggSplit aggsplit,
3249 : List *groupClause,
3250 : List *qual,
3251 : const AggClauseCosts *aggcosts,
3252 : double numGroups)
3253 : {
3254 57736 : AggPath *pathnode = makeNode(AggPath);
3255 :
3256 57736 : pathnode->path.pathtype = T_Agg;
3257 57736 : pathnode->path.parent = rel;
3258 57736 : pathnode->path.pathtarget = target;
3259 : /* For now, assume we are above any joins, so no parameterization */
3260 57736 : pathnode->path.param_info = NULL;
3261 57736 : pathnode->path.parallel_aware = false;
3262 96658 : pathnode->path.parallel_safe = rel->consider_parallel &&
3263 38922 : subpath->parallel_safe;
3264 57736 : pathnode->path.parallel_workers = subpath->parallel_workers;
3265 :
3266 57736 : if (aggstrategy == AGG_SORTED)
3267 : {
3268 : /*
3269 : * Attempt to preserve the order of the subpath. Additional pathkeys
3270 : * may have been added in adjust_group_pathkeys_for_groupagg() to
3271 : * support ORDER BY / DISTINCT aggregates. Pathkeys added there
3272 : * belong to columns within the aggregate function, so we must strip
3273 : * these additional pathkeys off as those columns are unavailable
3274 : * above the aggregate node.
3275 : */
3276 7820 : if (list_length(subpath->pathkeys) > root->num_groupby_pathkeys)
3277 328 : pathnode->path.pathkeys = list_copy_head(subpath->pathkeys,
3278 : root->num_groupby_pathkeys);
3279 : else
3280 7492 : pathnode->path.pathkeys = subpath->pathkeys; /* preserves order */
3281 : }
3282 : else
3283 49916 : pathnode->path.pathkeys = NIL; /* output is unordered */
3284 :
3285 57736 : pathnode->subpath = subpath;
3286 :
3287 57736 : pathnode->aggstrategy = aggstrategy;
3288 57736 : pathnode->aggsplit = aggsplit;
3289 57736 : pathnode->numGroups = numGroups;
3290 57736 : pathnode->transitionSpace = aggcosts ? aggcosts->transitionSpace : 0;
3291 57736 : pathnode->groupClause = groupClause;
3292 57736 : pathnode->qual = qual;
3293 :
3294 57736 : cost_agg(&pathnode->path, root,
3295 : aggstrategy, aggcosts,
3296 : list_length(groupClause), numGroups,
3297 : qual,
3298 : subpath->disabled_nodes,
3299 : subpath->startup_cost, subpath->total_cost,
3300 57736 : subpath->rows, subpath->pathtarget->width);
3301 :
3302 : /* add tlist eval cost for each output row */
3303 57736 : pathnode->path.startup_cost += target->cost.startup;
3304 57736 : pathnode->path.total_cost += target->cost.startup +
3305 57736 : target->cost.per_tuple * pathnode->path.rows;
3306 :
3307 57736 : return pathnode;
3308 : }
3309 :
3310 : /*
3311 : * create_groupingsets_path
3312 : * Creates a pathnode that represents performing GROUPING SETS aggregation
3313 : *
3314 : * GroupingSetsPath represents sorted grouping with one or more grouping sets.
3315 : * The input path's result must be sorted to match the last entry in
3316 : * rollup_groupclauses.
3317 : *
3318 : * 'rel' is the parent relation associated with the result
3319 : * 'subpath' is the path representing the source of data
3320 : * 'target' is the PathTarget to be computed
3321 : * 'having_qual' is the HAVING quals if any
3322 : * 'rollups' is a list of RollupData nodes
3323 : * 'agg_costs' contains cost info about the aggregate functions to be computed
3324 : */
3325 : GroupingSetsPath *
3326 2128 : create_groupingsets_path(PlannerInfo *root,
3327 : RelOptInfo *rel,
3328 : Path *subpath,
3329 : List *having_qual,
3330 : AggStrategy aggstrategy,
3331 : List *rollups,
3332 : const AggClauseCosts *agg_costs)
3333 : {
3334 2128 : GroupingSetsPath *pathnode = makeNode(GroupingSetsPath);
3335 2128 : PathTarget *target = rel->reltarget;
3336 : ListCell *lc;
3337 2128 : bool is_first = true;
3338 2128 : bool is_first_sort = true;
3339 :
3340 : /* The topmost generated Plan node will be an Agg */
3341 2128 : pathnode->path.pathtype = T_Agg;
3342 2128 : pathnode->path.parent = rel;
3343 2128 : pathnode->path.pathtarget = target;
3344 2128 : pathnode->path.param_info = subpath->param_info;
3345 2128 : pathnode->path.parallel_aware = false;
3346 3118 : pathnode->path.parallel_safe = rel->consider_parallel &&
3347 990 : subpath->parallel_safe;
3348 2128 : pathnode->path.parallel_workers = subpath->parallel_workers;
3349 2128 : pathnode->subpath = subpath;
3350 :
3351 : /*
3352 : * Simplify callers by downgrading AGG_SORTED to AGG_PLAIN, and AGG_MIXED
3353 : * to AGG_HASHED, here if possible.
3354 : */
3355 3036 : if (aggstrategy == AGG_SORTED &&
3356 908 : list_length(rollups) == 1 &&
3357 458 : ((RollupData *) linitial(rollups))->groupClause == NIL)
3358 42 : aggstrategy = AGG_PLAIN;
3359 :
3360 3044 : if (aggstrategy == AGG_MIXED &&
3361 916 : list_length(rollups) == 1)
3362 0 : aggstrategy = AGG_HASHED;
3363 :
3364 : /*
3365 : * Output will be in sorted order by group_pathkeys if, and only if, there
3366 : * is a single rollup operation on a non-empty list of grouping
3367 : * expressions.
3368 : */
3369 2128 : if (aggstrategy == AGG_SORTED && list_length(rollups) == 1)
3370 416 : pathnode->path.pathkeys = root->group_pathkeys;
3371 : else
3372 1712 : pathnode->path.pathkeys = NIL;
3373 :
3374 2128 : pathnode->aggstrategy = aggstrategy;
3375 2128 : pathnode->rollups = rollups;
3376 2128 : pathnode->qual = having_qual;
3377 2128 : pathnode->transitionSpace = agg_costs ? agg_costs->transitionSpace : 0;
3378 :
3379 : Assert(rollups != NIL);
3380 : Assert(aggstrategy != AGG_PLAIN || list_length(rollups) == 1);
3381 : Assert(aggstrategy != AGG_MIXED || list_length(rollups) > 1);
3382 :
3383 7380 : foreach(lc, rollups)
3384 : {
3385 5252 : RollupData *rollup = lfirst(lc);
3386 5252 : List *gsets = rollup->gsets;
3387 5252 : int numGroupCols = list_length(linitial(gsets));
3388 :
3389 : /*
3390 : * In AGG_SORTED or AGG_PLAIN mode, the first rollup takes the
3391 : * (already-sorted) input, and following ones do their own sort.
3392 : *
3393 : * In AGG_HASHED mode, there is one rollup for each grouping set.
3394 : *
3395 : * In AGG_MIXED mode, the first rollups are hashed, the first
3396 : * non-hashed one takes the (already-sorted) input, and following ones
3397 : * do their own sort.
3398 : */
3399 5252 : if (is_first)
3400 : {
3401 2128 : cost_agg(&pathnode->path, root,
3402 : aggstrategy,
3403 : agg_costs,
3404 : numGroupCols,
3405 : rollup->numGroups,
3406 : having_qual,
3407 : subpath->disabled_nodes,
3408 : subpath->startup_cost,
3409 : subpath->total_cost,
3410 : subpath->rows,
3411 2128 : subpath->pathtarget->width);
3412 2128 : is_first = false;
3413 2128 : if (!rollup->is_hashed)
3414 908 : is_first_sort = false;
3415 : }
3416 : else
3417 : {
3418 : Path sort_path; /* dummy for result of cost_sort */
3419 : Path agg_path; /* dummy for result of cost_agg */
3420 :
3421 3124 : if (rollup->is_hashed || is_first_sort)
3422 : {
3423 : /*
3424 : * Account for cost of aggregation, but don't charge input
3425 : * cost again
3426 : */
3427 2398 : cost_agg(&agg_path, root,
3428 2398 : rollup->is_hashed ? AGG_HASHED : AGG_SORTED,
3429 : agg_costs,
3430 : numGroupCols,
3431 : rollup->numGroups,
3432 : having_qual,
3433 : 0, 0.0, 0.0,
3434 : subpath->rows,
3435 2398 : subpath->pathtarget->width);
3436 2398 : if (!rollup->is_hashed)
3437 916 : is_first_sort = false;
3438 : }
3439 : else
3440 : {
3441 : /* Account for cost of sort, but don't charge input cost again */
3442 726 : cost_sort(&sort_path, root, NIL, 0,
3443 : 0.0,
3444 : subpath->rows,
3445 726 : subpath->pathtarget->width,
3446 : 0.0,
3447 : work_mem,
3448 : -1.0);
3449 :
3450 : /* Account for cost of aggregation */
3451 :
3452 726 : cost_agg(&agg_path, root,
3453 : AGG_SORTED,
3454 : agg_costs,
3455 : numGroupCols,
3456 : rollup->numGroups,
3457 : having_qual,
3458 : sort_path.disabled_nodes,
3459 : sort_path.startup_cost,
3460 : sort_path.total_cost,
3461 : sort_path.rows,
3462 726 : subpath->pathtarget->width);
3463 : }
3464 :
3465 3124 : pathnode->path.disabled_nodes += agg_path.disabled_nodes;
3466 3124 : pathnode->path.total_cost += agg_path.total_cost;
3467 3124 : pathnode->path.rows += agg_path.rows;
3468 : }
3469 : }
3470 :
3471 : /* add tlist eval cost for each output row */
3472 2128 : pathnode->path.startup_cost += target->cost.startup;
3473 2128 : pathnode->path.total_cost += target->cost.startup +
3474 2128 : target->cost.per_tuple * pathnode->path.rows;
3475 :
3476 2128 : return pathnode;
3477 : }
3478 :
3479 : /*
3480 : * create_minmaxagg_path
3481 : * Creates a pathnode that represents computation of MIN/MAX aggregates
3482 : *
3483 : * 'rel' is the parent relation associated with the result
3484 : * 'target' is the PathTarget to be computed
3485 : * 'mmaggregates' is a list of MinMaxAggInfo structs
3486 : * 'quals' is the HAVING quals if any
3487 : */
3488 : MinMaxAggPath *
3489 410 : create_minmaxagg_path(PlannerInfo *root,
3490 : RelOptInfo *rel,
3491 : PathTarget *target,
3492 : List *mmaggregates,
3493 : List *quals)
3494 : {
3495 410 : MinMaxAggPath *pathnode = makeNode(MinMaxAggPath);
3496 : Cost initplan_cost;
3497 410 : int initplan_disabled_nodes = 0;
3498 : ListCell *lc;
3499 :
3500 : /* The topmost generated Plan node will be a Result */
3501 410 : pathnode->path.pathtype = T_Result;
3502 410 : pathnode->path.parent = rel;
3503 410 : pathnode->path.pathtarget = target;
3504 : /* For now, assume we are above any joins, so no parameterization */
3505 410 : pathnode->path.param_info = NULL;
3506 410 : pathnode->path.parallel_aware = false;
3507 410 : pathnode->path.parallel_safe = true; /* might change below */
3508 410 : pathnode->path.parallel_workers = 0;
3509 : /* Result is one unordered row */
3510 410 : pathnode->path.rows = 1;
3511 410 : pathnode->path.pathkeys = NIL;
3512 :
3513 410 : pathnode->mmaggregates = mmaggregates;
3514 410 : pathnode->quals = quals;
3515 :
3516 : /* Calculate cost of all the initplans, and check parallel safety */
3517 410 : initplan_cost = 0;
3518 856 : foreach(lc, mmaggregates)
3519 : {
3520 446 : MinMaxAggInfo *mminfo = (MinMaxAggInfo *) lfirst(lc);
3521 :
3522 446 : initplan_disabled_nodes += mminfo->path->disabled_nodes;
3523 446 : initplan_cost += mminfo->pathcost;
3524 446 : if (!mminfo->path->parallel_safe)
3525 110 : pathnode->path.parallel_safe = false;
3526 : }
3527 :
3528 : /* add tlist eval cost for each output row, plus cpu_tuple_cost */
3529 410 : pathnode->path.disabled_nodes = initplan_disabled_nodes;
3530 410 : pathnode->path.startup_cost = initplan_cost + target->cost.startup;
3531 410 : pathnode->path.total_cost = initplan_cost + target->cost.startup +
3532 410 : target->cost.per_tuple + cpu_tuple_cost;
3533 :
3534 : /*
3535 : * Add cost of qual, if any --- but we ignore its selectivity, since our
3536 : * rowcount estimate should be 1 no matter what the qual is.
3537 : */
3538 410 : if (quals)
3539 : {
3540 : QualCost qual_cost;
3541 :
3542 0 : cost_qual_eval(&qual_cost, quals, root);
3543 0 : pathnode->path.startup_cost += qual_cost.startup;
3544 0 : pathnode->path.total_cost += qual_cost.startup + qual_cost.per_tuple;
3545 : }
3546 :
3547 : /*
3548 : * If the initplans were all parallel-safe, also check safety of the
3549 : * target and quals. (The Result node itself isn't parallelizable, but if
3550 : * we are in a subquery then it can be useful for the outer query to know
3551 : * that this one is parallel-safe.)
3552 : */
3553 410 : if (pathnode->path.parallel_safe)
3554 300 : pathnode->path.parallel_safe =
3555 600 : is_parallel_safe(root, (Node *) target->exprs) &&
3556 300 : is_parallel_safe(root, (Node *) quals);
3557 :
3558 410 : return pathnode;
3559 : }
3560 :
3561 : /*
3562 : * create_windowagg_path
3563 : * Creates a pathnode that represents computation of window functions
3564 : *
3565 : * 'rel' is the parent relation associated with the result
3566 : * 'subpath' is the path representing the source of data
3567 : * 'target' is the PathTarget to be computed
3568 : * 'windowFuncs' is a list of WindowFunc structs
3569 : * 'runCondition' is a list of OpExprs to short-circuit WindowAgg execution
3570 : * 'winclause' is a WindowClause that is common to all the WindowFuncs
3571 : * 'qual' WindowClause.runconditions from lower-level WindowAggPaths.
3572 : * Must always be NIL when topwindow == false
3573 : * 'topwindow' pass as true only for the top-level WindowAgg. False for all
3574 : * intermediate WindowAggs.
3575 : *
3576 : * The input must be sorted according to the WindowClause's PARTITION keys
3577 : * plus ORDER BY keys.
3578 : */
3579 : WindowAggPath *
3580 2754 : create_windowagg_path(PlannerInfo *root,
3581 : RelOptInfo *rel,
3582 : Path *subpath,
3583 : PathTarget *target,
3584 : List *windowFuncs,
3585 : List *runCondition,
3586 : WindowClause *winclause,
3587 : List *qual,
3588 : bool topwindow)
3589 : {
3590 2754 : WindowAggPath *pathnode = makeNode(WindowAggPath);
3591 :
3592 : /* qual can only be set for the topwindow */
3593 : Assert(qual == NIL || topwindow);
3594 :
3595 2754 : pathnode->path.pathtype = T_WindowAgg;
3596 2754 : pathnode->path.parent = rel;
3597 2754 : pathnode->path.pathtarget = target;
3598 : /* For now, assume we are above any joins, so no parameterization */
3599 2754 : pathnode->path.param_info = NULL;
3600 2754 : pathnode->path.parallel_aware = false;
3601 2754 : pathnode->path.parallel_safe = rel->consider_parallel &&
3602 0 : subpath->parallel_safe;
3603 2754 : pathnode->path.parallel_workers = subpath->parallel_workers;
3604 : /* WindowAgg preserves the input sort order */
3605 2754 : pathnode->path.pathkeys = subpath->pathkeys;
3606 :
3607 2754 : pathnode->subpath = subpath;
3608 2754 : pathnode->winclause = winclause;
3609 2754 : pathnode->qual = qual;
3610 2754 : pathnode->runCondition = runCondition;
3611 2754 : pathnode->topwindow = topwindow;
3612 :
3613 : /*
3614 : * For costing purposes, assume that there are no redundant partitioning
3615 : * or ordering columns; it's not worth the trouble to deal with that
3616 : * corner case here. So we just pass the unmodified list lengths to
3617 : * cost_windowagg.
3618 : */
3619 2754 : cost_windowagg(&pathnode->path, root,
3620 : windowFuncs,
3621 : winclause,
3622 : subpath->disabled_nodes,
3623 : subpath->startup_cost,
3624 : subpath->total_cost,
3625 : subpath->rows);
3626 :
3627 : /* add tlist eval cost for each output row */
3628 2754 : pathnode->path.startup_cost += target->cost.startup;
3629 2754 : pathnode->path.total_cost += target->cost.startup +
3630 2754 : target->cost.per_tuple * pathnode->path.rows;
3631 :
3632 2754 : return pathnode;
3633 : }
3634 :
3635 : /*
3636 : * create_setop_path
3637 : * Creates a pathnode that represents computation of INTERSECT or EXCEPT
3638 : *
3639 : * 'rel' is the parent relation associated with the result
3640 : * 'leftpath' is the path representing the left-hand source of data
3641 : * 'rightpath' is the path representing the right-hand source of data
3642 : * 'cmd' is the specific semantics (INTERSECT or EXCEPT, with/without ALL)
3643 : * 'strategy' is the implementation strategy (sorted or hashed)
3644 : * 'groupList' is a list of SortGroupClause's representing the grouping
3645 : * 'numGroups' is the estimated number of distinct groups in left-hand input
3646 : * 'outputRows' is the estimated number of output rows
3647 : *
3648 : * leftpath and rightpath must produce the same columns. Moreover, if
3649 : * strategy is SETOP_SORTED, leftpath and rightpath must both be sorted
3650 : * by all the grouping columns.
3651 : */
3652 : SetOpPath *
3653 1264 : create_setop_path(PlannerInfo *root,
3654 : RelOptInfo *rel,
3655 : Path *leftpath,
3656 : Path *rightpath,
3657 : SetOpCmd cmd,
3658 : SetOpStrategy strategy,
3659 : List *groupList,
3660 : double numGroups,
3661 : double outputRows)
3662 : {
3663 1264 : SetOpPath *pathnode = makeNode(SetOpPath);
3664 :
3665 1264 : pathnode->path.pathtype = T_SetOp;
3666 1264 : pathnode->path.parent = rel;
3667 1264 : pathnode->path.pathtarget = rel->reltarget;
3668 : /* For now, assume we are above any joins, so no parameterization */
3669 1264 : pathnode->path.param_info = NULL;
3670 1264 : pathnode->path.parallel_aware = false;
3671 2528 : pathnode->path.parallel_safe = rel->consider_parallel &&
3672 1264 : leftpath->parallel_safe && rightpath->parallel_safe;
3673 1264 : pathnode->path.parallel_workers =
3674 1264 : leftpath->parallel_workers + rightpath->parallel_workers;
3675 : /* SetOp preserves the input sort order if in sort mode */
3676 1264 : pathnode->path.pathkeys =
3677 1264 : (strategy == SETOP_SORTED) ? leftpath->pathkeys : NIL;
3678 :
3679 1264 : pathnode->leftpath = leftpath;
3680 1264 : pathnode->rightpath = rightpath;
3681 1264 : pathnode->cmd = cmd;
3682 1264 : pathnode->strategy = strategy;
3683 1264 : pathnode->groupList = groupList;
3684 1264 : pathnode->numGroups = numGroups;
3685 :
3686 : /*
3687 : * Compute cost estimates. As things stand, we end up with the same total
3688 : * cost in this node for sort and hash methods, but different startup
3689 : * costs. This could be refined perhaps, but it'll do for now.
3690 : */
3691 1264 : pathnode->path.disabled_nodes =
3692 1264 : leftpath->disabled_nodes + rightpath->disabled_nodes;
3693 1264 : if (strategy == SETOP_SORTED)
3694 : {
3695 : /*
3696 : * In sorted mode, we can emit output incrementally. Charge one
3697 : * cpu_operator_cost per comparison per input tuple. Like cost_group,
3698 : * we assume all columns get compared at most of the tuples.
3699 : */
3700 662 : pathnode->path.startup_cost =
3701 662 : leftpath->startup_cost + rightpath->startup_cost;
3702 662 : pathnode->path.total_cost =
3703 1324 : leftpath->total_cost + rightpath->total_cost +
3704 662 : cpu_operator_cost * (leftpath->rows + rightpath->rows) * list_length(groupList);
3705 :
3706 : /*
3707 : * Also charge a small amount per extracted tuple. Like cost_sort,
3708 : * charge only operator cost not cpu_tuple_cost, since SetOp does no
3709 : * qual-checking or projection.
3710 : */
3711 662 : pathnode->path.total_cost += cpu_operator_cost * outputRows;
3712 : }
3713 : else
3714 : {
3715 : Size hashentrysize;
3716 :
3717 : /*
3718 : * In hashed mode, we must read all the input before we can emit
3719 : * anything. Also charge comparison costs to represent the cost of
3720 : * hash table lookups.
3721 : */
3722 602 : pathnode->path.startup_cost =
3723 1204 : leftpath->total_cost + rightpath->total_cost +
3724 602 : cpu_operator_cost * (leftpath->rows + rightpath->rows) * list_length(groupList);
3725 602 : pathnode->path.total_cost = pathnode->path.startup_cost;
3726 :
3727 : /*
3728 : * Also charge a small amount per extracted tuple. Like cost_sort,
3729 : * charge only operator cost not cpu_tuple_cost, since SetOp does no
3730 : * qual-checking or projection.
3731 : */
3732 602 : pathnode->path.total_cost += cpu_operator_cost * outputRows;
3733 :
3734 : /*
3735 : * Mark the path as disabled if enable_hashagg is off. While this
3736 : * isn't exactly a HashAgg node, it seems close enough to justify
3737 : * letting that switch control it.
3738 : */
3739 602 : if (!enable_hashagg)
3740 114 : pathnode->path.disabled_nodes++;
3741 :
3742 : /*
3743 : * Also disable if it doesn't look like the hashtable will fit into
3744 : * hash_mem.
3745 : */
3746 602 : hashentrysize = MAXALIGN(leftpath->pathtarget->width) +
3747 : MAXALIGN(SizeofMinimalTupleHeader);
3748 602 : if (hashentrysize * numGroups > get_hash_memory_limit())
3749 0 : pathnode->path.disabled_nodes++;
3750 : }
3751 1264 : pathnode->path.rows = outputRows;
3752 :
3753 1264 : return pathnode;
3754 : }
3755 :
3756 : /*
3757 : * create_recursiveunion_path
3758 : * Creates a pathnode that represents a recursive UNION node
3759 : *
3760 : * 'rel' is the parent relation associated with the result
3761 : * 'leftpath' is the source of data for the non-recursive term
3762 : * 'rightpath' is the source of data for the recursive term
3763 : * 'target' is the PathTarget to be computed
3764 : * 'distinctList' is a list of SortGroupClause's representing the grouping
3765 : * 'wtParam' is the ID of Param representing work table
3766 : * 'numGroups' is the estimated number of groups
3767 : *
3768 : * For recursive UNION ALL, distinctList is empty and numGroups is zero
3769 : */
3770 : RecursiveUnionPath *
3771 1006 : create_recursiveunion_path(PlannerInfo *root,
3772 : RelOptInfo *rel,
3773 : Path *leftpath,
3774 : Path *rightpath,
3775 : PathTarget *target,
3776 : List *distinctList,
3777 : int wtParam,
3778 : double numGroups)
3779 : {
3780 1006 : RecursiveUnionPath *pathnode = makeNode(RecursiveUnionPath);
3781 :
3782 1006 : pathnode->path.pathtype = T_RecursiveUnion;
3783 1006 : pathnode->path.parent = rel;
3784 1006 : pathnode->path.pathtarget = target;
3785 : /* For now, assume we are above any joins, so no parameterization */
3786 1006 : pathnode->path.param_info = NULL;
3787 1006 : pathnode->path.parallel_aware = false;
3788 2012 : pathnode->path.parallel_safe = rel->consider_parallel &&
3789 1006 : leftpath->parallel_safe && rightpath->parallel_safe;
3790 : /* Foolish, but we'll do it like joins for now: */
3791 1006 : pathnode->path.parallel_workers = leftpath->parallel_workers;
3792 : /* RecursiveUnion result is always unsorted */
3793 1006 : pathnode->path.pathkeys = NIL;
3794 :
3795 1006 : pathnode->leftpath = leftpath;
3796 1006 : pathnode->rightpath = rightpath;
3797 1006 : pathnode->distinctList = distinctList;
3798 1006 : pathnode->wtParam = wtParam;
3799 1006 : pathnode->numGroups = numGroups;
3800 :
3801 1006 : cost_recursive_union(&pathnode->path, leftpath, rightpath);
3802 :
3803 1006 : return pathnode;
3804 : }
3805 :
3806 : /*
3807 : * create_lockrows_path
3808 : * Creates a pathnode that represents acquiring row locks
3809 : *
3810 : * 'rel' is the parent relation associated with the result
3811 : * 'subpath' is the path representing the source of data
3812 : * 'rowMarks' is a list of PlanRowMark's
3813 : * 'epqParam' is the ID of Param for EvalPlanQual re-eval
3814 : */
3815 : LockRowsPath *
3816 8192 : create_lockrows_path(PlannerInfo *root, RelOptInfo *rel,
3817 : Path *subpath, List *rowMarks, int epqParam)
3818 : {
3819 8192 : LockRowsPath *pathnode = makeNode(LockRowsPath);
3820 :
3821 8192 : pathnode->path.pathtype = T_LockRows;
3822 8192 : pathnode->path.parent = rel;
3823 : /* LockRows doesn't project, so use source path's pathtarget */
3824 8192 : pathnode->path.pathtarget = subpath->pathtarget;
3825 : /* For now, assume we are above any joins, so no parameterization */
3826 8192 : pathnode->path.param_info = NULL;
3827 8192 : pathnode->path.parallel_aware = false;
3828 8192 : pathnode->path.parallel_safe = false;
3829 8192 : pathnode->path.parallel_workers = 0;
3830 8192 : pathnode->path.rows = subpath->rows;
3831 :
3832 : /*
3833 : * The result cannot be assumed sorted, since locking might cause the sort
3834 : * key columns to be replaced with new values.
3835 : */
3836 8192 : pathnode->path.pathkeys = NIL;
3837 :
3838 8192 : pathnode->subpath = subpath;
3839 8192 : pathnode->rowMarks = rowMarks;
3840 8192 : pathnode->epqParam = epqParam;
3841 :
3842 : /*
3843 : * We should charge something extra for the costs of row locking and
3844 : * possible refetches, but it's hard to say how much. For now, use
3845 : * cpu_tuple_cost per row.
3846 : */
3847 8192 : pathnode->path.disabled_nodes = subpath->disabled_nodes;
3848 8192 : pathnode->path.startup_cost = subpath->startup_cost;
3849 8192 : pathnode->path.total_cost = subpath->total_cost +
3850 8192 : cpu_tuple_cost * subpath->rows;
3851 :
3852 8192 : return pathnode;
3853 : }
3854 :
3855 : /*
3856 : * create_modifytable_path
3857 : * Creates a pathnode that represents performing INSERT/UPDATE/DELETE/MERGE
3858 : * mods
3859 : *
3860 : * 'rel' is the parent relation associated with the result
3861 : * 'subpath' is a Path producing source data
3862 : * 'operation' is the operation type
3863 : * 'canSetTag' is true if we set the command tag/es_processed
3864 : * 'nominalRelation' is the parent RT index for use of EXPLAIN
3865 : * 'rootRelation' is the partitioned/inherited table root RTI, or 0 if none
3866 : * 'partColsUpdated' is true if any partitioning columns are being updated,
3867 : * either from the target relation or a descendent partitioned table.
3868 : * 'resultRelations' is an integer list of actual RT indexes of target rel(s)
3869 : * 'updateColnosLists' is a list of UPDATE target column number lists
3870 : * (one sublist per rel); or NIL if not an UPDATE
3871 : * 'withCheckOptionLists' is a list of WCO lists (one per rel)
3872 : * 'returningLists' is a list of RETURNING tlists (one per rel)
3873 : * 'rowMarks' is a list of PlanRowMarks (non-locking only)
3874 : * 'onconflict' is the ON CONFLICT clause, or NULL
3875 : * 'epqParam' is the ID of Param for EvalPlanQual re-eval
3876 : * 'mergeActionLists' is a list of lists of MERGE actions (one per rel)
3877 : * 'mergeJoinConditions' is a list of join conditions for MERGE (one per rel)
3878 : */
3879 : ModifyTablePath *
3880 92456 : create_modifytable_path(PlannerInfo *root, RelOptInfo *rel,
3881 : Path *subpath,
3882 : CmdType operation, bool canSetTag,
3883 : Index nominalRelation, Index rootRelation,
3884 : bool partColsUpdated,
3885 : List *resultRelations,
3886 : List *updateColnosLists,
3887 : List *withCheckOptionLists, List *returningLists,
3888 : List *rowMarks, OnConflictExpr *onconflict,
3889 : List *mergeActionLists, List *mergeJoinConditions,
3890 : int epqParam)
3891 : {
3892 92456 : ModifyTablePath *pathnode = makeNode(ModifyTablePath);
3893 :
3894 : Assert(operation == CMD_MERGE ||
3895 : (operation == CMD_UPDATE ?
3896 : list_length(resultRelations) == list_length(updateColnosLists) :
3897 : updateColnosLists == NIL));
3898 : Assert(withCheckOptionLists == NIL ||
3899 : list_length(resultRelations) == list_length(withCheckOptionLists));
3900 : Assert(returningLists == NIL ||
3901 : list_length(resultRelations) == list_length(returningLists));
3902 :
3903 92456 : pathnode->path.pathtype = T_ModifyTable;
3904 92456 : pathnode->path.parent = rel;
3905 : /* pathtarget is not interesting, just make it minimally valid */
3906 92456 : pathnode->path.pathtarget = rel->reltarget;
3907 : /* For now, assume we are above any joins, so no parameterization */
3908 92456 : pathnode->path.param_info = NULL;
3909 92456 : pathnode->path.parallel_aware = false;
3910 92456 : pathnode->path.parallel_safe = false;
3911 92456 : pathnode->path.parallel_workers = 0;
3912 92456 : pathnode->path.pathkeys = NIL;
3913 :
3914 : /*
3915 : * Compute cost & rowcount as subpath cost & rowcount (if RETURNING)
3916 : *
3917 : * Currently, we don't charge anything extra for the actual table
3918 : * modification work, nor for the WITH CHECK OPTIONS or RETURNING
3919 : * expressions if any. It would only be window dressing, since
3920 : * ModifyTable is always a top-level node and there is no way for the
3921 : * costs to change any higher-level planning choices. But we might want
3922 : * to make it look better sometime.
3923 : */
3924 92456 : pathnode->path.disabled_nodes = subpath->disabled_nodes;
3925 92456 : pathnode->path.startup_cost = subpath->startup_cost;
3926 92456 : pathnode->path.total_cost = subpath->total_cost;
3927 92456 : if (returningLists != NIL)
3928 : {
3929 2890 : pathnode->path.rows = subpath->rows;
3930 :
3931 : /*
3932 : * Set width to match the subpath output. XXX this is totally wrong:
3933 : * we should return an average of the RETURNING tlist widths. But
3934 : * it's what happened historically, and improving it is a task for
3935 : * another day. (Again, it's mostly window dressing.)
3936 : */
3937 2890 : pathnode->path.pathtarget->width = subpath->pathtarget->width;
3938 : }
3939 : else
3940 : {
3941 89566 : pathnode->path.rows = 0;
3942 89566 : pathnode->path.pathtarget->width = 0;
3943 : }
3944 :
3945 92456 : pathnode->subpath = subpath;
3946 92456 : pathnode->operation = operation;
3947 92456 : pathnode->canSetTag = canSetTag;
3948 92456 : pathnode->nominalRelation = nominalRelation;
3949 92456 : pathnode->rootRelation = rootRelation;
3950 92456 : pathnode->partColsUpdated = partColsUpdated;
3951 92456 : pathnode->resultRelations = resultRelations;
3952 92456 : pathnode->updateColnosLists = updateColnosLists;
3953 92456 : pathnode->withCheckOptionLists = withCheckOptionLists;
3954 92456 : pathnode->returningLists = returningLists;
3955 92456 : pathnode->rowMarks = rowMarks;
3956 92456 : pathnode->onconflict = onconflict;
3957 92456 : pathnode->epqParam = epqParam;
3958 92456 : pathnode->mergeActionLists = mergeActionLists;
3959 92456 : pathnode->mergeJoinConditions = mergeJoinConditions;
3960 :
3961 92456 : return pathnode;
3962 : }
3963 :
3964 : /*
3965 : * create_limit_path
3966 : * Creates a pathnode that represents performing LIMIT/OFFSET
3967 : *
3968 : * In addition to providing the actual OFFSET and LIMIT expressions,
3969 : * the caller must provide estimates of their values for costing purposes.
3970 : * The estimates are as computed by preprocess_limit(), ie, 0 represents
3971 : * the clause not being present, and -1 means it's present but we could
3972 : * not estimate its value.
3973 : *
3974 : * 'rel' is the parent relation associated with the result
3975 : * 'subpath' is the path representing the source of data
3976 : * 'limitOffset' is the actual OFFSET expression, or NULL
3977 : * 'limitCount' is the actual LIMIT expression, or NULL
3978 : * 'offset_est' is the estimated value of the OFFSET expression
3979 : * 'count_est' is the estimated value of the LIMIT expression
3980 : */
3981 : LimitPath *
3982 5974 : create_limit_path(PlannerInfo *root, RelOptInfo *rel,
3983 : Path *subpath,
3984 : Node *limitOffset, Node *limitCount,
3985 : LimitOption limitOption,
3986 : int64 offset_est, int64 count_est)
3987 : {
3988 5974 : LimitPath *pathnode = makeNode(LimitPath);
3989 :
3990 5974 : pathnode->path.pathtype = T_Limit;
3991 5974 : pathnode->path.parent = rel;
3992 : /* Limit doesn't project, so use source path's pathtarget */
3993 5974 : pathnode->path.pathtarget = subpath->pathtarget;
3994 : /* For now, assume we are above any joins, so no parameterization */
3995 5974 : pathnode->path.param_info = NULL;
3996 5974 : pathnode->path.parallel_aware = false;
3997 8368 : pathnode->path.parallel_safe = rel->consider_parallel &&
3998 2394 : subpath->parallel_safe;
3999 5974 : pathnode->path.parallel_workers = subpath->parallel_workers;
4000 5974 : pathnode->path.rows = subpath->rows;
4001 5974 : pathnode->path.disabled_nodes = subpath->disabled_nodes;
4002 5974 : pathnode->path.startup_cost = subpath->startup_cost;
4003 5974 : pathnode->path.total_cost = subpath->total_cost;
4004 5974 : pathnode->path.pathkeys = subpath->pathkeys;
4005 5974 : pathnode->subpath = subpath;
4006 5974 : pathnode->limitOffset = limitOffset;
4007 5974 : pathnode->limitCount = limitCount;
4008 5974 : pathnode->limitOption = limitOption;
4009 :
4010 : /*
4011 : * Adjust the output rows count and costs according to the offset/limit.
4012 : */
4013 5974 : adjust_limit_rows_costs(&pathnode->path.rows,
4014 : &pathnode->path.startup_cost,
4015 : &pathnode->path.total_cost,
4016 : offset_est, count_est);
4017 :
4018 5974 : return pathnode;
4019 : }
4020 :
4021 : /*
4022 : * adjust_limit_rows_costs
4023 : * Adjust the size and cost estimates for a LimitPath node according to the
4024 : * offset/limit.
4025 : *
4026 : * This is only a cosmetic issue if we are at top level, but if we are
4027 : * building a subquery then it's important to report correct info to the outer
4028 : * planner.
4029 : *
4030 : * When the offset or count couldn't be estimated, use 10% of the estimated
4031 : * number of rows emitted from the subpath.
4032 : *
4033 : * XXX we don't bother to add eval costs of the offset/limit expressions
4034 : * themselves to the path costs. In theory we should, but in most cases those
4035 : * expressions are trivial and it's just not worth the trouble.
4036 : */
4037 : void
4038 6158 : adjust_limit_rows_costs(double *rows, /* in/out parameter */
4039 : Cost *startup_cost, /* in/out parameter */
4040 : Cost *total_cost, /* in/out parameter */
4041 : int64 offset_est,
4042 : int64 count_est)
4043 : {
4044 6158 : double input_rows = *rows;
4045 6158 : Cost input_startup_cost = *startup_cost;
4046 6158 : Cost input_total_cost = *total_cost;
4047 :
4048 6158 : if (offset_est != 0)
4049 : {
4050 : double offset_rows;
4051 :
4052 694 : if (offset_est > 0)
4053 670 : offset_rows = (double) offset_est;
4054 : else
4055 24 : offset_rows = clamp_row_est(input_rows * 0.10);
4056 694 : if (offset_rows > *rows)
4057 38 : offset_rows = *rows;
4058 694 : if (input_rows > 0)
4059 694 : *startup_cost +=
4060 694 : (input_total_cost - input_startup_cost)
4061 694 : * offset_rows / input_rows;
4062 694 : *rows -= offset_rows;
4063 694 : if (*rows < 1)
4064 46 : *rows = 1;
4065 : }
4066 :
4067 6158 : if (count_est != 0)
4068 : {
4069 : double count_rows;
4070 :
4071 6100 : if (count_est > 0)
4072 6094 : count_rows = (double) count_est;
4073 : else
4074 6 : count_rows = clamp_row_est(input_rows * 0.10);
4075 6100 : if (count_rows > *rows)
4076 248 : count_rows = *rows;
4077 6100 : if (input_rows > 0)
4078 6100 : *total_cost = *startup_cost +
4079 6100 : (input_total_cost - input_startup_cost)
4080 6100 : * count_rows / input_rows;
4081 6100 : *rows = count_rows;
4082 6100 : if (*rows < 1)
4083 0 : *rows = 1;
4084 : }
4085 6158 : }
4086 :
4087 :
4088 : /*
4089 : * reparameterize_path
4090 : * Attempt to modify a Path to have greater parameterization
4091 : *
4092 : * We use this to attempt to bring all child paths of an appendrel to the
4093 : * same parameterization level, ensuring that they all enforce the same set
4094 : * of join quals (and thus that that parameterization can be attributed to
4095 : * an append path built from such paths). Currently, only a few path types
4096 : * are supported here, though more could be added at need. We return NULL
4097 : * if we can't reparameterize the given path.
4098 : *
4099 : * Note: we intentionally do not pass created paths to add_path(); it would
4100 : * possibly try to delete them on the grounds of being cost-inferior to the
4101 : * paths they were made from, and we don't want that. Paths made here are
4102 : * not necessarily of general-purpose usefulness, but they can be useful
4103 : * as members of an append path.
4104 : */
4105 : Path *
4106 350 : reparameterize_path(PlannerInfo *root, Path *path,
4107 : Relids required_outer,
4108 : double loop_count)
4109 : {
4110 350 : RelOptInfo *rel = path->parent;
4111 :
4112 : /* Can only increase, not decrease, path's parameterization */
4113 350 : if (!bms_is_subset(PATH_REQ_OUTER(path), required_outer))
4114 0 : return NULL;
4115 350 : switch (path->pathtype)
4116 : {
4117 264 : case T_SeqScan:
4118 264 : return create_seqscan_path(root, rel, required_outer, 0);
4119 0 : case T_SampleScan:
4120 0 : return (Path *) create_samplescan_path(root, rel, required_outer);
4121 0 : case T_IndexScan:
4122 : case T_IndexOnlyScan:
4123 : {
4124 0 : IndexPath *ipath = (IndexPath *) path;
4125 0 : IndexPath *newpath = makeNode(IndexPath);
4126 :
4127 : /*
4128 : * We can't use create_index_path directly, and would not want
4129 : * to because it would re-compute the indexqual conditions
4130 : * which is wasted effort. Instead we hack things a bit:
4131 : * flat-copy the path node, revise its param_info, and redo
4132 : * the cost estimate.
4133 : */
4134 0 : memcpy(newpath, ipath, sizeof(IndexPath));
4135 0 : newpath->path.param_info =
4136 0 : get_baserel_parampathinfo(root, rel, required_outer);
4137 0 : cost_index(newpath, root, loop_count, false);
4138 0 : return (Path *) newpath;
4139 : }
4140 0 : case T_BitmapHeapScan:
4141 : {
4142 0 : BitmapHeapPath *bpath = (BitmapHeapPath *) path;
4143 :
4144 0 : return (Path *) create_bitmap_heap_path(root,
4145 : rel,
4146 : bpath->bitmapqual,
4147 : required_outer,
4148 : loop_count, 0);
4149 : }
4150 0 : case T_SubqueryScan:
4151 : {
4152 0 : SubqueryScanPath *spath = (SubqueryScanPath *) path;
4153 0 : Path *subpath = spath->subpath;
4154 : bool trivial_pathtarget;
4155 :
4156 : /*
4157 : * If existing node has zero extra cost, we must have decided
4158 : * its target is trivial. (The converse is not true, because
4159 : * it might have a trivial target but quals to enforce; but in
4160 : * that case the new node will too, so it doesn't matter
4161 : * whether we get the right answer here.)
4162 : */
4163 0 : trivial_pathtarget =
4164 0 : (subpath->total_cost == spath->path.total_cost);
4165 :
4166 0 : return (Path *) create_subqueryscan_path(root,
4167 : rel,
4168 : subpath,
4169 : trivial_pathtarget,
4170 : spath->path.pathkeys,
4171 : required_outer);
4172 : }
4173 54 : case T_Result:
4174 : /* Supported only for RTE_RESULT scan paths */
4175 54 : if (IsA(path, Path))
4176 54 : return create_resultscan_path(root, rel, required_outer);
4177 0 : break;
4178 0 : case T_Append:
4179 : {
4180 0 : AppendPath *apath = (AppendPath *) path;
4181 0 : List *childpaths = NIL;
4182 0 : List *partialpaths = NIL;
4183 : int i;
4184 : ListCell *lc;
4185 :
4186 : /* Reparameterize the children */
4187 0 : i = 0;
4188 0 : foreach(lc, apath->subpaths)
4189 : {
4190 0 : Path *spath = (Path *) lfirst(lc);
4191 :
4192 0 : spath = reparameterize_path(root, spath,
4193 : required_outer,
4194 : loop_count);
4195 0 : if (spath == NULL)
4196 0 : return NULL;
4197 : /* We have to re-split the regular and partial paths */
4198 0 : if (i < apath->first_partial_path)
4199 0 : childpaths = lappend(childpaths, spath);
4200 : else
4201 0 : partialpaths = lappend(partialpaths, spath);
4202 0 : i++;
4203 : }
4204 0 : return (Path *)
4205 0 : create_append_path(root, rel, childpaths, partialpaths,
4206 : apath->path.pathkeys, required_outer,
4207 : apath->path.parallel_workers,
4208 0 : apath->path.parallel_aware,
4209 : -1);
4210 : }
4211 0 : case T_Material:
4212 : {
4213 0 : MaterialPath *mpath = (MaterialPath *) path;
4214 0 : Path *spath = mpath->subpath;
4215 :
4216 0 : spath = reparameterize_path(root, spath,
4217 : required_outer,
4218 : loop_count);
4219 0 : if (spath == NULL)
4220 0 : return NULL;
4221 0 : return (Path *) create_material_path(rel, spath);
4222 : }
4223 0 : case T_Memoize:
4224 : {
4225 0 : MemoizePath *mpath = (MemoizePath *) path;
4226 0 : Path *spath = mpath->subpath;
4227 :
4228 0 : spath = reparameterize_path(root, spath,
4229 : required_outer,
4230 : loop_count);
4231 0 : if (spath == NULL)
4232 0 : return NULL;
4233 0 : return (Path *) create_memoize_path(root, rel,
4234 : spath,
4235 : mpath->param_exprs,
4236 : mpath->hash_operators,
4237 0 : mpath->singlerow,
4238 0 : mpath->binary_mode,
4239 : mpath->calls);
4240 : }
4241 32 : default:
4242 32 : break;
4243 : }
4244 32 : return NULL;
4245 : }
4246 :
4247 : /*
4248 : * reparameterize_path_by_child
4249 : * Given a path parameterized by the parent of the given child relation,
4250 : * translate the path to be parameterized by the given child relation.
4251 : *
4252 : * Most fields in the path are not changed, but any expressions must be
4253 : * adjusted to refer to the correct varnos, and any subpaths must be
4254 : * recursively reparameterized. Other fields that refer to specific relids
4255 : * also need adjustment.
4256 : *
4257 : * The cost, number of rows, width and parallel path properties depend upon
4258 : * path->parent, which does not change during the translation. So we need
4259 : * not change those.
4260 : *
4261 : * Currently, only a few path types are supported here, though more could be
4262 : * added at need. We return NULL if we can't reparameterize the given path.
4263 : *
4264 : * Note that this function can change referenced RangeTblEntries, RelOptInfos
4265 : * and IndexOptInfos as well as the Path structures. Therefore, it's only safe
4266 : * to call during create_plan(), when we have made a final choice of which Path
4267 : * to use for each RangeTblEntry/RelOptInfo/IndexOptInfo.
4268 : *
4269 : * Keep this code in sync with path_is_reparameterizable_by_child()!
4270 : */
4271 : Path *
4272 95094 : reparameterize_path_by_child(PlannerInfo *root, Path *path,
4273 : RelOptInfo *child_rel)
4274 : {
4275 : Path *new_path;
4276 : ParamPathInfo *new_ppi;
4277 : ParamPathInfo *old_ppi;
4278 : Relids required_outer;
4279 :
4280 : #define ADJUST_CHILD_ATTRS(node) \
4281 : ((node) = (void *) adjust_appendrel_attrs_multilevel(root, \
4282 : (Node *) (node), \
4283 : child_rel, \
4284 : child_rel->top_parent))
4285 :
4286 : #define REPARAMETERIZE_CHILD_PATH(path) \
4287 : do { \
4288 : (path) = reparameterize_path_by_child(root, (path), child_rel); \
4289 : if ((path) == NULL) \
4290 : return NULL; \
4291 : } while(0)
4292 :
4293 : #define REPARAMETERIZE_CHILD_PATH_LIST(pathlist) \
4294 : do { \
4295 : if ((pathlist) != NIL) \
4296 : { \
4297 : (pathlist) = reparameterize_pathlist_by_child(root, (pathlist), \
4298 : child_rel); \
4299 : if ((pathlist) == NIL) \
4300 : return NULL; \
4301 : } \
4302 : } while(0)
4303 :
4304 : /*
4305 : * If the path is not parameterized by the parent of the given relation,
4306 : * it doesn't need reparameterization.
4307 : */
4308 95094 : if (!path->param_info ||
4309 46642 : !bms_overlap(PATH_REQ_OUTER(path), child_rel->top_parent_relids))
4310 94134 : return path;
4311 :
4312 : /*
4313 : * If possible, reparameterize the given path.
4314 : *
4315 : * This function is currently only applied to the inner side of a nestloop
4316 : * join that is being partitioned by the partitionwise-join code. Hence,
4317 : * we need only support path types that plausibly arise in that context.
4318 : * (In particular, supporting sorted path types would be a waste of code
4319 : * and cycles: even if we translated them here, they'd just lose in
4320 : * subsequent cost comparisons.) If we do see an unsupported path type,
4321 : * that just means we won't be able to generate a partitionwise-join plan
4322 : * using that path type.
4323 : */
4324 960 : switch (nodeTag(path))
4325 : {
4326 228 : case T_Path:
4327 228 : new_path = path;
4328 228 : ADJUST_CHILD_ATTRS(new_path->parent->baserestrictinfo);
4329 228 : if (path->pathtype == T_SampleScan)
4330 : {
4331 48 : Index scan_relid = path->parent->relid;
4332 : RangeTblEntry *rte;
4333 :
4334 : /* it should be a base rel with a tablesample clause... */
4335 : Assert(scan_relid > 0);
4336 48 : rte = planner_rt_fetch(scan_relid, root);
4337 : Assert(rte->rtekind == RTE_RELATION);
4338 : Assert(rte->tablesample != NULL);
4339 :
4340 48 : ADJUST_CHILD_ATTRS(rte->tablesample);
4341 : }
4342 228 : break;
4343 :
4344 480 : case T_IndexPath:
4345 : {
4346 480 : IndexPath *ipath = (IndexPath *) path;
4347 :
4348 480 : ADJUST_CHILD_ATTRS(ipath->indexinfo->indrestrictinfo);
4349 480 : ADJUST_CHILD_ATTRS(ipath->indexclauses);
4350 480 : new_path = (Path *) ipath;
4351 : }
4352 480 : break;
4353 :
4354 48 : case T_BitmapHeapPath:
4355 : {
4356 48 : BitmapHeapPath *bhpath = (BitmapHeapPath *) path;
4357 :
4358 48 : ADJUST_CHILD_ATTRS(bhpath->path.parent->baserestrictinfo);
4359 48 : REPARAMETERIZE_CHILD_PATH(bhpath->bitmapqual);
4360 48 : new_path = (Path *) bhpath;
4361 : }
4362 48 : break;
4363 :
4364 24 : case T_BitmapAndPath:
4365 : {
4366 24 : BitmapAndPath *bapath = (BitmapAndPath *) path;
4367 :
4368 24 : REPARAMETERIZE_CHILD_PATH_LIST(bapath->bitmapquals);
4369 24 : new_path = (Path *) bapath;
4370 : }
4371 24 : break;
4372 :
4373 24 : case T_BitmapOrPath:
4374 : {
4375 24 : BitmapOrPath *bopath = (BitmapOrPath *) path;
4376 :
4377 24 : REPARAMETERIZE_CHILD_PATH_LIST(bopath->bitmapquals);
4378 24 : new_path = (Path *) bopath;
4379 : }
4380 24 : break;
4381 :
4382 0 : case T_ForeignPath:
4383 : {
4384 0 : ForeignPath *fpath = (ForeignPath *) path;
4385 : ReparameterizeForeignPathByChild_function rfpc_func;
4386 :
4387 0 : ADJUST_CHILD_ATTRS(fpath->path.parent->baserestrictinfo);
4388 0 : if (fpath->fdw_outerpath)
4389 0 : REPARAMETERIZE_CHILD_PATH(fpath->fdw_outerpath);
4390 0 : if (fpath->fdw_restrictinfo)
4391 0 : ADJUST_CHILD_ATTRS(fpath->fdw_restrictinfo);
4392 :
4393 : /* Hand over to FDW if needed. */
4394 0 : rfpc_func =
4395 0 : path->parent->fdwroutine->ReparameterizeForeignPathByChild;
4396 0 : if (rfpc_func)
4397 0 : fpath->fdw_private = rfpc_func(root, fpath->fdw_private,
4398 : child_rel);
4399 0 : new_path = (Path *) fpath;
4400 : }
4401 0 : break;
4402 :
4403 0 : case T_CustomPath:
4404 : {
4405 0 : CustomPath *cpath = (CustomPath *) path;
4406 :
4407 0 : ADJUST_CHILD_ATTRS(cpath->path.parent->baserestrictinfo);
4408 0 : REPARAMETERIZE_CHILD_PATH_LIST(cpath->custom_paths);
4409 0 : if (cpath->custom_restrictinfo)
4410 0 : ADJUST_CHILD_ATTRS(cpath->custom_restrictinfo);
4411 0 : if (cpath->methods &&
4412 0 : cpath->methods->ReparameterizeCustomPathByChild)
4413 0 : cpath->custom_private =
4414 0 : cpath->methods->ReparameterizeCustomPathByChild(root,
4415 : cpath->custom_private,
4416 : child_rel);
4417 0 : new_path = (Path *) cpath;
4418 : }
4419 0 : break;
4420 :
4421 36 : case T_NestPath:
4422 : {
4423 36 : NestPath *npath = (NestPath *) path;
4424 36 : JoinPath *jpath = (JoinPath *) npath;
4425 :
4426 36 : REPARAMETERIZE_CHILD_PATH(jpath->outerjoinpath);
4427 36 : REPARAMETERIZE_CHILD_PATH(jpath->innerjoinpath);
4428 36 : ADJUST_CHILD_ATTRS(jpath->joinrestrictinfo);
4429 36 : new_path = (Path *) npath;
4430 : }
4431 36 : break;
4432 :
4433 0 : case T_MergePath:
4434 : {
4435 0 : MergePath *mpath = (MergePath *) path;
4436 0 : JoinPath *jpath = (JoinPath *) mpath;
4437 :
4438 0 : REPARAMETERIZE_CHILD_PATH(jpath->outerjoinpath);
4439 0 : REPARAMETERIZE_CHILD_PATH(jpath->innerjoinpath);
4440 0 : ADJUST_CHILD_ATTRS(jpath->joinrestrictinfo);
4441 0 : ADJUST_CHILD_ATTRS(mpath->path_mergeclauses);
4442 0 : new_path = (Path *) mpath;
4443 : }
4444 0 : break;
4445 :
4446 48 : case T_HashPath:
4447 : {
4448 48 : HashPath *hpath = (HashPath *) path;
4449 48 : JoinPath *jpath = (JoinPath *) hpath;
4450 :
4451 48 : REPARAMETERIZE_CHILD_PATH(jpath->outerjoinpath);
4452 48 : REPARAMETERIZE_CHILD_PATH(jpath->innerjoinpath);
4453 48 : ADJUST_CHILD_ATTRS(jpath->joinrestrictinfo);
4454 48 : ADJUST_CHILD_ATTRS(hpath->path_hashclauses);
4455 48 : new_path = (Path *) hpath;
4456 : }
4457 48 : break;
4458 :
4459 24 : case T_AppendPath:
4460 : {
4461 24 : AppendPath *apath = (AppendPath *) path;
4462 :
4463 24 : REPARAMETERIZE_CHILD_PATH_LIST(apath->subpaths);
4464 24 : new_path = (Path *) apath;
4465 : }
4466 24 : break;
4467 :
4468 0 : case T_MaterialPath:
4469 : {
4470 0 : MaterialPath *mpath = (MaterialPath *) path;
4471 :
4472 0 : REPARAMETERIZE_CHILD_PATH(mpath->subpath);
4473 0 : new_path = (Path *) mpath;
4474 : }
4475 0 : break;
4476 :
4477 48 : case T_MemoizePath:
4478 : {
4479 48 : MemoizePath *mpath = (MemoizePath *) path;
4480 :
4481 48 : REPARAMETERIZE_CHILD_PATH(mpath->subpath);
4482 48 : ADJUST_CHILD_ATTRS(mpath->param_exprs);
4483 48 : new_path = (Path *) mpath;
4484 : }
4485 48 : break;
4486 :
4487 0 : case T_GatherPath:
4488 : {
4489 0 : GatherPath *gpath = (GatherPath *) path;
4490 :
4491 0 : REPARAMETERIZE_CHILD_PATH(gpath->subpath);
4492 0 : new_path = (Path *) gpath;
4493 : }
4494 0 : break;
4495 :
4496 0 : default:
4497 : /* We don't know how to reparameterize this path. */
4498 0 : return NULL;
4499 : }
4500 :
4501 : /*
4502 : * Adjust the parameterization information, which refers to the topmost
4503 : * parent. The topmost parent can be multiple levels away from the given
4504 : * child, hence use multi-level expression adjustment routines.
4505 : */
4506 960 : old_ppi = new_path->param_info;
4507 : required_outer =
4508 960 : adjust_child_relids_multilevel(root, old_ppi->ppi_req_outer,
4509 : child_rel,
4510 960 : child_rel->top_parent);
4511 :
4512 : /* If we already have a PPI for this parameterization, just return it */
4513 960 : new_ppi = find_param_path_info(new_path->parent, required_outer);
4514 :
4515 : /*
4516 : * If not, build a new one and link it to the list of PPIs. For the same
4517 : * reason as explained in mark_dummy_rel(), allocate new PPI in the same
4518 : * context the given RelOptInfo is in.
4519 : */
4520 960 : if (new_ppi == NULL)
4521 : {
4522 : MemoryContext oldcontext;
4523 816 : RelOptInfo *rel = path->parent;
4524 :
4525 816 : oldcontext = MemoryContextSwitchTo(GetMemoryChunkContext(rel));
4526 :
4527 816 : new_ppi = makeNode(ParamPathInfo);
4528 816 : new_ppi->ppi_req_outer = bms_copy(required_outer);
4529 816 : new_ppi->ppi_rows = old_ppi->ppi_rows;
4530 816 : new_ppi->ppi_clauses = old_ppi->ppi_clauses;
4531 816 : ADJUST_CHILD_ATTRS(new_ppi->ppi_clauses);
4532 816 : new_ppi->ppi_serials = bms_copy(old_ppi->ppi_serials);
4533 816 : rel->ppilist = lappend(rel->ppilist, new_ppi);
4534 :
4535 816 : MemoryContextSwitchTo(oldcontext);
4536 : }
4537 960 : bms_free(required_outer);
4538 :
4539 960 : new_path->param_info = new_ppi;
4540 :
4541 : /*
4542 : * Adjust the path target if the parent of the outer relation is
4543 : * referenced in the targetlist. This can happen when only the parent of
4544 : * outer relation is laterally referenced in this relation.
4545 : */
4546 960 : if (bms_overlap(path->parent->lateral_relids,
4547 960 : child_rel->top_parent_relids))
4548 : {
4549 480 : new_path->pathtarget = copy_pathtarget(new_path->pathtarget);
4550 480 : ADJUST_CHILD_ATTRS(new_path->pathtarget->exprs);
4551 : }
4552 :
4553 960 : return new_path;
4554 : }
4555 :
4556 : /*
4557 : * path_is_reparameterizable_by_child
4558 : * Given a path parameterized by the parent of the given child relation,
4559 : * see if it can be translated to be parameterized by the child relation.
4560 : *
4561 : * This must return true if and only if reparameterize_path_by_child()
4562 : * would succeed on this path. Currently it's sufficient to verify that
4563 : * the path and all of its subpaths (if any) are of the types handled by
4564 : * that function. However, subpaths that are not parameterized can be
4565 : * disregarded since they won't require translation.
4566 : */
4567 : bool
4568 34680 : path_is_reparameterizable_by_child(Path *path, RelOptInfo *child_rel)
4569 : {
4570 : #define REJECT_IF_PATH_NOT_REPARAMETERIZABLE(path) \
4571 : do { \
4572 : if (!path_is_reparameterizable_by_child(path, child_rel)) \
4573 : return false; \
4574 : } while(0)
4575 :
4576 : #define REJECT_IF_PATH_LIST_NOT_REPARAMETERIZABLE(pathlist) \
4577 : do { \
4578 : if (!pathlist_is_reparameterizable_by_child(pathlist, child_rel)) \
4579 : return false; \
4580 : } while(0)
4581 :
4582 : /*
4583 : * If the path is not parameterized by the parent of the given relation,
4584 : * it doesn't need reparameterization.
4585 : */
4586 34680 : if (!path->param_info ||
4587 34272 : !bms_overlap(PATH_REQ_OUTER(path), child_rel->top_parent_relids))
4588 984 : return true;
4589 :
4590 : /*
4591 : * Check that the path type is one that reparameterize_path_by_child() can
4592 : * handle, and recursively check subpaths.
4593 : */
4594 33696 : switch (nodeTag(path))
4595 : {
4596 22584 : case T_Path:
4597 : case T_IndexPath:
4598 22584 : break;
4599 :
4600 48 : case T_BitmapHeapPath:
4601 : {
4602 48 : BitmapHeapPath *bhpath = (BitmapHeapPath *) path;
4603 :
4604 48 : REJECT_IF_PATH_NOT_REPARAMETERIZABLE(bhpath->bitmapqual);
4605 : }
4606 48 : break;
4607 :
4608 24 : case T_BitmapAndPath:
4609 : {
4610 24 : BitmapAndPath *bapath = (BitmapAndPath *) path;
4611 :
4612 24 : REJECT_IF_PATH_LIST_NOT_REPARAMETERIZABLE(bapath->bitmapquals);
4613 : }
4614 24 : break;
4615 :
4616 24 : case T_BitmapOrPath:
4617 : {
4618 24 : BitmapOrPath *bopath = (BitmapOrPath *) path;
4619 :
4620 24 : REJECT_IF_PATH_LIST_NOT_REPARAMETERIZABLE(bopath->bitmapquals);
4621 : }
4622 24 : break;
4623 :
4624 148 : case T_ForeignPath:
4625 : {
4626 148 : ForeignPath *fpath = (ForeignPath *) path;
4627 :
4628 148 : if (fpath->fdw_outerpath)
4629 0 : REJECT_IF_PATH_NOT_REPARAMETERIZABLE(fpath->fdw_outerpath);
4630 : }
4631 148 : break;
4632 :
4633 0 : case T_CustomPath:
4634 : {
4635 0 : CustomPath *cpath = (CustomPath *) path;
4636 :
4637 0 : REJECT_IF_PATH_LIST_NOT_REPARAMETERIZABLE(cpath->custom_paths);
4638 : }
4639 0 : break;
4640 :
4641 1248 : case T_NestPath:
4642 : case T_MergePath:
4643 : case T_HashPath:
4644 : {
4645 1248 : JoinPath *jpath = (JoinPath *) path;
4646 :
4647 1248 : REJECT_IF_PATH_NOT_REPARAMETERIZABLE(jpath->outerjoinpath);
4648 1248 : REJECT_IF_PATH_NOT_REPARAMETERIZABLE(jpath->innerjoinpath);
4649 : }
4650 1248 : break;
4651 :
4652 192 : case T_AppendPath:
4653 : {
4654 192 : AppendPath *apath = (AppendPath *) path;
4655 :
4656 192 : REJECT_IF_PATH_LIST_NOT_REPARAMETERIZABLE(apath->subpaths);
4657 : }
4658 192 : break;
4659 :
4660 0 : case T_MaterialPath:
4661 : {
4662 0 : MaterialPath *mpath = (MaterialPath *) path;
4663 :
4664 0 : REJECT_IF_PATH_NOT_REPARAMETERIZABLE(mpath->subpath);
4665 : }
4666 0 : break;
4667 :
4668 9428 : case T_MemoizePath:
4669 : {
4670 9428 : MemoizePath *mpath = (MemoizePath *) path;
4671 :
4672 9428 : REJECT_IF_PATH_NOT_REPARAMETERIZABLE(mpath->subpath);
4673 : }
4674 9428 : break;
4675 :
4676 0 : case T_GatherPath:
4677 : {
4678 0 : GatherPath *gpath = (GatherPath *) path;
4679 :
4680 0 : REJECT_IF_PATH_NOT_REPARAMETERIZABLE(gpath->subpath);
4681 : }
4682 0 : break;
4683 :
4684 0 : default:
4685 : /* We don't know how to reparameterize this path. */
4686 0 : return false;
4687 : }
4688 :
4689 33696 : return true;
4690 : }
4691 :
4692 : /*
4693 : * reparameterize_pathlist_by_child
4694 : * Helper function to reparameterize a list of paths by given child rel.
4695 : *
4696 : * Returns NIL to indicate failure, so pathlist had better not be NIL.
4697 : */
4698 : static List *
4699 72 : reparameterize_pathlist_by_child(PlannerInfo *root,
4700 : List *pathlist,
4701 : RelOptInfo *child_rel)
4702 : {
4703 : ListCell *lc;
4704 72 : List *result = NIL;
4705 :
4706 216 : foreach(lc, pathlist)
4707 : {
4708 144 : Path *path = reparameterize_path_by_child(root, lfirst(lc),
4709 : child_rel);
4710 :
4711 144 : if (path == NULL)
4712 : {
4713 0 : list_free(result);
4714 0 : return NIL;
4715 : }
4716 :
4717 144 : result = lappend(result, path);
4718 : }
4719 :
4720 72 : return result;
4721 : }
4722 :
4723 : /*
4724 : * pathlist_is_reparameterizable_by_child
4725 : * Helper function to check if a list of paths can be reparameterized.
4726 : */
4727 : static bool
4728 240 : pathlist_is_reparameterizable_by_child(List *pathlist, RelOptInfo *child_rel)
4729 : {
4730 : ListCell *lc;
4731 :
4732 720 : foreach(lc, pathlist)
4733 : {
4734 480 : Path *path = (Path *) lfirst(lc);
4735 :
4736 480 : if (!path_is_reparameterizable_by_child(path, child_rel))
4737 0 : return false;
4738 : }
4739 :
4740 240 : return true;
4741 : }
|
