OSDN Git Service

PostgreSQL本体から流用の必要のある関数を追加した。
authorMitsuru Hasegawa <hasegawa@metrosystems.co.jp>
Thu, 23 Feb 2012 02:43:11 +0000 (11:43 +0900)
committerMitsuru Hasegawa <hasegawa@metrosystems.co.jp>
Thu, 23 Feb 2012 02:43:11 +0000 (11:43 +0900)
.gitignore
pg_hint_plan.c

index 9f22fc1..827ef37 100644 (file)
@@ -1,5 +1,6 @@
 # Global excludes across all subdirectories
 *.so
+tags
 
 # Generated subdirectories
 /log/
index d6410a8..bd67a94 100644 (file)
@@ -16,6 +16,9 @@
 #include "utils/builtins.h"
 #include "utils/memutils.h"
 #include "optimizer/cost.h"
+#include "optimizer/joininfo.h"
+#include "optimizer/pathnode.h"
+#include "optimizer/paths.h"
 
 #ifdef PG_MODULE_MAGIC
 PG_MODULE_MAGIC;
@@ -75,6 +78,13 @@ static unsigned int calc_hash(TidList *tidlist);
 static HashEntry *search_ent(TidList *tidlist);
 #endif
 
+static join_search_hook_type org_join_search = NULL;
+
+static RelOptInfo *my_join_search(PlannerInfo *root, int levels_needed,
+                                                         List *initial_rels);
+static void build_join_hints(PlannerInfo *root, int level, List *initial_rels);
+static void my_join_search_one_level(PlannerInfo *root, int level);
+
 PG_FUNCTION_INFO_V1(pg_add_hint);
 PG_FUNCTION_INFO_V1(pg_clear_hint);
 
@@ -90,6 +100,8 @@ _PG_init(void)
        org_cost_hook = cost_hook;
        cost_hook = my_cost_hook;
 #endif
+       org_join_search = join_search_hook;
+       join_search_hook = my_join_search;
        
        for (i = 0 ; i < HASH_ENTRIES ; i++)
                hashent[i] = NULL;
@@ -106,6 +118,7 @@ _PG_fini(void)
 #ifdef NOT_USED
        cost_hook = org_cost_hook;
 #endif
+       join_search_hook = org_join_search;
 
        for (i = 0 ; i < HASH_ENTRIES ; i++)
        {
@@ -639,3 +652,468 @@ pg_dump_hint(PG_FUNCTION_ARGS)
        
        PG_RETURN_CSTRING(cstring_to_text(buf));
 }
+
+/*
+ * pg_add_hint()で登録したヒントから、今回のクエリで使用するもののみ抽出し、
+ * 使用しやすい構造に変換する。
+ */
+static void
+build_join_hints(PlannerInfo *root, int level, List *initial_rels)
+{
+}
+
+/*
+ * src/backend/optimizer/path/joinrels.c
+ * export make_join_rel() をラップする関数
+ * 
+ * ヒントにしたがって、enabele_* パラメータを変更した上で、make_join_rel()を
+ * 呼び出す。
+ */
+static RelOptInfo *
+my_make_join_rel(PlannerInfo *root, RelOptInfo *rel1, RelOptInfo *rel2)
+{
+       return make_join_rel(root, rel1, rel2);
+}
+
+/*
+ * PostgreSQL 本体から流用した関数
+ */
+
+/*
+ * src/backend/optimizer/path/allpaths.c
+ * export standard_join_search() を流用
+ * 
+ * 変更箇所
+ *  build_join_hints() の呼び出しを追加
+ */
+/*
+ * standard_join_search
+ *       Find possible joinpaths for a query by successively finding ways
+ *       to join component relations into join relations.
+ *
+ * 'levels_needed' is the number of iterations needed, ie, the number of
+ *             independent jointree items in the query.  This is > 1.
+ *
+ * 'initial_rels' is a list of RelOptInfo nodes for each independent
+ *             jointree item.  These are the components to be joined together.
+ *             Note that levels_needed == list_length(initial_rels).
+ *
+ * Returns the final level of join relations, i.e., the relation that is
+ * the result of joining all the original relations together.
+ * At least one implementation path must be provided for this relation and
+ * all required sub-relations.
+ *
+ * To support loadable plugins that modify planner behavior by changing the
+ * join searching algorithm, we provide a hook variable that lets a plugin
+ * replace or supplement this function.  Any such hook must return the same
+ * final join relation as the standard code would, but it might have a
+ * different set of implementation paths attached, and only the sub-joinrels
+ * needed for these paths need have been instantiated.
+ *
+ * Note to plugin authors: the functions invoked during standard_join_search()
+ * modify root->join_rel_list and root->join_rel_hash. If you want to do more
+ * than one join-order search, you'll probably need to save and restore the
+ * original states of those data structures.  See geqo_eval() for an example.
+ */
+static RelOptInfo *
+my_join_search(PlannerInfo *root, int levels_needed, List *initial_rels)
+{
+       int                     lev;
+       RelOptInfo *rel;
+
+       /*
+        * This function cannot be invoked recursively within any one planning
+        * problem, so join_rel_level[] can't be in use already.
+        */
+       Assert(root->join_rel_level == NULL);
+
+       /*
+        * We employ a simple "dynamic programming" algorithm: we first find all
+        * ways to build joins of two jointree items, then all ways to build joins
+        * of three items (from two-item joins and single items), then four-item
+        * joins, and so on until we have considered all ways to join all the
+        * items into one rel.
+        *
+        * root->join_rel_level[j] is a list of all the j-item rels.  Initially we
+        * set root->join_rel_level[1] to represent all the single-jointree-item
+        * relations.
+        */
+       root->join_rel_level = (List **) palloc0((levels_needed + 1) * sizeof(List *));
+
+       root->join_rel_level[1] = initial_rels;
+
+       build_join_hints(root, levels_needed, initial_rels);
+
+       for (lev = 2; lev <= levels_needed; lev++)
+       {
+               ListCell   *lc;
+
+               /*
+                * Determine all possible pairs of relations to be joined at this
+                * level, and build paths for making each one from every available
+                * pair of lower-level relations.
+                */
+               my_join_search_one_level(root, lev);
+
+               /*
+                * Do cleanup work on each just-processed rel.
+                */
+               foreach(lc, root->join_rel_level[lev])
+               {
+                       rel = (RelOptInfo *) lfirst(lc);
+
+                       /* Find and save the cheapest paths for this rel */
+                       set_cheapest(rel);
+
+#ifdef OPTIMIZER_DEBUG
+                       debug_print_rel(root, rel);
+#endif
+               }
+       }
+
+       /*
+        * We should have a single rel at the final level.
+        */
+       if (root->join_rel_level[levels_needed] == NIL)
+               elog(ERROR, "failed to build any %d-way joins", levels_needed);
+       Assert(list_length(root->join_rel_level[levels_needed]) == 1);
+
+       rel = (RelOptInfo *) linitial(root->join_rel_level[levels_needed]);
+
+       root->join_rel_level = NULL;
+
+       return rel;
+}
+
+/*
+ * src/backend/optimizer/path/joinrels.c
+ * static join_search_one_level() を流用
+ * 
+ * 変更箇所
+ *  make_join_rel() の呼び出しをラップする、my_make_join_rel()の呼び出しに変更
+ */
+/*
+ * join_search_one_level
+ *       Consider ways to produce join relations containing exactly 'level'
+ *       jointree items.  (This is one step of the dynamic-programming method
+ *       embodied in standard_join_search.)  Join rel nodes for each feasible
+ *       combination of lower-level rels are created and returned in a list.
+ *       Implementation paths are created for each such joinrel, too.
+ *
+ * level: level of rels we want to make this time
+ * root->join_rel_level[j], 1 <= j < level, is a list of rels containing j items
+ *
+ * The result is returned in root->join_rel_level[level].
+ */
+static void
+my_join_search_one_level(PlannerInfo *root, int level)
+{
+       List      **joinrels = root->join_rel_level;
+       ListCell   *r;
+       int                     k;
+
+       Assert(joinrels[level] == NIL);
+
+       /* Set join_cur_level so that new joinrels are added to proper list */
+       root->join_cur_level = level;
+
+       /*
+        * First, consider left-sided and right-sided plans, in which rels of
+        * exactly level-1 member relations are joined against initial relations.
+        * We prefer to join using join clauses, but if we find a rel of level-1
+        * members that has no join clauses, we will generate Cartesian-product
+        * joins against all initial rels not already contained in it.
+        *
+        * In the first pass (level == 2), we try to join each initial rel to each
+        * initial rel that appears later in joinrels[1].  (The mirror-image joins
+        * are handled automatically by make_join_rel.)  In later passes, we try
+        * to join rels of size level-1 from joinrels[level-1] to each initial rel
+        * in joinrels[1].
+        */
+       foreach(r, joinrels[level - 1])
+       {
+               RelOptInfo *old_rel = (RelOptInfo *) lfirst(r);
+               ListCell   *other_rels;
+
+               if (level == 2)
+                       other_rels = lnext(r);          /* only consider remaining initial
+                                                                                * rels */
+               else
+                       other_rels = list_head(joinrels[1]);            /* consider all initial
+                                                                                                                * rels */
+
+               if (old_rel->joininfo != NIL || old_rel->has_eclass_joins ||
+                       has_join_restriction(root, old_rel))
+               {
+                       /*
+                        * Note that if all available join clauses for this rel require
+                        * more than one other rel, we will fail to make any joins against
+                        * it here.  In most cases that's OK; it'll be considered by
+                        * "bushy plan" join code in a higher-level pass where we have
+                        * those other rels collected into a join rel.
+                        *
+                        * See also the last-ditch case below.
+                        */
+                       make_rels_by_clause_joins(root,
+                                                                         old_rel,
+                                                                         other_rels);
+               }
+               else
+               {
+                       /*
+                        * Oops, we have a relation that is not joined to any other
+                        * relation, either directly or by join-order restrictions.
+                        * Cartesian product time.
+                        */
+                       make_rels_by_clauseless_joins(root,
+                                                                                 old_rel,
+                                                                                 other_rels);
+               }
+       }
+
+       /*
+        * Now, consider "bushy plans" in which relations of k initial rels are
+        * joined to relations of level-k initial rels, for 2 <= k <= level-2.
+        *
+        * We only consider bushy-plan joins for pairs of rels where there is a
+        * suitable join clause (or join order restriction), in order to avoid
+        * unreasonable growth of planning time.
+        */
+       for (k = 2;; k++)
+       {
+               int                     other_level = level - k;
+
+               /*
+                * Since make_join_rel(x, y) handles both x,y and y,x cases, we only
+                * need to go as far as the halfway point.
+                */
+               if (k > other_level)
+                       break;
+
+               foreach(r, joinrels[k])
+               {
+                       RelOptInfo *old_rel = (RelOptInfo *) lfirst(r);
+                       ListCell   *other_rels;
+                       ListCell   *r2;
+
+                       /*
+                        * We can ignore clauseless joins here, *except* when they
+                        * participate in join-order restrictions --- then we might have
+                        * to force a bushy join plan.
+                        */
+                       if (old_rel->joininfo == NIL && !old_rel->has_eclass_joins &&
+                               !has_join_restriction(root, old_rel))
+                               continue;
+
+                       if (k == other_level)
+                               other_rels = lnext(r);  /* only consider remaining rels */
+                       else
+                               other_rels = list_head(joinrels[other_level]);
+
+                       for_each_cell(r2, other_rels)
+                       {
+                               RelOptInfo *new_rel = (RelOptInfo *) lfirst(r2);
+
+                               if (!bms_overlap(old_rel->relids, new_rel->relids))
+                               {
+                                       /*
+                                        * OK, we can build a rel of the right level from this
+                                        * pair of rels.  Do so if there is at least one usable
+                                        * join clause or a relevant join restriction.
+                                        */
+                                       if (have_relevant_joinclause(root, old_rel, new_rel) ||
+                                               have_join_order_restriction(root, old_rel, new_rel))
+                                       {
+                                               (void) my_make_join_rel(root, old_rel, new_rel);
+                                       }
+                               }
+                       }
+               }
+       }
+
+       /*
+        * Last-ditch effort: if we failed to find any usable joins so far, force
+        * a set of cartesian-product joins to be generated.  This handles the
+        * special case where all the available rels have join clauses but we
+        * cannot use any of those clauses yet.  An example is
+        *
+        * SELECT * FROM a,b,c WHERE (a.f1 + b.f2 + c.f3) = 0;
+        *
+        * The join clause will be usable at level 3, but at level 2 we have no
+        * choice but to make cartesian joins.  We consider only left-sided and
+        * right-sided cartesian joins in this case (no bushy).
+        */
+       if (joinrels[level] == NIL)
+       {
+               /*
+                * This loop is just like the first one, except we always call
+                * make_rels_by_clauseless_joins().
+                */
+               foreach(r, joinrels[level - 1])
+               {
+                       RelOptInfo *old_rel = (RelOptInfo *) lfirst(r);
+                       ListCell   *other_rels;
+
+                       if (level == 2)
+                               other_rels = lnext(r);  /* only consider remaining initial
+                                                                                * rels */
+                       else
+                               other_rels = list_head(joinrels[1]);    /* consider all initial
+                                                                                                                * rels */
+
+                       make_rels_by_clauseless_joins(root,
+                                                                                 old_rel,
+                                                                                 other_rels);
+               }
+
+               /*----------
+                * When special joins are involved, there may be no legal way
+                * to make an N-way join for some values of N.  For example consider
+                *
+                * SELECT ... FROM t1 WHERE
+                *       x IN (SELECT ... FROM t2,t3 WHERE ...) AND
+                *       y IN (SELECT ... FROM t4,t5 WHERE ...)
+                *
+                * We will flatten this query to a 5-way join problem, but there are
+                * no 4-way joins that join_is_legal() will consider legal.  We have
+                * to accept failure at level 4 and go on to discover a workable
+                * bushy plan at level 5.
+                *
+                * However, if there are no special joins then join_is_legal() should
+                * never fail, and so the following sanity check is useful.
+                *----------
+                */
+               if (joinrels[level] == NIL && root->join_info_list == NIL)
+                       elog(ERROR, "failed to build any %d-way joins", level);
+       }
+}
+
+/*
+ * src/backend/optimizer/path/joinrels.c
+ * static make_rels_by_clause_joins() を流用
+ * 
+ * 変更箇所
+ *  make_join_rel() の呼び出しをラップする、my_make_join_rel()の呼び出しに変更
+ */
+/*
+ * make_rels_by_clause_joins
+ *       Build joins between the given relation 'old_rel' and other relations
+ *       that participate in join clauses that 'old_rel' also participates in
+ *       (or participate in join-order restrictions with it).
+ *       The join rels are returned in root->join_rel_level[join_cur_level].
+ *
+ * Note: at levels above 2 we will generate the same joined relation in
+ * multiple ways --- for example (a join b) join c is the same RelOptInfo as
+ * (b join c) join a, though the second case will add a different set of Paths
+ * to it.  This is the reason for using the join_rel_level mechanism, which
+ * automatically ensures that each new joinrel is only added to the list once.
+ *
+ * 'old_rel' is the relation entry for the relation to be joined
+ * 'other_rels': the first cell in a linked list containing the other
+ * rels to be considered for joining
+ *
+ * Currently, this is only used with initial rels in other_rels, but it
+ * will work for joining to joinrels too.
+ */
+static void
+make_rels_by_clause_joins(PlannerInfo *root,
+                                                 RelOptInfo *old_rel,
+                                                 ListCell *other_rels)
+{
+       ListCell   *l;
+
+       for_each_cell(l, other_rels)
+       {
+               RelOptInfo *other_rel = (RelOptInfo *) lfirst(l);
+
+               if (!bms_overlap(old_rel->relids, other_rel->relids) &&
+                       (have_relevant_joinclause(root, old_rel, other_rel) ||
+                        have_join_order_restriction(root, old_rel, other_rel)))
+               {
+                       (void) my_make_join_rel(root, old_rel, other_rel);
+               }
+       }
+}
+
+/*
+ * src/backend/optimizer/path/joinrels.c
+ * static make_rels_by_clauseless_joins() を流用
+ * 
+ * 変更箇所
+ *  make_join_rel() の呼び出しをラップする、my_make_join_rel()の呼び出しに変更
+ */
+/*
+ * make_rels_by_clauseless_joins
+ *       Given a relation 'old_rel' and a list of other relations
+ *       'other_rels', create a join relation between 'old_rel' and each
+ *       member of 'other_rels' that isn't already included in 'old_rel'.
+ *       The join rels are returned in root->join_rel_level[join_cur_level].
+ *
+ * 'old_rel' is the relation entry for the relation to be joined
+ * 'other_rels': the first cell of a linked list containing the
+ * other rels to be considered for joining
+ *
+ * Currently, this is only used with initial rels in other_rels, but it would
+ * work for joining to joinrels too.
+ */
+static void
+make_rels_by_clauseless_joins(PlannerInfo *root,
+                                                         RelOptInfo *old_rel,
+                                                         ListCell *other_rels)
+{
+       ListCell   *l;
+
+       for_each_cell(l, other_rels)
+       {
+               RelOptInfo *other_rel = (RelOptInfo *) lfirst(l);
+
+               if (!bms_overlap(other_rel->relids, old_rel->relids))
+               {
+                       (void) my_make_join_rel(root, old_rel, other_rel);
+               }
+       }
+}
+
+/*
+ * src/backend/optimizer/path/joinrels.c
+ * static has_join_restriction() を流用
+ * 
+ * 変更箇所
+ *  なし
+ */
+/*
+ * has_join_restriction
+ *             Detect whether the specified relation has join-order restrictions
+ *             due to being inside an outer join or an IN (sub-SELECT).
+ *
+ * Essentially, this tests whether have_join_order_restriction() could
+ * succeed with this rel and some other one.  It's OK if we sometimes
+ * say "true" incorrectly.     (Therefore, we don't bother with the relatively
+ * expensive has_legal_joinclause test.)
+ */
+static bool
+has_join_restriction(PlannerInfo *root, RelOptInfo *rel)
+{
+       ListCell   *l;
+
+       foreach(l, root->join_info_list)
+       {
+               SpecialJoinInfo *sjinfo = (SpecialJoinInfo *) lfirst(l);
+
+               /* ignore full joins --- other mechanisms preserve their ordering */
+               if (sjinfo->jointype == JOIN_FULL)
+                       continue;
+
+               /* ignore if SJ is already contained in rel */
+               if (bms_is_subset(sjinfo->min_lefthand, rel->relids) &&
+                       bms_is_subset(sjinfo->min_righthand, rel->relids))
+                       continue;
+
+               /* restricted if it overlaps LHS or RHS, but doesn't contain SJ */
+               if (bms_overlap(sjinfo->min_lefthand, rel->relids) ||
+                       bms_overlap(sjinfo->min_righthand, rel->relids))
+                       return true;
+       }
+
+       return false;
+}