/* Copyright (c) 2000, 2011, Oracle and/or its affiliates. Copyright (c) 2010, 2011, Monty Program Ab This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; version 2 of the License. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ /* TODO: Fix that MAYBE_KEY are stored in the tree so that we can detect use of full hash keys for queries like: select s.id, kws.keyword_id from sites as s,kws where s.id=kws.site_id and kws.keyword_id in (204,205); */ /* This file contains: RangeAnalysisModule A module that accepts a condition, index (or partitioning) description, and builds lists of intervals (in index/partitioning space), such that all possible records that match the condition are contained within the intervals. The entry point for the range analysis module is get_mm_tree() function. The lists are returned in form of complicated structure of interlinked SEL_TREE/SEL_IMERGE/SEL_ARG objects. See quick_range_seq_next, find_used_partitions for examples of how to walk this structure. All direct "users" of this module are located within this file, too. PartitionPruningModule A module that accepts a partitioned table, condition, and finds which partitions we will need to use in query execution. Search down for "PartitionPruningModule" for description. The module has single entry point - prune_partitions() function. Range/index_merge/groupby-minmax optimizer module A module that accepts a table, condition, and returns - a QUICK_*_SELECT object that can be used to retrieve rows that match the specified condition, or a "no records will match the condition" statement. The module entry points are test_quick_select() get_quick_select_for_ref() Record retrieval code for range/index_merge/groupby-min-max. Implementations of QUICK_*_SELECT classes. KeyTupleFormat ~~~~~~~~~~~~~~ The code in this file (and elsewhere) makes operations on key value tuples. Those tuples are stored in the following format: The tuple is a sequence of key part values. The length of key part value depends only on its type (and not depends on the what value is stored) KeyTuple: keypart1-data, keypart2-data, ... The value of each keypart is stored in the following format: keypart_data: [isnull_byte] keypart-value-bytes If a keypart may have a NULL value (key_part->field->real_maybe_null() can be used to check this), then the first byte is a NULL indicator with the following valid values: 1 - keypart has NULL value. 0 - keypart has non-NULL value. If isnull_byte==1 (NULL value), then the following keypart->length bytes must be 0. keypart-value-bytes holds the value. Its format depends on the field type. The length of keypart-value-bytes may or may not depend on the value being stored. The default is that length is static and equal to KEY_PART_INFO::length. Key parts with (key_part_flag & HA_BLOB_PART) have length depending of the value: keypart-value-bytes: value_length value_bytes The value_length part itself occupies HA_KEY_BLOB_LENGTH=2 bytes. See key_copy() and key_restore() for code to move data between index tuple and table record CAUTION: the above description is only sergefp's understanding of the subject and may omit some details. */ #ifdef USE_PRAGMA_IMPLEMENTATION #pragma implementation // gcc: Class implementation #endif #include "sql_priv.h" #include "key.h" // is_key_used, key_copy, key_cmp, key_restore #include "sql_parse.h" // check_stack_overrun #include "sql_partition.h" // get_part_id_func, PARTITION_ITERATOR, // struct partition_info #include "sql_base.h" // free_io_cache #include "records.h" // init_read_record, end_read_record #include #include "sql_select.h" #ifndef EXTRA_DEBUG #define test_rb_tree(A,B) {} #define test_use_count(A) {} #endif /* Convert double value to #rows. Currently this does floor(), and we might consider using round() instead. */ #define double2rows(x) ((ha_rows)(x)) static int sel_cmp(Field *f,uchar *a,uchar *b,uint8 a_flag,uint8 b_flag); static uchar is_null_string[2]= {1,0}; class RANGE_OPT_PARAM; /* A construction block of the SEL_ARG-graph. The following description only covers graphs of SEL_ARG objects with sel_arg->type==KEY_RANGE: One SEL_ARG object represents an "elementary interval" in form min_value <=? table.keypartX <=? max_value The interval is a non-empty interval of any kind: with[out] minimum/maximum bound, [half]open/closed, single-point interval, etc. 1. SEL_ARG GRAPH STRUCTURE SEL_ARG objects are linked together in a graph. The meaning of the graph is better demostrated by an example: tree->keys[i] | | $ $ | part=1 $ part=2 $ part=3 | $ $ | +-------+ $ +-------+ $ +--------+ | | kp1<1 |--$-->| kp2=5 |--$-->| kp3=10 | | +-------+ $ +-------+ $ +--------+ | | $ $ | | | $ $ +--------+ | | $ $ | kp3=12 | | | $ $ +--------+ | +-------+ $ $ \->| kp1=2 |--$--------------$-+ +-------+ $ $ | +--------+ | $ $ ==>| kp3=11 | +-------+ $ $ | +--------+ | kp1=3 |--$--------------$-+ | +-------+ $ $ +--------+ | $ $ | kp3=14 | ... $ $ +--------+ The entire graph is partitioned into "interval lists". An interval list is a sequence of ordered disjoint intervals over the same key part. SEL_ARG are linked via "next" and "prev" pointers. Additionally, all intervals in the list form an RB-tree, linked via left/right/parent pointers. The RB-tree root SEL_ARG object will be further called "root of the interval list". In the example pic, there are 4 interval lists: "kp<1 OR kp1=2 OR kp1=3", "kp2=5", "kp3=10 OR kp3=12", "kp3=11 OR kp3=13". The vertical lines represent SEL_ARG::next/prev pointers. In an interval list, each member X may have SEL_ARG::next_key_part pointer pointing to the root of another interval list Y. The pointed interval list must cover a key part with greater number (i.e. Y->part > X->part). In the example pic, the next_key_part pointers are represented by horisontal lines. 2. SEL_ARG GRAPH SEMANTICS It represents a condition in a special form (we don't have a name for it ATM) The SEL_ARG::next/prev is "OR", and next_key_part is "AND". For example, the picture represents the condition in form: (kp1 < 1 AND kp2=5 AND (kp3=10 OR kp3=12)) OR (kp1=2 AND (kp3=11 OR kp3=14)) OR (kp1=3 AND (kp3=11 OR kp3=14)) 3. SEL_ARG GRAPH USE Use get_mm_tree() to construct SEL_ARG graph from WHERE condition. Then walk the SEL_ARG graph and get a list of dijsoint ordered key intervals (i.e. intervals in form (constA1, .., const1_K) < (keypart1,.., keypartK) < (constB1, .., constB_K) Those intervals can be used to access the index. The uses are in: - check_quick_select() - Walk the SEL_ARG graph and find an estimate of how many table records are contained within all intervals. - get_quick_select() - Walk the SEL_ARG, materialize the key intervals, and create QUICK_RANGE_SELECT object that will read records within these intervals. 4. SPACE COMPLEXITY NOTES SEL_ARG graph is a representation of an ordered disjoint sequence of intervals over the ordered set of index tuple values. For multi-part keys, one can construct a WHERE expression such that its list of intervals will be of combinatorial size. Here is an example: (keypart1 IN (1,2, ..., n1)) AND (keypart2 IN (1,2, ..., n2)) AND (keypart3 IN (1,2, ..., n3)) For this WHERE clause the list of intervals will have n1*n2*n3 intervals of form (keypart1, keypart2, keypart3) = (k1, k2, k3), where 1 <= k{i} <= n{i} SEL_ARG graph structure aims to reduce the amount of required space by "sharing" the elementary intervals when possible (the pic at the beginning of this comment has examples of such sharing). The sharing may prevent combinatorial blowup: There are WHERE clauses that have combinatorial-size interval lists but will be represented by a compact SEL_ARG graph. Example: (keypartN IN (1,2, ..., n1)) AND ... (keypart2 IN (1,2, ..., n2)) AND (keypart1 IN (1,2, ..., n3)) but not in all cases: - There are WHERE clauses that do have a compact SEL_ARG-graph representation but get_mm_tree() and its callees will construct a graph of combinatorial size. Example: (keypart1 IN (1,2, ..., n1)) AND (keypart2 IN (1,2, ..., n2)) AND ... (keypartN IN (1,2, ..., n3)) - There are WHERE clauses for which the minimal possible SEL_ARG graph representation will have combinatorial size. Example: By induction: Let's take any interval on some keypart in the middle: kp15=c0 Then let's AND it with this interval 'structure' from preceding and following keyparts: (kp14=c1 AND kp16=c3) OR keypart14=c2) (*) We will obtain this SEL_ARG graph: kp14 $ kp15 $ kp16 $ $ +---------+ $ +---------+ $ +---------+ | kp14=c1 |--$-->| kp15=c0 |--$-->| kp16=c3 | +---------+ $ +---------+ $ +---------+ | $ $ +---------+ $ +---------+ $ | kp14=c2 |--$-->| kp15=c0 | $ +---------+ $ +---------+ $ $ $ Note that we had to duplicate "kp15=c0" and there was no way to avoid that. The induction step: AND the obtained expression with another "wrapping" expression like (*). When the process ends because of the limit on max. number of keyparts we'll have: WHERE clause length is O(3*#max_keyparts) SEL_ARG graph size is O(2^(#max_keyparts/2)) (it is also possible to construct a case where instead of 2 in 2^n we have a bigger constant, e.g. 4, and get a graph with 4^(31/2)= 2^31 nodes) We avoid consuming too much memory by setting a limit on the number of SEL_ARG object we can construct during one range analysis invocation. */ class SEL_ARG :public Sql_alloc { public: uint8 min_flag,max_flag,maybe_flag; uint8 part; // Which key part uint8 maybe_null; /* The ordinal number the least significant component encountered in the ranges of the SEL_ARG tree (the first component has number 1) */ uint16 max_part_no; /* Number of children of this element in the RB-tree, plus 1 for this element itself. */ uint16 elements; /* Valid only for elements which are RB-tree roots: Number of times this RB-tree is referred to (it is referred by SEL_ARG::next_key_part or by SEL_TREE::keys[i] or by a temporary SEL_ARG* variable) */ ulong use_count; Field *field; uchar *min_value,*max_value; // Pointer to range /* eq_tree() requires that left == right == 0 if the type is MAYBE_KEY. */ SEL_ARG *left,*right; /* R-B tree children */ SEL_ARG *next,*prev; /* Links for bi-directional interval list */ SEL_ARG *parent; /* R-B tree parent */ SEL_ARG *next_key_part; enum leaf_color { BLACK,RED } color; enum Type { IMPOSSIBLE, MAYBE, MAYBE_KEY, KEY_RANGE } type; enum { MAX_SEL_ARGS = 16000 }; SEL_ARG() {} SEL_ARG(SEL_ARG &); SEL_ARG(Field *,const uchar *, const uchar *); SEL_ARG(Field *field, uint8 part, uchar *min_value, uchar *max_value, uint8 min_flag, uint8 max_flag, uint8 maybe_flag); SEL_ARG(enum Type type_arg) :min_flag(0), max_part_no(0) /* first key part means 1. 0 mean 'no parts'*/, elements(1),use_count(1),left(0),right(0), next_key_part(0), color(BLACK), type(type_arg) {} inline bool is_same(SEL_ARG *arg) { if (type != arg->type || part != arg->part) return 0; if (type != KEY_RANGE) return 1; return cmp_min_to_min(arg) == 0 && cmp_max_to_max(arg) == 0; } inline void merge_flags(SEL_ARG *arg) { maybe_flag|=arg->maybe_flag; } inline void maybe_smaller() { maybe_flag=1; } /* Return true iff it's a single-point null interval */ inline bool is_null_interval() { return maybe_null && max_value[0] == 1; } inline int cmp_min_to_min(SEL_ARG* arg) { return sel_cmp(field,min_value, arg->min_value, min_flag, arg->min_flag); } inline int cmp_min_to_max(SEL_ARG* arg) { return sel_cmp(field,min_value, arg->max_value, min_flag, arg->max_flag); } inline int cmp_max_to_max(SEL_ARG* arg) { return sel_cmp(field,max_value, arg->max_value, max_flag, arg->max_flag); } inline int cmp_max_to_min(SEL_ARG* arg) { return sel_cmp(field,max_value, arg->min_value, max_flag, arg->min_flag); } SEL_ARG *clone_and(SEL_ARG* arg) { // Get overlapping range uchar *new_min,*new_max; uint8 flag_min,flag_max; if (cmp_min_to_min(arg) >= 0) { new_min=min_value; flag_min=min_flag; } else { new_min=arg->min_value; flag_min=arg->min_flag; /* purecov: deadcode */ } if (cmp_max_to_max(arg) <= 0) { new_max=max_value; flag_max=max_flag; } else { new_max=arg->max_value; flag_max=arg->max_flag; } return new SEL_ARG(field, part, new_min, new_max, flag_min, flag_max, test(maybe_flag && arg->maybe_flag)); } SEL_ARG *clone_first(SEL_ARG *arg) { // min <= X < arg->min return new SEL_ARG(field,part, min_value, arg->min_value, min_flag, arg->min_flag & NEAR_MIN ? 0 : NEAR_MAX, maybe_flag | arg->maybe_flag); } SEL_ARG *clone_last(SEL_ARG *arg) { // min <= X <= key_max return new SEL_ARG(field, part, min_value, arg->max_value, min_flag, arg->max_flag, maybe_flag | arg->maybe_flag); } SEL_ARG *clone(RANGE_OPT_PARAM *param, SEL_ARG *new_parent, SEL_ARG **next); bool copy_min(SEL_ARG* arg) { // Get overlapping range if (cmp_min_to_min(arg) > 0) { min_value=arg->min_value; min_flag=arg->min_flag; if ((max_flag & (NO_MAX_RANGE | NO_MIN_RANGE)) == (NO_MAX_RANGE | NO_MIN_RANGE)) return 1; // Full range } maybe_flag|=arg->maybe_flag; return 0; } bool copy_max(SEL_ARG* arg) { // Get overlapping range if (cmp_max_to_max(arg) <= 0) { max_value=arg->max_value; max_flag=arg->max_flag; if ((max_flag & (NO_MAX_RANGE | NO_MIN_RANGE)) == (NO_MAX_RANGE | NO_MIN_RANGE)) return 1; // Full range } maybe_flag|=arg->maybe_flag; return 0; } void copy_min_to_min(SEL_ARG *arg) { min_value=arg->min_value; min_flag=arg->min_flag; } void copy_min_to_max(SEL_ARG *arg) { max_value=arg->min_value; max_flag=arg->min_flag & NEAR_MIN ? 0 : NEAR_MAX; } void copy_max_to_min(SEL_ARG *arg) { min_value=arg->max_value; min_flag=arg->max_flag & NEAR_MAX ? 0 : NEAR_MIN; } /* returns a number of keypart values (0 or 1) appended to the key buffer */ int store_min(uint length, uchar **min_key,uint min_key_flag) { /* "(kp1 > c1) AND (kp2 OP c2) AND ..." -> (kp1 > c1) */ if ((min_flag & GEOM_FLAG) || (!(min_flag & NO_MIN_RANGE) && !(min_key_flag & (NO_MIN_RANGE | NEAR_MIN)))) { if (maybe_null && *min_value) { **min_key=1; bzero(*min_key+1,length-1); } else memcpy(*min_key,min_value,length); (*min_key)+= length; return 1; } return 0; } /* returns a number of keypart values (0 or 1) appended to the key buffer */ int store_max(uint length, uchar **max_key, uint max_key_flag) { if (!(max_flag & NO_MAX_RANGE) && !(max_key_flag & (NO_MAX_RANGE | NEAR_MAX))) { if (maybe_null && *max_value) { **max_key=1; bzero(*max_key+1,length-1); } else memcpy(*max_key,max_value,length); (*max_key)+= length; return 1; } return 0; } /* Returns a number of keypart values appended to the key buffer for min key and max key. This function is used by both Range Analysis and Partition pruning. For partition pruning we have to ensure that we don't store also subpartition fields. Thus we have to stop at the last partition part and not step into the subpartition fields. For Range Analysis we set last_part to MAX_KEY which we should never reach. */ int store_min_key(KEY_PART *key, uchar **range_key, uint *range_key_flag, uint last_part) { SEL_ARG *key_tree= first(); uint res= key_tree->store_min(key[key_tree->part].store_length, range_key, *range_key_flag); *range_key_flag|= key_tree->min_flag; if (key_tree->next_key_part && key_tree->next_key_part->type == SEL_ARG::KEY_RANGE && key_tree->part != last_part && key_tree->next_key_part->part == key_tree->part+1 && !(*range_key_flag & (NO_MIN_RANGE | NEAR_MIN))) res+= key_tree->next_key_part->store_min_key(key, range_key, range_key_flag, last_part); return res; } /* returns a number of keypart values appended to the key buffer */ int store_max_key(KEY_PART *key, uchar **range_key, uint *range_key_flag, uint last_part) { SEL_ARG *key_tree= last(); uint res=key_tree->store_max(key[key_tree->part].store_length, range_key, *range_key_flag); (*range_key_flag)|= key_tree->max_flag; if (key_tree->next_key_part && key_tree->next_key_part->type == SEL_ARG::KEY_RANGE && key_tree->part != last_part && key_tree->next_key_part->part == key_tree->part+1 && !(*range_key_flag & (NO_MAX_RANGE | NEAR_MAX))) res+= key_tree->next_key_part->store_max_key(key, range_key, range_key_flag, last_part); return res; } SEL_ARG *insert(SEL_ARG *key); SEL_ARG *tree_delete(SEL_ARG *key); SEL_ARG *find_range(SEL_ARG *key); SEL_ARG *rb_insert(SEL_ARG *leaf); friend SEL_ARG *rb_delete_fixup(SEL_ARG *root,SEL_ARG *key, SEL_ARG *par); #ifdef EXTRA_DEBUG friend int test_rb_tree(SEL_ARG *element,SEL_ARG *parent); void test_use_count(SEL_ARG *root); #endif SEL_ARG *first(); SEL_ARG *last(); void make_root(); inline bool simple_key() { return !next_key_part && elements == 1; } void increment_use_count(long count) { if (next_key_part) { next_key_part->use_count+=count; count*= (next_key_part->use_count-count); for (SEL_ARG *pos=next_key_part->first(); pos ; pos=pos->next) if (pos->next_key_part) pos->increment_use_count(count); } } void incr_refs() { increment_use_count(1); use_count++; } void free_tree() { for (SEL_ARG *pos=first(); pos ; pos=pos->next) if (pos->next_key_part) { pos->next_key_part->use_count--; pos->next_key_part->free_tree(); } } inline SEL_ARG **parent_ptr() { return parent->left == this ? &parent->left : &parent->right; } /* Check if this SEL_ARG object represents a single-point interval SYNOPSIS is_singlepoint() DESCRIPTION Check if this SEL_ARG object (not tree) represents a single-point interval, i.e. if it represents a "keypart = const" or "keypart IS NULL". RETURN TRUE This SEL_ARG object represents a singlepoint interval FALSE Otherwise */ bool is_singlepoint() { /* Check for NEAR_MIN ("strictly less") and NO_MIN_RANGE (-inf < field) flags, and the same for right edge. */ if (min_flag || max_flag) return FALSE; uchar *min_val= min_value; uchar *max_val= max_value; if (maybe_null) { /* First byte is a NULL value indicator */ if (*min_val != *max_val) return FALSE; if (*min_val) return TRUE; /* This "x IS NULL" */ min_val++; max_val++; } return !field->key_cmp(min_val, max_val); } SEL_ARG *clone_tree(RANGE_OPT_PARAM *param); }; class SEL_IMERGE; #define CLONE_KEY1_MAYBE 1 #define CLONE_KEY2_MAYBE 2 #define swap_clone_flag(A) ((A & 1) << 1) | ((A & 2) >> 1) /* While objects of the class SEL_ARG represent ranges for indexes or index infixes (including ranges for index prefixes and index suffixes), objects of the class SEL_TREE represent AND/OR formulas of such ranges. Currently an AND/OR formula represented by a SEL_TREE object can have at most three levels: ::= [ AND ] [ [ AND ...] ] ::= [ AND ... ] ::= [ OR ] As we can see from the above definitions: - SEL_RANGE_TREE formula is a conjunction of SEL_ARG formulas - SEL_IMERGE formula is a disjunction of SEL_RANGE_TREE formulas - SEL_TREE formula is a conjunction of a SEL_RANGE_TREE formula and SEL_IMERGE formulas. It's required above that a SEL_TREE formula has at least one conjunct. Usually we will consider normalized SEL_RANGE_TREE formulas where we use TRUE as conjunct members for those indexes whose SEL_ARG trees are empty. We will call an SEL_TREE object simply 'tree'. The part of a tree that represents SEL_RANGE_TREE formula is called 'range part' of the tree while the remaining part is called 'imerge part'. If a tree contains only a range part then we call such a tree 'range tree'. Components of a range tree that represent SEL_ARG formulas are called ranges. If a tree does not contain any range part we call such a tree 'imerge tree'. Components of the imerge part of a tree that represent SEL_IMERGE formula are called imerges. Usually we'll designate: SEL_TREE formulas by T_1,...,T_k SEL_ARG formulas by R_1,...,R_k SEL_RANGE_TREE formulas by RT_1,...,RT_k SEL_IMERGE formulas by M_1,...,M_k Accordingly we'll use: t_1,...,t_k - to designate trees representing T_1,...,T_k r_1,...,r_k - to designate ranges representing R_1,...,R_k rt_1,...,r_tk - to designate range trees representing RT_1,...,RT_k m_1,...,m_k - to designate imerges representing M_1,...,M_k SEL_TREE objects are usually built from WHERE conditions or ON expressions. A SEL_TREE object always represents an inference of the condition it is built from. Therefore, if a row satisfies a SEL_TREE formula it also satisfies the condition it is built from. The following transformations of tree t representing SEL_TREE formula T yield a new tree t1 thar represents an inference of T: T=>T1. (1) remove any of SEL_ARG tree from the range part of t (2) remove any imerge from the tree t (3) remove any of SEL_ARG tree from any range tree contained in any imerge of tree Since the basic blocks of any SEL_TREE objects are ranges, SEL_TREE objects in many cases can be effectively used to filter out a big part of table rows that do not satisfy WHERE/IN conditions utilizing only single or multiple range index scans. A single range index scan is constructed for a range tree that contains only one SEL_ARG object for an index or an index prefix. An index intersection scan can be constructed for a range tree that contains several SEL_ARG objects. Currently index intersection scans are constructed only for single-point ranges. An index merge scan is constructed for a imerge tree that contains only one imerge. If range trees of this imerge contain only single-point merges than a union of index intersections can be built. Usually the tree built by the range optimizer for a query table contains more than one range in the range part, and additionally may contain some imerges in the imerge part. The range optimizer evaluates all of them one by one and chooses the range or the imerge that provides the cheapest single or multiple range index scan of the table. According to rules (1)-(3) this scan always filter out only those rows that do not satisfy the query conditions. For any condition the SEL_TREE object for it is built in a bottom up manner starting from the range trees for the predicates. The tree_and function builds a tree for any conjunction of formulas from the trees for its conjuncts. The tree_or function builds a tree for any disjunction of formulas from the trees for its disjuncts. */ class SEL_TREE :public Sql_alloc { public: /* Starting an effort to document this field: (for some i, keys[i]->type == SEL_ARG::IMPOSSIBLE) => (type == SEL_TREE::IMPOSSIBLE) */ enum Type { IMPOSSIBLE, ALWAYS, MAYBE, KEY, KEY_SMALLER } type; SEL_TREE(enum Type type_arg) :type(type_arg) {} SEL_TREE() :type(KEY) { keys_map.clear_all(); bzero((char*) keys,sizeof(keys)); } SEL_TREE(SEL_TREE *arg, bool without_merges, RANGE_OPT_PARAM *param); /* Note: there may exist SEL_TREE objects with sel_tree->type=KEY and keys[i]=0 for all i. (SergeyP: it is not clear whether there is any merit in range analyzer functions (e.g. get_mm_parts) returning a pointer to such SEL_TREE instead of NULL) */ SEL_ARG *keys[MAX_KEY]; key_map keys_map; /* bitmask of non-NULL elements in keys */ /* Possible ways to read rows using index_merge. The list is non-empty only if type==KEY. Currently can be non empty only if keys_map.is_clear_all(). */ List merges; /* The members below are filled/used only after get_mm_tree is done */ key_map ror_scans_map; /* bitmask of ROR scan-able elements in keys */ uint n_ror_scans; /* number of set bits in ror_scans_map */ struct st_index_scan_info **index_scans; /* list of index scans */ struct st_index_scan_info **index_scans_end; /* last index scan */ struct st_ror_scan_info **ror_scans; /* list of ROR key scans */ struct st_ror_scan_info **ror_scans_end; /* last ROR scan */ /* Note that #records for each key scan is stored in table->quick_rows */ bool without_ranges() { return keys_map.is_clear_all(); } bool without_imerges() { return merges.is_empty(); } }; class RANGE_OPT_PARAM { public: THD *thd; /* Current thread handle */ TABLE *table; /* Table being analyzed */ COND *cond; /* Used inside get_mm_tree(). */ table_map prev_tables; table_map read_tables; table_map current_table; /* Bit of the table being analyzed */ /* Array of parts of all keys for which range analysis is performed */ KEY_PART *key_parts; KEY_PART *key_parts_end; MEM_ROOT *mem_root; /* Memory that will be freed when range analysis completes */ MEM_ROOT *old_root; /* Memory that will last until the query end */ /* Number of indexes used in range analysis (In SEL_TREE::keys only first #keys elements are not empty) */ uint keys; /* If true, the index descriptions describe real indexes (and it is ok to call field->optimize_range(real_keynr[...], ...). Otherwise index description describes fake indexes. */ bool using_real_indexes; /* Aggressively remove "scans" that do not have conditions on first keyparts. Such scans are usable when doing partition pruning but not regular range optimization. */ bool remove_jump_scans; /* used_key_no -> table_key_no translation table. Only makes sense if using_real_indexes==TRUE */ uint real_keynr[MAX_KEY]; /* Used to store 'current key tuples', in both range analysis and partitioning (list) analysis */ uchar min_key[MAX_KEY_LENGTH+MAX_FIELD_WIDTH], max_key[MAX_KEY_LENGTH+MAX_FIELD_WIDTH]; /* Number of SEL_ARG objects allocated by SEL_ARG::clone_tree operations */ uint alloced_sel_args; bool force_default_mrr; KEY_PART *key[MAX_KEY]; /* First key parts of keys used in the query */ }; class PARAM : public RANGE_OPT_PARAM { public: ha_rows quick_rows[MAX_KEY]; longlong baseflag; uint max_key_part, range_count; bool quick; // Don't calulate possible keys uint fields_bitmap_size; MY_BITMAP needed_fields; /* bitmask of fields needed by the query */ MY_BITMAP tmp_covered_fields; key_map *needed_reg; /* ptr to SQL_SELECT::needed_reg */ uint *imerge_cost_buff; /* buffer for index_merge cost estimates */ uint imerge_cost_buff_size; /* size of the buffer */ /* TRUE if last checked tree->key can be used for ROR-scan */ bool is_ror_scan; /* Number of ranges in the last checked tree->key */ uint n_ranges; uint8 first_null_comp; /* first null component if any, 0 - otherwise */ }; class TABLE_READ_PLAN; class TRP_RANGE; class TRP_ROR_INTERSECT; class TRP_ROR_UNION; class TRP_INDEX_INTERSECT; class TRP_INDEX_MERGE; class TRP_GROUP_MIN_MAX; struct st_index_scan_info; struct st_ror_scan_info; static SEL_TREE * get_mm_parts(RANGE_OPT_PARAM *param,COND *cond_func,Field *field, Item_func::Functype type,Item *value, Item_result cmp_type); static SEL_ARG *get_mm_leaf(RANGE_OPT_PARAM *param,COND *cond_func,Field *field, KEY_PART *key_part, Item_func::Functype type,Item *value); static SEL_TREE *get_mm_tree(RANGE_OPT_PARAM *param,COND *cond); static bool is_key_scan_ror(PARAM *param, uint keynr, uint8 nparts); static ha_rows check_quick_select(PARAM *param, uint idx, bool index_only, SEL_ARG *tree, bool update_tbl_stats, uint *mrr_flags, uint *bufsize, COST_VECT *cost); QUICK_RANGE_SELECT *get_quick_select(PARAM *param,uint index, SEL_ARG *key_tree, uint mrr_flags, uint mrr_buf_size, MEM_ROOT *alloc); static TRP_RANGE *get_key_scans_params(PARAM *param, SEL_TREE *tree, bool index_read_must_be_used, bool update_tbl_stats, double read_time); static TRP_INDEX_INTERSECT *get_best_index_intersect(PARAM *param, SEL_TREE *tree, double read_time); static TRP_ROR_INTERSECT *get_best_ror_intersect(const PARAM *param, SEL_TREE *tree, double read_time, bool *are_all_covering); static TRP_ROR_INTERSECT *get_best_covering_ror_intersect(PARAM *param, SEL_TREE *tree, double read_time); static TABLE_READ_PLAN *get_best_disjunct_quick(PARAM *param, SEL_IMERGE *imerge, double read_time); static TABLE_READ_PLAN *merge_same_index_scans(PARAM *param, SEL_IMERGE *imerge, TRP_INDEX_MERGE *imerge_trp, double read_time); static TRP_GROUP_MIN_MAX *get_best_group_min_max(PARAM *param, SEL_TREE *tree, double read_time); #ifndef DBUG_OFF static void print_sel_tree(PARAM *param, SEL_TREE *tree, key_map *tree_map, const char *msg); static void print_ror_scans_arr(TABLE *table, const char *msg, struct st_ror_scan_info **start, struct st_ror_scan_info **end); static void print_quick(QUICK_SELECT_I *quick, const key_map *needed_reg); #endif static SEL_TREE *tree_and(RANGE_OPT_PARAM *param, SEL_TREE *tree1, SEL_TREE *tree2); static SEL_TREE *tree_or(RANGE_OPT_PARAM *param, SEL_TREE *tree1,SEL_TREE *tree2); static SEL_ARG *sel_add(SEL_ARG *key1,SEL_ARG *key2); static SEL_ARG *key_or(RANGE_OPT_PARAM *param, SEL_ARG *key1, SEL_ARG *key2); static SEL_ARG *key_and(RANGE_OPT_PARAM *param, SEL_ARG *key1, SEL_ARG *key2, uint clone_flag); static bool get_range(SEL_ARG **e1,SEL_ARG **e2,SEL_ARG *root1); bool get_quick_keys(PARAM *param,QUICK_RANGE_SELECT *quick,KEY_PART *key, SEL_ARG *key_tree, uchar *min_key,uint min_key_flag, uchar *max_key,uint max_key_flag); static bool eq_tree(SEL_ARG* a,SEL_ARG *b); static SEL_ARG null_element(SEL_ARG::IMPOSSIBLE); static bool null_part_in_key(KEY_PART *key_part, const uchar *key, uint length); static bool is_key_scan_ror(PARAM *param, uint keynr, uint8 nparts); #include "opt_range_mrr.cc" static bool sel_trees_have_common_keys(SEL_TREE *tree1, SEL_TREE *tree2, key_map *common_keys); static void eliminate_single_tree_imerges(RANGE_OPT_PARAM *param, SEL_TREE *tree); static bool sel_trees_can_be_ored(RANGE_OPT_PARAM* param, SEL_TREE *tree1, SEL_TREE *tree2, key_map *common_keys); static bool sel_trees_must_be_ored(RANGE_OPT_PARAM* param, SEL_TREE *tree1, SEL_TREE *tree2, key_map common_keys); static int and_range_trees(RANGE_OPT_PARAM *param, SEL_TREE *tree1, SEL_TREE *tree2, SEL_TREE *result); static bool remove_nonrange_trees(RANGE_OPT_PARAM *param, SEL_TREE *tree); /* SEL_IMERGE is a list of possible ways to do index merge, i.e. it is a condition in the following form: (t_1||t_2||...||t_N) && (next) where all t_i are SEL_TREEs, next is another SEL_IMERGE and no pair (t_i,t_j) contains SEL_ARGS for the same index. SEL_TREE contained in SEL_IMERGE always has merges=NULL. This class relies on memory manager to do the cleanup. */ class SEL_IMERGE : public Sql_alloc { enum { PREALLOCED_TREES= 10}; public: SEL_TREE *trees_prealloced[PREALLOCED_TREES]; SEL_TREE **trees; /* trees used to do index_merge */ SEL_TREE **trees_next; /* last of these trees */ SEL_TREE **trees_end; /* end of allocated space */ SEL_ARG ***best_keys; /* best keys to read in SEL_TREEs */ SEL_IMERGE() : trees(&trees_prealloced[0]), trees_next(trees), trees_end(trees + PREALLOCED_TREES) {} SEL_IMERGE (SEL_IMERGE *arg, uint cnt, RANGE_OPT_PARAM *param); int or_sel_tree(RANGE_OPT_PARAM *param, SEL_TREE *tree); bool have_common_keys(RANGE_OPT_PARAM *param, SEL_TREE *tree); int and_sel_tree(RANGE_OPT_PARAM *param, SEL_TREE *tree, SEL_IMERGE *new_imerge); int or_sel_tree_with_checks(RANGE_OPT_PARAM *param, uint n_init_trees, SEL_TREE *new_tree, bool is_first_check_pass, bool *is_last_check_pass); int or_sel_imerge_with_checks(RANGE_OPT_PARAM *param, uint n_init_trees, SEL_IMERGE* imerge, bool is_first_check_pass, bool *is_last_check_pass); }; /* Add a range tree to the range trees of this imerge SYNOPSIS or_sel_tree() param Context info for the operation tree SEL_TREE to add to this imerge DESCRIPTION The function just adds the range tree 'tree' to the range trees of this imerge. RETURN 0 if the operation is success -1 if the function runs out memory */ int SEL_IMERGE::or_sel_tree(RANGE_OPT_PARAM *param, SEL_TREE *tree) { if (trees_next == trees_end) { const int realloc_ratio= 2; /* Double size for next round */ uint old_elements= (trees_end - trees); uint old_size= sizeof(SEL_TREE**) * old_elements; uint new_size= old_size * realloc_ratio; SEL_TREE **new_trees; if (!(new_trees= (SEL_TREE**)alloc_root(param->mem_root, new_size))) return -1; memcpy(new_trees, trees, old_size); trees= new_trees; trees_next= trees + old_elements; trees_end= trees + old_elements * realloc_ratio; } *(trees_next++)= tree; return 0; } /* Check if any of the range trees of this imerge intersects with a given tree SYNOPSIS have_common_keys() param Context info for the function tree SEL_TREE intersection with the imerge range trees is checked for DESCRIPTION The function checks whether there is any range tree rt_i in this imerge such that there are some indexes for which ranges are defined in both rt_i and the range part of the SEL_TREE tree. To check this the function calls the function sel_trees_have_common_keys. RETURN TRUE if there are such range trees in this imerge FALSE otherwise */ bool SEL_IMERGE::have_common_keys(RANGE_OPT_PARAM *param, SEL_TREE *tree) { for (SEL_TREE** or_tree= trees, **bound= trees_next; or_tree != bound; or_tree++) { key_map common_keys; if (sel_trees_have_common_keys(*or_tree, tree, &common_keys)) return TRUE; } return FALSE; } /* Perform AND operation for this imerge and the range part of a tree SYNOPSIS and_sel_tree() param Context info for the operation tree SEL_TREE for the second operand of the operation new_imerge OUT imerge for the result of the operation DESCRIPTION This function performs AND operation for this imerge m and the range part of the SEL_TREE tree rt. In other words the function pushes rt into this imerge. The resulting imerge is returned in the parameter new_imerge. If this imerge m represent the formula RT_1 OR ... OR RT_k then the resulting imerge of the function represents the formula (RT_1 AND RT) OR ... OR (RT_k AND RT) The function calls the function and_range_trees to construct the range tree representing (RT_i AND RT). NOTE The function may return an empty imerge without any range trees. This happens when each call of and_range_trees returns an impossible range tree (SEL_TREE::IMPOSSIBLE). Example: (key1 < 2 AND key2 > 10) AND (key1 > 4 OR key2 < 6). RETURN 0 if the operation is a success -1 otherwise: there is not enough memory to perform the operation */ int SEL_IMERGE::and_sel_tree(RANGE_OPT_PARAM *param, SEL_TREE *tree, SEL_IMERGE *new_imerge) { for (SEL_TREE** or_tree= trees; or_tree != trees_next; or_tree++) { SEL_TREE *res_or_tree= 0; if (!(res_or_tree= new SEL_TREE())) return (-1); if (!and_range_trees(param, *or_tree, tree, res_or_tree)) { if (new_imerge->or_sel_tree(param, res_or_tree)) return (-1); } } return 0; } /* Perform OR operation on this imerge and the range part of a tree SYNOPSIS or_sel_tree_with_checks() param Context info for the operation n_trees Number of trees in this imerge to check for oring tree SEL_TREE whose range part is to be ored is_first_check_pass <=> the first call of the function for this imerge is_last_check_pass OUT <=> no more calls of the function for this imerge DESCRIPTION The function performs OR operation on this imerge m and the range part of the SEL_TREE tree rt. It always replaces this imerge with the result of the operation. The operation can be performed in two different modes: with is_first_check_pass==TRUE and is_first_check_pass==FALSE, transforming this imerge differently. Given this imerge represents the formula RT_1 OR ... OR RT_k: 1. In the first mode, when is_first_check_pass==TRUE : 1.1. If rt must be ored(see the function sel_trees_must_be_ored) with some rt_j (there may be only one such range tree in the imerge) then the function produces an imerge representing the formula RT_1 OR ... OR (RT_j OR RT) OR ... OR RT_k, where the tree for (RT_j OR RT) is built by oring the pairs of SEL_ARG trees for the corresponding indexes 1.2. Otherwise the function produces the imerge representing the formula: RT_1 OR ... OR RT_k OR RT. 2. In the second mode, when is_first_check_pass==FALSE : 2.1. For each rt_j in the imerge that can be ored (see the function sel_trees_can_be_ored) with rt the function replaces rt_j for a range tree such that for each index for which ranges are defined in both in rt_j and rt the tree contains the result of oring of these ranges. 2.2. In other cases the function does not produce any imerge. When is_first_check==TRUE the function returns FALSE in the parameter is_last_check_pass if there is no rt_j such that rt_j can be ored with rt, but, at the same time, it's not true that rt_j must be ored with rt. When is_first_check==FALSE the function always returns FALSE in the parameter is_last_check_pass. RETURN 1 The result of oring of rt_j and rt that must be ored returns the the range tree with type==SEL_TREE::ALWAYS (in this case the imerge m should be discarded) -1 The function runs out of memory 0 in all other cases */ int SEL_IMERGE::or_sel_tree_with_checks(RANGE_OPT_PARAM *param, uint n_trees, SEL_TREE *tree, bool is_first_check_pass, bool *is_last_check_pass) { bool was_ored= FALSE; *is_last_check_pass= is_first_check_pass; SEL_TREE** or_tree = trees; for (uint i= 0; i < n_trees; i++, or_tree++) { SEL_TREE *result= 0; key_map result_keys; key_map ored_keys; if (sel_trees_can_be_ored(param, *or_tree, tree, &ored_keys)) { bool must_be_ored= sel_trees_must_be_ored(param, *or_tree, tree, ored_keys); if (must_be_ored || !is_first_check_pass) { result_keys.clear_all(); result= *or_tree; for (uint key_no= 0; key_no < param->keys; key_no++) { if (!ored_keys.is_set(key_no)) { result->keys[key_no]= 0; continue; } SEL_ARG *key1= (*or_tree)->keys[key_no]; SEL_ARG *key2= tree->keys[key_no]; key2->incr_refs(); if ((result->keys[key_no]= key_or(param, key1, key2))) { result_keys.set_bit(key_no); #ifdef EXTRA_DEBUG if (param->alloced_sel_args < SEL_ARG::MAX_SEL_ARGS) { key1= result->keys[key_no]; (key1)->test_use_count(key1); } #endif } } } else if(is_first_check_pass) *is_last_check_pass= FALSE; } if (result) { result->keys_map= result_keys; if (result_keys.is_clear_all()) result->type= SEL_TREE::ALWAYS; if ((result->type == SEL_TREE::MAYBE) || (result->type == SEL_TREE::ALWAYS)) return 1; /* SEL_TREE::IMPOSSIBLE is impossible here */ *or_tree= result; was_ored= TRUE; } } if (was_ored) return 0; if (is_first_check_pass && !*is_last_check_pass && !(tree= new SEL_TREE(tree, FALSE, param))) return (-1); return or_sel_tree(param, tree); } /* Perform OR operation on this imerge and and another imerge SYNOPSIS or_sel_imerge_with_checks() param Context info for the operation n_trees Number of trees in this imerge to check for oring imerge The second operand of the operation is_first_check_pass <=> the first call of the function for this imerge is_last_check_pass OUT <=> no more calls of the function for this imerge DESCRIPTION For each range tree rt from 'imerge' the function calls the method SEL_IMERGE::or_sel_tree_with_checks that performs OR operation on this SEL_IMERGE object m and the tree rt. The mode of the operation is specified by the parameter is_first_check_pass. Each call of SEL_IMERGE::or_sel_tree_with_checks transforms this SEL_IMERGE object m. The function returns FALSE in the prameter is_last_check_pass if at least one of the calls of SEL_IMERGE::or_sel_tree_with_checks returns FALSE as the value of its last parameter. RETURN 1 One of the calls of SEL_IMERGE::or_sel_tree_with_checks returns 1. (in this case the imerge m should be discarded) -1 The function runs out of memory 0 in all other cases */ int SEL_IMERGE::or_sel_imerge_with_checks(RANGE_OPT_PARAM *param, uint n_trees, SEL_IMERGE* imerge, bool is_first_check_pass, bool *is_last_check_pass) { *is_last_check_pass= TRUE; SEL_TREE** tree= imerge->trees; SEL_TREE** tree_end= imerge->trees_next; for ( ; tree < tree_end; tree++) { uint rc; bool is_last= TRUE; rc= or_sel_tree_with_checks(param, n_trees, *tree, is_first_check_pass, &is_last); if (!is_last) *is_last_check_pass= FALSE; if (rc) return rc; } return 0; } /* Copy constructor for SEL_TREE objects SYNOPSIS SEL_TREE arg The source tree for the constructor without_merges <=> only the range part of the tree arg is copied param Context info for the operation DESCRIPTION The constructor creates a full copy of the SEL_TREE arg if the prameter without_merges==FALSE. Otherwise a tree is created that contains the copy only of the range part of the tree arg. */ SEL_TREE::SEL_TREE(SEL_TREE *arg, bool without_merges, RANGE_OPT_PARAM *param): Sql_alloc() { keys_map= arg->keys_map; type= arg->type; for (uint idx= 0; idx < param->keys; idx++) { if ((keys[idx]= arg->keys[idx])) keys[idx]->incr_refs(); } if (without_merges) return; List_iterator it(arg->merges); for (SEL_IMERGE *el= it++; el; el= it++) { SEL_IMERGE *merge= new SEL_IMERGE(el, 0, param); if (!merge || merge->trees == merge->trees_next) { merges.empty(); return; } merges.push_back (merge); } } /* Copy constructor for SEL_IMERGE objects SYNOPSIS SEL_IMERGE arg The source imerge for the constructor cnt How many trees from arg are to be copied param Context info for the operation DESCRIPTION The cnt==0 then the constructor creates a full copy of the imerge arg. Otherwise only the first cnt trees of the imerge are copied. */ SEL_IMERGE::SEL_IMERGE(SEL_IMERGE *arg, uint cnt, RANGE_OPT_PARAM *param) : Sql_alloc() { uint elements= (arg->trees_end - arg->trees); if (elements > PREALLOCED_TREES) { uint size= elements * sizeof (SEL_TREE **); if (!(trees= (SEL_TREE **)alloc_root(param->mem_root, size))) goto mem_err; } else trees= &trees_prealloced[0]; trees_next= trees + (cnt ? cnt : arg->trees_next-arg->trees); trees_end= trees + elements; for (SEL_TREE **tree = trees, **arg_tree= arg->trees; tree < trees_next; tree++, arg_tree++) { if (!(*tree= new SEL_TREE(*arg_tree, FALSE, param))) goto mem_err; } return; mem_err: trees= &trees_prealloced[0]; trees_next= trees; trees_end= trees; } /* Perform AND operation on two imerge lists SYNOPSIS imerge_list_and_list() param Context info for the operation im1 The first imerge list for the operation im2 The second imerge list for the operation DESCRIPTION The function just appends the imerge list im2 to the imerge list im1 RETURN VALUE none */ inline void imerge_list_and_list(List *im1, List *im2) { im1->concat(im2); } /* Perform OR operation on two imerge lists SYNOPSIS imerge_list_or_list() param Context info for the operation im1 The first imerge list for the operation im2 The second imerge list for the operation DESCRIPTION Assuming that the first imerge list represents the formula F1= M1_1 AND ... AND M1_k1 while the second imerge list represents the formula F2= M2_1 AND ... AND M2_k2, where M1_i= RT1_i_1 OR ... OR RT1_i_l1i (i in [1..k1]) and M2_i = RT2_i_1 OR ... OR RT2_i_l2i (i in [1..k2]), the function builds a list of imerges for some formula that can be inferred from the formula (F1 OR F2). More exactly the function builds imerges for the formula (M1_1 OR M2_1). Note that (F1 OR F2) = (M1_1 AND ... AND M1_k1) OR (M2_1 AND ... AND M2_k2) = AND (M1_i OR M2_j) (i in [1..k1], j in [1..k2]) => M1_1 OR M2_1. So (M1_1 OR M2_1) is indeed an inference formula for (F1 OR F2). To build imerges for the formula (M1_1 OR M2_1) the function invokes, possibly twice, the method SEL_IMERGE::or_sel_imerge_with_checks for the imerge m1_1. At its first invocation the method SEL_IMERGE::or_sel_imerge_with_checks performs OR operation on the imerge m1_1 and the range tree rt2_1_1 by calling SEL_IMERGE::or_sel_tree_with_checks with is_first_pass_check==TRUE. The resulting imerge of the operation is ored with the next range tree of the imerge m2_1. This oring continues until the last range tree from m2_1 has been ored. At its second invocation the method SEL_IMERGE::or_sel_imerge_with_checks performs the same sequence of OR operations, but now calling SEL_IMERGE::or_sel_tree_with_checks with is_first_pass_check==FALSE. The imerges that the operation produces replace those in the list im1 RETURN 0 if the operation is a success -1 if the function has run out of memory */ int imerge_list_or_list(RANGE_OPT_PARAM *param, List *im1, List *im2) { uint rc; bool is_last_check_pass= FALSE; SEL_IMERGE *imerge= im1->head(); uint elems= imerge->trees_next-imerge->trees; im1->empty(); im1->push_back(imerge); rc= imerge->or_sel_imerge_with_checks(param, elems, im2->head(), TRUE, &is_last_check_pass); if (rc) { if (rc == 1) { im1->empty(); rc= 0; } return rc; } if (!is_last_check_pass) { SEL_IMERGE* new_imerge= new SEL_IMERGE(imerge, elems, param); if (new_imerge) { is_last_check_pass= TRUE; rc= new_imerge->or_sel_imerge_with_checks(param, elems, im2->head(), FALSE, &is_last_check_pass); if (!rc) im1->push_back(new_imerge); } } return rc; } /* Perform OR operation for each imerge from a list and the range part of a tree SYNOPSIS imerge_list_or_tree() param Context info for the operation merges The list of imerges to be ored with the range part of tree tree SEL_TREE whose range part is to be ored with the imerges DESCRIPTION For each imerge mi from the list 'merges' the function performes OR operation with mi and the range part of 'tree' rt, producing one or two imerges. Given the merge mi represent the formula RTi_1 OR ... OR RTi_k, the function forms the merges by the following rules: 1. If rt cannot be ored with any of the trees rti the function just produces an imerge that represents the formula RTi_1 OR ... RTi_k OR RT. 2. If there exist a tree rtj that must be ored with rt the function produces an imerge the represents the formula RTi_1 OR ... OR (RTi_j OR RT) OR ... OR RTi_k, where the range tree for (RTi_j OR RT) is constructed by oring the SEL_ARG trees that must be ored. 3. For each rti_j that can be ored with rt the function produces the new tree rti_j' and substitutes rti_j for this new range tree. In any case the function removes mi from the list and then adds all produced imerges. To build imerges by rules 1-3 the function calls the method SEL_IMERGE::or_sel_tree_with_checks, possibly twice. With the first call it passes TRUE for the third parameter of the function. At this first call imerges by rules 1-2 are built. If the call returns FALSE as the return value of its fourth parameter then the function are called for the second time. At this call the imerge of rule 3 is produced. If a call of SEL_IMERGE::or_sel_tree_with_checks returns 1 then then it means that the produced tree contains an always true range tree and the whole imerge can be discarded. RETURN 1 if no imerges are produced 0 otherwise */ static int imerge_list_or_tree(RANGE_OPT_PARAM *param, List *merges, SEL_TREE *tree) { SEL_IMERGE *imerge; List additional_merges; List_iterator it(*merges); while ((imerge= it++)) { bool is_last_check_pass; int rc= 0; int rc1= 0; SEL_TREE *or_tree= new SEL_TREE (tree, FALSE, param); if (or_tree) { uint elems= imerge->trees_next-imerge->trees; rc= imerge->or_sel_tree_with_checks(param, elems, or_tree, TRUE, &is_last_check_pass); if (!is_last_check_pass) { SEL_IMERGE *new_imerge= new SEL_IMERGE(imerge, elems, param); if (new_imerge) { rc1= new_imerge->or_sel_tree_with_checks(param, elems, or_tree, FALSE, &is_last_check_pass); if (!rc1) additional_merges.push_back(new_imerge); } } } if (rc || rc1 || !or_tree) it.remove(); } merges->concat(&additional_merges); return merges->is_empty(); } /* Perform pushdown operation of the range part of a tree into given imerges SYNOPSIS imerge_list_and_tree() param Context info for the operation merges IN/OUT List of imerges to push the range part of 'tree' into tree SEL_TREE whose range part is to be pushed into imerges DESCRIPTION For each imerge from the list merges the function pushes the range part rt of 'tree' into the imerge. More exactly if the imerge mi from the list represents the formula RTi_1 OR ... OR RTi_k the function bulds a new imerge that represents the formula (RTi_1 AND RT) OR ... OR (RTi_k AND RT) and adds this imerge to the list merges. To perform this pushdown operation the function calls the method SEL_IMERGE::and_sel_tree. For any imerge mi the new imerge is not created if for each pair of trees rti_j and rt the intersection of the indexes with defined ranges is empty. If the result of the pushdown operation for the imerge mi returns an imerge with no trees then then not only nothing is added to the list merges but mi itself is removed from the list. RETURN 1 if no imerges are left in the list merges 0 otherwise */ static int imerge_list_and_tree(RANGE_OPT_PARAM *param, List *merges, SEL_TREE *tree) { SEL_IMERGE *imerge; SEL_IMERGE *new_imerge= NULL; List new_merges; List_iterator it(*merges); while ((imerge= it++)) { if (!new_imerge) new_imerge= new SEL_IMERGE(); if (imerge->have_common_keys(param, tree) && new_imerge && !imerge->and_sel_tree(param, tree, new_imerge)) { if (new_imerge->trees == new_imerge->trees_next) it.remove(); else { new_merges.push_back(new_imerge); new_imerge= NULL; } } } imerge_list_and_list(&new_merges, merges); *merges= new_merges; return merges->is_empty(); } /*************************************************************************** ** Basic functions for SQL_SELECT and QUICK_RANGE_SELECT ***************************************************************************/ /* make a select from mysql info Error is set as following: 0 = ok 1 = Got some error (out of memory?) */ SQL_SELECT *make_select(TABLE *head, table_map const_tables, table_map read_tables, COND *conds, bool allow_null_cond, int *error) { SQL_SELECT *select; DBUG_ENTER("make_select"); *error=0; if (!conds && !allow_null_cond) DBUG_RETURN(0); if (!(select= new SQL_SELECT)) { *error= 1; // out of memory DBUG_RETURN(0); /* purecov: inspected */ } select->read_tables=read_tables; select->const_tables=const_tables; select->head=head; select->cond= conds; if (head->sort.io_cache) { select->file= *head->sort.io_cache; select->records=(ha_rows) (select->file.end_of_file/ head->file->ref_length); my_free(head->sort.io_cache); head->sort.io_cache=0; } DBUG_RETURN(select); } SQL_SELECT::SQL_SELECT() :quick(0),cond(0),pre_idx_push_select_cond(NULL),free_cond(0) { quick_keys.clear_all(); needed_reg.clear_all(); my_b_clear(&file); } void SQL_SELECT::cleanup() { delete quick; quick= 0; if (free_cond) { free_cond=0; delete cond; cond= 0; } close_cached_file(&file); } SQL_SELECT::~SQL_SELECT() { cleanup(); } #undef index // Fix for Unixware 7 QUICK_SELECT_I::QUICK_SELECT_I() :max_used_key_length(0), used_key_parts(0) {} QUICK_RANGE_SELECT::QUICK_RANGE_SELECT(THD *thd, TABLE *table, uint key_nr, bool no_alloc, MEM_ROOT *parent_alloc, bool *create_error) :doing_key_read(0),free_file(0),cur_range(NULL),last_range(0),dont_free(0) { my_bitmap_map *bitmap; DBUG_ENTER("QUICK_RANGE_SELECT::QUICK_RANGE_SELECT"); in_ror_merged_scan= 0; index= key_nr; head= table; key_part_info= head->key_info[index].key_part; my_init_dynamic_array(&ranges, sizeof(QUICK_RANGE*), 16, 16); /* 'thd' is not accessible in QUICK_RANGE_SELECT::reset(). */ mrr_buf_size= thd->variables.mrr_buff_size; mrr_buf_desc= NULL; if (!no_alloc && !parent_alloc) { // Allocates everything through the internal memroot init_sql_alloc(&alloc, thd->variables.range_alloc_block_size, 0); thd->mem_root= &alloc; } else bzero((char*) &alloc,sizeof(alloc)); file= head->file; record= head->record[0]; save_read_set= head->read_set; save_write_set= head->write_set; /* Allocate a bitmap for used columns (Q: why not on MEM_ROOT?) */ if (!(bitmap= (my_bitmap_map*) my_malloc(head->s->column_bitmap_size, MYF(MY_WME)))) { column_bitmap.bitmap= 0; *create_error= 1; } else bitmap_init(&column_bitmap, bitmap, head->s->fields, FALSE); DBUG_VOID_RETURN; } void QUICK_RANGE_SELECT::need_sorted_output() { if (!(mrr_flags & HA_MRR_SORTED)) { /* Native implementation can't produce sorted output. We'll have to switch to default */ mrr_flags |= HA_MRR_USE_DEFAULT_IMPL; } mrr_flags |= HA_MRR_SORTED; } int QUICK_RANGE_SELECT::init() { DBUG_ENTER("QUICK_RANGE_SELECT::init"); if (file->inited != handler::NONE) file->ha_index_or_rnd_end(); DBUG_RETURN(FALSE); } void QUICK_RANGE_SELECT::range_end() { if (file->inited != handler::NONE) file->ha_index_or_rnd_end(); } QUICK_RANGE_SELECT::~QUICK_RANGE_SELECT() { DBUG_ENTER("QUICK_RANGE_SELECT::~QUICK_RANGE_SELECT"); if (!dont_free) { /* file is NULL for CPK scan on covering ROR-intersection */ if (file) { range_end(); if (doing_key_read) file->extra(HA_EXTRA_NO_KEYREAD); if (free_file) { DBUG_PRINT("info", ("Freeing separate handler 0x%lx (free: %d)", (long) file, free_file)); file->ha_external_lock(current_thd, F_UNLCK); file->ha_close(); delete file; } } delete_dynamic(&ranges); /* ranges are allocated in alloc */ free_root(&alloc,MYF(0)); my_free(column_bitmap.bitmap); } head->column_bitmaps_set(save_read_set, save_write_set); my_free(mrr_buf_desc); DBUG_VOID_RETURN; } /* QUICK_INDEX_SORT_SELECT works as follows: - Do index scans, accumulate rowids in the Unique object (Unique will also sort and de-duplicate rowids) - Use rowids from unique to run a disk-ordered sweep */ QUICK_INDEX_SORT_SELECT::QUICK_INDEX_SORT_SELECT(THD *thd_param, TABLE *table) :unique(NULL), pk_quick_select(NULL), thd(thd_param) { DBUG_ENTER("QUICK_INDEX_SORT_SELECT::QUICK_INDEX_SORT_SELECT"); index= MAX_KEY; head= table; bzero(&read_record, sizeof(read_record)); init_sql_alloc(&alloc, thd->variables.range_alloc_block_size, 0); DBUG_VOID_RETURN; } int QUICK_INDEX_SORT_SELECT::init() { DBUG_ENTER("QUICK_INDEX_SORT_SELECT::init"); DBUG_RETURN(0); } int QUICK_INDEX_SORT_SELECT::reset() { DBUG_ENTER("QUICK_INDEX_SORT_SELECT::reset"); DBUG_RETURN(read_keys_and_merge()); } bool QUICK_INDEX_SORT_SELECT::push_quick_back(QUICK_RANGE_SELECT *quick_sel_range) { DBUG_ENTER("QUICK_INDEX_SORT_SELECT::push_quick_back"); if (head->file->primary_key_is_clustered() && quick_sel_range->index == head->s->primary_key) { /* A quick_select over a clustered primary key is handled specifically Here we assume: - PK columns are included in any other merged index - Scan on the PK is disk-ordered. (not meeting #2 will only cause performance degradation) We could treat clustered PK as any other index, but that would be inefficient. There is no point in doing scan on CPK, remembering the rowid, then making rnd_pos() call with that rowid. */ pk_quick_select= quick_sel_range; DBUG_RETURN(0); } DBUG_RETURN(quick_selects.push_back(quick_sel_range)); } QUICK_INDEX_SORT_SELECT::~QUICK_INDEX_SORT_SELECT() { List_iterator_fast quick_it(quick_selects); QUICK_RANGE_SELECT* quick; DBUG_ENTER("QUICK_INDEX_SORT_SELECT::~QUICK_INDEX_SORT_SELECT"); delete unique; quick_it.rewind(); while ((quick= quick_it++)) quick->file= NULL; quick_selects.delete_elements(); delete pk_quick_select; /* It's ok to call the next two even if they are already deinitialized */ end_read_record(&read_record); free_io_cache(head); free_root(&alloc,MYF(0)); DBUG_VOID_RETURN; } QUICK_ROR_INTERSECT_SELECT::QUICK_ROR_INTERSECT_SELECT(THD *thd_param, TABLE *table, bool retrieve_full_rows, MEM_ROOT *parent_alloc) : cpk_quick(NULL), thd(thd_param), need_to_fetch_row(retrieve_full_rows), scans_inited(FALSE) { index= MAX_KEY; head= table; record= head->record[0]; if (!parent_alloc) init_sql_alloc(&alloc, thd->variables.range_alloc_block_size, 0); else bzero(&alloc, sizeof(MEM_ROOT)); last_rowid= (uchar*) alloc_root(parent_alloc? parent_alloc : &alloc, head->file->ref_length); } /* Do post-constructor initialization. SYNOPSIS QUICK_ROR_INTERSECT_SELECT::init() RETURN 0 OK other Error code */ int QUICK_ROR_INTERSECT_SELECT::init() { DBUG_ENTER("QUICK_ROR_INTERSECT_SELECT::init"); /* Check if last_rowid was successfully allocated in ctor */ DBUG_RETURN(!last_rowid); } /* Initialize this quick select to be a ROR-merged scan. SYNOPSIS QUICK_RANGE_SELECT::init_ror_merged_scan() reuse_handler If TRUE, use head->file, otherwise create a separate handler object NOTES This function creates and prepares for subsequent use a separate handler object if it can't reuse head->file. The reason for this is that during ROR-merge several key scans are performed simultaneously, and a single handler is only capable of preserving context of a single key scan. In ROR-merge the quick select doing merge does full records retrieval, merged quick selects read only keys. RETURN 0 ROR child scan initialized, ok to use. 1 error */ int QUICK_RANGE_SELECT::init_ror_merged_scan(bool reuse_handler) { handler *save_file= file, *org_file; my_bool org_key_read; THD *thd; DBUG_ENTER("QUICK_RANGE_SELECT::init_ror_merged_scan"); in_ror_merged_scan= 1; if (reuse_handler) { DBUG_PRINT("info", ("Reusing handler 0x%lx", (long) file)); if (init() || reset()) { DBUG_RETURN(1); } head->column_bitmaps_set(&column_bitmap, &column_bitmap); goto end; } /* Create a separate handler object for this quick select */ if (free_file) { /* already have own 'handler' object. */ DBUG_RETURN(0); } thd= head->in_use; if (!(file= head->file->clone(head->s->normalized_path.str, thd->mem_root))) { /* Manually set the error flag. Note: there seems to be quite a few places where a failure could cause the server to "hang" the client by sending no response to a query. ATM those are not real errors because the storage engine calls in question happen to never fail with the existing storage engines. */ my_error(ER_OUT_OF_RESOURCES, MYF(0)); /* purecov: inspected */ /* Caller will free the memory */ goto failure; /* purecov: inspected */ } head->column_bitmaps_set(&column_bitmap, &column_bitmap); if (file->ha_external_lock(thd, F_RDLCK)) goto failure; if (init() || reset()) { file->ha_external_lock(thd, F_UNLCK); file->ha_close(); goto failure; } free_file= TRUE; last_rowid= file->ref; end: /* We are only going to read key fields and call position() on 'file' The following sets head->tmp_set to only use this key and then updates head->read_set and head->write_set to use this bitmap. The now bitmap is stored in 'column_bitmap' which is used in ::get_next() */ org_file= head->file; org_key_read= head->key_read; head->file= file; head->key_read= 0; if (!head->no_keyread) { doing_key_read= 1; head->mark_columns_used_by_index(index); } head->prepare_for_position(); head->file= org_file; head->key_read= org_key_read; bitmap_copy(&column_bitmap, head->read_set); head->column_bitmaps_set(&column_bitmap, &column_bitmap); DBUG_RETURN(0); failure: head->column_bitmaps_set(save_read_set, save_write_set); delete file; file= save_file; DBUG_RETURN(1); } /* Initialize this quick select to be a part of a ROR-merged scan. SYNOPSIS QUICK_ROR_INTERSECT_SELECT::init_ror_merged_scan() reuse_handler If TRUE, use head->file, otherwise create separate handler object. RETURN 0 OK other error code */ int QUICK_ROR_INTERSECT_SELECT::init_ror_merged_scan(bool reuse_handler) { List_iterator_fast quick_it(quick_selects); QUICK_SELECT_WITH_RECORD *cur; QUICK_RANGE_SELECT *quick; DBUG_ENTER("QUICK_ROR_INTERSECT_SELECT::init_ror_merged_scan"); /* Initialize all merged "children" quick selects */ DBUG_ASSERT(!need_to_fetch_row || reuse_handler); if (!need_to_fetch_row && reuse_handler) { cur= quick_it++; quick= cur->quick; /* There is no use of this->file. Use it for the first of merged range selects. */ if (quick->init_ror_merged_scan(TRUE)) DBUG_RETURN(1); quick->file->extra(HA_EXTRA_KEYREAD_PRESERVE_FIELDS); } while ((cur= quick_it++)) { quick= cur->quick; if (quick->init_ror_merged_scan(FALSE)) DBUG_RETURN(1); quick->file->extra(HA_EXTRA_KEYREAD_PRESERVE_FIELDS); /* All merged scans share the same record buffer in intersection. */ quick->record= head->record[0]; } if (need_to_fetch_row && head->file->ha_rnd_init_with_error(1)) { DBUG_PRINT("error", ("ROR index_merge rnd_init call failed")); DBUG_RETURN(1); } DBUG_RETURN(0); } /* Initialize quick select for row retrieval. SYNOPSIS reset() RETURN 0 OK other Error code */ int QUICK_ROR_INTERSECT_SELECT::reset() { DBUG_ENTER("QUICK_ROR_INTERSECT_SELECT::reset"); if (!scans_inited && init_ror_merged_scan(TRUE)) DBUG_RETURN(1); scans_inited= TRUE; List_iterator_fast it(quick_selects); QUICK_SELECT_WITH_RECORD *qr; while ((qr= it++)) qr->quick->reset(); DBUG_RETURN(0); } /* Add a merged quick select to this ROR-intersection quick select. SYNOPSIS QUICK_ROR_INTERSECT_SELECT::push_quick_back() alloc Mem root to create auxiliary structures on quick Quick select to be added. The quick select must return rows in rowid order. NOTES This call can only be made before init() is called. RETURN FALSE OK TRUE Out of memory. */ bool QUICK_ROR_INTERSECT_SELECT::push_quick_back(MEM_ROOT *alloc, QUICK_RANGE_SELECT *quick) { QUICK_SELECT_WITH_RECORD *qr; if (!(qr= new QUICK_SELECT_WITH_RECORD) || !(qr->key_tuple= (uchar*)alloc_root(alloc, quick->max_used_key_length))) return TRUE; qr->quick= quick; return quick_selects.push_back(qr); } QUICK_ROR_INTERSECT_SELECT::~QUICK_ROR_INTERSECT_SELECT() { DBUG_ENTER("QUICK_ROR_INTERSECT_SELECT::~QUICK_ROR_INTERSECT_SELECT"); quick_selects.delete_elements(); delete cpk_quick; free_root(&alloc,MYF(0)); if (need_to_fetch_row && head->file->inited != handler::NONE) head->file->ha_rnd_end(); DBUG_VOID_RETURN; } QUICK_ROR_UNION_SELECT::QUICK_ROR_UNION_SELECT(THD *thd_param, TABLE *table) : thd(thd_param), scans_inited(FALSE) { index= MAX_KEY; head= table; rowid_length= table->file->ref_length; record= head->record[0]; init_sql_alloc(&alloc, thd->variables.range_alloc_block_size, 0); thd_param->mem_root= &alloc; } /* Comparison function to be used QUICK_ROR_UNION_SELECT::queue priority queue. SYNPOSIS QUICK_ROR_UNION_SELECT_queue_cmp() arg Pointer to QUICK_ROR_UNION_SELECT val1 First merged select val2 Second merged select */ C_MODE_START static int QUICK_ROR_UNION_SELECT_queue_cmp(void *arg, uchar *val1, uchar *val2) { QUICK_ROR_UNION_SELECT *self= (QUICK_ROR_UNION_SELECT*)arg; return self->head->file->cmp_ref(((QUICK_SELECT_I*)val1)->last_rowid, ((QUICK_SELECT_I*)val2)->last_rowid); } C_MODE_END /* Do post-constructor initialization. SYNOPSIS QUICK_ROR_UNION_SELECT::init() RETURN 0 OK other Error code */ int QUICK_ROR_UNION_SELECT::init() { DBUG_ENTER("QUICK_ROR_UNION_SELECT::init"); if (init_queue(&queue, quick_selects.elements, 0, FALSE , QUICK_ROR_UNION_SELECT_queue_cmp, (void*) this, 0, 0)) { bzero(&queue, sizeof(QUEUE)); DBUG_RETURN(1); } if (!(cur_rowid= (uchar*) alloc_root(&alloc, 2*head->file->ref_length))) DBUG_RETURN(1); prev_rowid= cur_rowid + head->file->ref_length; DBUG_RETURN(0); } /* Initialize quick select for row retrieval. SYNOPSIS reset() RETURN 0 OK other Error code */ int QUICK_ROR_UNION_SELECT::reset() { QUICK_SELECT_I *quick; int error; DBUG_ENTER("QUICK_ROR_UNION_SELECT::reset"); have_prev_rowid= FALSE; if (!scans_inited) { List_iterator_fast it(quick_selects); while ((quick= it++)) { if (quick->init_ror_merged_scan(FALSE)) DBUG_RETURN(1); } scans_inited= TRUE; } queue_remove_all(&queue); /* Initialize scans for merged quick selects and put all merged quick selects into the queue. */ List_iterator_fast it(quick_selects); while ((quick= it++)) { if (quick->reset()) DBUG_RETURN(1); if ((error= quick->get_next())) { if (error == HA_ERR_END_OF_FILE) continue; DBUG_RETURN(error); } quick->save_last_pos(); queue_insert(&queue, (uchar*)quick); } if (head->file->ha_rnd_init_with_error(1)) { DBUG_PRINT("error", ("ROR index_merge rnd_init call failed")); DBUG_RETURN(1); } DBUG_RETURN(0); } bool QUICK_ROR_UNION_SELECT::push_quick_back(QUICK_SELECT_I *quick_sel_range) { return quick_selects.push_back(quick_sel_range); } QUICK_ROR_UNION_SELECT::~QUICK_ROR_UNION_SELECT() { DBUG_ENTER("QUICK_ROR_UNION_SELECT::~QUICK_ROR_UNION_SELECT"); delete_queue(&queue); quick_selects.delete_elements(); if (head->file->inited != handler::NONE) head->file->ha_rnd_end(); free_root(&alloc,MYF(0)); DBUG_VOID_RETURN; } QUICK_RANGE::QUICK_RANGE() :min_key(0),max_key(0),min_length(0),max_length(0), flag(NO_MIN_RANGE | NO_MAX_RANGE), min_keypart_map(0), max_keypart_map(0) {} SEL_ARG::SEL_ARG(SEL_ARG &arg) :Sql_alloc() { type=arg.type; min_flag=arg.min_flag; max_flag=arg.max_flag; maybe_flag=arg.maybe_flag; maybe_null=arg.maybe_null; part=arg.part; field=arg.field; min_value=arg.min_value; max_value=arg.max_value; next_key_part=arg.next_key_part; max_part_no= arg.max_part_no; use_count=1; elements=1; } inline void SEL_ARG::make_root() { left=right= &null_element; color=BLACK; next=prev=0; use_count=0; elements=1; } SEL_ARG::SEL_ARG(Field *f,const uchar *min_value_arg, const uchar *max_value_arg) :min_flag(0), max_flag(0), maybe_flag(0), maybe_null(f->real_maybe_null()), elements(1), use_count(1), field(f), min_value((uchar*) min_value_arg), max_value((uchar*) max_value_arg), next(0),prev(0), next_key_part(0), color(BLACK), type(KEY_RANGE) { left=right= &null_element; max_part_no= 1; } SEL_ARG::SEL_ARG(Field *field_,uint8 part_, uchar *min_value_, uchar *max_value_, uint8 min_flag_,uint8 max_flag_,uint8 maybe_flag_) :min_flag(min_flag_),max_flag(max_flag_),maybe_flag(maybe_flag_), part(part_),maybe_null(field_->real_maybe_null()), elements(1),use_count(1), field(field_), min_value(min_value_), max_value(max_value_), next(0),prev(0),next_key_part(0),color(BLACK),type(KEY_RANGE) { max_part_no= part+1; left=right= &null_element; } SEL_ARG *SEL_ARG::clone(RANGE_OPT_PARAM *param, SEL_ARG *new_parent, SEL_ARG **next_arg) { SEL_ARG *tmp; /* Bail out if we have already generated too many SEL_ARGs */ if (++param->alloced_sel_args > MAX_SEL_ARGS) return 0; if (type != KEY_RANGE) { if (!(tmp= new (param->mem_root) SEL_ARG(type))) return 0; // out of memory tmp->prev= *next_arg; // Link into next/prev chain (*next_arg)->next=tmp; (*next_arg)= tmp; tmp->part= this->part; } else { if (!(tmp= new (param->mem_root) SEL_ARG(field,part, min_value,max_value, min_flag, max_flag, maybe_flag))) return 0; // OOM tmp->parent=new_parent; tmp->next_key_part=next_key_part; if (left != &null_element) if (!(tmp->left=left->clone(param, tmp, next_arg))) return 0; // OOM tmp->prev= *next_arg; // Link into next/prev chain (*next_arg)->next=tmp; (*next_arg)= tmp; if (right != &null_element) if (!(tmp->right= right->clone(param, tmp, next_arg))) return 0; // OOM } increment_use_count(1); tmp->color= color; tmp->elements= this->elements; tmp->max_part_no= max_part_no; return tmp; } SEL_ARG *SEL_ARG::first() { SEL_ARG *next_arg=this; if (!next_arg->left) return 0; // MAYBE_KEY while (next_arg->left != &null_element) next_arg=next_arg->left; return next_arg; } SEL_ARG *SEL_ARG::last() { SEL_ARG *next_arg=this; if (!next_arg->right) return 0; // MAYBE_KEY while (next_arg->right != &null_element) next_arg=next_arg->right; return next_arg; } /* Check if a compare is ok, when one takes ranges in account Returns -2 or 2 if the ranges where 'joined' like < 2 and >= 2 */ static int sel_cmp(Field *field, uchar *a, uchar *b, uint8 a_flag, uint8 b_flag) { int cmp; /* First check if there was a compare to a min or max element */ if (a_flag & (NO_MIN_RANGE | NO_MAX_RANGE)) { if ((a_flag & (NO_MIN_RANGE | NO_MAX_RANGE)) == (b_flag & (NO_MIN_RANGE | NO_MAX_RANGE))) return 0; return (a_flag & NO_MIN_RANGE) ? -1 : 1; } if (b_flag & (NO_MIN_RANGE | NO_MAX_RANGE)) return (b_flag & NO_MIN_RANGE) ? 1 : -1; if (field->real_maybe_null()) // If null is part of key { if (*a != *b) { return *a ? -1 : 1; } if (*a) goto end; // NULL where equal a++; b++; // Skip NULL marker } cmp=field->key_cmp(a , b); if (cmp) return cmp < 0 ? -1 : 1; // The values differed // Check if the compared equal arguments was defined with open/closed range end: if (a_flag & (NEAR_MIN | NEAR_MAX)) { if ((a_flag & (NEAR_MIN | NEAR_MAX)) == (b_flag & (NEAR_MIN | NEAR_MAX))) return 0; if (!(b_flag & (NEAR_MIN | NEAR_MAX))) return (a_flag & NEAR_MIN) ? 2 : -2; return (a_flag & NEAR_MIN) ? 1 : -1; } if (b_flag & (NEAR_MIN | NEAR_MAX)) return (b_flag & NEAR_MIN) ? -2 : 2; return 0; // The elements where equal } SEL_ARG *SEL_ARG::clone_tree(RANGE_OPT_PARAM *param) { SEL_ARG tmp_link,*next_arg,*root; next_arg= &tmp_link; if (!(root= clone(param, (SEL_ARG *) 0, &next_arg))) return 0; next_arg->next=0; // Fix last link tmp_link.next->prev=0; // Fix first link if (root) // If not OOM root->use_count= 0; return root; } /* Table rows retrieval plan. Range optimizer creates QUICK_SELECT_I-derived objects from table read plans. */ class TABLE_READ_PLAN { public: /* Plan read cost, with or without cost of full row retrieval, depending on plan creation parameters. */ double read_cost; ha_rows records; /* estimate of #rows to be examined */ /* If TRUE, the scan returns rows in rowid order. This is used only for scans that can be both ROR and non-ROR. */ bool is_ror; /* Create quick select for this plan. SYNOPSIS make_quick() param Parameter from test_quick_select retrieve_full_rows If TRUE, created quick select will do full record retrieval. parent_alloc Memory pool to use, if any. NOTES retrieve_full_rows is ignored by some implementations. RETURN created quick select NULL on any error. */ virtual QUICK_SELECT_I *make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc=NULL) = 0; /* Table read plans are allocated on MEM_ROOT and are never deleted */ static void *operator new(size_t size, MEM_ROOT *mem_root) { return (void*) alloc_root(mem_root, (uint) size); } static void operator delete(void *ptr,size_t size) { TRASH(ptr, size); } static void operator delete(void *ptr, MEM_ROOT *mem_root) { /* Never called */ } virtual ~TABLE_READ_PLAN() {} /* Remove gcc warning */ }; class TRP_ROR_INTERSECT; class TRP_ROR_UNION; class TRP_INDEX_MERGE; /* Plan for a QUICK_RANGE_SELECT scan. TRP_RANGE::make_quick ignores retrieve_full_rows parameter because QUICK_RANGE_SELECT doesn't distinguish between 'index only' scans and full record retrieval scans. */ class TRP_RANGE : public TABLE_READ_PLAN { public: SEL_ARG *key; /* set of intervals to be used in "range" method retrieval */ uint key_idx; /* key number in PARAM::key */ uint mrr_flags; uint mrr_buf_size; TRP_RANGE(SEL_ARG *key_arg, uint idx_arg, uint mrr_flags_arg) : key(key_arg), key_idx(idx_arg), mrr_flags(mrr_flags_arg) {} virtual ~TRP_RANGE() {} /* Remove gcc warning */ QUICK_SELECT_I *make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc) { DBUG_ENTER("TRP_RANGE::make_quick"); QUICK_RANGE_SELECT *quick; if ((quick= get_quick_select(param, key_idx, key, mrr_flags, mrr_buf_size, parent_alloc))) { quick->records= records; quick->read_time= read_cost; } DBUG_RETURN(quick); } }; /* Plan for QUICK_ROR_INTERSECT_SELECT scan. */ class TRP_ROR_INTERSECT : public TABLE_READ_PLAN { public: TRP_ROR_INTERSECT() {} /* Remove gcc warning */ virtual ~TRP_ROR_INTERSECT() {} /* Remove gcc warning */ QUICK_SELECT_I *make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc); /* Array of pointers to ROR range scans used in this intersection */ struct st_ror_scan_info **first_scan; struct st_ror_scan_info **last_scan; /* End of the above array */ struct st_ror_scan_info *cpk_scan; /* Clustered PK scan, if there is one */ bool is_covering; /* TRUE if no row retrieval phase is necessary */ double index_scan_costs; /* SUM(cost(index_scan)) */ }; /* Plan for QUICK_ROR_UNION_SELECT scan. QUICK_ROR_UNION_SELECT always retrieves full rows, so retrieve_full_rows is ignored by make_quick. */ class TRP_ROR_UNION : public TABLE_READ_PLAN { public: TRP_ROR_UNION() {} /* Remove gcc warning */ virtual ~TRP_ROR_UNION() {} /* Remove gcc warning */ QUICK_SELECT_I *make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc); TABLE_READ_PLAN **first_ror; /* array of ptrs to plans for merged scans */ TABLE_READ_PLAN **last_ror; /* end of the above array */ }; /* Plan for QUICK_INDEX_INTERSECT_SELECT scan. QUICK_INDEX_INTERSECT_SELECT always retrieves full rows, so retrieve_full_rows is ignored by make_quick. */ class TRP_INDEX_INTERSECT : public TABLE_READ_PLAN { public: TRP_INDEX_INTERSECT() {} /* Remove gcc warning */ virtual ~TRP_INDEX_INTERSECT() {} /* Remove gcc warning */ QUICK_SELECT_I *make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc); TRP_RANGE **range_scans; /* array of ptrs to plans of intersected scans */ TRP_RANGE **range_scans_end; /* end of the array */ /* keys whose scans are to be filtered by cpk conditions */ key_map filtered_scans; }; /* Plan for QUICK_INDEX_MERGE_SELECT scan. QUICK_ROR_INTERSECT_SELECT always retrieves full rows, so retrieve_full_rows is ignored by make_quick. */ class TRP_INDEX_MERGE : public TABLE_READ_PLAN { public: TRP_INDEX_MERGE() {} /* Remove gcc warning */ virtual ~TRP_INDEX_MERGE() {} /* Remove gcc warning */ QUICK_SELECT_I *make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc); TRP_RANGE **range_scans; /* array of ptrs to plans of merged scans */ TRP_RANGE **range_scans_end; /* end of the array */ }; /* Plan for a QUICK_GROUP_MIN_MAX_SELECT scan. */ class TRP_GROUP_MIN_MAX : public TABLE_READ_PLAN { private: bool have_min, have_max, have_agg_distinct; KEY_PART_INFO *min_max_arg_part; uint group_prefix_len; uint used_key_parts; uint group_key_parts; KEY *index_info; uint index; uint key_infix_len; uchar key_infix[MAX_KEY_LENGTH]; SEL_TREE *range_tree; /* Represents all range predicates in the query. */ SEL_ARG *index_tree; /* The SEL_ARG sub-tree corresponding to index_info. */ uint param_idx; /* Index of used key in param->key. */ bool is_index_scan; /* Use index_next() instead of random read */ public: /* Number of records selected by the ranges in index_tree. */ ha_rows quick_prefix_records; public: TRP_GROUP_MIN_MAX(bool have_min_arg, bool have_max_arg, bool have_agg_distinct_arg, KEY_PART_INFO *min_max_arg_part_arg, uint group_prefix_len_arg, uint used_key_parts_arg, uint group_key_parts_arg, KEY *index_info_arg, uint index_arg, uint key_infix_len_arg, uchar *key_infix_arg, SEL_TREE *tree_arg, SEL_ARG *index_tree_arg, uint param_idx_arg, ha_rows quick_prefix_records_arg) : have_min(have_min_arg), have_max(have_max_arg), have_agg_distinct(have_agg_distinct_arg), min_max_arg_part(min_max_arg_part_arg), group_prefix_len(group_prefix_len_arg), used_key_parts(used_key_parts_arg), group_key_parts(group_key_parts_arg), index_info(index_info_arg), index(index_arg), key_infix_len(key_infix_len_arg), range_tree(tree_arg), index_tree(index_tree_arg), param_idx(param_idx_arg), is_index_scan(FALSE), quick_prefix_records(quick_prefix_records_arg) { if (key_infix_len) memcpy(this->key_infix, key_infix_arg, key_infix_len); } virtual ~TRP_GROUP_MIN_MAX() {} /* Remove gcc warning */ QUICK_SELECT_I *make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc); void use_index_scan() { is_index_scan= TRUE; } }; typedef struct st_index_scan_info { uint idx; /* # of used key in param->keys */ uint keynr; /* # of used key in table */ uint range_count; ha_rows records; /* estimate of # records this scan will return */ /* Set of intervals over key fields that will be used for row retrieval. */ SEL_ARG *sel_arg; KEY *key_info; uint used_key_parts; /* Estimate of # records filtered out by intersection with cpk */ ha_rows filtered_out; /* Bitmap of fields used in index intersection */ MY_BITMAP used_fields; /* Fields used in the query and covered by ROR scan. */ MY_BITMAP covered_fields; uint used_fields_covered; /* # of set bits in covered_fields */ int key_rec_length; /* length of key record (including rowid) */ /* Cost of reading all index records with values in sel_arg intervals set (assuming there is no need to access full table records) */ double index_read_cost; uint first_uncovered_field; /* first unused bit in covered_fields */ uint key_components; /* # of parts in the key */ } INDEX_SCAN_INFO; /* Fill param->needed_fields with bitmap of fields used in the query. SYNOPSIS fill_used_fields_bitmap() param Parameter from test_quick_select function. NOTES Clustered PK members are not put into the bitmap as they are implicitly present in all keys (and it is impossible to avoid reading them). RETURN 0 Ok 1 Out of memory. */ static int fill_used_fields_bitmap(PARAM *param) { TABLE *table= param->table; my_bitmap_map *tmp; uint pk; param->tmp_covered_fields.bitmap= 0; param->fields_bitmap_size= table->s->column_bitmap_size; if (!(tmp= (my_bitmap_map*) alloc_root(param->mem_root, param->fields_bitmap_size)) || bitmap_init(¶m->needed_fields, tmp, table->s->fields, FALSE)) return 1; bitmap_copy(¶m->needed_fields, table->read_set); bitmap_union(¶m->needed_fields, table->write_set); pk= param->table->s->primary_key; if (pk != MAX_KEY && param->table->file->primary_key_is_clustered()) { /* The table uses clustered PK and it is not internally generated */ KEY_PART_INFO *key_part= param->table->key_info[pk].key_part; KEY_PART_INFO *key_part_end= key_part + param->table->key_info[pk].key_parts; for (;key_part != key_part_end; ++key_part) bitmap_clear_bit(¶m->needed_fields, key_part->fieldnr-1); } return 0; } /* Test if a key can be used in different ranges SYNOPSIS SQL_SELECT::test_quick_select() thd Current thread keys_to_use Keys to use for range retrieval prev_tables Tables assumed to be already read when the scan is performed (but not read at the moment of this call) limit Query limit force_quick_range Prefer to use range (instead of full table scan) even if it is more expensive. NOTES Updates the following in the select parameter: needed_reg - Bits for keys with may be used if all prev regs are read quick - Parameter to use when reading records. In the table struct the following information is updated: quick_keys - Which keys can be used quick_rows - How many rows the key matches quick_condition_rows - E(# rows that will satisfy the table condition) IMPLEMENTATION quick_condition_rows value is obtained as follows: It is a minimum of E(#output rows) for all considered table access methods (range and index_merge accesses over various indexes). The obtained value is not a true E(#rows that satisfy table condition) but rather a pessimistic estimate. To obtain a true E(#...) one would need to combine estimates of various access methods, taking into account correlations between sets of rows they will return. For example, if values of tbl.key1 and tbl.key2 are independent (a right assumption if we have no information about their correlation) then the correct estimate will be: E(#rows("tbl.key1 < c1 AND tbl.key2 < c2")) = = E(#rows(tbl.key1 < c1)) / total_rows(tbl) * E(#rows(tbl.key2 < c2) which is smaller than MIN(E(#rows(tbl.key1 < c1), E(#rows(tbl.key2 < c2))) which is currently produced. TODO * Change the value returned in quick_condition_rows from a pessimistic estimate to true E(#rows that satisfy table condition). (we can re-use some of E(#rows) calcuation code from index_merge/intersection for this) * Check if this function really needs to modify keys_to_use, and change the code to pass it by reference if it doesn't. * In addition to force_quick_range other means can be (an usually are) used to make this function prefer range over full table scan. Figure out if force_quick_range is really needed. RETURN -1 if impossible select (i.e. certainly no rows will be selected) 0 if can't use quick_select 1 if found usable ranges and quick select has been successfully created. */ int SQL_SELECT::test_quick_select(THD *thd, key_map keys_to_use, table_map prev_tables, ha_rows limit, bool force_quick_range, bool ordered_output) { uint idx; double scan_time; DBUG_ENTER("SQL_SELECT::test_quick_select"); DBUG_PRINT("enter",("keys_to_use: %lu prev_tables: %lu const_tables: %lu", (ulong) keys_to_use.to_ulonglong(), (ulong) prev_tables, (ulong) const_tables)); DBUG_PRINT("info", ("records: %lu", (ulong) head->file->stats.records)); delete quick; quick=0; needed_reg.clear_all(); quick_keys.clear_all(); DBUG_ASSERT(!head->is_filled_at_execution()); if (keys_to_use.is_clear_all() || head->is_filled_at_execution()) DBUG_RETURN(0); records= head->file->stats.records; if (!records) records++; /* purecov: inspected */ scan_time= (double) records / TIME_FOR_COMPARE + 1; read_time= (double) head->file->scan_time() + scan_time + 1.1; if (head->force_index) scan_time= read_time= DBL_MAX; if (limit < records) read_time= (double) records + scan_time + 1; // Force to use index else if (read_time <= 2.0 && !force_quick_range) DBUG_RETURN(0); /* No need for quick select */ DBUG_PRINT("info",("Time to scan table: %g", read_time)); keys_to_use.intersect(head->keys_in_use_for_query); if (!keys_to_use.is_clear_all()) { uchar buff[STACK_BUFF_ALLOC]; MEM_ROOT alloc; SEL_TREE *tree= NULL; KEY_PART *key_parts; KEY *key_info; PARAM param; if (check_stack_overrun(thd, 2*STACK_MIN_SIZE + sizeof(PARAM), buff)) DBUG_RETURN(0); // Fatal error flag is set /* set up parameter that is passed to all functions */ param.thd= thd; param.baseflag= head->file->ha_table_flags(); param.prev_tables=prev_tables | const_tables; param.read_tables=read_tables; param.current_table= head->map; param.table=head; param.keys=0; param.mem_root= &alloc; param.old_root= thd->mem_root; param.needed_reg= &needed_reg; param.imerge_cost_buff_size= 0; param.using_real_indexes= TRUE; param.remove_jump_scans= TRUE; param.force_default_mrr= ordered_output; thd->no_errors=1; // Don't warn about NULL init_sql_alloc(&alloc, thd->variables.range_alloc_block_size, 0); if (!(param.key_parts= (KEY_PART*) alloc_root(&alloc, sizeof(KEY_PART)* head->s->key_parts)) || fill_used_fields_bitmap(¶m)) { thd->no_errors=0; free_root(&alloc,MYF(0)); // Return memory & allocator DBUG_RETURN(0); // Can't use range } key_parts= param.key_parts; thd->mem_root= &alloc; /* Make an array with description of all key parts of all table keys. This is used in get_mm_parts function. */ key_info= head->key_info; for (idx=0 ; idx < head->s->keys ; idx++, key_info++) { KEY_PART_INFO *key_part_info; if (!keys_to_use.is_set(idx)) continue; if (key_info->flags & HA_FULLTEXT) continue; // ToDo: ft-keys in non-ft ranges, if possible SerG param.key[param.keys]=key_parts; key_part_info= key_info->key_part; for (uint part=0 ; part < key_info->key_parts ; part++, key_parts++, key_part_info++) { key_parts->key= param.keys; key_parts->part= part; key_parts->length= key_part_info->length; key_parts->store_length= key_part_info->store_length; key_parts->field= key_part_info->field; key_parts->null_bit= key_part_info->null_bit; key_parts->image_type = (key_info->flags & HA_SPATIAL) ? Field::itMBR : Field::itRAW; /* Only HA_PART_KEY_SEG is used */ key_parts->flag= (uint8) key_part_info->key_part_flag; } param.real_keynr[param.keys++]=idx; } param.key_parts_end=key_parts; param.alloced_sel_args= 0; /* Calculate cost of full index read for the shortest covering index */ if (!head->covering_keys.is_clear_all()) { int key_for_use= find_shortest_key(head, &head->covering_keys); double key_read_time= head->file->keyread_time(key_for_use, 1, records) + (double) records / TIME_FOR_COMPARE; DBUG_PRINT("info", ("'all'+'using index' scan will be using key %d, " "read time %g", key_for_use, key_read_time)); if (key_read_time < read_time) read_time= key_read_time; } TABLE_READ_PLAN *best_trp= NULL; TRP_GROUP_MIN_MAX *group_trp; double best_read_time= read_time; if (cond) { if ((tree= get_mm_tree(¶m,cond))) { if (tree->type == SEL_TREE::IMPOSSIBLE) { records=0L; /* Return -1 from this function. */ read_time= (double) HA_POS_ERROR; goto free_mem; } /* If the tree can't be used for range scans, proceed anyway, as we can construct a group-min-max quick select */ if (tree->type != SEL_TREE::KEY && tree->type != SEL_TREE::KEY_SMALLER) tree= NULL; } } /* Try to construct a QUICK_GROUP_MIN_MAX_SELECT. Notice that it can be constructed no matter if there is a range tree. */ group_trp= get_best_group_min_max(¶m, tree, best_read_time); if (group_trp) { param.table->quick_condition_rows= min(group_trp->records, head->file->stats.records); if (group_trp->read_cost < best_read_time) { best_trp= group_trp; best_read_time= best_trp->read_cost; } } if (tree) { /* It is possible to use a range-based quick select (but it might be slower than 'all' table scan). */ TRP_RANGE *range_trp; TRP_ROR_INTERSECT *rori_trp; TRP_INDEX_INTERSECT *intersect_trp; bool can_build_covering= FALSE; remove_nonrange_trees(¶m, tree); /* Get best 'range' plan and prepare data for making other plans */ if ((range_trp= get_key_scans_params(¶m, tree, FALSE, TRUE, best_read_time))) { best_trp= range_trp; best_read_time= best_trp->read_cost; } /* Simultaneous key scans and row deletes on several handler objects are not allowed so don't use ROR-intersection for table deletes. */ if ((thd->lex->sql_command != SQLCOM_DELETE) && optimizer_flag(thd, OPTIMIZER_SWITCH_INDEX_MERGE)) { /* Get best non-covering ROR-intersection plan and prepare data for building covering ROR-intersection. */ if ((rori_trp= get_best_ror_intersect(¶m, tree, best_read_time, &can_build_covering))) { best_trp= rori_trp; best_read_time= best_trp->read_cost; /* Try constructing covering ROR-intersect only if it looks possible and worth doing. */ if (!rori_trp->is_covering && can_build_covering && (rori_trp= get_best_covering_ror_intersect(¶m, tree, best_read_time))) best_trp= rori_trp; } } /* Do not look for an index intersection plan if there is a covering index. The scan by this covering index will be always cheaper than any index intersection. */ if (param.table->covering_keys.is_clear_all() && optimizer_flag(thd, OPTIMIZER_SWITCH_INDEX_MERGE) && optimizer_flag(thd, OPTIMIZER_SWITCH_INDEX_MERGE_SORT_INTERSECT)) { if ((intersect_trp= get_best_index_intersect(¶m, tree, best_read_time))) { best_trp= intersect_trp; best_read_time= best_trp->read_cost; set_if_smaller(param.table->quick_condition_rows, intersect_trp->records); } } if (optimizer_flag(thd, OPTIMIZER_SWITCH_INDEX_MERGE)) { /* Try creating index_merge/ROR-union scan. */ SEL_IMERGE *imerge; TABLE_READ_PLAN *best_conj_trp= NULL, *new_conj_trp; LINT_INIT(new_conj_trp); /* no empty index_merge lists possible */ DBUG_PRINT("info",("No range reads possible," " trying to construct index_merge")); List_iterator_fast it(tree->merges); while ((imerge= it++)) { new_conj_trp= get_best_disjunct_quick(¶m, imerge, best_read_time); if (new_conj_trp) set_if_smaller(param.table->quick_condition_rows, new_conj_trp->records); if (new_conj_trp && (!best_conj_trp || new_conj_trp->read_cost < best_conj_trp->read_cost)) { best_conj_trp= new_conj_trp; best_read_time= best_conj_trp->read_cost; } } if (best_conj_trp) best_trp= best_conj_trp; } } thd->mem_root= param.old_root; /* If we got a read plan, create a quick select from it. */ if (best_trp) { records= best_trp->records; if (!(quick= best_trp->make_quick(¶m, TRUE)) || quick->init()) { delete quick; quick= NULL; } } free_mem: free_root(&alloc,MYF(0)); // Return memory & allocator thd->mem_root= param.old_root; thd->no_errors=0; } DBUG_EXECUTE("info", print_quick(quick, &needed_reg);); /* Assume that if the user is using 'limit' we will only need to scan limit rows if we are using a key */ DBUG_RETURN(records ? test(quick) : -1); } /**************************************************************************** * Partition pruning module ****************************************************************************/ #ifdef WITH_PARTITION_STORAGE_ENGINE /* PartitionPruningModule This part of the code does partition pruning. Partition pruning solves the following problem: given a query over partitioned tables, find partitions that we will not need to access (i.e. partitions that we can assume to be empty) when executing the query. The set of partitions to prune doesn't depend on which query execution plan will be used to execute the query. HOW IT WORKS Partition pruning module makes use of RangeAnalysisModule. The following examples show how the problem of partition pruning can be reduced to the range analysis problem: EXAMPLE 1 Consider a query: SELECT * FROM t1 WHERE (t1.a < 5 OR t1.a = 10) AND t1.a > 3 AND t1.b='z' where table t1 is partitioned using PARTITION BY RANGE(t1.a). An apparent way to find the used (i.e. not pruned away) partitions is as follows: 1. analyze the WHERE clause and extract the list of intervals over t1.a for the above query we will get this list: {(3 < t1.a < 5), (t1.a=10)} 2. for each interval I { find partitions that have non-empty intersection with I; mark them as used; } EXAMPLE 2 Suppose the table is partitioned by HASH(part_func(t1.a, t1.b)). Then we need to: 1. Analyze the WHERE clause and get a list of intervals over (t1.a, t1.b). The list of intervals we'll obtain will look like this: ((t1.a, t1.b) = (1,'foo')), ((t1.a, t1.b) = (2,'bar')), ((t1,a, t1.b) > (10,'zz')) 2. for each interval I { if (the interval has form "(t1.a, t1.b) = (const1, const2)" ) { calculate HASH(part_func(t1.a, t1.b)); find which partition has records with this hash value and mark it as used; } else { mark all partitions as used; break; } } For both examples the step #1 is exactly what RangeAnalysisModule could be used to do, if it was provided with appropriate index description (array of KEY_PART structures). In example #1, we need to provide it with description of index(t1.a), in example #2, we need to provide it with description of index(t1.a, t1.b). These index descriptions are further called "partitioning index descriptions". Note that it doesn't matter if such indexes really exist, as range analysis module only uses the description. Putting it all together, partitioning module works as follows: prune_partitions() { call create_partition_index_description(); call get_mm_tree(); // invoke the RangeAnalysisModule // analyze the obtained interval list and get used partitions call find_used_partitions(); } */ struct st_part_prune_param; struct st_part_opt_info; typedef void (*mark_full_part_func)(partition_info*, uint32); /* Partition pruning operation context */ typedef struct st_part_prune_param { RANGE_OPT_PARAM range_param; /* Range analyzer parameters */ /*************************************************************** Following fields are filled in based solely on partitioning definition and not modified after that: **************************************************************/ partition_info *part_info; /* Copy of table->part_info */ /* Function to get partition id from partitioning fields only */ get_part_id_func get_top_partition_id_func; /* Function to mark a partition as used (w/all subpartitions if they exist)*/ mark_full_part_func mark_full_partition_used; /* Partitioning 'index' description, array of key parts */ KEY_PART *key; /* Number of fields in partitioning 'index' definition created for partitioning (0 if partitioning 'index' doesn't include partitioning fields) */ uint part_fields; uint subpart_fields; /* Same as above for subpartitioning */ /* Number of the last partitioning field keypart in the index, or -1 if partitioning index definition doesn't include partitioning fields. */ int last_part_partno; int last_subpart_partno; /* Same as above for supartitioning */ /* is_part_keypart[i] == test(keypart #i in partitioning index is a member used in partitioning) Used to maintain current values of cur_part_fields and cur_subpart_fields */ my_bool *is_part_keypart; /* Same as above for subpartitioning */ my_bool *is_subpart_keypart; my_bool ignore_part_fields; /* Ignore rest of partioning fields */ /*************************************************************** Following fields form find_used_partitions() recursion context: **************************************************************/ SEL_ARG **arg_stack; /* "Stack" of SEL_ARGs */ SEL_ARG **arg_stack_end; /* Top of the stack */ /* Number of partitioning fields for which we have a SEL_ARG* in arg_stack */ uint cur_part_fields; /* Same as cur_part_fields, but for subpartitioning */ uint cur_subpart_fields; /* Iterator to be used to obtain the "current" set of used partitions */ PARTITION_ITERATOR part_iter; /* Initialized bitmap of num_subparts size */ MY_BITMAP subparts_bitmap; uchar *cur_min_key; uchar *cur_max_key; uint cur_min_flag, cur_max_flag; } PART_PRUNE_PARAM; static bool create_partition_index_description(PART_PRUNE_PARAM *prune_par); static int find_used_partitions(PART_PRUNE_PARAM *ppar, SEL_ARG *key_tree); static int find_used_partitions_imerge(PART_PRUNE_PARAM *ppar, SEL_IMERGE *imerge); static int find_used_partitions_imerge_list(PART_PRUNE_PARAM *ppar, List &merges); static void mark_all_partitions_as_used(partition_info *part_info); #ifndef DBUG_OFF static void print_partitioning_index(KEY_PART *parts, KEY_PART *parts_end); static void dbug_print_field(Field *field); static void dbug_print_segment_range(SEL_ARG *arg, KEY_PART *part); static void dbug_print_singlepoint_range(SEL_ARG **start, uint num); #endif /* Perform partition pruning for a given table and condition. SYNOPSIS prune_partitions() thd Thread handle table Table to perform partition pruning for pprune_cond Condition to use for partition pruning DESCRIPTION This function assumes that all partitions are marked as unused when it is invoked. The function analyzes the condition, finds partitions that need to be used to retrieve the records that match the condition, and marks them as used by setting appropriate bit in part_info->used_partitions In the worst case all partitions are marked as used. NOTE This function returns promptly if called for non-partitioned table. RETURN TRUE We've inferred that no partitions need to be used (i.e. no table records will satisfy pprune_cond) FALSE Otherwise */ bool prune_partitions(THD *thd, TABLE *table, Item *pprune_cond) { bool retval= FALSE; partition_info *part_info = table->part_info; DBUG_ENTER("prune_partitions"); if (!part_info) DBUG_RETURN(FALSE); /* not a partitioned table */ if (!pprune_cond) { mark_all_partitions_as_used(part_info); DBUG_RETURN(FALSE); } PART_PRUNE_PARAM prune_param; MEM_ROOT alloc; RANGE_OPT_PARAM *range_par= &prune_param.range_param; my_bitmap_map *old_sets[2]; prune_param.part_info= part_info; init_sql_alloc(&alloc, thd->variables.range_alloc_block_size, 0); range_par->mem_root= &alloc; range_par->old_root= thd->mem_root; if (create_partition_index_description(&prune_param)) { mark_all_partitions_as_used(part_info); free_root(&alloc,MYF(0)); // Return memory & allocator DBUG_RETURN(FALSE); } dbug_tmp_use_all_columns(table, old_sets, table->read_set, table->write_set); range_par->thd= thd; range_par->table= table; /* range_par->cond doesn't need initialization */ range_par->prev_tables= range_par->read_tables= 0; range_par->current_table= table->map; range_par->keys= 1; // one index range_par->using_real_indexes= FALSE; range_par->remove_jump_scans= FALSE; range_par->real_keynr[0]= 0; range_par->alloced_sel_args= 0; thd->no_errors=1; // Don't warn about NULL thd->mem_root=&alloc; bitmap_clear_all(&part_info->used_partitions); prune_param.key= prune_param.range_param.key_parts; SEL_TREE *tree; int res; tree= get_mm_tree(range_par, pprune_cond); if (!tree) goto all_used; if (tree->type == SEL_TREE::IMPOSSIBLE) { retval= TRUE; goto end; } if (tree->type != SEL_TREE::KEY && tree->type != SEL_TREE::KEY_SMALLER) goto all_used; if (tree->merges.is_empty()) { /* Range analysis has produced a single list of intervals. */ prune_param.arg_stack_end= prune_param.arg_stack; prune_param.cur_part_fields= 0; prune_param.cur_subpart_fields= 0; prune_param.cur_min_key= prune_param.range_param.min_key; prune_param.cur_max_key= prune_param.range_param.max_key; prune_param.cur_min_flag= prune_param.cur_max_flag= 0; init_all_partitions_iterator(part_info, &prune_param.part_iter); if (!tree->keys[0] || (-1 == (res= find_used_partitions(&prune_param, tree->keys[0])))) goto all_used; } else { if (tree->merges.elements == 1) { /* Range analysis has produced a "merge" of several intervals lists, a SEL_TREE that represents an expression in form sel_imerge = (tree1 OR tree2 OR ... OR treeN) that cannot be reduced to one tree. This can only happen when partitioning index has several keyparts and the condition is OR of conditions that refer to different key parts. For example, we'll get here for "partitioning_field=const1 OR subpartitioning_field=const2" */ if (-1 == (res= find_used_partitions_imerge(&prune_param, tree->merges.head()))) goto all_used; } else { /* Range analysis has produced a list of several imerges, i.e. a structure that represents a condition in form imerge_list= (sel_imerge1 AND sel_imerge2 AND ... AND sel_imergeN) This is produced for complicated WHERE clauses that range analyzer can't really analyze properly. */ if (-1 == (res= find_used_partitions_imerge_list(&prune_param, tree->merges))) goto all_used; } } /* res == 0 => no used partitions => retval=TRUE res == 1 => some used partitions => retval=FALSE res == -1 - we jump over this line to all_used: */ retval= test(!res); goto end; all_used: retval= FALSE; // some partitions are used mark_all_partitions_as_used(prune_param.part_info); end: dbug_tmp_restore_column_maps(table->read_set, table->write_set, old_sets); thd->no_errors=0; thd->mem_root= range_par->old_root; free_root(&alloc,MYF(0)); // Return memory & allocator DBUG_RETURN(retval); } /* Store field key image to table record SYNOPSIS store_key_image_to_rec() field Field which key image should be stored ptr Field value in key format len Length of the value, in bytes DESCRIPTION Copy the field value from its key image to the table record. The source is the value in key image format, occupying len bytes in buffer pointed by ptr. The destination is table record, in "field value in table record" format. */ void store_key_image_to_rec(Field *field, uchar *ptr, uint len) { /* Do the same as print_key() does */ my_bitmap_map *old_map; if (field->real_maybe_null()) { if (*ptr) { field->set_null(); return; } field->set_notnull(); ptr++; } old_map= dbug_tmp_use_all_columns(field->table, field->table->write_set); field->set_key_image(ptr, len); dbug_tmp_restore_column_map(field->table->write_set, old_map); } /* For SEL_ARG* array, store sel_arg->min values into table record buffer SYNOPSIS store_selargs_to_rec() ppar Partition pruning context start Array of SEL_ARG* for which the minimum values should be stored num Number of elements in the array DESCRIPTION For each SEL_ARG* interval in the specified array, store the left edge field value (sel_arg->min, key image format) into the table record. */ static void store_selargs_to_rec(PART_PRUNE_PARAM *ppar, SEL_ARG **start, int num) { KEY_PART *parts= ppar->range_param.key_parts; for (SEL_ARG **end= start + num; start != end; start++) { SEL_ARG *sel_arg= (*start); store_key_image_to_rec(sel_arg->field, sel_arg->min_value, parts[sel_arg->part].length); } } /* Mark a partition as used in the case when there are no subpartitions */ static void mark_full_partition_used_no_parts(partition_info* part_info, uint32 part_id) { DBUG_ENTER("mark_full_partition_used_no_parts"); DBUG_PRINT("enter", ("Mark partition %u as used", part_id)); bitmap_set_bit(&part_info->used_partitions, part_id); DBUG_VOID_RETURN; } /* Mark a partition as used in the case when there are subpartitions */ static void mark_full_partition_used_with_parts(partition_info *part_info, uint32 part_id) { uint32 start= part_id * part_info->num_subparts; uint32 end= start + part_info->num_subparts; DBUG_ENTER("mark_full_partition_used_with_parts"); for (; start != end; start++) { DBUG_PRINT("info", ("1:Mark subpartition %u as used", start)); bitmap_set_bit(&part_info->used_partitions, start); } DBUG_VOID_RETURN; } /* Find the set of used partitions for List SYNOPSIS find_used_partitions_imerge_list ppar Partition pruning context. key_tree Intervals tree to perform pruning for. DESCRIPTION List represents "imerge1 AND imerge2 AND ...". The set of used partitions is an intersection of used partitions sets for imerge_{i}. We accumulate this intersection in a separate bitmap. RETURN See find_used_partitions() */ static int find_used_partitions_imerge_list(PART_PRUNE_PARAM *ppar, List &merges) { MY_BITMAP all_merges; uint bitmap_bytes; my_bitmap_map *bitmap_buf; uint n_bits= ppar->part_info->used_partitions.n_bits; bitmap_bytes= bitmap_buffer_size(n_bits); if (!(bitmap_buf= (my_bitmap_map*) alloc_root(ppar->range_param.mem_root, bitmap_bytes))) { /* Fallback, process just the first SEL_IMERGE. This can leave us with more partitions marked as used then actually needed. */ return find_used_partitions_imerge(ppar, merges.head()); } bitmap_init(&all_merges, bitmap_buf, n_bits, FALSE); bitmap_set_prefix(&all_merges, n_bits); List_iterator it(merges); SEL_IMERGE *imerge; while ((imerge=it++)) { int res= find_used_partitions_imerge(ppar, imerge); if (!res) { /* no used partitions on one ANDed imerge => no used partitions at all */ return 0; } if (res != -1) bitmap_intersect(&all_merges, &ppar->part_info->used_partitions); if (bitmap_is_clear_all(&all_merges)) return 0; bitmap_clear_all(&ppar->part_info->used_partitions); } memcpy(ppar->part_info->used_partitions.bitmap, all_merges.bitmap, bitmap_bytes); return 1; } /* Find the set of used partitions for SEL_IMERGE structure SYNOPSIS find_used_partitions_imerge() ppar Partition pruning context. key_tree Intervals tree to perform pruning for. DESCRIPTION SEL_IMERGE represents "tree1 OR tree2 OR ...". The implementation is trivial - just use mark used partitions for each tree and bail out early if for some tree_{i} all partitions are used. RETURN See find_used_partitions(). */ static int find_used_partitions_imerge(PART_PRUNE_PARAM *ppar, SEL_IMERGE *imerge) { int res= 0; for (SEL_TREE **ptree= imerge->trees; ptree < imerge->trees_next; ptree++) { ppar->arg_stack_end= ppar->arg_stack; ppar->cur_part_fields= 0; ppar->cur_subpart_fields= 0; ppar->cur_min_key= ppar->range_param.min_key; ppar->cur_max_key= ppar->range_param.max_key; ppar->cur_min_flag= ppar->cur_max_flag= 0; init_all_partitions_iterator(ppar->part_info, &ppar->part_iter); SEL_ARG *key_tree= (*ptree)->keys[0]; if (!key_tree || (-1 == (res |= find_used_partitions(ppar, key_tree)))) return -1; } return res; } /* Collect partitioning ranges for the SEL_ARG tree and mark partitions as used SYNOPSIS find_used_partitions() ppar Partition pruning context. key_tree SEL_ARG range tree to perform pruning for DESCRIPTION This function * recursively walks the SEL_ARG* tree collecting partitioning "intervals" * finds the partitions one needs to use to get rows in these intervals * marks these partitions as used. The next session desribes the process in greater detail. IMPLEMENTATION TYPES OF RESTRICTIONS THAT WE CAN OBTAIN PARTITIONS FOR We can find out which [sub]partitions to use if we obtain restrictions on [sub]partitioning fields in the following form: 1. "partition_field1=const1 AND ... AND partition_fieldN=constN" 1.1 Same as (1) but for subpartition fields If partitioning supports interval analysis (i.e. partitioning is a function of a single table field, and partition_info:: get_part_iter_for_interval != NULL), then we can also use condition in this form: 2. "const1 <=? partition_field <=? const2" 2.1 Same as (2) but for subpartition_field INFERRING THE RESTRICTIONS FROM SEL_ARG TREE The below is an example of what SEL_ARG tree may represent: (start) | $ | Partitioning keyparts $ subpartitioning keyparts | $ | ... ... $ | | | $ | +---------+ +---------+ $ +-----------+ +-----------+ \-| par1=c1 |--| par2=c2 |-----| subpar1=c3|--| subpar2=c5| +---------+ +---------+ $ +-----------+ +-----------+ | $ | | | $ | +-----------+ | $ | | subpar2=c6| | $ | +-----------+ | $ | | $ +-----------+ +-----------+ | $ | subpar1=c4|--| subpar2=c8| | $ +-----------+ +-----------+ | $ | $ +---------+ $ +------------+ +------------+ | par1=c2 |------------------| subpar1=c10|--| subpar2=c12| +---------+ $ +------------+ +------------+ | $ ... $ The up-down connections are connections via SEL_ARG::left and SEL_ARG::right. A horizontal connection to the right is the SEL_ARG::next_key_part connection. find_used_partitions() traverses the entire tree via recursion on * SEL_ARG::next_key_part (from left to right on the picture) * SEL_ARG::left|right (up/down on the pic). Left-right recursion is performed for each depth level. Recursion descent on SEL_ARG::next_key_part is used to accumulate (in ppar->arg_stack) constraints on partitioning and subpartitioning fields. For the example in the above picture, one of stack states is: in find_used_partitions(key_tree = "subpar2=c5") (***) in find_used_partitions(key_tree = "subpar1=c3") in find_used_partitions(key_tree = "par2=c2") (**) in find_used_partitions(key_tree = "par1=c1") in prune_partitions(...) We apply partitioning limits as soon as possible, e.g. when we reach the depth (**), we find which partition(s) correspond to "par1=c1 AND par2=c2", and save them in ppar->part_iter. When we reach the depth (***), we find which subpartition(s) correspond to "subpar1=c3 AND subpar2=c5", and then mark appropriate subpartitions in appropriate subpartitions as used. It is possible that constraints on some partitioning fields are missing. For the above example, consider this stack state: in find_used_partitions(key_tree = "subpar2=c12") (***) in find_used_partitions(key_tree = "subpar1=c10") in find_used_partitions(key_tree = "par1=c2") in prune_partitions(...) Here we don't have constraints for all partitioning fields. Since we've never set the ppar->part_iter to contain used set of partitions, we use its default "all partitions" value. We get subpartition id for "subpar1=c3 AND subpar2=c5", and mark that subpartition as used in every partition. The inverse is also possible: we may get constraints on partitioning fields, but not constraints on subpartitioning fields. In that case, calls to find_used_partitions() with depth below (**) will return -1, and we will mark entire partition as used. TODO Replace recursion on SEL_ARG::left and SEL_ARG::right with a loop RETURN 1 OK, one or more [sub]partitions are marked as used. 0 The passed condition doesn't match any partitions -1 Couldn't infer any partition pruning "intervals" from the passed SEL_ARG* tree (which means that all partitions should be marked as used) Marking partitions as used is the responsibility of the caller. */ static int find_used_partitions(PART_PRUNE_PARAM *ppar, SEL_ARG *key_tree) { int res, left_res=0, right_res=0; int key_tree_part= (int)key_tree->part; bool set_full_part_if_bad_ret= FALSE; bool ignore_part_fields= ppar->ignore_part_fields; bool did_set_ignore_part_fields= FALSE; RANGE_OPT_PARAM *range_par= &(ppar->range_param); if (check_stack_overrun(range_par->thd, 3*STACK_MIN_SIZE, NULL)) return -1; if (key_tree->left != &null_element) { if (-1 == (left_res= find_used_partitions(ppar,key_tree->left))) return -1; } /* Push SEL_ARG's to stack to enable looking backwards as well */ ppar->cur_part_fields+= ppar->is_part_keypart[key_tree_part]; ppar->cur_subpart_fields+= ppar->is_subpart_keypart[key_tree_part]; *(ppar->arg_stack_end++)= key_tree; if (key_tree->type == SEL_ARG::KEY_RANGE) { if (ppar->part_info->get_part_iter_for_interval && key_tree->part <= ppar->last_part_partno) { if (ignore_part_fields) { /* We come here when a condition on the first partitioning fields led to evaluating the partitioning condition (due to finding a condition of the type a < const or b > const). Thus we must ignore the rest of the partitioning fields but we still want to analyse the subpartitioning fields. */ if (key_tree->next_key_part) res= find_used_partitions(ppar, key_tree->next_key_part); else res= -1; goto pop_and_go_right; } /* Collect left and right bound, their lengths and flags */ uchar *min_key= ppar->cur_min_key; uchar *max_key= ppar->cur_max_key; uchar *tmp_min_key= min_key; uchar *tmp_max_key= max_key; key_tree->store_min(ppar->key[key_tree->part].store_length, &tmp_min_key, ppar->cur_min_flag); key_tree->store_max(ppar->key[key_tree->part].store_length, &tmp_max_key, ppar->cur_max_flag); uint flag; if (key_tree->next_key_part && key_tree->next_key_part->part == key_tree->part+1 && key_tree->next_key_part->part <= ppar->last_part_partno && key_tree->next_key_part->type == SEL_ARG::KEY_RANGE) { /* There are more key parts for partition pruning to handle This mainly happens when the condition is an equality condition. */ if ((tmp_min_key - min_key) == (tmp_max_key - max_key) && (memcmp(min_key, max_key, (uint)(tmp_max_key - max_key)) == 0) && !key_tree->min_flag && !key_tree->max_flag) { /* Set 'parameters' */ ppar->cur_min_key= tmp_min_key; ppar->cur_max_key= tmp_max_key; uint save_min_flag= ppar->cur_min_flag; uint save_max_flag= ppar->cur_max_flag; ppar->cur_min_flag|= key_tree->min_flag; ppar->cur_max_flag|= key_tree->max_flag; res= find_used_partitions(ppar, key_tree->next_key_part); /* Restore 'parameters' back */ ppar->cur_min_key= min_key; ppar->cur_max_key= max_key; ppar->cur_min_flag= save_min_flag; ppar->cur_max_flag= save_max_flag; goto pop_and_go_right; } /* We have arrived at the last field in the partition pruning */ uint tmp_min_flag= key_tree->min_flag, tmp_max_flag= key_tree->max_flag; if (!tmp_min_flag) key_tree->next_key_part->store_min_key(ppar->key, &tmp_min_key, &tmp_min_flag, ppar->last_part_partno); if (!tmp_max_flag) key_tree->next_key_part->store_max_key(ppar->key, &tmp_max_key, &tmp_max_flag, ppar->last_part_partno); flag= tmp_min_flag | tmp_max_flag; } else flag= key_tree->min_flag | key_tree->max_flag; if (tmp_min_key != range_par->min_key) flag&= ~NO_MIN_RANGE; else flag|= NO_MIN_RANGE; if (tmp_max_key != range_par->max_key) flag&= ~NO_MAX_RANGE; else flag|= NO_MAX_RANGE; /* We need to call the interval mapper if we have a condition which makes sense to prune on. In the example of COLUMNS on a and b it makes sense if we have a condition on a, or conditions on both a and b. If we only have conditions on b it might make sense but this is a harder case we will solve later. For the harder case this clause then turns into use of all partitions and thus we simply set res= -1 as if the mapper had returned that. TODO: What to do here is defined in WL#4065. */ if (ppar->arg_stack[0]->part == 0) { uint32 i; uint32 store_length_array[MAX_KEY]; uint32 num_keys= ppar->part_fields; for (i= 0; i < num_keys; i++) store_length_array[i]= ppar->key[i].store_length; res= ppar->part_info-> get_part_iter_for_interval(ppar->part_info, FALSE, store_length_array, range_par->min_key, range_par->max_key, tmp_min_key - range_par->min_key, tmp_max_key - range_par->max_key, flag, &ppar->part_iter); if (!res) goto pop_and_go_right; /* res==0 --> no satisfying partitions */ } else res= -1; if (res == -1) { /* get a full range iterator */ init_all_partitions_iterator(ppar->part_info, &ppar->part_iter); } /* Save our intent to mark full partition as used if we will not be able to obtain further limits on subpartitions */ if (key_tree_part < ppar->last_part_partno) { /* We need to ignore the rest of the partitioning fields in all evaluations after this */ did_set_ignore_part_fields= TRUE; ppar->ignore_part_fields= TRUE; } set_full_part_if_bad_ret= TRUE; goto process_next_key_part; } if (key_tree_part == ppar->last_subpart_partno && (NULL != ppar->part_info->get_subpart_iter_for_interval)) { PARTITION_ITERATOR subpart_iter; DBUG_EXECUTE("info", dbug_print_segment_range(key_tree, range_par->key_parts);); res= ppar->part_info-> get_subpart_iter_for_interval(ppar->part_info, TRUE, NULL, /* Currently not used here */ key_tree->min_value, key_tree->max_value, 0, 0, /* Those are ignored here */ key_tree->min_flag | key_tree->max_flag, &subpart_iter); DBUG_ASSERT(res); /* We can't get "no satisfying subpartitions" */ if (res == -1) goto pop_and_go_right; /* all subpartitions satisfy */ uint32 subpart_id; bitmap_clear_all(&ppar->subparts_bitmap); while ((subpart_id= subpart_iter.get_next(&subpart_iter)) != NOT_A_PARTITION_ID) bitmap_set_bit(&ppar->subparts_bitmap, subpart_id); /* Mark each partition as used in each subpartition. */ uint32 part_id; while ((part_id= ppar->part_iter.get_next(&ppar->part_iter)) != NOT_A_PARTITION_ID) { for (uint i= 0; i < ppar->part_info->num_subparts; i++) if (bitmap_is_set(&ppar->subparts_bitmap, i)) bitmap_set_bit(&ppar->part_info->used_partitions, part_id * ppar->part_info->num_subparts + i); } goto pop_and_go_right; } if (key_tree->is_singlepoint()) { if (key_tree_part == ppar->last_part_partno && ppar->cur_part_fields == ppar->part_fields && ppar->part_info->get_part_iter_for_interval == NULL) { /* Ok, we've got "fieldN<=>constN"-type SEL_ARGs for all partitioning fields. Save all constN constants into table record buffer. */ store_selargs_to_rec(ppar, ppar->arg_stack, ppar->part_fields); DBUG_EXECUTE("info", dbug_print_singlepoint_range(ppar->arg_stack, ppar->part_fields);); uint32 part_id; longlong func_value; /* Find in which partition the {const1, ...,constN} tuple goes */ if (ppar->get_top_partition_id_func(ppar->part_info, &part_id, &func_value)) { res= 0; /* No satisfying partitions */ goto pop_and_go_right; } /* Rembember the limit we got - single partition #part_id */ init_single_partition_iterator(part_id, &ppar->part_iter); /* If there are no subpartitions/we fail to get any limit for them, then we'll mark full partition as used. */ set_full_part_if_bad_ret= TRUE; goto process_next_key_part; } if (key_tree_part == ppar->last_subpart_partno && ppar->cur_subpart_fields == ppar->subpart_fields) { /* Ok, we've got "fieldN<=>constN"-type SEL_ARGs for all subpartitioning fields. Save all constN constants into table record buffer. */ store_selargs_to_rec(ppar, ppar->arg_stack_end - ppar->subpart_fields, ppar->subpart_fields); DBUG_EXECUTE("info", dbug_print_singlepoint_range(ppar->arg_stack_end- ppar->subpart_fields, ppar->subpart_fields);); /* Find the subpartition (it's HASH/KEY so we always have one) */ partition_info *part_info= ppar->part_info; uint32 part_id, subpart_id; if (part_info->get_subpartition_id(part_info, &subpart_id)) return 0; /* Mark this partition as used in each subpartition. */ while ((part_id= ppar->part_iter.get_next(&ppar->part_iter)) != NOT_A_PARTITION_ID) { bitmap_set_bit(&part_info->used_partitions, part_id * part_info->num_subparts + subpart_id); } res= 1; /* Some partitions were marked as used */ goto pop_and_go_right; } } else { /* Can't handle condition on current key part. If we're that deep that we're processing subpartititoning's key parts, this means we'll not be able to infer any suitable condition, so bail out. */ if (key_tree_part >= ppar->last_part_partno) { res= -1; goto pop_and_go_right; } } } process_next_key_part: if (key_tree->next_key_part) res= find_used_partitions(ppar, key_tree->next_key_part); else res= -1; if (did_set_ignore_part_fields) { /* We have returned from processing all key trees linked to our next key part. We are ready to be moving down (using right pointers) and this tree is a new evaluation requiring its own decision on whether to ignore partitioning fields. */ ppar->ignore_part_fields= FALSE; } if (set_full_part_if_bad_ret) { if (res == -1) { /* Got "full range" for subpartitioning fields */ uint32 part_id; bool found= FALSE; while ((part_id= ppar->part_iter.get_next(&ppar->part_iter)) != NOT_A_PARTITION_ID) { ppar->mark_full_partition_used(ppar->part_info, part_id); found= TRUE; } res= test(found); } /* Restore the "used partitions iterator" to the default setting that specifies iteration over all partitions. */ init_all_partitions_iterator(ppar->part_info, &ppar->part_iter); } pop_and_go_right: /* Pop this key part info off the "stack" */ ppar->arg_stack_end--; ppar->cur_part_fields-= ppar->is_part_keypart[key_tree_part]; ppar->cur_subpart_fields-= ppar->is_subpart_keypart[key_tree_part]; if (res == -1) return -1; if (key_tree->right != &null_element) { if (-1 == (right_res= find_used_partitions(ppar,key_tree->right))) return -1; } return (left_res || right_res || res); } static void mark_all_partitions_as_used(partition_info *part_info) { bitmap_set_all(&part_info->used_partitions); } /* Check if field types allow to construct partitioning index description SYNOPSIS fields_ok_for_partition_index() pfield NULL-terminated array of pointers to fields. DESCRIPTION For an array of fields, check if we can use all of the fields to create partitioning index description. We can't process GEOMETRY fields - for these fields singlepoint intervals cant be generated, and non-singlepoint are "special" kinds of intervals to which our processing logic can't be applied. It is not known if we could process ENUM fields, so they are disabled to be on the safe side. RETURN TRUE Yes, fields can be used in partitioning index FALSE Otherwise */ static bool fields_ok_for_partition_index(Field **pfield) { if (!pfield) return FALSE; for (; (*pfield); pfield++) { enum_field_types ftype= (*pfield)->real_type(); if (ftype == MYSQL_TYPE_ENUM || ftype == MYSQL_TYPE_GEOMETRY) return FALSE; } return TRUE; } /* Create partition index description and fill related info in the context struct SYNOPSIS create_partition_index_description() prune_par INOUT Partition pruning context DESCRIPTION Create partition index description. Partition index description is: part_index(used_fields_list(part_expr), used_fields_list(subpart_expr)) If partitioning/sub-partitioning uses BLOB or Geometry fields, then corresponding fields_list(...) is not included into index description and we don't perform partition pruning for partitions/subpartitions. RETURN TRUE Out of memory or can't do partition pruning at all FALSE OK */ static bool create_partition_index_description(PART_PRUNE_PARAM *ppar) { RANGE_OPT_PARAM *range_par= &(ppar->range_param); partition_info *part_info= ppar->part_info; uint used_part_fields, used_subpart_fields; used_part_fields= fields_ok_for_partition_index(part_info->part_field_array) ? part_info->num_part_fields : 0; used_subpart_fields= fields_ok_for_partition_index(part_info->subpart_field_array)? part_info->num_subpart_fields : 0; uint total_parts= used_part_fields + used_subpart_fields; ppar->ignore_part_fields= FALSE; ppar->part_fields= used_part_fields; ppar->last_part_partno= (int)used_part_fields - 1; ppar->subpart_fields= used_subpart_fields; ppar->last_subpart_partno= used_subpart_fields?(int)(used_part_fields + used_subpart_fields - 1): -1; if (part_info->is_sub_partitioned()) { ppar->mark_full_partition_used= mark_full_partition_used_with_parts; ppar->get_top_partition_id_func= part_info->get_part_partition_id; } else { ppar->mark_full_partition_used= mark_full_partition_used_no_parts; ppar->get_top_partition_id_func= part_info->get_partition_id; } KEY_PART *key_part; MEM_ROOT *alloc= range_par->mem_root; if (!total_parts || !(key_part= (KEY_PART*)alloc_root(alloc, sizeof(KEY_PART)* total_parts)) || !(ppar->arg_stack= (SEL_ARG**)alloc_root(alloc, sizeof(SEL_ARG*)* total_parts)) || !(ppar->is_part_keypart= (my_bool*)alloc_root(alloc, sizeof(my_bool)* total_parts)) || !(ppar->is_subpart_keypart= (my_bool*)alloc_root(alloc, sizeof(my_bool)* total_parts))) return TRUE; if (ppar->subpart_fields) { my_bitmap_map *buf; uint32 bufsize= bitmap_buffer_size(ppar->part_info->num_subparts); if (!(buf= (my_bitmap_map*) alloc_root(alloc, bufsize))) return TRUE; bitmap_init(&ppar->subparts_bitmap, buf, ppar->part_info->num_subparts, FALSE); } range_par->key_parts= key_part; Field **field= (ppar->part_fields)? part_info->part_field_array : part_info->subpart_field_array; bool in_subpart_fields= FALSE; for (uint part= 0; part < total_parts; part++, key_part++) { key_part->key= 0; key_part->part= part; key_part->length= (uint16)(*field)->key_length(); key_part->store_length= (uint16)get_partition_field_store_length(*field); DBUG_PRINT("info", ("part %u length %u store_length %u", part, key_part->length, key_part->store_length)); key_part->field= (*field); key_part->image_type = Field::itRAW; /* We set keypart flag to 0 here as the only HA_PART_KEY_SEG is checked in the RangeAnalysisModule. */ key_part->flag= 0; /* We don't set key_parts->null_bit as it will not be used */ ppar->is_part_keypart[part]= !in_subpart_fields; ppar->is_subpart_keypart[part]= in_subpart_fields; /* Check if this was last field in this array, in this case we switch to subpartitioning fields. (This will only happens if there are subpartitioning fields to cater for). */ if (!*(++field)) { field= part_info->subpart_field_array; in_subpart_fields= TRUE; } } range_par->key_parts_end= key_part; DBUG_EXECUTE("info", print_partitioning_index(range_par->key_parts, range_par->key_parts_end);); return FALSE; } #ifndef DBUG_OFF static void print_partitioning_index(KEY_PART *parts, KEY_PART *parts_end) { DBUG_ENTER("print_partitioning_index"); DBUG_LOCK_FILE; fprintf(DBUG_FILE, "partitioning INDEX("); for (KEY_PART *p=parts; p != parts_end; p++) { fprintf(DBUG_FILE, "%s%s", p==parts?"":" ,", p->field->field_name); } fputs(");\n", DBUG_FILE); DBUG_UNLOCK_FILE; DBUG_VOID_RETURN; } /* Print field value into debug trace, in NULL-aware way. */ static void dbug_print_field(Field *field) { if (field->is_real_null()) fprintf(DBUG_FILE, "NULL"); else { char buf[256]; String str(buf, sizeof(buf), &my_charset_bin); str.length(0); String *pstr; pstr= field->val_str(&str); fprintf(DBUG_FILE, "'%s'", pstr->c_ptr_safe()); } } /* Print a "c1 < keypartX < c2" - type interval into debug trace. */ static void dbug_print_segment_range(SEL_ARG *arg, KEY_PART *part) { DBUG_ENTER("dbug_print_segment_range"); DBUG_LOCK_FILE; if (!(arg->min_flag & NO_MIN_RANGE)) { store_key_image_to_rec(part->field, arg->min_value, part->length); dbug_print_field(part->field); if (arg->min_flag & NEAR_MIN) fputs(" < ", DBUG_FILE); else fputs(" <= ", DBUG_FILE); } fprintf(DBUG_FILE, "%s", part->field->field_name); if (!(arg->max_flag & NO_MAX_RANGE)) { if (arg->max_flag & NEAR_MAX) fputs(" < ", DBUG_FILE); else fputs(" <= ", DBUG_FILE); store_key_image_to_rec(part->field, arg->max_value, part->length); dbug_print_field(part->field); } fputs("\n", DBUG_FILE); DBUG_UNLOCK_FILE; DBUG_VOID_RETURN; } /* Print a singlepoint multi-keypart range interval to debug trace SYNOPSIS dbug_print_singlepoint_range() start Array of SEL_ARG* ptrs representing conditions on key parts num Number of elements in the array. DESCRIPTION This function prints a "keypartN=constN AND ... AND keypartK=constK"-type interval to debug trace. */ static void dbug_print_singlepoint_range(SEL_ARG **start, uint num) { DBUG_ENTER("dbug_print_singlepoint_range"); DBUG_LOCK_FILE; SEL_ARG **end= start + num; for (SEL_ARG **arg= start; arg != end; arg++) { Field *field= (*arg)->field; fprintf(DBUG_FILE, "%s%s=", (arg==start)?"":", ", field->field_name); dbug_print_field(field); } fputs("\n", DBUG_FILE); DBUG_UNLOCK_FILE; DBUG_VOID_RETURN; } #endif /**************************************************************************** * Partition pruning code ends ****************************************************************************/ #endif /* Get cost of 'sweep' full records retrieval. SYNOPSIS get_sweep_read_cost() param Parameter from test_quick_select records # of records to be retrieved RETURN cost of sweep */ double get_sweep_read_cost(const PARAM *param, ha_rows records) { double result; DBUG_ENTER("get_sweep_read_cost"); if (param->table->file->primary_key_is_clustered()) { /* We are using the primary key to find the rows. Calculate the cost for this. */ result= param->table->file->read_time(param->table->s->primary_key, (uint)records, records); } else { /* Rows will be retreived with rnd_pos(). Caluclate the expected cost for this. */ double n_blocks= ceil(ulonglong2double(param->table->file->stats.data_file_length) / IO_SIZE); double busy_blocks= n_blocks * (1.0 - pow(1.0 - 1.0/n_blocks, rows2double(records))); if (busy_blocks < 1.0) busy_blocks= 1.0; DBUG_PRINT("info",("sweep: nblocks: %g, busy_blocks: %g", n_blocks, busy_blocks)); /* Disabled: Bail out if # of blocks to read is bigger than # of blocks in table data file. if (max_cost != DBL_MAX && (busy_blocks+index_reads_cost) >= n_blocks) return 1; */ JOIN *join= param->thd->lex->select_lex.join; if (!join || join->table_count == 1) { /* No join, assume reading is done in one 'sweep' */ result= busy_blocks*(DISK_SEEK_BASE_COST + DISK_SEEK_PROP_COST*n_blocks/busy_blocks); } else { /* Possibly this is a join with source table being non-last table, so assume that disk seeks are random here. */ result= busy_blocks; } } DBUG_PRINT("return",("cost: %g", result)); DBUG_RETURN(result); } /* Get best plan for a SEL_IMERGE disjunctive expression. SYNOPSIS get_best_disjunct_quick() param Parameter from check_quick_select function imerge Expression to use read_time Don't create scans with cost > read_time NOTES index_merge cost is calculated as follows: index_merge_cost = cost(index_reads) + (see #1) cost(rowid_to_row_scan) + (see #2) cost(unique_use) (see #3) 1. cost(index_reads) =SUM_i(cost(index_read_i)) For non-CPK scans, cost(index_read_i) = {cost of ordinary 'index only' scan} For CPK scan, cost(index_read_i) = {cost of non-'index only' scan} 2. cost(rowid_to_row_scan) If table PK is clustered then cost(rowid_to_row_scan) = {cost of ordinary clustered PK scan with n_ranges=n_rows} Otherwise, we use the following model to calculate costs: We need to retrieve n_rows rows from file that occupies n_blocks blocks. We assume that offsets of rows we need are independent variates with uniform distribution in [0..max_file_offset] range. We'll denote block as "busy" if it contains row(s) we need to retrieve and "empty" if doesn't contain rows we need. Probability that a block is empty is (1 - 1/n_blocks)^n_rows (this applies to any block in file). Let x_i be a variate taking value 1 if block #i is empty and 0 otherwise. Then E(x_i) = (1 - 1/n_blocks)^n_rows; E(n_empty_blocks) = E(sum(x_i)) = sum(E(x_i)) = = n_blocks * ((1 - 1/n_blocks)^n_rows) = ~= n_blocks * exp(-n_rows/n_blocks). E(n_busy_blocks) = n_blocks*(1 - (1 - 1/n_blocks)^n_rows) = ~= n_blocks * (1 - exp(-n_rows/n_blocks)). Average size of "hole" between neighbor non-empty blocks is E(hole_size) = n_blocks/E(n_busy_blocks). The total cost of reading all needed blocks in one "sweep" is: E(n_busy_blocks)* (DISK_SEEK_BASE_COST + DISK_SEEK_PROP_COST*n_blocks/E(n_busy_blocks)). 3. Cost of Unique use is calculated in Unique::get_use_cost function. ROR-union cost is calculated in the same way index_merge, but instead of Unique a priority queue is used. RETURN Created read plan NULL - Out of memory or no read scan could be built. */ static TABLE_READ_PLAN *get_best_disjunct_quick(PARAM *param, SEL_IMERGE *imerge, double read_time) { SEL_TREE **ptree; TRP_INDEX_MERGE *imerge_trp= NULL; TRP_RANGE **range_scans; TRP_RANGE **cur_child; TRP_RANGE **cpk_scan= NULL; bool imerge_too_expensive= FALSE; double imerge_cost= 0.0; ha_rows cpk_scan_records= 0; ha_rows non_cpk_scan_records= 0; bool pk_is_clustered= param->table->file->primary_key_is_clustered(); bool all_scans_ror_able= TRUE; bool all_scans_rors= TRUE; uint unique_calc_buff_size; TABLE_READ_PLAN **roru_read_plans; TABLE_READ_PLAN **cur_roru_plan; double roru_index_costs; ha_rows roru_total_records; double roru_intersect_part= 1.0; DBUG_ENTER("get_best_disjunct_quick"); DBUG_PRINT("info", ("Full table scan cost: %g", read_time)); /* In every tree of imerge remove SEL_ARG trees that do not make ranges. If after this removal some SEL_ARG tree becomes empty discard imerge. */ for (ptree= imerge->trees; ptree != imerge->trees_next; ptree++) { if (remove_nonrange_trees(param, *ptree)) { imerge->trees_next= imerge->trees; break; } } uint n_child_scans= imerge->trees_next - imerge->trees; if (!n_child_scans) DBUG_RETURN(NULL); if (!(range_scans= (TRP_RANGE**)alloc_root(param->mem_root, sizeof(TRP_RANGE*)* n_child_scans))) DBUG_RETURN(NULL); /* Collect best 'range' scan for each of disjuncts, and, while doing so, analyze possibility of ROR scans. Also calculate some values needed by other parts of the code. */ for (ptree= imerge->trees, cur_child= range_scans; ptree != imerge->trees_next; ptree++, cur_child++) { DBUG_EXECUTE("info", print_sel_tree(param, *ptree, &(*ptree)->keys_map, "tree in SEL_IMERGE");); if (!(*cur_child= get_key_scans_params(param, *ptree, TRUE, FALSE, read_time))) { /* One of index scans in this index_merge is more expensive than entire table read for another available option. The entire index_merge (and any possible ROR-union) will be more expensive then, too. We continue here only to update SQL_SELECT members. */ imerge_too_expensive= TRUE; } if (imerge_too_expensive) continue; imerge_cost += (*cur_child)->read_cost; all_scans_ror_able &= ((*ptree)->n_ror_scans > 0); all_scans_rors &= (*cur_child)->is_ror; if (pk_is_clustered && param->real_keynr[(*cur_child)->key_idx] == param->table->s->primary_key) { cpk_scan= cur_child; cpk_scan_records= (*cur_child)->records; } else non_cpk_scan_records += (*cur_child)->records; } DBUG_PRINT("info", ("index_merge scans cost %g", imerge_cost)); if (imerge_too_expensive || (imerge_cost > read_time) || ((non_cpk_scan_records+cpk_scan_records >= param->table->file->stats.records) && read_time != DBL_MAX)) { /* Bail out if it is obvious that both index_merge and ROR-union will be more expensive */ DBUG_PRINT("info", ("Sum of index_merge scans is more expensive than " "full table scan, bailing out")); DBUG_RETURN(NULL); } /* If all scans happen to be ROR, proceed to generate a ROR-union plan (it's guaranteed to be cheaper than non-ROR union), unless ROR-unions are disabled in @@optimizer_switch */ if (all_scans_rors && optimizer_flag(param->thd, OPTIMIZER_SWITCH_INDEX_MERGE_UNION)) { roru_read_plans= (TABLE_READ_PLAN**)range_scans; goto skip_to_ror_scan; } if (cpk_scan) { /* Add one ROWID comparison for each row retrieved on non-CPK scan. (it is done in QUICK_RANGE_SELECT::row_in_ranges) */ imerge_cost += non_cpk_scan_records / TIME_FOR_COMPARE_ROWID; } /* Calculate cost(rowid_to_row_scan) */ imerge_cost += get_sweep_read_cost(param, non_cpk_scan_records); DBUG_PRINT("info",("index_merge cost with rowid-to-row scan: %g", imerge_cost)); if (imerge_cost > read_time || !optimizer_flag(param->thd, OPTIMIZER_SWITCH_INDEX_MERGE_SORT_UNION)) { goto build_ror_index_merge; } /* Add Unique operations cost */ unique_calc_buff_size= Unique::get_cost_calc_buff_size((ulong)non_cpk_scan_records, param->table->file->ref_length, param->thd->variables.sortbuff_size); if (param->imerge_cost_buff_size < unique_calc_buff_size) { if (!(param->imerge_cost_buff= (uint*)alloc_root(param->mem_root, unique_calc_buff_size))) DBUG_RETURN(NULL); param->imerge_cost_buff_size= unique_calc_buff_size; } imerge_cost += Unique::get_use_cost(param->imerge_cost_buff, (uint)non_cpk_scan_records, param->table->file->ref_length, param->thd->variables.sortbuff_size, TIME_FOR_COMPARE_ROWID, FALSE, NULL); DBUG_PRINT("info",("index_merge total cost: %g (wanted: less then %g)", imerge_cost, read_time)); if (imerge_cost < read_time) { if ((imerge_trp= new (param->mem_root)TRP_INDEX_MERGE)) { imerge_trp->read_cost= imerge_cost; imerge_trp->records= non_cpk_scan_records + cpk_scan_records; imerge_trp->records= min(imerge_trp->records, param->table->file->stats.records); imerge_trp->range_scans= range_scans; imerge_trp->range_scans_end= range_scans + n_child_scans; read_time= imerge_cost; } if (imerge_trp) { TABLE_READ_PLAN *trp= merge_same_index_scans(param, imerge, imerge_trp, read_time); if (trp != imerge_trp) DBUG_RETURN(trp); } } build_ror_index_merge: if (!all_scans_ror_able || param->thd->lex->sql_command == SQLCOM_DELETE || !optimizer_flag(param->thd, OPTIMIZER_SWITCH_INDEX_MERGE_UNION)) DBUG_RETURN(imerge_trp); /* Ok, it is possible to build a ROR-union, try it. */ bool dummy; if (!(roru_read_plans= (TABLE_READ_PLAN**)alloc_root(param->mem_root, sizeof(TABLE_READ_PLAN*)* n_child_scans))) DBUG_RETURN(imerge_trp); skip_to_ror_scan: roru_index_costs= 0.0; roru_total_records= 0; cur_roru_plan= roru_read_plans; /* Find 'best' ROR scan for each of trees in disjunction */ for (ptree= imerge->trees, cur_child= range_scans; ptree != imerge->trees_next; ptree++, cur_child++, cur_roru_plan++) { /* Assume the best ROR scan is the one that has cheapest full-row-retrieval scan cost. Also accumulate index_only scan costs as we'll need them to calculate overall index_intersection cost. */ double cost; if ((*cur_child)->is_ror) { /* Ok, we have index_only cost, now get full rows scan cost */ cost= param->table->file-> read_time(param->real_keynr[(*cur_child)->key_idx], 1, (*cur_child)->records) + rows2double((*cur_child)->records) / TIME_FOR_COMPARE; } else cost= read_time; TABLE_READ_PLAN *prev_plan= *cur_child; if (!(*cur_roru_plan= get_best_ror_intersect(param, *ptree, cost, &dummy))) { if (prev_plan->is_ror) *cur_roru_plan= prev_plan; else DBUG_RETURN(imerge_trp); roru_index_costs += (*cur_roru_plan)->read_cost; } else roru_index_costs += ((TRP_ROR_INTERSECT*)(*cur_roru_plan))->index_scan_costs; roru_total_records += (*cur_roru_plan)->records; roru_intersect_part *= (*cur_roru_plan)->records / param->table->file->stats.records; } /* rows to retrieve= SUM(rows_in_scan_i) - table_rows * PROD(rows_in_scan_i / table_rows). This is valid because index_merge construction guarantees that conditions in disjunction do not share key parts. */ roru_total_records -= (ha_rows)(roru_intersect_part* param->table->file->stats.records); /* ok, got a ROR read plan for each of the disjuncts Calculate cost: cost(index_union_scan(scan_1, ... scan_n)) = SUM_i(cost_of_index_only_scan(scan_i)) + queue_use_cost(rowid_len, n) + cost_of_row_retrieval See get_merge_buffers_cost function for queue_use_cost formula derivation. */ double roru_total_cost; roru_total_cost= roru_index_costs + rows2double(roru_total_records)*log((double)n_child_scans) / (TIME_FOR_COMPARE_ROWID * M_LN2) + get_sweep_read_cost(param, roru_total_records); DBUG_PRINT("info", ("ROR-union: cost %g, %d members", roru_total_cost, n_child_scans)); TRP_ROR_UNION* roru; if (roru_total_cost < read_time) { if ((roru= new (param->mem_root) TRP_ROR_UNION)) { roru->first_ror= roru_read_plans; roru->last_ror= roru_read_plans + n_child_scans; roru->read_cost= roru_total_cost; roru->records= roru_total_records; DBUG_RETURN(roru); } } DBUG_RETURN(imerge_trp); } /* Merge index scans for the same indexes in an index merge plan SYNOPSIS merge_same_index_scans() param Context info for the operation imerge IN/OUT SEL_IMERGE from which imerge_trp has been extracted imerge_trp The index merge plan where index scans for the same indexes are to be merges read_time The upper bound for the cost of the plan to be evaluated DESRIPTION For the given index merge plan imerge_trp extracted from the SEL_MERGE imerge the function looks for range scans with the same indexes and merges them into SEL_ARG trees. Then for each such SEL_ARG tree r_i the function creates a range tree rt_i that contains only r_i. All rt_i are joined into one index merge that replaces the original index merge imerge. The function calls get_best_disjunct_quick for the new index merge to get a new index merge plan that contains index scans only for different indexes. If there are no index scans for the same index in the original index merge plan the function does not change the original imerge and returns imerge_trp as its result. RETURN The original or or improved index merge plan */ static TABLE_READ_PLAN *merge_same_index_scans(PARAM *param, SEL_IMERGE *imerge, TRP_INDEX_MERGE *imerge_trp, double read_time) { uint16 first_scan_tree_idx[MAX_KEY]; SEL_TREE **tree; TRP_RANGE **cur_child; uint removed_cnt= 0; DBUG_ENTER("merge_same_index_scans"); bzero(first_scan_tree_idx, sizeof(first_scan_tree_idx[0])*param->keys); for (tree= imerge->trees, cur_child= imerge_trp->range_scans; tree != imerge->trees_next; tree++, cur_child++) { DBUG_ASSERT(tree); uint key_idx= (*cur_child)->key_idx; uint16 *tree_idx_ptr= &first_scan_tree_idx[key_idx]; if (!*tree_idx_ptr) *tree_idx_ptr= (uint16) (tree-imerge->trees+1); else { SEL_TREE **changed_tree= imerge->trees+(*tree_idx_ptr-1); SEL_ARG *key= (*changed_tree)->keys[key_idx]; bzero((*changed_tree)->keys, sizeof((*changed_tree)->keys[0])*param->keys); (*changed_tree)->keys_map.clear_all(); if (((*changed_tree)->keys[key_idx]= key_or(param, key, (*tree)->keys[key_idx]))) (*changed_tree)->keys_map.set_bit(key_idx); *tree= NULL; removed_cnt++; } } if (!removed_cnt) DBUG_RETURN(imerge_trp); TABLE_READ_PLAN *trp= NULL; SEL_TREE **new_trees_next= imerge->trees; for (tree= new_trees_next; tree != imerge->trees_next; tree++) { if (!*tree) continue; if (tree > new_trees_next) *new_trees_next= *tree; new_trees_next++; } imerge->trees_next= new_trees_next; DBUG_ASSERT(imerge->trees_next>imerge->trees); if (imerge->trees_next-imerge->trees > 1) trp= get_best_disjunct_quick(param, imerge, read_time); else { /* This alternative theoretically can be reached when the cost of the index merge for such a formula as (key1 BETWEEN c1_1 AND c1_2) AND key2 > c2 OR (key1 BETWEEN c1_3 AND c1_4) AND key3 > c3 is estimated as being cheaper than the cost of index scan for the formula (key1 BETWEEN c1_1 AND c1_2) OR (key1 BETWEEN c1_3 AND c1_4) In the current code this may happen for two reasons: 1. for a single index range scan data records are accessed in a random order 2. the functions that estimate the cost of a range scan and an index merge retrievals are not well calibrated */ trp= get_key_scans_params(param, *imerge->trees, FALSE, TRUE, read_time); } DBUG_RETURN(trp); } /* This structure contains the info common for all steps of a partial index intersection plan. Morever it contains also the info common for index intersect plans. This info is filled in by the function prepare_search_best just before searching for the best index intersection plan. */ typedef struct st_common_index_intersect_info { PARAM *param; /* context info for range optimizations */ uint key_size; /* size of a ROWID element stored in Unique object */ uint compare_factor; /* 1/compare - cost to compare two ROWIDs */ ulonglong max_memory_size; /* maximum space allowed for Unique objects */ ha_rows table_cardinality; /* estimate of the number of records in table */ double cutoff_cost; /* discard index intersects with greater costs */ INDEX_SCAN_INFO *cpk_scan; /* clustered primary key used in intersection */ bool in_memory; /* unique object for intersection is completely in memory */ INDEX_SCAN_INFO **search_scans; /* scans possibly included in intersect */ uint n_search_scans; /* number of elements in search_scans */ bool best_uses_cpk; /* current best intersect uses clustered primary key */ double best_cost; /* cost of the current best index intersection */ /* estimate of the number of records in the current best intersection */ ha_rows best_records; uint best_length; /* number of indexes in the current best intersection */ INDEX_SCAN_INFO **best_intersect; /* the current best index intersection */ /* scans from the best intersect to be filtrered by cpk conditions */ key_map filtered_scans; uint *buff_elems; /* buffer to calculate cost of index intersection */ } COMMON_INDEX_INTERSECT_INFO; /* This structure contains the info specific for one step of an index intersection plan. The structure is filled in by the function check_index_intersect_extension. */ typedef struct st_partial_index_intersect_info { COMMON_INDEX_INTERSECT_INFO *common_info; /* shared by index intersects */ uint length; /* number of index scans in the partial intersection */ ha_rows records; /* estimate of the number of records in intersection */ double cost; /* cost of the partial index intersection */ /* estimate of total number of records of all scans of the partial index intersect sent to the Unique object used for the intersection */ ha_rows records_sent_to_unique; /* total cost of the scans of indexes from the partial index intersection */ double index_read_cost; bool use_cpk_filter; /* cpk filter is to be used for this scan */ bool in_memory; /* uses unique object in memory */ double in_memory_cost; /* cost of using unique object in memory */ key_map filtered_scans; /* scans to be filtered by cpk conditions */ MY_BITMAP *intersect_fields; /* bitmap of fields used in intersection */ } PARTIAL_INDEX_INTERSECT_INFO; /* Check whether two indexes have the same first n components */ static bool same_index_prefix(KEY *key1, KEY *key2, uint used_parts) { KEY_PART_INFO *part1= key1->key_part; KEY_PART_INFO *part2= key2->key_part; for(uint i= 0; i < used_parts; i++, part1++, part2++) { if (part1->fieldnr != part2->fieldnr) return FALSE; } return TRUE; } /* Create a bitmap for all fields of a table */ static bool create_fields_bitmap(PARAM *param, MY_BITMAP *fields_bitmap) { my_bitmap_map *bitmap_buf; if (!(bitmap_buf= (my_bitmap_map *) alloc_root(param->mem_root, param->fields_bitmap_size))) return TRUE; if (bitmap_init(fields_bitmap, bitmap_buf, param->table->s->fields, FALSE)) return TRUE; return FALSE; } /* Compare two indexes scans for sort before search for the best intersection */ static int cmp_intersect_index_scan(INDEX_SCAN_INFO **a, INDEX_SCAN_INFO **b) { return (*a)->records < (*b)->records ? -1 : (*a)->records == (*b)->records ? 0 : 1; } static inline void set_field_bitmap_for_index_prefix(MY_BITMAP *field_bitmap, KEY_PART_INFO *key_part, uint used_key_parts) { bitmap_clear_all(field_bitmap); for (KEY_PART_INFO *key_part_end= key_part+used_key_parts; key_part < key_part_end; key_part++) { bitmap_set_bit(field_bitmap, key_part->fieldnr-1); } } /* Round up table cardinality read from statistics provided by engine. This function should go away when mysql test will allow to handle more or less easily in the test suites deviations of InnoDB statistical data. */ static inline ha_rows get_table_cardinality_for_index_intersect(TABLE *table) { if (table->file->ha_table_flags() & HA_STATS_RECORDS_IS_EXACT) return table->file->stats.records; else { ha_rows d; double q; for (q= (double)table->file->stats.records, d= 1 ; q >= 10; q/= 10, d*= 10 ) ; return (ha_rows) (floor(q+0.5) * d); } } static ha_rows records_in_index_intersect_extension(PARTIAL_INDEX_INTERSECT_INFO *curr, INDEX_SCAN_INFO *ext_index_scan); /* Prepare to search for the best index intersection SYNOPSIS prepare_search_best_index_intersect() param common info about index ranges tree tree of ranges for indexes than can be intersected common OUT info needed for search to be filled by the function init OUT info for an initial pseudo step of the intersection plans cutoff_cost cut off cost of the interesting index intersection DESCRIPTION The function initializes all fields of the structure 'common' to be used when searching for the best intersection plan. It also allocates memory to store the most cheap index intersection. NOTES When selecting candidates for index intersection we always take only one representative out of any set of indexes that share the same range conditions. These indexes always have the same prefixes and the components of this prefixes are exactly those used in these range conditions. Range conditions over clustered primary key (cpk) is always used only as the condition that filters out some rowids retrieved by the scans for secondary indexes. The cpk index will be handled in special way by the function that search for the best index intersection. RETURN FALSE in the case of success TRUE otherwise */ static bool prepare_search_best_index_intersect(PARAM *param, SEL_TREE *tree, COMMON_INDEX_INTERSECT_INFO *common, PARTIAL_INDEX_INTERSECT_INFO *init, double cutoff_cost) { uint i; uint n_search_scans; double cost; INDEX_SCAN_INFO **index_scan; INDEX_SCAN_INFO **scan_ptr; INDEX_SCAN_INFO *cpk_scan= NULL; TABLE *table= param->table; uint n_index_scans= tree->index_scans_end - tree->index_scans; if (!n_index_scans) return 1; bzero(init, sizeof(*init)); init->common_info= common; init->cost= cutoff_cost; common->param= param; common->key_size= table->file->ref_length; common->compare_factor= TIME_FOR_COMPARE_ROWID; common->max_memory_size= param->thd->variables.sortbuff_size; common->cutoff_cost= cutoff_cost; common->cpk_scan= NULL; common->table_cardinality= get_table_cardinality_for_index_intersect(table); if (n_index_scans <= 1) return TRUE; if (table->file->primary_key_is_clustered()) { INDEX_SCAN_INFO **index_scan_end; index_scan= tree->index_scans; index_scan_end= index_scan+n_index_scans; for ( ; index_scan < index_scan_end; index_scan++) { if ((*index_scan)->keynr == table->s->primary_key) { common->cpk_scan= cpk_scan= *index_scan; break; } } } i= n_index_scans - test(cpk_scan != NULL) + 1; if (!(common->search_scans = (INDEX_SCAN_INFO **) alloc_root (param->mem_root, sizeof(INDEX_SCAN_INFO *) * i))) return TRUE; bzero(common->search_scans, sizeof(INDEX_SCAN_INFO *) * i); INDEX_SCAN_INFO **selected_index_scans= common->search_scans; for (i=0, index_scan= tree->index_scans; i < n_index_scans; i++, index_scan++) { uint used_key_parts= (*index_scan)->used_key_parts; KEY *key_info= (*index_scan)->key_info; if (*index_scan == cpk_scan) continue; if (cpk_scan && cpk_scan->used_key_parts >= used_key_parts && same_index_prefix(cpk_scan->key_info, key_info, used_key_parts)) continue; cost= table->file->keyread_time((*index_scan)->keynr, (*index_scan)->range_count, (*index_scan)->records); if (cost >= cutoff_cost) continue; for (scan_ptr= selected_index_scans; *scan_ptr ; scan_ptr++) { /* When we have range conditions for two different indexes with the same beginning it does not make sense to consider both of them for index intersection if the range conditions are covered by common initial components of the indexes. Actually in this case the indexes are guaranteed to have the same range conditions. */ if ((*scan_ptr)->used_key_parts == used_key_parts && same_index_prefix((*scan_ptr)->key_info, key_info, used_key_parts)) break; } if (!*scan_ptr || cost < (*scan_ptr)->index_read_cost) { *scan_ptr= *index_scan; (*scan_ptr)->index_read_cost= cost; } } ha_rows records_in_scans= 0; for (scan_ptr=selected_index_scans, i= 0; *scan_ptr; scan_ptr++, i++) { if (create_fields_bitmap(param, &(*scan_ptr)->used_fields)) return TRUE; records_in_scans+= (*scan_ptr)->records; } n_search_scans= i; if (cpk_scan && create_fields_bitmap(param, &cpk_scan->used_fields)) return TRUE; if (!(common->n_search_scans= n_search_scans)) return TRUE; common->best_uses_cpk= FALSE; common->best_cost= cutoff_cost + COST_EPS; common->best_length= 0; if (!(common->best_intersect= (INDEX_SCAN_INFO **) alloc_root (param->mem_root, sizeof(INDEX_SCAN_INFO *) * (i + test(cpk_scan != NULL))))) return TRUE; size_t calc_cost_buff_size= Unique::get_cost_calc_buff_size((size_t)records_in_scans, common->key_size, common->max_memory_size); if (!(common->buff_elems= (uint *) alloc_root(param->mem_root, calc_cost_buff_size))) return TRUE; my_qsort(selected_index_scans, n_search_scans, sizeof(INDEX_SCAN_INFO *), (qsort_cmp) cmp_intersect_index_scan); if (cpk_scan) { PARTIAL_INDEX_INTERSECT_INFO curr; set_field_bitmap_for_index_prefix(&cpk_scan->used_fields, cpk_scan->key_info->key_part, cpk_scan->used_key_parts); curr.common_info= common; curr.intersect_fields= &cpk_scan->used_fields; curr.records= cpk_scan->records; curr.length= 1; for (scan_ptr=selected_index_scans; *scan_ptr; scan_ptr++) { ha_rows scan_records= (*scan_ptr)->records; ha_rows records= records_in_index_intersect_extension(&curr, *scan_ptr); (*scan_ptr)->filtered_out= records >= scan_records ? 0 : scan_records-records; } } else { for (scan_ptr=selected_index_scans; *scan_ptr; scan_ptr++) (*scan_ptr)->filtered_out= 0; } return FALSE; } /* On Estimation of the Number of Records in an Index Intersection =============================================================== Consider query Q over table t. Let C be the WHERE condition of this query, and, idx1(a1_1,...,a1_k1) and idx2(a2_1,...,a2_k2) be some indexes defined on table t. Let rt1 and rt2 be the range trees extracted by the range optimizer from C for idx1 and idx2 respectively. Let #t be the estimate of the number of records in table t provided for the optimizer. Let #r1 and #r2 be the estimates of the number of records in the range trees rt1 and rt2, respectively, obtained by the range optimizer. We need to get an estimate for the number of records in the index intersection of rt1 and rt2. In other words, we need to estimate the cardinality of the set of records that are in both trees. Let's designate this number by #r. If we do not make any assumptions then we can only state that #r<=min(#r1,#r2). With this estimate we can't say that the index intersection scan will be cheaper than the cheapest index scan. Let Rt1 and Rt2 be AND/OR conditions representing rt and rt2 respectively. The probability that a record belongs to rt1 is sel(Rt1)=#r1/#t. The probability that a record belongs to rt2 is sel(Rt2)=#r2/#t. If we assume that the values in columns of idx1 and idx2 are independent then #r/#t=sel(Rt1&Rt2)=sel(Rt1)*sel(Rt2)=(#r1/#t)*(#r2/#t). So in this case we have: #r=#r1*#r2/#t. The above assumption of independence of the columns in idx1 and idx2 means that: - all columns are different - values from one column do not correlate with values from any other column. We can't help with the case when column correlate with each other. Yet, if they are assumed to be uncorrelated the value of #r theoretically can be evaluated . Unfortunately this evaluation, in general, is rather complex. Let's consider two indexes idx1:(dept, manager), idx2:(dept, building) over table 'employee' and two range conditions over these indexes: Rt1: dept=10 AND manager LIKE 'S%' Rt2: dept=10 AND building LIKE 'L%'. We can state that: sel(Rt1&Rt2)=sel(dept=10)*sel(manager LIKE 'S%')*sel(building LIKE 'L%') =sel(Rt1)*sel(Rt2)/sel(dept=10). sel(Rt1/2_0:dept=10) can be estimated if we know the cardinality #r1_0 of the range for sub-index idx1_0 (dept) of the index idx1 or the cardinality #rt2_0 of the same range for sub-index idx2_0(dept) of the index idx2. The current code does not make an estimate either for #rt1_0, or for #rt2_0, but it can be adjusted to provide those numbers. Alternatively, min(rec_per_key) for (dept) could be used to get an upper bound for the value of sel(Rt1&Rt2). Yet this statistics is not provided now. Let's consider two other indexes idx1:(dept, last_name), idx2:(first_name, last_name) and two range conditions over these indexes: Rt1: dept=5 AND last_name='Sm%' Rt2: first_name='Robert' AND last_name='Sm%'. sel(Rt1&Rt2)=sel(dept=5)*sel(last_name='Sm5')*sel(first_name='Robert') =sel(Rt2)*sel(dept=5) Here max(rec_per_key) for (dept) could be used to get an upper bound for the value of sel(Rt1&Rt2). When the intersected indexes have different major columns, but some minor column are common the picture may be more complicated. Let's consider the following range conditions for the same indexes as in the previous example: Rt1: (Rt11: dept=5 AND last_name='So%') OR (Rt12: dept=7 AND last_name='Saw%') Rt2: (Rt21: first_name='Robert' AND last_name='Saw%') OR (Rt22: first_name='Bob' AND last_name='So%') Here we have: sel(Rt1&Rt2)= sel(Rt11)*sel(Rt21)+sel(Rt22)*sel(dept=5) + sel(Rt21)*sel(dept=7)+sel(Rt12)*sel(Rt22) Now consider the range condition: Rt1_0: (dept=5 OR dept=7) For this condition we can state that: sel(Rt1_0&Rt2)=(sel(dept=5)+sel(dept=7))*(sel(Rt21)+sel(Rt22))= sel(dept=5)*sel(Rt21)+sel(dept=7)*sel(Rt21)+ sel(dept=5)*sel(Rt22)+sel(dept=7)*sel(Rt22)= sel(dept=5)*sel(Rt21)+sel(Rt21)*sel(dept=7)+ sel(Rt22)*sel(dept=5)+sel(dept=7)*sel(Rt22) > sel(Rt11)*sel(Rt21)+sel(Rt22)*sel(dept=5)+ sel(Rt21)*sel(dept=7)+sel(Rt12)*sel(Rt22) > sel(Rt1 & Rt2) We've just demonstrated for an example what is intuitively almost obvious in general. We can remove the ending parts fromrange trees getting less selective range conditions for sub-indexes. So if not a most major component with the number k of an index idx is encountered in the index with which we intersect we can use the sub-index idx_k-1 that includes the components of idx up to the i-th component and the range tree for idx_k-1 to make an upper bound estimate for the number of records in the index intersection. The range tree for idx_k-1 we use here is the subtree of the original range tree for idx that contains only parts from the first k-1 components. As it was mentioned above the range optimizer currently does not provide an estimate for the number of records in the ranges for sub-indexes. However, some reasonable upper bound estimate can be obtained. Let's consider the following range tree: Rt: (first_name='Robert' AND last_name='Saw%') OR (first_name='Bob' AND last_name='So%') Let #r be the number of records in Rt. Let f_1 be the fan-out of column last_name: f_1 = rec_per_key[first_name]/rec_per_key[last_name]. The the number of records in the range tree: Rt_0: (first_name='Robert' OR first_name='Bob') for the sub-index (first_name) is not greater than max(#r*f_1, #t). Strictly speaking, we can state only that it's not greater than max(#r*max_f_1, #t), where max_f_1= max_rec_per_key[first_name]/min_rec_per_key[last_name]. Yet, if #r/#t is big enough (and this is the case of an index intersection, because using this index range with a single index scan is cheaper than the cost of the intersection when #r/#t is small) then almost safely we can use here f_1 instead of max_f_1. The above considerations can be used in future development. Now, they are used partly in the function that provides a rough upper bound estimate for the number of records in an index intersection that follow below. */ /* Estimate the number of records selected by an extension a partial intersection SYNOPSIS records_in_index_intersect_extension() curr partial intersection plan to be extended ext_index_scan the evaluated extension of this partial plan DESCRIPTION The function provides an estimate for the number of records in the intersection of the partial index intersection curr with the index ext_index_scan. If all intersected indexes does not have common columns then the function returns an exact estimate (assuming there are no correlations between values in the columns). If the intersected indexes have common columns the function returns an upper bound for the number of records in the intersection provided that the intersection of curr with ext_index_scan can is expected to have less records than the expected number of records in the partial intersection curr. In this case the function also assigns the bitmap of the columns in the extended intersection to ext_index_scan->used_fields. If the function cannot expect that the number of records in the extended intersection is less that the expected number of records #r in curr then the function returns a number bigger than #r. NOTES See the comment before the desription of the function that explains the reasoning used by this function. RETURN The expected number of rows in the extended index intersection */ static ha_rows records_in_index_intersect_extension(PARTIAL_INDEX_INTERSECT_INFO *curr, INDEX_SCAN_INFO *ext_index_scan) { KEY *key_info= ext_index_scan->key_info; KEY_PART_INFO* key_part= key_info->key_part; uint used_key_parts= ext_index_scan->used_key_parts; MY_BITMAP *used_fields= &ext_index_scan->used_fields; if (!curr->length) { /* If this the first index in the intersection just mark the fields in the used_fields bitmap and return the expected number of records in the range scan for the index provided by the range optimizer. */ set_field_bitmap_for_index_prefix(used_fields, key_part, used_key_parts); return ext_index_scan->records; } uint i; bool better_selectivity= FALSE; ha_rows records= curr->records; MY_BITMAP *curr_intersect_fields= curr->intersect_fields; for (i= 0; i < used_key_parts; i++, key_part++) { if (bitmap_is_set(curr_intersect_fields, key_part->fieldnr-1)) break; } if (i) { ha_rows table_cardinality= curr->common_info->table_cardinality; ha_rows ext_records= ext_index_scan->records; if (i < used_key_parts) { ulong *rec_per_key= key_info->rec_per_key+i-1; ulong f1= rec_per_key[0] ? rec_per_key[0] : 1; ulong f2= rec_per_key[1] ? rec_per_key[1] : 1; ext_records= (ha_rows) ((double) ext_records / f2 * f1); } if (ext_records < table_cardinality) { better_selectivity= TRUE; records= (ha_rows) ((double) records / table_cardinality * ext_records); bitmap_copy(used_fields, curr_intersect_fields); key_part= key_info->key_part; for (uint j= 0; j < used_key_parts; j++, key_part++) bitmap_set_bit(used_fields, key_part->fieldnr-1); } } return !better_selectivity ? records+1 : !records ? 1 : records; } /* Estimate the cost a binary search within disjoint cpk range intervals Number of comparisons to check whether a cpk value satisfies the cpk range condition = log2(cpk_scan->range_count). */ static inline double get_cpk_filter_cost(ha_rows filtered_records, INDEX_SCAN_INFO *cpk_scan, double compare_factor) { return log((double) (cpk_scan->range_count+1)) / (compare_factor * M_LN2) * filtered_records; } /* Check whether a patial index intersection plan can be extended SYNOPSIS check_index_intersect_extension() curr partial intersection plan to be extended ext_index_scan a possible extension of this plan to be checked next OUT the structure to be filled for the extended plan DESCRIPTION The function checks whether it makes sense to extend the index intersection plan adding the index ext_index_scan, and, if this the case, the function fills in the structure for the extended plan. RETURN TRUE if it makes sense to extend the given plan FALSE otherwise */ static bool check_index_intersect_extension(PARTIAL_INDEX_INTERSECT_INFO *curr, INDEX_SCAN_INFO *ext_index_scan, PARTIAL_INDEX_INTERSECT_INFO *next) { ha_rows records; ha_rows records_sent_to_unique; double cost; ha_rows ext_index_scan_records= ext_index_scan->records; ha_rows records_filtered_out_by_cpk= ext_index_scan->filtered_out; COMMON_INDEX_INTERSECT_INFO *common_info= curr->common_info; double cutoff_cost= common_info->cutoff_cost; uint idx= curr->length; next->index_read_cost= curr->index_read_cost+ext_index_scan->index_read_cost; if (next->index_read_cost > cutoff_cost) return FALSE; if ((next->in_memory= curr->in_memory)) next->in_memory_cost= curr->in_memory_cost; next->intersect_fields= &ext_index_scan->used_fields; next->filtered_scans= curr->filtered_scans; records_sent_to_unique= curr->records_sent_to_unique; next->use_cpk_filter= FALSE; /* Calculate the cost of using a Unique object for index intersection */ if (idx && next->in_memory) { /* All rowids received from the first scan are expected in one unique tree */ ha_rows elems_in_tree= common_info->search_scans[0]->records- common_info->search_scans[0]->filtered_out ; next->in_memory_cost+= Unique::get_search_cost(elems_in_tree, common_info->compare_factor)* ext_index_scan_records; cost= next->in_memory_cost; } else { uint *buff_elems= common_info->buff_elems; uint key_size= common_info->key_size; uint compare_factor= common_info->compare_factor; ulonglong max_memory_size= common_info->max_memory_size; records_sent_to_unique+= ext_index_scan_records; cost= Unique::get_use_cost(buff_elems, (size_t) records_sent_to_unique, key_size, max_memory_size, compare_factor, TRUE, &next->in_memory); if (records_filtered_out_by_cpk) { /* Check whether using cpk filter for this scan is beneficial */ double cost2; bool in_memory2; ha_rows records2= records_sent_to_unique-records_filtered_out_by_cpk; cost2= Unique::get_use_cost(buff_elems, (size_t) records2, key_size, max_memory_size, compare_factor, TRUE, &in_memory2); cost2+= get_cpk_filter_cost(ext_index_scan_records, common_info->cpk_scan, compare_factor); if (cost > cost2 + COST_EPS) { cost= cost2; next->in_memory= in_memory2; next->use_cpk_filter= TRUE; records_sent_to_unique= records2; } } if (next->in_memory) next->in_memory_cost= cost; } if (next->use_cpk_filter) { next->filtered_scans.set_bit(ext_index_scan->keynr); bitmap_union(&ext_index_scan->used_fields, &common_info->cpk_scan->used_fields); } next->records_sent_to_unique= records_sent_to_unique; records= records_in_index_intersect_extension(curr, ext_index_scan); if (idx && records > curr->records) return FALSE; if (next->use_cpk_filter && curr->filtered_scans.is_clear_all()) records-= records_filtered_out_by_cpk; next->records= records; cost+= next->index_read_cost; if (cost >= cutoff_cost) return FALSE; cost+= get_sweep_read_cost(common_info->param, records); next->cost= cost; next->length= curr->length+1; return TRUE; } /* Search for the cheapest extensions of range scans used to access a table SYNOPSIS find_index_intersect_best_extension() curr partial intersection to evaluate all possible extension for DESCRIPTION The function tries to extend the partial plan curr in all possible ways to look for a cheapest index intersection whose cost less than the cut off value set in curr->common_info.cutoff_cost. */ static void find_index_intersect_best_extension(PARTIAL_INDEX_INTERSECT_INFO *curr) { PARTIAL_INDEX_INTERSECT_INFO next; COMMON_INDEX_INTERSECT_INFO *common_info= curr->common_info; INDEX_SCAN_INFO **index_scans= common_info->search_scans; uint idx= curr->length; INDEX_SCAN_INFO **rem_first_index_scan_ptr= &index_scans[idx]; double cost= curr->cost; if (cost + COST_EPS < common_info->best_cost) { common_info->best_cost= cost; common_info->best_length= curr->length; common_info->best_records= curr->records; common_info->filtered_scans= curr->filtered_scans; /* common_info->best_uses_cpk <=> at least one scan uses a cpk filter */ common_info->best_uses_cpk= !curr->filtered_scans.is_clear_all(); uint sz= sizeof(INDEX_SCAN_INFO *) * curr->length; memcpy(common_info->best_intersect, common_info->search_scans, sz); common_info->cutoff_cost= cost; } if (!(*rem_first_index_scan_ptr)) return; next.common_info= common_info; INDEX_SCAN_INFO *rem_first_index_scan= *rem_first_index_scan_ptr; for (INDEX_SCAN_INFO **index_scan_ptr= rem_first_index_scan_ptr; *index_scan_ptr; index_scan_ptr++) { *rem_first_index_scan_ptr= *index_scan_ptr; *index_scan_ptr= rem_first_index_scan; if (check_index_intersect_extension(curr, *rem_first_index_scan_ptr, &next)) find_index_intersect_best_extension(&next); *index_scan_ptr= *rem_first_index_scan_ptr; *rem_first_index_scan_ptr= rem_first_index_scan; } } /* Get the plan of the best intersection of range scans used to access a table SYNOPSIS get_best_index_intersect() param common info about index ranges tree tree of ranges for indexes than can be intersected read_time cut off value for the evaluated plans DESCRIPTION The function looks for the cheapest index intersection of the range scans to access a table. The info about the ranges for all indexes is provided by the range optimizer and is passed through the parameters param and tree. Any plan whose cost is greater than read_time is rejected. After the best index intersection is found the function constructs the structure that manages the execution by the chosen plan. RETURN Pointer to the generated execution structure if a success, 0 - otherwise. */ static TRP_INDEX_INTERSECT *get_best_index_intersect(PARAM *param, SEL_TREE *tree, double read_time) { uint i; uint count; TRP_RANGE **cur_range; TRP_RANGE **range_scans; INDEX_SCAN_INFO *index_scan; COMMON_INDEX_INTERSECT_INFO common; PARTIAL_INDEX_INTERSECT_INFO init; TRP_INDEX_INTERSECT *intersect_trp= NULL; TABLE *table= param->table; DBUG_ENTER("get_best_index_intersect"); if (prepare_search_best_index_intersect(param, tree, &common, &init, read_time)) DBUG_RETURN(NULL); find_index_intersect_best_extension(&init); if (common.best_length <= 1 && !common.best_uses_cpk) DBUG_RETURN(NULL); if (common.best_uses_cpk) { memmove((char *) (common.best_intersect+1), (char *) common.best_intersect, sizeof(INDEX_SCAN_INFO *) * common.best_length); common.best_intersect[0]= common.cpk_scan; common.best_length++; } count= common.best_length; if (!(range_scans= (TRP_RANGE**)alloc_root(param->mem_root, sizeof(TRP_RANGE *)* count))) DBUG_RETURN(NULL); for (i= 0, cur_range= range_scans; i < count; i++) { index_scan= common.best_intersect[i]; if ((*cur_range= new (param->mem_root) TRP_RANGE(index_scan->sel_arg, index_scan->idx, 0))) { TRP_RANGE *trp= *cur_range; trp->read_cost= index_scan->index_read_cost; trp->records= index_scan->records; trp->is_ror= FALSE; trp->mrr_buf_size= 0; table->intersect_keys.set_bit(index_scan->keynr); cur_range++; } } count= tree->index_scans_end - tree->index_scans; for (i= 0; i < count; i++) { index_scan= tree->index_scans[i]; if (!table->intersect_keys.is_set(index_scan->keynr)) { for (uint j= 0; j < common.best_length; j++) { INDEX_SCAN_INFO *scan= common.best_intersect[j]; if (same_index_prefix(index_scan->key_info, scan->key_info, scan->used_key_parts)) { table->intersect_keys.set_bit(index_scan->keynr); break; } } } } if ((intersect_trp= new (param->mem_root)TRP_INDEX_INTERSECT)) { intersect_trp->read_cost= common.best_cost; intersect_trp->records= common.best_records; intersect_trp->range_scans= range_scans; intersect_trp->range_scans_end= cur_range; intersect_trp->filtered_scans= common.filtered_scans; } DBUG_RETURN(intersect_trp); } typedef struct st_ror_scan_info : INDEX_SCAN_INFO { } ROR_SCAN_INFO; /* Create ROR_SCAN_INFO* structure with a single ROR scan on index idx using sel_arg set of intervals. SYNOPSIS make_ror_scan() param Parameter from test_quick_select function idx Index of key in param->keys sel_arg Set of intervals for a given key RETURN NULL - out of memory ROR scan structure containing a scan for {idx, sel_arg} */ static ROR_SCAN_INFO *make_ror_scan(const PARAM *param, int idx, SEL_ARG *sel_arg) { ROR_SCAN_INFO *ror_scan; my_bitmap_map *bitmap_buf; uint keynr; DBUG_ENTER("make_ror_scan"); if (!(ror_scan= (ROR_SCAN_INFO*)alloc_root(param->mem_root, sizeof(ROR_SCAN_INFO)))) DBUG_RETURN(NULL); ror_scan->idx= idx; ror_scan->keynr= keynr= param->real_keynr[idx]; ror_scan->key_rec_length= (param->table->key_info[keynr].key_length + param->table->file->ref_length); ror_scan->sel_arg= sel_arg; ror_scan->records= param->quick_rows[keynr]; if (!(bitmap_buf= (my_bitmap_map*) alloc_root(param->mem_root, param->fields_bitmap_size))) DBUG_RETURN(NULL); if (bitmap_init(&ror_scan->covered_fields, bitmap_buf, param->table->s->fields, FALSE)) DBUG_RETURN(NULL); bitmap_clear_all(&ror_scan->covered_fields); KEY_PART_INFO *key_part= param->table->key_info[keynr].key_part; KEY_PART_INFO *key_part_end= key_part + param->table->key_info[keynr].key_parts; for (;key_part != key_part_end; ++key_part) { if (bitmap_is_set(¶m->needed_fields, key_part->fieldnr-1)) bitmap_set_bit(&ror_scan->covered_fields, key_part->fieldnr-1); } ror_scan->index_read_cost= param->table->file->keyread_time(ror_scan->keynr, 1, ror_scan->records); DBUG_RETURN(ror_scan); } /* Compare two ROR_SCAN_INFO** by E(#records_matched) * key_record_length. SYNOPSIS cmp_ror_scan_info() a ptr to first compared value b ptr to second compared value RETURN -1 a < b 0 a = b 1 a > b */ static int cmp_ror_scan_info(ROR_SCAN_INFO** a, ROR_SCAN_INFO** b) { double val1= rows2double((*a)->records) * (*a)->key_rec_length; double val2= rows2double((*b)->records) * (*b)->key_rec_length; return (val1 < val2)? -1: (val1 == val2)? 0 : 1; } /* Compare two ROR_SCAN_INFO** by (#covered fields in F desc, #components asc, number of first not covered component asc) SYNOPSIS cmp_ror_scan_info_covering() a ptr to first compared value b ptr to second compared value RETURN -1 a < b 0 a = b 1 a > b */ static int cmp_ror_scan_info_covering(ROR_SCAN_INFO** a, ROR_SCAN_INFO** b) { if ((*a)->used_fields_covered > (*b)->used_fields_covered) return -1; if ((*a)->used_fields_covered < (*b)->used_fields_covered) return 1; if ((*a)->key_components < (*b)->key_components) return -1; if ((*a)->key_components > (*b)->key_components) return 1; if ((*a)->first_uncovered_field < (*b)->first_uncovered_field) return -1; if ((*a)->first_uncovered_field > (*b)->first_uncovered_field) return 1; return 0; } /* Auxiliary structure for incremental ROR-intersection creation */ typedef struct { const PARAM *param; MY_BITMAP covered_fields; /* union of fields covered by all scans */ /* Fraction of table records that satisfies conditions of all scans. This is the number of full records that will be retrieved if a non-index_only index intersection will be employed. */ double out_rows; /* TRUE if covered_fields is a superset of needed_fields */ bool is_covering; ha_rows index_records; /* sum(#records to look in indexes) */ double index_scan_costs; /* SUM(cost of 'index-only' scans) */ double total_cost; } ROR_INTERSECT_INFO; /* Allocate a ROR_INTERSECT_INFO and initialize it to contain zero scans. SYNOPSIS ror_intersect_init() param Parameter from test_quick_select RETURN allocated structure NULL on error */ static ROR_INTERSECT_INFO* ror_intersect_init(const PARAM *param) { ROR_INTERSECT_INFO *info; my_bitmap_map* buf; if (!(info= (ROR_INTERSECT_INFO*)alloc_root(param->mem_root, sizeof(ROR_INTERSECT_INFO)))) return NULL; info->param= param; if (!(buf= (my_bitmap_map*) alloc_root(param->mem_root, param->fields_bitmap_size))) return NULL; if (bitmap_init(&info->covered_fields, buf, param->table->s->fields, FALSE)) return NULL; info->is_covering= FALSE; info->index_scan_costs= 0.0; info->index_records= 0; info->out_rows= (double) param->table->file->stats.records; bitmap_clear_all(&info->covered_fields); return info; } void ror_intersect_cpy(ROR_INTERSECT_INFO *dst, const ROR_INTERSECT_INFO *src) { dst->param= src->param; memcpy(dst->covered_fields.bitmap, src->covered_fields.bitmap, no_bytes_in_map(&src->covered_fields)); dst->out_rows= src->out_rows; dst->is_covering= src->is_covering; dst->index_records= src->index_records; dst->index_scan_costs= src->index_scan_costs; dst->total_cost= src->total_cost; } /* Get selectivity of a ROR scan wrt ROR-intersection. SYNOPSIS ror_scan_selectivity() info ROR-interection scan ROR scan NOTES Suppose we have a condition on several keys cond=k_11=c_11 AND k_12=c_12 AND ... // parts of first key k_21=c_21 AND k_22=c_22 AND ... // parts of second key ... k_n1=c_n1 AND k_n3=c_n3 AND ... (1) //parts of the key used by *scan where k_ij may be the same as any k_pq (i.e. keys may have common parts). A full row is retrieved if entire condition holds. The recursive procedure for finding P(cond) is as follows: First step: Pick 1st part of 1st key and break conjunction (1) into two parts: cond= (k_11=c_11 AND R) Here R may still contain condition(s) equivalent to k_11=c_11. Nevertheless, the following holds: P(k_11=c_11 AND R) = P(k_11=c_11) * P(R | k_11=c_11). Mark k_11 as fixed field (and satisfied condition) F, save P(F), save R to be cond and proceed to recursion step. Recursion step: We have a set of fixed fields/satisfied conditions) F, probability P(F), and remaining conjunction R Pick next key part on current key and its condition "k_ij=c_ij". We will add "k_ij=c_ij" into F and update P(F). Lets denote k_ij as t, R = t AND R1, where R1 may still contain t. Then P((t AND R1)|F) = P(t|F) * P(R1|t|F) = P(t|F) * P(R1|(t AND F)) (2) (where '|' mean conditional probability, not "or") Consider the first multiplier in (2). One of the following holds: a) F contains condition on field used in t (i.e. t AND F = F). Then P(t|F) = 1 b) F doesn't contain condition on field used in t. Then F and t are considered independent. P(t|F) = P(t|(fields_before_t_in_key AND other_fields)) = = P(t|fields_before_t_in_key). P(t|fields_before_t_in_key) = #records(fields_before_t_in_key) / #records(fields_before_t_in_key, t) The second multiplier is calculated by applying this step recursively. IMPLEMENTATION This function calculates the result of application of the "recursion step" described above for all fixed key members of a single key, accumulating set of covered fields, selectivity, etc. The calculation is conducted as follows: Lets denote #records(keypart1, ... keypartK) as n_k. We need to calculate n_{k1} n_{k2} --------- * --------- * .... (3) n_{k1-1} n_{k2-1} where k1,k2,... are key parts which fields were not yet marked as fixed ( this is result of application of option b) of the recursion step for parts of a single key). Since it is reasonable to expect that most of the fields are not marked as fixed, we calculate (3) as n_{i1} n_{i2} (3) = n_{max_key_part} / ( --------- * --------- * .... ) n_{i1-1} n_{i2-1} where i1,i2, .. are key parts that were already marked as fixed. In order to minimize number of expensive records_in_range calls we group and reduce adjacent fractions. RETURN Selectivity of given ROR scan. */ static double ror_scan_selectivity(const ROR_INTERSECT_INFO *info, const ROR_SCAN_INFO *scan) { double selectivity_mult= 1.0; KEY_PART_INFO *key_part= info->param->table->key_info[scan->keynr].key_part; uchar key_val[MAX_KEY_LENGTH+MAX_FIELD_WIDTH]; /* key values tuple */ uchar *key_ptr= key_val; SEL_ARG *sel_arg, *tuple_arg= NULL; key_part_map keypart_map= 0; bool cur_covered; bool prev_covered= test(bitmap_is_set(&info->covered_fields, key_part->fieldnr-1)); key_range min_range; key_range max_range; min_range.key= key_val; min_range.flag= HA_READ_KEY_EXACT; max_range.key= key_val; max_range.flag= HA_READ_AFTER_KEY; ha_rows prev_records= info->param->table->file->stats.records; DBUG_ENTER("ror_scan_selectivity"); for (sel_arg= scan->sel_arg; sel_arg; sel_arg= sel_arg->next_key_part) { DBUG_PRINT("info",("sel_arg step")); cur_covered= test(bitmap_is_set(&info->covered_fields, key_part[sel_arg->part].fieldnr-1)); if (cur_covered != prev_covered) { /* create (part1val, ..., part{n-1}val) tuple. */ ha_rows records; if (!tuple_arg) { tuple_arg= scan->sel_arg; /* Here we use the length of the first key part */ tuple_arg->store_min(key_part->store_length, &key_ptr, 0); keypart_map= 1; } while (tuple_arg->next_key_part != sel_arg) { tuple_arg= tuple_arg->next_key_part; tuple_arg->store_min(key_part[tuple_arg->part].store_length, &key_ptr, 0); keypart_map= (keypart_map << 1) | 1; } min_range.length= max_range.length= (size_t) (key_ptr - key_val); min_range.keypart_map= max_range.keypart_map= keypart_map; records= (info->param->table->file-> records_in_range(scan->keynr, &min_range, &max_range)); if (cur_covered) { /* uncovered -> covered */ double tmp= rows2double(records)/rows2double(prev_records); DBUG_PRINT("info", ("Selectivity multiplier: %g", tmp)); selectivity_mult *= tmp; prev_records= HA_POS_ERROR; } else { /* covered -> uncovered */ prev_records= records; } } prev_covered= cur_covered; } if (!prev_covered) { double tmp= rows2double(info->param->quick_rows[scan->keynr]) / rows2double(prev_records); DBUG_PRINT("info", ("Selectivity multiplier: %g", tmp)); selectivity_mult *= tmp; } DBUG_PRINT("info", ("Returning multiplier: %g", selectivity_mult)); DBUG_RETURN(selectivity_mult); } /* Check if adding a ROR scan to a ROR-intersection reduces its cost of ROR-intersection and if yes, update parameters of ROR-intersection, including its cost. SYNOPSIS ror_intersect_add() param Parameter from test_quick_select info ROR-intersection structure to add the scan to. ror_scan ROR scan info to add. is_cpk_scan If TRUE, add the scan as CPK scan (this can be inferred from other parameters and is passed separately only to avoid duplicating the inference code) NOTES Adding a ROR scan to ROR-intersect "makes sense" iff the cost of ROR- intersection decreases. The cost of ROR-intersection is calculated as follows: cost= SUM_i(key_scan_cost_i) + cost_of_full_rows_retrieval When we add a scan the first increases and the second decreases. cost_of_full_rows_retrieval= (union of indexes used covers all needed fields) ? cost_of_sweep_read(E(rows_to_retrieve), rows_in_table) : 0 E(rows_to_retrieve) = #rows_in_table * ror_scan_selectivity(null, scan1) * ror_scan_selectivity({scan1}, scan2) * ... * ror_scan_selectivity({scan1,...}, scanN). RETURN TRUE ROR scan added to ROR-intersection, cost updated. FALSE It doesn't make sense to add this ROR scan to this ROR-intersection. */ static bool ror_intersect_add(ROR_INTERSECT_INFO *info, ROR_SCAN_INFO* ror_scan, bool is_cpk_scan) { double selectivity_mult= 1.0; DBUG_ENTER("ror_intersect_add"); DBUG_PRINT("info", ("Current out_rows= %g", info->out_rows)); DBUG_PRINT("info", ("Adding scan on %s", info->param->table->key_info[ror_scan->keynr].name)); DBUG_PRINT("info", ("is_cpk_scan: %d",is_cpk_scan)); selectivity_mult = ror_scan_selectivity(info, ror_scan); if (selectivity_mult == 1.0) { /* Don't add this scan if it doesn't improve selectivity. */ DBUG_PRINT("info", ("The scan doesn't improve selectivity.")); DBUG_RETURN(FALSE); } info->out_rows *= selectivity_mult; if (is_cpk_scan) { /* CPK scan is used to filter out rows. We apply filtering for each record of every scan. Assuming 1/TIME_FOR_COMPARE_ROWID per check this gives us: */ info->index_scan_costs += rows2double(info->index_records) / TIME_FOR_COMPARE_ROWID; } else { info->index_records += info->param->quick_rows[ror_scan->keynr]; info->index_scan_costs += ror_scan->index_read_cost; bitmap_union(&info->covered_fields, &ror_scan->covered_fields); if (!info->is_covering && bitmap_is_subset(&info->param->needed_fields, &info->covered_fields)) { DBUG_PRINT("info", ("ROR-intersect is covering now")); info->is_covering= TRUE; } } info->total_cost= info->index_scan_costs; DBUG_PRINT("info", ("info->total_cost: %g", info->total_cost)); if (!info->is_covering) { info->total_cost += get_sweep_read_cost(info->param, double2rows(info->out_rows)); DBUG_PRINT("info", ("info->total_cost= %g", info->total_cost)); } DBUG_PRINT("info", ("New out_rows: %g", info->out_rows)); DBUG_PRINT("info", ("New cost: %g, %scovering", info->total_cost, info->is_covering?"" : "non-")); DBUG_RETURN(TRUE); } /* Get best ROR-intersection plan using non-covering ROR-intersection search algorithm. The returned plan may be covering. SYNOPSIS get_best_ror_intersect() param Parameter from test_quick_select function. tree Transformed restriction condition to be used to look for ROR scans. read_time Do not return read plans with cost > read_time. are_all_covering [out] set to TRUE if union of all scans covers all fields needed by the query (and it is possible to build a covering ROR-intersection) NOTES get_key_scans_params must be called before this function can be called. When this function is called by ROR-union construction algorithm it assumes it is building an uncovered ROR-intersection (and thus # of full records to be retrieved is wrong here). This is a hack. IMPLEMENTATION The approximate best non-covering plan search algorithm is as follows: find_min_ror_intersection_scan() { R= select all ROR scans; order R by (E(#records_matched) * key_record_length). S= first(R); -- set of scans that will be used for ROR-intersection R= R-first(S); min_cost= cost(S); min_scan= make_scan(S); while (R is not empty) { firstR= R - first(R); if (!selectivity(S + firstR < selectivity(S))) continue; S= S + first(R); if (cost(S) < min_cost) { min_cost= cost(S); min_scan= make_scan(S); } } return min_scan; } See ror_intersect_add function for ROR intersection costs. Special handling for Clustered PK scans Clustered PK contains all table fields, so using it as a regular scan in index intersection doesn't make sense: a range scan on CPK will be less expensive in this case. Clustered PK scan has special handling in ROR-intersection: it is not used to retrieve rows, instead its condition is used to filter row references we get from scans on other keys. RETURN ROR-intersection table read plan NULL if out of memory or no suitable plan found. */ static TRP_ROR_INTERSECT *get_best_ror_intersect(const PARAM *param, SEL_TREE *tree, double read_time, bool *are_all_covering) { uint idx; double min_cost= DBL_MAX; DBUG_ENTER("get_best_ror_intersect"); if ((tree->n_ror_scans < 2) || !param->table->file->stats.records || !optimizer_flag(param->thd, OPTIMIZER_SWITCH_INDEX_MERGE_INTERSECT)) DBUG_RETURN(NULL); /* Step1: Collect ROR-able SEL_ARGs and create ROR_SCAN_INFO for each of them. Also find and save clustered PK scan if there is one. */ ROR_SCAN_INFO **cur_ror_scan; ROR_SCAN_INFO *cpk_scan= NULL; uint cpk_no; if (!(tree->ror_scans= (ROR_SCAN_INFO**)alloc_root(param->mem_root, sizeof(ROR_SCAN_INFO*)* param->keys))) return NULL; cpk_no= ((param->table->file->primary_key_is_clustered()) ? param->table->s->primary_key : MAX_KEY); for (idx= 0, cur_ror_scan= tree->ror_scans; idx < param->keys; idx++) { ROR_SCAN_INFO *scan; uint key_no; if (!tree->ror_scans_map.is_set(idx)) continue; key_no= param->real_keynr[idx]; if (key_no != cpk_no && param->table->file->index_flags(key_no,0,0) & HA_CLUSTERED_INDEX) { /* Ignore clustering keys */ tree->n_ror_scans--; continue; } if (!(scan= make_ror_scan(param, idx, tree->keys[idx]))) return NULL; if (key_no == cpk_no) { cpk_scan= scan; tree->n_ror_scans--; } else *(cur_ror_scan++)= scan; } tree->ror_scans_end= cur_ror_scan; DBUG_EXECUTE("info",print_ror_scans_arr(param->table, "original", tree->ror_scans, tree->ror_scans_end);); /* Ok, [ror_scans, ror_scans_end) is array of ptrs to initialized ROR_SCAN_INFO's. Step 2: Get best ROR-intersection using an approximate algorithm. */ my_qsort(tree->ror_scans, tree->n_ror_scans, sizeof(ROR_SCAN_INFO*), (qsort_cmp)cmp_ror_scan_info); DBUG_EXECUTE("info",print_ror_scans_arr(param->table, "ordered", tree->ror_scans, tree->ror_scans_end);); ROR_SCAN_INFO **intersect_scans; /* ROR scans used in index intersection */ ROR_SCAN_INFO **intersect_scans_end; if (!(intersect_scans= (ROR_SCAN_INFO**)alloc_root(param->mem_root, sizeof(ROR_SCAN_INFO*)* tree->n_ror_scans))) return NULL; intersect_scans_end= intersect_scans; /* Create and incrementally update ROR intersection. */ ROR_INTERSECT_INFO *intersect, *intersect_best; if (!(intersect= ror_intersect_init(param)) || !(intersect_best= ror_intersect_init(param))) return NULL; /* [intersect_scans,intersect_scans_best) will hold the best intersection */ ROR_SCAN_INFO **intersect_scans_best; cur_ror_scan= tree->ror_scans; intersect_scans_best= intersect_scans; while (cur_ror_scan != tree->ror_scans_end && !intersect->is_covering) { /* S= S + first(R); R= R - first(R); */ if (!ror_intersect_add(intersect, *cur_ror_scan, FALSE)) { cur_ror_scan++; continue; } *(intersect_scans_end++)= *(cur_ror_scan++); if (intersect->total_cost < min_cost) { /* Local minimum found, save it */ ror_intersect_cpy(intersect_best, intersect); intersect_scans_best= intersect_scans_end; min_cost = intersect->total_cost; } } if (intersect_scans_best == intersect_scans) { DBUG_PRINT("info", ("None of scans increase selectivity")); DBUG_RETURN(NULL); } DBUG_EXECUTE("info",print_ror_scans_arr(param->table, "best ROR-intersection", intersect_scans, intersect_scans_best);); *are_all_covering= intersect->is_covering; uint best_num= intersect_scans_best - intersect_scans; ror_intersect_cpy(intersect, intersect_best); /* Ok, found the best ROR-intersection of non-CPK key scans. Check if we should add a CPK scan. If the obtained ROR-intersection is covering, it doesn't make sense to add CPK scan. */ if (cpk_scan && !intersect->is_covering) { if (ror_intersect_add(intersect, cpk_scan, TRUE) && (intersect->total_cost < min_cost)) intersect_best= intersect; //just set pointer here } else cpk_scan= 0; // Don't use cpk_scan /* Ok, return ROR-intersect plan if we have found one */ TRP_ROR_INTERSECT *trp= NULL; if (min_cost < read_time && (cpk_scan || best_num > 1)) { if (!(trp= new (param->mem_root) TRP_ROR_INTERSECT)) DBUG_RETURN(trp); if (!(trp->first_scan= (ROR_SCAN_INFO**)alloc_root(param->mem_root, sizeof(ROR_SCAN_INFO*)*best_num))) DBUG_RETURN(NULL); memcpy(trp->first_scan, intersect_scans, best_num*sizeof(ROR_SCAN_INFO*)); trp->last_scan= trp->first_scan + best_num; trp->is_covering= intersect_best->is_covering; trp->read_cost= intersect_best->total_cost; /* Prevent divisons by zero */ ha_rows best_rows = double2rows(intersect_best->out_rows); if (!best_rows) best_rows= 1; set_if_smaller(param->table->quick_condition_rows, best_rows); trp->records= best_rows; trp->index_scan_costs= intersect_best->index_scan_costs; trp->cpk_scan= cpk_scan; DBUG_PRINT("info", ("Returning non-covering ROR-intersect plan:" "cost %g, records %lu", trp->read_cost, (ulong) trp->records)); } DBUG_RETURN(trp); } /* Get best covering ROR-intersection. SYNOPSIS get_best_ntersectcovering_ror_intersect() param Parameter from test_quick_select function. tree SEL_TREE with sets of intervals for different keys. read_time Don't return table read plans with cost > read_time. RETURN Best covering ROR-intersection plan NULL if no plan found. NOTES get_best_ror_intersect must be called for a tree before calling this function for it. This function invalidates tree->ror_scans member values. The following approximate algorithm is used: I=set of all covering indexes F=set of all fields to cover S={} do { Order I by (#covered fields in F desc, #components asc, number of first not covered component asc); F=F-covered by first(I); S=S+first(I); I=I-first(I); } while F is not empty. */ static TRP_ROR_INTERSECT *get_best_covering_ror_intersect(PARAM *param, SEL_TREE *tree, double read_time) { ROR_SCAN_INFO **ror_scan_mark; ROR_SCAN_INFO **ror_scans_end= tree->ror_scans_end; DBUG_ENTER("get_best_covering_ror_intersect"); if (!optimizer_flag(param->thd, OPTIMIZER_SWITCH_INDEX_MERGE_INTERSECT)) DBUG_RETURN(NULL); for (ROR_SCAN_INFO **scan= tree->ror_scans; scan != ror_scans_end; ++scan) (*scan)->key_components= param->table->key_info[(*scan)->keynr].key_parts; /* Run covering-ROR-search algorithm. Assume set I is [ror_scan .. ror_scans_end) */ /*I=set of all covering indexes */ ror_scan_mark= tree->ror_scans; MY_BITMAP *covered_fields= ¶m->tmp_covered_fields; if (!covered_fields->bitmap) covered_fields->bitmap= (my_bitmap_map*)alloc_root(param->mem_root, param->fields_bitmap_size); if (!covered_fields->bitmap || bitmap_init(covered_fields, covered_fields->bitmap, param->table->s->fields, FALSE)) DBUG_RETURN(0); bitmap_clear_all(covered_fields); double total_cost= 0.0f; ha_rows records=0; bool all_covered; DBUG_PRINT("info", ("Building covering ROR-intersection")); DBUG_EXECUTE("info", print_ror_scans_arr(param->table, "building covering ROR-I", ror_scan_mark, ror_scans_end);); do { /* Update changed sorting info: #covered fields, number of first not covered component Calculate and save these values for each of remaining scans. */ for (ROR_SCAN_INFO **scan= ror_scan_mark; scan != ror_scans_end; ++scan) { bitmap_subtract(&(*scan)->covered_fields, covered_fields); (*scan)->used_fields_covered= bitmap_bits_set(&(*scan)->covered_fields); (*scan)->first_uncovered_field= bitmap_get_first(&(*scan)->covered_fields); } my_qsort(ror_scan_mark, ror_scans_end-ror_scan_mark, sizeof(ROR_SCAN_INFO*), (qsort_cmp)cmp_ror_scan_info_covering); DBUG_EXECUTE("info", print_ror_scans_arr(param->table, "remaining scans", ror_scan_mark, ror_scans_end);); /* I=I-first(I) */ total_cost += (*ror_scan_mark)->index_read_cost; records += (*ror_scan_mark)->records; DBUG_PRINT("info", ("Adding scan on %s", param->table->key_info[(*ror_scan_mark)->keynr].name)); if (total_cost > read_time) DBUG_RETURN(NULL); /* F=F-covered by first(I) */ bitmap_union(covered_fields, &(*ror_scan_mark)->covered_fields); all_covered= bitmap_is_subset(¶m->needed_fields, covered_fields); } while ((++ror_scan_mark < ror_scans_end) && !all_covered); if (!all_covered || (ror_scan_mark - tree->ror_scans) == 1) DBUG_RETURN(NULL); /* Ok, [tree->ror_scans .. ror_scan) holds covering index_intersection with cost total_cost. */ DBUG_PRINT("info", ("Covering ROR-intersect scans cost: %g", total_cost)); DBUG_EXECUTE("info", print_ror_scans_arr(param->table, "creating covering ROR-intersect", tree->ror_scans, ror_scan_mark);); /* Add priority queue use cost. */ total_cost += rows2double(records)* log((double)(ror_scan_mark - tree->ror_scans)) / (TIME_FOR_COMPARE_ROWID * M_LN2); DBUG_PRINT("info", ("Covering ROR-intersect full cost: %g", total_cost)); if (total_cost > read_time) DBUG_RETURN(NULL); TRP_ROR_INTERSECT *trp; if (!(trp= new (param->mem_root) TRP_ROR_INTERSECT)) DBUG_RETURN(trp); uint best_num= (ror_scan_mark - tree->ror_scans); if (!(trp->first_scan= (ROR_SCAN_INFO**)alloc_root(param->mem_root, sizeof(ROR_SCAN_INFO*)* best_num))) DBUG_RETURN(NULL); memcpy(trp->first_scan, tree->ror_scans, best_num*sizeof(ROR_SCAN_INFO*)); trp->last_scan= trp->first_scan + best_num; trp->is_covering= TRUE; trp->read_cost= total_cost; trp->records= records; trp->cpk_scan= NULL; set_if_smaller(param->table->quick_condition_rows, records); DBUG_PRINT("info", ("Returning covering ROR-intersect plan: cost %g, records %lu", trp->read_cost, (ulong) trp->records)); DBUG_RETURN(trp); } /* Get best "range" table read plan for given SEL_TREE. Also update PARAM members and store ROR scans info in the SEL_TREE. SYNOPSIS get_key_scans_params param parameters from test_quick_select tree make range select for this SEL_TREE index_read_must_be_used if TRUE, assume 'index only' option will be set (except for clustered PK indexes) read_time don't create read plans with cost > read_time. RETURN Best range read plan NULL if no plan found or error occurred */ static TRP_RANGE *get_key_scans_params(PARAM *param, SEL_TREE *tree, bool index_read_must_be_used, bool update_tbl_stats, double read_time) { uint idx; SEL_ARG **key,**end, **key_to_read= NULL; ha_rows UNINIT_VAR(best_records); /* protected by key_to_read */ uint UNINIT_VAR(best_mrr_flags), /* protected by key_to_read */ UNINIT_VAR(best_buf_size); /* protected by key_to_read */ TRP_RANGE* read_plan= NULL; DBUG_ENTER("get_key_scans_params"); /* Note that there may be trees that have type SEL_TREE::KEY but contain no key reads at all, e.g. tree for expression "key1 is not null" where key1 is defined as "not null". */ DBUG_EXECUTE("info", print_sel_tree(param, tree, &tree->keys_map, "tree scans");); tree->ror_scans_map.clear_all(); tree->n_ror_scans= 0; tree->index_scans= 0; if (!tree->keys_map.is_clear_all()) { tree->index_scans= (INDEX_SCAN_INFO **) alloc_root(param->mem_root, sizeof(INDEX_SCAN_INFO *) * param->keys); } tree->index_scans_end= tree->index_scans; for (idx= 0,key=tree->keys, end=key+param->keys; key != end; key++,idx++) { if (*key) { ha_rows found_records; COST_VECT cost; double found_read_time; uint mrr_flags, buf_size; INDEX_SCAN_INFO *index_scan; uint keynr= param->real_keynr[idx]; if ((*key)->type == SEL_ARG::MAYBE_KEY || (*key)->maybe_flag) param->needed_reg->set_bit(keynr); bool read_index_only= index_read_must_be_used ? TRUE : (bool) param->table->covering_keys.is_set(keynr); found_records= check_quick_select(param, idx, read_index_only, *key, update_tbl_stats, &mrr_flags, &buf_size, &cost); if (found_records != HA_POS_ERROR && tree->index_scans && (index_scan= (INDEX_SCAN_INFO *)alloc_root(param->mem_root, sizeof(INDEX_SCAN_INFO)))) { index_scan->idx= idx; index_scan->keynr= keynr; index_scan->key_info= ¶m->table->key_info[keynr]; index_scan->used_key_parts= param->max_key_part+1; index_scan->range_count= param->range_count; index_scan->records= found_records; index_scan->sel_arg= *key; *tree->index_scans_end++= index_scan; } if ((found_records != HA_POS_ERROR) && param->is_ror_scan) { tree->n_ror_scans++; tree->ror_scans_map.set_bit(idx); } if (found_records != HA_POS_ERROR && read_time > (found_read_time= cost.total_cost())) { read_time= found_read_time; best_records= found_records; key_to_read= key; best_mrr_flags= mrr_flags; best_buf_size= buf_size; } } } DBUG_EXECUTE("info", print_sel_tree(param, tree, &tree->ror_scans_map, "ROR scans");); if (key_to_read) { idx= key_to_read - tree->keys; if ((read_plan= new (param->mem_root) TRP_RANGE(*key_to_read, idx, best_mrr_flags))) { read_plan->records= best_records; read_plan->is_ror= tree->ror_scans_map.is_set(idx); read_plan->read_cost= read_time; read_plan->mrr_buf_size= best_buf_size; DBUG_PRINT("info", ("Returning range plan for key %s, cost %g, records %lu", param->table->key_info[param->real_keynr[idx]].name, read_plan->read_cost, (ulong) read_plan->records)); } } else DBUG_PRINT("info", ("No 'range' table read plan found")); DBUG_RETURN(read_plan); } QUICK_SELECT_I *TRP_INDEX_MERGE::make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc) { QUICK_INDEX_MERGE_SELECT *quick_imerge; QUICK_RANGE_SELECT *quick; /* index_merge always retrieves full rows, ignore retrieve_full_rows */ if (!(quick_imerge= new QUICK_INDEX_MERGE_SELECT(param->thd, param->table))) return NULL; quick_imerge->records= records; quick_imerge->read_time= read_cost; for (TRP_RANGE **range_scan= range_scans; range_scan != range_scans_end; range_scan++) { if (!(quick= (QUICK_RANGE_SELECT*) ((*range_scan)->make_quick(param, FALSE, &quick_imerge->alloc)))|| quick_imerge->push_quick_back(quick)) { delete quick; delete quick_imerge; return NULL; } } return quick_imerge; } QUICK_SELECT_I *TRP_INDEX_INTERSECT::make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc) { QUICK_INDEX_INTERSECT_SELECT *quick_intersect; QUICK_RANGE_SELECT *quick; /* index_merge always retrieves full rows, ignore retrieve_full_rows */ if (!(quick_intersect= new QUICK_INDEX_INTERSECT_SELECT(param->thd, param->table))) return NULL; quick_intersect->records= records; quick_intersect->read_time= read_cost; quick_intersect->filtered_scans= filtered_scans; for (TRP_RANGE **range_scan= range_scans; range_scan != range_scans_end; range_scan++) { if (!(quick= (QUICK_RANGE_SELECT*) ((*range_scan)->make_quick(param, FALSE, &quick_intersect->alloc)))|| quick_intersect->push_quick_back(quick)) { delete quick; delete quick_intersect; return NULL; } } return quick_intersect; } QUICK_SELECT_I *TRP_ROR_INTERSECT::make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc) { QUICK_ROR_INTERSECT_SELECT *quick_intrsect; QUICK_RANGE_SELECT *quick; DBUG_ENTER("TRP_ROR_INTERSECT::make_quick"); MEM_ROOT *alloc; if ((quick_intrsect= new QUICK_ROR_INTERSECT_SELECT(param->thd, param->table, (retrieve_full_rows? (!is_covering) : FALSE), parent_alloc))) { DBUG_EXECUTE("info", print_ror_scans_arr(param->table, "creating ROR-intersect", first_scan, last_scan);); alloc= parent_alloc? parent_alloc: &quick_intrsect->alloc; for (; first_scan != last_scan;++first_scan) { if (!(quick= get_quick_select(param, (*first_scan)->idx, (*first_scan)->sel_arg, HA_MRR_USE_DEFAULT_IMPL | HA_MRR_SORTED, 0, alloc)) || quick_intrsect->push_quick_back(alloc, quick)) { delete quick_intrsect; DBUG_RETURN(NULL); } } if (cpk_scan) { if (!(quick= get_quick_select(param, cpk_scan->idx, cpk_scan->sel_arg, HA_MRR_USE_DEFAULT_IMPL | HA_MRR_SORTED, 0, alloc))) { delete quick_intrsect; DBUG_RETURN(NULL); } quick->file= NULL; quick_intrsect->cpk_quick= quick; } quick_intrsect->records= records; quick_intrsect->read_time= read_cost; } DBUG_RETURN(quick_intrsect); } QUICK_SELECT_I *TRP_ROR_UNION::make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc) { QUICK_ROR_UNION_SELECT *quick_roru; TABLE_READ_PLAN **scan; QUICK_SELECT_I *quick; DBUG_ENTER("TRP_ROR_UNION::make_quick"); /* It is impossible to construct a ROR-union that will not retrieve full rows, ignore retrieve_full_rows parameter. */ if ((quick_roru= new QUICK_ROR_UNION_SELECT(param->thd, param->table))) { for (scan= first_ror; scan != last_ror; scan++) { if (!(quick= (*scan)->make_quick(param, FALSE, &quick_roru->alloc)) || quick_roru->push_quick_back(quick)) DBUG_RETURN(NULL); } quick_roru->records= records; quick_roru->read_time= read_cost; } DBUG_RETURN(quick_roru); } /* Build a SEL_TREE for <> or NOT BETWEEN predicate SYNOPSIS get_ne_mm_tree() param PARAM from SQL_SELECT::test_quick_select cond_func item for the predicate field field in the predicate lt_value constant that field should be smaller gt_value constant that field should be greaterr cmp_type compare type for the field RETURN # Pointer to tree built tree 0 on error */ static SEL_TREE *get_ne_mm_tree(RANGE_OPT_PARAM *param, Item_func *cond_func, Field *field, Item *lt_value, Item *gt_value, Item_result cmp_type) { SEL_TREE *tree; tree= get_mm_parts(param, cond_func, field, Item_func::LT_FUNC, lt_value, cmp_type); if (tree) { tree= tree_or(param, tree, get_mm_parts(param, cond_func, field, Item_func::GT_FUNC, gt_value, cmp_type)); } return tree; } /* Build a SEL_TREE for a simple predicate SYNOPSIS get_func_mm_tree() param PARAM from SQL_SELECT::test_quick_select cond_func item for the predicate field field in the predicate value constant in the predicate cmp_type compare type for the field inv TRUE <> NOT cond_func is considered (makes sense only when cond_func is BETWEEN or IN) RETURN Pointer to the tree built tree */ static SEL_TREE *get_func_mm_tree(RANGE_OPT_PARAM *param, Item_func *cond_func, Field *field, Item *value, Item_result cmp_type, bool inv) { SEL_TREE *tree= 0; DBUG_ENTER("get_func_mm_tree"); switch (cond_func->functype()) { case Item_func::NE_FUNC: tree= get_ne_mm_tree(param, cond_func, field, value, value, cmp_type); break; case Item_func::BETWEEN: { if (!value) { if (inv) { tree= get_ne_mm_tree(param, cond_func, field, cond_func->arguments()[1], cond_func->arguments()[2], cmp_type); } else { tree= get_mm_parts(param, cond_func, field, Item_func::GE_FUNC, cond_func->arguments()[1],cmp_type); if (tree) { tree= tree_and(param, tree, get_mm_parts(param, cond_func, field, Item_func::LE_FUNC, cond_func->arguments()[2], cmp_type)); } } } else tree= get_mm_parts(param, cond_func, field, (inv ? (value == (Item*)1 ? Item_func::GT_FUNC : Item_func::LT_FUNC): (value == (Item*)1 ? Item_func::LE_FUNC : Item_func::GE_FUNC)), cond_func->arguments()[0], cmp_type); break; } case Item_func::IN_FUNC: { Item_func_in *func=(Item_func_in*) cond_func; /* Array for IN() is constructed when all values have the same result type. Tree won't be built for values with different result types, so we check it here to avoid unnecessary work. */ if (!func->arg_types_compatible) break; if (inv) { if (func->array && func->array->result_type() != ROW_RESULT) { /* We get here for conditions in form "t.key NOT IN (c1, c2, ...)", where c{i} are constants. Our goal is to produce a SEL_TREE that represents intervals: ($MINmem_root; param->thd->mem_root= param->old_root; /* Create one Item_type constant object. We'll need it as get_mm_parts only accepts constant values wrapped in Item_Type objects. We create the Item on param->mem_root which points to per-statement mem_root (while thd->mem_root is currently pointing to mem_root local to range optimizer). */ Item *value_item= func->array->create_item(); param->thd->mem_root= tmp_root; if (func->array->count > NOT_IN_IGNORE_THRESHOLD || !value_item) break; /* Get a SEL_TREE for "(-inf|NULL) < X < c_0" interval. */ uint i=0; do { func->array->value_to_item(i, value_item); tree= get_mm_parts(param, cond_func, field, Item_func::LT_FUNC, value_item, cmp_type); if (!tree) break; i++; } while (i < func->array->count && tree->type == SEL_TREE::IMPOSSIBLE); if (!tree || tree->type == SEL_TREE::IMPOSSIBLE) { /* We get here in cases like "t.unsigned NOT IN (-1,-2,-3) */ tree= NULL; break; } SEL_TREE *tree2; for (; i < func->array->count; i++) { if (func->array->compare_elems(i, i-1)) { /* Get a SEL_TREE for "-inf < X < c_i" interval */ func->array->value_to_item(i, value_item); tree2= get_mm_parts(param, cond_func, field, Item_func::LT_FUNC, value_item, cmp_type); if (!tree2) { tree= NULL; break; } /* Change all intervals to be "c_{i-1} < X < c_i" */ for (uint idx= 0; idx < param->keys; idx++) { SEL_ARG *new_interval, *last_val; if (((new_interval= tree2->keys[idx])) && (tree->keys[idx]) && ((last_val= tree->keys[idx]->last()))) { new_interval->min_value= last_val->max_value; new_interval->min_flag= NEAR_MIN; } } /* The following doesn't try to allocate memory so no need to check for NULL. */ tree= tree_or(param, tree, tree2); } } if (tree && tree->type != SEL_TREE::IMPOSSIBLE) { /* Get the SEL_TREE for the last "c_last < X < +inf" interval (value_item cotains c_last already) */ tree2= get_mm_parts(param, cond_func, field, Item_func::GT_FUNC, value_item, cmp_type); tree= tree_or(param, tree, tree2); } } else { tree= get_ne_mm_tree(param, cond_func, field, func->arguments()[1], func->arguments()[1], cmp_type); if (tree) { Item **arg, **end; for (arg= func->arguments()+2, end= arg+func->argument_count()-2; arg < end ; arg++) { tree= tree_and(param, tree, get_ne_mm_tree(param, cond_func, field, *arg, *arg, cmp_type)); } } } } else { tree= get_mm_parts(param, cond_func, field, Item_func::EQ_FUNC, func->arguments()[1], cmp_type); if (tree) { Item **arg, **end; for (arg= func->arguments()+2, end= arg+func->argument_count()-2; arg < end ; arg++) { tree= tree_or(param, tree, get_mm_parts(param, cond_func, field, Item_func::EQ_FUNC, *arg, cmp_type)); } } } break; } default: { /* Here the function for the following predicates are processed: <, <=, =, >=, >, LIKE, IS NULL, IS NOT NULL. If the predicate is of the form (value op field) it is handled as the equivalent predicate (field rev_op value), e.g. 2 <= a is handled as a >= 2. */ Item_func::Functype func_type= (value != cond_func->arguments()[0]) ? cond_func->functype() : ((Item_bool_func2*) cond_func)->rev_functype(); tree= get_mm_parts(param, cond_func, field, func_type, value, cmp_type); } } DBUG_RETURN(tree); } /* Build conjunction of all SEL_TREEs for a simple predicate applying equalities SYNOPSIS get_full_func_mm_tree() param PARAM from SQL_SELECT::test_quick_select cond_func item for the predicate field_item field in the predicate value constant in the predicate (for BETWEEN it contains the number of the field argument, for IN it's always 0) inv TRUE <> NOT cond_func is considered (makes sense only when cond_func is BETWEEN or IN) DESCRIPTION For a simple SARGable predicate of the form (f op c), where f is a field and c is a constant, the function builds a conjunction of all SEL_TREES that can be obtained by the substitution of f for all different fields equal to f. NOTES If the WHERE condition contains a predicate (fi op c), then not only SELL_TREE for this predicate is built, but the trees for the results of substitution of fi for each fj belonging to the same multiple equality as fi are built as well. E.g. for WHERE t1.a=t2.a AND t2.a > 10 a SEL_TREE for t2.a > 10 will be built for quick select from t2 and a SEL_TREE for t1.a > 10 will be built for quick select from t1. A BETWEEN predicate of the form (fi [NOT] BETWEEN c1 AND c2) is treated in a similar way: we build a conjuction of trees for the results of all substitutions of fi for equal fj. Yet a predicate of the form (c BETWEEN f1i AND f2i) is processed differently. It is considered as a conjuction of two SARGable predicates (f1i <= c) and (f2i <=c) and the function get_full_func_mm_tree is called for each of them separately producing trees for AND j (f1j <=c ) and AND j (f2j <= c) After this these two trees are united in one conjunctive tree. It's easy to see that the same tree is obtained for AND j,k (f1j <=c AND f2k<=c) which is equivalent to AND j,k (c BETWEEN f1j AND f2k). The validity of the processing of the predicate (c NOT BETWEEN f1i AND f2i) which equivalent to (f1i > c OR f2i < c) is not so obvious. Here the function get_full_func_mm_tree is called for (f1i > c) and (f2i < c) producing trees for AND j (f1j > c) and AND j (f2j < c). Then this two trees are united in one OR-tree. The expression (AND j (f1j > c) OR AND j (f2j < c) is equivalent to the expression AND j,k (f1j > c OR f2k < c) which is just a translation of AND j,k (c NOT BETWEEN f1j AND f2k) In the cases when one of the items f1, f2 is a constant c1 we do not create a tree for it at all. It works for BETWEEN predicates but does not work for NOT BETWEEN predicates as we have to evaluate the expression with it. If it is TRUE then the other tree can be completely ignored. We do not do it now and no trees are built in these cases for NOT BETWEEN predicates. As to IN predicates only ones of the form (f IN (c1,...,cn)), where f1 is a field and c1,...,cn are constant, are considered as SARGable. We never try to narrow the index scan using predicates of the form (c IN (c1,...,f,...,cn)). RETURN Pointer to the tree representing the built conjunction of SEL_TREEs */ static SEL_TREE *get_full_func_mm_tree(RANGE_OPT_PARAM *param, Item_func *cond_func, Item_field *field_item, Item *value, bool inv) { SEL_TREE *tree= 0; SEL_TREE *ftree= 0; table_map ref_tables= 0; table_map param_comp= ~(param->prev_tables | param->read_tables | param->current_table); DBUG_ENTER("get_full_func_mm_tree"); for (uint i= 0; i < cond_func->arg_count; i++) { Item *arg= cond_func->arguments()[i]->real_item(); if (arg != field_item) ref_tables|= arg->used_tables(); } Field *field= field_item->field; Item_result cmp_type= field->cmp_type(); if (!((ref_tables | field->table->map) & param_comp)) ftree= get_func_mm_tree(param, cond_func, field, value, cmp_type, inv); Item_equal *item_equal= field_item->item_equal; if (item_equal) { Item_equal_fields_iterator it(*item_equal); while (it++) { Field *f= it.get_curr_field(); if (field->eq(f)) continue; if (!((ref_tables | f->table->map) & param_comp)) { tree= get_func_mm_tree(param, cond_func, f, value, cmp_type, inv); ftree= !ftree ? tree : tree_and(param, ftree, tree); } } } DBUG_RETURN(ftree); } /* make a select tree of all keys in condition */ static SEL_TREE *get_mm_tree(RANGE_OPT_PARAM *param,COND *cond) { SEL_TREE *tree=0; SEL_TREE *ftree= 0; Item_field *field_item= 0; bool inv= FALSE; Item *value= 0; DBUG_ENTER("get_mm_tree"); if (cond->type() == Item::COND_ITEM) { List_iterator li(*((Item_cond*) cond)->argument_list()); if (((Item_cond*) cond)->functype() == Item_func::COND_AND_FUNC) { tree=0; Item *item; while ((item=li++)) { SEL_TREE *new_tree=get_mm_tree(param,item); if (param->thd->is_fatal_error || param->alloced_sel_args > SEL_ARG::MAX_SEL_ARGS) DBUG_RETURN(0); // out of memory tree=tree_and(param,tree,new_tree); if (tree && tree->type == SEL_TREE::IMPOSSIBLE) break; } } else { // COND OR tree=get_mm_tree(param,li++); if (tree) { Item *item; while ((item=li++)) { SEL_TREE *new_tree=get_mm_tree(param,item); if (!new_tree) DBUG_RETURN(0); // out of memory tree=tree_or(param,tree,new_tree); if (!tree || tree->type == SEL_TREE::ALWAYS) break; } } } DBUG_RETURN(tree); } /* Here when simple cond */ if (cond->const_item() && !cond->is_expensive()) { /* During the cond->val_int() evaluation we can come across a subselect item which may allocate memory on the thd->mem_root and assumes all the memory allocated has the same life span as the subselect item itself. So we have to restore the thread's mem_root here. */ MEM_ROOT *tmp_root= param->mem_root; param->thd->mem_root= param->old_root; tree= cond->val_int() ? new(tmp_root) SEL_TREE(SEL_TREE::ALWAYS) : new(tmp_root) SEL_TREE(SEL_TREE::IMPOSSIBLE); param->thd->mem_root= tmp_root; DBUG_RETURN(tree); } table_map ref_tables= 0; table_map param_comp= ~(param->prev_tables | param->read_tables | param->current_table); if (cond->type() != Item::FUNC_ITEM) { // Should be a field ref_tables= cond->used_tables(); if ((ref_tables & param->current_table) || (ref_tables & ~(param->prev_tables | param->read_tables))) DBUG_RETURN(0); DBUG_RETURN(new SEL_TREE(SEL_TREE::MAYBE)); } Item_func *cond_func= (Item_func*) cond; if (cond_func->functype() == Item_func::BETWEEN || cond_func->functype() == Item_func::IN_FUNC) inv= ((Item_func_opt_neg *) cond_func)->negated; else if (cond_func->select_optimize() == Item_func::OPTIMIZE_NONE) DBUG_RETURN(0); param->cond= cond; switch (cond_func->functype()) { case Item_func::BETWEEN: if (cond_func->arguments()[0]->real_item()->type() == Item::FIELD_ITEM) { field_item= (Item_field*) (cond_func->arguments()[0]->real_item()); ftree= get_full_func_mm_tree(param, cond_func, field_item, NULL, inv); } /* Concerning the code below see the NOTES section in the comments for the function get_full_func_mm_tree() */ for (uint i= 1 ; i < cond_func->arg_count ; i++) { if (cond_func->arguments()[i]->real_item()->type() == Item::FIELD_ITEM) { field_item= (Item_field*) (cond_func->arguments()[i]->real_item()); SEL_TREE *tmp= get_full_func_mm_tree(param, cond_func, field_item, (Item*)(intptr)i, inv); if (inv) { tree= !tree ? tmp : tree_or(param, tree, tmp); if (tree == NULL) break; } else tree= tree_and(param, tree, tmp); } else if (inv) { tree= 0; break; } } ftree = tree_and(param, ftree, tree); break; case Item_func::IN_FUNC: { Item_func_in *func=(Item_func_in*) cond_func; if (func->key_item()->real_item()->type() != Item::FIELD_ITEM) DBUG_RETURN(0); field_item= (Item_field*) (func->key_item()->real_item()); ftree= get_full_func_mm_tree(param, cond_func, field_item, NULL, inv); break; } case Item_func::MULT_EQUAL_FUNC: { Item_equal *item_equal= (Item_equal *) cond; if (!(value= item_equal->get_const()) || value->is_expensive()) DBUG_RETURN(0); Item_equal_fields_iterator it(*item_equal); ref_tables= value->used_tables(); while (it++) { Field *field= it.get_curr_field(); Item_result cmp_type= field->cmp_type(); if (!((ref_tables | field->table->map) & param_comp)) { tree= get_mm_parts(param, cond, field, Item_func::EQ_FUNC, value,cmp_type); ftree= !ftree ? tree : tree_and(param, ftree, tree); } } DBUG_RETURN(ftree); } default: if (cond_func->arguments()[0]->real_item()->type() == Item::FIELD_ITEM) { field_item= (Item_field*) (cond_func->arguments()[0]->real_item()); value= cond_func->arg_count > 1 ? cond_func->arguments()[1] : 0; } else if (cond_func->have_rev_func() && cond_func->arguments()[1]->real_item()->type() == Item::FIELD_ITEM) { field_item= (Item_field*) (cond_func->arguments()[1]->real_item()); value= cond_func->arguments()[0]; } else DBUG_RETURN(0); if (value && value->is_expensive()) DBUG_RETURN(0); ftree= get_full_func_mm_tree(param, cond_func, field_item, value, inv); } DBUG_RETURN(ftree); } static SEL_TREE * get_mm_parts(RANGE_OPT_PARAM *param, COND *cond_func, Field *field, Item_func::Functype type, Item *value, Item_result cmp_type) { DBUG_ENTER("get_mm_parts"); if (field->table != param->table) DBUG_RETURN(0); KEY_PART *key_part = param->key_parts; KEY_PART *end = param->key_parts_end; SEL_TREE *tree=0; if (value && value->used_tables() & ~(param->prev_tables | param->read_tables)) DBUG_RETURN(0); for (; key_part != end ; key_part++) { if (field->eq(key_part->field)) { SEL_ARG *sel_arg=0; if (!tree && !(tree=new SEL_TREE())) DBUG_RETURN(0); // OOM if (!value || !(value->used_tables() & ~param->read_tables)) { sel_arg=get_mm_leaf(param,cond_func, key_part->field,key_part,type,value); if (!sel_arg) continue; if (sel_arg->type == SEL_ARG::IMPOSSIBLE) { tree->type=SEL_TREE::IMPOSSIBLE; DBUG_RETURN(tree); } } else { // This key may be used later if (!(sel_arg= new SEL_ARG(SEL_ARG::MAYBE_KEY))) DBUG_RETURN(0); // OOM } sel_arg->part=(uchar) key_part->part; sel_arg->max_part_no= sel_arg->part+1; tree->keys[key_part->key]=sel_add(tree->keys[key_part->key],sel_arg); tree->keys_map.set_bit(key_part->key); } } if (tree && tree->merges.is_empty() && tree->keys_map.is_clear_all()) tree= NULL; DBUG_RETURN(tree); } static SEL_ARG * get_mm_leaf(RANGE_OPT_PARAM *param, COND *conf_func, Field *field, KEY_PART *key_part, Item_func::Functype type,Item *value) { uint maybe_null=(uint) field->real_maybe_null(); bool optimize_range; SEL_ARG *tree= 0; MEM_ROOT *alloc= param->mem_root; uchar *str; int err; DBUG_ENTER("get_mm_leaf"); /* We need to restore the runtime mem_root of the thread in this function because it evaluates the value of its argument, while the argument can be any, e.g. a subselect. The subselect items, in turn, assume that all the memory allocated during the evaluation has the same life span as the item itself. TODO: opt_range.cc should not reset thd->mem_root at all. */ param->thd->mem_root= param->old_root; if (!value) // IS NULL or IS NOT NULL { if (field->table->maybe_null) // Can't use a key on this goto end; if (!maybe_null) // Not null field { if (type == Item_func::ISNULL_FUNC) tree= &null_element; goto end; } if (!(tree= new (alloc) SEL_ARG(field,is_null_string,is_null_string))) goto end; // out of memory if (type == Item_func::ISNOTNULL_FUNC) { tree->min_flag=NEAR_MIN; /* IS NOT NULL -> X > NULL */ tree->max_flag=NO_MAX_RANGE; } goto end; } /* 1. Usually we can't use an index if the column collation differ from the operation collation. 2. However, we can reuse a case insensitive index for the binary searches: WHERE latin1_swedish_ci_column = 'a' COLLATE lati1_bin; WHERE latin1_swedish_ci_colimn = BINARY 'a ' */ if (field->result_type() == STRING_RESULT && ((Field_str*) field)->match_collation_to_optimize_range() && value->result_type() == STRING_RESULT && key_part->image_type == Field::itRAW && ((Field_str*)field)->charset() != conf_func->compare_collation() && !(conf_func->compare_collation()->state & MY_CS_BINSORT && (type == Item_func::EQUAL_FUNC || type == Item_func::EQ_FUNC))) goto end; if (key_part->image_type == Field::itMBR) { switch (type) { case Item_func::SP_EQUALS_FUNC: case Item_func::SP_DISJOINT_FUNC: case Item_func::SP_INTERSECTS_FUNC: case Item_func::SP_TOUCHES_FUNC: case Item_func::SP_CROSSES_FUNC: case Item_func::SP_WITHIN_FUNC: case Item_func::SP_CONTAINS_FUNC: case Item_func::SP_OVERLAPS_FUNC: break; default: /* We cannot involve spatial indexes for queries that don't use MBREQUALS(), MBRDISJOINT(), etc. functions. */ goto end; } } if (param->using_real_indexes) optimize_range= field->optimize_range(param->real_keynr[key_part->key], key_part->part); else optimize_range= TRUE; if (type == Item_func::LIKE_FUNC) { bool like_error; char buff1[MAX_FIELD_WIDTH]; uchar *min_str,*max_str; String tmp(buff1,sizeof(buff1),value->collation.collation),*res; size_t length, offset, min_length, max_length; uint field_length= field->pack_length()+maybe_null; if (!optimize_range) goto end; if (!(res= value->val_str(&tmp))) { tree= &null_element; goto end; } /* TODO: Check if this was a function. This should have be optimized away in the sql_select.cc */ if (res != &tmp) { tmp.copy(*res); // Get own copy res= &tmp; } if (field->cmp_type() != STRING_RESULT) goto end; // Can only optimize strings offset=maybe_null; length=key_part->store_length; if (length != key_part->length + maybe_null) { /* key packed with length prefix */ offset+= HA_KEY_BLOB_LENGTH; field_length= length - HA_KEY_BLOB_LENGTH; } else { if (unlikely(length < field_length)) { /* This can only happen in a table created with UNIREG where one key overlaps many fields */ length= field_length; } else field_length= length; } length+=offset; if (!(min_str= (uchar*) alloc_root(alloc, length*2))) goto end; max_str=min_str+length; if (maybe_null) max_str[0]= min_str[0]=0; field_length-= maybe_null; like_error= my_like_range(field->charset(), res->ptr(), res->length(), ((Item_func_like*)(param->cond))->escape, wild_one, wild_many, field_length, (char*) min_str+offset, (char*) max_str+offset, &min_length, &max_length); if (like_error) // Can't optimize with LIKE goto end; if (offset != maybe_null) // BLOB or VARCHAR { int2store(min_str+maybe_null,min_length); int2store(max_str+maybe_null,max_length); } tree= new (alloc) SEL_ARG(field, min_str, max_str); goto end; } if (!optimize_range && type != Item_func::EQ_FUNC && type != Item_func::EQUAL_FUNC) goto end; // Can't optimize this /* We can't always use indexes when comparing a string index to a number cmp_type() is checked to allow compare of dates to numbers */ if (field->cmp_type() == STRING_RESULT && value->cmp_type() != STRING_RESULT) goto end; err= value->save_in_field_no_warnings(field, 1); if (err > 0) { if (field->cmp_type() != value->result_type()) { if ((type == Item_func::EQ_FUNC || type == Item_func::EQUAL_FUNC) && value->result_type() == item_cmp_type(field->result_type(), value->result_type())) { tree= new (alloc) SEL_ARG(field, 0, 0); tree->type= SEL_ARG::IMPOSSIBLE; goto end; } else { /* TODO: We should return trees of the type SEL_ARG::IMPOSSIBLE for the cases like int_field > 999999999999999999999999 as well. */ tree= 0; if (err == 3 && field->type() == FIELD_TYPE_DATE && (type == Item_func::GT_FUNC || type == Item_func::GE_FUNC || type == Item_func::LT_FUNC || type == Item_func::LE_FUNC) ) { /* We were saving DATETIME into a DATE column, the conversion went ok but a non-zero time part was cut off. In MySQL's SQL dialect, DATE and DATETIME are compared as datetime values. Index over a DATE column uses DATE comparison. Changing from one comparison to the other is possible: datetime(date_col)< '2007-12-10 12:34:55' -> date_col<='2007-12-10' datetime(date_col)<='2007-12-10 12:34:55' -> date_col<='2007-12-10' datetime(date_col)> '2007-12-10 12:34:55' -> date_col>='2007-12-10' datetime(date_col)>='2007-12-10 12:34:55' -> date_col>='2007-12-10' but we'll need to convert '>' to '>=' and '<' to '<='. This will be done together with other types at the end of this function (grep for stored_field_cmp_to_item) */ } else goto end; } } /* guaranteed at this point: err > 0; field and const of same type If an integer got bounded (e.g. to within 0..255 / -128..127) for < or >, set flags as for <= or >= (no NEAR_MAX / NEAR_MIN) */ else if (err == 1 && field->result_type() == INT_RESULT) { if (type == Item_func::LT_FUNC && (value->val_int() > 0)) type = Item_func::LE_FUNC; else if (type == Item_func::GT_FUNC && (field->type() != FIELD_TYPE_BIT) && !((Field_num*)field)->unsigned_flag && !((Item_int*)value)->unsigned_flag && (value->val_int() < 0)) type = Item_func::GE_FUNC; } } else if (err < 0) { /* This happens when we try to insert a NULL field in a not null column */ tree= &null_element; // cmp with NULL is never TRUE goto end; } /* Any sargable predicate except "<=>" involving NULL as a constant is always FALSE */ if (type != Item_func::EQUAL_FUNC && field->is_real_null()) { tree= &null_element; goto end; } str= (uchar*) alloc_root(alloc, key_part->store_length+1); if (!str) goto end; if (maybe_null) *str= (uchar) field->is_real_null(); // Set to 1 if null field->get_key_image(str+maybe_null, key_part->length, key_part->image_type); if (!(tree= new (alloc) SEL_ARG(field, str, str))) goto end; // out of memory /* Check if we are comparing an UNSIGNED integer with a negative constant. In this case we know that: (a) (unsigned_int [< | <=] negative_constant) == FALSE (b) (unsigned_int [> | >=] negative_constant) == TRUE In case (a) the condition is false for all values, and in case (b) it is true for all values, so we can avoid unnecessary retrieval and condition testing, and we also get correct comparison of unsinged integers with negative integers (which otherwise fails because at query execution time negative integers are cast to unsigned if compared with unsigned). */ if (field->result_type() == INT_RESULT && value->result_type() == INT_RESULT && ((field->type() == FIELD_TYPE_BIT || ((Field_num *) field)->unsigned_flag) && !((Item_int*) value)->unsigned_flag)) { longlong item_val= value->val_int(); if (item_val < 0) { if (type == Item_func::LT_FUNC || type == Item_func::LE_FUNC) { tree->type= SEL_ARG::IMPOSSIBLE; goto end; } if (type == Item_func::GT_FUNC || type == Item_func::GE_FUNC) { tree= 0; goto end; } } } switch (type) { case Item_func::LT_FUNC: if (stored_field_cmp_to_item(param->thd, field, value) == 0) tree->max_flag=NEAR_MAX; /* fall through */ case Item_func::LE_FUNC: if (!maybe_null) tree->min_flag=NO_MIN_RANGE; /* From start */ else { // > NULL tree->min_value=is_null_string; tree->min_flag=NEAR_MIN; } break; case Item_func::GT_FUNC: /* Don't use open ranges for partial key_segments */ if ((!(key_part->flag & HA_PART_KEY_SEG)) && (stored_field_cmp_to_item(param->thd, field, value) <= 0)) tree->min_flag=NEAR_MIN; tree->max_flag= NO_MAX_RANGE; break; case Item_func::GE_FUNC: /* Don't use open ranges for partial key_segments */ if ((!(key_part->flag & HA_PART_KEY_SEG)) && (stored_field_cmp_to_item(param->thd, field, value) < 0)) tree->min_flag= NEAR_MIN; tree->max_flag=NO_MAX_RANGE; break; case Item_func::SP_EQUALS_FUNC: tree->min_flag=GEOM_FLAG | HA_READ_MBR_EQUAL;// NEAR_MIN;//512; tree->max_flag=NO_MAX_RANGE; break; case Item_func::SP_DISJOINT_FUNC: tree->min_flag=GEOM_FLAG | HA_READ_MBR_DISJOINT;// NEAR_MIN;//512; tree->max_flag=NO_MAX_RANGE; break; case Item_func::SP_INTERSECTS_FUNC: tree->min_flag=GEOM_FLAG | HA_READ_MBR_INTERSECT;// NEAR_MIN;//512; tree->max_flag=NO_MAX_RANGE; break; case Item_func::SP_TOUCHES_FUNC: tree->min_flag=GEOM_FLAG | HA_READ_MBR_INTERSECT;// NEAR_MIN;//512; tree->max_flag=NO_MAX_RANGE; break; case Item_func::SP_CROSSES_FUNC: tree->min_flag=GEOM_FLAG | HA_READ_MBR_INTERSECT;// NEAR_MIN;//512; tree->max_flag=NO_MAX_RANGE; break; case Item_func::SP_WITHIN_FUNC: tree->min_flag=GEOM_FLAG | HA_READ_MBR_WITHIN;// NEAR_MIN;//512; tree->max_flag=NO_MAX_RANGE; break; case Item_func::SP_CONTAINS_FUNC: tree->min_flag=GEOM_FLAG | HA_READ_MBR_CONTAIN;// NEAR_MIN;//512; tree->max_flag=NO_MAX_RANGE; break; case Item_func::SP_OVERLAPS_FUNC: tree->min_flag=GEOM_FLAG | HA_READ_MBR_INTERSECT;// NEAR_MIN;//512; tree->max_flag=NO_MAX_RANGE; break; default: break; } end: param->thd->mem_root= alloc; DBUG_RETURN(tree); } /****************************************************************************** ** Tree manipulation functions ** If tree is 0 it means that the condition can't be tested. It refers ** to a non existent table or to a field in current table with isn't a key. ** The different tree flags: ** IMPOSSIBLE: Condition is never TRUE ** ALWAYS: Condition is always TRUE ** MAYBE: Condition may exists when tables are read ** MAYBE_KEY: Condition refers to a key that may be used in join loop ** KEY_RANGE: Condition uses a key ******************************************************************************/ /* Add a new key test to a key when scanning through all keys This will never be called for same key parts. */ static SEL_ARG * sel_add(SEL_ARG *key1,SEL_ARG *key2) { SEL_ARG *root,**key_link; if (!key1) return key2; if (!key2) return key1; key_link= &root; while (key1 && key2) { if (key1->part < key2->part) { *key_link= key1; key_link= &key1->next_key_part; key1=key1->next_key_part; } else { *key_link= key2; key_link= &key2->next_key_part; key2=key2->next_key_part; } } *key_link=key1 ? key1 : key2; return root; } /* Build a range tree for the conjunction of the range parts of two trees SYNOPSIS and_range_trees() param Context info for the operation tree1 SEL_TREE for the first conjunct tree2 SEL_TREE for the second conjunct result SEL_TREE for the result DESCRIPTION This function takes range parts of two trees tree1 and tree2 and builds a range tree for the conjunction of the formulas that these two range parts represent. More exactly: if the range part of tree1 represents the normalized formula R1_1 AND ... AND R1_k, and the range part of tree2 represents the normalized formula R2_1 AND ... AND R2_k, then the range part of the result represents the formula: RT = R_1 AND ... AND R_k, where R_i=(R1_i AND R2_i) for each i from [1..k] The function assumes that tree1 is never equal to tree2. At the same time the tree result can be the same as tree1 (but never as tree2). If result==tree1 then rt replaces the range part of tree1 leaving imerges as they are. if result!=tree1 than it is assumed that the SEL_ARG trees in tree1 and tree2 should be preserved. Otherwise they can be destroyed. RETURN 1 if the type the result tree is SEL_TREE::IMPOSSIBLE 0 otherwise */ static int and_range_trees(RANGE_OPT_PARAM *param, SEL_TREE *tree1, SEL_TREE *tree2, SEL_TREE *result) { DBUG_ENTER("and_ranges"); key_map result_keys; result_keys.clear_all(); key_map anded_keys= tree1->keys_map; anded_keys.merge(tree2->keys_map); int key_no; key_map::Iterator it(anded_keys); while ((key_no= it++) != key_map::Iterator::BITMAP_END) { uint flag=0; SEL_ARG *key1= tree1->keys[key_no]; SEL_ARG *key2= tree2->keys[key_no]; if (key1 && !key1->simple_key()) flag|= CLONE_KEY1_MAYBE; if (key2 && !key2->simple_key()) flag|=CLONE_KEY2_MAYBE; if (result != tree1) { if (key1) key1->incr_refs(); if (key2) key2->incr_refs(); } SEL_ARG *key; if ((result->keys[key_no]= key =key_and(param, key1, key2, flag))) { if (key && key->type == SEL_ARG::IMPOSSIBLE) { result->type= SEL_TREE::IMPOSSIBLE; DBUG_RETURN(1); } result_keys.set_bit(key_no); #ifdef EXTRA_DEBUG if (param->alloced_sel_args < SEL_ARG::MAX_SEL_ARGS) key->test_use_count(key); #endif } } result->keys_map= result_keys; DBUG_RETURN(0); } /* Build a SEL_TREE for a conjunction out of such trees for the conjuncts SYNOPSIS tree_and() param Context info for the operation tree1 SEL_TREE for the first conjunct tree2 SEL_TREE for the second conjunct DESCRIPTION This function builds a tree for the formula (A AND B) out of the trees tree1 and tree2 that has been built for the formulas A and B respectively. In a general case tree1 represents the formula RT1 AND MT1, where RT1 = R1_1 AND ... AND R1_k1, MT1=M1_1 AND ... AND M1_l1; tree2 represents the formula RT2 AND MT2 where RT2 = R2_1 AND ... AND R2_k2, MT2=M2_1 and ... and M2_l2. The result tree will represent the formula of the the following structure: RT AND MT1 AND MT2 AND RT1MT2 AND RT2MT1, such that rt is a tree obtained by range intersection of trees tree1 and tree2, RT1MT2 = RT1M2_1 AND ... AND RT1M2_l2, RT2MT1 = RT2M1_1 AND ... AND RT2M1_l1, where rt1m2_i (i=1,...,l2) is the result of the pushdown operation of range tree rt1 into imerge m2_i, while rt2m1_j (j=1,...,l1) is the result of the pushdown operation of range tree rt2 into imerge m1_j. RT1MT2/RT2MT is empty if MT2/MT1 is empty. The range intersection of two range trees is produced by the function and_range_trees. The pushdown of a range tree to a imerge is performed by the function imerge_list_and_tree. This function may produce imerges containing only one range tree. Such trees are intersected with rt and the result of intersection is returned as the range part of the result tree, while the corresponding imerges are removed altogether from its imerge part. NOTE. The pushdown operation of range trees into imerges is needed to be able to construct valid imerges for the condition like this: key1_p1=c1 AND (key1_p2 BETWEEN c21 AND c22 OR key2 < c2) RETURN The result tree, if a success 0 - otherwise. */ static SEL_TREE *tree_and(RANGE_OPT_PARAM *param, SEL_TREE *tree1, SEL_TREE *tree2) { DBUG_ENTER("tree_and"); if (!tree1) DBUG_RETURN(tree2); if (!tree2) DBUG_RETURN(tree1); if (tree1->type == SEL_TREE::IMPOSSIBLE || tree2->type == SEL_TREE::ALWAYS) DBUG_RETURN(tree1); if (tree2->type == SEL_TREE::IMPOSSIBLE || tree1->type == SEL_TREE::ALWAYS) DBUG_RETURN(tree2); if (tree1->type == SEL_TREE::MAYBE) { if (tree2->type == SEL_TREE::KEY) tree2->type=SEL_TREE::KEY_SMALLER; DBUG_RETURN(tree2); } if (tree2->type == SEL_TREE::MAYBE) { tree1->type=SEL_TREE::KEY_SMALLER; DBUG_RETURN(tree1); } if (!tree1->merges.is_empty()) imerge_list_and_tree(param, &tree1->merges, tree2); if (!tree2->merges.is_empty()) imerge_list_and_tree(param, &tree2->merges, tree1); if (and_range_trees(param, tree1, tree2, tree1)) DBUG_RETURN(tree1); imerge_list_and_list(&tree1->merges, &tree2->merges); eliminate_single_tree_imerges(param, tree1); DBUG_RETURN(tree1); } /* Eliminate single tree imerges in a SEL_TREE objects SYNOPSIS eliminate_single_tree_imerges() param Context info for the function tree SEL_TREE where single tree imerges are to be eliminated DESCRIPTION For each imerge in 'tree' that contains only one disjunct tree, i.e. for any imerge of the form m=rt, the function performs and operation the range part of tree, replaces rt the with the result of anding and removes imerge m from the the merge part of 'tree'. RETURN VALUE none */ static void eliminate_single_tree_imerges(RANGE_OPT_PARAM *param, SEL_TREE *tree) { SEL_IMERGE *imerge; List merges= tree->merges; List_iterator it(merges); tree->merges.empty(); while ((imerge= it++)) { if (imerge->trees+1 == imerge->trees_next) { tree= tree_and(param, tree, *imerge->trees); it.remove(); } } tree->merges= merges; } /* For two trees check that there are indexes with ranges in both of them SYNOPSIS sel_trees_have_common_keys() tree1 SEL_TREE for the first tree tree2 SEL_TREE for the second tree common_keys OUT bitmap of all indexes with ranges in both trees DESCRIPTION For two trees tree1 and tree1 the function checks if there are indexes in their range parts such that SEL_ARG trees are defined for them in the range parts of both trees. The function returns the bitmap of such indexes in the parameter common_keys. RETURN TRUE if there are such indexes (common_keys is nor empty) FALSE otherwise */ static bool sel_trees_have_common_keys(SEL_TREE *tree1, SEL_TREE *tree2, key_map *common_keys) { *common_keys= tree1->keys_map; common_keys->intersect(tree2->keys_map); return !common_keys->is_clear_all(); } /* Check whether range parts of two trees can be ored for some indexes SYNOPSIS sel_trees_can_be_ored() param Context info for the function tree1 SEL_TREE for the first tree tree2 SEL_TREE for the second tree common_keys IN/OUT IN: bitmap of all indexes with SEL_ARG in both trees OUT: bitmap of all indexes that can be ored DESCRIPTION For two trees tree1 and tree2 and the bitmap common_keys containing bits for indexes that have SEL_ARG trees in range parts of both trees the function checks if there are indexes for which SEL_ARG trees can be ored. Two SEL_ARG trees for the same index can be ored if the most major components of the index used in these trees coincide. If the SEL_ARG trees for an index cannot be ored the function clears the bit for this index in the bitmap common_keys. The function does not verify that indexes marked in common_keys really have SEL_ARG trees in both tree1 and tree2. It assumes that this is true. NOTE The function sel_trees_can_be_ored is usually used in pair with the function sel_trees_have_common_keys. RETURN TRUE if there are indexes for which SEL_ARG trees can be ored FALSE otherwise */ static bool sel_trees_can_be_ored(RANGE_OPT_PARAM* param, SEL_TREE *tree1, SEL_TREE *tree2, key_map *common_keys) { DBUG_ENTER("sel_trees_can_be_ored"); if (!sel_trees_have_common_keys(tree1, tree2, common_keys)) DBUG_RETURN(FALSE); int key_no; key_map::Iterator it(*common_keys); while ((key_no= it++) != key_map::Iterator::BITMAP_END) { DBUG_ASSERT(tree1->keys[key_no] && tree2->keys[key_no]); /* Trees have a common key, check if they refer to the same key part */ if (tree1->keys[key_no]->part != tree2->keys[key_no]->part) common_keys->clear_bit(key_no); } DBUG_RETURN(!common_keys->is_clear_all()); } /* Check whether range parts of two trees must be ored for some indexes SYNOPSIS sel_trees_must_be_ored() param Context info for the function tree1 SEL_TREE for the first tree tree2 SEL_TREE for the second tree ordable_keys bitmap of SEL_ARG trees that can be ored DESCRIPTION For two trees tree1 and tree2 the function checks whether they must be ored. The function assumes that the bitmap ordable_keys contains bits for those corresponding pairs of SEL_ARG trees from tree1 and tree2 that can be ored. We believe that tree1 and tree2 must be ored if any pair of SEL_ARG trees r1 and r2, such that r1 is from tree1 and r2 is from tree2 and both of them are marked in ordable_keys, can be merged. NOTE The function sel_trees_must_be_ored as a rule is used in pair with the function sel_trees_can_be_ored. RETURN TRUE if there are indexes for which SEL_ARG trees must be ored FALSE otherwise */ static bool sel_trees_must_be_ored(RANGE_OPT_PARAM* param, SEL_TREE *tree1, SEL_TREE *tree2, key_map oredable_keys) { key_map tmp; DBUG_ENTER("sel_trees_must_be_ored"); tmp= tree1->keys_map; tmp.merge(tree2->keys_map); tmp.subtract(oredable_keys); if (!tmp.is_clear_all()) DBUG_RETURN(FALSE); int idx1, idx2; key_map::Iterator it1(oredable_keys); while ((idx1= it1++) != key_map::Iterator::BITMAP_END) { KEY_PART *key1_init= param->key[idx1]+tree1->keys[idx1]->part; KEY_PART *key1_end= param->key[idx1]+tree1->keys[idx1]->max_part_no; key_map::Iterator it2(oredable_keys); while ((idx2= it2++) != key_map::Iterator::BITMAP_END) { if (idx2 <= idx1) continue; KEY_PART *key2_init= param->key[idx2]+tree2->keys[idx2]->part; KEY_PART *key2_end= param->key[idx2]+tree2->keys[idx2]->max_part_no; KEY_PART *part1, *part2; for (part1= key1_init, part2= key2_init; part1 < key1_end && part2 < key2_end; part1++, part2++) { if (!part1->field->eq(part2->field)) DBUG_RETURN(FALSE); } } } DBUG_RETURN(TRUE); } /* Remove the trees that are not suitable for record retrieval SYNOPSIS remove_nonrange_trees() param Context info for the function tree Tree to be processed, tree->type is KEY or KEY_SMALLER DESCRIPTION This function walks through tree->keys[] and removes the SEL_ARG* trees that are not "maybe" trees (*) and cannot be used to construct quick range selects. (*) - have type MAYBE or MAYBE_KEY. Perhaps we should remove trees of these types here as well. A SEL_ARG* tree cannot be used to construct quick select if it has tree->part != 0. (e.g. it could represent "keypart2 < const"). Normally we allow construction of SEL_TREE objects that have SEL_ARG trees that do not allow quick range select construction. For example: for " keypart1=1 AND keypart2=2 " the execution will proceed as follows: tree1= SEL_TREE { SEL_ARG{keypart1=1} } tree2= SEL_TREE { SEL_ARG{keypart2=2} } -- can't make quick range select from this call tree_and(tree1, tree2) -- this joins SEL_ARGs into a usable SEL_ARG tree. Another example: tree3= SEL_TREE { SEL_ARG{key1part1 = 1} } tree4= SEL_TREE { SEL_ARG{key2part2 = 2} } -- can't make quick range select from this call tree_or(tree3, tree4) -- creates a SEL_MERGE ot of which no index merge can be constructed, but it is potentially useful, as anding it with tree5= SEL_TREE { SEL_ARG{key2part1 = 3} } creates an index merge that represents the formula key1part1=1 AND key2part1=3 OR key2part1=3 AND key2part2=2 for which an index merge can be built. Any final SEL_TREE may contain SEL_ARG trees for which no quick select can be built. Such SEL_ARG trees should be removed from the range part before different range scans are evaluated. Such SEL_ARG trees also should be removed from all range trees of each index merge before different possible index merge plans are evaluated. If after this removal one of the range trees in the index merge becomes empty the whole index merge must be discarded. RETURN 0 Ok, some suitable trees left 1 No tree->keys[] left. */ static bool remove_nonrange_trees(RANGE_OPT_PARAM *param, SEL_TREE *tree) { bool res= FALSE; for (uint i=0; i < param->keys; i++) { if (tree->keys[i]) { if (tree->keys[i]->part) { tree->keys[i]= NULL; tree->keys_map.clear_bit(i); } else res= TRUE; } } return !res; } /* Build a SEL_TREE for a disjunction out of such trees for the disjuncts SYNOPSIS tree_or() param Context info for the operation tree1 SEL_TREE for the first disjunct tree2 SEL_TREE for the second disjunct DESCRIPTION This function builds a tree for the formula (A OR B) out of the trees tree1 and tree2 that has been built for the formulas A and B respectively. In a general case tree1 represents the formula RT1 AND MT1, where RT1=R1_1 AND ... AND R1_k1, MT1=M1_1 AND ... AND M1_l1; tree2 represents the formula RT2 AND MT2 where RT2=R2_1 AND ... AND R2_k2, MT2=M2_1 and ... and M2_l2. The function constructs the result tree according the formula (RT1 OR RT2) AND (MT1 OR RT1) AND (MT2 OR RT2) AND (MT1 OR MT2) that is equivalent to the formula (RT1 AND MT1) OR (RT2 AND MT2). To limit the number of produced imerges the function considers a weaker formula than the original one: (RT1 AND M1_1) OR (RT2 AND M2_1) that is equivalent to: (RT1 OR RT2) (1) AND (M1_1 OR M2_1) (2) AND (M1_1 OR RT2) (3) AND (M2_1 OR RT1) (4) For the first conjunct (1) the function builds a tree with a range part and, possibly, one imerge. For the other conjuncts (2-4)the function produces sets of imerges. All constructed imerges are included into the result tree. For the formula (1) the function produces the tree representing a formula of the structure RT [AND M], such that: - the range tree rt contains the result of oring SEL_ARG trees from rt1 and rt2 - the imerge m consists of two range trees rt1 and rt2. The imerge m is added if it's not true that rt1 and rt2 must be ored If rt1 and rt2 can't be ored rt is empty and only m is produced for (1). To produce imerges for the formula (2) the function calls the function imerge_list_or_list passing it the merge parts of tree1 and tree2 as parameters. To produce imerges for the formula (3) the function calls the function imerge_list_or_tree passing it the imerge m1_1 and the range tree rt2 as parameters. Similarly, to produce imerges for the formula (4) the function calls the function imerge_list_or_tree passing it the imerge m2_1 and the range tree rt1. If rt1 is empty then the trees for (1) and (4) are empty. If rt2 is empty then the trees for (1) and (3) are empty. If mt1 is empty then the trees for (2) and (3) are empty. If mt2 is empty then the trees for (2) and (4) are empty. RETURN The result tree for the operation if a success 0 - otherwise */ static SEL_TREE * tree_or(RANGE_OPT_PARAM *param,SEL_TREE *tree1,SEL_TREE *tree2) { DBUG_ENTER("tree_or"); if (!tree1 || !tree2) DBUG_RETURN(0); if (tree1->type == SEL_TREE::IMPOSSIBLE || tree2->type == SEL_TREE::ALWAYS) DBUG_RETURN(tree2); if (tree2->type == SEL_TREE::IMPOSSIBLE || tree1->type == SEL_TREE::ALWAYS) DBUG_RETURN(tree1); if (tree1->type == SEL_TREE::MAYBE) DBUG_RETURN(tree1); // Can't use this if (tree2->type == SEL_TREE::MAYBE) DBUG_RETURN(tree2); SEL_TREE *result= NULL; key_map result_keys; key_map ored_keys; SEL_TREE *rtree[2]= {NULL,NULL}; SEL_IMERGE *imerge[2]= {NULL, NULL}; bool no_ranges1= tree1->without_ranges(); bool no_ranges2= tree2->without_ranges(); bool no_merges1= tree1->without_imerges(); bool no_merges2= tree2->without_imerges(); if (!no_ranges1 && !no_merges2) { rtree[0]= new SEL_TREE(tree1, TRUE, param); imerge[1]= new SEL_IMERGE(tree2->merges.head(), 0, param); } if (!no_ranges2 && !no_merges1) { rtree[1]= new SEL_TREE(tree2, TRUE, param); imerge[0]= new SEL_IMERGE(tree1->merges.head(), 0, param); } bool no_imerge_from_ranges= FALSE; if (!(result= new SEL_TREE())) DBUG_RETURN(result); /* Build the range part of the tree for the formula (1) */ if (sel_trees_can_be_ored(param, tree1, tree2, &ored_keys)) { bool must_be_ored= sel_trees_must_be_ored(param, tree1, tree2, ored_keys); no_imerge_from_ranges= must_be_ored; key_map::Iterator it(ored_keys); int key_no; while ((key_no= it++) != key_map::Iterator::BITMAP_END) { SEL_ARG *key1= tree1->keys[key_no]; SEL_ARG *key2= tree2->keys[key_no]; if (!must_be_ored) { key1->incr_refs(); key2->incr_refs(); } if ((result->keys[key_no]= key_or(param, key1, key2))) result->keys_map.set_bit(key_no); } result->type= tree1->type; } if (no_imerge_from_ranges && no_merges1 && no_merges2) { if (result->keys_map.is_clear_all()) result->type= SEL_TREE::ALWAYS; DBUG_RETURN(result); } SEL_IMERGE *imerge_from_ranges; if (!(imerge_from_ranges= new SEL_IMERGE())) result= NULL; else if (!no_ranges1 && !no_ranges2 && !no_imerge_from_ranges) { /* Build the imerge part of the tree for the formula (1) */ SEL_TREE *rt1= tree1; SEL_TREE *rt2= tree2; if (no_merges1) rt1= new SEL_TREE(tree1, TRUE, param); if (no_merges2) rt2= new SEL_TREE(tree2, TRUE, param); if (!rt1 || !rt2 || result->merges.push_back(imerge_from_ranges) || imerge_from_ranges->or_sel_tree(param, rt1) || imerge_from_ranges->or_sel_tree(param, rt2)) result= NULL; } if (!result) DBUG_RETURN(result); result->type= tree1->type; if (!no_merges1 && !no_merges2 && !imerge_list_or_list(param, &tree1->merges, &tree2->merges)) { /* Build the imerges for the formula (2) */ imerge_list_and_list(&result->merges, &tree1->merges); } /* Build the imerges for the formulas (3) and (4) */ for (uint i=0; i < 2; i++) { List merges; SEL_TREE *rt= rtree[i]; SEL_IMERGE *im= imerge[1-i]; if (rt && im && !merges.push_back(im) && !imerge_list_or_tree(param, &merges, rt)) imerge_list_and_list(&result->merges, &merges); } DBUG_RETURN(result); } /* And key trees where key1->part < key2 -> part */ static SEL_ARG * and_all_keys(RANGE_OPT_PARAM *param, SEL_ARG *key1, SEL_ARG *key2, uint clone_flag) { SEL_ARG *next; ulong use_count=key1->use_count; if (key1->elements != 1) { key2->use_count+=key1->elements-1; //psergey: why we don't count that key1 has n-k-p? key2->increment_use_count((int) key1->elements-1); } if (key1->type == SEL_ARG::MAYBE_KEY) { key1->right= key1->left= &null_element; key1->next= key1->prev= 0; } for (next=key1->first(); next ; next=next->next) { if (next->next_key_part) { SEL_ARG *tmp= key_and(param, next->next_key_part, key2, clone_flag); if (tmp && tmp->type == SEL_ARG::IMPOSSIBLE) { key1=key1->tree_delete(next); continue; } next->next_key_part=tmp; if (use_count) next->increment_use_count(use_count); if (param->alloced_sel_args > SEL_ARG::MAX_SEL_ARGS) break; } else next->next_key_part=key2; } if (!key1) return &null_element; // Impossible ranges key1->use_count++; key1->max_part_no= max(key2->max_part_no, key2->part+1); return key1; } /* Produce a SEL_ARG graph that represents "key1 AND key2" SYNOPSIS key_and() param Range analysis context (needed to track if we have allocated too many SEL_ARGs) key1 First argument, root of its RB-tree key2 Second argument, root of its RB-tree RETURN RB-tree root of the resulting SEL_ARG graph. NULL if the result of AND operation is an empty interval {0}. */ static SEL_ARG * key_and(RANGE_OPT_PARAM *param, SEL_ARG *key1, SEL_ARG *key2, uint clone_flag) { if (!key1) return key2; if (!key2) return key1; if (key1->part != key2->part) { if (key1->part > key2->part) { swap_variables(SEL_ARG *, key1, key2); clone_flag=swap_clone_flag(clone_flag); } // key1->part < key2->part key1->use_count--; if (key1->use_count > 0) if (!(key1= key1->clone_tree(param))) return 0; // OOM return and_all_keys(param, key1, key2, clone_flag); } if (((clone_flag & CLONE_KEY2_MAYBE) && !(clone_flag & CLONE_KEY1_MAYBE) && key2->type != SEL_ARG::MAYBE_KEY) || key1->type == SEL_ARG::MAYBE_KEY) { // Put simple key in key2 swap_variables(SEL_ARG *, key1, key2); clone_flag=swap_clone_flag(clone_flag); } /* If one of the key is MAYBE_KEY then the found region may be smaller */ if (key2->type == SEL_ARG::MAYBE_KEY) { if (key1->use_count > 1) { key1->use_count--; if (!(key1=key1->clone_tree(param))) return 0; // OOM key1->use_count++; } if (key1->type == SEL_ARG::MAYBE_KEY) { // Both are maybe key key1->next_key_part=key_and(param, key1->next_key_part, key2->next_key_part, clone_flag); if (key1->next_key_part && key1->next_key_part->type == SEL_ARG::IMPOSSIBLE) return key1; } else { key1->maybe_smaller(); if (key2->next_key_part) { key1->use_count--; // Incremented in and_all_keys return and_all_keys(param, key1, key2, clone_flag); } key2->use_count--; // Key2 doesn't have a tree } return key1; } if ((key1->min_flag | key2->min_flag) & GEOM_FLAG) { /* TODO: why not leave one of the trees? */ key1->free_tree(); key2->free_tree(); return 0; // Can't optimize this } key1->use_count--; key2->use_count--; SEL_ARG *e1=key1->first(), *e2=key2->first(), *new_tree=0; uint max_part_no= max(key1->max_part_no, key2->max_part_no); while (e1 && e2) { int cmp=e1->cmp_min_to_min(e2); if (cmp < 0) { if (get_range(&e1,&e2,key1)) continue; } else if (get_range(&e2,&e1,key2)) continue; SEL_ARG *next=key_and(param, e1->next_key_part, e2->next_key_part, clone_flag); e1->incr_refs(); e2->incr_refs(); if (!next || next->type != SEL_ARG::IMPOSSIBLE) { SEL_ARG *new_arg= e1->clone_and(e2); if (!new_arg) return &null_element; // End of memory new_arg->next_key_part=next; if (!new_tree) { new_tree=new_arg; } else new_tree=new_tree->insert(new_arg); } if (e1->cmp_max_to_max(e2) < 0) e1=e1->next; // e1 can't overlapp next e2 else e2=e2->next; } key1->free_tree(); key2->free_tree(); if (!new_tree) return &null_element; // Impossible range new_tree->max_part_no= max_part_no; return new_tree; } static bool get_range(SEL_ARG **e1,SEL_ARG **e2,SEL_ARG *root1) { (*e1)=root1->find_range(*e2); // first e1->min < e2->min if ((*e1)->cmp_max_to_min(*e2) < 0) { if (!((*e1)=(*e1)->next)) return 1; if ((*e1)->cmp_min_to_max(*e2) > 0) { (*e2)=(*e2)->next; return 1; } } return 0; } /** Combine two range expression under a common OR. On a logical level, the transformation is key_or( expr1, expr2 ) => expr1 OR expr2. Both expressions are assumed to be in the SEL_ARG format. In a logic sense, theformat is reminiscent of DNF, since an expression such as the following ( 1 < kp1 < 10 AND p1 ) OR ( 10 <= kp2 < 20 AND p2 ) where there is a key consisting of keyparts ( kp1, kp2, ..., kpn ) and p1 and p2 are valid SEL_ARG expressions over keyparts kp2 ... kpn, is a valid SEL_ARG condition. The disjuncts appear ordered by the minimum endpoint of the first range and ranges must not overlap. It follows that they are also ordered by maximum endpoints. Thus ( 1 < kp1 <= 2 AND ( kp2 = 2 OR kp2 = 3 ) ) OR kp1 = 3 Is a a valid SER_ARG expression for a key of at least 2 keyparts. For simplicity, we will assume that expr2 is a single range predicate, i.e. on the form ( a < x < b AND ... ). It is easy to generalize to a disjunction of several predicates by subsequently call key_or for each disjunct. The algorithm iterates over each disjunct of expr1, and for each disjunct where the first keypart's range overlaps with the first keypart's range in expr2: If the predicates are equal for the rest of the keyparts, or if there are no more, the range in expr2 has its endpoints copied in, and the SEL_ARG node in expr2 is deallocated. If more ranges became connected in expr1, the surplus is also dealocated. If they differ, two ranges are created. - The range leading up to the overlap. Empty if endpoints are equal. - The overlapping sub-range. May be the entire range if they are equal. Finally, there may be one more range if expr2's first keypart's range has a greater maximum endpoint than the last range in expr1. For the overlapping sub-range, we recursively call key_or. Thus in order to compute key_or of (1) ( 1 < kp1 < 10 AND 1 < kp2 < 10 ) (2) ( 2 < kp1 < 20 AND 4 < kp2 < 20 ) We create the ranges 1 < kp <= 2, 2 < kp1 < 10, 10 <= kp1 < 20. For the first one, we simply hook on the condition for the second keypart from (1) : 1 < kp2 < 10. For the second range 2 < kp1 < 10, key_or( 1 < kp2 < 10, 4 < kp2 < 20 ) is called, yielding 1 < kp2 < 20. For the last range, we reuse the range 4 < kp2 < 20 from (2) for the second keypart. The result is thus ( 1 < kp1 <= 2 AND 1 < kp2 < 10 ) OR ( 2 < kp1 < 10 AND 1 < kp2 < 20 ) OR ( 10 <= kp1 < 20 AND 4 < kp2 < 20 ) */ static SEL_ARG * key_or(RANGE_OPT_PARAM *param, SEL_ARG *key1,SEL_ARG *key2) { if (!key1) { if (key2) { key2->use_count--; key2->free_tree(); } return 0; } if (!key2) { key1->use_count--; key1->free_tree(); return 0; } key1->use_count--; key2->use_count--; if (key1->part != key2->part || (key1->min_flag | key2->min_flag) & GEOM_FLAG) { key1->free_tree(); key2->free_tree(); return 0; // Can't optimize this } // If one of the key is MAYBE_KEY then the found region may be bigger if (key1->type == SEL_ARG::MAYBE_KEY) { key2->free_tree(); key1->use_count++; return key1; } if (key2->type == SEL_ARG::MAYBE_KEY) { key1->free_tree(); key2->use_count++; return key2; } if (key1->use_count > 0) { if (key2->use_count == 0 || key1->elements > key2->elements) { swap_variables(SEL_ARG *,key1,key2); } if (key1->use_count > 0 && !(key1=key1->clone_tree(param))) return 0; // OOM } // Add tree at key2 to tree at key1 bool key2_shared=key2->use_count != 0; key1->maybe_flag|=key2->maybe_flag; /* Notation for illustrations used in the rest of this function: Range: [--------] ^ ^ start stop Two overlapping ranges: [-----] [----] [--] [---] or [---] or [-------] Ambiguity: *** The range starts or stops somewhere in the "***" range. Example: a starts before b and may end before/the same plase/after b a: [----***] b: [---] Adjacent ranges: Ranges that meet but do not overlap. Example: a = "x < 3", b = "x >= 3" a: ----] b: [---- */ uint max_part_no= max(key1->max_part_no, key2->max_part_no); for (key2=key2->first(); key2; ) { /* key1 consists of one or more ranges. tmp is the range currently being handled. initialize tmp to the latest range in key1 that starts the same place or before the range in key2 starts key2: [------] key1: [---] [-----] [----] ^ tmp */ SEL_ARG *tmp=key1->find_range(key2); /* Used to describe how two key values are positioned compared to each other. Consider key_value_a.(key_value_b): -2: key_value_a is smaller than key_value_b, and they are adjacent -1: key_value_a is smaller than key_value_b (not adjacent) 0: the key values are equal 1: key_value_a is bigger than key_value_b (not adjacent) -2: key_value_a is bigger than key_value_b, and they are adjacent Example: "cmp= tmp->cmp_max_to_min(key2)" key2: [-------- (10 <= x ...) tmp: -----] (... x < 10) => cmp==-2 tmp: ----] (... x <= 9) => cmp==-1 tmp: ------] (... x = 10) => cmp== 0 tmp: --------] (... x <= 12) => cmp== 1 (cmp == 2 does not make sense for cmp_max_to_min()) */ int cmp= 0; if (!tmp) { /* The range in key2 starts before the first range in key1. Use the first range in key1 as tmp. key2: [--------] key1: [****--] [----] [-------] ^ tmp */ tmp=key1->first(); cmp= -1; } else if ((cmp= tmp->cmp_max_to_min(key2)) < 0) { /* This is the case: key2: [-------] tmp: [----**] */ SEL_ARG *next=tmp->next; if (cmp == -2 && eq_tree(tmp->next_key_part,key2->next_key_part)) { /* Adjacent (cmp==-2) and equal next_key_parts => ranges can be merged This is the case: key2: [-------] tmp: [----] Result: key2: [-------------] => inserted into key1 below tmp: => deleted */ SEL_ARG *key2_next=key2->next; if (key2_shared) { if (!(key2=new SEL_ARG(*key2))) return 0; // out of memory key2->increment_use_count(key1->use_count+1); key2->next=key2_next; // New copy of key2 } key2->copy_min(tmp); if (!(key1=key1->tree_delete(tmp))) { // Only one key in tree key1=key2; key1->make_root(); key2=key2_next; break; } } if (!(tmp=next)) // Move to next range in key1. Now tmp.min > key2.min break; // No more ranges in key1. Copy rest of key2 } if (cmp < 0) { /* This is the case: key2: [--***] tmp: [----] */ int tmp_cmp; if ((tmp_cmp=tmp->cmp_min_to_max(key2)) > 0) { /* This is the case: key2: [------**] tmp: [----] */ if (tmp_cmp == 2 && eq_tree(tmp->next_key_part,key2->next_key_part)) { /* Adjacent ranges with equal next_key_part. Merge like this: This is the case: key2: [------] tmp: [-----] Result: key2: [------] tmp: [-------------] Then move on to next key2 range. */ tmp->copy_min_to_min(key2); key1->merge_flags(key2); if (tmp->min_flag & NO_MIN_RANGE && tmp->max_flag & NO_MAX_RANGE) { if (key1->maybe_flag) return new SEL_ARG(SEL_ARG::MAYBE_KEY); return 0; } key2->increment_use_count(-1); // Free not used tree key2=key2->next; continue; } else { /* key2 not adjacent to tmp or has different next_key_part. Insert into key1 and move to next range in key2 This is the case: key2: [------**] tmp: [----] Result: key1_ [------**][----] ^ ^ insert tmp */ SEL_ARG *next=key2->next; if (key2_shared) { SEL_ARG *cpy= new SEL_ARG(*key2); // Must make copy if (!cpy) return 0; // OOM key1=key1->insert(cpy); key2->increment_use_count(key1->use_count+1); } else key1=key1->insert(key2); // Will destroy key2_root key2=next; continue; } } } /* The ranges in tmp and key2 are overlapping: key2: [----------] tmp: [*****-----*****] Corollary: tmp.min <= key2.max */ if (eq_tree(tmp->next_key_part,key2->next_key_part)) { // Merge overlapping ranges with equal next_key_part if (tmp->is_same(key2)) { /* Found exact match of key2 inside key1. Use the relevant range in key1. */ tmp->merge_flags(key2); // Copy maybe flags key2->increment_use_count(-1); // Free not used tree } else { SEL_ARG *last= tmp; SEL_ARG *first= tmp; /* Find the last range in key1 that overlaps key2 and where all ranges first...last have the same next_key_part as key2. key2: [****----------------------*******] key1: [--] [----] [---] [-----] [xxxx] ^ ^ ^ first last different next_key_part Since key2 covers them, the ranges between first and last are merged into one range by deleting first...last-1 from the key1 tree. In the figure, this applies to first and the two consecutive ranges. The range of last is then extended: * last.min: Set to min(key2.min, first.min) * last.max: If there is a last->next that overlaps key2 (i.e., last->next has a different next_key_part): Set adjacent to last->next.min Otherwise: Set to max(key2.max, last.max) Result: key2: [****----------------------*******] [--] [----] [---] => deleted from key1 key1: [**------------------------***][xxxx] ^ ^ tmp=last different next_key_part */ while (last->next && last->next->cmp_min_to_max(key2) <= 0 && eq_tree(last->next->next_key_part,key2->next_key_part)) { /* last->next is covered by key2 and has same next_key_part. last can be deleted */ SEL_ARG *save=last; last=last->next; key1=key1->tree_delete(save); } // Redirect tmp to last which will cover the entire range tmp= last; /* We need the minimum endpoint of first so we can compare it with the minimum endpoint of the enclosing key2 range. */ last->copy_min(first); bool full_range= last->copy_min(key2); if (!full_range) { if (last->next && key2->cmp_max_to_min(last->next) >= 0) { /* This is the case: key2: [-------------] key1: [***------] [xxxx] ^ ^ last different next_key_part Extend range of last up to last->next: key2: [-------------] key1: [***--------][xxxx] */ last->copy_min_to_max(last->next); } else /* This is the case: key2: [--------*****] key1: [***---------] [xxxx] ^ ^ last different next_key_part Extend range of last up to max(last.max, key2.max): key2: [--------*****] key1: [***----------**] [xxxx] */ full_range= last->copy_max(key2); } if (full_range) { // Full range key1->free_tree(); for (; key2 ; key2=key2->next) key2->increment_use_count(-1); // Free not used tree if (key1->maybe_flag) return new SEL_ARG(SEL_ARG::MAYBE_KEY); return 0; } } } if (cmp >= 0 && tmp->cmp_min_to_min(key2) < 0) { /* This is the case ("cmp>=0" means that tmp.max >= key2.min): key2: [----] tmp: [------------*****] */ if (!tmp->next_key_part) { /* tmp->next_key_part is empty: cut the range that is covered by tmp from key2. Reason: (key2->next_key_part OR tmp->next_key_part) will be empty and therefore equal to tmp->next_key_part. Thus, this part of the key2 range is completely covered by tmp. */ if (tmp->cmp_max_to_max(key2) >= 0) { /* tmp covers the entire range in key2. key2: [----] tmp: [-----------------] Move on to next range in key2 */ key2->increment_use_count(-1); // Free not used tree key2=key2->next; continue; } else { /* This is the case: key2: [-------] tmp: [---------] Result: key2: [---] tmp: [---------] */ if (key2->use_count) { SEL_ARG *key2_cpy= new SEL_ARG(*key2); if (key2_cpy) return 0; key2= key2_cpy; } key2->copy_max_to_min(tmp); continue; } } /* The ranges are overlapping but have not been merged because next_key_part of tmp and key2 differ. key2: [----] tmp: [------------*****] Split tmp in two where key2 starts: key2: [----] key1: [--------][--*****] ^ ^ insert tmp */ SEL_ARG *new_arg=tmp->clone_first(key2); if (!new_arg) return 0; // OOM if ((new_arg->next_key_part= tmp->next_key_part)) new_arg->increment_use_count(key1->use_count+1); tmp->copy_min_to_min(key2); key1=key1->insert(new_arg); } // tmp.min >= key2.min due to this if() /* Now key2.min <= tmp.min <= key2.max: key2: [---------] tmp: [****---*****] */ SEL_ARG key2_cpy(*key2); // Get copy we can modify for (;;) { if (tmp->cmp_min_to_min(&key2_cpy) > 0) { /* This is the case: key2_cpy: [------------] key1: [-*****] ^ tmp Result: key2_cpy: [---] key1: [-------][-*****] ^ ^ insert tmp */ SEL_ARG *new_arg=key2_cpy.clone_first(tmp); if (!new_arg) return 0; // OOM if ((new_arg->next_key_part=key2_cpy.next_key_part)) new_arg->increment_use_count(key1->use_count+1); key1=key1->insert(new_arg); key2_cpy.copy_min_to_min(tmp); } // Now key2_cpy.min == tmp.min if ((cmp= tmp->cmp_max_to_max(&key2_cpy)) <= 0) { /* tmp.max <= key2_cpy.max: key2_cpy: a) [-------] or b) [----] tmp: [----] [----] Steps: 1) Update next_key_part of tmp: OR it with key2_cpy->next_key_part. 2) If case a: Insert range [tmp.max, key2_cpy.max] into key1 using next_key_part of key2_cpy Result: key1: a) [----][-] or b) [----] */ tmp->maybe_flag|= key2_cpy.maybe_flag; key2_cpy.increment_use_count(key1->use_count+1); tmp->next_key_part= key_or(param, tmp->next_key_part, key2_cpy.next_key_part); if (!cmp) break; // case b: done with this key2 range // Make key2_cpy the range [tmp.max, key2_cpy.max] key2_cpy.copy_max_to_min(tmp); if (!(tmp=tmp->next)) { /* No more ranges in key1. Insert key2_cpy and go to "end" label to insert remaining ranges in key2 if any. */ SEL_ARG *tmp2= new SEL_ARG(key2_cpy); if (!tmp2) return 0; // OOM key1=key1->insert(tmp2); key2=key2->next; goto end; } if (tmp->cmp_min_to_max(&key2_cpy) > 0) { /* The next range in key1 does not overlap with key2_cpy. Insert this range into key1 and move on to the next range in key2. */ SEL_ARG *tmp2= new SEL_ARG(key2_cpy); if (!tmp2) return 0; // OOM key1=key1->insert(tmp2); break; } /* key2_cpy overlaps with the next range in key1 and the case is now "key2.min <= tmp.min <= key2.max". Go back to for(;;) to handle this situation. */ continue; } else { /* This is the case: key2_cpy: [-------] tmp: [------------] Result: key1: [-------][---] ^ ^ new_arg tmp Steps: 0) If tmp->next_key_part is empty: do nothing. Reason: (key2_cpy->next_key_part OR tmp->next_key_part) will be empty and therefore equal to tmp->next_key_part. Thus, the range in key2_cpy is completely covered by tmp 1) Make new_arg with range [tmp.min, key2_cpy.max]. new_arg->next_key_part is OR between next_key_part of tmp and key2_cpy 2) Make tmp the range [key2.max, tmp.max] 3) Insert new_arg into key1 */ if (!tmp->next_key_part) // Step 0 { key2_cpy.increment_use_count(-1); // Free not used tree break; } SEL_ARG *new_arg=tmp->clone_last(&key2_cpy); if (!new_arg) return 0; // OOM tmp->copy_max_to_min(&key2_cpy); tmp->increment_use_count(key1->use_count+1); /* Increment key count as it may be used for next loop */ key2_cpy.increment_use_count(1); new_arg->next_key_part= key_or(param, tmp->next_key_part, key2_cpy.next_key_part); key1=key1->insert(new_arg); break; } } // Move on to next range in key2 key2=key2->next; } end: /* Add key2 ranges that are non-overlapping with and higher than the highest range in key1. */ while (key2) { SEL_ARG *next=key2->next; if (key2_shared) { SEL_ARG *tmp=new SEL_ARG(*key2); // Must make copy if (!tmp) return 0; key2->increment_use_count(key1->use_count+1); key1=key1->insert(tmp); } else key1=key1->insert(key2); // Will destroy key2_root key2=next; } key1->use_count++; key1->max_part_no= max_part_no; return key1; } /* Compare if two trees are equal */ static bool eq_tree(SEL_ARG* a,SEL_ARG *b) { if (a == b) return 1; if (!a || !b || !a->is_same(b)) return 0; if (a->left != &null_element && b->left != &null_element) { if (!eq_tree(a->left,b->left)) return 0; } else if (a->left != &null_element || b->left != &null_element) return 0; if (a->right != &null_element && b->right != &null_element) { if (!eq_tree(a->right,b->right)) return 0; } else if (a->right != &null_element || b->right != &null_element) return 0; if (a->next_key_part != b->next_key_part) { // Sub range if (!a->next_key_part != !b->next_key_part || !eq_tree(a->next_key_part, b->next_key_part)) return 0; } return 1; } SEL_ARG * SEL_ARG::insert(SEL_ARG *key) { SEL_ARG *element,**UNINIT_VAR(par),*UNINIT_VAR(last_element); for (element= this; element != &null_element ; ) { last_element=element; if (key->cmp_min_to_min(element) > 0) { par= &element->right; element= element->right; } else { par = &element->left; element= element->left; } } *par=key; key->parent=last_element; /* Link in list */ if (par == &last_element->left) { key->next=last_element; if ((key->prev=last_element->prev)) key->prev->next=key; last_element->prev=key; } else { if ((key->next=last_element->next)) key->next->prev=key; key->prev=last_element; last_element->next=key; } key->left=key->right= &null_element; SEL_ARG *root=rb_insert(key); // rebalance tree root->use_count=this->use_count; // copy root info root->elements= this->elements+1; root->maybe_flag=this->maybe_flag; return root; } /* ** Find best key with min <= given key ** Because the call context this should never return 0 to get_range */ SEL_ARG * SEL_ARG::find_range(SEL_ARG *key) { SEL_ARG *element=this,*found=0; for (;;) { if (element == &null_element) return found; int cmp=element->cmp_min_to_min(key); if (cmp == 0) return element; if (cmp < 0) { found=element; element=element->right; } else element=element->left; } } /* Remove a element from the tree SYNOPSIS tree_delete() key Key that is to be deleted from tree (this) NOTE This also frees all sub trees that is used by the element RETURN root of new tree (with key deleted) */ SEL_ARG * SEL_ARG::tree_delete(SEL_ARG *key) { enum leaf_color remove_color; SEL_ARG *root,*nod,**par,*fix_par; DBUG_ENTER("tree_delete"); root=this; this->parent= 0; /* Unlink from list */ if (key->prev) key->prev->next=key->next; if (key->next) key->next->prev=key->prev; key->increment_use_count(-1); if (!key->parent) par= &root; else par=key->parent_ptr(); if (key->left == &null_element) { *par=nod=key->right; fix_par=key->parent; if (nod != &null_element) nod->parent=fix_par; remove_color= key->color; } else if (key->right == &null_element) { *par= nod=key->left; nod->parent=fix_par=key->parent; remove_color= key->color; } else { SEL_ARG *tmp=key->next; // next bigger key (exist!) nod= *tmp->parent_ptr()= tmp->right; // unlink tmp from tree fix_par=tmp->parent; if (nod != &null_element) nod->parent=fix_par; remove_color= tmp->color; tmp->parent=key->parent; // Move node in place of key (tmp->left=key->left)->parent=tmp; if ((tmp->right=key->right) != &null_element) tmp->right->parent=tmp; tmp->color=key->color; *par=tmp; if (fix_par == key) // key->right == key->next fix_par=tmp; // new parent of nod } if (root == &null_element) DBUG_RETURN(0); // Maybe root later if (remove_color == BLACK) root=rb_delete_fixup(root,nod,fix_par); test_rb_tree(root,root->parent); root->use_count=this->use_count; // Fix root counters root->elements=this->elements-1; root->maybe_flag=this->maybe_flag; DBUG_RETURN(root); } /* Functions to fix up the tree after insert and delete */ static void left_rotate(SEL_ARG **root,SEL_ARG *leaf) { SEL_ARG *y=leaf->right; leaf->right=y->left; if (y->left != &null_element) y->left->parent=leaf; if (!(y->parent=leaf->parent)) *root=y; else *leaf->parent_ptr()=y; y->left=leaf; leaf->parent=y; } static void right_rotate(SEL_ARG **root,SEL_ARG *leaf) { SEL_ARG *y=leaf->left; leaf->left=y->right; if (y->right != &null_element) y->right->parent=leaf; if (!(y->parent=leaf->parent)) *root=y; else *leaf->parent_ptr()=y; y->right=leaf; leaf->parent=y; } SEL_ARG * SEL_ARG::rb_insert(SEL_ARG *leaf) { SEL_ARG *y,*par,*par2,*root; root= this; root->parent= 0; leaf->color=RED; while (leaf != root && (par= leaf->parent)->color == RED) { // This can't be root or 1 level under if (par == (par2= leaf->parent->parent)->left) { y= par2->right; if (y->color == RED) { par->color=BLACK; y->color=BLACK; leaf=par2; leaf->color=RED; /* And the loop continues */ } else { if (leaf == par->right) { left_rotate(&root,leaf->parent); par=leaf; /* leaf is now parent to old leaf */ } par->color=BLACK; par2->color=RED; right_rotate(&root,par2); break; } } else { y= par2->left; if (y->color == RED) { par->color=BLACK; y->color=BLACK; leaf=par2; leaf->color=RED; /* And the loop continues */ } else { if (leaf == par->left) { right_rotate(&root,par); par=leaf; } par->color=BLACK; par2->color=RED; left_rotate(&root,par2); break; } } } root->color=BLACK; test_rb_tree(root,root->parent); return root; } SEL_ARG *rb_delete_fixup(SEL_ARG *root,SEL_ARG *key,SEL_ARG *par) { SEL_ARG *x,*w; root->parent=0; x= key; while (x != root && x->color == SEL_ARG::BLACK) { if (x == par->left) { w=par->right; if (w->color == SEL_ARG::RED) { w->color=SEL_ARG::BLACK; par->color=SEL_ARG::RED; left_rotate(&root,par); w=par->right; } if (w->left->color == SEL_ARG::BLACK && w->right->color == SEL_ARG::BLACK) { w->color=SEL_ARG::RED; x=par; } else { if (w->right->color == SEL_ARG::BLACK) { w->left->color=SEL_ARG::BLACK; w->color=SEL_ARG::RED; right_rotate(&root,w); w=par->right; } w->color=par->color; par->color=SEL_ARG::BLACK; w->right->color=SEL_ARG::BLACK; left_rotate(&root,par); x=root; break; } } else { w=par->left; if (w->color == SEL_ARG::RED) { w->color=SEL_ARG::BLACK; par->color=SEL_ARG::RED; right_rotate(&root,par); w=par->left; } if (w->right->color == SEL_ARG::BLACK && w->left->color == SEL_ARG::BLACK) { w->color=SEL_ARG::RED; x=par; } else { if (w->left->color == SEL_ARG::BLACK) { w->right->color=SEL_ARG::BLACK; w->color=SEL_ARG::RED; left_rotate(&root,w); w=par->left; } w->color=par->color; par->color=SEL_ARG::BLACK; w->left->color=SEL_ARG::BLACK; right_rotate(&root,par); x=root; break; } } par=x->parent; } x->color=SEL_ARG::BLACK; return root; } /* Test that the properties for a red-black tree hold */ #ifdef EXTRA_DEBUG int test_rb_tree(SEL_ARG *element,SEL_ARG *parent) { int count_l,count_r; if (element == &null_element) return 0; // Found end of tree if (element->parent != parent) { sql_print_error("Wrong tree: Parent doesn't point at parent"); return -1; } if (element->color == SEL_ARG::RED && (element->left->color == SEL_ARG::RED || element->right->color == SEL_ARG::RED)) { sql_print_error("Wrong tree: Found two red in a row"); return -1; } if (element->left == element->right && element->left != &null_element) { // Dummy test sql_print_error("Wrong tree: Found right == left"); return -1; } count_l=test_rb_tree(element->left,element); count_r=test_rb_tree(element->right,element); if (count_l >= 0 && count_r >= 0) { if (count_l == count_r) return count_l+(element->color == SEL_ARG::BLACK); sql_print_error("Wrong tree: Incorrect black-count: %d - %d", count_l,count_r); } return -1; // Error, no more warnings } /** Count how many times SEL_ARG graph "root" refers to its part "key" via transitive closure. @param root An RB-Root node in a SEL_ARG graph. @param key Another RB-Root node in that SEL_ARG graph. The passed "root" node may refer to "key" node via root->next_key_part, root->next->n This function counts how many times the node "key" is referred (via SEL_ARG::next_key_part) by - intervals of RB-tree pointed by "root", - intervals of RB-trees that are pointed by SEL_ARG::next_key_part from intervals of RB-tree pointed by "root", - and so on. Here is an example (horizontal links represent next_key_part pointers, vertical links - next/prev prev pointers): +----+ $ |root|-----------------+ +----+ $ | | $ | | $ | +----+ +---+ $ | +---+ Here the return value | |- ... -| |---$-+--+->|key| will be 4. +----+ +---+ $ | | +---+ | $ | | ... $ | | | $ | | +----+ +---+ $ | | | |---| |---------+ | +----+ +---+ $ | | | $ | ... +---+ $ | | |------------+ +---+ $ @return Number of links to "key" from nodes reachable from "root". */ static ulong count_key_part_usage(SEL_ARG *root, SEL_ARG *key) { ulong count= 0; for (root=root->first(); root ; root=root->next) { if (root->next_key_part) { if (root->next_key_part == key) count++; if (root->next_key_part->part < key->part) count+=count_key_part_usage(root->next_key_part,key); } } return count; } /* Check if SEL_ARG::use_count value is correct SYNOPSIS SEL_ARG::test_use_count() root The root node of the SEL_ARG graph (an RB-tree root node that has the least value of sel_arg->part in the entire graph, and thus is the "origin" of the graph) DESCRIPTION Check if SEL_ARG::use_count value is correct. See the definition of use_count for what is "correct". */ void SEL_ARG::test_use_count(SEL_ARG *root) { uint e_count=0; if (this->type != SEL_ARG::KEY_RANGE) return; for (SEL_ARG *pos=first(); pos ; pos=pos->next) { e_count++; if (pos->next_key_part) { ulong count=count_key_part_usage(root,pos->next_key_part); if (count > pos->next_key_part->use_count) { sql_print_information("Use_count: Wrong count for key at 0x%lx, %lu " "should be %lu", (long unsigned int)pos, pos->next_key_part->use_count, count); return; } pos->next_key_part->test_use_count(root); } } if (e_count != elements) sql_print_warning("Wrong use count: %u (should be %u) for tree at 0x%lx", e_count, elements, (long unsigned int) this); } #endif /* Calculate cost and E(#rows) for a given index and intervals tree SYNOPSIS check_quick_select() param Parameter from test_quick_select idx Number of index to use in PARAM::key SEL_TREE::key index_only TRUE - assume only index tuples will be accessed FALSE - assume full table rows will be read tree Transformed selection condition, tree->key[idx] holds the intervals for the given index. update_tbl_stats TRUE <=> update table->quick_* with information about range scan we've evaluated. mrr_flags INOUT MRR access flags cost OUT Scan cost NOTES param->is_ror_scan is set to reflect if the key scan is a ROR (see is_key_scan_ror function for more info) param->table->quick_*, param->range_count (and maybe others) are updated with data of given key scan, see quick_range_seq_next for details. RETURN Estimate # of records to be retrieved. HA_POS_ERROR if estimate calculation failed due to table handler problems. */ static ha_rows check_quick_select(PARAM *param, uint idx, bool index_only, SEL_ARG *tree, bool update_tbl_stats, uint *mrr_flags, uint *bufsize, COST_VECT *cost) { SEL_ARG_RANGE_SEQ seq; RANGE_SEQ_IF seq_if = {NULL, sel_arg_range_seq_init, sel_arg_range_seq_next, 0, 0}; handler *file= param->table->file; ha_rows rows= HA_POS_ERROR; uint keynr= param->real_keynr[idx]; DBUG_ENTER("check_quick_select"); /* Handle cases when we don't have a valid non-empty list of range */ if (!tree) DBUG_RETURN(HA_POS_ERROR); if (tree->type == SEL_ARG::IMPOSSIBLE) DBUG_RETURN(0L); if (tree->type != SEL_ARG::KEY_RANGE || tree->part != 0) DBUG_RETURN(HA_POS_ERROR); seq.keyno= idx; seq.real_keyno= keynr; seq.param= param; seq.start= tree; param->range_count=0; param->max_key_part=0; param->is_ror_scan= TRUE; if (file->index_flags(keynr, 0, TRUE) & HA_KEY_SCAN_NOT_ROR) param->is_ror_scan= FALSE; *mrr_flags= param->force_default_mrr? HA_MRR_USE_DEFAULT_IMPL: 0; /* Pass HA_MRR_SORTED to see if MRR implementation can handle sorting. */ *mrr_flags|= HA_MRR_NO_ASSOCIATION | HA_MRR_SORTED; bool pk_is_clustered= file->primary_key_is_clustered(); if (index_only && (file->index_flags(keynr, param->max_key_part, 1) & HA_KEYREAD_ONLY) && !(file->index_flags(keynr, param->max_key_part, 1) & HA_CLUSTERED_INDEX)) *mrr_flags |= HA_MRR_INDEX_ONLY; if (param->thd->lex->sql_command != SQLCOM_SELECT) *mrr_flags |= HA_MRR_USE_DEFAULT_IMPL; *bufsize= param->thd->variables.mrr_buff_size; /* Skip materialized derived table/view result table from MRR check as they aren't contain any data yet. */ if (param->table->pos_in_table_list->is_non_derived()) rows= file->multi_range_read_info_const(keynr, &seq_if, (void*)&seq, 0, bufsize, mrr_flags, cost); if (rows != HA_POS_ERROR) { param->quick_rows[keynr]= rows; if (update_tbl_stats) { param->table->quick_keys.set_bit(keynr); param->table->quick_key_parts[keynr]= param->max_key_part+1; param->table->quick_n_ranges[keynr]= param->range_count; param->table->quick_condition_rows= min(param->table->quick_condition_rows, rows); param->table->quick_rows[keynr]= rows; } } /* Figure out if the key scan is ROR (returns rows in ROWID order) or not */ enum ha_key_alg key_alg= param->table->key_info[seq.real_keyno].algorithm; if ((key_alg != HA_KEY_ALG_BTREE) && (key_alg!= HA_KEY_ALG_UNDEF)) { /* All scans are non-ROR scans for those index types. TODO: Don't have this logic here, make table engines return appropriate flags instead. */ param->is_ror_scan= FALSE; } else if (param->table->s->primary_key == keynr && pk_is_clustered) { /* Clustered PK scan is always a ROR scan (TODO: same as above) */ param->is_ror_scan= TRUE; } else if (param->range_count > 1) { /* Scaning multiple key values in the index: the records are ROR for each value, but not between values. E.g, "SELECT ... x IN (1,3)" returns ROR order for all records with x=1, then ROR order for records with x=3 */ param->is_ror_scan= FALSE; } DBUG_PRINT("exit", ("Records: %lu", (ulong) rows)); DBUG_RETURN(rows); //psergey-merge:todo: maintain first_null_comp. } /* Check if key scan on given index with equality conditions on first n key parts is a ROR scan. SYNOPSIS is_key_scan_ror() param Parameter from test_quick_select keynr Number of key in the table. The key must not be a clustered primary key. nparts Number of first key parts for which equality conditions are present. NOTES ROR (Rowid Ordered Retrieval) key scan is a key scan that produces ordered sequence of rowids (ha_xxx::cmp_ref is the comparison function) This function is needed to handle a practically-important special case: an index scan is a ROR scan if it is done using a condition in form "key1_1=c_1 AND ... AND key1_n=c_n" where the index is defined on (key1_1, ..., key1_N [,a_1, ..., a_n]) and the table has a clustered Primary Key defined as PRIMARY KEY(a_1, ..., a_n, b1, ..., b_k) i.e. the first key parts of it are identical to uncovered parts ot the key being scanned. This function assumes that the index flags do not include HA_KEY_SCAN_NOT_ROR flag (that is checked elsewhere). Check (1) is made in quick_range_seq_next() RETURN TRUE The scan is ROR-scan FALSE Otherwise */ static bool is_key_scan_ror(PARAM *param, uint keynr, uint8 nparts) { KEY *table_key= param->table->key_info + keynr; KEY_PART_INFO *key_part= table_key->key_part + nparts; KEY_PART_INFO *key_part_end= (table_key->key_part + table_key->key_parts); uint pk_number; for (KEY_PART_INFO *kp= table_key->key_part; kp < key_part; kp++) { uint16 fieldnr= param->table->key_info[keynr]. key_part[kp - table_key->key_part].fieldnr - 1; if (param->table->field[fieldnr]->key_length() != kp->length) return FALSE; } if (key_part == key_part_end) return TRUE; key_part= table_key->key_part + nparts; pk_number= param->table->s->primary_key; if (!param->table->file->primary_key_is_clustered() || pk_number == MAX_KEY) return FALSE; KEY_PART_INFO *pk_part= param->table->key_info[pk_number].key_part; KEY_PART_INFO *pk_part_end= pk_part + param->table->key_info[pk_number].key_parts; for (;(key_part!=key_part_end) && (pk_part != pk_part_end); ++key_part, ++pk_part) { if ((key_part->field != pk_part->field) || (key_part->length != pk_part->length)) return FALSE; } return (key_part == key_part_end); } /* Create a QUICK_RANGE_SELECT from given key and SEL_ARG tree for that key. SYNOPSIS get_quick_select() param idx Index of used key in param->key. key_tree SEL_ARG tree for the used key mrr_flags MRR parameter for quick select mrr_buf_size MRR parameter for quick select parent_alloc If not NULL, use it to allocate memory for quick select data. Otherwise use quick->alloc. NOTES The caller must call QUICK_SELECT::init for returned quick select. CAUTION! This function may change thd->mem_root to a MEM_ROOT which will be deallocated when the returned quick select is deleted. RETURN NULL on error otherwise created quick select */ QUICK_RANGE_SELECT * get_quick_select(PARAM *param,uint idx,SEL_ARG *key_tree, uint mrr_flags, uint mrr_buf_size, MEM_ROOT *parent_alloc) { QUICK_RANGE_SELECT *quick; bool create_err= FALSE; DBUG_ENTER("get_quick_select"); if (param->table->key_info[param->real_keynr[idx]].flags & HA_SPATIAL) quick=new QUICK_RANGE_SELECT_GEOM(param->thd, param->table, param->real_keynr[idx], test(parent_alloc), parent_alloc, &create_err); else quick=new QUICK_RANGE_SELECT(param->thd, param->table, param->real_keynr[idx], test(parent_alloc), NULL, &create_err); if (quick) { if (create_err || get_quick_keys(param,quick,param->key[idx],key_tree,param->min_key,0, param->max_key,0)) { delete quick; quick=0; } else { quick->mrr_flags= mrr_flags; quick->mrr_buf_size= mrr_buf_size; quick->key_parts=(KEY_PART*) memdup_root(parent_alloc? parent_alloc : &quick->alloc, (char*) param->key[idx], sizeof(KEY_PART)* param->table->key_info[param->real_keynr[idx]].key_parts); } } DBUG_RETURN(quick); } /* ** Fix this to get all possible sub_ranges */ bool get_quick_keys(PARAM *param,QUICK_RANGE_SELECT *quick,KEY_PART *key, SEL_ARG *key_tree, uchar *min_key,uint min_key_flag, uchar *max_key, uint max_key_flag) { QUICK_RANGE *range; uint flag; int min_part= key_tree->part-1, // # of keypart values in min_key buffer max_part= key_tree->part-1; // # of keypart values in max_key buffer if (key_tree->left != &null_element) { if (get_quick_keys(param,quick,key,key_tree->left, min_key,min_key_flag, max_key, max_key_flag)) return 1; } uchar *tmp_min_key=min_key,*tmp_max_key=max_key; min_part+= key_tree->store_min(key[key_tree->part].store_length, &tmp_min_key,min_key_flag); max_part+= key_tree->store_max(key[key_tree->part].store_length, &tmp_max_key,max_key_flag); if (key_tree->next_key_part && key_tree->next_key_part->type == SEL_ARG::KEY_RANGE && key_tree->next_key_part->part == key_tree->part+1) { // const key as prefix if ((tmp_min_key - min_key) == (tmp_max_key - max_key) && memcmp(min_key, max_key, (uint)(tmp_max_key - max_key))==0 && key_tree->min_flag==0 && key_tree->max_flag==0) { if (get_quick_keys(param,quick,key,key_tree->next_key_part, tmp_min_key, min_key_flag | key_tree->min_flag, tmp_max_key, max_key_flag | key_tree->max_flag)) return 1; goto end; // Ugly, but efficient } { uint tmp_min_flag=key_tree->min_flag,tmp_max_flag=key_tree->max_flag; if (!tmp_min_flag) min_part+= key_tree->next_key_part->store_min_key(key, &tmp_min_key, &tmp_min_flag, MAX_KEY); if (!tmp_max_flag) max_part+= key_tree->next_key_part->store_max_key(key, &tmp_max_key, &tmp_max_flag, MAX_KEY); flag=tmp_min_flag | tmp_max_flag; } } else { flag = (key_tree->min_flag & GEOM_FLAG) ? key_tree->min_flag : key_tree->min_flag | key_tree->max_flag; } /* Ensure that some part of min_key and max_key are used. If not, regard this as no lower/upper range */ if ((flag & GEOM_FLAG) == 0) { if (tmp_min_key != param->min_key) flag&= ~NO_MIN_RANGE; else flag|= NO_MIN_RANGE; if (tmp_max_key != param->max_key) flag&= ~NO_MAX_RANGE; else flag|= NO_MAX_RANGE; } if (flag == 0) { uint length= (uint) (tmp_min_key - param->min_key); if (length == (uint) (tmp_max_key - param->max_key) && !memcmp(param->min_key,param->max_key,length)) { KEY *table_key=quick->head->key_info+quick->index; flag=EQ_RANGE; if ((table_key->flags & HA_NOSAME) && key->part == table_key->key_parts-1) { if (!(table_key->flags & HA_NULL_PART_KEY) || !null_part_in_key(key, param->min_key, (uint) (tmp_min_key - param->min_key))) flag|= UNIQUE_RANGE; else flag|= NULL_RANGE; } } } /* Get range for retrieving rows in QUICK_SELECT::get_next */ if (!(range= new QUICK_RANGE(param->min_key, (uint) (tmp_min_key - param->min_key), min_part >=0 ? make_keypart_map(min_part) : 0, param->max_key, (uint) (tmp_max_key - param->max_key), max_part >=0 ? make_keypart_map(max_part) : 0, flag))) return 1; // out of memory set_if_bigger(quick->max_used_key_length, range->min_length); set_if_bigger(quick->max_used_key_length, range->max_length); set_if_bigger(quick->used_key_parts, (uint) key_tree->part+1); if (insert_dynamic(&quick->ranges, (uchar*) &range)) return 1; end: if (key_tree->right != &null_element) return get_quick_keys(param,quick,key,key_tree->right, min_key,min_key_flag, max_key,max_key_flag); return 0; } /* Return 1 if there is only one range and this uses the whole primary key */ bool QUICK_RANGE_SELECT::unique_key_range() { if (ranges.elements == 1) { QUICK_RANGE *tmp= *((QUICK_RANGE**)ranges.buffer); if ((tmp->flag & (EQ_RANGE | NULL_RANGE)) == EQ_RANGE) { KEY *key=head->key_info+index; return (key->flags & HA_NOSAME) && key->key_length == tmp->min_length; } } return 0; } /* Return TRUE if any part of the key is NULL SYNOPSIS null_part_in_key() key_part Array of key parts (index description) key Key values tuple length Length of key values tuple in bytes. RETURN TRUE The tuple has at least one "keypartX is NULL" FALSE Otherwise */ static bool null_part_in_key(KEY_PART *key_part, const uchar *key, uint length) { for (const uchar *end=key+length ; key < end; key+= key_part++->store_length) { if (key_part->null_bit && *key) return 1; } return 0; } bool QUICK_SELECT_I::is_keys_used(const MY_BITMAP *fields) { return is_key_used(head, index, fields); } bool QUICK_INDEX_SORT_SELECT::is_keys_used(const MY_BITMAP *fields) { QUICK_RANGE_SELECT *quick; List_iterator_fast it(quick_selects); while ((quick= it++)) { if (is_key_used(head, quick->index, fields)) return 1; } return 0; } bool QUICK_ROR_INTERSECT_SELECT::is_keys_used(const MY_BITMAP *fields) { QUICK_SELECT_WITH_RECORD *qr; List_iterator_fast it(quick_selects); while ((qr= it++)) { if (is_key_used(head, qr->quick->index, fields)) return 1; } return 0; } bool QUICK_ROR_UNION_SELECT::is_keys_used(const MY_BITMAP *fields) { QUICK_SELECT_I *quick; List_iterator_fast it(quick_selects); while ((quick= it++)) { if (quick->is_keys_used(fields)) return 1; } return 0; } FT_SELECT *get_ft_select(THD *thd, TABLE *table, uint key) { bool create_err= FALSE; FT_SELECT *fts= new FT_SELECT(thd, table, key, &create_err); if (create_err) { delete fts; return NULL; } else return fts; } /* Create quick select from ref/ref_or_null scan. SYNOPSIS get_quick_select_for_ref() thd Thread handle table Table to access ref ref[_or_null] scan parameters records Estimate of number of records (needed only to construct quick select) NOTES This allocates things in a new memory root, as this may be called many times during a query. RETURN Quick select that retrieves the same rows as passed ref scan NULL on error. */ QUICK_RANGE_SELECT *get_quick_select_for_ref(THD *thd, TABLE *table, TABLE_REF *ref, ha_rows records) { MEM_ROOT *old_root, *alloc; QUICK_RANGE_SELECT *quick; KEY *key_info = &table->key_info[ref->key]; KEY_PART *key_part; QUICK_RANGE *range; uint part; bool create_err= FALSE; COST_VECT cost; old_root= thd->mem_root; /* The following call may change thd->mem_root */ quick= new QUICK_RANGE_SELECT(thd, table, ref->key, 0, 0, &create_err); /* save mem_root set by QUICK_RANGE_SELECT constructor */ alloc= thd->mem_root; /* return back default mem_root (thd->mem_root) changed by QUICK_RANGE_SELECT constructor */ thd->mem_root= old_root; if (!quick || create_err) return 0; /* no ranges found */ if (quick->init()) goto err; quick->records= records; if ((cp_buffer_from_ref(thd, table, ref) && thd->is_fatal_error) || !(range= new(alloc) QUICK_RANGE())) goto err; // out of memory range->min_key= range->max_key= ref->key_buff; range->min_length= range->max_length= ref->key_length; range->min_keypart_map= range->max_keypart_map= make_prev_keypart_map(ref->key_parts); range->flag= (ref->key_length == key_info->key_length ? EQ_RANGE : 0); if (!(quick->key_parts=key_part=(KEY_PART *) alloc_root(&quick->alloc,sizeof(KEY_PART)*ref->key_parts))) goto err; for (part=0 ; part < ref->key_parts ;part++,key_part++) { key_part->part=part; key_part->field= key_info->key_part[part].field; key_part->length= key_info->key_part[part].length; key_part->store_length= key_info->key_part[part].store_length; key_part->null_bit= key_info->key_part[part].null_bit; key_part->flag= (uint8) key_info->key_part[part].key_part_flag; } if (insert_dynamic(&quick->ranges,(uchar*)&range)) goto err; /* Add a NULL range if REF_OR_NULL optimization is used. For example: if we have "WHERE A=2 OR A IS NULL" we created the (A=2) range above and have ref->null_ref_key set. Will create a new NULL range here. */ if (ref->null_ref_key) { QUICK_RANGE *null_range; *ref->null_ref_key= 1; // Set null byte then create a range if (!(null_range= new (alloc) QUICK_RANGE(ref->key_buff, ref->key_length, make_prev_keypart_map(ref->key_parts), ref->key_buff, ref->key_length, make_prev_keypart_map(ref->key_parts), EQ_RANGE))) goto err; *ref->null_ref_key= 0; // Clear null byte if (insert_dynamic(&quick->ranges,(uchar*)&null_range)) goto err; } /* Call multi_range_read_info() to get the MRR flags and buffer size */ quick->mrr_flags= HA_MRR_NO_ASSOCIATION | (table->key_read ? HA_MRR_INDEX_ONLY : 0); if (thd->lex->sql_command != SQLCOM_SELECT) quick->mrr_flags |= HA_MRR_USE_DEFAULT_IMPL; quick->mrr_buf_size= thd->variables.mrr_buff_size; if (table->file->multi_range_read_info(quick->index, 1, (uint)records, ~0, &quick->mrr_buf_size, &quick->mrr_flags, &cost)) goto err; return quick; err: delete quick; return 0; } /* Perform key scans for all used indexes (except CPK), get rowids and merge them into an ordered non-recurrent sequence of rowids. The merge/duplicate removal is performed using Unique class. We put all rowids into Unique, get the sorted sequence and destroy the Unique. If table has a clustered primary key that covers all rows (TRUE for bdb and innodb currently) and one of the index_merge scans is a scan on PK, then rows that will be retrieved by PK scan are not put into Unique and primary key scan is not performed here, it is performed later separately. RETURN 0 OK other error */ int read_keys_and_merge_scans(THD *thd, TABLE *head, List quick_selects, QUICK_RANGE_SELECT *pk_quick_select, READ_RECORD *read_record, bool intersection, key_map *filtered_scans, Unique **unique_ptr) { List_iterator_fast cur_quick_it(quick_selects); QUICK_RANGE_SELECT* cur_quick; int result; Unique *unique= *unique_ptr; handler *file= head->file; bool with_cpk_filter= pk_quick_select != NULL; DBUG_ENTER("read_keys_and_merge"); /* We're going to just read rowids. */ if (!head->key_read) { head->enable_keyread(); } head->prepare_for_position(); cur_quick_it.rewind(); cur_quick= cur_quick_it++; bool first_quick= TRUE; DBUG_ASSERT(cur_quick != 0); /* We reuse the same instance of handler so we need to call both init and reset here. */ if (cur_quick->init() || cur_quick->reset()) goto err; if (unique == NULL) { DBUG_EXECUTE_IF("index_merge_may_not_create_a_Unique", DBUG_ABORT(); ); DBUG_EXECUTE_IF("only_one_Unique_may_be_created", DBUG_SET("+d,index_merge_may_not_create_a_Unique"); ); unique= new Unique(refpos_order_cmp, (void *)file, file->ref_length, thd->variables.sortbuff_size, intersection ? quick_selects.elements : 0); if (!unique) goto err; *unique_ptr= unique; } else unique->reset(); DBUG_ASSERT(file->ref_length == unique->get_size()); DBUG_ASSERT(thd->variables.sortbuff_size == unique->get_max_in_memory_size()); for (;;) { while ((result= cur_quick->get_next()) == HA_ERR_END_OF_FILE) { if (intersection) with_cpk_filter= filtered_scans->is_set(cur_quick->index); if (first_quick) { first_quick= FALSE; if (intersection && unique->is_in_memory()) unique->close_for_expansion(); } cur_quick->range_end(); cur_quick= cur_quick_it++; if (!cur_quick) break; if (cur_quick->file->inited != handler::NONE) cur_quick->file->ha_index_end(); if (cur_quick->init() || cur_quick->reset()) goto err; } if (result) { if (result != HA_ERR_END_OF_FILE) { cur_quick->range_end(); goto err; } break; } if (thd->killed) goto err; if (with_cpk_filter && pk_quick_select->row_in_ranges() != intersection ) continue; cur_quick->file->position(cur_quick->record); if (unique->unique_add((char*)cur_quick->file->ref)) goto err; } /* Ok all rowids are in the Unique now. The next call will initialize head->sort structure so it can be used to iterate through the rowids sequence. */ result= unique->get(head); /* index merge currently doesn't support "using index" at all */ head->disable_keyread(); if (init_read_record(read_record, thd, head, (SQL_SELECT*) 0, 1 , 1, TRUE)) result= 1; DBUG_RETURN(result); err: head->disable_keyread(); DBUG_RETURN(1); } int QUICK_INDEX_MERGE_SELECT::read_keys_and_merge() { int result; DBUG_ENTER("QUICK_INDEX_MERGE_SELECT::read_keys_and_merge"); result= read_keys_and_merge_scans(thd, head, quick_selects, pk_quick_select, &read_record, FALSE, NULL, &unique); doing_pk_scan= FALSE; DBUG_RETURN(result); } /* Get next row for index_merge. NOTES The rows are read from 1. rowids stored in Unique. 2. QUICK_RANGE_SELECT with clustered primary key (if any). The sets of rows retrieved in 1) and 2) are guaranteed to be disjoint. */ int QUICK_INDEX_MERGE_SELECT::get_next() { int result; DBUG_ENTER("QUICK_INDEX_MERGE_SELECT::get_next"); if (doing_pk_scan) DBUG_RETURN(pk_quick_select->get_next()); if ((result= read_record.read_record(&read_record)) == -1) { result= HA_ERR_END_OF_FILE; end_read_record(&read_record); free_io_cache(head); /* All rows from Unique have been retrieved, do a clustered PK scan */ if (pk_quick_select) { doing_pk_scan= TRUE; if ((result= pk_quick_select->init()) || (result= pk_quick_select->reset())) DBUG_RETURN(result); DBUG_RETURN(pk_quick_select->get_next()); } } DBUG_RETURN(result); } int QUICK_INDEX_INTERSECT_SELECT::read_keys_and_merge() { int result; DBUG_ENTER("QUICK_INDEX_INTERSECT_SELECT::read_keys_and_merge"); result= read_keys_and_merge_scans(thd, head, quick_selects, pk_quick_select, &read_record, TRUE, &filtered_scans, &unique); DBUG_RETURN(result); } int QUICK_INDEX_INTERSECT_SELECT::get_next() { int result; DBUG_ENTER("QUICK_INDEX_INTERSECT_SELECT::get_next"); if ((result= read_record.read_record(&read_record)) == -1) { result= HA_ERR_END_OF_FILE; end_read_record(&read_record); free_io_cache(head); } DBUG_RETURN(result); } /* Retrieve next record. SYNOPSIS QUICK_ROR_INTERSECT_SELECT::get_next() NOTES Invariant on enter/exit: all intersected selects have retrieved all index records with rowid <= some_rowid_val and no intersected select has retrieved any index records with rowid > some_rowid_val. We start fresh and loop until we have retrieved the same rowid in each of the key scans or we got an error. If a Clustered PK scan is present, it is used only to check if row satisfies its condition (and never used for row retrieval). RETURN 0 - Ok other - Error code if any error occurred. */ int QUICK_ROR_INTERSECT_SELECT::get_next() { List_iterator_fast quick_it(quick_selects); QUICK_SELECT_WITH_RECORD *qr; QUICK_RANGE_SELECT* quick; int error, cmp; uint last_rowid_count=0; DBUG_ENTER("QUICK_ROR_INTERSECT_SELECT::get_next"); do { /* Get a rowid for first quick and save it as a 'candidate' */ qr= quick_it++; quick= qr->quick; error= quick->get_next(); if (cpk_quick) { while (!error && !cpk_quick->row_in_ranges()) error= quick->get_next(); } if (error) DBUG_RETURN(error); /* Save the read key tuple */ key_copy(qr->key_tuple, record, head->key_info + quick->index, quick->max_used_key_length); quick->file->position(quick->record); memcpy(last_rowid, quick->file->ref, head->file->ref_length); last_rowid_count= 1; while (last_rowid_count < quick_selects.elements) { if (!(qr= quick_it++)) { quick_it.rewind(); qr= quick_it++; } quick= qr->quick; do { if ((error= quick->get_next())) DBUG_RETURN(error); quick->file->position(quick->record); cmp= head->file->cmp_ref(quick->file->ref, last_rowid); } while (cmp < 0); key_copy(qr->key_tuple, record, head->key_info + quick->index, quick->max_used_key_length); /* Ok, current select 'caught up' and returned ref >= cur_ref */ if (cmp > 0) { /* Found a row with ref > cur_ref. Make it a new 'candidate' */ if (cpk_quick) { while (!cpk_quick->row_in_ranges()) { if ((error= quick->get_next())) DBUG_RETURN(error); } quick->file->position(quick->record); } memcpy(last_rowid, quick->file->ref, head->file->ref_length); last_rowid_count= 1; //save the fields here key_copy(qr->key_tuple, record, head->key_info + quick->index, quick->max_used_key_length); } else { /* current 'candidate' row confirmed by this select */ last_rowid_count++; } } /* We get here if we got the same row ref in all scans. */ if (need_to_fetch_row) error= head->file->ha_rnd_pos(head->record[0], last_rowid); } while (error == HA_ERR_RECORD_DELETED); if (!need_to_fetch_row) { /* Restore the columns we've read/saved with other quick selects */ quick_it.rewind(); while ((qr= quick_it++)) { if (qr->quick != quick) { key_restore(record, qr->key_tuple, head->key_info + qr->quick->index, qr->quick->max_used_key_length); } } } DBUG_RETURN(error); } /* Retrieve next record. SYNOPSIS QUICK_ROR_UNION_SELECT::get_next() NOTES Enter/exit invariant: For each quick select in the queue a {key,rowid} tuple has been retrieved but the corresponding row hasn't been passed to output. RETURN 0 - Ok other - Error code if any error occurred. */ int QUICK_ROR_UNION_SELECT::get_next() { int error, dup_row; QUICK_SELECT_I *quick; uchar *tmp; DBUG_ENTER("QUICK_ROR_UNION_SELECT::get_next"); do { do { if (!queue.elements) DBUG_RETURN(HA_ERR_END_OF_FILE); /* Ok, we have a queue with >= 1 scans */ quick= (QUICK_SELECT_I*)queue_top(&queue); memcpy(cur_rowid, quick->last_rowid, rowid_length); /* put into queue rowid from the same stream as top element */ if ((error= quick->get_next())) { if (error != HA_ERR_END_OF_FILE) DBUG_RETURN(error); queue_remove_top(&queue); } else { quick->save_last_pos(); queue_replace_top(&queue); } if (!have_prev_rowid) { /* No rows have been returned yet */ dup_row= FALSE; have_prev_rowid= TRUE; } else dup_row= !head->file->cmp_ref(cur_rowid, prev_rowid); } while (dup_row); tmp= cur_rowid; cur_rowid= prev_rowid; prev_rowid= tmp; error= head->file->ha_rnd_pos(quick->record, prev_rowid); } while (error == HA_ERR_RECORD_DELETED); DBUG_RETURN(error); } int QUICK_RANGE_SELECT::reset() { uint buf_size; uchar *mrange_buff; int error; HANDLER_BUFFER empty_buf; DBUG_ENTER("QUICK_RANGE_SELECT::reset"); last_range= NULL; cur_range= (QUICK_RANGE**) ranges.buffer; if (file->inited == handler::NONE) { if (in_ror_merged_scan) head->column_bitmaps_set_no_signal(&column_bitmap, &column_bitmap); if ((error= file->ha_index_init(index,1))) DBUG_RETURN(error); } /* Allocate buffer if we need one but haven't allocated it yet */ if (mrr_buf_size && !mrr_buf_desc) { buf_size= mrr_buf_size; while (buf_size && !my_multi_malloc(MYF(MY_WME), &mrr_buf_desc, sizeof(*mrr_buf_desc), &mrange_buff, buf_size, NullS)) { /* Try to shrink the buffers until both are 0. */ buf_size/= 2; } if (!mrr_buf_desc) DBUG_RETURN(HA_ERR_OUT_OF_MEM); /* Initialize the handler buffer. */ mrr_buf_desc->buffer= mrange_buff; mrr_buf_desc->buffer_end= mrange_buff + buf_size; mrr_buf_desc->end_of_used_area= mrange_buff; #ifdef HAVE_valgrind /* We need this until ndb will use the buffer efficiently (Now ndb stores complete row in here, instead of only the used fields which gives us valgrind warnings in compare_record[]) */ bzero((char*) mrange_buff, buf_size); #endif } if (!mrr_buf_desc) empty_buf.buffer= empty_buf.buffer_end= empty_buf.end_of_used_area= NULL; RANGE_SEQ_IF seq_funcs= {NULL, quick_range_seq_init, quick_range_seq_next, 0, 0}; error= file->multi_range_read_init(&seq_funcs, (void*)this, ranges.elements, mrr_flags, mrr_buf_desc? mrr_buf_desc: &empty_buf); DBUG_RETURN(error); } /* Get next possible record using quick-struct. SYNOPSIS QUICK_RANGE_SELECT::get_next() NOTES Record is read into table->record[0] RETURN 0 Found row HA_ERR_END_OF_FILE No (more) rows in range # Error code */ int QUICK_RANGE_SELECT::get_next() { range_id_t dummy; DBUG_ENTER("QUICK_RANGE_SELECT::get_next"); if (in_ror_merged_scan) { /* We don't need to signal the bitmap change as the bitmap is always the same for this head->file */ head->column_bitmaps_set_no_signal(&column_bitmap, &column_bitmap); } int result= file->multi_range_read_next(&dummy); if (in_ror_merged_scan) { /* Restore bitmaps set on entry */ head->column_bitmaps_set_no_signal(save_read_set, save_write_set); } DBUG_RETURN(result); } /* Get the next record with a different prefix. @param prefix_length length of cur_prefix @param group_key_parts The number of key parts in the group prefix @param cur_prefix prefix of a key to be searched for Each subsequent call to the method retrieves the first record that has a prefix with length prefix_length and which is different from cur_prefix, such that the record with the new prefix is within the ranges described by this->ranges. The record found is stored into the buffer pointed by this->record. The method is useful for GROUP-BY queries with range conditions to discover the prefix of the next group that satisfies the range conditions. @todo This method is a modified copy of QUICK_RANGE_SELECT::get_next(), so both methods should be unified into a more general one to reduce code duplication. @retval 0 on success @retval HA_ERR_END_OF_FILE if returned all keys @retval other if some error occurred */ int QUICK_RANGE_SELECT::get_next_prefix(uint prefix_length, uint group_key_parts, uchar *cur_prefix) { DBUG_ENTER("QUICK_RANGE_SELECT::get_next_prefix"); const key_part_map keypart_map= make_prev_keypart_map(group_key_parts); for (;;) { int result; if (last_range) { /* Read the next record in the same range with prefix after cur_prefix. */ DBUG_ASSERT(cur_prefix != NULL); result= file->ha_index_read_map(record, cur_prefix, keypart_map, HA_READ_AFTER_KEY); if (result || last_range->max_keypart_map == 0) DBUG_RETURN(result); key_range previous_endpoint; last_range->make_max_endpoint(&previous_endpoint, prefix_length, keypart_map); if (file->compare_key(&previous_endpoint) <= 0) DBUG_RETURN(0); } uint count= ranges.elements - (cur_range - (QUICK_RANGE**) ranges.buffer); if (count == 0) { /* Ranges have already been used up before. None is left for read. */ last_range= 0; DBUG_RETURN(HA_ERR_END_OF_FILE); } last_range= *(cur_range++); key_range start_key, end_key; last_range->make_min_endpoint(&start_key, prefix_length, keypart_map); last_range->make_max_endpoint(&end_key, prefix_length, keypart_map); result= file->read_range_first(last_range->min_keypart_map ? &start_key : 0, last_range->max_keypart_map ? &end_key : 0, test(last_range->flag & EQ_RANGE), TRUE); if (last_range->flag == (UNIQUE_RANGE | EQ_RANGE)) last_range= 0; // Stop searching if (result != HA_ERR_END_OF_FILE) DBUG_RETURN(result); last_range= 0; // No matching rows; go to next range } } /* Get next for geometrical indexes */ int QUICK_RANGE_SELECT_GEOM::get_next() { DBUG_ENTER("QUICK_RANGE_SELECT_GEOM::get_next"); for (;;) { int result; if (last_range) { // Already read through key result= file->ha_index_next_same(record, last_range->min_key, last_range->min_length); if (result != HA_ERR_END_OF_FILE) DBUG_RETURN(result); } uint count= ranges.elements - (cur_range - (QUICK_RANGE**) ranges.buffer); if (count == 0) { /* Ranges have already been used up before. None is left for read. */ last_range= 0; DBUG_RETURN(HA_ERR_END_OF_FILE); } last_range= *(cur_range++); result= file->ha_index_read_map(record, last_range->min_key, last_range->min_keypart_map, (ha_rkey_function)(last_range->flag ^ GEOM_FLAG)); if (result != HA_ERR_KEY_NOT_FOUND && result != HA_ERR_END_OF_FILE) DBUG_RETURN(result); last_range= 0; // Not found, to next range } } /* Check if current row will be retrieved by this QUICK_RANGE_SELECT NOTES It is assumed that currently a scan is being done on another index which reads all necessary parts of the index that is scanned by this quick select. The implementation does a binary search on sorted array of disjoint ranges, without taking size of range into account. This function is used to filter out clustered PK scan rows in index_merge quick select. RETURN TRUE if current row will be retrieved by this quick select FALSE if not */ bool QUICK_RANGE_SELECT::row_in_ranges() { QUICK_RANGE *res; uint min= 0; uint max= ranges.elements - 1; uint mid= (max + min)/2; while (min != max) { if (cmp_next(*(QUICK_RANGE**)dynamic_array_ptr(&ranges, mid))) { /* current row value > mid->max */ min= mid + 1; } else max= mid; mid= (min + max) / 2; } res= *(QUICK_RANGE**)dynamic_array_ptr(&ranges, mid); return (!cmp_next(res) && !cmp_prev(res)); } /* This is a hack: we inherit from QUICK_RANGE_SELECT so that we can use the get_next() interface, but we have to hold a pointer to the original QUICK_RANGE_SELECT because its data are used all over the place. What should be done is to factor out the data that is needed into a base class (QUICK_SELECT), and then have two subclasses (_ASC and _DESC) which handle the ranges and implement the get_next() function. But for now, this seems to work right at least. */ QUICK_SELECT_DESC::QUICK_SELECT_DESC(QUICK_RANGE_SELECT *q, uint used_key_parts_arg) :QUICK_RANGE_SELECT(*q), rev_it(rev_ranges), used_key_parts (used_key_parts_arg) { QUICK_RANGE *r; /* Use default MRR implementation for reverse scans. No table engine currently can do an MRR scan with output in reverse index order. */ mrr_buf_desc= NULL; mrr_flags |= HA_MRR_USE_DEFAULT_IMPL; mrr_buf_size= 0; QUICK_RANGE **pr= (QUICK_RANGE**)ranges.buffer; QUICK_RANGE **end_range= pr + ranges.elements; for (; pr!=end_range; pr++) rev_ranges.push_front(*pr); /* Remove EQ_RANGE flag for keys that are not using the full key */ for (r = rev_it++; r; r = rev_it++) { if ((r->flag & EQ_RANGE) && head->key_info[index].key_length != r->max_length) r->flag&= ~EQ_RANGE; } rev_it.rewind(); q->dont_free=1; // Don't free shared mem } int QUICK_SELECT_DESC::get_next() { DBUG_ENTER("QUICK_SELECT_DESC::get_next"); /* The max key is handled as follows: * - if there is NO_MAX_RANGE, start at the end and move backwards * - if it is an EQ_RANGE, which means that max key covers the entire * key, go directly to the key and read through it (sorting backwards is * same as sorting forwards) * - if it is NEAR_MAX, go to the key or next, step back once, and * move backwards * - otherwise (not NEAR_MAX == include the key), go after the key, * step back once, and move backwards */ for (;;) { int result; if (last_range) { // Already read through key result = ((last_range->flag & EQ_RANGE && used_key_parts <= head->key_info[index].key_parts) ? file->ha_index_next_same(record, last_range->min_key, last_range->min_length) : file->ha_index_prev(record)); if (!result) { if (cmp_prev(*rev_it.ref()) == 0) DBUG_RETURN(0); } else if (result != HA_ERR_END_OF_FILE) DBUG_RETURN(result); } if (!(last_range= rev_it++)) DBUG_RETURN(HA_ERR_END_OF_FILE); // All ranges used if (last_range->flag & NO_MAX_RANGE) // Read last record { int local_error; if ((local_error= file->ha_index_last(record))) DBUG_RETURN(local_error); // Empty table if (cmp_prev(last_range) == 0) DBUG_RETURN(0); last_range= 0; // No match; go to next range continue; } if (last_range->flag & EQ_RANGE && used_key_parts <= head->key_info[index].key_parts) { result= file->ha_index_read_map(record, last_range->max_key, last_range->max_keypart_map, HA_READ_KEY_EXACT); } else { DBUG_ASSERT(last_range->flag & NEAR_MAX || (last_range->flag & EQ_RANGE && used_key_parts > head->key_info[index].key_parts) || range_reads_after_key(last_range)); result= file->ha_index_read_map(record, last_range->max_key, last_range->max_keypart_map, ((last_range->flag & NEAR_MAX) ? HA_READ_BEFORE_KEY : HA_READ_PREFIX_LAST_OR_PREV)); } if (result) { if (result != HA_ERR_KEY_NOT_FOUND && result != HA_ERR_END_OF_FILE) DBUG_RETURN(result); last_range= 0; // Not found, to next range continue; } if (cmp_prev(last_range) == 0) { if (last_range->flag == (UNIQUE_RANGE | EQ_RANGE)) last_range= 0; // Stop searching DBUG_RETURN(0); // Found key is in range } last_range= 0; // To next range } } /** Create a compatible quick select with the result ordered in an opposite way @param used_key_parts_arg Number of used key parts @retval NULL in case of errors (OOM etc) @retval pointer to a newly created QUICK_SELECT_DESC if success */ QUICK_SELECT_I *QUICK_RANGE_SELECT::make_reverse(uint used_key_parts_arg) { QUICK_SELECT_DESC *new_quick= new QUICK_SELECT_DESC(this, used_key_parts_arg); if (new_quick == NULL) { delete new_quick; return NULL; } return new_quick; } /* Compare if found key is over max-value Returns 0 if key <= range->max_key TODO: Figure out why can't this function be as simple as cmp_prev(). */ int QUICK_RANGE_SELECT::cmp_next(QUICK_RANGE *range_arg) { if (range_arg->flag & NO_MAX_RANGE) return 0; /* key can't be to large */ KEY_PART *key_part=key_parts; uint store_length; for (uchar *key=range_arg->max_key, *end=key+range_arg->max_length; key < end; key+= store_length, key_part++) { int cmp; store_length= key_part->store_length; if (key_part->null_bit) { if (*key) { if (!key_part->field->is_null()) return 1; continue; } else if (key_part->field->is_null()) return 0; key++; // Skip null byte store_length--; } if ((cmp=key_part->field->key_cmp(key, key_part->length)) < 0) return 0; if (cmp > 0) return 1; } return (range_arg->flag & NEAR_MAX) ? 1 : 0; // Exact match } /* Returns 0 if found key is inside range (found key >= range->min_key). */ int QUICK_RANGE_SELECT::cmp_prev(QUICK_RANGE *range_arg) { int cmp; if (range_arg->flag & NO_MIN_RANGE) return 0; /* key can't be to small */ cmp= key_cmp(key_part_info, range_arg->min_key, range_arg->min_length); if (cmp > 0 || (cmp == 0 && !(range_arg->flag & NEAR_MIN))) return 0; return 1; // outside of range } /* * TRUE if this range will require using HA_READ_AFTER_KEY See comment in get_next() about this */ bool QUICK_SELECT_DESC::range_reads_after_key(QUICK_RANGE *range_arg) { return ((range_arg->flag & (NO_MAX_RANGE | NEAR_MAX)) || !(range_arg->flag & EQ_RANGE) || head->key_info[index].key_length != range_arg->max_length) ? 1 : 0; } void QUICK_SELECT_I::add_key_name(String *str, bool *first) { KEY *key_info= head->key_info + index; if (*first) *first= FALSE; else str->append(','); str->append(key_info->name); } void QUICK_RANGE_SELECT::add_info_string(String *str) { bool first= TRUE; add_key_name(str, &first); } void QUICK_INDEX_MERGE_SELECT::add_info_string(String *str) { QUICK_RANGE_SELECT *quick; bool first= TRUE; List_iterator_fast it(quick_selects); str->append(STRING_WITH_LEN("sort_union(")); while ((quick= it++)) { quick->add_key_name(str, &first); } if (pk_quick_select) pk_quick_select->add_key_name(str, &first); str->append(')'); } void QUICK_INDEX_INTERSECT_SELECT::add_info_string(String *str) { QUICK_RANGE_SELECT *quick; bool first= TRUE; List_iterator_fast it(quick_selects); str->append(STRING_WITH_LEN("sort_intersect(")); if (pk_quick_select) pk_quick_select->add_key_name(str, &first); while ((quick= it++)) { quick->add_key_name(str, &first); } str->append(')'); } void QUICK_ROR_INTERSECT_SELECT::add_info_string(String *str) { bool first= TRUE; QUICK_SELECT_WITH_RECORD *qr; List_iterator_fast it(quick_selects); str->append(STRING_WITH_LEN("intersect(")); while ((qr= it++)) { qr->quick->add_key_name(str, &first); } if (cpk_quick) cpk_quick->add_key_name(str, &first); str->append(')'); } void QUICK_ROR_UNION_SELECT::add_info_string(String *str) { QUICK_SELECT_I *quick; bool first= TRUE; List_iterator_fast it(quick_selects); str->append(STRING_WITH_LEN("union(")); while ((quick= it++)) { if (first) first= FALSE; else str->append(','); quick->add_info_string(str); } str->append(')'); } void QUICK_SELECT_I::add_key_and_length(String *key_names, String *used_lengths, bool *first) { char buf[64]; uint length; KEY *key_info= head->key_info + index; if (*first) *first= FALSE; else { key_names->append(','); used_lengths->append(','); } key_names->append(key_info->name); length= longlong10_to_str(max_used_key_length, buf, 10) - buf; used_lengths->append(buf, length); } void QUICK_RANGE_SELECT::add_keys_and_lengths(String *key_names, String *used_lengths) { bool first= TRUE; add_key_and_length(key_names, used_lengths, &first); } void QUICK_INDEX_MERGE_SELECT::add_keys_and_lengths(String *key_names, String *used_lengths) { QUICK_RANGE_SELECT *quick; bool first= TRUE; List_iterator_fast it(quick_selects); while ((quick= it++)) { quick->add_key_and_length(key_names, used_lengths, &first); } if (pk_quick_select) pk_quick_select->add_key_and_length(key_names, used_lengths, &first); } void QUICK_INDEX_INTERSECT_SELECT::add_keys_and_lengths(String *key_names, String *used_lengths) { QUICK_RANGE_SELECT *quick; bool first= TRUE; List_iterator_fast it(quick_selects); if (pk_quick_select) pk_quick_select->add_key_and_length(key_names, used_lengths, &first); while ((quick= it++)) { quick->add_key_and_length(key_names, used_lengths, &first); } } void QUICK_ROR_INTERSECT_SELECT::add_keys_and_lengths(String *key_names, String *used_lengths) { QUICK_SELECT_WITH_RECORD *qr; bool first= TRUE; List_iterator_fast it(quick_selects); while ((qr= it++)) { qr->quick->add_key_and_length(key_names, used_lengths, &first); } if (cpk_quick) cpk_quick->add_key_and_length(key_names, used_lengths, &first); } void QUICK_ROR_UNION_SELECT::add_keys_and_lengths(String *key_names, String *used_lengths) { QUICK_SELECT_I *quick; bool first= TRUE; List_iterator_fast it(quick_selects); while ((quick= it++)) { if (first) first= FALSE; else { used_lengths->append(','); key_names->append(','); } quick->add_keys_and_lengths(key_names, used_lengths); } } /******************************************************************************* * Implementation of QUICK_GROUP_MIN_MAX_SELECT *******************************************************************************/ static inline uint get_field_keypart(KEY *index, Field *field); static inline SEL_ARG * get_index_range_tree(uint index, SEL_TREE* range_tree, PARAM *param, uint *param_idx); static bool get_constant_key_infix(KEY *index_info, SEL_ARG *index_range_tree, KEY_PART_INFO *first_non_group_part, KEY_PART_INFO *min_max_arg_part, KEY_PART_INFO *last_part, THD *thd, uchar *key_infix, uint *key_infix_len, KEY_PART_INFO **first_non_infix_part); static bool check_group_min_max_predicates(Item *cond, Item_field *min_max_arg_item, Field::imagetype image_type); static void cost_group_min_max(TABLE* table, KEY *index_info, uint used_key_parts, uint group_key_parts, SEL_TREE *range_tree, SEL_ARG *index_tree, ha_rows quick_prefix_records, bool have_min, bool have_max, double *read_cost, ha_rows *records); /** Test if this access method is applicable to a GROUP query with MIN/MAX functions, and if so, construct a new TRP object. DESCRIPTION Test whether a query can be computed via a QUICK_GROUP_MIN_MAX_SELECT. Queries computable via a QUICK_GROUP_MIN_MAX_SELECT must satisfy the following conditions: A) Table T has at least one compound index I of the form: I = B) Query conditions: B0. Q is over a single table T. B1. The attributes referenced by Q are a subset of the attributes of I. B2. All attributes QA in Q can be divided into 3 overlapping groups: - SA = {S_1, ..., S_l, [C]} - from the SELECT clause, where C is referenced by any number of MIN and/or MAX functions if present. - WA = {W_1, ..., W_p} - from the WHERE clause - GA = - from the GROUP BY clause (if any) = SA - if Q is a DISTINCT query (based on the equivalence of DISTINCT and GROUP queries. - NGA = QA - (GA union C) = {NG_1, ..., NG_m} - the ones not in GROUP BY and not referenced by MIN/MAX functions. with the following properties specified below. B3. If Q has a GROUP BY WITH ROLLUP clause the access method is not applicable. SA1. There is at most one attribute in SA referenced by any number of MIN and/or MAX functions which, which if present, is denoted as C. SA2. The position of the C attribute in the index is after the last A_k. SA3. The attribute C can be referenced in the WHERE clause only in predicates of the forms: - (C {< | <= | > | >= | =} const) - (const {< | <= | > | >= | =} C) - (C between const_i and const_j) - C IS NULL - C IS NOT NULL - C != const SA4. If Q has a GROUP BY clause, there are no other aggregate functions except MIN and MAX. For queries with DISTINCT, aggregate functions are allowed. SA5. The select list in DISTINCT queries should not contain expressions. GA1. If Q has a GROUP BY clause, then GA is a prefix of I. That is, if G_i = A_j => i = j. GA2. If Q has a DISTINCT clause, then there is a permutation of SA that forms a prefix of I. This permutation is used as the GROUP clause when the DISTINCT query is converted to a GROUP query. GA3. The attributes in GA may participate in arbitrary predicates, divided into two groups: - RNG(G_1,...,G_q ; where q <= k) is a range condition over the attributes of a prefix of GA - PA(G_i1,...G_iq) is an arbitrary predicate over an arbitrary subset of GA. Since P is applied to only GROUP attributes it filters some groups, and thus can be applied after the grouping. GA4. There are no expressions among G_i, just direct column references. NGA1.If in the index I there is a gap between the last GROUP attribute G_k, and the MIN/MAX attribute C, then NGA must consist of exactly the index attributes that constitute the gap. As a result there is a permutation of NGA that coincides with the gap in the index . NGA2.If BA <> {}, then the WHERE clause must contain a conjunction EQ of equality conditions for all NG_i of the form (NG_i = const) or (const = NG_i), such that each NG_i is referenced in exactly one conjunct. Informally, the predicates provide constants to fill the gap in the index. WA1. There are no other attributes in the WHERE clause except the ones referenced in predicates RNG, PA, PC, EQ defined above. Therefore WA is subset of (GA union NGA union C) for GA,NGA,C that pass the above tests. By transitivity then it also follows that each WA_i participates in the index I (if this was already tested for GA, NGA and C). C) Overall query form: SELECT EXPR([A_1,...,A_k], [B_1,...,B_m], [MIN(C)], [MAX(C)]) FROM T WHERE [RNG(A_1,...,A_p ; where p <= k)] [AND EQ(B_1,...,B_m)] [AND PC(C)] [AND PA(A_i1,...,A_iq)] GROUP BY A_1,...,A_k [HAVING PH(A_1, ..., B_1,..., C)] where EXPR(...) is an arbitrary expression over some or all SELECT fields, or: SELECT DISTINCT A_i1,...,A_ik FROM T WHERE [RNG(A_1,...,A_p ; where p <= k)] [AND PA(A_i1,...,A_iq)]; NOTES If the current query satisfies the conditions above, and if (mem_root! = NULL), then the function constructs and returns a new TRP object, that is later used to construct a new QUICK_GROUP_MIN_MAX_SELECT. If (mem_root == NULL), then the function only tests whether the current query satisfies the conditions above, and, if so, sets is_applicable = TRUE. Queries with DISTINCT for which index access can be used are transformed into equivalent group-by queries of the form: SELECT A_1,...,A_k FROM T WHERE [RNG(A_1,...,A_p ; where p <= k)] [AND PA(A_i1,...,A_iq)] GROUP BY A_1,...,A_k; The group-by list is a permutation of the select attributes, according to their order in the index. TODO - What happens if the query groups by the MIN/MAX field, and there is no other field as in: "select min(a) from t1 group by a" ? - We assume that the general correctness of the GROUP-BY query was checked before this point. Is this correct, or do we have to check it completely? - Lift the limitation in condition (B3), that is, make this access method applicable to ROLLUP queries. @param param Parameter from test_quick_select @param sel_tree Range tree generated by get_mm_tree @param read_time Best read time so far (=table/index scan time) @return table read plan @retval NULL Loose index scan not applicable or mem_root == NULL @retval !NULL Loose index scan table read plan */ static TRP_GROUP_MIN_MAX * get_best_group_min_max(PARAM *param, SEL_TREE *tree, double read_time) { THD *thd= param->thd; JOIN *join= thd->lex->current_select->join; TABLE *table= param->table; bool have_min= FALSE; /* TRUE if there is a MIN function. */ bool have_max= FALSE; /* TRUE if there is a MAX function. */ Item_field *min_max_arg_item= NULL; // The argument of all MIN/MAX functions KEY_PART_INFO *min_max_arg_part= NULL; /* The corresponding keypart. */ uint group_prefix_len= 0; /* Length (in bytes) of the key prefix. */ KEY *index_info= NULL; /* The index chosen for data access. */ uint index= 0; /* The id of the chosen index. */ uint group_key_parts= 0; // Number of index key parts in the group prefix. uint used_key_parts= 0; /* Number of index key parts used for access. */ uchar key_infix[MAX_KEY_LENGTH]; /* Constants from equality predicates.*/ uint key_infix_len= 0; /* Length of key_infix. */ TRP_GROUP_MIN_MAX *read_plan= NULL; /* The eventually constructed TRP. */ uint key_part_nr; ORDER *tmp_group; Item *item; Item_field *item_field; bool is_agg_distinct; List agg_distinct_flds; DBUG_ENTER("get_best_group_min_max"); /* Perform few 'cheap' tests whether this access method is applicable. */ if (!join) DBUG_RETURN(NULL); /* This is not a select statement. */ if ((join->table_count != 1) || /* The query must reference one table. */ (join->select_lex->olap == ROLLUP_TYPE)) /* Check (B3) for ROLLUP */ DBUG_RETURN(NULL); if (table->s->keys == 0) /* There are no indexes to use. */ DBUG_RETURN(NULL); /* Check (SA1,SA4) and store the only MIN/MAX argument - the C attribute.*/ if (join->make_sum_func_list(join->all_fields, join->fields_list, 1)) DBUG_RETURN(NULL); List_iterator select_items_it(join->fields_list); is_agg_distinct = is_indexed_agg_distinct(join, &agg_distinct_flds); if ((!join->group_list) && /* Neither GROUP BY nor a DISTINCT query. */ (!join->select_distinct) && !is_agg_distinct) DBUG_RETURN(NULL); /* Analyze the query in more detail. */ if (join->sum_funcs[0]) { Item_sum *min_max_item; Item_sum **func_ptr= join->sum_funcs; while ((min_max_item= *(func_ptr++))) { if (min_max_item->sum_func() == Item_sum::MIN_FUNC) have_min= TRUE; else if (min_max_item->sum_func() == Item_sum::MAX_FUNC) have_max= TRUE; else if (min_max_item->sum_func() == Item_sum::COUNT_DISTINCT_FUNC || min_max_item->sum_func() == Item_sum::SUM_DISTINCT_FUNC || min_max_item->sum_func() == Item_sum::AVG_DISTINCT_FUNC) continue; else DBUG_RETURN(NULL); /* The argument of MIN/MAX. */ Item *expr= min_max_item->get_arg(0)->real_item(); if (expr->type() == Item::FIELD_ITEM) /* Is it an attribute? */ { if (! min_max_arg_item) min_max_arg_item= (Item_field*) expr; else if (! min_max_arg_item->eq(expr, 1)) DBUG_RETURN(NULL); } else DBUG_RETURN(NULL); } } /* Check (SA5). */ if (join->select_distinct) { while ((item= select_items_it++)) { if (item->real_item()->type() != Item::FIELD_ITEM) DBUG_RETURN(NULL); } } /* Check (GA4) - that there are no expressions among the group attributes. */ for (tmp_group= join->group_list; tmp_group; tmp_group= tmp_group->next) { if ((*tmp_group->item)->real_item()->type() != Item::FIELD_ITEM) DBUG_RETURN(NULL); } /* Check that table has at least one compound index such that the conditions (GA1,GA2) are all TRUE. If there is more than one such index, select the first one. Here we set the variables: group_prefix_len and index_info. */ KEY *cur_index_info= table->key_info; KEY *cur_index_info_end= cur_index_info + table->s->keys; /* Cost-related variables for the best index so far. */ double best_read_cost= DBL_MAX; ha_rows best_records= 0; SEL_ARG *best_index_tree= NULL; ha_rows best_quick_prefix_records= 0; uint best_param_idx= 0; const uint pk= param->table->s->primary_key; uint max_key_part; SEL_ARG *cur_index_tree= NULL; ha_rows cur_quick_prefix_records= 0; uint cur_param_idx=MAX_KEY; for (uint cur_index= 0 ; cur_index_info != cur_index_info_end ; cur_index_info++, cur_index++) { KEY_PART_INFO *cur_part; KEY_PART_INFO *end_part; /* Last part for loops. */ /* Last index part. */ KEY_PART_INFO *last_part; KEY_PART_INFO *first_non_group_part; KEY_PART_INFO *first_non_infix_part; uint key_infix_parts; uint cur_group_key_parts= 0; uint cur_group_prefix_len= 0; double cur_read_cost; ha_rows cur_records; key_map used_key_parts_map; uint cur_key_infix_len= 0; uchar cur_key_infix[MAX_KEY_LENGTH]; uint cur_used_key_parts; /* Check (B1) - if current index is covering. */ if (!table->covering_keys.is_set(cur_index)) goto next_index; /* If the current storage manager is such that it appends the primary key to each index, then the above condition is insufficient to check if the index is covering. In such cases it may happen that some fields are covered by the PK index, but not by the current index. Since we can't use the concatenation of both indexes for index lookup, such an index does not qualify as covering in our case. If this is the case, below we check that all query fields are indeed covered by 'cur_index'. */ if (pk < MAX_KEY && cur_index != pk && (table->file->ha_table_flags() & HA_PRIMARY_KEY_IN_READ_INDEX)) { /* For each table field */ for (uint i= 0; i < table->s->fields; i++) { Field *cur_field= table->field[i]; /* If the field is used in the current query ensure that it's part of 'cur_index' */ if (bitmap_is_set(table->read_set, cur_field->field_index) && !cur_field->part_of_key_not_clustered.is_set(cur_index)) goto next_index; // Field was not part of key } } max_key_part= 0; used_key_parts_map.clear_all(); /* Check (GA1) for GROUP BY queries. */ if (join->group_list) { cur_part= cur_index_info->key_part; end_part= cur_part + cur_index_info->key_parts; /* Iterate in parallel over the GROUP list and the index parts. */ for (tmp_group= join->group_list; tmp_group && (cur_part != end_part); tmp_group= tmp_group->next, cur_part++) { /* TODO: tmp_group::item is an array of Item, is it OK to consider only the first Item? If so, then why? What is the array for? */ /* Above we already checked that all group items are fields. */ DBUG_ASSERT((*tmp_group->item)->real_item()->type() == Item::FIELD_ITEM); Item_field *group_field= (Item_field *) (*tmp_group->item)->real_item(); if (group_field->field->eq(cur_part->field)) { cur_group_prefix_len+= cur_part->store_length; ++cur_group_key_parts; max_key_part= cur_part - cur_index_info->key_part + 1; used_key_parts_map.set_bit(max_key_part); } else goto next_index; } } /* Check (GA2) if this is a DISTINCT query. If GA2, then Store a new ORDER object in group_fields_array at the position of the key part of item_field->field. Thus we get the ORDER objects for each field ordered as the corresponding key parts. Later group_fields_array of ORDER objects is used to convert the query to a GROUP query. */ if ((!join->group_list && join->select_distinct) || is_agg_distinct) { if (!is_agg_distinct) { select_items_it.rewind(); } List_iterator agg_distinct_flds_it (agg_distinct_flds); while (NULL != (item = (is_agg_distinct ? (Item *) agg_distinct_flds_it++ : select_items_it++))) { /* (SA5) already checked above. */ item_field= (Item_field*) item->real_item(); DBUG_ASSERT(item->real_item()->type() == Item::FIELD_ITEM); /* not doing loose index scan for derived tables */ if (!item_field->field) goto next_index; /* Find the order of the key part in the index. */ key_part_nr= get_field_keypart(cur_index_info, item_field->field); /* Check if this attribute was already present in the select list. If it was present, then its corresponding key part was alredy used. */ if (used_key_parts_map.is_set(key_part_nr)) continue; if (key_part_nr < 1 || (!is_agg_distinct && key_part_nr > join->fields_list.elements)) goto next_index; cur_part= cur_index_info->key_part + key_part_nr - 1; cur_group_prefix_len+= cur_part->store_length; used_key_parts_map.set_bit(key_part_nr); ++cur_group_key_parts; max_key_part= max(max_key_part,key_part_nr); } /* Check that used key parts forms a prefix of the index. To check this we compare bits in all_parts and cur_parts. all_parts have all bits set from 0 to (max_key_part-1). cur_parts have bits set for only used keyparts. */ ulonglong all_parts, cur_parts; all_parts= (1<> 1; if (all_parts != cur_parts) goto next_index; } /* Check (SA2). */ if (min_max_arg_item) { key_part_nr= get_field_keypart(cur_index_info, min_max_arg_item->field); if (key_part_nr <= cur_group_key_parts) goto next_index; min_max_arg_part= cur_index_info->key_part + key_part_nr - 1; } /* Check (NGA1, NGA2) and extract a sequence of constants to be used as part of all search keys. */ /* If there is MIN/MAX, each keypart between the last group part and the MIN/MAX part must participate in one equality with constants, and all keyparts after the MIN/MAX part must not be referenced in the query. If there is no MIN/MAX, the keyparts after the last group part can be referenced only in equalities with constants, and the referenced keyparts must form a sequence without any gaps that starts immediately after the last group keypart. */ last_part= cur_index_info->key_part + cur_index_info->key_parts; first_non_group_part= (cur_group_key_parts < cur_index_info->key_parts) ? cur_index_info->key_part + cur_group_key_parts : NULL; first_non_infix_part= min_max_arg_part ? (min_max_arg_part < last_part) ? min_max_arg_part : NULL : NULL; if (first_non_group_part && (!min_max_arg_part || (min_max_arg_part - first_non_group_part > 0))) { if (tree) { uint dummy; SEL_ARG *index_range_tree= get_index_range_tree(cur_index, tree, param, &dummy); if (!get_constant_key_infix(cur_index_info, index_range_tree, first_non_group_part, min_max_arg_part, last_part, thd, cur_key_infix, &cur_key_infix_len, &first_non_infix_part)) goto next_index; } else if (min_max_arg_part && (min_max_arg_part - first_non_group_part > 0)) { /* There is a gap but no range tree, thus no predicates at all for the non-group keyparts. */ goto next_index; } else if (first_non_group_part && join->conds) { /* If there is no MIN/MAX function in the query, but some index key part is referenced in the WHERE clause, then this index cannot be used because the WHERE condition over the keypart's field cannot be 'pushed' to the index (because there is no range 'tree'), and the WHERE clause must be evaluated before GROUP BY/DISTINCT. */ /* Store the first and last keyparts that need to be analyzed into one array that can be passed as parameter. */ KEY_PART_INFO *key_part_range[2]; key_part_range[0]= first_non_group_part; key_part_range[1]= last_part; /* Check if cur_part is referenced in the WHERE clause. */ if (join->conds->walk(&Item::find_item_in_field_list_processor, 0, (uchar*) key_part_range)) goto next_index; } } /* Test (WA1) partially - that no other keypart after the last infix part is referenced in the query. */ if (first_non_infix_part) { cur_part= first_non_infix_part + (min_max_arg_part && (min_max_arg_part < last_part)); for (; cur_part != last_part; cur_part++) { if (bitmap_is_set(table->read_set, cur_part->field->field_index)) goto next_index; } } /* If we got to this point, cur_index_info passes the test. */ key_infix_parts= cur_key_infix_len ? (uint) (first_non_infix_part - first_non_group_part) : 0; cur_used_key_parts= cur_group_key_parts + key_infix_parts; /* Compute the cost of using this index. */ if (tree) { /* Find the SEL_ARG sub-tree that corresponds to the chosen index. */ cur_index_tree= get_index_range_tree(cur_index, tree, param, &cur_param_idx); /* Check if this range tree can be used for prefix retrieval. */ COST_VECT dummy_cost; uint mrr_flags= HA_MRR_USE_DEFAULT_IMPL; uint mrr_bufsize=0; cur_quick_prefix_records= check_quick_select(param, cur_param_idx, FALSE /*don't care*/, cur_index_tree, TRUE, &mrr_flags, &mrr_bufsize, &dummy_cost); } cost_group_min_max(table, cur_index_info, cur_used_key_parts, cur_group_key_parts, tree, cur_index_tree, cur_quick_prefix_records, have_min, have_max, &cur_read_cost, &cur_records); /* If cur_read_cost is lower than best_read_cost use cur_index. Do not compare doubles directly because they may have different representations (64 vs. 80 bits). */ if (cur_read_cost < best_read_cost - (DBL_EPSILON * cur_read_cost)) { index_info= cur_index_info; index= cur_index; best_read_cost= cur_read_cost; best_records= cur_records; best_index_tree= cur_index_tree; best_quick_prefix_records= cur_quick_prefix_records; best_param_idx= cur_param_idx; group_key_parts= cur_group_key_parts; group_prefix_len= cur_group_prefix_len; key_infix_len= cur_key_infix_len; if (key_infix_len) memcpy (key_infix, cur_key_infix, sizeof (key_infix)); used_key_parts= cur_used_key_parts; } next_index:; } if (!index_info) /* No usable index found. */ DBUG_RETURN(NULL); /* Check (SA3) for the where clause. */ if (join->conds && min_max_arg_item && !check_group_min_max_predicates(join->conds, min_max_arg_item, (index_info->flags & HA_SPATIAL) ? Field::itMBR : Field::itRAW)) DBUG_RETURN(NULL); /* The query passes all tests, so construct a new TRP object. */ read_plan= new (param->mem_root) TRP_GROUP_MIN_MAX(have_min, have_max, is_agg_distinct, min_max_arg_part, group_prefix_len, used_key_parts, group_key_parts, index_info, index, key_infix_len, (key_infix_len > 0) ? key_infix : NULL, tree, best_index_tree, best_param_idx, best_quick_prefix_records); if (read_plan) { if (tree && read_plan->quick_prefix_records == 0) DBUG_RETURN(NULL); read_plan->read_cost= best_read_cost; read_plan->records= best_records; if (read_time < best_read_cost && is_agg_distinct) { read_plan->read_cost= 0; read_plan->use_index_scan(); } DBUG_PRINT("info", ("Returning group min/max plan: cost: %g, records: %lu", read_plan->read_cost, (ulong) read_plan->records)); } DBUG_RETURN(read_plan); } /* Check that the MIN/MAX attribute participates only in range predicates with constants. SYNOPSIS check_group_min_max_predicates() cond tree (or subtree) describing all or part of the WHERE clause being analyzed min_max_arg_item the field referenced by the MIN/MAX function(s) min_max_arg_part the keypart of the MIN/MAX argument if any DESCRIPTION The function walks recursively over the cond tree representing a WHERE clause, and checks condition (SA3) - if a field is referenced by a MIN/MAX aggregate function, it is referenced only by one of the following predicates: {=, !=, <, <=, >, >=, between, is null, is not null}. RETURN TRUE if cond passes the test FALSE o/w */ static bool check_group_min_max_predicates(Item *cond, Item_field *min_max_arg_item, Field::imagetype image_type) { DBUG_ENTER("check_group_min_max_predicates"); DBUG_ASSERT(cond && min_max_arg_item); cond= cond->real_item(); Item::Type cond_type= cond->real_type(); if (cond_type == Item::COND_ITEM) /* 'AND' or 'OR' */ { DBUG_PRINT("info", ("Analyzing: %s", ((Item_func*) cond)->func_name())); List_iterator_fast li(*((Item_cond*) cond)->argument_list()); Item *and_or_arg; while ((and_or_arg= li++)) { if (!check_group_min_max_predicates(and_or_arg, min_max_arg_item, image_type)) DBUG_RETURN(FALSE); } DBUG_RETURN(TRUE); } /* Disallow loose index scan if the MIN/MAX argument field is referenced by a subquery in the WHERE clause. */ if (cond_type == Item::SUBSELECT_ITEM) { Item_subselect *subs_cond= (Item_subselect*) cond; if (subs_cond->is_correlated) { DBUG_ASSERT(subs_cond->upper_refs.elements > 0); List_iterator_fast li(subs_cond->upper_refs); Item_subselect::Ref_to_outside *dep; while ((dep= li++)) { if (dep->item->eq(min_max_arg_item, FALSE)) DBUG_RETURN(FALSE); } } DBUG_RETURN(TRUE); } /* Condition of the form 'field' is equivalent to 'field <> 0' and thus satisfies the SA3 condition. */ if (cond_type == Item::FIELD_ITEM) { DBUG_PRINT("info", ("Analyzing: %s", cond->full_name())); DBUG_RETURN(TRUE); } /* We presume that at this point there are no other Items than functions. */ DBUG_ASSERT(cond_type == Item::FUNC_ITEM); /* Test if cond references only group-by or non-group fields. */ Item_func *pred= (Item_func*) cond; Item **arguments= pred->arguments(); Item *cur_arg; DBUG_PRINT("info", ("Analyzing: %s", pred->func_name())); for (uint arg_idx= 0; arg_idx < pred->argument_count (); arg_idx++) { cur_arg= arguments[arg_idx]->real_item(); DBUG_PRINT("info", ("cur_arg: %s", cur_arg->full_name())); if (cur_arg->type() == Item::FIELD_ITEM) { if (min_max_arg_item->eq(cur_arg, 1)) { /* If pred references the MIN/MAX argument, check whether pred is a range condition that compares the MIN/MAX argument with a constant. */ Item_func::Functype pred_type= pred->functype(); if (pred_type != Item_func::EQUAL_FUNC && pred_type != Item_func::LT_FUNC && pred_type != Item_func::LE_FUNC && pred_type != Item_func::GT_FUNC && pred_type != Item_func::GE_FUNC && pred_type != Item_func::BETWEEN && pred_type != Item_func::ISNULL_FUNC && pred_type != Item_func::ISNOTNULL_FUNC && pred_type != Item_func::EQ_FUNC && pred_type != Item_func::NE_FUNC) DBUG_RETURN(FALSE); /* Check that pred compares min_max_arg_item with a constant. */ Item *args[3]; bzero(args, 3 * sizeof(Item*)); bool inv; /* Test if this is a comparison of a field and a constant. */ if (!simple_pred(pred, args, &inv)) DBUG_RETURN(FALSE); /* Check for compatible string comparisons - similar to get_mm_leaf. */ if (args[0] && args[1] && !args[2] && // this is a binary function min_max_arg_item->result_type() == STRING_RESULT && /* Don't use an index when comparing strings of different collations. */ ((args[1]->result_type() == STRING_RESULT && image_type == Field::itRAW && ((Field_str*) min_max_arg_item->field)->charset() != pred->compare_collation()) || /* We can't always use indexes when comparing a string index to a number. */ (args[1]->result_type() != STRING_RESULT && min_max_arg_item->field->cmp_type() != args[1]->result_type()))) DBUG_RETURN(FALSE); } } else if (cur_arg->type() == Item::FUNC_ITEM) { if (!check_group_min_max_predicates(cur_arg, min_max_arg_item, image_type)) DBUG_RETURN(FALSE); } else if (cur_arg->const_item()) { /* For predicates of the form "const OP expr" we also have to check 'expr' to make a decision. */ continue; } else DBUG_RETURN(FALSE); } DBUG_RETURN(TRUE); } /* Extract a sequence of constants from a conjunction of equality predicates. SYNOPSIS get_constant_key_infix() index_info [in] Descriptor of the chosen index. index_range_tree [in] Range tree for the chosen index first_non_group_part [in] First index part after group attribute parts min_max_arg_part [in] The keypart of the MIN/MAX argument if any last_part [in] Last keypart of the index thd [in] Current thread key_infix [out] Infix of constants to be used for index lookup key_infix_len [out] Lenghth of the infix first_non_infix_part [out] The first keypart after the infix (if any) DESCRIPTION Test conditions (NGA1, NGA2) from get_best_group_min_max(). Namely, for each keypart field NGF_i not in GROUP-BY, check that there is a constant equality predicate among conds with the form (NGF_i = const_ci) or (const_ci = NGF_i). Thus all the NGF_i attributes must fill the 'gap' between the last group-by attribute and the MIN/MAX attribute in the index (if present). If these conditions hold, copy each constant from its corresponding predicate into key_infix, in the order its NG_i attribute appears in the index, and update key_infix_len with the total length of the key parts in key_infix. RETURN TRUE if the index passes the test FALSE o/w */ static bool get_constant_key_infix(KEY *index_info, SEL_ARG *index_range_tree, KEY_PART_INFO *first_non_group_part, KEY_PART_INFO *min_max_arg_part, KEY_PART_INFO *last_part, THD *thd, uchar *key_infix, uint *key_infix_len, KEY_PART_INFO **first_non_infix_part) { SEL_ARG *cur_range; KEY_PART_INFO *cur_part; /* End part for the first loop below. */ KEY_PART_INFO *end_part= min_max_arg_part ? min_max_arg_part : last_part; *key_infix_len= 0; uchar *key_ptr= key_infix; for (cur_part= first_non_group_part; cur_part != end_part; cur_part++) { /* Find the range tree for the current keypart. We assume that index_range_tree points to the leftmost keypart in the index. */ for (cur_range= index_range_tree; cur_range && cur_range->type == SEL_ARG::KEY_RANGE; cur_range= cur_range->next_key_part) { if (cur_range->field->eq(cur_part->field)) break; } if (!cur_range || cur_range->type != SEL_ARG::KEY_RANGE) { if (min_max_arg_part) return FALSE; /* The current keypart has no range predicates at all. */ else { *first_non_infix_part= cur_part; return TRUE; } } /* Check that the current range tree is a single point interval. */ if (cur_range->prev || cur_range->next) return FALSE; /* This is not the only range predicate for the field. */ if ((cur_range->min_flag & NO_MIN_RANGE) || (cur_range->max_flag & NO_MAX_RANGE) || (cur_range->min_flag & NEAR_MIN) || (cur_range->max_flag & NEAR_MAX)) return FALSE; uint field_length= cur_part->store_length; if (cur_range->maybe_null && cur_range->min_value[0] && cur_range->max_value[0]) { /* cur_range specifies 'IS NULL'. In this case the argument points to a "null value" (is_null_string) that may not always be long enough for a direct memcpy to a field. */ DBUG_ASSERT (field_length > 0); *key_ptr= 1; bzero(key_ptr+1,field_length-1); key_ptr+= field_length; *key_infix_len+= field_length; } else if (memcmp(cur_range->min_value, cur_range->max_value, field_length) == 0) { /* cur_range specifies an equality condition. */ memcpy(key_ptr, cur_range->min_value, field_length); key_ptr+= field_length; *key_infix_len+= field_length; } else return FALSE; } if (!min_max_arg_part && (cur_part == last_part)) *first_non_infix_part= last_part; return TRUE; } /* Find the key part referenced by a field. SYNOPSIS get_field_keypart() index descriptor of an index field field that possibly references some key part in index NOTES The return value can be used to get a KEY_PART_INFO pointer by part= index->key_part + get_field_keypart(...) - 1; RETURN Positive number which is the consecutive number of the key part, or 0 if field does not reference any index field. */ static inline uint get_field_keypart(KEY *index, Field *field) { KEY_PART_INFO *part, *end; for (part= index->key_part, end= part + index->key_parts; part < end; part++) { if (field->eq(part->field)) return part - index->key_part + 1; } return 0; } /* Find the SEL_ARG sub-tree that corresponds to the chosen index. SYNOPSIS get_index_range_tree() index [in] The ID of the index being looked for range_tree[in] Tree of ranges being searched param [in] PARAM from SQL_SELECT::test_quick_select param_idx [out] Index in the array PARAM::key that corresponds to 'index' DESCRIPTION A SEL_TREE contains range trees for all usable indexes. This procedure finds the SEL_ARG sub-tree for 'index'. The members of a SEL_TREE are ordered in the same way as the members of PARAM::key, thus we first find the corresponding index in the array PARAM::key. This index is returned through the variable param_idx, to be used later as argument of check_quick_select(). RETURN Pointer to the SEL_ARG subtree that corresponds to index. */ SEL_ARG * get_index_range_tree(uint index, SEL_TREE* range_tree, PARAM *param, uint *param_idx) { uint idx= 0; /* Index nr in param->key_parts */ while (idx < param->keys) { if (index == param->real_keynr[idx]) break; idx++; } *param_idx= idx; return(range_tree->keys[idx]); } /* Compute the cost of a quick_group_min_max_select for a particular index. SYNOPSIS cost_group_min_max() table [in] The table being accessed index_info [in] The index used to access the table used_key_parts [in] Number of key parts used to access the index group_key_parts [in] Number of index key parts in the group prefix range_tree [in] Tree of ranges for all indexes index_tree [in] The range tree for the current index quick_prefix_records [in] Number of records retrieved by the internally used quick range select if any have_min [in] True if there is a MIN function have_max [in] True if there is a MAX function read_cost [out] The cost to retrieve rows via this quick select records [out] The number of rows retrieved DESCRIPTION This method computes the access cost of a TRP_GROUP_MIN_MAX instance and the number of rows returned. It updates this->read_cost and this->records. NOTES The cost computation distinguishes several cases: 1) No equality predicates over non-group attributes (thus no key_infix). If groups are bigger than blocks on the average, then we assume that it is very unlikely that block ends are aligned with group ends, thus even if we look for both MIN and MAX keys, all pairs of neighbor MIN/MAX keys, except for the first MIN and the last MAX keys, will be in the same block. If groups are smaller than blocks, then we are going to read all blocks. 2) There are equality predicates over non-group attributes. In this case the group prefix is extended by additional constants, and as a result the min/max values are inside sub-groups of the original groups. The number of blocks that will be read depends on whether the ends of these sub-groups will be contained in the same or in different blocks. We compute the probability for the two ends of a subgroup to be in two different blocks as the ratio of: - the number of positions of the left-end of a subgroup inside a group, such that the right end of the subgroup is past the end of the buffer containing the left-end, and - the total number of possible positions for the left-end of the subgroup, which is the number of keys in the containing group. We assume it is very unlikely that two ends of subsequent subgroups are in the same block. 3) The are range predicates over the group attributes. Then some groups may be filtered by the range predicates. We use the selectivity of the range predicates to decide how many groups will be filtered. TODO - Take into account the optional range predicates over the MIN/MAX argument. - Check if we have a PK index and we use all cols - then each key is a group, and it will be better to use an index scan. RETURN None */ void cost_group_min_max(TABLE* table, KEY *index_info, uint used_key_parts, uint group_key_parts, SEL_TREE *range_tree, SEL_ARG *index_tree, ha_rows quick_prefix_records, bool have_min, bool have_max, double *read_cost, ha_rows *records) { ha_rows table_records; uint num_groups; uint num_blocks; uint keys_per_block; uint keys_per_group; uint keys_per_subgroup; /* Average number of keys in sub-groups */ /* formed by a key infix. */ double p_overlap; /* Probability that a sub-group overlaps two blocks. */ double quick_prefix_selectivity; double io_cost; double cpu_cost= 0; /* TODO: CPU cost of index_read calls? */ DBUG_ENTER("cost_group_min_max"); table_records= table->file->stats.records; keys_per_block= (table->file->stats.block_size / 2 / (index_info->key_length + table->file->ref_length) + 1); num_blocks= (uint)(table_records / keys_per_block) + 1; /* Compute the number of keys in a group. */ keys_per_group= index_info->rec_per_key[group_key_parts - 1]; if (keys_per_group == 0) /* If there is no statistics try to guess */ /* each group contains 10% of all records */ keys_per_group= (uint)(table_records / 10) + 1; num_groups= (uint)(table_records / keys_per_group) + 1; /* Apply the selectivity of the quick select for group prefixes. */ if (range_tree && (quick_prefix_records != HA_POS_ERROR)) { quick_prefix_selectivity= (double) quick_prefix_records / (double) table_records; num_groups= (uint) rint(num_groups * quick_prefix_selectivity); set_if_bigger(num_groups, 1); } if (used_key_parts > group_key_parts) { /* Compute the probability that two ends of a subgroup are inside different blocks. */ keys_per_subgroup= index_info->rec_per_key[used_key_parts - 1]; if (keys_per_subgroup >= keys_per_block) /* If a subgroup is bigger than */ p_overlap= 1.0; /* a block, it will overlap at least two blocks. */ else { double blocks_per_group= (double) num_blocks / (double) num_groups; p_overlap= (blocks_per_group * (keys_per_subgroup - 1)) / keys_per_group; p_overlap= min(p_overlap, 1.0); } io_cost= (double) min(num_groups * (1 + p_overlap), num_blocks); } else io_cost= (keys_per_group > keys_per_block) ? (have_min && have_max) ? (double) (num_groups + 1) : (double) num_groups : (double) num_blocks; /* TODO: If there is no WHERE clause and no other expressions, there should be no CPU cost. We leave it here to make this cost comparable to that of index scan as computed in SQL_SELECT::test_quick_select(). */ cpu_cost= (double) num_groups / TIME_FOR_COMPARE; *read_cost= io_cost + cpu_cost; *records= num_groups; DBUG_PRINT("info", ("table rows: %lu keys/block: %u keys/group: %u result rows: %lu blocks: %u", (ulong)table_records, keys_per_block, keys_per_group, (ulong) *records, num_blocks)); DBUG_VOID_RETURN; } /* Construct a new quick select object for queries with group by with min/max. SYNOPSIS TRP_GROUP_MIN_MAX::make_quick() param Parameter from test_quick_select retrieve_full_rows ignored parent_alloc Memory pool to use, if any. NOTES Make_quick ignores the retrieve_full_rows parameter because QUICK_GROUP_MIN_MAX_SELECT always performs 'index only' scans. The other parameter are ignored as well because all necessary data to create the QUICK object is computed at this TRP creation time. RETURN New QUICK_GROUP_MIN_MAX_SELECT object if successfully created, NULL otherwise. */ QUICK_SELECT_I * TRP_GROUP_MIN_MAX::make_quick(PARAM *param, bool retrieve_full_rows, MEM_ROOT *parent_alloc) { QUICK_GROUP_MIN_MAX_SELECT *quick; DBUG_ENTER("TRP_GROUP_MIN_MAX::make_quick"); quick= new QUICK_GROUP_MIN_MAX_SELECT(param->table, param->thd->lex->current_select->join, have_min, have_max, have_agg_distinct, min_max_arg_part, group_prefix_len, group_key_parts, used_key_parts, index_info, index, read_cost, records, key_infix_len, key_infix, parent_alloc, is_index_scan); if (!quick) DBUG_RETURN(NULL); if (quick->init()) { delete quick; DBUG_RETURN(NULL); } if (range_tree) { DBUG_ASSERT(quick_prefix_records > 0); if (quick_prefix_records == HA_POS_ERROR) quick->quick_prefix_select= NULL; /* Can't construct a quick select. */ else /* Make a QUICK_RANGE_SELECT to be used for group prefix retrieval. */ quick->quick_prefix_select= get_quick_select(param, param_idx, index_tree, HA_MRR_USE_DEFAULT_IMPL, 0, &quick->alloc); /* Extract the SEL_ARG subtree that contains only ranges for the MIN/MAX attribute, and create an array of QUICK_RANGES to be used by the new quick select. */ if (min_max_arg_part) { SEL_ARG *min_max_range= index_tree; while (min_max_range) /* Find the tree for the MIN/MAX key part. */ { if (min_max_range->field->eq(min_max_arg_part->field)) break; min_max_range= min_max_range->next_key_part; } /* Scroll to the leftmost interval for the MIN/MAX argument. */ while (min_max_range && min_max_range->prev) min_max_range= min_max_range->prev; /* Create an array of QUICK_RANGEs for the MIN/MAX argument. */ while (min_max_range) { if (quick->add_range(min_max_range)) { delete quick; quick= NULL; DBUG_RETURN(NULL); } min_max_range= min_max_range->next; } } } else quick->quick_prefix_select= NULL; quick->update_key_stat(); quick->adjust_prefix_ranges(); DBUG_RETURN(quick); } /* Construct new quick select for group queries with min/max. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::QUICK_GROUP_MIN_MAX_SELECT() table The table being accessed join Descriptor of the current query have_min TRUE if the query selects a MIN function have_max TRUE if the query selects a MAX function min_max_arg_part The only argument field of all MIN/MAX functions group_prefix_len Length of all key parts in the group prefix prefix_key_parts All key parts in the group prefix index_info The index chosen for data access use_index The id of index_info read_cost Cost of this access method records Number of records returned key_infix_len Length of the key infix appended to the group prefix key_infix Infix of constants from equality predicates parent_alloc Memory pool for this and quick_prefix_select data is_index_scan get the next different key not by jumping on it via index read, but by scanning until the end of the rows with equal key value. RETURN None */ QUICK_GROUP_MIN_MAX_SELECT:: QUICK_GROUP_MIN_MAX_SELECT(TABLE *table, JOIN *join_arg, bool have_min_arg, bool have_max_arg, bool have_agg_distinct_arg, KEY_PART_INFO *min_max_arg_part_arg, uint group_prefix_len_arg, uint group_key_parts_arg, uint used_key_parts_arg, KEY *index_info_arg, uint use_index, double read_cost_arg, ha_rows records_arg, uint key_infix_len_arg, uchar *key_infix_arg, MEM_ROOT *parent_alloc, bool is_index_scan_arg) :file(table->file), join(join_arg), index_info(index_info_arg), group_prefix_len(group_prefix_len_arg), group_key_parts(group_key_parts_arg), have_min(have_min_arg), have_max(have_max_arg), have_agg_distinct(have_agg_distinct_arg), seen_first_key(FALSE), doing_key_read(FALSE), min_max_arg_part(min_max_arg_part_arg), key_infix(key_infix_arg), key_infix_len(key_infix_len_arg), min_functions_it(NULL), max_functions_it(NULL), is_index_scan(is_index_scan_arg) { head= table; index= use_index; record= head->record[0]; tmp_record= head->record[1]; read_time= read_cost_arg; records= records_arg; used_key_parts= used_key_parts_arg; real_key_parts= used_key_parts_arg; real_prefix_len= group_prefix_len + key_infix_len; group_prefix= NULL; min_max_arg_len= min_max_arg_part ? min_max_arg_part->store_length : 0; /* We can't have parent_alloc set as the init function can't handle this case yet. */ DBUG_ASSERT(!parent_alloc); if (!parent_alloc) { init_sql_alloc(&alloc, join->thd->variables.range_alloc_block_size, 0); join->thd->mem_root= &alloc; } else bzero(&alloc, sizeof(MEM_ROOT)); // ensure that it's not used } /* Do post-constructor initialization. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::init() DESCRIPTION The method performs initialization that cannot be done in the constructor such as memory allocations that may fail. It allocates memory for the group prefix and inifix buffers, and for the lists of MIN/MAX item to be updated during execution. RETURN 0 OK other Error code */ int QUICK_GROUP_MIN_MAX_SELECT::init() { if (group_prefix) /* Already initialized. */ return 0; if (!(last_prefix= (uchar*) alloc_root(&alloc, group_prefix_len))) return 1; /* We may use group_prefix to store keys with all select fields, so allocate enough space for it. */ if (!(group_prefix= (uchar*) alloc_root(&alloc, real_prefix_len + min_max_arg_len))) return 1; if (key_infix_len > 0) { /* The memory location pointed to by key_infix will be deleted soon, so allocate a new buffer and copy the key_infix into it. */ uchar *tmp_key_infix= (uchar*) alloc_root(&alloc, key_infix_len); if (!tmp_key_infix) return 1; memcpy(tmp_key_infix, this->key_infix, key_infix_len); this->key_infix= tmp_key_infix; } if (min_max_arg_part) { if (my_init_dynamic_array(&min_max_ranges, sizeof(QUICK_RANGE*), 16, 16)) return 1; if (have_min) { if (!(min_functions= new List)) return 1; } else min_functions= NULL; if (have_max) { if (!(max_functions= new List)) return 1; } else max_functions= NULL; Item_sum *min_max_item; Item_sum **func_ptr= join->sum_funcs; while ((min_max_item= *(func_ptr++))) { if (have_min && (min_max_item->sum_func() == Item_sum::MIN_FUNC)) min_functions->push_back(min_max_item); else if (have_max && (min_max_item->sum_func() == Item_sum::MAX_FUNC)) max_functions->push_back(min_max_item); } if (have_min) { if (!(min_functions_it= new List_iterator(*min_functions))) return 1; } if (have_max) { if (!(max_functions_it= new List_iterator(*max_functions))) return 1; } } else min_max_ranges.elements= 0; return 0; } QUICK_GROUP_MIN_MAX_SELECT::~QUICK_GROUP_MIN_MAX_SELECT() { DBUG_ENTER("QUICK_GROUP_MIN_MAX_SELECT::~QUICK_GROUP_MIN_MAX_SELECT"); if (file->inited != handler::NONE) { DBUG_ASSERT(file == head->file); if (doing_key_read) head->disable_keyread(); file->ha_index_end(); } if (min_max_arg_part) delete_dynamic(&min_max_ranges); free_root(&alloc,MYF(0)); delete min_functions_it; delete max_functions_it; delete quick_prefix_select; DBUG_VOID_RETURN; } /* Eventually create and add a new quick range object. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::add_range() sel_range Range object from which a NOTES Construct a new QUICK_RANGE object from a SEL_ARG object, and add it to the array min_max_ranges. If sel_arg is an infinite range, e.g. (x < 5 or x > 4), then skip it and do not construct a quick range. RETURN FALSE on success TRUE otherwise */ bool QUICK_GROUP_MIN_MAX_SELECT::add_range(SEL_ARG *sel_range) { QUICK_RANGE *range; uint range_flag= sel_range->min_flag | sel_range->max_flag; /* Skip (-inf,+inf) ranges, e.g. (x < 5 or x > 4). */ if ((range_flag & NO_MIN_RANGE) && (range_flag & NO_MAX_RANGE)) return FALSE; if (!(sel_range->min_flag & NO_MIN_RANGE) && !(sel_range->max_flag & NO_MAX_RANGE)) { if (sel_range->maybe_null && sel_range->min_value[0] && sel_range->max_value[0]) range_flag|= NULL_RANGE; /* IS NULL condition */ else if (memcmp(sel_range->min_value, sel_range->max_value, min_max_arg_len) == 0) range_flag|= EQ_RANGE; /* equality condition */ } range= new QUICK_RANGE(sel_range->min_value, min_max_arg_len, make_keypart_map(sel_range->part), sel_range->max_value, min_max_arg_len, make_keypart_map(sel_range->part), range_flag); if (!range) return TRUE; if (insert_dynamic(&min_max_ranges, (uchar*)&range)) return TRUE; return FALSE; } /* Opens the ranges if there are more conditions in quick_prefix_select than the ones used for jumping through the prefixes. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::adjust_prefix_ranges() NOTES quick_prefix_select is made over the conditions on the whole key. It defines a number of ranges of length x. However when jumping through the prefixes we use only the the first few most significant keyparts in the range key. However if there are more keyparts to follow the ones we are using we must make the condition on the key inclusive (because x < "ab" means x[0] < 'a' OR (x[0] == 'a' AND x[1] < 'b'). To achive the above we must turn off the NEAR_MIN/NEAR_MAX */ void QUICK_GROUP_MIN_MAX_SELECT::adjust_prefix_ranges () { if (quick_prefix_select && group_prefix_len < quick_prefix_select->max_used_key_length) { DYNAMIC_ARRAY *arr; uint inx; for (inx= 0, arr= &quick_prefix_select->ranges; inx < arr->elements; inx++) { QUICK_RANGE *range; get_dynamic(arr, (uchar*)&range, inx); range->flag &= ~(NEAR_MIN | NEAR_MAX); } } } /* Determine the total number and length of the keys that will be used for index lookup. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::update_key_stat() DESCRIPTION The total length of the keys used for index lookup depends on whether there are any predicates referencing the min/max argument, and/or if the min/max argument field can be NULL. This function does an optimistic analysis whether the search key might be extended by a constant for the min/max keypart. It is 'optimistic' because during actual execution it may happen that a particular range is skipped, and then a shorter key will be used. However this is data dependent and can't be easily estimated here. RETURN None */ void QUICK_GROUP_MIN_MAX_SELECT::update_key_stat() { max_used_key_length= real_prefix_len; if (min_max_ranges.elements > 0) { QUICK_RANGE *cur_range; if (have_min) { /* Check if the right-most range has a lower boundary. */ get_dynamic(&min_max_ranges, (uchar*)&cur_range, min_max_ranges.elements - 1); if (!(cur_range->flag & NO_MIN_RANGE)) { max_used_key_length+= min_max_arg_len; used_key_parts++; return; } } if (have_max) { /* Check if the left-most range has an upper boundary. */ get_dynamic(&min_max_ranges, (uchar*)&cur_range, 0); if (!(cur_range->flag & NO_MAX_RANGE)) { max_used_key_length+= min_max_arg_len; used_key_parts++; return; } } } else if (have_min && min_max_arg_part && min_max_arg_part->field->real_maybe_null()) { /* If a MIN/MAX argument value is NULL, we can quickly determine that we're in the beginning of the next group, because NULLs are always < any other value. This allows us to quickly determine the end of the current group and jump to the next group (see next_min()) and thus effectively increases the usable key length. */ max_used_key_length+= min_max_arg_len; used_key_parts++; } } /* Initialize a quick group min/max select for key retrieval. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::reset() DESCRIPTION Initialize the index chosen for access and find and store the prefix of the last group. The method is expensive since it performs disk access. RETURN 0 OK other Error code */ int QUICK_GROUP_MIN_MAX_SELECT::reset(void) { int result; DBUG_ENTER("QUICK_GROUP_MIN_MAX_SELECT::reset"); if (!head->key_read) { doing_key_read= 1; head->enable_keyread(); /* We need only the key attributes */ } if ((result= file->ha_index_init(index,1))) DBUG_RETURN(result); if (quick_prefix_select && quick_prefix_select->reset()) DBUG_RETURN(1); result= file->ha_index_last(record); if (result == HA_ERR_END_OF_FILE) DBUG_RETURN(0); /* Save the prefix of the last group. */ key_copy(last_prefix, record, index_info, group_prefix_len); DBUG_RETURN(0); } /* Get the next key containing the MIN and/or MAX key for the next group. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::get_next() DESCRIPTION The method finds the next subsequent group of records that satisfies the query conditions and finds the keys that contain the MIN/MAX values for the key part referenced by the MIN/MAX function(s). Once a group and its MIN/MAX values are found, store these values in the Item_sum objects for the MIN/MAX functions. The rest of the values in the result row are stored in the Item_field::result_field of each select field. If the query does not contain MIN and/or MAX functions, then the function only finds the group prefix, which is a query answer itself. NOTES If both MIN and MAX are computed, then we use the fact that if there is no MIN key, there can't be a MAX key as well, so we can skip looking for a MAX key in this case. RETURN 0 on success HA_ERR_END_OF_FILE if returned all keys other if some error occurred */ int QUICK_GROUP_MIN_MAX_SELECT::get_next() { int min_res= 0; int max_res= 0; #ifdef HPUX11 /* volatile is required by a bug in the HP compiler due to which the last test of result fails. */ volatile int result; #else int result; #endif int is_last_prefix= 0; DBUG_ENTER("QUICK_GROUP_MIN_MAX_SELECT::get_next"); /* Loop until a group is found that satisfies all query conditions or the last group is reached. */ do { result= next_prefix(); /* Check if this is the last group prefix. Notice that at this point this->record contains the current prefix in record format. */ if (!result) { is_last_prefix= key_cmp(index_info->key_part, last_prefix, group_prefix_len); DBUG_ASSERT(is_last_prefix <= 0); } else { if (result == HA_ERR_KEY_NOT_FOUND) continue; break; } if (have_min) { min_res= next_min(); if (min_res == 0) update_min_result(); } /* If there is no MIN in the group, there is no MAX either. */ if ((have_max && !have_min) || (have_max && have_min && (min_res == 0))) { max_res= next_max(); if (max_res == 0) update_max_result(); /* If a MIN was found, a MAX must have been found as well. */ DBUG_ASSERT((have_max && !have_min) || (have_max && have_min && (max_res == 0))); } /* If this is just a GROUP BY or DISTINCT without MIN or MAX and there are equality predicates for the key parts after the group, find the first sub-group with the extended prefix. */ if (!have_min && !have_max && key_infix_len > 0) result= file->ha_index_read_map(record, group_prefix, make_prev_keypart_map(real_key_parts), HA_READ_KEY_EXACT); result= have_min ? min_res : have_max ? max_res : result; } while ((result == HA_ERR_KEY_NOT_FOUND || result == HA_ERR_END_OF_FILE) && is_last_prefix != 0); if (result == HA_ERR_KEY_NOT_FOUND) result= HA_ERR_END_OF_FILE; DBUG_RETURN(result); } /* Retrieve the minimal key in the next group. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::next_min() DESCRIPTION Find the minimal key within this group such that the key satisfies the query conditions and NULL semantics. The found key is loaded into this->record. IMPLEMENTATION Depending on the values of min_max_ranges.elements, key_infix_len, and whether there is a NULL in the MIN field, this function may directly return without any data access. In this case we use the key loaded into this->record by the call to this->next_prefix() just before this call. RETURN 0 on success HA_ERR_KEY_NOT_FOUND if no MIN key was found that fulfills all conditions. HA_ERR_END_OF_FILE - "" - other if some error occurred */ int QUICK_GROUP_MIN_MAX_SELECT::next_min() { int result= 0; DBUG_ENTER("QUICK_GROUP_MIN_MAX_SELECT::next_min"); /* Find the MIN key using the eventually extended group prefix. */ if (min_max_ranges.elements > 0) { if ((result= next_min_in_range())) DBUG_RETURN(result); } else { /* Apply the constant equality conditions to the non-group select fields */ if (key_infix_len > 0) { if ((result= file->ha_index_read_map(record, group_prefix, make_prev_keypart_map(real_key_parts), HA_READ_KEY_EXACT))) DBUG_RETURN(result); } /* If the min/max argument field is NULL, skip subsequent rows in the same group with NULL in it. Notice that: - if the first row in a group doesn't have a NULL in the field, no row in the same group has (because NULL < any other value), - min_max_arg_part->field->ptr points to some place in 'record'. */ if (min_max_arg_part && min_max_arg_part->field->is_null()) { /* Find the first subsequent record without NULL in the MIN/MAX field. */ key_copy(tmp_record, record, index_info, 0); result= file->ha_index_read_map(record, tmp_record, make_keypart_map(real_key_parts), HA_READ_AFTER_KEY); /* Check if the new record belongs to the current group by comparing its prefix with the group's prefix. If it is from the next group, then the whole group has NULLs in the MIN/MAX field, so use the first record in the group as a result. TODO: It is possible to reuse this new record as the result candidate for the next call to next_min(), and to save one lookup in the next call. For this add a new member 'this->next_group_prefix'. */ if (!result) { if (key_cmp(index_info->key_part, group_prefix, real_prefix_len)) key_restore(record, tmp_record, index_info, 0); } else if (result == HA_ERR_KEY_NOT_FOUND || result == HA_ERR_END_OF_FILE) result= 0; /* There is a result in any case. */ } } /* If the MIN attribute is non-nullable, this->record already contains the MIN key in the group, so just return. */ DBUG_RETURN(result); } /* Retrieve the maximal key in the next group. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::next_max() DESCRIPTION Lookup the maximal key of the group, and store it into this->record. RETURN 0 on success HA_ERR_KEY_NOT_FOUND if no MAX key was found that fulfills all conditions. HA_ERR_END_OF_FILE - "" - other if some error occurred */ int QUICK_GROUP_MIN_MAX_SELECT::next_max() { int result; DBUG_ENTER("QUICK_GROUP_MIN_MAX_SELECT::next_max"); /* Get the last key in the (possibly extended) group. */ if (min_max_ranges.elements > 0) result= next_max_in_range(); else result= file->ha_index_read_map(record, group_prefix, make_prev_keypart_map(real_key_parts), HA_READ_PREFIX_LAST); DBUG_RETURN(result); } /** Find the next different key value by skiping all the rows with the same key value. Implements a specialized loose index access method for queries containing aggregate functions with distinct of the form: SELECT [SUM|COUNT|AVG](DISTINCT a,...) FROM t This method comes to replace the index scan + Unique class (distinct selection) for loose index scan that visits all the rows of a covering index instead of jumping in the begining of each group. TODO: Placeholder function. To be replaced by a handler API call @param is_index_scan hint to use index scan instead of random index read to find the next different value. @param file table handler @param key_part group key to compare @param record row data @param group_prefix current key prefix data @param group_prefix_len length of the current key prefix data @param group_key_parts number of the current key prefix columns @return status @retval 0 success @retval !0 failure */ static int index_next_different (bool is_index_scan, handler *file, KEY_PART_INFO *key_part, uchar * record, const uchar * group_prefix, uint group_prefix_len, uint group_key_parts) { if (is_index_scan) { int result= 0; while (!key_cmp (key_part, group_prefix, group_prefix_len)) { result= file->ha_index_next(record); if (result) return(result); } return result; } else return file->ha_index_read_map(record, group_prefix, make_prev_keypart_map(group_key_parts), HA_READ_AFTER_KEY); } /* Determine the prefix of the next group. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::next_prefix() DESCRIPTION Determine the prefix of the next group that satisfies the query conditions. If there is a range condition referencing the group attributes, use a QUICK_RANGE_SELECT object to retrieve the *first* key that satisfies the condition. If there is a key infix of constants, append this infix immediately after the group attributes. The possibly extended prefix is stored in this->group_prefix. The first key of the found group is stored in this->record, on which relies this->next_min(). RETURN 0 on success HA_ERR_KEY_NOT_FOUND if there is no key with the formed prefix HA_ERR_END_OF_FILE if there are no more keys other if some error occurred */ int QUICK_GROUP_MIN_MAX_SELECT::next_prefix() { int result; DBUG_ENTER("QUICK_GROUP_MIN_MAX_SELECT::next_prefix"); if (quick_prefix_select) { uchar *cur_prefix= seen_first_key ? group_prefix : NULL; if ((result= quick_prefix_select->get_next_prefix(group_prefix_len, group_key_parts, cur_prefix))) DBUG_RETURN(result); seen_first_key= TRUE; } else { if (!seen_first_key) { result= file->ha_index_first(record); if (result) DBUG_RETURN(result); seen_first_key= TRUE; } else { /* Load the first key in this group into record. */ result= index_next_different (is_index_scan, file, index_info->key_part, record, group_prefix, group_prefix_len, group_key_parts); if (result) DBUG_RETURN(result); } } /* Save the prefix of this group for subsequent calls. */ key_copy(group_prefix, record, index_info, group_prefix_len); /* Append key_infix to group_prefix. */ if (key_infix_len > 0) memcpy(group_prefix + group_prefix_len, key_infix, key_infix_len); DBUG_RETURN(0); } /* Find the minimal key in a group that satisfies some range conditions for the min/max argument field. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::next_min_in_range() DESCRIPTION Given the sequence of ranges min_max_ranges, find the minimal key that is in the left-most possible range. If there is no such key, then the current group does not have a MIN key that satisfies the WHERE clause. If a key is found, its value is stored in this->record. RETURN 0 on success HA_ERR_KEY_NOT_FOUND if there is no key with the given prefix in any of the ranges HA_ERR_END_OF_FILE - "" - other if some error */ int QUICK_GROUP_MIN_MAX_SELECT::next_min_in_range() { ha_rkey_function find_flag; key_part_map keypart_map; QUICK_RANGE *cur_range; bool found_null= FALSE; int result= HA_ERR_KEY_NOT_FOUND; DBUG_ASSERT(min_max_ranges.elements > 0); for (uint range_idx= 0; range_idx < min_max_ranges.elements; range_idx++) { /* Search from the left-most range to the right. */ get_dynamic(&min_max_ranges, (uchar*)&cur_range, range_idx); /* If the current value for the min/max argument is bigger than the right boundary of cur_range, there is no need to check this range. */ if (range_idx != 0 && !(cur_range->flag & NO_MAX_RANGE) && (key_cmp(min_max_arg_part, (const uchar*) cur_range->max_key, min_max_arg_len) == 1)) continue; if (cur_range->flag & NO_MIN_RANGE) { keypart_map= make_prev_keypart_map(real_key_parts); find_flag= HA_READ_KEY_EXACT; } else { /* Extend the search key with the lower boundary for this range. */ memcpy(group_prefix + real_prefix_len, cur_range->min_key, cur_range->min_length); keypart_map= make_keypart_map(real_key_parts); find_flag= (cur_range->flag & (EQ_RANGE | NULL_RANGE)) ? HA_READ_KEY_EXACT : (cur_range->flag & NEAR_MIN) ? HA_READ_AFTER_KEY : HA_READ_KEY_OR_NEXT; } result= file->ha_index_read_map(record, group_prefix, keypart_map, find_flag); if (result) { if ((result == HA_ERR_KEY_NOT_FOUND || result == HA_ERR_END_OF_FILE) && (cur_range->flag & (EQ_RANGE | NULL_RANGE))) continue; /* Check the next range. */ /* In all other cases (HA_ERR_*, HA_READ_KEY_EXACT with NO_MIN_RANGE, HA_READ_AFTER_KEY, HA_READ_KEY_OR_NEXT) if the lookup failed for this range, it can't succeed for any other subsequent range. */ break; } /* A key was found. */ if (cur_range->flag & EQ_RANGE) break; /* No need to perform the checks below for equal keys. */ if (cur_range->flag & NULL_RANGE) { /* Remember this key, and continue looking for a non-NULL key that satisfies some other condition. */ memcpy(tmp_record, record, head->s->rec_buff_length); found_null= TRUE; continue; } /* Check if record belongs to the current group. */ if (key_cmp(index_info->key_part, group_prefix, real_prefix_len)) { result= HA_ERR_KEY_NOT_FOUND; continue; } /* If there is an upper limit, check if the found key is in the range. */ if ( !(cur_range->flag & NO_MAX_RANGE) ) { /* Compose the MAX key for the range. */ uchar *max_key= (uchar*) my_alloca(real_prefix_len + min_max_arg_len); memcpy(max_key, group_prefix, real_prefix_len); memcpy(max_key + real_prefix_len, cur_range->max_key, cur_range->max_length); /* Compare the found key with max_key. */ int cmp_res= key_cmp(index_info->key_part, max_key, real_prefix_len + min_max_arg_len); my_afree(max_key); /* The key is outside of the range if: the interval is open and the key is equal to the maximum boundry or the key is greater than the maximum */ if (((cur_range->flag & NEAR_MAX) && cmp_res == 0) || cmp_res > 0) { result= HA_ERR_KEY_NOT_FOUND; continue; } } /* If we got to this point, the current key qualifies as MIN. */ DBUG_ASSERT(result == 0); break; } /* If there was a key with NULL in the MIN/MAX field, and there was no other key without NULL from the same group that satisfies some other condition, then use the key with the NULL. */ if (found_null && result) { memcpy(record, tmp_record, head->s->rec_buff_length); result= 0; } return result; } /* Find the maximal key in a group that satisfies some range conditions for the min/max argument field. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::next_max_in_range() DESCRIPTION Given the sequence of ranges min_max_ranges, find the maximal key that is in the right-most possible range. If there is no such key, then the current group does not have a MAX key that satisfies the WHERE clause. If a key is found, its value is stored in this->record. RETURN 0 on success HA_ERR_KEY_NOT_FOUND if there is no key with the given prefix in any of the ranges HA_ERR_END_OF_FILE - "" - other if some error */ int QUICK_GROUP_MIN_MAX_SELECT::next_max_in_range() { ha_rkey_function find_flag; key_part_map keypart_map; QUICK_RANGE *cur_range; int result; DBUG_ASSERT(min_max_ranges.elements > 0); for (uint range_idx= min_max_ranges.elements; range_idx > 0; range_idx--) { /* Search from the right-most range to the left. */ get_dynamic(&min_max_ranges, (uchar*)&cur_range, range_idx - 1); /* If the current value for the min/max argument is smaller than the left boundary of cur_range, there is no need to check this range. */ if (range_idx != min_max_ranges.elements && !(cur_range->flag & NO_MIN_RANGE) && (key_cmp(min_max_arg_part, (const uchar*) cur_range->min_key, min_max_arg_len) == -1)) continue; if (cur_range->flag & NO_MAX_RANGE) { keypart_map= make_prev_keypart_map(real_key_parts); find_flag= HA_READ_PREFIX_LAST; } else { /* Extend the search key with the upper boundary for this range. */ memcpy(group_prefix + real_prefix_len, cur_range->max_key, cur_range->max_length); keypart_map= make_keypart_map(real_key_parts); find_flag= (cur_range->flag & EQ_RANGE) ? HA_READ_KEY_EXACT : (cur_range->flag & NEAR_MAX) ? HA_READ_BEFORE_KEY : HA_READ_PREFIX_LAST_OR_PREV; } result= file->ha_index_read_map(record, group_prefix, keypart_map, find_flag); if (result) { if ((result == HA_ERR_KEY_NOT_FOUND || result == HA_ERR_END_OF_FILE) && (cur_range->flag & EQ_RANGE)) continue; /* Check the next range. */ /* In no key was found with this upper bound, there certainly are no keys in the ranges to the left. */ return result; } /* A key was found. */ if (cur_range->flag & EQ_RANGE) return 0; /* No need to perform the checks below for equal keys. */ /* Check if record belongs to the current group. */ if (key_cmp(index_info->key_part, group_prefix, real_prefix_len)) continue; // Row not found /* If there is a lower limit, check if the found key is in the range. */ if ( !(cur_range->flag & NO_MIN_RANGE) ) { /* Compose the MIN key for the range. */ uchar *min_key= (uchar*) my_alloca(real_prefix_len + min_max_arg_len); memcpy(min_key, group_prefix, real_prefix_len); memcpy(min_key + real_prefix_len, cur_range->min_key, cur_range->min_length); /* Compare the found key with min_key. */ int cmp_res= key_cmp(index_info->key_part, min_key, real_prefix_len + min_max_arg_len); my_afree(min_key); /* The key is outside of the range if: the interval is open and the key is equal to the minimum boundry or the key is less than the minimum */ if (((cur_range->flag & NEAR_MIN) && cmp_res == 0) || cmp_res < 0) continue; } /* If we got to this point, the current key qualifies as MAX. */ return result; } return HA_ERR_KEY_NOT_FOUND; } /* Update all MIN function results with the newly found value. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::update_min_result() DESCRIPTION The method iterates through all MIN functions and updates the result value of each function by calling Item_sum::reset(), which in turn picks the new result value from this->head->record[0], previously updated by next_min(). The updated value is stored in a member variable of each of the Item_sum objects, depending on the value type. IMPLEMENTATION The update must be done separately for MIN and MAX, immediately after next_min() was called and before next_max() is called, because both MIN and MAX take their result value from the same buffer this->head->record[0] (i.e. this->record). RETURN None */ void QUICK_GROUP_MIN_MAX_SELECT::update_min_result() { Item_sum *min_func; min_functions_it->rewind(); while ((min_func= (*min_functions_it)++)) min_func->reset_and_add(); } /* Update all MAX function results with the newly found value. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::update_max_result() DESCRIPTION The method iterates through all MAX functions and updates the result value of each function by calling Item_sum::reset(), which in turn picks the new result value from this->head->record[0], previously updated by next_max(). The updated value is stored in a member variable of each of the Item_sum objects, depending on the value type. IMPLEMENTATION The update must be done separately for MIN and MAX, immediately after next_max() was called, because both MIN and MAX take their result value from the same buffer this->head->record[0] (i.e. this->record). RETURN None */ void QUICK_GROUP_MIN_MAX_SELECT::update_max_result() { Item_sum *max_func; max_functions_it->rewind(); while ((max_func= (*max_functions_it)++)) max_func->reset_and_add(); } /* Append comma-separated list of keys this quick select uses to key_names; append comma-separated list of corresponding used lengths to used_lengths. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::add_keys_and_lengths() key_names [out] Names of used indexes used_lengths [out] Corresponding lengths of the index names DESCRIPTION This method is used by select_describe to extract the names of the indexes used by a quick select. */ void QUICK_GROUP_MIN_MAX_SELECT::add_keys_and_lengths(String *key_names, String *used_lengths) { bool first= TRUE; add_key_and_length(key_names, used_lengths, &first); } #ifndef DBUG_OFF static void print_sel_tree(PARAM *param, SEL_TREE *tree, key_map *tree_map, const char *msg) { SEL_ARG **key,**end; int idx; char buff[1024]; DBUG_ENTER("print_sel_tree"); String tmp(buff,sizeof(buff),&my_charset_bin); tmp.length(0); for (idx= 0,key=tree->keys, end=key+param->keys ; key != end ; key++,idx++) { if (tree_map->is_set(idx)) { uint keynr= param->real_keynr[idx]; if (tmp.length()) tmp.append(','); tmp.append(param->table->key_info[keynr].name); } } if (!tmp.length()) tmp.append(STRING_WITH_LEN("(empty)")); DBUG_PRINT("info", ("SEL_TREE: 0x%lx (%s) scans: %s", (long) tree, msg, tmp.c_ptr_safe())); DBUG_VOID_RETURN; } static void print_ror_scans_arr(TABLE *table, const char *msg, struct st_ror_scan_info **start, struct st_ror_scan_info **end) { DBUG_ENTER("print_ror_scans_arr"); char buff[1024]; String tmp(buff,sizeof(buff),&my_charset_bin); tmp.length(0); for (;start != end; start++) { if (tmp.length()) tmp.append(','); tmp.append(table->key_info[(*start)->keynr].name); } if (!tmp.length()) tmp.append(STRING_WITH_LEN("(empty)")); DBUG_PRINT("info", ("ROR key scans (%s): %s", msg, tmp.c_ptr())); DBUG_VOID_RETURN; } /***************************************************************************** ** Print a quick range for debugging ** TODO: ** This should be changed to use a String to store each row instead ** of locking the DEBUG stream ! *****************************************************************************/ static void print_key(KEY_PART *key_part, const uchar *key, uint used_length) { char buff[1024]; const uchar *key_end= key+used_length; uint store_length; TABLE *table= key_part->field->table; my_bitmap_map *old_sets[2]; dbug_tmp_use_all_columns(table, old_sets, table->read_set, table->write_set); for (; key < key_end; key+=store_length, key_part++) { String tmp(buff,sizeof(buff),&my_charset_bin); Field *field= key_part->field; store_length= key_part->store_length; if (field->real_maybe_null()) { if (*key) { fwrite("NULL",sizeof(char),4,DBUG_FILE); continue; } key++; // Skip null byte store_length--; } field->set_key_image(key, key_part->length); if (field->type() == MYSQL_TYPE_BIT) (void) field->val_int_as_str(&tmp, 1); else field->val_str(&tmp); fwrite(tmp.ptr(),sizeof(char),tmp.length(),DBUG_FILE); if (key+store_length < key_end) fputc('/',DBUG_FILE); } dbug_tmp_restore_column_maps(table->read_set, table->write_set, old_sets); } static void print_quick(QUICK_SELECT_I *quick, const key_map *needed_reg) { char buf[MAX_KEY/8+1]; TABLE *table; my_bitmap_map *old_sets[2]; DBUG_ENTER("print_quick"); if (!quick) DBUG_VOID_RETURN; DBUG_LOCK_FILE; table= quick->head; dbug_tmp_use_all_columns(table, old_sets, table->read_set, table->write_set); quick->dbug_dump(0, TRUE); dbug_tmp_restore_column_maps(table->read_set, table->write_set, old_sets); fprintf(DBUG_FILE,"other_keys: 0x%s:\n", needed_reg->print(buf)); DBUG_UNLOCK_FILE; DBUG_VOID_RETURN; } void QUICK_RANGE_SELECT::dbug_dump(int indent, bool verbose) { /* purecov: begin inspected */ fprintf(DBUG_FILE, "%*squick range select, key %s, length: %d\n", indent, "", head->key_info[index].name, max_used_key_length); if (verbose) { QUICK_RANGE *range; QUICK_RANGE **pr= (QUICK_RANGE**)ranges.buffer; QUICK_RANGE **end_range= pr + ranges.elements; for (; pr != end_range; ++pr) { fprintf(DBUG_FILE, "%*s", indent + 2, ""); range= *pr; if (!(range->flag & NO_MIN_RANGE)) { print_key(key_parts, range->min_key, range->min_length); if (range->flag & NEAR_MIN) fputs(" < ",DBUG_FILE); else fputs(" <= ",DBUG_FILE); } fputs("X",DBUG_FILE); if (!(range->flag & NO_MAX_RANGE)) { if (range->flag & NEAR_MAX) fputs(" < ",DBUG_FILE); else fputs(" <= ",DBUG_FILE); print_key(key_parts, range->max_key, range->max_length); } fputs("\n",DBUG_FILE); } } /* purecov: end */ } void QUICK_INDEX_SORT_SELECT::dbug_dump(int indent, bool verbose) { List_iterator_fast it(quick_selects); QUICK_RANGE_SELECT *quick; fprintf(DBUG_FILE, "%*squick index_merge select\n", indent, ""); fprintf(DBUG_FILE, "%*smerged scans {\n", indent, ""); while ((quick= it++)) quick->dbug_dump(indent+2, verbose); if (pk_quick_select) { fprintf(DBUG_FILE, "%*sclustered PK quick:\n", indent, ""); pk_quick_select->dbug_dump(indent+2, verbose); } fprintf(DBUG_FILE, "%*s}\n", indent, ""); } void QUICK_ROR_INTERSECT_SELECT::dbug_dump(int indent, bool verbose) { List_iterator_fast it(quick_selects); QUICK_SELECT_WITH_RECORD *qr; fprintf(DBUG_FILE, "%*squick ROR-intersect select, %scovering\n", indent, "", need_to_fetch_row? "":"non-"); fprintf(DBUG_FILE, "%*smerged scans {\n", indent, ""); while ((qr= it++)) qr->quick->dbug_dump(indent+2, verbose); if (cpk_quick) { fprintf(DBUG_FILE, "%*sclustered PK quick:\n", indent, ""); cpk_quick->dbug_dump(indent+2, verbose); } fprintf(DBUG_FILE, "%*s}\n", indent, ""); } void QUICK_ROR_UNION_SELECT::dbug_dump(int indent, bool verbose) { List_iterator_fast it(quick_selects); QUICK_SELECT_I *quick; fprintf(DBUG_FILE, "%*squick ROR-union select\n", indent, ""); fprintf(DBUG_FILE, "%*smerged scans {\n", indent, ""); while ((quick= it++)) quick->dbug_dump(indent+2, verbose); fprintf(DBUG_FILE, "%*s}\n", indent, ""); } /* Print quick select information to DBUG_FILE. SYNOPSIS QUICK_GROUP_MIN_MAX_SELECT::dbug_dump() indent Indentation offset verbose If TRUE show more detailed output. DESCRIPTION Print the contents of this quick select to DBUG_FILE. The method also calls dbug_dump() for the used quick select if any. IMPLEMENTATION Caller is responsible for locking DBUG_FILE before this call and unlocking it afterwards. RETURN None */ void QUICK_GROUP_MIN_MAX_SELECT::dbug_dump(int indent, bool verbose) { fprintf(DBUG_FILE, "%*squick_group_min_max_select: index %s (%d), length: %d\n", indent, "", index_info->name, index, max_used_key_length); if (key_infix_len > 0) { fprintf(DBUG_FILE, "%*susing key_infix with length %d:\n", indent, "", key_infix_len); } if (quick_prefix_select) { fprintf(DBUG_FILE, "%*susing quick_range_select:\n", indent, ""); quick_prefix_select->dbug_dump(indent + 2, verbose); } if (min_max_ranges.elements > 0) { fprintf(DBUG_FILE, "%*susing %d quick_ranges for MIN/MAX:\n", indent, "", min_max_ranges.elements); } } #endif /* !DBUG_OFF */ /***************************************************************************** ** Instantiate templates *****************************************************************************/ #ifdef HAVE_EXPLICIT_TEMPLATE_INSTANTIATION template class List; template class List_iterator; #endif