PostgreSQL 源码解读（58）- 查询语句#43（make_one_rel函数#8-B...

这一小节继续介绍查询物理优化中的create_index_paths->create_bitmap_heap_path函数,该函数创建位图堆扫描访问路径节点。
关于BitmapHeapScan的相关知识,请参照PostgreSQL DBA(6) - SeqScan vs IndexScan vs BitmapHeapScan这篇文章.
本节没有描述具体的Cost成本计算方法(公式)，后续再行详述。

一、数据结构

Cost相关
注意:实际使用的参数值通过系统配置文件定义,而不是这里的常量定义!

 typedef double Cost; /* execution cost (in page-access units) */ /* defaults for costsize.c's Cost parameters */ /* NB: cost-estimation code should use the variables, not these constants! */ /* 注意:实际值通过系统配置文件定义,而不是这里的常量定义! */ /* If you change these, update backend/utils/misc/postgresql.sample.conf */ #define DEFAULT_SEQ_PAGE_COST  1.0       //顺序扫描page的成本 #define DEFAULT_RANDOM_PAGE_COST  4.0      //随机扫描page的成本 #define DEFAULT_CPU_TUPLE_COST  0.01     //处理一个元组的CPU成本 #define DEFAULT_CPU_INDEX_TUPLE_COST 0.005   //处理一个索引元组的CPU成本 #define DEFAULT_CPU_OPERATOR_COST  0.0025    //执行一次操作或函数的CPU成本 #define DEFAULT_PARALLEL_TUPLE_COST 0.1    //并行执行,从一个worker传输一个元组到另一个worker的成本 #define DEFAULT_PARALLEL_SETUP_COST  1000.0  //构建并行执行环境的成本  #define DEFAULT_EFFECTIVE_CACHE_SIZE  524288    /*先前已有介绍, measured in pages */ double      seq_page_cost = DEFAULT_SEQ_PAGE_COST; double      random_page_cost = DEFAULT_RANDOM_PAGE_COST; double      cpu_tuple_cost = DEFAULT_CPU_TUPLE_COST; double      cpu_index_tuple_cost = DEFAULT_CPU_INDEX_TUPLE_COST; double      cpu_operator_cost = DEFAULT_CPU_OPERATOR_COST; double      parallel_tuple_cost = DEFAULT_PARALLEL_TUPLE_COST; double      parallel_setup_cost = DEFAULT_PARALLEL_SETUP_COST;  int         effective_cache_size = DEFAULT_EFFECTIVE_CACHE_SIZE; Cost        disable_cost = 1.0e10;//1后面10个0,通过设置一个巨大的成本,让优化器自动放弃此路径  int         max_parallel_workers_per_gather = 2;//每次gather使用的worker数

二、源码解读

create_bitmap_heap_path函数
create_index_paths->create_bitmap_heap_path函数,创建位图堆扫描访问路径节点.

 /*  * create_bitmap_heap_path  *    Creates a path node for a bitmap scan.  *    创建位图堆扫描访问路径节点  *  * 'bitmapqual' is a tree of IndexPath, BitmapAndPath, and BitmapOrPath nodes.  * bitmapqual-IndexPath, BitmapAndPath, and BitmapOrPath节点组成的树  * 'required_outer' is the set of outer relids for a parameterized path.  * required_outer-参数化路径中依赖的外部relids  * 'loop_count' is the number of repetitions of the indexscan to factor into  *      estimates of caching behavior.  * loop_count-上一节已介绍  *  * loop_count should match the value used when creating the component  * IndexPaths.  */ BitmapHeapPath * create_bitmap_heap_path(PlannerInfo *root,                         RelOptInfo *rel,                         Path *bitmapqual,                         Relids required_outer,                         double loop_count,                         int parallel_degree) {     BitmapHeapPath *pathnode = makeNode(BitmapHeapPath);//创建节点      pathnode->path.pathtype = T_BitmapHeapScan;     pathnode->path.parent = rel;     pathnode->path.pathtarget = rel->reltarget;     pathnode->path.param_info = get_baserel_parampathinfo(root, rel,                                                           required_outer);     pathnode->path.parallel_aware = parallel_degree > 0 ? true : false;     pathnode->path.parallel_safe = rel->consider_parallel;     pathnode->path.parallel_workers = parallel_degree;     pathnode->path.pathkeys = NIL;  /* always unordered */      pathnode->bitmapqual = bitmapqual;      cost_bitmap_heap_scan(&pathnode->path, root, rel,                           pathnode->path.param_info,                           bitmapqual, loop_count);//成本估算      return pathnode;//返回结果 } //-------------------------------------------------------- cost_bitmap_heap_scan /*  * cost_bitmap_heap_scan  *    Determines and returns the cost of scanning a relation using a bitmap  *    index-then-heap plan.  *    确定并返回使用BitmapIndexScan和BitmapHeapScan的成本.  *  * 'baserel' is the relation to be scanned  * baserel-需扫描的Relation  * 'param_info' is the ParamPathInfo if this is a parameterized path, else NULL  * param_info-参数化信息  * 'bitmapqual' is a tree of IndexPaths, BitmapAndPaths, and BitmapOrPaths  * bitmapqual-位图条件表达式,IndexPath, BitmapAndPath, and BitmapOrPath节点组成的树  * 'loop_count' is the number of repetitions of the indexscan to factor into  *      estimates of caching behavior  *  * Note: the component IndexPaths in bitmapqual should have been costed  * using the same loop_count.  */ void cost_bitmap_heap_scan(Path *path, PlannerInfo *root, RelOptInfo *baserel,                       ParamPathInfo *param_info,                       Path *bitmapqual, double loop_count) {     Cost        startup_cost = 0;//启动成本     Cost        run_cost = 0;//执行成本     Cost        indexTotalCost;//索引扫描总成本     QualCost    qpqual_cost;//表达式成本     Cost        cpu_per_tuple;     Cost        cost_per_page;     Cost        cpu_run_cost;     double      tuples_fetched;     double      pages_fetched;     double      spc_seq_page_cost,                 spc_random_page_cost;     double      T;      /* Should only be applied to base relations */     Assert(IsA(baserel, RelOptInfo));     Assert(baserel->relid > 0);     Assert(baserel->rtekind == RTE_RELATION);      /* Mark the path with the correct row estimate */     if (param_info)         path->rows = param_info->ppi_rows;     else         path->rows = baserel->rows;      if (!enable_bitmapscan)//不允许位图扫描         startup_cost += disable_cost;//禁用之      pages_fetched = compute_bitmap_pages(root, baserel, bitmapqual,                                          loop_count, &indexTotalCost,                                          &tuples_fetched);//计算页面数      startup_cost += indexTotalCost;//启动成本为BitmapIndexScan的总成本     T = (baserel->pages > 1) ? (double) baserel->pages : 1.0;//页面数      /* Fetch estimated page costs for tablespace containing table. */     get_tablespace_page_costs(baserel->reltablespace,                               &spc_random_page_cost,                               &spc_seq_page_cost);//访问表空间页面成本      /*      * For small numbers of pages we should charge spc_random_page_cost      * apiece, while if nearly all the table's pages are being read, it's more      * appropriate to charge spc_seq_page_cost apiece.  The effect is      * nonlinear, too. For lack of a better idea, interpolate like this to      * determine the cost per page.      * 对于少量的页面，每个页面的成本为spc_random_page_cost，      * 而如果几乎所有的页面都被读取，则每个页面的成本为spc_seq_page_cost。      * 这种影响也是非线性的。由于缺乏更好的方法，通过插值法确定每页的成本。      */     if (pages_fetched >= 2.0)         cost_per_page = spc_random_page_cost -             (spc_random_page_cost - spc_seq_page_cost)             * sqrt(pages_fetched / T);     else         cost_per_page = spc_random_page_cost;      run_cost += pages_fetched * cost_per_page;//执行成本      /*      * Estimate CPU costs per tuple.      * 为每个元组估算CPU成本(Rechck步骤的成本)      *       * Often the indexquals don't need to be rechecked at each tuple ... but      * not always, especially not if there are enough tuples involved that the      * bitmaps become lossy.  For the moment, just assume they will be      * rechecked always.  This means we charge the full freight for all the      * scan clauses.       * 通常情况下，索引约束条件不需要在每个元组上重新检查,但现实并非如此理想，      * 尤其是当涉及到较多的元组时。就目前而言，      * 优化器会假设它们总是会被重新检查。这意味着我们需要为所有扫描条件计算成本。      */     get_restriction_qual_cost(root, baserel, param_info, &qpqual_cost);//获取条件表达式      startup_cost += qpqual_cost.startup;//增加启动成本     cpu_per_tuple = cpu_tuple_cost + qpqual_cost.per_tuple;//增加处理每个元组的CPU成本     cpu_run_cost = cpu_per_tuple * tuples_fetched;//CPU运行成本      /* Adjust costing for parallelism, if used. */     if (path->parallel_workers > 0)//是否并行?     {         double      parallel_divisor = get_parallel_divisor(path);          /* The CPU cost is divided among all the workers. */         cpu_run_cost /= parallel_divisor;          path->rows = clamp_row_est(path->rows / parallel_divisor);     }      //计算最终成本     run_cost += cpu_run_cost;      /* tlist eval costs are paid per output row, not per tuple scanned */     startup_cost += path->pathtarget->cost.startup;     run_cost += path->pathtarget->cost.per_tuple * path->rows;      path->startup_cost = startup_cost;     path->total_cost = startup_cost + run_cost; }//--------------------------------------- compute_bitmap_pages /*  * compute_bitmap_pages  *   * compute number of pages fetched from heap in bitmap heap scan.  * 计算页面数  */ double compute_bitmap_pages(PlannerInfo *root, RelOptInfo *baserel, Path *bitmapqual,                      int loop_count, Cost *cost, double *tuple) {     Cost        indexTotalCost;     Selectivity indexSelectivity;     double      T;     double      pages_fetched;     double      tuples_fetched;     double      heap_pages;     long        maxentries;      /*      * Fetch total cost of obtaining the bitmap, as well as its total      * selectivity.      * 获取位图的总成本，以及它的总选择性。      */     cost_bitmap_tree_node(bitmapqual, &indexTotalCost, &indexSelectivity);      /*      * Estimate number of main-table pages fetched.      * 估算主表的页面数      */     tuples_fetched = clamp_row_est(indexSelectivity * baserel->tuples);//计算总元组数      T = (baserel->pages > 1) ? (double) baserel->pages : 1.0;      /*      * For a single scan, the number of heap pages that need to be fetched is      * the same as the Mackert and Lohman formula for the case T <= b (ie, no      * re-reads needed).      * 对于单个扫描，需要获取的堆页面数量与T <= b(即不需要重新读取)的Mackert和Lohman公式相同。      */     pages_fetched = (2.0 * T * tuples_fetched) / (2.0 * T + tuples_fetched);      /*      * Calculate the number of pages fetched from the heap.  Then based on      * current work_mem estimate get the estimated maxentries in the bitmap.      * (Note that we always do this calculation based on the number of pages      * that would be fetched in a single iteration, even if loop_count > 1.      * That's correct, because only that number of entries will be stored in      * the bitmap at one time.)      * 计算从堆中读取的页面数.      * 根据当前的work_mem估算得到位图中粗略的最大访问入口(entries)。      * (请注意，我们总是根据单个迭代中获取的页面数来进行计算，      * 即使loop_count > 1也是如此。因为只有该数量的条目在位图中只存储一次。      */     heap_pages = Min(pages_fetched, baserel->pages);//堆页面数     maxentries = tbm_calculate_entries(work_mem * 1024L);//位图最大入口数      if (loop_count > 1)     {         /*          * For repeated bitmap scans, scale up the number of tuples fetched in          * the Mackert and Lohman formula by the number of scans, so that we          * estimate the number of pages fetched by all the scans. Then          * pro-rate for one scan.          */         pages_fetched = index_pages_fetched(tuples_fetched * loop_count,                                             baserel->pages,                                             get_indexpath_pages(bitmapqual),                                             root);         pages_fetched /= loop_count;     }      if (pages_fetched >= T)         pages_fetched = T;//数据字典中的页面数     else         pages_fetched = ceil(pages_fetched);      if (maxentries < heap_pages)//最大入口数小于堆页面数     {         double      exact_pages;         double      lossy_pages;          /*          * Crude approximation of the number of lossy pages.  Because of the          * way tbm_lossify() is coded, the number of lossy pages increases          * very sharply as soon as we run short of memory; this formula has          * that property and seems to perform adequately in testing, but it's          * possible we could do better somehow.          * 粗略估计缺页的数目。由于tbm_lossify()的编码方式，          * 一旦内存不足，缺页的数量就会急剧增加;          * 这个公式有这个性质，在测试中表现得很好，但有可能做得更好。          */         lossy_pages = Max(0, heap_pages - maxentries / 2);         exact_pages = heap_pages - lossy_pages;          /*          * If there are lossy pages then recompute the  number of tuples          * processed by the bitmap heap node.  We assume here that the chance          * of a given tuple coming from an exact page is the same as the          * chance that a given page is exact.  This might not be true, but          * it's not clear how we can do any better.          * 如果存在缺页面，则重新计算位图堆节点处理的元组数量。          * 这里假设给定元组来自精确页面的概率与给定页面的概率相同。          * 但这可能不符合实际情况，但我们不清楚如何才能做得更好:(          */         if (lossy_pages > 0)             tuples_fetched =                 clamp_row_est(indexSelectivity *                               (exact_pages / heap_pages) * baserel->tuples +                               (lossy_pages / heap_pages) * baserel->tuples);     }      if (cost)         *cost = indexTotalCost;     if (tuple)         *tuple = tuples_fetched;      return pages_fetched; }//--------------------------- tbm_calculate_entries /*  * tbm_calculate_entries  *  * Estimate number of hashtable entries we can have within maxbytes.  */ long tbm_calculate_entries(double maxbytes) {     long        nbuckets;      /*      * Estimate number of hashtable entries we can have within maxbytes. This      * estimates the hash cost as sizeof(PagetableEntry), which is good enough      * for our purpose.  Also count an extra Pointer per entry for the arrays      * created during iteration readout.      * 估计maxbytes中可以包含的哈希表条目的数量。      * 这将散列成本估计为sizeof(PagetableEntry)，这已经足够好了。      * 还要为迭代读出期间创建的数组中每个条目计算额外的指针。      */     nbuckets = maxbytes /         (sizeof(PagetableEntry) + sizeof(Pointer) + sizeof(Pointer));//桶数     nbuckets = Min(nbuckets, INT_MAX - 1);  /* safety limit */     nbuckets = Max(nbuckets, 16);   /* sanity limit */      return nbuckets; }//--------------------------- cost_bitmap_tree_node /*  * cost_bitmap_tree_node  *      Extract cost and selectivity from a bitmap tree node (index/and/or)  */ void cost_bitmap_tree_node(Path *path, Cost *cost, Selectivity *selec) {     if (IsA(path, IndexPath))//索引访问路径     {         *cost = ((IndexPath *) path)->indextotalcost;         *selec = ((IndexPath *) path)->indexselectivity;          /*          * Charge a small amount per retrieved tuple to reflect the costs of          * manipulating the bitmap.  This is mostly to make sure that a bitmap          * scan doesn't look to be the same cost as an indexscan to retrieve a          * single tuple.          * 对每个检索到的元组计算少量成本，以反映操作位图的成本。          * 这主要是为了确保位图扫描与索引扫描检索单个元组的成本不一样。          */         *cost += 0.1 * cpu_operator_cost * path->rows;     }     else if (IsA(path, BitmapAndPath))//BitmapAndPath     {         *cost = path->total_cost;         *selec = ((BitmapAndPath *) path)->bitmapselectivity;     }     else if (IsA(path, BitmapOrPath))//BitmapOrPath     {         *cost = path->total_cost;         *selec = ((BitmapOrPath *) path)->bitmapselectivity;     }     else     {         elog(ERROR, "unrecognized node type: %d", nodeTag(path));         *cost = *selec = 0;     /* keep compiler quiet */     } }

三、跟踪分析

测试脚本如下

select t1.* from t_dwxx t1 where dwbh > '10000' and dwbh < '30000';

启动gdb跟踪

(gdb) b create_bitmap_heap_pathBreakpoint 1 at 0x78f1c1: file pathnode.c, line 1090.(gdb) cContinuing.Breakpoint 1, create_bitmap_heap_path (root=0x23d93d8, rel=0x248a788, bitmapqual=0x2473a08, required_outer=0x0,     loop_count=1, parallel_degree=0) at pathnode.c:10901090        BitmapHeapPath *pathnode = makeNode(BitmapHeapPath);

创建节点,并赋值

1090        BitmapHeapPath *pathnode = makeNode(BitmapHeapPath);(gdb) n1092        pathnode->path.pathtype = T_BitmapHeapScan;(gdb) n1093        pathnode->path.parent = rel;(gdb) n1094        pathnode->path.pathtarget = rel->reltarget;(gdb) n1095        pathnode->path.param_info = get_baserel_parampathinfo(root, rel,(gdb) 1097        pathnode->path.parallel_aware = parallel_degree > 0 ? true : false;(gdb) 1098        pathnode->path.parallel_safe = rel->consider_parallel;(gdb) 1099        pathnode->path.parallel_workers = parallel_degree;(gdb) 1100        pathnode->path.pathkeys = NIL;  /* always unordered */(gdb) 1102        pathnode->bitmapqual = bitmapqual;

进入cost_bitmap_heap_scan函数

(gdb) 1104        cost_bitmap_heap_scan(&pathnode->path, root, rel,(gdb) stepcost_bitmap_heap_scan (path=0x24737d8, root=0x23d93d8, baserel=0x248a788, param_info=0x0, bitmapqual=0x2473a08,     loop_count=1) at costsize.c:949949     Cost        startup_cost = 0;

输入参数,其中bitmapqual为T_IndexPath节点
路径的其他关键信息:rows = 2223, startup_cost = 0.28500000000000003, total_cost = 169.23871600907944

(gdb) p *(IndexPath *)bitmapqual$2 = {path = {type = T_IndexPath, pathtype = T_IndexScan, parent = 0x248a788, pathtarget = 0x248a998, param_info = 0x0,     parallel_aware = false, parallel_safe = true, parallel_workers = 0, rows = 2223, startup_cost = 0.28500000000000003,     total_cost = 169.23871600907944, pathkeys = 0x0}, indexinfo = 0x23b63b8, indexclauses = 0x2473948,   indexquals = 0x2473b38, indexqualcols = 0x2473b88, indexorderbys = 0x0, indexorderbycols = 0x0,   indexscandir = ForwardScanDirection, indextotalcost = 50.515000000000001, indexselectivity = 0.22227191011235958}

开始计算成本

...980     startup_cost += indexTotalCost;(gdb) p indexTotalCost$16 = 51.070750000000004(gdb) p startup_cost$17 = 0(gdb) p pages_fetched$18 = 64(gdb) p baserel->pages$19 = 64...(gdb) p qpqual_cost$20 = {startup = 0, per_tuple = 0.0050000000000000001}

最终的访问路径信息

(gdb) p *(BitmapHeapPath *)path$22 = {path = {type = T_BitmapHeapPath, pathtype = T_BitmapHeapScan, parent = 0x248a788, pathtarget = 0x248a998,     param_info = 0x0, parallel_aware = false, parallel_safe = true, parallel_workers = 0, rows = 2223,     startup_cost = 51.070750000000004, total_cost = 148.41575, pathkeys = 0x0}, bitmapqual = 0x2473a08}

除了BitmapHeapPath,还有BitmapOr和BitmapAnd,这两种Path的解析后续再详述.

四、参考资料

allpaths.c
cost.h
costsize.c
PG Document:Query Planning