PostgreSQL 源码解读（75）- 查询语句#60（Review - standard_...

本节Review standard_planner函数的实现逻辑，该函数查询优化器的主入口。

一、源码解读

standard_planner函数由exec_simple_query->pg_plan_queries->pg_plan_query->planner函数调用,其调用栈如下:

(gdb) b planner
Breakpoint 1 at 0x7c7442: file planner.c, line 259.
(gdb) c
Continuing.

Breakpoint 1, planner (parse=0x2eee0b0, cursorOptions=256, boundParams=0x0) at planner.c:259
259     if (planner_hook)
(gdb) bt
#0  planner (parse=0x2eee0b0, cursorOptions=256, boundParams=0x0) at planner.c:259
#1  0x00000000008c5824 in pg_plan_query (querytree=0x2eee0b0, cursorOptions=256, boundParams=0x0) at postgres.c:809
#2  0x00000000008c595d in pg_plan_queries (querytrees=0x2f10ac8, cursorOptions=256, boundParams=0x0) at postgres.c:875
#3  0x00000000008c5c35 in exec_simple_query (query_string=0x2eed1a8 "select * from t1;") at postgres.c:1050
#4  0x00000000008ca0f8 in PostgresMain (arGC=1, argv=0x2f16cb8, dbname=0x2f16b20 "testdb", username=0x2ee9d48 "xdb")
    at postgres.c:4159
#5  0x0000000000825880 in BackendRun (port=0x2f0eb00) at postmaster.c:4361
#6  0x0000000000824fe4 in BackendStartup (port=0x2f0eb00) at postmaster.c:4033
#7  0x0000000000821371 in ServerLoop () at postmaster.c:1706
#8  0x0000000000820c09 in PostmasterMain (argc=1, argv=0x2ee7d00) at postmaster.c:1379
#9  0x0000000000747ea7 in main (argc=1, argv=0x2ee7d00) at main.c:228

该函数根据输入的查询树(Query)生成已规划的语句,可用于执行器实际执行.

PlannedStmt *
planner(Query *parse, int cursorOptions, ParamListInfo boundParams)
{
    PlannedStmt *result;

    if (planner_hook)//钩子函数
        result = (*planner_hook) (parse, cursorOptions, boundParams);
    else
        result = standard_planner(parse, cursorOptions, boundParams);
    return result;
}

PlannedStmt *
standard_planner(Query *parse, int cursorOptions, ParamListInfo boundParams)//标准优化过程
{
    PlannedStmt *result;//返回结果
    PlannerGlobal *glob;//全局的Plan信息-Global infORMation for planning/optimization
    double      tuple_fraction;//
    PlannerInfo *root;//每个Query的Plan信息-Per-query information for planning/optimization
    RelOptInfo *final_rel;//Plan中的每个Relation信息-Per-relation information for planning/optimization
    Path       *best_path;//最优路径
    Plan       *top_plan;//最上层的Plan
    ListCell   *lp,//临时变量
               *lr;

    
    glob = makenode(PlannerGlobal);//构建PlannerGlobal
     //初始化参数
    glob->boundParams = boundParams;
    glob->subplans = NIL;
    glob->subroots = NIL;
    glob->rewindPlanIDs = NULL;
    glob->finalrtable = NIL;
    glob->finalrowmarks = NIL;
    glob->resultRelations = NIL;
    glob->nonleafResultRelations = NIL;
    glob->rootResultRelations = NIL;
    glob->relationOids = NIL;
    glob->invalItems = NIL;
    glob->paramExecTypes = NIL;
    glob->lastPHId = 0;
    glob->lastRowMarkId = 0;
    glob->lastPlanNodeId = 0;
    glob->transientPlan = false;
    glob->dependsOnRole = false;

    
    if ((cursorOptions & CURSOR_OPT_PARALLEL_OK) != 0 &&
        IsUnderPostmaster &&    
        parse->commandType == CMD_SELECT &&
        !parse->hasModifyingCTE &&
        max_parallel_workers_per_gather > 0 &&
        !IsParallelWorker() &&
        !IsolationIsSerializable())//并行模式的判断
    {
        
        glob->maxParallelHazard = max_parallel_hazard(parse);
        glob->parallelModeOK = (glob->maxParallelHazard != PROPARALLEL_UNSAFE);
    }
    else
    {
        
        glob->maxParallelHazard = PROPARALLEL_UNSAFE;
        glob->parallelModeOK = false;
    }

    
    glob->parallelModeNeeded = glob->parallelModeOK &&
        (force_parallel_mode != FORCE_PARALLEL_OFF);

    
    //确定扫描比例
    if (cursorOptions & CURSOR_OPT_FAST_PLAN)
    {
        
        tuple_fraction = cursor_tuple_fraction;//使用GUC 参数

        
        if (tuple_fraction >= 1.0)
            tuple_fraction = 0.0;
        else if (tuple_fraction <= 0.0)
            tuple_fraction = 1e-10;
    }
    else
    {
        
        //默认假设:需要所有元组
        tuple_fraction = 0.0;
    }

    
    //主规划过程入口(可能会递归执行)
    root = subquery_planner(glob, parse, NULL,
                            false, tuple_fraction);//获取PlannerInfo根节点

    
    //选择最优路径并把该路径转换为执行计划
    final_rel = fetch_upper_rel(root, UPPERREL_FINAL, NULL);//获取顶层的RelOptInfo
    best_path = get_cheapest_fractional_path(final_rel, tuple_fraction);//选择最佳路径

    top_plan = create_plan(root, best_path);//生成执行计划

    
    if (cursorOptions & CURSOR_OPT_SCROLL)
    {
        if (!ExecSupportsBackwardScan(top_plan))
            top_plan = materialize_finished_plan(top_plan);
    }

    
    if (force_parallel_mode != FORCE_PARALLEL_OFF && top_plan->parallel_safe)
    {
        Gather     *gather = makeNode(Gather);

        
        gather->plan.initPlan = top_plan->initPlan;
        top_plan->initPlan = NIL;

        gather->plan.targetlist = top_plan->targetlist;
        gather->plan.qual = NIL;
        gather->plan.lefttree = top_plan;
        gather->plan.righttree = NULL;
        gather->num_workers = 1;
        gather->single_copy = true;
        gather->invisible = (force_parallel_mode == FORCE_PARALLEL_REGRESS);

        
        gather->rescan_param = -1;

        
        gather->plan.startup_cost = top_plan->startup_cost +
            parallel_setup_cost;
        gather->plan.total_cost = top_plan->total_cost +
            parallel_setup_cost + parallel_tuple_cost * top_plan->plan_rows;
        gather->plan.plan_rows = top_plan->plan_rows;
        gather->plan.plan_width = top_plan->plan_width;
        gather->plan.parallel_aware = false;
        gather->plan.parallel_safe = false;

        
        //使用并行模式
        root->glob->parallelModeNeeded = true;

        top_plan = &gather->plan;
    }

    
    if (glob->paramExecTypes != NIL)
    {
        Assert(list_length(glob->subplans) == list_length(glob->subroots));
        forboth(lp, glob->subplans, lr, glob->subroots)
        {
            Plan       *subplan = (Plan *) lfirst(lp);
            PlannerInfo *subroot = lfirst_node(PlannerInfo, lr);

            SS_finalize_plan(subroot, subplan);
        }
        SS_finalize_plan(root, top_plan);
    }

    
    //最后的清理工作
    Assert(glob->finalrtable == NIL);
    Assert(glob->finalrowmarks == NIL);
    Assert(glob->resultRelations == NIL);
    Assert(glob->nonleafResultRelations == NIL);
    Assert(glob->rootResultRelations == NIL);
    top_plan = set_plan_references(root, top_plan);
    
    Assert(list_length(glob->subplans) == list_length(glob->subroots));
    forboth(lp, glob->subplans, lr, glob->subroots)
    {
        Plan       *subplan = (Plan *) lfirst(lp);
        PlannerInfo *subroot = lfirst_node(PlannerInfo, lr);

        lfirst(lp) = set_plan_references(subroot, subplan);
    }

    
    //构建PlannedStmt结构
    result = makeNode(PlannedStmt);

    result->commandType = parse->commandType;//命令类型
    result->queryId = parse->queryId;
    result->hasReturning = (parse->returningList != NIL);
    result->hasModifyingCTE = parse->hasModifyingCTE;
    result->canSetTag = parse->canSetTag;
    result->transientPlan = glob->transientPlan;
    result->dependsOnRole = glob->dependsOnRole;
    result->parallelModeNeeded = glob->parallelModeNeeded;
    result->planTree = top_plan;//执行计划(这是后续执行sql使用到的最重要的地方)
    result->rtable = glob->finalrtable;
    result->resultRelations = glob->resultRelations;
    result->nonleafResultRelations = glob->nonleafResultRelations;
    result->rootResultRelations = glob->rootResultRelations;
    result->subplans = glob->subplans;
    result->rewindPlanIDs = glob->rewindPlanIDs;
    result->rowMarks = glob->finalrowmarks;
    result->relationOids = glob->relationOids;
    result->invalItems = glob->invalItems;
    result->paramExecTypes = glob->paramExecTypes;
    
    result->utilityStmt = parse->utilityStmt;
    result->stmt_location = parse->stmt_location;
    result->stmt_len = parse->stmt_len;

    result->jitFlags = PGJIT_NONE;
    if (jit_enabled && jit_above_cost >= 0 &&
        top_plan->total_cost > jit_above_cost)
    {
        result->jitFlags |= PGJIT_PERFORM;

        
        if (jit_optimize_above_cost >= 0 &&
            top_plan->total_cost > jit_optimize_above_cost)
            result->jitFlags |= PGJIT_OPT3;
        if (jit_inline_above_cost >= 0 &&
            top_plan->total_cost > jit_inline_above_cost)
            result->jitFlags |= PGJIT_INLINE;

        
        if (jit_expressions)
            result->jitFlags |= PGJIT_EXPR;
        if (jit_tuple_deforming)
            result->jitFlags |= PGJIT_DEFORM;
    }

    return result;
} 

pg_plan_queries&pg_plan_query

 
 List *
 pg_plan_queries(List *querytrees, int cursorOptions, ParamListInfo boundParams)
 {
     List       *stmt_list = NIL;
     ListCell   *query_list;
 
     foreach(query_list, querytrees)
     {
         Query      *query = lfirst_node(Query, query_list);
         PlannedStmt *stmt;
 
         if (query->commandType == CMD_UTILITY)
         {
             
             stmt = makeNode(PlannedStmt);
             stmt->commandType = CMD_UTILITY;
             stmt->canSetTag = query->canSetTag;
             stmt->utilityStmt = query->utilityStmt;
             stmt->stmt_location = query->stmt_location;
             stmt->stmt_len = query->stmt_len;
         }
         else
         {
             stmt = pg_plan_query(query, cursorOptions, boundParams);
         }
 
         stmt_list = lappend(stmt_list, stmt);
     }
 
     return stmt_list;
 }


PlannedStmt *
pg_plan_query(Query *querytree, int cursorOptions, ParamListInfo boundParams)
{
    PlannedStmt *plan;

    
    if (querytree->commandType == CMD_UTILITY)
        return NULL;

    
    Assert(ActiveSnapshotSet());

    TRACE_postgresql_QUERY_PLAN_START();

    if (log_planner_stats)
        ResetUsage();

    
    plan = planner(querytree, cursorOptions, boundParams);

    if (log_planner_stats)
        ShowUsage("PLANNER STATISTICS");

#ifdef COPY_PARSE_PLAN_TREES
    
    {
        PlannedStmt *new_plan = copyObject(plan);

        
#ifdef NOT_USED
        
        if (!equal(new_plan, plan))
            elog(WARNING, "copyObject() failed to produce an equal plan tree");
        else
#endif
            plan = new_plan;
    }
#endif

#ifdef WRITE_READ_PARSE_PLAN_TREES
    
    {
        char       *str;
        PlannedStmt *new_plan;

        str = nodeToString(plan);
        new_plan = stringToNodeWithLocations(str);
        pfree(str);

        
#ifdef NOT_USED
        
        if (!equal(new_plan, plan))
            elog(WARNING, "outfuncs/readfuncs failed to produce an equal plan tree");
        else
#endif
            plan = new_plan;
    }
#endif

    
    if (Debug_print_plan)
        elog_node_display(LOG, "plan", plan, Debug_pretty_print);

    TRACE_POSTGRESQL_QUERY_PLAN_DONE();

    return plan;
}

二、参考资料

allpaths.c
PG Document:Query Planning