转载来自：
PostgreSQL服务过程中的那些事二：Pg服务进程处理简单查询六：执行器执行
blog.csdn.net/beiigang
beigang.iteye.com
话说查询“select cname,comp from test1,test2 where test1.id=test2.id;”发送到服务器端，走查询分支exec_simple_query，先调用start_xact_command初始化了事务管理相关对象和资源，接着调用pg_parse_query，通过Lex和Yacc对传入sql语句进行词法语法解析，生成解析树。下来调用GetTransactionSnapshot方法做内存快照，然后调用pg_analyze_and_rewrite方法，进行语义分析把parsetree转换成querytree，然后对该querytree进行重写。接着调用pg_plan_queries方法，根据querytree做查询规划，生成查询计划树plantree。然后调用了一系列方法CreatePortal、PortalStart、PortalSetResultFormat、CreateDestReceiver、PortalRun、PortalDrop，创建portal、初始化portal、设置结果列格式、创建目的地接收者、运行portal执行查询并返回结果，然后释放portal相关资源，再调用finish_xact_command释放事务相关资源。这个例子终于执行完了。
1
下面是执行plantree的调用序列图。
执行plantree的调用序列图
上图红色方框中显示了执行plantree的过程，主要分为portal创建和初始化（CreatePortal、PortalStart）、执行（PortalRun）、释放资源（PortalDrop）三个部分。portal是个抽象概念，它表示一个正在运行或可运行query的执行状态。Portal支持sql层的CURSOR和协议层的portal。从上图中可以看出，在各个部分，portal调用了执行器executor的对应方法ExecutorStart、ExecutorRun、ExecutorDrop完成了相关操作。
简单描述一下执行过程，方法CreatePortal在内存上下文PortalMemory里创建portal，并在portal管理器"Portal hash"中注册（参见《Postgresql服务过程中的那些事一：启动postgres服务进程一.七：初始化portal管理环境》）；调用PortalDefineQuery为新建的portal设置sourceText（就是例子里的源sql语句）、stmts（就是上一节讨论的plantree列表）等字段，将portal状态设置为PORTAL_DEFINED；再调用PortalStart初始化portal，在该例子里选择PORTAL_ONE_SELECT策略，调用CreateQueryDesc方法，为portal创建查询描述符，将portal状态设置为PORTAL_READY；调用方法PortalRun，根据选择策略调用相应方法，根据portal相关由plantree转化来的各节点结构由下至上逐层进行处理，涉及扫描数据、进行投射、结果缓存等；最后调用PortalDrop释放portal相关资源。
portal有多个执行策略，依赖于执行什么样的query（其实就是根据不同的sql语句调用不同的方法以完成相应操作）。（注意：在所有情况下，一个portal仅执行一个源sql query，并且从用户的视点，仅产生一个结果。但是，规则重写器可以扩展一个源query成0或多个实际的query。）下面是portal的四种策略
PORTAL_ONE_SELECT：这个portal包含一个SELECT查询。我们增加运行一个执行器作为被要求的结果。这个策略还支持可持有的cursor（为了事务结束后访问结果，执行器结果可以被转储到一个tuplestore里）。
PORTAL_ONE_RETURNING：这个portal包含一个伴有RETURNING子句的INSERT/UPDATE/DELETE查询（由规则重写器重写增加的可能的辅助查询/query）。在第一次执行时，我们运行portal以完成和转储主query的结果到portal的tuplestore里；然后这个结果按要求返回给用户。（我们不支持query的部分遍历悬挂，因为AFTER触发器代码不能处理，并且还因为我们不想在执行所有辅助query时冒失败风险。）
PORTAL_ONE_MOD_WITH：portal包含一个SELECT qiery，但它包含数据修改CTE。这个目前和处理PORTAL_ONE_RETURNING的情况一样，因为需要触发触发器的可能性。将来处理这种情况的行为可以更像PORTAL_ONE_SELECT。
PORTAL_UTIL_SELECT：portal包含一个utility语句，其返回一个象SELECT那样的结果（例如，WXPLAIN或者SHOW）。在第一次执行时，我们运行这个语句任何转储其结果到portal的tuplestore里；然后这个结果按要求的返回给客户端。
PORTAL_MULTI_QUERY：所有其它情况。这儿，我们不支持portal执行：portal的query会被运行以完成第一次调用。
下面是portal和其执行状态、策略的类型定义及执行器状态结构定义，其他涉及到节点结构定义略去。
typedefenumPortalStrategy
{
PORTAL_ONE_SELECT,
PORTAL_ONE_RETURNING,
PORTAL_ONE_MOD_WITH,
PORTAL_UTIL_SELECT,
PORTAL_MULTI_QUERY
}PortalStrategy;
typedefenumPortalStatus
{
PORTAL_NEW,/* freshly created */
PORTAL_DEFINED,/* PortalDefineQuery done */
PORTAL_READY,/* PortalStart complete,can run it */
PORTAL_ACTIVE,/* portal is running (can't delete it) */
PORTAL_DONE,/* portal is finished (don't re-run it) */
PORTAL_Failed/* portal got error (can't re-run it) */
}PortalStatus;
typedefstructPortalData *Portal;
typedefstructPortalData
{
/* Bookkeeping data */
constchar*name;/* portal's name */
prepStmtName;/* source prepared statement (NULL if none) */
MemoryContextheap;/* subsidiary memory for portal */
ResourceOwnerresowner;/* resources owned by portal */
void(*cleanup) (Portalportal);/* cleanup hook */
SubTransactionIdcreateSubid;/* the ID of the creatingsubxact*/
/*
* if createSubid is InvalidSubTransactionId,the portal is held over from
* a prevIoUs transaction
*/
/* The query or queries the portal will execute */
sourceText;/* text of query (as of 8.4,never NULL) */
commandTag;/* command tag for original query */
List*stmts;/* PlannedStmts and/or utility statements */
CachedPlan*cplan;/* CachedPlan,ifstmtsare from one */
ParamListInfoportalParams;/*paramsto pass to query */
/* Features/options */
PortalStrategystrategy;/* see above */
intcursorOptions;/* DECLARE CURSOR option bits */
/* Status data */
PortalStatusstatus;/* see above */
boolportalPinned;/* a pinned portal can't be dropped */
/* If not NULL,Executor is active; call ExecutorEnd eventually: */
QueryDesc*queryDesc;/* info needed for executor invocation */
/* If portal returns tuples,this is theirtupdesc: */
TupleDesctupDesc;/* descriptor for result tuples */
/* and these are the format codes to use for the columns: */
int16*formats;/* a format code for each column */
* Where we store tuples for a held cursor or a PORTAL_ONE_RETURNING or
* PORTAL_UTIL_SELECT query.(A cursor held past the end of its
* transaction no longer has any active executor state.)
*/
Tuplestorestate*holdStore;/* store forholdablecursors */
holdContext;/* memory containing holdStore */
* atStart,atEnd and portalPos indicate the current cursor position.
* portalPos is zero before the first row,N after fetching N'th row of
* query.After we run off the end,portalPos = # of rows in query,and
* atEnd is true.If portalPos overflows,set posOverflow (this causes us
* to stop relying on its value for navigation).Note that atStart
* implies portalPos == 0,but not the reverse (portalPos could have
* overflowed).
atStart;
atEnd;
posOverflow;
longportalPos;
/* Presentation data,primarily used by the pg_cursors system view */
TimestampTzcreation_time;/* time at which this portal was defined */
visible;/* include this portal in pg_cursors? */
}PortalData;
执行器调用的主工作状态
typedefstructEState
{
NodeTagtype;
/* Basic state for all query types: */
ScanDirectiones_direction;/* current scan direction */
Snapshotes_snapshot;/* timequalto use */
es_crosscheck_snapshot;/*crosschecktimequalfor RI */
es_range_table;/* List of RangeTblEntry */
PlannedStmt*es_plannedstmt;/* link to top of plan tree */
JunkFilter*es_junkFilter;/* top-level junk filter,if any */
/* If query can insert/delete tuples,the command ID to mark them with */
CommandIdes_output_cid;
/* Info about target table(s) for insert/update/delete queries: */
ResultRelInfo*es_result_relations;/* array of ResultRelInfos */
es_num_result_relations;/* length of array */
es_result_relation_info;/* currently active arrayelt*/
/* Stuff used for firing triggers: */
es_trig_target_relations;/* trigger-only ResultRelInfos */
TupleTableSlot*es_trig_tuple_slot;/* for trigger output tuples */
es_trig_oldtup_slot;/* for TriggerEnabled */
/* Parameter info: */
es_param_list_info;/* values of externalparams*/
ParamExecData*es_param_exec_vals;/* values of internalparams*/
/* Other working state: */
es_query_cxt;/* per-query context in which EState lives */
es_tupleTable;/* List of TupleTableSlots */
es_rowMarks;/* List of ExecRowMarks */
uint32es_processed;/* # of tuples processed */
Oides_lastoid;/* lastoidprocessed (by INSERT) */
es_top_eflags;/*eflagspassed to ExecutorStart */
es_instrument;/* OR of InstrumentOption flags */
es_select_into;/* true if doing SELECT INTO */
es_into_oids;/* true to generate OIDs in SELECT INTO */
es_finished;/* true when ExecutorFinish is done */
es_exprcontexts;/* List of ExprContexts within EState */
es_subplanstates;/* List of PlanState for SubPlans */
es_auxmodifytables;/* List of secondary ModifyTableStates */
* this ExprContext is for per-output-tuple operations,such as constraint
* checks and index-value computations.It will be reset for each output
* tuple.Note that it will be created only if needed.
ExprContext*es_per_tuple_exprcontext;
* These fields are for re-evaluating planqualswhen an updated tuple is
* substituted in READ COMMITTED mode.es_epqTuple[] contains tuples that
* scan plan nodes should return instead of whatever they'd normally
* return,or NULL if nothing to return; es_epqTupleSet[] is true if a
* particular array entry is valid; and es_epqScanDone[] is state to
* remember if the tuple has been returned already.Arrays are of size
* list_length(es_range_table) and are indexed by scan nodescanrelid- 1.
HeapTuple*es_epqTuple;/* array of EPQ substitute tuples */
bool*es_epqTupleSet;/* true if EPQ tuple is provided */
es_epqScanDone;/* true if EPQ tuple has been fetched */
* this field added at end ofstructto avoid post-release ABI breakage in
* existing release branches.It'll be in a more logical place in 9.2.
es_trig_newtup_slot;/* for TriggerEnabled */
}EState;
下面是执行这个查询的portal相关内存结构图：
portal相关内存结构图
就到这儿吧。
