Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * execParallel.c
4 : * Support routines for parallel execution.
5 : *
6 : * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group
7 : * Portions Copyright (c) 1994, Regents of the University of California
8 : *
9 : * This file contains routines that are intended to support setting up,
10 : * using, and tearing down a ParallelContext from within the PostgreSQL
11 : * executor. The ParallelContext machinery will handle starting the
12 : * workers and ensuring that their state generally matches that of the
13 : * leader; see src/backend/access/transam/README.parallel for details.
14 : * However, we must save and restore relevant executor state, such as
15 : * any ParamListInfo associated with the query, buffer usage info, and
16 : * the actual plan to be passed down to the worker.
17 : *
18 : * IDENTIFICATION
19 : * src/backend/executor/execParallel.c
20 : *
21 : *-------------------------------------------------------------------------
22 : */
23 :
24 : #include "postgres.h"
25 :
26 : #include "executor/execParallel.h"
27 : #include "executor/executor.h"
28 : #include "executor/nodeAppend.h"
29 : #include "executor/nodeBitmapHeapscan.h"
30 : #include "executor/nodeCustom.h"
31 : #include "executor/nodeForeignscan.h"
32 : #include "executor/nodeHash.h"
33 : #include "executor/nodeHashjoin.h"
34 : #include "executor/nodeIndexonlyscan.h"
35 : #include "executor/nodeIndexscan.h"
36 : #include "executor/nodeSeqscan.h"
37 : #include "executor/nodeSort.h"
38 : #include "executor/nodeSubplan.h"
39 : #include "executor/tqueue.h"
40 : #include "jit/jit.h"
41 : #include "nodes/nodeFuncs.h"
42 : #include "pgstat.h"
43 : #include "storage/spin.h"
44 : #include "tcop/tcopprot.h"
45 : #include "utils/datum.h"
46 : #include "utils/dsa.h"
47 : #include "utils/lsyscache.h"
48 : #include "utils/memutils.h"
49 : #include "utils/snapmgr.h"
50 :
51 : /*
52 : * Magic numbers for parallel executor communication. We use constants
53 : * greater than any 32-bit integer here so that values < 2^32 can be used
54 : * by individual parallel nodes to store their own state.
55 : */
56 : #define PARALLEL_KEY_EXECUTOR_FIXED UINT64CONST(0xE000000000000001)
57 : #define PARALLEL_KEY_PLANNEDSTMT UINT64CONST(0xE000000000000002)
58 : #define PARALLEL_KEY_PARAMLISTINFO UINT64CONST(0xE000000000000003)
59 : #define PARALLEL_KEY_BUFFER_USAGE UINT64CONST(0xE000000000000004)
60 : #define PARALLEL_KEY_TUPLE_QUEUE UINT64CONST(0xE000000000000005)
61 : #define PARALLEL_KEY_INSTRUMENTATION UINT64CONST(0xE000000000000006)
62 : #define PARALLEL_KEY_DSA UINT64CONST(0xE000000000000007)
63 : #define PARALLEL_KEY_QUERY_TEXT UINT64CONST(0xE000000000000008)
64 : #define PARALLEL_KEY_JIT_INSTRUMENTATION UINT64CONST(0xE000000000000009)
65 :
66 : #define PARALLEL_TUPLE_QUEUE_SIZE 65536
67 :
68 : /*
69 : * Fixed-size random stuff that we need to pass to parallel workers.
70 : */
71 : typedef struct FixedParallelExecutorState
72 : {
73 : int64 tuples_needed; /* tuple bound, see ExecSetTupleBound */
74 : dsa_pointer param_exec;
75 : int eflags;
76 : int jit_flags;
77 : } FixedParallelExecutorState;
78 :
79 : /*
80 : * DSM structure for accumulating per-PlanState instrumentation.
81 : *
82 : * instrument_options: Same meaning here as in instrument.c.
83 : *
84 : * instrument_offset: Offset, relative to the start of this structure,
85 : * of the first Instrumentation object. This will depend on the length of
86 : * the plan_node_id array.
87 : *
88 : * num_workers: Number of workers.
89 : *
90 : * num_plan_nodes: Number of plan nodes.
91 : *
92 : * plan_node_id: Array of plan nodes for which we are gathering instrumentation
93 : * from parallel workers. The length of this array is given by num_plan_nodes.
94 : */
95 : struct SharedExecutorInstrumentation
96 : {
97 : int instrument_options;
98 : int instrument_offset;
99 : int num_workers;
100 : int num_plan_nodes;
101 : int plan_node_id[FLEXIBLE_ARRAY_MEMBER];
102 : /* array of num_plan_nodes * num_workers Instrumentation objects follows */
103 : };
104 : #define GetInstrumentationArray(sei) \
105 : (AssertVariableIsOfTypeMacro(sei, SharedExecutorInstrumentation *), \
106 : (Instrumentation *) (((char *) sei) + sei->instrument_offset))
107 :
108 : /* Context object for ExecParallelEstimate. */
109 : typedef struct ExecParallelEstimateContext
110 : {
111 : ParallelContext *pcxt;
112 : int nnodes;
113 : } ExecParallelEstimateContext;
114 :
115 : /* Context object for ExecParallelInitializeDSM. */
116 : typedef struct ExecParallelInitializeDSMContext
117 : {
118 : ParallelContext *pcxt;
119 : SharedExecutorInstrumentation *instrumentation;
120 : int nnodes;
121 : } ExecParallelInitializeDSMContext;
122 :
123 : /* Helper functions that run in the parallel leader. */
124 : static char *ExecSerializePlan(Plan *plan, EState *estate);
125 : static bool ExecParallelEstimate(PlanState *node,
126 : ExecParallelEstimateContext *e);
127 : static bool ExecParallelInitializeDSM(PlanState *node,
128 : ExecParallelInitializeDSMContext *d);
129 : static shm_mq_handle **ExecParallelSetupTupleQueues(ParallelContext *pcxt,
130 : bool reinitialize);
131 : static bool ExecParallelReInitializeDSM(PlanState *planstate,
132 : ParallelContext *pcxt);
133 : static bool ExecParallelRetrieveInstrumentation(PlanState *planstate,
134 : SharedExecutorInstrumentation *instrumentation);
135 :
136 : /* Helper function that runs in the parallel worker. */
137 : static DestReceiver *ExecParallelGetReceiver(dsm_segment *seg, shm_toc *toc);
138 :
139 : /*
140 : * Create a serialized representation of the plan to be sent to each worker.
141 : */
142 : static char *
143 362 : ExecSerializePlan(Plan *plan, EState *estate)
144 : {
145 : PlannedStmt *pstmt;
146 : ListCell *lc;
147 :
148 : /* We can't scribble on the original plan, so make a copy. */
149 362 : plan = copyObject(plan);
150 :
151 : /*
152 : * The worker will start its own copy of the executor, and that copy will
153 : * insert a junk filter if the toplevel node has any resjunk entries. We
154 : * don't want that to happen, because while resjunk columns shouldn't be
155 : * sent back to the user, here the tuples are coming back to another
156 : * backend which may very well need them. So mutate the target list
157 : * accordingly. This is sort of a hack; there might be better ways to do
158 : * this...
159 : */
160 950 : foreach(lc, plan->targetlist)
161 : {
162 588 : TargetEntry *tle = lfirst_node(TargetEntry, lc);
163 :
164 588 : tle->resjunk = false;
165 : }
166 :
167 : /*
168 : * Create a dummy PlannedStmt. Most of the fields don't need to be valid
169 : * for our purposes, but the worker will need at least a minimal
170 : * PlannedStmt to start the executor.
171 : */
172 362 : pstmt = makeNode(PlannedStmt);
173 362 : pstmt->commandType = CMD_SELECT;
174 362 : pstmt->queryId = UINT64CONST(0);
175 362 : pstmt->hasReturning = false;
176 362 : pstmt->hasModifyingCTE = false;
177 362 : pstmt->canSetTag = true;
178 362 : pstmt->transientPlan = false;
179 362 : pstmt->dependsOnRole = false;
180 362 : pstmt->parallelModeNeeded = false;
181 362 : pstmt->planTree = plan;
182 362 : pstmt->rtable = estate->es_range_table;
183 362 : pstmt->resultRelations = NIL;
184 :
185 : /*
186 : * Transfer only parallel-safe subplans, leaving a NULL "hole" in the list
187 : * for unsafe ones (so that the list indexes of the safe ones are
188 : * preserved). This positively ensures that the worker won't try to run,
189 : * or even do ExecInitNode on, an unsafe subplan. That's important to
190 : * protect, eg, non-parallel-aware FDWs from getting into trouble.
191 : */
192 362 : pstmt->subplans = NIL;
193 382 : foreach(lc, estate->es_plannedstmt->subplans)
194 : {
195 20 : Plan *subplan = (Plan *) lfirst(lc);
196 :
197 20 : if (subplan && !subplan->parallel_safe)
198 16 : subplan = NULL;
199 20 : pstmt->subplans = lappend(pstmt->subplans, subplan);
200 : }
201 :
202 362 : pstmt->rewindPlanIDs = NULL;
203 362 : pstmt->rowMarks = NIL;
204 362 : pstmt->relationOids = NIL;
205 362 : pstmt->invalItems = NIL; /* workers can't replan anyway... */
206 362 : pstmt->paramExecTypes = estate->es_plannedstmt->paramExecTypes;
207 362 : pstmt->utilityStmt = NULL;
208 362 : pstmt->stmt_location = -1;
209 362 : pstmt->stmt_len = -1;
210 :
211 : /* Return serialized copy of our dummy PlannedStmt. */
212 362 : return nodeToString(pstmt);
213 : }
214 :
215 : /*
216 : * Parallel-aware plan nodes (and occasionally others) may need some state
217 : * which is shared across all parallel workers. Before we size the DSM, give
218 : * them a chance to call shm_toc_estimate_chunk or shm_toc_estimate_keys on
219 : * &pcxt->estimator.
220 : *
221 : * While we're at it, count the number of PlanState nodes in the tree, so
222 : * we know how many Instrumentation structures we need.
223 : */
224 : static bool
225 1662 : ExecParallelEstimate(PlanState *planstate, ExecParallelEstimateContext *e)
226 : {
227 1662 : if (planstate == NULL)
228 0 : return false;
229 :
230 : /* Count this node. */
231 1662 : e->nnodes++;
232 :
233 1662 : switch (nodeTag(planstate))
234 : {
235 : case T_SeqScanState:
236 640 : if (planstate->plan->parallel_aware)
237 492 : ExecSeqScanEstimate((SeqScanState *) planstate,
238 : e->pcxt);
239 640 : break;
240 : case T_IndexScanState:
241 192 : if (planstate->plan->parallel_aware)
242 8 : ExecIndexScanEstimate((IndexScanState *) planstate,
243 : e->pcxt);
244 192 : break;
245 : case T_IndexOnlyScanState:
246 28 : if (planstate->plan->parallel_aware)
247 24 : ExecIndexOnlyScanEstimate((IndexOnlyScanState *) planstate,
248 : e->pcxt);
249 28 : break;
250 : case T_ForeignScanState:
251 0 : if (planstate->plan->parallel_aware)
252 0 : ExecForeignScanEstimate((ForeignScanState *) planstate,
253 : e->pcxt);
254 0 : break;
255 : case T_AppendState:
256 100 : if (planstate->plan->parallel_aware)
257 68 : ExecAppendEstimate((AppendState *) planstate,
258 : e->pcxt);
259 100 : break;
260 : case T_CustomScanState:
261 0 : if (planstate->plan->parallel_aware)
262 0 : ExecCustomScanEstimate((CustomScanState *) planstate,
263 : e->pcxt);
264 0 : break;
265 : case T_BitmapHeapScanState:
266 10 : if (planstate->plan->parallel_aware)
267 8 : ExecBitmapHeapEstimate((BitmapHeapScanState *) planstate,
268 : e->pcxt);
269 10 : break;
270 : case T_HashJoinState:
271 112 : if (planstate->plan->parallel_aware)
272 64 : ExecHashJoinEstimate((HashJoinState *) planstate,
273 : e->pcxt);
274 112 : break;
275 : case T_HashState:
276 : /* even when not parallel-aware, for EXPLAIN ANALYZE */
277 112 : ExecHashEstimate((HashState *) planstate, e->pcxt);
278 112 : break;
279 : case T_SortState:
280 : /* even when not parallel-aware, for EXPLAIN ANALYZE */
281 86 : ExecSortEstimate((SortState *) planstate, e->pcxt);
282 86 : break;
283 :
284 : default:
285 382 : break;
286 : }
287 :
288 1662 : return planstate_tree_walker(planstate, ExecParallelEstimate, e);
289 : }
290 :
291 : /*
292 : * Estimate the amount of space required to serialize the indicated parameters.
293 : */
294 : static Size
295 8 : EstimateParamExecSpace(EState *estate, Bitmapset *params)
296 : {
297 : int paramid;
298 8 : Size sz = sizeof(int);
299 :
300 8 : paramid = -1;
301 28 : while ((paramid = bms_next_member(params, paramid)) >= 0)
302 : {
303 : Oid typeOid;
304 : int16 typLen;
305 : bool typByVal;
306 : ParamExecData *prm;
307 :
308 12 : prm = &(estate->es_param_exec_vals[paramid]);
309 12 : typeOid = list_nth_oid(estate->es_plannedstmt->paramExecTypes,
310 : paramid);
311 :
312 12 : sz = add_size(sz, sizeof(int)); /* space for paramid */
313 :
314 : /* space for datum/isnull */
315 12 : if (OidIsValid(typeOid))
316 12 : get_typlenbyval(typeOid, &typLen, &typByVal);
317 : else
318 : {
319 : /* If no type OID, assume by-value, like copyParamList does. */
320 0 : typLen = sizeof(Datum);
321 0 : typByVal = true;
322 : }
323 24 : sz = add_size(sz,
324 12 : datumEstimateSpace(prm->value, prm->isnull,
325 : typByVal, typLen));
326 : }
327 8 : return sz;
328 : }
329 :
330 : /*
331 : * Serialize specified PARAM_EXEC parameters.
332 : *
333 : * We write the number of parameters first, as a 4-byte integer, and then
334 : * write details for each parameter in turn. The details for each parameter
335 : * consist of a 4-byte paramid (location of param in execution time internal
336 : * parameter array) and then the datum as serialized by datumSerialize().
337 : */
338 : static dsa_pointer
339 8 : SerializeParamExecParams(EState *estate, Bitmapset *params, dsa_area *area)
340 : {
341 : Size size;
342 : int nparams;
343 : int paramid;
344 : ParamExecData *prm;
345 : dsa_pointer handle;
346 : char *start_address;
347 :
348 : /* Allocate enough space for the current parameter values. */
349 8 : size = EstimateParamExecSpace(estate, params);
350 8 : handle = dsa_allocate(area, size);
351 8 : start_address = dsa_get_address(area, handle);
352 :
353 : /* First write the number of parameters as a 4-byte integer. */
354 8 : nparams = bms_num_members(params);
355 8 : memcpy(start_address, &nparams, sizeof(int));
356 8 : start_address += sizeof(int);
357 :
358 : /* Write details for each parameter in turn. */
359 8 : paramid = -1;
360 28 : while ((paramid = bms_next_member(params, paramid)) >= 0)
361 : {
362 : Oid typeOid;
363 : int16 typLen;
364 : bool typByVal;
365 :
366 12 : prm = &(estate->es_param_exec_vals[paramid]);
367 12 : typeOid = list_nth_oid(estate->es_plannedstmt->paramExecTypes,
368 : paramid);
369 :
370 : /* Write paramid. */
371 12 : memcpy(start_address, ¶mid, sizeof(int));
372 12 : start_address += sizeof(int);
373 :
374 : /* Write datum/isnull */
375 12 : if (OidIsValid(typeOid))
376 12 : get_typlenbyval(typeOid, &typLen, &typByVal);
377 : else
378 : {
379 : /* If no type OID, assume by-value, like copyParamList does. */
380 0 : typLen = sizeof(Datum);
381 0 : typByVal = true;
382 : }
383 12 : datumSerialize(prm->value, prm->isnull, typByVal, typLen,
384 : &start_address);
385 : }
386 :
387 8 : return handle;
388 : }
389 :
390 : /*
391 : * Restore specified PARAM_EXEC parameters.
392 : */
393 : static void
394 24 : RestoreParamExecParams(char *start_address, EState *estate)
395 : {
396 : int nparams;
397 : int i;
398 : int paramid;
399 :
400 24 : memcpy(&nparams, start_address, sizeof(int));
401 24 : start_address += sizeof(int);
402 :
403 56 : for (i = 0; i < nparams; i++)
404 : {
405 : ParamExecData *prm;
406 :
407 : /* Read paramid */
408 32 : memcpy(¶mid, start_address, sizeof(int));
409 32 : start_address += sizeof(int);
410 32 : prm = &(estate->es_param_exec_vals[paramid]);
411 :
412 : /* Read datum/isnull. */
413 32 : prm->value = datumRestore(&start_address, &prm->isnull);
414 32 : prm->execPlan = NULL;
415 : }
416 24 : }
417 :
418 : /*
419 : * Initialize the dynamic shared memory segment that will be used to control
420 : * parallel execution.
421 : */
422 : static bool
423 1662 : ExecParallelInitializeDSM(PlanState *planstate,
424 : ExecParallelInitializeDSMContext *d)
425 : {
426 1662 : if (planstate == NULL)
427 0 : return false;
428 :
429 : /* If instrumentation is enabled, initialize slot for this node. */
430 1662 : if (d->instrumentation != NULL)
431 1272 : d->instrumentation->plan_node_id[d->nnodes] =
432 636 : planstate->plan->plan_node_id;
433 :
434 : /* Count this node. */
435 1662 : d->nnodes++;
436 :
437 : /*
438 : * Call initializers for DSM-using plan nodes.
439 : *
440 : * Most plan nodes won't do anything here, but plan nodes that allocated
441 : * DSM may need to initialize shared state in the DSM before parallel
442 : * workers are launched. They can allocate the space they previously
443 : * estimated using shm_toc_allocate, and add the keys they previously
444 : * estimated using shm_toc_insert, in each case targeting pcxt->toc.
445 : */
446 1662 : switch (nodeTag(planstate))
447 : {
448 : case T_SeqScanState:
449 640 : if (planstate->plan->parallel_aware)
450 492 : ExecSeqScanInitializeDSM((SeqScanState *) planstate,
451 : d->pcxt);
452 640 : break;
453 : case T_IndexScanState:
454 192 : if (planstate->plan->parallel_aware)
455 8 : ExecIndexScanInitializeDSM((IndexScanState *) planstate,
456 : d->pcxt);
457 192 : break;
458 : case T_IndexOnlyScanState:
459 28 : if (planstate->plan->parallel_aware)
460 24 : ExecIndexOnlyScanInitializeDSM((IndexOnlyScanState *) planstate,
461 : d->pcxt);
462 28 : break;
463 : case T_ForeignScanState:
464 0 : if (planstate->plan->parallel_aware)
465 0 : ExecForeignScanInitializeDSM((ForeignScanState *) planstate,
466 : d->pcxt);
467 0 : break;
468 : case T_AppendState:
469 100 : if (planstate->plan->parallel_aware)
470 68 : ExecAppendInitializeDSM((AppendState *) planstate,
471 : d->pcxt);
472 100 : break;
473 : case T_CustomScanState:
474 0 : if (planstate->plan->parallel_aware)
475 0 : ExecCustomScanInitializeDSM((CustomScanState *) planstate,
476 : d->pcxt);
477 0 : break;
478 : case T_BitmapHeapScanState:
479 10 : if (planstate->plan->parallel_aware)
480 8 : ExecBitmapHeapInitializeDSM((BitmapHeapScanState *) planstate,
481 : d->pcxt);
482 10 : break;
483 : case T_HashJoinState:
484 112 : if (planstate->plan->parallel_aware)
485 64 : ExecHashJoinInitializeDSM((HashJoinState *) planstate,
486 : d->pcxt);
487 112 : break;
488 : case T_HashState:
489 : /* even when not parallel-aware, for EXPLAIN ANALYZE */
490 112 : ExecHashInitializeDSM((HashState *) planstate, d->pcxt);
491 112 : break;
492 : case T_SortState:
493 : /* even when not parallel-aware, for EXPLAIN ANALYZE */
494 86 : ExecSortInitializeDSM((SortState *) planstate, d->pcxt);
495 86 : break;
496 :
497 : default:
498 382 : break;
499 : }
500 :
501 1662 : return planstate_tree_walker(planstate, ExecParallelInitializeDSM, d);
502 : }
503 :
504 : /*
505 : * It sets up the response queues for backend workers to return tuples
506 : * to the main backend and start the workers.
507 : */
508 : static shm_mq_handle **
509 530 : ExecParallelSetupTupleQueues(ParallelContext *pcxt, bool reinitialize)
510 : {
511 : shm_mq_handle **responseq;
512 : char *tqueuespace;
513 : int i;
514 :
515 : /* Skip this if no workers. */
516 530 : if (pcxt->nworkers == 0)
517 0 : return NULL;
518 :
519 : /* Allocate memory for shared memory queue handles. */
520 530 : responseq = (shm_mq_handle **)
521 530 : palloc(pcxt->nworkers * sizeof(shm_mq_handle *));
522 :
523 : /*
524 : * If not reinitializing, allocate space from the DSM for the queues;
525 : * otherwise, find the already allocated space.
526 : */
527 530 : if (!reinitialize)
528 362 : tqueuespace =
529 362 : shm_toc_allocate(pcxt->toc,
530 : mul_size(PARALLEL_TUPLE_QUEUE_SIZE,
531 362 : pcxt->nworkers));
532 : else
533 168 : tqueuespace = shm_toc_lookup(pcxt->toc, PARALLEL_KEY_TUPLE_QUEUE, false);
534 :
535 : /* Create the queues, and become the receiver for each. */
536 2016 : for (i = 0; i < pcxt->nworkers; ++i)
537 : {
538 : shm_mq *mq;
539 :
540 1486 : mq = shm_mq_create(tqueuespace +
541 1486 : ((Size) i) * PARALLEL_TUPLE_QUEUE_SIZE,
542 : (Size) PARALLEL_TUPLE_QUEUE_SIZE);
543 :
544 1486 : shm_mq_set_receiver(mq, MyProc);
545 1486 : responseq[i] = shm_mq_attach(mq, pcxt->seg, NULL);
546 : }
547 :
548 : /* Add array of queues to shm_toc, so others can find it. */
549 530 : if (!reinitialize)
550 362 : shm_toc_insert(pcxt->toc, PARALLEL_KEY_TUPLE_QUEUE, tqueuespace);
551 :
552 : /* Return array of handles. */
553 530 : return responseq;
554 : }
555 :
556 : /*
557 : * Sets up the required infrastructure for backend workers to perform
558 : * execution and return results to the main backend.
559 : */
560 : ParallelExecutorInfo *
561 362 : ExecInitParallelPlan(PlanState *planstate, EState *estate,
562 : Bitmapset *sendParams, int nworkers,
563 : int64 tuples_needed)
564 : {
565 : ParallelExecutorInfo *pei;
566 : ParallelContext *pcxt;
567 : ExecParallelEstimateContext e;
568 : ExecParallelInitializeDSMContext d;
569 : FixedParallelExecutorState *fpes;
570 : char *pstmt_data;
571 : char *pstmt_space;
572 : char *paramlistinfo_space;
573 : BufferUsage *bufusage_space;
574 362 : SharedExecutorInstrumentation *instrumentation = NULL;
575 362 : SharedJitInstrumentation *jit_instrumentation = NULL;
576 : int pstmt_len;
577 : int paramlistinfo_len;
578 362 : int instrumentation_len = 0;
579 362 : int jit_instrumentation_len = 0;
580 362 : int instrument_offset = 0;
581 362 : Size dsa_minsize = dsa_minimum_size();
582 : char *query_string;
583 : int query_len;
584 :
585 : /*
586 : * Force any initplan outputs that we're going to pass to workers to be
587 : * evaluated, if they weren't already.
588 : *
589 : * For simplicity, we use the EState's per-output-tuple ExprContext here.
590 : * That risks intra-query memory leakage, since we might pass through here
591 : * many times before that ExprContext gets reset; but ExecSetParamPlan
592 : * doesn't normally leak any memory in the context (see its comments), so
593 : * it doesn't seem worth complicating this function's API to pass it a
594 : * shorter-lived ExprContext. This might need to change someday.
595 : */
596 362 : ExecSetParamPlanMulti(sendParams, GetPerTupleExprContext(estate));
597 :
598 : /* Allocate object for return value. */
599 362 : pei = palloc0(sizeof(ParallelExecutorInfo));
600 362 : pei->finished = false;
601 362 : pei->planstate = planstate;
602 :
603 : /* Fix up and serialize plan to be sent to workers. */
604 362 : pstmt_data = ExecSerializePlan(planstate->plan, estate);
605 :
606 : /* Create a parallel context. */
607 362 : pcxt = CreateParallelContext("postgres", "ParallelQueryMain", nworkers);
608 362 : pei->pcxt = pcxt;
609 :
610 : /*
611 : * Before telling the parallel context to create a dynamic shared memory
612 : * segment, we need to figure out how big it should be. Estimate space
613 : * for the various things we need to store.
614 : */
615 :
616 : /* Estimate space for fixed-size state. */
617 362 : shm_toc_estimate_chunk(&pcxt->estimator,
618 : sizeof(FixedParallelExecutorState));
619 362 : shm_toc_estimate_keys(&pcxt->estimator, 1);
620 :
621 : /* Estimate space for query text. */
622 362 : query_len = strlen(estate->es_sourceText);
623 362 : shm_toc_estimate_chunk(&pcxt->estimator, query_len + 1);
624 362 : shm_toc_estimate_keys(&pcxt->estimator, 1);
625 :
626 : /* Estimate space for serialized PlannedStmt. */
627 362 : pstmt_len = strlen(pstmt_data) + 1;
628 362 : shm_toc_estimate_chunk(&pcxt->estimator, pstmt_len);
629 362 : shm_toc_estimate_keys(&pcxt->estimator, 1);
630 :
631 : /* Estimate space for serialized ParamListInfo. */
632 362 : paramlistinfo_len = EstimateParamListSpace(estate->es_param_list_info);
633 362 : shm_toc_estimate_chunk(&pcxt->estimator, paramlistinfo_len);
634 362 : shm_toc_estimate_keys(&pcxt->estimator, 1);
635 :
636 : /*
637 : * Estimate space for BufferUsage.
638 : *
639 : * If EXPLAIN is not in use and there are no extensions loaded that care,
640 : * we could skip this. But we have no way of knowing whether anyone's
641 : * looking at pgBufferUsage, so do it unconditionally.
642 : */
643 362 : shm_toc_estimate_chunk(&pcxt->estimator,
644 : mul_size(sizeof(BufferUsage), pcxt->nworkers));
645 362 : shm_toc_estimate_keys(&pcxt->estimator, 1);
646 :
647 : /* Estimate space for tuple queues. */
648 362 : shm_toc_estimate_chunk(&pcxt->estimator,
649 : mul_size(PARALLEL_TUPLE_QUEUE_SIZE, pcxt->nworkers));
650 362 : shm_toc_estimate_keys(&pcxt->estimator, 1);
651 :
652 : /*
653 : * Give parallel-aware nodes a chance to add to the estimates, and get a
654 : * count of how many PlanState nodes there are.
655 : */
656 362 : e.pcxt = pcxt;
657 362 : e.nnodes = 0;
658 362 : ExecParallelEstimate(planstate, &e);
659 :
660 : /* Estimate space for instrumentation, if required. */
661 362 : if (estate->es_instrument)
662 : {
663 108 : instrumentation_len =
664 : offsetof(SharedExecutorInstrumentation, plan_node_id) +
665 108 : sizeof(int) * e.nnodes;
666 108 : instrumentation_len = MAXALIGN(instrumentation_len);
667 108 : instrument_offset = instrumentation_len;
668 108 : instrumentation_len +=
669 216 : mul_size(sizeof(Instrumentation),
670 108 : mul_size(e.nnodes, nworkers));
671 108 : shm_toc_estimate_chunk(&pcxt->estimator, instrumentation_len);
672 108 : shm_toc_estimate_keys(&pcxt->estimator, 1);
673 :
674 : /* Estimate space for JIT instrumentation, if required. */
675 108 : if (estate->es_jit_flags != PGJIT_NONE)
676 : {
677 16 : jit_instrumentation_len =
678 16 : offsetof(SharedJitInstrumentation, jit_instr) +
679 : sizeof(JitInstrumentation) * nworkers;
680 16 : shm_toc_estimate_chunk(&pcxt->estimator, jit_instrumentation_len);
681 16 : shm_toc_estimate_keys(&pcxt->estimator, 1);
682 : }
683 : }
684 :
685 : /* Estimate space for DSA area. */
686 362 : shm_toc_estimate_chunk(&pcxt->estimator, dsa_minsize);
687 362 : shm_toc_estimate_keys(&pcxt->estimator, 1);
688 :
689 : /* Everyone's had a chance to ask for space, so now create the DSM. */
690 362 : InitializeParallelDSM(pcxt);
691 :
692 : /*
693 : * OK, now we have a dynamic shared memory segment, and it should be big
694 : * enough to store all of the data we estimated we would want to put into
695 : * it, plus whatever general stuff (not specifically executor-related) the
696 : * ParallelContext itself needs to store there. None of the space we
697 : * asked for has been allocated or initialized yet, though, so do that.
698 : */
699 :
700 : /* Store fixed-size state. */
701 362 : fpes = shm_toc_allocate(pcxt->toc, sizeof(FixedParallelExecutorState));
702 362 : fpes->tuples_needed = tuples_needed;
703 362 : fpes->param_exec = InvalidDsaPointer;
704 362 : fpes->eflags = estate->es_top_eflags;
705 362 : fpes->jit_flags = estate->es_jit_flags;
706 362 : shm_toc_insert(pcxt->toc, PARALLEL_KEY_EXECUTOR_FIXED, fpes);
707 :
708 : /* Store query string */
709 362 : query_string = shm_toc_allocate(pcxt->toc, query_len + 1);
710 362 : memcpy(query_string, estate->es_sourceText, query_len + 1);
711 362 : shm_toc_insert(pcxt->toc, PARALLEL_KEY_QUERY_TEXT, query_string);
712 :
713 : /* Store serialized PlannedStmt. */
714 362 : pstmt_space = shm_toc_allocate(pcxt->toc, pstmt_len);
715 362 : memcpy(pstmt_space, pstmt_data, pstmt_len);
716 362 : shm_toc_insert(pcxt->toc, PARALLEL_KEY_PLANNEDSTMT, pstmt_space);
717 :
718 : /* Store serialized ParamListInfo. */
719 362 : paramlistinfo_space = shm_toc_allocate(pcxt->toc, paramlistinfo_len);
720 362 : shm_toc_insert(pcxt->toc, PARALLEL_KEY_PARAMLISTINFO, paramlistinfo_space);
721 362 : SerializeParamList(estate->es_param_list_info, ¶mlistinfo_space);
722 :
723 : /* Allocate space for each worker's BufferUsage; no need to initialize. */
724 362 : bufusage_space = shm_toc_allocate(pcxt->toc,
725 362 : mul_size(sizeof(BufferUsage), pcxt->nworkers));
726 362 : shm_toc_insert(pcxt->toc, PARALLEL_KEY_BUFFER_USAGE, bufusage_space);
727 362 : pei->buffer_usage = bufusage_space;
728 :
729 : /* Set up the tuple queues that the workers will write into. */
730 362 : pei->tqueue = ExecParallelSetupTupleQueues(pcxt, false);
731 :
732 : /* We don't need the TupleQueueReaders yet, though. */
733 362 : pei->reader = NULL;
734 :
735 : /*
736 : * If instrumentation options were supplied, allocate space for the data.
737 : * It only gets partially initialized here; the rest happens during
738 : * ExecParallelInitializeDSM.
739 : */
740 362 : if (estate->es_instrument)
741 : {
742 : Instrumentation *instrument;
743 : int i;
744 :
745 108 : instrumentation = shm_toc_allocate(pcxt->toc, instrumentation_len);
746 108 : instrumentation->instrument_options = estate->es_instrument;
747 108 : instrumentation->instrument_offset = instrument_offset;
748 108 : instrumentation->num_workers = nworkers;
749 108 : instrumentation->num_plan_nodes = e.nnodes;
750 108 : instrument = GetInstrumentationArray(instrumentation);
751 1112 : for (i = 0; i < nworkers * e.nnodes; ++i)
752 1004 : InstrInit(&instrument[i], estate->es_instrument);
753 108 : shm_toc_insert(pcxt->toc, PARALLEL_KEY_INSTRUMENTATION,
754 : instrumentation);
755 108 : pei->instrumentation = instrumentation;
756 :
757 108 : if (estate->es_jit_flags != PGJIT_NONE)
758 : {
759 16 : jit_instrumentation = shm_toc_allocate(pcxt->toc,
760 : jit_instrumentation_len);
761 16 : jit_instrumentation->num_workers = nworkers;
762 16 : memset(jit_instrumentation->jit_instr, 0,
763 : sizeof(JitInstrumentation) * nworkers);
764 16 : shm_toc_insert(pcxt->toc, PARALLEL_KEY_JIT_INSTRUMENTATION,
765 : jit_instrumentation);
766 16 : pei->jit_instrumentation = jit_instrumentation;
767 : }
768 : }
769 :
770 : /*
771 : * Create a DSA area that can be used by the leader and all workers.
772 : * (However, if we failed to create a DSM and are using private memory
773 : * instead, then skip this.)
774 : */
775 362 : if (pcxt->seg != NULL)
776 : {
777 : char *area_space;
778 :
779 362 : area_space = shm_toc_allocate(pcxt->toc, dsa_minsize);
780 362 : shm_toc_insert(pcxt->toc, PARALLEL_KEY_DSA, area_space);
781 362 : pei->area = dsa_create_in_place(area_space, dsa_minsize,
782 : LWTRANCHE_PARALLEL_QUERY_DSA,
783 : pcxt->seg);
784 :
785 : /*
786 : * Serialize parameters, if any, using DSA storage. We don't dare use
787 : * the main parallel query DSM for this because we might relaunch
788 : * workers after the values have changed (and thus the amount of
789 : * storage required has changed).
790 : */
791 362 : if (!bms_is_empty(sendParams))
792 : {
793 8 : pei->param_exec = SerializeParamExecParams(estate, sendParams,
794 : pei->area);
795 8 : fpes->param_exec = pei->param_exec;
796 : }
797 : }
798 :
799 : /*
800 : * Give parallel-aware nodes a chance to initialize their shared data.
801 : * This also initializes the elements of instrumentation->ps_instrument,
802 : * if it exists.
803 : */
804 362 : d.pcxt = pcxt;
805 362 : d.instrumentation = instrumentation;
806 362 : d.nnodes = 0;
807 :
808 : /* Install our DSA area while initializing the plan. */
809 362 : estate->es_query_dsa = pei->area;
810 362 : ExecParallelInitializeDSM(planstate, &d);
811 362 : estate->es_query_dsa = NULL;
812 :
813 : /*
814 : * Make sure that the world hasn't shifted under our feet. This could
815 : * probably just be an Assert(), but let's be conservative for now.
816 : */
817 362 : if (e.nnodes != d.nnodes)
818 0 : elog(ERROR, "inconsistent count of PlanState nodes");
819 :
820 : /* OK, we're ready to rock and roll. */
821 362 : return pei;
822 : }
823 :
824 : /*
825 : * Set up tuple queue readers to read the results of a parallel subplan.
826 : *
827 : * This is separate from ExecInitParallelPlan() because we can launch the
828 : * worker processes and let them start doing something before we do this.
829 : */
830 : void
831 518 : ExecParallelCreateReaders(ParallelExecutorInfo *pei)
832 : {
833 518 : int nworkers = pei->pcxt->nworkers_launched;
834 : int i;
835 :
836 : Assert(pei->reader == NULL);
837 :
838 518 : if (nworkers > 0)
839 : {
840 518 : pei->reader = (TupleQueueReader **)
841 518 : palloc(nworkers * sizeof(TupleQueueReader *));
842 :
843 1954 : for (i = 0; i < nworkers; i++)
844 : {
845 1436 : shm_mq_set_handle(pei->tqueue[i],
846 1436 : pei->pcxt->worker[i].bgwhandle);
847 1436 : pei->reader[i] = CreateTupleQueueReader(pei->tqueue[i]);
848 : }
849 : }
850 518 : }
851 :
852 : /*
853 : * Re-initialize the parallel executor shared memory state before launching
854 : * a fresh batch of workers.
855 : */
856 : void
857 168 : ExecParallelReinitialize(PlanState *planstate,
858 : ParallelExecutorInfo *pei,
859 : Bitmapset *sendParams)
860 : {
861 168 : EState *estate = planstate->state;
862 : FixedParallelExecutorState *fpes;
863 :
864 : /* Old workers must already be shut down */
865 : Assert(pei->finished);
866 :
867 : /*
868 : * Force any initplan outputs that we're going to pass to workers to be
869 : * evaluated, if they weren't already (see comments in
870 : * ExecInitParallelPlan).
871 : */
872 168 : ExecSetParamPlanMulti(sendParams, GetPerTupleExprContext(estate));
873 :
874 168 : ReinitializeParallelDSM(pei->pcxt);
875 168 : pei->tqueue = ExecParallelSetupTupleQueues(pei->pcxt, true);
876 168 : pei->reader = NULL;
877 168 : pei->finished = false;
878 :
879 168 : fpes = shm_toc_lookup(pei->pcxt->toc, PARALLEL_KEY_EXECUTOR_FIXED, false);
880 :
881 : /* Free any serialized parameters from the last round. */
882 168 : if (DsaPointerIsValid(fpes->param_exec))
883 : {
884 0 : dsa_free(pei->area, fpes->param_exec);
885 0 : fpes->param_exec = InvalidDsaPointer;
886 : }
887 :
888 : /* Serialize current parameter values if required. */
889 168 : if (!bms_is_empty(sendParams))
890 : {
891 0 : pei->param_exec = SerializeParamExecParams(estate, sendParams,
892 : pei->area);
893 0 : fpes->param_exec = pei->param_exec;
894 : }
895 :
896 : /* Traverse plan tree and let each child node reset associated state. */
897 168 : estate->es_query_dsa = pei->area;
898 168 : ExecParallelReInitializeDSM(planstate, pei->pcxt);
899 168 : estate->es_query_dsa = NULL;
900 168 : }
901 :
902 : /*
903 : * Traverse plan tree to reinitialize per-node dynamic shared memory state
904 : */
905 : static bool
906 436 : ExecParallelReInitializeDSM(PlanState *planstate,
907 : ParallelContext *pcxt)
908 : {
909 436 : if (planstate == NULL)
910 0 : return false;
911 :
912 : /*
913 : * Call reinitializers for DSM-using plan nodes.
914 : */
915 436 : switch (nodeTag(planstate))
916 : {
917 : case T_SeqScanState:
918 180 : if (planstate->plan->parallel_aware)
919 148 : ExecSeqScanReInitializeDSM((SeqScanState *) planstate,
920 : pcxt);
921 180 : break;
922 : case T_IndexScanState:
923 8 : if (planstate->plan->parallel_aware)
924 8 : ExecIndexScanReInitializeDSM((IndexScanState *) planstate,
925 : pcxt);
926 8 : break;
927 : case T_IndexOnlyScanState:
928 8 : if (planstate->plan->parallel_aware)
929 8 : ExecIndexOnlyScanReInitializeDSM((IndexOnlyScanState *) planstate,
930 : pcxt);
931 8 : break;
932 : case T_ForeignScanState:
933 0 : if (planstate->plan->parallel_aware)
934 0 : ExecForeignScanReInitializeDSM((ForeignScanState *) planstate,
935 : pcxt);
936 0 : break;
937 : case T_AppendState:
938 0 : if (planstate->plan->parallel_aware)
939 0 : ExecAppendReInitializeDSM((AppendState *) planstate, pcxt);
940 0 : break;
941 : case T_CustomScanState:
942 0 : if (planstate->plan->parallel_aware)
943 0 : ExecCustomScanReInitializeDSM((CustomScanState *) planstate,
944 : pcxt);
945 0 : break;
946 : case T_BitmapHeapScanState:
947 36 : if (planstate->plan->parallel_aware)
948 36 : ExecBitmapHeapReInitializeDSM((BitmapHeapScanState *) planstate,
949 : pcxt);
950 36 : break;
951 : case T_HashJoinState:
952 64 : if (planstate->plan->parallel_aware)
953 32 : ExecHashJoinReInitializeDSM((HashJoinState *) planstate,
954 : pcxt);
955 64 : break;
956 : case T_HashState:
957 : case T_SortState:
958 : /* these nodes have DSM state, but no reinitialization is required */
959 80 : break;
960 :
961 : default:
962 60 : break;
963 : }
964 :
965 436 : return planstate_tree_walker(planstate, ExecParallelReInitializeDSM, pcxt);
966 : }
967 :
968 : /*
969 : * Copy instrumentation information about this node and its descendants from
970 : * dynamic shared memory.
971 : */
972 : static bool
973 636 : ExecParallelRetrieveInstrumentation(PlanState *planstate,
974 : SharedExecutorInstrumentation *instrumentation)
975 : {
976 : Instrumentation *instrument;
977 : int i;
978 : int n;
979 : int ibytes;
980 636 : int plan_node_id = planstate->plan->plan_node_id;
981 : MemoryContext oldcontext;
982 :
983 : /* Find the instrumentation for this node. */
984 2908 : for (i = 0; i < instrumentation->num_plan_nodes; ++i)
985 2908 : if (instrumentation->plan_node_id[i] == plan_node_id)
986 636 : break;
987 636 : if (i >= instrumentation->num_plan_nodes)
988 0 : elog(ERROR, "plan node %d not found", plan_node_id);
989 :
990 : /* Accumulate the statistics from all workers. */
991 636 : instrument = GetInstrumentationArray(instrumentation);
992 636 : instrument += i * instrumentation->num_workers;
993 1640 : for (n = 0; n < instrumentation->num_workers; ++n)
994 1004 : InstrAggNode(planstate->instrument, &instrument[n]);
995 :
996 : /*
997 : * Also store the per-worker detail.
998 : *
999 : * Worker instrumentation should be allocated in the same context as the
1000 : * regular instrumentation information, which is the per-query context.
1001 : * Switch into per-query memory context.
1002 : */
1003 636 : oldcontext = MemoryContextSwitchTo(planstate->state->es_query_cxt);
1004 636 : ibytes = mul_size(instrumentation->num_workers, sizeof(Instrumentation));
1005 636 : planstate->worker_instrument =
1006 636 : palloc(ibytes + offsetof(WorkerInstrumentation, instrument));
1007 636 : MemoryContextSwitchTo(oldcontext);
1008 :
1009 636 : planstate->worker_instrument->num_workers = instrumentation->num_workers;
1010 636 : memcpy(&planstate->worker_instrument->instrument, instrument, ibytes);
1011 :
1012 : /* Perform any node-type-specific work that needs to be done. */
1013 636 : switch (nodeTag(planstate))
1014 : {
1015 : case T_SortState:
1016 4 : ExecSortRetrieveInstrumentation((SortState *) planstate);
1017 4 : break;
1018 : case T_HashState:
1019 56 : ExecHashRetrieveInstrumentation((HashState *) planstate);
1020 56 : break;
1021 : default:
1022 576 : break;
1023 : }
1024 :
1025 636 : return planstate_tree_walker(planstate, ExecParallelRetrieveInstrumentation,
1026 : instrumentation);
1027 : }
1028 :
1029 : /*
1030 : * Add up the workers' JIT instrumentation from dynamic shared memory.
1031 : */
1032 : static void
1033 16 : ExecParallelRetrieveJitInstrumentation(PlanState *planstate,
1034 : SharedJitInstrumentation *shared_jit)
1035 : {
1036 : JitInstrumentation *combined;
1037 : int ibytes;
1038 :
1039 : int n;
1040 :
1041 : /*
1042 : * Accumulate worker JIT instrumentation into the combined JIT
1043 : * instrumentation, allocating it if required.
1044 : */
1045 16 : if (!planstate->state->es_jit_worker_instr)
1046 32 : planstate->state->es_jit_worker_instr =
1047 16 : MemoryContextAllocZero(planstate->state->es_query_cxt, sizeof(JitInstrumentation));
1048 16 : combined = planstate->state->es_jit_worker_instr;
1049 :
1050 : /* Accumulate all the workers' instrumentations. */
1051 48 : for (n = 0; n < shared_jit->num_workers; ++n)
1052 32 : InstrJitAgg(combined, &shared_jit->jit_instr[n]);
1053 :
1054 : /*
1055 : * Store the per-worker detail.
1056 : *
1057 : * Similar to ExecParallelRetrieveInstrumentation(), allocate the
1058 : * instrumentation in per-query context.
1059 : */
1060 16 : ibytes = offsetof(SharedJitInstrumentation, jit_instr)
1061 16 : + mul_size(shared_jit->num_workers, sizeof(JitInstrumentation));
1062 16 : planstate->worker_jit_instrument =
1063 16 : MemoryContextAlloc(planstate->state->es_query_cxt, ibytes);
1064 :
1065 16 : memcpy(planstate->worker_jit_instrument, shared_jit, ibytes);
1066 16 : }
1067 :
1068 : /*
1069 : * Finish parallel execution. We wait for parallel workers to finish, and
1070 : * accumulate their buffer usage.
1071 : */
1072 : void
1073 964 : ExecParallelFinish(ParallelExecutorInfo *pei)
1074 : {
1075 964 : int nworkers = pei->pcxt->nworkers_launched;
1076 : int i;
1077 :
1078 : /* Make this be a no-op if called twice in a row. */
1079 964 : if (pei->finished)
1080 438 : return;
1081 :
1082 : /*
1083 : * Detach from tuple queues ASAP, so that any still-active workers will
1084 : * notice that no further results are wanted.
1085 : */
1086 526 : if (pei->tqueue != NULL)
1087 : {
1088 1958 : for (i = 0; i < nworkers; i++)
1089 1432 : shm_mq_detach(pei->tqueue[i]);
1090 526 : pfree(pei->tqueue);
1091 526 : pei->tqueue = NULL;
1092 : }
1093 :
1094 : /*
1095 : * While we're waiting for the workers to finish, let's get rid of the
1096 : * tuple queue readers. (Any other local cleanup could be done here too.)
1097 : */
1098 526 : if (pei->reader != NULL)
1099 : {
1100 1946 : for (i = 0; i < nworkers; i++)
1101 1432 : DestroyTupleQueueReader(pei->reader[i]);
1102 514 : pfree(pei->reader);
1103 514 : pei->reader = NULL;
1104 : }
1105 :
1106 : /* Now wait for the workers to finish. */
1107 526 : WaitForParallelWorkersToFinish(pei->pcxt);
1108 :
1109 : /*
1110 : * Next, accumulate buffer usage. (This must wait for the workers to
1111 : * finish, or we might get incomplete data.)
1112 : */
1113 1958 : for (i = 0; i < nworkers; i++)
1114 1432 : InstrAccumParallelQuery(&pei->buffer_usage[i]);
1115 :
1116 526 : pei->finished = true;
1117 : }
1118 :
1119 : /*
1120 : * Accumulate instrumentation, and then clean up whatever ParallelExecutorInfo
1121 : * resources still exist after ExecParallelFinish. We separate these
1122 : * routines because someone might want to examine the contents of the DSM
1123 : * after ExecParallelFinish and before calling this routine.
1124 : */
1125 : void
1126 358 : ExecParallelCleanup(ParallelExecutorInfo *pei)
1127 : {
1128 : /* Accumulate instrumentation, if any. */
1129 358 : if (pei->instrumentation)
1130 108 : ExecParallelRetrieveInstrumentation(pei->planstate,
1131 : pei->instrumentation);
1132 :
1133 : /* Accumulate JIT instrumentation, if any. */
1134 358 : if (pei->jit_instrumentation)
1135 16 : ExecParallelRetrieveJitInstrumentation(pei->planstate,
1136 16 : pei->jit_instrumentation);
1137 :
1138 : /* Free any serialized parameters. */
1139 358 : if (DsaPointerIsValid(pei->param_exec))
1140 : {
1141 8 : dsa_free(pei->area, pei->param_exec);
1142 8 : pei->param_exec = InvalidDsaPointer;
1143 : }
1144 358 : if (pei->area != NULL)
1145 : {
1146 358 : dsa_detach(pei->area);
1147 358 : pei->area = NULL;
1148 : }
1149 358 : if (pei->pcxt != NULL)
1150 : {
1151 358 : DestroyParallelContext(pei->pcxt);
1152 358 : pei->pcxt = NULL;
1153 : }
1154 358 : pfree(pei);
1155 358 : }
1156 :
1157 : /*
1158 : * Create a DestReceiver to write tuples we produce to the shm_mq designated
1159 : * for that purpose.
1160 : */
1161 : static DestReceiver *
1162 1436 : ExecParallelGetReceiver(dsm_segment *seg, shm_toc *toc)
1163 : {
1164 : char *mqspace;
1165 : shm_mq *mq;
1166 :
1167 1436 : mqspace = shm_toc_lookup(toc, PARALLEL_KEY_TUPLE_QUEUE, false);
1168 1436 : mqspace += ParallelWorkerNumber * PARALLEL_TUPLE_QUEUE_SIZE;
1169 1436 : mq = (shm_mq *) mqspace;
1170 1436 : shm_mq_set_sender(mq, MyProc);
1171 1436 : return CreateTupleQueueDestReceiver(shm_mq_attach(mq, seg, NULL));
1172 : }
1173 :
1174 : /*
1175 : * Create a QueryDesc for the PlannedStmt we are to execute, and return it.
1176 : */
1177 : static QueryDesc *
1178 1436 : ExecParallelGetQueryDesc(shm_toc *toc, DestReceiver *receiver,
1179 : int instrument_options)
1180 : {
1181 : char *pstmtspace;
1182 : char *paramspace;
1183 : PlannedStmt *pstmt;
1184 : ParamListInfo paramLI;
1185 : char *queryString;
1186 :
1187 : /* Get the query string from shared memory */
1188 1436 : queryString = shm_toc_lookup(toc, PARALLEL_KEY_QUERY_TEXT, false);
1189 :
1190 : /* Reconstruct leader-supplied PlannedStmt. */
1191 1436 : pstmtspace = shm_toc_lookup(toc, PARALLEL_KEY_PLANNEDSTMT, false);
1192 1436 : pstmt = (PlannedStmt *) stringToNode(pstmtspace);
1193 :
1194 : /* Reconstruct ParamListInfo. */
1195 1436 : paramspace = shm_toc_lookup(toc, PARALLEL_KEY_PARAMLISTINFO, false);
1196 1436 : paramLI = RestoreParamList(¶mspace);
1197 :
1198 : /* Create a QueryDesc for the query. */
1199 1436 : return CreateQueryDesc(pstmt,
1200 : queryString,
1201 : GetActiveSnapshot(), InvalidSnapshot,
1202 : receiver, paramLI, NULL, instrument_options);
1203 : }
1204 :
1205 : /*
1206 : * Copy instrumentation information from this node and its descendants into
1207 : * dynamic shared memory, so that the parallel leader can retrieve it.
1208 : */
1209 : static bool
1210 1468 : ExecParallelReportInstrumentation(PlanState *planstate,
1211 : SharedExecutorInstrumentation *instrumentation)
1212 : {
1213 : int i;
1214 1468 : int plan_node_id = planstate->plan->plan_node_id;
1215 : Instrumentation *instrument;
1216 :
1217 1468 : InstrEndLoop(planstate->instrument);
1218 :
1219 : /*
1220 : * If we shuffled the plan_node_id values in ps_instrument into sorted
1221 : * order, we could use binary search here. This might matter someday if
1222 : * we're pushing down sufficiently large plan trees. For now, do it the
1223 : * slow, dumb way.
1224 : */
1225 4796 : for (i = 0; i < instrumentation->num_plan_nodes; ++i)
1226 4796 : if (instrumentation->plan_node_id[i] == plan_node_id)
1227 1468 : break;
1228 1468 : if (i >= instrumentation->num_plan_nodes)
1229 0 : elog(ERROR, "plan node %d not found", plan_node_id);
1230 :
1231 : /*
1232 : * Add our statistics to the per-node, per-worker totals. It's possible
1233 : * that this could happen more than once if we relaunched workers.
1234 : */
1235 1468 : instrument = GetInstrumentationArray(instrumentation);
1236 1468 : instrument += i * instrumentation->num_workers;
1237 : Assert(IsParallelWorker());
1238 : Assert(ParallelWorkerNumber < instrumentation->num_workers);
1239 1468 : InstrAggNode(&instrument[ParallelWorkerNumber], planstate->instrument);
1240 :
1241 1468 : return planstate_tree_walker(planstate, ExecParallelReportInstrumentation,
1242 : instrumentation);
1243 : }
1244 :
1245 : /*
1246 : * Initialize the PlanState and its descendants with the information
1247 : * retrieved from shared memory. This has to be done once the PlanState
1248 : * is allocated and initialized by executor; that is, after ExecutorStart().
1249 : */
1250 : static bool
1251 4718 : ExecParallelInitializeWorker(PlanState *planstate, ParallelWorkerContext *pwcxt)
1252 : {
1253 4718 : if (planstate == NULL)
1254 0 : return false;
1255 :
1256 4718 : switch (nodeTag(planstate))
1257 : {
1258 : case T_SeqScanState:
1259 1878 : if (planstate->plan->parallel_aware)
1260 1482 : ExecSeqScanInitializeWorker((SeqScanState *) planstate, pwcxt);
1261 1878 : break;
1262 : case T_IndexScanState:
1263 248 : if (planstate->plan->parallel_aware)
1264 64 : ExecIndexScanInitializeWorker((IndexScanState *) planstate,
1265 : pwcxt);
1266 248 : break;
1267 : case T_IndexOnlyScanState:
1268 144 : if (planstate->plan->parallel_aware)
1269 128 : ExecIndexOnlyScanInitializeWorker((IndexOnlyScanState *) planstate,
1270 : pwcxt);
1271 144 : break;
1272 : case T_ForeignScanState:
1273 0 : if (planstate->plan->parallel_aware)
1274 0 : ExecForeignScanInitializeWorker((ForeignScanState *) planstate,
1275 : pwcxt);
1276 0 : break;
1277 : case T_AppendState:
1278 188 : if (planstate->plan->parallel_aware)
1279 148 : ExecAppendInitializeWorker((AppendState *) planstate, pwcxt);
1280 188 : break;
1281 : case T_CustomScanState:
1282 0 : if (planstate->plan->parallel_aware)
1283 0 : ExecCustomScanInitializeWorker((CustomScanState *) planstate,
1284 : pwcxt);
1285 0 : break;
1286 : case T_BitmapHeapScanState:
1287 178 : if (planstate->plan->parallel_aware)
1288 176 : ExecBitmapHeapInitializeWorker((BitmapHeapScanState *) planstate,
1289 : pwcxt);
1290 178 : break;
1291 : case T_HashJoinState:
1292 340 : if (planstate->plan->parallel_aware)
1293 180 : ExecHashJoinInitializeWorker((HashJoinState *) planstate,
1294 : pwcxt);
1295 340 : break;
1296 : case T_HashState:
1297 : /* even when not parallel-aware, for EXPLAIN ANALYZE */
1298 340 : ExecHashInitializeWorker((HashState *) planstate, pwcxt);
1299 340 : break;
1300 : case T_SortState:
1301 : /* even when not parallel-aware, for EXPLAIN ANALYZE */
1302 262 : ExecSortInitializeWorker((SortState *) planstate, pwcxt);
1303 262 : break;
1304 :
1305 : default:
1306 1140 : break;
1307 : }
1308 :
1309 4718 : return planstate_tree_walker(planstate, ExecParallelInitializeWorker,
1310 : pwcxt);
1311 : }
1312 :
1313 : /*
1314 : * Main entrypoint for parallel query worker processes.
1315 : *
1316 : * We reach this function from ParallelWorkerMain, so the setup necessary to
1317 : * create a sensible parallel environment has already been done;
1318 : * ParallelWorkerMain worries about stuff like the transaction state, combo
1319 : * CID mappings, and GUC values, so we don't need to deal with any of that
1320 : * here.
1321 : *
1322 : * Our job is to deal with concerns specific to the executor. The parallel
1323 : * group leader will have stored a serialized PlannedStmt, and it's our job
1324 : * to execute that plan and write the resulting tuples to the appropriate
1325 : * tuple queue. Various bits of supporting information that we need in order
1326 : * to do this are also stored in the dsm_segment and can be accessed through
1327 : * the shm_toc.
1328 : */
1329 : void
1330 1436 : ParallelQueryMain(dsm_segment *seg, shm_toc *toc)
1331 : {
1332 : FixedParallelExecutorState *fpes;
1333 : BufferUsage *buffer_usage;
1334 : DestReceiver *receiver;
1335 : QueryDesc *queryDesc;
1336 : SharedExecutorInstrumentation *instrumentation;
1337 : SharedJitInstrumentation *jit_instrumentation;
1338 1436 : int instrument_options = 0;
1339 : void *area_space;
1340 : dsa_area *area;
1341 : ParallelWorkerContext pwcxt;
1342 :
1343 : /* Get fixed-size state. */
1344 1436 : fpes = shm_toc_lookup(toc, PARALLEL_KEY_EXECUTOR_FIXED, false);
1345 :
1346 : /* Set up DestReceiver, SharedExecutorInstrumentation, and QueryDesc. */
1347 1436 : receiver = ExecParallelGetReceiver(seg, toc);
1348 1436 : instrumentation = shm_toc_lookup(toc, PARALLEL_KEY_INSTRUMENTATION, true);
1349 1436 : if (instrumentation != NULL)
1350 452 : instrument_options = instrumentation->instrument_options;
1351 1436 : jit_instrumentation = shm_toc_lookup(toc, PARALLEL_KEY_JIT_INSTRUMENTATION,
1352 : true);
1353 1436 : queryDesc = ExecParallelGetQueryDesc(toc, receiver, instrument_options);
1354 :
1355 : /* Setting debug_query_string for individual workers */
1356 1436 : debug_query_string = queryDesc->sourceText;
1357 :
1358 : /* Report workers' query for monitoring purposes */
1359 1436 : pgstat_report_activity(STATE_RUNNING, debug_query_string);
1360 :
1361 : /* Attach to the dynamic shared memory area. */
1362 1436 : area_space = shm_toc_lookup(toc, PARALLEL_KEY_DSA, false);
1363 1436 : area = dsa_attach_in_place(area_space, seg);
1364 :
1365 : /* Start up the executor */
1366 1436 : queryDesc->plannedstmt->jitFlags = fpes->jit_flags;
1367 1436 : ExecutorStart(queryDesc, fpes->eflags);
1368 :
1369 : /* Special executor initialization steps for parallel workers */
1370 1436 : queryDesc->planstate->state->es_query_dsa = area;
1371 1436 : if (DsaPointerIsValid(fpes->param_exec))
1372 : {
1373 : char *paramexec_space;
1374 :
1375 24 : paramexec_space = dsa_get_address(area, fpes->param_exec);
1376 24 : RestoreParamExecParams(paramexec_space, queryDesc->estate);
1377 :
1378 : }
1379 1436 : pwcxt.toc = toc;
1380 1436 : pwcxt.seg = seg;
1381 1436 : ExecParallelInitializeWorker(queryDesc->planstate, &pwcxt);
1382 :
1383 : /* Pass down any tuple bound */
1384 1436 : ExecSetTupleBound(fpes->tuples_needed, queryDesc->planstate);
1385 :
1386 : /*
1387 : * Prepare to track buffer usage during query execution.
1388 : *
1389 : * We do this after starting up the executor to match what happens in the
1390 : * leader, which also doesn't count buffer accesses that occur during
1391 : * executor startup.
1392 : */
1393 1436 : InstrStartParallelQuery();
1394 :
1395 : /*
1396 : * Run the plan. If we specified a tuple bound, be careful not to demand
1397 : * more tuples than that.
1398 : */
1399 1436 : ExecutorRun(queryDesc,
1400 : ForwardScanDirection,
1401 1436 : fpes->tuples_needed < 0 ? (int64) 0 : fpes->tuples_needed,
1402 : true);
1403 :
1404 : /* Shut down the executor */
1405 1432 : ExecutorFinish(queryDesc);
1406 :
1407 : /* Report buffer usage during parallel execution. */
1408 1432 : buffer_usage = shm_toc_lookup(toc, PARALLEL_KEY_BUFFER_USAGE, false);
1409 1432 : InstrEndParallelQuery(&buffer_usage[ParallelWorkerNumber]);
1410 :
1411 : /* Report instrumentation data if any instrumentation options are set. */
1412 1432 : if (instrumentation != NULL)
1413 452 : ExecParallelReportInstrumentation(queryDesc->planstate,
1414 : instrumentation);
1415 :
1416 : /* Report JIT instrumentation data if any */
1417 1432 : if (queryDesc->estate->es_jit && jit_instrumentation != NULL)
1418 : {
1419 : Assert(ParallelWorkerNumber < jit_instrumentation->num_workers);
1420 192 : jit_instrumentation->jit_instr[ParallelWorkerNumber] =
1421 96 : queryDesc->estate->es_jit->instr;
1422 : }
1423 :
1424 : /* Must do this after capturing instrumentation. */
1425 1432 : ExecutorEnd(queryDesc);
1426 :
1427 : /* Cleanup. */
1428 1432 : dsa_detach(area);
1429 1432 : FreeQueryDesc(queryDesc);
1430 1432 : receiver->rDestroy(receiver);
1431 1432 : }
|
