| 1 |
/*------------------------------------------------------------------------- |
| 2 |
* |
| 3 |
* nodeGatherMerge.c |
| 4 |
* Scan a plan in multiple workers, and do order-preserving merge. |
| 5 |
* |
| 6 |
* Portions Copyright (c) 1996-2025, PostgreSQL Global Development Group |
| 7 |
* Portions Copyright (c) 1994, Regents of the University of California |
| 8 |
* |
| 9 |
* IDENTIFICATION |
| 10 |
* src/backend/executor/nodeGatherMerge.c |
| 11 |
* |
| 12 |
*------------------------------------------------------------------------- |
| 13 |
*/ |
| 14 |
|
| 15 |
#include "postgres.h" |
| 16 |
|
| 17 |
#include "executor/executor.h" |
| 18 |
#include "executor/execParallel.h" |
| 19 |
#include "executor/nodeGatherMerge.h" |
| 20 |
#include "executor/tqueue.h" |
| 21 |
#include "lib/binaryheap.h" |
| 22 |
#include "miscadmin.h" |
| 23 |
#include "optimizer/optimizer.h" |
| 24 |
|
| 25 |
/* |
| 26 |
* When we read tuples from workers, it's a good idea to read several at once |
| 27 |
* for efficiency when possible: this minimizes context-switching overhead. |
| 28 |
* But reading too many at a time wastes memory without improving performance. |
| 29 |
* We'll read up to MAX_TUPLE_STORE tuples (in addition to the first one). |
| 30 |
*/ |
| 31 |
#define MAX_TUPLE_STORE 10 |
| 32 |
|
| 33 |
/* |
| 34 |
* Pending-tuple array for each worker. This holds additional tuples that |
| 35 |
* we were able to fetch from the worker, but can't process yet. In addition, |
| 36 |
* this struct holds the "done" flag indicating the worker is known to have |
| 37 |
* no more tuples. (We do not use this struct for the leader; we don't keep |
| 38 |
* any pending tuples for the leader, and the need_to_scan_locally flag serves |
| 39 |
* as its "done" indicator.) |
| 40 |
*/ |
| 41 |
typedef struct GMReaderTupleBuffer |
| 42 |
{ |
| 43 |
MinimalTuple *tuple; /* array of length MAX_TUPLE_STORE */ |
| 44 |
int nTuples; /* number of tuples currently stored */ |
| 45 |
int readCounter; /* index of next tuple to extract */ |
| 46 |
bool done; /* true if reader is known exhausted */ |
| 47 |
} GMReaderTupleBuffer; |
| 48 |
|
| 49 |
static TupleTableSlot *ExecGatherMerge(PlanState *pstate); |
| 50 |
static int32 heap_compare_slots(Datum a, Datum b, void *arg); |
| 51 |
static TupleTableSlot *gather_merge_getnext(GatherMergeState *gm_state); |
| 52 |
static MinimalTuple gm_readnext_tuple(GatherMergeState *gm_state, int nreader, |
| 53 |
bool nowait, bool *done); |
| 54 |
static void ExecShutdownGatherMergeWorkers(GatherMergeState *node); |
| 55 |
static void gather_merge_setup(GatherMergeState *gm_state); |
| 56 |
static void gather_merge_init(GatherMergeState *gm_state); |
| 57 |
static void gather_merge_clear_tuples(GatherMergeState *gm_state); |
| 58 |
static bool gather_merge_readnext(GatherMergeState *gm_state, int reader, |
| 59 |
bool nowait); |
| 60 |
static void load_tuple_array(GatherMergeState *gm_state, int reader); |
| 61 |
|
| 62 |
/* ---------------------------------------------------------------- |
| 63 |
* ExecInitGather |
| 64 |
* ---------------------------------------------------------------- |
| 65 |
*/ |
| 66 |
GatherMergeState * |
| 67 |
ExecInitGatherMerge(GatherMerge *node, EState *estate, int eflags) |
| 68 |
{ |
| 69 |
GatherMergeState *gm_state; |
| 70 |
Plan *outerNode; |
| 71 |
TupleDesc tupDesc; |
| 72 |
|
| 73 |
/* Gather merge node doesn't have innerPlan node. */ |
| 74 |
Assert(innerPlan(node) == NULL); |
| 75 |
|
| 76 |
/* |
| 77 |
* create state structure |
| 78 |
*/ |
| 79 |
gm_state = makeNode(GatherMergeState); |
| 80 |
gm_state->ps.plan = (Plan *) node; |
| 81 |
gm_state->ps.state = estate; |
| 82 |
gm_state->ps.ExecProcNode = ExecGatherMerge; |
| 83 |
|
| 84 |
gm_state->initialized = false; |
| 85 |
gm_state->gm_initialized = false; |
| 86 |
gm_state->tuples_needed = -1; |
| 87 |
|
| 88 |
/* |
| 89 |
* Miscellaneous initialization |
| 90 |
* |
| 91 |
* create expression context for node |
| 92 |
*/ |
| 93 |
ExecAssignExprContext(estate, &gm_state->ps); |
| 94 |
|
| 95 |
/* |
| 96 |
* GatherMerge doesn't support checking a qual (it's always more efficient |
| 97 |
* to do it in the child node). |
| 98 |
*/ |
| 99 |
Assert(!node->plan.qual); |
| 100 |
|
| 101 |
/* |
| 102 |
* now initialize outer plan |
| 103 |
*/ |
| 104 |
outerNode = outerPlan(node); |
| 105 |
outerPlanState(gm_state) = ExecInitNode(outerNode, estate, eflags); |
| 106 |
|
| 107 |
/* |
| 108 |
* Leader may access ExecProcNode result directly (if |
| 109 |
* need_to_scan_locally), or from workers via tuple queue. So we can't |
| 110 |
* trivially rely on the slot type being fixed for expressions evaluated |
| 111 |
* within this node. |
| 112 |
*/ |
| 113 |
gm_state->ps.outeropsset = true; |
| 114 |
gm_state->ps.outeropsfixed = false; |
| 115 |
|
| 116 |
/* |
| 117 |
* Store the tuple descriptor into gather merge state, so we can use it |
| 118 |
* while initializing the gather merge slots. |
| 119 |
*/ |
| 120 |
tupDesc = ExecGetResultType(outerPlanState(gm_state)); |
| 121 |
gm_state->tupDesc = tupDesc; |
| 122 |
|
| 123 |
/* |
| 124 |
* Initialize result type and projection. |
| 125 |
*/ |
| 126 |
ExecInitResultTypeTL(&gm_state->ps); |
| 127 |
ExecConditionalAssignProjectionInfo(&gm_state->ps, tupDesc, OUTER_VAR); |
| 128 |
|
| 129 |
/* |
| 130 |
* Without projections result slot type is not trivially known, see |
| 131 |
* comment above. |
| 132 |
*/ |
| 133 |
if (gm_state->ps.ps_ProjInfo == NULL) |
| 134 |
{ |
| 135 |
gm_state->ps.resultopsset = true; |
| 136 |
gm_state->ps.resultopsfixed = false; |
| 137 |
} |
| 138 |
|
| 139 |
/* |
| 140 |
* initialize sort-key information |
| 141 |
*/ |
| 142 |
if (node->numCols) |
| 143 |
{ |
| 144 |
int i; |
| 145 |
|
| 146 |
gm_state->gm_nkeys = node->numCols; |
| 147 |
gm_state->gm_sortkeys = |
| 148 |
palloc0(sizeof(SortSupportData) * node->numCols); |
| 149 |
|
| 150 |
for (i = 0; i < node->numCols; i++) |
| 151 |
{ |
| 152 |
SortSupport sortKey = gm_state->gm_sortkeys + i; |
| 153 |
|
| 154 |
sortKey->ssup_cxt = CurrentMemoryContext; |
| 155 |
sortKey->ssup_collation = node->collations[i]; |
| 156 |
sortKey->ssup_nulls_first = node->nullsFirst[i]; |
| 157 |
sortKey->ssup_attno = node->sortColIdx[i]; |
| 158 |
|
| 159 |
/* |
| 160 |
* We don't perform abbreviated key conversion here, for the same |
| 161 |
* reasons that it isn't used in MergeAppend |
| 162 |
*/ |
| 163 |
sortKey->abbreviate = false; |
| 164 |
|
| 165 |
PrepareSortSupportFromOrderingOp(node->sortOperators[i], sortKey); |
| 166 |
} |
| 167 |
} |
| 168 |
|
| 169 |
/* Now allocate the workspace for gather merge */ |
| 170 |
gather_merge_setup(gm_state); |
| 171 |
|
| 172 |
return gm_state; |
| 173 |
} |
| 174 |
|
| 175 |
/* ---------------------------------------------------------------- |
| 176 |
* ExecGatherMerge(node) |
| 177 |
* |
| 178 |
* Scans the relation via multiple workers and returns |
| 179 |
* the next qualifying tuple. |
| 180 |
* ---------------------------------------------------------------- |
| 181 |
*/ |
| 182 |
static TupleTableSlot * |
| 183 |
ExecGatherMerge(PlanState *pstate) |
| 184 |
{ |
| 185 |
GatherMergeState *node = castNode(GatherMergeState, pstate); |
| 186 |
TupleTableSlot *slot; |
| 187 |
ExprContext *econtext; |
| 188 |
|
| 189 |
CHECK_FOR_INTERRUPTS(); |
| 190 |
|
| 191 |
/* |
| 192 |
* As with Gather, we don't launch workers until this node is actually |
| 193 |
* executed. |
| 194 |
*/ |
| 195 |
if (!node->initialized) |
| 196 |
{ |
| 197 |
EState *estate = node->ps.state; |
| 198 |
GatherMerge *gm = castNode(GatherMerge, node->ps.plan); |
| 199 |
|
| 200 |
/* |
| 201 |
* Sometimes we might have to run without parallelism; but if parallel |
| 202 |
* mode is active then we can try to fire up some workers. |
| 203 |
*/ |
| 204 |
if (gm->num_workers > 0 && estate->es_use_parallel_mode) |
| 205 |
{ |
| 206 |
ParallelContext *pcxt; |
| 207 |
|
| 208 |
/* Initialize, or re-initialize, shared state needed by workers. */ |
| 209 |
if (!node->pei) |
| 210 |
node->pei = ExecInitParallelPlan(outerPlanState(node), |
| 211 |
estate, |
| 212 |
gm->initParam, |
| 213 |
gm->num_workers, |
| 214 |
node->tuples_needed); |
| 215 |
else |
| 216 |
ExecParallelReinitialize(outerPlanState(node), |
| 217 |
node->pei, |
| 218 |
gm->initParam); |
| 219 |
|
| 220 |
/* Try to launch workers. */ |
| 221 |
pcxt = node->pei->pcxt; |
| 222 |
LaunchParallelWorkers(pcxt); |
| 223 |
/* We save # workers launched for the benefit of EXPLAIN */ |
| 224 |
node->nworkers_launched = pcxt->nworkers_launched; |
| 225 |
|
| 226 |
/* |
| 227 |
* Count number of workers originally wanted and actually |
| 228 |
* launched. |
| 229 |
*/ |
| 230 |
estate->es_parallel_workers_to_launch += pcxt->nworkers_to_launch; |
| 231 |
estate->es_parallel_workers_launched += pcxt->nworkers_launched; |
| 232 |
|
| 233 |
/* Set up tuple queue readers to read the results. */ |
| 234 |
if (pcxt->nworkers_launched > 0) |
| 235 |
{ |
| 236 |
ExecParallelCreateReaders(node->pei); |
| 237 |
/* Make a working array showing the active readers */ |
| 238 |
node->nreaders = pcxt->nworkers_launched; |
| 239 |
node->reader = (TupleQueueReader **) |
| 240 |
palloc(node->nreaders * sizeof(TupleQueueReader *)); |
| 241 |
memcpy(node->reader, node->pei->reader, |
| 242 |
node->nreaders * sizeof(TupleQueueReader *)); |
| 243 |
} |
| 244 |
else |
| 245 |
{ |
| 246 |
/* No workers? Then never mind. */ |
| 247 |
node->nreaders = 0; |
| 248 |
node->reader = NULL; |
| 249 |
} |
| 250 |
} |
| 251 |
|
| 252 |
/* allow leader to participate if enabled or no choice */ |
| 253 |
if (parallel_leader_participation || node->nreaders == 0) |
| 254 |
node->need_to_scan_locally = true; |
| 255 |
node->initialized = true; |
| 256 |
} |
| 257 |
|
| 258 |
/* |
| 259 |
* Reset per-tuple memory context to free any expression evaluation |
| 260 |
* storage allocated in the previous tuple cycle. |
| 261 |
*/ |
| 262 |
econtext = node->ps.ps_ExprContext; |
| 263 |
ResetExprContext(econtext); |
| 264 |
|
| 265 |
/* |
| 266 |
* Get next tuple, either from one of our workers, or by running the plan |
| 267 |
* ourselves. |
| 268 |
*/ |
| 269 |
slot = gather_merge_getnext(node); |
| 270 |
if (TupIsNull(slot)) |
| 271 |
return NULL; |
| 272 |
|
| 273 |
/* If no projection is required, we're done. */ |
| 274 |
if (node->ps.ps_ProjInfo == NULL) |
| 275 |
return slot; |
| 276 |
|
| 277 |
/* |
| 278 |
* Form the result tuple using ExecProject(), and return it. |
| 279 |
*/ |
| 280 |
econtext->ecxt_outertuple = slot; |
| 281 |
return ExecProject(node->ps.ps_ProjInfo); |
| 282 |
} |
| 283 |
|
| 284 |
/* ---------------------------------------------------------------- |
| 285 |
* ExecEndGatherMerge |
| 286 |
* |
| 287 |
* frees any storage allocated through C routines. |
| 288 |
* ---------------------------------------------------------------- |
| 289 |
*/ |
| 290 |
void |
| 291 |
ExecEndGatherMerge(GatherMergeState *node) |
| 292 |
{ |
| 293 |
ExecEndNode(outerPlanState(node)); /* let children clean up first */ |
| 294 |
ExecShutdownGatherMerge(node); |
| 295 |
} |
| 296 |
|
| 297 |
/* ---------------------------------------------------------------- |
| 298 |
* ExecShutdownGatherMerge |
| 299 |
* |
| 300 |
* Destroy the setup for parallel workers including parallel context. |
| 301 |
* ---------------------------------------------------------------- |
| 302 |
*/ |
| 303 |
void |
| 304 |
ExecShutdownGatherMerge(GatherMergeState *node) |
| 305 |
{ |
| 306 |
ExecShutdownGatherMergeWorkers(node); |
| 307 |
|
| 308 |
/* Now destroy the parallel context. */ |
| 309 |
if (node->pei != NULL) |
| 310 |
{ |
| 311 |
ExecParallelCleanup(node->pei); |
| 312 |
node->pei = NULL; |
| 313 |
} |
| 314 |
} |
| 315 |
|
| 316 |
/* ---------------------------------------------------------------- |
| 317 |
* ExecShutdownGatherMergeWorkers |
| 318 |
* |
| 319 |
* Stop all the parallel workers. |
| 320 |
* ---------------------------------------------------------------- |
| 321 |
*/ |
| 322 |
static void |
| 323 |
ExecShutdownGatherMergeWorkers(GatherMergeState *node) |
| 324 |
{ |
| 325 |
if (node->pei != NULL) |
| 326 |
ExecParallelFinish(node->pei); |
| 327 |
|
| 328 |
/* Flush local copy of reader array */ |
| 329 |
if (node->reader) |
| 330 |
pfree(node->reader); |
| 331 |
node->reader = NULL; |
| 332 |
} |
| 333 |
|
| 334 |
/* ---------------------------------------------------------------- |
| 335 |
* ExecReScanGatherMerge |
| 336 |
* |
| 337 |
* Prepare to re-scan the result of a GatherMerge. |
| 338 |
* ---------------------------------------------------------------- |
| 339 |
*/ |
| 340 |
void |
| 341 |
ExecReScanGatherMerge(GatherMergeState *node) |
| 342 |
{ |
| 343 |
GatherMerge *gm = (GatherMerge *) node->ps.plan; |
| 344 |
PlanState *outerPlan = outerPlanState(node); |
| 345 |
|
| 346 |
/* Make sure any existing workers are gracefully shut down */ |
| 347 |
ExecShutdownGatherMergeWorkers(node); |
| 348 |
|
| 349 |
/* Free any unused tuples, so we don't leak memory across rescans */ |
| 350 |
gather_merge_clear_tuples(node); |
| 351 |
|
| 352 |
/* Mark node so that shared state will be rebuilt at next call */ |
| 353 |
node->initialized = false; |
| 354 |
node->gm_initialized = false; |
| 355 |
|
| 356 |
/* |
| 357 |
* Set child node's chgParam to tell it that the next scan might deliver a |
| 358 |
* different set of rows within the leader process. (The overall rowset |
| 359 |
* shouldn't change, but the leader process's subset might; hence nodes |
| 360 |
* between here and the parallel table scan node mustn't optimize on the |
| 361 |
* assumption of an unchanging rowset.) |
| 362 |
*/ |
| 363 |
if (gm->rescan_param >= 0) |
| 364 |
outerPlan->chgParam = bms_add_member(outerPlan->chgParam, |
| 365 |
gm->rescan_param); |
| 366 |
|
| 367 |
/* |
| 368 |
* If chgParam of subnode is not null then plan will be re-scanned by |
| 369 |
* first ExecProcNode. Note: because this does nothing if we have a |
| 370 |
* rescan_param, it's currently guaranteed that parallel-aware child nodes |
| 371 |
* will not see a ReScan call until after they get a ReInitializeDSM call. |
| 372 |
* That ordering might not be something to rely on, though. A good rule |
| 373 |
* of thumb is that ReInitializeDSM should reset only shared state, ReScan |
| 374 |
* should reset only local state, and anything that depends on both of |
| 375 |
* those steps being finished must wait until the first ExecProcNode call. |
| 376 |
*/ |
| 377 |
if (outerPlan->chgParam == NULL) |
| 378 |
ExecReScan(outerPlan); |
| 379 |
} |
| 380 |
|
| 381 |
/* |
| 382 |
* Set up the data structures that we'll need for Gather Merge. |
| 383 |
* |
| 384 |
* We allocate these once on the basis of gm->num_workers, which is an |
| 385 |
* upper bound for the number of workers we'll actually have. During |
| 386 |
* a rescan, we reset the structures to empty. This approach simplifies |
| 387 |
* not leaking memory across rescans. |
| 388 |
* |
| 389 |
* In the gm_slots[] array, index 0 is for the leader, and indexes 1 to n |
| 390 |
* are for workers. The values placed into gm_heap correspond to indexes |
| 391 |
* in gm_slots[]. The gm_tuple_buffers[] array, however, is indexed from |
| 392 |
* 0 to n-1; it has no entry for the leader. |
| 393 |
*/ |
| 394 |
static void |
| 395 |
gather_merge_setup(GatherMergeState *gm_state) |
| 396 |
{ |
| 397 |
GatherMerge *gm = castNode(GatherMerge, gm_state->ps.plan); |
| 398 |
int nreaders = gm->num_workers; |
| 399 |
int i; |
| 400 |
|
| 401 |
/* |
| 402 |
* Allocate gm_slots for the number of workers + one more slot for leader. |
| 403 |
* Slot 0 is always for the leader. Leader always calls ExecProcNode() to |
| 404 |
* read the tuple, and then stores it directly into its gm_slots entry. |
| 405 |
* For other slots, code below will call ExecInitExtraTupleSlot() to |
| 406 |
* create a slot for the worker's results. Note that during any single |
| 407 |
* scan, we might have fewer than num_workers available workers, in which |
| 408 |
* case the extra array entries go unused. |
| 409 |
*/ |
| 410 |
gm_state->gm_slots = (TupleTableSlot **) |
| 411 |
palloc0((nreaders + 1) * sizeof(TupleTableSlot *)); |
| 412 |
|
| 413 |
/* Allocate the tuple slot and tuple array for each worker */ |
| 414 |
gm_state->gm_tuple_buffers = (GMReaderTupleBuffer *) |
| 415 |
palloc0(nreaders * sizeof(GMReaderTupleBuffer)); |
| 416 |
|
| 417 |
for (i = 0; i < nreaders; i++) |
| 418 |
{ |
| 419 |
/* Allocate the tuple array with length MAX_TUPLE_STORE */ |
| 420 |
gm_state->gm_tuple_buffers[i].tuple = |
| 421 |
(MinimalTuple *) palloc0(sizeof(MinimalTuple) * MAX_TUPLE_STORE); |
| 422 |
|
| 423 |
/* Initialize tuple slot for worker */ |
| 424 |
gm_state->gm_slots[i + 1] = |
| 425 |
ExecInitExtraTupleSlot(gm_state->ps.state, gm_state->tupDesc, |
| 426 |
&TTSOpsMinimalTuple); |
| 427 |
} |
| 428 |
|
| 429 |
/* Allocate the resources for the merge */ |
| 430 |
gm_state->gm_heap = binaryheap_allocate(nreaders + 1, |
| 431 |
heap_compare_slots, |
| 432 |
gm_state); |
| 433 |
} |
| 434 |
|
| 435 |
/* |
| 436 |
* Initialize the Gather Merge. |
| 437 |
* |
| 438 |
* Reset data structures to ensure they're empty. Then pull at least one |
| 439 |
* tuple from leader + each worker (or set its "done" indicator), and set up |
| 440 |
* the heap. |
| 441 |
*/ |
| 442 |
static void |
| 443 |
gather_merge_init(GatherMergeState *gm_state) |
| 444 |
{ |
| 445 |
int nreaders = gm_state->nreaders; |
| 446 |
bool nowait = true; |
| 447 |
int i; |
| 448 |
|
| 449 |
/* Assert that gather_merge_setup made enough space */ |
| 450 |
Assert(nreaders <= castNode(GatherMerge, gm_state->ps.plan)->num_workers); |
| 451 |
|
| 452 |
/* Reset leader's tuple slot to empty */ |
| 453 |
gm_state->gm_slots[0] = NULL; |
| 454 |
|
| 455 |
/* Reset the tuple slot and tuple array for each worker */ |
| 456 |
for (i = 0; i < nreaders; i++) |
| 457 |
{ |
| 458 |
/* Reset tuple array to empty */ |
| 459 |
gm_state->gm_tuple_buffers[i].nTuples = 0; |
| 460 |
gm_state->gm_tuple_buffers[i].readCounter = 0; |
| 461 |
/* Reset done flag to not-done */ |
| 462 |
gm_state->gm_tuple_buffers[i].done = false; |
| 463 |
/* Ensure output slot is empty */ |
| 464 |
ExecClearTuple(gm_state->gm_slots[i + 1]); |
| 465 |
} |
| 466 |
|
| 467 |
/* Reset binary heap to empty */ |
| 468 |
binaryheap_reset(gm_state->gm_heap); |
| 469 |
|
| 470 |
/* |
| 471 |
* First, try to read a tuple from each worker (including leader) in |
| 472 |
* nowait mode. After this, if not all workers were able to produce a |
| 473 |
* tuple (or a "done" indication), then re-read from remaining workers, |
| 474 |
* this time using wait mode. Add all live readers (those producing at |
| 475 |
* least one tuple) to the heap. |
| 476 |
*/ |
| 477 |
reread: |
| 478 |
for (i = 0; i <= nreaders; i++) |
| 479 |
{ |
| 480 |
CHECK_FOR_INTERRUPTS(); |
| 481 |
|
| 482 |
/* skip this source if already known done */ |
| 483 |
if ((i == 0) ? gm_state->need_to_scan_locally : |
| 484 |
!gm_state->gm_tuple_buffers[i - 1].done) |
| 485 |
{ |
| 486 |
if (TupIsNull(gm_state->gm_slots[i])) |
| 487 |
{ |
| 488 |
/* Don't have a tuple yet, try to get one */ |
| 489 |
if (gather_merge_readnext(gm_state, i, nowait)) |
| 490 |
binaryheap_add_unordered(gm_state->gm_heap, |
| 491 |
Int32GetDatum(i)); |
| 492 |
} |
| 493 |
else |
| 494 |
{ |
| 495 |
/* |
| 496 |
* We already got at least one tuple from this worker, but |
| 497 |
* might as well see if it has any more ready by now. |
| 498 |
*/ |
| 499 |
load_tuple_array(gm_state, i); |
| 500 |
} |
| 501 |
} |
| 502 |
} |
| 503 |
|
| 504 |
/* need not recheck leader, since nowait doesn't matter for it */ |
| 505 |
for (i = 1; i <= nreaders; i++) |
| 506 |
{ |
| 507 |
if (!gm_state->gm_tuple_buffers[i - 1].done && |
| 508 |
TupIsNull(gm_state->gm_slots[i])) |
| 509 |
{ |
| 510 |
nowait = false; |
| 511 |
goto reread; |
| 512 |
} |
| 513 |
} |
| 514 |
|
| 515 |
/* Now heapify the heap. */ |
| 516 |
binaryheap_build(gm_state->gm_heap); |
| 517 |
|
| 518 |
gm_state->gm_initialized = true; |
| 519 |
} |
| 520 |
|
| 521 |
/* |
| 522 |
* Clear out the tuple table slot, and any unused pending tuples, |
| 523 |
* for each gather merge input. |
| 524 |
*/ |
| 525 |
static void |
| 526 |
gather_merge_clear_tuples(GatherMergeState *gm_state) |
| 527 |
{ |
| 528 |
int i; |
| 529 |
|
| 530 |
for (i = 0; i < gm_state->nreaders; i++) |
| 531 |
{ |
| 532 |
GMReaderTupleBuffer *tuple_buffer = &gm_state->gm_tuple_buffers[i]; |
| 533 |
|
| 534 |
while (tuple_buffer->readCounter < tuple_buffer->nTuples) |
| 535 |
pfree(tuple_buffer->tuple[tuple_buffer->readCounter++]); |
| 536 |
|
| 537 |
ExecClearTuple(gm_state->gm_slots[i + 1]); |
| 538 |
} |
| 539 |
} |
| 540 |
|
| 541 |
/* |
| 542 |
* Read the next tuple for gather merge. |
| 543 |
* |
| 544 |
* Fetch the sorted tuple out of the heap. |
| 545 |
*/ |
| 546 |
static TupleTableSlot * |
| 547 |
gather_merge_getnext(GatherMergeState *gm_state) |
| 548 |
{ |
| 549 |
int i; |
| 550 |
|
| 551 |
if (!gm_state->gm_initialized) |
| 552 |
{ |
| 553 |
/* |
| 554 |
* First time through: pull the first tuple from each participant, and |
| 555 |
* set up the heap. |
| 556 |
*/ |
| 557 |
gather_merge_init(gm_state); |
| 558 |
} |
| 559 |
else |
| 560 |
{ |
| 561 |
/* |
| 562 |
* Otherwise, pull the next tuple from whichever participant we |
| 563 |
* returned from last time, and reinsert that participant's index into |
| 564 |
* the heap, because it might now compare differently against the |
| 565 |
* other elements of the heap. |
| 566 |
*/ |
| 567 |
i = DatumGetInt32(binaryheap_first(gm_state->gm_heap)); |
| 568 |
|
| 569 |
if (gather_merge_readnext(gm_state, i, false)) |
| 570 |
binaryheap_replace_first(gm_state->gm_heap, Int32GetDatum(i)); |
| 571 |
else |
| 572 |
{ |
| 573 |
/* reader exhausted, remove it from heap */ |
| 574 |
(void) binaryheap_remove_first(gm_state->gm_heap); |
| 575 |
} |
| 576 |
} |
| 577 |
|
| 578 |
if (binaryheap_empty(gm_state->gm_heap)) |
| 579 |
{ |
| 580 |
/* All the queues are exhausted, and so is the heap */ |
| 581 |
gather_merge_clear_tuples(gm_state); |
| 582 |
return NULL; |
| 583 |
} |
| 584 |
else |
| 585 |
{ |
| 586 |
/* Return next tuple from whichever participant has the leading one */ |
| 587 |
i = DatumGetInt32(binaryheap_first(gm_state->gm_heap)); |
| 588 |
return gm_state->gm_slots[i]; |
| 589 |
} |
| 590 |
} |
| 591 |
|
| 592 |
/* |
| 593 |
* Read tuple(s) for given reader in nowait mode, and load into its tuple |
| 594 |
* array, until we have MAX_TUPLE_STORE of them or would have to block. |
| 595 |
*/ |
| 596 |
static void |
| 597 |
load_tuple_array(GatherMergeState *gm_state, int reader) |
| 598 |
{ |
| 599 |
GMReaderTupleBuffer *tuple_buffer; |
| 600 |
int i; |
| 601 |
|
| 602 |
/* Don't do anything if this is the leader. */ |
| 603 |
if (reader == 0) |
| 604 |
return; |
| 605 |
|
| 606 |
tuple_buffer = &gm_state->gm_tuple_buffers[reader - 1]; |
| 607 |
|
| 608 |
/* If there's nothing in the array, reset the counters to zero. */ |
| 609 |
if (tuple_buffer->nTuples == tuple_buffer->readCounter) |
| 610 |
tuple_buffer->nTuples = tuple_buffer->readCounter = 0; |
| 611 |
|
| 612 |
/* Try to fill additional slots in the array. */ |
| 613 |
for (i = tuple_buffer->nTuples; i < MAX_TUPLE_STORE; i++) |
| 614 |
{ |
| 615 |
MinimalTuple tuple; |
| 616 |
|
| 617 |
tuple = gm_readnext_tuple(gm_state, |
| 618 |
reader, |
| 619 |
true, |
| 620 |
&tuple_buffer->done); |
| 621 |
if (!tuple) |
| 622 |
break; |
| 623 |
tuple_buffer->tuple[i] = tuple; |
| 624 |
tuple_buffer->nTuples++; |
| 625 |
} |
| 626 |
} |
| 627 |
|
| 628 |
/* |
| 629 |
* Store the next tuple for a given reader into the appropriate slot. |
| 630 |
* |
| 631 |
* Returns true if successful, false if not (either reader is exhausted, |
| 632 |
* or we didn't want to wait for a tuple). Sets done flag if reader |
| 633 |
* is found to be exhausted. |
| 634 |
*/ |
| 635 |
static bool |
| 636 |
gather_merge_readnext(GatherMergeState *gm_state, int reader, bool nowait) |
| 637 |
{ |
| 638 |
GMReaderTupleBuffer *tuple_buffer; |
| 639 |
MinimalTuple tup; |
| 640 |
|
| 641 |
/* |
| 642 |
* If we're being asked to generate a tuple from the leader, then we just |
| 643 |
* call ExecProcNode as normal to produce one. |
| 644 |
*/ |
| 645 |
if (reader == 0) |
| 646 |
{ |
| 647 |
if (gm_state->need_to_scan_locally) |
| 648 |
{ |
| 649 |
PlanState *outerPlan = outerPlanState(gm_state); |
| 650 |
TupleTableSlot *outerTupleSlot; |
| 651 |
EState *estate = gm_state->ps.state; |
| 652 |
|
| 653 |
/* Install our DSA area while executing the plan. */ |
| 654 |
estate->es_query_dsa = gm_state->pei ? gm_state->pei->area : NULL; |
| 655 |
outerTupleSlot = ExecProcNode(outerPlan); |
| 656 |
estate->es_query_dsa = NULL; |
| 657 |
|
| 658 |
if (!TupIsNull(outerTupleSlot)) |
| 659 |
{ |
| 660 |
gm_state->gm_slots[0] = outerTupleSlot; |
| 661 |
return true; |
| 662 |
} |
| 663 |
/* need_to_scan_locally serves as "done" flag for leader */ |
| 664 |
gm_state->need_to_scan_locally = false; |
| 665 |
} |
| 666 |
return false; |
| 667 |
} |
| 668 |
|
| 669 |
/* Otherwise, check the state of the relevant tuple buffer. */ |
| 670 |
tuple_buffer = &gm_state->gm_tuple_buffers[reader - 1]; |
| 671 |
|
| 672 |
if (tuple_buffer->nTuples > tuple_buffer->readCounter) |
| 673 |
{ |
| 674 |
/* Return any tuple previously read that is still buffered. */ |
| 675 |
tup = tuple_buffer->tuple[tuple_buffer->readCounter++]; |
| 676 |
} |
| 677 |
else if (tuple_buffer->done) |
| 678 |
{ |
| 679 |
/* Reader is known to be exhausted. */ |
| 680 |
return false; |
| 681 |
} |
| 682 |
else |
| 683 |
{ |
| 684 |
/* Read and buffer next tuple. */ |
| 685 |
tup = gm_readnext_tuple(gm_state, |
| 686 |
reader, |
| 687 |
nowait, |
| 688 |
&tuple_buffer->done); |
| 689 |
if (!tup) |
| 690 |
return false; |
| 691 |
|
| 692 |
/* |
| 693 |
* Attempt to read more tuples in nowait mode and store them in the |
| 694 |
* pending-tuple array for the reader. |
| 695 |
*/ |
| 696 |
load_tuple_array(gm_state, reader); |
| 697 |
} |
| 698 |
|
| 699 |
Assert(tup); |
| 700 |
|
| 701 |
/* Build the TupleTableSlot for the given tuple */ |
| 702 |
ExecStoreMinimalTuple(tup, /* tuple to store */ |
| 703 |
gm_state->gm_slots[reader], /* slot in which to |
| 704 |
* store the tuple */ |
| 705 |
true); /* pfree tuple when done with it */ |
| 706 |
|
| 707 |
return true; |
| 708 |
} |
| 709 |
|
| 710 |
/* |
| 711 |
* Attempt to read a tuple from given worker. |
| 712 |
*/ |
| 713 |
static MinimalTuple |
| 714 |
gm_readnext_tuple(GatherMergeState *gm_state, int nreader, bool nowait, |
| 715 |
bool *done) |
| 716 |
{ |
| 717 |
TupleQueueReader *reader; |
| 718 |
MinimalTuple tup; |
| 719 |
|
| 720 |
/* Check for async events, particularly messages from workers. */ |
| 721 |
CHECK_FOR_INTERRUPTS(); |
| 722 |
|
| 723 |
/* |
| 724 |
* Attempt to read a tuple. |
| 725 |
* |
| 726 |
* Note that TupleQueueReaderNext will just return NULL for a worker which |
| 727 |
* fails to initialize. We'll treat that worker as having produced no |
| 728 |
* tuples; WaitForParallelWorkersToFinish will error out when we get |
| 729 |
* there. |
| 730 |
*/ |
| 731 |
reader = gm_state->reader[nreader - 1]; |
| 732 |
tup = TupleQueueReaderNext(reader, nowait, done); |
| 733 |
|
| 734 |
/* |
| 735 |
* Since we'll be buffering these across multiple calls, we need to make a |
| 736 |
* copy. |
| 737 |
*/ |
| 738 |
return tup ? heap_copy_minimal_tuple(tup, 0) : NULL; |
| 739 |
} |
| 740 |
|
| 741 |
/* |
| 742 |
* We have one slot for each item in the heap array. We use SlotNumber |
| 743 |
* to store slot indexes. This doesn't actually provide any formal |
| 744 |
* type-safety, but it makes the code more self-documenting. |
| 745 |
*/ |
| 746 |
typedef int32 SlotNumber; |
| 747 |
|
| 748 |
/* |
| 749 |
* Compare the tuples in the two given slots. |
| 750 |
*/ |
| 751 |
static int32 |
| 752 |
heap_compare_slots(Datum a, Datum b, void *arg) |
| 753 |
{ |
| 754 |
GatherMergeState *node = (GatherMergeState *) arg; |
| 755 |
SlotNumber slot1 = DatumGetInt32(a); |
| 756 |
SlotNumber slot2 = DatumGetInt32(b); |
| 757 |
|
| 758 |
TupleTableSlot *s1 = node->gm_slots[slot1]; |
| 759 |
TupleTableSlot *s2 = node->gm_slots[slot2]; |
| 760 |
int nkey; |
| 761 |
|
| 762 |
Assert(!TupIsNull(s1)); |
| 763 |
Assert(!TupIsNull(s2)); |
| 764 |
|
| 765 |
for (nkey = 0; nkey < node->gm_nkeys; nkey++) |
| 766 |
{ |
| 767 |
SortSupport sortKey = node->gm_sortkeys + nkey; |
| 768 |
AttrNumber attno = sortKey->ssup_attno; |
| 769 |
Datum datum1, |
| 770 |
datum2; |
| 771 |
bool isNull1, |
| 772 |
isNull2; |
| 773 |
int compare; |
| 774 |
|
| 775 |
datum1 = slot_getattr(s1, attno, &isNull1); |
| 776 |
datum2 = slot_getattr(s2, attno, &isNull2); |
| 777 |
|
| 778 |
compare = ApplySortComparator(datum1, isNull1, |
| 779 |
datum2, isNull2, |
| 780 |
sortKey); |
| 781 |
if (compare != 0) |
| 782 |
{ |
| 783 |
INVERT_COMPARE_RESULT(compare); |
| 784 |
return compare; |
| 785 |
} |
| 786 |
} |
| 787 |
return 0; |
| 788 |
} |