nodeHashjoin.c File Reference

#include "postgres.h"
#include "access/htup_details.h"
#include "access/parallel.h"
#include "executor/executor.h"
#include "executor/hashjoin.h"
#include "executor/nodeHash.h"
#include "executor/nodeHashjoin.h"
#include "miscadmin.h"
#include "utils/lsyscache.h"
#include "utils/sharedtuplestore.h"
#include "utils/wait_event.h"

Include dependency graph for nodeHashjoin.c:

Go to the source code of this file.

Macros
#define	HJ_BUILD_HASHTABLE   1
#define	HJ_NEED_NEW_OUTER   2
#define	HJ_SCAN_BUCKET   3
#define	HJ_FILL_OUTER_TUPLE   4
#define	HJ_FILL_INNER_TUPLES   5
#define	HJ_NEED_NEW_BATCH   6
#define	HJ_FILL_OUTER(hjstate)   ((hjstate)->hj_NullInnerTupleSlot != NULL)
#define	HJ_FILL_INNER(hjstate)   ((hjstate)->hj_NullOuterTupleSlot != NULL)

Functions
static TupleTableSlot *	ExecHashJoinOuterGetTuple (PlanState *outerNode, HashJoinState *hjstate, uint32 *hashvalue)
static TupleTableSlot *	ExecParallelHashJoinOuterGetTuple (PlanState *outerNode, HashJoinState *hjstate, uint32 *hashvalue)
static TupleTableSlot *	ExecHashJoinGetSavedTuple (HashJoinState *hjstate, BufFile *file, uint32 *hashvalue, TupleTableSlot *tupleSlot)
static bool	ExecHashJoinNewBatch (HashJoinState *hjstate)
static bool	ExecParallelHashJoinNewBatch (HashJoinState *hjstate)
static void	ExecParallelHashJoinPartitionOuter (HashJoinState *hjstate)
static pg_attribute_always_inline TupleTableSlot *	ExecHashJoinImpl (PlanState *pstate, bool parallel)
static TupleTableSlot *	ExecHashJoin (PlanState *pstate)
static TupleTableSlot *	ExecParallelHashJoin (PlanState *pstate)
HashJoinState *	ExecInitHashJoin (HashJoin *node, EState *estate, int eflags)
void	ExecEndHashJoin (HashJoinState *node)
void	ExecHashJoinSaveTuple (MinimalTuple tuple, uint32 hashvalue, BufFile **fileptr, HashJoinTable hashtable)
void	ExecReScanHashJoin (HashJoinState *node)
void	ExecShutdownHashJoin (HashJoinState *node)
void	ExecHashJoinEstimate (HashJoinState *state, ParallelContext *pcxt)
void	ExecHashJoinInitializeDSM (HashJoinState *state, ParallelContext *pcxt)
void	ExecHashJoinReInitializeDSM (HashJoinState *state, ParallelContext *pcxt)
void	ExecHashJoinInitializeWorker (HashJoinState *state, ParallelWorkerContext *pwcxt)

Macro Definition Documentation

HJ_BUILD_HASHTABLE

#define HJ_BUILD_HASHTABLE   1

Definition at line 180 of file nodeHashjoin.c.

HJ_FILL_INNER

#define HJ_FILL_INNER

(

hjstate

)

((hjstate)->hj_NullOuterTupleSlot != NULL)

Definition at line 190 of file nodeHashjoin.c.

HJ_FILL_INNER_TUPLES

#define HJ_FILL_INNER_TUPLES   5

Definition at line 184 of file nodeHashjoin.c.

HJ_FILL_OUTER

#define HJ_FILL_OUTER

(

hjstate

)

((hjstate)->hj_NullInnerTupleSlot != NULL)

Definition at line 188 of file nodeHashjoin.c.

HJ_FILL_OUTER_TUPLE

#define HJ_FILL_OUTER_TUPLE   4

Definition at line 183 of file nodeHashjoin.c.

HJ_NEED_NEW_BATCH

#define HJ_NEED_NEW_BATCH   6

Definition at line 185 of file nodeHashjoin.c.

HJ_NEED_NEW_OUTER

#define HJ_NEED_NEW_OUTER   2

Definition at line 181 of file nodeHashjoin.c.

HJ_SCAN_BUCKET

#define HJ_SCAN_BUCKET   3

Definition at line 182 of file nodeHashjoin.c.

Function Documentation

ExecEndHashJoin()

void ExecEndHashJoin

(

HashJoinState *

node

)

Definition at line 948 of file nodeHashjoin.c.

{
    /*
     * Free hash table
     */
    if (node->hj_HashTable)
    {
        ExecHashTableDestroy(node->hj_HashTable);
        node->hj_HashTable = NULL;
    }
 
    /*
     * clean up subtrees
     */
    ExecEndNode(outerPlanState(node));
    ExecEndNode(innerPlanState(node));
}

References ExecEndNode(), ExecHashTableDestroy(), fb(), HashJoinState::hj_HashTable, innerPlanState, and outerPlanState.

Referenced by ExecEndNode().

ExecHashJoin()

static TupleTableSlot * ExecHashJoin ( PlanState *  pstate )

static

Definition at line 684 of file nodeHashjoin.c.

{
    /*
     * On sufficiently smart compilers this should be inlined with the
     * parallel-aware branches removed.
     */
    return ExecHashJoinImpl(pstate, false);
}

References ExecHashJoinImpl().

Referenced by ExecInitHashJoin().

ExecHashJoinEstimate()

void ExecHashJoinEstimate	(	HashJoinState *	state,
		ParallelContext *	pcxt
	)

Definition at line 1648 of file nodeHashjoin.c.

{
    shm_toc_estimate_chunk(&pcxt->estimator, sizeof(ParallelHashJoinState));
    shm_toc_estimate_keys(&pcxt->estimator, 1);
}

References ParallelContext::estimator, shm_toc_estimate_chunk, and shm_toc_estimate_keys.

Referenced by ExecParallelEstimate().

ExecHashJoinGetSavedTuple()

static TupleTableSlot * ExecHashJoinGetSavedTuple ( HashJoinState *  hjstate, BufFile *  file, uint32 *  hashvalue, TupleTableSlot *  tupleSlot  )

static

Definition at line 1455 of file nodeHashjoin.c.

{
    uint32      header[2];
    size_t      nread;
    MinimalTuple tuple;
 
    /*
     * We check for interrupts here because this is typically taken as an
     * alternative code path to an ExecProcNode() call, which would include
     * such a check.
     */
    CHECK_FOR_INTERRUPTS();
 
    /*
     * Since both the hash value and the MinimalTuple length word are uint32,
     * we can read them both in one BufFileRead() call without any type
     * cheating.
     */
    nread = BufFileReadMaybeEOF(file, header, sizeof(header), true);
    if (nread == 0)             /* end of file */
    {
        ExecClearTuple(tupleSlot);
        return NULL;
    }
    *hashvalue = header[0];
    tuple = (MinimalTuple) palloc(header[1]);
    tuple->t_len = header[1];
    BufFileReadExact(file,
                     (char *) tuple + sizeof(uint32),
                     header[1] - sizeof(uint32));
    ExecForceStoreMinimalTuple(tuple, tupleSlot, true);
    return tupleSlot;
}

References BufFileReadExact(), BufFileReadMaybeEOF(), CHECK_FOR_INTERRUPTS, ExecClearTuple(), ExecForceStoreMinimalTuple(), fb(), palloc(), and MinimalTupleData::t_len.

Referenced by ExecHashJoinNewBatch(), and ExecHashJoinOuterGetTuple().

ExecHashJoinImpl()

static pg_attribute_always_inline TupleTableSlot * ExecHashJoinImpl ( PlanState *  pstate, bool  parallel  )

static

Definition at line 221 of file nodeHashjoin.c.

{
    HashJoinState *node = castNode(HashJoinState, pstate);
    PlanState  *outerNode;
    HashState  *hashNode;
    ExprState  *joinqual;
    ExprState  *otherqual;
    ExprContext *econtext;
    HashJoinTable hashtable;
    TupleTableSlot *outerTupleSlot;
    uint32      hashvalue;
    int         batchno;
    ParallelHashJoinState *parallel_state;
 
    /*
     * get information from HashJoin node
     */
    joinqual = node->js.joinqual;
    otherqual = node->js.ps.qual;
    hashNode = (HashState *) innerPlanState(node);
    outerNode = outerPlanState(node);
    hashtable = node->hj_HashTable;
    econtext = node->js.ps.ps_ExprContext;
    parallel_state = hashNode->parallel_state;
 
    /*
     * Reset per-tuple memory context to free any expression evaluation
     * storage allocated in the previous tuple cycle.
     */
    ResetExprContext(econtext);
 
    /*
     * run the hash join state machine
     */
    for (;;)
    {
        /*
         * It's possible to iterate this loop many times before returning a
         * tuple, in some pathological cases such as needing to move much of
         * the current batch to a later batch.  So let's check for interrupts
         * each time through.
         */
        CHECK_FOR_INTERRUPTS();
 
        switch (node->hj_JoinState)
        {
            case HJ_BUILD_HASHTABLE:
 
                /*
                 * First time through: build hash table for inner relation.
                 */
                Assert(hashtable == NULL);
 
                /*
                 * If the outer relation is completely empty, and it's not
                 * right/right-anti/full join, we can quit without building
                 * the hash table.  However, for an inner join it is only a
                 * win to check this when the outer relation's startup cost is
                 * less than the projected cost of building the hash table.
                 * Otherwise it's best to build the hash table first and see
                 * if the inner relation is empty.  (When it's a left join, we
                 * should always make this check, since we aren't going to be
                 * able to skip the join on the strength of an empty inner
                 * relation anyway.)
                 *
                 * If we are rescanning the join, we make use of information
                 * gained on the previous scan: don't bother to try the
                 * prefetch if the previous scan found the outer relation
                 * nonempty. This is not 100% reliable since with new
                 * parameters the outer relation might yield different
                 * results, but it's a good heuristic.
                 *
                 * The only way to make the check is to try to fetch a tuple
                 * from the outer plan node.  If we succeed, we have to stash
                 * it away for later consumption by ExecHashJoinOuterGetTuple.
                 */
                if (HJ_FILL_INNER(node))
                {
                    /* no chance to not build the hash table */
                    node->hj_FirstOuterTupleSlot = NULL;
                }
                else if (parallel)
                {
                    /*
                     * The empty-outer optimization is not implemented for
                     * shared hash tables, because no one participant can
                     * determine that there are no outer tuples, and it's not
                     * yet clear that it's worth the synchronization overhead
                     * of reaching consensus to figure that out.  So we have
                     * to build the hash table.
                     */
                    node->hj_FirstOuterTupleSlot = NULL;
                }
                else if (HJ_FILL_OUTER(node) ||
                         (outerNode->plan->startup_cost < hashNode->ps.plan->total_cost &&
                          !node->hj_OuterNotEmpty))
                {
                    node->hj_FirstOuterTupleSlot = ExecProcNode(outerNode);
                    if (TupIsNull(node->hj_FirstOuterTupleSlot))
                    {
                        node->hj_OuterNotEmpty = false;
                        return NULL;
                    }
                    else
                        node->hj_OuterNotEmpty = true;
                }
                else
                    node->hj_FirstOuterTupleSlot = NULL;
 
                /*
                 * Create the hash table.  If using Parallel Hash, then
                 * whoever gets here first will create the hash table and any
                 * later arrivals will merely attach to it.
                 */
                hashtable = ExecHashTableCreate(hashNode);
                node->hj_HashTable = hashtable;
 
                /*
                 * Execute the Hash node, to build the hash table.  If using
                 * Parallel Hash, then we'll try to help hashing unless we
                 * arrived too late.
                 */
                hashNode->hashtable = hashtable;
                (void) MultiExecProcNode((PlanState *) hashNode);
 
                /*
                 * If the inner relation is completely empty, and we're not
                 * doing a left outer join, we can quit without scanning the
                 * outer relation.
                 */
                if (hashtable->totalTuples == 0 && !HJ_FILL_OUTER(node))
                {
                    if (parallel)
                    {
                        /*
                         * Advance the build barrier to PHJ_BUILD_RUN before
                         * proceeding so we can negotiate resource cleanup.
                         */
                        Barrier    *build_barrier = &parallel_state->build_barrier;
 
                        while (BarrierPhase(build_barrier) < PHJ_BUILD_RUN)
                            BarrierArriveAndWait(build_barrier, 0);
                    }
                    return NULL;
                }
 
                /*
                 * need to remember whether nbatch has increased since we
                 * began scanning the outer relation
                 */
                hashtable->nbatch_outstart = hashtable->nbatch;
 
                /*
                 * Reset OuterNotEmpty for scan.  (It's OK if we fetched a
                 * tuple above, because ExecHashJoinOuterGetTuple will
                 * immediately set it again.)
                 */
                node->hj_OuterNotEmpty = false;
 
                if (parallel)
                {
                    Barrier    *build_barrier;
 
                    build_barrier = &parallel_state->build_barrier;
                    Assert(BarrierPhase(build_barrier) == PHJ_BUILD_HASH_OUTER ||
                           BarrierPhase(build_barrier) == PHJ_BUILD_RUN ||
                           BarrierPhase(build_barrier) == PHJ_BUILD_FREE);
                    if (BarrierPhase(build_barrier) == PHJ_BUILD_HASH_OUTER)
                    {
                        /*
                         * If multi-batch, we need to hash the outer relation
                         * up front.
                         */
                        if (hashtable->nbatch > 1)
                            ExecParallelHashJoinPartitionOuter(node);
                        BarrierArriveAndWait(build_barrier,
                                             WAIT_EVENT_HASH_BUILD_HASH_OUTER);
                    }
                    else if (BarrierPhase(build_barrier) == PHJ_BUILD_FREE)
                    {
                        /*
                         * If we attached so late that the job is finished and
                         * the batch state has been freed, we can return
                         * immediately.
                         */
                        return NULL;
                    }
 
                    /* Each backend should now select a batch to work on. */
                    Assert(BarrierPhase(build_barrier) == PHJ_BUILD_RUN);
                    hashtable->curbatch = -1;
                    node->hj_JoinState = HJ_NEED_NEW_BATCH;
 
                    continue;
                }
                else
                    node->hj_JoinState = HJ_NEED_NEW_OUTER;
 
                /* FALL THRU */
 
            case HJ_NEED_NEW_OUTER:
 
                /*
                 * We don't have an outer tuple, try to get the next one
                 */
                if (parallel)
                    outerTupleSlot =
                        ExecParallelHashJoinOuterGetTuple(outerNode, node,
                                                          &hashvalue);
                else
                    outerTupleSlot =
                        ExecHashJoinOuterGetTuple(outerNode, node, &hashvalue);
 
                if (TupIsNull(outerTupleSlot))
                {
                    /* end of batch, or maybe whole join */
                    if (HJ_FILL_INNER(node))
                    {
                        /* set up to scan for unmatched inner tuples */
                        if (parallel)
                        {
                            /*
                             * Only one process is currently allow to handle
                             * each batch's unmatched tuples, in a parallel
                             * join.
                             */
                            if (ExecParallelPrepHashTableForUnmatched(node))
                                node->hj_JoinState = HJ_FILL_INNER_TUPLES;
                            else
                                node->hj_JoinState = HJ_NEED_NEW_BATCH;
                        }
                        else
                        {
                            ExecPrepHashTableForUnmatched(node);
                            node->hj_JoinState = HJ_FILL_INNER_TUPLES;
                        }
                    }
                    else
                        node->hj_JoinState = HJ_NEED_NEW_BATCH;
                    continue;
                }
 
                econtext->ecxt_outertuple = outerTupleSlot;
                node->hj_MatchedOuter = false;
 
                /*
                 * Find the corresponding bucket for this tuple in the main
                 * hash table or skew hash table.
                 */
                node->hj_CurHashValue = hashvalue;
                ExecHashGetBucketAndBatch(hashtable, hashvalue,
                                          &node->hj_CurBucketNo, &batchno);
                node->hj_CurSkewBucketNo = ExecHashGetSkewBucket(hashtable,
                                                                 hashvalue);
                node->hj_CurTuple = NULL;
 
                /*
                 * The tuple might not belong to the current batch (where
                 * "current batch" includes the skew buckets if any).
                 */
                if (batchno != hashtable->curbatch &&
                    node->hj_CurSkewBucketNo == INVALID_SKEW_BUCKET_NO)
                {
                    bool        shouldFree;
                    MinimalTuple mintuple = ExecFetchSlotMinimalTuple(outerTupleSlot,
                                                                      &shouldFree);
 
                    /*
                     * Need to postpone this outer tuple to a later batch.
                     * Save it in the corresponding outer-batch file.
                     */
                    Assert(parallel_state == NULL);
                    Assert(batchno > hashtable->curbatch);
                    ExecHashJoinSaveTuple(mintuple, hashvalue,
                                          &hashtable->outerBatchFile[batchno],
                                          hashtable);
 
                    if (shouldFree)
                        heap_free_minimal_tuple(mintuple);
 
                    /* Loop around, staying in HJ_NEED_NEW_OUTER state */
                    continue;
                }
 
                /* OK, let's scan the bucket for matches */
                node->hj_JoinState = HJ_SCAN_BUCKET;
 
                /* FALL THRU */
 
            case HJ_SCAN_BUCKET:
 
                /*
                 * Scan the selected hash bucket for matches to current outer
                 */
                if (parallel)
                {
                    if (!ExecParallelScanHashBucket(node, econtext))
                    {
                        /* out of matches; check for possible outer-join fill */
                        node->hj_JoinState = HJ_FILL_OUTER_TUPLE;
                        continue;
                    }
                }
                else
                {
                    if (!ExecScanHashBucket(node, econtext))
                    {
                        /* out of matches; check for possible outer-join fill */
                        node->hj_JoinState = HJ_FILL_OUTER_TUPLE;
                        continue;
                    }
                }
 
                /*
                 * In a right-semijoin, we only need the first match for each
                 * inner tuple.
                 */
                if (node->js.jointype == JOIN_RIGHT_SEMI &&
                    HeapTupleHeaderHasMatch(HJTUPLE_MINTUPLE(node->hj_CurTuple)))
                    continue;
 
                /*
                 * We've got a match, but still need to test non-hashed quals.
                 * ExecScanHashBucket already set up all the state needed to
                 * call ExecQual.
                 *
                 * If we pass the qual, then save state for next call and have
                 * ExecProject form the projection, store it in the tuple
                 * table, and return the slot.
                 *
                 * Only the joinquals determine tuple match status, but all
                 * quals must pass to actually return the tuple.
                 */
                if (joinqual == NULL || ExecQual(joinqual, econtext))
                {
                    node->hj_MatchedOuter = true;
 
                    /*
                     * This is really only needed if HJ_FILL_INNER(node) or if
                     * we are in a right-semijoin, but we'll avoid the branch
                     * and just set it always.
                     */
                    if (!HeapTupleHeaderHasMatch(HJTUPLE_MINTUPLE(node->hj_CurTuple)))
                        HeapTupleHeaderSetMatch(HJTUPLE_MINTUPLE(node->hj_CurTuple));
 
                    /* In an antijoin, we never return a matched tuple */
                    if (node->js.jointype == JOIN_ANTI)
                    {
                        node->hj_JoinState = HJ_NEED_NEW_OUTER;
                        continue;
                    }
 
                    /*
                     * If we only need to consider the first matching inner
                     * tuple, then advance to next outer tuple after we've
                     * processed this one.
                     */
                    if (node->js.single_match)
                        node->hj_JoinState = HJ_NEED_NEW_OUTER;
 
                    /*
                     * In a right-antijoin, we never return a matched tuple.
                     * If it's not an inner_unique join, we need to stay on
                     * the current outer tuple to continue scanning the inner
                     * side for matches.
                     */
                    if (node->js.jointype == JOIN_RIGHT_ANTI)
                        continue;
 
                    if (otherqual == NULL || ExecQual(otherqual, econtext))
                        return ExecProject(node->js.ps.ps_ProjInfo);
                    else
                        InstrCountFiltered2(node, 1);
                }
                else
                    InstrCountFiltered1(node, 1);
                break;
 
            case HJ_FILL_OUTER_TUPLE:
 
                /*
                 * The current outer tuple has run out of matches, so check
                 * whether to emit a dummy outer-join tuple.  Whether we emit
                 * one or not, the next state is NEED_NEW_OUTER.
                 */
                node->hj_JoinState = HJ_NEED_NEW_OUTER;
 
                if (!node->hj_MatchedOuter &&
                    HJ_FILL_OUTER(node))
                {
                    /*
                     * Generate a fake join tuple with nulls for the inner
                     * tuple, and return it if it passes the non-join quals.
                     */
                    econtext->ecxt_innertuple = node->hj_NullInnerTupleSlot;
 
                    if (otherqual == NULL || ExecQual(otherqual, econtext))
                        return ExecProject(node->js.ps.ps_ProjInfo);
                    else
                        InstrCountFiltered2(node, 1);
                }
                break;
 
            case HJ_FILL_INNER_TUPLES:
 
                /*
                 * We have finished a batch, but we are doing
                 * right/right-anti/full join, so any unmatched inner tuples
                 * in the hashtable have to be emitted before we continue to
                 * the next batch.
                 */
                if (!(parallel ? ExecParallelScanHashTableForUnmatched(node, econtext)
                      : ExecScanHashTableForUnmatched(node, econtext)))
                {
                    /* no more unmatched tuples */
                    node->hj_JoinState = HJ_NEED_NEW_BATCH;
                    continue;
                }
 
                /*
                 * Generate a fake join tuple with nulls for the outer tuple,
                 * and return it if it passes the non-join quals.
                 */
                econtext->ecxt_outertuple = node->hj_NullOuterTupleSlot;
 
                if (otherqual == NULL || ExecQual(otherqual, econtext))
                    return ExecProject(node->js.ps.ps_ProjInfo);
                else
                    InstrCountFiltered2(node, 1);
                break;
 
            case HJ_NEED_NEW_BATCH:
 
                /*
                 * Try to advance to next batch.  Done if there are no more.
                 */
                if (parallel)
                {
                    if (!ExecParallelHashJoinNewBatch(node))
                        return NULL;    /* end of parallel-aware join */
                }
                else
                {
                    if (!ExecHashJoinNewBatch(node))
                        return NULL;    /* end of parallel-oblivious join */
                }
                node->hj_JoinState = HJ_NEED_NEW_OUTER;
                break;
 
            default:
                elog(ERROR, "unrecognized hashjoin state: %d",
                     (int) node->hj_JoinState);
        }
    }
}

References Assert, BarrierArriveAndWait(), BarrierPhase(), ParallelHashJoinState::build_barrier, castNode, CHECK_FOR_INTERRUPTS, HashJoinTableData::curbatch, ExprContext::ecxt_innertuple, ExprContext::ecxt_outertuple, elog, ERROR, ExecFetchSlotMinimalTuple(), ExecHashGetBucketAndBatch(), ExecHashGetSkewBucket(), ExecHashJoinNewBatch(), ExecHashJoinOuterGetTuple(), ExecHashJoinSaveTuple(), ExecHashTableCreate(), ExecParallelHashJoinNewBatch(), ExecParallelHashJoinOuterGetTuple(), ExecParallelHashJoinPartitionOuter(), ExecParallelPrepHashTableForUnmatched(), ExecParallelScanHashBucket(), ExecParallelScanHashTableForUnmatched(), ExecPrepHashTableForUnmatched(), ExecProcNode(), ExecProject(), ExecQual(), ExecScanHashBucket(), ExecScanHashTableForUnmatched(), fb(), heap_free_minimal_tuple(), HeapTupleHeaderHasMatch(), HeapTupleHeaderSetMatch(), HJ_BUILD_HASHTABLE, HashJoinState::hj_CurBucketNo, HashJoinState::hj_CurHashValue, HashJoinState::hj_CurSkewBucketNo, HashJoinState::hj_CurTuple, HJ_FILL_INNER, HJ_FILL_INNER_TUPLES, HJ_FILL_OUTER, HJ_FILL_OUTER_TUPLE, HashJoinState::hj_FirstOuterTupleSlot, HashJoinState::hj_HashTable, HashJoinState::hj_JoinState, HashJoinState::hj_MatchedOuter, HJ_NEED_NEW_BATCH, HJ_NEED_NEW_OUTER, HashJoinState::hj_NullInnerTupleSlot, HashJoinState::hj_NullOuterTupleSlot, HashJoinState::hj_OuterNotEmpty, HJ_SCAN_BUCKET, HJTUPLE_MINTUPLE, innerPlanState, InstrCountFiltered1, InstrCountFiltered2, INVALID_SKEW_BUCKET_NO, JOIN_ANTI, JOIN_RIGHT_ANTI, JOIN_RIGHT_SEMI, JoinState::joinqual, JoinState::jointype, HashJoinState::js, MultiExecProcNode(), HashJoinTableData::nbatch, HashJoinTableData::nbatch_outstart, HashJoinTableData::outerBatchFile, outerPlanState, PHJ_BUILD_FREE, PHJ_BUILD_HASH_OUTER, PHJ_BUILD_RUN, JoinState::ps, PlanState::ps_ExprContext, PlanState::ps_ProjInfo, PlanState::qual, ResetExprContext, JoinState::single_match, HashJoinTableData::totalTuples, and TupIsNull.

Referenced by ExecHashJoin(), and ExecParallelHashJoin().

ExecHashJoinInitializeDSM()

void ExecHashJoinInitializeDSM	(	HashJoinState *	state,
		ParallelContext *	pcxt
	)

Definition at line 1655 of file nodeHashjoin.c.

{
    int         plan_node_id = state->js.ps.plan->plan_node_id;
    HashState  *hashNode;
    ParallelHashJoinState *pstate;
 
    /*
     * Disable shared hash table mode if we failed to create a real DSM
     * segment, because that means that we don't have a DSA area to work with.
     */
    if (pcxt->seg == NULL)
        return;
 
    ExecSetExecProcNode(&state->js.ps, ExecParallelHashJoin);
 
    /*
     * Set up the state needed to coordinate access to the shared hash
     * table(s), using the plan node ID as the toc key.
     */
    pstate = shm_toc_allocate(pcxt->toc, sizeof(ParallelHashJoinState));
    shm_toc_insert(pcxt->toc, plan_node_id, pstate);
 
    /*
     * Set up the shared hash join state with no batches initially.
     * ExecHashTableCreate() will prepare at least one later and set nbatch
     * and space_allowed.
     */
    pstate->nbatch = 0;
    pstate->space_allowed = 0;
    pstate->batches = InvalidDsaPointer;
    pstate->old_batches = InvalidDsaPointer;
    pstate->nbuckets = 0;
    pstate->growth = PHJ_GROWTH_OK;
    pstate->chunk_work_queue = InvalidDsaPointer;
    pg_atomic_init_u32(&pstate->distributor, 0);
    pstate->nparticipants = pcxt->nworkers + 1;
    pstate->total_tuples = 0;
    LWLockInitialize(&pstate->lock,
                     LWTRANCHE_PARALLEL_HASH_JOIN);
    BarrierInit(&pstate->build_barrier, 0);
    BarrierInit(&pstate->grow_batches_barrier, 0);
    BarrierInit(&pstate->grow_buckets_barrier, 0);
 
    /* Set up the space we'll use for shared temporary files. */
    SharedFileSetInit(&pstate->fileset, pcxt->seg);
 
    /* Initialize the shared state in the hash node. */
    hashNode = (HashState *) innerPlanState(state);
    hashNode->parallel_state = pstate;
}

References BarrierInit(), ParallelHashJoinState::batches, ParallelHashJoinState::build_barrier, ParallelHashJoinState::chunk_work_queue, ParallelHashJoinState::distributor, ExecParallelHashJoin(), ExecSetExecProcNode(), fb(), ParallelHashJoinState::fileset, ParallelHashJoinState::grow_batches_barrier, ParallelHashJoinState::grow_buckets_barrier, ParallelHashJoinState::growth, innerPlanState, InvalidDsaPointer, ParallelHashJoinState::lock, LWLockInitialize(), ParallelHashJoinState::nbatch, ParallelHashJoinState::nbuckets, ParallelHashJoinState::nparticipants, ParallelContext::nworkers, ParallelHashJoinState::old_batches, pg_atomic_init_u32(), PHJ_GROWTH_OK, ParallelContext::seg, SharedFileSetInit(), shm_toc_allocate(), shm_toc_insert(), ParallelHashJoinState::space_allowed, ParallelContext::toc, and ParallelHashJoinState::total_tuples.

Referenced by ExecParallelInitializeDSM().

ExecHashJoinInitializeWorker()

void ExecHashJoinInitializeWorker	(	HashJoinState *	state,
		ParallelWorkerContext *	pwcxt
	)

Definition at line 1751 of file nodeHashjoin.c.

{
    HashState  *hashNode;
    int         plan_node_id = state->js.ps.plan->plan_node_id;
    ParallelHashJoinState *pstate =
        shm_toc_lookup(pwcxt->toc, plan_node_id, false);
 
    /* Attach to the space for shared temporary files. */
    SharedFileSetAttach(&pstate->fileset, pwcxt->seg);
 
    /* Attach to the shared state in the hash node. */
    hashNode = (HashState *) innerPlanState(state);
    hashNode->parallel_state = pstate;
 
    ExecSetExecProcNode(&state->js.ps, ExecParallelHashJoin);
}

References ExecParallelHashJoin(), ExecSetExecProcNode(), fb(), ParallelHashJoinState::fileset, innerPlanState, SharedFileSetAttach(), and shm_toc_lookup().

Referenced by ExecParallelInitializeWorker().

ExecHashJoinNewBatch()

static bool ExecHashJoinNewBatch ( HashJoinState *  hjstate )

static

Definition at line 1130 of file nodeHashjoin.c.

{
    HashJoinTable hashtable = hjstate->hj_HashTable;
    int         nbatch;
    int         curbatch;
    BufFile    *innerFile;
    TupleTableSlot *slot;
    uint32      hashvalue;
 
    nbatch = hashtable->nbatch;
    curbatch = hashtable->curbatch;
 
    if (curbatch > 0)
    {
        /*
         * We no longer need the previous outer batch file; close it right
         * away to free disk space.
         */
        if (hashtable->outerBatchFile[curbatch])
            BufFileClose(hashtable->outerBatchFile[curbatch]);
        hashtable->outerBatchFile[curbatch] = NULL;
    }
    else                        /* we just finished the first batch */
    {
        /*
         * Reset some of the skew optimization state variables, since we no
         * longer need to consider skew tuples after the first batch. The
         * memory context reset we are about to do will release the skew
         * hashtable itself.
         */
        hashtable->skewEnabled = false;
        hashtable->skewBucket = NULL;
        hashtable->skewBucketNums = NULL;
        hashtable->nSkewBuckets = 0;
        hashtable->spaceUsedSkew = 0;
    }
 
    /*
     * We can always skip over any batches that are completely empty on both
     * sides.  We can sometimes skip over batches that are empty on only one
     * side, but there are exceptions:
     *
     * 1. In a left/full outer join, we have to process outer batches even if
     * the inner batch is empty.  Similarly, in a right/right-anti/full outer
     * join, we have to process inner batches even if the outer batch is
     * empty.
     *
     * 2. If we have increased nbatch since the initial estimate, we have to
     * scan inner batches since they might contain tuples that need to be
     * reassigned to later inner batches.
     *
     * 3. Similarly, if we have increased nbatch since starting the outer
     * scan, we have to rescan outer batches in case they contain tuples that
     * need to be reassigned.
     */
    curbatch++;
    while (curbatch < nbatch &&
           (hashtable->outerBatchFile[curbatch] == NULL ||
            hashtable->innerBatchFile[curbatch] == NULL))
    {
        if (hashtable->outerBatchFile[curbatch] &&
            HJ_FILL_OUTER(hjstate))
            break;              /* must process due to rule 1 */
        if (hashtable->innerBatchFile[curbatch] &&
            HJ_FILL_INNER(hjstate))
            break;              /* must process due to rule 1 */
        if (hashtable->innerBatchFile[curbatch] &&
            nbatch != hashtable->nbatch_original)
            break;              /* must process due to rule 2 */
        if (hashtable->outerBatchFile[curbatch] &&
            nbatch != hashtable->nbatch_outstart)
            break;              /* must process due to rule 3 */
        /* We can ignore this batch. */
        /* Release associated temp files right away. */
        if (hashtable->innerBatchFile[curbatch])
            BufFileClose(hashtable->innerBatchFile[curbatch]);
        hashtable->innerBatchFile[curbatch] = NULL;
        if (hashtable->outerBatchFile[curbatch])
            BufFileClose(hashtable->outerBatchFile[curbatch]);
        hashtable->outerBatchFile[curbatch] = NULL;
        curbatch++;
    }
 
    if (curbatch >= nbatch)
        return false;           /* no more batches */
 
    hashtable->curbatch = curbatch;
 
    /*
     * Reload the hash table with the new inner batch (which could be empty)
     */
    ExecHashTableReset(hashtable);
 
    innerFile = hashtable->innerBatchFile[curbatch];
 
    if (innerFile != NULL)
    {
        if (BufFileSeek(innerFile, 0, 0, SEEK_SET))
            ereport(ERROR,
                    (errcode_for_file_access(),
                     errmsg("could not rewind hash-join temporary file")));
 
        while ((slot = ExecHashJoinGetSavedTuple(hjstate,
                                                 innerFile,
                                                 &hashvalue,
                                                 hjstate->hj_HashTupleSlot)))
        {
            /*
             * NOTE: some tuples may be sent to future batches.  Also, it is
             * possible for hashtable->nbatch to be increased here!
             */
            ExecHashTableInsert(hashtable, slot, hashvalue);
        }
 
        /*
         * after we build the hash table, the inner batch file is no longer
         * needed
         */
        BufFileClose(innerFile);
        hashtable->innerBatchFile[curbatch] = NULL;
    }
 
    /*
     * Rewind outer batch file (if present), so that we can start reading it.
     */
    if (hashtable->outerBatchFile[curbatch] != NULL)
    {
        if (BufFileSeek(hashtable->outerBatchFile[curbatch], 0, 0, SEEK_SET))
            ereport(ERROR,
                    (errcode_for_file_access(),
                     errmsg("could not rewind hash-join temporary file")));
    }
 
    return true;
}

References BufFileClose(), BufFileSeek(), HashJoinTableData::curbatch, ereport, errcode_for_file_access(), errmsg(), ERROR, ExecHashJoinGetSavedTuple(), ExecHashTableInsert(), ExecHashTableReset(), fb(), HJ_FILL_INNER, HJ_FILL_OUTER, HashJoinTableData::innerBatchFile, HashJoinTableData::nbatch, HashJoinTableData::nbatch_original, HashJoinTableData::nbatch_outstart, HashJoinTableData::nSkewBuckets, HashJoinTableData::outerBatchFile, HashJoinTableData::skewBucket, HashJoinTableData::skewBucketNums, HashJoinTableData::skewEnabled, and HashJoinTableData::spaceUsedSkew.

Referenced by ExecHashJoinImpl().

ExecHashJoinOuterGetTuple()

static TupleTableSlot * ExecHashJoinOuterGetTuple ( PlanState *  outerNode, HashJoinState *  hjstate, uint32 *  hashvalue  )

static

Definition at line 979 of file nodeHashjoin.c.

{
    HashJoinTable hashtable = hjstate->hj_HashTable;
    int         curbatch = hashtable->curbatch;
    TupleTableSlot *slot;
 
    if (curbatch == 0)          /* if it is the first pass */
    {
        /*
         * Check to see if first outer tuple was already fetched by
         * ExecHashJoin() and not used yet.
         */
        slot = hjstate->hj_FirstOuterTupleSlot;
        if (!TupIsNull(slot))
            hjstate->hj_FirstOuterTupleSlot = NULL;
        else
            slot = ExecProcNode(outerNode);
 
        while (!TupIsNull(slot))
        {
            bool        isnull;
 
            /*
             * We have to compute the tuple's hash value.
             */
            ExprContext *econtext = hjstate->js.ps.ps_ExprContext;
 
            econtext->ecxt_outertuple = slot;
 
            ResetExprContext(econtext);
 
            *hashvalue = DatumGetUInt32(ExecEvalExprSwitchContext(hjstate->hj_OuterHash,
                                                                  econtext,
                                                                  &isnull));
 
            if (!isnull)
            {
                /* remember outer relation is not empty for possible rescan */
                hjstate->hj_OuterNotEmpty = true;
 
                return slot;
            }
 
            /*
             * That tuple couldn't match because of a NULL, so discard it and
             * continue with the next one.
             */
            slot = ExecProcNode(outerNode);
        }
    }
    else if (curbatch < hashtable->nbatch)
    {
        BufFile    *file = hashtable->outerBatchFile[curbatch];
 
        /*
         * In outer-join cases, we could get here even though the batch file
         * is empty.
         */
        if (file == NULL)
            return NULL;
 
        slot = ExecHashJoinGetSavedTuple(hjstate,
                                         file,
                                         hashvalue,
                                         hjstate->hj_OuterTupleSlot);
        if (!TupIsNull(slot))
            return slot;
    }
 
    /* End of this batch */
    return NULL;
}

References HashJoinTableData::curbatch, DatumGetUInt32(), ExprContext::ecxt_outertuple, ExecEvalExprSwitchContext(), ExecHashJoinGetSavedTuple(), ExecProcNode(), fb(), HashJoinTableData::outerBatchFile, ResetExprContext, and TupIsNull.

Referenced by ExecHashJoinImpl().

ExecHashJoinReInitializeDSM()

void ExecHashJoinReInitializeDSM	(	HashJoinState *	state,
		ParallelContext *	pcxt
	)

Definition at line 1713 of file nodeHashjoin.c.

{
    int         plan_node_id = state->js.ps.plan->plan_node_id;
    ParallelHashJoinState *pstate;
 
    /* Nothing to do if we failed to create a DSM segment. */
    if (pcxt->seg == NULL)
        return;
 
    pstate = shm_toc_lookup(pcxt->toc, plan_node_id, false);
 
    /*
     * It would be possible to reuse the shared hash table in single-batch
     * cases by resetting and then fast-forwarding build_barrier to
     * PHJ_BUILD_FREE and batch 0's batch_barrier to PHJ_BATCH_PROBE, but
     * currently shared hash tables are already freed by now (by the last
     * participant to detach from the batch).  We could consider keeping it
     * around for single-batch joins.  We'd also need to adjust
     * finalize_plan() so that it doesn't record a dummy dependency for
     * Parallel Hash nodes, preventing the rescan optimization.  For now we
     * don't try.
     */
 
    /* Detach, freeing any remaining shared memory. */
    if (state->hj_HashTable != NULL)
    {
        ExecHashTableDetachBatch(state->hj_HashTable);
        ExecHashTableDetach(state->hj_HashTable);
    }
 
    /* Clear any shared batch files. */
    SharedFileSetDeleteAll(&pstate->fileset);
 
    /* Reset build_barrier to PHJ_BUILD_ELECT so we can go around again. */
    BarrierInit(&pstate->build_barrier, 0);
}

References BarrierInit(), ParallelHashJoinState::build_barrier, ExecHashTableDetach(), ExecHashTableDetachBatch(), fb(), ParallelHashJoinState::fileset, ParallelContext::seg, SharedFileSetDeleteAll(), shm_toc_lookup(), and ParallelContext::toc.

Referenced by ExecParallelReInitializeDSM().

ExecHashJoinSaveTuple()

void ExecHashJoinSaveTuple	(	MinimalTuple	tuple,
		uint32	hashvalue,
		BufFile **	fileptr,
		HashJoinTable	hashtable
	)

Definition at line 1414 of file nodeHashjoin.c.

{
    BufFile    *file = *fileptr;
 
    /*
     * The batch file is lazily created. If this is the first tuple written to
     * this batch, the batch file is created and its buffer is allocated in
     * the spillCxt context, NOT in the batchCxt.
     *
     * During the build phase, buffered files are created for inner batches.
     * Each batch's buffered file is closed (and its buffer freed) after the
     * batch is loaded into memory during the outer side scan. Therefore, it
     * is necessary to allocate the batch file buffer in a memory context
     * which outlives the batch itself.
     *
     * Also, we use spillCxt instead of hashCxt for a better accounting of the
     * spilling memory consumption.
     */
    if (file == NULL)
    {
        MemoryContext oldctx = MemoryContextSwitchTo(hashtable->spillCxt);
 
        file = BufFileCreateTemp(false);
        *fileptr = file;
 
        MemoryContextSwitchTo(oldctx);
    }
 
    BufFileWrite(file, &hashvalue, sizeof(uint32));
    BufFileWrite(file, tuple, tuple->t_len);
}

References BufFileCreateTemp(), BufFileWrite(), fb(), MemoryContextSwitchTo(), HashJoinTableData::spillCxt, and MinimalTupleData::t_len.

Referenced by ExecHashIncreaseNumBatches(), ExecHashJoinImpl(), ExecHashRemoveNextSkewBucket(), and ExecHashTableInsert().

ExecInitHashJoin()

HashJoinState * ExecInitHashJoin	(	HashJoin *	node,
		EState *	estate,
		int	eflags
	)

Definition at line 716 of file nodeHashjoin.c.

{
    HashJoinState *hjstate;
    Plan       *outerNode;
    Hash       *hashNode;
    TupleDesc   outerDesc,
                innerDesc;
    const TupleTableSlotOps *ops;
 
    /* check for unsupported flags */
    Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
 
    /*
     * create state structure
     */
    hjstate = makeNode(HashJoinState);
    hjstate->js.ps.plan = (Plan *) node;
    hjstate->js.ps.state = estate;
 
    /*
     * See ExecHashJoinInitializeDSM() and ExecHashJoinInitializeWorker()
     * where this function may be replaced with a parallel version, if we
     * managed to launch a parallel query.
     */
    hjstate->js.ps.ExecProcNode = ExecHashJoin;
    hjstate->js.jointype = node->join.jointype;
 
    /*
     * Miscellaneous initialization
     *
     * create expression context for node
     */
    ExecAssignExprContext(estate, &hjstate->js.ps);
 
    /*
     * initialize child nodes
     *
     * Note: we could suppress the REWIND flag for the inner input, which
     * would amount to betting that the hash will be a single batch.  Not
     * clear if this would be a win or not.
     */
    outerNode = outerPlan(node);
    hashNode = (Hash *) innerPlan(node);
 
    outerPlanState(hjstate) = ExecInitNode(outerNode, estate, eflags);
    outerDesc = ExecGetResultType(outerPlanState(hjstate));
    innerPlanState(hjstate) = ExecInitNode((Plan *) hashNode, estate, eflags);
    innerDesc = ExecGetResultType(innerPlanState(hjstate));
 
    /*
     * Initialize result slot, type and projection.
     */
    ExecInitResultTupleSlotTL(&hjstate->js.ps, &TTSOpsVirtual);
    ExecAssignProjectionInfo(&hjstate->js.ps, NULL);
 
    /*
     * tuple table initialization
     */
    ops = ExecGetResultSlotOps(outerPlanState(hjstate), NULL);
    hjstate->hj_OuterTupleSlot = ExecInitExtraTupleSlot(estate, outerDesc,
                                                        ops);
 
    /*
     * detect whether we need only consider the first matching inner tuple
     */
    hjstate->js.single_match = (node->join.inner_unique ||
                                node->join.jointype == JOIN_SEMI);
 
    /* set up null tuples for outer joins, if needed */
    switch (node->join.jointype)
    {
        case JOIN_INNER:
        case JOIN_SEMI:
        case JOIN_RIGHT_SEMI:
            break;
        case JOIN_LEFT:
        case JOIN_ANTI:
            hjstate->hj_NullInnerTupleSlot =
                ExecInitNullTupleSlot(estate, innerDesc, &TTSOpsVirtual);
            break;
        case JOIN_RIGHT:
        case JOIN_RIGHT_ANTI:
            hjstate->hj_NullOuterTupleSlot =
                ExecInitNullTupleSlot(estate, outerDesc, &TTSOpsVirtual);
            break;
        case JOIN_FULL:
            hjstate->hj_NullOuterTupleSlot =
                ExecInitNullTupleSlot(estate, outerDesc, &TTSOpsVirtual);
            hjstate->hj_NullInnerTupleSlot =
                ExecInitNullTupleSlot(estate, innerDesc, &TTSOpsVirtual);
            break;
        default:
            elog(ERROR, "unrecognized join type: %d",
                 (int) node->join.jointype);
    }
 
    /*
     * now for some voodoo.  our temporary tuple slot is actually the result
     * tuple slot of the Hash node (which is our inner plan).  we can do this
     * because Hash nodes don't return tuples via ExecProcNode() -- instead
     * the hash join node uses ExecScanHashBucket() to get at the contents of
     * the hash table.  -cim 6/9/91
     */
    {
        HashState  *hashstate = (HashState *) innerPlanState(hjstate);
        Hash       *hash = (Hash *) hashstate->ps.plan;
        TupleTableSlot *slot = hashstate->ps.ps_ResultTupleSlot;
        Oid        *outer_hashfuncid;
        Oid        *inner_hashfuncid;
        bool       *hash_strict;
        ListCell   *lc;
        int         nkeys;
 
 
        hjstate->hj_HashTupleSlot = slot;
 
        /*
         * Build ExprStates to obtain hash values for either side of the join.
         * This must be done here as ExecBuildHash32Expr needs to know how to
         * handle NULL inputs and the required handling of that depends on the
         * jointype.  We don't know the join type in ExecInitHash() and we
         * must build the ExprStates before ExecHashTableCreate() so we
         * properly attribute any SubPlans that exist in the hash expressions
         * to the correct PlanState.
         */
        nkeys = list_length(node->hashoperators);
 
        outer_hashfuncid = palloc_array(Oid, nkeys);
        inner_hashfuncid = palloc_array(Oid, nkeys);
        hash_strict = palloc_array(bool, nkeys);
 
        /*
         * Determine the hash function for each side of the join for the given
         * hash operator.
         */
        foreach(lc, node->hashoperators)
        {
            Oid         hashop = lfirst_oid(lc);
            int         i = foreach_current_index(lc);
 
            if (!get_op_hash_functions(hashop,
                                       &outer_hashfuncid[i],
                                       &inner_hashfuncid[i]))
                elog(ERROR,
                     "could not find hash function for hash operator %u",
                     hashop);
            hash_strict[i] = op_strict(hashop);
        }
 
        /*
         * Build an ExprState to generate the hash value for the expressions
         * on the outer of the join.  This ExprState must finish generating
         * the hash value when HJ_FILL_OUTER() is true.  Otherwise,
         * ExecBuildHash32Expr will set up the ExprState to abort early if it
         * finds a NULL.  In these cases, we don't need to store these tuples
         * in the hash table as the jointype does not require it.
         */
        hjstate->hj_OuterHash =
            ExecBuildHash32Expr(hjstate->js.ps.ps_ResultTupleDesc,
                                hjstate->js.ps.resultops,
                                outer_hashfuncid,
                                node->hashcollations,
                                node->hashkeys,
                                hash_strict,
                                &hjstate->js.ps,
                                0,
                                HJ_FILL_OUTER(hjstate));
 
        /* As above, but for the inner side of the join */
        hashstate->hash_expr =
            ExecBuildHash32Expr(hashstate->ps.ps_ResultTupleDesc,
                                hashstate->ps.resultops,
                                inner_hashfuncid,
                                node->hashcollations,
                                hash->hashkeys,
                                hash_strict,
                                &hashstate->ps,
                                0,
                                HJ_FILL_INNER(hjstate));
 
        /*
         * Set up the skew table hash function while we have a record of the
         * first key's hash function Oid.
         */
        if (OidIsValid(hash->skewTable))
        {
            hashstate->skew_hashfunction = palloc0_object(FmgrInfo);
            hashstate->skew_collation = linitial_oid(node->hashcollations);
            fmgr_info(outer_hashfuncid[0], hashstate->skew_hashfunction);
        }
 
        /* no need to keep these */
        pfree(outer_hashfuncid);
        pfree(inner_hashfuncid);
        pfree(hash_strict);
    }
 
    /*
     * initialize child expressions
     */
    hjstate->js.ps.qual =
        ExecInitQual(node->join.plan.qual, (PlanState *) hjstate);
    hjstate->js.joinqual =
        ExecInitQual(node->join.joinqual, (PlanState *) hjstate);
    hjstate->hashclauses =
        ExecInitQual(node->hashclauses, (PlanState *) hjstate);
 
    /*
     * initialize hash-specific info
     */
    hjstate->hj_HashTable = NULL;
    hjstate->hj_FirstOuterTupleSlot = NULL;
 
    hjstate->hj_CurHashValue = 0;
    hjstate->hj_CurBucketNo = 0;
    hjstate->hj_CurSkewBucketNo = INVALID_SKEW_BUCKET_NO;
    hjstate->hj_CurTuple = NULL;
 
    hjstate->hj_JoinState = HJ_BUILD_HASHTABLE;
    hjstate->hj_MatchedOuter = false;
    hjstate->hj_OuterNotEmpty = false;
 
    return hjstate;
}

References Assert, elog, ERROR, EXEC_FLAG_BACKWARD, EXEC_FLAG_MARK, ExecAssignExprContext(), ExecAssignProjectionInfo(), ExecBuildHash32Expr(), ExecGetResultSlotOps(), ExecGetResultType(), ExecHashJoin(), ExecInitExtraTupleSlot(), ExecInitNode(), ExecInitNullTupleSlot(), ExecInitQual(), ExecInitResultTupleSlotTL(), fb(), fmgr_info(), foreach_current_index, get_op_hash_functions(), hash(), HashJoin::hashclauses, HashJoin::hashcollations, HashJoin::hashkeys, HashJoin::hashoperators, HJ_BUILD_HASHTABLE, HJ_FILL_INNER, HJ_FILL_OUTER, i, Join::inner_unique, innerPlan, innerPlanState, INVALID_SKEW_BUCKET_NO, HashJoin::join, JOIN_ANTI, JOIN_FULL, JOIN_INNER, JOIN_LEFT, JOIN_RIGHT, JOIN_RIGHT_ANTI, JOIN_RIGHT_SEMI, JOIN_SEMI, Join::joinqual, Join::jointype, lfirst_oid, linitial_oid, list_length(), makeNode, OidIsValid, op_strict(), outerPlan, outerPlanState, palloc0_object, palloc_array, pfree(), Hash::plan, and TTSOpsVirtual.

Referenced by ExecInitNode().

ExecParallelHashJoin()

static TupleTableSlot * ExecParallelHashJoin ( PlanState *  pstate )

static

Definition at line 700 of file nodeHashjoin.c.

{
    /*
     * On sufficiently smart compilers this should be inlined with the
     * parallel-oblivious branches removed.
     */
    return ExecHashJoinImpl(pstate, true);
}

References ExecHashJoinImpl().

Referenced by ExecHashJoinInitializeDSM(), and ExecHashJoinInitializeWorker().

ExecParallelHashJoinNewBatch()

static bool ExecParallelHashJoinNewBatch ( HashJoinState *  hjstate )

static

Definition at line 1271 of file nodeHashjoin.c.

{
    HashJoinTable hashtable = hjstate->hj_HashTable;
    int         start_batchno;
    int         batchno;
 
    /*
     * If we were already attached to a batch, remember not to bother checking
     * it again, and detach from it (possibly freeing the hash table if we are
     * last to detach).
     */
    if (hashtable->curbatch >= 0)
    {
        hashtable->batches[hashtable->curbatch].done = true;
        ExecHashTableDetachBatch(hashtable);
    }
 
    /*
     * Search for a batch that isn't done.  We use an atomic counter to start
     * our search at a different batch in every participant when there are
     * more batches than participants.
     */
    batchno = start_batchno =
        pg_atomic_fetch_add_u32(&hashtable->parallel_state->distributor, 1) %
        hashtable->nbatch;
    do
    {
        uint32      hashvalue;
        MinimalTuple tuple;
        TupleTableSlot *slot;
 
        if (!hashtable->batches[batchno].done)
        {
            SharedTuplestoreAccessor *inner_tuples;
            Barrier    *batch_barrier =
                &hashtable->batches[batchno].shared->batch_barrier;
 
            switch (BarrierAttach(batch_barrier))
            {
                case PHJ_BATCH_ELECT:
 
                    /* One backend allocates the hash table. */
                    if (BarrierArriveAndWait(batch_barrier,
                                             WAIT_EVENT_HASH_BATCH_ELECT))
                        ExecParallelHashTableAlloc(hashtable, batchno);
                    /* Fall through. */
 
                case PHJ_BATCH_ALLOCATE:
                    /* Wait for allocation to complete. */
                    BarrierArriveAndWait(batch_barrier,
                                         WAIT_EVENT_HASH_BATCH_ALLOCATE);
                    /* Fall through. */
 
                case PHJ_BATCH_LOAD:
                    /* Start (or join in) loading tuples. */
                    ExecParallelHashTableSetCurrentBatch(hashtable, batchno);
                    inner_tuples = hashtable->batches[batchno].inner_tuples;
                    sts_begin_parallel_scan(inner_tuples);
                    while ((tuple = sts_parallel_scan_next(inner_tuples,
                                                           &hashvalue)))
                    {
                        ExecForceStoreMinimalTuple(tuple,
                                                   hjstate->hj_HashTupleSlot,
                                                   false);
                        slot = hjstate->hj_HashTupleSlot;
                        ExecParallelHashTableInsertCurrentBatch(hashtable, slot,
                                                                hashvalue);
                    }
                    sts_end_parallel_scan(inner_tuples);
                    BarrierArriveAndWait(batch_barrier,
                                         WAIT_EVENT_HASH_BATCH_LOAD);
                    /* Fall through. */
 
                case PHJ_BATCH_PROBE:
 
                    /*
                     * This batch is ready to probe.  Return control to
                     * caller. We stay attached to batch_barrier so that the
                     * hash table stays alive until everyone's finished
                     * probing it, but no participant is allowed to wait at
                     * this barrier again (or else a deadlock could occur).
                     * All attached participants must eventually detach from
                     * the barrier and one worker must advance the phase so
                     * that the final phase is reached.
                     */
                    ExecParallelHashTableSetCurrentBatch(hashtable, batchno);
                    sts_begin_parallel_scan(hashtable->batches[batchno].outer_tuples);
 
                    return true;
                case PHJ_BATCH_SCAN:
 
                    /*
                     * In principle, we could help scan for unmatched tuples,
                     * since that phase is already underway (the thing we
                     * can't do under current deadlock-avoidance rules is wait
                     * for others to arrive at PHJ_BATCH_SCAN, because
                     * PHJ_BATCH_PROBE emits tuples, but in this case we just
                     * got here without waiting).  That is not yet done.  For
                     * now, we just detach and go around again.  We have to
                     * use ExecHashTableDetachBatch() because there's a small
                     * chance we'll be the last to detach, and then we're
                     * responsible for freeing memory.
                     */
                    ExecParallelHashTableSetCurrentBatch(hashtable, batchno);
                    hashtable->batches[batchno].done = true;
                    ExecHashTableDetachBatch(hashtable);
                    break;
 
                case PHJ_BATCH_FREE:
 
                    /*
                     * Already done.  Detach and go around again (if any
                     * remain).
                     */
                    BarrierDetach(batch_barrier);
                    hashtable->batches[batchno].done = true;
                    hashtable->curbatch = -1;
                    break;
 
                default:
                    elog(ERROR, "unexpected batch phase %d",
                         BarrierPhase(batch_barrier));
            }
        }
        batchno = (batchno + 1) % hashtable->nbatch;
    } while (batchno != start_batchno);
 
    return false;
}

References BarrierArriveAndWait(), BarrierAttach(), BarrierDetach(), BarrierPhase(), ParallelHashJoinBatch::batch_barrier, HashJoinTableData::batches, HashJoinTableData::curbatch, ParallelHashJoinState::distributor, ParallelHashJoinBatchAccessor::done, elog, ERROR, ExecForceStoreMinimalTuple(), ExecHashTableDetachBatch(), ExecParallelHashTableAlloc(), ExecParallelHashTableInsertCurrentBatch(), ExecParallelHashTableSetCurrentBatch(), fb(), ParallelHashJoinBatchAccessor::inner_tuples, HashJoinTableData::nbatch, ParallelHashJoinBatchAccessor::outer_tuples, HashJoinTableData::parallel_state, pg_atomic_fetch_add_u32(), PHJ_BATCH_ALLOCATE, PHJ_BATCH_ELECT, PHJ_BATCH_FREE, PHJ_BATCH_LOAD, PHJ_BATCH_PROBE, PHJ_BATCH_SCAN, ParallelHashJoinBatchAccessor::shared, sts_begin_parallel_scan(), sts_end_parallel_scan(), and sts_parallel_scan_next().

Referenced by ExecHashJoinImpl().

ExecParallelHashJoinOuterGetTuple()

static TupleTableSlot * ExecParallelHashJoinOuterGetTuple ( PlanState *  outerNode, HashJoinState *  hjstate, uint32 *  hashvalue  )

static

Definition at line 1058 of file nodeHashjoin.c.

{
    HashJoinTable hashtable = hjstate->hj_HashTable;
    int         curbatch = hashtable->curbatch;
    TupleTableSlot *slot;
 
    /*
     * In the Parallel Hash case we only run the outer plan directly for
     * single-batch hash joins.  Otherwise we have to go to batch files, even
     * for batch 0.
     */
    if (curbatch == 0 && hashtable->nbatch == 1)
    {
        slot = ExecProcNode(outerNode);
 
        while (!TupIsNull(slot))
        {
            bool        isnull;
 
            ExprContext *econtext = hjstate->js.ps.ps_ExprContext;
 
            econtext->ecxt_outertuple = slot;
 
            ResetExprContext(econtext);
 
            *hashvalue = DatumGetUInt32(ExecEvalExprSwitchContext(hjstate->hj_OuterHash,
                                                                  econtext,
                                                                  &isnull));
 
            if (!isnull)
                return slot;
 
            /*
             * That tuple couldn't match because of a NULL, so discard it and
             * continue with the next one.
             */
            slot = ExecProcNode(outerNode);
        }
    }
    else if (curbatch < hashtable->nbatch)
    {
        MinimalTuple tuple;
 
        tuple = sts_parallel_scan_next(hashtable->batches[curbatch].outer_tuples,
                                       hashvalue);
        if (tuple != NULL)
        {
            ExecForceStoreMinimalTuple(tuple,
                                       hjstate->hj_OuterTupleSlot,
                                       false);
            slot = hjstate->hj_OuterTupleSlot;
            return slot;
        }
        else
            ExecClearTuple(hjstate->hj_OuterTupleSlot);
    }
 
    /* End of this batch */
    hashtable->batches[curbatch].outer_eof = true;
 
    return NULL;
}

References HashJoinTableData::batches, HashJoinTableData::curbatch, DatumGetUInt32(), ExprContext::ecxt_outertuple, ExecClearTuple(), ExecEvalExprSwitchContext(), ExecForceStoreMinimalTuple(), ExecProcNode(), fb(), HashJoinTableData::nbatch, ParallelHashJoinBatchAccessor::outer_eof, ParallelHashJoinBatchAccessor::outer_tuples, ResetExprContext, sts_parallel_scan_next(), and TupIsNull.

Referenced by ExecHashJoinImpl().

ExecParallelHashJoinPartitionOuter()

static void ExecParallelHashJoinPartitionOuter ( HashJoinState *  hjstate )

static

Definition at line 1597 of file nodeHashjoin.c.

{
    PlanState  *outerState = outerPlanState(hjstate);
    ExprContext *econtext = hjstate->js.ps.ps_ExprContext;
    HashJoinTable hashtable = hjstate->hj_HashTable;
    TupleTableSlot *slot;
    uint32      hashvalue;
    int         i;
 
    Assert(hjstate->hj_FirstOuterTupleSlot == NULL);
 
    /* Execute outer plan, writing all tuples to shared tuplestores. */
    for (;;)
    {
        bool        isnull;
 
        slot = ExecProcNode(outerState);
        if (TupIsNull(slot))
            break;
        econtext->ecxt_outertuple = slot;
 
        ResetExprContext(econtext);
 
        hashvalue = DatumGetUInt32(ExecEvalExprSwitchContext(hjstate->hj_OuterHash,
                                                             econtext,
                                                             &isnull));
 
        if (!isnull)
        {
            int         batchno;
            int         bucketno;
            bool        shouldFree;
            MinimalTuple mintup = ExecFetchSlotMinimalTuple(slot, &shouldFree);
 
            ExecHashGetBucketAndBatch(hashtable, hashvalue, &bucketno,
                                      &batchno);
            sts_puttuple(hashtable->batches[batchno].outer_tuples,
                         &hashvalue, mintup);
 
            if (shouldFree)
                heap_free_minimal_tuple(mintup);
        }
        CHECK_FOR_INTERRUPTS();
    }
 
    /* Make sure all outer partitions are readable by any backend. */
    for (i = 0; i < hashtable->nbatch; ++i)
        sts_end_write(hashtable->batches[i].outer_tuples);
}

References Assert, HashJoinTableData::batches, CHECK_FOR_INTERRUPTS, DatumGetUInt32(), ExprContext::ecxt_outertuple, ExecEvalExprSwitchContext(), ExecFetchSlotMinimalTuple(), ExecHashGetBucketAndBatch(), ExecProcNode(), fb(), heap_free_minimal_tuple(), i, HashJoinTableData::nbatch, ParallelHashJoinBatchAccessor::outer_tuples, outerPlanState, ResetExprContext, sts_end_write(), sts_puttuple(), and TupIsNull.

Referenced by ExecHashJoinImpl().

ExecReScanHashJoin()

void ExecReScanHashJoin

(

HashJoinState *

node

)

Definition at line 1494 of file nodeHashjoin.c.

{
    PlanState  *outerPlan = outerPlanState(node);
    PlanState  *innerPlan = innerPlanState(node);
 
    /*
     * In a multi-batch join, we currently have to do rescans the hard way,
     * primarily because batch temp files may have already been released. But
     * if it's a single-batch join, and there is no parameter change for the
     * inner subnode, then we can just re-use the existing hash table without
     * rebuilding it.
     */
    if (node->hj_HashTable != NULL)
    {
        if (node->hj_HashTable->nbatch == 1 &&
            innerPlan->chgParam == NULL)
        {
            /*
             * Okay to reuse the hash table; needn't rescan inner, either.
             *
             * However, if it's a right/right-anti/right-semi/full join, we'd
             * better reset the inner-tuple match flags contained in the
             * table.
             */
            if (HJ_FILL_INNER(node) || node->js.jointype == JOIN_RIGHT_SEMI)
                ExecHashTableResetMatchFlags(node->hj_HashTable);
 
            /*
             * Also, we need to reset our state about the emptiness of the
             * outer relation, so that the new scan of the outer will update
             * it correctly if it turns out to be empty this time. (There's no
             * harm in clearing it now because ExecHashJoin won't need the
             * info.  In the other cases, where the hash table doesn't exist
             * or we are destroying it, we leave this state alone because
             * ExecHashJoin will need it the first time through.)
             */
            node->hj_OuterNotEmpty = false;
 
            /* ExecHashJoin can skip the BUILD_HASHTABLE step */
            node->hj_JoinState = HJ_NEED_NEW_OUTER;
        }
        else
        {
            /* must destroy and rebuild hash table */
            HashState  *hashNode = castNode(HashState, innerPlan);
 
            Assert(hashNode->hashtable == node->hj_HashTable);
            /* accumulate stats from old hash table, if wanted */
            /* (this should match ExecShutdownHash) */
            if (hashNode->ps.instrument && !hashNode->hinstrument)
                hashNode->hinstrument = palloc0_object(HashInstrumentation);
            if (hashNode->hinstrument)
                ExecHashAccumInstrumentation(hashNode->hinstrument,
                                             hashNode->hashtable);
            /* for safety, be sure to clear child plan node's pointer too */
            hashNode->hashtable = NULL;
 
            ExecHashTableDestroy(node->hj_HashTable);
            node->hj_HashTable = NULL;
            node->hj_JoinState = HJ_BUILD_HASHTABLE;
 
            /*
             * if chgParam of subnode is not null then plan will be re-scanned
             * by first ExecProcNode.
             */
            if (innerPlan->chgParam == NULL)
                ExecReScan(innerPlan);
        }
    }
 
    /* Always reset intra-tuple state */
    node->hj_CurHashValue = 0;
    node->hj_CurBucketNo = 0;
    node->hj_CurSkewBucketNo = INVALID_SKEW_BUCKET_NO;
    node->hj_CurTuple = NULL;
 
    node->hj_MatchedOuter = false;
    node->hj_FirstOuterTupleSlot = NULL;
 
    /*
     * if chgParam of subnode is not null then plan will be re-scanned by
     * first ExecProcNode.
     */
    if (outerPlan->chgParam == NULL)
        ExecReScan(outerPlan);
}

References Assert, castNode, ExecHashAccumInstrumentation(), ExecHashTableDestroy(), ExecHashTableResetMatchFlags(), ExecReScan(), fb(), HJ_BUILD_HASHTABLE, HashJoinState::hj_CurBucketNo, HashJoinState::hj_CurHashValue, HashJoinState::hj_CurSkewBucketNo, HashJoinState::hj_CurTuple, HJ_FILL_INNER, HashJoinState::hj_FirstOuterTupleSlot, HashJoinState::hj_HashTable, HashJoinState::hj_JoinState, HashJoinState::hj_MatchedOuter, HJ_NEED_NEW_OUTER, HashJoinState::hj_OuterNotEmpty, innerPlan, innerPlanState, INVALID_SKEW_BUCKET_NO, JOIN_RIGHT_SEMI, JoinState::jointype, HashJoinState::js, HashJoinTableData::nbatch, outerPlan, outerPlanState, and palloc0_object.

Referenced by ExecReScan().

ExecShutdownHashJoin()

void ExecShutdownHashJoin

(

HashJoinState *

node

)

Definition at line 1582 of file nodeHashjoin.c.

{
    if (node->hj_HashTable)
    {
        /*
         * Detach from shared state before DSM memory goes away.  This makes
         * sure that we don't have any pointers into DSM memory by the time
         * ExecEndHashJoin runs.
         */
        ExecHashTableDetachBatch(node->hj_HashTable);
        ExecHashTableDetach(node->hj_HashTable);
    }
}

References ExecHashTableDetach(), ExecHashTableDetachBatch(), and HashJoinState::hj_HashTable.

Referenced by ExecShutdownNode_walker().