* indexam.c
* general index access method routines
*
- * Portions Copyright (c) 1996-2008, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1996-2017, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
*
* IDENTIFICATION
- * $PostgreSQL: pgsql/src/backend/access/index/indexam.c,v 1.103 2008/03/26 18:48:59 alvherre Exp $
+ * src/backend/access/index/indexam.c
*
* INTERFACE ROUTINES
* index_open - open an index relation by relation OID
* index_close - close an index relation
* index_beginscan - start a scan of an index with amgettuple
- * index_beginscan_multi - start a scan of an index with amgetmulti
+ * index_beginscan_bitmap - start a scan of an index with amgetbitmap
* index_rescan - restart a scan of an index
* index_endscan - end a scan
* index_insert - insert an index tuple into a relation
* index_markpos - mark a scan position
* index_restrpos - restore a scan position
- * index_getnext - get the next tuple from a scan
- * index_getmulti - get multiple tuples from a scan
+ * index_parallelscan_estimate - estimate shared memory for parallel scan
+ * index_parallelscan_initialize - initialize parallel scan
+ * index_parallelrescan - (re)start a parallel scan of an index
+ * index_beginscan_parallel - join parallel index scan
+ * index_getnext_tid - get the next TID from a scan
+ * index_fetch_heap - get the scan's next heap tuple
+ * index_getnext - get the next heap tuple from a scan
+ * index_getbitmap - get all tuples from a scan
* index_bulk_delete - bulk deletion of index tuples
* index_vacuum_cleanup - post-deletion cleanup of an index
+ * index_can_return - does index support index-only scans?
* index_getprocid - get a support procedure OID
* index_getprocinfo - get a support procedure's lookup info
*
#include "postgres.h"
-#include "access/genam.h"
-#include "access/heapam.h"
+#include "access/amapi.h"
+#include "access/relscan.h"
#include "access/transam.h"
+#include "access/xlog.h"
+#include "catalog/catalog.h"
+#include "catalog/index.h"
#include "pgstat.h"
-#include "utils/relcache.h"
+#include "storage/bufmgr.h"
+#include "storage/lmgr.h"
+#include "storage/predicate.h"
#include "utils/snapmgr.h"
+#include "utils/tqual.h"
/* ----------------------------------------------------------------
* macros used in index_ routines
+ *
+ * Note: the ReindexIsProcessingIndex() check in RELATION_CHECKS is there
+ * to check that we don't try to scan or do retail insertions into an index
+ * that is currently being rebuilt or pending rebuild. This helps to catch
+ * things that don't work when reindexing system catalogs. The assertion
+ * doesn't prevent the actual rebuild because we don't use RELATION_CHECKS
+ * when calling the index AM's ambuild routine, and there is no reason for
+ * ambuild to call its subsidiary routines through this file.
* ----------------------------------------------------------------
*/
#define RELATION_CHECKS \
( \
AssertMacro(RelationIsValid(indexRelation)), \
- AssertMacro(PointerIsValid(indexRelation->rd_am)) \
+ AssertMacro(PointerIsValid(indexRelation->rd_amroutine)), \
+ AssertMacro(!ReindexIsProcessingIndex(RelationGetRelid(indexRelation))) \
)
#define SCAN_CHECKS \
( \
AssertMacro(IndexScanIsValid(scan)), \
AssertMacro(RelationIsValid(scan->indexRelation)), \
- AssertMacro(PointerIsValid(scan->indexRelation->rd_am)) \
+ AssertMacro(PointerIsValid(scan->indexRelation->rd_amroutine)) \
)
-#define GET_REL_PROCEDURE(pname) \
+#define CHECK_REL_PROCEDURE(pname) \
do { \
- procedure = &indexRelation->rd_aminfo->pname; \
- if (!OidIsValid(procedure->fn_oid)) \
- { \
- RegProcedure procOid = indexRelation->rd_am->pname; \
- if (!RegProcedureIsValid(procOid)) \
- elog(ERROR, "invalid %s regproc", CppAsString(pname)); \
- fmgr_info_cxt(procOid, procedure, indexRelation->rd_indexcxt); \
- } \
+ if (indexRelation->rd_amroutine->pname == NULL) \
+ elog(ERROR, "function %s is not defined for index %s", \
+ CppAsString(pname), RelationGetRelationName(indexRelation)); \
} while(0)
-#define GET_SCAN_PROCEDURE(pname) \
+#define CHECK_SCAN_PROCEDURE(pname) \
do { \
- procedure = &scan->indexRelation->rd_aminfo->pname; \
- if (!OidIsValid(procedure->fn_oid)) \
- { \
- RegProcedure procOid = scan->indexRelation->rd_am->pname; \
- if (!RegProcedureIsValid(procOid)) \
- elog(ERROR, "invalid %s regproc", CppAsString(pname)); \
- fmgr_info_cxt(procOid, procedure, scan->indexRelation->rd_indexcxt); \
- } \
+ if (scan->indexRelation->rd_amroutine->pname == NULL) \
+ elog(ERROR, "function %s is not defined for index %s", \
+ CppAsString(pname), RelationGetRelationName(scan->indexRelation)); \
} while(0)
static IndexScanDesc index_beginscan_internal(Relation indexRelation,
- int nkeys, ScanKey key);
+ int nkeys, int norderbys, Snapshot snapshot,
+ ParallelIndexScanDesc pscan, bool temp_snap);
/* ----------------------------------------------------------------
* index_open - open an index relation by relation OID
*
* If lockmode is not "NoLock", the specified kind of lock is
- * obtained on the index. (Generally, NoLock should only be
+ * obtained on the index. (Generally, NoLock should only be
* used if the caller knows it has some appropriate lock on the
* index already.)
*
bool *isnull,
ItemPointer heap_t_ctid,
Relation heapRelation,
- bool check_uniqueness)
+ IndexUniqueCheck checkUnique)
{
- FmgrInfo *procedure;
-
RELATION_CHECKS;
- GET_REL_PROCEDURE(aminsert);
+ CHECK_REL_PROCEDURE(aminsert);
- /*
- * have the am's insert proc do all the work.
- */
- return DatumGetBool(FunctionCall6(procedure,
- PointerGetDatum(indexRelation),
- PointerGetDatum(values),
- PointerGetDatum(isnull),
- PointerGetDatum(heap_t_ctid),
- PointerGetDatum(heapRelation),
- BoolGetDatum(check_uniqueness)));
+ if (!(indexRelation->rd_amroutine->ampredlocks))
+ CheckForSerializableConflictIn(indexRelation,
+ (HeapTuple) NULL,
+ InvalidBuffer);
+
+ return indexRelation->rd_amroutine->aminsert(indexRelation, values, isnull,
+ heap_t_ctid, heapRelation,
+ checkUnique);
}
/*
* index_beginscan - start a scan of an index with amgettuple
*
- * Note: heapRelation may be NULL if there is no intention of calling
- * index_getnext on this scan; index_getnext_indexitem will not use the
- * heapRelation link (nor the snapshot). However, the caller had better
- * be holding some kind of lock on the heap relation in any case, to ensure
- * no one deletes it (or the index) out from under us. Caller must also
- * be holding a lock on the index.
+ * Caller must be holding suitable locks on the heap and the index.
*/
IndexScanDesc
index_beginscan(Relation heapRelation,
Relation indexRelation,
Snapshot snapshot,
- int nkeys, ScanKey key)
+ int nkeys, int norderbys)
{
IndexScanDesc scan;
- scan = index_beginscan_internal(indexRelation, nkeys, key);
+ scan = index_beginscan_internal(indexRelation, nkeys, norderbys, snapshot, NULL, false);
/*
* Save additional parameters into the scandesc. Everything else was set
* up by RelationGetIndexScan.
*/
- scan->is_multiscan = false;
scan->heapRelation = heapRelation;
scan->xs_snapshot = snapshot;
}
/*
- * index_beginscan_multi - start a scan of an index with amgetmulti
+ * index_beginscan_bitmap - start a scan of an index with amgetbitmap
*
* As above, caller had better be holding some lock on the parent heap
* relation, even though it's not explicitly mentioned here.
*/
IndexScanDesc
-index_beginscan_multi(Relation indexRelation,
- Snapshot snapshot,
- int nkeys, ScanKey key)
+index_beginscan_bitmap(Relation indexRelation,
+ Snapshot snapshot,
+ int nkeys)
{
IndexScanDesc scan;
- scan = index_beginscan_internal(indexRelation, nkeys, key);
+ scan = index_beginscan_internal(indexRelation, nkeys, 0, snapshot, NULL, false);
/*
* Save additional parameters into the scandesc. Everything else was set
* up by RelationGetIndexScan.
*/
- scan->is_multiscan = true;
scan->xs_snapshot = snapshot;
return scan;
*/
static IndexScanDesc
index_beginscan_internal(Relation indexRelation,
- int nkeys, ScanKey key)
+ int nkeys, int norderbys, Snapshot snapshot,
+ ParallelIndexScanDesc pscan, bool temp_snap)
{
IndexScanDesc scan;
- FmgrInfo *procedure;
RELATION_CHECKS;
- GET_REL_PROCEDURE(ambeginscan);
+ CHECK_REL_PROCEDURE(ambeginscan);
+
+ if (!(indexRelation->rd_amroutine->ampredlocks))
+ PredicateLockRelation(indexRelation, snapshot);
/*
* We hold a reference count to the relcache entry throughout the scan.
/*
* Tell the AM to open a scan.
*/
- scan = (IndexScanDesc)
- DatumGetPointer(FunctionCall3(procedure,
- PointerGetDatum(indexRelation),
- Int32GetDatum(nkeys),
- PointerGetDatum(key)));
+ scan = indexRelation->rd_amroutine->ambeginscan(indexRelation, nkeys,
+ norderbys);
+ /* Initialize information for parallel scan. */
+ scan->parallel_scan = pscan;
+ scan->xs_temp_snap = temp_snap;
return scan;
}
/* ----------------
* index_rescan - (re)start a scan of an index
*
- * The caller may specify a new set of scankeys (but the number of keys
- * cannot change). To restart the scan without changing keys, pass NULL
- * for the key array.
- *
- * Note that this is also called when first starting an indexscan;
- * see RelationGetIndexScan. Keys *must* be passed in that case,
- * unless scan->numberOfKeys is zero.
+ * During a restart, the caller may specify a new set of scankeys and/or
+ * orderbykeys; but the number of keys cannot differ from what index_beginscan
+ * was told. (Later we might relax that to "must not exceed", but currently
+ * the index AMs tend to assume that scan->numberOfKeys is what to believe.)
+ * To restart the scan without changing keys, pass NULL for the key arrays.
+ * (Of course, keys *must* be passed on the first call, unless
+ * scan->numberOfKeys is zero.)
* ----------------
*/
void
-index_rescan(IndexScanDesc scan, ScanKey key)
+index_rescan(IndexScanDesc scan,
+ ScanKey keys, int nkeys,
+ ScanKey orderbys, int norderbys)
{
- FmgrInfo *procedure;
-
SCAN_CHECKS;
- GET_SCAN_PROCEDURE(amrescan);
+ CHECK_SCAN_PROCEDURE(amrescan);
+
+ Assert(nkeys == scan->numberOfKeys);
+ Assert(norderbys == scan->numberOfOrderBys);
/* Release any held pin on a heap page */
if (BufferIsValid(scan->xs_cbuf))
scan->xs_cbuf = InvalidBuffer;
}
- scan->xs_next_hot = InvalidOffsetNumber;
+ scan->xs_continue_hot = false;
scan->kill_prior_tuple = false; /* for safety */
- FunctionCall2(procedure,
- PointerGetDatum(scan),
- PointerGetDatum(key));
+ scan->indexRelation->rd_amroutine->amrescan(scan, keys, nkeys,
+ orderbys, norderbys);
}
/* ----------------
void
index_endscan(IndexScanDesc scan)
{
- FmgrInfo *procedure;
-
SCAN_CHECKS;
- GET_SCAN_PROCEDURE(amendscan);
+ CHECK_SCAN_PROCEDURE(amendscan);
/* Release any held pin on a heap page */
if (BufferIsValid(scan->xs_cbuf))
}
/* End the AM's scan */
- FunctionCall1(procedure, PointerGetDatum(scan));
+ scan->indexRelation->rd_amroutine->amendscan(scan);
/* Release index refcount acquired by index_beginscan */
RelationDecrementReferenceCount(scan->indexRelation);
+ if (scan->xs_temp_snap)
+ UnregisterSnapshot(scan->xs_snapshot);
+
/* Release the scan data structure itself */
IndexScanEnd(scan);
}
void
index_markpos(IndexScanDesc scan)
{
- FmgrInfo *procedure;
-
SCAN_CHECKS;
- GET_SCAN_PROCEDURE(ammarkpos);
+ CHECK_SCAN_PROCEDURE(ammarkpos);
- FunctionCall1(procedure, PointerGetDatum(scan));
+ scan->indexRelation->rd_amroutine->ammarkpos(scan);
}
/* ----------------
* returnable tuple in each HOT chain, and so restoring the prior state at the
* granularity of the index AM is sufficient. Since the only current user
* of mark/restore functionality is nodeMergejoin.c, this effectively means
- * that merge-join plans only work for MVCC snapshots. This could be fixed
+ * that merge-join plans only work for MVCC snapshots. This could be fixed
* if necessary, but for now it seems unimportant.
* ----------------
*/
void
index_restrpos(IndexScanDesc scan)
{
- FmgrInfo *procedure;
-
Assert(IsMVCCSnapshot(scan->xs_snapshot));
SCAN_CHECKS;
- GET_SCAN_PROCEDURE(amrestrpos);
+ CHECK_SCAN_PROCEDURE(amrestrpos);
- scan->xs_next_hot = InvalidOffsetNumber;
+ scan->xs_continue_hot = false;
scan->kill_prior_tuple = false; /* for safety */
- FunctionCall1(procedure, PointerGetDatum(scan));
+ scan->indexRelation->rd_amroutine->amrestrpos(scan);
}
-/* ----------------
- * index_getnext - get the next heap tuple from a scan
+/*
+ * index_parallelscan_estimate - estimate shared memory for parallel scan
*
- * The result is the next heap tuple satisfying the scan keys and the
- * snapshot, or NULL if no more matching tuples exist. On success,
- * the buffer containing the heap tuple is pinned (the pin will be dropped
- * at the next index_getnext or index_endscan).
- * ----------------
+ * Currently, we don't pass any information to the AM-specific estimator,
+ * so it can probably only return a constant. In the future, we might need
+ * to pass more information.
*/
-HeapTuple
-index_getnext(IndexScanDesc scan, ScanDirection direction)
+Size
+index_parallelscan_estimate(Relation indexRelation, Snapshot snapshot)
{
- HeapTuple heapTuple = &scan->xs_ctup;
- ItemPointer tid = &heapTuple->t_self;
- FmgrInfo *procedure;
+ Size nbytes;
- SCAN_CHECKS;
- GET_SCAN_PROCEDURE(amgettuple);
+ RELATION_CHECKS;
+
+ nbytes = offsetof(ParallelIndexScanDescData, ps_snapshot_data);
+ nbytes = add_size(nbytes, EstimateSnapshotSpace(snapshot));
+ nbytes = MAXALIGN(nbytes);
/*
- * We always reset xs_hot_dead; if we are here then either we are just
- * starting the scan, or we previously returned a visible tuple, and in
- * either case it's inappropriate to kill the prior index entry.
+ * If amestimateparallelscan is not provided, assume there is no
+ * AM-specific data needed. (It's hard to believe that could work, but
+ * it's easy enough to cater to it here.)
*/
- scan->xs_hot_dead = false;
+ if (indexRelation->rd_amroutine->amestimateparallelscan != NULL)
+ nbytes = add_size(nbytes,
+ indexRelation->rd_amroutine->amestimateparallelscan());
- for (;;)
- {
- OffsetNumber offnum;
- bool at_chain_start;
- Page dp;
+ return nbytes;
+}
- if (scan->xs_next_hot != InvalidOffsetNumber)
- {
- /*
- * We are resuming scan of a HOT chain after having returned an
- * earlier member. Must still hold pin on current heap page.
- */
- Assert(BufferIsValid(scan->xs_cbuf));
- Assert(ItemPointerGetBlockNumber(tid) ==
- BufferGetBlockNumber(scan->xs_cbuf));
- Assert(TransactionIdIsValid(scan->xs_prev_xmax));
- offnum = scan->xs_next_hot;
- at_chain_start = false;
- scan->xs_next_hot = InvalidOffsetNumber;
- }
- else
- {
- bool found;
- Buffer prev_buf;
+/*
+ * index_parallelscan_initialize - initialize parallel scan
+ *
+ * We initialize both the ParallelIndexScanDesc proper and the AM-specific
+ * information which follows it.
+ *
+ * This function calls access method specific initialization routine to
+ * initialize am specific information. Call this just once in the leader
+ * process; then, individual workers attach via index_beginscan_parallel.
+ */
+void
+index_parallelscan_initialize(Relation heapRelation, Relation indexRelation,
+ Snapshot snapshot, ParallelIndexScanDesc target)
+{
+ Size offset;
- /*
- * If we scanned a whole HOT chain and found only dead tuples,
- * tell index AM to kill its entry for that TID.
- */
- scan->kill_prior_tuple = scan->xs_hot_dead;
+ RELATION_CHECKS;
- /*
- * The AM's gettuple proc finds the next index entry matching the
- * scan keys, and puts the TID in xs_ctup.t_self (ie, *tid).
- */
- found = DatumGetBool(FunctionCall2(procedure,
- PointerGetDatum(scan),
- Int32GetDatum(direction)));
+ offset = add_size(offsetof(ParallelIndexScanDescData, ps_snapshot_data),
+ EstimateSnapshotSpace(snapshot));
+ offset = MAXALIGN(offset);
- /* Reset kill flag immediately for safety */
- scan->kill_prior_tuple = false;
+ target->ps_relid = RelationGetRelid(heapRelation);
+ target->ps_indexid = RelationGetRelid(indexRelation);
+ target->ps_offset = offset;
+ SerializeSnapshot(snapshot, target->ps_snapshot_data);
- /* If we're out of index entries, break out of outer loop */
- if (!found)
- break;
+ /* aminitparallelscan is optional; assume no-op if not provided by AM */
+ if (indexRelation->rd_amroutine->aminitparallelscan != NULL)
+ {
+ void *amtarget;
- pgstat_count_index_tuples(scan->indexRelation, 1);
+ amtarget = OffsetToPointer(target, offset);
+ indexRelation->rd_amroutine->aminitparallelscan(amtarget);
+ }
+}
- /* Switch to correct buffer if we don't have it already */
- prev_buf = scan->xs_cbuf;
- scan->xs_cbuf = ReleaseAndReadBuffer(scan->xs_cbuf,
- scan->heapRelation,
- ItemPointerGetBlockNumber(tid));
+/* ----------------
+ * index_parallelrescan - (re)start a parallel scan of an index
+ * ----------------
+ */
+void
+index_parallelrescan(IndexScanDesc scan)
+{
+ SCAN_CHECKS;
- /*
- * Prune page, but only if we weren't already on this page
- */
- if (prev_buf != scan->xs_cbuf)
- heap_page_prune_opt(scan->heapRelation, scan->xs_cbuf,
- RecentGlobalXmin);
+ /* amparallelrescan is optional; assume no-op if not provided by AM */
+ if (scan->indexRelation->rd_amroutine->amparallelrescan != NULL)
+ scan->indexRelation->rd_amroutine->amparallelrescan(scan);
+}
- /* Prepare to scan HOT chain starting at index-referenced offnum */
- offnum = ItemPointerGetOffsetNumber(tid);
- at_chain_start = true;
+/*
+ * index_beginscan_parallel - join parallel index scan
+ *
+ * Caller must be holding suitable locks on the heap and the index.
+ */
+IndexScanDesc
+index_beginscan_parallel(Relation heaprel, Relation indexrel, int nkeys,
+ int norderbys, ParallelIndexScanDesc pscan)
+{
+ Snapshot snapshot;
+ IndexScanDesc scan;
- /* We don't know what the first tuple's xmin should be */
- scan->xs_prev_xmax = InvalidTransactionId;
+ Assert(RelationGetRelid(heaprel) == pscan->ps_relid);
+ snapshot = RestoreSnapshot(pscan->ps_snapshot_data);
+ RegisterSnapshot(snapshot);
+ scan = index_beginscan_internal(indexrel, nkeys, norderbys, snapshot,
+ pscan, true);
- /* Initialize flag to detect if all entries are dead */
- scan->xs_hot_dead = true;
- }
+ /*
+ * Save additional parameters into the scandesc. Everything else was set
+ * up by index_beginscan_internal.
+ */
+ scan->heapRelation = heaprel;
+ scan->xs_snapshot = snapshot;
- /* Obtain share-lock on the buffer so we can examine visibility */
- LockBuffer(scan->xs_cbuf, BUFFER_LOCK_SHARE);
+ return scan;
+}
- dp = (Page) BufferGetPage(scan->xs_cbuf);
+/* ----------------
+ * index_getnext_tid - get the next TID from a scan
+ *
+ * The result is the next TID satisfying the scan keys,
+ * or NULL if no more matching tuples exist.
+ * ----------------
+ */
+ItemPointer
+index_getnext_tid(IndexScanDesc scan, ScanDirection direction)
+{
+ bool found;
- /* Scan through possible multiple members of HOT-chain */
- for (;;)
- {
- ItemId lp;
- ItemPointer ctid;
+ SCAN_CHECKS;
+ CHECK_SCAN_PROCEDURE(amgettuple);
- /* check for bogus TID */
- if (offnum < FirstOffsetNumber ||
- offnum > PageGetMaxOffsetNumber(dp))
- break;
+ Assert(TransactionIdIsValid(RecentGlobalXmin));
- lp = PageGetItemId(dp, offnum);
-
- /* check for unused, dead, or redirected items */
- if (!ItemIdIsNormal(lp))
- {
- /* We should only see a redirect at start of chain */
- if (ItemIdIsRedirected(lp) && at_chain_start)
- {
- /* Follow the redirect */
- offnum = ItemIdGetRedirect(lp);
- at_chain_start = false;
- continue;
- }
- /* else must be end of chain */
- break;
- }
+ /*
+ * The AM's amgettuple proc finds the next index entry matching the scan
+ * keys, and puts the TID into scan->xs_ctup.t_self. It should also set
+ * scan->xs_recheck and possibly scan->xs_itup, though we pay no attention
+ * to those fields here.
+ */
+ found = scan->indexRelation->rd_amroutine->amgettuple(scan, direction);
- /*
- * We must initialize all of *heapTuple (ie, scan->xs_ctup) since
- * it is returned to the executor on success.
- */
- heapTuple->t_data = (HeapTupleHeader) PageGetItem(dp, lp);
- heapTuple->t_len = ItemIdGetLength(lp);
- ItemPointerSetOffsetNumber(tid, offnum);
- heapTuple->t_tableOid = RelationGetRelid(scan->heapRelation);
- ctid = &heapTuple->t_data->t_ctid;
+ /* Reset kill flag immediately for safety */
+ scan->kill_prior_tuple = false;
- /*
- * Shouldn't see a HEAP_ONLY tuple at chain start. (This test
- * should be unnecessary, since the chain root can't be removed
- * while we have pin on the index entry, but let's make it
- * anyway.)
- */
- if (at_chain_start && HeapTupleIsHeapOnly(heapTuple))
- break;
+ /* If we're out of index entries, we're done */
+ if (!found)
+ {
+ /* ... but first, release any held pin on a heap page */
+ if (BufferIsValid(scan->xs_cbuf))
+ {
+ ReleaseBuffer(scan->xs_cbuf);
+ scan->xs_cbuf = InvalidBuffer;
+ }
+ return NULL;
+ }
- /*
- * The xmin should match the previous xmax value, else chain is
- * broken. (Note: this test is not optional because it protects
- * us against the case where the prior chain member's xmax aborted
- * since we looked at it.)
- */
- if (TransactionIdIsValid(scan->xs_prev_xmax) &&
- !TransactionIdEquals(scan->xs_prev_xmax,
- HeapTupleHeaderGetXmin(heapTuple->t_data)))
- break;
+ pgstat_count_index_tuples(scan->indexRelation, 1);
- /* If it's visible per the snapshot, we must return it */
- if (HeapTupleSatisfiesVisibility(heapTuple, scan->xs_snapshot,
- scan->xs_cbuf))
- {
- /*
- * If the snapshot is MVCC, we know that it could accept at
- * most one member of the HOT chain, so we can skip examining
- * any more members. Otherwise, check for continuation of the
- * HOT-chain, and set state for next time.
- */
- if (IsMVCCSnapshot(scan->xs_snapshot))
- scan->xs_next_hot = InvalidOffsetNumber;
- else if (HeapTupleIsHotUpdated(heapTuple))
- {
- Assert(ItemPointerGetBlockNumber(ctid) ==
- ItemPointerGetBlockNumber(tid));
- scan->xs_next_hot = ItemPointerGetOffsetNumber(ctid);
- scan->xs_prev_xmax = HeapTupleHeaderGetXmax(heapTuple->t_data);
- }
- else
- scan->xs_next_hot = InvalidOffsetNumber;
-
- LockBuffer(scan->xs_cbuf, BUFFER_LOCK_UNLOCK);
-
- pgstat_count_heap_fetch(scan->indexRelation);
-
- return heapTuple;
- }
+ /* Return the TID of the tuple we found. */
+ return &scan->xs_ctup.t_self;
+}
- /*
- * If we can't see it, maybe no one else can either. Check to see
- * if the tuple is dead to all transactions. If we find that all
- * the tuples in the HOT chain are dead, we'll signal the index AM
- * to not return that TID on future indexscans.
- */
- if (scan->xs_hot_dead &&
- HeapTupleSatisfiesVacuum(heapTuple->t_data, RecentGlobalXmin,
- scan->xs_cbuf) != HEAPTUPLE_DEAD)
- scan->xs_hot_dead = false;
+/* ----------------
+ * index_fetch_heap - get the scan's next heap tuple
+ *
+ * The result is a visible heap tuple associated with the index TID most
+ * recently fetched by index_getnext_tid, or NULL if no more matching tuples
+ * exist. (There can be more than one matching tuple because of HOT chains,
+ * although when using an MVCC snapshot it should be impossible for more than
+ * one such tuple to exist.)
+ *
+ * On success, the buffer containing the heap tup is pinned (the pin will be
+ * dropped in a future index_getnext_tid, index_fetch_heap or index_endscan
+ * call).
+ *
+ * Note: caller must check scan->xs_recheck, and perform rechecking of the
+ * scan keys if required. We do not do that here because we don't have
+ * enough information to do it efficiently in the general case.
+ * ----------------
+ */
+HeapTuple
+index_fetch_heap(IndexScanDesc scan)
+{
+ ItemPointer tid = &scan->xs_ctup.t_self;
+ bool all_dead = false;
+ bool got_heap_tuple;
- /*
- * Check to see if HOT chain continues past this tuple; if so
- * fetch the next offnum (we don't bother storing it into
- * xs_next_hot, but must store xs_prev_xmax), and loop around.
- */
- if (HeapTupleIsHotUpdated(heapTuple))
- {
- Assert(ItemPointerGetBlockNumber(ctid) ==
- ItemPointerGetBlockNumber(tid));
- offnum = ItemPointerGetOffsetNumber(ctid);
- at_chain_start = false;
- scan->xs_prev_xmax = HeapTupleHeaderGetXmax(heapTuple->t_data);
- }
- else
- break; /* end of chain */
- } /* loop over a single HOT chain */
-
- LockBuffer(scan->xs_cbuf, BUFFER_LOCK_UNLOCK);
-
- /* Loop around to ask index AM for another TID */
- scan->xs_next_hot = InvalidOffsetNumber;
+ /* We can skip the buffer-switching logic if we're in mid-HOT chain. */
+ if (!scan->xs_continue_hot)
+ {
+ /* Switch to correct buffer if we don't have it already */
+ Buffer prev_buf = scan->xs_cbuf;
+
+ scan->xs_cbuf = ReleaseAndReadBuffer(scan->xs_cbuf,
+ scan->heapRelation,
+ ItemPointerGetBlockNumber(tid));
+
+ /*
+ * Prune page, but only if we weren't already on this page
+ */
+ if (prev_buf != scan->xs_cbuf)
+ heap_page_prune_opt(scan->heapRelation, scan->xs_cbuf);
}
- /* Release any held pin on a heap page */
- if (BufferIsValid(scan->xs_cbuf))
+ /* Obtain share-lock on the buffer so we can examine visibility */
+ LockBuffer(scan->xs_cbuf, BUFFER_LOCK_SHARE);
+ got_heap_tuple = heap_hot_search_buffer(tid, scan->heapRelation,
+ scan->xs_cbuf,
+ scan->xs_snapshot,
+ &scan->xs_ctup,
+ &all_dead,
+ !scan->xs_continue_hot);
+ LockBuffer(scan->xs_cbuf, BUFFER_LOCK_UNLOCK);
+
+ if (got_heap_tuple)
{
- ReleaseBuffer(scan->xs_cbuf);
- scan->xs_cbuf = InvalidBuffer;
+ /*
+ * Only in a non-MVCC snapshot can more than one member of the HOT
+ * chain be visible.
+ */
+ scan->xs_continue_hot = !IsMVCCSnapshot(scan->xs_snapshot);
+ pgstat_count_heap_fetch(scan->indexRelation);
+ return &scan->xs_ctup;
}
- return NULL; /* failure exit */
+ /* We've reached the end of the HOT chain. */
+ scan->xs_continue_hot = false;
+
+ /*
+ * If we scanned a whole HOT chain and found only dead tuples, tell index
+ * AM to kill its entry for that TID (this will take effect in the next
+ * amgettuple call, in index_getnext_tid). We do not do this when in
+ * recovery because it may violate MVCC to do so. See comments in
+ * RelationGetIndexScan().
+ */
+ if (!scan->xactStartedInRecovery)
+ scan->kill_prior_tuple = all_dead;
+
+ return NULL;
}
/* ----------------
- * index_getnext_indexitem - get the next index tuple from a scan
+ * index_getnext - get the next heap tuple from a scan
*
- * Finds the next index tuple satisfying the scan keys. Note that the
- * corresponding heap tuple is not accessed, and thus no time qual (snapshot)
- * check is done, other than the index AM's internal check for killed tuples
- * (which most callers of this routine will probably want to suppress by
- * setting scan->ignore_killed_tuples = false).
+ * The result is the next heap tuple satisfying the scan keys and the
+ * snapshot, or NULL if no more matching tuples exist.
+ *
+ * On success, the buffer containing the heap tup is pinned (the pin will be
+ * dropped in a future index_getnext_tid, index_fetch_heap or index_endscan
+ * call).
*
- * On success (TRUE return), the heap TID of the found index entry is in
- * scan->xs_ctup.t_self. scan->xs_cbuf is untouched.
+ * Note: caller must check scan->xs_recheck, and perform rechecking of the
+ * scan keys if required. We do not do that here because we don't have
+ * enough information to do it efficiently in the general case.
* ----------------
*/
-bool
-index_getnext_indexitem(IndexScanDesc scan,
- ScanDirection direction)
+HeapTuple
+index_getnext(IndexScanDesc scan, ScanDirection direction)
{
- FmgrInfo *procedure;
- bool found;
+ HeapTuple heapTuple;
+ ItemPointer tid;
- SCAN_CHECKS;
- GET_SCAN_PROCEDURE(amgettuple);
-
- /* just make sure this is false... */
- scan->kill_prior_tuple = false;
+ for (;;)
+ {
+ if (scan->xs_continue_hot)
+ {
+ /*
+ * We are resuming scan of a HOT chain after having returned an
+ * earlier member. Must still hold pin on current heap page.
+ */
+ Assert(BufferIsValid(scan->xs_cbuf));
+ Assert(ItemPointerGetBlockNumber(&scan->xs_ctup.t_self) ==
+ BufferGetBlockNumber(scan->xs_cbuf));
+ }
+ else
+ {
+ /* Time to fetch the next TID from the index */
+ tid = index_getnext_tid(scan, direction);
- /*
- * have the am's gettuple proc do all the work.
- */
- found = DatumGetBool(FunctionCall2(procedure,
- PointerGetDatum(scan),
- Int32GetDatum(direction)));
+ /* If we're out of index entries, we're done */
+ if (tid == NULL)
+ break;
+ }
- if (found)
- pgstat_count_index_tuples(scan->indexRelation, 1);
+ /*
+ * Fetch the next (or only) visible heap tuple for this index entry.
+ * If we don't find anything, loop around and grab the next TID from
+ * the index.
+ */
+ heapTuple = index_fetch_heap(scan);
+ if (heapTuple != NULL)
+ return heapTuple;
+ }
- return found;
+ return NULL; /* failure exit */
}
/* ----------------
- * index_getmulti - get multiple tuples from an index scan
+ * index_getbitmap - get all tuples at once from an index scan
*
- * Collects the TIDs of multiple heap tuples satisfying the scan keys.
+ * Adds the TIDs of all heap tuples satisfying the scan keys to a bitmap.
* Since there's no interlock between the index scan and the eventual heap
* access, this is only safe to use with MVCC-based snapshots: the heap
* item slot could have been replaced by a newer tuple by the time we get
* to it.
*
- * A TRUE result indicates more calls should occur; a FALSE result says the
- * scan is done. *returned_tids could be zero or nonzero in either case.
+ * Returns the number of matching tuples found. (Note: this might be only
+ * approximate, so it should only be used for statistical purposes.)
* ----------------
*/
-bool
-index_getmulti(IndexScanDesc scan,
- ItemPointer tids, int32 max_tids,
- int32 *returned_tids)
+int64
+index_getbitmap(IndexScanDesc scan, TIDBitmap *bitmap)
{
- FmgrInfo *procedure;
- bool found;
+ int64 ntids;
SCAN_CHECKS;
- GET_SCAN_PROCEDURE(amgetmulti);
+ CHECK_SCAN_PROCEDURE(amgetbitmap);
/* just make sure this is false... */
scan->kill_prior_tuple = false;
/*
- * have the am's getmulti proc do all the work.
+ * have the am's getbitmap proc do all the work.
*/
- found = DatumGetBool(FunctionCall4(procedure,
- PointerGetDatum(scan),
- PointerGetDatum(tids),
- Int32GetDatum(max_tids),
- PointerGetDatum(returned_tids)));
+ ntids = scan->indexRelation->rd_amroutine->amgetbitmap(scan, bitmap);
- pgstat_count_index_tuples(scan->indexRelation, *returned_tids);
+ pgstat_count_index_tuples(scan->indexRelation, ntids);
- return found;
+ return ntids;
}
/* ----------------
void *callback_state)
{
Relation indexRelation = info->index;
- FmgrInfo *procedure;
- IndexBulkDeleteResult *result;
RELATION_CHECKS;
- GET_REL_PROCEDURE(ambulkdelete);
-
- result = (IndexBulkDeleteResult *)
- DatumGetPointer(FunctionCall4(procedure,
- PointerGetDatum(info),
- PointerGetDatum(stats),
- PointerGetDatum((Pointer) callback),
- PointerGetDatum(callback_state)));
+ CHECK_REL_PROCEDURE(ambulkdelete);
- return result;
+ return indexRelation->rd_amroutine->ambulkdelete(info, stats,
+ callback, callback_state);
}
/* ----------------
IndexBulkDeleteResult *stats)
{
Relation indexRelation = info->index;
- FmgrInfo *procedure;
- IndexBulkDeleteResult *result;
RELATION_CHECKS;
- GET_REL_PROCEDURE(amvacuumcleanup);
+ CHECK_REL_PROCEDURE(amvacuumcleanup);
+
+ return indexRelation->rd_amroutine->amvacuumcleanup(info, stats);
+}
+
+/* ----------------
+ * index_can_return
+ *
+ * Does the index access method support index-only scans for the given
+ * column?
+ * ----------------
+ */
+bool
+index_can_return(Relation indexRelation, int attno)
+{
+ RELATION_CHECKS;
- result = (IndexBulkDeleteResult *)
- DatumGetPointer(FunctionCall2(procedure,
- PointerGetDatum(info),
- PointerGetDatum(stats)));
+ /* amcanreturn is optional; assume FALSE if not provided by AM */
+ if (indexRelation->rd_amroutine->amcanreturn == NULL)
+ return false;
- return result;
+ return indexRelation->rd_amroutine->amcanreturn(indexRelation, attno);
}
/* ----------------
* particular indexed attribute are those with both types equal to
* the index opclass' opcintype (note that this is subtly different
* from the indexed attribute's own type: it may be a binary-compatible
- * type instead). Only the default functions are stored in relcache
+ * type instead). Only the default functions are stored in relcache
* entries --- access methods can use the syscache to look up non-default
* functions.
*
int nproc;
int procindex;
- nproc = irel->rd_am->amsupport;
+ nproc = irel->rd_amroutine->amsupport;
Assert(procnum > 0 && procnum <= (uint16) nproc);
* index_getprocinfo
*
* This routine allows index AMs to keep fmgr lookup info for
- * support procs in the relcache. As above, only the "default"
+ * support procs in the relcache. As above, only the "default"
* functions for any particular indexed attribute are cached.
*
* Note: the return value points into cached data that will be lost during
int nproc;
int procindex;
- nproc = irel->rd_am->amsupport;
+ nproc = irel->rd_amroutine->amsupport;
Assert(procnum > 0 && procnum <= (uint16) nproc);
projects / postgresql / blobdiff