* nbtutils.c
* Utility code for Postgres btree implementation.
*
- * Portions Copyright (c) 1996-2007, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1996-2018, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
*
* IDENTIFICATION
- * $PostgreSQL: pgsql/src/backend/access/nbtree/nbtutils.c,v 1.83 2007/03/30 00:12:59 tgl Exp $
+ * src/backend/access/nbtree/nbtutils.c
*
*-------------------------------------------------------------------------
*/
#include <time.h>
-#include "access/genam.h"
#include "access/nbtree.h"
#include "access/reloptions.h"
-#include "executor/execdebug.h"
+#include "access/relscan.h"
#include "miscadmin.h"
-#include "storage/lwlock.h"
-#include "storage/shmem.h"
+#include "utils/array.h"
#include "utils/lsyscache.h"
+#include "utils/memutils.h"
+#include "utils/rel.h"
+typedef struct BTSortArrayContext
+{
+ FmgrInfo flinfo;
+ Oid collation;
+ bool reverse;
+} BTSortArrayContext;
+
+static Datum _bt_find_extreme_element(IndexScanDesc scan, ScanKey skey,
+ StrategyNumber strat,
+ Datum *elems, int nelems);
+static int _bt_sort_array_elements(IndexScanDesc scan, ScanKey skey,
+ bool reverse,
+ Datum *elems, int nelems);
+static int _bt_compare_array_elements(const void *a, const void *b, void *arg);
static bool _bt_compare_scankey_args(IndexScanDesc scan, ScanKey op,
ScanKey leftarg, ScanKey rightarg,
bool *result);
-static void _bt_mark_scankey_with_indoption(ScanKey skey, int16 *indoption);
+static bool _bt_fix_scankey_strategy(ScanKey skey, int16 *indoption);
static void _bt_mark_scankey_required(ScanKey skey);
static bool _bt_check_rowcompare(ScanKey skey,
IndexTuple tuple, TupleDesc tupdesc,
{
ScanKey skey;
TupleDesc itupdesc;
- int natts;
+ int indnatts PG_USED_FOR_ASSERTS_ONLY;
+ int indnkeyatts;
int16 *indoption;
int i;
itupdesc = RelationGetDescr(rel);
- natts = RelationGetNumberOfAttributes(rel);
+ indnatts = IndexRelationGetNumberOfAttributes(rel);
+ indnkeyatts = IndexRelationGetNumberOfKeyAttributes(rel);
indoption = rel->rd_indoption;
- skey = (ScanKey) palloc(natts * sizeof(ScanKeyData));
+ Assert(indnkeyatts > 0);
+ Assert(indnkeyatts <= indnatts);
+ Assert(BTreeTupleGetNAtts(itup, rel) == indnatts ||
+ BTreeTupleGetNAtts(itup, rel) == indnkeyatts);
+
+ /*
+ * We'll execute search using scan key constructed on key columns. Non-key
+ * (INCLUDE index) columns are always omitted from scan keys.
+ */
+ skey = (ScanKey) palloc(indnkeyatts * sizeof(ScanKeyData));
- for (i = 0; i < natts; i++)
+ for (i = 0; i < indnkeyatts; i++)
{
FmgrInfo *procinfo;
Datum arg;
(AttrNumber) (i + 1),
InvalidStrategy,
InvalidOid,
+ rel->rd_indcollation[i],
procinfo,
arg);
}
* comparison data ultimately used must match the key datatypes.
*
* The result cannot be used with _bt_compare(), unless comparison
- * data is first stored into the key entries. Currently this
+ * data is first stored into the key entries. Currently this
* routine is only called by nbtsort.c and tuplesort.c, which have
* their own comparison routines.
*/
_bt_mkscankey_nodata(Relation rel)
{
ScanKey skey;
- int natts;
+ int indnkeyatts;
int16 *indoption;
int i;
- natts = RelationGetNumberOfAttributes(rel);
+ indnkeyatts = IndexRelationGetNumberOfKeyAttributes(rel);
indoption = rel->rd_indoption;
- skey = (ScanKey) palloc(natts * sizeof(ScanKeyData));
+ skey = (ScanKey) palloc(indnkeyatts * sizeof(ScanKeyData));
- for (i = 0; i < natts; i++)
+ for (i = 0; i < indnkeyatts; i++)
{
FmgrInfo *procinfo;
int flags;
(AttrNumber) (i + 1),
InvalidStrategy,
InvalidOid,
+ rel->rd_indcollation[i],
procinfo,
(Datum) 0);
}
}
+/*
+ * _bt_preprocess_array_keys() -- Preprocess SK_SEARCHARRAY scan keys
+ *
+ * If there are any SK_SEARCHARRAY scan keys, deconstruct the array(s) and
+ * set up BTArrayKeyInfo info for each one that is an equality-type key.
+ * Prepare modified scan keys in so->arrayKeyData, which will hold the current
+ * array elements during each primitive indexscan operation. For inequality
+ * array keys, it's sufficient to find the extreme element value and replace
+ * the whole array with that scalar value.
+ *
+ * Note: the reason we need so->arrayKeyData, rather than just scribbling
+ * on scan->keyData, is that callers are permitted to call btrescan without
+ * supplying a new set of scankey data.
+ */
+void
+_bt_preprocess_array_keys(IndexScanDesc scan)
+{
+ BTScanOpaque so = (BTScanOpaque) scan->opaque;
+ int numberOfKeys = scan->numberOfKeys;
+ int16 *indoption = scan->indexRelation->rd_indoption;
+ int numArrayKeys;
+ ScanKey cur;
+ int i;
+ MemoryContext oldContext;
+
+ /* Quick check to see if there are any array keys */
+ numArrayKeys = 0;
+ for (i = 0; i < numberOfKeys; i++)
+ {
+ cur = &scan->keyData[i];
+ if (cur->sk_flags & SK_SEARCHARRAY)
+ {
+ numArrayKeys++;
+ Assert(!(cur->sk_flags & (SK_ROW_HEADER | SK_SEARCHNULL | SK_SEARCHNOTNULL)));
+ /* If any arrays are null as a whole, we can quit right now. */
+ if (cur->sk_flags & SK_ISNULL)
+ {
+ so->numArrayKeys = -1;
+ so->arrayKeyData = NULL;
+ return;
+ }
+ }
+ }
+
+ /* Quit if nothing to do. */
+ if (numArrayKeys == 0)
+ {
+ so->numArrayKeys = 0;
+ so->arrayKeyData = NULL;
+ return;
+ }
+
+ /*
+ * Make a scan-lifespan context to hold array-associated data, or reset it
+ * if we already have one from a previous rescan cycle.
+ */
+ if (so->arrayContext == NULL)
+ so->arrayContext = AllocSetContextCreate(CurrentMemoryContext,
+ "BTree array context",
+ ALLOCSET_SMALL_SIZES);
+ else
+ MemoryContextReset(so->arrayContext);
+
+ oldContext = MemoryContextSwitchTo(so->arrayContext);
+
+ /* Create modifiable copy of scan->keyData in the workspace context */
+ so->arrayKeyData = (ScanKey) palloc(scan->numberOfKeys * sizeof(ScanKeyData));
+ memcpy(so->arrayKeyData,
+ scan->keyData,
+ scan->numberOfKeys * sizeof(ScanKeyData));
+
+ /* Allocate space for per-array data in the workspace context */
+ so->arrayKeys = (BTArrayKeyInfo *) palloc0(numArrayKeys * sizeof(BTArrayKeyInfo));
+
+ /* Now process each array key */
+ numArrayKeys = 0;
+ for (i = 0; i < numberOfKeys; i++)
+ {
+ ArrayType *arrayval;
+ int16 elmlen;
+ bool elmbyval;
+ char elmalign;
+ int num_elems;
+ Datum *elem_values;
+ bool *elem_nulls;
+ int num_nonnulls;
+ int j;
+
+ cur = &so->arrayKeyData[i];
+ if (!(cur->sk_flags & SK_SEARCHARRAY))
+ continue;
+
+ /*
+ * First, deconstruct the array into elements. Anything allocated
+ * here (including a possibly detoasted array value) is in the
+ * workspace context.
+ */
+ arrayval = DatumGetArrayTypeP(cur->sk_argument);
+ /* We could cache this data, but not clear it's worth it */
+ get_typlenbyvalalign(ARR_ELEMTYPE(arrayval),
+ &elmlen, &elmbyval, &elmalign);
+ deconstruct_array(arrayval,
+ ARR_ELEMTYPE(arrayval),
+ elmlen, elmbyval, elmalign,
+ &elem_values, &elem_nulls, &num_elems);
+
+ /*
+ * Compress out any null elements. We can ignore them since we assume
+ * all btree operators are strict.
+ */
+ num_nonnulls = 0;
+ for (j = 0; j < num_elems; j++)
+ {
+ if (!elem_nulls[j])
+ elem_values[num_nonnulls++] = elem_values[j];
+ }
+
+ /* We could pfree(elem_nulls) now, but not worth the cycles */
+
+ /* If there's no non-nulls, the scan qual is unsatisfiable */
+ if (num_nonnulls == 0)
+ {
+ numArrayKeys = -1;
+ break;
+ }
+
+ /*
+ * If the comparison operator is not equality, then the array qual
+ * degenerates to a simple comparison against the smallest or largest
+ * non-null array element, as appropriate.
+ */
+ switch (cur->sk_strategy)
+ {
+ case BTLessStrategyNumber:
+ case BTLessEqualStrategyNumber:
+ cur->sk_argument =
+ _bt_find_extreme_element(scan, cur,
+ BTGreaterStrategyNumber,
+ elem_values, num_nonnulls);
+ continue;
+ case BTEqualStrategyNumber:
+ /* proceed with rest of loop */
+ break;
+ case BTGreaterEqualStrategyNumber:
+ case BTGreaterStrategyNumber:
+ cur->sk_argument =
+ _bt_find_extreme_element(scan, cur,
+ BTLessStrategyNumber,
+ elem_values, num_nonnulls);
+ continue;
+ default:
+ elog(ERROR, "unrecognized StrategyNumber: %d",
+ (int) cur->sk_strategy);
+ break;
+ }
+
+ /*
+ * Sort the non-null elements and eliminate any duplicates. We must
+ * sort in the same ordering used by the index column, so that the
+ * successive primitive indexscans produce data in index order.
+ */
+ num_elems = _bt_sort_array_elements(scan, cur,
+ (indoption[cur->sk_attno - 1] & INDOPTION_DESC) != 0,
+ elem_values, num_nonnulls);
+
+ /*
+ * And set up the BTArrayKeyInfo data.
+ */
+ so->arrayKeys[numArrayKeys].scan_key = i;
+ so->arrayKeys[numArrayKeys].num_elems = num_elems;
+ so->arrayKeys[numArrayKeys].elem_values = elem_values;
+ numArrayKeys++;
+ }
+
+ so->numArrayKeys = numArrayKeys;
+
+ MemoryContextSwitchTo(oldContext);
+}
+
+/*
+ * _bt_find_extreme_element() -- get least or greatest array element
+ *
+ * scan and skey identify the index column, whose opfamily determines the
+ * comparison semantics. strat should be BTLessStrategyNumber to get the
+ * least element, or BTGreaterStrategyNumber to get the greatest.
+ */
+static Datum
+_bt_find_extreme_element(IndexScanDesc scan, ScanKey skey,
+ StrategyNumber strat,
+ Datum *elems, int nelems)
+{
+ Relation rel = scan->indexRelation;
+ Oid elemtype,
+ cmp_op;
+ RegProcedure cmp_proc;
+ FmgrInfo flinfo;
+ Datum result;
+ int i;
+
+ /*
+ * Determine the nominal datatype of the array elements. We have to
+ * support the convention that sk_subtype == InvalidOid means the opclass
+ * input type; this is a hack to simplify life for ScanKeyInit().
+ */
+ elemtype = skey->sk_subtype;
+ if (elemtype == InvalidOid)
+ elemtype = rel->rd_opcintype[skey->sk_attno - 1];
+
+ /*
+ * Look up the appropriate comparison operator in the opfamily.
+ *
+ * Note: it's possible that this would fail, if the opfamily is
+ * incomplete, but it seems quite unlikely that an opfamily would omit
+ * non-cross-type comparison operators for any datatype that it supports
+ * at all.
+ */
+ cmp_op = get_opfamily_member(rel->rd_opfamily[skey->sk_attno - 1],
+ elemtype,
+ elemtype,
+ strat);
+ if (!OidIsValid(cmp_op))
+ elog(ERROR, "missing operator %d(%u,%u) in opfamily %u",
+ strat, elemtype, elemtype,
+ rel->rd_opfamily[skey->sk_attno - 1]);
+ cmp_proc = get_opcode(cmp_op);
+ if (!RegProcedureIsValid(cmp_proc))
+ elog(ERROR, "missing oprcode for operator %u", cmp_op);
+
+ fmgr_info(cmp_proc, &flinfo);
+
+ Assert(nelems > 0);
+ result = elems[0];
+ for (i = 1; i < nelems; i++)
+ {
+ if (DatumGetBool(FunctionCall2Coll(&flinfo,
+ skey->sk_collation,
+ elems[i],
+ result)))
+ result = elems[i];
+ }
+
+ return result;
+}
+
+/*
+ * _bt_sort_array_elements() -- sort and de-dup array elements
+ *
+ * The array elements are sorted in-place, and the new number of elements
+ * after duplicate removal is returned.
+ *
+ * scan and skey identify the index column, whose opfamily determines the
+ * comparison semantics. If reverse is true, we sort in descending order.
+ */
+static int
+_bt_sort_array_elements(IndexScanDesc scan, ScanKey skey,
+ bool reverse,
+ Datum *elems, int nelems)
+{
+ Relation rel = scan->indexRelation;
+ Oid elemtype;
+ RegProcedure cmp_proc;
+ BTSortArrayContext cxt;
+ int last_non_dup;
+ int i;
+
+ if (nelems <= 1)
+ return nelems; /* no work to do */
+
+ /*
+ * Determine the nominal datatype of the array elements. We have to
+ * support the convention that sk_subtype == InvalidOid means the opclass
+ * input type; this is a hack to simplify life for ScanKeyInit().
+ */
+ elemtype = skey->sk_subtype;
+ if (elemtype == InvalidOid)
+ elemtype = rel->rd_opcintype[skey->sk_attno - 1];
+
+ /*
+ * Look up the appropriate comparison function in the opfamily.
+ *
+ * Note: it's possible that this would fail, if the opfamily is
+ * incomplete, but it seems quite unlikely that an opfamily would omit
+ * non-cross-type support functions for any datatype that it supports at
+ * all.
+ */
+ cmp_proc = get_opfamily_proc(rel->rd_opfamily[skey->sk_attno - 1],
+ elemtype,
+ elemtype,
+ BTORDER_PROC);
+ if (!RegProcedureIsValid(cmp_proc))
+ elog(ERROR, "missing support function %d(%u,%u) in opfamily %u",
+ BTORDER_PROC, elemtype, elemtype,
+ rel->rd_opfamily[skey->sk_attno - 1]);
+
+ /* Sort the array elements */
+ fmgr_info(cmp_proc, &cxt.flinfo);
+ cxt.collation = skey->sk_collation;
+ cxt.reverse = reverse;
+ qsort_arg((void *) elems, nelems, sizeof(Datum),
+ _bt_compare_array_elements, (void *) &cxt);
+
+ /* Now scan the sorted elements and remove duplicates */
+ last_non_dup = 0;
+ for (i = 1; i < nelems; i++)
+ {
+ int32 compare;
+
+ compare = DatumGetInt32(FunctionCall2Coll(&cxt.flinfo,
+ cxt.collation,
+ elems[last_non_dup],
+ elems[i]));
+ if (compare != 0)
+ elems[++last_non_dup] = elems[i];
+ }
+
+ return last_non_dup + 1;
+}
+
+/*
+ * qsort_arg comparator for sorting array elements
+ */
+static int
+_bt_compare_array_elements(const void *a, const void *b, void *arg)
+{
+ Datum da = *((const Datum *) a);
+ Datum db = *((const Datum *) b);
+ BTSortArrayContext *cxt = (BTSortArrayContext *) arg;
+ int32 compare;
+
+ compare = DatumGetInt32(FunctionCall2Coll(&cxt->flinfo,
+ cxt->collation,
+ da, db));
+ if (cxt->reverse)
+ compare = -compare;
+ return compare;
+}
+
+/*
+ * _bt_start_array_keys() -- Initialize array keys at start of a scan
+ *
+ * Set up the cur_elem counters and fill in the first sk_argument value for
+ * each array scankey. We can't do this until we know the scan direction.
+ */
+void
+_bt_start_array_keys(IndexScanDesc scan, ScanDirection dir)
+{
+ BTScanOpaque so = (BTScanOpaque) scan->opaque;
+ int i;
+
+ for (i = 0; i < so->numArrayKeys; i++)
+ {
+ BTArrayKeyInfo *curArrayKey = &so->arrayKeys[i];
+ ScanKey skey = &so->arrayKeyData[curArrayKey->scan_key];
+
+ Assert(curArrayKey->num_elems > 0);
+ if (ScanDirectionIsBackward(dir))
+ curArrayKey->cur_elem = curArrayKey->num_elems - 1;
+ else
+ curArrayKey->cur_elem = 0;
+ skey->sk_argument = curArrayKey->elem_values[curArrayKey->cur_elem];
+ }
+}
+
+/*
+ * _bt_advance_array_keys() -- Advance to next set of array elements
+ *
+ * Returns true if there is another set of values to consider, false if not.
+ * On true result, the scankeys are initialized with the next set of values.
+ */
+bool
+_bt_advance_array_keys(IndexScanDesc scan, ScanDirection dir)
+{
+ BTScanOpaque so = (BTScanOpaque) scan->opaque;
+ bool found = false;
+ int i;
+
+ /*
+ * We must advance the last array key most quickly, since it will
+ * correspond to the lowest-order index column among the available
+ * qualifications. This is necessary to ensure correct ordering of output
+ * when there are multiple array keys.
+ */
+ for (i = so->numArrayKeys - 1; i >= 0; i--)
+ {
+ BTArrayKeyInfo *curArrayKey = &so->arrayKeys[i];
+ ScanKey skey = &so->arrayKeyData[curArrayKey->scan_key];
+ int cur_elem = curArrayKey->cur_elem;
+ int num_elems = curArrayKey->num_elems;
+
+ if (ScanDirectionIsBackward(dir))
+ {
+ if (--cur_elem < 0)
+ {
+ cur_elem = num_elems - 1;
+ found = false; /* need to advance next array key */
+ }
+ else
+ found = true;
+ }
+ else
+ {
+ if (++cur_elem >= num_elems)
+ {
+ cur_elem = 0;
+ found = false; /* need to advance next array key */
+ }
+ else
+ found = true;
+ }
+
+ curArrayKey->cur_elem = cur_elem;
+ skey->sk_argument = curArrayKey->elem_values[cur_elem];
+ if (found)
+ break;
+ }
+
+ /* advance parallel scan */
+ if (scan->parallel_scan != NULL)
+ _bt_parallel_advance_array_keys(scan);
+
+ return found;
+}
+
+/*
+ * _bt_mark_array_keys() -- Handle array keys during btmarkpos
+ *
+ * Save the current state of the array keys as the "mark" position.
+ */
+void
+_bt_mark_array_keys(IndexScanDesc scan)
+{
+ BTScanOpaque so = (BTScanOpaque) scan->opaque;
+ int i;
+
+ for (i = 0; i < so->numArrayKeys; i++)
+ {
+ BTArrayKeyInfo *curArrayKey = &so->arrayKeys[i];
+
+ curArrayKey->mark_elem = curArrayKey->cur_elem;
+ }
+}
+
+/*
+ * _bt_restore_array_keys() -- Handle array keys during btrestrpos
+ *
+ * Restore the array keys to where they were when the mark was set.
+ */
+void
+_bt_restore_array_keys(IndexScanDesc scan)
+{
+ BTScanOpaque so = (BTScanOpaque) scan->opaque;
+ bool changed = false;
+ int i;
+
+ /* Restore each array key to its position when the mark was set */
+ for (i = 0; i < so->numArrayKeys; i++)
+ {
+ BTArrayKeyInfo *curArrayKey = &so->arrayKeys[i];
+ ScanKey skey = &so->arrayKeyData[curArrayKey->scan_key];
+ int mark_elem = curArrayKey->mark_elem;
+
+ if (curArrayKey->cur_elem != mark_elem)
+ {
+ curArrayKey->cur_elem = mark_elem;
+ skey->sk_argument = curArrayKey->elem_values[mark_elem];
+ changed = true;
+ }
+ }
+
+ /*
+ * If we changed any keys, we must redo _bt_preprocess_keys. That might
+ * sound like overkill, but in cases with multiple keys per index column
+ * it seems necessary to do the full set of pushups.
+ */
+ if (changed)
+ {
+ _bt_preprocess_keys(scan);
+ /* The mark should have been set on a consistent set of keys... */
+ Assert(so->qual_ok);
+ }
+}
+
+
/*
* _bt_preprocess_keys() -- Preprocess scan keys
*
- * The caller-supplied search-type keys (in scan->keyData[]) are copied to
- * so->keyData[] with possible transformation. scan->numberOfKeys is
- * the number of input keys, so->numberOfKeys gets the number of output
- * keys (possibly less, never greater).
+ * The given search-type keys (in scan->keyData[] or so->arrayKeyData[])
+ * are copied to so->keyData[] with possible transformation.
+ * scan->numberOfKeys is the number of input keys, so->numberOfKeys gets
+ * the number of output keys (possibly less, never greater).
*
* The output keys are marked with additional sk_flag bits beyond the
* system-standard bits supplied by the caller. The DESC and NULLS_FIRST
* Also, for a DESC column, we commute (flip) all the sk_strategy numbers
* so that the index sorts in the desired direction.
*
- * One key purpose of this routine is to discover how many scan keys
- * must be satisfied to continue the scan. It also attempts to eliminate
- * redundant keys and detect contradictory keys. (If the index opfamily
- * provides incomplete sets of cross-type operators, we may fail to detect
- * redundant or contradictory keys, but we can survive that.)
+ * One key purpose of this routine is to discover which scan keys must be
+ * satisfied to continue the scan. It also attempts to eliminate redundant
+ * keys and detect contradictory keys. (If the index opfamily provides
+ * incomplete sets of cross-type operators, we may fail to detect redundant
+ * or contradictory keys, but we can survive that.)
*
* The output keys must be sorted by index attribute. Presently we expect
* (but verify) that the input keys are already so sorted --- this is done
- * by group_clauses_by_indexkey() in indxpath.c. Some reordering of the keys
+ * by match_clauses_to_index() in indxpath.c. Some reordering of the keys
* within each attribute may be done as a byproduct of the processing here,
* but no other code depends on that.
*
* </<= keys if we can't compare them. The logic about required keys still
* works if we don't eliminate redundant keys.
*
+ * Note that one reason we need direction-sensitive required-key flags is
+ * precisely that we may not be able to eliminate redundant keys. Suppose
+ * we have "x > 4::int AND x > 10::bigint", and we are unable to determine
+ * which key is more restrictive for lack of a suitable cross-type operator.
+ * _bt_first will arbitrarily pick one of the keys to do the initial
+ * positioning with. If it picks x > 4, then the x > 10 condition will fail
+ * until we reach index entries > 10; but we can't stop the scan just because
+ * x > 10 is failing. On the other hand, if we are scanning backwards, then
+ * failure of either key is indeed enough to stop the scan. (In general, when
+ * inequality keys are present, the initial-positioning code only promises to
+ * position before the first possible match, not exactly at the first match,
+ * for a forward scan; or after the last match for a backward scan.)
+ *
* As a byproduct of this work, we can detect contradictory quals such
- * as "x = 1 AND x > 2". If we see that, we return so->qual_ok = FALSE,
+ * as "x = 1 AND x > 2". If we see that, we return so->qual_ok = false,
* indicating the scan need not be run at all since no tuples can match.
* (In this case we do not bother completing the output key array!)
* Again, missing cross-type operators might cause us to fail to prove the
*
* Note: the reason we have to copy the preprocessed scan keys into private
* storage is that we are modifying the array based on comparisons of the
- * key argument values, which could change on a rescan. Therefore we can't
- * overwrite the caller's data structure.
+ * key argument values, which could change on a rescan or after moving to
+ * new elements of array keys. Therefore we can't overwrite the source data.
*/
void
_bt_preprocess_keys(IndexScanDesc scan)
if (numberOfKeys < 1)
return; /* done if qual-less scan */
- inkeys = scan->keyData;
+ /*
+ * Read so->arrayKeyData if array keys are present, else scan->keyData
+ */
+ if (so->arrayKeyData != NULL)
+ inkeys = so->arrayKeyData;
+ else
+ inkeys = scan->keyData;
+
outkeys = so->keyData;
cur = &inkeys[0];
/* we check that input keys are correctly ordered */
/* We can short-circuit most of the work if there's just one key */
if (numberOfKeys == 1)
{
- /*
- * We don't use indices for 'A is null' and 'A is not null' currently
- * and 'A < = > <> NULL' will always fail - so qual is not OK if
- * comparison value is NULL. - vadim 03/21/97
- */
- if (cur->sk_flags & SK_ISNULL)
+ /* Apply indoption to scankey (might change sk_strategy!) */
+ if (!_bt_fix_scankey_strategy(cur, indoption))
so->qual_ok = false;
- _bt_mark_scankey_with_indoption(cur, indoption);
memcpy(outkeys, cur, sizeof(ScanKeyData));
so->numberOfKeys = 1;
/* We can mark the qual as required if it's for first index col */
/*
* Initialize for processing of keys for attr 1.
*
- * xform[i] points to the currently best scan key of strategy type i+1;
- * it is NULL if we haven't yet found such a key for this attr.
+ * xform[i] points to the currently best scan key of strategy type i+1; it
+ * is NULL if we haven't yet found such a key for this attr.
*/
attno = 1;
memset(xform, 0, sizeof(xform));
{
if (i < numberOfKeys)
{
- /* See comments above: any NULL implies cannot match qual */
- /* Note: we assume SK_ISNULL is never set in a row header key */
- if (cur->sk_flags & SK_ISNULL)
+ /* Apply indoption to scankey (might change sk_strategy!) */
+ if (!_bt_fix_scankey_strategy(cur, indoption))
{
+ /* NULL can't be matched, so give up */
so->qual_ok = false;
-
- /*
- * Quit processing so we don't try to invoke comparison
- * routines on NULLs.
- */
return;
}
}
/*
* If = has been specified, all other keys can be eliminated as
- * redundant. In case of key > 2 && key == 1 we can set qual_ok
- * to false and abandon further processing.
+ * redundant. If we have a case like key = 1 AND key > 2, we can
+ * set qual_ok to false and abandon further processing.
+ *
+ * We also have to deal with the case of "key IS NULL", which is
+ * unsatisfiable in combination with any other index condition. By
+ * the time we get here, that's been classified as an equality
+ * check, and we've rejected any combination of it with a regular
+ * equality condition; but not with other types of conditions.
*/
if (xform[BTEqualStrategyNumber - 1])
{
if (!chk || j == (BTEqualStrategyNumber - 1))
continue;
+
+ if (eq->sk_flags & SK_SEARCHNULL)
+ {
+ /* IS NULL is contradictory to anything else */
+ so->qual_ok = false;
+ return;
+ }
+
if (_bt_compare_scankey_args(scan, chk, eq, chk,
&test_result))
{
/*
* Emit the cleaned-up keys into the outkeys[] array, and then
- * mark them if they are required. They are required (possibly
+ * mark them if they are required. They are required (possibly
* only in one direction) if all attrs before this one had "=".
*/
for (j = BTMaxStrategyNumber; --j >= 0;)
memset(xform, 0, sizeof(xform));
}
- /* apply indoption to scankey (might change sk_strategy!) */
- _bt_mark_scankey_with_indoption(cur, indoption);
-
/* check strategy this key's operator corresponds to */
j = cur->sk_strategy - 1;
memcpy(outkey, cur, sizeof(ScanKeyData));
if (numberOfEqualCols == attno - 1)
_bt_mark_scankey_required(outkey);
+
/*
* We don't support RowCompare using equality; such a qual would
* mess up the numberOfEqualCols tracking.
else
{
/*
- * We can't determine which key is more restrictive. Keep
- * the previous one in xform[j] and push this one directly
- * to the output array.
+ * We can't determine which key is more restrictive. Keep the
+ * previous one in xform[j] and push this one directly to the
+ * output array.
*/
ScanKey outkey = &outkeys[new_numberOfKeys++];
}
/*
- * Compare two scankey values using a specified operator. Both values
- * must be already known non-NULL.
+ * Compare two scankey values using a specified operator.
*
* The test we want to perform is logically "leftarg op rightarg", where
* leftarg and rightarg are the sk_argument values in those ScanKeys, and
*
* If the opfamily doesn't supply a complete set of cross-type operators we
* may not be able to make the comparison. If we can make the comparison
- * we store the operator result in *result and return TRUE. We return FALSE
+ * we store the operator result in *result and return true. We return false
* if the comparison could not be made.
*
* Note: op always points at the same ScanKey as either leftarg or rightarg.
*
* Note: this routine needs to be insensitive to any DESC option applied
* to the index column. For example, "x < 4" is a tighter constraint than
- * "x < 5" regardless of which way the index is sorted. We don't worry about
- * NULLS FIRST/LAST either, since the given values are never nulls.
+ * "x < 5" regardless of which way the index is sorted.
*/
static bool
_bt_compare_scankey_args(IndexScanDesc scan, ScanKey op,
cmp_op;
StrategyNumber strat;
+ /*
+ * First, deal with cases where one or both args are NULL. This should
+ * only happen when the scankeys represent IS NULL/NOT NULL conditions.
+ */
+ if ((leftarg->sk_flags | rightarg->sk_flags) & SK_ISNULL)
+ {
+ bool leftnull,
+ rightnull;
+
+ if (leftarg->sk_flags & SK_ISNULL)
+ {
+ Assert(leftarg->sk_flags & (SK_SEARCHNULL | SK_SEARCHNOTNULL));
+ leftnull = true;
+ }
+ else
+ leftnull = false;
+ if (rightarg->sk_flags & SK_ISNULL)
+ {
+ Assert(rightarg->sk_flags & (SK_SEARCHNULL | SK_SEARCHNOTNULL));
+ rightnull = true;
+ }
+ else
+ rightnull = false;
+
+ /*
+ * We treat NULL as either greater than or less than all other values.
+ * Since true > false, the tests below work correctly for NULLS LAST
+ * logic. If the index is NULLS FIRST, we need to flip the strategy.
+ */
+ strat = op->sk_strategy;
+ if (op->sk_flags & SK_BT_NULLS_FIRST)
+ strat = BTCommuteStrategyNumber(strat);
+
+ switch (strat)
+ {
+ case BTLessStrategyNumber:
+ *result = (leftnull < rightnull);
+ break;
+ case BTLessEqualStrategyNumber:
+ *result = (leftnull <= rightnull);
+ break;
+ case BTEqualStrategyNumber:
+ *result = (leftnull == rightnull);
+ break;
+ case BTGreaterEqualStrategyNumber:
+ *result = (leftnull >= rightnull);
+ break;
+ case BTGreaterStrategyNumber:
+ *result = (leftnull > rightnull);
+ break;
+ default:
+ elog(ERROR, "unrecognized StrategyNumber: %d", (int) strat);
+ *result = false; /* keep compiler quiet */
+ break;
+ }
+ return true;
+ }
+
/*
* The opfamily we need to worry about is identified by the index column.
*/
*/
if (lefttype == opcintype && righttype == optype)
{
- *result = DatumGetBool(FunctionCall2(&op->sk_func,
- leftarg->sk_argument,
- rightarg->sk_argument));
+ *result = DatumGetBool(FunctionCall2Coll(&op->sk_func,
+ op->sk_collation,
+ leftarg->sk_argument,
+ rightarg->sk_argument));
return true;
}
* indexscan initiated by syscache lookup will use cross-data-type
* operators.)
*
- * If the sk_strategy was flipped by _bt_mark_scankey_with_indoption,
- * we have to un-flip it to get the correct opfamily member.
+ * If the sk_strategy was flipped by _bt_fix_scankey_strategy, we have to
+ * un-flip it to get the correct opfamily member.
*/
strat = op->sk_strategy;
if (op->sk_flags & SK_BT_DESC)
if (RegProcedureIsValid(cmp_proc))
{
- *result = DatumGetBool(OidFunctionCall2(cmp_proc,
- leftarg->sk_argument,
- rightarg->sk_argument));
+ *result = DatumGetBool(OidFunctionCall2Coll(cmp_proc,
+ op->sk_collation,
+ leftarg->sk_argument,
+ rightarg->sk_argument));
return true;
}
}
}
/*
- * Mark a scankey with info from the index's indoption array.
+ * Adjust a scankey's strategy and flags setting as needed for indoptions.
*
* We copy the appropriate indoption value into the scankey sk_flags
* (shifting to avoid clobbering system-defined flag bits). Also, if
* the DESC option is set, commute (flip) the operator strategy number.
*
+ * A secondary purpose is to check for IS NULL/NOT NULL scankeys and set up
+ * the strategy field correctly for them.
+ *
+ * Lastly, for ordinary scankeys (not IS NULL/NOT NULL), we check for a
+ * NULL comparison value. Since all btree operators are assumed strict,
+ * a NULL means that the qual cannot be satisfied. We return true if the
+ * comparison value isn't NULL, or false if the scan should be abandoned.
+ *
* This function is applied to the *input* scankey structure; therefore
* on a rescan we will be looking at already-processed scankeys. Hence
* we have to be careful not to re-commute the strategy if we already did it.
* there shouldn't be any problem, since the index's indoptions are certainly
* not going to change while the scankey survives.
*/
-static void
-_bt_mark_scankey_with_indoption(ScanKey skey, int16 *indoption)
+static bool
+_bt_fix_scankey_strategy(ScanKey skey, int16 *indoption)
{
- int addflags;
+ int addflags;
addflags = indoption[skey->sk_attno - 1] << SK_BT_INDOPTION_SHIFT;
+
+ /*
+ * We treat all btree operators as strict (even if they're not so marked
+ * in pg_proc). This means that it is impossible for an operator condition
+ * with a NULL comparison constant to succeed, and we can reject it right
+ * away.
+ *
+ * However, we now also support "x IS NULL" clauses as search conditions,
+ * so in that case keep going. The planner has not filled in any
+ * particular strategy in this case, so set it to BTEqualStrategyNumber
+ * --- we can treat IS NULL as an equality operator for purposes of search
+ * strategy.
+ *
+ * Likewise, "x IS NOT NULL" is supported. We treat that as either "less
+ * than NULL" in a NULLS LAST index, or "greater than NULL" in a NULLS
+ * FIRST index.
+ *
+ * Note: someday we might have to fill in sk_collation from the index
+ * column's collation. At the moment this is a non-issue because we'll
+ * never actually call the comparison operator on a NULL.
+ */
+ if (skey->sk_flags & SK_ISNULL)
+ {
+ /* SK_ISNULL shouldn't be set in a row header scankey */
+ Assert(!(skey->sk_flags & SK_ROW_HEADER));
+
+ /* Set indoption flags in scankey (might be done already) */
+ skey->sk_flags |= addflags;
+
+ /* Set correct strategy for IS NULL or NOT NULL search */
+ if (skey->sk_flags & SK_SEARCHNULL)
+ {
+ skey->sk_strategy = BTEqualStrategyNumber;
+ skey->sk_subtype = InvalidOid;
+ skey->sk_collation = InvalidOid;
+ }
+ else if (skey->sk_flags & SK_SEARCHNOTNULL)
+ {
+ if (skey->sk_flags & SK_BT_NULLS_FIRST)
+ skey->sk_strategy = BTGreaterStrategyNumber;
+ else
+ skey->sk_strategy = BTLessStrategyNumber;
+ skey->sk_subtype = InvalidOid;
+ skey->sk_collation = InvalidOid;
+ }
+ else
+ {
+ /* regular qual, so it cannot be satisfied */
+ return false;
+ }
+
+ /* Needn't do the rest */
+ return true;
+ }
+
+ /* Adjust strategy for DESC, if we didn't already */
if ((addflags & SK_BT_DESC) && !(skey->sk_flags & SK_BT_DESC))
skey->sk_strategy = BTCommuteStrategyNumber(skey->sk_strategy);
skey->sk_flags |= addflags;
+ /* If it's a row header, fix row member flags and strategies similarly */
if (skey->sk_flags & SK_ROW_HEADER)
{
ScanKey subkey = (ScanKey) DatumGetPointer(skey->sk_argument);
subkey++;
}
}
+
+ return true;
}
/*
* Mark a scankey as "required to continue the scan".
*
* Depending on the operator type, the key may be required for both scan
- * directions or just one. Also, if the key is a row comparison header,
- * we have to mark the appropriate subsidiary ScanKeys as required. In
- * such cases, the first subsidiary key is required, but subsequent ones
- * are required only as long as they correspond to successive index columns
- * and match the leading column as to sort direction.
- * Otherwise the row comparison ordering is different from the index ordering
- * and so we can't stop the scan on the basis of those lower-order columns.
+ * directions or just one. Also, if the key is a row comparison header,
+ * we have to mark its first subsidiary ScanKey as required. (Subsequent
+ * subsidiary ScanKeys are normally for lower-order columns, and thus
+ * cannot be required, since they're after the first non-equality scankey.)
*
* Note: when we set required-key flag bits in a subsidiary scankey, we are
* scribbling on a data structure belonging to the index AM's caller, not on
* our private copy. This should be OK because the marking will not change
* from scan to scan within a query, and so we'd just re-mark the same way
- * anyway on a rescan. Something to keep an eye on though.
+ * anyway on a rescan. Something to keep an eye on though.
*/
static void
_bt_mark_scankey_required(ScanKey skey)
if (skey->sk_flags & SK_ROW_HEADER)
{
ScanKey subkey = (ScanKey) DatumGetPointer(skey->sk_argument);
- AttrNumber attno = skey->sk_attno;
- /* First subkey should be same as the header says */
- Assert(subkey->sk_attno == attno);
-
- for (;;)
- {
- Assert(subkey->sk_flags & SK_ROW_MEMBER);
- if (subkey->sk_attno != attno)
- break; /* non-adjacent key, so not required */
- if (subkey->sk_strategy != skey->sk_strategy)
- break; /* wrong direction, so not required */
- subkey->sk_flags |= addflags;
- if (subkey->sk_flags & SK_ROW_END)
- break;
- subkey++;
- attno++;
- }
+ /* First subkey should be same column/operator as the header */
+ Assert(subkey->sk_flags & SK_ROW_MEMBER);
+ Assert(subkey->sk_attno == skey->sk_attno);
+ Assert(subkey->sk_strategy == skey->sk_strategy);
+ subkey->sk_flags |= addflags;
}
}
/*
* Test whether an indextuple satisfies all the scankey conditions.
*
- * If so, copy its TID into scan->xs_ctup.t_self, and return TRUE.
- * If not, return FALSE (xs_ctup is not changed).
+ * If so, return the address of the index tuple on the index page.
+ * If not, return NULL.
*
* If the tuple fails to pass the qual, we also determine whether there's
* any need to continue the scan beyond this tuple, and set *continuescan
* offnum: offset number of index tuple (must be a valid item!)
* dir: direction we are scanning in
* continuescan: output parameter (will be set correctly in all cases)
+ *
+ * Caller must hold pin and lock on the index page.
*/
-bool
+IndexTuple
_bt_checkkeys(IndexScanDesc scan,
Page page, OffsetNumber offnum,
ScanDirection dir, bool *continuescan)
{
ItemId iid = PageGetItemId(page, offnum);
- bool tuple_valid;
+ bool tuple_alive;
IndexTuple tuple;
TupleDesc tupdesc;
BTScanOpaque so;
* However, if this is the last tuple on the page, we should check the
* index keys to prevent uselessly advancing to the next page.
*/
- if (scan->ignore_killed_tuples && ItemIdDeleted(iid))
+ if (scan->ignore_killed_tuples && ItemIdIsDead(iid))
{
/* return immediately if there are more tuples on the page */
if (ScanDirectionIsForward(dir))
{
if (offnum < PageGetMaxOffsetNumber(page))
- return false;
+ return NULL;
}
else
{
BTPageOpaque opaque = (BTPageOpaque) PageGetSpecialPointer(page);
if (offnum > P_FIRSTDATAKEY(opaque))
- return false;
+ return NULL;
}
/*
- * OK, we want to check the keys, but we'll return FALSE even if the
- * tuple passes the key tests.
+ * OK, we want to check the keys so we can set continuescan correctly,
+ * but we'll return NULL even if the tuple passes the key tests.
*/
- tuple_valid = false;
+ tuple_alive = false;
}
else
- tuple_valid = true;
+ tuple_alive = true;
tuple = (IndexTuple) PageGetItem(page, iid);
- IncrIndexProcessed();
-
tupdesc = RelationGetDescr(scan->indexRelation);
so = (BTScanOpaque) scan->opaque;
keysz = so->numberOfKeys;
bool isNull;
Datum test;
+ Assert(key->sk_attno <= BTreeTupleGetNAtts(tuple, scan->indexRelation));
/* row-comparison keys need special processing */
if (key->sk_flags & SK_ROW_HEADER)
{
if (_bt_check_rowcompare(key, tuple, tupdesc, dir, continuescan))
continue;
- return false;
+ return NULL;
}
datum = index_getattr(tuple,
tupdesc,
&isNull);
- /* btree doesn't support 'A is null' clauses, yet */
if (key->sk_flags & SK_ISNULL)
{
- /* we shouldn't get here, really; see _bt_preprocess_keys() */
- *continuescan = false;
- return false;
+ /* Handle IS NULL/NOT NULL tests */
+ if (key->sk_flags & SK_SEARCHNULL)
+ {
+ if (isNull)
+ continue; /* tuple satisfies this qual */
+ }
+ else
+ {
+ Assert(key->sk_flags & SK_SEARCHNOTNULL);
+ if (!isNull)
+ continue; /* tuple satisfies this qual */
+ }
+
+ /*
+ * Tuple fails this qual. If it's a required qual for the current
+ * scan direction, then we can conclude no further tuples will
+ * pass, either.
+ */
+ if ((key->sk_flags & SK_BT_REQFWD) &&
+ ScanDirectionIsForward(dir))
+ *continuescan = false;
+ else if ((key->sk_flags & SK_BT_REQBKWD) &&
+ ScanDirectionIsBackward(dir))
+ *continuescan = false;
+
+ /*
+ * In any case, this indextuple doesn't match the qual.
+ */
+ return NULL;
}
if (isNull)
/*
* Since NULLs are sorted before non-NULLs, we know we have
* reached the lower limit of the range of values for this
- * index attr. On a backward scan, we can stop if this qual is
- * one of the "must match" subset. On a forward scan,
- * however, we should keep going.
+ * index attr. On a backward scan, we can stop if this qual
+ * is one of the "must match" subset. We can stop regardless
+ * of whether the qual is > or <, so long as it's required,
+ * because it's not possible for any future tuples to pass. On
+ * a forward scan, however, we must keep going, because we may
+ * have initially positioned to the start of the index.
*/
- if ((key->sk_flags & SK_BT_REQBKWD) &&
+ if ((key->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD)) &&
ScanDirectionIsBackward(dir))
*continuescan = false;
}
* Since NULLs are sorted after non-NULLs, we know we have
* reached the upper limit of the range of values for this
* index attr. On a forward scan, we can stop if this qual is
- * one of the "must match" subset. On a backward scan,
- * however, we should keep going.
+ * one of the "must match" subset. We can stop regardless of
+ * whether the qual is > or <, so long as it's required,
+ * because it's not possible for any future tuples to pass. On
+ * a backward scan, however, we must keep going, because we
+ * may have initially positioned to the end of the index.
*/
- if ((key->sk_flags & SK_BT_REQFWD) &&
+ if ((key->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD)) &&
ScanDirectionIsForward(dir))
*continuescan = false;
}
/*
* In any case, this indextuple doesn't match the qual.
*/
- return false;
+ return NULL;
}
- test = FunctionCall2(&key->sk_func, datum, key->sk_argument);
+ test = FunctionCall2Coll(&key->sk_func, key->sk_collation,
+ datum, key->sk_argument);
if (!DatumGetBool(test))
{
/*
* In any case, this indextuple doesn't match the qual.
*/
- return false;
+ return NULL;
}
}
- /* If we get here, the tuple passes all index quals. */
- if (tuple_valid)
- scan->xs_ctup.t_self = tuple->t_tid;
+ /* Check for failure due to it being a killed tuple. */
+ if (!tuple_alive)
+ return NULL;
- return tuple_valid;
+ /* If we get here, the tuple passes all index quals. */
+ return tuple;
}
/*
/*
* Since NULLs are sorted before non-NULLs, we know we have
* reached the lower limit of the range of values for this
- * index attr. On a backward scan, we can stop if this qual is
- * one of the "must match" subset. On a forward scan,
- * however, we should keep going.
+ * index attr. On a backward scan, we can stop if this qual
+ * is one of the "must match" subset. We can stop regardless
+ * of whether the qual is > or <, so long as it's required,
+ * because it's not possible for any future tuples to pass. On
+ * a forward scan, however, we must keep going, because we may
+ * have initially positioned to the start of the index.
*/
- if ((subkey->sk_flags & SK_BT_REQBKWD) &&
+ if ((subkey->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD)) &&
ScanDirectionIsBackward(dir))
*continuescan = false;
}
/*
* Since NULLs are sorted after non-NULLs, we know we have
* reached the upper limit of the range of values for this
- * index attr. On a forward scan, we can stop if this qual is
- * one of the "must match" subset. On a backward scan,
- * however, we should keep going.
+ * index attr. On a forward scan, we can stop if this qual is
+ * one of the "must match" subset. We can stop regardless of
+ * whether the qual is > or <, so long as it's required,
+ * because it's not possible for any future tuples to pass. On
+ * a backward scan, however, we must keep going, because we
+ * may have initially positioned to the end of the index.
*/
- if ((subkey->sk_flags & SK_BT_REQFWD) &&
+ if ((subkey->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD)) &&
ScanDirectionIsForward(dir))
*continuescan = false;
}
{
/*
* Unlike the simple-scankey case, this isn't a disallowed case.
- * But it can never match. If all the earlier row comparison
+ * But it can never match. If all the earlier row comparison
* columns are required for the scan direction, we can stop the
* scan, because there can't be another tuple that will succeed.
*/
}
/* Perform the test --- three-way comparison not bool operator */
- cmpresult = DatumGetInt32(FunctionCall2(&subkey->sk_func,
- datum,
- subkey->sk_argument));
+ cmpresult = DatumGetInt32(FunctionCall2Coll(&subkey->sk_func,
+ subkey->sk_collation,
+ datum,
+ subkey->sk_argument));
if (subkey->sk_flags & SK_BT_DESC)
cmpresult = -cmpresult;
/*
* Tuple fails this qual. If it's a required qual for the current
* scan direction, then we can conclude no further tuples will pass,
- * either. Note we have to look at the deciding column, not
+ * either. Note we have to look at the deciding column, not
* necessarily the first or last column of the row condition.
*/
if ((subkey->sk_flags & SK_BT_REQFWD) &&
}
/*
- * _bt_killitems - set LP_DELETE bit for items an indexscan caller has
+ * _bt_killitems - set LP_DEAD state for items an indexscan caller has
* told us were killed
*
- * scan->so contains information about the current page and killed tuples
- * thereon (generally, this should only be called if so->numKilled > 0).
+ * scan->opaque, referenced locally through so, contains information about the
+ * current page and killed tuples thereon (generally, this should only be
+ * called if so->numKilled > 0).
*
- * The caller must have pin on so->currPos.buf, but may or may not have
- * read-lock, as indicated by haveLock. Note that we assume read-lock
- * is sufficient for setting LP_DELETE hint bits.
+ * The caller does not have a lock on the page and may or may not have the
+ * page pinned in a buffer. Note that read-lock is sufficient for setting
+ * LP_DEAD status (which is only a hint).
*
* We match items by heap TID before assuming they are the right ones to
- * delete. We cope with cases where items have moved right due to insertions.
+ * delete. We cope with cases where items have moved right due to insertions.
* If an item has moved off the current page due to a split, we'll fail to
* find it and do nothing (this is not an error case --- we assume the item
- * will eventually get marked in a future indexscan). Note that because we
- * hold pin on the target page continuously from initially reading the items
- * until applying this function, VACUUM cannot have deleted any items from
- * the page, and so there is no need to search left from the recorded offset.
- * (This observation also guarantees that the item is still the right one
- * to delete, which might otherwise be questionable since heap TIDs can get
- * recycled.)
+ * will eventually get marked in a future indexscan).
+ *
+ * Note that if we hold a pin on the target page continuously from initially
+ * reading the items until applying this function, VACUUM cannot have deleted
+ * any items from the page, and so there is no need to search left from the
+ * recorded offset. (This observation also guarantees that the item is still
+ * the right one to delete, which might otherwise be questionable since heap
+ * TIDs can get recycled.) This holds true even if the page has been modified
+ * by inserts and page splits, so there is no need to consult the LSN.
+ *
+ * If the pin was released after reading the page, then we re-read it. If it
+ * has been modified since we read it (as determined by the LSN), we dare not
+ * flag any entries because it is possible that the old entry was vacuumed
+ * away and the TID was re-used by a completely different heap tuple.
*/
void
-_bt_killitems(IndexScanDesc scan, bool haveLock)
+_bt_killitems(IndexScanDesc scan)
{
BTScanOpaque so = (BTScanOpaque) scan->opaque;
Page page;
OffsetNumber minoff;
OffsetNumber maxoff;
int i;
+ int numKilled = so->numKilled;
bool killedsomething = false;
- Assert(BufferIsValid(so->currPos.buf));
+ Assert(BTScanPosIsValid(so->currPos));
+
+ /*
+ * Always reset the scan state, so we don't look for same items on other
+ * pages.
+ */
+ so->numKilled = 0;
- if (!haveLock)
+ if (BTScanPosIsPinned(so->currPos))
+ {
+ /*
+ * We have held the pin on this page since we read the index tuples,
+ * so all we need to do is lock it. The pin will have prevented
+ * re-use of any TID on the page, so there is no need to check the
+ * LSN.
+ */
LockBuffer(so->currPos.buf, BT_READ);
- page = BufferGetPage(so->currPos.buf);
+ page = BufferGetPage(so->currPos.buf);
+ }
+ else
+ {
+ Buffer buf;
+
+ /* Attempt to re-read the buffer, getting pin and lock. */
+ buf = _bt_getbuf(scan->indexRelation, so->currPos.currPage, BT_READ);
+
+ /* It might not exist anymore; in which case we can't hint it. */
+ if (!BufferIsValid(buf))
+ return;
+
+ page = BufferGetPage(buf);
+ if (BufferGetLSNAtomic(buf) == so->currPos.lsn)
+ so->currPos.buf = buf;
+ else
+ {
+ /* Modified while not pinned means hinting is not safe. */
+ _bt_relbuf(scan->indexRelation, buf);
+ return;
+ }
+ }
+
opaque = (BTPageOpaque) PageGetSpecialPointer(page);
minoff = P_FIRSTDATAKEY(opaque);
maxoff = PageGetMaxOffsetNumber(page);
- for (i = 0; i < so->numKilled; i++)
+ for (i = 0; i < numKilled; i++)
{
int itemIndex = so->killedItems[i];
BTScanPosItem *kitem = &so->currPos.items[itemIndex];
if (ItemPointerEquals(&ituple->t_tid, &kitem->heapTid))
{
/* found the item */
- iid->lp_flags |= LP_DELETE;
+ ItemIdMarkDead(iid);
killedsomething = true;
break; /* out of inner search loop */
}
}
/*
- * Since this can be redone later if needed, it's treated the same as a
- * commit-hint-bit status update for heap tuples: we mark the buffer dirty
- * but don't make a WAL log entry.
+ * Since this can be redone later if needed, mark as dirty hint.
*
- * Whenever we mark anything LP_DELETEd, we also set the page's
+ * Whenever we mark anything LP_DEAD, we also set the page's
* BTP_HAS_GARBAGE flag, which is likewise just a hint.
*/
if (killedsomething)
{
opaque->btpo_flags |= BTP_HAS_GARBAGE;
- SetBufferCommitInfoNeedsSave(so->currPos.buf);
+ MarkBufferDirtyHint(so->currPos.buf, true);
}
- if (!haveLock)
- LockBuffer(so->currPos.buf, BUFFER_LOCK_UNLOCK);
-
- /*
- * Always reset the scan state, so we don't look for same items on other
- * pages.
- */
- so->numKilled = 0;
+ LockBuffer(so->currPos.buf, BUFFER_LOCK_UNLOCK);
}
/*
* The following routines manage a shared-memory area in which we track
* assignment of "vacuum cycle IDs" to currently-active btree vacuuming
- * operations. There is a single counter which increments each time we
- * start a vacuum to assign it a cycle ID. Since multiple vacuums could
+ * operations. There is a single counter which increments each time we
+ * start a vacuum to assign it a cycle ID. Since multiple vacuums could
* be active concurrently, we have to track the cycle ID for each active
* vacuum; this requires at most MaxBackends entries (usually far fewer).
* We assume at most one vacuum can be active for a given index.
BTCycleId cycle_ctr; /* cycle ID most recently assigned */
int num_vacuums; /* number of currently active VACUUMs */
int max_vacuums; /* allocated length of vacuums[] array */
- BTOneVacInfo vacuums[1]; /* VARIABLE LENGTH ARRAY */
+ BTOneVacInfo vacuums[FLEXIBLE_ARRAY_MEMBER];
} BTVacInfo;
static BTVacInfo *btvacinfo;
/*
* _bt_start_vacuum --- assign a cycle ID to a just-starting VACUUM operation
*
- * Note: the caller must guarantee (via PG_TRY) that it will eventually call
- * _bt_end_vacuum, else we'll permanently leak an array slot.
+ * Note: the caller must guarantee that it will eventually call
+ * _bt_end_vacuum, else we'll permanently leak an array slot. To ensure
+ * that this happens even in elog(FATAL) scenarios, the appropriate coding
+ * is not just a PG_TRY, but
+ * PG_ENSURE_ERROR_CLEANUP(_bt_end_vacuum_callback, PointerGetDatum(rel))
*/
BTCycleId
_bt_start_vacuum(Relation rel)
LWLockAcquire(BtreeVacuumLock, LW_EXCLUSIVE);
- /* Assign the next cycle ID, being careful to avoid zero */
- do
- {
- result = ++(btvacinfo->cycle_ctr);
- } while (result == 0);
+ /*
+ * Assign the next cycle ID, being careful to avoid zero as well as the
+ * reserved high values.
+ */
+ result = ++(btvacinfo->cycle_ctr);
+ if (result == 0 || result > MAX_BT_CYCLE_ID)
+ result = btvacinfo->cycle_ctr = 1;
/* Let's just make sure there's no entry already for this index */
for (i = 0; i < btvacinfo->num_vacuums; i++)
LWLockRelease(BtreeVacuumLock);
}
+/*
+ * _bt_end_vacuum wrapped as an on_shmem_exit callback function
+ */
+void
+_bt_end_vacuum_callback(int code, Datum arg)
+{
+ _bt_end_vacuum((Relation) DatumGetPointer(arg));
+}
+
/*
* BTreeShmemSize --- report amount of shared memory space needed
*/
{
Size size;
- size = offsetof(BTVacInfo, vacuums[0]);
+ size = offsetof(BTVacInfo, vacuums);
size = add_size(size, mul_size(MaxBackends, sizeof(BTOneVacInfo)));
return size;
}
Assert(found);
}
-Datum
-btoptions(PG_FUNCTION_ARGS)
+bytea *
+btoptions(Datum reloptions, bool validate)
+{
+ return default_reloptions(reloptions, validate, RELOPT_KIND_BTREE);
+}
+
+/*
+ * btproperty() -- Check boolean properties of indexes.
+ *
+ * This is optional, but handling AMPROP_RETURNABLE here saves opening the rel
+ * to call btcanreturn.
+ */
+bool
+btproperty(Oid index_oid, int attno,
+ IndexAMProperty prop, const char *propname,
+ bool *res, bool *isnull)
{
- Datum reloptions = PG_GETARG_DATUM(0);
- bool validate = PG_GETARG_BOOL(1);
- bytea *result;
-
- result = default_reloptions(reloptions, validate,
- BTREE_MIN_FILLFACTOR,
- BTREE_DEFAULT_FILLFACTOR);
- if (result)
- PG_RETURN_BYTEA_P(result);
- PG_RETURN_NULL();
+ switch (prop)
+ {
+ case AMPROP_RETURNABLE:
+ /* answer only for columns, not AM or whole index */
+ if (attno == 0)
+ return false;
+ /* otherwise, btree can always return data */
+ *res = true;
+ return true;
+
+ default:
+ return false; /* punt to generic code */
+ }
+}
+
+/*
+ * _bt_nonkey_truncate() -- create tuple without non-key suffix attributes.
+ *
+ * Returns truncated index tuple allocated in caller's memory context, with key
+ * attributes copied from caller's itup argument. Currently, suffix truncation
+ * is only performed to create pivot tuples in INCLUDE indexes, but some day it
+ * could be generalized to remove suffix attributes after the first
+ * distinguishing key attribute.
+ *
+ * Truncated tuple is guaranteed to be no larger than the original, which is
+ * important for staying under the 1/3 of a page restriction on tuple size.
+ *
+ * Note that returned tuple's t_tid offset will hold the number of attributes
+ * present, so the original item pointer offset is not represented. Caller
+ * should only change truncated tuple's downlink.
+ */
+IndexTuple
+_bt_nonkey_truncate(Relation rel, IndexTuple itup)
+{
+ int nkeyattrs = IndexRelationGetNumberOfKeyAttributes(rel);
+ IndexTuple truncated;
+
+ /*
+ * We should only ever truncate leaf index tuples, which must have both key
+ * and non-key attributes. It's never okay to truncate a second time.
+ */
+ Assert(BTreeTupleGetNAtts(itup, rel) ==
+ IndexRelationGetNumberOfAttributes(rel));
+
+ truncated = index_truncate_tuple(RelationGetDescr(rel), itup, nkeyattrs);
+ BTreeTupleSetNAtts(truncated, nkeyattrs);
+
+ return truncated;
+}
+
+/*
+ * _bt_check_natts() -- Verify tuple has expected number of attributes.
+ *
+ * Returns value indicating if the expected number of attributes were found
+ * for a particular offset on page. This can be used as a general purpose
+ * sanity check.
+ *
+ * Testing a tuple directly with BTreeTupleGetNAtts() should generally be
+ * preferred to calling here. That's usually more convenient, and is always
+ * more explicit. Call here instead when offnum's tuple may be a negative
+ * infinity tuple that uses the pre-v11 on-disk representation, or when a low
+ * context check is appropriate.
+ */
+bool
+_bt_check_natts(Relation rel, Page page, OffsetNumber offnum)
+{
+ int16 natts = IndexRelationGetNumberOfAttributes(rel);
+ int16 nkeyatts = IndexRelationGetNumberOfKeyAttributes(rel);
+ BTPageOpaque opaque = (BTPageOpaque) PageGetSpecialPointer(page);
+ IndexTuple itup;
+
+ /*
+ * We cannot reliably test a deleted or half-deleted page, since they have
+ * dummy high keys
+ */
+ if (P_IGNORE(opaque))
+ return true;
+
+ Assert(offnum >= FirstOffsetNumber &&
+ offnum <= PageGetMaxOffsetNumber(page));
+ /*
+ * Mask allocated for number of keys in index tuple must be able to fit
+ * maximum possible number of index attributes
+ */
+ StaticAssertStmt(BT_N_KEYS_OFFSET_MASK >= INDEX_MAX_KEYS,
+ "BT_N_KEYS_OFFSET_MASK can't fit INDEX_MAX_KEYS");
+
+ itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, offnum));
+
+ if (P_ISLEAF(opaque))
+ {
+ if (offnum >= P_FIRSTDATAKEY(opaque))
+ {
+ /*
+ * Leaf tuples that are not the page high key (non-pivot tuples)
+ * should never be truncated
+ */
+ return BTreeTupleGetNAtts(itup, rel) == natts;
+ }
+ else
+ {
+ /*
+ * Rightmost page doesn't contain a page high key, so tuple was
+ * checked above as ordinary leaf tuple
+ */
+ Assert(!P_RIGHTMOST(opaque));
+
+ /* Page high key tuple contains only key attributes */
+ return BTreeTupleGetNAtts(itup, rel) == nkeyatts;
+ }
+ }
+ else /* !P_ISLEAF(opaque) */
+ {
+ if (offnum == P_FIRSTDATAKEY(opaque))
+ {
+ /*
+ * The first tuple on any internal page (possibly the first after
+ * its high key) is its negative infinity tuple. Negative infinity
+ * tuples are always truncated to zero attributes. They are a
+ * particular kind of pivot tuple.
+ *
+ * The number of attributes won't be explicitly represented if the
+ * negative infinity tuple was generated during a page split that
+ * occurred with a version of Postgres before v11. There must be a
+ * problem when there is an explicit representation that is
+ * non-zero, or when there is no explicit representation and the
+ * tuple is evidently not a pre-pg_upgrade tuple.
+ *
+ * Prior to v11, downlinks always had P_HIKEY as their offset. Use
+ * that to decide if the tuple is a pre-v11 tuple.
+ */
+ return BTreeTupleGetNAtts(itup, rel) == 0 ||
+ ((itup->t_info & INDEX_ALT_TID_MASK) == 0 &&
+ ItemPointerGetOffsetNumber(&(itup->t_tid)) == P_HIKEY);
+ }
+ else
+ {
+ /*
+ * Tuple contains only key attributes despite on is it page high
+ * key or not
+ */
+ return BTreeTupleGetNAtts(itup, rel) == nkeyatts;
+ }
+
+ }
}
projects / postgresql / blobdiff