/*
* systable_beginscan_ordered --- set up for ordered catalog scan
*
* These routines have essentially the same API as systable_beginscan etc,
* except that they guarantee to return multiple matching tuples in
* index order. Also, for largely historical reasons, the index to use
* is opened and locked by the caller, not here.
*
* Currently we do not support non-index-based scans here. (In principle
* we could do a heapscan and sort, but the uses are in places that
* probably don't need to still work with corrupted catalog indexes.)
* For the moment, therefore, these functions are merely the thinnest of
* wrappers around index_beginscan/index_getnext. The main reason for their
* existence is to centralize possible future support of lossy operators
* in catalog scans.
*/
SysScanDesc
systable_beginscan_ordered(Relation heapRelation,
Relation indexRelation,
Snapshot snapshot,
int nkeys, ScanKey key)
{
SysScanDesc sysscan;
int i;
/* REINDEX can probably be a hard error here ... */
if (ReindexIsProcessingIndex(RelationGetRelid(indexRelation)))
elog(ERROR, "cannot do ordered scan on index \"%s\", because it is being reindexed",
RelationGetRelationName(indexRelation));
/* ... but we only throw a warning about violating IgnoreSystemIndexes */
if (IgnoreSystemIndexes)
elog(WARNING, "using index \"%s\" despite IgnoreSystemIndexes",
RelationGetRelationName(indexRelation));
sysscan = (SysScanDesc) palloc(sizeof(SysScanDescData));
sysscan->heap_rel = heapRelation;
sysscan->irel = indexRelation;
if (snapshot == NULL)
{
Oid relid = RelationGetRelid(heapRelation);
snapshot = RegisterSnapshot(GetCatalogSnapshot(relid));
sysscan->snapshot = snapshot;
}
else
{
/* Caller is responsible for any snapshot. */
sysscan->snapshot = NULL;
}
/* Change attribute numbers to be index column numbers. */
for (i = 0; i < nkeys; i++)
{
int j;
for (j = 0; j < indexRelation->rd_index->indnatts; j++)
{
if (key[i].sk_attno == indexRelation->rd_index->indkey.values[j])
{
key[i].sk_attno = j + 1;
break;
}
}
if (j == indexRelation->rd_index->indnatts)
elog(ERROR, "column is not in index");
}
sysscan->iscan = index_beginscan(heapRelation, indexRelation,
snapshot, nkeys, 0);
index_rescan(sysscan->iscan, key, nkeys, NULL, 0);
sysscan->scan = NULL;
return sysscan;
}
/*
* GetNewOidWithIndex
* Guts of GetNewOid: use the supplied index
*
* This is exported separately because there are cases where we want to use
* an index that will not be recognized by RelationGetOidIndex: TOAST tables
* and pg_largeobject have indexes that are usable, but have multiple columns
* and are on ordinary columns rather than a true OID column. This code
* will work anyway, so long as the OID is the index's first column.
*
* Caller must have a suitable lock on the relation.
*/
Oid
GetNewOidWithIndex(Relation relation, Relation indexrel)
{
Oid newOid;
SnapshotData SnapshotDirty;
IndexScanDesc scan;
ScanKeyData key;
bool collides;
InitDirtySnapshot(SnapshotDirty);
/* Generate new OIDs until we find one not in the table */
do
{
CHECK_FOR_INTERRUPTS();
newOid = GetNewObjectId();
ScanKeyInit(&key,
(AttrNumber) 1,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(newOid));
/* see notes above about using SnapshotDirty */
scan = index_beginscan(relation, indexrel,
&SnapshotDirty, 1, &key);
collides = HeapTupleIsValid(index_getnext(scan, ForwardScanDirection));
index_endscan(scan);
} while (collides);
return newOid;
}
/*
* Determine size of a large object
*
* NOTE: LOs can contain gaps, just like Unix files. We actually return
* the offset of the last byte + 1.
*/
static uint32
inv_getsize(LargeObjectDesc *obj_desc)
{
bool found = false;
uint32 lastbyte = 0;
ScanKeyData skey[1];
IndexScanDesc sd;
HeapTuple tuple;
Assert(PointerIsValid(obj_desc));
open_lo_relation();
ScanKeyInit(&skey[0],
Anum_pg_largeobject_loid,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(obj_desc->id));
sd = index_beginscan(lo_heap_r, lo_index_r,
obj_desc->snapshot, 1, skey);
/*
* Because the pg_largeobject index is on both loid and pageno, but we
* constrain only loid, a backwards scan should visit all pages of the
* large object in reverse pageno order. So, it's sufficient to examine
* the first valid tuple (== last valid page).
*/
while ((tuple = index_getnext(sd, BackwardScanDirection)) != NULL)
{
Form_pg_largeobject data;
bytea *datafield;
bool pfreeit;
found = true;
if (HeapTupleHasNulls(tuple)) /* paranoia */
elog(ERROR, "null field found in pg_largeobject");
data = (Form_pg_largeobject) GETSTRUCT(tuple);
datafield = &(data->data); /* see note at top of file */
pfreeit = false;
if (VARATT_IS_EXTENDED(datafield))
{
datafield = (bytea *)
heap_tuple_untoast_attr((struct varlena *) datafield);
pfreeit = true;
}
lastbyte = data->pageno * LOBLKSIZE + getbytealen(datafield);
if (pfreeit)
pfree(datafield);
break;
}
index_endscan(sd);
if (!found)
ereport(ERROR,
(errcode(ERRCODE_UNDEFINED_OBJECT),
errmsg("large object %u does not exist", obj_desc->id)));
return lastbyte;
}
/*
* systable_beginscan --- set up for heap-or-index scan
*
* rel: catalog to scan, already opened and suitably locked
* indexRelname: name of index to conditionally use
* indexOK: if false, forces a heap scan (see notes below)
* snapshot: time qual to use (usually should be SnapshotNow)
* nkeys, key: scan keys
*
* The attribute numbers in the scan key should be set for the heap case.
* If we choose to index, we reset them to 1..n to reference the index
* columns. Note this means there must be one scankey qualification per
* index column! This is checked by the Asserts in the normal, index-using
* case, but won't be checked if the heapscan path is taken.
*
* The routine checks the normal cases for whether an indexscan is safe,
* but caller can make additional checks and pass indexOK=false if needed.
* In standard case indexOK can simply be constant TRUE.
*/
SysScanDesc
systable_beginscan(Relation heapRelation,
const char *indexRelname,
bool indexOK,
Snapshot snapshot,
int nkeys, ScanKey key)
{
SysScanDesc sysscan;
Relation irel;
if (indexOK && !IsIgnoringSystemIndexes())
{
/* We assume it's a system index, so index_openr is OK */
irel = index_openr(indexRelname);
if (ReindexIsProcessingIndex(RelationGetRelid(irel)))
{
/* oops, can't use index that's being rebuilt */
index_close(irel);
irel = NULL;
}
}
else
irel = NULL;
sysscan = (SysScanDesc) palloc(sizeof(SysScanDescData));
sysscan->heap_rel = heapRelation;
sysscan->irel = irel;
if (irel)
{
int i;
/*
* Change attribute numbers to be index column numbers.
*
* This code could be generalized to search for the index key numbers
* to substitute, but for now there's no need.
*/
for (i = 0; i < nkeys; i++)
{
Assert(key[i].sk_attno == irel->rd_index->indkey[i]);
key[i].sk_attno = i + 1;
}
sysscan->iscan = index_beginscan(heapRelation, irel, snapshot,
nkeys, key);
sysscan->scan = NULL;
}
else
{
sysscan->scan = heap_beginscan(heapRelation, snapshot, nkeys, key);
sysscan->iscan = NULL;
}
return sysscan;
}
/*
* systable_beginscan --- set up for heap-or-index scan
*
* rel: catalog to scan, already opened and suitably locked
* indexId: OID of index to conditionally use
* indexOK: if false, forces a heap scan (see notes below)
* snapshot: time qual to use (usually should be SnapshotNow)
* nkeys, key: scan keys
*
* The attribute numbers in the scan key should be set for the heap case.
* If we choose to index, we reset them to 1..n to reference the index
* columns. Note this means there must be one scankey qualification per
* index column! This is checked by the Asserts in the normal, index-using
* case, but won't be checked if the heapscan path is taken.
*
* The routine checks the normal cases for whether an indexscan is safe,
* but caller can make additional checks and pass indexOK=false if needed.
* In standard case indexOK can simply be constant TRUE.
*/
SysScanDesc
systable_beginscan(Relation heapRelation,
Oid indexId,
bool indexOK,
Snapshot snapshot,
int nkeys, ScanKey key)
{
SysScanDesc sysscan;
Relation irel;
if (indexOK &&
!IgnoreSystemIndexes &&
!ReindexIsProcessingIndex(indexId))
irel = index_open(indexId, AccessShareLock);
else
irel = NULL;
sysscan = (SysScanDesc) palloc(sizeof(SysScanDescData));
sysscan->heap_rel = heapRelation;
sysscan->irel = irel;
if (irel)
{
int i;
/* Change attribute numbers to be index column numbers. */
for (i = 0; i < nkeys; i++)
{
int j;
for (j = 0; j < irel->rd_index->indnatts; j++)
{
if (key[i].sk_attno == irel->rd_index->indkey.values[j])
{
key[i].sk_attno = j + 1;
break;
}
}
if (j == irel->rd_index->indnatts)
elog(ERROR, "column is not in index");
}
sysscan->iscan = index_beginscan(heapRelation, irel,
snapshot, nkeys, key);
sysscan->scan = NULL;
}
else
{
sysscan->scan = heap_beginscan(heapRelation, snapshot, nkeys, key);
sysscan->iscan = NULL;
}
return sysscan;
}
/* ----------
* toast_delete_datum -
*
* Delete a single external stored value.
* ----------
*/
static void
toast_delete_datum(Relation rel __attribute__((unused)), Datum value)
{
varattrib *attr = (varattrib *) DatumGetPointer(value);
Relation toastrel;
Relation toastidx;
ScanKeyData toastkey;
IndexScanDesc toastscan;
HeapTuple toasttup;
if (!VARATT_IS_EXTERNAL(attr))
return;
/*
* Open the toast relation and its index
*/
toastrel = heap_open(attr->va_external.va_toastrelid,
RowExclusiveLock);
toastidx = index_open(toastrel->rd_rel->reltoastidxid, RowExclusiveLock);
/*
* Setup a scan key to fetch from the index by va_valueid (we don't
* particularly care whether we see them in sequence or not)
*/
ScanKeyInit(&toastkey,
(AttrNumber) 1,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(attr->va_external.va_valueid));
/*
* Find all the chunks. (We don't actually care whether we see them in
* sequence or not, but since we've already locked the index we might as
* well use systable_beginscan_ordered.)
*/
toastscan = index_beginscan(toastrel, toastidx,
SnapshotToast, 1, &toastkey);
while ((toasttup = index_getnext(toastscan, ForwardScanDirection)) != NULL)
{
/*
* Have a chunk, delete it
*/
simple_heap_delete(toastrel, &toasttup->t_self);
}
/*
* End scan and close relations
*/
index_endscan(toastscan);
index_close(toastidx, RowExclusiveLock);
heap_close(toastrel, RowExclusiveLock);
}
/* ----------
* toast_delete_datum -
*
* Delete a single external stored value.
* ----------
*/
static void
toast_delete_datum(Relation rel, Datum value)
{
varattrib *attr = (varattrib *) DatumGetPointer(value);
Relation toastrel;
Relation toastidx;
ScanKeyData toastkey;
IndexScanDesc toastscan;
HeapTuple toasttup;
if (!VARATT_IS_EXTERNAL(attr))
return;
/*
* Open the toast relation and it's index
*/
toastrel = heap_open(attr->va_content.va_external.va_toastrelid,
RowExclusiveLock);
toastidx = index_open(toastrel->rd_rel->reltoastidxid);
/*
* Setup a scan key to fetch from the index by va_valueid (we don't
* particularly care whether we see them in sequence or not)
*/
ScanKeyEntryInitialize(&toastkey,
(bits16) 0,
(AttrNumber) 1,
(RegProcedure) F_OIDEQ,
ObjectIdGetDatum(attr->va_content.va_external.va_valueid));
/*
* Find the chunks by index
*/
toastscan = index_beginscan(toastrel, toastidx, SnapshotToast,
1, &toastkey);
while ((toasttup = index_getnext(toastscan, ForwardScanDirection)) != NULL)
{
/*
* Have a chunk, delete it
*/
simple_heap_delete(toastrel, &toasttup->t_self);
}
/*
* End scan and close relations
*/
index_endscan(toastscan);
index_close(toastidx);
heap_close(toastrel, RowExclusiveLock);
}
/*
* systable_beginscan --- set up for heap-or-index scan
*
* rel: catalog to scan, already opened and suitably locked
* indexId: OID of index to conditionally use
* indexOK: if false, forces a heap scan (see notes below)
* snapshot: time qual to use (usually should be SnapshotNow)
* nkeys, key: scan keys
*
* The attribute numbers in the scan key should be set for the heap case.
* If we choose to index, we reset them to 1..n to reference the index
* columns. Note this means there must be one scankey qualification per
* index column! This is checked by the Asserts in the normal, index-using
* case, but won't be checked if the heapscan path is taken.
*
* The routine checks the normal cases for whether an indexscan is safe,
* but caller can make additional checks and pass indexOK=false if needed.
* In standard case indexOK can simply be constant TRUE.
*/
SysScanDesc
systable_beginscan(Relation heapRelation,
Oid indexId,
bool indexOK,
Snapshot snapshot,
int nkeys, ScanKey key)
{
SysScanDesc sysscan;
Relation irel;
if (indexOK &&
!IgnoreSystemIndexes &&
!ReindexIsProcessingIndex(indexId))
irel = index_open(indexId, AccessShareLock);
else
irel = NULL;
sysscan = (SysScanDesc) palloc(sizeof(SysScanDescData));
sysscan->heap_rel = heapRelation;
sysscan->irel = irel;
if (irel)
{
int i;
/*
* Change attribute numbers to be index column numbers.
*
* This code could be generalized to search for the index key numbers
* to substitute, but for now there's no need.
*/
for (i = 0; i < nkeys; i++)
{
Assert(key[i].sk_attno == irel->rd_index->indkey.values[i]);
key[i].sk_attno = i + 1;
}
sysscan->iscan = index_beginscan(heapRelation, irel,
snapshot, nkeys, key);
sysscan->scan = NULL;
}
else
{
sysscan->scan = heap_beginscan(heapRelation, snapshot, nkeys, key);
sysscan->iscan = NULL;
}
return sysscan;
}
/*
* Starts a scan over the visimap store.
*
* Parameter keys may be NULL iff nkeys is zero.
*/
IndexScanDesc
AppendOnlyVisimapStore_BeginScan(
AppendOnlyVisimapStore *visiMapStore,
int nkeys,
ScanKey keys)
{
Assert(visiMapStore);
Assert(RelationIsValid(visiMapStore->visimapRelation));
return index_beginscan(
visiMapStore->visimapRelation,
visiMapStore->visimapIndex,
visiMapStore->snapshot,
nkeys,
keys);
}
/*
* GetNewOidWithIndex
* Guts of GetNewOid: use the supplied index
*
* This is exported separately because there are cases where we want to use
* an index that will not be recognized by RelationGetOidIndex: TOAST tables
* and pg_largeobject have indexes that are usable, but have multiple columns
* and are on ordinary columns rather than a true OID column. This code
* will work anyway, so long as the OID is the index's first column.
*
* Caller must have a suitable lock on the relation.
*/
Oid
GetNewOidWithIndex(Relation relation, Relation indexrel)
{
Oid newOid;
IndexScanDesc scan;
ScanKeyData key;
bool collides;
/* Generate new OIDs until we find one not in the table */
do
{
CHECK_FOR_INTERRUPTS();
newOid = GetNewObjectId();
ScanKeyInit(&key,
(AttrNumber) 1,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(newOid));
/* see notes above about using SnapshotDirty */
scan = index_beginscan(relation, indexrel,
SnapshotDirty, 1, &key);
collides = HeapTupleIsValid(index_getnext(scan, ForwardScanDirection));
index_endscan(scan);
} while (collides);
if (IsSystemNamespace(RelationGetNamespace(relation)))
{
if (Gp_role == GP_ROLE_EXECUTE)
{
if (relation->rd_id != 2604 /* pg_attrdef */ && relation->rd_id != 2606 /* pg_constraint */ && relation->rd_id != 2615 /* pg_namespace */)
elog(DEBUG1,"Allocating Oid %u with index on relid %u %s in EXECUTE mode",newOid,relation->rd_id, RelationGetRelationName(relation));
else
elog(DEBUG4,"Allocating Oid %u with index on relid %u %s in EXECUTE mode",newOid,relation->rd_id, RelationGetRelationName(relation));
}
if (Gp_role == GP_ROLE_DISPATCH)
{
elog(DEBUG5,"Allocating Oid %u with index on relid %u %s in DISPATCH mode",newOid,relation->rd_id, RelationGetRelationName(relation));
}
}
return newOid;
}
/*
* Initialize the index scan descriptor if it is not initialized.
*/
static inline void
initScanDesc(IndexScanState *indexstate)
{
Relation currentRelation = indexstate->ss.ss_currentRelation;
EState *estate = indexstate->ss.ps.state;
if (indexstate->iss_ScanDesc == NULL)
{
/*
* Initialize scan descriptor.
*/
indexstate->iss_ScanDesc = index_beginscan(currentRelation,
indexstate->iss_RelationDesc,
estate->es_snapshot,
indexstate->iss_NumScanKeys,
indexstate->iss_ScanKeys);
}
}
/*
* Starts a scan over the visimap store.
*
* Parameter keys may be NULL iff nkeys is zero.
*/
IndexScanDesc
AppendOnlyVisimapStore_BeginScan(AppendOnlyVisimapStore *visiMapStore,
int nkeys,
ScanKey keys)
{
IndexScanDesc scandesc;
Assert(visiMapStore);
Assert(RelationIsValid(visiMapStore->visimapRelation));
scandesc = index_beginscan(visiMapStore->visimapRelation,
visiMapStore->visimapIndex,
visiMapStore->snapshot,
nkeys,
0);
index_rescan(scandesc, keys, nkeys, NULL, 0);
return scandesc;
}
/*
* AppendOnlyBlockDirectory_DeleteSegmentFile
*
* Deletes all block directory entries for given segment file of an
* append-only relation.
*/
void
AppendOnlyBlockDirectory_DeleteSegmentFile(
AppendOnlyEntry *aoEntry,
Snapshot snapshot,
int segno,
int columnGroupNo)
{
Assert(aoEntry);
Assert(OidIsValid(aoEntry->blkdirrelid));
Assert(OidIsValid(aoEntry->blkdiridxid));
Relation blkdirRel = heap_open(aoEntry->blkdirrelid, RowExclusiveLock);
Relation blkdirIdx = index_open(aoEntry->blkdiridxid, RowExclusiveLock);
ScanKeyData scanKey;
ScanKeyInit(&scanKey,
1, /* segno */
BTEqualStrategyNumber,
F_INT4EQ,
Int32GetDatum(segno));
IndexScanDesc indexScan = index_beginscan(
blkdirRel,
blkdirIdx,
snapshot,
1,
&scanKey);
HeapTuple tuple = NULL;
while ((tuple = index_getnext(indexScan, ForwardScanDirection)) != NULL)
{
simple_heap_delete(blkdirRel,
&tuple->t_self);
}
index_endscan(indexScan);
index_close(blkdirIdx, RowExclusiveLock);
heap_close(blkdirRel, RowExclusiveLock);
}
void
rebuildheap(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex)
{
Relation LocalNewHeap, LocalOldHeap, LocalOldIndex;
IndexScanDesc ScanDesc;
RetrieveIndexResult ScanResult;
ItemPointer HeapTid;
HeapTuple LocalHeapTuple;
Buffer LocalBuffer;
Oid OIDNewHeapInsert;
/*
* Open the relations I need. Scan through the OldHeap on the OldIndex and
* insert each tuple into the NewHeap.
*/
LocalNewHeap=(Relation)heap_open(OIDNewHeap);
LocalOldHeap=(Relation)heap_open(OIDOldHeap);
LocalOldIndex=(Relation)index_open(OIDOldIndex);
ScanDesc=index_beginscan(LocalOldIndex, false, 0, (ScanKey) NULL);
while ((ScanResult =
index_getnext(ScanDesc, ForwardScanDirection)) != NULL) {
HeapTid = &ScanResult->heap_iptr;
LocalHeapTuple = heap_fetch(LocalOldHeap, 0, HeapTid, &LocalBuffer);
OIDNewHeapInsert =
heap_insert(LocalNewHeap, LocalHeapTuple);
pfree(ScanResult);
ReleaseBuffer(LocalBuffer);
}
index_close(LocalOldIndex);
heap_close(LocalOldHeap);
heap_close(LocalNewHeap);
}
/*
* _bitmap_init_buildstate() -- initialize the build state before building
* a bitmap index.
*/
void
_bitmap_init_buildstate(Relation index, BMBuildState *bmstate)
{
MIRROREDLOCK_BUFMGR_DECLARE;
BMMetaPage mp;
HASHCTL hash_ctl;
int hash_flags;
int i;
Buffer metabuf;
/* initialize the build state */
bmstate->bm_tupDesc = RelationGetDescr(index);
bmstate->bm_tidLocsBuffer = (BMTidBuildBuf *)
palloc(sizeof(BMTidBuildBuf));
bmstate->bm_tidLocsBuffer->byte_size = 0;
bmstate->bm_tidLocsBuffer->lov_blocks = NIL;
bmstate->bm_tidLocsBuffer->max_lov_block = InvalidBlockNumber;
// -------- MirroredLock ----------
MIRROREDLOCK_BUFMGR_LOCK;
metabuf = _bitmap_getbuf(index, BM_METAPAGE, BM_READ);
mp = _bitmap_get_metapage_data(index, metabuf);
_bitmap_open_lov_heapandindex(index, mp, &(bmstate->bm_lov_heap),
&(bmstate->bm_lov_index),
RowExclusiveLock);
_bitmap_relbuf(metabuf);
MIRROREDLOCK_BUFMGR_UNLOCK;
// -------- MirroredLock ----------
cur_bmbuild = (BMBuildHashData *)palloc(sizeof(BMBuildHashData));
cur_bmbuild->hash_funcs = (FmgrInfo *)
palloc(sizeof(FmgrInfo) * bmstate->bm_tupDesc->natts);
cur_bmbuild->eq_funcs = (FmgrInfo *)
palloc(sizeof(FmgrInfo) * bmstate->bm_tupDesc->natts);
cur_bmbuild->hash_func_is_strict = (bool *)
palloc(sizeof(bool) * bmstate->bm_tupDesc->natts);
for (i = 0; i < bmstate->bm_tupDesc->natts; i++)
{
Oid typid = bmstate->bm_tupDesc->attrs[i]->atttypid;
Operator optup;
Oid eq_opr;
Oid eq_function;
Oid left_hash_function;
Oid right_hash_function;
optup = equality_oper(typid, false);
eq_opr = oprid(optup);
eq_function = oprfuncid(optup);
ReleaseOperator(optup);
if (!get_op_hash_functions(eq_opr,
&left_hash_function,
&right_hash_function))
{
pfree(cur_bmbuild);
cur_bmbuild = NULL;
break;
}
Assert(left_hash_function == right_hash_function);
fmgr_info(eq_function, &cur_bmbuild->eq_funcs[i]);
fmgr_info(right_hash_function, &cur_bmbuild->hash_funcs[i]);
cur_bmbuild->hash_func_is_strict[i] = func_strict(right_hash_function);
}
if (cur_bmbuild)
{
cur_bmbuild->natts = bmstate->bm_tupDesc->natts;
cur_bmbuild->tmpcxt = AllocSetContextCreate(CurrentMemoryContext,
"Bitmap build temp space",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
/* setup the hash table */
MemSet(&hash_ctl, 0, sizeof(hash_ctl));
/**
* Reserve enough space for the hash key header and then the data segments (values followed by nulls)
*/
hash_ctl.keysize = MAXALIGN(sizeof(BMBuildHashKey)) +
MAXALIGN(sizeof(Datum) * cur_bmbuild->natts) +
MAXALIGN(sizeof(bool) * cur_bmbuild->natts);
hash_ctl.entrysize = hash_ctl.keysize + sizeof(BMBuildLovData) + 200;
hash_ctl.hash = build_hash_key;
hash_ctl.match = build_match_key;
hash_ctl.keycopy = build_keycopy;
hash_ctl.hcxt = AllocSetContextCreate(CurrentMemoryContext,
"Bitmap build hash table",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
cur_bmbuild->hash_cxt = hash_ctl.hcxt;
hash_flags = HASH_ELEM | HASH_FUNCTION | HASH_COMPARE | HASH_CONTEXT | HASH_KEYCOPY;
bmstate->lovitem_hash = hash_create("Bitmap index build lov item hash",
100, &hash_ctl, hash_flags);
bmstate->lovitem_hashKeySize = hash_ctl.keysize;
}
else
{
int attno;
bmstate->lovitem_hash = NULL;
bmstate->lovitem_hashKeySize = 0;
bmstate->bm_lov_scanKeys =
(ScanKey)palloc0(bmstate->bm_tupDesc->natts * sizeof(ScanKeyData));
for (attno = 0; attno < bmstate->bm_tupDesc->natts; attno++)
{
RegProcedure opfuncid;
Oid atttypid;
atttypid = bmstate->bm_tupDesc->attrs[attno]->atttypid;
opfuncid = equality_oper_funcid(atttypid);
ScanKeyEntryInitialize(&(bmstate->bm_lov_scanKeys[attno]), SK_ISNULL,
attno + 1, BTEqualStrategyNumber, InvalidOid,
opfuncid, 0);
}
bmstate->bm_lov_scanDesc = index_beginscan(bmstate->bm_lov_heap,
bmstate->bm_lov_index, ActiveSnapshot,
bmstate->bm_tupDesc->natts,
bmstate->bm_lov_scanKeys);
}
/*
* We need to log index creation in WAL iff WAL archiving is enabled
* AND it's not a temp index. Currently, since building an index
* writes page to the shared buffer, we can't disable WAL archiving.
* We will add this shortly.
*/
bmstate->use_wal = !XLog_UnconvertedCanBypassWal() && !index->rd_istemp;
}
bool
CheckNewRelFileNodeIsOk(Oid newOid, Oid reltablespace, bool relisshared,
Relation pg_class)
{
RelFileNode rnode;
char *rpath;
int fd;
bool collides;
if (pg_class)
{
Oid oidIndex;
Relation indexrel;
IndexScanDesc scan;
ScanKeyData key;
Assert(!IsBootstrapProcessingMode());
Assert(pg_class->rd_rel->relhasoids);
/* The relcache will cache the identity of the OID index for us */
oidIndex = RelationGetOidIndex(pg_class);
Assert(OidIsValid(oidIndex));
indexrel = index_open(oidIndex, AccessShareLock);
ScanKeyInit(&key,
(AttrNumber) 1,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(newOid));
scan = index_beginscan(pg_class, indexrel, SnapshotDirty, 1, &key);
collides = HeapTupleIsValid(index_getnext(scan, ForwardScanDirection));
index_endscan(scan);
index_close(indexrel, AccessShareLock);
if (collides)
elog(ERROR, "relfilenode %d already in use in \"pg_class\"",
newOid);
}
/* This should match RelationInitPhysicalAddr */
rnode.spcNode = reltablespace ? reltablespace : MyDatabaseTableSpace;
rnode.dbNode = relisshared ? InvalidOid : MyDatabaseId;
rnode.relNode = newOid;
/* Check for existing file of same name */
rpath = relpath(rnode);
fd = BasicOpenFile(rpath, O_RDONLY | PG_BINARY, 0);
if (fd >= 0)
{
/* definite collision */
gp_retry_close(fd);
collides = true;
}
else
collides = false;
pfree(rpath);
if (collides && !relisshared)
elog(ERROR, "oid %d already in use", newOid);
while(GetNewObjectId() < newOid);
return !collides;
}
int
inv_read(LargeObjectDesc *obj_desc, char *buf, int nbytes)
{
int nread = 0;
int n;
int off;
int len;
int32 pageno = (int32) (obj_desc->offset / LOBLKSIZE);
uint32 pageoff;
ScanKeyData skey[2];
IndexScanDesc sd;
HeapTuple tuple;
Assert(PointerIsValid(obj_desc));
Assert(buf != NULL);
if (nbytes <= 0)
return 0;
open_lo_relation();
ScanKeyInit(&skey[0],
Anum_pg_largeobject_loid,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(obj_desc->id));
ScanKeyInit(&skey[1],
Anum_pg_largeobject_pageno,
BTGreaterEqualStrategyNumber, F_INT4GE,
Int32GetDatum(pageno));
sd = index_beginscan(lo_heap_r, lo_index_r,
SnapshotNow, 2, skey);
while ((tuple = index_getnext(sd, ForwardScanDirection)) != NULL)
{
Form_pg_largeobject data;
bytea *datafield;
bool pfreeit;
data = (Form_pg_largeobject) GETSTRUCT(tuple);
/*
* We assume the indexscan will deliver pages in order. However,
* there may be missing pages if the LO contains unwritten
* "holes". We want missing sections to read out as zeroes.
*/
pageoff = ((uint32) data->pageno) * LOBLKSIZE;
if (pageoff > obj_desc->offset)
{
n = pageoff - obj_desc->offset;
n = (n <= (nbytes - nread)) ? n : (nbytes - nread);
MemSet(buf + nread, 0, n);
nread += n;
obj_desc->offset += n;
}
if (nread < nbytes)
{
Assert(obj_desc->offset >= pageoff);
off = (int) (obj_desc->offset - pageoff);
Assert(off >= 0 && off < LOBLKSIZE);
datafield = &(data->data);
pfreeit = false;
if (VARATT_IS_EXTENDED(datafield))
{
datafield = (bytea *)
heap_tuple_untoast_attr((varattrib *) datafield);
pfreeit = true;
}
len = getbytealen(datafield);
if (len > off)
{
n = len - off;
n = (n <= (nbytes - nread)) ? n : (nbytes - nread);
memcpy(buf + nread, VARDATA(datafield) + off, n);
nread += n;
obj_desc->offset += n;
}
if (pfreeit)
pfree(datafield);
}
if (nread >= nbytes)
break;
}
index_endscan(sd);
return nread;
}
/*
* systable_beginscan --- set up for heap-or-index scan
*
* rel: catalog to scan, already opened and suitably locked
* indexId: OID of index to conditionally use
* indexOK: if false, forces a heap scan (see notes below)
* snapshot: time qual to use (NULL for a recent catalog snapshot)
* nkeys, key: scan keys
*
* The attribute numbers in the scan key should be set for the heap case.
* If we choose to index, we reset them to 1..n to reference the index
* columns. Note this means there must be one scankey qualification per
* index column! This is checked by the Asserts in the normal, index-using
* case, but won't be checked if the heapscan path is taken.
*
* The routine checks the normal cases for whether an indexscan is safe,
* but caller can make additional checks and pass indexOK=false if needed.
* In standard case indexOK can simply be constant TRUE.
*/
SysScanDesc
systable_beginscan(Relation heapRelation,
Oid indexId,
bool indexOK,
Snapshot snapshot,
int nkeys, ScanKey key)
{
SysScanDesc sysscan;
Relation irel;
if (indexOK &&
!IgnoreSystemIndexes &&
!ReindexIsProcessingIndex(indexId))
irel = index_open(indexId, AccessShareLock);
else
irel = NULL;
sysscan = (SysScanDesc) palloc(sizeof(SysScanDescData));
sysscan->heap_rel = heapRelation;
sysscan->irel = irel;
if (snapshot == NULL)
{
Oid relid = RelationGetRelid(heapRelation);
snapshot = RegisterSnapshot(GetCatalogSnapshot(relid));
sysscan->snapshot = snapshot;
}
else
{
/* Caller is responsible for any snapshot. */
sysscan->snapshot = NULL;
}
if (irel)
{
int i;
/* Change attribute numbers to be index column numbers. */
for (i = 0; i < nkeys; i++)
{
int j;
for (j = 0; j < irel->rd_index->indnatts; j++)
{
if (key[i].sk_attno == irel->rd_index->indkey.values[j])
{
key[i].sk_attno = j + 1;
break;
}
}
if (j == irel->rd_index->indnatts)
elog(ERROR, "column is not in index");
}
sysscan->iscan = index_beginscan(heapRelation, irel,
snapshot, nkeys, 0);
index_rescan(sysscan->iscan, key, nkeys, NULL, 0);
sysscan->scan = NULL;
}
else
{
/*
* We disallow synchronized scans when forced to use a heapscan on a
* catalog. In most cases the desired rows are near the front, so
* that the unpredictable start point of a syncscan is a serious
* disadvantage; and there are no compensating advantages, because
* it's unlikely that such scans will occur in parallel.
*/
sysscan->scan = heap_beginscan_strat(heapRelation, snapshot,
nkeys, key,
true, false);
sysscan->iscan = NULL;
}
return sysscan;
}
int
inv_write(LargeObjectDesc *obj_desc, char *buf, int nbytes)
{
int nwritten = 0;
int n;
int off;
int len;
int32 pageno = (int32) (obj_desc->offset / LOBLKSIZE);
ScanKeyData skey[2];
IndexScanDesc sd;
HeapTuple oldtuple;
Form_pg_largeobject olddata;
bool neednextpage;
bytea *datafield;
bool pfreeit;
struct
{
bytea hdr;
char data[LOBLKSIZE];
} workbuf;
char *workb = VARATT_DATA(&workbuf.hdr);
HeapTuple newtup;
Datum values[Natts_pg_largeobject];
char nulls[Natts_pg_largeobject];
char replace[Natts_pg_largeobject];
CatalogIndexState indstate;
Assert(PointerIsValid(obj_desc));
Assert(buf != NULL);
if (nbytes <= 0)
return 0;
open_lo_relation();
indstate = CatalogOpenIndexes(lo_heap_r);
ScanKeyInit(&skey[0],
Anum_pg_largeobject_loid,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(obj_desc->id));
ScanKeyInit(&skey[1],
Anum_pg_largeobject_pageno,
BTGreaterEqualStrategyNumber, F_INT4GE,
Int32GetDatum(pageno));
sd = index_beginscan(lo_heap_r, lo_index_r,
SnapshotNow, 2, skey);
oldtuple = NULL;
olddata = NULL;
neednextpage = true;
while (nwritten < nbytes)
{
/*
* If possible, get next pre-existing page of the LO. We assume
* the indexscan will deliver these in order --- but there may be
* holes.
*/
if (neednextpage)
{
if ((oldtuple = index_getnext(sd, ForwardScanDirection)) != NULL)
{
olddata = (Form_pg_largeobject) GETSTRUCT(oldtuple);
Assert(olddata->pageno >= pageno);
}
neednextpage = false;
}
/*
* If we have a pre-existing page, see if it is the page we want
* to write, or a later one.
*/
if (olddata != NULL && olddata->pageno == pageno)
{
/*
* Update an existing page with fresh data.
*
* First, load old data into workbuf
*/
datafield = &(olddata->data);
pfreeit = false;
if (VARATT_IS_EXTENDED(datafield))
{
datafield = (bytea *)
heap_tuple_untoast_attr((varattrib *) datafield);
pfreeit = true;
}
len = getbytealen(datafield);
Assert(len <= LOBLKSIZE);
memcpy(workb, VARDATA(datafield), len);
if (pfreeit)
pfree(datafield);
/*
* Fill any hole
*/
off = (int) (obj_desc->offset % LOBLKSIZE);
if (off > len)
MemSet(workb + len, 0, off - len);
/*
* Insert appropriate portion of new data
*/
n = LOBLKSIZE - off;
n = (n <= (nbytes - nwritten)) ? n : (nbytes - nwritten);
memcpy(workb + off, buf + nwritten, n);
nwritten += n;
obj_desc->offset += n;
off += n;
/* compute valid length of new page */
len = (len >= off) ? len : off;
VARATT_SIZEP(&workbuf.hdr) = len + VARHDRSZ;
/*
* Form and insert updated tuple
*/
memset(values, 0, sizeof(values));
memset(nulls, ' ', sizeof(nulls));
memset(replace, ' ', sizeof(replace));
values[Anum_pg_largeobject_data - 1] = PointerGetDatum(&workbuf);
replace[Anum_pg_largeobject_data - 1] = 'r';
newtup = heap_modifytuple(oldtuple, lo_heap_r,
values, nulls, replace);
simple_heap_update(lo_heap_r, &newtup->t_self, newtup);
CatalogIndexInsert(indstate, newtup);
heap_freetuple(newtup);
/*
* We're done with this old page.
*/
oldtuple = NULL;
olddata = NULL;
neednextpage = true;
}
else
{
/*
* Write a brand new page.
*
* First, fill any hole
*/
off = (int) (obj_desc->offset % LOBLKSIZE);
if (off > 0)
MemSet(workb, 0, off);
/*
* Insert appropriate portion of new data
*/
n = LOBLKSIZE - off;
n = (n <= (nbytes - nwritten)) ? n : (nbytes - nwritten);
memcpy(workb + off, buf + nwritten, n);
nwritten += n;
obj_desc->offset += n;
/* compute valid length of new page */
len = off + n;
VARATT_SIZEP(&workbuf.hdr) = len + VARHDRSZ;
/*
* Form and insert updated tuple
*/
memset(values, 0, sizeof(values));
memset(nulls, ' ', sizeof(nulls));
values[Anum_pg_largeobject_loid - 1] = ObjectIdGetDatum(obj_desc->id);
values[Anum_pg_largeobject_pageno - 1] = Int32GetDatum(pageno);
values[Anum_pg_largeobject_data - 1] = PointerGetDatum(&workbuf);
newtup = heap_formtuple(lo_heap_r->rd_att, values, nulls);
simple_heap_insert(lo_heap_r, newtup);
CatalogIndexInsert(indstate, newtup);
heap_freetuple(newtup);
}
pageno++;
}
index_endscan(sd);
CatalogCloseIndexes(indstate);
/*
* Advance command counter so that my tuple updates will be seen by
* later large-object operations in this transaction.
*/
CommandCounterIncrement();
return nwritten;
}
/*
* GetNewSequenceRelationOid
* Get a sequence relation Oid and verify it is valid against
* the pg_class relation by doing an index lookup. The caller
* should have a suitable lock on pg_class.
*/
Oid
GetNewSequenceRelationOid(Relation relation)
{
Oid newOid;
Oid oidIndex;
Relation indexrel;
SnapshotData SnapshotDirty;
IndexScanDesc scan;
ScanKeyData key;
bool collides;
RelFileNode rnode;
char *rpath;
int fd;
/* This should match RelationInitPhysicalAddr */
rnode.spcNode = relation->rd_rel->reltablespace ? relation->rd_rel->reltablespace : MyDatabaseTableSpace;
rnode.dbNode = relation->rd_rel->relisshared ? InvalidOid : MyDatabaseId;
/* We should only be using pg_class */
Assert(RelationGetRelid(relation) == RelationRelationId);
/* The relcache will cache the identity of the OID index for us */
oidIndex = RelationGetOidIndex(relation);
/* Otherwise, use the index to find a nonconflicting OID */
indexrel = index_open(oidIndex, AccessShareLock);
InitDirtySnapshot(SnapshotDirty);
/* Generate new sequence relation OIDs until we find one not in the table */
do
{
CHECK_FOR_INTERRUPTS();
newOid = GetNewSequenceRelationObjectId();
ScanKeyInit(&key,
(AttrNumber) 1,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(newOid));
/* see notes above about using SnapshotDirty */
scan = index_beginscan(relation, indexrel,
&SnapshotDirty, 1, &key);
collides = HeapTupleIsValid(index_getnext(scan, ForwardScanDirection));
index_endscan(scan);
if (!collides)
{
/* Check for existing file of same name */
rpath = relpath(rnode);
fd = BasicOpenFile(rpath, O_RDONLY | PG_BINARY, 0);
if (fd >= 0)
{
/* definite collision */
gp_retry_close(fd);
collides = true;
}
else
{
/*
* Here we have a little bit of a dilemma: if errno is something
* other than ENOENT, should we declare a collision and loop? In
* particular one might think this advisable for, say, EPERM.
* However there really shouldn't be any unreadable files in a
* tablespace directory, and if the EPERM is actually complaining
* that we can't read the directory itself, we'd be in an infinite
* loop. In practice it seems best to go ahead regardless of the
* errno. If there is a colliding file we will get an smgr
* failure when we attempt to create the new relation file.
*/
collides = false;
}
}
/*
* Also check that the OID hasn't been pre-assigned for a different
* relation.
*
* We're a bit sloppy between OIDs and relfilenodes here; it would be
* OK to use a value that's been reserved for use as a type or
* relation OID here, as long as the relfilenode is free. But there's
* no harm in skipping over those too, so we don't bother to
* distinguish them.
*/
if (!collides && !IsOidAcceptable(newOid))
collides = true;
} while (collides);
index_close(indexrel, AccessShareLock);
return newOid;
}
/* ----------
* toast_fetch_datum -
*
* Reconstruct an in memory varattrib from the chunks saved
* in the toast relation
* ----------
*/
static varattrib *
toast_fetch_datum(varattrib *attr)
{
Relation toastrel;
Relation toastidx;
ScanKeyData toastkey;
IndexScanDesc toastscan;
HeapTuple ttup;
TupleDesc toasttupDesc;
varattrib *result;
int32 ressize;
int32 residx,
nextidx;
int32 numchunks;
Pointer chunk;
bool isnull;
int32 chunksize;
ressize = attr->va_content.va_external.va_extsize;
numchunks = ((ressize - 1) / TOAST_MAX_CHUNK_SIZE) + 1;
result = (varattrib *) palloc(ressize + VARHDRSZ);
VARATT_SIZEP(result) = ressize + VARHDRSZ;
if (VARATT_IS_COMPRESSED(attr))
VARATT_SIZEP(result) |= VARATT_FLAG_COMPRESSED;
/*
* Open the toast relation and its index
*/
toastrel = heap_open(attr->va_content.va_external.va_toastrelid,
AccessShareLock);
toasttupDesc = toastrel->rd_att;
toastidx = index_open(toastrel->rd_rel->reltoastidxid);
/*
* Setup a scan key to fetch from the index by va_valueid
*/
ScanKeyEntryInitialize(&toastkey,
(bits16) 0,
(AttrNumber) 1,
(RegProcedure) F_OIDEQ,
ObjectIdGetDatum(attr->va_content.va_external.va_valueid));
/*
* Read the chunks by index
*
* Note that because the index is actually on (valueid, chunkidx) we will
* see the chunks in chunkidx order, even though we didn't explicitly
* ask for it.
*/
nextidx = 0;
toastscan = index_beginscan(toastrel, toastidx, SnapshotToast,
1, &toastkey);
while ((ttup = index_getnext(toastscan, ForwardScanDirection)) != NULL)
{
/*
* Have a chunk, extract the sequence number and the data
*/
residx = DatumGetInt32(heap_getattr(ttup, 2, toasttupDesc, &isnull));
Assert(!isnull);
chunk = DatumGetPointer(heap_getattr(ttup, 3, toasttupDesc, &isnull));
Assert(!isnull);
chunksize = VARATT_SIZE(chunk) - VARHDRSZ;
/*
* Some checks on the data we've found
*/
if (residx != nextidx)
elog(ERROR, "unexpected chunk number %d (expected %d) for toast value %u",
residx, nextidx,
attr->va_content.va_external.va_valueid);
if (residx < numchunks - 1)
{
if (chunksize != TOAST_MAX_CHUNK_SIZE)
elog(ERROR, "unexpected chunk size %d in chunk %d for toast value %u",
chunksize, residx,
attr->va_content.va_external.va_valueid);
}
else if (residx < numchunks)
{
if ((residx * TOAST_MAX_CHUNK_SIZE + chunksize) != ressize)
elog(ERROR, "unexpected chunk size %d in chunk %d for toast value %u",
chunksize, residx,
attr->va_content.va_external.va_valueid);
}
else
elog(ERROR, "unexpected chunk number %d for toast value %u",
residx,
attr->va_content.va_external.va_valueid);
/*
* Copy the data into proper place in our result
*/
memcpy(((char *) VARATT_DATA(result)) + residx * TOAST_MAX_CHUNK_SIZE,
VARATT_DATA(chunk),
chunksize);
nextidx++;
}
/*
* Final checks that we successfully fetched the datum
*/
if (nextidx != numchunks)
elog(ERROR, "missing chunk number %d for toast value %u",
nextidx,
attr->va_content.va_external.va_valueid);
/*
* End scan and close relations
*/
index_endscan(toastscan);
index_close(toastidx);
heap_close(toastrel, AccessShareLock);
return result;
}
/* ----------
* toast_fetch_datum -
*
* Reconstruct an in memory Datum from the chunks saved
* in the toast relation
* ----------
*/
static struct varlena *
toast_fetch_datum(struct varlena *attr)
{
Relation toastrel;
Relation toastidx;
ScanKeyData toastkey;
IndexScanDesc toastscan;
HeapTuple ttup;
TupleDesc toasttupDesc;
varattrib *result;
int32 ressize;
int32 residx,
nextidx;
int32 numchunks;
Pointer chunk;
bool isnull;
int32 chunksize;
void *chunkdata;
ressize = ((varattrib *)attr)->va_external.va_extsize;
numchunks = ((ressize - 1) / TOAST_MAX_CHUNK_SIZE) + 1;
result = (varattrib *) palloc(ressize + VARHDRSZ);
SET_VARSIZE(result, ressize + VARHDRSZ);
if (VARATT_EXTERNAL_IS_COMPRESSED(attr))
VARATT_SET_COMPRESSED(result);
/*
* Open the toast relation and its index
*/
toastrel = heap_open(((varattrib *)attr)->va_external.va_toastrelid, AccessShareLock);
toasttupDesc = toastrel->rd_att;
toastidx = index_open(toastrel->rd_rel->reltoastidxid, AccessShareLock);
/*
* Setup a scan key to fetch from the index by va_valueid
*/
ScanKeyInit(&toastkey,
(AttrNumber) 1,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(((varattrib *)attr)->va_external.va_valueid));
/*
* Read the chunks by index
*
* Note that because the index is actually on (valueid, chunkidx) we will
* see the chunks in chunkidx order, even though we didn't explicitly ask
* for it.
*/
nextidx = 0;
toastscan = index_beginscan(toastrel, toastidx,
SnapshotToast, 1, &toastkey);
while ((ttup = index_getnext(toastscan, ForwardScanDirection)) != NULL)
{
/*
* Have a chunk, extract the sequence number and the data
*/
residx = DatumGetInt32(fastgetattr(ttup, 2, toasttupDesc, &isnull));
Assert(!isnull);
chunk = DatumGetPointer(fastgetattr(ttup, 3, toasttupDesc, &isnull));
Assert(!isnull);
if (VARATT_IS_SHORT(chunk))
{
chunksize = VARSIZE_SHORT(chunk) - VARHDRSZ_SHORT;
chunkdata = VARDATA_SHORT(chunk);
}
else if (!VARATT_IS_EXTENDED(chunk))
{
chunksize = VARSIZE(chunk) - VARHDRSZ;
chunkdata = VARDATA(chunk);
}
else
{
elog(ERROR, "found toasted toast chunk?");
chunksize = 0; /* shut compiler up */
chunkdata = NULL;
}
/*
* Some checks on the data we've found
*/
if (residx != nextidx)
elog(ERROR, "unexpected chunk number %d (expected %d) for toast value %u",
residx, nextidx,
((varattrib *)attr)->va_external.va_valueid);
if (residx < numchunks - 1)
{
if (chunksize != TOAST_MAX_CHUNK_SIZE)
elog(ERROR, "unexpected chunk size %d in chunk %d of %d for toast value %u (expected %d)",
chunksize, residx,
((varattrib *)attr)->va_external.va_valueid, numchunks-1,
(int)TOAST_MAX_CHUNK_SIZE);
}
else if (residx == numchunks-1)
{
if ((residx * TOAST_MAX_CHUNK_SIZE + chunksize) != ressize)
elog(ERROR, "unexpected chunk size %d in final chunk %d for toast value %u (expected %d)",
chunksize, residx,
((varattrib *)attr)->va_external.va_valueid,
ressize - residx*(int)TOAST_MAX_CHUNK_SIZE);
}
else
elog(ERROR, "unexpected chunk number %d for toast value %u (expected in %d..%d)",
residx,
((varattrib *)attr)->va_external.va_valueid,
0, numchunks-1);
/*
* Copy the data into proper place in our result
*/
memcpy(((char *) VARDATA(result)) + residx * TOAST_MAX_CHUNK_SIZE,
chunkdata,
chunksize);
nextidx++;
}
/*
* Final checks that we successfully fetched the datum
*/
if (nextidx != numchunks)
elog(ERROR, "missing chunk number %d for toast value %u",
nextidx,
((varattrib *)attr)->va_external.va_valueid);
/*
* End scan and close relations
*/
index_endscan(toastscan);
index_close(toastidx, AccessShareLock);
heap_close(toastrel, AccessShareLock);
return (struct varlena *)result;
}
/*
* Search the relation 'rel' for tuple using the index.
*
* If a matching tuple is found, lock it with lockmode, fill the slot with its
* contents, and return true. Return false otherwise.
*/
bool
RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
LockTupleMode lockmode,
TupleTableSlot *searchslot,
TupleTableSlot *outslot)
{
HeapTuple scantuple;
ScanKeyData skey[INDEX_MAX_KEYS];
IndexScanDesc scan;
SnapshotData snap;
TransactionId xwait;
Relation idxrel;
bool found;
/* Open the index. */
idxrel = index_open(idxoid, RowExclusiveLock);
/* Start an index scan. */
InitDirtySnapshot(snap);
scan = index_beginscan(rel, idxrel, &snap,
RelationGetNumberOfAttributes(idxrel),
0);
/* Build scan key. */
build_replindex_scan_key(skey, rel, idxrel, searchslot);
retry:
found = false;
index_rescan(scan, skey, RelationGetNumberOfAttributes(idxrel), NULL, 0);
/* Try to find the tuple */
if ((scantuple = index_getnext(scan, ForwardScanDirection)) != NULL)
{
found = true;
ExecStoreTuple(scantuple, outslot, InvalidBuffer, false);
ExecMaterializeSlot(outslot);
xwait = TransactionIdIsValid(snap.xmin) ?
snap.xmin : snap.xmax;
/*
* If the tuple is locked, wait for locking transaction to finish and
* retry.
*/
if (TransactionIdIsValid(xwait))
{
XactLockTableWait(xwait, NULL, NULL, XLTW_None);
goto retry;
}
}
/* Found tuple, try to lock it in the lockmode. */
if (found)
{
Buffer buf;
HeapUpdateFailureData hufd;
HTSU_Result res;
HeapTupleData locktup;
ItemPointerCopy(&outslot->tts_tuple->t_self, &locktup.t_self);
PushActiveSnapshot(GetLatestSnapshot());
res = heap_lock_tuple(rel, &locktup, GetCurrentCommandId(false),
lockmode,
LockWaitBlock,
false /* don't follow updates */ ,
&buf, &hufd);
/* the tuple slot already has the buffer pinned */
ReleaseBuffer(buf);
PopActiveSnapshot();
switch (res)
{
case HeapTupleMayBeUpdated:
break;
case HeapTupleUpdated:
/* XXX: Improve handling here */
ereport(LOG,
(errcode(ERRCODE_T_R_SERIALIZATION_FAILURE),
errmsg("concurrent update, retrying")));
goto retry;
case HeapTupleInvisible:
elog(ERROR, "attempted to lock invisible tuple");
default:
elog(ERROR, "unexpected heap_lock_tuple status: %u", res);
break;
}
}
index_endscan(scan);
/* Don't release lock until commit. */
index_close(idxrel, NoLock);
return found;
}
/* ----------------------------------------------------------------
* ExecInitIndexOnlyScan
*
* Initializes the index scan's state information, creates
* scan keys, and opens the base and index relations.
*
* Note: index scans have 2 sets of state information because
* we have to keep track of the base relation and the
* index relation.
* ----------------------------------------------------------------
*/
IndexOnlyScanState *
ExecInitIndexOnlyScan(IndexOnlyScan *node, EState *estate, int eflags)
{
IndexOnlyScanState *indexstate;
Relation currentRelation;
bool relistarget;
TupleDesc tupDesc;
/*
* create state structure
*/
indexstate = makeNode(IndexOnlyScanState);
indexstate->ss.ps.plan = (Plan *) node;
indexstate->ss.ps.state = estate;
indexstate->ioss_HeapFetches = 0;
/*
* Miscellaneous initialization
*
* create expression context for node
*/
ExecAssignExprContext(estate, &indexstate->ss.ps);
indexstate->ss.ps.ps_TupFromTlist = false;
/*
* initialize child expressions
*
* Note: we don't initialize all of the indexorderby expression, only the
* sub-parts corresponding to runtime keys (see below).
*/
indexstate->ss.ps.targetlist = (List *)
ExecInitExpr((Expr *) node->scan.plan.targetlist,
(PlanState *) indexstate);
indexstate->ss.ps.qual = (List *)
ExecInitExpr((Expr *) node->scan.plan.qual,
(PlanState *) indexstate);
indexstate->indexqual = (List *)
ExecInitExpr((Expr *) node->indexqual,
(PlanState *) indexstate);
/*
* tuple table initialization
*/
ExecInitResultTupleSlot(estate, &indexstate->ss.ps);
ExecInitScanTupleSlot(estate, &indexstate->ss);
/*
* open the base relation and acquire appropriate lock on it.
*/
currentRelation = ExecOpenScanRelation(estate, node->scan.scanrelid, eflags);
indexstate->ss.ss_currentRelation = currentRelation;
indexstate->ss.ss_currentScanDesc = NULL; /* no heap scan here */
/*
* Build the scan tuple type using the indextlist generated by the
* planner. We use this, rather than the index's physical tuple
* descriptor, because the latter contains storage column types not the
* types of the original datums. (It's the AM's responsibility to return
* suitable data anyway.)
*/
tupDesc = ExecTypeFromTL(node->indextlist, false);
ExecAssignScanType(&indexstate->ss, tupDesc);
/*
* Initialize result tuple type and projection info. The node's
* targetlist will contain Vars with varno = INDEX_VAR, referencing the
* scan tuple.
*/
ExecAssignResultTypeFromTL(&indexstate->ss.ps);
ExecAssignScanProjectionInfoWithVarno(&indexstate->ss, INDEX_VAR);
/*
* If we are just doing EXPLAIN (ie, aren't going to run the plan), stop
* here. This allows an index-advisor plugin to EXPLAIN a plan containing
* references to nonexistent indexes.
*/
if (eflags & EXEC_FLAG_EXPLAIN_ONLY)
return indexstate;
/*
* Open the index relation.
*
* If the parent table is one of the target relations of the query, then
* InitPlan already opened and write-locked the index, so we can avoid
* taking another lock here. Otherwise we need a normal reader's lock.
*/
relistarget = ExecRelationIsTargetRelation(estate, node->scan.scanrelid);
indexstate->ioss_RelationDesc = index_open(node->indexid,
relistarget ? NoLock : AccessShareLock);
/*
* Initialize index-specific scan state
*/
indexstate->ioss_RuntimeKeysReady = false;
indexstate->ioss_RuntimeKeys = NULL;
indexstate->ioss_NumRuntimeKeys = 0;
/*
* build the index scan keys from the index qualification
*/
ExecIndexBuildScanKeys((PlanState *) indexstate,
indexstate->ioss_RelationDesc,
node->indexqual,
false,
&indexstate->ioss_ScanKeys,
&indexstate->ioss_NumScanKeys,
&indexstate->ioss_RuntimeKeys,
&indexstate->ioss_NumRuntimeKeys,
NULL, /* no ArrayKeys */
NULL);
/*
* any ORDER BY exprs have to be turned into scankeys in the same way
*/
ExecIndexBuildScanKeys((PlanState *) indexstate,
indexstate->ioss_RelationDesc,
node->indexorderby,
true,
&indexstate->ioss_OrderByKeys,
&indexstate->ioss_NumOrderByKeys,
&indexstate->ioss_RuntimeKeys,
&indexstate->ioss_NumRuntimeKeys,
NULL, /* no ArrayKeys */
NULL);
/*
* If we have runtime keys, we need an ExprContext to evaluate them. The
* node's standard context won't do because we want to reset that context
* for every tuple. So, build another context just like the other one...
* -tgl 7/11/00
*/
if (indexstate->ioss_NumRuntimeKeys != 0)
{
ExprContext *stdecontext = indexstate->ss.ps.ps_ExprContext;
ExecAssignExprContext(estate, &indexstate->ss.ps);
indexstate->ioss_RuntimeContext = indexstate->ss.ps.ps_ExprContext;
indexstate->ss.ps.ps_ExprContext = stdecontext;
}
else
{
indexstate->ioss_RuntimeContext = NULL;
}
/*
* Initialize scan descriptor.
*/
indexstate->ioss_ScanDesc = index_beginscan(currentRelation,
indexstate->ioss_RelationDesc,
estate->es_snapshot,
indexstate->ioss_NumScanKeys,
indexstate->ioss_NumOrderByKeys);
/* Set it up for index-only scan */
indexstate->ioss_ScanDesc->xs_want_itup = true;
indexstate->ioss_VMBuffer = InvalidBuffer;
/*
* If no run-time keys to calculate, go ahead and pass the scankeys to the
* index AM.
*/
if (indexstate->ioss_NumRuntimeKeys == 0)
index_rescan(indexstate->ioss_ScanDesc,
indexstate->ioss_ScanKeys,
indexstate->ioss_NumScanKeys,
indexstate->ioss_OrderByKeys,
indexstate->ioss_NumOrderByKeys);
/*
* all done.
*/
return indexstate;
}
/* ----------
* toast_fetch_datum_slice -
*
* Reconstruct a segment of a Datum from the chunks saved
* in the toast relation
* ----------
*/
static struct varlena *
toast_fetch_datum_slice(struct varlena *attr, int32 sliceoffset, int32 length)
{
Relation toastrel;
Relation toastidx;
ScanKeyData toastkey[3];
int nscankeys;
IndexScanDesc toastscan;
HeapTuple ttup;
TupleDesc toasttupDesc;
varattrib *result;
int32 attrsize;
int32 residx;
int32 nextidx;
int numchunks;
int startchunk;
int endchunk;
int32 startoffset;
int32 endoffset;
int totalchunks;
Pointer chunk;
bool isnull;
int32 chunksize;
int32 chcpystrt;
int32 chcpyend;
attrsize = ((varattrib *)attr)->va_external.va_extsize;
totalchunks = ((attrsize - 1) / TOAST_MAX_CHUNK_SIZE) + 1;
if (sliceoffset >= attrsize)
{
sliceoffset = 0;
length = 0;
}
if (((sliceoffset + length) > attrsize) || length < 0)
length = attrsize - sliceoffset;
result = (varattrib *) palloc(length + VARHDRSZ);
SET_VARSIZE(result, length + VARHDRSZ);
if (VARATT_EXTERNAL_IS_COMPRESSED(attr))
VARATT_SET_COMPRESSED(result);
if (length == 0)
return (struct varlena *)result; /* Can save a lot of work at this point! */
startchunk = sliceoffset / TOAST_MAX_CHUNK_SIZE;
endchunk = (sliceoffset + length - 1) / TOAST_MAX_CHUNK_SIZE;
numchunks = (endchunk - startchunk) + 1;
startoffset = sliceoffset % TOAST_MAX_CHUNK_SIZE;
endoffset = (sliceoffset + length - 1) % TOAST_MAX_CHUNK_SIZE;
/*
* Open the toast relation and its index
*/
toastrel = heap_open(((varattrib *)attr)->va_external.va_toastrelid, AccessShareLock);
toasttupDesc = toastrel->rd_att;
toastidx = index_open(toastrel->rd_rel->reltoastidxid, AccessShareLock);
/*
* Setup a scan key to fetch from the index. This is either two keys or
* three depending on the number of chunks.
*/
ScanKeyInit(&toastkey[0],
(AttrNumber) 1,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(((varattrib *)attr)->va_external.va_valueid));
/*
* Use equality condition for one chunk, a range condition otherwise:
*/
if (numchunks == 1)
{
ScanKeyInit(&toastkey[1],
(AttrNumber) 2,
BTEqualStrategyNumber, F_INT4EQ,
Int32GetDatum(startchunk));
nscankeys = 2;
}
else
{
ScanKeyInit(&toastkey[1],
(AttrNumber) 2,
BTGreaterEqualStrategyNumber, F_INT4GE,
Int32GetDatum(startchunk));
ScanKeyInit(&toastkey[2],
(AttrNumber) 2,
BTLessEqualStrategyNumber, F_INT4LE,
Int32GetDatum(endchunk));
nscankeys = 3;
}
/*
* Read the chunks by index
*
* The index is on (valueid, chunkidx) so they will come in order
*/
nextidx = startchunk;
toastscan = index_beginscan(toastrel, toastidx,
SnapshotToast, nscankeys, toastkey);
while ((ttup = index_getnext(toastscan, ForwardScanDirection)) != NULL)
{
/*
* Have a chunk, extract the sequence number and the data
*/
residx = DatumGetInt32(fastgetattr(ttup, 2, toasttupDesc, &isnull));
Assert(!isnull);
chunk = DatumGetPointer(fastgetattr(ttup, 3, toasttupDesc, &isnull));
Assert(!isnull);
if (VARATT_IS_SHORT((varattrib *)chunk))
chunksize = VARSIZE_SHORT((varattrib *)chunk) - VARHDRSZ_SHORT;
else if (!VARATT_IS_EXTENDED((varattrib *)chunk))
chunksize = VARSIZE((varattrib *)chunk) - VARHDRSZ;
else {
elog(ERROR, "found toasted toast chunk?");
chunksize = 0; /* shut compiler up */
}
/*
* Some checks on the data we've found
*/
if ((residx != nextidx) || (residx > endchunk) || (residx < startchunk))
elog(ERROR, "unexpected chunk number %d (expected %d) for toast value %u",
residx, nextidx,
((varattrib *)attr)->va_external.va_valueid);
if (residx < totalchunks - 1)
{
if (chunksize != TOAST_MAX_CHUNK_SIZE)
elog(ERROR, "unexpected chunk size %d in chunk %d for toast value %u of %d when fetching slice (expected %d)",
chunksize, residx,
((varattrib *)attr)->va_external.va_valueid, totalchunks-1,
(int)TOAST_MAX_CHUNK_SIZE);
}
else if (residx == totalchunks-1)
{
if ((residx * TOAST_MAX_CHUNK_SIZE + chunksize) != attrsize)
elog(ERROR, "unexpected chunk size %d in chunk %d for final toast value %u when fetching slice (expected %d)",
chunksize, residx,
((varattrib *)attr)->va_external.va_valueid,
attrsize - residx * (int)TOAST_MAX_CHUNK_SIZE);
}
else
{
elog(ERROR, "unexpected chunk");
}
/*
* Copy the data into proper place in our result
*/
chcpystrt = 0;
chcpyend = chunksize - 1;
if (residx == startchunk)
chcpystrt = startoffset;
if (residx == endchunk)
chcpyend = endoffset;
memcpy(((char *) VARDATA(result)) +
(residx * TOAST_MAX_CHUNK_SIZE - sliceoffset) + chcpystrt,
VARDATA((varattrib *)chunk) + chcpystrt,
(chcpyend - chcpystrt) + 1);
nextidx++;
}
/*
* Final checks that we successfully fetched the datum
*/
if (nextidx != (endchunk + 1))
elog(ERROR, "missing chunk number %d for toast value %u",
nextidx,
((varattrib *)attr)->va_external.va_valueid);
/*
* End scan and close relations
*/
index_endscan(toastscan);
index_close(toastidx, AccessShareLock);
heap_close(toastrel, AccessShareLock);
return (struct varlena *)result;
}
/* ----------------------------------------------------------------
* IndexOnlyNext
*
* Retrieve a tuple from the IndexOnlyScan node's index.
* ----------------------------------------------------------------
*/
static TupleTableSlot *
IndexOnlyNext(IndexOnlyScanState *node)
{
EState *estate;
ExprContext *econtext;
ScanDirection direction;
IndexScanDesc scandesc;
TupleTableSlot *slot;
ItemPointer tid;
/*
* extract necessary information from index scan node
*/
estate = node->ss.ps.state;
direction = estate->es_direction;
/* flip direction if this is an overall backward scan */
if (ScanDirectionIsBackward(((IndexOnlyScan *) node->ss.ps.plan)->indexorderdir))
{
if (ScanDirectionIsForward(direction))
direction = BackwardScanDirection;
else if (ScanDirectionIsBackward(direction))
direction = ForwardScanDirection;
}
scandesc = node->ioss_ScanDesc;
econtext = node->ss.ps.ps_ExprContext;
slot = node->ss.ss_ScanTupleSlot;
if (scandesc == NULL)
{
/*
* We reach here if the index only scan is not parallel, or if we're
* serially executing an index only scan that was planned to be
* parallel.
*/
scandesc = index_beginscan(node->ss.ss_currentRelation,
node->ioss_RelationDesc,
estate->es_snapshot,
node->ioss_NumScanKeys,
node->ioss_NumOrderByKeys);
node->ioss_ScanDesc = scandesc;
/* Set it up for index-only scan */
node->ioss_ScanDesc->xs_want_itup = true;
node->ioss_VMBuffer = InvalidBuffer;
/*
* If no run-time keys to calculate or they are ready, go ahead and
* pass the scankeys to the index AM.
*/
if (node->ioss_NumRuntimeKeys == 0 || node->ioss_RuntimeKeysReady)
index_rescan(scandesc,
node->ioss_ScanKeys,
node->ioss_NumScanKeys,
node->ioss_OrderByKeys,
node->ioss_NumOrderByKeys);
}
/*
* OK, now that we have what we need, fetch the next tuple.
*/
while ((tid = index_getnext_tid(scandesc, direction)) != NULL)
{
HeapTuple tuple = NULL;
CHECK_FOR_INTERRUPTS();
/*
* We can skip the heap fetch if the TID references a heap page on
* which all tuples are known visible to everybody. In any case,
* we'll use the index tuple not the heap tuple as the data source.
*
* Note on Memory Ordering Effects: visibilitymap_get_status does not
* lock the visibility map buffer, and therefore the result we read
* here could be slightly stale. However, it can't be stale enough to
* matter.
*
* We need to detect clearing a VM bit due to an insert right away,
* because the tuple is present in the index page but not visible. The
* reading of the TID by this scan (using a shared lock on the index
* buffer) is serialized with the insert of the TID into the index
* (using an exclusive lock on the index buffer). Because the VM bit
* is cleared before updating the index, and locking/unlocking of the
* index page acts as a full memory barrier, we are sure to see the
* cleared bit if we see a recently-inserted TID.
*
* Deletes do not update the index page (only VACUUM will clear out
* the TID), so the clearing of the VM bit by a delete is not
* serialized with this test below, and we may see a value that is
* significantly stale. However, we don't care about the delete right
* away, because the tuple is still visible until the deleting
* transaction commits or the statement ends (if it's our
* transaction). In either case, the lock on the VM buffer will have
* been released (acting as a write barrier) after clearing the bit.
* And for us to have a snapshot that includes the deleting
* transaction (making the tuple invisible), we must have acquired
* ProcArrayLock after that time, acting as a read barrier.
*
* It's worth going through this complexity to avoid needing to lock
* the VM buffer, which could cause significant contention.
*/
if (!VM_ALL_VISIBLE(scandesc->heapRelation,
ItemPointerGetBlockNumber(tid),
&node->ioss_VMBuffer))
{
/*
* Rats, we have to visit the heap to check visibility.
*/
InstrCountTuples2(node, 1);
tuple = index_fetch_heap(scandesc);
if (tuple == NULL)
continue; /* no visible tuple, try next index entry */
/*
* Only MVCC snapshots are supported here, so there should be no
* need to keep following the HOT chain once a visible entry has
* been found. If we did want to allow that, we'd need to keep
* more state to remember not to call index_getnext_tid next time.
*/
if (scandesc->xs_continue_hot)
elog(ERROR, "non-MVCC snapshots are not supported in index-only scans");
/*
* Note: at this point we are holding a pin on the heap page, as
* recorded in scandesc->xs_cbuf. We could release that pin now,
* but it's not clear whether it's a win to do so. The next index
* entry might require a visit to the same heap page.
*/
}
/*
* Fill the scan tuple slot with data from the index. This might be
* provided in either HeapTuple or IndexTuple format. Conceivably an
* index AM might fill both fields, in which case we prefer the heap
* format, since it's probably a bit cheaper to fill a slot from.
*/
if (scandesc->xs_hitup)
{
/*
* We don't take the trouble to verify that the provided tuple has
* exactly the slot's format, but it seems worth doing a quick
* check on the number of fields.
*/
Assert(slot->tts_tupleDescriptor->natts ==
scandesc->xs_hitupdesc->natts);
ExecStoreHeapTuple(scandesc->xs_hitup, slot, false);
}
else if (scandesc->xs_itup)
StoreIndexTuple(slot, scandesc->xs_itup, scandesc->xs_itupdesc);
else
elog(ERROR, "no data returned for index-only scan");
/*
* If the index was lossy, we have to recheck the index quals.
* (Currently, this can never happen, but we should support the case
* for possible future use, eg with GiST indexes.)
*/
if (scandesc->xs_recheck)
{
econtext->ecxt_scantuple = slot;
if (!ExecQualAndReset(node->indexqual, econtext))
{
/* Fails recheck, so drop it and loop back for another */
InstrCountFiltered2(node, 1);
continue;
}
}
/*
* We don't currently support rechecking ORDER BY distances. (In
* principle, if the index can support retrieval of the originally
* indexed value, it should be able to produce an exact distance
* calculation too. So it's not clear that adding code here for
* recheck/re-sort would be worth the trouble. But we should at least
* throw an error if someone tries it.)
*/
if (scandesc->numberOfOrderBys > 0 && scandesc->xs_recheckorderby)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("lossy distance functions are not supported in index-only scans")));
/*
* Predicate locks for index-only scans must be acquired at the page
* level when the heap is not accessed, since tuple-level predicate
* locks need the tuple's xmin value. If we had to visit the tuple
* anyway, then we already have the tuple-level lock and can skip the
* page lock.
*/
if (tuple == NULL)
PredicateLockPage(scandesc->heapRelation,
ItemPointerGetBlockNumber(tid),
estate->es_snapshot);
return slot;
}
/*
* if we get here it means the index scan failed so we are at the end of
* the scan..
*/
return ExecClearTuple(slot);
}
/* ----------
* toast_fetch_datum_slice -
*
* Reconstruct a segment of a varattrib from the chunks saved
* in the toast relation
* ----------
*/
static varattrib *
toast_fetch_datum_slice(varattrib *attr, int32 sliceoffset, int32 length)
{
Relation toastrel;
Relation toastidx;
ScanKeyData toastkey[3];
int nscankeys;
IndexScanDesc toastscan;
HeapTuple ttup;
TupleDesc toasttupDesc;
varattrib *result;
int32 attrsize;
int32 residx;
int32 nextidx;
int numchunks;
int startchunk;
int endchunk;
int32 startoffset;
int32 endoffset;
int totalchunks;
Pointer chunk;
bool isnull;
int32 chunksize;
int32 chcpystrt;
int32 chcpyend;
attrsize = attr->va_content.va_external.va_extsize;
totalchunks = ((attrsize - 1) / TOAST_MAX_CHUNK_SIZE) + 1;
if (sliceoffset >= attrsize)
{
sliceoffset = 0;
length = 0;
}
if (((sliceoffset + length) > attrsize) || length < 0)
length = attrsize - sliceoffset;
result = (varattrib *) palloc(length + VARHDRSZ);
VARATT_SIZEP(result) = length + VARHDRSZ;
if (VARATT_IS_COMPRESSED(attr))
VARATT_SIZEP(result) |= VARATT_FLAG_COMPRESSED;
if (length == 0)
return (result); /* Can save a lot of work at this point! */
startchunk = sliceoffset / TOAST_MAX_CHUNK_SIZE;
endchunk = (sliceoffset + length - 1) / TOAST_MAX_CHUNK_SIZE;
numchunks = (endchunk - startchunk) + 1;
startoffset = sliceoffset % TOAST_MAX_CHUNK_SIZE;
endoffset = (sliceoffset + length - 1) % TOAST_MAX_CHUNK_SIZE;
/*
* Open the toast relation and it's index
*/
toastrel = heap_open(attr->va_content.va_external.va_toastrelid,
AccessShareLock);
toasttupDesc = toastrel->rd_att;
toastidx = index_open(toastrel->rd_rel->reltoastidxid);
/*
* Setup a scan key to fetch from the index. This is either two keys
* or three depending on the number of chunks.
*/
ScanKeyEntryInitialize(&toastkey[0],
(bits16) 0,
(AttrNumber) 1,
(RegProcedure) F_OIDEQ,
ObjectIdGetDatum(attr->va_content.va_external.va_valueid));
/*
* Now dependent on number of chunks:
*/
if (numchunks == 1)
{
ScanKeyEntryInitialize(&toastkey[1],
(bits16) 0,
(AttrNumber) 2,
(RegProcedure) F_INT4EQ,
Int32GetDatum(startchunk));
nscankeys = 2;
}
else
{
ScanKeyEntryInitialize(&toastkey[1],
(bits16) 0,
(AttrNumber) 2,
(RegProcedure) F_INT4GE,
Int32GetDatum(startchunk));
ScanKeyEntryInitialize(&toastkey[2],
(bits16) 0,
(AttrNumber) 2,
(RegProcedure) F_INT4LE,
Int32GetDatum(endchunk));
nscankeys = 3;
}
/*
* Read the chunks by index
*
* The index is on (valueid, chunkidx) so they will come in order
*/
nextidx = startchunk;
toastscan = index_beginscan(toastrel, toastidx, SnapshotToast,
nscankeys, toastkey);
while ((ttup = index_getnext(toastscan, ForwardScanDirection)) != NULL)
{
/*
* Have a chunk, extract the sequence number and the data
*/
residx = DatumGetInt32(heap_getattr(ttup, 2, toasttupDesc, &isnull));
Assert(!isnull);
chunk = DatumGetPointer(heap_getattr(ttup, 3, toasttupDesc, &isnull));
Assert(!isnull);
chunksize = VARATT_SIZE(chunk) - VARHDRSZ;
/*
* Some checks on the data we've found
*/
if ((residx != nextidx) || (residx > endchunk) || (residx < startchunk))
elog(ERROR, "unexpected chunk number %d (expected %d) for toast value %u",
residx, nextidx,
attr->va_content.va_external.va_valueid);
if (residx < totalchunks - 1)
{
if (chunksize != TOAST_MAX_CHUNK_SIZE)
elog(ERROR, "unexpected chunk size %d in chunk %d for toast value %u",
chunksize, residx,
attr->va_content.va_external.va_valueid);
}
else
{
if ((residx * TOAST_MAX_CHUNK_SIZE + chunksize) != attrsize)
elog(ERROR, "unexpected chunk size %d in chunk %d for toast value %u",
chunksize, residx,
attr->va_content.va_external.va_valueid);
}
/*
* Copy the data into proper place in our result
*/
chcpystrt = 0;
chcpyend = chunksize - 1;
if (residx == startchunk)
chcpystrt = startoffset;
if (residx == endchunk)
chcpyend = endoffset;
memcpy(((char *) VARATT_DATA(result)) +
(residx * TOAST_MAX_CHUNK_SIZE - sliceoffset) + chcpystrt,
VARATT_DATA(chunk) + chcpystrt,
(chcpyend - chcpystrt) + 1);
nextidx++;
}
/*
* Final checks that we successfully fetched the datum
*/
if (nextidx != (endchunk + 1))
elog(ERROR, "missing chunk number %d for toast value %u",
nextidx,
attr->va_content.va_external.va_valueid);
/*
* End scan and close relations
*/
index_endscan(toastscan);
index_close(toastidx);
heap_close(toastrel, AccessShareLock);
return result;
}
/*
* Bulk deletion of all index entries pointing to a set of heap tuples.
* The set of target tuples is specified via a callback routine that tells
* whether any given heap tuple (identified by ItemPointer) is being deleted.
*
* Result: a palloc'd struct containing statistical info for VACUUM displays.
*/
Datum
rtbulkdelete(PG_FUNCTION_ARGS)
{
Relation rel = (Relation) PG_GETARG_POINTER(0);
IndexBulkDeleteCallback callback = (IndexBulkDeleteCallback) PG_GETARG_POINTER(1);
void *callback_state = (void *) PG_GETARG_POINTER(2);
IndexBulkDeleteResult *result;
BlockNumber num_pages;
double tuples_removed;
double num_index_tuples;
IndexScanDesc iscan;
tuples_removed = 0;
num_index_tuples = 0;
/*
* Since rtree is not marked "amconcurrent" in pg_am, caller should have
* acquired exclusive lock on index relation. We need no locking here.
*/
/*
* XXX generic implementation --- should be improved!
*/
/* walk through the entire index */
iscan = index_beginscan(NULL, rel, SnapshotAny, 0, NULL);
/* including killed tuples */
iscan->ignore_killed_tuples = false;
while (index_getnext_indexitem(iscan, ForwardScanDirection))
{
vacuum_delay_point();
if (callback(&iscan->xs_ctup.t_self, callback_state))
{
ItemPointerData indextup = iscan->currentItemData;
BlockNumber blkno;
OffsetNumber offnum;
Buffer buf;
Page page;
blkno = ItemPointerGetBlockNumber(&indextup);
offnum = ItemPointerGetOffsetNumber(&indextup);
/* adjust any scans that will be affected by this deletion */
/* (namely, my own scan) */
rtadjscans(rel, RTOP_DEL, blkno, offnum);
/* delete the index tuple */
buf = ReadBuffer(rel, blkno);
page = BufferGetPage(buf);
PageIndexTupleDelete(page, offnum);
WriteBuffer(buf);
tuples_removed += 1;
}
else
num_index_tuples += 1;
}
index_endscan(iscan);
/* return statistics */
num_pages = RelationGetNumberOfBlocks(rel);
result = (IndexBulkDeleteResult *) palloc0(sizeof(IndexBulkDeleteResult));
result->num_pages = num_pages;
result->num_index_tuples = num_index_tuples;
result->tuples_removed = tuples_removed;
PG_RETURN_POINTER(result);
}
/* ----------------------------------------------------------------
* ExecInitIndexScan
*
* Initializes the index scan's state information, creates
* scan keys, and opens the base and index relations.
*
* Note: index scans have 2 sets of state information because
* we have to keep track of the base relation and the
* index relation.
* ----------------------------------------------------------------
*/
IndexScanState *
ExecInitIndexScan(IndexScan *node, EState *estate, int eflags)
{
IndexScanState *indexstate;
Relation currentRelation;
bool relistarget;
/*
* create state structure
*/
indexstate = makeNode(IndexScanState);
indexstate->ss.ps.plan = (Plan *) node;
indexstate->ss.ps.state = estate;
/*
* Miscellaneous initialization
*
* create expression context for node
*/
ExecAssignExprContext(estate, &indexstate->ss.ps);
indexstate->ss.ps.ps_TupFromTlist = false;
/*
* initialize child expressions
*
* Note: we don't initialize all of the indexqual expression, only the
* sub-parts corresponding to runtime keys (see below). Likewise for
* indexorderby, if any. But the indexqualorig expression is always
* initialized even though it will only be used in some uncommon cases ---
* would be nice to improve that. (Problem is that any SubPlans present
* in the expression must be found now...)
*/
indexstate->ss.ps.targetlist = (List *)
ExecInitExpr((Expr *) node->scan.plan.targetlist,
(PlanState *) indexstate);
indexstate->ss.ps.qual = (List *)
ExecInitExpr((Expr *) node->scan.plan.qual,
(PlanState *) indexstate);
indexstate->indexqualorig = (List *)
ExecInitExpr((Expr *) node->indexqualorig,
(PlanState *) indexstate);
/*
* tuple table initialization
*/
ExecInitResultTupleSlot(estate, &indexstate->ss.ps);
ExecInitScanTupleSlot(estate, &indexstate->ss);
/*
* open the base relation and acquire appropriate lock on it.
*/
currentRelation = ExecOpenScanRelation(estate, node->scan.scanrelid);
indexstate->ss.ss_currentRelation = currentRelation;
indexstate->ss.ss_currentScanDesc = NULL; /* no heap scan here */
/*
* get the scan type from the relation descriptor.
*/
ExecAssignScanType(&indexstate->ss, RelationGetDescr(currentRelation));
/*
* Initialize result tuple type and projection info.
*/
ExecAssignResultTypeFromTL(&indexstate->ss.ps);
ExecAssignScanProjectionInfo(&indexstate->ss);
/*
* If we are just doing EXPLAIN (ie, aren't going to run the plan), stop
* here. This allows an index-advisor plugin to EXPLAIN a plan containing
* references to nonexistent indexes.
*/
if (eflags & EXEC_FLAG_EXPLAIN_ONLY)
return indexstate;
/*
* Open the index relation.
*
* If the parent table is one of the target relations of the query, then
* InitPlan already opened and write-locked the index, so we can avoid
* taking another lock here. Otherwise we need a normal reader's lock.
*/
relistarget = ExecRelationIsTargetRelation(estate, node->scan.scanrelid);
indexstate->iss_RelationDesc = index_open(node->indexid,
relistarget ? NoLock : AccessShareLock);
/*
* Initialize index-specific scan state
*/
indexstate->iss_RuntimeKeysReady = false;
indexstate->iss_RuntimeKeys = NULL;
indexstate->iss_NumRuntimeKeys = 0;
/*
* build the index scan keys from the index qualification
*/
ExecIndexBuildScanKeys((PlanState *) indexstate,
indexstate->iss_RelationDesc,
node->scan.scanrelid,
node->indexqual,
false,
&indexstate->iss_ScanKeys,
&indexstate->iss_NumScanKeys,
&indexstate->iss_RuntimeKeys,
&indexstate->iss_NumRuntimeKeys,
NULL, /* no ArrayKeys */
NULL);
/*
* any ORDER BY exprs have to be turned into scankeys in the same way
*/
ExecIndexBuildScanKeys((PlanState *) indexstate,
indexstate->iss_RelationDesc,
node->scan.scanrelid,
node->indexorderby,
true,
&indexstate->iss_OrderByKeys,
&indexstate->iss_NumOrderByKeys,
&indexstate->iss_RuntimeKeys,
&indexstate->iss_NumRuntimeKeys,
NULL, /* no ArrayKeys */
NULL);
/*
* If we have runtime keys, we need an ExprContext to evaluate them. The
* node's standard context won't do because we want to reset that context
* for every tuple. So, build another context just like the other one...
* -tgl 7/11/00
*/
if (indexstate->iss_NumRuntimeKeys != 0)
{
ExprContext *stdecontext = indexstate->ss.ps.ps_ExprContext;
ExecAssignExprContext(estate, &indexstate->ss.ps);
indexstate->iss_RuntimeContext = indexstate->ss.ps.ps_ExprContext;
indexstate->ss.ps.ps_ExprContext = stdecontext;
}
else
{
indexstate->iss_RuntimeContext = NULL;
}
/*
* Initialize scan descriptor.
*/
indexstate->iss_ScanDesc = index_beginscan(currentRelation,
indexstate->iss_RelationDesc,
estate->es_snapshot,
indexstate->iss_NumScanKeys,
indexstate->iss_NumOrderByKeys);
/*
* If no run-time keys to calculate, go ahead and pass the scankeys to the
* index AM.
*/
if (indexstate->iss_NumRuntimeKeys == 0)
index_rescan(indexstate->iss_ScanDesc,
indexstate->iss_ScanKeys, indexstate->iss_NumScanKeys,
indexstate->iss_OrderByKeys, indexstate->iss_NumOrderByKeys);
/*
* all done.
*/
return indexstate;
}
/*
* This function initializes a part and returns true if a new index has been prepared for scanning.
*/
static bool
initNextIndexToScan(DynamicIndexScanState *node)
{
IndexScanState *indexState = &(node->indexScanState);
DynamicIndexScan *dynamicIndexScan = (DynamicIndexScan *)node->indexScanState.ss.ps.plan;
/* Load new index when the scanning of the previous index is done. */
if (indexState->ss.scan_state == SCAN_INIT ||
indexState->ss.scan_state == SCAN_DONE)
{
/* This is the oid of a partition of the table (*not* index) */
Oid *pid = hash_seq_search(&node->pidxStatus);
if (pid == NULL)
{
/* Return if all parts have been scanned. */
node->shouldCallHashSeqTerm = false;
return false;
}
/* Collect number of partitions scanned in EXPLAIN ANALYZE */
if(NULL != indexState->ss.ps.instrument)
{
Instrumentation *instr = indexState->ss.ps.instrument;
instr->numPartScanned ++;
}
DynamicIndexScan_ReMapColumns(node, *pid);
/*
* The is the oid of the partition of an *index*. Note: a partitioned table
* has a root and a set of partitions (may be multi-level). An index
* on a partitioned table also has a root and a set of index partitions.
* We started at table level, and now we are fetching the oid of an index
* partition.
*/
Oid pindex = getPhysicalIndexRelid(dynamicIndexScan->logicalIndexInfo,
*pid);
Assert(OidIsValid(pindex));
Relation currentRelation = OpenScanRelationByOid(*pid);
indexState->ss.ss_currentRelation = currentRelation;
for (int i=0; i < DYNAMICINDEXSCAN_NSLOTS; i++)
{
indexState->ss.ss_ScanTupleSlot[i].tts_tableOid = *pid;
}
ExecAssignScanType(&indexState->ss, RelationGetDescr(currentRelation));
ScanState *scanState = (ScanState *)node;
MemoryContextReset(node->partitionMemoryContext);
MemoryContext oldCxt = MemoryContextSwitchTo(node->partitionMemoryContext);
/* Initialize child expressions */
scanState->ps.qual = (List *)ExecInitExpr((Expr *)scanState->ps.plan->qual, (PlanState*)scanState);
scanState->ps.targetlist = (List *)ExecInitExpr((Expr *)scanState->ps.plan->targetlist, (PlanState*)scanState);
ExecAssignScanProjectionInfo(scanState);
EState *estate = indexState->ss.ps.state;
indexState->iss_RelationDesc =
OpenIndexRelation(estate, pindex, *pid);
/*
* build the index scan keys from the index qualification
*/
ExecIndexBuildScanKeys((PlanState *) indexState,
indexState->iss_RelationDesc,
dynamicIndexScan->indexqual,
dynamicIndexScan->indexstrategy,
dynamicIndexScan->indexsubtype,
&indexState->iss_ScanKeys,
&indexState->iss_NumScanKeys,
&indexState->iss_RuntimeKeys,
&indexState->iss_NumRuntimeKeys,
NULL,
NULL);
MemoryContextSwitchTo(oldCxt);
ExprContext *econtext = indexState->iss_RuntimeContext; /* context for runtime keys */
if (indexState->iss_NumRuntimeKeys != 0)
{
ExecIndexEvalRuntimeKeys(econtext,
indexState->iss_RuntimeKeys,
indexState->iss_NumRuntimeKeys);
}
indexState->iss_RuntimeKeysReady = true;
/*
* Initialize result tuple type and projection info.
*/
TupleDesc td = indexState->ss.ps.ps_ResultTupleSlot->tts_tupleDescriptor;
if (td)
{
pfree(td);
td = NULL;
}
ExecAssignResultTypeFromTL(&indexState->ss.ps);
ExecAssignScanProjectionInfo(&indexState->ss);
indexState->iss_ScanDesc = index_beginscan(currentRelation,
indexState->iss_RelationDesc,
estate->es_snapshot,
indexState->iss_NumScanKeys,
indexState->iss_ScanKeys);
indexState->ss.scan_state = SCAN_SCAN;
}
return true;
}
