/*
* pgstat_heap -- returns live/dead tuples info in a heap
*/
static Datum
pgstat_heap(Relation rel, FunctionCallInfo fcinfo)
{
HeapScanDesc scan;
HeapTuple tuple;
BlockNumber nblocks;
BlockNumber block = 0; /* next block to count free space in */
BlockNumber tupblock;
Buffer buffer;
pgstattuple_type stat = {0};
SnapshotData SnapshotDirty;
/* Disable syncscan because we assume we scan from block zero upwards */
scan = heap_beginscan_strat(rel, SnapshotAny, 0, NULL, true, false);
InitDirtySnapshot(SnapshotDirty);
nblocks = scan->rs_nblocks; /* # blocks to be scanned */
/* scan the relation */
while ((tuple = heap_getnext(scan, ForwardScanDirection)) != NULL)
{
CHECK_FOR_INTERRUPTS();
/* must hold a buffer lock to call HeapTupleSatisfiesVisibility */
LockBuffer(scan->rs_cbuf, BUFFER_LOCK_SHARE);
if (HeapTupleSatisfiesVisibility(tuple, &SnapshotDirty, scan->rs_cbuf))
{
stat.tuple_len += tuple->t_len;
stat.tuple_count++;
}
else
{
stat.dead_tuple_len += tuple->t_len;
stat.dead_tuple_count++;
}
LockBuffer(scan->rs_cbuf, BUFFER_LOCK_UNLOCK);
/*
* To avoid physically reading the table twice, try to do the
* free-space scan in parallel with the heap scan. However,
* heap_getnext may find no tuples on a given page, so we cannot
* simply examine the pages returned by the heap scan.
*/
tupblock = BlockIdGetBlockNumber(&tuple->t_self.ip_blkid);
while (block <= tupblock)
{
CHECK_FOR_INTERRUPTS();
buffer = ReadBufferExtended(rel, MAIN_FORKNUM, block,
RBM_NORMAL, scan->rs_strategy);
LockBuffer(buffer, BUFFER_LOCK_SHARE);
stat.free_space += PageGetHeapFreeSpace((Page) BufferGetPage(buffer));
UnlockReleaseBuffer(buffer);
block++;
}
}
while (block < nblocks)
{
CHECK_FOR_INTERRUPTS();
buffer = ReadBufferExtended(rel, MAIN_FORKNUM, block,
RBM_NORMAL, scan->rs_strategy);
LockBuffer(buffer, BUFFER_LOCK_SHARE);
stat.free_space += PageGetHeapFreeSpace((Page) BufferGetPage(buffer));
UnlockReleaseBuffer(buffer);
block++;
}
heap_endscan(scan);
relation_close(rel, AccessShareLock);
stat.table_len = (uint64) nblocks *BLCKSZ;
return build_pgstattuple_type(&stat, fcinfo);
}
/*
* Search the relation 'rel' for tuple using the sequential scan.
*
* If a matching tuple is found, lock it with lockmode, fill the slot with its
* contents, and return true. Return false otherwise.
*
* Note that this stops on the first matching tuple.
*
* This can obviously be quite slow on tables that have more than few rows.
*/
bool
RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
TupleTableSlot *searchslot, TupleTableSlot *outslot)
{
HeapTuple scantuple;
HeapScanDesc scan;
SnapshotData snap;
TransactionId xwait;
bool found;
TupleDesc desc = RelationGetDescr(rel);
Assert(equalTupleDescs(desc, outslot->tts_tupleDescriptor));
/* Start an index scan. */
InitDirtySnapshot(snap);
scan = heap_beginscan(rel, &snap, 0, NULL);
retry:
found = false;
heap_rescan(scan, NULL);
/* Try to find the tuple */
while ((scantuple = heap_getnext(scan, ForwardScanDirection)) != NULL)
{
if (!tuple_equals_slot(desc, scantuple, searchslot))
continue;
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;
}
}
heap_endscan(scan);
return found;
}
/*
* 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;
}
/*
* 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;
}
bool
CheckNewRelFileNodeIsOk(Oid newOid, Oid reltablespace, bool relisshared,
Relation pg_class)
{
RelFileNode rnode;
char *rpath;
int fd;
bool collides;
SnapshotData SnapshotDirty;
InitDirtySnapshot(SnapshotDirty);
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;
}
