Extend relations multiple blocks at a time to improve scalability.

Contention on the relation extension lock can become quite fierce when
multiple processes are inserting data into the same relation at the same
time at a high rate.  Experimentation shows the extending the relation
multiple blocks at a time improves scalability.

Dilip Kumar, reviewed by Petr Jelinek, Amit Kapila, and me.

3 files changed, 153 insertions, 0 deletions

diff --git a/src/backend/storage/freespace/freespace.c b/src/backend/storage/freespace/freespace.c
index 813990ea707..2ffa8ff24d0 100644
--- a/src/backend/storage/freespace/freespace.c
+++ b/src/backend/storage/freespace/freespace.c
@@ -109,6 +109,8 @@ static int fsm_set_and_search(Relation rel, FSMAddress addr, uint16 slot,
 				   uint8 newValue, uint8 minValue);
 static BlockNumber fsm_search(Relation rel, uint8 min_cat);
 static uint8 fsm_vacuum_page(Relation rel, FSMAddress addr, bool *eof);
+static BlockNumber fsm_get_lastblckno(Relation rel, FSMAddress addr);
+static void fsm_update_recursive(Relation rel, FSMAddress addr, uint8 new_cat);
 
 
 /******** Public API ********/
@@ -189,6 +191,46 @@ RecordPageWithFreeSpace(Relation rel, BlockNumber heapBlk, Size spaceAvail)
 }
 
 /*
+ * Update the upper levels of the free space map all the way up to the root
+ * to make sure we don't lose track of new blocks we just inserted.  This is
+ * intended to be used after adding many new blocks to the relation; we judge
+ * it not worth updating the upper levels of the tree every time data for
+ * a single page changes, but for a bulk-extend it's worth it.
+ */
+void
+UpdateFreeSpaceMap(Relation rel, BlockNumber startBlkNum,
+					BlockNumber endBlkNum, Size freespace)
+{
+	int			new_cat = fsm_space_avail_to_cat(freespace);
+	FSMAddress	addr;
+	uint16		slot;
+	BlockNumber	blockNum;
+	BlockNumber	lastBlkOnPage;
+
+	blockNum = startBlkNum;
+
+	while (blockNum <= endBlkNum)
+	{
+		/*
+		 * Find FSM address for this block; update tree all the way to the
+		 * root.
+		 */
+		addr = fsm_get_location(blockNum, &slot);
+		fsm_update_recursive(rel, addr, new_cat);
+
+		/*
+		 * Get the last block number on this FSM page.  If that's greater
+		 * than or equal to our endBlkNum, we're done.  Otherwise, advance
+		 * to the first block on the next page.
+		 */
+		lastBlkOnPage = fsm_get_lastblckno(rel, addr);
+		if (lastBlkOnPage >= endBlkNum)
+			break;
+		blockNum = lastBlkOnPage + 1;
+	}
+}
+
+/*
  * XLogRecordPageWithFreeSpace - like RecordPageWithFreeSpace, for use in
  *		WAL replay
  */
@@ -788,3 +830,42 @@ fsm_vacuum_page(Relation rel, FSMAddress addr, bool *eof_p)
 
 	return max_avail;
 }
+
+/*
+ * This function will return the last block number stored on given
+ * FSM page address.
+ */
+static BlockNumber
+fsm_get_lastblckno(Relation rel, FSMAddress addr)
+{
+	int			slot;
+
+	/*
+	 * Get the last slot number on the given address and convert that to
+	 * block number
+	 */
+	slot = SlotsPerFSMPage - 1;
+	return fsm_get_heap_blk(addr, slot);
+}
+
+/*
+ * Recursively update the FSM tree from given address to
+ * all the way up to root.
+ */
+static void
+fsm_update_recursive(Relation rel, FSMAddress addr, uint8 new_cat)
+{
+	uint16		parentslot;
+	FSMAddress	parent;
+
+	if (addr.level == FSM_ROOT_LEVEL)
+		return;
+
+	/*
+	 * Get the parent page and our slot in the parent page, and
+	 * update the information in that.
+	 */
+	parent = fsm_get_parent(addr, &parentslot);
+	fsm_set_and_search(rel, parent, parentslot, new_cat, 0);
+	fsm_update_recursive(rel, parent, new_cat);
+}
diff --git a/src/backend/storage/lmgr/lmgr.c b/src/backend/storage/lmgr/lmgr.c
index 0632fc009e5..7b08555b071 100644
--- a/src/backend/storage/lmgr/lmgr.c
+++ b/src/backend/storage/lmgr/lmgr.c
@@ -341,6 +341,41 @@ LockRelationForExtension(Relation relation, LOCKMODE lockmode)
 }
 
 /*
+ *		ConditionalLockRelationForExtension
+ *
+ * As above, but only lock if we can get the lock without blocking.
+ * Returns TRUE iff the lock was acquired.
+ */
+bool
+ConditionalLockRelationForExtension(Relation relation, LOCKMODE lockmode)
+{
+	LOCKTAG		tag;
+
+	SET_LOCKTAG_RELATION_EXTEND(tag,
+								relation->rd_lockInfo.lockRelId.dbId,
+								relation->rd_lockInfo.lockRelId.relId);
+
+	return (LockAcquire(&tag, lockmode, false, true) != LOCKACQUIRE_NOT_AVAIL);
+}
+
+/*
+ *		RelationExtensionLockWaiterCount
+ *
+ * Count the number of processes waiting for the given relation extension lock.
+ */
+int
+RelationExtensionLockWaiterCount(Relation relation)
+{
+	LOCKTAG		tag;
+
+	SET_LOCKTAG_RELATION_EXTEND(tag,
+								relation->rd_lockInfo.lockRelId.dbId,
+								relation->rd_lockInfo.lockRelId.relId);
+
+	return LockWaiterCount(&tag);
+}
+
+/*
  *		UnlockRelationForExtension
  */
 void
diff --git a/src/backend/storage/lmgr/lock.c b/src/backend/storage/lmgr/lock.c
index b30b7b1009b..41f69306459 100644
--- a/src/backend/storage/lmgr/lock.c
+++ b/src/backend/storage/lmgr/lock.c
@@ -4380,3 +4380,40 @@ VirtualXactLock(VirtualTransactionId vxid, bool wait)
 	LockRelease(&tag, ShareLock, false);
 	return true;
 }
+
+/*
+ * LockWaiterCount
+ *
+ * Find the number of lock requester on this locktag
+ */
+int
+LockWaiterCount(const LOCKTAG *locktag)
+{
+	LOCKMETHODID lockmethodid = locktag->locktag_lockmethodid;
+	LOCK	   *lock;
+	bool		found;
+	uint32		hashcode;
+	LWLock	   *partitionLock;
+	int			waiters = 0;
+
+	if (lockmethodid <= 0 || lockmethodid >= lengthof(LockMethods))
+		elog(ERROR, "unrecognized lock method: %d", lockmethodid);
+
+	hashcode = LockTagHashCode(locktag);
+	partitionLock = LockHashPartitionLock(hashcode);
+	LWLockAcquire(partitionLock, LW_EXCLUSIVE);
+
+	lock = (LOCK *) hash_search_with_hash_value(LockMethodLockHash,
+												(const void *) locktag,
+												hashcode,
+												HASH_FIND,
+												&found);
+	if (found)
+	{
+		Assert(lock != NULL);
+		waiters = lock->nRequested;
+	}
+	LWLockRelease(partitionLock);
+
+	return waiters;
+}