BRIN de-summarization

When the BRIN summary tuple for a page range becomes too "wide" for the
values actually stored in the table (because the tuples that were
present originally are no longer present due to updates or deletes), it
can be useful to remove the outdated summary tuple, so that a future
summarization can install a tighter summary.

This commit introduces a SQL-callable interface to do so.

Author: Álvaro Herrera
Reviewed-by: Eiji Seki
Discussion: https://postgr.es/m/20170228045643.n2ri74ara4fhhfxf@alvherre.pgsql

doc/src/sgml/brin.sgml

@@ -80,6 +80,10 @@
    or by automatic summarization executed by autovacuum, as insertions
    occur.  (This last trigger is disabled by default and can be enabled
    with the <literal>autosummarize</literal> parameter.)
+   Conversely, a range can be de-summarized using the
+   <function>brin_desummarize_range(regclass, bigint)</function> range,
+   which is useful when the index tuple is no longer a very good
+   representation because the existing values have changed.
   </para>
 
  </sect2>

doc/src/sgml/func.sgml

@@ -19660,6 +19660,14 @@ postgres=# SELECT * FROM pg_walfile_name_offset(pg_stop_backup());
     <primary>gin_clean_pending_list</primary>
    </indexterm>
 
+   <indexterm>
+    <primary>brin_summarize_range</primary>
+   </indexterm>
+
+   <indexterm>
+    <primary>brin_desummarize_range</primary>
+   </indexterm>
+
    <para>
     <xref linkend="functions-admin-index-table"> shows the functions
     available for index maintenance tasks.
@@ -19690,6 +19698,13 @@ postgres=# SELECT * FROM pg_walfile_name_offset(pg_stop_backup());
        <entry><type>integer</type></entry>
        <entry>summarize the page range covering the given block, if not already summarized</entry>
       </row>
+      <row>
+       <entry>
+        <literal><function>brin_desummarize_range(<parameter>index</> <type>regclass</>, <parameter>blockNumber</> <type>bigint</type>)</function></literal>
+       </entry>
+       <entry><type>integer</type></entry>
+       <entry>de-summarize the page range covering the given block, if summarized</entry>
+      </row>
       <row>
        <entry>
         <literal><function>gin_clean_pending_list(<parameter>index</> <type>regclass</>)</function></literal>

src/backend/access/brin/brin.c

@@ -908,6 +908,80 @@ brin_summarize_range(PG_FUNCTION_ARGS)
 	PG_RETURN_INT32((int32) numSummarized);
 }
 
+/*
+ * SQL-callable interface to mark a range as no longer summarized
+ */
+Datum
+brin_desummarize_range(PG_FUNCTION_ARGS)
+{
+	Oid		indexoid = PG_GETARG_OID(0);
+	int64	heapBlk64 = PG_GETARG_INT64(1);
+	BlockNumber heapBlk;
+	Oid		heapoid;
+	Relation heapRel;
+	Relation indexRel;
+	bool	done;
+
+	if (heapBlk64 > MaxBlockNumber || heapBlk64 < 0)
+	{
+		char	   *blk = psprintf(INT64_FORMAT, heapBlk64);
+
+		ereport(ERROR,
+				(errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE),
+				 errmsg("block number out of range: %s", blk)));
+	}
+	heapBlk = (BlockNumber) heapBlk64;
+
+	/*
+	 * We must lock table before index to avoid deadlocks.  However, if the
+	 * passed indexoid isn't an index then IndexGetRelation() will fail.
+	 * Rather than emitting a not-very-helpful error message, postpone
+	 * complaining, expecting that the is-it-an-index test below will fail.
+	 */
+	heapoid = IndexGetRelation(indexoid, true);
+	if (OidIsValid(heapoid))
+		heapRel = heap_open(heapoid, ShareUpdateExclusiveLock);
+	else
+		heapRel = NULL;
+
+	indexRel = index_open(indexoid, ShareUpdateExclusiveLock);
+
+	/* Must be a BRIN index */
+	if (indexRel->rd_rel->relkind != RELKIND_INDEX ||
+		indexRel->rd_rel->relam != BRIN_AM_OID)
+		ereport(ERROR,
+				(errcode(ERRCODE_WRONG_OBJECT_TYPE),
+				 errmsg("\"%s\" is not a BRIN index",
+						RelationGetRelationName(indexRel))));
+
+	/* User must own the index (comparable to privileges needed for VACUUM) */
+	if (!pg_class_ownercheck(indexoid, GetUserId()))
+		aclcheck_error(ACLCHECK_NOT_OWNER, ACL_KIND_CLASS,
+					   RelationGetRelationName(indexRel));
+
+	/*
+	 * Since we did the IndexGetRelation call above without any lock, it's
+	 * barely possible that a race against an index drop/recreation could have
+	 * netted us the wrong table.  Recheck.
+	 */
+	if (heapRel == NULL || heapoid != IndexGetRelation(indexoid, false))
+		ereport(ERROR,
+				(errcode(ERRCODE_UNDEFINED_TABLE),
+				 errmsg("could not open parent table of index %s",
+						RelationGetRelationName(indexRel))));
+
+	/* the revmap does the hard work */
+	do {
+		done = brinRevmapDesummarizeRange(indexRel, heapBlk);
+	}
+	while (!done);
+
+	relation_close(indexRel, ShareUpdateExclusiveLock);
+	relation_close(heapRel, ShareUpdateExclusiveLock);
+
+	PG_RETURN_VOID();
+}
+
 /*
  * Build a BrinDesc used to create or scan a BRIN index
  */

src/backend/access/brin/brin_revmap.c

@@ -168,9 +168,12 @@ brinSetHeapBlockItemptr(Buffer buf, BlockNumber pagesPerRange,
 	iptr = (ItemPointerData *) contents->rm_tids;
 	iptr += HEAPBLK_TO_REVMAP_INDEX(pagesPerRange, heapBlk);
 
-	ItemPointerSet(iptr,
-				   ItemPointerGetBlockNumber(&tid),
-				   ItemPointerGetOffsetNumber(&tid));
+	if (ItemPointerIsValid(&tid))
+		ItemPointerSet(iptr,
+					   ItemPointerGetBlockNumber(&tid),
+					   ItemPointerGetOffsetNumber(&tid));
+	else
+		ItemPointerSetInvalid(iptr);
 }
 
 /*
@@ -304,6 +307,137 @@ brinGetTupleForHeapBlock(BrinRevmap *revmap, BlockNumber heapBlk,
 	return NULL;
 }
 
+/*
+ * Delete an index tuple, marking a page range as unsummarized.
+ *
+ * Index must be locked in ShareUpdateExclusiveLock mode.
+ *
+ * Return FALSE if caller should retry.
+ */
+bool
+brinRevmapDesummarizeRange(Relation idxrel, BlockNumber heapBlk)
+{
+	BrinRevmap *revmap;
+	BlockNumber	pagesPerRange;
+	RevmapContents *contents;
+	ItemPointerData *iptr;
+	ItemPointerData	invalidIptr;
+	BlockNumber	revmapBlk;
+	Buffer		revmapBuf;
+	Buffer		regBuf;
+	Page		revmapPg;
+	Page		regPg;
+	OffsetNumber revmapOffset;
+	OffsetNumber regOffset;
+	ItemId		lp;
+	BrinTuple  *tup;
+
+	revmap = brinRevmapInitialize(idxrel, &pagesPerRange, NULL);
+
+	revmapBlk = revmap_get_blkno(revmap, heapBlk);
+	if (!BlockNumberIsValid(revmapBlk))
+	{
+		/* revmap page doesn't exist: range not summarized, we're done */
+		brinRevmapTerminate(revmap);
+		return true;
+	}
+
+	/* Lock the revmap page, obtain the index tuple pointer from it */
+	revmapBuf = brinLockRevmapPageForUpdate(revmap, heapBlk);
+	revmapPg = BufferGetPage(revmapBuf);
+	revmapOffset = HEAPBLK_TO_REVMAP_INDEX(revmap->rm_pagesPerRange, heapBlk);
+
+	contents = (RevmapContents *) PageGetContents(revmapPg);
+	iptr = contents->rm_tids;
+	iptr += revmapOffset;
+
+	if (!ItemPointerIsValid(iptr))
+	{
+		/* no index tuple: range not summarized, we're done */
+		LockBuffer(revmapBuf, BUFFER_LOCK_UNLOCK);
+		brinRevmapTerminate(revmap);
+		return true;
+	}
+
+	regBuf = ReadBuffer(idxrel, ItemPointerGetBlockNumber(iptr));
+	LockBuffer(regBuf, BUFFER_LOCK_EXCLUSIVE);
+	regPg = BufferGetPage(regBuf);
+
+	/* if this is no longer a regular page, tell caller to start over */
+	if (!BRIN_IS_REGULAR_PAGE(regPg))
+	{
+		LockBuffer(revmapBuf, BUFFER_LOCK_UNLOCK);
+		LockBuffer(regBuf, BUFFER_LOCK_UNLOCK);
+		brinRevmapTerminate(revmap);
+		return false;
+	}
+
+	regOffset = ItemPointerGetOffsetNumber(iptr);
+	if (regOffset > PageGetMaxOffsetNumber(regPg))
+		ereport(ERROR,
+				(errcode(ERRCODE_INDEX_CORRUPTED),
+				 errmsg("corrupted BRIN index: inconsistent range map")));
+
+	lp = PageGetItemId(regPg, regOffset);
+	if (!ItemIdIsUsed(lp))
+		ereport(ERROR,
+				(errcode(ERRCODE_INDEX_CORRUPTED),
+				 errmsg("corrupted BRIN index: inconsistent range map")));
+	tup = (BrinTuple *) PageGetItem(regPg, lp);
+	/* XXX apply sanity checks?  Might as well delete a bogus tuple ... */
+
+	/*
+	 * We're only removing data, not reading it, so there's no need to
+	 * TestForOldSnapshot here.
+	 */
+
+	/*
+	 * Because of SUE lock, this function shouldn't run concurrently with
+	 * summarization.  Placeholder tuples can only exist as leftovers from
+	 * crashed summarization, so if we detect any, we complain but proceed.
+	 */
+	if (BrinTupleIsPlaceholder(tup))
+		ereport(WARNING,
+				(errmsg("leftover placeholder tuple detected in BRIN index \"%s\", deleting",
+						RelationGetRelationName(idxrel))));
+
+	START_CRIT_SECTION();
+
+	ItemPointerSetInvalid(&invalidIptr);
+	brinSetHeapBlockItemptr(revmapBuf, revmap->rm_pagesPerRange, heapBlk,
+							invalidIptr);
+	PageIndexTupleDeleteNoCompact(regPg, regOffset);
+	/* XXX record free space in FSM? */
+
+	MarkBufferDirty(regBuf);
+	MarkBufferDirty(revmapBuf);
+
+	if (RelationNeedsWAL(idxrel))
+	{
+		xl_brin_desummarize xlrec;
+		XLogRecPtr		recptr;
+
+		xlrec.heapBlk = heapBlk;
+		xlrec.regOffset = regOffset;
+
+		XLogBeginInsert();
+		XLogRegisterData((char *) &xlrec, SizeOfBrinDesummarize);
+		XLogRegisterBuffer(0, revmapBuf, 0);
+		XLogRegisterBuffer(1, regBuf, REGBUF_STANDARD);
+		recptr = XLogInsert(RM_BRIN_ID, XLOG_BRIN_DESUMMARIZE);
+		PageSetLSN(revmapPg, recptr);
+		PageSetLSN(regPg, recptr);
+	}
+
+	END_CRIT_SECTION();
+
+	UnlockReleaseBuffer(regBuf);
+	LockBuffer(revmapBuf, BUFFER_LOCK_UNLOCK);
+	brinRevmapTerminate(revmap);
+
+	return true;
+}
+
 /*
  * Given a heap block number, find the corresponding physical revmap block
  * number and return it.  If the revmap page hasn't been allocated yet, return

src/backend/access/brin/brin_xlog.c

@@ -254,6 +254,46 @@ brin_xlog_revmap_extend(XLogReaderState *record)
 		UnlockReleaseBuffer(metabuf);
 }
 
+static void
+brin_xlog_desummarize_page(XLogReaderState *record)
+{
+	XLogRecPtr	lsn = record->EndRecPtr;
+	xl_brin_desummarize *xlrec;
+	Buffer		buffer;
+	XLogRedoAction action;
+
+	xlrec = (xl_brin_desummarize *) XLogRecGetData(record);
+
+	/* Update the revmap */
+	action = XLogReadBufferForRedo(record, 0, &buffer);
+	if (action == BLK_NEEDS_REDO)
+	{
+		ItemPointerData	iptr;
+
+		ItemPointerSetInvalid(&iptr);
+		brinSetHeapBlockItemptr(buffer, xlrec->pagesPerRange, xlrec->heapBlk, iptr);
+
+		PageSetLSN(BufferGetPage(buffer), lsn);
+		MarkBufferDirty(buffer);
+	}
+	if (BufferIsValid(buffer))
+		UnlockReleaseBuffer(buffer);
+
+	/* remove the leftover entry from the regular page */
+	action = XLogReadBufferForRedo(record, 1, &buffer);
+	if (action == BLK_NEEDS_REDO)
+	{
+		Page	regPg = BufferGetPage(buffer);
+
+		PageIndexTupleDeleteNoCompact(regPg, xlrec->regOffset);
+
+		PageSetLSN(regPg, lsn);
+		MarkBufferDirty(buffer);
+	}
+	if (BufferIsValid(buffer))
+		UnlockReleaseBuffer(buffer);
+}
+
 void
 brin_redo(XLogReaderState *record)
 {
@@ -276,6 +316,9 @@ brin_redo(XLogReaderState *record)
 		case XLOG_BRIN_REVMAP_EXTEND:
 			brin_xlog_revmap_extend(record);
 			break;
+		case XLOG_BRIN_DESUMMARIZE:
+			brin_xlog_desummarize_page(record);
+			break;
 		default:
 			elog(PANIC, "brin_redo: unknown op code %u", info);
 	}

src/backend/access/rmgrdesc/brindesc.c

@@ -61,6 +61,13 @@ brin_desc(StringInfo buf, XLogReaderState *record)
 
 		appendStringInfo(buf, "targetBlk %u", xlrec->targetBlk);
 	}
+	else if (info == XLOG_BRIN_DESUMMARIZE)
+	{
+		xl_brin_desummarize *xlrec = (xl_brin_desummarize *) rec;
+
+		appendStringInfo(buf, "pagesPerRange %u, heapBlk %u, page offset %u",
+						 xlrec->pagesPerRange, xlrec->heapBlk, xlrec->regOffset);
+	}
 }
 
 const char *
@@ -91,6 +98,9 @@ brin_identify(uint8 info)
 		case XLOG_BRIN_REVMAP_EXTEND:
 			id = "REVMAP_EXTEND";
 			break;
+		case XLOG_BRIN_DESUMMARIZE:
+			id = "DESUMMARIZE";
+			break;
 	}
 
 	return id;

src/include/access/brin_revmap.h

@@ -36,5 +36,6 @@ extern void brinSetHeapBlockItemptr(Buffer rmbuf, BlockNumber pagesPerRange,
 extern BrinTuple *brinGetTupleForHeapBlock(BrinRevmap *revmap,
 						 BlockNumber heapBlk, Buffer *buf, OffsetNumber *off,
 						 Size *size, int mode, Snapshot snapshot);
+extern bool brinRevmapDesummarizeRange(Relation idxrel, BlockNumber heapBlk);
 
 #endif   /* BRIN_REVMAP_H */

src/include/access/brin_xlog.h

@@ -33,7 +33,7 @@
 #define XLOG_BRIN_UPDATE			0x20
 #define XLOG_BRIN_SAMEPAGE_UPDATE	0x30
 #define XLOG_BRIN_REVMAP_EXTEND		0x40
-#define XLOG_BRIN_REVMAP_VACUUM		0x50
+#define XLOG_BRIN_DESUMMARIZE		0x50
 
 #define XLOG_BRIN_OPMASK			0x70
 /*
@@ -124,6 +124,24 @@ typedef struct xl_brin_revmap_extend
 #define SizeOfBrinRevmapExtend	(offsetof(xl_brin_revmap_extend, targetBlk) + \
 								 sizeof(BlockNumber))
 
+/*
+ * This is what we need to know about a range de-summarization
+ *
+ * Backup block 0: revmap page
+ * Backup block 1: regular page
+ */
+typedef struct xl_brin_desummarize
+{
+	BlockNumber	pagesPerRange;
+	/* page number location to set to invalid */
+	OffsetNumber heapBlk;
+	/* offset of item to delete in regular index page */
+	OffsetNumber regOffset;
+} xl_brin_desummarize;
+
+#define SizeOfBrinDesummarize	(offsetof(xl_brin_desummarize, regOffset) + \
+								 sizeof(OffsetNumber))
+
 
 extern void brin_redo(XLogReaderState *record);
 extern void brin_desc(StringInfo buf, XLogReaderState *record);

src/include/catalog/catversion.h

@@ -53,6 +53,6 @@
  */
 
 /*							yyyymmddN */
-#define CATALOG_VERSION_NO	201704011
+#define CATALOG_VERSION_NO	201704012
 
 #endif

src/include/catalog/pg_proc.h

@@ -566,6 +566,8 @@ DATA(insert OID = 3952 (  brin_summarize_new_values PGNSP PGUID 12 1 0 0 0 f f f
 DESCR("brin: standalone scan new table pages");
 DATA(insert OID = 3999 (  brin_summarize_range PGNSP PGUID 12 1 0 0 0 f f f f t f v s 2 0 23 "2205 20" _null_ _null_ _null_ _null_ _null_ brin_summarize_range _null_ _null_ _null_ ));
 DESCR("brin: standalone scan new table pages");
+DATA(insert OID = 4014 (  brin_desummarize_range PGNSP PGUID 12 1 0 0 0 f f f f t f v s 2 0 2278 "2205 20" _null_ _null_ _null_ _null_ _null_ brin_desummarize_range _null_ _null_ _null_ ));
+DESCR("brin: desummarize page range");
 
 DATA(insert OID = 338 (  amvalidate		PGNSP PGUID 12 1 0 0 0 f f f f t f v s 1 0 16 "26" _null_ _null_ _null_ _null_ _null_	amvalidate _null_ _null_ _null_ ));
 DESCR("validate an operator class");

src/test/regress/expected/brin.out

@@ -392,6 +392,12 @@ INSERT INTO brintest SELECT
 	format('%s/%s%s', odd, even, tenthous)::pg_lsn,
 	box(point(odd, even), point(thousand, twothousand))
 FROM tenk1 ORDER BY unique2 LIMIT 5 OFFSET 5;
+SELECT brin_desummarize_range('brinidx', 0);
+ brin_desummarize_range 
+------------------------
+ 
+(1 row)
+
 VACUUM brintest;  -- force a summarization cycle in brinidx
 UPDATE brintest SET int8col = int8col * int4col;
 UPDATE brintest SET textcol = '' WHERE textcol IS NOT NULL;
@@ -406,6 +412,27 @@ SELECT brin_summarize_new_values('brinidx'); -- ok, no change expected
                          0
 (1 row)
 
+-- Tests for brin_desummarize_range
+SELECT brin_desummarize_range('brinidx', -1); -- error, invalid range
+ERROR:  block number out of range: -1
+SELECT brin_desummarize_range('brinidx', 0);
+ brin_desummarize_range 
+------------------------
+ 
+(1 row)
+
+SELECT brin_desummarize_range('brinidx', 0);
+ brin_desummarize_range 
+------------------------
+ 
+(1 row)
+
+SELECT brin_desummarize_range('brinidx', 100000000);
+ brin_desummarize_range 
+------------------------
+ 
+(1 row)
+
 -- Test brin_summarize_range
 CREATE TABLE brin_summarize (
     value int

src/test/regress/sql/brin.sql

@@ -400,6 +400,7 @@ INSERT INTO brintest SELECT
 	box(point(odd, even), point(thousand, twothousand))
 FROM tenk1 ORDER BY unique2 LIMIT 5 OFFSET 5;
 
+SELECT brin_desummarize_range('brinidx', 0);
 VACUUM brintest;  -- force a summarization cycle in brinidx
 
 UPDATE brintest SET int8col = int8col * int4col;
@@ -410,6 +411,12 @@ SELECT brin_summarize_new_values('brintest'); -- error, not an index
 SELECT brin_summarize_new_values('tenk1_unique1'); -- error, not a BRIN index
 SELECT brin_summarize_new_values('brinidx'); -- ok, no change expected
 
+-- Tests for brin_desummarize_range
+SELECT brin_desummarize_range('brinidx', -1); -- error, invalid range
+SELECT brin_desummarize_range('brinidx', 0);
+SELECT brin_desummarize_range('brinidx', 0);
+SELECT brin_desummarize_range('brinidx', 100000000);
+
 -- Test brin_summarize_range
 CREATE TABLE brin_summarize (
     value int