Arrange to squeeze out the MINIMAL_TUPLE_PADDING in the tuple representation

tglsfdc · tglsfdc · commit d26bf23f3448 · 2008-10-28T15:51:03.000Z
written to temp files by tuplesort.c and tuplestore.c.  This saves 2 bytes per
row for 32-bit machines, and 6 bytes per row for 64-bit machines, which seems
worth the slight additional uglification of the tuple read/write routines.
diff --git a/src/backend/utils/sort/tuplesort.c b/src/backend/utils/sort/tuplesort.c
@@ -91,7 +91,7 @@
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/utils/sort/tuplesort.c,v 1.87 2008/09/15 18:43:41 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/utils/sort/tuplesort.c,v 1.88 2008/10/28 15:51:03 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -2632,18 +2632,20 @@ copytup_heap(Tuplesortstate *state, SortTuple *stup, void *tup)
 								&stup->isnull1);
 }
 
-/*
- * Since MinimalTuple already has length in its first word, we don't need
- * to write that separately.
- */
 static void
 writetup_heap(Tuplesortstate *state, int tapenum, SortTuple *stup)
 {
 	MinimalTuple tuple = (MinimalTuple) stup->tuple;
-	unsigned int tuplen = tuple->t_len;
+	/* the part of the MinimalTuple we'll write: */
+	char	   *tupbody = (char *) tuple + MINIMAL_TUPLE_DATA_OFFSET;
+	unsigned int tupbodylen = tuple->t_len - MINIMAL_TUPLE_DATA_OFFSET;
+	/* total on-disk footprint: */
+	unsigned int tuplen = tupbodylen + sizeof(int);
 
 	LogicalTapeWrite(state->tapeset, tapenum,
-					 (void *) tuple, tuplen);
+					 (void *) &tuplen, sizeof(tuplen));
+	LogicalTapeWrite(state->tapeset, tapenum,
+					 (void *) tupbody, tupbodylen);
 	if (state->randomAccess)	/* need trailing length word? */
 		LogicalTapeWrite(state->tapeset, tapenum,
 						 (void *) &tuplen, sizeof(tuplen));
@@ -2656,16 +2658,18 @@ static void
 readtup_heap(Tuplesortstate *state, SortTuple *stup,
 			 int tapenum, unsigned int len)
 {
-	MinimalTuple tuple = (MinimalTuple) palloc(len);
-	unsigned int tuplen;
+	unsigned int tupbodylen = len - sizeof(int);
+	unsigned int tuplen = tupbodylen + MINIMAL_TUPLE_DATA_OFFSET;
+	MinimalTuple tuple = (MinimalTuple) palloc(tuplen);
+	char	   *tupbody = (char *) tuple + MINIMAL_TUPLE_DATA_OFFSET;
 	HeapTupleData htup;
 
 	USEMEM(state, GetMemoryChunkSpace(tuple));
 	/* read in the tuple proper */
-	tuple->t_len = len;
+	tuple->t_len = tuplen;
 	if (LogicalTapeRead(state->tapeset, tapenum,
-						(void *) ((char *) tuple + sizeof(int)),
-						len - sizeof(int)) != (size_t) (len - sizeof(int)))
+						(void *) tupbody,
+						tupbodylen) != (size_t) tupbodylen)
 		elog(ERROR, "unexpected end of data");
 	if (state->randomAccess)	/* need trailing length word? */
 		if (LogicalTapeRead(state->tapeset, tapenum, (void *) &tuplen,
diff --git a/src/backend/utils/sort/tuplestore.c b/src/backend/utils/sort/tuplestore.c
@@ -46,7 +46,7 @@
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/utils/sort/tuplestore.c,v 1.42 2008/10/07 00:05:55 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/utils/sort/tuplestore.c,v 1.43 2008/10/28 15:51:03 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -1173,9 +1173,17 @@ static void
 writetup_heap(Tuplestorestate *state, void *tup)
 {
 	MinimalTuple tuple = (MinimalTuple) tup;
-	unsigned int tuplen = tuple->t_len;
-
-	if (BufFileWrite(state->myfile, (void *) tuple, tuplen) != (size_t) tuplen)
+	/* the part of the MinimalTuple we'll write: */
+	char	   *tupbody = (char *) tuple + MINIMAL_TUPLE_DATA_OFFSET;
+	unsigned int tupbodylen = tuple->t_len - MINIMAL_TUPLE_DATA_OFFSET;
+	/* total on-disk footprint: */
+	unsigned int tuplen = tupbodylen + sizeof(int);
+
+	if (BufFileWrite(state->myfile, (void *) &tuplen,
+					 sizeof(tuplen)) != sizeof(tuplen))
+		elog(ERROR, "write failed");
+	if (BufFileWrite(state->myfile, (void *) tupbody,
+					 tupbodylen) != (size_t) tupbodylen)
 		elog(ERROR, "write failed");
 	if (state->backward)		/* need trailing length word? */
 		if (BufFileWrite(state->myfile, (void *) &tuplen,
@@ -1189,14 +1197,16 @@ writetup_heap(Tuplestorestate *state, void *tup)
 static void *
 readtup_heap(Tuplestorestate *state, unsigned int len)
 {
-	MinimalTuple tuple = (MinimalTuple) palloc(len);
-	unsigned int tuplen;
+	unsigned int tupbodylen = len - sizeof(int);
+	unsigned int tuplen = tupbodylen + MINIMAL_TUPLE_DATA_OFFSET;
+	MinimalTuple tuple = (MinimalTuple) palloc(tuplen);
+	char	   *tupbody = (char *) tuple + MINIMAL_TUPLE_DATA_OFFSET;
 
 	USEMEM(state, GetMemoryChunkSpace(tuple));
 	/* read in the tuple proper */
-	tuple->t_len = len;
-	if (BufFileRead(state->myfile, (void *) ((char *) tuple + sizeof(int)),
-					len - sizeof(int)) != (size_t) (len - sizeof(int)))
+	tuple->t_len = tuplen;
+	if (BufFileRead(state->myfile, (void *) tupbody,
+					 tupbodylen) != (size_t) tupbodylen)
 		elog(ERROR, "unexpected end of data");
 	if (state->backward)		/* need trailing length word? */
 		if (BufFileRead(state->myfile, (void *) &tuplen,
diff --git a/src/include/access/htup.h b/src/include/access/htup.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2008, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/htup.h,v 1.101 2008/08/11 11:05:11 heikki Exp $
+ * $PostgreSQL: pgsql/src/include/access/htup.h,v 1.102 2008/10/28 15:51:03 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -420,11 +420,17 @@ do { \
  *
  * Note that t_hoff is computed the same as in a full tuple, hence it includes
  * the MINIMAL_TUPLE_OFFSET distance.  t_len does not include that, however.
+ *
+ * MINIMAL_TUPLE_DATA_OFFSET is the offset to the first useful (non-pad) data
+ * other than the length word.  tuplesort.c and tuplestore.c use this to avoid
+ * writing the padding to disk.
  */
 #define MINIMAL_TUPLE_OFFSET \
 	((offsetof(HeapTupleHeaderData, t_infomask2) - sizeof(uint32)) / MAXIMUM_ALIGNOF * MAXIMUM_ALIGNOF)
 #define MINIMAL_TUPLE_PADDING \
 	((offsetof(HeapTupleHeaderData, t_infomask2) - sizeof(uint32)) % MAXIMUM_ALIGNOF)
+#define MINIMAL_TUPLE_DATA_OFFSET \
+	offsetof(MinimalTupleData, t_infomask2)
 
 typedef struct MinimalTupleData
 {
