logtape.c: do not preallocate for tapes when sorting

The preallocation logic is only useful for HashAgg, so disable it when
sorting.

Also, adjust an out-of-date comment.

Reviewed-by: Peter Geoghegan
Discussion: https://postgr.es/m/CAH2-Wzn_o7tE2+hRVvwSFghRb75AJ5g-nqGzDUqLYMexjOAe=g@mail.gmail.com
Backpatch-through: 13

src/backend/executor/nodeAgg.c

@@ -2882,7 +2882,7 @@ hashagg_tapeinfo_init(AggState *aggstate)
 	HashTapeInfo *tapeinfo = palloc(sizeof(HashTapeInfo));
 	int			init_tapes = 16;	/* expanded dynamically */
 
-	tapeinfo->tapeset = LogicalTapeSetCreate(init_tapes, NULL, NULL, -1);
+	tapeinfo->tapeset = LogicalTapeSetCreate(init_tapes, true, NULL, NULL, -1);
 	tapeinfo->ntapes = init_tapes;
 	tapeinfo->nfreetapes = init_tapes;
 	tapeinfo->freetapes_alloc = init_tapes;

src/backend/utils/sort/logtape.c

@@ -210,6 +210,7 @@ struct LogicalTapeSet
 	long	   *freeBlocks;		/* resizable array holding minheap */
 	long		nFreeBlocks;	/* # of currently free blocks */
 	Size		freeBlocksLen;	/* current allocated length of freeBlocks[] */
+	bool		enable_prealloc;	/* preallocate write blocks? */
 
 	/* The array of logical tapes. */
 	int			nTapes;			/* # of logical tapes in set */
@@ -218,6 +219,7 @@ struct LogicalTapeSet
 
 static void ltsWriteBlock(LogicalTapeSet *lts, long blocknum, void *buffer);
 static void ltsReadBlock(LogicalTapeSet *lts, long blocknum, void *buffer);
+static long ltsGetBlock(LogicalTapeSet *lts, LogicalTape *lt);
 static long ltsGetFreeBlock(LogicalTapeSet *lts);
 static long ltsGetPreallocBlock(LogicalTapeSet *lts, LogicalTape *lt);
 static void ltsReleaseBlock(LogicalTapeSet *lts, long blocknum);
@@ -240,12 +242,8 @@ ltsWriteBlock(LogicalTapeSet *lts, long blocknum, void *buffer)
 	 * that's past the current end of file, fill the space between the current
 	 * end of file and the target block with zeros.
 	 *
-	 * This should happen rarely, otherwise you are not writing very
+	 * This can happen either when tapes preallocate blocks; or for the last
-	 * sequentially.  In current use, this only happens when the sort ends
+	 * block of a tape which might not have been flushed.
-	 * writing a run, and switches to another tape.  The last block of the
-	 * previous tape isn't flushed to disk until the end of the sort, so you
-	 * get one-block hole, where the last block of the previous tape will
-	 * later go.
 	 *
 	 * Note that BufFile concatenation can leave "holes" in BufFile between
 	 * worker-owned block ranges.  These are tracked for reporting purposes
@@ -371,8 +369,20 @@ parent_offset(unsigned long i)
 }
 
 /*
- * Select the lowest currently unused block by taking the first element from
+ * Get the next block for writing.
- * the freelist min heap.
+ */
+static long
+ltsGetBlock(LogicalTapeSet *lts, LogicalTape *lt)
+{
+	if (lts->enable_prealloc)
+		return ltsGetPreallocBlock(lts, lt);
+	else
+		return ltsGetFreeBlock(lts);
+}
+
+/*
+ * Select the lowest currently unused block from the tape set's global free
+ * list min heap.
  */
 static long
 ltsGetFreeBlock(LogicalTapeSet *lts)
@@ -428,7 +438,8 @@ ltsGetFreeBlock(LogicalTapeSet *lts)
 
 /*
  * Return the lowest free block number from the tape's preallocation list.
- * Refill the preallocation list if necessary.
+ * Refill the preallocation list with blocks from the tape set's free list if
+ * necessary.
  */
 static long
 ltsGetPreallocBlock(LogicalTapeSet *lts, LogicalTape *lt)
@@ -669,8 +680,8 @@ ltsInitReadBuffer(LogicalTapeSet *lts, LogicalTape *lt)
  * infrastructure that may be lifted in the future.
  */
 LogicalTapeSet *
-LogicalTapeSetCreate(int ntapes, TapeShare *shared, SharedFileSet *fileset,
+LogicalTapeSetCreate(int ntapes, bool preallocate, TapeShare *shared,
-					 int worker)
+					 SharedFileSet *fileset, int worker)
 {
 	LogicalTapeSet *lts;
 	int			i;
@@ -687,6 +698,7 @@ LogicalTapeSetCreate(int ntapes, TapeShare *shared, SharedFileSet *fileset,
 	lts->freeBlocksLen = 32;	/* reasonable initial guess */
 	lts->freeBlocks = (long *) palloc(lts->freeBlocksLen * sizeof(long));
 	lts->nFreeBlocks = 0;
+	lts->enable_prealloc = preallocate;
 	lts->nTapes = ntapes;
 	lts->tapes = (LogicalTape *) palloc(ntapes * sizeof(LogicalTape));
 
@@ -780,7 +792,7 @@ LogicalTapeWrite(LogicalTapeSet *lts, int tapenum,
 		Assert(lt->firstBlockNumber == -1);
 		Assert(lt->pos == 0);
 
-		lt->curBlockNumber = ltsGetPreallocBlock(lts, lt);
+		lt->curBlockNumber = ltsGetBlock(lts, lt);
 		lt->firstBlockNumber = lt->curBlockNumber;
 
 		TapeBlockGetTrailer(lt->buffer)->prev = -1L;
@@ -804,7 +816,7 @@ LogicalTapeWrite(LogicalTapeSet *lts, int tapenum,
 			 * First allocate the next block, so that we can store it in the
 			 * 'next' pointer of this block.
 			 */
-			nextBlockNumber = ltsGetPreallocBlock(lts, lt);
+			nextBlockNumber = ltsGetBlock(lts, lt);
 
 			/* set the next-pointer and dump the current block. */
 			TapeBlockGetTrailer(lt->buffer)->next = nextBlockNumber;

src/backend/utils/sort/tuplesort.c

@@ -2591,7 +2591,7 @@ inittapes(Tuplesortstate *state, bool mergeruns)
 	/* Create the tape set and allocate the per-tape data arrays */
 	inittapestate(state, maxTapes);
 	state->tapeset =
-		LogicalTapeSetCreate(maxTapes, NULL,
+		LogicalTapeSetCreate(maxTapes, false, NULL,
 							 state->shared ? &state->shared->fileset : NULL,
 							 state->worker);
 
@@ -4657,8 +4657,9 @@ leader_takeover_tapes(Tuplesortstate *state)
 	 * randomAccess is disallowed for parallel sorts.
 	 */
 	inittapestate(state, nParticipants + 1);
-	state->tapeset = LogicalTapeSetCreate(nParticipants + 1, shared->tapes,
+	state->tapeset = LogicalTapeSetCreate(nParticipants + 1, false,
-										  &shared->fileset, state->worker);
+										  shared->tapes, &shared->fileset,
+										  state->worker);
 
 	/* mergeruns() relies on currentRun for # of runs (in one-pass cases) */
 	state->currentRun = nParticipants;

src/include/utils/logtape.h

@@ -54,7 +54,8 @@ typedef struct TapeShare
  * prototypes for functions in logtape.c
  */
 
-extern LogicalTapeSet *LogicalTapeSetCreate(int ntapes, TapeShare *shared,
+extern LogicalTapeSet *LogicalTapeSetCreate(int ntapes, bool preallocate,
+											TapeShare *shared,
 											SharedFileSet *fileset, int worker);
 extern void LogicalTapeSetClose(LogicalTapeSet *lts);
 extern void LogicalTapeSetForgetFreeSpace(LogicalTapeSet *lts);