Fix platform-specific performance regression in logtape.c.

jeff-davis · jeff-davis · commit 1fbb6c93df30 · 2020-06-07T09:25:55.000-07:00
Commit 24d8595 made a change that indirectly caused a performance regression by triggering a change in the way GCC optimizes memcpy() on some platforms. The behavior seemed to contradict a GCC document, so I filed a report: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95556 This patch implements a narrow workaround which eliminates the regression I observed. The workaround is benign enough that it seems unlikely to cause a different regression on another platform. Discussion: https://postgr.es/m/99b2eab335c1592c925d8143979c8e9e81e1575f.camel@j-davis.com
diff --git a/src/backend/utils/sort/logtape.c b/src/backend/utils/sort/logtape.c
@@ -782,7 +782,7 @@ LogicalTapeWrite(LogicalTapeSet *lts, int tapenum,
 	Assert(lt->buffer_size == BLCKSZ);
 	while (size > 0)
 	{
-		if (lt->pos >= TapeBlockPayloadSize)
+		if (lt->pos >= (int) TapeBlockPayloadSize)
 		{
 			/* Buffer full, dump it out */
 			long		nextBlockNumber;

Original file line number	Diff line number	Diff line change
`@@ -782,7 +782,7 @@ LogicalTapeWrite(LogicalTapeSet *lts, int tapenum,`
`782`	`782`	`Assert(lt->buffer_size == BLCKSZ);`
`783`	`783`	`while (size > 0)`
`784`	`784`	`{`
`785`		`- if (lt->pos >= TapeBlockPayloadSize)`
	`785`	`+ if (lt->pos >= (int) TapeBlockPayloadSize)`
`786`	`786`	`{`
`787`	`787`	`/* Buffer full, dump it out */`
`788`	`788`	`long nextBlockNumber;`