slru-replacement-fix.patch

diff --git a/src/backend/access/transam/slru.c b/src/backend/access/transam/slru.c
index 3049e01..63fd4f2 100644
--- a/src/backend/access/transam/slru.c
+++ b/src/backend/access/transam/slru.c
@@ -403,12 +403,6 @@ SimpleLruReadPage(SlruCtl ctl, int pageno, bool write_ok,
 		/* Acquire per-buffer lock (cannot deadlock, see notes at top) */
 		LWLockAcquire(shared->buffer_locks[slotno], LW_EXCLUSIVE);
 
-		/*
-		 * Temporarily mark page as recently-used to discourage
-		 * SlruSelectLRUPage from selecting it again for someone else.
-		 */
-		SlruRecentlyUsed(shared, slotno);
-
 		/* Release control lock while doing I/O */
 		LWLockRelease(shared->ControlLock);
 
@@ -909,9 +903,12 @@ SlruSelectLRUPage(SlruCtl ctl, int pageno)
 	{
 		int			slotno;
 		int			cur_count;
-		int			bestslot;
-		int			best_delta;
-		int			best_page_number;
+		int			bestvalidslot = 0;				/* keep compiler quiet */
+		int			best_valid_delta = -1;
+		int			best_valid_page_number = 0;		/* keep compiler quiet */
+		int			bestinvalidslot = 0;			/* keep compiler quiet */
+		int			best_invalid_delta = -1;
+		int			best_invalid_page_number = 0;	/* keep compiler quiet */
 
 		/* See if page already has a buffer assigned */
 		for (slotno = 0; slotno < shared->num_slots; slotno++)
@@ -933,7 +930,8 @@ SlruSelectLRUPage(SlruCtl ctl, int pageno)
 		 * choosing the furthest-back page.
 		 *
 		 * In no case will we select the slot containing latest_page_number
-		 * for replacement, even if it appears least recently used.
+		 * for replacement, even if it appears least recently used.  We also
+		 * try hard to avoid selecting a page on which an I/O is in progress.
 		 *
 		 * Notice that this next line forcibly advances cur_lru_count to a
 		 * value that is certainly beyond any value that will be in the
@@ -944,9 +942,6 @@ SlruSelectLRUPage(SlruCtl ctl, int pageno)
 		 * multiple pages with the same lru_count.
 		 */
 		cur_count = (shared->cur_lru_count)++;
-		best_delta = -1;
-		bestslot = 0;			/* no-op, just keeps compiler quiet */
-		best_page_number = 0;	/* ditto */
 		for (slotno = 0; slotno < shared->num_slots; slotno++)
 		{
 			int			this_delta;
@@ -968,34 +963,57 @@ SlruSelectLRUPage(SlruCtl ctl, int pageno)
 				this_delta = 0;
 			}
 			this_page_number = shared->page_number[slotno];
-			if ((this_delta > best_delta ||
-				 (this_delta == best_delta &&
-				  ctl->PagePrecedes(this_page_number, best_page_number))) &&
-				this_page_number != shared->latest_page_number)
+			if (this_page_number == shared->latest_page_number)
+				continue;
+			if (shared->page_status[slotno] == SLRU_PAGE_VALID)
+			{
+				if (this_delta > best_valid_delta ||
+					(this_delta == best_valid_delta &&
+					 ctl->PagePrecedes(this_page_number,
+									   best_valid_page_number)))
+				{
+					bestvalidslot = slotno;
+					best_valid_delta = this_delta;
+					best_valid_page_number = this_page_number;
+				}
+			}
+			else
 			{
-				bestslot = slotno;
-				best_delta = this_delta;
-				best_page_number = this_page_number;
+				if (this_delta > best_invalid_delta ||
+					(this_delta == best_invalid_delta &&
+					 ctl->PagePrecedes(this_page_number,
+									   best_invalid_page_number)))
+				{
+					bestinvalidslot = slotno;
+					best_invalid_delta = this_delta;
+					best_invalid_page_number = this_page_number;
+				}
 			}
 		}
 
 		/*
+		 * All pages (except possibly the latest one) are I/O busy. We'll have
+		 * to wait for an I/O to complete and then retry.  We choose to wait
+		 * for the I/O on the least recently used slot, on the assumption that
+		 * it was likely initiated first of all the I/Os in progress and may
+		 * therefore finish first.
+		 */
+		if (best_valid_delta < 0)
+		{
+			SimpleLruWaitIO(ctl, bestinvalidslot);
+			continue;
+		}
+
+		/*
 		 * If the selected page is clean, we're set.
 		 */
-		if (shared->page_status[bestslot] == SLRU_PAGE_VALID &&
-			!shared->page_dirty[bestslot])
-			return bestslot;
+		if (!shared->page_dirty[bestvalidslot])
+			return bestvalidslot;
 
 		/*
-		 * We need to wait for I/O.  Normal case is that it's dirty and we
-		 * must initiate a write, but it's possible that the page is already
-		 * write-busy, or in the worst case still read-busy.  In those cases
-		 * we wait for the existing I/O to complete.
+		 * Write the page.
 		 */
-		if (shared->page_status[bestslot] == SLRU_PAGE_VALID)
-			SlruInternalWritePage(ctl, bestslot, NULL);
-		else
-			SimpleLruWaitIO(ctl, bestslot);
+		SlruInternalWritePage(ctl, bestvalidslot, NULL);
 
 		/*
 		 * Now loop back and try again.  This is the easiest way of dealing