Avoid parallel nbtree index scan hangs with SAOPs.

petergeoghegan · petergeoghegan · commit a24bffc021d9 · 2024-09-17T11:10:33.000-04:00
Commit 5bf748b, which enhanced nbtree ScalarArrayOp execution, made parallel index scans work with the new design for arrays via explicit scheduling of primitive index scans. A backend that successfully scheduled the scan's next primitive index scan saved its backend local array keys in shared memory. Any backend could pick up the scheduled primitive scan within _bt_first. This scheme decouples scheduling a primitive scan from starting the scan (by performing another descent of the index via a _bt_search call from _bt_first) to make things robust. The scheme had a deadlock hazard, at least when the leader process participated in the scan. _bt_parallel_seize had a code path that made backends that were not in an immediate position to start a scheduled primitive index scan wait for some other backend to do so instead. Under the right circumstances, the leader process could wait here forever: the leader would wait for any other backend to start the primitive scan, while every worker was busy waiting on the leader to consume tuples from the scan's tuple queue. To fix, don't wait for a scheduled primitive index scan to be started by some other eligible backend from within _bt_parallel_seize (when the calling backend isn't in a position to do so itself). Return false instead, while recording that the scan has a scheduled primitive index scan in backend local state. This leaves the backend in the same state as the existing case where a backend schedules (or tries to schedule) another primitive index scan from within _bt_advance_array_keys, before calling _bt_parallel_seize. _bt_parallel_seize already handles that case by returning false without waiting, and without unsetting the backend local state. Leaving the backend in this state enables it to start a previously scheduled primitive index scan once it gets back to _bt_first. Oversight in commit 5bf748b, which enhanced nbtree ScalarArrayOp execution. Matthias van de Meent, with tweaks by me. Author: Matthias van de Meent <boekewurm+postgres@gmail.com> Reported-By: Tomas Vondra <tomas@vondra.me> Reviewed-By: Peter Geoghegan <pg@bowt.ie> Discussion: https://postgr.es/m/CAH2-WzmMGaPa32u9x_FvEbPTUkP5e95i=QxR8054nvCRydP-sw@mail.gmail.com Backpatch: 17-, where nbtree SAOP execution was enhanced.
diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c
@@ -585,7 +585,10 @@ btparallelrescan(IndexScanDesc scan)
  *		or _bt_parallel_done().
  *
  * The return value is true if we successfully seized the scan and false
- * if we did not.  The latter case occurs if no pages remain.
+ * if we did not.  The latter case occurs when no pages remain, or when
+ * another primitive index scan is scheduled that caller's backend cannot
+ * start just yet (only backends that call from _bt_first are capable of
+ * starting primitive index scans, which they indicate by passing first=true).
  *
  * If the return value is true, *pageno returns the next or current page
  * of the scan (depending on the scan direction).  An invalid block number
@@ -596,10 +599,6 @@ btparallelrescan(IndexScanDesc scan)
  * scan will return false.
  *
  * Callers should ignore the value of pageno if the return value is false.
- *
- * Callers that are in a position to start a new primitive index scan must
- * pass first=true (all other callers pass first=false).  We just return false
- * for first=false callers that require another primitive index scan.
  */
 bool
 _bt_parallel_seize(IndexScanDesc scan, BlockNumber *pageno, bool first)
@@ -616,22 +615,16 @@ _bt_parallel_seize(IndexScanDesc scan, BlockNumber *pageno, bool first)
 	{
 		/*
 		 * Initialize array related state when called from _bt_first, assuming
-		 * that this will either be the first primitive index scan for the
-		 * scan, or a previous explicitly scheduled primitive scan.
-		 *
-		 * Note: so->needPrimScan is only set when a scheduled primitive index
-		 * scan is set to be performed in caller's worker process.  It should
-		 * not be set here by us for the first primitive scan, nor should we
-		 * ever set it for a parallel scan that has no array keys.
+		 * that this will be the first primitive index scan for the scan
 		 */
 		so->needPrimScan = false;
 		so->scanBehind = false;
 	}
 	else
 	{
 		/*
-		 * Don't attempt to seize the scan when backend requires another
-		 * primitive index scan unless we're in a position to start it now
+		 * Don't attempt to seize the scan when it requires another primitive
+		 * index scan, since caller's backend cannot start it right now
 		 */
 		if (so->needPrimScan)
 			return false;
@@ -653,12 +646,9 @@ _bt_parallel_seize(IndexScanDesc scan, BlockNumber *pageno, bool first)
 		{
 			Assert(so->numArrayKeys);
 
-			/*
-			 * If we can start another primitive scan right away, do so.
-			 * Otherwise just wait.
-			 */
 			if (first)
 			{
+				/* Can start scheduled primitive scan right away, so do so */
 				btscan->btps_pageStatus = BTPARALLEL_ADVANCING;
 				for (int i = 0; i < so->numArrayKeys; i++)
 				{
@@ -668,11 +658,25 @@ _bt_parallel_seize(IndexScanDesc scan, BlockNumber *pageno, bool first)
 					array->cur_elem = btscan->btps_arrElems[i];
 					skey->sk_argument = array->elem_values[array->cur_elem];
 				}
-				so->needPrimScan = true;
-				so->scanBehind = false;
 				*pageno = InvalidBlockNumber;
 				exit_loop = true;
 			}
+			else
+			{
+				/*
+				 * Don't attempt to seize the scan when it requires another
+				 * primitive index scan, since caller's backend cannot start
+				 * it right now
+				 */
+				status = false;
+			}
+
+			/*
+			 * Either way, update backend local state to indicate that a
+			 * pending primitive scan is required
+			 */
+			so->needPrimScan = true;
+			so->scanBehind = false;
 		}
 		else if (btscan->btps_pageStatus != BTPARALLEL_ADVANCING)
 		{
@@ -731,6 +735,7 @@ _bt_parallel_release(IndexScanDesc scan, BlockNumber scan_page)
 void
 _bt_parallel_done(IndexScanDesc scan)
 {
+	BTScanOpaque so = (BTScanOpaque) scan->opaque;
 	ParallelIndexScanDesc parallel_scan = scan->parallel_scan;
 	BTParallelScanDesc btscan;
 	bool		status_changed = false;
@@ -739,6 +744,13 @@ _bt_parallel_done(IndexScanDesc scan)
 	if (parallel_scan == NULL)
 		return;
 
+	/*
+	 * Should not mark parallel scan done when there's still a pending
+	 * primitive index scan
+	 */
+	if (so->needPrimScan)
+		return;
+
 	btscan = (BTParallelScanDesc) OffsetToPointer((void *) parallel_scan,
 												  parallel_scan->ps_offset);
 
@@ -747,6 +759,7 @@ _bt_parallel_done(IndexScanDesc scan)
 	 * already
 	 */
 	SpinLockAcquire(&btscan->btps_mutex);
+	Assert(btscan->btps_pageStatus != BTPARALLEL_NEED_PRIMSCAN);
 	if (btscan->btps_pageStatus != BTPARALLEL_DONE)
 	{
 		btscan->btps_pageStatus = BTPARALLEL_DONE;
