Use condition variables to wait for checkpoints.

Previously we used a polling/sleeping loop to wait for checkpoints
to begin and end, which leads to up to a couple hundred milliseconds
of needless thumb-twiddling.  Use condition variables instead.

Author: Thomas Munro
Reviewed-by: Andres Freund
Discussion: https://postgr.es/m/CA%2BhUKGLY7sDe%2Bbg1K%3DbnEzOofGoo4bJHYh9%2BcDCXJepb6DQmLw%40mail.gmail.com

4 files changed, 34 insertions, 5 deletions

diff --git a/doc/src/sgml/monitoring.sgml b/doc/src/sgml/monitoring.sgml
index e2630fd368..60b89356f7 100644
--- a/doc/src/sgml/monitoring.sgml
+++ b/doc/src/sgml/monitoring.sgml
@@ -1281,7 +1281,7 @@ postgres   27093  0.0  0.0  30096  2752 ?        Ss   11:34   0:00 postgres: ser
          <entry>Waiting in an extension.</entry>
         </row>
         <row>
-         <entry morerows="34"><literal>IPC</literal></entry>
+         <entry morerows="36"><literal>IPC</literal></entry>
          <entry><literal>BgWorkerShutdown</literal></entry>
          <entry>Waiting for background worker to shut down.</entry>
         </row>
@@ -1294,6 +1294,14 @@ postgres   27093  0.0  0.0  30096  2752 ?        Ss   11:34   0:00 postgres: ser
          <entry>Waiting for the page number needed to continue a parallel B-tree scan to become available.</entry>
         </row>
         <row>
+         <entry><literal>CheckpointDone</literal></entry>
+         <entry>Waiting for a checkpoint to complete.</entry>
+        </row>
+        <row>
+         <entry><literal>CheckpointStart</literal></entry>
+         <entry>Waiting for a checkpoint to start.</entry>
+        </row>
+        <row>
          <entry><literal>ClogGroupUpdate</literal></entry>
          <entry>Waiting for group leader to update transaction status at transaction end.</entry>
         </row>
diff --git a/src/backend/postmaster/checkpointer.c b/src/backend/postmaster/checkpointer.c
index fe96c41359..3d5b382d04 100644
--- a/src/backend/postmaster/checkpointer.c
+++ b/src/backend/postmaster/checkpointer.c
@@ -126,6 +126,9 @@ typedef struct
 
 	int			ckpt_flags;		/* checkpoint flags, as defined in xlog.h */
 
+	ConditionVariable start_cv; /* signaled when ckpt_started advances */
+	ConditionVariable done_cv;	/* signaled when ckpt_done advances */
+
 	uint32		num_backend_writes; /* counts user backend buffer writes */
 	uint32		num_backend_fsync;	/* counts user backend fsync calls */
 
@@ -428,6 +431,8 @@ CheckpointerMain(void)
 			CheckpointerShmem->ckpt_started++;
 			SpinLockRelease(&CheckpointerShmem->ckpt_lck);
 
+			ConditionVariableBroadcast(&CheckpointerShmem->start_cv);
+
 			/*
 			 * The end-of-recovery checkpoint is a real checkpoint that's
 			 * performed while we're still in recovery.
@@ -488,6 +493,8 @@ CheckpointerMain(void)
 			CheckpointerShmem->ckpt_done = CheckpointerShmem->ckpt_started;
 			SpinLockRelease(&CheckpointerShmem->ckpt_lck);
 
+			ConditionVariableBroadcast(&CheckpointerShmem->done_cv);
+
 			if (ckpt_performed)
 			{
 				/*
@@ -915,6 +922,8 @@ CheckpointerShmemInit(void)
 		MemSet(CheckpointerShmem, 0, size);
 		SpinLockInit(&CheckpointerShmem->ckpt_lck);
 		CheckpointerShmem->max_requests = NBuffers;
+		ConditionVariableInit(&CheckpointerShmem->start_cv);
+		ConditionVariableInit(&CheckpointerShmem->done_cv);
 	}
 }
 
@@ -1023,6 +1032,7 @@ RequestCheckpoint(int flags)
 					new_failed;
 
 		/* Wait for a new checkpoint to start. */
+		ConditionVariablePrepareToSleep(&CheckpointerShmem->start_cv);
 		for (;;)
 		{
 			SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
@@ -1032,13 +1042,15 @@ RequestCheckpoint(int flags)
 			if (new_started != old_started)
 				break;
 
-			CHECK_FOR_INTERRUPTS();
-			pg_usleep(100000L);
+			ConditionVariableSleep(&CheckpointerShmem->start_cv,
+								   WAIT_EVENT_CHECKPOINT_START);
 		}
+		ConditionVariableCancelSleep();
 
 		/*
 		 * We are waiting for ckpt_done >= new_started, in a modulo sense.
 		 */
+		ConditionVariablePrepareToSleep(&CheckpointerShmem->done_cv);
 		for (;;)
 		{
 			int			new_done;
@@ -1051,9 +1063,10 @@ RequestCheckpoint(int flags)
 			if (new_done - new_started >= 0)
 				break;
 
-			CHECK_FOR_INTERRUPTS();
-			pg_usleep(100000L);
+			ConditionVariableSleep(&CheckpointerShmem->done_cv,
+								   WAIT_EVENT_CHECKPOINT_DONE);
 		}
+		ConditionVariableCancelSleep();
 
 		if (new_failed != old_failed)
 			ereport(ERROR,
diff --git a/src/backend/postmaster/pgstat.c b/src/backend/postmaster/pgstat.c
index ba31f532ea..2fbfadd9f0 100644
--- a/src/backend/postmaster/pgstat.c
+++ b/src/backend/postmaster/pgstat.c
@@ -3623,6 +3623,12 @@ pgstat_get_wait_ipc(WaitEventIPC w)
 		case WAIT_EVENT_BTREE_PAGE:
 			event_name = "BtreePage";
 			break;
+		case WAIT_EVENT_CHECKPOINT_DONE:
+			event_name = "CheckpointDone";
+			break;
+		case WAIT_EVENT_CHECKPOINT_START:
+			event_name = "CheckpointStart";
+			break;
 		case WAIT_EVENT_CLOG_GROUP_UPDATE:
 			event_name = "ClogGroupUpdate";
 			break;
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 725c8b0d64..ea6cc8b560 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -817,6 +817,8 @@ typedef enum
 	WAIT_EVENT_BGWORKER_STARTUP,
 	WAIT_EVENT_BTREE_PAGE,
 	WAIT_EVENT_CLOG_GROUP_UPDATE,
+	WAIT_EVENT_CHECKPOINT_DONE,
+	WAIT_EVENT_CHECKPOINT_START,
 	WAIT_EVENT_EXECUTE_GATHER,
 	WAIT_EVENT_HASH_BATCH_ALLOCATING,
 	WAIT_EVENT_HASH_BATCH_ELECTING,