style(cluster): clang-format-18 compliance for spec-5.6 sources

The spec-5.6 cluster sources were hand-formatted and diverged from the repo
.clang-format under clang-format 18 (the fast-gate version; local was 22 --
the L345 version skew).  Reformatted with clang-format 18.1.x; whitespace,
brace placement and cast spacing only -- no semantic change.  Backend builds
clean and the cf cluster_unit binaries pass.

Spec: spec-5.6-cf-enqueue-shared-controlfile-authority.md

Author: SqlRush

src/backend/cluster/cluster_cf_authority.c

@@ -56,8 +56,7 @@
 static bool
 build_path(char *dst, size_t dstlen, const char *relpath)
 {
-	if (cluster_shared_data_dir == NULL || cluster_shared_data_dir[0] == '\0')
-	{
+	if (cluster_shared_data_dir == NULL || cluster_shared_data_dir[0] == '\0') {
 		if (dstlen > 0)
 			dst[0] = '\0';
 		return false;
@@ -101,7 +100,7 @@ ClusterCfValidity
 cluster_cf_classify_buffer(const char *buf, size_t len, uint64 expected_sysid)
 {
 	ControlFileData cf;
-	pg_crc32c	crc;
+	pg_crc32c crc;
 
 	if (buf == NULL || len < sizeof(ControlFileData))
 		return CLUSTER_CF_INVALID_SHORT;
@@ -115,8 +114,7 @@ cluster_cf_classify_buffer(const char *buf, size_t len, uint64 expected_sysid)
 		return CLUSTER_CF_INVALID_CRC;
 
 	/* Foreign byte order: version is a nonzero multiple of 65536. */
-	if (cf.pg_control_version % 65536 == 0 &&
-		cf.pg_control_version / 65536 != 0)
+	if (cf.pg_control_version % 65536 == 0 && cf.pg_control_version / 65536 != 0)
 		return CLUSTER_CF_INVALID_BYTE_ORDER;
 
 	if (expected_sysid != 0 && cf.system_identifier != expected_sysid)
@@ -134,8 +132,7 @@ cluster_cf_classify_buffer(const char *buf, size_t len, uint64 expected_sysid)
  *	corrupt primary (spec §3.9 T3).  Otherwise fail-closed.
  */
 ClusterCfReadSource
-cluster_cf_decide_source(ClusterCfValidity primary, ClusterCfValidity bak,
-						 bool bak_strict_ok)
+cluster_cf_decide_source(ClusterCfValidity primary, ClusterCfValidity bak, bool bak_strict_ok)
 {
 	if (primary == CLUSTER_CF_VALID)
 		return CLUSTER_CF_SOURCE_PRIMARY;
@@ -174,8 +171,8 @@ cluster_cf_bak_strict_ok(const ControlFileData *bak, uint64 expected_sysid,
 static ClusterCfValidity
 read_image(const char *path, char *image)
 {
-	int			fd;
-	int			r;
+	int fd;
+	int r;
 
 	if (path == NULL)
 		return CLUSTER_CF_INVALID_SHORT;
@@ -187,7 +184,7 @@ read_image(const char *path, char *image)
 	r = read(fd, image, sizeof(ControlFileData));
 	CloseTransientFile(fd);
 
-	if (r != (int) sizeof(ControlFileData))
+	if (r != (int)sizeof(ControlFileData))
 		return CLUSTER_CF_INVALID_SHORT;
 
 	return cluster_cf_classify_buffer(image, sizeof(ControlFileData), 0);
@@ -203,16 +200,15 @@ read_image(const char *path, char *image)
 bool
 cluster_cf_authority_read(ControlFileData *out)
 {
-	char		primary_img[sizeof(ControlFileData)];
-	char		bak_img[sizeof(ControlFileData)];
+	char primary_img[sizeof(ControlFileData)];
+	char bak_img[sizeof(ControlFileData)];
 	ClusterCfValidity pv;
 	ClusterCfValidity bv;
-	bool		bak_strict_ok;
+	bool bak_strict_ok;
 	ClusterCfReadSource src;
 
 	pv = read_image(cluster_cf_shared_path(), primary_img);
-	if (pv == CLUSTER_CF_VALID)
-	{
+	if (pv == CLUSTER_CF_VALID) {
 		memcpy(out, primary_img, sizeof(ControlFileData));
 		return true;
 	}
@@ -230,29 +226,26 @@ cluster_cf_authority_read(ControlFileData *out)
 	 * is skipped here (0); the symlink/migrate gates already reject a foreign
 	 * authority at startup (§3.2 M3).
 	 */
-	if (bv == CLUSTER_CF_VALID)
-	{
+	if (bv == CLUSTER_CF_VALID) {
 		ControlFileData bak_cf;
 
 		memcpy(&bak_cf, bak_img, sizeof(ControlFileData));
-		bak_strict_ok = cluster_cf_bak_strict_ok(&bak_cf, 0,
-												  cluster_cf_bak_checkpoint_recoverable(&bak_cf));
-	}
-	else
+		bak_strict_ok
+			= cluster_cf_bak_strict_ok(&bak_cf, 0, cluster_cf_bak_checkpoint_recoverable(&bak_cf));
+	} else
 		bak_strict_ok = false;
 
 	src = cluster_cf_decide_source(pv, bv, bak_strict_ok);
-	switch (src)
-	{
-		case CLUSTER_CF_SOURCE_PRIMARY:
-			memcpy(out, primary_img, sizeof(ControlFileData));
-			return true;
-		case CLUSTER_CF_SOURCE_BAK:
-			cluster_cf_counter_inc(CLUSTER_CF_BAK_FALLBACK);
-			memcpy(out, bak_img, sizeof(ControlFileData));
-			return true;
-		case CLUSTER_CF_SOURCE_FAILCLOSED:
-			break;
+	switch (src) {
+	case CLUSTER_CF_SOURCE_PRIMARY:
+		memcpy(out, primary_img, sizeof(ControlFileData));
+		return true;
+	case CLUSTER_CF_SOURCE_BAK:
+		cluster_cf_counter_inc(CLUSTER_CF_BAK_FALLBACK);
+		memcpy(out, bak_img, sizeof(ControlFileData));
+		return true;
+	case CLUSTER_CF_SOURCE_FAILCLOSED:
+		break;
 	}
 	return false;
 }
@@ -265,39 +258,28 @@ cluster_cf_authority_read(ControlFileData *out)
 static void
 write_durable(const char *tmp, const char *final, const char *buf)
 {
-	int			fd;
+	int fd;
 
 	fd = OpenTransientFile(tmp, O_RDWR | O_CREAT | O_TRUNC | PG_BINARY);
 	if (fd < 0)
-		ereport(PANIC,
-				(errcode_for_file_access(),
-				 errmsg("could not open file \"%s\": %m", tmp)));
+		ereport(PANIC, (errcode_for_file_access(), errmsg("could not open file \"%s\": %m", tmp)));
 
 	errno = 0;
-	if (write(fd, buf, PG_CONTROL_FILE_SIZE) != PG_CONTROL_FILE_SIZE)
-	{
+	if (write(fd, buf, PG_CONTROL_FILE_SIZE) != PG_CONTROL_FILE_SIZE) {
 		if (errno == 0)
 			errno = ENOSPC;
-		ereport(PANIC,
-				(errcode_for_file_access(),
-				 errmsg("could not write file \"%s\": %m", tmp)));
+		ereport(PANIC, (errcode_for_file_access(), errmsg("could not write file \"%s\": %m", tmp)));
 	}
 
 	if (pg_fsync(fd) != 0)
-		ereport(PANIC,
-				(errcode_for_file_access(),
-				 errmsg("could not fsync file \"%s\": %m", tmp)));
+		ereport(PANIC, (errcode_for_file_access(), errmsg("could not fsync file \"%s\": %m", tmp)));
 
 	if (CloseTransientFile(fd) != 0)
-		ereport(PANIC,
-				(errcode_for_file_access(),
-				 errmsg("could not close file \"%s\": %m", tmp)));
+		ereport(PANIC, (errcode_for_file_access(), errmsg("could not close file \"%s\": %m", tmp)));
 
 	if (durable_rename(tmp, final, PANIC) != 0)
-		ereport(PANIC,
-				(errcode_for_file_access(),
-				 errmsg("could not rename file \"%s\" to \"%s\": %m",
-						tmp, final)));
+		ereport(PANIC, (errcode_for_file_access(),
+						errmsg("could not rename file \"%s\" to \"%s\": %m", tmp, final)));
 }
 
 /*
@@ -308,18 +290,18 @@ write_durable(const char *tmp, const char *final, const char *buf)
 static void
 roll_primary_to_bak(const char *primary, const char *bak, const char *baktmp)
 {
-	char		buf[PG_CONTROL_FILE_SIZE];
-	int			fd;
-	int			r;
+	char buf[PG_CONTROL_FILE_SIZE];
+	int fd;
+	int r;
 
 	fd = OpenTransientFile(primary, O_RDONLY | PG_BINARY);
 	if (fd < 0)
-		return;					/* first write: no prior primary to preserve */
+		return; /* first write: no prior primary to preserve */
 
 	r = read(fd, buf, PG_CONTROL_FILE_SIZE);
 	CloseTransientFile(fd);
 	if (r != PG_CONTROL_FILE_SIZE)
-		return;					/* short/odd primary: don't manufacture a .bak */
+		return; /* short/odd primary: don't manufacture a .bak */
 
 	write_durable(baktmp, bak, buf);
 }
@@ -332,24 +314,23 @@ roll_primary_to_bak(const char *primary, const char *bak, const char *baktmp)
 void
 cluster_cf_authority_write(const ControlFileData *cf)
 {
-	char		buffer[PG_CONTROL_FILE_SIZE];
-	char		primary[MAXPGPATH];
-	char		bak[MAXPGPATH];
-	char		tmp[MAXPGPATH];
-	char		baktmp[MAXPGPATH];
+	char buffer[PG_CONTROL_FILE_SIZE];
+	char primary[MAXPGPATH];
+	char bak[MAXPGPATH];
+	char tmp[MAXPGPATH];
+	char baktmp[MAXPGPATH];
 	ControlFileData local;
 
-	if (!build_path(primary, sizeof(primary), CLUSTER_CF_REL_PATH) ||
-		!build_path(bak, sizeof(bak), CLUSTER_CF_BAK_REL_PATH) ||
-		!build_path(tmp, sizeof(tmp), CLUSTER_CF_TMP_REL_PATH) ||
-		!build_path(baktmp, sizeof(baktmp), CLUSTER_CF_BAK_TMP_REL_PATH))
-		ereport(PANIC,
-				(errmsg("cluster shared_data_dir is not configured")));
+	if (!build_path(primary, sizeof(primary), CLUSTER_CF_REL_PATH)
+		|| !build_path(bak, sizeof(bak), CLUSTER_CF_BAK_REL_PATH)
+		|| !build_path(tmp, sizeof(tmp), CLUSTER_CF_TMP_REL_PATH)
+		|| !build_path(baktmp, sizeof(baktmp), CLUSTER_CF_BAK_TMP_REL_PATH))
+		ereport(PANIC, (errmsg("cluster shared_data_dir is not configured")));
 
 	/* Recompute CRC over a private copy (cf is const). */
 	memcpy(&local, cf, sizeof(local));
 	INIT_CRC32C(local.crc);
-	COMP_CRC32C(local.crc, (char *) &local, offsetof(ControlFileData, crc));
+	COMP_CRC32C(local.crc, (char *)&local, offsetof(ControlFileData, crc));
 	FIN_CRC32C(local.crc);
 
 	/* Zero-pad to the full on-disk size, as update_controlfile does. */

src/backend/cluster/cluster_cf_enqueue.c

@@ -43,11 +43,10 @@
  * release must not run S6 against a phantom holder.  There is one slot per
  * mode (X / S); CF is not reentrant within a backend for a given mode.
  */
-typedef struct CfHoldState
-{
-	bool		held;
-	bool		coordinated;
-	ClusterLockAcquireRequest req;	/* resid + holder + request_id for release */
+typedef struct CfHoldState {
+	bool held;
+	bool coordinated;
+	ClusterLockAcquireRequest req; /* resid + holder + request_id for release */
 } CfHoldState;
 
 static CfHoldState cf_hold_x;
@@ -117,67 +116,63 @@ cluster_cf_lock(LOCKMODE mode)
 	cluster_cf_resid_encode(&req.resid);
 	/* locktag left zeroed: not LOCKTAG_ADVISORY, so normal blocking semantics. */
 	req.lockmode = mode;
-	req.op = CLUSTER_LOCK_OP_REQUEST;	/* CF never converts (X/S independent) */
+	req.op = CLUSTER_LOCK_OP_REQUEST; /* CF never converts (X/S independent) */
 	req.current_mode = NoLock;
 	req.lockmethod_id = DEFAULT_LOCKMETHOD;
-	req.dontwait = false;				/* block until granted or timeout */
+	req.dontwait = false; /* block until granted or timeout */
 	req.sessionLock = false;
-	req.caller_local_start_ts_ms = (uint64) (GetCurrentTimestamp() / 1000);
+	req.caller_local_start_ts_ms = (uint64)(GetCurrentTimestamp() / 1000);
 	/* spec-5.6 Dc4b: bound the CF acquire wait and label it ClusterCfEnqueueWait. */
 	req.timeout_ms = cluster_cf_enqueue_timeout_ms;
 	req.wait_event = WAIT_EVENT_CLUSTER_CF_ENQUEUE;
 
 	r = cluster_lock_acquire_seven_step(&req);
 
-	switch (r)
-	{
-		case CLUSTER_LOCK_ACQUIRE_OK_NATIVE:
+	switch (r) {
+	case CLUSTER_LOCK_ACQUIRE_OK_NATIVE:
 
-			/*
+		/*
 			 * Cluster/LMS layer inactive (single-node or gate off): no cross-
 			 * node coordination is needed and nothing was registered in the
 			 * GRD.  Treat as held but uncoordinated so release is a no-op.
 			 */
-			slot->held = true;
-			slot->coordinated = false;
-			slot->req = req;
-			cluster_cf_counter_inc(mode == ExclusiveLock ? CLUSTER_CF_X_ACQUIRE
-								   : CLUSTER_CF_S_ACQUIRE);
-			return true;
+		slot->held = true;
+		slot->coordinated = false;
+		slot->req = req;
+		cluster_cf_counter_inc(mode == ExclusiveLock ? CLUSTER_CF_X_ACQUIRE : CLUSTER_CF_S_ACQUIRE);
+		return true;
 
-		case CLUSTER_LOCK_ACQUIRE_NEED_PG_NATIVE_LOCK:
-		case CLUSTER_LOCK_ACQUIRE_OK_GRANTED:
-		case CLUSTER_LOCK_ACQUIRE_OK_CONVERTED:
+	case CLUSTER_LOCK_ACQUIRE_NEED_PG_NATIVE_LOCK:
+	case CLUSTER_LOCK_ACQUIRE_OK_GRANTED:
+	case CLUSTER_LOCK_ACQUIRE_OK_CONVERTED:
 
-			/*
+		/*
 			 * Granted at the cluster level.  CF has no PG-native heavyweight
 			 * lock to take, so run the S5 promote directly to turn the S3
 			 * reservation into a registered GRD holder (cross-node conflict
 			 * visibility).  S5 failure cancels the reservation (S7) and we
 			 * fail closed.
 			 */
-			if (cluster_lock_acquire_s5_promote(&req) != CLUSTER_LOCK_ACQUIRE_OK_GRANTED)
-			{
-				cluster_cf_counter_inc(CLUSTER_CF_FAILCLOSED);
-				return false;
-			}
-			slot->held = true;
-			slot->coordinated = true;
-			slot->req = req;			/* holder + request_id for the release */
-			cluster_cf_counter_inc(mode == ExclusiveLock ? CLUSTER_CF_X_ACQUIRE
-								   : CLUSTER_CF_S_ACQUIRE);
-			return true;
+		if (cluster_lock_acquire_s5_promote(&req) != CLUSTER_LOCK_ACQUIRE_OK_GRANTED) {
+			cluster_cf_counter_inc(CLUSTER_CF_FAILCLOSED);
+			return false;
+		}
+		slot->held = true;
+		slot->coordinated = true;
+		slot->req = req; /* holder + request_id for the release */
+		cluster_cf_counter_inc(mode == ExclusiveLock ? CLUSTER_CF_X_ACQUIRE : CLUSTER_CF_S_ACQUIRE);
+		return true;
 
-		default:
+	default:
 
-			/*
+		/*
 			 * NOT_AVAIL (try conflict; CF blocks so this is unexpected),
 			 * timeout, LMS-unavailable, deadlock, internal, etc.  The lock
 			 * could not be proven held: fail closed.  The caller raises the
 			 * appropriate FATAL/ERROR (CF correctness, spec §3.1 A3).
 			 */
-			cluster_cf_counter_inc(CLUSTER_CF_FAILCLOSED);
-			return false;
+		cluster_cf_counter_inc(CLUSTER_CF_FAILCLOSED);
+		return false;
 	}
 }
 
@@ -199,7 +194,7 @@ cluster_cf_unlock(LOCKMODE mode)
 	 * re-derives the resid from a PG LOCKTAG that CF does not have.
 	 */
 	if (slot->coordinated)
-		(void) cluster_lock_acquire_s6_release(&slot->req);
+		(void)cluster_lock_acquire_s6_release(&slot->req);
 
 	slot->held = false;
 	slot->coordinated = false;