codekingpro/portable-devtools

1/*-------------------------------------------------------------------------2 *3 * old_snapshot.h4 *		Data structures for 'snapshot too old'5 *6 * Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group7 * Portions Copyright (c) 1994, Regents of the University of California8 *9 * IDENTIFICATION10 *	  src/include/utils/old_snapshot.h11 *12 *-------------------------------------------------------------------------13 */14 15#ifndef OLD_SNAPSHOT_H16#define OLD_SNAPSHOT_H17 18#include "datatype/timestamp.h"19#include "storage/s_lock.h"20 21/*22 * Structure for dealing with old_snapshot_threshold implementation.23 */24typedef struct OldSnapshotControlData25{26	/*27	 * Variables for old snapshot handling are shared among processes and are28	 * only allowed to move forward.29	 */30	slock_t		mutex_current;	/* protect current_timestamp */31	TimestampTz current_timestamp;	/* latest snapshot timestamp */32	slock_t		mutex_latest_xmin;	/* protect latest_xmin and next_map_update */33	TransactionId latest_xmin;	/* latest snapshot xmin */34	TimestampTz next_map_update;	/* latest snapshot valid up to */35	slock_t		mutex_threshold;	/* protect threshold fields */36	TimestampTz threshold_timestamp;	/* earlier snapshot is old */37	TransactionId threshold_xid;	/* earlier xid may be gone */38 39	/*40	 * Keep one xid per minute for old snapshot error handling.41	 *42	 * Use a circular buffer with a head offset, a count of entries currently43	 * used, and a timestamp corresponding to the xid at the head offset.  A44	 * count_used value of zero means that there are no times stored; a45	 * count_used value of OLD_SNAPSHOT_TIME_MAP_ENTRIES means that the buffer46	 * is full and the head must be advanced to add new entries.  Use47	 * timestamps aligned to minute boundaries, since that seems less48	 * surprising than aligning based on the first usage timestamp.  The49	 * latest bucket is effectively stored within latest_xmin.  The circular50	 * buffer is updated when we get a new xmin value that doesn't fall into51	 * the same interval.52	 *53	 * It is OK if the xid for a given time slot is from earlier than54	 * calculated by adding the number of minutes corresponding to the55	 * (possibly wrapped) distance from the head offset to the time of the56	 * head entry, since that just results in the vacuuming of old tuples57	 * being slightly less aggressive.  It would not be OK for it to be off in58	 * the other direction, since it might result in vacuuming tuples that are59	 * still expected to be there.60	 *61	 * Use of an SLRU was considered but not chosen because it is more62	 * heavyweight than is needed for this, and would probably not be any less63	 * code to implement.64	 *65	 * Persistence is not needed.66	 */67	int			head_offset;	/* subscript of oldest tracked time */68	TimestampTz head_timestamp; /* time corresponding to head xid */69	int			count_used;		/* how many slots are in use */70	TransactionId xid_by_minute[FLEXIBLE_ARRAY_MEMBER];71} OldSnapshotControlData;72 73extern PGDLLIMPORT volatile OldSnapshotControlData *oldSnapshotControl;74 75#endif76