#include "postgres.h"
#include <ctype.h>
#include <math.h>
#include <time.h>
#include <fcntl.h>
#include <sys/stat.h>
#include <unistd.h>
#include "access/clog.h"
#include "access/commit_ts.h"
#include "access/heaptoast.h"
#include "access/multixact.h"
#include "access/rewriteheap.h"
#include "access/subtrans.h"
#include "access/timeline.h"
#include "access/transam.h"
#include "access/twophase.h"
#include "access/xact.h"
#include "access/xlog_internal.h"
#include "access/xlogarchive.h"
#include "access/xloginsert.h"
#include "access/xlogreader.h"
#include "access/xlogrecovery.h"
#include "access/xlogutils.h"
#include "backup/basebackup.h"
#include "catalog/catversion.h"
#include "catalog/pg_control.h"
#include "catalog/pg_database.h"
#include "common/controldata_utils.h"
#include "common/file_utils.h"
#include "executor/instrument.h"
#include "miscadmin.h"
#include "pg_trace.h"
#include "pgstat.h"
#include "port/atomics.h"
#include "port/pg_iovec.h"
#include "postmaster/bgwriter.h"
#include "postmaster/startup.h"
#include "postmaster/walsummarizer.h"
#include "postmaster/walwriter.h"
#include "replication/origin.h"
#include "replication/slot.h"
#include "replication/snapbuild.h"
#include "replication/walreceiver.h"
#include "replication/walsender.h"
#include "storage/bufmgr.h"
#include "storage/fd.h"
#include "storage/ipc.h"
#include "storage/large_object.h"
#include "storage/latch.h"
#include "storage/predicate.h"
#include "storage/proc.h"
#include "storage/procarray.h"
#include "storage/reinit.h"
#include "storage/spin.h"
#include "storage/sync.h"
#include "utils/guc_hooks.h"
#include "utils/guc_tables.h"
#include "utils/injection_point.h"
#include "utils/memutils.h"
#include "utils/ps_status.h"
#include "utils/relmapper.h"
#include "utils/snapmgr.h"
#include "utils/timeout.h"
#include "utils/timestamp.h"
#include "utils/varlena.h"

Include dependency graph for xlog.c:

Data Structures
struct	XLogwrtRqst

struct	XLogwrtResult

struct	WALInsertLock

union	WALInsertLockPadded

struct	XLogCtlInsert

struct	XLogCtlData

Macros
#define	BootstrapTimeLineID 1

#define	NUM_XLOGINSERT_LOCKS 8

#define	INSERT_FREESPACE(endptr) (((endptr) % XLOG_BLCKSZ == 0) ? 0 : (XLOG_BLCKSZ - (endptr) % XLOG_BLCKSZ))

#define	NextBufIdx(idx) (((idx) == XLogCtl->XLogCacheBlck) ? 0 : ((idx) + 1))

#define	XLogRecPtrToBufIdx(recptr) (((recptr) / XLOG_BLCKSZ) % (XLogCtl->XLogCacheBlck + 1))

#define	UsableBytesInPage (XLOG_BLCKSZ - SizeOfXLogShortPHD)

#define	ConvertToXSegs(x, segsize) XLogMBVarToSegs((x), (segsize))

#define	RefreshXLogWriteResult(_target)

Typedefs
typedef struct XLogwrtRqst	XLogwrtRqst

typedef struct XLogwrtResult	XLogwrtResult

typedef union WALInsertLockPadded	WALInsertLockPadded

typedef struct XLogCtlInsert	XLogCtlInsert

typedef struct XLogCtlData	XLogCtlData

Enumerations
enum	WalInsertClass { WALINSERT_NORMAL , WALINSERT_SPECIAL_SWITCH , WALINSERT_SPECIAL_CHECKPOINT }

Functions
static void	CleanupAfterArchiveRecovery (TimeLineID EndOfLogTLI, XLogRecPtr EndOfLog, TimeLineID newTLI)

static void	CheckRequiredParameterValues (void)

static void	XLogReportParameters (void)

static int	LocalSetXLogInsertAllowed (void)

static void	CreateEndOfRecoveryRecord (void)

static XLogRecPtr	CreateOverwriteContrecordRecord (XLogRecPtr aborted_lsn, XLogRecPtr pagePtr, TimeLineID newTLI)

static void	CheckPointGuts (XLogRecPtr checkPointRedo, int flags)

static void	KeepLogSeg (XLogRecPtr recptr, XLogSegNo *logSegNo)

static XLogRecPtr	XLogGetReplicationSlotMinimumLSN (void)

static void	AdvanceXLInsertBuffer (XLogRecPtr upto, TimeLineID tli, bool opportunistic)

static void	XLogWrite (XLogwrtRqst WriteRqst, TimeLineID tli, bool flexible)

static bool	InstallXLogFileSegment (XLogSegNo segno, char tmppath, bool find_free, XLogSegNo max_segno, TimeLineID tli)

static void	XLogFileClose (void)

static void	PreallocXlogFiles (XLogRecPtr endptr, TimeLineID tli)

static void	RemoveTempXlogFiles (void)

static void	RemoveOldXlogFiles (XLogSegNo segno, XLogRecPtr lastredoptr, XLogRecPtr endptr, TimeLineID insertTLI)

static void	RemoveXlogFile (const struct dirent segment_de, XLogSegNo recycleSegNo, XLogSegNo endlogSegNo, TimeLineID insertTLI)

static void	UpdateLastRemovedPtr (char *filename)

static void	ValidateXLOGDirectoryStructure (void)

static void	CleanupBackupHistory (void)

static void	UpdateMinRecoveryPoint (XLogRecPtr lsn, bool force)

static bool	PerformRecoveryXLogAction (void)

static void	InitControlFile (uint64 sysidentifier, uint32 data_checksum_version)

static void	WriteControlFile (void)

static void	ReadControlFile (void)

static void	UpdateControlFile (void)

static char *	str_time (pg_time_t tnow)

static int	get_sync_bit (int method)

static void	CopyXLogRecordToWAL (int write_len, bool isLogSwitch, XLogRecData *rdata, XLogRecPtr StartPos, XLogRecPtr EndPos, TimeLineID tli)

static void	ReserveXLogInsertLocation (int size, XLogRecPtr StartPos, XLogRecPtr EndPos, XLogRecPtr *PrevPtr)

static bool	ReserveXLogSwitch (XLogRecPtr StartPos, XLogRecPtr EndPos, XLogRecPtr *PrevPtr)

static XLogRecPtr	WaitXLogInsertionsToFinish (XLogRecPtr upto)

static char *	GetXLogBuffer (XLogRecPtr ptr, TimeLineID tli)

static XLogRecPtr	XLogBytePosToRecPtr (uint64 bytepos)

static XLogRecPtr	XLogBytePosToEndRecPtr (uint64 bytepos)

static uint64	XLogRecPtrToBytePos (XLogRecPtr ptr)

static void	WALInsertLockAcquire (void)

static void	WALInsertLockAcquireExclusive (void)

static void	WALInsertLockRelease (void)

static void	WALInsertLockUpdateInsertingAt (XLogRecPtr insertingAt)

XLogRecPtr	XLogInsertRecord (XLogRecData *rdata, XLogRecPtr fpw_lsn, uint8 flags, int num_fpi, bool topxid_included)

Size	WALReadFromBuffers (char *dstbuf, XLogRecPtr startptr, Size count, TimeLineID tli)

static void	CalculateCheckpointSegments (void)

void	assign_max_wal_size (int newval, void *extra)

void	assign_checkpoint_completion_target (double newval, void *extra)

bool	check_wal_segment_size (int newval, void *extra, GucSource source)

bool	check_max_slot_wal_keep_size (int newval, void *extra, GucSource source)

static XLogSegNo	XLOGfileslop (XLogRecPtr lastredoptr)

bool	XLogCheckpointNeeded (XLogSegNo new_segno)

void	XLogSetAsyncXactLSN (XLogRecPtr asyncXactLSN)

void	XLogSetReplicationSlotMinimumLSN (XLogRecPtr lsn)

void	XLogFlush (XLogRecPtr record)

bool	XLogBackgroundFlush (void)

bool	XLogNeedsFlush (XLogRecPtr record)

static int	XLogFileInitInternal (XLogSegNo logsegno, TimeLineID logtli, bool added, char path)

int	XLogFileInit (XLogSegNo logsegno, TimeLineID logtli)

static void	XLogFileCopy (TimeLineID destTLI, XLogSegNo destsegno, TimeLineID srcTLI, XLogSegNo srcsegno, int upto)

int	XLogFileOpen (XLogSegNo segno, TimeLineID tli)

void	CheckXLogRemoved (XLogSegNo segno, TimeLineID tli)

XLogSegNo	XLogGetLastRemovedSegno (void)

XLogSegNo	XLogGetOldestSegno (TimeLineID tli)

void	RemoveNonParentXlogFiles (XLogRecPtr switchpoint, TimeLineID newTLI)

uint64	GetSystemIdentifier (void)

char *	GetMockAuthenticationNonce (void)

bool	DataChecksumsEnabled (void)

XLogRecPtr	GetFakeLSNForUnloggedRel (void)

static int	XLOGChooseNumBuffers (void)

bool	check_wal_buffers (int newval, void *extra, GucSource source)

bool	check_wal_consistency_checking (char newval, void extra, GucSource source)

void	assign_wal_consistency_checking (const char newval, void extra)

void	InitializeWalConsistencyChecking (void)

const char *	show_archive_command (void)

const char *	show_in_hot_standby (void)

void	LocalProcessControlFile (bool reset)

WalLevel	GetActiveWalLevelOnStandby (void)

Size	XLOGShmemSize (void)

void	XLOGShmemInit (void)

void	BootStrapXLOG (uint32 data_checksum_version)

static void	XLogInitNewTimeline (TimeLineID endTLI, XLogRecPtr endOfLog, TimeLineID newTLI)

void	StartupXLOG (void)

void	SwitchIntoArchiveRecovery (XLogRecPtr EndRecPtr, TimeLineID replayTLI)

void	ReachedEndOfBackup (XLogRecPtr EndRecPtr, TimeLineID tli)

bool	RecoveryInProgress (void)

RecoveryState	GetRecoveryState (void)

bool	XLogInsertAllowed (void)

XLogRecPtr	GetRedoRecPtr (void)

void	GetFullPageWriteInfo (XLogRecPtr RedoRecPtr_p, bool doPageWrites_p)

XLogRecPtr	GetInsertRecPtr (void)

XLogRecPtr	GetFlushRecPtr (TimeLineID *insertTLI)

TimeLineID	GetWALInsertionTimeLine (void)

XLogRecPtr	GetLastImportantRecPtr (void)

pg_time_t	GetLastSegSwitchData (XLogRecPtr *lastSwitchLSN)

void	ShutdownXLOG (int code, Datum arg)

static void	LogCheckpointStart (int flags, bool restartpoint)

static void	LogCheckpointEnd (bool restartpoint)

static void	UpdateCheckPointDistanceEstimate (uint64 nbytes)

static void	update_checkpoint_display (int flags, bool restartpoint, bool reset)

void	CreateCheckPoint (int flags)

static void	RecoveryRestartPoint (const CheckPoint checkPoint, XLogReaderState record)

bool	CreateRestartPoint (int flags)

WALAvailability	GetWALAvailability (XLogRecPtr targetLSN)

void	XLogPutNextOid (Oid nextOid)

XLogRecPtr	RequestXLogSwitch (bool mark_unimportant)

XLogRecPtr	XLogRestorePoint (const char *rpName)

void	UpdateFullPageWrites (void)

void	xlog_redo (XLogReaderState *record)

void	assign_wal_sync_method (int new_wal_sync_method, void *extra)

void	issue_xlog_fsync (int fd, XLogSegNo segno, TimeLineID tli)

void	do_pg_backup_start (const char backupidstr, bool fast, List tablespaces, BackupState state, StringInfo tblspcmapfile)

SessionBackupState	get_backup_status (void)

void	do_pg_backup_stop (BackupState *state, bool waitforarchive)

void	do_pg_abort_backup (int code, Datum arg)

void	register_persistent_abort_backup_handler (void)

XLogRecPtr	GetXLogInsertRecPtr (void)

XLogRecPtr	GetXLogWriteRecPtr (void)

void	GetOldestRestartPoint (XLogRecPtr oldrecptr, TimeLineID oldtli)

void	XLogShutdownWalRcv (void)

void	SetInstallXLogFileSegmentActive (void)

bool	IsInstallXLogFileSegmentActive (void)

void	SetWalWriterSleeping (bool sleeping)

Variables
int	max_wal_size_mb = 1024

int	min_wal_size_mb = 80

int	wal_keep_size_mb = 0

int	XLOGbuffers = -1

int	XLogArchiveTimeout = 0

int	XLogArchiveMode = ARCHIVE_MODE_OFF

char *	XLogArchiveCommand = NULL

bool	EnableHotStandby = false

bool	fullPageWrites = true

bool	wal_log_hints = false

int	wal_compression = WAL_COMPRESSION_NONE

char *	wal_consistency_checking_string = NULL

bool *	wal_consistency_checking = NULL

bool	wal_init_zero = true

bool	wal_recycle = true

bool	log_checkpoints = true

int	wal_sync_method = DEFAULT_WAL_SYNC_METHOD

int	wal_level = WAL_LEVEL_REPLICA

int	CommitDelay = 0

int	CommitSiblings = 5

int	wal_retrieve_retry_interval = 5000

int	max_slot_wal_keep_size_mb = -1

int	wal_decode_buffer_size = 512 * 1024

bool	track_wal_io_timing = false

int	wal_segment_size = DEFAULT_XLOG_SEG_SIZE

int	CheckPointSegments

static double	CheckPointDistanceEstimate = 0

static double	PrevCheckPointDistance = 0

static bool	check_wal_consistency_checking_deferred = false

const struct config_enum_entry	wal_sync_method_options []

const struct config_enum_entry	archive_mode_options []

CheckpointStatsData	CheckpointStats

static bool	lastFullPageWrites

static bool	LocalRecoveryInProgress = true

static int	LocalXLogInsertAllowed = -1

XLogRecPtr	ProcLastRecPtr = InvalidXLogRecPtr

XLogRecPtr	XactLastRecEnd = InvalidXLogRecPtr

XLogRecPtr	XactLastCommitEnd = InvalidXLogRecPtr

static XLogRecPtr	RedoRecPtr

static bool	doPageWrites

static SessionBackupState	sessionBackupState = SESSION_BACKUP_NONE

static XLogCtlData *	XLogCtl = NULL

static WALInsertLockPadded *	WALInsertLocks = NULL

static ControlFileData *	ControlFile = NULL

static int	UsableBytesInSegment

static XLogwrtResult	LogwrtResult = {0, 0}

static int	openLogFile = -1

static XLogSegNo	openLogSegNo = 0

static TimeLineID	openLogTLI = 0

static XLogRecPtr	LocalMinRecoveryPoint

static TimeLineID	LocalMinRecoveryPointTLI

static bool	updateMinRecoveryPoint = true

static int	MyLockNo = 0

static bool	holdingAllLocks = false

Macro Definition Documentation

◆ BootstrapTimeLineID

#define BootstrapTimeLineID 1

Definition at line 109 of file xlog.c.

◆ ConvertToXSegs

#define ConvertToXSegs	(	x,
		segsize
	)	XLogMBVarToSegs((x), (segsize))

Definition at line 596 of file xlog.c.

◆ INSERT_FREESPACE

#define INSERT_FREESPACE ( endptr ) (((endptr) % XLOG_BLCKSZ == 0) ? 0 : (XLOG_BLCKSZ - (endptr) % XLOG_BLCKSZ))

Definition at line 573 of file xlog.c.

◆ NextBufIdx

#define NextBufIdx ( idx ) (((idx) == XLogCtl->XLogCacheBlck) ? 0 : ((idx) + 1))

Definition at line 577 of file xlog.c.

◆ NUM_XLOGINSERT_LOCKS

#define NUM_XLOGINSERT_LOCKS 8

Definition at line 148 of file xlog.c.

◆ RefreshXLogWriteResult

#define RefreshXLogWriteResult ( _target )

Value:

    do { \
        _target.Flush = pg_atomic_read_u64(&XLogCtl->logFlushResult); \
        pg_read_barrier(); \
        _target.Write = pg_atomic_read_u64(&XLogCtl->logWriteResult); \
    } while (0)

Definition at line 613 of file xlog.c.

◆ UsableBytesInPage

#define UsableBytesInPage (XLOG_BLCKSZ - SizeOfXLogShortPHD)

Definition at line 590 of file xlog.c.

◆ XLogRecPtrToBufIdx

#define XLogRecPtrToBufIdx ( recptr ) (((recptr) / XLOG_BLCKSZ) % (XLogCtl->XLogCacheBlck + 1))

Definition at line 584 of file xlog.c.

Typedef Documentation

◆ WALInsertLockPadded

typedef union WALInsertLockPadded WALInsertLockPadded

◆ XLogCtlData

typedef struct XLogCtlData XLogCtlData

◆ XLogCtlInsert

typedef struct XLogCtlInsert XLogCtlInsert

◆ XLogwrtResult

typedef struct XLogwrtResult XLogwrtResult

◆ XLogwrtRqst

typedef struct XLogwrtRqst XLogwrtRqst

Enumeration Type Documentation

◆ WalInsertClass

enum WalInsertClass

Enumerator
WALINSERT_NORMAL
WALINSERT_SPECIAL_SWITCH
WALINSERT_SPECIAL_CHECKPOINT

Definition at line 552 of file xlog.c.

 {
     WALINSERT_NORMAL,
     WALINSERT_SPECIAL_SWITCH,
     WALINSERT_SPECIAL_CHECKPOINT
 } WalInsertClass;

Function Documentation

◆ AdvanceXLInsertBuffer()

static void AdvanceXLInsertBuffer	(	XLogRecPtr	upto,
		TimeLineID	tli,
		bool	opportunistic
	)

static

Definition at line 1978 of file xlog.c.

 {
     XLogCtlInsert *Insert = &XLogCtl->Insert;
     int         nextidx;
     XLogRecPtr  OldPageRqstPtr;
     XLogwrtRqst WriteRqst;
     XLogRecPtr  NewPageEndPtr = InvalidXLogRecPtr;
     XLogRecPtr  NewPageBeginPtr;
     XLogPageHeader NewPage;
     int         npages pg_attribute_unused() = 0;
  
     LWLockAcquire(WALBufMappingLock, LW_EXCLUSIVE);
  
     /*
      * Now that we have the lock, check if someone initialized the page
      * already.
      */
     while (upto >= XLogCtl->InitializedUpTo || opportunistic)
     {
         nextidx = XLogRecPtrToBufIdx(XLogCtl->InitializedUpTo);
  
         /*
          * Get ending-offset of the buffer page we need to replace (this may
          * be zero if the buffer hasn't been used yet).  Fall through if it's
          * already written out.
          */
         OldPageRqstPtr = pg_atomic_read_u64(&XLogCtl->xlblocks[nextidx]);
         if (LogwrtResult.Write < OldPageRqstPtr)
         {
             /*
              * Nope, got work to do. If we just want to pre-initialize as much
              * as we can without flushing, give up now.
              */
             if (opportunistic)
                 break;
  
             /* Advance shared memory write request position */
             SpinLockAcquire(&XLogCtl->info_lck);
             if (XLogCtl->LogwrtRqst.Write < OldPageRqstPtr)
                 XLogCtl->LogwrtRqst.Write = OldPageRqstPtr;
             SpinLockRelease(&XLogCtl->info_lck);
  
             /*
              * Acquire an up-to-date LogwrtResult value and see if we still
              * need to write it or if someone else already did.
              */
             RefreshXLogWriteResult(LogwrtResult);
             if (LogwrtResult.Write < OldPageRqstPtr)
             {
                 /*
                  * Must acquire write lock. Release WALBufMappingLock first,
                  * to make sure that all insertions that we need to wait for
                  * can finish (up to this same position). Otherwise we risk
                  * deadlock.
                  */
                 LWLockRelease(WALBufMappingLock);
  
                 WaitXLogInsertionsToFinish(OldPageRqstPtr);
  
                 LWLockAcquire(WALWriteLock, LW_EXCLUSIVE);
  
                 RefreshXLogWriteResult(LogwrtResult);
                 if (LogwrtResult.Write >= OldPageRqstPtr)
                 {
                     /* OK, someone wrote it already */
                     LWLockRelease(WALWriteLock);
                 }
                 else
                 {
                     /* Have to write it ourselves */
                     TRACE_POSTGRESQL_WAL_BUFFER_WRITE_DIRTY_START();
                     WriteRqst.Write = OldPageRqstPtr;
                     WriteRqst.Flush = 0;
                     XLogWrite(WriteRqst, tli, false);
                     LWLockRelease(WALWriteLock);
                     PendingWalStats.wal_buffers_full++;
                     TRACE_POSTGRESQL_WAL_BUFFER_WRITE_DIRTY_DONE();
                 }
                 /* Re-acquire WALBufMappingLock and retry */
                 LWLockAcquire(WALBufMappingLock, LW_EXCLUSIVE);
                 continue;
             }
         }
  
         /*
          * Now the next buffer slot is free and we can set it up to be the
          * next output page.
          */
         NewPageBeginPtr = XLogCtl->InitializedUpTo;
         NewPageEndPtr = NewPageBeginPtr + XLOG_BLCKSZ;
  
         Assert(XLogRecPtrToBufIdx(NewPageBeginPtr) == nextidx);
  
         NewPage = (XLogPageHeader) (XLogCtl->pages + nextidx * (Size) XLOG_BLCKSZ);
  
         /*
          * Mark the xlblock with InvalidXLogRecPtr and issue a write barrier
          * before initializing. Otherwise, the old page may be partially
          * zeroed but look valid.
          */
         pg_atomic_write_u64(&XLogCtl->xlblocks[nextidx], InvalidXLogRecPtr);
         pg_write_barrier();
  
         /*
          * Be sure to re-zero the buffer so that bytes beyond what we've
          * written will look like zeroes and not valid XLOG records...
          */
         MemSet((char *) NewPage, 0, XLOG_BLCKSZ);
  
         /*
          * Fill the new page's header
          */
         NewPage->xlp_magic = XLOG_PAGE_MAGIC;
  
         /* NewPage->xlp_info = 0; */    /* done by memset */
         NewPage->xlp_tli = tli;
         NewPage->xlp_pageaddr = NewPageBeginPtr;
  
         /* NewPage->xlp_rem_len = 0; */ /* done by memset */
  
         /*
          * If online backup is not in progress, mark the header to indicate
          * that WAL records beginning in this page have removable backup
          * blocks.  This allows the WAL archiver to know whether it is safe to
          * compress archived WAL data by transforming full-block records into
          * the non-full-block format.  It is sufficient to record this at the
          * page level because we force a page switch (in fact a segment
          * switch) when starting a backup, so the flag will be off before any
          * records can be written during the backup.  At the end of a backup,
          * the last page will be marked as all unsafe when perhaps only part
          * is unsafe, but at worst the archiver would miss the opportunity to
          * compress a few records.
          */
         if (Insert->runningBackups == 0)
             NewPage->xlp_info |= XLP_BKP_REMOVABLE;
  
         /*
          * If first page of an XLOG segment file, make it a long header.
          */
         if ((XLogSegmentOffset(NewPage->xlp_pageaddr, wal_segment_size)) == 0)
         {
             XLogLongPageHeader NewLongPage = (XLogLongPageHeader) NewPage;
  
             NewLongPage->xlp_sysid = ControlFile->system_identifier;
             NewLongPage->xlp_seg_size = wal_segment_size;
             NewLongPage->xlp_xlog_blcksz = XLOG_BLCKSZ;
             NewPage->xlp_info |= XLP_LONG_HEADER;
         }
  
         /*
          * Make sure the initialization of the page becomes visible to others
          * before the xlblocks update. GetXLogBuffer() reads xlblocks without
          * holding a lock.
          */
         pg_write_barrier();
  
         pg_atomic_write_u64(&XLogCtl->xlblocks[nextidx], NewPageEndPtr);
         XLogCtl->InitializedUpTo = NewPageEndPtr;
  
         npages++;
     }
     LWLockRelease(WALBufMappingLock);
  
 #ifdef WAL_DEBUG
     if (XLOG_DEBUG && npages > 0)
     {
         elog(DEBUG1, "initialized %d pages, up to %X/%X",
              npages, LSN_FORMAT_ARGS(NewPageEndPtr));
     }
 #endif
 }

References Assert, ControlFile, DEBUG1, elog, XLogwrtRqst::Flush, XLogCtlData::info_lck, XLogCtlData::InitializedUpTo, XLogCtlData::Insert, Insert(), InvalidXLogRecPtr, LogwrtResult, XLogCtlData::LogwrtRqst, LSN_FORMAT_ARGS, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), MemSet, XLogCtlData::pages, PendingWalStats, pg_atomic_read_u64(), pg_atomic_write_u64(), pg_attribute_unused, pg_write_barrier, RefreshXLogWriteResult, SpinLockAcquire, SpinLockRelease, ControlFileData::system_identifier, WaitXLogInsertionsToFinish(), PgStat_PendingWalStats::wal_buffers_full, wal_segment_size, XLogwrtRqst::Write, XLogwrtResult::Write, XLogCtlData::xlblocks, XLOG_PAGE_MAGIC, XLogCtl, XLogRecPtrToBufIdx, XLogSegmentOffset, XLogWrite(), XLP_BKP_REMOVABLE, XLP_LONG_HEADER, XLogLongPageHeaderData::xlp_seg_size, XLogLongPageHeaderData::xlp_sysid, and XLogLongPageHeaderData::xlp_xlog_blcksz.

Referenced by GetXLogBuffer(), and XLogBackgroundFlush().

◆ assign_checkpoint_completion_target()

void assign_checkpoint_completion_target	(	double	newval,
		void *	extra
	)

Definition at line 2191 of file xlog.c.

 {
     CheckPointCompletionTarget = newval;
     CalculateCheckpointSegments();
 }

References CalculateCheckpointSegments(), CheckPointCompletionTarget, and newval.

◆ assign_max_wal_size()

void assign_max_wal_size	(	int	newval,
		void *	extra
	)

Definition at line 2184 of file xlog.c.

 {
     max_wal_size_mb = newval;
     CalculateCheckpointSegments();
 }

References CalculateCheckpointSegments(), max_wal_size_mb, and newval.

◆ assign_wal_consistency_checking()

void assign_wal_consistency_checking	(	const char *	newval,
		void *	extra
	)

Definition at line 4722 of file xlog.c.

 {
     /*
      * If some checks were deferred, it's possible that the checks will fail
      * later during InitializeWalConsistencyChecking(). But in that case, the
      * postmaster will exit anyway, so it's safe to proceed with the
      * assignment.
      *
      * Any built-in resource managers specified are assigned immediately,
      * which affects WAL created before shared_preload_libraries are
      * processed. Any custom resource managers specified won't be assigned
      * until after shared_preload_libraries are processed, but that's OK
      * because WAL for a custom resource manager can't be written before the
      * module is loaded anyway.
      */
     wal_consistency_checking = extra;
 }

References wal_consistency_checking.

◆ assign_wal_sync_method()

void assign_wal_sync_method	(	int	new_wal_sync_method,
		void *	extra
	)

Definition at line 8571 of file xlog.c.

 {
     if (wal_sync_method != new_wal_sync_method)
     {
         /*
          * To ensure that no blocks escape unsynced, force an fsync on the
          * currently open log segment (if any).  Also, if the open flag is
          * changing, close the log file so it will be reopened (with new flag
          * bit) at next use.
          */
         if (openLogFile >= 0)
         {
             pgstat_report_wait_start(WAIT_EVENT_WAL_SYNC_METHOD_ASSIGN);
             if (pg_fsync(openLogFile) != 0)
             {
                 char        xlogfname[MAXFNAMELEN];
                 int         save_errno;
  
                 save_errno = errno;
                 XLogFileName(xlogfname, openLogTLI, openLogSegNo,
                              wal_segment_size);
                 errno = save_errno;
                 ereport(PANIC,
                         (errcode_for_file_access(),
                          errmsg("could not fsync file \"%s\": %m", xlogfname)));
             }
  
             pgstat_report_wait_end();
             if (get_sync_bit(wal_sync_method) != get_sync_bit(new_wal_sync_method))
                 XLogFileClose();
         }
     }
 }

References ereport, errcode_for_file_access(), errmsg(), get_sync_bit(), MAXFNAMELEN, openLogFile, openLogSegNo, openLogTLI, PANIC, pg_fsync(), pgstat_report_wait_end(), pgstat_report_wait_start(), wal_segment_size, wal_sync_method, XLogFileClose(), and XLogFileName().

◆ BootStrapXLOG()

void BootStrapXLOG ( uint32 data_checksum_version )

Definition at line 4998 of file xlog.c.

 {
     CheckPoint  checkPoint;
     char       *buffer;
     XLogPageHeader page;
     XLogLongPageHeader longpage;
     XLogRecord *record;
     char       *recptr;
     uint64      sysidentifier;
     struct timeval tv;
     pg_crc32c   crc;
  
     /* allow ordinary WAL segment creation, like StartupXLOG() would */
     SetInstallXLogFileSegmentActive();
  
     /*
      * Select a hopefully-unique system identifier code for this installation.
      * We use the result of gettimeofday(), including the fractional seconds
      * field, as being about as unique as we can easily get.  (Think not to
      * use random(), since it hasn't been seeded and there's no portable way
      * to seed it other than the system clock value...)  The upper half of the
      * uint64 value is just the tv_sec part, while the lower half contains the
      * tv_usec part (which must fit in 20 bits), plus 12 bits from our current
      * PID for a little extra uniqueness.  A person knowing this encoding can
      * determine the initialization time of the installation, which could
      * perhaps be useful sometimes.
      */
     gettimeofday(&tv, NULL);
     sysidentifier = ((uint64) tv.tv_sec) << 32;
     sysidentifier |= ((uint64) tv.tv_usec) << 12;
     sysidentifier |= getpid() & 0xFFF;
  
     /* page buffer must be aligned suitably for O_DIRECT */
     buffer = (char *) palloc(XLOG_BLCKSZ + XLOG_BLCKSZ);
     page = (XLogPageHeader) TYPEALIGN(XLOG_BLCKSZ, buffer);
     memset(page, 0, XLOG_BLCKSZ);
  
     /*
      * Set up information for the initial checkpoint record
      *
      * The initial checkpoint record is written to the beginning of the WAL
      * segment with logid=0 logseg=1. The very first WAL segment, 0/0, is not
      * used, so that we can use 0/0 to mean "before any valid WAL segment".
      */
     checkPoint.redo = wal_segment_size + SizeOfXLogLongPHD;
     checkPoint.ThisTimeLineID = BootstrapTimeLineID;
     checkPoint.PrevTimeLineID = BootstrapTimeLineID;
     checkPoint.fullPageWrites = fullPageWrites;
     checkPoint.wal_level = wal_level;
     checkPoint.nextXid =
         FullTransactionIdFromEpochAndXid(0, FirstNormalTransactionId);
     checkPoint.nextOid = FirstGenbkiObjectId;
     checkPoint.nextMulti = FirstMultiXactId;
     checkPoint.nextMultiOffset = 0;
     checkPoint.oldestXid = FirstNormalTransactionId;
     checkPoint.oldestXidDB = Template1DbOid;
     checkPoint.oldestMulti = FirstMultiXactId;
     checkPoint.oldestMultiDB = Template1DbOid;
     checkPoint.oldestCommitTsXid = InvalidTransactionId;
     checkPoint.newestCommitTsXid = InvalidTransactionId;
     checkPoint.time = (pg_time_t) time(NULL);
     checkPoint.oldestActiveXid = InvalidTransactionId;
  
     TransamVariables->nextXid = checkPoint.nextXid;
     TransamVariables->nextOid = checkPoint.nextOid;
     TransamVariables->oidCount = 0;
     MultiXactSetNextMXact(checkPoint.nextMulti, checkPoint.nextMultiOffset);
     AdvanceOldestClogXid(checkPoint.oldestXid);
     SetTransactionIdLimit(checkPoint.oldestXid, checkPoint.oldestXidDB);
     SetMultiXactIdLimit(checkPoint.oldestMulti, checkPoint.oldestMultiDB, true);
     SetCommitTsLimit(InvalidTransactionId, InvalidTransactionId);
  
     /* Set up the XLOG page header */
     page->xlp_magic = XLOG_PAGE_MAGIC;
     page->xlp_info = XLP_LONG_HEADER;
     page->xlp_tli = BootstrapTimeLineID;
     page->xlp_pageaddr = wal_segment_size;
     longpage = (XLogLongPageHeader) page;
     longpage->xlp_sysid = sysidentifier;
     longpage->xlp_seg_size = wal_segment_size;
     longpage->xlp_xlog_blcksz = XLOG_BLCKSZ;
  
     /* Insert the initial checkpoint record */
     recptr = ((char *) page + SizeOfXLogLongPHD);
     record = (XLogRecord *) recptr;
     record->xl_prev = 0;
     record->xl_xid = InvalidTransactionId;
     record->xl_tot_len = SizeOfXLogRecord + SizeOfXLogRecordDataHeaderShort + sizeof(checkPoint);
     record->xl_info = XLOG_CHECKPOINT_SHUTDOWN;
     record->xl_rmid = RM_XLOG_ID;
     recptr += SizeOfXLogRecord;
     /* fill the XLogRecordDataHeaderShort struct */
     *(recptr++) = (char) XLR_BLOCK_ID_DATA_SHORT;
     *(recptr++) = sizeof(checkPoint);
     memcpy(recptr, &checkPoint, sizeof(checkPoint));
     recptr += sizeof(checkPoint);
     Assert(recptr - (char *) record == record->xl_tot_len);
  
     INIT_CRC32C(crc);
     COMP_CRC32C(crc, ((char *) record) + SizeOfXLogRecord, record->xl_tot_len - SizeOfXLogRecord);
     COMP_CRC32C(crc, (char *) record, offsetof(XLogRecord, xl_crc));
     FIN_CRC32C(crc);
     record->xl_crc = crc;
  
     /* Create first XLOG segment file */
     openLogTLI = BootstrapTimeLineID;
     openLogFile = XLogFileInit(1, BootstrapTimeLineID);
  
     /*
      * We needn't bother with Reserve/ReleaseExternalFD here, since we'll
      * close the file again in a moment.
      */
  
     /* Write the first page with the initial record */
     errno = 0;
     pgstat_report_wait_start(WAIT_EVENT_WAL_BOOTSTRAP_WRITE);
     if (write(openLogFile, page, XLOG_BLCKSZ) != XLOG_BLCKSZ)
     {
         /* if write didn't set errno, assume problem is no disk space */
         if (errno == 0)
             errno = ENOSPC;
         ereport(PANIC,
                 (errcode_for_file_access(),
                  errmsg("could not write bootstrap write-ahead log file: %m")));
     }
     pgstat_report_wait_end();
  
     pgstat_report_wait_start(WAIT_EVENT_WAL_BOOTSTRAP_SYNC);
     if (pg_fsync(openLogFile) != 0)
         ereport(PANIC,
                 (errcode_for_file_access(),
                  errmsg("could not fsync bootstrap write-ahead log file: %m")));
     pgstat_report_wait_end();
  
     if (close(openLogFile) != 0)
         ereport(PANIC,
                 (errcode_for_file_access(),
                  errmsg("could not close bootstrap write-ahead log file: %m")));
  
     openLogFile = -1;
  
     /* Now create pg_control */
     InitControlFile(sysidentifier, data_checksum_version);
     ControlFile->time = checkPoint.time;
     ControlFile->checkPoint = checkPoint.redo;
     ControlFile->checkPointCopy = checkPoint;
  
     /* some additional ControlFile fields are set in WriteControlFile() */
     WriteControlFile();
  
     /* Bootstrap the commit log, too */
     BootStrapCLOG();
     BootStrapCommitTs();
     BootStrapSUBTRANS();
     BootStrapMultiXact();
  
     pfree(buffer);
  
     /*
      * Force control file to be read - in contrast to normal processing we'd
      * otherwise never run the checks and GUC related initializations therein.
      */
     ReadControlFile();
 }

References AdvanceOldestClogXid(), Assert, BootStrapCLOG(), BootStrapCommitTs(), BootStrapMultiXact(), BootStrapSUBTRANS(), BootstrapTimeLineID, ControlFileData::checkPoint, ControlFileData::checkPointCopy, close, COMP_CRC32C, ControlFile, crc, ereport, errcode_for_file_access(), errmsg(), FIN_CRC32C, FirstGenbkiObjectId, FirstMultiXactId, FirstNormalTransactionId, fullPageWrites, CheckPoint::fullPageWrites, FullTransactionIdFromEpochAndXid(), gettimeofday(), INIT_CRC32C, InitControlFile(), InvalidTransactionId, MultiXactSetNextMXact(), CheckPoint::newestCommitTsXid, CheckPoint::nextMulti, CheckPoint::nextMultiOffset, TransamVariablesData::nextOid, CheckPoint::nextOid, TransamVariablesData::nextXid, CheckPoint::nextXid, TransamVariablesData::oidCount, CheckPoint::oldestActiveXid, CheckPoint::oldestCommitTsXid, CheckPoint::oldestMulti, CheckPoint::oldestMultiDB, CheckPoint::oldestXid, CheckPoint::oldestXidDB, openLogFile, openLogTLI, palloc(), PANIC, pfree(), pg_fsync(), pgstat_report_wait_end(), pgstat_report_wait_start(), CheckPoint::PrevTimeLineID, ReadControlFile(), CheckPoint::redo, SetCommitTsLimit(), SetInstallXLogFileSegmentActive(), SetMultiXactIdLimit(), SetTransactionIdLimit(), SizeOfXLogLongPHD, SizeOfXLogRecord, SizeOfXLogRecordDataHeaderShort, CheckPoint::ThisTimeLineID, CheckPoint::time, ControlFileData::time, TransamVariables, TYPEALIGN, wal_level, CheckPoint::wal_level, wal_segment_size, write, WriteControlFile(), XLogRecord::xl_crc, XLogRecord::xl_info, XLogRecord::xl_prev, XLogRecord::xl_rmid, XLogRecord::xl_tot_len, XLogRecord::xl_xid, XLOG_CHECKPOINT_SHUTDOWN, XLOG_PAGE_MAGIC, XLogFileInit(), XLogPageHeaderData::xlp_info, XLP_LONG_HEADER, XLogPageHeaderData::xlp_magic, XLogPageHeaderData::xlp_pageaddr, XLogLongPageHeaderData::xlp_seg_size, XLogLongPageHeaderData::xlp_sysid, XLogPageHeaderData::xlp_tli, XLogLongPageHeaderData::xlp_xlog_blcksz, and XLR_BLOCK_ID_DATA_SHORT.

Referenced by BootstrapModeMain().

◆ CalculateCheckpointSegments()

static void CalculateCheckpointSegments ( void )

static

Definition at line 2155 of file xlog.c.

 {
     double      target;
  
     /*-------
      * Calculate the distance at which to trigger a checkpoint, to avoid
      * exceeding max_wal_size_mb. This is based on two assumptions:
      *
      * a) we keep WAL for only one checkpoint cycle (prior to PG11 we kept
      *    WAL for two checkpoint cycles to allow us to recover from the
      *    secondary checkpoint if the first checkpoint failed, though we
      *    only did this on the primary anyway, not on standby. Keeping just
      *    one checkpoint simplifies processing and reduces disk space in
      *    many smaller databases.)
      * b) during checkpoint, we consume checkpoint_completion_target *
      *    number of segments consumed between checkpoints.
      *-------
      */
     target = (double) ConvertToXSegs(max_wal_size_mb, wal_segment_size) /
         (1.0 + CheckPointCompletionTarget);
  
     /* round down */
     CheckPointSegments = (int) target;
  
     if (CheckPointSegments < 1)
         CheckPointSegments = 1;
 }

References CheckPointCompletionTarget, CheckPointSegments, ConvertToXSegs, max_wal_size_mb, and wal_segment_size.

Referenced by assign_checkpoint_completion_target(), assign_max_wal_size(), and ReadControlFile().

◆ check_max_slot_wal_keep_size()

bool check_max_slot_wal_keep_size	(	int *	newval,
		void **	extra,
		GucSource	source
	)

Definition at line 2216 of file xlog.c.

 {
     if (IsBinaryUpgrade && *newval != -1)
     {
         GUC_check_errdetail("\"%s\" must be set to -1 during binary upgrade mode.",
                             "max_slot_wal_keep_size");
         return false;
     }
  
     return true;
 }

References GUC_check_errdetail, IsBinaryUpgrade, and newval.

◆ check_wal_buffers()

bool check_wal_buffers	(	int *	newval,
		void **	extra,
		GucSource	source
	)

Definition at line 4602 of file xlog.c.

 {
     /*
      * -1 indicates a request for auto-tune.
      */
     if (*newval == -1)
     {
         /*
          * If we haven't yet changed the boot_val default of -1, just let it
          * be.  We'll fix it when XLOGShmemSize is called.
          */
         if (XLOGbuffers == -1)
             return true;
  
         /* Otherwise, substitute the auto-tune value */
         *newval = XLOGChooseNumBuffers();
     }
  
     /*
      * We clamp manually-set values to at least 4 blocks.  Prior to PostgreSQL
      * 9.1, a minimum of 4 was enforced by guc.c, but since that is no longer
      * the case, we just silently treat such values as a request for the
      * minimum.  (We could throw an error instead, but that doesn't seem very
      * helpful.)
      */
     if (*newval < 4)
         *newval = 4;
  
     return true;
 }

References newval, XLOGbuffers, and XLOGChooseNumBuffers().

◆ check_wal_consistency_checking()

bool check_wal_consistency_checking	(	char **	newval,
		void **	extra,
		GucSource	source
	)

Definition at line 4637 of file xlog.c.

 {
     char       *rawstring;
     List       *elemlist;
     ListCell   *l;
     bool        newwalconsistency[RM_MAX_ID + 1];
  
     /* Initialize the array */
     MemSet(newwalconsistency, 0, (RM_MAX_ID + 1) * sizeof(bool));
  
     /* Need a modifiable copy of string */
     rawstring = pstrdup(*newval);
  
     /* Parse string into list of identifiers */
     if (!SplitIdentifierString(rawstring, ',', &elemlist))
     {
         /* syntax error in list */
         GUC_check_errdetail("List syntax is invalid.");
         pfree(rawstring);
         list_free(elemlist);
         return false;
     }
  
     foreach(l, elemlist)
     {
         char       *tok = (char *) lfirst(l);
         int         rmid;
  
         /* Check for 'all'. */
         if (pg_strcasecmp(tok, "all") == 0)
         {
             for (rmid = 0; rmid <= RM_MAX_ID; rmid++)
                 if (RmgrIdExists(rmid) && GetRmgr(rmid).rm_mask != NULL)
                     newwalconsistency[rmid] = true;
         }
         else
         {
             /* Check if the token matches any known resource manager. */
             bool        found = false;
  
             for (rmid = 0; rmid <= RM_MAX_ID; rmid++)
             {
                 if (RmgrIdExists(rmid) && GetRmgr(rmid).rm_mask != NULL &&
                     pg_strcasecmp(tok, GetRmgr(rmid).rm_name) == 0)
                 {
                     newwalconsistency[rmid] = true;
                     found = true;
                     break;
                 }
             }
             if (!found)
             {
                 /*
                  * During startup, it might be a not-yet-loaded custom
                  * resource manager.  Defer checking until
                  * InitializeWalConsistencyChecking().
                  */
                 if (!process_shared_preload_libraries_done)
                 {
                     check_wal_consistency_checking_deferred = true;
                 }
                 else
                 {
                     GUC_check_errdetail("Unrecognized key word: \"%s\".", tok);
                     pfree(rawstring);
                     list_free(elemlist);
                     return false;
                 }
             }
         }
     }
  
     pfree(rawstring);
     list_free(elemlist);
  
     /* assign new value */
     *extra = guc_malloc(ERROR, (RM_MAX_ID + 1) * sizeof(bool));
     memcpy(*extra, newwalconsistency, (RM_MAX_ID + 1) * sizeof(bool));
     return true;
 }

References check_wal_consistency_checking_deferred, ERROR, GetRmgr(), GUC_check_errdetail, guc_malloc(), lfirst, list_free(), MemSet, newval, pfree(), pg_strcasecmp(), process_shared_preload_libraries_done, pstrdup(), RmgrData::rm_mask, RM_MAX_ID, RmgrIdExists(), and SplitIdentifierString().

◆ check_wal_segment_size()

bool check_wal_segment_size	(	int *	newval,
		void **	extra,
		GucSource	source
	)

Definition at line 2198 of file xlog.c.

 {
     if (!IsValidWalSegSize(*newval))
     {
         GUC_check_errdetail("The WAL segment size must be a power of two between 1 MB and 1 GB.");
         return false;
     }
  
     return true;
 }

References GUC_check_errdetail, IsValidWalSegSize, and newval.

◆ CheckPointGuts()

static void CheckPointGuts	(	XLogRecPtr	checkPointRedo,
		int	flags
	)

static

Definition at line 7440 of file xlog.c.

 {
     CheckPointRelationMap();
     CheckPointReplicationSlots(flags & CHECKPOINT_IS_SHUTDOWN);
     CheckPointSnapBuild();
     CheckPointLogicalRewriteHeap();
     CheckPointReplicationOrigin();
  
     /* Write out all dirty data in SLRUs and the main buffer pool */
     TRACE_POSTGRESQL_BUFFER_CHECKPOINT_START(flags);
     CheckpointStats.ckpt_write_t = GetCurrentTimestamp();
     CheckPointCLOG();
     CheckPointCommitTs();
     CheckPointSUBTRANS();
     CheckPointMultiXact();
     CheckPointPredicate();
     CheckPointBuffers(flags);
  
     /* Perform all queued up fsyncs */
     TRACE_POSTGRESQL_BUFFER_CHECKPOINT_SYNC_START();
     CheckpointStats.ckpt_sync_t = GetCurrentTimestamp();
     ProcessSyncRequests();
     CheckpointStats.ckpt_sync_end_t = GetCurrentTimestamp();
     TRACE_POSTGRESQL_BUFFER_CHECKPOINT_DONE();
  
     /* We deliberately delay 2PC checkpointing as long as possible */
     CheckPointTwoPhase(checkPointRedo);
 }

References CHECKPOINT_IS_SHUTDOWN, CheckPointBuffers(), CheckPointCLOG(), CheckPointCommitTs(), CheckPointLogicalRewriteHeap(), CheckPointMultiXact(), CheckPointPredicate(), CheckPointRelationMap(), CheckPointReplicationOrigin(), CheckPointReplicationSlots(), CheckPointSnapBuild(), CheckpointStats, CheckPointSUBTRANS(), CheckPointTwoPhase(), CheckpointStatsData::ckpt_sync_end_t, CheckpointStatsData::ckpt_sync_t, CheckpointStatsData::ckpt_write_t, GetCurrentTimestamp(), and ProcessSyncRequests().

Referenced by CreateCheckPoint(), and CreateRestartPoint().

◆ CheckRequiredParameterValues()

static void CheckRequiredParameterValues ( void )

static

Definition at line 5343 of file xlog.c.

 {
     /*
      * For archive recovery, the WAL must be generated with at least 'replica'
      * wal_level.
      */
     if (ArchiveRecoveryRequested && ControlFile->wal_level == WAL_LEVEL_MINIMAL)
     {
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("WAL was generated with \"wal_level=minimal\", cannot continue recovering"),
                  errdetail("This happens if you temporarily set \"wal_level=minimal\" on the server."),
                  errhint("Use a backup taken after setting \"wal_level\" to higher than \"minimal\".")));
     }
  
     /*
      * For Hot Standby, the WAL must be generated with 'replica' mode, and we
      * must have at least as many backend slots as the primary.
      */
     if (ArchiveRecoveryRequested && EnableHotStandby)
     {
         /* We ignore autovacuum_max_workers when we make this test. */
         RecoveryRequiresIntParameter("max_connections",
                                      MaxConnections,
                                      ControlFile->MaxConnections);
         RecoveryRequiresIntParameter("max_worker_processes",
                                      max_worker_processes,
                                      ControlFile->max_worker_processes);
         RecoveryRequiresIntParameter("max_wal_senders",
                                      max_wal_senders,
                                      ControlFile->max_wal_senders);
         RecoveryRequiresIntParameter("max_prepared_transactions",
                                      max_prepared_xacts,
                                      ControlFile->max_prepared_xacts);
         RecoveryRequiresIntParameter("max_locks_per_transaction",
                                      max_locks_per_xact,
                                      ControlFile->max_locks_per_xact);
     }
 }

References ArchiveRecoveryRequested, ControlFile, EnableHotStandby, ereport, errcode(), errdetail(), errhint(), errmsg(), FATAL, max_locks_per_xact, ControlFileData::max_locks_per_xact, max_prepared_xacts, ControlFileData::max_prepared_xacts, max_wal_senders, ControlFileData::max_wal_senders, max_worker_processes, ControlFileData::max_worker_processes, MaxConnections, ControlFileData::MaxConnections, RecoveryRequiresIntParameter(), ControlFileData::wal_level, and WAL_LEVEL_MINIMAL.

Referenced by StartupXLOG(), and xlog_redo().

◆ CheckXLogRemoved()

void CheckXLogRemoved	(	XLogSegNo	segno,
		TimeLineID	tli
	)

Definition at line 3714 of file xlog.c.

 {
     int         save_errno = errno;
     XLogSegNo   lastRemovedSegNo;
  
     SpinLockAcquire(&XLogCtl->info_lck);
     lastRemovedSegNo = XLogCtl->lastRemovedSegNo;
     SpinLockRelease(&XLogCtl->info_lck);
  
     if (segno <= lastRemovedSegNo)
     {
         char        filename[MAXFNAMELEN];
  
         XLogFileName(filename, tli, segno, wal_segment_size);
         errno = save_errno;
         ereport(ERROR,
                 (errcode_for_file_access(),
                  errmsg("requested WAL segment %s has already been removed",
                         filename)));
     }
     errno = save_errno;
 }

References ereport, errcode_for_file_access(), errmsg(), ERROR, filename, XLogCtlData::info_lck, XLogCtlData::lastRemovedSegNo, MAXFNAMELEN, SpinLockAcquire, SpinLockRelease, wal_segment_size, XLogCtl, and XLogFileName().

Referenced by logical_read_xlog_page(), perform_base_backup(), and XLogSendPhysical().

◆ CleanupAfterArchiveRecovery()

static void CleanupAfterArchiveRecovery	(	TimeLineID	EndOfLogTLI,
		XLogRecPtr	EndOfLog,
		TimeLineID	newTLI
	)

static

Definition at line 5254 of file xlog.c.

 {
     /*
      * Execute the recovery_end_command, if any.
      */
     if (recoveryEndCommand && strcmp(recoveryEndCommand, "") != 0)
         ExecuteRecoveryCommand(recoveryEndCommand,
                                "recovery_end_command",
                                true,
                                WAIT_EVENT_RECOVERY_END_COMMAND);
  
     /*
      * We switched to a new timeline. Clean up segments on the old timeline.
      *
      * If there are any higher-numbered segments on the old timeline, remove
      * them. They might contain valid WAL, but they might also be
      * pre-allocated files containing garbage. In any case, they are not part
      * of the new timeline's history so we don't need them.
      */
     RemoveNonParentXlogFiles(EndOfLog, newTLI);
  
     /*
      * If the switch happened in the middle of a segment, what to do with the
      * last, partial segment on the old timeline? If we don't archive it, and
      * the server that created the WAL never archives it either (e.g. because
      * it was hit by a meteor), it will never make it to the archive. That's
      * OK from our point of view, because the new segment that we created with
      * the new TLI contains all the WAL from the old timeline up to the switch
      * point. But if you later try to do PITR to the "missing" WAL on the old
      * timeline, recovery won't find it in the archive. It's physically
      * present in the new file with new TLI, but recovery won't look there
      * when it's recovering to the older timeline. On the other hand, if we
      * archive the partial segment, and the original server on that timeline
      * is still running and archives the completed version of the same segment
      * later, it will fail. (We used to do that in 9.4 and below, and it
      * caused such problems).
      *
      * As a compromise, we rename the last segment with the .partial suffix,
      * and archive it. Archive recovery will never try to read .partial
      * segments, so they will normally go unused. But in the odd PITR case,
      * the administrator can copy them manually to the pg_wal directory
      * (removing the suffix). They can be useful in debugging, too.
      *
      * If a .done or .ready file already exists for the old timeline, however,
      * we had already determined that the segment is complete, so we can let
      * it be archived normally. (In particular, if it was restored from the
      * archive to begin with, it's expected to have a .done file).
      */
     if (XLogSegmentOffset(EndOfLog, wal_segment_size) != 0 &&
         XLogArchivingActive())
     {
         char        origfname[MAXFNAMELEN];
         XLogSegNo   endLogSegNo;
  
         XLByteToPrevSeg(EndOfLog, endLogSegNo, wal_segment_size);
         XLogFileName(origfname, EndOfLogTLI, endLogSegNo, wal_segment_size);
  
         if (!XLogArchiveIsReadyOrDone(origfname))
         {
             char        origpath[MAXPGPATH];
             char        partialfname[MAXFNAMELEN];
             char        partialpath[MAXPGPATH];
  
             XLogFilePath(origpath, EndOfLogTLI, endLogSegNo, wal_segment_size);
             snprintf(partialfname, MAXFNAMELEN, "%s.partial", origfname);
             snprintf(partialpath, MAXPGPATH, "%s.partial", origpath);
  
             /*
              * Make sure there's no .done or .ready file for the .partial
              * file.
              */
             XLogArchiveCleanup(partialfname);
  
             durable_rename(origpath, partialpath, ERROR);
             XLogArchiveNotify(partialfname);
         }
     }
 }

References durable_rename(), ERROR, ExecuteRecoveryCommand(), MAXFNAMELEN, MAXPGPATH, recoveryEndCommand, RemoveNonParentXlogFiles(), snprintf, wal_segment_size, XLByteToPrevSeg, XLogArchiveCleanup(), XLogArchiveIsReadyOrDone(), XLogArchiveNotify(), XLogArchivingActive, XLogFileName(), XLogFilePath(), and XLogSegmentOffset.

Referenced by StartupXLOG().

◆ CleanupBackupHistory()

static void CleanupBackupHistory ( void )

static

Definition at line 4148 of file xlog.c.

 {
     DIR        *xldir;
     struct dirent *xlde;
     char        path[MAXPGPATH + sizeof(XLOGDIR)];
  
     xldir = AllocateDir(XLOGDIR);
  
     while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL)
     {
         if (IsBackupHistoryFileName(xlde->d_name))
         {
             if (XLogArchiveCheckDone(xlde->d_name))
             {
                 elog(DEBUG2, "removing WAL backup history file \"%s\"",
                      xlde->d_name);
                 snprintf(path, sizeof(path), XLOGDIR "/%s", xlde->d_name);
                 unlink(path);
                 XLogArchiveCleanup(xlde->d_name);
             }
         }
     }
  
     FreeDir(xldir);
 }

References AllocateDir(), dirent::d_name, DEBUG2, elog, FreeDir(), IsBackupHistoryFileName(), MAXPGPATH, ReadDir(), snprintf, XLogArchiveCheckDone(), XLogArchiveCleanup(), and XLOGDIR.

Referenced by do_pg_backup_stop().

◆ CopyXLogRecordToWAL()

static void CopyXLogRecordToWAL	(	int	write_len,
		bool	isLogSwitch,
		XLogRecData *	rdata,
		XLogRecPtr	StartPos,
		XLogRecPtr	EndPos,
		TimeLineID	tli
	)

static

Definition at line 1218 of file xlog.c.

 {
     char       *currpos;
     int         freespace;
     int         written;
     XLogRecPtr  CurrPos;
     XLogPageHeader pagehdr;
  
     /*
      * Get a pointer to the right place in the right WAL buffer to start
      * inserting to.
      */
     CurrPos = StartPos;
     currpos = GetXLogBuffer(CurrPos, tli);
     freespace = INSERT_FREESPACE(CurrPos);
  
     /*
      * there should be enough space for at least the first field (xl_tot_len)
      * on this page.
      */
     Assert(freespace >= sizeof(uint32));
  
     /* Copy record data */
     written = 0;
     while (rdata != NULL)
     {
         char       *rdata_data = rdata->data;
         int         rdata_len = rdata->len;
  
         while (rdata_len > freespace)
         {
             /*
              * Write what fits on this page, and continue on the next page.
              */
             Assert(CurrPos % XLOG_BLCKSZ >= SizeOfXLogShortPHD || freespace == 0);
             memcpy(currpos, rdata_data, freespace);
             rdata_data += freespace;
             rdata_len -= freespace;
             written += freespace;
             CurrPos += freespace;
  
             /*
              * Get pointer to beginning of next page, and set the xlp_rem_len
              * in the page header. Set XLP_FIRST_IS_CONTRECORD.
              *
              * It's safe to set the contrecord flag and xlp_rem_len without a
              * lock on the page. All the other flags were already set when the
              * page was initialized, in AdvanceXLInsertBuffer, and we're the
              * only backend that needs to set the contrecord flag.
              */
             currpos = GetXLogBuffer(CurrPos, tli);
             pagehdr = (XLogPageHeader) currpos;
             pagehdr->xlp_rem_len = write_len - written;
             pagehdr->xlp_info |= XLP_FIRST_IS_CONTRECORD;
  
             /* skip over the page header */
             if (XLogSegmentOffset(CurrPos, wal_segment_size) == 0)
             {
                 CurrPos += SizeOfXLogLongPHD;
                 currpos += SizeOfXLogLongPHD;
             }
             else
             {
                 CurrPos += SizeOfXLogShortPHD;
                 currpos += SizeOfXLogShortPHD;
             }
             freespace = INSERT_FREESPACE(CurrPos);
         }
  
         Assert(CurrPos % XLOG_BLCKSZ >= SizeOfXLogShortPHD || rdata_len == 0);
         memcpy(currpos, rdata_data, rdata_len);
         currpos += rdata_len;
         CurrPos += rdata_len;
         freespace -= rdata_len;
         written += rdata_len;
  
         rdata = rdata->next;
     }
     Assert(written == write_len);
  
     /*
      * If this was an xlog-switch, it's not enough to write the switch record,
      * we also have to consume all the remaining space in the WAL segment.  We
      * have already reserved that space, but we need to actually fill it.
      */
     if (isLogSwitch && XLogSegmentOffset(CurrPos, wal_segment_size) != 0)
     {
         /* An xlog-switch record doesn't contain any data besides the header */
         Assert(write_len == SizeOfXLogRecord);
  
         /* Assert that we did reserve the right amount of space */
         Assert(XLogSegmentOffset(EndPos, wal_segment_size) == 0);
  
         /* Use up all the remaining space on the current page */
         CurrPos += freespace;
  
         /*
          * Cause all remaining pages in the segment to be flushed, leaving the
          * XLog position where it should be, at the start of the next segment.
          * We do this one page at a time, to make sure we don't deadlock
          * against ourselves if wal_buffers < wal_segment_size.
          */
         while (CurrPos < EndPos)
         {
             /*
              * The minimal action to flush the page would be to call
              * WALInsertLockUpdateInsertingAt(CurrPos) followed by
              * AdvanceXLInsertBuffer(...).  The page would be left initialized
              * mostly to zeros, except for the page header (always the short
              * variant, as this is never a segment's first page).
              *
              * The large vistas of zeros are good for compressibility, but the
              * headers interrupting them every XLOG_BLCKSZ (with values that
              * differ from page to page) are not.  The effect varies with
              * compression tool, but bzip2 for instance compresses about an
              * order of magnitude worse if those headers are left in place.
              *
              * Rather than complicating AdvanceXLInsertBuffer itself (which is
              * called in heavily-loaded circumstances as well as this lightly-
              * loaded one) with variant behavior, we just use GetXLogBuffer
              * (which itself calls the two methods we need) to get the pointer
              * and zero most of the page.  Then we just zero the page header.
              */
             currpos = GetXLogBuffer(CurrPos, tli);
             MemSet(currpos, 0, SizeOfXLogShortPHD);
  
             CurrPos += XLOG_BLCKSZ;
         }
     }
     else
     {
         /* Align the end position, so that the next record starts aligned */
         CurrPos = MAXALIGN64(CurrPos);
     }
  
     if (CurrPos != EndPos)
         ereport(PANIC,
                 errcode(ERRCODE_DATA_CORRUPTED),
                 errmsg_internal("space reserved for WAL record does not match what was written"));
 }

References Assert, XLogRecData::data, ereport, errcode(), ERRCODE_DATA_CORRUPTED, errmsg_internal(), GetXLogBuffer(), INSERT_FREESPACE, XLogRecData::len, MAXALIGN64, MemSet, XLogRecData::next, PANIC, SizeOfXLogLongPHD, SizeOfXLogRecord, SizeOfXLogShortPHD, wal_segment_size, XLogSegmentOffset, XLP_FIRST_IS_CONTRECORD, XLogPageHeaderData::xlp_info, and XLogPageHeaderData::xlp_rem_len.

Referenced by XLogInsertRecord().

◆ CreateCheckPoint()

void CreateCheckPoint ( int flags )

Definition at line 6821 of file xlog.c.

 {
     bool        shutdown;
     CheckPoint  checkPoint;
     XLogRecPtr  recptr;
     XLogSegNo   _logSegNo;
     XLogCtlInsert *Insert = &XLogCtl->Insert;
     uint32      freespace;
     XLogRecPtr  PriorRedoPtr;
     XLogRecPtr  last_important_lsn;
     VirtualTransactionId *vxids;
     int         nvxids;
     int         oldXLogAllowed = 0;
  
     /*
      * An end-of-recovery checkpoint is really a shutdown checkpoint, just
      * issued at a different time.
      */
     if (flags & (CHECKPOINT_IS_SHUTDOWN | CHECKPOINT_END_OF_RECOVERY))
         shutdown = true;
     else
         shutdown = false;
  
     /* sanity check */
     if (RecoveryInProgress() && (flags & CHECKPOINT_END_OF_RECOVERY) == 0)
         elog(ERROR, "can't create a checkpoint during recovery");
  
     /*
      * Prepare to accumulate statistics.
      *
      * Note: because it is possible for log_checkpoints to change while a
      * checkpoint proceeds, we always accumulate stats, even if
      * log_checkpoints is currently off.
      */
     MemSet(&CheckpointStats, 0, sizeof(CheckpointStats));
     CheckpointStats.ckpt_start_t = GetCurrentTimestamp();
  
     /*
      * Let smgr prepare for checkpoint; this has to happen outside the
      * critical section and before we determine the REDO pointer.  Note that
      * smgr must not do anything that'd have to be undone if we decide no
      * checkpoint is needed.
      */
     SyncPreCheckpoint();
  
     /*
      * Use a critical section to force system panic if we have trouble.
      */
     START_CRIT_SECTION();
  
     if (shutdown)
     {
         LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
         ControlFile->state = DB_SHUTDOWNING;
         UpdateControlFile();
         LWLockRelease(ControlFileLock);
     }
  
     /* Begin filling in the checkpoint WAL record */
     MemSet(&checkPoint, 0, sizeof(checkPoint));
     checkPoint.time = (pg_time_t) time(NULL);
  
     /*
      * For Hot Standby, derive the oldestActiveXid before we fix the redo
      * pointer. This allows us to begin accumulating changes to assemble our
      * starting snapshot of locks and transactions.
      */
     if (!shutdown && XLogStandbyInfoActive())
         checkPoint.oldestActiveXid = GetOldestActiveTransactionId();
     else
         checkPoint.oldestActiveXid = InvalidTransactionId;
  
     /*
      * Get location of last important record before acquiring insert locks (as
      * GetLastImportantRecPtr() also locks WAL locks).
      */
     last_important_lsn = GetLastImportantRecPtr();
  
     /*
      * If this isn't a shutdown or forced checkpoint, and if there has been no
      * WAL activity requiring a checkpoint, skip it.  The idea here is to
      * avoid inserting duplicate checkpoints when the system is idle.
      */
     if ((flags & (CHECKPOINT_IS_SHUTDOWN | CHECKPOINT_END_OF_RECOVERY |
                   CHECKPOINT_FORCE)) == 0)
     {
         if (last_important_lsn == ControlFile->checkPoint)
         {
             END_CRIT_SECTION();
             ereport(DEBUG1,
                     (errmsg_internal("checkpoint skipped because system is idle")));
             return;
         }
     }
  
     /*
      * An end-of-recovery checkpoint is created before anyone is allowed to
      * write WAL. To allow us to write the checkpoint record, temporarily
      * enable XLogInsertAllowed.
      */
     if (flags & CHECKPOINT_END_OF_RECOVERY)
         oldXLogAllowed = LocalSetXLogInsertAllowed();
  
     checkPoint.ThisTimeLineID = XLogCtl->InsertTimeLineID;
     if (flags & CHECKPOINT_END_OF_RECOVERY)
         checkPoint.PrevTimeLineID = XLogCtl->PrevTimeLineID;
     else
         checkPoint.PrevTimeLineID = checkPoint.ThisTimeLineID;
  
     /*
      * We must block concurrent insertions while examining insert state.
      */
     WALInsertLockAcquireExclusive();
  
     checkPoint.fullPageWrites = Insert->fullPageWrites;
     checkPoint.wal_level = wal_level;
  
     if (shutdown)
     {
         XLogRecPtr  curInsert = XLogBytePosToRecPtr(Insert->CurrBytePos);
  
         /*
          * Compute new REDO record ptr = location of next XLOG record.
          *
          * Since this is a shutdown checkpoint, there can't be any concurrent
          * WAL insertion.
          */
         freespace = INSERT_FREESPACE(curInsert);
         if (freespace == 0)
         {
             if (XLogSegmentOffset(curInsert, wal_segment_size) == 0)
                 curInsert += SizeOfXLogLongPHD;
             else
                 curInsert += SizeOfXLogShortPHD;
         }
         checkPoint.redo = curInsert;
  
         /*
          * Here we update the shared RedoRecPtr for future XLogInsert calls;
          * this must be done while holding all the insertion locks.
          *
          * Note: if we fail to complete the checkpoint, RedoRecPtr will be
          * left pointing past where it really needs to point.  This is okay;
          * the only consequence is that XLogInsert might back up whole buffers
          * that it didn't really need to.  We can't postpone advancing
          * RedoRecPtr because XLogInserts that happen while we are dumping
          * buffers must assume that their buffer changes are not included in
          * the checkpoint.
          */
         RedoRecPtr = XLogCtl->Insert.RedoRecPtr = checkPoint.redo;
     }
  
     /*
      * Now we can release the WAL insertion locks, allowing other xacts to
      * proceed while we are flushing disk buffers.
      */
     WALInsertLockRelease();
  
     /*
      * If this is an online checkpoint, we have not yet determined the redo
      * point. We do so now by inserting the special XLOG_CHECKPOINT_REDO
      * record; the LSN at which it starts becomes the new redo pointer. We
      * don't do this for a shutdown checkpoint, because in that case no WAL
      * can be written between the redo point and the insertion of the
      * checkpoint record itself, so the checkpoint record itself serves to
      * mark the redo point.
      */
     if (!shutdown)
     {
         /* Include WAL level in record for WAL summarizer's benefit. */
         XLogBeginInsert();
         XLogRegisterData((char *) &wal_level, sizeof(wal_level));
         (void) XLogInsert(RM_XLOG_ID, XLOG_CHECKPOINT_REDO);
  
         /*
          * XLogInsertRecord will have updated XLogCtl->Insert.RedoRecPtr in
          * shared memory and RedoRecPtr in backend-local memory, but we need
          * to copy that into the record that will be inserted when the
          * checkpoint is complete.
          */
         checkPoint.redo = RedoRecPtr;
     }
  
     /* Update the info_lck-protected copy of RedoRecPtr as well */
     SpinLockAcquire(&XLogCtl->info_lck);
     XLogCtl->RedoRecPtr = checkPoint.redo;
     SpinLockRelease(&XLogCtl->info_lck);
  
     /*
      * If enabled, log checkpoint start.  We postpone this until now so as not
      * to log anything if we decided to skip the checkpoint.
      */
     if (log_checkpoints)
         LogCheckpointStart(flags, false);
  
     /* Update the process title */
     update_checkpoint_display(flags, false, false);
  
     TRACE_POSTGRESQL_CHECKPOINT_START(flags);
  
     /*
      * Get the other info we need for the checkpoint record.
      *
      * We don't need to save oldestClogXid in the checkpoint, it only matters
      * for the short period in which clog is being truncated, and if we crash
      * during that we'll redo the clog truncation and fix up oldestClogXid
      * there.
      */
     LWLockAcquire(XidGenLock, LW_SHARED);
     checkPoint.nextXid = TransamVariables->nextXid;
     checkPoint.oldestXid = TransamVariables->oldestXid;
     checkPoint.oldestXidDB = TransamVariables->oldestXidDB;
     LWLockRelease(XidGenLock);
  
     LWLockAcquire(CommitTsLock, LW_SHARED);
     checkPoint.oldestCommitTsXid = TransamVariables->oldestCommitTsXid;
     checkPoint.newestCommitTsXid = TransamVariables->newestCommitTsXid;
     LWLockRelease(CommitTsLock);
  
     LWLockAcquire(OidGenLock, LW_SHARED);
     checkPoint.nextOid = TransamVariables->nextOid;
     if (!shutdown)
         checkPoint.nextOid += TransamVariables->oidCount;
     LWLockRelease(OidGenLock);
  
     MultiXactGetCheckptMulti(shutdown,
                              &checkPoint.nextMulti,
                              &checkPoint.nextMultiOffset,
                              &checkPoint.oldestMulti,
                              &checkPoint.oldestMultiDB);
  
     /*
      * Having constructed the checkpoint record, ensure all shmem disk buffers
      * and commit-log buffers are flushed to disk.
      *
      * This I/O could fail for various reasons.  If so, we will fail to
      * complete the checkpoint, but there is no reason to force a system
      * panic. Accordingly, exit critical section while doing it.
      */
     END_CRIT_SECTION();
  
     /*
      * In some cases there are groups of actions that must all occur on one
      * side or the other of a checkpoint record. Before flushing the
      * checkpoint record we must explicitly wait for any backend currently
      * performing those groups of actions.
      *
      * One example is end of transaction, so we must wait for any transactions
      * that are currently in commit critical sections.  If an xact inserted
      * its commit record into XLOG just before the REDO point, then a crash
      * restart from the REDO point would not replay that record, which means
      * that our flushing had better include the xact's update of pg_xact.  So
      * we wait till he's out of his commit critical section before proceeding.
      * See notes in RecordTransactionCommit().
      *
      * Because we've already released the insertion locks, this test is a bit
      * fuzzy: it is possible that we will wait for xacts we didn't really need
      * to wait for.  But the delay should be short and it seems better to make
      * checkpoint take a bit longer than to hold off insertions longer than
      * necessary. (In fact, the whole reason we have this issue is that xact.c
      * does commit record XLOG insertion and clog update as two separate steps
      * protected by different locks, but again that seems best on grounds of
      * minimizing lock contention.)
      *
      * A transaction that has not yet set delayChkptFlags when we look cannot
      * be at risk, since it has not inserted its commit record yet; and one
      * that's already cleared it is not at risk either, since it's done fixing
      * clog and we will correctly flush the update below.  So we cannot miss
      * any xacts we need to wait for.
      */
     vxids = GetVirtualXIDsDelayingChkpt(&nvxids, DELAY_CHKPT_START);
     if (nvxids > 0)
     {
         do
         {
             /*
              * Keep absorbing fsync requests while we wait. There could even
              * be a deadlock if we don't, if the process that prevents the
              * checkpoint is trying to add a request to the queue.
              */
             AbsorbSyncRequests();
  
             pgstat_report_wait_start(WAIT_EVENT_CHECKPOINT_DELAY_START);
             pg_usleep(10000L);  /* wait for 10 msec */
             pgstat_report_wait_end();
         } while (HaveVirtualXIDsDelayingChkpt(vxids, nvxids,
                                               DELAY_CHKPT_START));
     }
     pfree(vxids);
  
     CheckPointGuts(checkPoint.redo, flags);
  
     vxids = GetVirtualXIDsDelayingChkpt(&nvxids, DELAY_CHKPT_COMPLETE);
     if (nvxids > 0)
     {
         do
         {
             AbsorbSyncRequests();
  
             pgstat_report_wait_start(WAIT_EVENT_CHECKPOINT_DELAY_COMPLETE);
             pg_usleep(10000L);  /* wait for 10 msec */
             pgstat_report_wait_end();
         } while (HaveVirtualXIDsDelayingChkpt(vxids, nvxids,
                                               DELAY_CHKPT_COMPLETE));
     }
     pfree(vxids);
  
     /*
      * Take a snapshot of running transactions and write this to WAL. This
      * allows us to reconstruct the state of running transactions during
      * archive recovery, if required. Skip, if this info disabled.
      *
      * If we are shutting down, or Startup process is completing crash
      * recovery we don't need to write running xact data.
      */
     if (!shutdown && XLogStandbyInfoActive())
         LogStandbySnapshot();
  
     START_CRIT_SECTION();
  
     /*
      * Now insert the checkpoint record into XLOG.
      */
     XLogBeginInsert();
     XLogRegisterData((char *) (&checkPoint), sizeof(checkPoint));
     recptr = XLogInsert(RM_XLOG_ID,
                         shutdown ? XLOG_CHECKPOINT_SHUTDOWN :
                         XLOG_CHECKPOINT_ONLINE);
  
     XLogFlush(recptr);
  
     /*
      * We mustn't write any new WAL after a shutdown checkpoint, or it will be
      * overwritten at next startup.  No-one should even try, this just allows
      * sanity-checking.  In the case of an end-of-recovery checkpoint, we want
      * to just temporarily disable writing until the system has exited
      * recovery.
      */
     if (shutdown)
     {
         if (flags & CHECKPOINT_END_OF_RECOVERY)
             LocalXLogInsertAllowed = oldXLogAllowed;
         else
             LocalXLogInsertAllowed = 0; /* never again write WAL */
     }
  
     /*
      * We now have ProcLastRecPtr = start of actual checkpoint record, recptr
      * = end of actual checkpoint record.
      */
     if (shutdown && checkPoint.redo != ProcLastRecPtr)
         ereport(PANIC,
                 (errmsg("concurrent write-ahead log activity while database system is shutting down")));
  
     /*
      * Remember the prior checkpoint's redo ptr for
      * UpdateCheckPointDistanceEstimate()
      */
     PriorRedoPtr = ControlFile->checkPointCopy.redo;
  
     /*
      * Update the control file.
      */
     LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
     if (shutdown)
         ControlFile->state = DB_SHUTDOWNED;
     ControlFile->checkPoint = ProcLastRecPtr;
     ControlFile->checkPointCopy = checkPoint;
     /* crash recovery should always recover to the end of WAL */
     ControlFile->minRecoveryPoint = InvalidXLogRecPtr;
     ControlFile->minRecoveryPointTLI = 0;
  
     /*
      * Persist unloggedLSN value. It's reset on crash recovery, so this goes
      * unused on non-shutdown checkpoints, but seems useful to store it always
      * for debugging purposes.
      */
     ControlFile->unloggedLSN = pg_atomic_read_membarrier_u64(&XLogCtl->unloggedLSN);
  
     UpdateControlFile();
     LWLockRelease(ControlFileLock);
  
     /* Update shared-memory copy of checkpoint XID/epoch */
     SpinLockAcquire(&XLogCtl->info_lck);
     XLogCtl->ckptFullXid = checkPoint.nextXid;
     SpinLockRelease(&XLogCtl->info_lck);
  
     /*
      * We are now done with critical updates; no need for system panic if we
      * have trouble while fooling with old log segments.
      */
     END_CRIT_SECTION();
  
     /*
      * WAL summaries end when the next XLOG_CHECKPOINT_REDO or
      * XLOG_CHECKPOINT_SHUTDOWN record is reached. This is the first point
      * where (a) we're not inside of a critical section and (b) we can be
      * certain that the relevant record has been flushed to disk, which must
      * happen before it can be summarized.
      *
      * If this is a shutdown checkpoint, then this happens reasonably
      * promptly: we've only just inserted and flushed the
      * XLOG_CHECKPOINT_SHUTDOWN record. If this is not a shutdown checkpoint,
      * then this might not be very prompt at all: the XLOG_CHECKPOINT_REDO
      * record was written before we began flushing data to disk, and that
      * could be many minutes ago at this point. However, we don't XLogFlush()
      * after inserting that record, so we're not guaranteed that it's on disk
      * until after the above call that flushes the XLOG_CHECKPOINT_ONLINE
      * record.
      */
     SetWalSummarizerLatch();
  
     /*
      * Let smgr do post-checkpoint cleanup (eg, deleting old files).
      */
     SyncPostCheckpoint();
  
     /*
      * Update the average distance between checkpoints if the prior checkpoint
      * exists.
      */
     if (PriorRedoPtr != InvalidXLogRecPtr)
         UpdateCheckPointDistanceEstimate(RedoRecPtr - PriorRedoPtr);
  
     /*
      * Delete old log files, those no longer needed for last checkpoint to
      * prevent the disk holding the xlog from growing full.
      */
     XLByteToSeg(RedoRecPtr, _logSegNo, wal_segment_size);
     KeepLogSeg(recptr, &_logSegNo);
     if (InvalidateObsoleteReplicationSlots(RS_INVAL_WAL_REMOVED,
                                            _logSegNo, InvalidOid,
                                            InvalidTransactionId))
     {
         /*
          * Some slots have been invalidated; recalculate the old-segment
          * horizon, starting again from RedoRecPtr.
          */
         XLByteToSeg(RedoRecPtr, _logSegNo, wal_segment_size);
         KeepLogSeg(recptr, &_logSegNo);
     }
     _logSegNo--;
     RemoveOldXlogFiles(_logSegNo, RedoRecPtr, recptr,
                        checkPoint.ThisTimeLineID);
  
     /*
      * Make more log segments if needed.  (Do this after recycling old log
      * segments, since that may supply some of the needed files.)
      */
     if (!shutdown)
         PreallocXlogFiles(recptr, checkPoint.ThisTimeLineID);
  
     /*
      * Truncate pg_subtrans if possible.  We can throw away all data before
      * the oldest XMIN of any running transaction.  No future transaction will
      * attempt to reference any pg_subtrans entry older than that (see Asserts
      * in subtrans.c).  During recovery, though, we mustn't do this because
      * StartupSUBTRANS hasn't been called yet.
      */
     if (!RecoveryInProgress())
         TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning());
  
     /* Real work is done; log and update stats. */
     LogCheckpointEnd(false);
  
     /* Reset the process title */
     update_checkpoint_display(flags, false, true);
  
     TRACE_POSTGRESQL_CHECKPOINT_DONE(CheckpointStats.ckpt_bufs_written,
                                      NBuffers,
                                      CheckpointStats.ckpt_segs_added,
                                      CheckpointStats.ckpt_segs_removed,
                                      CheckpointStats.ckpt_segs_recycled);
 }

References AbsorbSyncRequests(), ControlFileData::checkPoint, CHECKPOINT_END_OF_RECOVERY, CHECKPOINT_FORCE, CHECKPOINT_IS_SHUTDOWN, ControlFileData::checkPointCopy, CheckPointGuts(), CheckpointStats, CheckpointStatsData::ckpt_bufs_written, CheckpointStatsData::ckpt_segs_added, CheckpointStatsData::ckpt_segs_recycled, CheckpointStatsData::ckpt_segs_removed, CheckpointStatsData::ckpt_start_t, XLogCtlData::ckptFullXid, ControlFile, DB_SHUTDOWNED, DB_SHUTDOWNING, DEBUG1, DELAY_CHKPT_COMPLETE, DELAY_CHKPT_START, elog, END_CRIT_SECTION, ereport, errmsg(), errmsg_internal(), ERROR, CheckPoint::fullPageWrites, GetCurrentTimestamp(), GetLastImportantRecPtr(), GetOldestActiveTransactionId(), GetOldestTransactionIdConsideredRunning(), GetVirtualXIDsDelayingChkpt(), HaveVirtualXIDsDelayingChkpt(), XLogCtlData::info_lck, XLogCtlData::Insert, Insert(), INSERT_FREESPACE, XLogCtlData::InsertTimeLineID, InvalidateObsoleteReplicationSlots(), InvalidOid, InvalidTransactionId, InvalidXLogRecPtr, KeepLogSeg(), LocalSetXLogInsertAllowed(), LocalXLogInsertAllowed, log_checkpoints, LogCheckpointEnd(), LogCheckpointStart(), LogStandbySnapshot(), LW_EXCLUSIVE, LW_SHARED, LWLockAcquire(), LWLockRelease(), MemSet, ControlFileData::minRecoveryPoint, ControlFileData::minRecoveryPointTLI, MultiXactGetCheckptMulti(), NBuffers, TransamVariablesData::newestCommitTsXid, CheckPoint::newestCommitTsXid, CheckPoint::nextMulti, CheckPoint::nextMultiOffset, TransamVariablesData::nextOid, CheckPoint::nextOid, TransamVariablesData::nextXid, CheckPoint::nextXid, TransamVariablesData::oidCount, CheckPoint::oldestActiveXid, TransamVariablesData::oldestCommitTsXid, CheckPoint::oldestCommitTsXid, CheckPoint::oldestMulti, CheckPoint::oldestMultiDB, TransamVariablesData::oldestXid, CheckPoint::oldestXid, TransamVariablesData::oldestXidDB, CheckPoint::oldestXidDB, PANIC, pfree(), pg_atomic_read_membarrier_u64(), pg_usleep(), pgstat_report_wait_end(), pgstat_report_wait_start(), PreallocXlogFiles(), XLogCtlData::PrevTimeLineID, CheckPoint::PrevTimeLineID, ProcLastRecPtr, RecoveryInProgress(), CheckPoint::redo, RedoRecPtr, XLogCtlInsert::RedoRecPtr, XLogCtlData::RedoRecPtr, RemoveOldXlogFiles(), RS_INVAL_WAL_REMOVED, SetWalSummarizerLatch(), SizeOfXLogLongPHD, SizeOfXLogShortPHD, SpinLockAcquire, SpinLockRelease, START_CRIT_SECTION, ControlFileData::state, SyncPostCheckpoint(), SyncPreCheckpoint(), CheckPoint::ThisTimeLineID, CheckPoint::time, TransamVariables, TruncateSUBTRANS(), XLogCtlData::unloggedLSN, ControlFileData::unloggedLSN, update_checkpoint_display(), UpdateCheckPointDistanceEstimate(), UpdateControlFile(), wal_level, CheckPoint::wal_level, wal_segment_size, WALInsertLockAcquireExclusive(), WALInsertLockRelease(), XLByteToSeg, XLOG_CHECKPOINT_ONLINE, XLOG_CHECKPOINT_REDO, XLOG_CHECKPOINT_SHUTDOWN, XLogBeginInsert(), XLogBytePosToRecPtr(), XLogCtl, XLogFlush(), XLogInsert(), XLogRegisterData(), XLogSegmentOffset, and XLogStandbyInfoActive.

Referenced by CheckpointerMain(), RequestCheckpoint(), and ShutdownXLOG().

◆ CreateEndOfRecoveryRecord()

static void CreateEndOfRecoveryRecord ( void )

static

Definition at line 7305 of file xlog.c.

 {
     xl_end_of_recovery xlrec;
     XLogRecPtr  recptr;
  
     /* sanity check */
     if (!RecoveryInProgress())
         elog(ERROR, "can only be used to end recovery");
  
     xlrec.end_time = GetCurrentTimestamp();
     xlrec.wal_level = wal_level;
  
     WALInsertLockAcquireExclusive();
     xlrec.ThisTimeLineID = XLogCtl->InsertTimeLineID;
     xlrec.PrevTimeLineID = XLogCtl->PrevTimeLineID;
     WALInsertLockRelease();
  
     START_CRIT_SECTION();
  
     XLogBeginInsert();
     XLogRegisterData((char *) &xlrec, sizeof(xl_end_of_recovery));
     recptr = XLogInsert(RM_XLOG_ID, XLOG_END_OF_RECOVERY);
  
     XLogFlush(recptr);
  
     /*
      * Update the control file so that crash recovery can follow the timeline
      * changes to this point.
      */
     LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
     ControlFile->minRecoveryPoint = recptr;
     ControlFile->minRecoveryPointTLI = xlrec.ThisTimeLineID;
     UpdateControlFile();
     LWLockRelease(ControlFileLock);
  
     END_CRIT_SECTION();
 }

References ControlFile, elog, END_CRIT_SECTION, xl_end_of_recovery::end_time, ERROR, GetCurrentTimestamp(), XLogCtlData::InsertTimeLineID, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), ControlFileData::minRecoveryPoint, ControlFileData::minRecoveryPointTLI, XLogCtlData::PrevTimeLineID, xl_end_of_recovery::PrevTimeLineID, RecoveryInProgress(), START_CRIT_SECTION, xl_end_of_recovery::ThisTimeLineID, UpdateControlFile(), wal_level, xl_end_of_recovery::wal_level, WALInsertLockAcquireExclusive(), WALInsertLockRelease(), XLOG_END_OF_RECOVERY, XLogBeginInsert(), XLogCtl, XLogFlush(), XLogInsert(), and XLogRegisterData().

Referenced by PerformRecoveryXLogAction().

◆ CreateOverwriteContrecordRecord()

static XLogRecPtr CreateOverwriteContrecordRecord	(	XLogRecPtr	aborted_lsn,
		XLogRecPtr	pagePtr,
		TimeLineID	newTLI
	)

static

Definition at line 7370 of file xlog.c.

 {
     xl_overwrite_contrecord xlrec;
     XLogRecPtr  recptr;
     XLogPageHeader pagehdr;
     XLogRecPtr  startPos;
  
     /* sanity checks */
     if (!RecoveryInProgress())
         elog(ERROR, "can only be used at end of recovery");
     if (pagePtr % XLOG_BLCKSZ != 0)
         elog(ERROR, "invalid position for missing continuation record %X/%X",
              LSN_FORMAT_ARGS(pagePtr));
  
     /* The current WAL insert position should be right after the page header */
     startPos = pagePtr;
     if (XLogSegmentOffset(startPos, wal_segment_size) == 0)
         startPos += SizeOfXLogLongPHD;
     else
         startPos += SizeOfXLogShortPHD;
     recptr = GetXLogInsertRecPtr();
     if (recptr != startPos)
         elog(ERROR, "invalid WAL insert position %X/%X for OVERWRITE_CONTRECORD",
              LSN_FORMAT_ARGS(recptr));
  
     START_CRIT_SECTION();
  
     /*
      * Initialize the XLOG page header (by GetXLogBuffer), and set the
      * XLP_FIRST_IS_OVERWRITE_CONTRECORD flag.
      *
      * No other backend is allowed to write WAL yet, so acquiring the WAL
      * insertion lock is just pro forma.
      */
     WALInsertLockAcquire();
     pagehdr = (XLogPageHeader) GetXLogBuffer(pagePtr, newTLI);
     pagehdr->xlp_info |= XLP_FIRST_IS_OVERWRITE_CONTRECORD;
     WALInsertLockRelease();
  
     /*
      * Insert the XLOG_OVERWRITE_CONTRECORD record as the first record on the
      * page.  We know it becomes the first record, because no other backend is
      * allowed to write WAL yet.
      */
     XLogBeginInsert();
     xlrec.overwritten_lsn = aborted_lsn;
     xlrec.overwrite_time = GetCurrentTimestamp();
     XLogRegisterData((char *) &xlrec, sizeof(xl_overwrite_contrecord));
     recptr = XLogInsert(RM_XLOG_ID, XLOG_OVERWRITE_CONTRECORD);
  
     /* check that the record was inserted to the right place */
     if (ProcLastRecPtr != startPos)
         elog(ERROR, "OVERWRITE_CONTRECORD was inserted to unexpected position %X/%X",
              LSN_FORMAT_ARGS(ProcLastRecPtr));
  
     XLogFlush(recptr);
  
     END_CRIT_SECTION();
  
     return recptr;
 }

References elog, END_CRIT_SECTION, ERROR, GetCurrentTimestamp(), GetXLogBuffer(), GetXLogInsertRecPtr(), LSN_FORMAT_ARGS, xl_overwrite_contrecord::overwrite_time, xl_overwrite_contrecord::overwritten_lsn, ProcLastRecPtr, RecoveryInProgress(), SizeOfXLogLongPHD, SizeOfXLogShortPHD, START_CRIT_SECTION, wal_segment_size, WALInsertLockAcquire(), WALInsertLockRelease(), XLOG_OVERWRITE_CONTRECORD, XLogBeginInsert(), XLogFlush(), XLogInsert(), XLogRegisterData(), XLogSegmentOffset, XLP_FIRST_IS_OVERWRITE_CONTRECORD, and XLogPageHeaderData::xlp_info.

Referenced by StartupXLOG().

◆ CreateRestartPoint()

bool CreateRestartPoint ( int flags )

Definition at line 7521 of file xlog.c.

 {
     XLogRecPtr  lastCheckPointRecPtr;
     XLogRecPtr  lastCheckPointEndPtr;
     CheckPoint  lastCheckPoint;
     XLogRecPtr  PriorRedoPtr;
     XLogRecPtr  receivePtr;
     XLogRecPtr  replayPtr;
     TimeLineID  replayTLI;
     XLogRecPtr  endptr;
     XLogSegNo   _logSegNo;
     TimestampTz xtime;
  
     /* Concurrent checkpoint/restartpoint cannot happen */
     Assert(!IsUnderPostmaster || MyBackendType == B_CHECKPOINTER);
  
     /* Get a local copy of the last safe checkpoint record. */
     SpinLockAcquire(&XLogCtl->info_lck);
     lastCheckPointRecPtr = XLogCtl->lastCheckPointRecPtr;
     lastCheckPointEndPtr = XLogCtl->lastCheckPointEndPtr;
     lastCheckPoint = XLogCtl->lastCheckPoint;
     SpinLockRelease(&XLogCtl->info_lck);
  
     /*
      * Check that we're still in recovery mode. It's ok if we exit recovery
      * mode after this check, the restart point is valid anyway.
      */
     if (!RecoveryInProgress())
     {
         ereport(DEBUG2,
                 (errmsg_internal("skipping restartpoint, recovery has already ended")));
         return false;
     }
  
     /*
      * If the last checkpoint record we've replayed is already our last
      * restartpoint, we can't perform a new restart point. We still update
      * minRecoveryPoint in that case, so that if this is a shutdown restart
      * point, we won't start up earlier than before. That's not strictly
      * necessary, but when hot standby is enabled, it would be rather weird if
      * the database opened up for read-only connections at a point-in-time
      * before the last shutdown. Such time travel is still possible in case of
      * immediate shutdown, though.
      *
      * We don't explicitly advance minRecoveryPoint when we do create a
      * restartpoint. It's assumed that flushing the buffers will do that as a
      * side-effect.
      */
     if (XLogRecPtrIsInvalid(lastCheckPointRecPtr) ||
         lastCheckPoint.redo <= ControlFile->checkPointCopy.redo)
     {
         ereport(DEBUG2,
                 (errmsg_internal("skipping restartpoint, already performed at %X/%X",
                                  LSN_FORMAT_ARGS(lastCheckPoint.redo))));
  
         UpdateMinRecoveryPoint(InvalidXLogRecPtr, true);
         if (flags & CHECKPOINT_IS_SHUTDOWN)
         {
             LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
             ControlFile->state = DB_SHUTDOWNED_IN_RECOVERY;
             UpdateControlFile();
             LWLockRelease(ControlFileLock);
         }
         return false;
     }
  
     /*
      * Update the shared RedoRecPtr so that the startup process can calculate
      * the number of segments replayed since last restartpoint, and request a
      * restartpoint if it exceeds CheckPointSegments.
      *
      * Like in CreateCheckPoint(), hold off insertions to update it, although
      * during recovery this is just pro forma, because no WAL insertions are
      * happening.
      */
     WALInsertLockAcquireExclusive();
     RedoRecPtr = XLogCtl->Insert.RedoRecPtr = lastCheckPoint.redo;
     WALInsertLockRelease();
  
     /* Also update the info_lck-protected copy */
     SpinLockAcquire(&XLogCtl->info_lck);
     XLogCtl->RedoRecPtr = lastCheckPoint.redo;
     SpinLockRelease(&XLogCtl->info_lck);
  
     /*
      * Prepare to accumulate statistics.
      *
      * Note: because it is possible for log_checkpoints to change while a
      * checkpoint proceeds, we always accumulate stats, even if
      * log_checkpoints is currently off.
      */
     MemSet(&CheckpointStats, 0, sizeof(CheckpointStats));
     CheckpointStats.ckpt_start_t = GetCurrentTimestamp();
  
     if (log_checkpoints)
         LogCheckpointStart(flags, true);
  
     /* Update the process title */
     update_checkpoint_display(flags, true, false);
  
     CheckPointGuts(lastCheckPoint.redo, flags);
  
     /*
      * This location needs to be after CheckPointGuts() to ensure that some
      * work has already happened during this checkpoint.
      */
     INJECTION_POINT("create-restart-point");
  
     /*
      * Remember the prior checkpoint's redo ptr for
      * UpdateCheckPointDistanceEstimate()
      */
     PriorRedoPtr = ControlFile->checkPointCopy.redo;
  
     /*
      * Update pg_control, using current time.  Check that it still shows an
      * older checkpoint, else do nothing; this is a quick hack to make sure
      * nothing really bad happens if somehow we get here after the
      * end-of-recovery checkpoint.
      */
     LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
     if (ControlFile->checkPointCopy.redo < lastCheckPoint.redo)
     {
         /*
          * Update the checkpoint information.  We do this even if the cluster
          * does not show DB_IN_ARCHIVE_RECOVERY to match with the set of WAL
          * segments recycled below.
          */
         ControlFile->checkPoint = lastCheckPointRecPtr;
         ControlFile->checkPointCopy = lastCheckPoint;
  
         /*
          * Ensure minRecoveryPoint is past the checkpoint record and update it
          * if the control file still shows DB_IN_ARCHIVE_RECOVERY.  Normally,
          * this will have happened already while writing out dirty buffers,
          * but not necessarily - e.g. because no buffers were dirtied.  We do
          * this because a backup performed in recovery uses minRecoveryPoint
          * to determine which WAL files must be included in the backup, and
          * the file (or files) containing the checkpoint record must be
          * included, at a minimum.  Note that for an ordinary restart of
          * recovery there's no value in having the minimum recovery point any
          * earlier than this anyway, because redo will begin just after the
          * checkpoint record.
          */
         if (ControlFile->state == DB_IN_ARCHIVE_RECOVERY)
         {
             if (ControlFile->minRecoveryPoint < lastCheckPointEndPtr)
             {
                 ControlFile->minRecoveryPoint = lastCheckPointEndPtr;
                 ControlFile->minRecoveryPointTLI = lastCheckPoint.ThisTimeLineID;
  
                 /* update local copy */
                 LocalMinRecoveryPoint = ControlFile->minRecoveryPoint;
                 LocalMinRecoveryPointTLI = ControlFile->minRecoveryPointTLI;
             }
             if (flags & CHECKPOINT_IS_SHUTDOWN)
                 ControlFile->state = DB_SHUTDOWNED_IN_RECOVERY;
         }
         UpdateControlFile();
     }
     LWLockRelease(ControlFileLock);
  
     /*
      * Update the average distance between checkpoints/restartpoints if the
      * prior checkpoint exists.
      */
     if (PriorRedoPtr != InvalidXLogRecPtr)
         UpdateCheckPointDistanceEstimate(RedoRecPtr - PriorRedoPtr);
  
     /*
      * Delete old log files, those no longer needed for last restartpoint to
      * prevent the disk holding the xlog from growing full.
      */
     XLByteToSeg(RedoRecPtr, _logSegNo, wal_segment_size);
  
     /*
      * Retreat _logSegNo using the current end of xlog replayed or received,
      * whichever is later.
      */
     receivePtr = GetWalRcvFlushRecPtr(NULL, NULL);
     replayPtr = GetXLogReplayRecPtr(&replayTLI);
     endptr = (receivePtr < replayPtr) ? replayPtr : receivePtr;
     KeepLogSeg(endptr, &_logSegNo);
     if (InvalidateObsoleteReplicationSlots(RS_INVAL_WAL_REMOVED,
                                            _logSegNo, InvalidOid,
                                            InvalidTransactionId))
     {
         /*
          * Some slots have been invalidated; recalculate the old-segment
          * horizon, starting again from RedoRecPtr.
          */
         XLByteToSeg(RedoRecPtr, _logSegNo, wal_segment_size);
         KeepLogSeg(endptr, &_logSegNo);
     }
     _logSegNo--;
  
     /*
      * Try to recycle segments on a useful timeline. If we've been promoted
      * since the beginning of this restartpoint, use the new timeline chosen
      * at end of recovery.  If we're still in recovery, use the timeline we're
      * currently replaying.
      *
      * There is no guarantee that the WAL segments will be useful on the
      * current timeline; if recovery proceeds to a new timeline right after
      * this, the pre-allocated WAL segments on this timeline will not be used,
      * and will go wasted until recycled on the next restartpoint. We'll live
      * with that.
      */
     if (!RecoveryInProgress())
         replayTLI = XLogCtl->InsertTimeLineID;
  
     RemoveOldXlogFiles(_logSegNo, RedoRecPtr, endptr, replayTLI);
  
     /*
      * Make more log segments if needed.  (Do this after recycling old log
      * segments, since that may supply some of the needed files.)
      */
     PreallocXlogFiles(endptr, replayTLI);
  
     /*
      * Truncate pg_subtrans if possible.  We can throw away all data before
      * the oldest XMIN of any running transaction.  No future transaction will
      * attempt to reference any pg_subtrans entry older than that (see Asserts
      * in subtrans.c).  When hot standby is disabled, though, we mustn't do
      * this because StartupSUBTRANS hasn't been called yet.
      */
     if (EnableHotStandby)
         TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning());
  
     /* Real work is done; log and update stats. */
     LogCheckpointEnd(true);
  
     /* Reset the process title */
     update_checkpoint_display(flags, true, true);
  
     xtime = GetLatestXTime();
     ereport((log_checkpoints ? LOG : DEBUG2),
             (errmsg("recovery restart point at %X/%X",
                     LSN_FORMAT_ARGS(lastCheckPoint.redo)),
              xtime ? errdetail("Last completed transaction was at log time %s.",
                                timestamptz_to_str(xtime)) : 0));
  
     /*
      * Finally, execute archive_cleanup_command, if any.
      */
     if (archiveCleanupCommand && strcmp(archiveCleanupCommand, "") != 0)
         ExecuteRecoveryCommand(archiveCleanupCommand,
                                "archive_cleanup_command",
                                false,
                                WAIT_EVENT_ARCHIVE_CLEANUP_COMMAND);
  
     return true;
 }

References archiveCleanupCommand, Assert, B_CHECKPOINTER, ControlFileData::checkPoint, CHECKPOINT_IS_SHUTDOWN, ControlFileData::checkPointCopy, CheckPointGuts(), CheckpointStats, CheckpointStatsData::ckpt_start_t, ControlFile, DB_IN_ARCHIVE_RECOVERY, DB_SHUTDOWNED_IN_RECOVERY, DEBUG2, EnableHotStandby, ereport, errdetail(), errmsg(), errmsg_internal(), ExecuteRecoveryCommand(), GetCurrentTimestamp(), GetLatestXTime(), GetOldestTransactionIdConsideredRunning(), GetWalRcvFlushRecPtr(), GetXLogReplayRecPtr(), XLogCtlData::info_lck, INJECTION_POINT, XLogCtlData::Insert, XLogCtlData::InsertTimeLineID, InvalidateObsoleteReplicationSlots(), InvalidOid, InvalidTransactionId, InvalidXLogRecPtr, IsUnderPostmaster, KeepLogSeg(), XLogCtlData::lastCheckPoint, XLogCtlData::lastCheckPointEndPtr, XLogCtlData::lastCheckPointRecPtr, LocalMinRecoveryPoint, LocalMinRecoveryPointTLI, LOG, log_checkpoints, LogCheckpointEnd(), LogCheckpointStart(), LSN_FORMAT_ARGS, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), MemSet, ControlFileData::minRecoveryPoint, ControlFileData::minRecoveryPointTLI, MyBackendType, PreallocXlogFiles(), RecoveryInProgress(), CheckPoint::redo, RedoRecPtr, XLogCtlInsert::RedoRecPtr, XLogCtlData::RedoRecPtr, RemoveOldXlogFiles(), RS_INVAL_WAL_REMOVED, SpinLockAcquire, SpinLockRelease, ControlFileData::state, CheckPoint::ThisTimeLineID, timestamptz_to_str(), TruncateSUBTRANS(), update_checkpoint_display(), UpdateCheckPointDistanceEstimate(), UpdateControlFile(), UpdateMinRecoveryPoint(), wal_segment_size, WALInsertLockAcquireExclusive(), WALInsertLockRelease(), XLByteToSeg, XLogCtl, and XLogRecPtrIsInvalid.

Referenced by CheckpointerMain(), and ShutdownXLOG().

◆ DataChecksumsEnabled()

bool DataChecksumsEnabled ( void )

Definition at line 4553 of file xlog.c.

 {
     Assert(ControlFile != NULL);
     return (ControlFile->data_checksum_version > 0);
 }

References Assert, ControlFile, and ControlFileData::data_checksum_version.

Referenced by PageIsVerifiedExtended(), PageSetChecksumCopy(), PageSetChecksumInplace(), pg_stat_get_db_checksum_failures(), pg_stat_get_db_checksum_last_failure(), ReadControlFile(), and sendFile().

◆ do_pg_abort_backup()

void do_pg_abort_backup	(	int	code,
		Datum	arg
	)

Definition at line 9324 of file xlog.c.

 {
     bool        during_backup_start = DatumGetBool(arg);
  
     /* If called during backup start, there shouldn't be one already running */
     Assert(!during_backup_start || sessionBackupState == SESSION_BACKUP_NONE);
  
     if (during_backup_start || sessionBackupState != SESSION_BACKUP_NONE)
     {
         WALInsertLockAcquireExclusive();
         Assert(XLogCtl->Insert.runningBackups > 0);
         XLogCtl->Insert.runningBackups--;
  
         sessionBackupState = SESSION_BACKUP_NONE;
         WALInsertLockRelease();
  
         if (!during_backup_start)
             ereport(WARNING,
                     errmsg("aborting backup due to backend exiting before pg_backup_stop was called"));
     }
 }

References arg, Assert, DatumGetBool(), ereport, errmsg(), XLogCtlData::Insert, XLogCtlInsert::runningBackups, SESSION_BACKUP_NONE, sessionBackupState, WALInsertLockAcquireExclusive(), WALInsertLockRelease(), WARNING, and XLogCtl.

Referenced by do_pg_backup_start(), perform_base_backup(), and register_persistent_abort_backup_handler().

◆ do_pg_backup_start()

void do_pg_backup_start	(	const char *	backupidstr,
		bool	fast,
		List **	tablespaces,
		BackupState *	state,
		StringInfo	tblspcmapfile
	)

Definition at line 8722 of file xlog.c.

 {
     bool        backup_started_in_recovery;
  
     Assert(state != NULL);
     backup_started_in_recovery = RecoveryInProgress();
  
     /*
      * During recovery, we don't need to check WAL level. Because, if WAL
      * level is not sufficient, it's impossible to get here during recovery.
      */
     if (!backup_started_in_recovery && !XLogIsNeeded())
         ereport(ERROR,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("WAL level not sufficient for making an online backup"),
                  errhint("\"wal_level\" must be set to \"replica\" or \"logical\" at server start.")));
  
     if (strlen(backupidstr) > MAXPGPATH)
         ereport(ERROR,
                 (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
                  errmsg("backup label too long (max %d bytes)",
                         MAXPGPATH)));
  
     strlcpy(state->name, backupidstr, sizeof(state->name));
  
     /*
      * Mark backup active in shared memory.  We must do full-page WAL writes
      * during an on-line backup even if not doing so at other times, because
      * it's quite possible for the backup dump to obtain a "torn" (partially
      * written) copy of a database page if it reads the page concurrently with
      * our write to the same page.  This can be fixed as long as the first
      * write to the page in the WAL sequence is a full-page write. Hence, we
      * increment runningBackups then force a CHECKPOINT, to ensure there are
      * no dirty pages in shared memory that might get dumped while the backup
      * is in progress without having a corresponding WAL record.  (Once the
      * backup is complete, we need not force full-page writes anymore, since
      * we expect that any pages not modified during the backup interval must
      * have been correctly captured by the backup.)
      *
      * Note that forcing full-page writes has no effect during an online
      * backup from the standby.
      *
      * We must hold all the insertion locks to change the value of
      * runningBackups, to ensure adequate interlocking against
      * XLogInsertRecord().
      */
     WALInsertLockAcquireExclusive();
     XLogCtl->Insert.runningBackups++;
     WALInsertLockRelease();
  
     /*
      * Ensure we decrement runningBackups if we fail below. NB -- for this to
      * work correctly, it is critical that sessionBackupState is only updated
      * after this block is over.
      */
     PG_ENSURE_ERROR_CLEANUP(do_pg_abort_backup, DatumGetBool(true));
     {
         bool        gotUniqueStartpoint = false;
         DIR        *tblspcdir;
         struct dirent *de;
         tablespaceinfo *ti;
         int         datadirpathlen;
  
         /*
          * Force an XLOG file switch before the checkpoint, to ensure that the
          * WAL segment the checkpoint is written to doesn't contain pages with
          * old timeline IDs.  That would otherwise happen if you called
          * pg_backup_start() right after restoring from a PITR archive: the
          * first WAL segment containing the startup checkpoint has pages in
          * the beginning with the old timeline ID.  That can cause trouble at
          * recovery: we won't have a history file covering the old timeline if
          * pg_wal directory was not included in the base backup and the WAL
          * archive was cleared too before starting the backup.
          *
          * This also ensures that we have emitted a WAL page header that has
          * XLP_BKP_REMOVABLE off before we emit the checkpoint record.
          * Therefore, if a WAL archiver (such as pglesslog) is trying to
          * compress out removable backup blocks, it won't remove any that
          * occur after this point.
          *
          * During recovery, we skip forcing XLOG file switch, which means that
          * the backup taken during recovery is not available for the special
          * recovery case described above.
          */
         if (!backup_started_in_recovery)
             RequestXLogSwitch(false);
  
         do
         {
             bool        checkpointfpw;
  
             /*
              * Force a CHECKPOINT.  Aside from being necessary to prevent torn
              * page problems, this guarantees that two successive backup runs
              * will have different checkpoint positions and hence different
              * history file names, even if nothing happened in between.
              *
              * During recovery, establish a restartpoint if possible. We use
              * the last restartpoint as the backup starting checkpoint. This
              * means that two successive backup runs can have same checkpoint
              * positions.
              *
              * Since the fact that we are executing do_pg_backup_start()
              * during recovery means that checkpointer is running, we can use
              * RequestCheckpoint() to establish a restartpoint.
              *
              * We use CHECKPOINT_IMMEDIATE only if requested by user (via
              * passing fast = true).  Otherwise this can take awhile.
              */
             RequestCheckpoint(CHECKPOINT_FORCE | CHECKPOINT_WAIT |
                               (fast ? CHECKPOINT_IMMEDIATE : 0));
  
             /*
              * Now we need to fetch the checkpoint record location, and also
              * its REDO pointer.  The oldest point in WAL that would be needed
              * to restore starting from the checkpoint is precisely the REDO
              * pointer.
              */
             LWLockAcquire(ControlFileLock, LW_SHARED);
             state->checkpointloc = ControlFile->checkPoint;
             state->startpoint = ControlFile->checkPointCopy.redo;
             state->starttli = ControlFile->checkPointCopy.ThisTimeLineID;
             checkpointfpw = ControlFile->checkPointCopy.fullPageWrites;
             LWLockRelease(ControlFileLock);
  
             if (backup_started_in_recovery)
             {
                 XLogRecPtr  recptr;
  
                 /*
                  * Check to see if all WAL replayed during online backup
                  * (i.e., since last restartpoint used as backup starting
                  * checkpoint) contain full-page writes.
                  */
                 SpinLockAcquire(&XLogCtl->info_lck);
                 recptr = XLogCtl->lastFpwDisableRecPtr;
                 SpinLockRelease(&XLogCtl->info_lck);
  
                 if (!checkpointfpw || state->startpoint <= recptr)
                     ereport(ERROR,
                             (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                              errmsg("WAL generated with \"full_page_writes=off\" was replayed "
                                     "since last restartpoint"),
                              errhint("This means that the backup being taken on the standby "
                                      "is corrupt and should not be used. "
                                      "Enable \"full_page_writes\" and run CHECKPOINT on the primary, "
                                      "and then try an online backup again.")));
  
                 /*
                  * During recovery, since we don't use the end-of-backup WAL
                  * record and don't write the backup history file, the
                  * starting WAL location doesn't need to be unique. This means
                  * that two base backups started at the same time might use
                  * the same checkpoint as starting locations.
                  */
                 gotUniqueStartpoint = true;
             }
  
             /*
              * If two base backups are started at the same time (in WAL sender
              * processes), we need to make sure that they use different
              * checkpoints as starting locations, because we use the starting
              * WAL location as a unique identifier for the base backup in the
              * end-of-backup WAL record and when we write the backup history
              * file. Perhaps it would be better generate a separate unique ID
              * for each backup instead of forcing another checkpoint, but
              * taking a checkpoint right after another is not that expensive
              * either because only few buffers have been dirtied yet.
              */
             WALInsertLockAcquireExclusive();
             if (XLogCtl->Insert.lastBackupStart < state->startpoint)
             {
                 XLogCtl->Insert.lastBackupStart = state->startpoint;
                 gotUniqueStartpoint = true;
             }
             WALInsertLockRelease();
         } while (!gotUniqueStartpoint);
  
         /*
          * Construct tablespace_map file.
          */
         datadirpathlen = strlen(DataDir);
  
         /* Collect information about all tablespaces */
         tblspcdir = AllocateDir("pg_tblspc");
         while ((de = ReadDir(tblspcdir, "pg_tblspc")) != NULL)
         {
             char        fullpath[MAXPGPATH + 10];
             char        linkpath[MAXPGPATH];
             char       *relpath = NULL;
             char       *s;
             PGFileType  de_type;
             char       *badp;
             Oid         tsoid;
  
             /*
              * Try to parse the directory name as an unsigned integer.
              *
              * Tablespace directories should be positive integers that can be
              * represented in 32 bits, with no leading zeroes or trailing
              * garbage. If we come across a name that doesn't meet those
              * criteria, skip it.
              */
             if (de->d_name[0] < '1' || de->d_name[1] > '9')
                 continue;
             errno = 0;
             tsoid = strtoul(de->d_name, &badp, 10);
             if (*badp != '\0' || errno == EINVAL || errno == ERANGE)
                 continue;
  
             snprintf(fullpath, sizeof(fullpath), "pg_tblspc/%s", de->d_name);
  
             de_type = get_dirent_type(fullpath, de, false, ERROR);
  
             if (de_type == PGFILETYPE_LNK)
             {
                 StringInfoData escapedpath;
                 int         rllen;
  
                 rllen = readlink(fullpath, linkpath, sizeof(linkpath));
                 if (rllen < 0)
                 {
                     ereport(WARNING,
                             (errmsg("could not read symbolic link \"%s\": %m",
                                     fullpath)));
                     continue;
                 }
                 else if (rllen >= sizeof(linkpath))
                 {
                     ereport(WARNING,
                             (errmsg("symbolic link \"%s\" target is too long",
                                     fullpath)));
                     continue;
                 }
                 linkpath[rllen] = '\0';
  
                 /*
                  * Relpath holds the relative path of the tablespace directory
                  * when it's located within PGDATA, or NULL if it's located
                  * elsewhere.
                  */
                 if (rllen > datadirpathlen &&
                     strncmp(linkpath, DataDir, datadirpathlen) == 0 &&
                     IS_DIR_SEP(linkpath[datadirpathlen]))
                     relpath = pstrdup(linkpath + datadirpathlen + 1);
  
                 /*
                  * Add a backslash-escaped version of the link path to the
                  * tablespace map file.
                  */
                 initStringInfo(&escapedpath);
                 for (s = linkpath; *s; s++)
                 {
                     if (*s == '\n' || *s == '\r' || *s == '\\')
                         appendStringInfoChar(&escapedpath, '\\');
                     appendStringInfoChar(&escapedpath, *s);
                 }
                 appendStringInfo(tblspcmapfile, "%s %s\n",
                                  de->d_name, escapedpath.data);
                 pfree(escapedpath.data);
             }
             else if (de_type == PGFILETYPE_DIR)
             {
                 /*
                  * It's possible to use allow_in_place_tablespaces to create
                  * directories directly under pg_tblspc, for testing purposes
                  * only.
                  *
                  * In this case, we store a relative path rather than an
                  * absolute path into the tablespaceinfo.
                  */
                 snprintf(linkpath, sizeof(linkpath), "pg_tblspc/%s",
                          de->d_name);
                 relpath = pstrdup(linkpath);
             }
             else
             {
                 /* Skip any other file type that appears here. */
                 continue;
             }
  
             ti = palloc(sizeof(tablespaceinfo));
             ti->oid = tsoid;
             ti->path = pstrdup(linkpath);
             ti->rpath = relpath;
             ti->size = -1;
  
             if (tablespaces)
                 *tablespaces = lappend(*tablespaces, ti);
         }
         FreeDir(tblspcdir);
  
         state->starttime = (pg_time_t) time(NULL);
     }
     PG_END_ENSURE_ERROR_CLEANUP(do_pg_abort_backup, DatumGetBool(true));
  
     state->started_in_recovery = backup_started_in_recovery;
  
     /*
      * Mark that the start phase has correctly finished for the backup.
      */
     sessionBackupState = SESSION_BACKUP_RUNNING;
 }

References AllocateDir(), appendStringInfo(), appendStringInfoChar(), Assert, backup_started_in_recovery, ControlFileData::checkPoint, CHECKPOINT_FORCE, CHECKPOINT_IMMEDIATE, CHECKPOINT_WAIT, ControlFileData::checkPointCopy, ControlFile, dirent::d_name, StringInfoData::data, DataDir, DatumGetBool(), do_pg_abort_backup(), ereport, errcode(), errhint(), errmsg(), ERROR, FreeDir(), CheckPoint::fullPageWrites, get_dirent_type(), XLogCtlData::info_lck, initStringInfo(), XLogCtlData::Insert, IS_DIR_SEP, lappend(), XLogCtlInsert::lastBackupStart, XLogCtlData::lastFpwDisableRecPtr, LW_SHARED, LWLockAcquire(), LWLockRelease(), MAXPGPATH, tablespaceinfo::oid, palloc(), tablespaceinfo::path, pfree(), PG_END_ENSURE_ERROR_CLEANUP, PG_ENSURE_ERROR_CLEANUP, PGFILETYPE_DIR, PGFILETYPE_LNK, pstrdup(), ReadDir(), readlink, RecoveryInProgress(), CheckPoint::redo, relpath, RequestCheckpoint(), RequestXLogSwitch(), tablespaceinfo::rpath, XLogCtlInsert::runningBackups, SESSION_BACKUP_RUNNING, sessionBackupState, tablespaceinfo::size, snprintf, SpinLockAcquire, SpinLockRelease, strlcpy(), CheckPoint::ThisTimeLineID, WALInsertLockAcquireExclusive(), WALInsertLockRelease(), WARNING, XLogCtl, and XLogIsNeeded.

Referenced by perform_base_backup(), and pg_backup_start().

◆ do_pg_backup_stop()

void do_pg_backup_stop	(	BackupState *	state,
		bool	waitforarchive
	)

Definition at line 9050 of file xlog.c.

 {
     bool        backup_stopped_in_recovery = false;
     char        histfilepath[MAXPGPATH];
     char        lastxlogfilename[MAXFNAMELEN];
     char        histfilename[MAXFNAMELEN];
     XLogSegNo   _logSegNo;
     FILE       *fp;
     int         seconds_before_warning;
     int         waits = 0;
     bool        reported_waiting = false;
  
     Assert(state != NULL);
  
     backup_stopped_in_recovery = RecoveryInProgress();
  
     /*
      * During recovery, we don't need to check WAL level. Because, if WAL
      * level is not sufficient, it's impossible to get here during recovery.
      */
     if (!backup_stopped_in_recovery && !XLogIsNeeded())
         ereport(ERROR,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("WAL level not sufficient for making an online backup"),
                  errhint("wal_level must be set to \"replica\" or \"logical\" at server start.")));
  
     /*
      * OK to update backup counter and session-level lock.
      *
      * Note that CHECK_FOR_INTERRUPTS() must not occur while updating them,
      * otherwise they can be updated inconsistently, which might cause
      * do_pg_abort_backup() to fail.
      */
     WALInsertLockAcquireExclusive();
  
     /*
      * It is expected that each do_pg_backup_start() call is matched by
      * exactly one do_pg_backup_stop() call.
      */
     Assert(XLogCtl->Insert.runningBackups > 0);
     XLogCtl->Insert.runningBackups--;
  
     /*
      * Clean up session-level lock.
      *
      * You might think that WALInsertLockRelease() can be called before
      * cleaning up session-level lock because session-level lock doesn't need
      * to be protected with WAL insertion lock. But since
      * CHECK_FOR_INTERRUPTS() can occur in it, session-level lock must be
      * cleaned up before it.
      */
     sessionBackupState = SESSION_BACKUP_NONE;
  
     WALInsertLockRelease();
  
     /*
      * If we are taking an online backup from the standby, we confirm that the
      * standby has not been promoted during the backup.
      */
     if (state->started_in_recovery && !backup_stopped_in_recovery)
         ereport(ERROR,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("the standby was promoted during online backup"),
                  errhint("This means that the backup being taken is corrupt "
                          "and should not be used. "
                          "Try taking another online backup.")));
  
     /*
      * During recovery, we don't write an end-of-backup record. We assume that
      * pg_control was backed up last and its minimum recovery point can be
      * available as the backup end location. Since we don't have an
      * end-of-backup record, we use the pg_control value to check whether
      * we've reached the end of backup when starting recovery from this
      * backup. We have no way of checking if pg_control wasn't backed up last
      * however.
      *
      * We don't force a switch to new WAL file but it is still possible to
      * wait for all the required files to be archived if waitforarchive is
      * true. This is okay if we use the backup to start a standby and fetch
      * the missing WAL using streaming replication. But in the case of an
      * archive recovery, a user should set waitforarchive to true and wait for
      * them to be archived to ensure that all the required files are
      * available.
      *
      * We return the current minimum recovery point as the backup end
      * location. Note that it can be greater than the exact backup end
      * location if the minimum recovery point is updated after the backup of
      * pg_control. This is harmless for current uses.
      *
      * XXX currently a backup history file is for informational and debug
      * purposes only. It's not essential for an online backup. Furthermore,
      * even if it's created, it will not be archived during recovery because
      * an archiver is not invoked. So it doesn't seem worthwhile to write a
      * backup history file during recovery.
      */
     if (backup_stopped_in_recovery)
     {
         XLogRecPtr  recptr;
  
         /*
          * Check to see if all WAL replayed during online backup contain
          * full-page writes.
          */
         SpinLockAcquire(&XLogCtl->info_lck);
         recptr = XLogCtl->lastFpwDisableRecPtr;
         SpinLockRelease(&XLogCtl->info_lck);
  
         if (state->startpoint <= recptr)
             ereport(ERROR,
                     (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                      errmsg("WAL generated with \"full_page_writes=off\" was replayed "
                             "during online backup"),
                      errhint("This means that the backup being taken on the standby "
                              "is corrupt and should not be used. "
                              "Enable \"full_page_writes\" and run CHECKPOINT on the primary, "
                              "and then try an online backup again.")));
  
  
         LWLockAcquire(ControlFileLock, LW_SHARED);
         state->stoppoint = ControlFile->minRecoveryPoint;
         state->stoptli = ControlFile->minRecoveryPointTLI;
         LWLockRelease(ControlFileLock);
     }
     else
     {
         char       *history_file;
  
         /*
          * Write the backup-end xlog record
          */
         XLogBeginInsert();
         XLogRegisterData((char *) (&state->startpoint),
                          sizeof(state->startpoint));
         state->stoppoint = XLogInsert(RM_XLOG_ID, XLOG_BACKUP_END);
  
         /*
          * Given that we're not in recovery, InsertTimeLineID is set and can't
          * change, so we can read it without a lock.
          */
         state->stoptli = XLogCtl->InsertTimeLineID;
  
         /*
          * Force a switch to a new xlog segment file, so that the backup is
          * valid as soon as archiver moves out the current segment file.
          */
         RequestXLogSwitch(false);
  
         state->stoptime = (pg_time_t) time(NULL);
  
         /*
          * Write the backup history file
          */
         XLByteToSeg(state->startpoint, _logSegNo, wal_segment_size);
         BackupHistoryFilePath(histfilepath, state->stoptli, _logSegNo,
                               state->startpoint, wal_segment_size);
         fp = AllocateFile(histfilepath, "w");
         if (!fp)
             ereport(ERROR,
                     (errcode_for_file_access(),
                      errmsg("could not create file \"%s\": %m",
                             histfilepath)));
  
         /* Build and save the contents of the backup history file */
         history_file = build_backup_content(state, true);
         fprintf(fp, "%s", history_file);
         pfree(history_file);
  
         if (fflush(fp) || ferror(fp) || FreeFile(fp))
             ereport(ERROR,
                     (errcode_for_file_access(),
                      errmsg("could not write file \"%s\": %m",
                             histfilepath)));
  
         /*
          * Clean out any no-longer-needed history files.  As a side effect,
          * this will post a .ready file for the newly created history file,
          * notifying the archiver that history file may be archived
          * immediately.
          */
         CleanupBackupHistory();
     }
  
     /*
      * If archiving is enabled, wait for all the required WAL files to be
      * archived before returning. If archiving isn't enabled, the required WAL
      * needs to be transported via streaming replication (hopefully with
      * wal_keep_size set high enough), or some more exotic mechanism like
      * polling and copying files from pg_wal with script. We have no knowledge
      * of those mechanisms, so it's up to the user to ensure that he gets all
      * the required WAL.
      *
      * We wait until both the last WAL file filled during backup and the
      * history file have been archived, and assume that the alphabetic sorting
      * property of the WAL files ensures any earlier WAL files are safely
      * archived as well.
      *
      * We wait forever, since archive_command is supposed to work and we
      * assume the admin wanted his backup to work completely. If you don't
      * wish to wait, then either waitforarchive should be passed in as false,
      * or you can set statement_timeout.  Also, some notices are issued to
      * clue in anyone who might be doing this interactively.
      */
  
     if (waitforarchive &&
         ((!backup_stopped_in_recovery && XLogArchivingActive()) ||
          (backup_stopped_in_recovery && XLogArchivingAlways())))
     {
         XLByteToPrevSeg(state->stoppoint, _logSegNo, wal_segment_size);
         XLogFileName(lastxlogfilename, state->stoptli, _logSegNo,
                      wal_segment_size);
  
         XLByteToSeg(state->startpoint, _logSegNo, wal_segment_size);
         BackupHistoryFileName(histfilename, state->stoptli, _logSegNo,
                               state->startpoint, wal_segment_size);
  
         seconds_before_warning = 60;
         waits = 0;
  
         while (XLogArchiveIsBusy(lastxlogfilename) ||
                XLogArchiveIsBusy(histfilename))
         {
             CHECK_FOR_INTERRUPTS();
  
             if (!reported_waiting && waits > 5)
             {
                 ereport(NOTICE,
                         (errmsg("base backup done, waiting for required WAL segments to be archived")));
                 reported_waiting = true;
             }
  
             (void) WaitLatch(MyLatch,
                              WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH,
                              1000L,
                              WAIT_EVENT_BACKUP_WAIT_WAL_ARCHIVE);
             ResetLatch(MyLatch);
  
             if (++waits >= seconds_before_warning)
             {
                 seconds_before_warning *= 2;    /* This wraps in >10 years... */
                 ereport(WARNING,
                         (errmsg("still waiting for all required WAL segments to be archived (%d seconds elapsed)",
                                 waits),
                          errhint("Check that your \"archive_command\" is executing properly.  "
                                  "You can safely cancel this backup, "
                                  "but the database backup will not be usable without all the WAL segments.")));
             }
         }
  
         ereport(NOTICE,
                 (errmsg("all required WAL segments have been archived")));
     }
     else if (waitforarchive)
         ereport(NOTICE,
                 (errmsg("WAL archiving is not enabled; you must ensure that all required WAL segments are copied through other means to complete the backup")));
 }

References AllocateFile(), Assert, BackupHistoryFileName(), BackupHistoryFilePath(), build_backup_content(), CHECK_FOR_INTERRUPTS, CleanupBackupHistory(), ControlFile, ereport, errcode(), errcode_for_file_access(), errhint(), errmsg(), ERROR, fflush(), fprintf, FreeFile(), XLogCtlData::info_lck, XLogCtlData::Insert, XLogCtlData::InsertTimeLineID, XLogCtlData::lastFpwDisableRecPtr, LW_SHARED, LWLockAcquire(), LWLockRelease(), MAXFNAMELEN, MAXPGPATH, ControlFileData::minRecoveryPoint, ControlFileData::minRecoveryPointTLI, MyLatch, NOTICE, pfree(), RecoveryInProgress(), RequestXLogSwitch(), ResetLatch(), XLogCtlInsert::runningBackups, SESSION_BACKUP_NONE, sessionBackupState, SpinLockAcquire, SpinLockRelease, WaitLatch(), wal_segment_size, WALInsertLockAcquireExclusive(), WALInsertLockRelease(), WARNING, WL_EXIT_ON_PM_DEATH, WL_LATCH_SET, WL_TIMEOUT, XLByteToPrevSeg, XLByteToSeg, XLOG_BACKUP_END, XLogArchiveIsBusy(), XLogArchivingActive, XLogArchivingAlways, XLogBeginInsert(), XLogCtl, XLogFileName(), XLogInsert(), XLogIsNeeded, and XLogRegisterData().

Referenced by perform_base_backup(), and pg_backup_stop().

◆ get_backup_status()

SessionBackupState get_backup_status ( void )

Definition at line 9031 of file xlog.c.

 {
     return sessionBackupState;
 }

References sessionBackupState.

Referenced by pg_backup_start(), pg_backup_stop(), and SendBaseBackup().

◆ get_sync_bit()

static int get_sync_bit ( int method )

static

Definition at line 8523 of file xlog.c.

 {
     int         o_direct_flag = 0;
  
     /*
      * Use O_DIRECT if requested, except in walreceiver process.  The WAL
      * written by walreceiver is normally read by the startup process soon
      * after it's written.  Also, walreceiver performs unaligned writes, which
      * don't work with O_DIRECT, so it is required for correctness too.
      */
     if ((io_direct_flags & IO_DIRECT_WAL) && !AmWalReceiverProcess())
         o_direct_flag = PG_O_DIRECT;
  
     /* If fsync is disabled, never open in sync mode */
     if (!enableFsync)
         return o_direct_flag;
  
     switch (method)
     {
             /*
              * enum values for all sync options are defined even if they are
              * not supported on the current platform.  But if not, they are
              * not included in the enum option array, and therefore will never
              * be seen here.
              */
         case WAL_SYNC_METHOD_FSYNC:
         case WAL_SYNC_METHOD_FSYNC_WRITETHROUGH:
         case WAL_SYNC_METHOD_FDATASYNC:
             return o_direct_flag;
 #ifdef O_SYNC
         case WAL_SYNC_METHOD_OPEN:
             return O_SYNC | o_direct_flag;
 #endif
 #ifdef O_DSYNC
         case WAL_SYNC_METHOD_OPEN_DSYNC:
             return O_DSYNC | o_direct_flag;
 #endif
         default:
             /* can't happen (unless we are out of sync with option array) */
             elog(ERROR, "unrecognized \"wal_sync_method\": %d", method);
             return 0;           /* silence warning */
     }
 }

References AmWalReceiverProcess, elog, enableFsync, ERROR, io_direct_flags, IO_DIRECT_WAL, O_DSYNC, PG_O_DIRECT, WAL_SYNC_METHOD_FDATASYNC, WAL_SYNC_METHOD_FSYNC, WAL_SYNC_METHOD_FSYNC_WRITETHROUGH, WAL_SYNC_METHOD_OPEN, and WAL_SYNC_METHOD_OPEN_DSYNC.

Referenced by assign_wal_sync_method(), XLogFileInit(), XLogFileInitInternal(), and XLogFileOpen().

◆ GetActiveWalLevelOnStandby()

WalLevel GetActiveWalLevelOnStandby ( void )

Definition at line 4824 of file xlog.c.

 {
     return ControlFile->wal_level;
 }

References ControlFile, and ControlFileData::wal_level.

Referenced by CheckLogicalDecodingRequirements().

◆ GetFakeLSNForUnloggedRel()

XLogRecPtr GetFakeLSNForUnloggedRel ( void )

Definition at line 4569 of file xlog.c.

 {
     return pg_atomic_fetch_add_u64(&XLogCtl->unloggedLSN, 1);
 }

References pg_atomic_fetch_add_u64(), XLogCtlData::unloggedLSN, and XLogCtl.

Referenced by gistGetFakeLSN().

◆ GetFlushRecPtr()

XLogRecPtr GetFlushRecPtr ( TimeLineID * insertTLI )

Definition at line 6455 of file xlog.c.

 {
     Assert(XLogCtl->SharedRecoveryState == RECOVERY_STATE_DONE);
  
     RefreshXLogWriteResult(LogwrtResult);
  
     /*
      * If we're writing and flushing WAL, the time line can't be changing, so
      * no lock is required.
      */
     if (insertTLI)
         *insertTLI = XLogCtl->InsertTimeLineID;
  
     return LogwrtResult.Flush;
 }

References Assert, XLogwrtResult::Flush, XLogCtlData::InsertTimeLineID, LogwrtResult, RECOVERY_STATE_DONE, RefreshXLogWriteResult, XLogCtlData::SharedRecoveryState, and XLogCtl.

Referenced by binary_upgrade_logical_slot_has_caught_up(), get_flush_position(), GetCurrentLSN(), GetLatestLSN(), IdentifySystem(), pg_current_wal_flush_lsn(), pg_logical_slot_get_changes_guts(), pg_replication_slot_advance(), read_local_xlog_page_guts(), StartReplication(), WalSndWaitForWal(), XLogSendLogical(), and XLogSendPhysical().

◆ GetFullPageWriteInfo()

void GetFullPageWriteInfo	(	XLogRecPtr *	RedoRecPtr_p,
		bool *	doPageWrites_p
	)

Definition at line 6423 of file xlog.c.

 {
     *RedoRecPtr_p = RedoRecPtr;
     *doPageWrites_p = doPageWrites;
 }

References doPageWrites, and RedoRecPtr.

Referenced by XLogCheckBufferNeedsBackup(), and XLogInsert().

◆ GetInsertRecPtr()

XLogRecPtr GetInsertRecPtr ( void )

Definition at line 6438 of file xlog.c.

 {
     XLogRecPtr  recptr;
  
     SpinLockAcquire(&XLogCtl->info_lck);
     recptr = XLogCtl->LogwrtRqst.Write;
     SpinLockRelease(&XLogCtl->info_lck);
  
     return recptr;
 }

References XLogCtlData::info_lck, XLogCtlData::LogwrtRqst, SpinLockAcquire, SpinLockRelease, XLogwrtRqst::Write, and XLogCtl.

Referenced by CheckpointerMain(), gistvacuumscan(), and IsCheckpointOnSchedule().

◆ GetLastImportantRecPtr()

XLogRecPtr GetLastImportantRecPtr ( void )

Definition at line 6493 of file xlog.c.

 {
     XLogRecPtr  res = InvalidXLogRecPtr;
     int         i;
  
     for (i = 0; i < NUM_XLOGINSERT_LOCKS; i++)
     {
         XLogRecPtr  last_important;
  
         /*
          * Need to take a lock to prevent torn reads of the LSN, which are
          * possible on some of the supported platforms. WAL insert locks only
          * support exclusive mode, so we have to use that.
          */
         LWLockAcquire(&WALInsertLocks[i].l.lock, LW_EXCLUSIVE);
         last_important = WALInsertLocks[i].l.lastImportantAt;
         LWLockRelease(&WALInsertLocks[i].l.lock);
  
         if (res < last_important)
             res = last_important;
     }
  
     return res;
 }

References i, InvalidXLogRecPtr, WALInsertLockPadded::l, WALInsertLock::lastImportantAt, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), NUM_XLOGINSERT_LOCKS, res, and WALInsertLocks.

Referenced by BackgroundWriterMain(), CheckArchiveTimeout(), and CreateCheckPoint().

◆ GetLastSegSwitchData()

pg_time_t GetLastSegSwitchData ( XLogRecPtr * lastSwitchLSN )

Definition at line 6522 of file xlog.c.

 {
     pg_time_t   result;
  
     /* Need WALWriteLock, but shared lock is sufficient */
     LWLockAcquire(WALWriteLock, LW_SHARED);
     result = XLogCtl->lastSegSwitchTime;
     *lastSwitchLSN = XLogCtl->lastSegSwitchLSN;
     LWLockRelease(WALWriteLock);
  
     return result;
 }

References XLogCtlData::lastSegSwitchLSN, XLogCtlData::lastSegSwitchTime, LW_SHARED, LWLockAcquire(), LWLockRelease(), and XLogCtl.

Referenced by CheckArchiveTimeout().

◆ GetMockAuthenticationNonce()

char* GetMockAuthenticationNonce ( void )

Definition at line 4543 of file xlog.c.

 {
     Assert(ControlFile != NULL);
     return ControlFile->mock_authentication_nonce;
 }

References Assert, ControlFile, and ControlFileData::mock_authentication_nonce.

Referenced by scram_mock_salt().

◆ GetOldestRestartPoint()

void GetOldestRestartPoint	(	XLogRecPtr *	oldrecptr,
		TimeLineID *	oldtli
	)

Definition at line 9393 of file xlog.c.

 {
     LWLockAcquire(ControlFileLock, LW_SHARED);
     *oldrecptr = ControlFile->checkPointCopy.redo;
     *oldtli = ControlFile->checkPointCopy.ThisTimeLineID;
     LWLockRelease(ControlFileLock);
 }

References ControlFileData::checkPointCopy, ControlFile, LW_SHARED, LWLockAcquire(), LWLockRelease(), CheckPoint::redo, and CheckPoint::ThisTimeLineID.

Referenced by ExecuteRecoveryCommand(), and RestoreArchivedFile().

◆ GetRecoveryState()

RecoveryState GetRecoveryState ( void )

Definition at line 6326 of file xlog.c.

 {
     RecoveryState retval;
  
     SpinLockAcquire(&XLogCtl->info_lck);
     retval = XLogCtl->SharedRecoveryState;
     SpinLockRelease(&XLogCtl->info_lck);
  
     return retval;
 }

References XLogCtlData::info_lck, XLogCtlData::SharedRecoveryState, SpinLockAcquire, SpinLockRelease, and XLogCtl.

Referenced by XLogArchiveCheckDone().

◆ GetRedoRecPtr()

XLogRecPtr GetRedoRecPtr ( void )

Definition at line 6393 of file xlog.c.

 {
     XLogRecPtr  ptr;
  
     /*
      * The possibly not up-to-date copy in XlogCtl is enough. Even if we
      * grabbed a WAL insertion lock to read the authoritative value in
      * Insert->RedoRecPtr, someone might update it just after we've released
      * the lock.
      */
     SpinLockAcquire(&XLogCtl->info_lck);
     ptr = XLogCtl->RedoRecPtr;
     SpinLockRelease(&XLogCtl->info_lck);
  
     if (RedoRecPtr < ptr)
         RedoRecPtr = ptr;
  
     return RedoRecPtr;
 }

References XLogCtlData::info_lck, RedoRecPtr, XLogCtlData::RedoRecPtr, SpinLockAcquire, SpinLockRelease, and XLogCtl.

Referenced by CheckPointLogicalRewriteHeap(), CheckPointSnapBuild(), MaybeRemoveOldWalSummaries(), nextval_internal(), ReplicationSlotReserveWal(), smgr_bulk_finish(), smgr_bulk_start_smgr(), XLogPageRead(), XLogSaveBufferForHint(), and XLogWrite().

◆ GetSystemIdentifier()

uint64 GetSystemIdentifier ( void )

Definition at line 4533 of file xlog.c.

 {
     Assert(ControlFile != NULL);
     return ControlFile->system_identifier;
 }

References Assert, ControlFile, and ControlFileData::system_identifier.

Referenced by IdentifySystem(), InitializeBackupManifest(), manifest_process_system_identifier(), ReplicationSlotNameForTablesync(), and WalReceiverMain().

◆ GetWALAvailability()

WALAvailability GetWALAvailability ( XLogRecPtr targetLSN )

Definition at line 7799 of file xlog.c.

 {
     XLogRecPtr  currpos;        /* current write LSN */
     XLogSegNo   currSeg;        /* segid of currpos */
     XLogSegNo   targetSeg;      /* segid of targetLSN */
     XLogSegNo   oldestSeg;      /* actual oldest segid */
     XLogSegNo   oldestSegMaxWalSize;    /* oldest segid kept by max_wal_size */
     XLogSegNo   oldestSlotSeg;  /* oldest segid kept by slot */
     uint64      keepSegs;
  
     /*
      * slot does not reserve WAL. Either deactivated, or has never been active
      */
     if (XLogRecPtrIsInvalid(targetLSN))
         return WALAVAIL_INVALID_LSN;
  
     /*
      * Calculate the oldest segment currently reserved by all slots,
      * considering wal_keep_size and max_slot_wal_keep_size.  Initialize
      * oldestSlotSeg to the current segment.
      */
     currpos = GetXLogWriteRecPtr();
     XLByteToSeg(currpos, oldestSlotSeg, wal_segment_size);
     KeepLogSeg(currpos, &oldestSlotSeg);
  
     /*
      * Find the oldest extant segment file. We get 1 until checkpoint removes
      * the first WAL segment file since startup, which causes the status being
      * wrong under certain abnormal conditions but that doesn't actually harm.
      */
     oldestSeg = XLogGetLastRemovedSegno() + 1;
  
     /* calculate oldest segment by max_wal_size */
     XLByteToSeg(currpos, currSeg, wal_segment_size);
     keepSegs = ConvertToXSegs(max_wal_size_mb, wal_segment_size) + 1;
  
     if (currSeg > keepSegs)
         oldestSegMaxWalSize = currSeg - keepSegs;
     else
         oldestSegMaxWalSize = 1;
  
     /* the segment we care about */
     XLByteToSeg(targetLSN, targetSeg, wal_segment_size);
  
     /*
      * No point in returning reserved or extended status values if the
      * targetSeg is known to be lost.
      */
     if (targetSeg >= oldestSlotSeg)
     {
         /* show "reserved" when targetSeg is within max_wal_size */
         if (targetSeg >= oldestSegMaxWalSize)
             return WALAVAIL_RESERVED;
  
         /* being retained by slots exceeding max_wal_size */
         return WALAVAIL_EXTENDED;
     }
  
     /* WAL segments are no longer retained but haven't been removed yet */
     if (targetSeg >= oldestSeg)
         return WALAVAIL_UNRESERVED;
  
     /* Definitely lost */
     return WALAVAIL_REMOVED;
 }

References ConvertToXSegs, GetXLogWriteRecPtr(), KeepLogSeg(), max_wal_size_mb, wal_segment_size, WALAVAIL_EXTENDED, WALAVAIL_INVALID_LSN, WALAVAIL_REMOVED, WALAVAIL_RESERVED, WALAVAIL_UNRESERVED, XLByteToSeg, XLogGetLastRemovedSegno(), and XLogRecPtrIsInvalid.

Referenced by pg_get_replication_slots().

◆ GetWALInsertionTimeLine()

TimeLineID GetWALInsertionTimeLine ( void )

Definition at line 6476 of file xlog.c.

 {
     Assert(XLogCtl->SharedRecoveryState == RECOVERY_STATE_DONE);
  
     /* Since the value can't be changing, no lock is required. */
     return XLogCtl->InsertTimeLineID;
 }

References Assert, XLogCtlData::InsertTimeLineID, RECOVERY_STATE_DONE, XLogCtlData::SharedRecoveryState, and XLogCtl.

Referenced by logical_read_xlog_page(), pg_walfile_name(), pg_walfile_name_offset(), ReadReplicationSlot(), WALReadFromBuffers(), and XLogSendPhysical().

◆ GetXLogBuffer()

static char * GetXLogBuffer	(	XLogRecPtr	ptr,
		TimeLineID	tli
	)

static

Definition at line 1625 of file xlog.c.

 {
     int         idx;
     XLogRecPtr  endptr;
     static uint64 cachedPage = 0;
     static char *cachedPos = NULL;
     XLogRecPtr  expectedEndPtr;
  
     /*
      * Fast path for the common case that we need to access again the same
      * page as last time.
      */
     if (ptr / XLOG_BLCKSZ == cachedPage)
     {
         Assert(((XLogPageHeader) cachedPos)->xlp_magic == XLOG_PAGE_MAGIC);
         Assert(((XLogPageHeader) cachedPos)->xlp_pageaddr == ptr - (ptr % XLOG_BLCKSZ));
         return cachedPos + ptr % XLOG_BLCKSZ;
     }
  
     /*
      * The XLog buffer cache is organized so that a page is always loaded to a
      * particular buffer.  That way we can easily calculate the buffer a given
      * page must be loaded into, from the XLogRecPtr alone.
      */
     idx = XLogRecPtrToBufIdx(ptr);
  
     /*
      * See what page is loaded in the buffer at the moment. It could be the
      * page we're looking for, or something older. It can't be anything newer
      * - that would imply the page we're looking for has already been written
      * out to disk and evicted, and the caller is responsible for making sure
      * that doesn't happen.
      *
      * We don't hold a lock while we read the value. If someone is just about
      * to initialize or has just initialized the page, it's possible that we
      * get InvalidXLogRecPtr. That's ok, we'll grab the mapping lock (in
      * AdvanceXLInsertBuffer) and retry if we see anything other than the page
      * we're looking for.
      */
     expectedEndPtr = ptr;
     expectedEndPtr += XLOG_BLCKSZ - ptr % XLOG_BLCKSZ;
  
     endptr = pg_atomic_read_u64(&XLogCtl->xlblocks[idx]);
     if (expectedEndPtr != endptr)
     {
         XLogRecPtr  initializedUpto;
  
         /*
          * Before calling AdvanceXLInsertBuffer(), which can block, let others
          * know how far we're finished with inserting the record.
          *
          * NB: If 'ptr' points to just after the page header, advertise a
          * position at the beginning of the page rather than 'ptr' itself. If
          * there are no other insertions running, someone might try to flush
          * up to our advertised location. If we advertised a position after
          * the page header, someone might try to flush the page header, even
          * though page might actually not be initialized yet. As the first
          * inserter on the page, we are effectively responsible for making
          * sure that it's initialized, before we let insertingAt to move past
          * the page header.
          */
         if (ptr % XLOG_BLCKSZ == SizeOfXLogShortPHD &&
             XLogSegmentOffset(ptr, wal_segment_size) > XLOG_BLCKSZ)
             initializedUpto = ptr - SizeOfXLogShortPHD;
         else if (ptr % XLOG_BLCKSZ == SizeOfXLogLongPHD &&
                  XLogSegmentOffset(ptr, wal_segment_size) < XLOG_BLCKSZ)
             initializedUpto = ptr - SizeOfXLogLongPHD;
         else
             initializedUpto = ptr;
  
         WALInsertLockUpdateInsertingAt(initializedUpto);
  
         AdvanceXLInsertBuffer(ptr, tli, false);
         endptr = pg_atomic_read_u64(&XLogCtl->xlblocks[idx]);
  
         if (expectedEndPtr != endptr)
             elog(PANIC, "could not find WAL buffer for %X/%X",
                  LSN_FORMAT_ARGS(ptr));
     }
     else
     {
         /*
          * Make sure the initialization of the page is visible to us, and
          * won't arrive later to overwrite the WAL data we write on the page.
          */
         pg_memory_barrier();
     }
  
     /*
      * Found the buffer holding this page. Return a pointer to the right
      * offset within the page.
      */
     cachedPage = ptr / XLOG_BLCKSZ;
     cachedPos = XLogCtl->pages + idx * (Size) XLOG_BLCKSZ;
  
     Assert(((XLogPageHeader) cachedPos)->xlp_magic == XLOG_PAGE_MAGIC);
     Assert(((XLogPageHeader) cachedPos)->xlp_pageaddr == ptr - (ptr % XLOG_BLCKSZ));
  
     return cachedPos + ptr % XLOG_BLCKSZ;
 }

References AdvanceXLInsertBuffer(), Assert, elog, idx(), LSN_FORMAT_ARGS, XLogCtlData::pages, PANIC, pg_atomic_read_u64(), pg_memory_barrier, SizeOfXLogLongPHD, SizeOfXLogShortPHD, wal_segment_size, WALInsertLockUpdateInsertingAt(), XLogCtlData::xlblocks, XLOG_PAGE_MAGIC, XLogCtl, XLogRecPtrToBufIdx, and XLogSegmentOffset.

Referenced by CopyXLogRecordToWAL(), and CreateOverwriteContrecordRecord().

◆ GetXLogInsertRecPtr()

XLogRecPtr GetXLogInsertRecPtr ( void )

Definition at line 9365 of file xlog.c.

 {
     XLogCtlInsert *Insert = &XLogCtl->Insert;
     uint64      current_bytepos;
  
     SpinLockAcquire(&Insert->insertpos_lck);
     current_bytepos = Insert->CurrBytePos;
     SpinLockRelease(&Insert->insertpos_lck);
  
     return XLogBytePosToRecPtr(current_bytepos);
 }

References XLogCtlData::Insert, Insert(), SpinLockAcquire, SpinLockRelease, XLogBytePosToRecPtr(), and XLogCtl.

Referenced by CreateOverwriteContrecordRecord(), gistGetFakeLSN(), logical_begin_heap_rewrite(), pg_current_wal_insert_lsn(), and ReplicationSlotReserveWal().

◆ GetXLogWriteRecPtr()

XLogRecPtr GetXLogWriteRecPtr ( void )

Definition at line 9381 of file xlog.c.

 {
     RefreshXLogWriteResult(LogwrtResult);
  
     return LogwrtResult.Write;
 }

References LogwrtResult, RefreshXLogWriteResult, and XLogwrtResult::Write.

Referenced by GetWALAvailability(), pg_attribute_noreturn(), pg_current_wal_lsn(), and pg_get_replication_slots().

◆ InitControlFile()

static void InitControlFile	(	uint64	sysidentifier,
		uint32	data_checksum_version
	)

static

Definition at line 4191 of file xlog.c.

 {
     char        mock_auth_nonce[MOCK_AUTH_NONCE_LEN];
  
     /*
      * Generate a random nonce. This is used for authentication requests that
      * will fail because the user does not exist. The nonce is used to create
      * a genuine-looking password challenge for the non-existent user, in lieu
      * of an actual stored password.
      */
     if (!pg_strong_random(mock_auth_nonce, MOCK_AUTH_NONCE_LEN))
         ereport(PANIC,
                 (errcode(ERRCODE_INTERNAL_ERROR),
                  errmsg("could not generate secret authorization token")));
  
     memset(ControlFile, 0, sizeof(ControlFileData));
     /* Initialize pg_control status fields */
     ControlFile->system_identifier = sysidentifier;
     memcpy(ControlFile->mock_authentication_nonce, mock_auth_nonce, MOCK_AUTH_NONCE_LEN);
     ControlFile->state = DB_SHUTDOWNED;
     ControlFile->unloggedLSN = FirstNormalUnloggedLSN;
  
     /* Set important parameter values for use when replaying WAL */
     ControlFile->MaxConnections = MaxConnections;
     ControlFile->max_worker_processes = max_worker_processes;
     ControlFile->max_wal_senders = max_wal_senders;
     ControlFile->max_prepared_xacts = max_prepared_xacts;
     ControlFile->max_locks_per_xact = max_locks_per_xact;
     ControlFile->wal_level = wal_level;
     ControlFile->wal_log_hints = wal_log_hints;
     ControlFile->track_commit_timestamp = track_commit_timestamp;
     ControlFile->data_checksum_version = data_checksum_version;
 }

References ControlFile, ControlFileData::data_checksum_version, DB_SHUTDOWNED, ereport, errcode(), errmsg(), FirstNormalUnloggedLSN, max_locks_per_xact, ControlFileData::max_locks_per_xact, max_prepared_xacts, ControlFileData::max_prepared_xacts, max_wal_senders, ControlFileData::max_wal_senders, max_worker_processes, ControlFileData::max_worker_processes, MaxConnections, ControlFileData::MaxConnections, MOCK_AUTH_NONCE_LEN, ControlFileData::mock_authentication_nonce, PANIC, pg_strong_random(), ControlFileData::state, ControlFileData::system_identifier, track_commit_timestamp, ControlFileData::track_commit_timestamp, ControlFileData::unloggedLSN, wal_level, ControlFileData::wal_level, wal_log_hints, and ControlFileData::wal_log_hints.

Referenced by BootStrapXLOG().

◆ InitializeWalConsistencyChecking()

void InitializeWalConsistencyChecking ( void )

Definition at line 4749 of file xlog.c.

 {
     Assert(process_shared_preload_libraries_done);
  
     if (check_wal_consistency_checking_deferred)
     {
         struct config_generic *guc;
  
         guc = find_option("wal_consistency_checking", false, false, ERROR);
  
         check_wal_consistency_checking_deferred = false;
  
         set_config_option_ext("wal_consistency_checking",
                               wal_consistency_checking_string,
                               guc->scontext, guc->source, guc->srole,
                               GUC_ACTION_SET, true, ERROR, false);
  
         /* checking should not be deferred again */
         Assert(!check_wal_consistency_checking_deferred);
     }
 }

References Assert, check_wal_consistency_checking_deferred, ERROR, find_option(), GUC_ACTION_SET, process_shared_preload_libraries_done, config_generic::scontext, set_config_option_ext(), config_generic::source, config_generic::srole, and wal_consistency_checking_string.

Referenced by PostgresSingleUserMain(), and PostmasterMain().

◆ InstallXLogFileSegment()

static bool InstallXLogFileSegment	(	XLogSegNo *	segno,
		char *	tmppath,
		bool	find_free,
		XLogSegNo	max_segno,
		TimeLineID	tli
	)

static

Definition at line 3550 of file xlog.c.

 {
     char        path[MAXPGPATH];
     struct stat stat_buf;
  
     Assert(tli != 0);
  
     XLogFilePath(path, tli, *segno, wal_segment_size);
  
     LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
     if (!XLogCtl->InstallXLogFileSegmentActive)
     {
         LWLockRelease(ControlFileLock);
         return false;
     }
  
     if (!find_free)
     {
         /* Force installation: get rid of any pre-existing segment file */
         durable_unlink(path, DEBUG1);
     }
     else
     {
         /* Find a free slot to put it in */
         while (stat(path, &stat_buf) == 0)
         {
             if ((*segno) >= max_segno)
             {
                 /* Failed to find a free slot within specified range */
                 LWLockRelease(ControlFileLock);
                 return false;
             }
             (*segno)++;
             XLogFilePath(path, tli, *segno, wal_segment_size);
         }
     }
  
     Assert(access(path, F_OK) != 0 && errno == ENOENT);
     if (durable_rename(tmppath, path, LOG) != 0)
     {
         LWLockRelease(ControlFileLock);
         /* durable_rename already emitted log message */
         return false;
     }
  
     LWLockRelease(ControlFileLock);
  
     return true;
 }

References Assert, DEBUG1, durable_rename(), durable_unlink(), XLogCtlData::InstallXLogFileSegmentActive, LOG, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), MAXPGPATH, stat, wal_segment_size, XLogCtl, and XLogFilePath().

Referenced by RemoveXlogFile(), XLogFileCopy(), and XLogFileInitInternal().

◆ IsInstallXLogFileSegmentActive()

bool IsInstallXLogFileSegmentActive ( void )

Definition at line 9422 of file xlog.c.

 {
     bool        result;
  
     LWLockAcquire(ControlFileLock, LW_SHARED);
     result = XLogCtl->InstallXLogFileSegmentActive;
     LWLockRelease(ControlFileLock);
  
     return result;
 }

References XLogCtlData::InstallXLogFileSegmentActive, LW_SHARED, LWLockAcquire(), LWLockRelease(), and XLogCtl.

Referenced by XLogFileRead().

◆ issue_xlog_fsync()

void issue_xlog_fsync	(	int	fd,
		XLogSegNo	segno,
		TimeLineID	tli
	)

Definition at line 8613 of file xlog.c.

 {
     char       *msg = NULL;
     instr_time  start;
  
     Assert(tli != 0);
  
     /*
      * Quick exit if fsync is disabled or write() has already synced the WAL
      * file.
      */
     if (!enableFsync ||
         wal_sync_method == WAL_SYNC_METHOD_OPEN ||
         wal_sync_method == WAL_SYNC_METHOD_OPEN_DSYNC)
         return;
  
     /* Measure I/O timing to sync the WAL file */
     if (track_wal_io_timing)
         INSTR_TIME_SET_CURRENT(start);
     else
         INSTR_TIME_SET_ZERO(start);
  
     pgstat_report_wait_start(WAIT_EVENT_WAL_SYNC);
     switch (wal_sync_method)
     {
         case WAL_SYNC_METHOD_FSYNC:
             if (pg_fsync_no_writethrough(fd) != 0)
                 msg = _("could not fsync file \"%s\": %m");
             break;
 #ifdef HAVE_FSYNC_WRITETHROUGH
         case WAL_SYNC_METHOD_FSYNC_WRITETHROUGH:
             if (pg_fsync_writethrough(fd) != 0)
                 msg = _("could not fsync write-through file \"%s\": %m");
             break;
 #endif
         case WAL_SYNC_METHOD_FDATASYNC:
             if (pg_fdatasync(fd) != 0)
                 msg = _("could not fdatasync file \"%s\": %m");
             break;
         case WAL_SYNC_METHOD_OPEN:
         case WAL_SYNC_METHOD_OPEN_DSYNC:
             /* not reachable */
             Assert(false);
             break;
         default:
             ereport(PANIC,
                     errcode(ERRCODE_INVALID_PARAMETER_VALUE),
                     errmsg_internal("unrecognized \"wal_sync_method\": %d", wal_sync_method));
             break;
     }
  
     /* PANIC if failed to fsync */
     if (msg)
     {
         char        xlogfname[MAXFNAMELEN];
         int         save_errno = errno;
  
         XLogFileName(xlogfname, tli, segno, wal_segment_size);
         errno = save_errno;
         ereport(PANIC,
                 (errcode_for_file_access(),
                  errmsg(msg, xlogfname)));
     }
  
     pgstat_report_wait_end();
  
     /*
      * Increment the I/O timing and the number of times WAL files were synced.
      */
     if (track_wal_io_timing)
     {
         instr_time  end;
  
         INSTR_TIME_SET_CURRENT(end);
         INSTR_TIME_ACCUM_DIFF(PendingWalStats.wal_sync_time, end, start);
     }
  
     PendingWalStats.wal_sync++;
 }

References _, Assert, enableFsync, ereport, errcode(), errcode_for_file_access(), errmsg(), errmsg_internal(), fd(), INSTR_TIME_ACCUM_DIFF, INSTR_TIME_SET_CURRENT, INSTR_TIME_SET_ZERO, MAXFNAMELEN, PANIC, PendingWalStats, pg_fdatasync(), pg_fsync_no_writethrough(), pg_fsync_writethrough(), pgstat_report_wait_end(), pgstat_report_wait_start(), start, track_wal_io_timing, wal_segment_size, PgStat_PendingWalStats::wal_sync, wal_sync_method, WAL_SYNC_METHOD_FDATASYNC, WAL_SYNC_METHOD_FSYNC, WAL_SYNC_METHOD_FSYNC_WRITETHROUGH, WAL_SYNC_METHOD_OPEN, WAL_SYNC_METHOD_OPEN_DSYNC, PgStat_PendingWalStats::wal_sync_time, and XLogFileName().

Referenced by XLogWalRcvFlush(), and XLogWrite().

◆ KeepLogSeg()

static void KeepLogSeg	(	XLogRecPtr	recptr,
		XLogSegNo *	logSegNo
	)

static

Definition at line 7883 of file xlog.c.

 {
     XLogSegNo   currSegNo;
     XLogSegNo   segno;
     XLogRecPtr  keep;
  
     XLByteToSeg(recptr, currSegNo, wal_segment_size);
     segno = currSegNo;
  
     /*
      * Calculate how many segments are kept by slots first, adjusting for
      * max_slot_wal_keep_size.
      */
     keep = XLogGetReplicationSlotMinimumLSN();
     if (keep != InvalidXLogRecPtr && keep < recptr)
     {
         XLByteToSeg(keep, segno, wal_segment_size);
  
         /* Cap by max_slot_wal_keep_size ... */
         if (max_slot_wal_keep_size_mb >= 0)
         {
             uint64      slot_keep_segs;
  
             slot_keep_segs =
                 ConvertToXSegs(max_slot_wal_keep_size_mb, wal_segment_size);
  
             if (currSegNo - segno > slot_keep_segs)
                 segno = currSegNo - slot_keep_segs;
         }
     }
  
     /*
      * If WAL summarization is in use, don't remove WAL that has yet to be
      * summarized.
      */
     keep = GetOldestUnsummarizedLSN(NULL, NULL);
     if (keep != InvalidXLogRecPtr)
     {
         XLogSegNo   unsummarized_segno;
  
         XLByteToSeg(keep, unsummarized_segno, wal_segment_size);
         if (unsummarized_segno < segno)
             segno = unsummarized_segno;
     }
  
     /* but, keep at least wal_keep_size if that's set */
     if (wal_keep_size_mb > 0)
     {
         uint64      keep_segs;
  
         keep_segs = ConvertToXSegs(wal_keep_size_mb, wal_segment_size);
         if (currSegNo - segno < keep_segs)
         {
             /* avoid underflow, don't go below 1 */
             if (currSegNo <= keep_segs)
                 segno = 1;
             else
                 segno = currSegNo - keep_segs;
         }
     }
  
     /* don't delete WAL segments newer than the calculated segment */
     if (segno < *logSegNo)
         *logSegNo = segno;
 }

References ConvertToXSegs, GetOldestUnsummarizedLSN(), InvalidXLogRecPtr, max_slot_wal_keep_size_mb, wal_keep_size_mb, wal_segment_size, XLByteToSeg, and XLogGetReplicationSlotMinimumLSN().

Referenced by CreateCheckPoint(), CreateRestartPoint(), and GetWALAvailability().

◆ LocalProcessControlFile()

void LocalProcessControlFile ( bool reset )

Definition at line 4811 of file xlog.c.

 {
     Assert(reset || ControlFile == NULL);
     ControlFile = palloc(sizeof(ControlFileData));
     ReadControlFile();
 }

References Assert, ControlFile, palloc(), ReadControlFile(), and reset().

Referenced by PostgresSingleUserMain(), PostmasterMain(), and PostmasterStateMachine().

◆ LocalSetXLogInsertAllowed()

static int LocalSetXLogInsertAllowed ( void )

static

Definition at line 6378 of file xlog.c.

 {
     int         oldXLogAllowed = LocalXLogInsertAllowed;
  
     LocalXLogInsertAllowed = 1;
  
     return oldXLogAllowed;
 }

References LocalXLogInsertAllowed.

Referenced by CreateCheckPoint(), and StartupXLOG().

◆ LogCheckpointEnd()

static void LogCheckpointEnd ( bool restartpoint )

static

Definition at line 6618 of file xlog.c.

 {
     long        write_msecs,
                 sync_msecs,
                 total_msecs,
                 longest_msecs,
                 average_msecs;
     uint64      average_sync_time;
  
     CheckpointStats.ckpt_end_t = GetCurrentTimestamp();
  
     write_msecs = TimestampDifferenceMilliseconds(CheckpointStats.ckpt_write_t,
                                                   CheckpointStats.ckpt_sync_t);
  
     sync_msecs = TimestampDifferenceMilliseconds(CheckpointStats.ckpt_sync_t,
                                                  CheckpointStats.ckpt_sync_end_t);
  
     /* Accumulate checkpoint timing summary data, in milliseconds. */
     PendingCheckpointerStats.write_time += write_msecs;
     PendingCheckpointerStats.sync_time += sync_msecs;
  
     /*
      * All of the published timing statistics are accounted for.  Only
      * continue if a log message is to be written.
      */
     if (!log_checkpoints)
         return;
  
     total_msecs = TimestampDifferenceMilliseconds(CheckpointStats.ckpt_start_t,
                                                   CheckpointStats.ckpt_end_t);
  
     /*
      * Timing values returned from CheckpointStats are in microseconds.
      * Convert to milliseconds for consistent printing.
      */
     longest_msecs = (long) ((CheckpointStats.ckpt_longest_sync + 999) / 1000);
  
     average_sync_time = 0;
     if (CheckpointStats.ckpt_sync_rels > 0)
         average_sync_time = CheckpointStats.ckpt_agg_sync_time /
             CheckpointStats.ckpt_sync_rels;
     average_msecs = (long) ((average_sync_time + 999) / 1000);
  
     /*
      * ControlFileLock is not required to see ControlFile->checkPoint and
      * ->checkPointCopy here as we are the only updator of those variables at
      * this moment.
      */
     if (restartpoint)
         ereport(LOG,
                 (errmsg("restartpoint complete: wrote %d buffers (%.1f%%); "
                         "%d WAL file(s) added, %d removed, %d recycled; "
                         "write=%ld.%03d s, sync=%ld.%03d s, total=%ld.%03d s; "
                         "sync files=%d, longest=%ld.%03d s, average=%ld.%03d s; "
                         "distance=%d kB, estimate=%d kB; "
                         "lsn=%X/%X, redo lsn=%X/%X",
                         CheckpointStats.ckpt_bufs_written,
                         (double) CheckpointStats.ckpt_bufs_written * 100 / NBuffers,
                         CheckpointStats.ckpt_segs_added,
                         CheckpointStats.ckpt_segs_removed,
                         CheckpointStats.ckpt_segs_recycled,
                         write_msecs / 1000, (int) (write_msecs % 1000),
                         sync_msecs / 1000, (int) (sync_msecs % 1000),
                         total_msecs / 1000, (int) (total_msecs % 1000),
                         CheckpointStats.ckpt_sync_rels,
                         longest_msecs / 1000, (int) (longest_msecs % 1000),
                         average_msecs / 1000, (int) (average_msecs % 1000),
                         (int) (PrevCheckPointDistance / 1024.0),
                         (int) (CheckPointDistanceEstimate / 1024.0),
                         LSN_FORMAT_ARGS(ControlFile->checkPoint),
                         LSN_FORMAT_ARGS(ControlFile->checkPointCopy.redo))));
     else
         ereport(LOG,
                 (errmsg("checkpoint complete: wrote %d buffers (%.1f%%); "
                         "%d WAL file(s) added, %d removed, %d recycled; "
                         "write=%ld.%03d s, sync=%ld.%03d s, total=%ld.%03d s; "
                         "sync files=%d, longest=%ld.%03d s, average=%ld.%03d s; "
                         "distance=%d kB, estimate=%d kB; "
                         "lsn=%X/%X, redo lsn=%X/%X",
                         CheckpointStats.ckpt_bufs_written,
                         (double) CheckpointStats.ckpt_bufs_written * 100 / NBuffers,
                         CheckpointStats.ckpt_segs_added,
                         CheckpointStats.ckpt_segs_removed,
                         CheckpointStats.ckpt_segs_recycled,
                         write_msecs / 1000, (int) (write_msecs % 1000),
                         sync_msecs / 1000, (int) (sync_msecs % 1000),
                         total_msecs / 1000, (int) (total_msecs % 1000),
                         CheckpointStats.ckpt_sync_rels,
                         longest_msecs / 1000, (int) (longest_msecs % 1000),
                         average_msecs / 1000, (int) (average_msecs % 1000),
                         (int) (PrevCheckPointDistance / 1024.0),
                         (int) (CheckPointDistanceEstimate / 1024.0),
                         LSN_FORMAT_ARGS(ControlFile->checkPoint),
                         LSN_FORMAT_ARGS(ControlFile->checkPointCopy.redo))));
 }

Referenced by CreateCheckPoint(), and CreateRestartPoint().

◆ LogCheckpointStart()

static void LogCheckpointStart	(	int	flags,
		bool	restartpoint
	)

static

Definition at line 6586 of file xlog.c.

 {
     if (restartpoint)
         ereport(LOG,
         /* translator: the placeholders show checkpoint options */
                 (errmsg("restartpoint starting:%s%s%s%s%s%s%s%s",
                         (flags & CHECKPOINT_IS_SHUTDOWN) ? " shutdown" : "",
                         (flags & CHECKPOINT_END_OF_RECOVERY) ? " end-of-recovery" : "",
                         (flags & CHECKPOINT_IMMEDIATE) ? " immediate" : "",
                         (flags & CHECKPOINT_FORCE) ? " force" : "",
                         (flags & CHECKPOINT_WAIT) ? " wait" : "",
                         (flags & CHECKPOINT_CAUSE_XLOG) ? " wal" : "",
                         (flags & CHECKPOINT_CAUSE_TIME) ? " time" : "",
                         (flags & CHECKPOINT_FLUSH_ALL) ? " flush-all" : "")));
     else
         ereport(LOG,
         /* translator: the placeholders show checkpoint options */
                 (errmsg("checkpoint starting:%s%s%s%s%s%s%s%s",
                         (flags & CHECKPOINT_IS_SHUTDOWN) ? " shutdown" : "",
                         (flags & CHECKPOINT_END_OF_RECOVERY) ? " end-of-recovery" : "",
                         (flags & CHECKPOINT_IMMEDIATE) ? " immediate" : "",
                         (flags & CHECKPOINT_FORCE) ? " force" : "",
                         (flags & CHECKPOINT_WAIT) ? " wait" : "",
                         (flags & CHECKPOINT_CAUSE_XLOG) ? " wal" : "",
                         (flags & CHECKPOINT_CAUSE_TIME) ? " time" : "",
                         (flags & CHECKPOINT_FLUSH_ALL) ? " flush-all" : "")));
 }

References CHECKPOINT_CAUSE_TIME, CHECKPOINT_CAUSE_XLOG, CHECKPOINT_END_OF_RECOVERY, CHECKPOINT_FLUSH_ALL, CHECKPOINT_FORCE, CHECKPOINT_IMMEDIATE, CHECKPOINT_IS_SHUTDOWN, CHECKPOINT_WAIT, ereport, errmsg(), and LOG.

Referenced by CreateCheckPoint(), and CreateRestartPoint().

◆ PerformRecoveryXLogAction()

static bool PerformRecoveryXLogAction ( void )

static

Definition at line 6240 of file xlog.c.

 {
     bool        promoted = false;
  
     /*
      * Perform a checkpoint to update all our recovery activity to disk.
      *
      * Note that we write a shutdown checkpoint rather than an on-line one.
      * This is not particularly critical, but since we may be assigning a new
      * TLI, using a shutdown checkpoint allows us to have the rule that TLI
      * only changes in shutdown checkpoints, which allows some extra error
      * checking in xlog_redo.
      *
      * In promotion, only create a lightweight end-of-recovery record instead
      * of a full checkpoint. A checkpoint is requested later, after we're
      * fully out of recovery mode and already accepting queries.
      */
     if (ArchiveRecoveryRequested && IsUnderPostmaster &&
         PromoteIsTriggered())
     {
         promoted = true;
  
         /*
          * Insert a special WAL record to mark the end of recovery, since we
          * aren't doing a checkpoint. That means that the checkpointer process
          * may likely be in the middle of a time-smoothed restartpoint and
          * could continue to be for minutes after this.  That sounds strange,
          * but the effect is roughly the same and it would be stranger to try
          * to come out of the restartpoint and then checkpoint. We request a
          * checkpoint later anyway, just for safety.
          */
         CreateEndOfRecoveryRecord();
     }
     else
     {
         RequestCheckpoint(CHECKPOINT_END_OF_RECOVERY |
                           CHECKPOINT_IMMEDIATE |
                           CHECKPOINT_WAIT);
     }
  
     return promoted;
 }

References ArchiveRecoveryRequested, CHECKPOINT_END_OF_RECOVERY, CHECKPOINT_IMMEDIATE, CHECKPOINT_WAIT, CreateEndOfRecoveryRecord(), IsUnderPostmaster, PromoteIsTriggered(), and RequestCheckpoint().

Referenced by StartupXLOG().

◆ PreallocXlogFiles()

static void PreallocXlogFiles	(	XLogRecPtr	endptr,
		TimeLineID	tli
	)

static

Definition at line 3677 of file xlog.c.

 {
     XLogSegNo   _logSegNo;
     int         lf;
     bool        added;
     char        path[MAXPGPATH];
     uint64      offset;
  
     if (!XLogCtl->InstallXLogFileSegmentActive)
         return;                 /* unlocked check says no */
  
     XLByteToPrevSeg(endptr, _logSegNo, wal_segment_size);
     offset = XLogSegmentOffset(endptr - 1, wal_segment_size);
     if (offset >= (uint32) (0.75 * wal_segment_size))
     {
         _logSegNo++;
         lf = XLogFileInitInternal(_logSegNo, tli, &added, path);
         if (lf >= 0)
             close(lf);
         if (added)
             CheckpointStats.ckpt_segs_added++;
     }
 }

References CheckpointStats, CheckpointStatsData::ckpt_segs_added, close, XLogCtlData::InstallXLogFileSegmentActive, MAXPGPATH, wal_segment_size, XLByteToPrevSeg, XLogCtl, XLogFileInitInternal(), and XLogSegmentOffset.

Referenced by CreateCheckPoint(), CreateRestartPoint(), and StartupXLOG().

◆ ReachedEndOfBackup()

void ReachedEndOfBackup	(	XLogRecPtr	EndRecPtr,
		TimeLineID	tli
	)

Definition at line 6203 of file xlog.c.

 {
     /*
      * We have reached the end of base backup, as indicated by pg_control. The
      * data on disk is now consistent (unless minRecoveryPoint is further
      * ahead, which can happen if we crashed during previous recovery).  Reset
      * backupStartPoint and backupEndPoint, and update minRecoveryPoint to
      * make sure we don't allow starting up at an earlier point even if
      * recovery is stopped and restarted soon after this.
      */
     LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
  
     if (ControlFile->minRecoveryPoint < EndRecPtr)
     {
         ControlFile->minRecoveryPoint = EndRecPtr;
         ControlFile->minRecoveryPointTLI = tli;
     }
  
     ControlFile->backupStartPoint = InvalidXLogRecPtr;
     ControlFile->backupEndPoint = InvalidXLogRecPtr;
     ControlFile->backupEndRequired = false;
     UpdateControlFile();
  
     LWLockRelease(ControlFileLock);
 }

References ControlFileData::backupEndPoint, ControlFileData::backupEndRequired, ControlFileData::backupStartPoint, ControlFile, InvalidXLogRecPtr, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), ControlFileData::minRecoveryPoint, ControlFileData::minRecoveryPointTLI, and UpdateControlFile().

Referenced by CheckRecoveryConsistency().

◆ ReadControlFile()

static void ReadControlFile ( void )

static

Definition at line 4308 of file xlog.c.

 {
     pg_crc32c   crc;
     int         fd;
     static char wal_segsz_str[20];
     int         r;
  
     /*
      * Read data...
      */
     fd = BasicOpenFile(XLOG_CONTROL_FILE,
                        O_RDWR | PG_BINARY);
     if (fd < 0)
         ereport(PANIC,
                 (errcode_for_file_access(),
                  errmsg("could not open file \"%s\": %m",
                         XLOG_CONTROL_FILE)));
  
     pgstat_report_wait_start(WAIT_EVENT_CONTROL_FILE_READ);
     r = read(fd, ControlFile, sizeof(ControlFileData));
     if (r != sizeof(ControlFileData))
     {
         if (r < 0)
             ereport(PANIC,
                     (errcode_for_file_access(),
                      errmsg("could not read file \"%s\": %m",
                             XLOG_CONTROL_FILE)));
         else
             ereport(PANIC,
                     (errcode(ERRCODE_DATA_CORRUPTED),
                      errmsg("could not read file \"%s\": read %d of %zu",
                             XLOG_CONTROL_FILE, r, sizeof(ControlFileData))));
     }
     pgstat_report_wait_end();
  
     close(fd);
  
     /*
      * Check for expected pg_control format version.  If this is wrong, the
      * CRC check will likely fail because we'll be checking the wrong number
      * of bytes.  Complaining about wrong version will probably be more
      * enlightening than complaining about wrong CRC.
      */
  
     if (ControlFile->pg_control_version != PG_CONTROL_VERSION && ControlFile->pg_control_version % 65536 == 0 && ControlFile->pg_control_version / 65536 != 0)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with PG_CONTROL_VERSION %d (0x%08x),"
                            " but the server was compiled with PG_CONTROL_VERSION %d (0x%08x).",
                            ControlFile->pg_control_version, ControlFile->pg_control_version,
                            PG_CONTROL_VERSION, PG_CONTROL_VERSION),
                  errhint("This could be a problem of mismatched byte ordering.  It looks like you need to initdb.")));
  
     if (ControlFile->pg_control_version != PG_CONTROL_VERSION)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with PG_CONTROL_VERSION %d,"
                            " but the server was compiled with PG_CONTROL_VERSION %d.",
                            ControlFile->pg_control_version, PG_CONTROL_VERSION),
                  errhint("It looks like you need to initdb.")));
  
     /* Now check the CRC. */
     INIT_CRC32C(crc);
     COMP_CRC32C(crc,
                 (char *) ControlFile,
                 offsetof(ControlFileData, crc));
     FIN_CRC32C(crc);
  
     if (!EQ_CRC32C(crc, ControlFile->crc))
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("incorrect checksum in control file")));
  
     /*
      * Do compatibility checking immediately.  If the database isn't
      * compatible with the backend executable, we want to abort before we can
      * possibly do any damage.
      */
     if (ControlFile->catalog_version_no != CATALOG_VERSION_NO)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with CATALOG_VERSION_NO %d,"
                            " but the server was compiled with CATALOG_VERSION_NO %d.",
                            ControlFile->catalog_version_no, CATALOG_VERSION_NO),
                  errhint("It looks like you need to initdb.")));
     if (ControlFile->maxAlign != MAXIMUM_ALIGNOF)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with MAXALIGN %d,"
                            " but the server was compiled with MAXALIGN %d.",
                            ControlFile->maxAlign, MAXIMUM_ALIGNOF),
                  errhint("It looks like you need to initdb.")));
     if (ControlFile->floatFormat != FLOATFORMAT_VALUE)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster appears to use a different floating-point number format than the server executable."),
                  errhint("It looks like you need to initdb.")));
     if (ControlFile->blcksz != BLCKSZ)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with BLCKSZ %d,"
                            " but the server was compiled with BLCKSZ %d.",
                            ControlFile->blcksz, BLCKSZ),
                  errhint("It looks like you need to recompile or initdb.")));
     if (ControlFile->relseg_size != RELSEG_SIZE)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with RELSEG_SIZE %d,"
                            " but the server was compiled with RELSEG_SIZE %d.",
                            ControlFile->relseg_size, RELSEG_SIZE),
                  errhint("It looks like you need to recompile or initdb.")));
     if (ControlFile->xlog_blcksz != XLOG_BLCKSZ)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with XLOG_BLCKSZ %d,"
                            " but the server was compiled with XLOG_BLCKSZ %d.",
                            ControlFile->xlog_blcksz, XLOG_BLCKSZ),
                  errhint("It looks like you need to recompile or initdb.")));
     if (ControlFile->nameDataLen != NAMEDATALEN)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with NAMEDATALEN %d,"
                            " but the server was compiled with NAMEDATALEN %d.",
                            ControlFile->nameDataLen, NAMEDATALEN),
                  errhint("It looks like you need to recompile or initdb.")));
     if (ControlFile->indexMaxKeys != INDEX_MAX_KEYS)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with INDEX_MAX_KEYS %d,"
                            " but the server was compiled with INDEX_MAX_KEYS %d.",
                            ControlFile->indexMaxKeys, INDEX_MAX_KEYS),
                  errhint("It looks like you need to recompile or initdb.")));
     if (ControlFile->toast_max_chunk_size != TOAST_MAX_CHUNK_SIZE)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with TOAST_MAX_CHUNK_SIZE %d,"
                            " but the server was compiled with TOAST_MAX_CHUNK_SIZE %d.",
                            ControlFile->toast_max_chunk_size, (int) TOAST_MAX_CHUNK_SIZE),
                  errhint("It looks like you need to recompile or initdb.")));
     if (ControlFile->loblksize != LOBLKSIZE)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with LOBLKSIZE %d,"
                            " but the server was compiled with LOBLKSIZE %d.",
                            ControlFile->loblksize, (int) LOBLKSIZE),
                  errhint("It looks like you need to recompile or initdb.")));
  
 #ifdef USE_FLOAT8_BYVAL
     if (ControlFile->float8ByVal != true)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized without USE_FLOAT8_BYVAL"
                            " but the server was compiled with USE_FLOAT8_BYVAL."),
                  errhint("It looks like you need to recompile or initdb.")));
 #else
     if (ControlFile->float8ByVal != false)
         ereport(FATAL,
                 (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                  errmsg("database files are incompatible with server"),
                  errdetail("The database cluster was initialized with USE_FLOAT8_BYVAL"
                            " but the server was compiled without USE_FLOAT8_BYVAL."),
                  errhint("It looks like you need to recompile or initdb.")));
 #endif
  
     wal_segment_size = ControlFile->xlog_seg_size;
  
     if (!IsValidWalSegSize(wal_segment_size))
         ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
                         errmsg_plural("invalid WAL segment size in control file (%d byte)",
                                       "invalid WAL segment size in control file (%d bytes)",
                                       wal_segment_size,
                                       wal_segment_size),
                         errdetail("The WAL segment size must be a power of two between 1 MB and 1 GB.")));
  
     snprintf(wal_segsz_str, sizeof(wal_segsz_str), "%d", wal_segment_size);
     SetConfigOption("wal_segment_size", wal_segsz_str, PGC_INTERNAL,
                     PGC_S_DYNAMIC_DEFAULT);
  
     /* check and update variables dependent on wal_segment_size */
     if (ConvertToXSegs(min_wal_size_mb, wal_segment_size) < 2)
         ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
                         errmsg("\"min_wal_size\" must be at least twice \"wal_segment_size\"")));
  
     if (ConvertToXSegs(max_wal_size_mb, wal_segment_size) < 2)
         ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
                         errmsg("\"max_wal_size\" must be at least twice \"wal_segment_size\"")));
  
     UsableBytesInSegment =
         (wal_segment_size / XLOG_BLCKSZ * UsableBytesInPage) -
         (SizeOfXLogLongPHD - SizeOfXLogShortPHD);
  
     CalculateCheckpointSegments();
  
     /* Make the initdb settings visible as GUC variables, too */
     SetConfigOption("data_checksums", DataChecksumsEnabled() ? "yes" : "no",
                     PGC_INTERNAL, PGC_S_DYNAMIC_DEFAULT);
 }

References BasicOpenFile(), ControlFileData::blcksz, CalculateCheckpointSegments(), CATALOG_VERSION_NO, ControlFileData::catalog_version_no, close, COMP_CRC32C, ControlFile, ConvertToXSegs, ControlFileData::crc, crc, DataChecksumsEnabled(), EQ_CRC32C, ereport, errcode(), ERRCODE_DATA_CORRUPTED, errcode_for_file_access(), errdetail(), errhint(), errmsg(), errmsg_plural(), ERROR, FATAL, fd(), FIN_CRC32C, ControlFileData::float8ByVal, ControlFileData::floatFormat, FLOATFORMAT_VALUE, INDEX_MAX_KEYS, ControlFileData::indexMaxKeys, INIT_CRC32C, IsValidWalSegSize, ControlFileData::loblksize, LOBLKSIZE, max_wal_size_mb, ControlFileData::maxAlign, min_wal_size_mb, ControlFileData::nameDataLen, NAMEDATALEN, PANIC, PG_BINARY, PG_CONTROL_VERSION, ControlFileData::pg_control_version, PGC_INTERNAL, PGC_S_DYNAMIC_DEFAULT, pgstat_report_wait_end(), pgstat_report_wait_start(), read, ControlFileData::relseg_size, SetConfigOption(), SizeOfXLogLongPHD, SizeOfXLogShortPHD, snprintf, TOAST_MAX_CHUNK_SIZE, ControlFileData::toast_max_chunk_size, UsableBytesInPage, UsableBytesInSegment, wal_segment_size, ControlFileData::xlog_blcksz, XLOG_CONTROL_FILE, and ControlFileData::xlog_seg_size.

Referenced by BootStrapXLOG(), and LocalProcessControlFile().

◆ RecoveryInProgress()

bool RecoveryInProgress ( void )

Definition at line 6290 of file xlog.c.

 {
     /*
      * We check shared state each time only until we leave recovery mode. We
      * can't re-enter recovery, so there's no need to keep checking after the
      * shared variable has once been seen false.
      */
     if (!LocalRecoveryInProgress)
         return false;
     else
     {
         /*
          * use volatile pointer to make sure we make a fresh read of the
          * shared variable.
          */
         volatile XLogCtlData *xlogctl = XLogCtl;
  
         LocalRecoveryInProgress = (xlogctl->SharedRecoveryState != RECOVERY_STATE_DONE);
  
         /*
          * Note: We don't need a memory barrier when we're still in recovery.
          * We might exit recovery immediately after return, so the caller
          * can't rely on 'true' meaning that we're still in recovery anyway.
          */
  
         return LocalRecoveryInProgress;
     }
 }

References LocalRecoveryInProgress, RECOVERY_STATE_DONE, XLogCtlData::SharedRecoveryState, and XLogCtl.

Referenced by BackgroundWriterMain(), BeginReportingGUCOptions(), brin_desummarize_range(), brin_summarize_range(), btree_index_mainfork_expected(), check_transaction_isolation(), check_transaction_read_only(), CheckArchiveTimeout(), CheckLogicalDecodingRequirements(), CheckpointerMain(), ComputeXidHorizons(), CreateCheckPoint(), CreateDecodingContext(), CreateEndOfRecoveryRecord(), CreateOverwriteContrecordRecord(), CreateRestartPoint(), do_pg_backup_start(), do_pg_backup_stop(), error_commit_ts_disabled(), get_relation_info(), GetCurrentLSN(), GetLatestLSN(), GetNewMultiXactId(), GetNewObjectId(), GetNewTransactionId(), GetOldestActiveTransactionId(), GetOldestSafeDecodingTransactionId(), GetRunningTransactionData(), GetSerializableTransactionSnapshot(), GetSerializableTransactionSnapshotInt(), GetSnapshotData(), GetStrictOldestNonRemovableTransactionId(), gin_clean_pending_list(), GlobalVisHorizonKindForRel(), heap_force_common(), heap_page_prune_opt(), IdentifySystem(), InitTempTableNamespace(), IsCheckpointOnSchedule(), LockAcquireExtended(), logical_read_xlog_page(), MaintainLatestCompletedXid(), MarkBufferDirtyHint(), perform_base_backup(), pg_create_restore_point(), pg_current_wal_flush_lsn(), pg_current_wal_insert_lsn(), pg_current_wal_lsn(), pg_get_wal_replay_pause_state(), pg_is_in_recovery(), pg_is_wal_replay_paused(), pg_log_standby_snapshot(), pg_logical_slot_get_changes_guts(), pg_promote(), pg_replication_slot_advance(), pg_sequence_last_value(), pg_switch_wal(), pg_sync_replication_slots(), pg_wal_replay_pause(), pg_wal_replay_resume(), pg_walfile_name(), pg_walfile_name_offset(), PhysicalWakeupLogicalWalSnd(), PrepareRedoAdd(), PrepareRedoRemove(), PreventCommandDuringRecovery(), ProcSleep(), read_local_xlog_page_guts(), ReadReplicationSlot(), recovery_create_dbdir(), ReplicationSlotAlter(), ReplicationSlotCreate(), ReplicationSlotDrop(), ReplicationSlotReserveWal(), replorigin_check_prerequisites(), ReportChangedGUCOptions(), sendDir(), SerialSetActiveSerXmin(), show_in_hot_standby(), ShutdownXLOG(), SnapBuildWaitSnapshot(), standard_ProcessUtility(), StandbySlotsHaveCaughtup(), StartLogicalReplication(), StartReplication(), StartTransaction(), TransactionIdIsInProgress(), TruncateMultiXact(), UpdateFullPageWrites(), verify_heapam(), WALReadFromBuffers(), WalReceiverMain(), WalSndWaitForWal(), xlog_decode(), XLogBackgroundFlush(), XLogFlush(), XLogInsertAllowed(), XLogNeedsFlush(), and XLogSendPhysical().

◆ RecoveryRestartPoint()

static void RecoveryRestartPoint	(	const CheckPoint *	checkPoint,
		XLogReaderState *	record
	)

static

Definition at line 7480 of file xlog.c.

 {
     /*
      * Also refrain from creating a restartpoint if we have seen any
      * references to non-existent pages. Restarting recovery from the
      * restartpoint would not see the references, so we would lose the
      * cross-check that the pages belonged to a relation that was dropped
      * later.
      */
     if (XLogHaveInvalidPages())
     {
         elog(DEBUG2,
              "could not record restart point at %X/%X because there "
              "are unresolved references to invalid pages",
              LSN_FORMAT_ARGS(checkPoint->redo));
         return;
     }
  
     /*
      * Copy the checkpoint record to shared memory, so that checkpointer can
      * work out the next time it wants to perform a restartpoint.
      */
     SpinLockAcquire(&XLogCtl->info_lck);
     XLogCtl->lastCheckPointRecPtr = record->ReadRecPtr;
     XLogCtl->lastCheckPointEndPtr = record->EndRecPtr;
     XLogCtl->lastCheckPoint = *checkPoint;
     SpinLockRelease(&XLogCtl->info_lck);
 }

References DEBUG2, elog, XLogReaderState::EndRecPtr, XLogCtlData::info_lck, XLogCtlData::lastCheckPoint, XLogCtlData::lastCheckPointEndPtr, XLogCtlData::lastCheckPointRecPtr, LSN_FORMAT_ARGS, XLogReaderState::ReadRecPtr, CheckPoint::redo, SpinLockAcquire, SpinLockRelease, XLogCtl, and XLogHaveInvalidPages().

Referenced by xlog_redo().

◆ register_persistent_abort_backup_handler()

void register_persistent_abort_backup_handler ( void )

Definition at line 9351 of file xlog.c.

 {
     static bool already_done = false;
  
     if (already_done)
         return;
     before_shmem_exit(do_pg_abort_backup, DatumGetBool(false));
     already_done = true;
 }

References before_shmem_exit(), DatumGetBool(), and do_pg_abort_backup().

Referenced by pg_backup_start().

◆ RemoveNonParentXlogFiles()

void RemoveNonParentXlogFiles	(	XLogRecPtr	switchpoint,
		TimeLineID	newTLI
	)

Definition at line 3927 of file xlog.c.

 {
     DIR        *xldir;
     struct dirent *xlde;
     char        switchseg[MAXFNAMELEN];
     XLogSegNo   endLogSegNo;
     XLogSegNo   switchLogSegNo;
     XLogSegNo   recycleSegNo;
  
     /*
      * Initialize info about where to begin the work.  This will recycle,
      * somewhat arbitrarily, 10 future segments.
      */
     XLByteToPrevSeg(switchpoint, switchLogSegNo, wal_segment_size);
     XLByteToSeg(switchpoint, endLogSegNo, wal_segment_size);
     recycleSegNo = endLogSegNo + 10;
  
     /*
      * Construct a filename of the last segment to be kept.
      */
     XLogFileName(switchseg, newTLI, switchLogSegNo, wal_segment_size);
  
     elog(DEBUG2, "attempting to remove WAL segments newer than log file %s",
          switchseg);
  
     xldir = AllocateDir(XLOGDIR);
  
     while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL)
     {
         /* Ignore files that are not XLOG segments */
         if (!IsXLogFileName(xlde->d_name))
             continue;
  
         /*
          * Remove files that are on a timeline older than the new one we're
          * switching to, but with a segment number >= the first segment on the
          * new timeline.
          */
         if (strncmp(xlde->d_name, switchseg, 8) < 0 &&
             strcmp(xlde->d_name + 8, switchseg + 8) > 0)
         {
             /*
              * If the file has already been marked as .ready, however, don't
              * remove it yet. It should be OK to remove it - files that are
              * not part of our timeline history are not required for recovery
              * - but seems safer to let them be archived and removed later.
              */
             if (!XLogArchiveIsReady(xlde->d_name))
                 RemoveXlogFile(xlde, recycleSegNo, &endLogSegNo, newTLI);
         }
     }
  
     FreeDir(xldir);
 }

References AllocateDir(), dirent::d_name, DEBUG2, elog, FreeDir(), IsXLogFileName(), MAXFNAMELEN, ReadDir(), RemoveXlogFile(), wal_segment_size, XLByteToPrevSeg, XLByteToSeg, XLogArchiveIsReady(), XLOGDIR, and XLogFileName().

Referenced by ApplyWalRecord(), and CleanupAfterArchiveRecovery().

◆ RemoveOldXlogFiles()

static void RemoveOldXlogFiles	(	XLogSegNo	segno,
		XLogRecPtr	lastredoptr,
		XLogRecPtr	endptr,
		TimeLineID	insertTLI
	)

static

Definition at line 3852 of file xlog.c.

 {
     DIR        *xldir;
     struct dirent *xlde;
     char        lastoff[MAXFNAMELEN];
     XLogSegNo   endlogSegNo;
     XLogSegNo   recycleSegNo;
  
     /* Initialize info about where to try to recycle to */
     XLByteToSeg(endptr, endlogSegNo, wal_segment_size);
     recycleSegNo = XLOGfileslop(lastredoptr);
  
     /*
      * Construct a filename of the last segment to be kept. The timeline ID
      * doesn't matter, we ignore that in the comparison. (During recovery,
      * InsertTimeLineID isn't set, so we can't use that.)
      */
     XLogFileName(lastoff, 0, segno, wal_segment_size);
  
     elog(DEBUG2, "attempting to remove WAL segments older than log file %s",
          lastoff);
  
     xldir = AllocateDir(XLOGDIR);
  
     while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL)
     {
         /* Ignore files that are not XLOG segments */
         if (!IsXLogFileName(xlde->d_name) &&
             !IsPartialXLogFileName(xlde->d_name))
             continue;
  
         /*
          * We ignore the timeline part of the XLOG segment identifiers in
          * deciding whether a segment is still needed.  This ensures that we
          * won't prematurely remove a segment from a parent timeline. We could
          * probably be a little more proactive about removing segments of
          * non-parent timelines, but that would be a whole lot more
          * complicated.
          *
          * We use the alphanumeric sorting property of the filenames to decide
          * which ones are earlier than the lastoff segment.
          */
         if (strcmp(xlde->d_name + 8, lastoff + 8) <= 0)
         {
             if (XLogArchiveCheckDone(xlde->d_name))
             {
                 /* Update the last removed location in shared memory first */
                 UpdateLastRemovedPtr(xlde->d_name);
  
                 RemoveXlogFile(xlde, recycleSegNo, &endlogSegNo, insertTLI);
             }
         }
     }
  
     FreeDir(xldir);
 }

References AllocateDir(), dirent::d_name, DEBUG2, elog, FreeDir(), IsPartialXLogFileName(), IsXLogFileName(), MAXFNAMELEN, ReadDir(), RemoveXlogFile(), UpdateLastRemovedPtr(), wal_segment_size, XLByteToSeg, XLogArchiveCheckDone(), XLOGDIR, XLogFileName(), and XLOGfileslop().

Referenced by CreateCheckPoint(), and CreateRestartPoint().

◆ RemoveTempXlogFiles()

static void RemoveTempXlogFiles ( void )

static

Definition at line 3819 of file xlog.c.

 {
     DIR        *xldir;
     struct dirent *xlde;
  
     elog(DEBUG2, "removing all temporary WAL segments");
  
     xldir = AllocateDir(XLOGDIR);
     while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL)
     {
         char        path[MAXPGPATH];
  
         if (strncmp(xlde->d_name, "xlogtemp.", 9) != 0)
             continue;
  
         snprintf(path, MAXPGPATH, XLOGDIR "/%s", xlde->d_name);
         unlink(path);
         elog(DEBUG2, "removed temporary WAL segment \"%s\"", path);
     }
     FreeDir(xldir);
 }

References AllocateDir(), dirent::d_name, DEBUG2, elog, FreeDir(), MAXPGPATH, ReadDir(), snprintf, and XLOGDIR.

Referenced by StartupXLOG().

◆ RemoveXlogFile()

static void RemoveXlogFile	(	const struct dirent *	segment_de,
		XLogSegNo	recycleSegNo,
		XLogSegNo *	endlogSegNo,
		TimeLineID	insertTLI
	)

static

Definition at line 3996 of file xlog.c.

 {
     char        path[MAXPGPATH];
 #ifdef WIN32
     char        newpath[MAXPGPATH];
 #endif
     const char *segname = segment_de->d_name;
  
     snprintf(path, MAXPGPATH, XLOGDIR "/%s", segname);
  
     /*
      * Before deleting the file, see if it can be recycled as a future log
      * segment. Only recycle normal files, because we don't want to recycle
      * symbolic links pointing to a separate archive directory.
      */
     if (wal_recycle &&
         *endlogSegNo <= recycleSegNo &&
         XLogCtl->InstallXLogFileSegmentActive &&    /* callee rechecks this */
         get_dirent_type(path, segment_de, false, DEBUG2) == PGFILETYPE_REG &&
         InstallXLogFileSegment(endlogSegNo, path,
                                true, recycleSegNo, insertTLI))
     {
         ereport(DEBUG2,
                 (errmsg_internal("recycled write-ahead log file \"%s\"",
                                  segname)));
         CheckpointStats.ckpt_segs_recycled++;
         /* Needn't recheck that slot on future iterations */
         (*endlogSegNo)++;
     }
     else
     {
         /* No need for any more future segments, or recycling failed ... */
         int         rc;
  
         ereport(DEBUG2,
                 (errmsg_internal("removing write-ahead log file \"%s\"",
                                  segname)));
  
 #ifdef WIN32
  
         /*
          * On Windows, if another process (e.g another backend) holds the file
          * open in FILE_SHARE_DELETE mode, unlink will succeed, but the file
          * will still show up in directory listing until the last handle is
          * closed. To avoid confusing the lingering deleted file for a live
          * WAL file that needs to be archived, rename it before deleting it.
          *
          * If another process holds the file open without FILE_SHARE_DELETE
          * flag, rename will fail. We'll try again at the next checkpoint.
          */
         snprintf(newpath, MAXPGPATH, "%s.deleted", path);
         if (rename(path, newpath) != 0)
         {
             ereport(LOG,
                     (errcode_for_file_access(),
                      errmsg("could not rename file \"%s\": %m",
                             path)));
             return;
         }
         rc = durable_unlink(newpath, LOG);
 #else
         rc = durable_unlink(path, LOG);
 #endif
         if (rc != 0)
         {
             /* Message already logged by durable_unlink() */
             return;
         }
         CheckpointStats.ckpt_segs_removed++;
     }
  
     XLogArchiveCleanup(segname);
 }

References CheckpointStats, CheckpointStatsData::ckpt_segs_recycled, CheckpointStatsData::ckpt_segs_removed, dirent::d_name, DEBUG2, durable_unlink(), ereport, errcode_for_file_access(), errmsg(), errmsg_internal(), get_dirent_type(), InstallXLogFileSegment(), XLogCtlData::InstallXLogFileSegmentActive, LOG, MAXPGPATH, PGFILETYPE_REG, snprintf, wal_recycle, XLogArchiveCleanup(), XLogCtl, and XLOGDIR.

Referenced by RemoveNonParentXlogFiles(), and RemoveOldXlogFiles().

◆ RequestXLogSwitch()

XLogRecPtr RequestXLogSwitch ( bool mark_unimportant )

Definition at line 7990 of file xlog.c.

 {
     XLogRecPtr  RecPtr;
  
     /* XLOG SWITCH has no data */
     XLogBeginInsert();
  
     if (mark_unimportant)
         XLogSetRecordFlags(XLOG_MARK_UNIMPORTANT);
     RecPtr = XLogInsert(RM_XLOG_ID, XLOG_SWITCH);
  
     return RecPtr;
 }

References XLOG_MARK_UNIMPORTANT, XLOG_SWITCH, XLogBeginInsert(), XLogInsert(), and XLogSetRecordFlags().

Referenced by CheckArchiveTimeout(), do_pg_backup_start(), do_pg_backup_stop(), pg_switch_wal(), and ShutdownXLOG().

◆ ReserveXLogInsertLocation()

static pg_attribute_always_inline void ReserveXLogInsertLocation	(	int	size,
		XLogRecPtr *	StartPos,
		XLogRecPtr *	EndPos,
		XLogRecPtr *	PrevPtr
	)

static

Definition at line 1101 of file xlog.c.

 {
     XLogCtlInsert *Insert = &XLogCtl->Insert;
     uint64      startbytepos;
     uint64      endbytepos;
     uint64      prevbytepos;
  
     size = MAXALIGN(size);
  
     /* All (non xlog-switch) records should contain data. */
     Assert(size > SizeOfXLogRecord);
  
     /*
      * The duration the spinlock needs to be held is minimized by minimizing
      * the calculations that have to be done while holding the lock. The
      * current tip of reserved WAL is kept in CurrBytePos, as a byte position
      * that only counts "usable" bytes in WAL, that is, it excludes all WAL
      * page headers. The mapping between "usable" byte positions and physical
      * positions (XLogRecPtrs) can be done outside the locked region, and
      * because the usable byte position doesn't include any headers, reserving
      * X bytes from WAL is almost as simple as "CurrBytePos += X".
      */
     SpinLockAcquire(&Insert->insertpos_lck);
  
     startbytepos = Insert->CurrBytePos;
     endbytepos = startbytepos + size;
     prevbytepos = Insert->PrevBytePos;
     Insert->CurrBytePos = endbytepos;
     Insert->PrevBytePos = startbytepos;
  
     SpinLockRelease(&Insert->insertpos_lck);
  
     *StartPos = XLogBytePosToRecPtr(startbytepos);
     *EndPos = XLogBytePosToEndRecPtr(endbytepos);
     *PrevPtr = XLogBytePosToRecPtr(prevbytepos);
  
     /*
      * Check that the conversions between "usable byte positions" and
      * XLogRecPtrs work consistently in both directions.
      */
     Assert(XLogRecPtrToBytePos(*StartPos) == startbytepos);
     Assert(XLogRecPtrToBytePos(*EndPos) == endbytepos);
     Assert(XLogRecPtrToBytePos(*PrevPtr) == prevbytepos);
 }

References Assert, XLogCtlData::Insert, Insert(), MAXALIGN, size, SizeOfXLogRecord, SpinLockAcquire, SpinLockRelease, XLogBytePosToEndRecPtr(), XLogBytePosToRecPtr(), XLogCtl, and XLogRecPtrToBytePos().

Referenced by XLogInsertRecord().

◆ ReserveXLogSwitch()

static bool ReserveXLogSwitch	(	XLogRecPtr *	StartPos,
		XLogRecPtr *	EndPos,
		XLogRecPtr *	PrevPtr
	)

static

Definition at line 1157 of file xlog.c.

 {
     XLogCtlInsert *Insert = &XLogCtl->Insert;
     uint64      startbytepos;
     uint64      endbytepos;
     uint64      prevbytepos;
     uint32      size = MAXALIGN(SizeOfXLogRecord);
     XLogRecPtr  ptr;
     uint32      segleft;
  
     /*
      * These calculations are a bit heavy-weight to be done while holding a
      * spinlock, but since we're holding all the WAL insertion locks, there
      * are no other inserters competing for it. GetXLogInsertRecPtr() does
      * compete for it, but that's not called very frequently.
      */
     SpinLockAcquire(&Insert->insertpos_lck);
  
     startbytepos = Insert->CurrBytePos;
  
     ptr = XLogBytePosToEndRecPtr(startbytepos);
     if (XLogSegmentOffset(ptr, wal_segment_size) == 0)
     {
         SpinLockRelease(&Insert->insertpos_lck);
         *EndPos = *StartPos = ptr;
         return false;
     }
  
     endbytepos = startbytepos + size;
     prevbytepos = Insert->PrevBytePos;
  
     *StartPos = XLogBytePosToRecPtr(startbytepos);
     *EndPos = XLogBytePosToEndRecPtr(endbytepos);
  
     segleft = wal_segment_size - XLogSegmentOffset(*EndPos, wal_segment_size);
     if (segleft != wal_segment_size)
     {
         /* consume the rest of the segment */
         *EndPos += segleft;
         endbytepos = XLogRecPtrToBytePos(*EndPos);
     }
     Insert->CurrBytePos = endbytepos;
     Insert->PrevBytePos = startbytepos;
  
     SpinLockRelease(&Insert->insertpos_lck);
  
     *PrevPtr = XLogBytePosToRecPtr(prevbytepos);
  
     Assert(XLogSegmentOffset(*EndPos, wal_segment_size) == 0);
     Assert(XLogRecPtrToBytePos(*EndPos) == endbytepos);
     Assert(XLogRecPtrToBytePos(*StartPos) == startbytepos);
     Assert(XLogRecPtrToBytePos(*PrevPtr) == prevbytepos);
  
     return true;
 }

References Assert, XLogCtlData::Insert, Insert(), MAXALIGN, size, SizeOfXLogRecord, SpinLockAcquire, SpinLockRelease, wal_segment_size, XLogBytePosToEndRecPtr(), XLogBytePosToRecPtr(), XLogCtl, XLogRecPtrToBytePos(), and XLogSegmentOffset.

Referenced by XLogInsertRecord().

◆ SetInstallXLogFileSegmentActive()

void SetInstallXLogFileSegmentActive ( void )

Definition at line 9414 of file xlog.c.

 {
     LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
     XLogCtl->InstallXLogFileSegmentActive = true;
     LWLockRelease(ControlFileLock);
 }

References XLogCtlData::InstallXLogFileSegmentActive, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), and XLogCtl.

Referenced by BootStrapXLOG(), StartupXLOG(), and WaitForWALToBecomeAvailable().

◆ SetWalWriterSleeping()

void SetWalWriterSleeping ( bool sleeping )

Definition at line 9437 of file xlog.c.

 {
     SpinLockAcquire(&XLogCtl->info_lck);
     XLogCtl->WalWriterSleeping = sleeping;
     SpinLockRelease(&XLogCtl->info_lck);
 }

References XLogCtlData::info_lck, SpinLockAcquire, SpinLockRelease, XLogCtlData::WalWriterSleeping, and XLogCtl.

Referenced by WalWriterMain().

◆ show_archive_command()

const char* show_archive_command ( void )

Definition at line 4775 of file xlog.c.

 {
     if (XLogArchivingActive())
         return XLogArchiveCommand;
     else
         return "(disabled)";
 }

References XLogArchiveCommand, and XLogArchivingActive.

◆ show_in_hot_standby()

const char* show_in_hot_standby ( void )

Definition at line 4787 of file xlog.c.

 {
     /*
      * We display the actual state based on shared memory, so that this GUC
      * reports up-to-date state if examined intra-query.  The underlying
      * variable (in_hot_standby_guc) changes only when we transmit a new value
      * to the client.
      */
     return RecoveryInProgress() ? "on" : "off";
 }

References RecoveryInProgress().

◆ ShutdownXLOG()

void ShutdownXLOG	(	int	code,
		Datum	arg
	)

Definition at line 6539 of file xlog.c.

 {
     /*
      * We should have an aux process resource owner to use, and we should not
      * be in a transaction that's installed some other resowner.
      */
     Assert(AuxProcessResourceOwner != NULL);
     Assert(CurrentResourceOwner == NULL ||
            CurrentResourceOwner == AuxProcessResourceOwner);
     CurrentResourceOwner = AuxProcessResourceOwner;
  
     /* Don't be chatty in standalone mode */
     ereport(IsPostmasterEnvironment ? LOG : NOTICE,
             (errmsg("shutting down")));
  
     /*
      * Signal walsenders to move to stopping state.
      */
     WalSndInitStopping();
  
     /*
      * Wait for WAL senders to be in stopping state.  This prevents commands
      * from writing new WAL.
      */
     WalSndWaitStopping();
  
     if (RecoveryInProgress())
         CreateRestartPoint(CHECKPOINT_IS_SHUTDOWN | CHECKPOINT_IMMEDIATE);
     else
     {
         /*
          * If archiving is enabled, rotate the last XLOG file so that all the
          * remaining records are archived (postmaster wakes up the archiver
          * process one more time at the end of shutdown). The checkpoint
          * record will go to the next XLOG file and won't be archived (yet).
          */
         if (XLogArchivingActive())
             RequestXLogSwitch(false);
  
         CreateCheckPoint(CHECKPOINT_IS_SHUTDOWN | CHECKPOINT_IMMEDIATE);
     }
 }

References Assert, AuxProcessResourceOwner, CHECKPOINT_IMMEDIATE, CHECKPOINT_IS_SHUTDOWN, CreateCheckPoint(), CreateRestartPoint(), CurrentResourceOwner, ereport, errmsg(), IsPostmasterEnvironment, LOG, NOTICE, RecoveryInProgress(), RequestXLogSwitch(), WalSndInitStopping(), WalSndWaitStopping(), and XLogArchivingActive.

Referenced by HandleCheckpointerInterrupts(), and InitPostgres().

◆ StartupXLOG()

void StartupXLOG ( void )

Definition at line 5387 of file xlog.c.

 {
     XLogCtlInsert *Insert;
     CheckPoint  checkPoint;
     bool        wasShutdown;
     bool        didCrash;
     bool        haveTblspcMap;
     bool        haveBackupLabel;
     XLogRecPtr  EndOfLog;
     TimeLineID  EndOfLogTLI;
     TimeLineID  newTLI;
     bool        performedWalRecovery;
     EndOfWalRecoveryInfo *endOfRecoveryInfo;
     XLogRecPtr  abortedRecPtr;
     XLogRecPtr  missingContrecPtr;
     TransactionId oldestActiveXID;
     bool        promoted = false;
  
     /*
      * We should have an aux process resource owner to use, and we should not
      * be in a transaction that's installed some other resowner.
      */
     Assert(AuxProcessResourceOwner != NULL);
     Assert(CurrentResourceOwner == NULL ||
            CurrentResourceOwner == AuxProcessResourceOwner);
     CurrentResourceOwner = AuxProcessResourceOwner;
  
     /*
      * Check that contents look valid.
      */
     if (!XRecOffIsValid(ControlFile->checkPoint))
         ereport(FATAL,
                 (errcode(ERRCODE_DATA_CORRUPTED),
                  errmsg("control file contains invalid checkpoint location")));
  
     switch (ControlFile->state)
     {
         case DB_SHUTDOWNED:
  
             /*
              * This is the expected case, so don't be chatty in standalone
              * mode
              */
             ereport(IsPostmasterEnvironment ? LOG : NOTICE,
                     (errmsg("database system was shut down at %s",
                             str_time(ControlFile->time))));
             break;
  
         case DB_SHUTDOWNED_IN_RECOVERY:
             ereport(LOG,
                     (errmsg("database system was shut down in recovery at %s",
                             str_time(ControlFile->time))));
             break;
  
         case DB_SHUTDOWNING:
             ereport(LOG,
                     (errmsg("database system shutdown was interrupted; last known up at %s",
                             str_time(ControlFile->time))));
             break;
  
         case DB_IN_CRASH_RECOVERY:
             ereport(LOG,
                     (errmsg("database system was interrupted while in recovery at %s",
                             str_time(ControlFile->time)),
                      errhint("This probably means that some data is corrupted and"
                              " you will have to use the last backup for recovery.")));
             break;
  
         case DB_IN_ARCHIVE_RECOVERY:
             ereport(LOG,
                     (errmsg("database system was interrupted while in recovery at log time %s",
                             str_time(ControlFile->checkPointCopy.time)),
                      errhint("If this has occurred more than once some data might be corrupted"
                              " and you might need to choose an earlier recovery target.")));
             break;
  
         case DB_IN_PRODUCTION:
             ereport(LOG,
                     (errmsg("database system was interrupted; last known up at %s",
                             str_time(ControlFile->time))));
             break;
  
         default:
             ereport(FATAL,
                     (errcode(ERRCODE_DATA_CORRUPTED),
                      errmsg("control file contains invalid database cluster state")));
     }
  
     /* This is just to allow attaching to startup process with a debugger */
 #ifdef XLOG_REPLAY_DELAY
     if (ControlFile->state != DB_SHUTDOWNED)
         pg_usleep(60000000L);
 #endif
  
     /*
      * Verify that pg_wal, pg_wal/archive_status, and pg_wal/summaries exist.
      * In cases where someone has performed a copy for PITR, these directories
      * may have been excluded and need to be re-created.
      */
     ValidateXLOGDirectoryStructure();
  
     /* Set up timeout handler needed to report startup progress. */
     if (!IsBootstrapProcessingMode())
         RegisterTimeout(STARTUP_PROGRESS_TIMEOUT,
                         startup_progress_timeout_handler);
  
     /*----------
      * If we previously crashed, perform a couple of actions:
      *
      * - The pg_wal directory may still include some temporary WAL segments
      *   used when creating a new segment, so perform some clean up to not
      *   bloat this path.  This is done first as there is no point to sync
      *   this temporary data.
      *
      * - There might be data which we had written, intending to fsync it, but
      *   which we had not actually fsync'd yet.  Therefore, a power failure in
      *   the near future might cause earlier unflushed writes to be lost, even
      *   though more recent data written to disk from here on would be
      *   persisted.  To avoid that, fsync the entire data directory.
      */
     if (ControlFile->state != DB_SHUTDOWNED &&
         ControlFile->state != DB_SHUTDOWNED_IN_RECOVERY)
     {
         RemoveTempXlogFiles();
         SyncDataDirectory();
         didCrash = true;
     }
     else
         didCrash = false;
  
     /*
      * Prepare for WAL recovery if needed.
      *
      * InitWalRecovery analyzes the control file and the backup label file, if
      * any.  It updates the in-memory ControlFile buffer according to the
      * starting checkpoint, and sets InRecovery and ArchiveRecoveryRequested.
      * It also applies the tablespace map file, if any.
      */
     InitWalRecovery(ControlFile, &wasShutdown,
                     &haveBackupLabel, &haveTblspcMap);
     checkPoint = ControlFile->checkPointCopy;
  
     /* initialize shared memory variables from the checkpoint record */
     TransamVariables->nextXid = checkPoint.nextXid;
     TransamVariables->nextOid = checkPoint.nextOid;
     TransamVariables->oidCount = 0;
     MultiXactSetNextMXact(checkPoint.nextMulti, checkPoint.nextMultiOffset);
     AdvanceOldestClogXid(checkPoint.oldestXid);
     SetTransactionIdLimit(checkPoint.oldestXid, checkPoint.oldestXidDB);
     SetMultiXactIdLimit(checkPoint.oldestMulti, checkPoint.oldestMultiDB, true);
     SetCommitTsLimit(checkPoint.oldestCommitTsXid,
                      checkPoint.newestCommitTsXid);
     XLogCtl->ckptFullXid = checkPoint.nextXid;
  
     /*
      * Clear out any old relcache cache files.  This is *necessary* if we do
      * any WAL replay, since that would probably result in the cache files
      * being out of sync with database reality.  In theory we could leave them
      * in place if the database had been cleanly shut down, but it seems
      * safest to just remove them always and let them be rebuilt during the
      * first backend startup.  These files needs to be removed from all
      * directories including pg_tblspc, however the symlinks are created only
      * after reading tablespace_map file in case of archive recovery from
      * backup, so needs to clear old relcache files here after creating
      * symlinks.
      */
     RelationCacheInitFileRemove();
  
     /*
      * Initialize replication slots, before there's a chance to remove
      * required resources.
      */
     StartupReplicationSlots();
  
     /*
      * Startup logical state, needs to be setup now so we have proper data
      * during crash recovery.
      */
     StartupReorderBuffer();
  
     /*
      * Startup CLOG. This must be done after TransamVariables->nextXid has
      * been initialized and before we accept connections or begin WAL replay.
      */
     StartupCLOG();
  
     /*
      * Startup MultiXact. We need to do this early to be able to replay
      * truncations.
      */
     StartupMultiXact();
  
     /*
      * Ditto for commit timestamps.  Activate the facility if the setting is
      * enabled in the control file, as there should be no tracking of commit
      * timestamps done when the setting was disabled.  This facility can be
      * started or stopped when replaying a XLOG_PARAMETER_CHANGE record.
      */
     if (ControlFile->track_commit_timestamp)
         StartupCommitTs();
  
     /*
      * Recover knowledge about replay progress of known replication partners.
      */
     StartupReplicationOrigin();
  
     /*
      * Initialize unlogged LSN. On a clean shutdown, it's restored from the
      * control file. On recovery, all unlogged relations are blown away, so
      * the unlogged LSN counter can be reset too.
      */
     if (ControlFile->state == DB_SHUTDOWNED)
         pg_atomic_write_membarrier_u64(&XLogCtl->unloggedLSN,
                                        ControlFile->unloggedLSN);
     else
         pg_atomic_write_membarrier_u64(&XLogCtl->unloggedLSN,
                                        FirstNormalUnloggedLSN);
  
     /*
      * Copy any missing timeline history files between 'now' and the recovery
      * target timeline from archive to pg_wal. While we don't need those files
      * ourselves - the history file of the recovery target timeline covers all
      * the previous timelines in the history too - a cascading standby server
      * might be interested in them. Or, if you archive the WAL from this
      * server to a different archive than the primary, it'd be good for all
      * the history files to get archived there after failover, so that you can
      * use one of the old timelines as a PITR target. Timeline history files
      * are small, so it's better to copy them unnecessarily than not copy them
      * and regret later.
      */
     restoreTimeLineHistoryFiles(checkPoint.ThisTimeLineID, recoveryTargetTLI);
  
     /*
      * Before running in recovery, scan pg_twophase and fill in its status to
      * be able to work on entries generated by redo.  Doing a scan before
      * taking any recovery action has the merit to discard any 2PC files that
      * are newer than the first record to replay, saving from any conflicts at
      * replay.  This avoids as well any subsequent scans when doing recovery
      * of the on-disk two-phase data.
      */
     restoreTwoPhaseData();
  
     /*
      * When starting with crash recovery, reset pgstat data - it might not be
      * valid. Otherwise restore pgstat data. It's safe to do this here,
      * because postmaster will not yet have started any other processes.
      *
      * NB: Restoring replication slot stats relies on slot state to have
      * already been restored from disk.
      *
      * TODO: With a bit of extra work we could just start with a pgstat file
      * associated with the checkpoint redo location we're starting from.
      */
     if (didCrash)
         pgstat_discard_stats();
     else
         pgstat_restore_stats();
  
     lastFullPageWrites = checkPoint.fullPageWrites;
  
     RedoRecPtr = XLogCtl->RedoRecPtr = XLogCtl->Insert.RedoRecPtr = checkPoint.redo;
     doPageWrites = lastFullPageWrites;
  
     /* REDO */
     if (InRecovery)
     {
         /* Initialize state for RecoveryInProgress() */
         SpinLockAcquire(&XLogCtl->info_lck);
         if (InArchiveRecovery)
             XLogCtl->SharedRecoveryState = RECOVERY_STATE_ARCHIVE;
         else
             XLogCtl->SharedRecoveryState = RECOVERY_STATE_CRASH;
         SpinLockRelease(&XLogCtl->info_lck);
  
         /*
          * Update pg_control to show that we are recovering and to show the
          * selected checkpoint as the place we are starting from. We also mark
          * pg_control with any minimum recovery stop point obtained from a
          * backup history file.
          *
          * No need to hold ControlFileLock yet, we aren't up far enough.
          */
         UpdateControlFile();
  
         /*
          * If there was a backup label file, it's done its job and the info
          * has now been propagated into pg_control.  We must get rid of the
          * label file so that if we crash during recovery, we'll pick up at
          * the latest recovery restartpoint instead of going all the way back
          * to the backup start point.  It seems prudent though to just rename
          * the file out of the way rather than delete it completely.
          */
         if (haveBackupLabel)
         {
             unlink(BACKUP_LABEL_OLD);
             durable_rename(BACKUP_LABEL_FILE, BACKUP_LABEL_OLD, FATAL);
         }
  
         /*
          * If there was a tablespace_map file, it's done its job and the
          * symlinks have been created.  We must get rid of the map file so
          * that if we crash during recovery, we don't create symlinks again.
          * It seems prudent though to just rename the file out of the way
          * rather than delete it completely.
          */
         if (haveTblspcMap)
         {
             unlink(TABLESPACE_MAP_OLD);
             durable_rename(TABLESPACE_MAP, TABLESPACE_MAP_OLD, FATAL);
         }
  
         /*
          * Initialize our local copy of minRecoveryPoint.  When doing crash
          * recovery we want to replay up to the end of WAL.  Particularly, in
          * the case of a promoted standby minRecoveryPoint value in the
          * control file is only updated after the first checkpoint.  However,
          * if the instance crashes before the first post-recovery checkpoint
          * is completed then recovery will use a stale location causing the
          * startup process to think that there are still invalid page
          * references when checking for data consistency.
          */
         if (InArchiveRecovery)
         {
             LocalMinRecoveryPoint = ControlFile->minRecoveryPoint;
             LocalMinRecoveryPointTLI = ControlFile->minRecoveryPointTLI;
         }
         else
         {
             LocalMinRecoveryPoint = InvalidXLogRecPtr;
             LocalMinRecoveryPointTLI = 0;
         }
  
         /* Check that the GUCs used to generate the WAL allow recovery */
         CheckRequiredParameterValues();
  
         /*
          * We're in recovery, so unlogged relations may be trashed and must be
          * reset.  This should be done BEFORE allowing Hot Standby
          * connections, so that read-only backends don't try to read whatever
          * garbage is left over from before.
          */
         ResetUnloggedRelations(UNLOGGED_RELATION_CLEANUP);
  
         /*
          * Likewise, delete any saved transaction snapshot files that got left
          * behind by crashed backends.
          */
         DeleteAllExportedSnapshotFiles();
  
         /*
          * Initialize for Hot Standby, if enabled. We won't let backends in
          * yet, not until we've reached the min recovery point specified in
          * control file and we've established a recovery snapshot from a
          * running-xacts WAL record.
          */
         if (ArchiveRecoveryRequested && EnableHotStandby)
         {
             TransactionId *xids;
             int         nxids;
  
             ereport(DEBUG1,
                     (errmsg_internal("initializing for hot standby")));
  
             InitRecoveryTransactionEnvironment();
  
             if (wasShutdown)
                 oldestActiveXID = PrescanPreparedTransactions(&xids, &nxids);
             else
                 oldestActiveXID = checkPoint.oldestActiveXid;
             Assert(TransactionIdIsValid(oldestActiveXID));
  
             /* Tell procarray about the range of xids it has to deal with */
             ProcArrayInitRecovery(XidFromFullTransactionId(TransamVariables->nextXid));
  
             /*
              * Startup subtrans only.  CLOG, MultiXact and commit timestamp
              * have already been started up and other SLRUs are not maintained
              * during recovery and need not be started yet.
              */
             StartupSUBTRANS(oldestActiveXID);
  
             /*
              * If we're beginning at a shutdown checkpoint, we know that
              * nothing was running on the primary at this point. So fake-up an
              * empty running-xacts record and use that here and now. Recover
              * additional standby state for prepared transactions.
              */
             if (wasShutdown)
             {
                 RunningTransactionsData running;
                 TransactionId latestCompletedXid;
  
                 /* Update pg_subtrans entries for any prepared transactions */
                 StandbyRecoverPreparedTransactions();
  
                 /*
                  * Construct a RunningTransactions snapshot representing a
                  * shut down server, with only prepared transactions still
                  * alive. We're never overflowed at this point because all
                  * subxids are listed with their parent prepared transactions.
                  */
                 running.xcnt = nxids;
                 running.subxcnt = 0;
                 running.subxid_status = SUBXIDS_IN_SUBTRANS;
                 running.nextXid = XidFromFullTransactionId(checkPoint.nextXid);
                 running.oldestRunningXid = oldestActiveXID;
                 latestCompletedXid = XidFromFullTransactionId(checkPoint.nextXid);
                 TransactionIdRetreat(latestCompletedXid);
                 Assert(TransactionIdIsNormal(latestCompletedXid));
                 running.latestCompletedXid = latestCompletedXid;
                 running.xids = xids;
  
                 ProcArrayApplyRecoveryInfo(&running);
             }
         }
  
         /*
          * We're all set for replaying the WAL now. Do it.
          */
         PerformWalRecovery();
         performedWalRecovery = true;
     }
     else
         performedWalRecovery = false;
  
     /*
      * Finish WAL recovery.
      */
     endOfRecoveryInfo = FinishWalRecovery();
     EndOfLog = endOfRecoveryInfo->endOfLog;
     EndOfLogTLI = endOfRecoveryInfo->endOfLogTLI;
     abortedRecPtr = endOfRecoveryInfo->abortedRecPtr;
     missingContrecPtr = endOfRecoveryInfo->missingContrecPtr;
  
     /*
      * Reset ps status display, so as no information related to recovery shows
      * up.
      */
     set_ps_display("");
  
     /*
      * When recovering from a backup (we are in recovery, and archive recovery
      * was requested), complain if we did not roll forward far enough to reach
      * the point where the database is consistent.  For regular online
      * backup-from-primary, that means reaching the end-of-backup WAL record
      * (at which point we reset backupStartPoint to be Invalid), for
      * backup-from-replica (which can't inject records into the WAL stream),
      * that point is when we reach the minRecoveryPoint in pg_control (which
      * we purposefully copy last when backing up from a replica).  For
      * pg_rewind (which creates a backup_label with a method of "pg_rewind")
      * or snapshot-style backups (which don't), backupEndRequired will be set
      * to false.
      *
      * Note: it is indeed okay to look at the local variable
      * LocalMinRecoveryPoint here, even though ControlFile->minRecoveryPoint
      * might be further ahead --- ControlFile->minRecoveryPoint cannot have
      * been advanced beyond the WAL we processed.
      */
     if (InRecovery &&
         (EndOfLog < LocalMinRecoveryPoint ||
          !XLogRecPtrIsInvalid(ControlFile->backupStartPoint)))
     {
         /*
          * Ran off end of WAL before reaching end-of-backup WAL record, or
          * minRecoveryPoint. That's a bad sign, indicating that you tried to
          * recover from an online backup but never called pg_backup_stop(), or
          * you didn't archive all the WAL needed.
          */
         if (ArchiveRecoveryRequested || ControlFile->backupEndRequired)
         {
             if (!XLogRecPtrIsInvalid(ControlFile->backupStartPoint) || ControlFile->backupEndRequired)
                 ereport(FATAL,
                         (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                          errmsg("WAL ends before end of online backup"),
                          errhint("All WAL generated while online backup was taken must be available at recovery.")));
             else
                 ereport(FATAL,
                         (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                          errmsg("WAL ends before consistent recovery point")));
         }
     }
  
     /*
      * Reset unlogged relations to the contents of their INIT fork. This is
      * done AFTER recovery is complete so as to include any unlogged relations
      * created during recovery, but BEFORE recovery is marked as having
      * completed successfully. Otherwise we'd not retry if any of the post
      * end-of-recovery steps fail.
      */
     if (InRecovery)
         ResetUnloggedRelations(UNLOGGED_RELATION_INIT);
  
     /*
      * Pre-scan prepared transactions to find out the range of XIDs present.
      * This information is not quite needed yet, but it is positioned here so
      * as potential problems are detected before any on-disk change is done.
      */
     oldestActiveXID = PrescanPreparedTransactions(NULL, NULL);
  
     /*
      * Allow ordinary WAL segment creation before possibly switching to a new
      * timeline, which creates a new segment, and after the last ReadRecord().
      */
     SetInstallXLogFileSegmentActive();
  
     /*
      * Consider whether we need to assign a new timeline ID.
      *
      * If we did archive recovery, we always assign a new ID.  This handles a
      * couple of issues.  If we stopped short of the end of WAL during
      * recovery, then we are clearly generating a new timeline and must assign
      * it a unique new ID.  Even if we ran to the end, modifying the current
      * last segment is problematic because it may result in trying to
      * overwrite an already-archived copy of that segment, and we encourage
      * DBAs to make their archive_commands reject that.  We can dodge the
      * problem by making the new active segment have a new timeline ID.
      *
      * In a normal crash recovery, we can just extend the timeline we were in.
      */
     newTLI = endOfRecoveryInfo->lastRecTLI;
     if (ArchiveRecoveryRequested)
     {
         newTLI = findNewestTimeLine(recoveryTargetTLI) + 1;
         ereport(LOG,
                 (errmsg("selected new timeline ID: %u", newTLI)));
  
         /*
          * Make a writable copy of the last WAL segment.  (Note that we also
          * have a copy of the last block of the old WAL in
          * endOfRecovery->lastPage; we will use that below.)
          */
         XLogInitNewTimeline(EndOfLogTLI, EndOfLog, newTLI);
  
         /*
          * Remove the signal files out of the way, so that we don't
          * accidentally re-enter archive recovery mode in a subsequent crash.
          */
         if (endOfRecoveryInfo->standby_signal_file_found)
             durable_unlink(STANDBY_SIGNAL_FILE, FATAL);
  
         if (endOfRecoveryInfo->recovery_signal_file_found)
             durable_unlink(RECOVERY_SIGNAL_FILE, FATAL);
  
         /*
          * Write the timeline history file, and have it archived. After this
          * point (or rather, as soon as the file is archived), the timeline
          * will appear as "taken" in the WAL archive and to any standby
          * servers.  If we crash before actually switching to the new
          * timeline, standby servers will nevertheless think that we switched
          * to the new timeline, and will try to connect to the new timeline.
          * To minimize the window for that, try to do as little as possible
          * between here and writing the end-of-recovery record.
          */
         writeTimeLineHistory(newTLI, recoveryTargetTLI,
                              EndOfLog, endOfRecoveryInfo->recoveryStopReason);
  
         ereport(LOG,
                 (errmsg("archive recovery complete")));
     }
  
     /* Save the selected TimeLineID in shared memory, too */
     XLogCtl->InsertTimeLineID = newTLI;
     XLogCtl->PrevTimeLineID = endOfRecoveryInfo->lastRecTLI;
  
     /*
      * Actually, if WAL ended in an incomplete record, skip the parts that
      * made it through and start writing after the portion that persisted.
      * (It's critical to first write an OVERWRITE_CONTRECORD message, which
      * we'll do as soon as we're open for writing new WAL.)
      */
     if (!XLogRecPtrIsInvalid(missingContrecPtr))
     {
         /*
          * We should only have a missingContrecPtr if we're not switching to a
          * new timeline. When a timeline switch occurs, WAL is copied from the
          * old timeline to the new only up to the end of the last complete
          * record, so there can't be an incomplete WAL record that we need to
          * disregard.
          */
         Assert(newTLI == endOfRecoveryInfo->lastRecTLI);
         Assert(!XLogRecPtrIsInvalid(abortedRecPtr));
         EndOfLog = missingContrecPtr;
     }
  
     /*
      * Prepare to write WAL starting at EndOfLog location, and init xlog
      * buffer cache using the block containing the last record from the
      * previous incarnation.
      */
     Insert = &XLogCtl->Insert;
     Insert->PrevBytePos = XLogRecPtrToBytePos(endOfRecoveryInfo->lastRec);
     Insert->CurrBytePos = XLogRecPtrToBytePos(EndOfLog);
  
     /*
      * Tricky point here: lastPage contains the *last* block that the LastRec
      * record spans, not the one it starts in.  The last block is indeed the
      * one we want to use.
      */
     if (EndOfLog % XLOG_BLCKSZ != 0)
     {
         char       *page;
         int         len;
         int         firstIdx;
  
         firstIdx = XLogRecPtrToBufIdx(EndOfLog);
         len = EndOfLog - endOfRecoveryInfo->lastPageBeginPtr;
         Assert(len < XLOG_BLCKSZ);
  
         /* Copy the valid part of the last block, and zero the rest */
         page = &XLogCtl->pages[firstIdx * XLOG_BLCKSZ];
         memcpy(page, endOfRecoveryInfo->lastPage, len);
         memset(page + len, 0, XLOG_BLCKSZ - len);
  
         pg_atomic_write_u64(&XLogCtl->xlblocks[firstIdx], endOfRecoveryInfo->lastPageBeginPtr + XLOG_BLCKSZ);
         XLogCtl->InitializedUpTo = endOfRecoveryInfo->lastPageBeginPtr + XLOG_BLCKSZ;
     }
     else
     {
         /*
          * There is no partial block to copy. Just set InitializedUpTo, and
          * let the first attempt to insert a log record to initialize the next
          * buffer.
          */
         XLogCtl->InitializedUpTo = EndOfLog;
     }
  
     /*
      * Update local and shared status.  This is OK to do without any locks
      * because no other process can be reading or writing WAL yet.
      */
     LogwrtResult.Write = LogwrtResult.Flush = EndOfLog;
     pg_atomic_write_u64(&XLogCtl->logInsertResult, EndOfLog);
     pg_atomic_write_u64(&XLogCtl->logWriteResult, EndOfLog);
     pg_atomic_write_u64(&XLogCtl->logFlushResult, EndOfLog);
     XLogCtl->LogwrtRqst.Write = EndOfLog;
     XLogCtl->LogwrtRqst.Flush = EndOfLog;
  
     /*
      * Preallocate additional log files, if wanted.
      */
     PreallocXlogFiles(EndOfLog, newTLI);
  
     /*
      * Okay, we're officially UP.
      */
     InRecovery = false;
  
     /* start the archive_timeout timer and LSN running */
     XLogCtl->lastSegSwitchTime = (pg_time_t) time(NULL);
     XLogCtl->lastSegSwitchLSN = EndOfLog;
  
     /* also initialize latestCompletedXid, to nextXid - 1 */
     LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE);
     TransamVariables->latestCompletedXid = TransamVariables->nextXid;
     FullTransactionIdRetreat(&TransamVariables->latestCompletedXid);
     LWLockRelease(ProcArrayLock);
  
     /*
      * Start up subtrans, if not already done for hot standby.  (commit
      * timestamps are started below, if necessary.)
      */
     if (standbyState == STANDBY_DISABLED)
         StartupSUBTRANS(oldestActiveXID);
  
     /*
      * Perform end of recovery actions for any SLRUs that need it.
      */
     TrimCLOG();
     TrimMultiXact();
  
     /*
      * Reload shared-memory state for prepared transactions.  This needs to
      * happen before renaming the last partial segment of the old timeline as
      * it may be possible that we have to recovery some transactions from it.
      */
     RecoverPreparedTransactions();
  
     /* Shut down xlogreader */
     ShutdownWalRecovery();
  
     /* Enable WAL writes for this backend only. */
     LocalSetXLogInsertAllowed();
  
     /* If necessary, write overwrite-contrecord before doing anything else */
     if (!XLogRecPtrIsInvalid(abortedRecPtr))
     {
         Assert(!XLogRecPtrIsInvalid(missingContrecPtr));
         CreateOverwriteContrecordRecord(abortedRecPtr, missingContrecPtr, newTLI);
     }
  
     /*
      * Update full_page_writes in shared memory and write an XLOG_FPW_CHANGE
      * record before resource manager writes cleanup WAL records or checkpoint
      * record is written.
      */
     Insert->fullPageWrites = lastFullPageWrites;
     UpdateFullPageWrites();
  
     /*
      * Emit checkpoint or end-of-recovery record in XLOG, if required.
      */
     if (performedWalRecovery)
         promoted = PerformRecoveryXLogAction();
  
     /*
      * If any of the critical GUCs have changed, log them before we allow
      * backends to write WAL.
      */
     XLogReportParameters();
  
     /* If this is archive recovery, perform post-recovery cleanup actions. */
     if (ArchiveRecoveryRequested)
         CleanupAfterArchiveRecovery(EndOfLogTLI, EndOfLog, newTLI);
  
     /*
      * Local WAL inserts enabled, so it's time to finish initialization of
      * commit timestamp.
      */
     CompleteCommitTsInitialization();
  
     /*
      * All done with end-of-recovery actions.
      *
      * Now allow backends to write WAL and update the control file status in
      * consequence.  SharedRecoveryState, that controls if backends can write
      * WAL, is updated while holding ControlFileLock to prevent other backends
      * to look at an inconsistent state of the control file in shared memory.
      * There is still a small window during which backends can write WAL and
      * the control file is still referring to a system not in DB_IN_PRODUCTION
      * state while looking at the on-disk control file.
      *
      * Also, we use info_lck to update SharedRecoveryState to ensure that
      * there are no race conditions concerning visibility of other recent
      * updates to shared memory.
      */
     LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
     ControlFile->state = DB_IN_PRODUCTION;
  
     SpinLockAcquire(&XLogCtl->info_lck);
     XLogCtl->SharedRecoveryState = RECOVERY_STATE_DONE;
     SpinLockRelease(&XLogCtl->info_lck);
  
     UpdateControlFile();
     LWLockRelease(ControlFileLock);
  
     /*
      * Shutdown the recovery environment.  This must occur after
      * RecoverPreparedTransactions() (see notes in lock_twophase_recover())
      * and after switching SharedRecoveryState to RECOVERY_STATE_DONE so as
      * any session building a snapshot will not rely on KnownAssignedXids as
      * RecoveryInProgress() would return false at this stage.  This is
      * particularly critical for prepared 2PC transactions, that would still
      * need to be included in snapshots once recovery has ended.
      */
     if (standbyState != STANDBY_DISABLED)
         ShutdownRecoveryTransactionEnvironment();
  
     /*
      * If there were cascading standby servers connected to us, nudge any wal
      * sender processes to notice that we've been promoted.
      */
     WalSndWakeup(true, true);
  
     /*
      * If this was a promotion, request an (online) checkpoint now. This isn't
      * required for consistency, but the last restartpoint might be far back,
      * and in case of a crash, recovering from it might take a longer than is
      * appropriate now that we're not in standby mode anymore.
      */
     if (promoted)
         RequestCheckpoint(CHECKPOINT_FORCE);
 }

Referenced by InitPostgres(), and StartupProcessMain().

◆ str_time()

static char * str_time ( pg_time_t tnow )

static

Definition at line 5164 of file xlog.c.

 {
     static char buf[128];
  
     pg_strftime(buf, sizeof(buf),
                 "%Y-%m-%d %H:%M:%S %Z",
                 pg_localtime(&tnow, log_timezone));
  
     return buf;
 }

References buf, log_timezone, pg_localtime(), and pg_strftime().

Referenced by StartupXLOG().

◆ SwitchIntoArchiveRecovery()

void SwitchIntoArchiveRecovery	(	XLogRecPtr	EndRecPtr,
		TimeLineID	replayTLI
	)

Definition at line 6165 of file xlog.c.

 {
     /* initialize minRecoveryPoint to this record */
     LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
     ControlFile->state = DB_IN_ARCHIVE_RECOVERY;
     if (ControlFile->minRecoveryPoint < EndRecPtr)
     {
         ControlFile->minRecoveryPoint = EndRecPtr;
         ControlFile->minRecoveryPointTLI = replayTLI;
     }
     /* update local copy */
     LocalMinRecoveryPoint = ControlFile->minRecoveryPoint;
     LocalMinRecoveryPointTLI = ControlFile->minRecoveryPointTLI;
  
     /*
      * The startup process can update its local copy of minRecoveryPoint from
      * this point.
      */
     updateMinRecoveryPoint = true;
  
     UpdateControlFile();
  
     /*
      * We update SharedRecoveryState while holding the lock on ControlFileLock
      * so both states are consistent in shared memory.
      */
     SpinLockAcquire(&XLogCtl->info_lck);
     XLogCtl->SharedRecoveryState = RECOVERY_STATE_ARCHIVE;
     SpinLockRelease(&XLogCtl->info_lck);
  
     LWLockRelease(ControlFileLock);
 }

References ControlFile, DB_IN_ARCHIVE_RECOVERY, XLogCtlData::info_lck, LocalMinRecoveryPoint, LocalMinRecoveryPointTLI, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), ControlFileData::minRecoveryPoint, ControlFileData::minRecoveryPointTLI, RECOVERY_STATE_ARCHIVE, XLogCtlData::SharedRecoveryState, SpinLockAcquire, SpinLockRelease, ControlFileData::state, UpdateControlFile(), updateMinRecoveryPoint, and XLogCtl.

Referenced by ReadRecord().

◆ update_checkpoint_display()

static void update_checkpoint_display	(	int	flags,
		bool	restartpoint,
		bool	reset
	)

static

Definition at line 6759 of file xlog.c.

 {
     /*
      * The status is reported only for end-of-recovery and shutdown
      * checkpoints or shutdown restartpoints.  Updating the ps display is
      * useful in those situations as it may not be possible to rely on
      * pg_stat_activity to see the status of the checkpointer or the startup
      * process.
      */
     if ((flags & (CHECKPOINT_END_OF_RECOVERY | CHECKPOINT_IS_SHUTDOWN)) == 0)
         return;
  
     if (reset)
         set_ps_display("");
     else
     {
         char        activitymsg[128];
  
         snprintf(activitymsg, sizeof(activitymsg), "performing %s%s%s",
                  (flags & CHECKPOINT_END_OF_RECOVERY) ? "end-of-recovery " : "",
                  (flags & CHECKPOINT_IS_SHUTDOWN) ? "shutdown " : "",
                  restartpoint ? "restartpoint" : "checkpoint");
         set_ps_display(activitymsg);
     }
 }

References CHECKPOINT_END_OF_RECOVERY, CHECKPOINT_IS_SHUTDOWN, reset(), set_ps_display(), and snprintf.

Referenced by CreateCheckPoint(), and CreateRestartPoint().

◆ UpdateCheckPointDistanceEstimate()

static void UpdateCheckPointDistanceEstimate ( uint64 nbytes )

static

Definition at line 6721 of file xlog.c.

 {
     /*
      * To estimate the number of segments consumed between checkpoints, keep a
      * moving average of the amount of WAL generated in previous checkpoint
      * cycles. However, if the load is bursty, with quiet periods and busy
      * periods, we want to cater for the peak load. So instead of a plain
      * moving average, let the average decline slowly if the previous cycle
      * used less WAL than estimated, but bump it up immediately if it used
      * more.
      *
      * When checkpoints are triggered by max_wal_size, this should converge to
      * CheckpointSegments * wal_segment_size,
      *
      * Note: This doesn't pay any attention to what caused the checkpoint.
      * Checkpoints triggered manually with CHECKPOINT command, or by e.g.
      * starting a base backup, are counted the same as those created
      * automatically. The slow-decline will largely mask them out, if they are
      * not frequent. If they are frequent, it seems reasonable to count them
      * in as any others; if you issue a manual checkpoint every 5 minutes and
      * never let a timed checkpoint happen, it makes sense to base the
      * preallocation on that 5 minute interval rather than whatever
      * checkpoint_timeout is set to.
      */
     PrevCheckPointDistance = nbytes;
     if (CheckPointDistanceEstimate < nbytes)
         CheckPointDistanceEstimate = nbytes;
     else
         CheckPointDistanceEstimate =
             (0.90 * CheckPointDistanceEstimate + 0.10 * (double) nbytes);
 }

References CheckPointDistanceEstimate, and PrevCheckPointDistance.

Referenced by CreateCheckPoint(), and CreateRestartPoint().

◆ UpdateControlFile()

static void UpdateControlFile ( void )

static

Definition at line 4524 of file xlog.c.

 {
     update_controlfile(DataDir, ControlFile, true);
 }

References ControlFile, DataDir, and update_controlfile().

Referenced by CreateCheckPoint(), CreateEndOfRecoveryRecord(), CreateRestartPoint(), ReachedEndOfBackup(), StartupXLOG(), SwitchIntoArchiveRecovery(), UpdateMinRecoveryPoint(), xlog_redo(), and XLogReportParameters().

◆ UpdateFullPageWrites()

void UpdateFullPageWrites ( void )

Definition at line 8096 of file xlog.c.

 {
     XLogCtlInsert *Insert = &XLogCtl->Insert;
     bool        recoveryInProgress;
  
     /*
      * Do nothing if full_page_writes has not been changed.
      *
      * It's safe to check the shared full_page_writes without the lock,
      * because we assume that there is no concurrently running process which
      * can update it.
      */
     if (fullPageWrites == Insert->fullPageWrites)
         return;
  
     /*
      * Perform this outside critical section so that the WAL insert
      * initialization done by RecoveryInProgress() doesn't trigger an
      * assertion failure.
      */
     recoveryInProgress = RecoveryInProgress();
  
     START_CRIT_SECTION();
  
     /*
      * It's always safe to take full page images, even when not strictly
      * required, but not the other round. So if we're setting full_page_writes
      * to true, first set it true and then write the WAL record. If we're
      * setting it to false, first write the WAL record and then set the global
      * flag.
      */
     if (fullPageWrites)
     {
         WALInsertLockAcquireExclusive();
         Insert->fullPageWrites = true;
         WALInsertLockRelease();
     }
  
     /*
      * Write an XLOG_FPW_CHANGE record. This allows us to keep track of
      * full_page_writes during archive recovery, if required.
      */
     if (XLogStandbyInfoActive() && !recoveryInProgress)
     {
         XLogBeginInsert();
         XLogRegisterData((char *) (&fullPageWrites), sizeof(bool));
  
         XLogInsert(RM_XLOG_ID, XLOG_FPW_CHANGE);
     }
  
     if (!fullPageWrites)
     {
         WALInsertLockAcquireExclusive();
         Insert->fullPageWrites = false;
         WALInsertLockRelease();
     }
     END_CRIT_SECTION();
 }

References END_CRIT_SECTION, fullPageWrites, XLogCtlData::Insert, Insert(), RecoveryInProgress(), START_CRIT_SECTION, WALInsertLockAcquireExclusive(), WALInsertLockRelease(), XLOG_FPW_CHANGE, XLogBeginInsert(), XLogCtl, XLogInsert(), XLogRegisterData(), and XLogStandbyInfoActive.

Referenced by StartupXLOG(), and UpdateSharedMemoryConfig().

◆ UpdateLastRemovedPtr()

static void UpdateLastRemovedPtr ( char * filename )

static

Definition at line 3799 of file xlog.c.

 {
     uint32      tli;
     XLogSegNo   segno;
  
     XLogFromFileName(filename, &tli, &segno, wal_segment_size);
  
     SpinLockAcquire(&XLogCtl->info_lck);
     if (segno > XLogCtl->lastRemovedSegNo)
         XLogCtl->lastRemovedSegNo = segno;
     SpinLockRelease(&XLogCtl->info_lck);
 }

References filename, XLogCtlData::info_lck, XLogCtlData::lastRemovedSegNo, SpinLockAcquire, SpinLockRelease, wal_segment_size, XLogCtl, and XLogFromFileName().

Referenced by RemoveOldXlogFiles().

◆ UpdateMinRecoveryPoint()

static void UpdateMinRecoveryPoint	(	XLogRecPtr	lsn,
		bool	force
	)

static

Definition at line 2709 of file xlog.c.

 {
     /* Quick check using our local copy of the variable */
     if (!updateMinRecoveryPoint || (!force && lsn <= LocalMinRecoveryPoint))
         return;
  
     /*
      * An invalid minRecoveryPoint means that we need to recover all the WAL,
      * i.e., we're doing crash recovery.  We never modify the control file's
      * value in that case, so we can short-circuit future checks here too. The
      * local values of minRecoveryPoint and minRecoveryPointTLI should not be
      * updated until crash recovery finishes.  We only do this for the startup
      * process as it should not update its own reference of minRecoveryPoint
      * until it has finished crash recovery to make sure that all WAL
      * available is replayed in this case.  This also saves from extra locks
      * taken on the control file from the startup process.
      */
     if (XLogRecPtrIsInvalid(LocalMinRecoveryPoint) && InRecovery)
     {
         updateMinRecoveryPoint = false;
         return;
     }
  
     LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
  
     /* update local copy */
     LocalMinRecoveryPoint = ControlFile->minRecoveryPoint;
     LocalMinRecoveryPointTLI = ControlFile->minRecoveryPointTLI;
  
     if (XLogRecPtrIsInvalid(LocalMinRecoveryPoint))
         updateMinRecoveryPoint = false;
     else if (force || LocalMinRecoveryPoint < lsn)
     {
         XLogRecPtr  newMinRecoveryPoint;
         TimeLineID  newMinRecoveryPointTLI;
  
         /*
          * To avoid having to update the control file too often, we update it
          * all the way to the last record being replayed, even though 'lsn'
          * would suffice for correctness.  This also allows the 'force' case
          * to not need a valid 'lsn' value.
          *
          * Another important reason for doing it this way is that the passed
          * 'lsn' value could be bogus, i.e., past the end of available WAL, if
          * the caller got it from a corrupted heap page.  Accepting such a
          * value as the min recovery point would prevent us from coming up at
          * all.  Instead, we just log a warning and continue with recovery.
          * (See also the comments about corrupt LSNs in XLogFlush.)
          */
         newMinRecoveryPoint = GetCurrentReplayRecPtr(&newMinRecoveryPointTLI);
         if (!force && newMinRecoveryPoint < lsn)
             elog(WARNING,
                  "xlog min recovery request %X/%X is past current point %X/%X",
                  LSN_FORMAT_ARGS(lsn), LSN_FORMAT_ARGS(newMinRecoveryPoint));
  
         /* update control file */
         if (ControlFile->minRecoveryPoint < newMinRecoveryPoint)
         {
             ControlFile->minRecoveryPoint = newMinRecoveryPoint;
             ControlFile->minRecoveryPointTLI = newMinRecoveryPointTLI;
             UpdateControlFile();
             LocalMinRecoveryPoint = newMinRecoveryPoint;
             LocalMinRecoveryPointTLI = newMinRecoveryPointTLI;
  
             ereport(DEBUG2,
                     (errmsg_internal("updated min recovery point to %X/%X on timeline %u",
                                      LSN_FORMAT_ARGS(newMinRecoveryPoint),
                                      newMinRecoveryPointTLI)));
         }
     }
     LWLockRelease(ControlFileLock);
 }

References ControlFile, DEBUG2, elog, ereport, errmsg_internal(), GetCurrentReplayRecPtr(), InRecovery, LocalMinRecoveryPoint, LocalMinRecoveryPointTLI, LSN_FORMAT_ARGS, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), ControlFileData::minRecoveryPoint, ControlFileData::minRecoveryPointTLI, UpdateControlFile(), updateMinRecoveryPoint, WARNING, and XLogRecPtrIsInvalid.

Referenced by CreateRestartPoint(), XLogFlush(), and XLogInitNewTimeline().

◆ ValidateXLOGDirectoryStructure()

static void ValidateXLOGDirectoryStructure ( void )

static

Definition at line 4086 of file xlog.c.

 {
     char        path[MAXPGPATH];
     struct stat stat_buf;
  
     /* Check for pg_wal; if it doesn't exist, error out */
     if (stat(XLOGDIR, &stat_buf) != 0 ||
         !S_ISDIR(stat_buf.st_mode))
         ereport(FATAL,
                 (errcode_for_file_access(),
                  errmsg("required WAL directory \"%s\" does not exist",
                         XLOGDIR)));
  
     /* Check for archive_status */
     snprintf(path, MAXPGPATH, XLOGDIR "/archive_status");
     if (stat(path, &stat_buf) == 0)
     {
         /* Check for weird cases where it exists but isn't a directory */
         if (!S_ISDIR(stat_buf.st_mode))
             ereport(FATAL,
                     (errcode_for_file_access(),
                      errmsg("required WAL directory \"%s\" does not exist",
                             path)));
     }
     else
     {
         ereport(LOG,
                 (errmsg("creating missing WAL directory \"%s\"", path)));
         if (MakePGDirectory(path) < 0)
             ereport(FATAL,
                     (errcode_for_file_access(),
                      errmsg("could not create missing directory \"%s\": %m",
                             path)));
     }
  
     /* Check for summaries */
     snprintf(path, MAXPGPATH, XLOGDIR "/summaries");
     if (stat(path, &stat_buf) == 0)
     {
         /* Check for weird cases where it exists but isn't a directory */
         if (!S_ISDIR(stat_buf.st_mode))
             ereport(FATAL,
                     (errmsg("required WAL directory \"%s\" does not exist",
                             path)));
     }
     else
     {
         ereport(LOG,
                 (errmsg("creating missing WAL directory \"%s\"", path)));
         if (MakePGDirectory(path) < 0)
             ereport(FATAL,
                     (errmsg("could not create missing directory \"%s\": %m",
                             path)));
     }
 }

References ereport, errcode_for_file_access(), errmsg(), FATAL, LOG, MakePGDirectory(), MAXPGPATH, S_ISDIR, snprintf, stat::st_mode, stat, and XLOGDIR.

Referenced by StartupXLOG().

◆ WaitXLogInsertionsToFinish()

static XLogRecPtr WaitXLogInsertionsToFinish ( XLogRecPtr upto )

static

Definition at line 1497 of file xlog.c.

 {
     uint64      bytepos;
     XLogRecPtr  inserted;
     XLogRecPtr  reservedUpto;
     XLogRecPtr  finishedUpto;
     XLogCtlInsert *Insert = &XLogCtl->Insert;
     int         i;
  
     if (MyProc == NULL)
         elog(PANIC, "cannot wait without a PGPROC structure");
  
     /*
      * Check if there's any work to do.  Use a barrier to ensure we get the
      * freshest value.
      */
     inserted = pg_atomic_read_membarrier_u64(&XLogCtl->logInsertResult);
     if (upto <= inserted)
         return inserted;
  
     /* Read the current insert position */
     SpinLockAcquire(&Insert->insertpos_lck);
     bytepos = Insert->CurrBytePos;
     SpinLockRelease(&Insert->insertpos_lck);
     reservedUpto = XLogBytePosToEndRecPtr(bytepos);
  
     /*
      * No-one should request to flush a piece of WAL that hasn't even been
      * reserved yet. However, it can happen if there is a block with a bogus
      * LSN on disk, for example. XLogFlush checks for that situation and
      * complains, but only after the flush. Here we just assume that to mean
      * that all WAL that has been reserved needs to be finished. In this
      * corner-case, the return value can be smaller than 'upto' argument.
      */
     if (upto > reservedUpto)
     {
         ereport(LOG,
                 (errmsg("request to flush past end of generated WAL; request %X/%X, current position %X/%X",
                         LSN_FORMAT_ARGS(upto), LSN_FORMAT_ARGS(reservedUpto))));
         upto = reservedUpto;
     }
  
     /*
      * Loop through all the locks, sleeping on any in-progress insert older
      * than 'upto'.
      *
      * finishedUpto is our return value, indicating the point upto which all
      * the WAL insertions have been finished. Initialize it to the head of
      * reserved WAL, and as we iterate through the insertion locks, back it
      * out for any insertion that's still in progress.
      */
     finishedUpto = reservedUpto;
     for (i = 0; i < NUM_XLOGINSERT_LOCKS; i++)
     {
         XLogRecPtr  insertingat = InvalidXLogRecPtr;
  
         do
         {
             /*
              * See if this insertion is in progress.  LWLockWaitForVar will
              * wait for the lock to be released, or for the 'value' to be set
              * by a LWLockUpdateVar call.  When a lock is initially acquired,
              * its value is 0 (InvalidXLogRecPtr), which means that we don't
              * know where it's inserting yet.  We will have to wait for it. If
              * it's a small insertion, the record will most likely fit on the
              * same page and the inserter will release the lock without ever
              * calling LWLockUpdateVar.  But if it has to sleep, it will
              * advertise the insertion point with LWLockUpdateVar before
              * sleeping.
              *
              * In this loop we are only waiting for insertions that started
              * before WaitXLogInsertionsToFinish was called.  The lack of
              * memory barriers in the loop means that we might see locks as
              * "unused" that have since become used.  This is fine because
              * they only can be used for later insertions that we would not
              * want to wait on anyway.  Not taking a lock to acquire the
              * current insertingAt value means that we might see older
              * insertingAt values.  This is also fine, because if we read a
              * value too old, we will add ourselves to the wait queue, which
              * contains atomic operations.
              */
             if (LWLockWaitForVar(&WALInsertLocks[i].l.lock,
                                  &WALInsertLocks[i].l.insertingAt,
                                  insertingat, &insertingat))
             {
                 /* the lock was free, so no insertion in progress */
                 insertingat = InvalidXLogRecPtr;
                 break;
             }
  
             /*
              * This insertion is still in progress. Have to wait, unless the
              * inserter has proceeded past 'upto'.
              */
         } while (insertingat < upto);
  
         if (insertingat != InvalidXLogRecPtr && insertingat < finishedUpto)
             finishedUpto = insertingat;
     }
  
     /*
      * Advance the limit we know to have been inserted and return the freshest
      * value we know of, which might be beyond what we requested if somebody
      * is concurrently doing this with an 'upto' pointer ahead of us.
      */
     finishedUpto = pg_atomic_monotonic_advance_u64(&XLogCtl->logInsertResult,
                                                    finishedUpto);
  
     return finishedUpto;
 }

References elog, ereport, errmsg(), i, XLogCtlData::Insert, Insert(), WALInsertLock::insertingAt, InvalidXLogRecPtr, WALInsertLockPadded::l, LOG, XLogCtlData::logInsertResult, LSN_FORMAT_ARGS, LWLockWaitForVar(), MyProc, NUM_XLOGINSERT_LOCKS, PANIC, pg_atomic_monotonic_advance_u64(), pg_atomic_read_membarrier_u64(), SpinLockAcquire, SpinLockRelease, WALInsertLocks, XLogBytePosToEndRecPtr(), and XLogCtl.

Referenced by AdvanceXLInsertBuffer(), XLogBackgroundFlush(), and XLogFlush().

◆ WALInsertLockAcquire()

static void WALInsertLockAcquire ( void )

static

Definition at line 1364 of file xlog.c.

 {
     bool        immed;
  
     /*
      * It doesn't matter which of the WAL insertion locks we acquire, so try
      * the one we used last time.  If the system isn't particularly busy, it's
      * a good bet that it's still available, and it's good to have some
      * affinity to a particular lock so that you don't unnecessarily bounce
      * cache lines between processes when there's no contention.
      *
      * If this is the first time through in this backend, pick a lock
      * (semi-)randomly.  This allows the locks to be used evenly if you have a
      * lot of very short connections.
      */
     static int  lockToTry = -1;
  
     if (lockToTry == -1)
         lockToTry = MyProcNumber % NUM_XLOGINSERT_LOCKS;
     MyLockNo = lockToTry;
  
     /*
      * The insertingAt value is initially set to 0, as we don't know our
      * insert location yet.
      */
     immed = LWLockAcquire(&WALInsertLocks[MyLockNo].l.lock, LW_EXCLUSIVE);
     if (!immed)
     {
         /*
          * If we couldn't get the lock immediately, try another lock next
          * time.  On a system with more insertion locks than concurrent
          * inserters, this causes all the inserters to eventually migrate to a
          * lock that no-one else is using.  On a system with more inserters
          * than locks, it still helps to distribute the inserters evenly
          * across the locks.
          */
         lockToTry = (lockToTry + 1) % NUM_XLOGINSERT_LOCKS;
     }
 }

References LW_EXCLUSIVE, LWLockAcquire(), MyLockNo, MyProcNumber, NUM_XLOGINSERT_LOCKS, and WALInsertLocks.

Referenced by CreateOverwriteContrecordRecord(), and XLogInsertRecord().

◆ WALInsertLockAcquireExclusive()

static void WALInsertLockAcquireExclusive ( void )

static

Definition at line 1409 of file xlog.c.

 {
     int         i;
  
     /*
      * When holding all the locks, all but the last lock's insertingAt
      * indicator is set to 0xFFFFFFFFFFFFFFFF, which is higher than any real
      * XLogRecPtr value, to make sure that no-one blocks waiting on those.
      */
     for (i = 0; i < NUM_XLOGINSERT_LOCKS - 1; i++)
     {
         LWLockAcquire(&WALInsertLocks[i].l.lock, LW_EXCLUSIVE);
         LWLockUpdateVar(&WALInsertLocks[i].l.lock,
                         &WALInsertLocks[i].l.insertingAt,
                         PG_UINT64_MAX);
     }
     /* Variable value reset to 0 at release */
     LWLockAcquire(&WALInsertLocks[i].l.lock, LW_EXCLUSIVE);
  
     holdingAllLocks = true;
 }

References holdingAllLocks, i, WALInsertLock::insertingAt, WALInsertLockPadded::l, LW_EXCLUSIVE, LWLockAcquire(), LWLockUpdateVar(), NUM_XLOGINSERT_LOCKS, PG_UINT64_MAX, and WALInsertLocks.

Referenced by CreateCheckPoint(), CreateEndOfRecoveryRecord(), CreateRestartPoint(), do_pg_abort_backup(), do_pg_backup_start(), do_pg_backup_stop(), UpdateFullPageWrites(), and XLogInsertRecord().

◆ WALInsertLockRelease()

static void WALInsertLockRelease ( void )

static

Definition at line 1438 of file xlog.c.

 {
     if (holdingAllLocks)
     {
         int         i;
  
         for (i = 0; i < NUM_XLOGINSERT_LOCKS; i++)
             LWLockReleaseClearVar(&WALInsertLocks[i].l.lock,
                                   &WALInsertLocks[i].l.insertingAt,
                                   0);
  
         holdingAllLocks = false;
     }
     else
     {
         LWLockReleaseClearVar(&WALInsertLocks[MyLockNo].l.lock,
                               &WALInsertLocks[MyLockNo].l.insertingAt,
                               0);
     }
 }

References holdingAllLocks, i, WALInsertLock::insertingAt, WALInsertLockPadded::l, LWLockReleaseClearVar(), MyLockNo, NUM_XLOGINSERT_LOCKS, and WALInsertLocks.

Referenced by CreateCheckPoint(), CreateEndOfRecoveryRecord(), CreateOverwriteContrecordRecord(), CreateRestartPoint(), do_pg_abort_backup(), do_pg_backup_start(), do_pg_backup_stop(), UpdateFullPageWrites(), and XLogInsertRecord().

◆ WALInsertLockUpdateInsertingAt()

static void WALInsertLockUpdateInsertingAt ( XLogRecPtr insertingAt )

static

Definition at line 1464 of file xlog.c.

 {
     if (holdingAllLocks)
     {
         /*
          * We use the last lock to mark our actual position, see comments in
          * WALInsertLockAcquireExclusive.
          */
         LWLockUpdateVar(&WALInsertLocks[NUM_XLOGINSERT_LOCKS - 1].l.lock,
                         &WALInsertLocks[NUM_XLOGINSERT_LOCKS - 1].l.insertingAt,
                         insertingAt);
     }
     else
         LWLockUpdateVar(&WALInsertLocks[MyLockNo].l.lock,
                         &WALInsertLocks[MyLockNo].l.insertingAt,
                         insertingAt);
 }

References holdingAllLocks, WALInsertLock::insertingAt, WALInsertLockPadded::l, LWLockUpdateVar(), MyLockNo, NUM_XLOGINSERT_LOCKS, and WALInsertLocks.

Referenced by GetXLogBuffer().

◆ WALReadFromBuffers()

Size WALReadFromBuffers	(	char *	dstbuf,
		XLogRecPtr	startptr,
		Size	count,
		TimeLineID	tli
	)

Definition at line 1741 of file xlog.c.

 {
     char       *pdst = dstbuf;
     XLogRecPtr  recptr = startptr;
     XLogRecPtr  inserted;
     Size        nbytes = count;
  
     if (RecoveryInProgress() || tli != GetWALInsertionTimeLine())
         return 0;
  
     Assert(!XLogRecPtrIsInvalid(startptr));
  
     /*
      * Caller should ensure that the requested data has been inserted into WAL
      * buffers before we try to read it.
      */
     inserted = pg_atomic_read_u64(&XLogCtl->logInsertResult);
     if (startptr + count > inserted)
         ereport(ERROR,
                 errmsg("cannot read past end of generated WAL: requested %X/%X, current position %X/%X",
                        LSN_FORMAT_ARGS(startptr + count),
                        LSN_FORMAT_ARGS(inserted)));
  
     /*
      * Loop through the buffers without a lock. For each buffer, atomically
      * read and verify the end pointer, then copy the data out, and finally
      * re-read and re-verify the end pointer.
      *
      * Once a page is evicted, it never returns to the WAL buffers, so if the
      * end pointer matches the expected end pointer before and after we copy
      * the data, then the right page must have been present during the data
      * copy. Read barriers are necessary to ensure that the data copy actually
      * happens between the two verification steps.
      *
      * If either verification fails, we simply terminate the loop and return
      * with the data that had been already copied out successfully.
      */
     while (nbytes > 0)
     {
         uint32      offset = recptr % XLOG_BLCKSZ;
         int         idx = XLogRecPtrToBufIdx(recptr);
         XLogRecPtr  expectedEndPtr;
         XLogRecPtr  endptr;
         const char *page;
         const char *psrc;
         Size        npagebytes;
  
         /*
          * Calculate the end pointer we expect in the xlblocks array if the
          * correct page is present.
          */
         expectedEndPtr = recptr + (XLOG_BLCKSZ - offset);
  
         /*
          * First verification step: check that the correct page is present in
          * the WAL buffers.
          */
         endptr = pg_atomic_read_u64(&XLogCtl->xlblocks[idx]);
         if (expectedEndPtr != endptr)
             break;
  
         /*
          * The correct page is present (or was at the time the endptr was
          * read; must re-verify later). Calculate pointer to source data and
          * determine how much data to read from this page.
          */
         page = XLogCtl->pages + idx * (Size) XLOG_BLCKSZ;
         psrc = page + offset;
         npagebytes = Min(nbytes, XLOG_BLCKSZ - offset);
  
         /*
          * Ensure that the data copy and the first verification step are not
          * reordered.
          */
         pg_read_barrier();
  
         /* data copy */
         memcpy(pdst, psrc, npagebytes);
  
         /*
          * Ensure that the data copy and the second verification step are not
          * reordered.
          */
         pg_read_barrier();
  
         /*
          * Second verification step: check that the page we read from wasn't
          * evicted while we were copying the data.
          */
         endptr = pg_atomic_read_u64(&XLogCtl->xlblocks[idx]);
         if (expectedEndPtr != endptr)
             break;
  
         pdst += npagebytes;
         recptr += npagebytes;
         nbytes -= npagebytes;
     }
  
     Assert(pdst - dstbuf <= count);
  
     return pdst - dstbuf;
 }

References Assert, ereport, errmsg(), ERROR, GetWALInsertionTimeLine(), idx(), XLogCtlData::logInsertResult, LSN_FORMAT_ARGS, Min, XLogCtlData::pages, pg_atomic_read_u64(), pg_read_barrier, RecoveryInProgress(), XLogCtlData::xlblocks, XLogCtl, XLogRecPtrIsInvalid, and XLogRecPtrToBufIdx.

Referenced by XLogSendPhysical().

◆ WriteControlFile()

static void WriteControlFile ( void )

static

Definition at line 4226 of file xlog.c.

 {
     int         fd;
     char        buffer[PG_CONTROL_FILE_SIZE];   /* need not be aligned */
  
     /*
      * Initialize version and compatibility-check fields
      */
     ControlFile->pg_control_version = PG_CONTROL_VERSION;
     ControlFile->catalog_version_no = CATALOG_VERSION_NO;
  
     ControlFile->maxAlign = MAXIMUM_ALIGNOF;
     ControlFile->floatFormat = FLOATFORMAT_VALUE;
  
     ControlFile->blcksz = BLCKSZ;
     ControlFile->relseg_size = RELSEG_SIZE;
     ControlFile->xlog_blcksz = XLOG_BLCKSZ;
     ControlFile->xlog_seg_size = wal_segment_size;
  
     ControlFile->nameDataLen = NAMEDATALEN;
     ControlFile->indexMaxKeys = INDEX_MAX_KEYS;
  
     ControlFile->toast_max_chunk_size = TOAST_MAX_CHUNK_SIZE;
     ControlFile->loblksize = LOBLKSIZE;
  
     ControlFile->float8ByVal = FLOAT8PASSBYVAL;
  
     /* Contents are protected with a CRC */
     INIT_CRC32C(ControlFile->crc);
     COMP_CRC32C(ControlFile->crc,
                 (char *) ControlFile,
                 offsetof(ControlFileData, crc));
     FIN_CRC32C(ControlFile->crc);
  
     /*
      * We write out PG_CONTROL_FILE_SIZE bytes into pg_control, zero-padding
      * the excess over sizeof(ControlFileData).  This reduces the odds of
      * premature-EOF errors when reading pg_control.  We'll still fail when we
      * check the contents of the file, but hopefully with a more specific
      * error than "couldn't read pg_control".
      */
     memset(buffer, 0, PG_CONTROL_FILE_SIZE);
     memcpy(buffer, ControlFile, sizeof(ControlFileData));
  
     fd = BasicOpenFile(XLOG_CONTROL_FILE,
                        O_RDWR | O_CREAT | O_EXCL | PG_BINARY);
     if (fd < 0)
         ereport(PANIC,
                 (errcode_for_file_access(),
                  errmsg("could not create file \"%s\": %m",
                         XLOG_CONTROL_FILE)));
  
     errno = 0;
     pgstat_report_wait_start(WAIT_EVENT_CONTROL_FILE_WRITE);
     if (write(fd, buffer, PG_CONTROL_FILE_SIZE) != PG_CONTROL_FILE_SIZE)
     {
         /* if write didn't set errno, assume problem is no disk space */
         if (errno == 0)
             errno = ENOSPC;
         ereport(PANIC,
                 (errcode_for_file_access(),
                  errmsg("could not write to file \"%s\": %m",
                         XLOG_CONTROL_FILE)));
     }
     pgstat_report_wait_end();
  
     pgstat_report_wait_start(WAIT_EVENT_CONTROL_FILE_SYNC);
     if (pg_fsync(fd) != 0)
         ereport(PANIC,
                 (errcode_for_file_access(),
                  errmsg("could not fsync file \"%s\": %m",
                         XLOG_CONTROL_FILE)));
     pgstat_report_wait_end();
  
     if (close(fd) != 0)
         ereport(PANIC,
                 (errcode_for_file_access(),
                  errmsg("could not close file \"%s\": %m",
                         XLOG_CONTROL_FILE)));
 }

References BasicOpenFile(), ControlFileData::blcksz, CATALOG_VERSION_NO, ControlFileData::catalog_version_no, close, COMP_CRC32C, ControlFile, ControlFileData::crc, crc, ereport, errcode_for_file_access(), errmsg(), fd(), FIN_CRC32C, ControlFileData::float8ByVal, FLOAT8PASSBYVAL, ControlFileData::floatFormat, FLOATFORMAT_VALUE, INDEX_MAX_KEYS, ControlFileData::indexMaxKeys, INIT_CRC32C, ControlFileData::loblksize, LOBLKSIZE, ControlFileData::maxAlign, ControlFileData::nameDataLen, NAMEDATALEN, PANIC, PG_BINARY, PG_CONTROL_FILE_SIZE, PG_CONTROL_VERSION, ControlFileData::pg_control_version, pg_fsync(), pgstat_report_wait_end(), pgstat_report_wait_start(), ControlFileData::relseg_size, TOAST_MAX_CHUNK_SIZE, ControlFileData::toast_max_chunk_size, wal_segment_size, write, ControlFileData::xlog_blcksz, XLOG_CONTROL_FILE, and ControlFileData::xlog_seg_size.

Referenced by BootStrapXLOG().

◆ xlog_redo()

void xlog_redo ( XLogReaderState * record )

Definition at line 8165 of file xlog.c.

 {
     uint8       info = XLogRecGetInfo(record) & ~XLR_INFO_MASK;
     XLogRecPtr  lsn = record->EndRecPtr;
  
     /*
      * In XLOG rmgr, backup blocks are only used by XLOG_FPI and
      * XLOG_FPI_FOR_HINT records.
      */
     Assert(info == XLOG_FPI || info == XLOG_FPI_FOR_HINT ||
            !XLogRecHasAnyBlockRefs(record));
  
     if (info == XLOG_NEXTOID)
     {
         Oid         nextOid;
  
         /*
          * We used to try to take the maximum of TransamVariables->nextOid and
          * the recorded nextOid, but that fails if the OID counter wraps
          * around.  Since no OID allocation should be happening during replay
          * anyway, better to just believe the record exactly.  We still take
          * OidGenLock while setting the variable, just in case.
          */
         memcpy(&nextOid, XLogRecGetData(record), sizeof(Oid));
         LWLockAcquire(OidGenLock, LW_EXCLUSIVE);
         TransamVariables->nextOid = nextOid;
         TransamVariables->oidCount = 0;
         LWLockRelease(OidGenLock);
     }
     else if (info == XLOG_CHECKPOINT_SHUTDOWN)
     {
         CheckPoint  checkPoint;
         TimeLineID  replayTLI;
  
         memcpy(&checkPoint, XLogRecGetData(record), sizeof(CheckPoint));
         /* In a SHUTDOWN checkpoint, believe the counters exactly */
         LWLockAcquire(XidGenLock, LW_EXCLUSIVE);
         TransamVariables->nextXid = checkPoint.nextXid;
         LWLockRelease(XidGenLock);
         LWLockAcquire(OidGenLock, LW_EXCLUSIVE);
         TransamVariables->nextOid = checkPoint.nextOid;
         TransamVariables->oidCount = 0;
         LWLockRelease(OidGenLock);
         MultiXactSetNextMXact(checkPoint.nextMulti,
                               checkPoint.nextMultiOffset);
  
         MultiXactAdvanceOldest(checkPoint.oldestMulti,
                                checkPoint.oldestMultiDB);
  
         /*
          * No need to set oldestClogXid here as well; it'll be set when we
          * redo an xl_clog_truncate if it changed since initialization.
          */
         SetTransactionIdLimit(checkPoint.oldestXid, checkPoint.oldestXidDB);
  
         /*
          * If we see a shutdown checkpoint while waiting for an end-of-backup
          * record, the backup was canceled and the end-of-backup record will
          * never arrive.
          */
         if (ArchiveRecoveryRequested &&
             !XLogRecPtrIsInvalid(ControlFile->backupStartPoint) &&
             XLogRecPtrIsInvalid(ControlFile->backupEndPoint))
             ereport(PANIC,
                     (errmsg("online backup was canceled, recovery cannot continue")));
  
         /*
          * If we see a shutdown checkpoint, we know that nothing was running
          * on the primary at this point. So fake-up an empty running-xacts
          * record and use that here and now. Recover additional standby state
          * for prepared transactions.
          */
         if (standbyState >= STANDBY_INITIALIZED)
         {
             TransactionId *xids;
             int         nxids;
             TransactionId oldestActiveXID;
             TransactionId latestCompletedXid;
             RunningTransactionsData running;
  
             oldestActiveXID = PrescanPreparedTransactions(&xids, &nxids);
  
             /* Update pg_subtrans entries for any prepared transactions */
             StandbyRecoverPreparedTransactions();
  
             /*
              * Construct a RunningTransactions snapshot representing a shut
              * down server, with only prepared transactions still alive. We're
              * never overflowed at this point because all subxids are listed
              * with their parent prepared transactions.
              */
             running.xcnt = nxids;
             running.subxcnt = 0;
             running.subxid_status = SUBXIDS_IN_SUBTRANS;
             running.nextXid = XidFromFullTransactionId(checkPoint.nextXid);
             running.oldestRunningXid = oldestActiveXID;
             latestCompletedXid = XidFromFullTransactionId(checkPoint.nextXid);
             TransactionIdRetreat(latestCompletedXid);
             Assert(TransactionIdIsNormal(latestCompletedXid));
             running.latestCompletedXid = latestCompletedXid;
             running.xids = xids;
  
             ProcArrayApplyRecoveryInfo(&running);
         }
  
         /* ControlFile->checkPointCopy always tracks the latest ckpt XID */
         LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
         ControlFile->checkPointCopy.nextXid = checkPoint.nextXid;
         LWLockRelease(ControlFileLock);
  
         /* Update shared-memory copy of checkpoint XID/epoch */
         SpinLockAcquire(&XLogCtl->info_lck);
         XLogCtl->ckptFullXid = checkPoint.nextXid;
         SpinLockRelease(&XLogCtl->info_lck);
  
         /*
          * We should've already switched to the new TLI before replaying this
          * record.
          */
         (void) GetCurrentReplayRecPtr(&replayTLI);
         if (checkPoint.ThisTimeLineID != replayTLI)
             ereport(PANIC,
                     (errmsg("unexpected timeline ID %u (should be %u) in shutdown checkpoint record",
                             checkPoint.ThisTimeLineID, replayTLI)));
  
         RecoveryRestartPoint(&checkPoint, record);
     }
     else if (info == XLOG_CHECKPOINT_ONLINE)
     {
         CheckPoint  checkPoint;
         TimeLineID  replayTLI;
  
         memcpy(&checkPoint, XLogRecGetData(record), sizeof(CheckPoint));
         /* In an ONLINE checkpoint, treat the XID counter as a minimum */
         LWLockAcquire(XidGenLock, LW_EXCLUSIVE);
         if (FullTransactionIdPrecedes(TransamVariables->nextXid,
                                       checkPoint.nextXid))
             TransamVariables->nextXid = checkPoint.nextXid;
         LWLockRelease(XidGenLock);
  
         /*
          * We ignore the nextOid counter in an ONLINE checkpoint, preferring
          * to track OID assignment through XLOG_NEXTOID records.  The nextOid
          * counter is from the start of the checkpoint and might well be stale
          * compared to later XLOG_NEXTOID records.  We could try to take the
          * maximum of the nextOid counter and our latest value, but since
          * there's no particular guarantee about the speed with which the OID
          * counter wraps around, that's a risky thing to do.  In any case,
          * users of the nextOid counter are required to avoid assignment of
          * duplicates, so that a somewhat out-of-date value should be safe.
          */
  
         /* Handle multixact */
         MultiXactAdvanceNextMXact(checkPoint.nextMulti,
                                   checkPoint.nextMultiOffset);
  
         /*
          * NB: This may perform multixact truncation when replaying WAL
          * generated by an older primary.
          */
         MultiXactAdvanceOldest(checkPoint.oldestMulti,
                                checkPoint.oldestMultiDB);
         if (TransactionIdPrecedes(TransamVariables->oldestXid,
                                   checkPoint.oldestXid))
             SetTransactionIdLimit(checkPoint.oldestXid,
                                   checkPoint.oldestXidDB);
         /* ControlFile->checkPointCopy always tracks the latest ckpt XID */
         LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
         ControlFile->checkPointCopy.nextXid = checkPoint.nextXid;
         LWLockRelease(ControlFileLock);
  
         /* Update shared-memory copy of checkpoint XID/epoch */
         SpinLockAcquire(&XLogCtl->info_lck);
         XLogCtl->ckptFullXid = checkPoint.nextXid;
         SpinLockRelease(&XLogCtl->info_lck);
  
         /* TLI should not change in an on-line checkpoint */
         (void) GetCurrentReplayRecPtr(&replayTLI);
         if (checkPoint.ThisTimeLineID != replayTLI)
             ereport(PANIC,
                     (errmsg("unexpected timeline ID %u (should be %u) in online checkpoint record",
                             checkPoint.ThisTimeLineID, replayTLI)));
  
         RecoveryRestartPoint(&checkPoint, record);
     }
     else if (info == XLOG_OVERWRITE_CONTRECORD)
     {
         /* nothing to do here, handled in xlogrecovery_redo() */
     }
     else if (info == XLOG_END_OF_RECOVERY)
     {
         xl_end_of_recovery xlrec;
         TimeLineID  replayTLI;
  
         memcpy(&xlrec, XLogRecGetData(record), sizeof(xl_end_of_recovery));
  
         /*
          * For Hot Standby, we could treat this like a Shutdown Checkpoint,
          * but this case is rarer and harder to test, so the benefit doesn't
          * outweigh the potential extra cost of maintenance.
          */
  
         /*
          * We should've already switched to the new TLI before replaying this
          * record.
          */
         (void) GetCurrentReplayRecPtr(&replayTLI);
         if (xlrec.ThisTimeLineID != replayTLI)
             ereport(PANIC,
                     (errmsg("unexpected timeline ID %u (should be %u) in end-of-recovery record",
                             xlrec.ThisTimeLineID, replayTLI)));
     }
     else if (info == XLOG_NOOP)
     {
         /* nothing to do here */
     }
     else if (info == XLOG_SWITCH)
     {
         /* nothing to do here */
     }
     else if (info == XLOG_RESTORE_POINT)
     {
         /* nothing to do here, handled in xlogrecovery.c */
     }
     else if (info == XLOG_FPI || info == XLOG_FPI_FOR_HINT)
     {
         /*
          * XLOG_FPI records contain nothing else but one or more block
          * references. Every block reference must include a full-page image
          * even if full_page_writes was disabled when the record was generated
          * - otherwise there would be no point in this record.
          *
          * XLOG_FPI_FOR_HINT records are generated when a page needs to be
          * WAL-logged because of a hint bit update. They are only generated
          * when checksums and/or wal_log_hints are enabled. They may include
          * no full-page images if full_page_writes was disabled when they were
          * generated. In this case there is nothing to do here.
          *
          * No recovery conflicts are generated by these generic records - if a
          * resource manager needs to generate conflicts, it has to define a
          * separate WAL record type and redo routine.
          */
         for (uint8 block_id = 0; block_id <= XLogRecMaxBlockId(record); block_id++)
         {
             Buffer      buffer;
  
             if (!XLogRecHasBlockImage(record, block_id))
             {
                 if (info == XLOG_FPI)
                     elog(ERROR, "XLOG_FPI record did not contain a full-page image");
                 continue;
             }
  
             if (XLogReadBufferForRedo(record, block_id, &buffer) != BLK_RESTORED)
                 elog(ERROR, "unexpected XLogReadBufferForRedo result when restoring backup block");
             UnlockReleaseBuffer(buffer);
         }
     }
     else if (info == XLOG_BACKUP_END)
     {
         /* nothing to do here, handled in xlogrecovery_redo() */
     }
     else if (info == XLOG_PARAMETER_CHANGE)
     {
         xl_parameter_change xlrec;
  
         /* Update our copy of the parameters in pg_control */
         memcpy(&xlrec, XLogRecGetData(record), sizeof(xl_parameter_change));
  
         /*
          * Invalidate logical slots if we are in hot standby and the primary
          * does not have a WAL level sufficient for logical decoding. No need
          * to search for potentially conflicting logically slots if standby is
          * running with wal_level lower than logical, because in that case, we
          * would have either disallowed creation of logical slots or
          * invalidated existing ones.
          */
         if (InRecovery && InHotStandby &&
             xlrec.wal_level < WAL_LEVEL_LOGICAL &&
             wal_level >= WAL_LEVEL_LOGICAL)
             InvalidateObsoleteReplicationSlots(RS_INVAL_WAL_LEVEL,
                                                0, InvalidOid,
                                                InvalidTransactionId);
  
         LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
         ControlFile->MaxConnections = xlrec.MaxConnections;
         ControlFile->max_worker_processes = xlrec.max_worker_processes;
         ControlFile->max_wal_senders = xlrec.max_wal_senders;
         ControlFile->max_prepared_xacts = xlrec.max_prepared_xacts;
         ControlFile->max_locks_per_xact = xlrec.max_locks_per_xact;
         ControlFile->wal_level = xlrec.wal_level;
         ControlFile->wal_log_hints = xlrec.wal_log_hints;
  
         /*
          * Update minRecoveryPoint to ensure that if recovery is aborted, we
          * recover back up to this point before allowing hot standby again.
          * This is important if the max_* settings are decreased, to ensure
          * you don't run queries against the WAL preceding the change. The
          * local copies cannot be updated as long as crash recovery is
          * happening and we expect all the WAL to be replayed.
          */
         if (InArchiveRecovery)
         {
             LocalMinRecoveryPoint = ControlFile->minRecoveryPoint;
             LocalMinRecoveryPointTLI = ControlFile->minRecoveryPointTLI;
         }
         if (LocalMinRecoveryPoint != InvalidXLogRecPtr && LocalMinRecoveryPoint < lsn)
         {
             TimeLineID  replayTLI;
  
             (void) GetCurrentReplayRecPtr(&replayTLI);
             ControlFile->minRecoveryPoint = lsn;
             ControlFile->minRecoveryPointTLI = replayTLI;
         }
  
         CommitTsParameterChange(xlrec.track_commit_timestamp,
                                 ControlFile->track_commit_timestamp);
         ControlFile->track_commit_timestamp = xlrec.track_commit_timestamp;
  
         UpdateControlFile();
         LWLockRelease(ControlFileLock);
  
         /* Check to see if any parameter change gives a problem on recovery */
         CheckRequiredParameterValues();
     }
     else if (info == XLOG_FPW_CHANGE)
     {
         bool        fpw;
  
         memcpy(&fpw, XLogRecGetData(record), sizeof(bool));
  
         /*
          * Update the LSN of the last replayed XLOG_FPW_CHANGE record so that
          * do_pg_backup_start() and do_pg_backup_stop() can check whether
          * full_page_writes has been disabled during online backup.
          */
         if (!fpw)
         {
             SpinLockAcquire(&XLogCtl->info_lck);
             if (XLogCtl->lastFpwDisableRecPtr < record->ReadRecPtr)
                 XLogCtl->lastFpwDisableRecPtr = record->ReadRecPtr;
             SpinLockRelease(&XLogCtl->info_lck);
         }
  
         /* Keep track of full_page_writes */
         lastFullPageWrites = fpw;
     }
     else if (info == XLOG_CHECKPOINT_REDO)
     {
         /* nothing to do here, just for informational purposes */
     }
 }

References ArchiveRecoveryRequested, Assert, ControlFileData::backupEndPoint, ControlFileData::backupStartPoint, BLK_RESTORED, ControlFileData::checkPointCopy, CheckRequiredParameterValues(), XLogCtlData::ckptFullXid, CommitTsParameterChange(), ControlFile, elog, XLogReaderState::EndRecPtr, ereport, errmsg(), ERROR, FullTransactionIdPrecedes, GetCurrentReplayRecPtr(), InArchiveRecovery, XLogCtlData::info_lck, InHotStandby, InRecovery, InvalidateObsoleteReplicationSlots(), InvalidOid, InvalidTransactionId, InvalidXLogRecPtr, XLogCtlData::lastFpwDisableRecPtr, lastFullPageWrites, RunningTransactionsData::latestCompletedXid, LocalMinRecoveryPoint, LocalMinRecoveryPointTLI, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), xl_parameter_change::max_locks_per_xact, ControlFileData::max_locks_per_xact, xl_parameter_change::max_prepared_xacts, ControlFileData::max_prepared_xacts, xl_parameter_change::max_wal_senders, ControlFileData::max_wal_senders, xl_parameter_change::max_worker_processes, ControlFileData::max_worker_processes, xl_parameter_change::MaxConnections, ControlFileData::MaxConnections, ControlFileData::minRecoveryPoint, ControlFileData::minRecoveryPointTLI, MultiXactAdvanceNextMXact(), MultiXactAdvanceOldest(), MultiXactSetNextMXact(), CheckPoint::nextMulti, CheckPoint::nextMultiOffset, TransamVariablesData::nextOid, CheckPoint::nextOid, TransamVariablesData::nextXid, CheckPoint::nextXid, RunningTransactionsData::nextXid, TransamVariablesData::oidCount, CheckPoint::oldestMulti, CheckPoint::oldestMultiDB, RunningTransactionsData::oldestRunningXid, TransamVariablesData::oldestXid, CheckPoint::oldestXid, CheckPoint::oldestXidDB, PANIC, PrescanPreparedTransactions(), ProcArrayApplyRecoveryInfo(), XLogReaderState::ReadRecPtr, RecoveryRestartPoint(), RS_INVAL_WAL_LEVEL, SetTransactionIdLimit(), SpinLockAcquire, SpinLockRelease, STANDBY_INITIALIZED, StandbyRecoverPreparedTransactions(), standbyState, RunningTransactionsData::subxcnt, RunningTransactionsData::subxid_status, SUBXIDS_IN_SUBTRANS, xl_end_of_recovery::ThisTimeLineID, CheckPoint::ThisTimeLineID, xl_parameter_change::track_commit_timestamp, ControlFileData::track_commit_timestamp, TransactionIdIsNormal, TransactionIdPrecedes(), TransactionIdRetreat, TransamVariables, UnlockReleaseBuffer(), UpdateControlFile(), wal_level, xl_parameter_change::wal_level, ControlFileData::wal_level, WAL_LEVEL_LOGICAL, xl_parameter_change::wal_log_hints, ControlFileData::wal_log_hints, RunningTransactionsData::xcnt, XidFromFullTransactionId, RunningTransactionsData::xids, XLOG_BACKUP_END, XLOG_CHECKPOINT_ONLINE, XLOG_CHECKPOINT_REDO, XLOG_CHECKPOINT_SHUTDOWN, XLOG_END_OF_RECOVERY, XLOG_FPI, XLOG_FPI_FOR_HINT, XLOG_FPW_CHANGE, XLOG_NEXTOID, XLOG_NOOP, XLOG_OVERWRITE_CONTRECORD, XLOG_PARAMETER_CHANGE, XLOG_RESTORE_POINT, XLOG_SWITCH, XLogCtl, XLogReadBufferForRedo(), XLogRecGetData, XLogRecGetInfo, XLogRecHasAnyBlockRefs, XLogRecHasBlockImage, XLogRecMaxBlockId, XLogRecPtrIsInvalid, and XLR_INFO_MASK.

◆ XLogBackgroundFlush()

bool XLogBackgroundFlush ( void )

Definition at line 2977 of file xlog.c.

 {
     XLogwrtRqst WriteRqst;
     bool        flexible = true;
     static TimestampTz lastflush;
     TimestampTz now;
     int         flushblocks;
     TimeLineID  insertTLI;
  
     /* XLOG doesn't need flushing during recovery */
     if (RecoveryInProgress())
         return false;
  
     /*
      * Since we're not in recovery, InsertTimeLineID is set and can't change,
      * so we can read it without a lock.
      */
     insertTLI = XLogCtl->InsertTimeLineID;
  
     /* read updated LogwrtRqst */
     SpinLockAcquire(&XLogCtl->info_lck);
     WriteRqst = XLogCtl->LogwrtRqst;
     SpinLockRelease(&XLogCtl->info_lck);
  
     /* back off to last completed page boundary */
     WriteRqst.Write -= WriteRqst.Write % XLOG_BLCKSZ;
  
     /* if we have already flushed that far, consider async commit records */
     RefreshXLogWriteResult(LogwrtResult);
     if (WriteRqst.Write <= LogwrtResult.Flush)
     {
         SpinLockAcquire(&XLogCtl->info_lck);
         WriteRqst.Write = XLogCtl->asyncXactLSN;
         SpinLockRelease(&XLogCtl->info_lck);
         flexible = false;       /* ensure it all gets written */
     }
  
     /*
      * If already known flushed, we're done. Just need to check if we are
      * holding an open file handle to a logfile that's no longer in use,
      * preventing the file from being deleted.
      */
     if (WriteRqst.Write <= LogwrtResult.Flush)
     {
         if (openLogFile >= 0)
         {
             if (!XLByteInPrevSeg(LogwrtResult.Write, openLogSegNo,
                                  wal_segment_size))
             {
                 XLogFileClose();
             }
         }
         return false;
     }
  
     /*
      * Determine how far to flush WAL, based on the wal_writer_delay and
      * wal_writer_flush_after GUCs.
      *
      * Note that XLogSetAsyncXactLSN() performs similar calculation based on
      * wal_writer_flush_after, to decide when to wake us up.  Make sure the
      * logic is the same in both places if you change this.
      */
     now = GetCurrentTimestamp();
     flushblocks =
         WriteRqst.Write / XLOG_BLCKSZ - LogwrtResult.Flush / XLOG_BLCKSZ;
  
     if (WalWriterFlushAfter == 0 || lastflush == 0)
     {
         /* first call, or block based limits disabled */
         WriteRqst.Flush = WriteRqst.Write;
         lastflush = now;
     }
     else if (TimestampDifferenceExceeds(lastflush, now, WalWriterDelay))
     {
         /*
          * Flush the writes at least every WalWriterDelay ms. This is
          * important to bound the amount of time it takes for an asynchronous
          * commit to hit disk.
          */
         WriteRqst.Flush = WriteRqst.Write;
         lastflush = now;
     }
     else if (flushblocks >= WalWriterFlushAfter)
     {
         /* exceeded wal_writer_flush_after blocks, flush */
         WriteRqst.Flush = WriteRqst.Write;
         lastflush = now;
     }
     else
     {
         /* no flushing, this time round */
         WriteRqst.Flush = 0;
     }
  
 #ifdef WAL_DEBUG
     if (XLOG_DEBUG)
         elog(LOG, "xlog bg flush request write %X/%X; flush: %X/%X, current is write %X/%X; flush %X/%X",
              LSN_FORMAT_ARGS(WriteRqst.Write),
              LSN_FORMAT_ARGS(WriteRqst.Flush),
              LSN_FORMAT_ARGS(LogwrtResult.Write),
              LSN_FORMAT_ARGS(LogwrtResult.Flush));
 #endif
  
     START_CRIT_SECTION();
  
     /* now wait for any in-progress insertions to finish and get write lock */
     WaitXLogInsertionsToFinish(WriteRqst.Write);
     LWLockAcquire(WALWriteLock, LW_EXCLUSIVE);
     RefreshXLogWriteResult(LogwrtResult);
     if (WriteRqst.Write > LogwrtResult.Write ||
         WriteRqst.Flush > LogwrtResult.Flush)
     {
         XLogWrite(WriteRqst, insertTLI, flexible);
     }
     LWLockRelease(WALWriteLock);
  
     END_CRIT_SECTION();
  
     /* wake up walsenders now that we've released heavily contended locks */
     WalSndWakeupProcessRequests(true, !RecoveryInProgress());
  
     /*
      * Great, done. To take some work off the critical path, try to initialize
      * as many of the no-longer-needed WAL buffers for future use as we can.
      */
     AdvanceXLInsertBuffer(InvalidXLogRecPtr, insertTLI, true);
  
     /*
      * If we determined that we need to write data, but somebody else
      * wrote/flushed already, it should be considered as being active, to
      * avoid hibernating too early.
      */
     return true;
 }

References AdvanceXLInsertBuffer(), XLogCtlData::asyncXactLSN, elog, END_CRIT_SECTION, XLogwrtRqst::Flush, XLogwrtResult::Flush, GetCurrentTimestamp(), XLogCtlData::info_lck, XLogCtlData::InsertTimeLineID, InvalidXLogRecPtr, LOG, LogwrtResult, XLogCtlData::LogwrtRqst, LSN_FORMAT_ARGS, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), now(), openLogFile, openLogSegNo, RecoveryInProgress(), RefreshXLogWriteResult, SpinLockAcquire, SpinLockRelease, START_CRIT_SECTION, TimestampDifferenceExceeds(), WaitXLogInsertionsToFinish(), wal_segment_size, WalSndWakeupProcessRequests(), WalWriterDelay, WalWriterFlushAfter, XLogwrtRqst::Write, XLogwrtResult::Write, XLByteInPrevSeg, XLogCtl, XLogFileClose(), and XLogWrite().

Referenced by WalSndWaitForWal(), and WalWriterMain().

◆ XLogBytePosToEndRecPtr()

static XLogRecPtr XLogBytePosToEndRecPtr ( uint64 bytepos )

static

Definition at line 1891 of file xlog.c.

 {
     uint64      fullsegs;
     uint64      fullpages;
     uint64      bytesleft;
     uint32      seg_offset;
     XLogRecPtr  result;
  
     fullsegs = bytepos / UsableBytesInSegment;
     bytesleft = bytepos % UsableBytesInSegment;
  
     if (bytesleft < XLOG_BLCKSZ - SizeOfXLogLongPHD)
     {
         /* fits on first page of segment */
         if (bytesleft == 0)
             seg_offset = 0;
         else
             seg_offset = bytesleft + SizeOfXLogLongPHD;
     }
     else
     {
         /* account for the first page on segment with long header */
         seg_offset = XLOG_BLCKSZ;
         bytesleft -= XLOG_BLCKSZ - SizeOfXLogLongPHD;
  
         fullpages = bytesleft / UsableBytesInPage;
         bytesleft = bytesleft % UsableBytesInPage;
  
         if (bytesleft == 0)
             seg_offset += fullpages * XLOG_BLCKSZ + bytesleft;
         else
             seg_offset += fullpages * XLOG_BLCKSZ + bytesleft + SizeOfXLogShortPHD;
     }
  
     XLogSegNoOffsetToRecPtr(fullsegs, seg_offset, wal_segment_size, result);
  
     return result;
 }

References SizeOfXLogLongPHD, SizeOfXLogShortPHD, UsableBytesInPage, UsableBytesInSegment, wal_segment_size, and XLogSegNoOffsetToRecPtr.

Referenced by ReserveXLogInsertLocation(), ReserveXLogSwitch(), and WaitXLogInsertionsToFinish().

◆ XLogBytePosToRecPtr()

static XLogRecPtr XLogBytePosToRecPtr ( uint64 bytepos )

static

Definition at line 1851 of file xlog.c.

 {
     uint64      fullsegs;
     uint64      fullpages;
     uint64      bytesleft;
     uint32      seg_offset;
     XLogRecPtr  result;
  
     fullsegs = bytepos / UsableBytesInSegment;
     bytesleft = bytepos % UsableBytesInSegment;
  
     if (bytesleft < XLOG_BLCKSZ - SizeOfXLogLongPHD)
     {
         /* fits on first page of segment */
         seg_offset = bytesleft + SizeOfXLogLongPHD;
     }
     else
     {
         /* account for the first page on segment with long header */
         seg_offset = XLOG_BLCKSZ;
         bytesleft -= XLOG_BLCKSZ - SizeOfXLogLongPHD;
  
         fullpages = bytesleft / UsableBytesInPage;
         bytesleft = bytesleft % UsableBytesInPage;
  
         seg_offset += fullpages * XLOG_BLCKSZ + bytesleft + SizeOfXLogShortPHD;
     }
  
     XLogSegNoOffsetToRecPtr(fullsegs, seg_offset, wal_segment_size, result);
  
     return result;
 }

References SizeOfXLogLongPHD, SizeOfXLogShortPHD, UsableBytesInPage, UsableBytesInSegment, wal_segment_size, and XLogSegNoOffsetToRecPtr.

Referenced by CreateCheckPoint(), GetXLogInsertRecPtr(), ReserveXLogInsertLocation(), and ReserveXLogSwitch().

◆ XLogCheckpointNeeded()

bool XLogCheckpointNeeded ( XLogSegNo new_segno )

Definition at line 2283 of file xlog.c.

 {
     XLogSegNo   old_segno;
  
     XLByteToSeg(RedoRecPtr, old_segno, wal_segment_size);
  
     if (new_segno >= old_segno + (uint64) (CheckPointSegments - 1))
         return true;
     return false;
 }

References CheckPointSegments, RedoRecPtr, wal_segment_size, and XLByteToSeg.

Referenced by XLogPageRead(), and XLogWrite().

◆ XLOGChooseNumBuffers()

static int XLOGChooseNumBuffers ( void )

static

Definition at line 4586 of file xlog.c.

 {
     int         xbuffers;
  
     xbuffers = NBuffers / 32;
     if (xbuffers > (wal_segment_size / XLOG_BLCKSZ))
         xbuffers = (wal_segment_size / XLOG_BLCKSZ);
     if (xbuffers < 8)
         xbuffers = 8;
     return xbuffers;
 }

References NBuffers, and wal_segment_size.

Referenced by check_wal_buffers(), and XLOGShmemSize().

◆ XLogFileClose()

static void XLogFileClose ( void )

static

Definition at line 3626 of file xlog.c.

 {
     Assert(openLogFile >= 0);
  
     /*
      * WAL segment files will not be re-read in normal operation, so we advise
      * the OS to release any cached pages.  But do not do so if WAL archiving
      * or streaming is active, because archiver and walsender process could
      * use the cache to read the WAL segment.
      */
 #if defined(USE_POSIX_FADVISE) && defined(POSIX_FADV_DONTNEED)
     if (!XLogIsNeeded() && (io_direct_flags & IO_DIRECT_WAL) == 0)
         (void) posix_fadvise(openLogFile, 0, 0, POSIX_FADV_DONTNEED);
 #endif
  
     if (close(openLogFile) != 0)
     {
         char        xlogfname[MAXFNAMELEN];
         int         save_errno = errno;
  
         XLogFileName(xlogfname, openLogTLI, openLogSegNo, wal_segment_size);
         errno = save_errno;
         ereport(PANIC,
                 (errcode_for_file_access(),
                  errmsg("could not close file \"%s\": %m", xlogfname)));
     }
  
     openLogFile = -1;
     ReleaseExternalFD();
 }

References Assert, close, ereport, errcode_for_file_access(), errmsg(), io_direct_flags, IO_DIRECT_WAL, MAXFNAMELEN, openLogFile, openLogSegNo, openLogTLI, PANIC, ReleaseExternalFD(), wal_segment_size, XLogFileName(), and XLogIsNeeded.

Referenced by assign_wal_sync_method(), XLogBackgroundFlush(), and XLogWrite().

◆ XLogFileCopy()

static void XLogFileCopy	(	TimeLineID	destTLI,
		XLogSegNo	destsegno,
		TimeLineID	srcTLI,
		XLogSegNo	srcsegno,
		int	upto
	)

static

Definition at line 3405 of file xlog.c.

 {
     char        path[MAXPGPATH];
     char        tmppath[MAXPGPATH];
     PGAlignedXLogBlock buffer;
     int         srcfd;
     int         fd;
     int         nbytes;
  
     /*
      * Open the source file
      */
     XLogFilePath(path, srcTLI, srcsegno, wal_segment_size);
     srcfd = OpenTransientFile(path, O_RDONLY | PG_BINARY);
     if (srcfd < 0)
         ereport(ERROR,
                 (errcode_for_file_access(),
                  errmsg("could not open file \"%s\": %m", path)));
  
     /*
      * Copy into a temp file name.
      */
     snprintf(tmppath, MAXPGPATH, XLOGDIR "/xlogtemp.%d", (int) getpid());
  
     unlink(tmppath);
  
     /* do not use get_sync_bit() here --- want to fsync only at end of fill */
     fd = OpenTransientFile(tmppath, O_RDWR | O_CREAT | O_EXCL | PG_BINARY);
     if (fd < 0)
         ereport(ERROR,
                 (errcode_for_file_access(),
                  errmsg("could not create file \"%s\": %m", tmppath)));
  
     /*
      * Do the data copying.
      */
     for (nbytes = 0; nbytes < wal_segment_size; nbytes += sizeof(buffer))
     {
         int         nread;
  
         nread = upto - nbytes;
  
         /*
          * The part that is not read from the source file is filled with
          * zeros.
          */
         if (nread < sizeof(buffer))
             memset(buffer.data, 0, sizeof(buffer));
  
         if (nread > 0)
         {
             int         r;
  
             if (nread > sizeof(buffer))
                 nread = sizeof(buffer);
             pgstat_report_wait_start(WAIT_EVENT_WAL_COPY_READ);
             r = read(srcfd, buffer.data, nread);
             if (r != nread)
             {
                 if (r < 0)
                     ereport(ERROR,
                             (errcode_for_file_access(),
                              errmsg("could not read file \"%s\": %m",
                                     path)));
                 else
                     ereport(ERROR,
                             (errcode(ERRCODE_DATA_CORRUPTED),
                              errmsg("could not read file \"%s\": read %d of %zu",
                                     path, r, (Size) nread)));
             }
             pgstat_report_wait_end();
         }
         errno = 0;
         pgstat_report_wait_start(WAIT_EVENT_WAL_COPY_WRITE);
         if ((int) write(fd, buffer.data, sizeof(buffer)) != (int) sizeof(buffer))
         {
             int         save_errno = errno;
  
             /*
              * If we fail to make the file, delete it to release disk space
              */
             unlink(tmppath);
             /* if write didn't set errno, assume problem is no disk space */
             errno = save_errno ? save_errno : ENOSPC;
  
             ereport(ERROR,
                     (errcode_for_file_access(),
                      errmsg("could not write to file \"%s\": %m", tmppath)));
         }
         pgstat_report_wait_end();
     }
  
     pgstat_report_wait_start(WAIT_EVENT_WAL_COPY_SYNC);
     if (pg_fsync(fd) != 0)
         ereport(data_sync_elevel(ERROR),
                 (errcode_for_file_access(),
                  errmsg("could not fsync file \"%s\": %m", tmppath)));
     pgstat_report_wait_end();
  
     if (CloseTransientFile(fd) != 0)
         ereport(ERROR,
                 (errcode_for_file_access(),
                  errmsg("could not close file \"%s\": %m", tmppath)));
  
     if (CloseTransientFile(srcfd) != 0)
         ereport(ERROR,
                 (errcode_for_file_access(),
                  errmsg("could not close file \"%s\": %m", path)));
  
     /*
      * Now move the segment into place with its final name.
      */
     if (!InstallXLogFileSegment(&destsegno, tmppath, false, 0, destTLI))
         elog(ERROR, "InstallXLogFileSegment should not have failed");
 }

References CloseTransientFile(), PGAlignedXLogBlock::data, data_sync_elevel(), elog, ereport, errcode(), ERRCODE_DATA_CORRUPTED, errcode_for_file_access(), errmsg(), ERROR, fd(), InstallXLogFileSegment(), MAXPGPATH, OpenTransientFile(), PG_BINARY, pg_fsync(), pgstat_report_wait_end(), pgstat_report_wait_start(), read, snprintf, wal_segment_size, write, XLOGDIR, and XLogFilePath().

Referenced by XLogInitNewTimeline().

◆ XLogFileInit()

int XLogFileInit	(	XLogSegNo	logsegno,
		TimeLineID	logtli
	)

Definition at line 3367 of file xlog.c.

 {
     bool        ignore_added;
     char        path[MAXPGPATH];
     int         fd;
  
     Assert(logtli != 0);
  
     fd = XLogFileInitInternal(logsegno, logtli, &ignore_added, path);
     if (fd >= 0)
         return fd;
  
     /* Now open original target segment (might not be file I just made) */
     fd = BasicOpenFile(path, O_RDWR | PG_BINARY | O_CLOEXEC |
                        get_sync_bit(wal_sync_method));
     if (fd < 0)
         ereport(ERROR,
                 (errcode_for_file_access(),
                  errmsg("could not open file \"%s\": %m", path)));
     return fd;
 }

References Assert, BasicOpenFile(), ereport, errcode_for_file_access(), errmsg(), ERROR, fd(), get_sync_bit(), MAXPGPATH, O_CLOEXEC, PG_BINARY, wal_sync_method, and XLogFileInitInternal().

Referenced by BootStrapXLOG(), XLogInitNewTimeline(), XLogWalRcvWrite(), and XLogWrite().

◆ XLogFileInitInternal()

static int XLogFileInitInternal	(	XLogSegNo	logsegno,
		TimeLineID	logtli,
		bool *	added,
		char *	path
	)

static

Definition at line 3197 of file xlog.c.

 {
     char        tmppath[MAXPGPATH];
     XLogSegNo   installed_segno;
     XLogSegNo   max_segno;
     int         fd;
     int         save_errno;
     int         open_flags = O_RDWR | O_CREAT | O_EXCL | PG_BINARY;
  
     Assert(logtli != 0);
  
     XLogFilePath(path, logtli, logsegno, wal_segment_size);
  
     /*
      * Try to use existent file (checkpoint maker may have created it already)
      */
     *added = false;
     fd = BasicOpenFile(path, O_RDWR | PG_BINARY | O_CLOEXEC |
                        get_sync_bit(wal_sync_method));
     if (fd < 0)
     {
         if (errno != ENOENT)
             ereport(ERROR,
                     (errcode_for_file_access(),
                      errmsg("could not open file \"%s\": %m", path)));
     }
     else
         return fd;
  
     /*
      * Initialize an empty (all zeroes) segment.  NOTE: it is possible that
      * another process is doing the same thing.  If so, we will end up
      * pre-creating an extra log segment.  That seems OK, and better than
      * holding the lock throughout this lengthy process.
      */
     elog(DEBUG2, "creating and filling new WAL file");
  
     snprintf(tmppath, MAXPGPATH, XLOGDIR "/xlogtemp.%d", (int) getpid());
  
     unlink(tmppath);
  
     if (io_direct_flags & IO_DIRECT_WAL_INIT)
         open_flags |= PG_O_DIRECT;
  
     /* do not use get_sync_bit() here --- want to fsync only at end of fill */
     fd = BasicOpenFile(tmppath, open_flags);
     if (fd < 0)
         ereport(ERROR,
                 (errcode_for_file_access(),
                  errmsg("could not create file \"%s\": %m", tmppath)));
  
     pgstat_report_wait_start(WAIT_EVENT_WAL_INIT_WRITE);
     save_errno = 0;
     if (wal_init_zero)
     {
         ssize_t     rc;
  
         /*
          * Zero-fill the file.  With this setting, we do this the hard way to
          * ensure that all the file space has really been allocated.  On
          * platforms that allow "holes" in files, just seeking to the end
          * doesn't allocate intermediate space.  This way, we know that we
          * have all the space and (after the fsync below) that all the
          * indirect blocks are down on disk.  Therefore, fdatasync(2) or
          * O_DSYNC will be sufficient to sync future writes to the log file.
          */
         rc = pg_pwrite_zeros(fd, wal_segment_size, 0);
  
         if (rc < 0)
             save_errno = errno;
     }
     else
     {
         /*
          * Otherwise, seeking to the end and writing a solitary byte is
          * enough.
          */
         errno = 0;
         if (pg_pwrite(fd, "\0", 1, wal_segment_size - 1) != 1)
         {
             /* if write didn't set errno, assume no disk space */
             save_errno = errno ? errno : ENOSPC;
         }
     }
     pgstat_report_wait_end();
  
     if (save_errno)
     {
         /*
          * If we fail to make the file, delete it to release disk space
          */
         unlink(tmppath);
  
         close(fd);
  
         errno = save_errno;
  
         ereport(ERROR,
                 (errcode_for_file_access(),
                  errmsg("could not write to file \"%s\": %m", tmppath)));
     }
  
     pgstat_report_wait_start(WAIT_EVENT_WAL_INIT_SYNC);
     if (pg_fsync(fd) != 0)
     {
         save_errno = errno;
         close(fd);
         errno = save_errno;
         ereport(ERROR,
                 (errcode_for_file_access(),
                  errmsg("could not fsync file \"%s\": %m", tmppath)));
     }
     pgstat_report_wait_end();
  
     if (close(fd) != 0)
         ereport(ERROR,
                 (errcode_for_file_access(),
                  errmsg("could not close file \"%s\": %m", tmppath)));
  
     /*
      * Now move the segment into place with its final name.  Cope with
      * possibility that someone else has created the file while we were
      * filling ours: if so, use ours to pre-create a future log segment.
      */
     installed_segno = logsegno;
  
     /*
      * XXX: What should we use as max_segno? We used to use XLOGfileslop when
      * that was a constant, but that was always a bit dubious: normally, at a
      * checkpoint, XLOGfileslop was the offset from the checkpoint record, but
      * here, it was the offset from the insert location. We can't do the
      * normal XLOGfileslop calculation here because we don't have access to
      * the prior checkpoint's redo location. So somewhat arbitrarily, just use
      * CheckPointSegments.
      */
     max_segno = logsegno + CheckPointSegments;
     if (InstallXLogFileSegment(&installed_segno, tmppath, true, max_segno,
                                logtli))
     {
         *added = true;
         elog(DEBUG2, "done creating and filling new WAL file");
     }
     else
     {
         /*
          * No need for any more future segments, or InstallXLogFileSegment()
          * failed to rename the file into place. If the rename failed, a
          * caller opening the file may fail.
          */
         unlink(tmppath);
         elog(DEBUG2, "abandoned new WAL file");
     }
  
     return -1;
 }

References Assert, BasicOpenFile(), CheckPointSegments, close, DEBUG2, elog, ereport, errcode_for_file_access(), errmsg(), ERROR, fd(), get_sync_bit(), InstallXLogFileSegment(), io_direct_flags, IO_DIRECT_WAL_INIT, MAXPGPATH, O_CLOEXEC, PG_BINARY, pg_fsync(), PG_O_DIRECT, pg_pwrite, pg_pwrite_zeros(), pgstat_report_wait_end(), pgstat_report_wait_start(), snprintf, wal_init_zero, wal_segment_size, wal_sync_method, XLOGDIR, and XLogFilePath().

Referenced by PreallocXlogFiles(), and XLogFileInit().

◆ XLogFileOpen()

int XLogFileOpen	(	XLogSegNo	segno,
		TimeLineID	tli
	)

Definition at line 3605 of file xlog.c.

 {
     char        path[MAXPGPATH];
     int         fd;
  
     XLogFilePath(path, tli, segno, wal_segment_size);
  
     fd = BasicOpenFile(path, O_RDWR | PG_BINARY | O_CLOEXEC |
                        get_sync_bit(wal_sync_method));
     if (fd < 0)
         ereport(PANIC,
                 (errcode_for_file_access(),
                  errmsg("could not open file \"%s\": %m", path)));
  
     return fd;
 }

References BasicOpenFile(), ereport, errcode_for_file_access(), errmsg(), fd(), get_sync_bit(), MAXPGPATH, O_CLOEXEC, PANIC, PG_BINARY, wal_segment_size, wal_sync_method, and XLogFilePath().

Referenced by XLogWrite().

◆ XLOGfileslop()

static XLogSegNo XLOGfileslop ( XLogRecPtr lastredoptr )

static

Definition at line 2233 of file xlog.c.

 {
     XLogSegNo   minSegNo;
     XLogSegNo   maxSegNo;
     double      distance;
     XLogSegNo   recycleSegNo;
  
     /*
      * Calculate the segment numbers that min_wal_size_mb and max_wal_size_mb
      * correspond to. Always recycle enough segments to meet the minimum, and
      * remove enough segments to stay below the maximum.
      */
     minSegNo = lastredoptr / wal_segment_size +
         ConvertToXSegs(min_wal_size_mb, wal_segment_size) - 1;
     maxSegNo = lastredoptr / wal_segment_size +
         ConvertToXSegs(max_wal_size_mb, wal_segment_size) - 1;
  
     /*
      * Between those limits, recycle enough segments to get us through to the
      * estimated end of next checkpoint.
      *
      * To estimate where the next checkpoint will finish, assume that the
      * system runs steadily consuming CheckPointDistanceEstimate bytes between
      * every checkpoint.
      */
     distance = (1.0 + CheckPointCompletionTarget) * CheckPointDistanceEstimate;
     /* add 10% for good measure. */
     distance *= 1.10;
  
     recycleSegNo = (XLogSegNo) ceil(((double) lastredoptr + distance) /
                                     wal_segment_size);
  
     if (recycleSegNo < minSegNo)
         recycleSegNo = minSegNo;
     if (recycleSegNo > maxSegNo)
         recycleSegNo = maxSegNo;
  
     return recycleSegNo;
 }

References CheckPointCompletionTarget, CheckPointDistanceEstimate, ConvertToXSegs, max_wal_size_mb, min_wal_size_mb, and wal_segment_size.

Referenced by RemoveOldXlogFiles().

◆ XLogFlush()

void XLogFlush ( XLogRecPtr record )

Definition at line 2789 of file xlog.c.

 {
     XLogRecPtr  WriteRqstPtr;
     XLogwrtRqst WriteRqst;
     TimeLineID  insertTLI = XLogCtl->InsertTimeLineID;
  
     /*
      * During REDO, we are reading not writing WAL.  Therefore, instead of
      * trying to flush the WAL, we should update minRecoveryPoint instead. We
      * test XLogInsertAllowed(), not InRecovery, because we need checkpointer
      * to act this way too, and because when it tries to write the
      * end-of-recovery checkpoint, it should indeed flush.
      */
     if (!XLogInsertAllowed())
     {
         UpdateMinRecoveryPoint(record, false);
         return;
     }
  
     /* Quick exit if already known flushed */
     if (record <= LogwrtResult.Flush)
         return;
  
 #ifdef WAL_DEBUG
     if (XLOG_DEBUG)
         elog(LOG, "xlog flush request %X/%X; write %X/%X; flush %X/%X",
              LSN_FORMAT_ARGS(record),
              LSN_FORMAT_ARGS(LogwrtResult.Write),
              LSN_FORMAT_ARGS(LogwrtResult.Flush));
 #endif
  
     START_CRIT_SECTION();
  
     /*
      * Since fsync is usually a horribly expensive operation, we try to
      * piggyback as much data as we can on each fsync: if we see any more data
      * entered into the xlog buffer, we'll write and fsync that too, so that
      * the final value of LogwrtResult.Flush is as large as possible. This
      * gives us some chance of avoiding another fsync immediately after.
      */
  
     /* initialize to given target; may increase below */
     WriteRqstPtr = record;
  
     /*
      * Now wait until we get the write lock, or someone else does the flush
      * for us.
      */
     for (;;)
     {
         XLogRecPtr  insertpos;
  
         /* done already? */
         RefreshXLogWriteResult(LogwrtResult);
         if (record <= LogwrtResult.Flush)
             break;
  
         /*
          * Before actually performing the write, wait for all in-flight
          * insertions to the pages we're about to write to finish.
          */
         SpinLockAcquire(&XLogCtl->info_lck);
         if (WriteRqstPtr < XLogCtl->LogwrtRqst.Write)
             WriteRqstPtr = XLogCtl->LogwrtRqst.Write;
         SpinLockRelease(&XLogCtl->info_lck);
         insertpos = WaitXLogInsertionsToFinish(WriteRqstPtr);
  
         /*
          * Try to get the write lock. If we can't get it immediately, wait
          * until it's released, and recheck if we still need to do the flush
          * or if the backend that held the lock did it for us already. This
          * helps to maintain a good rate of group committing when the system
          * is bottlenecked by the speed of fsyncing.
          */
         if (!LWLockAcquireOrWait(WALWriteLock, LW_EXCLUSIVE))
         {
             /*
              * The lock is now free, but we didn't acquire it yet. Before we
              * do, loop back to check if someone else flushed the record for
              * us already.
              */
             continue;
         }
  
         /* Got the lock; recheck whether request is satisfied */
         RefreshXLogWriteResult(LogwrtResult);
         if (record <= LogwrtResult.Flush)
         {
             LWLockRelease(WALWriteLock);
             break;
         }
  
         /*
          * Sleep before flush! By adding a delay here, we may give further
          * backends the opportunity to join the backlog of group commit
          * followers; this can significantly improve transaction throughput,
          * at the risk of increasing transaction latency.
          *
          * We do not sleep if enableFsync is not turned on, nor if there are
          * fewer than CommitSiblings other backends with active transactions.
          */
         if (CommitDelay > 0 && enableFsync &&
             MinimumActiveBackends(CommitSiblings))
         {
             pg_usleep(CommitDelay);
  
             /*
              * Re-check how far we can now flush the WAL. It's generally not
              * safe to call WaitXLogInsertionsToFinish while holding
              * WALWriteLock, because an in-progress insertion might need to
              * also grab WALWriteLock to make progress. But we know that all
              * the insertions up to insertpos have already finished, because
              * that's what the earlier WaitXLogInsertionsToFinish() returned.
              * We're only calling it again to allow insertpos to be moved
              * further forward, not to actually wait for anyone.
              */
             insertpos = WaitXLogInsertionsToFinish(insertpos);
         }
  
         /* try to write/flush later additions to XLOG as well */
         WriteRqst.Write = insertpos;
         WriteRqst.Flush = insertpos;
  
         XLogWrite(WriteRqst, insertTLI, false);
  
         LWLockRelease(WALWriteLock);
         /* done */
         break;
     }
  
     END_CRIT_SECTION();
  
     /* wake up walsenders now that we've released heavily contended locks */
     WalSndWakeupProcessRequests(true, !RecoveryInProgress());
  
     /*
      * If we still haven't flushed to the request point then we have a
      * problem; most likely, the requested flush point is past end of XLOG.
      * This has been seen to occur when a disk page has a corrupted LSN.
      *
      * Formerly we treated this as a PANIC condition, but that hurts the
      * system's robustness rather than helping it: we do not want to take down
      * the whole system due to corruption on one data page.  In particular, if
      * the bad page is encountered again during recovery then we would be
      * unable to restart the database at all!  (This scenario actually
      * happened in the field several times with 7.1 releases.)  As of 8.4, bad
      * LSNs encountered during recovery are UpdateMinRecoveryPoint's problem;
      * the only time we can reach here during recovery is while flushing the
      * end-of-recovery checkpoint record, and we don't expect that to have a
      * bad LSN.
      *
      * Note that for calls from xact.c, the ERROR will be promoted to PANIC
      * since xact.c calls this routine inside a critical section.  However,
      * calls from bufmgr.c are not within critical sections and so we will not
      * force a restart for a bad LSN on a data page.
      */
     if (LogwrtResult.Flush < record)
         elog(ERROR,
              "xlog flush request %X/%X is not satisfied --- flushed only to %X/%X",
              LSN_FORMAT_ARGS(record),
              LSN_FORMAT_ARGS(LogwrtResult.Flush));
 }

References CommitDelay, CommitSiblings, elog, enableFsync, END_CRIT_SECTION, ERROR, XLogwrtRqst::Flush, XLogwrtResult::Flush, XLogCtlData::info_lck, XLogCtlData::InsertTimeLineID, LOG, LogwrtResult, XLogCtlData::LogwrtRqst, LSN_FORMAT_ARGS, LW_EXCLUSIVE, LWLockAcquireOrWait(), LWLockRelease(), MinimumActiveBackends(), pg_usleep(), RecoveryInProgress(), RefreshXLogWriteResult, SpinLockAcquire, SpinLockRelease, START_CRIT_SECTION, UpdateMinRecoveryPoint(), WaitXLogInsertionsToFinish(), WalSndWakeupProcessRequests(), XLogwrtRqst::Write, XLogwrtResult::Write, XLogCtl, XLogInsertAllowed(), and XLogWrite().

Referenced by CheckPointReplicationOrigin(), CreateCheckPoint(), CreateEndOfRecoveryRecord(), CreateOverwriteContrecordRecord(), dropdb(), EndPrepare(), FlushBuffer(), LogLogicalMessage(), pg_attribute_noreturn(), RecordTransactionAbortPrepared(), RecordTransactionCommit(), RecordTransactionCommitPrepared(), RelationTruncate(), ReplicationSlotReserveWal(), replorigin_get_progress(), replorigin_session_get_progress(), SlruPhysicalWritePage(), smgr_redo(), write_relmap_file(), WriteMTruncateXlogRec(), WriteTruncateXlogRec(), xact_redo_abort(), xact_redo_commit(), XLogInsertRecord(), and XLogReportParameters().

◆ XLogGetLastRemovedSegno()

XLogSegNo XLogGetLastRemovedSegno ( void )

Definition at line 3745 of file xlog.c.

 {
     XLogSegNo   lastRemovedSegNo;
  
     SpinLockAcquire(&XLogCtl->info_lck);
     lastRemovedSegNo = XLogCtl->lastRemovedSegNo;
     SpinLockRelease(&XLogCtl->info_lck);
  
     return lastRemovedSegNo;
 }

References XLogCtlData::info_lck, XLogCtlData::lastRemovedSegNo, SpinLockAcquire, SpinLockRelease, and XLogCtl.

Referenced by copy_replication_slot(), GetWALAvailability(), ReplicationSlotReserveWal(), and reserve_wal_for_local_slot().

◆ XLogGetOldestSegno()

XLogSegNo XLogGetOldestSegno ( TimeLineID tli )

Definition at line 3761 of file xlog.c.

 {
     DIR        *xldir;
     struct dirent *xlde;
     XLogSegNo   oldest_segno = 0;
  
     xldir = AllocateDir(XLOGDIR);
     while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL)
     {
         TimeLineID  file_tli;
         XLogSegNo   file_segno;
  
         /* Ignore files that are not XLOG segments. */
         if (!IsXLogFileName(xlde->d_name))
             continue;
  
         /* Parse filename to get TLI and segno. */
         XLogFromFileName(xlde->d_name, &file_tli, &file_segno,
                          wal_segment_size);
  
         /* Ignore anything that's not from the TLI of interest. */
         if (tli != file_tli)
             continue;
  
         /* If it's the oldest so far, update oldest_segno. */
         if (oldest_segno == 0 || file_segno < oldest_segno)
             oldest_segno = file_segno;
     }
  
     FreeDir(xldir);
     return oldest_segno;
 }

References AllocateDir(), dirent::d_name, FreeDir(), IsXLogFileName(), ReadDir(), wal_segment_size, XLOGDIR, and XLogFromFileName().

Referenced by GetOldestUnsummarizedLSN(), MaybeRemoveOldWalSummaries(), and reserve_wal_for_local_slot().

◆ XLogGetReplicationSlotMinimumLSN()

static XLogRecPtr XLogGetReplicationSlotMinimumLSN ( void )

static

Definition at line 2688 of file xlog.c.

 {
     XLogRecPtr  retval;
  
     SpinLockAcquire(&XLogCtl->info_lck);
     retval = XLogCtl->replicationSlotMinLSN;
     SpinLockRelease(&XLogCtl->info_lck);
  
     return retval;
 }

References XLogCtlData::info_lck, XLogCtlData::replicationSlotMinLSN, SpinLockAcquire, SpinLockRelease, and XLogCtl.

Referenced by KeepLogSeg().

◆ XLogInitNewTimeline()

static void XLogInitNewTimeline	(	TimeLineID	endTLI,
		XLogRecPtr	endOfLog,
		TimeLineID	newTLI
	)

static

Definition at line 5179 of file xlog.c.

 {
     char        xlogfname[MAXFNAMELEN];
     XLogSegNo   endLogSegNo;
     XLogSegNo   startLogSegNo;
  
     /* we always switch to a new timeline after archive recovery */
     Assert(endTLI != newTLI);
  
     /*
      * Update min recovery point one last time.
      */
     UpdateMinRecoveryPoint(InvalidXLogRecPtr, true);
  
     /*
      * Calculate the last segment on the old timeline, and the first segment
      * on the new timeline. If the switch happens in the middle of a segment,
      * they are the same, but if the switch happens exactly at a segment
      * boundary, startLogSegNo will be endLogSegNo + 1.
      */
     XLByteToPrevSeg(endOfLog, endLogSegNo, wal_segment_size);
     XLByteToSeg(endOfLog, startLogSegNo, wal_segment_size);
  
     /*
      * Initialize the starting WAL segment for the new timeline. If the switch
      * happens in the middle of a segment, copy data from the last WAL segment
      * of the old timeline up to the switch point, to the starting WAL segment
      * on the new timeline.
      */
     if (endLogSegNo == startLogSegNo)
     {
         /*
          * Make a copy of the file on the new timeline.
          *
          * Writing WAL isn't allowed yet, so there are no locking
          * considerations. But we should be just as tense as XLogFileInit to
          * avoid emplacing a bogus file.
          */
         XLogFileCopy(newTLI, endLogSegNo, endTLI, endLogSegNo,
                      XLogSegmentOffset(endOfLog, wal_segment_size));
     }
     else
     {
         /*
          * The switch happened at a segment boundary, so just create the next
          * segment on the new timeline.
          */
         int         fd;
  
         fd = XLogFileInit(startLogSegNo, newTLI);
  
         if (close(fd) != 0)
         {
             int         save_errno = errno;
  
             XLogFileName(xlogfname, newTLI, startLogSegNo, wal_segment_size);
             errno = save_errno;
             ereport(ERROR,
                     (errcode_for_file_access(),
                      errmsg("could not close file \"%s\": %m", xlogfname)));
         }
     }
  
     /*
      * Let's just make real sure there are not .ready or .done flags posted
      * for the new segment.
      */
     XLogFileName(xlogfname, newTLI, startLogSegNo, wal_segment_size);
     XLogArchiveCleanup(xlogfname);
 }

References Assert, close, ereport, errcode_for_file_access(), errmsg(), ERROR, fd(), InvalidXLogRecPtr, MAXFNAMELEN, UpdateMinRecoveryPoint(), wal_segment_size, XLByteToPrevSeg, XLByteToSeg, XLogArchiveCleanup(), XLogFileCopy(), XLogFileInit(), XLogFileName(), and XLogSegmentOffset.

Referenced by StartupXLOG().

◆ XLogInsertAllowed()

bool XLogInsertAllowed ( void )

Definition at line 6345 of file xlog.c.

 {
     /*
      * If value is "unconditionally true" or "unconditionally false", just
      * return it.  This provides the normal fast path once recovery is known
      * done.
      */
     if (LocalXLogInsertAllowed >= 0)
         return (bool) LocalXLogInsertAllowed;
  
     /*
      * Else, must check to see if we're still in recovery.
      */
     if (RecoveryInProgress())
         return false;
  
     /*
      * On exit from recovery, reset to "unconditionally true", since there is
      * no need to keep checking.
      */
     LocalXLogInsertAllowed = 1;
     return true;
 }

References LocalXLogInsertAllowed, and RecoveryInProgress().

Referenced by XLogBeginInsert(), XLogFlush(), and XLogInsertRecord().

◆ XLogInsertRecord()

XLogRecPtr XLogInsertRecord	(	XLogRecData *	rdata,
		XLogRecPtr	fpw_lsn,
		uint8	flags,
		int	num_fpi,
		bool	topxid_included
	)

Definition at line 741 of file xlog.c.

 {
     XLogCtlInsert *Insert = &XLogCtl->Insert;
     pg_crc32c   rdata_crc;
     bool        inserted;
     XLogRecord *rechdr = (XLogRecord *) rdata->data;
     uint8       info = rechdr->xl_info & ~XLR_INFO_MASK;
     WalInsertClass class = WALINSERT_NORMAL;
     XLogRecPtr  StartPos;
     XLogRecPtr  EndPos;
     bool        prevDoPageWrites = doPageWrites;
     TimeLineID  insertTLI;
  
     /* Does this record type require special handling? */
     if (unlikely(rechdr->xl_rmid == RM_XLOG_ID))
     {
         if (info == XLOG_SWITCH)
             class = WALINSERT_SPECIAL_SWITCH;
         else if (info == XLOG_CHECKPOINT_REDO)
             class = WALINSERT_SPECIAL_CHECKPOINT;
     }
  
     /* we assume that all of the record header is in the first chunk */
     Assert(rdata->len >= SizeOfXLogRecord);
  
     /* cross-check on whether we should be here or not */
     if (!XLogInsertAllowed())
         elog(ERROR, "cannot make new WAL entries during recovery");
  
     /*
      * Given that we're not in recovery, InsertTimeLineID is set and can't
      * change, so we can read it without a lock.
      */
     insertTLI = XLogCtl->InsertTimeLineID;
  
     /*----------
      *
      * We have now done all the preparatory work we can without holding a
      * lock or modifying shared state. From here on, inserting the new WAL
      * record to the shared WAL buffer cache is a two-step process:
      *
      * 1. Reserve the right amount of space from the WAL. The current head of
      *    reserved space is kept in Insert->CurrBytePos, and is protected by
      *    insertpos_lck.
      *
      * 2. Copy the record to the reserved WAL space. This involves finding the
      *    correct WAL buffer containing the reserved space, and copying the
      *    record in place. This can be done concurrently in multiple processes.
      *
      * To keep track of which insertions are still in-progress, each concurrent
      * inserter acquires an insertion lock. In addition to just indicating that
      * an insertion is in progress, the lock tells others how far the inserter
      * has progressed. There is a small fixed number of insertion locks,
      * determined by NUM_XLOGINSERT_LOCKS. When an inserter crosses a page
      * boundary, it updates the value stored in the lock to the how far it has
      * inserted, to allow the previous buffer to be flushed.
      *
      * Holding onto an insertion lock also protects RedoRecPtr and
      * fullPageWrites from changing until the insertion is finished.
      *
      * Step 2 can usually be done completely in parallel. If the required WAL
      * page is not initialized yet, you have to grab WALBufMappingLock to
      * initialize it, but the WAL writer tries to do that ahead of insertions
      * to avoid that from happening in the critical path.
      *
      *----------
      */
     START_CRIT_SECTION();
  
     if (likely(class == WALINSERT_NORMAL))
     {
         WALInsertLockAcquire();
  
         /*
          * Check to see if my copy of RedoRecPtr is out of date. If so, may
          * have to go back and have the caller recompute everything. This can
          * only happen just after a checkpoint, so it's better to be slow in
          * this case and fast otherwise.
          *
          * Also check to see if fullPageWrites was just turned on or there's a
          * running backup (which forces full-page writes); if we weren't
          * already doing full-page writes then go back and recompute.
          *
          * If we aren't doing full-page writes then RedoRecPtr doesn't
          * actually affect the contents of the XLOG record, so we'll update
          * our local copy but not force a recomputation.  (If doPageWrites was
          * just turned off, we could recompute the record without full pages,
          * but we choose not to bother.)
          */
         if (RedoRecPtr != Insert->RedoRecPtr)
         {
             Assert(RedoRecPtr < Insert->RedoRecPtr);
             RedoRecPtr = Insert->RedoRecPtr;
         }
         doPageWrites = (Insert->fullPageWrites || Insert->runningBackups > 0);
  
         if (doPageWrites &&
             (!prevDoPageWrites ||
              (fpw_lsn != InvalidXLogRecPtr && fpw_lsn <= RedoRecPtr)))
         {
             /*
              * Oops, some buffer now needs to be backed up that the caller
              * didn't back up.  Start over.
              */
             WALInsertLockRelease();
             END_CRIT_SECTION();
             return InvalidXLogRecPtr;
         }
  
         /*
          * Reserve space for the record in the WAL. This also sets the xl_prev
          * pointer.
          */
         ReserveXLogInsertLocation(rechdr->xl_tot_len, &StartPos, &EndPos,
                                   &rechdr->xl_prev);
  
         /* Normal records are always inserted. */
         inserted = true;
     }
     else if (class == WALINSERT_SPECIAL_SWITCH)
     {
         /*
          * In order to insert an XLOG_SWITCH record, we need to hold all of
          * the WAL insertion locks, not just one, so that no one else can
          * begin inserting a record until we've figured out how much space
          * remains in the current WAL segment and claimed all of it.
          *
          * Nonetheless, this case is simpler than the normal cases handled
          * below, which must check for changes in doPageWrites and RedoRecPtr.
          * Those checks are only needed for records that can contain buffer
          * references, and an XLOG_SWITCH record never does.
          */
         Assert(fpw_lsn == InvalidXLogRecPtr);
         WALInsertLockAcquireExclusive();
         inserted = ReserveXLogSwitch(&StartPos, &EndPos, &rechdr->xl_prev);
     }
     else
     {
         Assert(class == WALINSERT_SPECIAL_CHECKPOINT);
  
         /*
          * We need to update both the local and shared copies of RedoRecPtr,
          * which means that we need to hold all the WAL insertion locks.
          * However, there can't be any buffer references, so as above, we need
          * not check RedoRecPtr before inserting the record; we just need to
          * update it afterwards.
          */
         Assert(fpw_lsn == InvalidXLogRecPtr);
         WALInsertLockAcquireExclusive();
         ReserveXLogInsertLocation(rechdr->xl_tot_len, &StartPos, &EndPos,
                                   &rechdr->xl_prev);
         RedoRecPtr = Insert->RedoRecPtr = StartPos;
         inserted = true;
     }
  
     if (inserted)
     {
         /*
          * Now that xl_prev has been filled in, calculate CRC of the record
          * header.
          */
         rdata_crc = rechdr->xl_crc;
         COMP_CRC32C(rdata_crc, rechdr, offsetof(XLogRecord, xl_crc));
         FIN_CRC32C(rdata_crc);
         rechdr->xl_crc = rdata_crc;
  
         /*
          * All the record data, including the header, is now ready to be
          * inserted. Copy the record in the space reserved.
          */
         CopyXLogRecordToWAL(rechdr->xl_tot_len,
                             class == WALINSERT_SPECIAL_SWITCH, rdata,
                             StartPos, EndPos, insertTLI);
  
         /*
          * Unless record is flagged as not important, update LSN of last
          * important record in the current slot. When holding all locks, just
          * update the first one.
          */
         if ((flags & XLOG_MARK_UNIMPORTANT) == 0)
         {
             int         lockno = holdingAllLocks ? 0 : MyLockNo;
  
             WALInsertLocks[lockno].l.lastImportantAt = StartPos;
         }
     }
     else
     {
         /*
          * This was an xlog-switch record, but the current insert location was
          * already exactly at the beginning of a segment, so there was no need
          * to do anything.
          */
     }
  
     /*
      * Done! Let others know that we're finished.
      */
     WALInsertLockRelease();
  
     END_CRIT_SECTION();
  
     MarkCurrentTransactionIdLoggedIfAny();
  
     /*
      * Mark top transaction id is logged (if needed) so that we should not try
      * to log it again with the next WAL record in the current subtransaction.
      */
     if (topxid_included)
         MarkSubxactTopXidLogged();
  
     /*
      * Update shared LogwrtRqst.Write, if we crossed page boundary.
      */
     if (StartPos / XLOG_BLCKSZ != EndPos / XLOG_BLCKSZ)
     {
         SpinLockAcquire(&XLogCtl->info_lck);
         /* advance global request to include new block(s) */
         if (XLogCtl->LogwrtRqst.Write < EndPos)
             XLogCtl->LogwrtRqst.Write = EndPos;
         SpinLockRelease(&XLogCtl->info_lck);
         RefreshXLogWriteResult(LogwrtResult);
     }
  
     /*
      * If this was an XLOG_SWITCH record, flush the record and the empty
      * padding space that fills the rest of the segment, and perform
      * end-of-segment actions (eg, notifying archiver).
      */
     if (class == WALINSERT_SPECIAL_SWITCH)
     {
         TRACE_POSTGRESQL_WAL_SWITCH();
         XLogFlush(EndPos);
  
         /*
          * Even though we reserved the rest of the segment for us, which is
          * reflected in EndPos, we return a pointer to just the end of the
          * xlog-switch record.
          */
         if (inserted)
         {
             EndPos = StartPos + SizeOfXLogRecord;
             if (StartPos / XLOG_BLCKSZ != EndPos / XLOG_BLCKSZ)
             {
                 uint64      offset = XLogSegmentOffset(EndPos, wal_segment_size);
  
                 if (offset == EndPos % XLOG_BLCKSZ)
                     EndPos += SizeOfXLogLongPHD;
                 else
                     EndPos += SizeOfXLogShortPHD;
             }
         }
     }
  
 #ifdef WAL_DEBUG
     if (XLOG_DEBUG)
     {
         static XLogReaderState *debug_reader = NULL;
         XLogRecord *record;
         DecodedXLogRecord *decoded;
         StringInfoData buf;
         StringInfoData recordBuf;
         char       *errormsg = NULL;
         MemoryContext oldCxt;
  
         oldCxt = MemoryContextSwitchTo(walDebugCxt);
  
         initStringInfo(&buf);
         appendStringInfo(&buf, "INSERT @ %X/%X: ", LSN_FORMAT_ARGS(EndPos));
  
         /*
          * We have to piece together the WAL record data from the XLogRecData
          * entries, so that we can pass it to the rm_desc function as one
          * contiguous chunk.
          */
         initStringInfo(&recordBuf);
         for (; rdata != NULL; rdata = rdata->next)
             appendBinaryStringInfo(&recordBuf, rdata->data, rdata->len);
  
         /* We also need temporary space to decode the record. */
         record = (XLogRecord *) recordBuf.data;
         decoded = (DecodedXLogRecord *)
             palloc(DecodeXLogRecordRequiredSpace(record->xl_tot_len));
  
         if (!debug_reader)
             debug_reader = XLogReaderAllocate(wal_segment_size, NULL,
                                               XL_ROUTINE(.page_read = NULL,
                                                          .segment_open = NULL,
                                                          .segment_close = NULL),
                                               NULL);
         if (!debug_reader)
         {
             appendStringInfoString(&buf, "error decoding record: out of memory while allocating a WAL reading processor");
         }
         else if (!DecodeXLogRecord(debug_reader,
                                    decoded,
                                    record,
                                    EndPos,
                                    &errormsg))
         {
             appendStringInfo(&buf, "error decoding record: %s",
                              errormsg ? errormsg : "no error message");
         }
         else
         {
             appendStringInfoString(&buf, " - ");
  
             debug_reader->record = decoded;
             xlog_outdesc(&buf, debug_reader);
             debug_reader->record = NULL;
         }
         elog(LOG, "%s", buf.data);
  
         pfree(decoded);
         pfree(buf.data);
         pfree(recordBuf.data);
         MemoryContextSwitchTo(oldCxt);
     }
 #endif
  
     /*
      * Update our global variables
      */
     ProcLastRecPtr = StartPos;
     XactLastRecEnd = EndPos;
  
     /* Report WAL traffic to the instrumentation. */
     if (inserted)
     {
         pgWalUsage.wal_bytes += rechdr->xl_tot_len;
         pgWalUsage.wal_records++;
         pgWalUsage.wal_fpi += num_fpi;
     }
  
     return EndPos;
 }

References appendBinaryStringInfo(), appendStringInfo(), appendStringInfoString(), Assert, buf, COMP_CRC32C, CopyXLogRecordToWAL(), XLogRecData::data, StringInfoData::data, DecodeXLogRecord(), DecodeXLogRecordRequiredSpace(), doPageWrites, elog, END_CRIT_SECTION, ERROR, FIN_CRC32C, holdingAllLocks, if(), XLogCtlData::info_lck, initStringInfo(), XLogCtlData::Insert, Insert(), XLogCtlData::InsertTimeLineID, InvalidXLogRecPtr, WALInsertLockPadded::l, WALInsertLock::lastImportantAt, XLogRecData::len, likely, LOG, LogwrtResult, XLogCtlData::LogwrtRqst, LSN_FORMAT_ARGS, MarkCurrentTransactionIdLoggedIfAny(), MarkSubxactTopXidLogged(), MemoryContextSwitchTo(), MyLockNo, XLogRecData::next, palloc(), pfree(), pgWalUsage, ProcLastRecPtr, XLogReaderState::record, RedoRecPtr, RefreshXLogWriteResult, ReserveXLogInsertLocation(), ReserveXLogSwitch(), SizeOfXLogLongPHD, SizeOfXLogRecord, SizeOfXLogShortPHD, SpinLockAcquire, SpinLockRelease, START_CRIT_SECTION, unlikely, WalUsage::wal_bytes, WalUsage::wal_fpi, WalUsage::wal_records, wal_segment_size, WALINSERT_NORMAL, WALINSERT_SPECIAL_CHECKPOINT, WALINSERT_SPECIAL_SWITCH, WALInsertLockAcquire(), WALInsertLockAcquireExclusive(), WALInsertLockRelease(), WALInsertLocks, XLogwrtRqst::Write, XactLastRecEnd, XLogRecord::xl_crc, XLogRecord::xl_info, XLogRecord::xl_prev, XLogRecord::xl_rmid, XL_ROUTINE, XLogRecord::xl_tot_len, XLOG_CHECKPOINT_REDO, XLOG_MARK_UNIMPORTANT, xlog_outdesc(), XLOG_SWITCH, XLogCtl, XLogFlush(), XLogInsertAllowed(), XLogReaderAllocate(), XLogSegmentOffset, and XLR_INFO_MASK.

Referenced by XLogInsert().

◆ XLogNeedsFlush()

bool XLogNeedsFlush ( XLogRecPtr record )

Definition at line 3120 of file xlog.c.

 {
     /*
      * During recovery, we don't flush WAL but update minRecoveryPoint
      * instead. So "needs flush" is taken to mean whether minRecoveryPoint
      * would need to be updated.
      */
     if (RecoveryInProgress())
     {
         /*
          * An invalid minRecoveryPoint means that we need to recover all the
          * WAL, i.e., we're doing crash recovery.  We never modify the control
          * file's value in that case, so we can short-circuit future checks
          * here too.  This triggers a quick exit path for the startup process,
          * which cannot update its local copy of minRecoveryPoint as long as
          * it has not replayed all WAL available when doing crash recovery.
          */
         if (XLogRecPtrIsInvalid(LocalMinRecoveryPoint) && InRecovery)
             updateMinRecoveryPoint = false;
  
         /* Quick exit if already known to be updated or cannot be updated */
         if (record <= LocalMinRecoveryPoint || !updateMinRecoveryPoint)
             return false;
  
         /*
          * Update local copy of minRecoveryPoint. But if the lock is busy,
          * just return a conservative guess.
          */
         if (!LWLockConditionalAcquire(ControlFileLock, LW_SHARED))
             return true;
         LocalMinRecoveryPoint = ControlFile->minRecoveryPoint;
         LocalMinRecoveryPointTLI = ControlFile->minRecoveryPointTLI;
         LWLockRelease(ControlFileLock);
  
         /*
          * Check minRecoveryPoint for any other process than the startup
          * process doing crash recovery, which should not update the control
          * file value if crash recovery is still running.
          */
         if (XLogRecPtrIsInvalid(LocalMinRecoveryPoint))
             updateMinRecoveryPoint = false;
  
         /* check again */
         if (record <= LocalMinRecoveryPoint || !updateMinRecoveryPoint)
             return false;
         else
             return true;
     }
  
     /* Quick exit if already known flushed */
     if (record <= LogwrtResult.Flush)
         return false;
  
     /* read LogwrtResult and update local state */
     RefreshXLogWriteResult(LogwrtResult);
  
     /* check again */
     if (record <= LogwrtResult.Flush)
         return false;
  
     return true;
 }

References ControlFile, XLogwrtResult::Flush, InRecovery, LocalMinRecoveryPoint, LocalMinRecoveryPointTLI, LogwrtResult, LW_SHARED, LWLockConditionalAcquire(), LWLockRelease(), ControlFileData::minRecoveryPoint, ControlFileData::minRecoveryPointTLI, RecoveryInProgress(), RefreshXLogWriteResult, updateMinRecoveryPoint, and XLogRecPtrIsInvalid.

Referenced by GetVictimBuffer(), and SetHintBits().

◆ XLogPutNextOid()

void XLogPutNextOid ( Oid nextOid )

Definition at line 7953 of file xlog.c.

 {
     XLogBeginInsert();
     XLogRegisterData((char *) (&nextOid), sizeof(Oid));
     (void) XLogInsert(RM_XLOG_ID, XLOG_NEXTOID);
  
     /*
      * We need not flush the NEXTOID record immediately, because any of the
      * just-allocated OIDs could only reach disk as part of a tuple insert or
      * update that would have its own XLOG record that must follow the NEXTOID
      * record.  Therefore, the standard buffer LSN interlock applied to those
      * records will ensure no such OID reaches disk before the NEXTOID record
      * does.
      *
      * Note, however, that the above statement only covers state "within" the
      * database.  When we use a generated OID as a file or directory name, we
      * are in a sense violating the basic WAL rule, because that filesystem
      * change may reach disk before the NEXTOID WAL record does.  The impact
      * of this is that if a database crash occurs immediately afterward, we
      * might after restart re-generate the same OID and find that it conflicts
      * with the leftover file or directory.  But since for safety's sake we
      * always loop until finding a nonconflicting filename, this poses no real
      * problem in practice. See pgsql-hackers discussion 27-Sep-2006.
      */
 }

References XLOG_NEXTOID, XLogBeginInsert(), XLogInsert(), and XLogRegisterData().

Referenced by GetNewObjectId().

◆ XLogRecPtrToBytePos()

static uint64 XLogRecPtrToBytePos ( XLogRecPtr ptr )

static

Definition at line 1934 of file xlog.c.

 {
     uint64      fullsegs;
     uint32      fullpages;
     uint32      offset;
     uint64      result;
  
     XLByteToSeg(ptr, fullsegs, wal_segment_size);
  
     fullpages = (XLogSegmentOffset(ptr, wal_segment_size)) / XLOG_BLCKSZ;
     offset = ptr % XLOG_BLCKSZ;
  
     if (fullpages == 0)
     {
         result = fullsegs * UsableBytesInSegment;
         if (offset > 0)
         {
             Assert(offset >= SizeOfXLogLongPHD);
             result += offset - SizeOfXLogLongPHD;
         }
     }
     else
     {
         result = fullsegs * UsableBytesInSegment +
             (XLOG_BLCKSZ - SizeOfXLogLongPHD) + /* account for first page */
             (fullpages - 1) * UsableBytesInPage;    /* full pages */
         if (offset > 0)
         {
             Assert(offset >= SizeOfXLogShortPHD);
             result += offset - SizeOfXLogShortPHD;
         }
     }
  
     return result;
 }

References Assert, SizeOfXLogLongPHD, SizeOfXLogShortPHD, UsableBytesInPage, UsableBytesInSegment, wal_segment_size, XLByteToSeg, and XLogSegmentOffset.

Referenced by ReserveXLogInsertLocation(), ReserveXLogSwitch(), and StartupXLOG().

◆ XLogReportParameters()

static void XLogReportParameters ( void )

static

Definition at line 8033 of file xlog.c.

 {
     if (wal_level != ControlFile->wal_level ||
         wal_log_hints != ControlFile->wal_log_hints ||
         MaxConnections != ControlFile->MaxConnections ||
         max_worker_processes != ControlFile->max_worker_processes ||
         max_wal_senders != ControlFile->max_wal_senders ||
         max_prepared_xacts != ControlFile->max_prepared_xacts ||
         max_locks_per_xact != ControlFile->max_locks_per_xact ||
         track_commit_timestamp != ControlFile->track_commit_timestamp)
     {
         /*
          * The change in number of backend slots doesn't need to be WAL-logged
          * if archiving is not enabled, as you can't start archive recovery
          * with wal_level=minimal anyway. We don't really care about the
          * values in pg_control either if wal_level=minimal, but seems better
          * to keep them up-to-date to avoid confusion.
          */
         if (wal_level != ControlFile->wal_level || XLogIsNeeded())
         {
             xl_parameter_change xlrec;
             XLogRecPtr  recptr;
  
             xlrec.MaxConnections = MaxConnections;
             xlrec.max_worker_processes = max_worker_processes;
             xlrec.max_wal_senders = max_wal_senders;
             xlrec.max_prepared_xacts = max_prepared_xacts;
             xlrec.max_locks_per_xact = max_locks_per_xact;
             xlrec.wal_level = wal_level;
             xlrec.wal_log_hints = wal_log_hints;
             xlrec.track_commit_timestamp = track_commit_timestamp;
  
             XLogBeginInsert();
             XLogRegisterData((char *) &xlrec, sizeof(xlrec));
  
             recptr = XLogInsert(RM_XLOG_ID, XLOG_PARAMETER_CHANGE);
             XLogFlush(recptr);
         }
  
         LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
  
         ControlFile->MaxConnections = MaxConnections;
         ControlFile->max_worker_processes = max_worker_processes;
         ControlFile->max_wal_senders = max_wal_senders;
         ControlFile->max_prepared_xacts = max_prepared_xacts;
         ControlFile->max_locks_per_xact = max_locks_per_xact;
         ControlFile->wal_level = wal_level;
         ControlFile->wal_log_hints = wal_log_hints;
         ControlFile->track_commit_timestamp = track_commit_timestamp;
         UpdateControlFile();
  
         LWLockRelease(ControlFileLock);
     }
 }

Referenced by StartupXLOG().

◆ XLogRestorePoint()

XLogRecPtr XLogRestorePoint ( const char * rpName )

Definition at line 8008 of file xlog.c.

 {
     XLogRecPtr  RecPtr;
     xl_restore_point xlrec;
  
     xlrec.rp_time = GetCurrentTimestamp();
     strlcpy(xlrec.rp_name, rpName, MAXFNAMELEN);
  
     XLogBeginInsert();
     XLogRegisterData((char *) &xlrec, sizeof(xl_restore_point));
  
     RecPtr = XLogInsert(RM_XLOG_ID, XLOG_RESTORE_POINT);
  
     ereport(LOG,
             (errmsg("restore point \"%s\" created at %X/%X",
                     rpName, LSN_FORMAT_ARGS(RecPtr))));
  
     return RecPtr;
 }

References ereport, errmsg(), GetCurrentTimestamp(), LOG, LSN_FORMAT_ARGS, MAXFNAMELEN, xl_restore_point::rp_name, xl_restore_point::rp_time, strlcpy(), XLOG_RESTORE_POINT, XLogBeginInsert(), XLogInsert(), and XLogRegisterData().

Referenced by pg_create_restore_point().

◆ XLogSetAsyncXactLSN()

void XLogSetAsyncXactLSN ( XLogRecPtr asyncXactLSN )

Definition at line 2624 of file xlog.c.

 {
     XLogRecPtr  WriteRqstPtr = asyncXactLSN;
     bool        sleeping;
     bool        wakeup = false;
     XLogRecPtr  prevAsyncXactLSN;
  
     SpinLockAcquire(&XLogCtl->info_lck);
     sleeping = XLogCtl->WalWriterSleeping;
     prevAsyncXactLSN = XLogCtl->asyncXactLSN;
     if (XLogCtl->asyncXactLSN < asyncXactLSN)
         XLogCtl->asyncXactLSN = asyncXactLSN;
     SpinLockRelease(&XLogCtl->info_lck);
  
     /*
      * If somebody else already called this function with a more aggressive
      * LSN, they will have done what we needed (and perhaps more).
      */
     if (asyncXactLSN <= prevAsyncXactLSN)
         return;
  
     /*
      * If the WALWriter is sleeping, kick it to make it come out of low-power
      * mode, so that this async commit will reach disk within the expected
      * amount of time.  Otherwise, determine whether it has enough WAL
      * available to flush, the same way that XLogBackgroundFlush() does.
      */
     if (sleeping)
         wakeup = true;
     else
     {
         int         flushblocks;
  
         RefreshXLogWriteResult(LogwrtResult);
  
         flushblocks =
             WriteRqstPtr / XLOG_BLCKSZ - LogwrtResult.Flush / XLOG_BLCKSZ;
  
         if (WalWriterFlushAfter == 0 || flushblocks >= WalWriterFlushAfter)
             wakeup = true;
     }
  
     if (wakeup && ProcGlobal->walwriterLatch)
         SetLatch(ProcGlobal->walwriterLatch);
 }

References XLogCtlData::asyncXactLSN, XLogwrtResult::Flush, XLogCtlData::info_lck, LogwrtResult, ProcGlobal, RefreshXLogWriteResult, SetLatch(), SpinLockAcquire, SpinLockRelease, wakeup, WalWriterFlushAfter, PROC_HDR::walwriterLatch, XLogCtlData::WalWriterSleeping, and XLogCtl.

Referenced by AbortTransaction(), LogCurrentRunningXacts(), RecordTransactionAbort(), and RecordTransactionCommit().

◆ XLogSetReplicationSlotMinimumLSN()

void XLogSetReplicationSlotMinimumLSN ( XLogRecPtr lsn )

Definition at line 2675 of file xlog.c.

 {
     SpinLockAcquire(&XLogCtl->info_lck);
     XLogCtl->replicationSlotMinLSN = lsn;
     SpinLockRelease(&XLogCtl->info_lck);
 }

References XLogCtlData::info_lck, XLogCtlData::replicationSlotMinLSN, SpinLockAcquire, SpinLockRelease, and XLogCtl.

Referenced by ReplicationSlotsComputeRequiredLSN().

◆ XLOGShmemInit()

void XLOGShmemInit ( void )

Definition at line 4883 of file xlog.c.

 {
     bool        foundCFile,
                 foundXLog;
     char       *allocptr;
     int         i;
     ControlFileData *localControlFile;
  
 #ifdef WAL_DEBUG
  
     /*
      * Create a memory context for WAL debugging that's exempt from the normal
      * "no pallocs in critical section" rule. Yes, that can lead to a PANIC if
      * an allocation fails, but wal_debug is not for production use anyway.
      */
     if (walDebugCxt == NULL)
     {
         walDebugCxt = AllocSetContextCreate(TopMemoryContext,
                                             "WAL Debug",
                                             ALLOCSET_DEFAULT_SIZES);
         MemoryContextAllowInCriticalSection(walDebugCxt, true);
     }
 #endif
  
  
     XLogCtl = (XLogCtlData *)
         ShmemInitStruct("XLOG Ctl", XLOGShmemSize(), &foundXLog);
  
     localControlFile = ControlFile;
     ControlFile = (ControlFileData *)
         ShmemInitStruct("Control File", sizeof(ControlFileData), &foundCFile);
  
     if (foundCFile || foundXLog)
     {
         /* both should be present or neither */
         Assert(foundCFile && foundXLog);
  
         /* Initialize local copy of WALInsertLocks */
         WALInsertLocks = XLogCtl->Insert.WALInsertLocks;
  
         if (localControlFile)
             pfree(localControlFile);
         return;
     }
     memset(XLogCtl, 0, sizeof(XLogCtlData));
  
     /*
      * Already have read control file locally, unless in bootstrap mode. Move
      * contents into shared memory.
      */
     if (localControlFile)
     {
         memcpy(ControlFile, localControlFile, sizeof(ControlFileData));
         pfree(localControlFile);
     }
  
     /*
      * Since XLogCtlData contains XLogRecPtr fields, its sizeof should be a
      * multiple of the alignment for same, so no extra alignment padding is
      * needed here.
      */
     allocptr = ((char *) XLogCtl) + sizeof(XLogCtlData);
     XLogCtl->xlblocks = (pg_atomic_uint64 *) allocptr;
     allocptr += sizeof(pg_atomic_uint64) * XLOGbuffers;
  
     for (i = 0; i < XLOGbuffers; i++)
     {
         pg_atomic_init_u64(&XLogCtl->xlblocks[i], InvalidXLogRecPtr);
     }
  
     /* WAL insertion locks. Ensure they're aligned to the full padded size */
     allocptr += sizeof(WALInsertLockPadded) -
         ((uintptr_t) allocptr) % sizeof(WALInsertLockPadded);
     WALInsertLocks = XLogCtl->Insert.WALInsertLocks =
         (WALInsertLockPadded *) allocptr;
     allocptr += sizeof(WALInsertLockPadded) * NUM_XLOGINSERT_LOCKS;
  
     for (i = 0; i < NUM_XLOGINSERT_LOCKS; i++)
     {
         LWLockInitialize(&WALInsertLocks[i].l.lock, LWTRANCHE_WAL_INSERT);
         pg_atomic_init_u64(&WALInsertLocks[i].l.insertingAt, InvalidXLogRecPtr);
         WALInsertLocks[i].l.lastImportantAt = InvalidXLogRecPtr;
     }
  
     /*
      * Align the start of the page buffers to a full xlog block size boundary.
      * This simplifies some calculations in XLOG insertion. It is also
      * required for O_DIRECT.
      */
     allocptr = (char *) TYPEALIGN(XLOG_BLCKSZ, allocptr);
     XLogCtl->pages = allocptr;
     memset(XLogCtl->pages, 0, (Size) XLOG_BLCKSZ * XLOGbuffers);
  
     /*
      * Do basic initialization of XLogCtl shared data. (StartupXLOG will fill
      * in additional info.)
      */
     XLogCtl->XLogCacheBlck = XLOGbuffers - 1;
     XLogCtl->SharedRecoveryState = RECOVERY_STATE_CRASH;
     XLogCtl->InstallXLogFileSegmentActive = false;
     XLogCtl->WalWriterSleeping = false;
  
     SpinLockInit(&XLogCtl->Insert.insertpos_lck);
     SpinLockInit(&XLogCtl->info_lck);
     pg_atomic_init_u64(&XLogCtl->logInsertResult, InvalidXLogRecPtr);
     pg_atomic_init_u64(&XLogCtl->logWriteResult, InvalidXLogRecPtr);
     pg_atomic_init_u64(&XLogCtl->logFlushResult, InvalidXLogRecPtr);
     pg_atomic_init_u64(&XLogCtl->unloggedLSN, InvalidXLogRecPtr);
 }

References ALLOCSET_DEFAULT_SIZES, AllocSetContextCreate, Assert, ControlFile, i, XLogCtlData::info_lck, XLogCtlData::Insert, XLogCtlInsert::insertpos_lck, XLogCtlData::InstallXLogFileSegmentActive, InvalidXLogRecPtr, WALInsertLockPadded::l, WALInsertLock::lastImportantAt, XLogCtlData::logFlushResult, XLogCtlData::logInsertResult, XLogCtlData::logWriteResult, LWLockInitialize(), LWTRANCHE_WAL_INSERT, MemoryContextAllowInCriticalSection(), NUM_XLOGINSERT_LOCKS, XLogCtlData::pages, pfree(), pg_atomic_init_u64(), RECOVERY_STATE_CRASH, XLogCtlData::SharedRecoveryState, ShmemInitStruct(), SpinLockInit, TopMemoryContext, TYPEALIGN, XLogCtlData::unloggedLSN, XLogCtlInsert::WALInsertLocks, WALInsertLocks, XLogCtlData::WalWriterSleeping, XLogCtlData::xlblocks, XLOGbuffers, XLogCtlData::XLogCacheBlck, XLogCtl, and XLOGShmemSize().

Referenced by CreateOrAttachShmemStructs().

◆ XLOGShmemSize()

Size XLOGShmemSize ( void )

Definition at line 4833 of file xlog.c.

 {
     Size        size;
  
     /*
      * If the value of wal_buffers is -1, use the preferred auto-tune value.
      * This isn't an amazingly clean place to do this, but we must wait till
      * NBuffers has received its final value, and must do it before using the
      * value of XLOGbuffers to do anything important.
      *
      * We prefer to report this value's source as PGC_S_DYNAMIC_DEFAULT.
      * However, if the DBA explicitly set wal_buffers = -1 in the config file,
      * then PGC_S_DYNAMIC_DEFAULT will fail to override that and we must force
      * the matter with PGC_S_OVERRIDE.
      */
     if (XLOGbuffers == -1)
     {
         char        buf[32];
  
         snprintf(buf, sizeof(buf), "%d", XLOGChooseNumBuffers());
         SetConfigOption("wal_buffers", buf, PGC_POSTMASTER,
                         PGC_S_DYNAMIC_DEFAULT);
         if (XLOGbuffers == -1)  /* failed to apply it? */
             SetConfigOption("wal_buffers", buf, PGC_POSTMASTER,
                             PGC_S_OVERRIDE);
     }
     Assert(XLOGbuffers > 0);
  
     /* XLogCtl */
     size = sizeof(XLogCtlData);
  
     /* WAL insertion locks, plus alignment */
     size = add_size(size, mul_size(sizeof(WALInsertLockPadded), NUM_XLOGINSERT_LOCKS + 1));
     /* xlblocks array */
     size = add_size(size, mul_size(sizeof(pg_atomic_uint64), XLOGbuffers));
     /* extra alignment padding for XLOG I/O buffers */
     size = add_size(size, Max(XLOG_BLCKSZ, PG_IO_ALIGN_SIZE));
     /* and the buffers themselves */
     size = add_size(size, mul_size(XLOG_BLCKSZ, XLOGbuffers));
  
     /*
      * Note: we don't count ControlFileData, it comes out of the "slop factor"
      * added by CreateSharedMemoryAndSemaphores.  This lets us use this
      * routine again below to compute the actual allocation size.
      */
  
     return size;
 }

References add_size(), Assert, buf, Max, mul_size(), NUM_XLOGINSERT_LOCKS, PG_IO_ALIGN_SIZE, PGC_POSTMASTER, PGC_S_DYNAMIC_DEFAULT, PGC_S_OVERRIDE, SetConfigOption(), size, snprintf, XLOGbuffers, and XLOGChooseNumBuffers().

Referenced by CalculateShmemSize(), and XLOGShmemInit().

◆ XLogShutdownWalRcv()

void XLogShutdownWalRcv ( void )

Definition at line 9403 of file xlog.c.

 {
     ShutdownWalRcv();
  
     LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
     XLogCtl->InstallXLogFileSegmentActive = false;
     LWLockRelease(ControlFileLock);
 }

References XLogCtlData::InstallXLogFileSegmentActive, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), ShutdownWalRcv(), and XLogCtl.

Referenced by FinishWalRecovery(), and WaitForWALToBecomeAvailable().

◆ XLogWrite()

static void XLogWrite	(	XLogwrtRqst	WriteRqst,
		TimeLineID	tli,
		bool	flexible
	)

static

Definition at line 2307 of file xlog.c.

 {
     bool        ispartialpage;
     bool        last_iteration;
     bool        finishing_seg;
     int         curridx;
     int         npages;
     int         startidx;
     uint32      startoffset;
  
     /* We should always be inside a critical section here */
     Assert(CritSectionCount > 0);
  
     /*
      * Update local LogwrtResult (caller probably did this already, but...)
      */
     RefreshXLogWriteResult(LogwrtResult);
  
     /*
      * Since successive pages in the xlog cache are consecutively allocated,
      * we can usually gather multiple pages together and issue just one
      * write() call.  npages is the number of pages we have determined can be
      * written together; startidx is the cache block index of the first one,
      * and startoffset is the file offset at which it should go. The latter
      * two variables are only valid when npages > 0, but we must initialize
      * all of them to keep the compiler quiet.
      */
     npages = 0;
     startidx = 0;
     startoffset = 0;
  
     /*
      * Within the loop, curridx is the cache block index of the page to
      * consider writing.  Begin at the buffer containing the next unwritten
      * page, or last partially written page.
      */
     curridx = XLogRecPtrToBufIdx(LogwrtResult.Write);
  
     while (LogwrtResult.Write < WriteRqst.Write)
     {
         /*
          * Make sure we're not ahead of the insert process.  This could happen
          * if we're passed a bogus WriteRqst.Write that is past the end of the
          * last page that's been initialized by AdvanceXLInsertBuffer.
          */
         XLogRecPtr  EndPtr = pg_atomic_read_u64(&XLogCtl->xlblocks[curridx]);
  
         if (LogwrtResult.Write >= EndPtr)
             elog(PANIC, "xlog write request %X/%X is past end of log %X/%X",
                  LSN_FORMAT_ARGS(LogwrtResult.Write),
                  LSN_FORMAT_ARGS(EndPtr));
  
         /* Advance LogwrtResult.Write to end of current buffer page */
         LogwrtResult.Write = EndPtr;
         ispartialpage = WriteRqst.Write < LogwrtResult.Write;
  
         if (!XLByteInPrevSeg(LogwrtResult.Write, openLogSegNo,
                              wal_segment_size))
         {
             /*
              * Switch to new logfile segment.  We cannot have any pending
              * pages here (since we dump what we have at segment end).
              */
             Assert(npages == 0);
             if (openLogFile >= 0)
                 XLogFileClose();
             XLByteToPrevSeg(LogwrtResult.Write, openLogSegNo,
                             wal_segment_size);
             openLogTLI = tli;
  
             /* create/use new log file */
             openLogFile = XLogFileInit(openLogSegNo, tli);
             ReserveExternalFD();
         }
  
         /* Make sure we have the current logfile open */
         if (openLogFile < 0)
         {
             XLByteToPrevSeg(LogwrtResult.Write, openLogSegNo,
                             wal_segment_size);
             openLogTLI = tli;
             openLogFile = XLogFileOpen(openLogSegNo, tli);
             ReserveExternalFD();
         }
  
         /* Add current page to the set of pending pages-to-dump */
         if (npages == 0)
         {
             /* first of group */
             startidx = curridx;
             startoffset = XLogSegmentOffset(LogwrtResult.Write - XLOG_BLCKSZ,
                                             wal_segment_size);
         }
         npages++;
  
         /*
          * Dump the set if this will be the last loop iteration, or if we are
          * at the last page of the cache area (since the next page won't be
          * contiguous in memory), or if we are at the end of the logfile
          * segment.
          */
         last_iteration = WriteRqst.Write <= LogwrtResult.Write;
  
         finishing_seg = !ispartialpage &&
             (startoffset + npages * XLOG_BLCKSZ) >= wal_segment_size;
  
         if (last_iteration ||
             curridx == XLogCtl->XLogCacheBlck ||
             finishing_seg)
         {
             char       *from;
             Size        nbytes;
             Size        nleft;
             ssize_t     written;
             instr_time  start;
  
             /* OK to write the page(s) */
             from = XLogCtl->pages + startidx * (Size) XLOG_BLCKSZ;
             nbytes = npages * (Size) XLOG_BLCKSZ;
             nleft = nbytes;
             do
             {
                 errno = 0;
  
                 /* Measure I/O timing to write WAL data */
                 if (track_wal_io_timing)
                     INSTR_TIME_SET_CURRENT(start);
                 else
                     INSTR_TIME_SET_ZERO(start);
  
                 pgstat_report_wait_start(WAIT_EVENT_WAL_WRITE);
                 written = pg_pwrite(openLogFile, from, nleft, startoffset);
                 pgstat_report_wait_end();
  
                 /*
                  * Increment the I/O timing and the number of times WAL data
                  * were written out to disk.
                  */
                 if (track_wal_io_timing)
                 {
                     instr_time  end;
  
                     INSTR_TIME_SET_CURRENT(end);
                     INSTR_TIME_ACCUM_DIFF(PendingWalStats.wal_write_time, end, start);
                 }
  
                 PendingWalStats.wal_write++;
  
                 if (written <= 0)
                 {
                     char        xlogfname[MAXFNAMELEN];
                     int         save_errno;
  
                     if (errno == EINTR)
                         continue;
  
                     save_errno = errno;
                     XLogFileName(xlogfname, tli, openLogSegNo,
                                  wal_segment_size);
                     errno = save_errno;
                     ereport(PANIC,
                             (errcode_for_file_access(),
                              errmsg("could not write to log file \"%s\" at offset %u, length %zu: %m",
                                     xlogfname, startoffset, nleft)));
                 }
                 nleft -= written;
                 from += written;
                 startoffset += written;
             } while (nleft > 0);
  
             npages = 0;
  
             /*
              * If we just wrote the whole last page of a logfile segment,
              * fsync the segment immediately.  This avoids having to go back
              * and re-open prior segments when an fsync request comes along
              * later. Doing it here ensures that one and only one backend will
              * perform this fsync.
              *
              * This is also the right place to notify the Archiver that the
              * segment is ready to copy to archival storage, and to update the
              * timer for archive_timeout, and to signal for a checkpoint if
              * too many logfile segments have been used since the last
              * checkpoint.
              */
             if (finishing_seg)
             {
                 issue_xlog_fsync(openLogFile, openLogSegNo, tli);
  
                 /* signal that we need to wakeup walsenders later */
                 WalSndWakeupRequest();
  
                 LogwrtResult.Flush = LogwrtResult.Write;    /* end of page */
  
                 if (XLogArchivingActive())
                     XLogArchiveNotifySeg(openLogSegNo, tli);
  
                 XLogCtl->lastSegSwitchTime = (pg_time_t) time(NULL);
                 XLogCtl->lastSegSwitchLSN = LogwrtResult.Flush;
  
                 /*
                  * Request a checkpoint if we've consumed too much xlog since
                  * the last one.  For speed, we first check using the local
                  * copy of RedoRecPtr, which might be out of date; if it looks
                  * like a checkpoint is needed, forcibly update RedoRecPtr and
                  * recheck.
                  */
                 if (IsUnderPostmaster && XLogCheckpointNeeded(openLogSegNo))
                 {
                     (void) GetRedoRecPtr();
                     if (XLogCheckpointNeeded(openLogSegNo))
                         RequestCheckpoint(CHECKPOINT_CAUSE_XLOG);
                 }
             }
         }
  
         if (ispartialpage)
         {
             /* Only asked to write a partial page */
             LogwrtResult.Write = WriteRqst.Write;
             break;
         }
         curridx = NextBufIdx(curridx);
  
         /* If flexible, break out of loop as soon as we wrote something */
         if (flexible && npages == 0)
             break;
     }
  
     Assert(npages == 0);
  
     /*
      * If asked to flush, do so
      */
     if (LogwrtResult.Flush < WriteRqst.Flush &&
         LogwrtResult.Flush < LogwrtResult.Write)
     {
         /*
          * Could get here without iterating above loop, in which case we might
          * have no open file or the wrong one.  However, we do not need to
          * fsync more than one file.
          */
         if (wal_sync_method != WAL_SYNC_METHOD_OPEN &&
             wal_sync_method != WAL_SYNC_METHOD_OPEN_DSYNC)
         {
             if (openLogFile >= 0 &&
                 !XLByteInPrevSeg(LogwrtResult.Write, openLogSegNo,
                                  wal_segment_size))
                 XLogFileClose();
             if (openLogFile < 0)
             {
                 XLByteToPrevSeg(LogwrtResult.Write, openLogSegNo,
                                 wal_segment_size);
                 openLogTLI = tli;
                 openLogFile = XLogFileOpen(openLogSegNo, tli);
                 ReserveExternalFD();
             }
  
             issue_xlog_fsync(openLogFile, openLogSegNo, tli);
         }
  
         /* signal that we need to wakeup walsenders later */
         WalSndWakeupRequest();
  
         LogwrtResult.Flush = LogwrtResult.Write;
     }
  
     /*
      * Update shared-memory status
      *
      * We make sure that the shared 'request' values do not fall behind the
      * 'result' values.  This is not absolutely essential, but it saves some
      * code in a couple of places.
      */
     SpinLockAcquire(&XLogCtl->info_lck);
     if (XLogCtl->LogwrtRqst.Write < LogwrtResult.Write)
         XLogCtl->LogwrtRqst.Write = LogwrtResult.Write;
     if (XLogCtl->LogwrtRqst.Flush < LogwrtResult.Flush)
         XLogCtl->LogwrtRqst.Flush = LogwrtResult.Flush;
     SpinLockRelease(&XLogCtl->info_lck);
  
     /*
      * We write Write first, bar, then Flush.  When reading, the opposite must
      * be done (with a matching barrier in between), so that we always see a
      * Flush value that trails behind the Write value seen.
      */
     pg_atomic_write_u64(&XLogCtl->logWriteResult, LogwrtResult.Write);
     pg_write_barrier();
     pg_atomic_write_u64(&XLogCtl->logFlushResult, LogwrtResult.Flush);
  
 #ifdef USE_ASSERT_CHECKING
     {
         XLogRecPtr  Flush;
         XLogRecPtr  Write;
         XLogRecPtr  Insert;
  
         Flush = pg_atomic_read_u64(&XLogCtl->logFlushResult);
         pg_read_barrier();
         Write = pg_atomic_read_u64(&XLogCtl->logWriteResult);
         pg_read_barrier();
         Insert = pg_atomic_read_u64(&XLogCtl->logInsertResult);
  
         /* WAL written to disk is always ahead of WAL flushed */
         Assert(Write >= Flush);
  
         /* WAL inserted to buffers is always ahead of WAL written */
         Assert(Insert >= Write);
     }
 #endif
 }

References Assert, CHECKPOINT_CAUSE_XLOG, CritSectionCount, EINTR, elog, ereport, errcode_for_file_access(), errmsg(), XLogwrtRqst::Flush, XLogwrtResult::Flush, Flush, GetRedoRecPtr(), XLogCtlData::info_lck, Insert(), INSTR_TIME_ACCUM_DIFF, INSTR_TIME_SET_CURRENT, INSTR_TIME_SET_ZERO, issue_xlog_fsync(), IsUnderPostmaster, XLogCtlData::lastSegSwitchLSN, XLogCtlData::lastSegSwitchTime, XLogCtlData::logFlushResult, XLogCtlData::logInsertResult, XLogCtlData::logWriteResult, LogwrtResult, XLogCtlData::LogwrtRqst, LSN_FORMAT_ARGS, MAXFNAMELEN, NextBufIdx, openLogFile, openLogSegNo, openLogTLI, XLogCtlData::pages, PANIC, PendingWalStats, pg_atomic_read_u64(), pg_atomic_write_u64(), pg_pwrite, pg_read_barrier, pg_write_barrier, pgstat_report_wait_end(), pgstat_report_wait_start(), RefreshXLogWriteResult, RequestCheckpoint(), ReserveExternalFD(), SpinLockAcquire, SpinLockRelease, start, track_wal_io_timing, wal_segment_size, wal_sync_method, WAL_SYNC_METHOD_OPEN, WAL_SYNC_METHOD_OPEN_DSYNC, PgStat_PendingWalStats::wal_write, PgStat_PendingWalStats::wal_write_time, WalSndWakeupRequest, XLogwrtRqst::Write, XLogwrtResult::Write, Write, XLogCtlData::xlblocks, XLByteInPrevSeg, XLByteToPrevSeg, XLogArchiveNotifySeg(), XLogArchivingActive, XLogCtlData::XLogCacheBlck, XLogCheckpointNeeded(), XLogCtl, XLogFileClose(), XLogFileInit(), XLogFileName(), XLogFileOpen(), XLogRecPtrToBufIdx, and XLogSegmentOffset.

Referenced by AdvanceXLInsertBuffer(), XLogBackgroundFlush(), and XLogFlush().

Variable Documentation

◆ archive_mode_options

const struct config_enum_entry archive_mode_options[]

Initial value:

= {
    {"always", ARCHIVE_MODE_ALWAYS, false},
    {"on", ARCHIVE_MODE_ON, false},
    {"off", ARCHIVE_MODE_OFF, false},
    {"true", ARCHIVE_MODE_ON, true},
    {"false", ARCHIVE_MODE_OFF, true},
    {"yes", ARCHIVE_MODE_ON, true},
    {"no", ARCHIVE_MODE_OFF, true},
    {"1", ARCHIVE_MODE_ON, true},
    {"0", ARCHIVE_MODE_OFF, true},
    {NULL, 0, false}
}

Definition at line 164 of file xlog.c.

◆ check_wal_consistency_checking_deferred

bool check_wal_consistency_checking_deferred = false

static

Definition at line 164 of file xlog.c.

Referenced by check_wal_consistency_checking(), and InitializeWalConsistencyChecking().

◆ CheckPointDistanceEstimate

double CheckPointDistanceEstimate = 0

static

Definition at line 157 of file xlog.c.

Referenced by LogCheckpointEnd(), UpdateCheckPointDistanceEstimate(), and XLOGfileslop().

◆ CheckPointSegments

int CheckPointSegments

Definition at line 154 of file xlog.c.

Referenced by CalculateCheckpointSegments(), IsCheckpointOnSchedule(), XLogCheckpointNeeded(), and XLogFileInitInternal().

◆ CheckpointStats

CheckpointStatsData CheckpointStats

Definition at line 207 of file xlog.c.

Referenced by BufferSync(), CheckPointGuts(), CreateCheckPoint(), CreateRestartPoint(), LogCheckpointEnd(), PreallocXlogFiles(), ProcessSyncRequests(), RemoveXlogFile(), and SlruInternalWritePage().

◆ CommitDelay

int CommitDelay = 0

Definition at line 130 of file xlog.c.

Referenced by XLogFlush().

◆ CommitSiblings

int CommitSiblings = 5

Definition at line 131 of file xlog.c.

Referenced by XLogFlush().

◆ ControlFile

ControlFileData* ControlFile = NULL

static

Definition at line 567 of file xlog.c.

Referenced by AdvanceXLInsertBuffer(), BootStrapXLOG(), checkControlFile(), CheckRequiredParameterValues(), CreateCheckPoint(), CreateEndOfRecoveryRecord(), CreateRestartPoint(), DataChecksumsEnabled(), digestControlFile(), do_pg_backup_start(), do_pg_backup_stop(), get_controlfile_by_exact_path(), GetActiveWalLevelOnStandby(), GetMockAuthenticationNonce(), GetOldestRestartPoint(), GetSystemIdentifier(), InitControlFile(), InitWalRecovery(), LocalProcessControlFile(), LogCheckpointEnd(), main(), pg_control_checkpoint(), pg_control_init(), pg_control_recovery(), pg_control_system(), ReachedEndOfBackup(), ReadControlFile(), StartupXLOG(), SwitchIntoArchiveRecovery(), update_controlfile(), UpdateControlFile(), UpdateMinRecoveryPoint(), WriteControlFile(), xlog_redo(), XLogNeedsFlush(), XLogReportParameters(), and XLOGShmemInit().

◆ doPageWrites

bool doPageWrites

static

Definition at line 284 of file xlog.c.

Referenced by GetFullPageWriteInfo(), StartupXLOG(), XLogCheckBufferNeedsBackup(), XLogInsert(), XLogInsertRecord(), and XLogRecordAssemble().

◆ EnableHotStandby

bool EnableHotStandby = false

Definition at line 119 of file xlog.c.

Referenced by BackendInitialize(), CheckRequiredParameterValues(), CreateRestartPoint(), CreateSharedProcArray(), ProcArrayShmemSize(), process_pm_pmsignal(), StartupXLOG(), and validateRecoveryParameters().

◆ fullPageWrites

bool fullPageWrites = true

Definition at line 120 of file xlog.c.

Referenced by BootStrapXLOG(), and UpdateFullPageWrites().

◆ holdingAllLocks

bool holdingAllLocks = false

static

Definition at line 645 of file xlog.c.

Referenced by WALInsertLockAcquireExclusive(), WALInsertLockRelease(), WALInsertLockUpdateInsertingAt(), and XLogInsertRecord().

◆ lastFullPageWrites

bool lastFullPageWrites

static

Definition at line 215 of file xlog.c.

Referenced by StartupXLOG(), and xlog_redo().

◆ LocalMinRecoveryPoint

XLogRecPtr LocalMinRecoveryPoint

static

Definition at line 639 of file xlog.c.

Referenced by CreateRestartPoint(), StartupXLOG(), SwitchIntoArchiveRecovery(), UpdateMinRecoveryPoint(), xlog_redo(), and XLogNeedsFlush().

◆ LocalMinRecoveryPointTLI

TimeLineID LocalMinRecoveryPointTLI

static

Definition at line 640 of file xlog.c.

Referenced by CreateRestartPoint(), StartupXLOG(), SwitchIntoArchiveRecovery(), UpdateMinRecoveryPoint(), xlog_redo(), and XLogNeedsFlush().

◆ LocalRecoveryInProgress

bool LocalRecoveryInProgress = true

static

Definition at line 222 of file xlog.c.

Referenced by RecoveryInProgress().

◆ LocalXLogInsertAllowed

int LocalXLogInsertAllowed = -1

static

Definition at line 234 of file xlog.c.

Referenced by CreateCheckPoint(), LocalSetXLogInsertAllowed(), and XLogInsertAllowed().

◆ log_checkpoints

bool log_checkpoints = true

Definition at line 127 of file xlog.c.

Referenced by CheckPointTwoPhase(), CreateCheckPoint(), CreateRestartPoint(), LogCheckpointEnd(), and ProcessSyncRequests().

◆ LogwrtResult

XLogwrtResult LogwrtResult = {0, 0}

static

Definition at line 605 of file xlog.c.

Referenced by AdvanceXLInsertBuffer(), GetFlushRecPtr(), GetXLogWriteRecPtr(), StartupXLOG(), XLogBackgroundFlush(), XLogFlush(), XLogInsertRecord(), XLogNeedsFlush(), XLogSetAsyncXactLSN(), and XLogWrite().

◆ max_slot_wal_keep_size_mb

int max_slot_wal_keep_size_mb = -1

Definition at line 133 of file xlog.c.

Referenced by KeepLogSeg(), and pg_get_replication_slots().

◆ max_wal_size_mb

int max_wal_size_mb = 1024

Definition at line 112 of file xlog.c.

Referenced by assign_max_wal_size(), CalculateCheckpointSegments(), GetWALAvailability(), ReadControlFile(), and XLOGfileslop().

◆ min_wal_size_mb

int min_wal_size_mb = 80

Definition at line 113 of file xlog.c.

Referenced by ReadControlFile(), and XLOGfileslop().

◆ MyLockNo

int MyLockNo = 0

static

Definition at line 644 of file xlog.c.

Referenced by WALInsertLockAcquire(), WALInsertLockRelease(), WALInsertLockUpdateInsertingAt(), and XLogInsertRecord().

◆ openLogFile

int openLogFile = -1

static

Definition at line 628 of file xlog.c.

Referenced by assign_wal_sync_method(), BootStrapXLOG(), XLogBackgroundFlush(), XLogFileClose(), and XLogWrite().

◆ openLogSegNo

XLogSegNo openLogSegNo = 0

static

Definition at line 629 of file xlog.c.

Referenced by assign_wal_sync_method(), XLogBackgroundFlush(), XLogFileClose(), and XLogWrite().

◆ openLogTLI

TimeLineID openLogTLI = 0

static

Definition at line 630 of file xlog.c.

Referenced by assign_wal_sync_method(), BootStrapXLOG(), XLogFileClose(), and XLogWrite().

◆ PrevCheckPointDistance

double PrevCheckPointDistance = 0

static

Definition at line 158 of file xlog.c.

Referenced by LogCheckpointEnd(), and UpdateCheckPointDistanceEstimate().

◆ ProcLastRecPtr

XLogRecPtr ProcLastRecPtr = InvalidXLogRecPtr

Definition at line 251 of file xlog.c.

Referenced by CreateCheckPoint(), CreateOverwriteContrecordRecord(), EndPrepare(), and XLogInsertRecord().

◆ RedoRecPtr

XLogRecPtr RedoRecPtr

static

Definition at line 271 of file xlog.c.

Referenced by CreateCheckPoint(), CreateRestartPoint(), GetFullPageWriteInfo(), GetRedoRecPtr(), StartupXLOG(), XLogCheckBufferNeedsBackup(), XLogCheckpointNeeded(), XLogInsert(), XLogInsertRecord(), XLogRecordAssemble(), and XLogSaveBufferForHint().

◆ sessionBackupState

SessionBackupState sessionBackupState = SESSION_BACKUP_NONE

static

Definition at line 389 of file xlog.c.

Referenced by do_pg_abort_backup(), do_pg_backup_start(), do_pg_backup_stop(), and get_backup_status().

◆ track_wal_io_timing

bool track_wal_io_timing = false

Definition at line 135 of file xlog.c.

Referenced by issue_xlog_fsync(), and XLogWrite().

◆ updateMinRecoveryPoint

bool updateMinRecoveryPoint = true

static

Definition at line 641 of file xlog.c.

Referenced by SwitchIntoArchiveRecovery(), UpdateMinRecoveryPoint(), and XLogNeedsFlush().

◆ UsableBytesInSegment

int UsableBytesInSegment

static

Definition at line 599 of file xlog.c.

Referenced by ReadControlFile(), XLogBytePosToEndRecPtr(), XLogBytePosToRecPtr(), and XLogRecPtrToBytePos().

◆ wal_compression

int wal_compression = WAL_COMPRESSION_NONE

Definition at line 122 of file xlog.c.

Referenced by XLogCompressBackupBlock(), and XLogRecordAssemble().

◆ wal_consistency_checking

bool* wal_consistency_checking = NULL

Definition at line 124 of file xlog.c.

Referenced by assign_wal_consistency_checking(), and XLogRecordAssemble().

◆ wal_consistency_checking_string

char* wal_consistency_checking_string = NULL

Definition at line 123 of file xlog.c.

Referenced by InitializeWalConsistencyChecking().

◆ wal_decode_buffer_size

int wal_decode_buffer_size = 512 * 1024

Definition at line 134 of file xlog.c.

Referenced by InitWalRecovery().

◆ wal_init_zero

bool wal_init_zero = true

Definition at line 125 of file xlog.c.

Referenced by XLogFileInitInternal().

◆ wal_keep_size_mb

int wal_keep_size_mb = 0

Definition at line 114 of file xlog.c.

Referenced by KeepLogSeg(), and pg_get_replication_slots().

◆ wal_level

◆ wal_log_hints

bool wal_log_hints = false

Definition at line 121 of file xlog.c.

Referenced by InitControlFile(), and XLogReportParameters().

◆ wal_recycle

bool wal_recycle = true

Definition at line 126 of file xlog.c.

Referenced by RemoveXlogFile().

◆ wal_retrieve_retry_interval

int wal_retrieve_retry_interval = 5000

Definition at line 132 of file xlog.c.

Referenced by ApplyLauncherMain(), process_syncing_tables_for_apply(), and WaitForWALToBecomeAvailable().

◆ wal_segment_size

int wal_segment_size = DEFAULT_XLOG_SEG_SIZE

Definition at line 141 of file xlog.c.

Referenced by AdvanceXLInsertBuffer(), assign_wal_sync_method(), BootStrapXLOG(), build_backup_content(), CalculateCheckpointSegments(), CheckArchiveTimeout(), CheckXLogRemoved(), CleanupAfterArchiveRecovery(), copy_replication_slot(), CopyXLogRecordToWAL(), CreateCheckPoint(), CreateOverwriteContrecordRecord(), CreateRestartPoint(), do_pg_backup_stop(), ExecuteRecoveryCommand(), FinishWalRecovery(), GetOldestUnsummarizedLSN(), GetWALAvailability(), GetXLogBuffer(), InitWalRecovery(), InitXLogReaderState(), InstallXLogFileSegment(), InvalidateObsoleteReplicationSlots(), IsCheckpointOnSchedule(), issue_xlog_fsync(), KeepLogSeg(), MaybeRemoveOldWalSummaries(), perform_base_backup(), pg_control_checkpoint(), pg_get_replication_slots(), pg_split_walfile_name(), pg_walfile_name(), pg_walfile_name_offset(), PreallocXlogFiles(), ReadControlFile(), ReadRecord(), RemoveNonParentXlogFiles(), RemoveOldXlogFiles(), ReorderBufferRestoreChanges(), ReorderBufferRestoreCleanup(), ReorderBufferSerializedPath(), ReorderBufferSerializeTXN(), ReplicationSlotReserveWal(), RequestXLogStreaming(), reserve_wal_for_local_slot(), ReserveXLogSwitch(), RestoreArchivedFile(), StartReplication(), StartupDecodingContext(), SummarizeWAL(), UpdateLastRemovedPtr(), WALReadRaiseError(), WalReceiverMain(), WalSndSegmentOpen(), WriteControlFile(), XLogArchiveNotifySeg(), XLogBackgroundFlush(), XLogBytePosToEndRecPtr(), XLogBytePosToRecPtr(), XLogCheckpointNeeded(), XLOGChooseNumBuffers(), XLogFileClose(), XLogFileCopy(), XLogFileInitInternal(), XLogFileOpen(), XLogFileRead(), XLogFileReadAnyTLI(), XLOGfileslop(), XLogGetOldestSegno(), XLogInitNewTimeline(), XLogInsertRecord(), XLogPageRead(), XLogReaderAllocate(), XlogReadTwoPhaseData(), XLogRecPtrToBytePos(), XLogWalRcvClose(), XLogWalRcvWrite(), and XLogWrite().

◆ wal_sync_method

int wal_sync_method = DEFAULT_WAL_SYNC_METHOD

Definition at line 128 of file xlog.c.

Referenced by assign_wal_sync_method(), issue_xlog_fsync(), pg_fsync(), XLogFileInit(), XLogFileInitInternal(), XLogFileOpen(), and XLogWrite().

◆ wal_sync_method_options

const struct config_enum_entry wal_sync_method_options[]

Initial value:

= {
    {"fsync", WAL_SYNC_METHOD_FSYNC, false},
 
    {"fdatasync", WAL_SYNC_METHOD_FDATASYNC, false},
 
    {NULL, 0, false}
}

Definition at line 164 of file xlog.c.

◆ WALInsertLocks

WALInsertLockPadded* WALInsertLocks = NULL

static

Definition at line 562 of file xlog.c.

Referenced by GetLastImportantRecPtr(), WaitXLogInsertionsToFinish(), WALInsertLockAcquire(), WALInsertLockAcquireExclusive(), WALInsertLockRelease(), WALInsertLockUpdateInsertingAt(), XLogInsertRecord(), and XLOGShmemInit().

◆ XactLastCommitEnd

XLogRecPtr XactLastCommitEnd = InvalidXLogRecPtr

Definition at line 253 of file xlog.c.

Referenced by apply_handle_commit_internal(), apply_handle_commit_prepared(), apply_handle_prepare(), apply_handle_rollback_prepared(), apply_handle_stream_commit(), apply_handle_stream_prepare(), and RecordTransactionCommit().

◆ XactLastRecEnd

XLogRecPtr XactLastRecEnd = InvalidXLogRecPtr

Definition at line 252 of file xlog.c.

Referenced by AbortTransaction(), CommitTransaction(), dropdb(), PrepareTransaction(), RecordTransactionAbort(), RecordTransactionAbortPrepared(), RecordTransactionCommit(), RecordTransactionCommitPrepared(), WaitForParallelWorkersToFinish(), and XLogInsertRecord().

◆ XLogArchiveCommand

char* XLogArchiveCommand = NULL

Definition at line 118 of file xlog.c.

Referenced by HandlePgArchInterrupts(), LoadArchiveLibrary(), shell_archive_configured(), shell_archive_file(), and show_archive_command().

◆ XLogArchiveMode

int XLogArchiveMode = ARCHIVE_MODE_OFF

Definition at line 117 of file xlog.c.

Referenced by KeepFileRestoredFromArchive(), PostmasterMain(), WalRcvFetchTimeLineHistoryFiles(), WalReceiverMain(), and XLogWalRcvClose().

◆ XLogArchiveTimeout

int XLogArchiveTimeout = 0

Definition at line 116 of file xlog.c.

Referenced by CheckArchiveTimeout(), and CheckpointerMain().

◆ XLOGbuffers

int XLOGbuffers = -1

Definition at line 115 of file xlog.c.

Referenced by check_wal_buffers(), XLOGShmemInit(), and XLOGShmemSize().

◆ XLogCtl

XLogCtlData* XLogCtl = NULL

static

Definition at line 559 of file xlog.c.

Referenced by AdvanceXLInsertBuffer(), CheckXLogRemoved(), CreateCheckPoint(), CreateEndOfRecoveryRecord(), CreateRestartPoint(), do_pg_abort_backup(), do_pg_backup_start(), do_pg_backup_stop(), GetFakeLSNForUnloggedRel(), GetFlushRecPtr(), GetInsertRecPtr(), GetLastSegSwitchData(), GetRecoveryState(), GetRedoRecPtr(), GetWALInsertionTimeLine(), GetXLogBuffer(), GetXLogInsertRecPtr(), InstallXLogFileSegment(), IsInstallXLogFileSegmentActive(), PreallocXlogFiles(), RecoveryInProgress(), RecoveryRestartPoint(), RemoveXlogFile(), ReserveXLogInsertLocation(), ReserveXLogSwitch(), SetInstallXLogFileSegmentActive(), SetWalWriterSleeping(), StartupXLOG(), SwitchIntoArchiveRecovery(), UpdateFullPageWrites(), UpdateLastRemovedPtr(), WaitXLogInsertionsToFinish(), WALReadFromBuffers(), xlog_redo(), XLogBackgroundFlush(), XLogFlush(), XLogGetLastRemovedSegno(), XLogGetReplicationSlotMinimumLSN(), XLogInsertRecord(), XLogSetAsyncXactLSN(), XLogSetReplicationSlotMinimumLSN(), XLOGShmemInit(), XLogShutdownWalRcv(), and XLogWrite().

Data Structures

Macros

Typedefs

Enumerations

Functions

Variables

Macro Definition Documentation

◆ BootstrapTimeLineID

◆ ConvertToXSegs

◆ INSERT_FREESPACE

◆ NextBufIdx

◆ NUM_XLOGINSERT_LOCKS

◆ RefreshXLogWriteResult

◆ UsableBytesInPage

◆ XLogRecPtrToBufIdx

Typedef Documentation

◆ WALInsertLockPadded

◆ XLogCtlData

◆ XLogCtlInsert

◆ XLogwrtResult

◆ XLogwrtRqst

Enumeration Type Documentation

◆ WalInsertClass

Function Documentation

◆ AdvanceXLInsertBuffer()

◆ assign_checkpoint_completion_target()

◆ assign_max_wal_size()

◆ assign_wal_consistency_checking()

◆ assign_wal_sync_method()

◆ BootStrapXLOG()

◆ CalculateCheckpointSegments()

◆ check_max_slot_wal_keep_size()

◆ check_wal_buffers()

◆ check_wal_consistency_checking()

◆ check_wal_segment_size()

◆ CheckPointGuts()

◆ CheckRequiredParameterValues()

◆ CheckXLogRemoved()

◆ CleanupAfterArchiveRecovery()

◆ CleanupBackupHistory()

◆ CopyXLogRecordToWAL()

◆ CreateCheckPoint()

◆ CreateEndOfRecoveryRecord()

◆ CreateOverwriteContrecordRecord()

◆ CreateRestartPoint()

◆ DataChecksumsEnabled()

◆ do_pg_abort_backup()

◆ do_pg_backup_start()

◆ do_pg_backup_stop()

◆ get_backup_status()

◆ get_sync_bit()

◆ GetActiveWalLevelOnStandby()

◆ GetFakeLSNForUnloggedRel()

◆ GetFlushRecPtr()

◆ GetFullPageWriteInfo()

◆ GetInsertRecPtr()

◆ GetLastImportantRecPtr()

◆ GetLastSegSwitchData()

◆ GetMockAuthenticationNonce()

◆ GetOldestRestartPoint()

◆ GetRecoveryState()

◆ GetRedoRecPtr()

◆ GetSystemIdentifier()

◆ GetWALAvailability()

◆ GetWALInsertionTimeLine()

◆ GetXLogBuffer()

◆ GetXLogInsertRecPtr()

◆ GetXLogWriteRecPtr()

◆ InitControlFile()

◆ InitializeWalConsistencyChecking()

◆ InstallXLogFileSegment()

◆ IsInstallXLogFileSegmentActive()

◆ issue_xlog_fsync()

◆ KeepLogSeg()

◆ LocalProcessControlFile()

◆ LocalSetXLogInsertAllowed()

◆ LogCheckpointEnd()

◆ LogCheckpointStart()

◆ PerformRecoveryXLogAction()

◆ PreallocXlogFiles()