fd.c

关系型数据库 Postgresql 6.5.2

/*-------------------------------------------------------------------------
 *
 * fd.c
 *	  Virtual file descriptor code.
 *
 * Copyright (c) 1994, Regents of the University of California
 *
 * IDENTIFICATION
 *	  $Id: fd.c,v 1.43 1999/07/08 02:46:39 momjian Exp $
 *
 * NOTES:
 *
 * This code manages a cache of 'virtual' file descriptors (VFDs).
 * The server opens many file descriptors for a variety of reasons,
 * including base tables, scratch files (e.g., sort and hash spool
 * files), and random calls to C library routines like system(3); it
 * is quite easy to exceed system limits on the number of open files a
 * single process can have.  (This is around 256 on many modern
 * operating systems, but can be as low as 32 on others.)
 *
 * VFDs are managed as an LRU pool, with actual OS file descriptors
 * being opened and closed as needed.  Obviously, if a routine is
 * opened using these interfaces, all subsequent operations must also
 * be through these interfaces (the File type is not a real file
 * descriptor).
 *
 * For this scheme to work, most (if not all) routines throughout the
 * server should use these interfaces instead of calling the C library
 * routines (e.g., open(2) and fopen(3)) themselves.  Otherwise, we
 * may find ourselves short of real file descriptors anyway.
 *
 * This file used to contain a bunch of stuff to support RAID levels 0
 * (jbod), 1 (duplex) and 5 (xor parity).  That stuff is all gone
 * because the parallel query processing code that called it is all
 * gone.  If you really need it you could get it from the original
 * POSTGRES source.
 *-------------------------------------------------------------------------
 */

#include <stdio.h>
#include <sys/types.h>
#include <sys/file.h>
#include <sys/param.h>
#include <sys/stat.h>
#include <errno.h>
#include <string.h>
#include <unistd.h>
#include <fcntl.h>

#include "postgres.h"
#include "miscadmin.h"			/* for DataDir */
#include "utils/palloc.h"
#include "storage/fd.h"
#include "utils/elog.h"

/*
 * Problem: Postgres does a system(ld...) to do dynamic loading.
 * This will open several extra files in addition to those used by
 * Postgres.  We need to guarantee that there are file descriptors free
 * for ld to use.
 *
 * The current solution is to limit the number of file descriptors
 * that this code will allocate at one time: it leaves RESERVE_FOR_LD free.
 *
 * (Even though most dynamic loaders now use dlopen(3) or the
 * equivalent, the OS must still open several files to perform the
 * dynamic loading.  Keep this here.)
 */
#ifndef RESERVE_FOR_LD
#define RESERVE_FOR_LD	10
#endif

/*
 * We need to ensure that we have at least some file descriptors
 * available to postgreSQL after we've reserved the ones for LD,
 * so we set that value here.
 *
 * I think 10 is an appropriate value so that's what it'll be
 * for now.
 */
#ifndef FD_MINFREE
#define FD_MINFREE 10
#endif

/* Debugging.... */

#ifdef FDDEBUG
#define DO_DB(A) A
#else
#define DO_DB(A)				/* A */
#endif

#define VFD_CLOSED (-1)

#define FileIsValid(file) \
	((file) > 0 && (file) < SizeVfdCache && VfdCache[file].fileName != NULL)

#define FileIsNotOpen(file) (VfdCache[file].fd == VFD_CLOSED)

typedef struct vfd
{
	signed short fd;			/* current FD, or VFD_CLOSED if none */
	unsigned short fdstate;		/* bitflags for VFD's state */

/* these are the assigned bits in fdstate: */
#define FD_DIRTY		(1 << 0)/* written to, but not yet fsync'd */
#define FD_TEMPORARY	(1 << 1)/* should be unlinked when closed */

	File		nextFree;		/* link to next free VFD, if in freelist */
	File		lruMoreRecently;/* doubly linked recency-of-use list */
	File		lruLessRecently;
	long		seekPos;		/* current logical file position */
	char	   *fileName;		/* name of file, or NULL for unused VFD */
	/* NB: fileName is malloc'd, and must be free'd when closing the VFD */
	int			fileFlags;		/* open(2) flags for opening the file */
	int			fileMode;		/* mode to pass to open(2) */
} Vfd;

/*
 * Virtual File Descriptor array pointer and size.	This grows as
 * needed.	'File' values are indexes into this array.
 * Note that VfdCache[0] is not a usable VFD, just a list header.
 */
static Vfd *VfdCache;
static Size SizeVfdCache = 0;

/*
 * Number of file descriptors known to be in use by VFD entries.
 */
static int	nfile = 0;

/*
 * List of stdio FILEs opened with AllocateFile.
 *
 * Since we don't want to encourage heavy use of AllocateFile, it seems
 * OK to put a pretty small maximum limit on the number of simultaneously
 * allocated files.
 */
#define MAX_ALLOCATED_FILES  32

static int	numAllocatedFiles = 0;
static FILE *allocatedFiles[MAX_ALLOCATED_FILES];

/*
 * Number of temporary files opened during the current transaction;
 * this is used in generation of tempfile names.
 */
static long tempFileCounter = 0;


/*--------------------
 *
 * Private Routines
 *
 * Delete		   - delete a file from the Lru ring
 * LruDelete	   - remove a file from the Lru ring and close its FD
 * Insert		   - put a file at the front of the Lru ring
 * LruInsert	   - put a file at the front of the Lru ring and open it
 * ReleaseLruFile  - Release an fd by closing the last entry in the Lru ring
 * AllocateVfd	   - grab a free (or new) file record (from VfdArray)
 * FreeVfd		   - free a file record
 *
 * The Least Recently Used ring is a doubly linked list that begins and
 * ends on element zero.  Element zero is special -- it doesn't represent
 * a file and its "fd" field always == VFD_CLOSED.	Element zero is just an
 * anchor that shows us the beginning/end of the ring.
 * Only VFD elements that are currently really open (have an FD assigned) are
 * in the Lru ring.  Elements that are "virtually" open can be recognized
 * by having a non-null fileName field.
 *
 * example:
 *
 *	   /--less----\				   /---------\
 *	   v		   \			  v			  \
 *	 #0 --more---> LeastRecentlyUsed --more-\ \
 *	  ^\									| |
 *	   \\less--> MostRecentlyUsedFile	<---/ |
 *		\more---/					 \--less--/
 *
 *--------------------
 */
static void Delete(File file);
static void LruDelete(File file);
static void Insert(File file);
static int	LruInsert(File file);
static void ReleaseLruFile(void);
static File AllocateVfd(void);
static void FreeVfd(File file);

static int	FileAccess(File file);
static File fileNameOpenFile(FileName fileName, int fileFlags, int fileMode);
static char *filepath(char *filename);
static long pg_nofile(void);
static int	BufFileFlush(BufFile *file);

/*
 * pg_fsync --- same as fsync except does nothing if -F switch was given
 */
int
pg_fsync(int fd)
{
	return disableFsync ? 0 : fsync(fd);
}

/*
 * pg_nofile: determine number of filedescriptors that fd.c is allowed to use
 */
static long
pg_nofile(void)
{
	static long no_files = 0;

	if (no_files == 0)
	{
		/* need do this calculation only once */
#ifndef HAVE_SYSCONF
		no_files = (long) NOFILE;
#else
		no_files = sysconf(_SC_OPEN_MAX);
		if (no_files == -1)
		{
			elog(DEBUG, "pg_nofile: Unable to get _SC_OPEN_MAX using sysconf(); using %d", NOFILE);
			no_files = (long) NOFILE;
		}
#endif

		if ((no_files - RESERVE_FOR_LD) < FD_MINFREE)
			elog(FATAL, "pg_nofile: insufficient File Descriptors in postmaster to start backend (%ld).\n"
				 "                   O/S allows %ld, Postmaster reserves %d, We need %d (MIN) after that.",
				 no_files - RESERVE_FOR_LD, no_files, RESERVE_FOR_LD, FD_MINFREE);

		no_files -= RESERVE_FOR_LD;
	}

	return no_files;
}

#if defined(FDDEBUG)

static void
_dump_lru()
{
	int			mru = VfdCache[0].lruLessRecently;
	Vfd		   *vfdP = &VfdCache[mru];
	char		buf[2048];

	sprintf(buf, "LRU: MOST %d ", mru);
	while (mru != 0)
	{
		mru = vfdP->lruLessRecently;
		vfdP = &VfdCache[mru];
		sprintf(buf + strlen(buf), "%d ", mru);
	}
	sprintf(buf + strlen(buf), "LEAST");
	elog(DEBUG, buf);
}

#endif	 /* FDDEBUG */

static void
Delete(File file)
{
	Vfd		   *vfdP;

	Assert(file != 0);

	DO_DB(elog(DEBUG, "Delete %d (%s)",
			   file, VfdCache[file].fileName));
	DO_DB(_dump_lru());

	vfdP = &VfdCache[file];

	VfdCache[vfdP->lruLessRecently].lruMoreRecently = vfdP->lruMoreRecently;
	VfdCache[vfdP->lruMoreRecently].lruLessRecently = vfdP->lruLessRecently;

	DO_DB(_dump_lru());
}

static void
LruDelete(File file)
{
	Vfd		   *vfdP;
	int			returnValue;

	Assert(file != 0);

	DO_DB(elog(DEBUG, "LruDelete %d (%s)",
			   file, VfdCache[file].fileName));

	vfdP = &VfdCache[file];

	/* delete the vfd record from the LRU ring */
	Delete(file);

	/* save the seek position */
	vfdP->seekPos = (long) lseek(vfdP->fd, 0L, SEEK_CUR);
	Assert(vfdP->seekPos != -1);

	/* if we have written to the file, sync it */
	if (vfdP->fdstate & FD_DIRTY)
	{
		returnValue = pg_fsync(vfdP->fd);
		Assert(returnValue != -1);
		vfdP->fdstate &= ~FD_DIRTY;
	}

	/* close the file */
	returnValue = close(vfdP->fd);
	Assert(returnValue != -1);

	--nfile;
	vfdP->fd = VFD_CLOSED;
}

static void
Insert(File file)
{
	Vfd		   *vfdP;

	Assert(file != 0);

	DO_DB(elog(DEBUG, "Insert %d (%s)",
			   file, VfdCache[file].fileName));
	DO_DB(_dump_lru());

	vfdP = &VfdCache[file];

	vfdP->lruMoreRecently = 0;
	vfdP->lruLessRecently = VfdCache[0].lruLessRecently;
	VfdCache[0].lruLessRecently = file;
	VfdCache[vfdP->lruLessRecently].lruMoreRecently = file;

	DO_DB(_dump_lru());
}

static int
LruInsert(File file)
{
	Vfd		   *vfdP;
	int			returnValue;

	Assert(file != 0);

	DO_DB(elog(DEBUG, "LruInsert %d (%s)",
			   file, VfdCache[file].fileName));

	vfdP = &VfdCache[file];

	if (FileIsNotOpen(file))
	{

		while (nfile + numAllocatedFiles >= pg_nofile())
			ReleaseLruFile();

		/*
		 * The open could still fail for lack of file descriptors, eg due
		 * to overall system file table being full.  So, be prepared to
		 * release another FD if necessary...
		 */
tryAgain:
		vfdP->fd = open(vfdP->fileName, vfdP->fileFlags, vfdP->fileMode);
		if (vfdP->fd < 0 && (errno == EMFILE || errno == ENFILE))
		{
			errno = 0;
			ReleaseLruFile();
			goto tryAgain;
		}

		if (vfdP->fd < 0)
		{
			DO_DB(elog(DEBUG, "RE_OPEN FAILED: %d",
					   errno));
			return vfdP->fd;
		}
		else
		{
			DO_DB(elog(DEBUG, "RE_OPEN SUCCESS"));
			++nfile;
		}

		/* seek to the right position */
		if (vfdP->seekPos != 0L)
		{
			returnValue = lseek(vfdP->fd, vfdP->seekPos, SEEK_SET);
			Assert(returnValue != -1);
		}

		/* Update state as appropriate for re-open (needed?) */
		vfdP->fdstate &= ~FD_DIRTY;
	}

	/*
	 * put it at the head of the Lru ring
	 */

	Insert(file);

	return 0;
}

static void
ReleaseLruFile()
{
	DO_DB(elog(DEBUG, "ReleaseLruFile. Opened %d", nfile));

	if (nfile <= 0)
		elog(FATAL, "ReleaseLruFile: No opened files - no one can be closed");

	/*
	 * There are opened files and so there should be at least one used vfd
	 * in the ring.
	 */
	Assert(VfdCache[0].lruMoreRecently != 0);
	LruDelete(VfdCache[0].lruMoreRecently);
}

static File
AllocateVfd()
{
	Index		i;
	File		file;

	DO_DB(elog(DEBUG, "AllocateVfd. Size %d", SizeVfdCache));

	if (SizeVfdCache == 0)
	{
		/* initialize header entry first time through */
		VfdCache = (Vfd *) malloc(sizeof(Vfd));
		Assert(VfdCache != NULL);
		MemSet((char *) &(VfdCache[0]), 0, sizeof(Vfd));
		VfdCache->fd = VFD_CLOSED;

		SizeVfdCache = 1;
	}

	if (VfdCache[0].nextFree == 0)
	{

		/*
		 * The free list is empty so it is time to increase the size of
		 * the array.  We choose to double it each time this happens.
		 * However, there's not much point in starting *real* small.
		 */
		Size		newCacheSize = SizeVfdCache * 2;

		if (newCacheSize < 32)
			newCacheSize = 32;

		VfdCache = (Vfd *) realloc(VfdCache, sizeof(Vfd) * newCacheSize);
		Assert(VfdCache != NULL);

		/*
		 * Initialize the new entries and link them into the free list.
		 */

		for (i = SizeVfdCache; i < newCacheSize; i++)
		{
			MemSet((char *) &(VfdCache[i]), 0, sizeof(Vfd));
			VfdCache[i].nextFree = i + 1;
			VfdCache[i].fd = VFD_CLOSED;
		}
		VfdCache[newCacheSize - 1].nextFree = 0;
		VfdCache[0].nextFree = SizeVfdCache;

		/*
		 * Record the new size
		 */

		SizeVfdCache = newCacheSize;
	}

	file = VfdCache[0].nextFree;

	VfdCache[0].nextFree = VfdCache[file].nextFree;

	return file;
}

static void
FreeVfd(File file)
{
	Vfd		   *vfdP = &VfdCache[file];

	DO_DB(elog(DEBUG, "FreeVfd: %d (%s)",
			   file, vfdP->fileName ? vfdP->fileName : ""));

	if (vfdP->fileName != NULL)
	{
		free(vfdP->fileName);
		vfdP->fileName = NULL;
	}

	vfdP->nextFree = VfdCache[0].nextFree;
	VfdCache[0].nextFree = file;
}

/* filepath()
 * Convert given pathname to absolute.
 * (Is this actually necessary, considering that we should be cd'd
 * into the database directory??)
 */
static char *
filepath(char *filename)
{
	char	   *buf;
	int			len;

	/* Not an absolute path name? Then fill in with database path... */
	if (*filename != SEP_CHAR)
	{
		len = strlen(DatabasePath) + strlen(filename) + 2;
		buf = (char *) palloc(len);
		sprintf(buf, "%s%c%s", DatabasePath, SEP_CHAR, filename);
	}
	else
	{
		buf = (char *) palloc(strlen(filename) + 1);
		strcpy(buf, filename);
	}

#ifdef FILEDEBUG
	printf("filepath: path is %s\n", buf);
#endif

	return buf;
}

static int
FileAccess(File file)
{
	int			returnValue;

	DO_DB(elog(DEBUG, "FileAccess %d (%s)",
			   file, VfdCache[file].fileName));

	/*
	 * Is the file open?  If not, open it and put it at the head of the
	 * LRU ring (possibly closing the least recently used file to get an
	 * FD).
	 */

	if (FileIsNotOpen(file))
	{
		returnValue = LruInsert(file);
		if (returnValue != 0)
			return returnValue;
	}
	else if (VfdCache[0].lruLessRecently != file)
	{

		/*
		 * We now know that the file is open and that it is not the last
		 * one accessed, so we need to move it to the head of the Lru
		 * ring.
		 */

		Delete(file);
		Insert(file);
	}

	return 0;
}

/*
 *	Called when we get a shared invalidation message on some relation.
 */
#ifdef NOT_USED
void
FileInvalidate(File file)
{
	Assert(FileIsValid(file));
	if (!FileIsNotOpen(file))
		LruDelete(file);
}

#endif

static File
fileNameOpenFile(FileName fileName,
				 int fileFlags,
				 int fileMode)
{
	File		file;
	Vfd		   *vfdP;

	if (fileName == NULL)
		elog(ERROR, "fileNameOpenFile: NULL fname");

	DO_DB(elog(DEBUG, "fileNameOpenFile: %s %x %o",
			   fileName, fileFlags, fileMode));

	file = AllocateVfd();
	vfdP = &VfdCache[file];

	while (nfile + numAllocatedFiles >= pg_nofile())
		ReleaseLruFile();

tryAgain:
	vfdP->fd = open(fileName, fileFlags, fileMode);
	if (vfdP->fd < 0 && (errno == EMFILE || errno == ENFILE))