mirrors/sqlite
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Refactoring the btree and pager routines into distinct two-phase commit

Browse Source 
routines.  We've always done a two-phase commit - this change is just
making that more apparent in the code. (CVS 3762)

FossilOrigin-Name: 66b3ad09ea657d25d48cb75ec2671ea2dc1b6005
This commit is contained in:
drh 2007-03-30 14:06:34 +00:00
parent b601861cc3
commit 80e35f4640
8 changed files with 255 additions and 229 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
manifest
View File

@ -1,5 +1,5 @@
C Make\syypMinor\savailable\sto\sthe\sstack\soverflow\scallbacks\sin\slemon\ngenerated\sparsers.\s\sThis\sdoes\snot\seffect\sSQLite.\s(CVS\s3761)
D 2007-03-30T13:35:06
C Refactoring\sthe\sbtree\sand\spager\sroutines\sinto\sdistinct\stwo-phase\scommit\nroutines.\s\sWe've\salways\sdone\sa\stwo-phase\scommit\s-\sthis\schange\sis\sjust\nmaking\sthat\smore\sapparent\sin\sthe\scode.\s(CVS\s3762)
D 2007-03-30T14:06:34
F Makefile.in 2f2c3bf69faf0ae7b8e8af4f94f1986849034530
F Makefile.linux-gcc 2d8574d1ba75f129aba2019f0b959db380a90935
F README 9c4e2d6706bdcc3efdd773ce752a8cdab4f90028
@ -58,8 +58,8 @@ F src/alter.c 2c79ec40f65e33deaf90ca493422c74586e481a3
F src/analyze.c 4bbf5ddf9680587c6d4917e02e378b6037be3651
F src/attach.c a16ada4a4654a0d126b8223ec9494ebb81bc5c3c
F src/auth.c 902f4722661c796b97f007d9606bd7529c02597f
F src/btree.c 9b41229e7ff4b0fe8556763ce2b466a90aaaa6a4
F src/btree.h 540dcbbf83435b77d4b6ef87f909c6cecad4dac9
F src/btree.c 931d8d6eb3e669d6e42305dce3f196bf071a53b6
F src/btree.h 9b2cc0d113c0bc2d37d244b9a394d56948c9acbf
F src/build.c ad3374b5409554e504300f77e1fbc6b4c106a57f
F src/callback.c 31d22b4919c7645cbcbb1591ce2453e8c677c558
F src/complete.c 7d1a44be8f37de125fcafd3d3a018690b3799675
@ -86,8 +86,8 @@ F src/os_unix.c 4291be23eec73d1ec04010ae702364b781b5f773
F src/os_unix.h 5768d56d28240d3fe4537fac08cc85e4fb52279e
F src/os_win.c c3a8403ea28bbb89d6507fa984c5919bd3fe7539
F src/os_win.h 41a946bea10f61c158ce8645e7646b29d44f122b
F src/pager.c b89ea0f592e499ee6d6cda10b84688f8e47a05ba
F src/pager.h f1b17bf848b3dce5d9afb2701186d3c9a8826f8c
F src/pager.c f9131543cc602202de8a436ca9207e4c28b3f41e
F src/pager.h e79a24cf200b8771366217f5bca414f5b7823f42
F src/parse.y 207ab04273ae13aa4a729b96008d294d5f334ab3
F src/pragma.c 8fd4f98822007a8d2c34e235ad3c35f1d77b3e51
F src/prepare.c 37207b2b2ccb41d379b01dd62231686bcc48ef1f
@ -102,7 +102,7 @@ F src/sqliteInt.h c8d0e5ce27a862836de70fc3eadc1e65cea7e3d8
F src/table.c 6d0da66dde26ee75614ed8f584a1996467088d06
F src/tclsqlite.c a8d1166319db5d505b25ac6a9820162afe63fc8a
F src/test1.c 439bba8da10fbc61c731019cf2894e6057578878
F src/test2.c dc48c84ce68b3bc2f2d01871d709f20dc77003b0
F src/test2.c 710a7252e22a8c690136a2b609e90fdad2697f35
F src/test3.c 65f92247cf8592854e9bf5115b3fb711f8b33280
F src/test4.c 8b784cd82de158a2317cb4ac4bc86f91ad315e25
F src/test5.c 7162f8526affb771c4ed256826eee7bb9eca265f
@ -127,7 +127,7 @@ F src/vdbe.c d5a4fc1e18af2a2873d022450c4a776f6c17bac3
F src/vdbe.h 0025259af1939fb264a545816c69e4b5b8d52691
F src/vdbeInt.h 4b19fd8febad3fd14c4c97adaefc06754d323132
F src/vdbeapi.c 7ac14f2e3b2283dcd35f5edefd31a8342cff783c
F src/vdbeaux.c d768c22c8aa0fbf9bf1408b4fc9ed0b38b1146ca
F src/vdbeaux.c 5efa6d0f6096f213fbdffa0b1bf0a4c8ae40d145
F src/vdbefifo.c 3ca8049c561d5d67cbcb94dc909ae9bb68c0bf8f
F src/vdbemem.c d3696b4b0e5f32272659816cdfa2348c650b1ba0
F src/vtab.c 7fbda947e28cbe7adb3ba752a76ca9ef29936750
@ -447,7 +447,7 @@ F www/tclsqlite.tcl bb0d1357328a42b1993d78573e587c6dcbc964b9
F www/vdbe.tcl 87a31ace769f20d3627a64fa1fade7fed47b90d0
F www/version3.tcl 890248cf7b70e60c383b0e84d77d5132b3ead42b
F www/whentouse.tcl 97e2b5cd296f7d8057e11f44427dea8a4c2db513
P 339941d83ae397d69084f41483afb1ea44d44967
R b47b2f2fe5cfddcd0105ea435c37f3d5
P 70c8c7e2ce5213778e63c200a6637849920deea6
R a5a112b3409ee86a088223323643c5e6
U drh
Z 511499bbee13f7559f6a7e8fb29eeb3d
Z 49017f06f92594291d8addf7b78eb7c9

2
manifest.uuid
View File

@ -1 +1 @@
70c8c7e2ce5213778e63c200a6637849920deea6
66b3ad09ea657d25d48cb75ec2671ea2dc1b6005

106
src/btree.c
View File

@ -9,7 +9,7 @@
** May you share freely, never taking more than you give.
**
*************************************************************************
** $Id: btree.c,v 1.346 2007/03/30 11:12:08 drh Exp $
** $Id: btree.c,v 1.347 2007/03/30 14:06:34 drh Exp $
**
** This file implements a external (disk-based) database using BTrees.
** For a detailed discussion of BTrees, refer to
@ -2407,6 +2407,50 @@ autovacuum_out:
}
#endif
/*
** This routine does the first phase of a two-phase commit. This routine
** causes a rollback journal to be created (if it does not already exist)
** and populated with enough information so that if a power loss occurs
** the database can be restored to its original state by playing back
** the journal. Then the contents of the journal are flushed out to
** the disk. After the journal is safely on oxide, the changes to the
** database are written into the database file and flushed to oxide.
** At the end of this call, the rollback journal still exists on the
** disk and we are still holding all locks, so the transaction has not
** committed. See sqlite3BtreeCommit() for the second phase of the
** commit process.
**
** This call is a no-op if no write-transaction is currently active on pBt.
**
** Otherwise, sync the database file for the btree pBt. zMaster points to
** the name of a master journal file that should be written into the
** individual journal file, or is NULL, indicating no master journal file
** (single database transaction).
**
** When this is called, the master journal should already have been
** created, populated with this journal pointer and synced to disk.
**
** Once this is routine has returned, the only thing required to commit
** the write-transaction for this database file is to delete the journal.
*/
int sqlite3BtreeCommitPhaseOne(Btree *p, const char *zMaster){
int rc = SQLITE_OK;
if( p->inTrans==TRANS_WRITE ){
BtShared *pBt = p->pBt;
Pgno nTrunc = 0;
#ifndef SQLITE_OMIT_AUTOVACUUM
if( pBt->autoVacuum ){
rc = autoVacuumCommit(pBt, &nTrunc);
if( rc!=SQLITE_OK ){
return rc;
}
}
#endif
rc = sqlite3PagerCommitPhaseOne(pBt->pPager, zMaster, nTrunc);
}
return rc;
}
/*
** Commit the transaction currently in progress.
**
@ -2421,7 +2465,7 @@ autovacuum_out:
** This will release the write lock on the database file. If there
** are no active cursors, it also releases the read lock.
*/
int sqlite3BtreeCommit(Btree *p){
int sqlite3BtreeCommitPhaseTwo(Btree *p){
BtShared *pBt = p->pBt;
btreeIntegrity(p);
@ -2433,7 +2477,7 @@ int sqlite3BtreeCommit(Btree *p){
int rc;
assert( pBt->inTransaction==TRANS_WRITE );
assert( pBt->nTransaction>0 );
rc = sqlite3PagerCommit(pBt->pPager);
rc = sqlite3PagerCommitPhaseTwo(pBt->pPager);
if( rc!=SQLITE_OK ){
return rc;
}
@ -2464,6 +2508,18 @@ int sqlite3BtreeCommit(Btree *p){
return SQLITE_OK;
}
/*
** Do both phases of a commit.
*/
int sqlite3BtreeCommit(Btree *p){
int rc;
rc = sqlite3BtreeCommitPhaseOne(p, 0);
if( rc==SQLITE_OK ){
rc = sqlite3BtreeCommitPhaseTwo(p);
}
return rc;
}
#ifndef NDEBUG
/*
** Return the number of write-cursors open on this handle. This is for use
@ -6522,50 +6578,6 @@ int sqlite3BtreeIsInReadTrans(Btree *p){
return (p && (p->inTrans!=TRANS_NONE));
}
/*
** This routine does the first phase of a 2-phase commit. This routine
** causes a rollback journal to be created (if it does not already exist)
** and populated with enough information so that if a power loss occurs
** the database can be restored to its original state by playing back
** the journal. Then the contents of the journal are flushed out to
** the disk. After the journal is safely on oxide, the changes to the
** database are written into the database file and flushed to oxide.
** At the end of this call, the rollback journal still exists on the
** disk and we are still holding all locks, so the transaction has not
** committed. See sqlite3BtreeCommit() for the second phase of the
** commit process.
**
** This call is a no-op if no write-transaction is currently active on pBt.
**
** Otherwise, sync the database file for the btree pBt. zMaster points to
** the name of a master journal file that should be written into the
** individual journal file, or is NULL, indicating no master journal file
** (single database transaction).
**
** When this is called, the master journal should already have been
** created, populated with this journal pointer and synced to disk.
**
** Once this is routine has returned, the only thing required to commit
** the write-transaction for this database file is to delete the journal.
*/
int sqlite3BtreeSync(Btree *p, const char *zMaster){
int rc = SQLITE_OK;
if( p->inTrans==TRANS_WRITE ){
BtShared *pBt = p->pBt;
Pgno nTrunc = 0;
#ifndef SQLITE_OMIT_AUTOVACUUM
if( pBt->autoVacuum ){
rc = autoVacuumCommit(pBt, &nTrunc);
if( rc!=SQLITE_OK ){
return rc;
}
}
#endif
rc = sqlite3PagerSync(pBt->pPager, zMaster, nTrunc);
}
return rc;
}
/*
** This function returns a pointer to a blob of memory associated with
** a single shared-btree. The memory is used by client code for it's own

5
src/btree.h
View File

@ -13,7 +13,7 @@
** subsystem. See comments in the source code for a detailed description
** of what each interface routine does.
**
** @(#) $Id: btree.h,v 1.73 2007/03/29 05:51:49 drh Exp $
** @(#) $Id: btree.h,v 1.74 2007/03/30 14:06:34 drh Exp $
*/
#ifndef _BTREE_H_
#define _BTREE_H_
@ -67,6 +67,8 @@ int sqlite3BtreeGetReserve(Btree*);
int sqlite3BtreeSetAutoVacuum(Btree *, int);
int sqlite3BtreeGetAutoVacuum(Btree *);
int sqlite3BtreeBeginTrans(Btree*,int);
int sqlite3BtreeCommitPhaseOne(Btree*, const char *zMaster);
int sqlite3BtreeCommitPhaseTwo(Btree*);
int sqlite3BtreeCommit(Btree*);
int sqlite3BtreeRollback(Btree*);
int sqlite3BtreeBeginStmt(Btree*);
@ -76,7 +78,6 @@ int sqlite3BtreeCreateTable(Btree*, int*, int flags);
int sqlite3BtreeIsInTrans(Btree*);
int sqlite3BtreeIsInStmt(Btree*);
int sqlite3BtreeIsInReadTrans(Btree*);
int sqlite3BtreeSync(Btree*, const char *zMaster);
void *sqlite3BtreeSchema(Btree *, int, void(*)(void *));
int sqlite3BtreeSchemaLocked(Btree *);
int sqlite3BtreeLockTable(Btree *, int, u8);

298
src/pager.c
View File

@ -18,7 +18,7 @@
** file simultaneously, or one process from reading the database while
** another is writing.
**
** @(#) $Id: pager.c,v 1.306 2007/03/29 18:19:52 drh Exp $
** @(#) $Id: pager.c,v 1.307 2007/03/30 14:06:34 drh Exp $
*/
#ifndef SQLITE_OMIT_DISKIO
#include "sqliteInt.h"
@ -160,7 +160,7 @@ struct PgHdr {
u8 inStmt; /* TRUE if in the statement subjournal */
u8 dirty; /* TRUE if we need to write back changes */
u8 needSync; /* Sync journal before writing this page */
u8 alwaysRollback; /* Disable dont_rollback() for this page */
u8 alwaysRollback; /* Disable DontRollback() for this page */
short int nRef; /* Number of users of this page */
PgHdr *pDirty, *pPrevDirty; /* Dirty pages */
u32 notUsed; /* Buffer space */
@ -236,9 +236,12 @@ struct Pager {
u8 readOnly; /* True for a read-only database */
u8 needSync; /* True if an fsync() is needed on the journal */
u8 dirtyCache; /* True if cached pages have changed */
u8 alwaysRollback; /* Disable dont_rollback() for all pages */
u8 alwaysRollback; /* Disable DontRollback() for all pages */
u8 memDb; /* True to inhibit all file I/O */
u8 setMaster; /* True if a m-j name has been written to jrnl */
u8 doNotSync; /* Boolean. While true, do not spill the cache */
u8 exclusiveMode; /* Boolean. True if locking_mode==EXCLUSIVE */
u8 changeCountDone; /* Set after incrementing the change-counter */
int errCode; /* One of several kinds of errors */
int dbSize; /* Number of pages in the file */
int origDbSize; /* dbSize before the current change */
@ -286,9 +289,6 @@ struct Pager {
#endif
char *pTmpSpace; /* Pager.pageSize bytes of space for tmp use */
u32 iChangeCount; /* Db change-counter for which cache is valid */
u8 doNotSync; /* Boolean. While true, do not spill the cache */
u8 exclusiveMode; /* Boolean. True if locking_mode==EXCLUSIVE */
u8 changeCountDone; /* Set after incrementing the change-counter */
};
/*
@ -910,16 +910,23 @@ static void pager_reset(Pager *pPager){
}
/*
** This routine ends a transaction. A transaction is ended by either
** a COMMIT or a ROLLBACK.
**
** When this routine is called, the pager has the journal file open and
** a RESERVED or EXCLUSIVE lock on the database. This routine releases
** the database lock and acquires a SHARED lock in its place. The journal
** file is deleted and closed.
** a RESERVED or EXCLUSIVE lock on the database. This routine will release
** the database lock and acquires a SHARED lock in its place if that is
** the appropriate thing to do. Release locks usually is appropriate,
** unless we are in exclusive access mode or unless this is a
** COMMIT AND BEGIN or ROLLBACK AND BEGIN operation.
**
** The journal file is either deleted or truncated.
**
** TODO: Consider keeping the journal file open for temporary databases.
** This might give a performance improvement on windows where opening
** a file is an expensive operation.
*/
static int pager_unwritelock(Pager *pPager){
static int pager_end_transaction(Pager *pPager){
PgHdr *pPg;
int rc = SQLITE_OK;
int rc2 = SQLITE_OK;
@ -1423,7 +1430,7 @@ static int pager_playback(Pager *pPager, int isHot){
end_playback:
if( rc==SQLITE_OK ){
rc = pager_unwritelock(pPager);
rc = pager_end_transaction(pPager);
}
if( zMaster ){
/* If there was a master journal and this routine will return success,
@ -2563,7 +2570,7 @@ static int pager_recycle(Pager *pPager, int syncOk, PgHdr **ppPg){
/* If the page we are recycling is marked as alwaysRollback, then
** set the global alwaysRollback flag, thus disabling the
** sqlite_dont_rollback() optimization for the rest of this transaction.
** sqlite3PagerDontRollback() optimization for the rest of this transaction.
** It is necessary to do this because the page marked alwaysRollback
** might be reloaded at a later time but at that point we won't remember
** that is was marked alwaysRollback. This means that all pages must
@ -3096,7 +3103,7 @@ static int pager_open_journal(Pager *pPager){
rc = sqlite3PagerStmtBegin(pPager);
}
if( rc!=SQLITE_OK && rc!=SQLITE_NOMEM ){
rc = pager_unwritelock(pPager);
rc = pager_end_transaction(pPager);
if( rc==SQLITE_OK ){
rc = SQLITE_FULL;
}
@ -3127,7 +3134,7 @@ failed_to_open_journal:
** Acquire a write-lock on the database. The lock is removed when
** the any of the following happen:
**
** * sqlite3PagerCommit() is called.
** * sqlite3PagerCommitPhaseTwo() is called.
** * sqlite3PagerRollback() is called.
** * sqlite3PagerClose() is called.
** * sqlite3PagerUnref() is called to on every outstanding page.
@ -3528,9 +3535,9 @@ int sqlite3PagerOverwrite(Pager *pPager, Pgno pgno, void *pData){
** a transaction then removed from the freelist during a later part
** of the same transaction and reused for some other purpose. When it
** is first added to the freelist, this routine is called. When reused,
** the dont_rollback() routine is called. But because the page contains
** critical data, we still need to be sure it gets rolled back in spite
** of the dont_rollback() call.
** the sqlite3PagerDontRollback() routine is called. But because the
** page contains critical data, we still need to be sure it gets
** rolled back in spite of the sqlite3PagerDontRollback() call.
*/
void sqlite3PagerDontWrite(Pager *pPager, Pgno pgno){
PgHdr *pPg;
@ -3594,6 +3601,129 @@ void sqlite3PagerDontRollback(DbPage *pPg){
}
/*
** This routine is called to increment the database file change-counter,
** stored at byte 24 of the pager file.
*/
static int pager_incr_changecounter(Pager *pPager){
PgHdr *pPgHdr;
u32 change_counter;
int rc;
if( !pPager->changeCountDone ){
/* Open page 1 of the file for writing. */
rc = sqlite3PagerGet(pPager, 1, &pPgHdr);
if( rc!=SQLITE_OK ) return rc;
rc = sqlite3PagerWrite(pPgHdr);
if( rc!=SQLITE_OK ) return rc;
/* Read the current value at byte 24. */
change_counter = retrieve32bits(pPgHdr, 24);
/* Increment the value just read and write it back to byte 24. */
change_counter++;
put32bits(((char*)PGHDR_TO_DATA(pPgHdr))+24, change_counter);
pPager->iChangeCount = change_counter;
/* Release the page reference. */
sqlite3PagerUnref(pPgHdr);
pPager->changeCountDone = 1;
}
return SQLITE_OK;
}
/*
** Sync the database file for the pager pPager. zMaster points to the name
** of a master journal file that should be written into the individual
** journal file. zMaster may be NULL, which is interpreted as no master
** journal (a single database transaction).
**
** This routine ensures that the journal is synced, all dirty pages written
** to the database file and the database file synced. The only thing that
** remains to commit the transaction is to delete the journal file (or
** master journal file if specified).
**
** Note that if zMaster==NULL, this does not overwrite a previous value
** passed to an sqlite3PagerCommitPhaseOne() call.
**
** If parameter nTrunc is non-zero, then the pager file is truncated to
** nTrunc pages (this is used by auto-vacuum databases).
*/
int sqlite3PagerCommitPhaseOne(Pager *pPager, const char *zMaster, Pgno nTrunc){
int rc = SQLITE_OK;
PAGERTRACE4("DATABASE SYNC: File=%s zMaster=%s nTrunc=%d\n",
pPager->zFilename, zMaster, nTrunc);
/* If this is an in-memory db, or no pages have been written to, or this
** function has already been called, it is a no-op.
*/
if( pPager->state!=PAGER_SYNCED && !MEMDB && pPager->dirtyCache ){
PgHdr *pPg;
assert( pPager->journalOpen );
/* If a master journal file name has already been written to the
** journal file, then no sync is required. This happens when it is
** written, then the process fails to upgrade from a RESERVED to an
** EXCLUSIVE lock. The next time the process tries to commit the
** transaction the m-j name will have already been written.
*/
if( !pPager->setMaster ){
rc = pager_incr_changecounter(pPager);
if( rc!=SQLITE_OK ) goto sync_exit;
#ifndef SQLITE_OMIT_AUTOVACUUM
if( nTrunc!=0 ){
/* If this transaction has made the database smaller, then all pages
** being discarded by the truncation must be written to the journal
** file.
*/
Pgno i;
int iSkip = PAGER_MJ_PGNO(pPager);
for( i=nTrunc+1; i<=pPager->origDbSize; i++ ){
if( !(pPager->aInJournal[i/8] & (1<<(i&7))) && i!=iSkip ){
rc = sqlite3PagerGet(pPager, i, &pPg);
if( rc!=SQLITE_OK ) goto sync_exit;
rc = sqlite3PagerWrite(pPg);
sqlite3PagerUnref(pPg);
if( rc!=SQLITE_OK ) goto sync_exit;
}
}
}
#endif
rc = writeMasterJournal(pPager, zMaster);
if( rc!=SQLITE_OK ) goto sync_exit;
rc = syncJournal(pPager);
if( rc!=SQLITE_OK ) goto sync_exit;
}
#ifndef SQLITE_OMIT_AUTOVACUUM
if( nTrunc!=0 ){
rc = sqlite3PagerTruncate(pPager, nTrunc);
if( rc!=SQLITE_OK ) goto sync_exit;
}
#endif
/* Write all dirty pages to the database file */
pPg = pager_get_all_dirty_pages(pPager);
rc = pager_write_pagelist(pPg);
if( rc!=SQLITE_OK ) goto sync_exit;
/* Sync the database file. */
if( !pPager->noSync ){
rc = sqlite3OsSync(pPager->fd, 0);
}
IOTRACE(("DBSYNC %p\n", pPager))
pPager->state = PAGER_SYNCED;
}else if( MEMDB && nTrunc!=0 ){
rc = sqlite3PagerTruncate(pPager, nTrunc);
}
sync_exit:
return rc;
}
/*
** Commit all changes to the database and release the write lock.
**
@ -3601,7 +3731,7 @@ void sqlite3PagerDontRollback(DbPage *pPg){
** and an error code is returned. If the commit worked, SQLITE_OK
** is returned.
*/
int sqlite3PagerCommit(Pager *pPager){
int sqlite3PagerCommitPhaseTwo(Pager *pPager){
int rc;
PgHdr *pPg;
@ -3640,12 +3770,12 @@ int sqlite3PagerCommit(Pager *pPager){
/* Exit early (without doing the time-consuming sqlite3OsSync() calls)
** if there have been no changes to the database file. */
assert( pPager->needSync==0 );
rc = pager_unwritelock(pPager);
rc = pager_end_transaction(pPager);
}else{
assert( pPager->journalOpen );
rc = sqlite3PagerSync(pPager, 0, 0);
rc = sqlite3PagerCommitPhaseOne(pPager, 0, 0);
if( rc==SQLITE_OK ){
rc = pager_unwritelock(pPager);
rc = pager_end_transaction(pPager);
}
}
return pager_error(pPager, rc);
@ -3703,7 +3833,7 @@ int sqlite3PagerRollback(Pager *pPager){
}
if( !pPager->dirtyCache || !pPager->journalOpen ){
rc = pager_unwritelock(pPager);
rc = pager_end_transaction(pPager);
return rc;
}
@ -3716,7 +3846,7 @@ int sqlite3PagerRollback(Pager *pPager){
if( pPager->state==PAGER_RESERVED ){
int rc2;
rc = pager_playback(pPager, 0);
rc2 = pager_unwritelock(pPager);
rc2 = pager_end_transaction(pPager);
if( rc==SQLITE_OK ){
rc = rc2;
}
@ -3926,128 +4056,6 @@ void sqlite3PagerSetCodec(
pPager->pCodecArg = pCodecArg;
}
/*
** This routine is called to increment the database file change-counter,
** stored at byte 24 of the pager file.
*/
static int pager_incr_changecounter(Pager *pPager){
PgHdr *pPgHdr;
u32 change_counter;
int rc;
if( !pPager->changeCountDone ){
/* Open page 1 of the file for writing. */
rc = sqlite3PagerGet(pPager, 1, &pPgHdr);
if( rc!=SQLITE_OK ) return rc;
rc = sqlite3PagerWrite(pPgHdr);
if( rc!=SQLITE_OK ) return rc;
/* Read the current value at byte 24. */
change_counter = retrieve32bits(pPgHdr, 24);
/* Increment the value just read and write it back to byte 24. */
change_counter++;
put32bits(((char*)PGHDR_TO_DATA(pPgHdr))+24, change_counter);
pPager->iChangeCount = change_counter;
/* Release the page reference. */
sqlite3PagerUnref(pPgHdr);
pPager->changeCountDone = 1;
}
return SQLITE_OK;
}
/*
** Sync the database file for the pager pPager. zMaster points to the name
** of a master journal file that should be written into the individual
** journal file. zMaster may be NULL, which is interpreted as no master
** journal (a single database transaction).
**
** This routine ensures that the journal is synced, all dirty pages written
** to the database file and the database file synced. The only thing that
** remains to commit the transaction is to delete the journal file (or
** master journal file if specified).
**
** Note that if zMaster==NULL, this does not overwrite a previous value
** passed to an sqlite3PagerSync() call.
**
** If parameter nTrunc is non-zero, then the pager file is truncated to
** nTrunc pages (this is used by auto-vacuum databases).
*/
int sqlite3PagerSync(Pager *pPager, const char *zMaster, Pgno nTrunc){
int rc = SQLITE_OK;
PAGERTRACE4("DATABASE SYNC: File=%s zMaster=%s nTrunc=%d\n",
pPager->zFilename, zMaster, nTrunc);
/* If this is an in-memory db, or no pages have been written to, or this
** function has already been called, it is a no-op.
*/
if( pPager->state!=PAGER_SYNCED && !MEMDB && pPager->dirtyCache ){
PgHdr *pPg;
assert( pPager->journalOpen );
/* If a master journal file name has already been written to the
** journal file, then no sync is required. This happens when it is
** written, then the process fails to upgrade from a RESERVED to an
** EXCLUSIVE lock. The next time the process tries to commit the
** transaction the m-j name will have already been written.
*/
if( !pPager->setMaster ){
rc = pager_incr_changecounter(pPager);
if( rc!=SQLITE_OK ) goto sync_exit;
#ifndef SQLITE_OMIT_AUTOVACUUM
if( nTrunc!=0 ){
/* If this transaction has made the database smaller, then all pages
** being discarded by the truncation must be written to the journal
** file.
*/
Pgno i;
int iSkip = PAGER_MJ_PGNO(pPager);
for( i=nTrunc+1; i<=pPager->origDbSize; i++ ){
if( !(pPager->aInJournal[i/8] & (1<<(i&7))) && i!=iSkip ){
rc = sqlite3PagerGet(pPager, i, &pPg);
if( rc!=SQLITE_OK ) goto sync_exit;
rc = sqlite3PagerWrite(pPg);
sqlite3PagerUnref(pPg);
if( rc!=SQLITE_OK ) goto sync_exit;
}
}
}
#endif
rc = writeMasterJournal(pPager, zMaster);
if( rc!=SQLITE_OK ) goto sync_exit;
rc = syncJournal(pPager);
if( rc!=SQLITE_OK ) goto sync_exit;
}
#ifndef SQLITE_OMIT_AUTOVACUUM
if( nTrunc!=0 ){
rc = sqlite3PagerTruncate(pPager, nTrunc);
if( rc!=SQLITE_OK ) goto sync_exit;
}
#endif
/* Write all dirty pages to the database file */
pPg = pager_get_all_dirty_pages(pPager);
rc = pager_write_pagelist(pPg);
if( rc!=SQLITE_OK ) goto sync_exit;
/* Sync the database file. */
if( !pPager->noSync ){
rc = sqlite3OsSync(pPager->fd, 0);
}
IOTRACE(("DBSYNC %p\n", pPager))
pPager->state = PAGER_SYNCED;
}else if( MEMDB && nTrunc!=0 ){
rc = sqlite3PagerTruncate(pPager, nTrunc);
}
sync_exit:
return rc;
}
#ifndef SQLITE_OMIT_AUTOVACUUM
/*
** Move the page identified by pData to location pgno in the file.

6
src/pager.h
View File

@ -13,7 +13,7 @@
** subsystem. The page cache subsystem reads and writes a file a page
** at a time and provides a journal for rollback.
**
** @(#) $Id: pager.h,v 1.56 2007/03/27 16:19:52 danielk1977 Exp $
** @(#) $Id: pager.h,v 1.57 2007/03/30 14:06:34 drh Exp $
*/
#ifndef _PAGER_H_
@ -101,8 +101,8 @@ int sqlite3PagerOverwrite(Pager *pPager, Pgno pgno, void*);
int sqlite3PagerPagecount(Pager*);
int sqlite3PagerTruncate(Pager*,Pgno);
int sqlite3PagerBegin(DbPage*, int exFlag);
int sqlite3PagerCommit(Pager*);
int sqlite3PagerSync(Pager*,const char *zMaster, Pgno);
int sqlite3PagerCommitPhaseOne(Pager*,const char *zMaster, Pgno);
int sqlite3PagerCommitPhaseTwo(Pager*);
int sqlite3PagerRollback(Pager*);
int sqlite3PagerIsreadonly(Pager*);
int sqlite3PagerStmtBegin(Pager*);

9
src/test2.c
View File

@ -13,7 +13,7 @@
** is not included in the SQLite library. It is used for automated
** testing of the SQLite library.
**
** $Id: test2.c,v 1.41 2007/03/19 17:44:28 danielk1977 Exp $
** $Id: test2.c,v 1.42 2007/03/30 14:06:34 drh Exp $
*/
#include "sqliteInt.h"
#include "os.h"
@ -163,7 +163,12 @@ static int pager_commit(
return TCL_ERROR;
}
pPager = sqlite3TextToPtr(argv[1]);
rc = sqlite3PagerCommit(pPager);
rc = sqlite3PagerCommitPhaseOne(pPager, 0, 0);
if( rc!=SQLITE_OK ){
Tcl_AppendResult(interp, errorName(rc), 0);
return TCL_ERROR;
}
rc = sqlite3PagerCommitPhaseTwo(pPager);
if( rc!=SQLITE_OK ){
Tcl_AppendResult(interp, errorName(rc), 0);
return TCL_ERROR;

36
src/vdbeaux.c
View File

@ -1090,19 +1090,19 @@ static int vdbeCommit(sqlite3 *db){
for(i=0; rc==SQLITE_OK && i<db->nDb; i++){
Btree *pBt = db->aDb[i].pBt;
if( pBt ){
rc = sqlite3BtreeSync(pBt, 0);
rc = sqlite3BtreeCommitPhaseOne(pBt, 0);
}
}
/* Do the commit only if all databases successfully synced.
** If one of the BtreeCommit() calls fails, this indicates an IO error
** while deleting or truncating a journal file. It is unlikely, but
** could happen. In this case abandon processing and return the error.
/* Do the commit only if all databases successfully complete phase 1.
** If one of the BtreeCommitPhaseOne() calls fails, this indicates an
** IO error while deleting or truncating a journal file. It is unlikely,
** but could happen. In this case abandon processing and return the error.
*/
for(i=0; rc==SQLITE_OK && i<db->nDb; i++){
Btree *pBt = db->aDb[i].pBt;
if( pBt ){
rc = sqlite3BtreeCommit(pBt);
rc = sqlite3BtreeCommitPhaseTwo(pBt);
}
}
if( rc==SQLITE_OK ){
@ -1182,16 +1182,16 @@ static int vdbeCommit(sqlite3 *db){
** sets the master journal pointer in each individual journal. If
** an error occurs here, do not delete the master journal file.
**
** If the error occurs during the first call to sqlite3BtreeSync(),
** then there is a chance that the master journal file will be
** orphaned. But we cannot delete it, in case the master journal
** file name was written into the journal file before the failure
** occured.
** If the error occurs during the first call to
** sqlite3BtreeCommitPhaseOne(), then there is a chance that the
** master journal file will be orphaned. But we cannot delete it,
** in case the master journal file name was written into the journal
** file before the failure occured.
*/
for(i=0; rc==SQLITE_OK && i<db->nDb; i++){
Btree *pBt = db->aDb[i].pBt;
if( pBt && sqlite3BtreeIsInTrans(pBt) ){
rc = sqlite3BtreeSync(pBt, zMaster);
rc = sqlite3BtreeCommitPhaseOne(pBt, zMaster);
}
}
sqlite3OsClose(&master);
@ -1223,17 +1223,17 @@ static int vdbeCommit(sqlite3 *db){
}
/* All files and directories have already been synced, so the following
** calls to sqlite3BtreeCommit() are only closing files and deleting
** journals. If something goes wrong while this is happening we don't
** really care. The integrity of the transaction is already guaranteed,
** but some stray 'cold' journals may be lying around. Returning an
** error code won't help matters.
** calls to sqlite3BtreeCommitPhaseTwo() are only closing files and
** deleting or truncating journals. If something goes wrong while
** this is happening we don't really care. The integrity of the
** transaction is already guaranteed, but some stray 'cold' journals
** may be lying around. Returning an error code won't help matters.
*/
disable_simulated_io_errors();
for(i=0; i<db->nDb; i++){
Btree *pBt = db->aDb[i].pBt;
if( pBt ){
sqlite3BtreeCommit(pBt);
sqlite3BtreeCommitPhaseTwo(pBt);
}
}
enable_simulated_io_errors();