Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
Overview
Comment: | Do not use the page containing locks for anything to avoid problems with the manditory file locking on windows. (CVS 1595) |
---|---|
Downloads: | Tarball | ZIP archive |
Timelines: | family | ancestors | descendants | both | trunk |
Files: | files | file ages | folders |
SHA1: |
5a814202777f381caf5337b37e11c9ab |
User & Date: | drh 2004-06-15 01:40:29.000 |
Context
2004-06-15
| ||
02:13 | Update comments in btree.c. No changes to code. (CVS 1596) (check-in: 1c6a070635 user: drh tags: trunk) | |
01:40 | Do not use the page containing locks for anything to avoid problems with the manditory file locking on windows. (CVS 1595) (check-in: 5a81420277 user: drh tags: trunk) | |
00:29 | Add the sqlite3OsFileModTime() interface. But it is still unused. The change counter in page 1 is always incremented. (CVS 1594) (check-in: 186c6f93e1 user: drh tags: trunk) | |
Changes
Changes to src/btree.c.
1 2 3 4 5 6 7 8 9 10 11 | /* ** 2004 April 6 ** ** The author disclaims copyright to this source code. In place of ** a legal notice, here is a blessing: ** ** May you do good and not evil. ** May you find forgiveness for yourself and forgive others. ** May you share freely, never taking more than you give. ** ************************************************************************* | | | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 | /* ** 2004 April 6 ** ** The author disclaims copyright to this source code. In place of ** a legal notice, here is a blessing: ** ** May you do good and not evil. ** May you find forgiveness for yourself and forgive others. ** May you share freely, never taking more than you give. ** ************************************************************************* ** $Id: btree.c,v 1.167 2004/06/15 01:40:29 drh Exp $ ** ** This file implements a external (disk-based) database using BTrees. ** For a detailed discussion of BTrees, refer to ** ** Donald E. Knuth, THE ART OF COMPUTER PROGRAMMING, Volume 3: ** "Sorting And Searching", pages 473-480. Addison-Wesley ** Publishing Company, Reading, Massachusetts. |
︙ | ︙ | |||
203 204 205 206 207 208 209 210 211 212 213 214 215 216 | ** 4 Page number of next trunk page ** 4 Number of leaf pointers on this page ** * zero or more pages numbers of leaves */ #include "sqliteInt.h" #include "pager.h" #include "btree.h" #include <assert.h> /* Maximum page size. The upper bound on this value is 65536 (a limit ** imposed by the 2-byte size of cell array pointers.) The ** maximum page size determines the amount of stack space allocated ** by many of the routines in this module. On embedded architectures | > | 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 | ** 4 Page number of next trunk page ** 4 Number of leaf pointers on this page ** * zero or more pages numbers of leaves */ #include "sqliteInt.h" #include "pager.h" #include "btree.h" #include "os.h" #include <assert.h> /* Maximum page size. The upper bound on this value is 65536 (a limit ** imposed by the 2-byte size of cell array pointers.) The ** maximum page size determines the amount of stack space allocated ** by many of the routines in this module. On embedded architectures |
︙ | ︙ | |||
4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 | sCheck.nPage = sqlite3pager_pagecount(sCheck.pPager); if( sCheck.nPage==0 ){ unlockBtreeIfUnused(pBt); return 0; } sCheck.anRef = sqliteMallocRaw( (sCheck.nPage+1)*sizeof(sCheck.anRef[0]) ); for(i=0; i<=sCheck.nPage; i++){ sCheck.anRef[i] = 0; } sCheck.zErrMsg = 0; /* Check the integrity of the freelist */ checkList(&sCheck, 1, get4byte(&pBt->pPage1->aData[32]), get4byte(&pBt->pPage1->aData[36]), "Main freelist: "); | > > > > | 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 | sCheck.nPage = sqlite3pager_pagecount(sCheck.pPager); if( sCheck.nPage==0 ){ unlockBtreeIfUnused(pBt); return 0; } sCheck.anRef = sqliteMallocRaw( (sCheck.nPage+1)*sizeof(sCheck.anRef[0]) ); for(i=0; i<=sCheck.nPage; i++){ sCheck.anRef[i] = 0; } i = PENDING_BYTE/pBt->pageSize + 1; if( i<=sCheck.nPage ){ sCheck.anRef[i] = 1; } sCheck.zErrMsg = 0; /* Check the integrity of the freelist */ checkList(&sCheck, 1, get4byte(&pBt->pPage1->aData[32]), get4byte(&pBt->pPage1->aData[36]), "Main freelist: "); |
︙ | ︙ |
Changes to src/os.h.
︙ | ︙ | |||
87 88 89 90 91 92 93 | #define NO_LOCK 0 #define SHARED_LOCK 1 #define RESERVED_LOCK 2 #define PENDING_LOCK 3 #define EXCLUSIVE_LOCK 4 /* | | < < < < < | 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 | #define NO_LOCK 0 #define SHARED_LOCK 1 #define RESERVED_LOCK 2 #define PENDING_LOCK 3 #define EXCLUSIVE_LOCK 4 /* ** File Locking Notes: (Mostly about windows but also some info for Unix) ** ** We cannot use LockFileEx() or UnlockFileEx() on Win95/98/ME because ** those functions are not available. So we use only LockFile() and ** UnlockFile(). ** ** LockFile() prevents not just writing but also reading by other processes. ** A SHARED_LOCK is obtained by locking a single randomly-chosen ** byte out of a specific range of bytes. The lock byte is obtained at ** random so two separate readers can probably access the file at the ** same time, unless they are unlucky and choose the same lock byte. ** An EXCLUSIVE_LOCK is obtained by locking all bytes in the range. ** There can only be one writer. A RESERVED_LOCK is obtained by locking ** a single byte of the file that is designated as the reserved lock byte. |
︙ | ︙ | |||
125 126 127 128 129 130 131 | ** SHARED_SIZE is the number of bytes available in the pool from which ** a random byte is selected for a shared lock. The pool of bytes for ** shared locks begins at SHARED_FIRST. ** ** These #defines are available in os.h so that Unix can use the same ** byte ranges for locking. This leaves open the possiblity of having ** clients on win95, winNT, and unix all talking to the same shared file | | > > > | > > > > > > > > > > > > | | | | > | 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 | ** SHARED_SIZE is the number of bytes available in the pool from which ** a random byte is selected for a shared lock. The pool of bytes for ** shared locks begins at SHARED_FIRST. ** ** These #defines are available in os.h so that Unix can use the same ** byte ranges for locking. This leaves open the possiblity of having ** clients on win95, winNT, and unix all talking to the same shared file ** and all locking correctly. To do so would require that samba (or whatever ** tool is being used for file sharing) implements locks correctly between ** windows and unix. I'm guessing that isn't likely to happen, but by ** using the same locking range we are at least open to the possibility. ** ** Locking in windows is manditory. For this reason, we cannot store ** actual data in the bytes used for locking. The pager never allocates ** the pages involved in locking therefore. SHARED_SIZE is selected so ** that all locks will fit on a single page even at the minimum page size. ** PENDING_BYTE defines the beginning of the locks. By default PENDING_BYTE ** is set high so that we don't have to allocate an unused page except ** for very large databases. But one should test the page skipping logic ** by setting PENDING_BYTE low and running the entire regression suite. ** ** Changing the value of PENDING_BYTE results in a subtly incompatible ** file format. Depending on how it is changed, you might not notice ** the incompatibility right away, even running a full regression test. ** The default location of PENDING_BYTE is the first byte past the ** 1GB boundary. ** */ #define PENDING_BYTE 0x40000000 /* First byte past the 1GB boundary */ /* #define PENDING_BYTE 0x5400 // Page 20 - for testing */ #define RESERVED_BYTE (PENDING_BYTE+1) #define SHARED_FIRST (PENDING_BYTE+2) #define SHARED_SIZE 510 int sqlite3OsDelete(const char*); int sqlite3OsFileExists(const char*); int sqliteOsFileRename(const char*, const char*); int sqlite3OsOpenReadWrite(const char*, OsFile*, int*); int sqlite3OsOpenExclusive(const char*, OsFile*, int); |
︙ | ︙ |
Changes to src/pager.c.
︙ | ︙ | |||
14 15 16 17 18 19 20 | ** The pager is used to access a database disk file. It implements ** atomic commit and rollback through the use of a journal file that ** is separate from the database file. The pager also implements file ** locking to prevent two processes from writing the same database ** file simultaneously, or one process from reading the database while ** another is writing. ** | | | 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 | ** The pager is used to access a database disk file. It implements ** atomic commit and rollback through the use of a journal file that ** is separate from the database file. The pager also implements file ** locking to prevent two processes from writing the same database ** file simultaneously, or one process from reading the database while ** another is writing. ** ** @(#) $Id: pager.c,v 1.127 2004/06/15 01:40:29 drh Exp $ */ #include "os.h" /* Must be first to enable large file support */ #include "sqliteInt.h" #include "pager.h" #include <assert.h> #include <string.h> |
︙ | ︙ | |||
1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 | return pPager->dbSize; } if( sqlite3OsFileSize(&pPager->fd, &n)!=SQLITE_OK ){ pPager->errMask |= PAGER_ERR_DISK; return 0; } n /= SQLITE_PAGE_SIZE; if( pPager->state!=PAGER_UNLOCK ){ pPager->dbSize = n; } return n; } /* | > > > | 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 | return pPager->dbSize; } if( sqlite3OsFileSize(&pPager->fd, &n)!=SQLITE_OK ){ pPager->errMask |= PAGER_ERR_DISK; return 0; } n /= SQLITE_PAGE_SIZE; if( !pPager->memDb && n==PENDING_BYTE/SQLITE_PAGE_SIZE ){ n++; } if( pPager->state!=PAGER_UNLOCK ){ pPager->dbSize = n; } return n; } /* |
︙ | ︙ | |||
2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 | page_add_to_stmt_list(pPg); } /* Update the database size and return. */ if( pPager->dbSize<(int)pPg->pgno ){ pPager->dbSize = pPg->pgno; } return rc; } /* ** Return TRUE if the page given in the argument was previously passed ** to sqlite3pager_write(). In other words, return TRUE if it is ok | > > > | 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 | page_add_to_stmt_list(pPg); } /* Update the database size and return. */ if( pPager->dbSize<(int)pPg->pgno ){ pPager->dbSize = pPg->pgno; if( !pPager->memDb && pPager->dbSize==PENDING_BYTE/pPager->pageSize ){ pPager->dbSize++; } } return rc; } /* ** Return TRUE if the page given in the argument was previously passed ** to sqlite3pager_write(). In other words, return TRUE if it is ok |
︙ | ︙ |