1 /* 2 ** 2004 April 6 3 ** 4 ** The author disclaims copyright to this source code. In place of 5 ** a legal notice, here is a blessing: 6 ** 7 ** May you do good and not evil. 8 ** May you find forgiveness for yourself and forgive others. 9 ** May you share freely, never taking more than you give. 10 ** 11 ************************************************************************* 12 ** This file implements an external (disk-based) database using BTrees. 13 ** For a detailed discussion of BTrees, refer to 14 ** 15 ** Donald E. Knuth, THE ART OF COMPUTER PROGRAMMING, Volume 3: 16 ** "Sorting And Searching", pages 473-480. Addison-Wesley 17 ** Publishing Company, Reading, Massachusetts. 18 ** 19 ** The basic idea is that each page of the file contains N database 20 ** entries and N+1 pointers to subpages. 21 ** 22 ** ---------------------------------------------------------------- 23 ** | Ptr(0) | Key(0) | Ptr(1) | Key(1) | ... | Key(N-1) | Ptr(N) | 24 ** ---------------------------------------------------------------- 25 ** 26 ** All of the keys on the page that Ptr(0) points to have values less 27 ** than Key(0). All of the keys on page Ptr(1) and its subpages have 28 ** values greater than Key(0) and less than Key(1). All of the keys 29 ** on Ptr(N) and its subpages have values greater than Key(N-1). And 30 ** so forth. 31 ** 32 ** Finding a particular key requires reading O(log(M)) pages from the 33 ** disk where M is the number of entries in the tree. 34 ** 35 ** In this implementation, a single file can hold one or more separate 36 ** BTrees. Each BTree is identified by the index of its root page. The 37 ** key and data for any entry are combined to form the "payload". A 38 ** fixed amount of payload can be carried directly on the database 39 ** page. If the payload is larger than the preset amount then surplus 40 ** bytes are stored on overflow pages. The payload for an entry 41 ** and the preceding pointer are combined to form a "Cell". Each 42 ** page has a small header which contains the Ptr(N) pointer and other 43 ** information such as the size of key and data. 44 ** 45 ** FORMAT DETAILS 46 ** 47 ** The file is divided into pages. The first page is called page 1, 48 ** the second is page 2, and so forth. A page number of zero indicates 49 ** "no such page". The page size can be any power of 2 between 512 and 65536. 50 ** Each page can be either a btree page, a freelist page, an overflow 51 ** page, or a pointer-map page. 52 ** 53 ** The first page is always a btree page. The first 100 bytes of the first 54 ** page contain a special header (the "file header") that describes the file. 55 ** The format of the file header is as follows: 56 ** 57 ** OFFSET SIZE DESCRIPTION 58 ** 0 16 Header string: "SQLite format 3\000" 59 ** 16 2 Page size in bytes. (1 means 65536) 60 ** 18 1 File format write version 61 ** 19 1 File format read version 62 ** 20 1 Bytes of unused space at the end of each page 63 ** 21 1 Max embedded payload fraction (must be 64) 64 ** 22 1 Min embedded payload fraction (must be 32) 65 ** 23 1 Min leaf payload fraction (must be 32) 66 ** 24 4 File change counter 67 ** 28 4 Reserved for future use 68 ** 32 4 First freelist page 69 ** 36 4 Number of freelist pages in the file 70 ** 40 60 15 4-byte meta values passed to higher layers 71 ** 72 ** 40 4 Schema cookie 73 ** 44 4 File format of schema layer 74 ** 48 4 Size of page cache 75 ** 52 4 Largest root-page (auto/incr_vacuum) 76 ** 56 4 1=UTF-8 2=UTF16le 3=UTF16be 77 ** 60 4 User version 78 ** 64 4 Incremental vacuum mode 79 ** 68 4 Application-ID 80 ** 72 20 unused 81 ** 92 4 The version-valid-for number 82 ** 96 4 SQLITE_VERSION_NUMBER 83 ** 84 ** All of the integer values are big-endian (most significant byte first). 85 ** 86 ** The file change counter is incremented when the database is changed 87 ** This counter allows other processes to know when the file has changed 88 ** and thus when they need to flush their cache. 89 ** 90 ** The max embedded payload fraction is the amount of the total usable 91 ** space in a page that can be consumed by a single cell for standard 92 ** B-tree (non-LEAFDATA) tables. A value of 255 means 100%. The default 93 ** is to limit the maximum cell size so that at least 4 cells will fit 94 ** on one page. Thus the default max embedded payload fraction is 64. 95 ** 96 ** If the payload for a cell is larger than the max payload, then extra 97 ** payload is spilled to overflow pages. Once an overflow page is allocated, 98 ** as many bytes as possible are moved into the overflow pages without letting 99 ** the cell size drop below the min embedded payload fraction. 100 ** 101 ** The min leaf payload fraction is like the min embedded payload fraction 102 ** except that it applies to leaf nodes in a LEAFDATA tree. The maximum 103 ** payload fraction for a LEAFDATA tree is always 100% (or 255) and it 104 ** not specified in the header. 105 ** 106 ** Each btree pages is divided into three sections: The header, the 107 ** cell pointer array, and the cell content area. Page 1 also has a 100-byte 108 ** file header that occurs before the page header. 109 ** 110 ** |----------------| 111 ** | file header | 100 bytes. Page 1 only. 112 ** |----------------| 113 ** | page header | 8 bytes for leaves. 12 bytes for interior nodes 114 ** |----------------| 115 ** | cell pointer | | 2 bytes per cell. Sorted order. 116 ** | array | | Grows downward 117 ** | | v 118 ** |----------------| 119 ** | unallocated | 120 ** | space | 121 ** |----------------| ^ Grows upwards 122 ** | cell content | | Arbitrary order interspersed with freeblocks. 123 ** | area | | and free space fragments. 124 ** |----------------| 125 ** 126 ** The page headers looks like this: 127 ** 128 ** OFFSET SIZE DESCRIPTION 129 ** 0 1 Flags. 1: intkey, 2: zerodata, 4: leafdata, 8: leaf 130 ** 1 2 byte offset to the first freeblock 131 ** 3 2 number of cells on this page 132 ** 5 2 first byte of the cell content area 133 ** 7 1 number of fragmented free bytes 134 ** 8 4 Right child (the Ptr(N) value). Omitted on leaves. 135 ** 136 ** The flags define the format of this btree page. The leaf flag means that 137 ** this page has no children. The zerodata flag means that this page carries 138 ** only keys and no data. The intkey flag means that the key is an integer 139 ** which is stored in the key size entry of the cell header rather than in 140 ** the payload area. 141 ** 142 ** The cell pointer array begins on the first byte after the page header. 143 ** The cell pointer array contains zero or more 2-byte numbers which are 144 ** offsets from the beginning of the page to the cell content in the cell 145 ** content area. The cell pointers occur in sorted order. The system strives 146 ** to keep free space after the last cell pointer so that new cells can 147 ** be easily added without having to defragment the page. 148 ** 149 ** Cell content is stored at the very end of the page and grows toward the 150 ** beginning of the page. 151 ** 152 ** Unused space within the cell content area is collected into a linked list of 153 ** freeblocks. Each freeblock is at least 4 bytes in size. The byte offset 154 ** to the first freeblock is given in the header. Freeblocks occur in 155 ** increasing order. Because a freeblock must be at least 4 bytes in size, 156 ** any group of 3 or fewer unused bytes in the cell content area cannot 157 ** exist on the freeblock chain. A group of 3 or fewer free bytes is called 158 ** a fragment. The total number of bytes in all fragments is recorded. 159 ** in the page header at offset 7. 160 ** 161 ** SIZE DESCRIPTION 162 ** 2 Byte offset of the next freeblock 163 ** 2 Bytes in this freeblock 164 ** 165 ** Cells are of variable length. Cells are stored in the cell content area at 166 ** the end of the page. Pointers to the cells are in the cell pointer array 167 ** that immediately follows the page header. Cells is not necessarily 168 ** contiguous or in order, but cell pointers are contiguous and in order. 169 ** 170 ** Cell content makes use of variable length integers. A variable 171 ** length integer is 1 to 9 bytes where the lower 7 bits of each 172 ** byte are used. The integer consists of all bytes that have bit 8 set and 173 ** the first byte with bit 8 clear. The most significant byte of the integer 174 ** appears first. A variable-length integer may not be more than 9 bytes long. 175 ** As a special case, all 8 bytes of the 9th byte are used as data. This 176 ** allows a 64-bit integer to be encoded in 9 bytes. 177 ** 178 ** 0x00 becomes 0x00000000 179 ** 0x7f becomes 0x0000007f 180 ** 0x81 0x00 becomes 0x00000080 181 ** 0x82 0x00 becomes 0x00000100 182 ** 0x80 0x7f becomes 0x0000007f 183 ** 0x8a 0x91 0xd1 0xac 0x78 becomes 0x12345678 184 ** 0x81 0x81 0x81 0x81 0x01 becomes 0x10204081 185 ** 186 ** Variable length integers are used for rowids and to hold the number of 187 ** bytes of key and data in a btree cell. 188 ** 189 ** The content of a cell looks like this: 190 ** 191 ** SIZE DESCRIPTION 192 ** 4 Page number of the left child. Omitted if leaf flag is set. 193 ** var Number of bytes of data. Omitted if the zerodata flag is set. 194 ** var Number of bytes of key. Or the key itself if intkey flag is set. 195 ** * Payload 196 ** 4 First page of the overflow chain. Omitted if no overflow 197 ** 198 ** Overflow pages form a linked list. Each page except the last is completely 199 ** filled with data (pagesize - 4 bytes). The last page can have as little 200 ** as 1 byte of data. 201 ** 202 ** SIZE DESCRIPTION 203 ** 4 Page number of next overflow page 204 ** * Data 205 ** 206 ** Freelist pages come in two subtypes: trunk pages and leaf pages. The 207 ** file header points to the first in a linked list of trunk page. Each trunk 208 ** page points to multiple leaf pages. The content of a leaf page is 209 ** unspecified. A trunk page looks like this: 210 ** 211 ** SIZE DESCRIPTION 212 ** 4 Page number of next trunk page 213 ** 4 Number of leaf pointers on this page 214 ** * zero or more pages numbers of leaves 215 */ 216 #include "sqliteInt.h" 217 218 219 /* The following value is the maximum cell size assuming a maximum page 220 ** size give above. 221 */ 222 #define MX_CELL_SIZE(pBt) ((int)(pBt->pageSize-8)) 223 224 /* The maximum number of cells on a single page of the database. This 225 ** assumes a minimum cell size of 6 bytes (4 bytes for the cell itself 226 ** plus 2 bytes for the index to the cell in the page header). Such 227 ** small cells will be rare, but they are possible. 228 */ 229 #define MX_CELL(pBt) ((pBt->pageSize-8)/6) 230 231 /* Forward declarations */ 232 typedef struct MemPage MemPage; 233 typedef struct BtLock BtLock; 234 typedef struct CellInfo CellInfo; 235 236 /* 237 ** This is a magic string that appears at the beginning of every 238 ** SQLite database in order to identify the file as a real database. 239 ** 240 ** You can change this value at compile-time by specifying a 241 ** -DSQLITE_FILE_HEADER="..." on the compiler command-line. The 242 ** header must be exactly 16 bytes including the zero-terminator so 243 ** the string itself should be 15 characters long. If you change 244 ** the header, then your custom library will not be able to read 245 ** databases generated by the standard tools and the standard tools 246 ** will not be able to read databases created by your custom library. 247 */ 248 #ifndef SQLITE_FILE_HEADER /* 123456789 123456 */ 249 # define SQLITE_FILE_HEADER "SQLite format 3" 250 #endif 251 252 /* 253 ** Page type flags. An ORed combination of these flags appear as the 254 ** first byte of on-disk image of every BTree page. 255 */ 256 #define PTF_INTKEY 0x01 257 #define PTF_ZERODATA 0x02 258 #define PTF_LEAFDATA 0x04 259 #define PTF_LEAF 0x08 260 261 /* 262 ** An instance of this object stores information about each a single database 263 ** page that has been loaded into memory. The information in this object 264 ** is derived from the raw on-disk page content. 265 ** 266 ** As each database page is loaded into memory, the pager allocats an 267 ** instance of this object and zeros the first 8 bytes. (This is the 268 ** "extra" information associated with each page of the pager.) 269 ** 270 ** Access to all fields of this structure is controlled by the mutex 271 ** stored in MemPage.pBt->mutex. 272 */ 273 struct MemPage { 274 u8 isInit; /* True if previously initialized. MUST BE FIRST! */ 275 u8 bBusy; /* Prevent endless loops on corrupt database files */ 276 u8 intKey; /* True if table b-trees. False for index b-trees */ 277 u8 intKeyLeaf; /* True if the leaf of an intKey table */ 278 Pgno pgno; /* Page number for this page */ 279 /* Only the first 8 bytes (above) are zeroed by pager.c when a new page 280 ** is allocated. All fields that follow must be initialized before use */ 281 u8 leaf; /* True if a leaf page */ 282 u8 hdrOffset; /* 100 for page 1. 0 otherwise */ 283 u8 childPtrSize; /* 0 if leaf==1. 4 if leaf==0 */ 284 u8 max1bytePayload; /* min(maxLocal,127) */ 285 u8 nOverflow; /* Number of overflow cell bodies in aCell[] */ 286 u16 maxLocal; /* Copy of BtShared.maxLocal or BtShared.maxLeaf */ 287 u16 minLocal; /* Copy of BtShared.minLocal or BtShared.minLeaf */ 288 u16 cellOffset; /* Index in aData of first cell pointer */ 289 int nFree; /* Number of free bytes on the page. -1 for unknown */ 290 u16 nCell; /* Number of cells on this page, local and ovfl */ 291 u16 maskPage; /* Mask for page offset */ 292 u16 aiOvfl[4]; /* Insert the i-th overflow cell before the aiOvfl-th 293 ** non-overflow cell */ 294 u8 *apOvfl[4]; /* Pointers to the body of overflow cells */ 295 BtShared *pBt; /* Pointer to BtShared that this page is part of */ 296 u8 *aData; /* Pointer to disk image of the page data */ 297 u8 *aDataEnd; /* One byte past the end of usable data */ 298 u8 *aCellIdx; /* The cell index area */ 299 u8 *aDataOfst; /* Same as aData for leaves. aData+4 for interior */ 300 DbPage *pDbPage; /* Pager page handle */ 301 u16 (*xCellSize)(MemPage*,u8*); /* cellSizePtr method */ 302 void (*xParseCell)(MemPage*,u8*,CellInfo*); /* btreeParseCell method */ 303 }; 304 305 /* 306 ** A linked list of the following structures is stored at BtShared.pLock. 307 ** Locks are added (or upgraded from READ_LOCK to WRITE_LOCK) when a cursor 308 ** is opened on the table with root page BtShared.iTable. Locks are removed 309 ** from this list when a transaction is committed or rolled back, or when 310 ** a btree handle is closed. 311 */ 312 struct BtLock { 313 Btree *pBtree; /* Btree handle holding this lock */ 314 Pgno iTable; /* Root page of table */ 315 u8 eLock; /* READ_LOCK or WRITE_LOCK */ 316 BtLock *pNext; /* Next in BtShared.pLock list */ 317 }; 318 319 /* Candidate values for BtLock.eLock */ 320 #define READ_LOCK 1 321 #define WRITE_LOCK 2 322 323 /* A Btree handle 324 ** 325 ** A database connection contains a pointer to an instance of 326 ** this object for every database file that it has open. This structure 327 ** is opaque to the database connection. The database connection cannot 328 ** see the internals of this structure and only deals with pointers to 329 ** this structure. 330 ** 331 ** For some database files, the same underlying database cache might be 332 ** shared between multiple connections. In that case, each connection 333 ** has it own instance of this object. But each instance of this object 334 ** points to the same BtShared object. The database cache and the 335 ** schema associated with the database file are all contained within 336 ** the BtShared object. 337 ** 338 ** All fields in this structure are accessed under sqlite3.mutex. 339 ** The pBt pointer itself may not be changed while there exists cursors 340 ** in the referenced BtShared that point back to this Btree since those 341 ** cursors have to go through this Btree to find their BtShared and 342 ** they often do so without holding sqlite3.mutex. 343 */ 344 struct Btree { 345 sqlite3 *db; /* The database connection holding this btree */ 346 BtShared *pBt; /* Sharable content of this btree */ 347 u8 inTrans; /* TRANS_NONE, TRANS_READ or TRANS_WRITE */ 348 u8 sharable; /* True if we can share pBt with another db */ 349 u8 locked; /* True if db currently has pBt locked */ 350 u8 hasIncrblobCur; /* True if there are one or more Incrblob cursors */ 351 int wantToLock; /* Number of nested calls to sqlite3BtreeEnter() */ 352 int nBackup; /* Number of backup operations reading this btree */ 353 u32 iDataVersion; /* Combines with pBt->pPager->iDataVersion */ 354 Btree *pNext; /* List of other sharable Btrees from the same db */ 355 Btree *pPrev; /* Back pointer of the same list */ 356 #ifdef SQLITE_DEBUG 357 u64 nSeek; /* Calls to sqlite3BtreeMovetoUnpacked() */ 358 #endif 359 #ifndef SQLITE_OMIT_SHARED_CACHE 360 BtLock lock; /* Object used to lock page 1 */ 361 #endif 362 }; 363 364 /* 365 ** Btree.inTrans may take one of the following values. 366 ** 367 ** If the shared-data extension is enabled, there may be multiple users 368 ** of the Btree structure. At most one of these may open a write transaction, 369 ** but any number may have active read transactions. 370 ** 371 ** These values must match SQLITE_TXN_NONE, SQLITE_TXN_READ, and 372 ** SQLITE_TXN_WRITE 373 */ 374 #define TRANS_NONE 0 375 #define TRANS_READ 1 376 #define TRANS_WRITE 2 377 378 #if TRANS_NONE!=SQLITE_TXN_NONE 379 # error wrong numeric code for no-transaction 380 #endif 381 #if TRANS_READ!=SQLITE_TXN_READ 382 # error wrong numeric code for read-transaction 383 #endif 384 #if TRANS_WRITE!=SQLITE_TXN_WRITE 385 # error wrong numeric code for write-transaction 386 #endif 387 388 389 /* 390 ** An instance of this object represents a single database file. 391 ** 392 ** A single database file can be in use at the same time by two 393 ** or more database connections. When two or more connections are 394 ** sharing the same database file, each connection has it own 395 ** private Btree object for the file and each of those Btrees points 396 ** to this one BtShared object. BtShared.nRef is the number of 397 ** connections currently sharing this database file. 398 ** 399 ** Fields in this structure are accessed under the BtShared.mutex 400 ** mutex, except for nRef and pNext which are accessed under the 401 ** global SQLITE_MUTEX_STATIC_MAIN mutex. The pPager field 402 ** may not be modified once it is initially set as long as nRef>0. 403 ** The pSchema field may be set once under BtShared.mutex and 404 ** thereafter is unchanged as long as nRef>0. 405 ** 406 ** isPending: 407 ** 408 ** If a BtShared client fails to obtain a write-lock on a database 409 ** table (because there exists one or more read-locks on the table), 410 ** the shared-cache enters 'pending-lock' state and isPending is 411 ** set to true. 412 ** 413 ** The shared-cache leaves the 'pending lock' state when either of 414 ** the following occur: 415 ** 416 ** 1) The current writer (BtShared.pWriter) concludes its transaction, OR 417 ** 2) The number of locks held by other connections drops to zero. 418 ** 419 ** while in the 'pending-lock' state, no connection may start a new 420 ** transaction. 421 ** 422 ** This feature is included to help prevent writer-starvation. 423 */ 424 struct BtShared { 425 Pager *pPager; /* The page cache */ 426 sqlite3 *db; /* Database connection currently using this Btree */ 427 BtCursor *pCursor; /* A list of all open cursors */ 428 MemPage *pPage1; /* First page of the database */ 429 u8 openFlags; /* Flags to sqlite3BtreeOpen() */ 430 #ifndef SQLITE_OMIT_AUTOVACUUM 431 u8 autoVacuum; /* True if auto-vacuum is enabled */ 432 u8 incrVacuum; /* True if incr-vacuum is enabled */ 433 u8 bDoTruncate; /* True to truncate db on commit */ 434 #endif 435 u8 inTransaction; /* Transaction state */ 436 u8 max1bytePayload; /* Maximum first byte of cell for a 1-byte payload */ 437 u8 nReserveWanted; /* Desired number of extra bytes per page */ 438 u16 btsFlags; /* Boolean parameters. See BTS_* macros below */ 439 u16 maxLocal; /* Maximum local payload in non-LEAFDATA tables */ 440 u16 minLocal; /* Minimum local payload in non-LEAFDATA tables */ 441 u16 maxLeaf; /* Maximum local payload in a LEAFDATA table */ 442 u16 minLeaf; /* Minimum local payload in a LEAFDATA table */ 443 u32 pageSize; /* Total number of bytes on a page */ 444 u32 usableSize; /* Number of usable bytes on each page */ 445 int nTransaction; /* Number of open transactions (read + write) */ 446 u32 nPage; /* Number of pages in the database */ 447 void *pSchema; /* Pointer to space allocated by sqlite3BtreeSchema() */ 448 void (*xFreeSchema)(void*); /* Destructor for BtShared.pSchema */ 449 sqlite3_mutex *mutex; /* Non-recursive mutex required to access this object */ 450 Bitvec *pHasContent; /* Set of pages moved to free-list this transaction */ 451 #ifndef SQLITE_OMIT_SHARED_CACHE 452 int nRef; /* Number of references to this structure */ 453 BtShared *pNext; /* Next on a list of sharable BtShared structs */ 454 BtLock *pLock; /* List of locks held on this shared-btree struct */ 455 Btree *pWriter; /* Btree with currently open write transaction */ 456 #endif 457 u8 *pTmpSpace; /* Temp space sufficient to hold a single cell */ 458 }; 459 460 /* 461 ** Allowed values for BtShared.btsFlags 462 */ 463 #define BTS_READ_ONLY 0x0001 /* Underlying file is readonly */ 464 #define BTS_PAGESIZE_FIXED 0x0002 /* Page size can no longer be changed */ 465 #define BTS_SECURE_DELETE 0x0004 /* PRAGMA secure_delete is enabled */ 466 #define BTS_OVERWRITE 0x0008 /* Overwrite deleted content with zeros */ 467 #define BTS_FAST_SECURE 0x000c /* Combination of the previous two */ 468 #define BTS_INITIALLY_EMPTY 0x0010 /* Database was empty at trans start */ 469 #define BTS_NO_WAL 0x0020 /* Do not open write-ahead-log files */ 470 #define BTS_EXCLUSIVE 0x0040 /* pWriter has an exclusive lock */ 471 #define BTS_PENDING 0x0080 /* Waiting for read-locks to clear */ 472 473 /* 474 ** An instance of the following structure is used to hold information 475 ** about a cell. The parseCellPtr() function fills in this structure 476 ** based on information extract from the raw disk page. 477 */ 478 struct CellInfo { 479 i64 nKey; /* The key for INTKEY tables, or nPayload otherwise */ 480 u8 *pPayload; /* Pointer to the start of payload */ 481 u32 nPayload; /* Bytes of payload */ 482 u16 nLocal; /* Amount of payload held locally, not on overflow */ 483 u16 nSize; /* Size of the cell content on the main b-tree page */ 484 }; 485 486 /* 487 ** Maximum depth of an SQLite B-Tree structure. Any B-Tree deeper than 488 ** this will be declared corrupt. This value is calculated based on a 489 ** maximum database size of 2^31 pages a minimum fanout of 2 for a 490 ** root-node and 3 for all other internal nodes. 491 ** 492 ** If a tree that appears to be taller than this is encountered, it is 493 ** assumed that the database is corrupt. 494 */ 495 #define BTCURSOR_MAX_DEPTH 20 496 497 /* 498 ** A cursor is a pointer to a particular entry within a particular 499 ** b-tree within a database file. 500 ** 501 ** The entry is identified by its MemPage and the index in 502 ** MemPage.aCell[] of the entry. 503 ** 504 ** A single database file can be shared by two more database connections, 505 ** but cursors cannot be shared. Each cursor is associated with a 506 ** particular database connection identified BtCursor.pBtree.db. 507 ** 508 ** Fields in this structure are accessed under the BtShared.mutex 509 ** found at self->pBt->mutex. 510 ** 511 ** skipNext meaning: 512 ** The meaning of skipNext depends on the value of eState: 513 ** 514 ** eState Meaning of skipNext 515 ** VALID skipNext is meaningless and is ignored 516 ** INVALID skipNext is meaningless and is ignored 517 ** SKIPNEXT sqlite3BtreeNext() is a no-op if skipNext>0 and 518 ** sqlite3BtreePrevious() is no-op if skipNext<0. 519 ** REQUIRESEEK restoreCursorPosition() restores the cursor to 520 ** eState=SKIPNEXT if skipNext!=0 521 ** FAULT skipNext holds the cursor fault error code. 522 */ 523 struct BtCursor { 524 u8 eState; /* One of the CURSOR_XXX constants (see below) */ 525 u8 curFlags; /* zero or more BTCF_* flags defined below */ 526 u8 curPagerFlags; /* Flags to send to sqlite3PagerGet() */ 527 u8 hints; /* As configured by CursorSetHints() */ 528 int skipNext; /* Prev() is noop if negative. Next() is noop if positive. 529 ** Error code if eState==CURSOR_FAULT */ 530 Btree *pBtree; /* The Btree to which this cursor belongs */ 531 Pgno *aOverflow; /* Cache of overflow page locations */ 532 void *pKey; /* Saved key that was cursor last known position */ 533 /* All fields above are zeroed when the cursor is allocated. See 534 ** sqlite3BtreeCursorZero(). Fields that follow must be manually 535 ** initialized. */ 536 #define BTCURSOR_FIRST_UNINIT pBt /* Name of first uninitialized field */ 537 BtShared *pBt; /* The BtShared this cursor points to */ 538 BtCursor *pNext; /* Forms a linked list of all cursors */ 539 CellInfo info; /* A parse of the cell we are pointing at */ 540 i64 nKey; /* Size of pKey, or last integer key */ 541 Pgno pgnoRoot; /* The root page of this tree */ 542 i8 iPage; /* Index of current page in apPage */ 543 u8 curIntKey; /* Value of apPage[0]->intKey */ 544 u16 ix; /* Current index for apPage[iPage] */ 545 u16 aiIdx[BTCURSOR_MAX_DEPTH-1]; /* Current index in apPage[i] */ 546 struct KeyInfo *pKeyInfo; /* Arg passed to comparison function */ 547 MemPage *pPage; /* Current page */ 548 MemPage *apPage[BTCURSOR_MAX_DEPTH-1]; /* Stack of parents of current page */ 549 }; 550 551 /* 552 ** Legal values for BtCursor.curFlags 553 */ 554 #define BTCF_WriteFlag 0x01 /* True if a write cursor */ 555 #define BTCF_ValidNKey 0x02 /* True if info.nKey is valid */ 556 #define BTCF_ValidOvfl 0x04 /* True if aOverflow is valid */ 557 #define BTCF_AtLast 0x08 /* Cursor is pointing ot the last entry */ 558 #define BTCF_Incrblob 0x10 /* True if an incremental I/O handle */ 559 #define BTCF_Multiple 0x20 /* Maybe another cursor on the same btree */ 560 #define BTCF_Pinned 0x40 /* Cursor is busy and cannot be moved */ 561 562 /* 563 ** Potential values for BtCursor.eState. 564 ** 565 ** CURSOR_INVALID: 566 ** Cursor does not point to a valid entry. This can happen (for example) 567 ** because the table is empty or because BtreeCursorFirst() has not been 568 ** called. 569 ** 570 ** CURSOR_VALID: 571 ** Cursor points to a valid entry. getPayload() etc. may be called. 572 ** 573 ** CURSOR_SKIPNEXT: 574 ** Cursor is valid except that the Cursor.skipNext field is non-zero 575 ** indicating that the next sqlite3BtreeNext() or sqlite3BtreePrevious() 576 ** operation should be a no-op. 577 ** 578 ** CURSOR_REQUIRESEEK: 579 ** The table that this cursor was opened on still exists, but has been 580 ** modified since the cursor was last used. The cursor position is saved 581 ** in variables BtCursor.pKey and BtCursor.nKey. When a cursor is in 582 ** this state, restoreCursorPosition() can be called to attempt to 583 ** seek the cursor to the saved position. 584 ** 585 ** CURSOR_FAULT: 586 ** An unrecoverable error (an I/O error or a malloc failure) has occurred 587 ** on a different connection that shares the BtShared cache with this 588 ** cursor. The error has left the cache in an inconsistent state. 589 ** Do nothing else with this cursor. Any attempt to use the cursor 590 ** should return the error code stored in BtCursor.skipNext 591 */ 592 #define CURSOR_VALID 0 593 #define CURSOR_INVALID 1 594 #define CURSOR_SKIPNEXT 2 595 #define CURSOR_REQUIRESEEK 3 596 #define CURSOR_FAULT 4 597 598 /* 599 ** The database page the PENDING_BYTE occupies. This page is never used. 600 */ 601 # define PENDING_BYTE_PAGE(pBt) PAGER_MJ_PGNO(pBt) 602 603 /* 604 ** These macros define the location of the pointer-map entry for a 605 ** database page. The first argument to each is the number of usable 606 ** bytes on each page of the database (often 1024). The second is the 607 ** page number to look up in the pointer map. 608 ** 609 ** PTRMAP_PAGENO returns the database page number of the pointer-map 610 ** page that stores the required pointer. PTRMAP_PTROFFSET returns 611 ** the offset of the requested map entry. 612 ** 613 ** If the pgno argument passed to PTRMAP_PAGENO is a pointer-map page, 614 ** then pgno is returned. So (pgno==PTRMAP_PAGENO(pgsz, pgno)) can be 615 ** used to test if pgno is a pointer-map page. PTRMAP_ISPAGE implements 616 ** this test. 617 */ 618 #define PTRMAP_PAGENO(pBt, pgno) ptrmapPageno(pBt, pgno) 619 #define PTRMAP_PTROFFSET(pgptrmap, pgno) (5*(pgno-pgptrmap-1)) 620 #define PTRMAP_ISPAGE(pBt, pgno) (PTRMAP_PAGENO((pBt),(pgno))==(pgno)) 621 622 /* 623 ** The pointer map is a lookup table that identifies the parent page for 624 ** each child page in the database file. The parent page is the page that 625 ** contains a pointer to the child. Every page in the database contains 626 ** 0 or 1 parent pages. (In this context 'database page' refers 627 ** to any page that is not part of the pointer map itself.) Each pointer map 628 ** entry consists of a single byte 'type' and a 4 byte parent page number. 629 ** The PTRMAP_XXX identifiers below are the valid types. 630 ** 631 ** The purpose of the pointer map is to facility moving pages from one 632 ** position in the file to another as part of autovacuum. When a page 633 ** is moved, the pointer in its parent must be updated to point to the 634 ** new location. The pointer map is used to locate the parent page quickly. 635 ** 636 ** PTRMAP_ROOTPAGE: The database page is a root-page. The page-number is not 637 ** used in this case. 638 ** 639 ** PTRMAP_FREEPAGE: The database page is an unused (free) page. The page-number 640 ** is not used in this case. 641 ** 642 ** PTRMAP_OVERFLOW1: The database page is the first page in a list of 643 ** overflow pages. The page number identifies the page that 644 ** contains the cell with a pointer to this overflow page. 645 ** 646 ** PTRMAP_OVERFLOW2: The database page is the second or later page in a list of 647 ** overflow pages. The page-number identifies the previous 648 ** page in the overflow page list. 649 ** 650 ** PTRMAP_BTREE: The database page is a non-root btree page. The page number 651 ** identifies the parent page in the btree. 652 */ 653 #define PTRMAP_ROOTPAGE 1 654 #define PTRMAP_FREEPAGE 2 655 #define PTRMAP_OVERFLOW1 3 656 #define PTRMAP_OVERFLOW2 4 657 #define PTRMAP_BTREE 5 658 659 /* A bunch of assert() statements to check the transaction state variables 660 ** of handle p (type Btree*) are internally consistent. 661 */ 662 #define btreeIntegrity(p) \ 663 assert( p->pBt->inTransaction!=TRANS_NONE || p->pBt->nTransaction==0 ); \ 664 assert( p->pBt->inTransaction>=p->inTrans ); 665 666 667 /* 668 ** The ISAUTOVACUUM macro is used within balance_nonroot() to determine 669 ** if the database supports auto-vacuum or not. Because it is used 670 ** within an expression that is an argument to another macro 671 ** (sqliteMallocRaw), it is not possible to use conditional compilation. 672 ** So, this macro is defined instead. 673 */ 674 #ifndef SQLITE_OMIT_AUTOVACUUM 675 #define ISAUTOVACUUM (pBt->autoVacuum) 676 #else 677 #define ISAUTOVACUUM 0 678 #endif 679 680 681 /* 682 ** This structure is passed around through all the sanity checking routines 683 ** in order to keep track of some global state information. 684 ** 685 ** The aRef[] array is allocated so that there is 1 bit for each page in 686 ** the database. As the integrity-check proceeds, for each page used in 687 ** the database the corresponding bit is set. This allows integrity-check to 688 ** detect pages that are used twice and orphaned pages (both of which 689 ** indicate corruption). 690 */ 691 typedef struct IntegrityCk IntegrityCk; 692 struct IntegrityCk { 693 BtShared *pBt; /* The tree being checked out */ 694 Pager *pPager; /* The associated pager. Also accessible by pBt->pPager */ 695 u8 *aPgRef; /* 1 bit per page in the db (see above) */ 696 Pgno nPage; /* Number of pages in the database */ 697 int mxErr; /* Stop accumulating errors when this reaches zero */ 698 int nErr; /* Number of messages written to zErrMsg so far */ 699 int bOomFault; /* A memory allocation error has occurred */ 700 const char *zPfx; /* Error message prefix */ 701 Pgno v1; /* Value for first %u substitution in zPfx */ 702 int v2; /* Value for second %d substitution in zPfx */ 703 StrAccum errMsg; /* Accumulate the error message text here */ 704 u32 *heap; /* Min-heap used for analyzing cell coverage */ 705 sqlite3 *db; /* Database connection running the check */ 706 }; 707 708 /* 709 ** Routines to read or write a two- and four-byte big-endian integer values. 710 */ 711 #define get2byte(x) ((x)[0]<<8 | (x)[1]) 712 #define put2byte(p,v) ((p)[0] = (u8)((v)>>8), (p)[1] = (u8)(v)) 713 #define get4byte sqlite3Get4byte 714 #define put4byte sqlite3Put4byte 715 716 /* 717 ** get2byteAligned(), unlike get2byte(), requires that its argument point to a 718 ** two-byte aligned address. get2bytea() is only used for accessing the 719 ** cell addresses in a btree header. 720 */ 721 #if SQLITE_BYTEORDER==4321 722 # define get2byteAligned(x) (*(u16*)(x)) 723 #elif SQLITE_BYTEORDER==1234 && GCC_VERSION>=4008000 724 # define get2byteAligned(x) __builtin_bswap16(*(u16*)(x)) 725 #elif SQLITE_BYTEORDER==1234 && MSVC_VERSION>=1300 726 # define get2byteAligned(x) _byteswap_ushort(*(u16*)(x)) 727 #else 728 # define get2byteAligned(x) ((x)[0]<<8 | (x)[1]) 729 #endif 730