xref: /sqlite-3.40.0/src/btreeInt.h (revision b39187ae)
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 ** $Id: btreeInt.h,v 1.38 2008/12/27 15:23:13 danielk1977 Exp $
13 **
14 ** This file implements a external (disk-based) database using BTrees.
15 ** For a detailed discussion of BTrees, refer to
16 **
17 **     Donald E. Knuth, THE ART OF COMPUTER PROGRAMMING, Volume 3:
18 **     "Sorting And Searching", pages 473-480. Addison-Wesley
19 **     Publishing Company, Reading, Massachusetts.
20 **
21 ** The basic idea is that each page of the file contains N database
22 ** entries and N+1 pointers to subpages.
23 **
24 **   ----------------------------------------------------------------
25 **   |  Ptr(0) | Key(0) | Ptr(1) | Key(1) | ... | Key(N-1) | Ptr(N) |
26 **   ----------------------------------------------------------------
27 **
28 ** All of the keys on the page that Ptr(0) points to have values less
29 ** than Key(0).  All of the keys on page Ptr(1) and its subpages have
30 ** values greater than Key(0) and less than Key(1).  All of the keys
31 ** on Ptr(N) and its subpages have values greater than Key(N-1).  And
32 ** so forth.
33 **
34 ** Finding a particular key requires reading O(log(M)) pages from the
35 ** disk where M is the number of entries in the tree.
36 **
37 ** In this implementation, a single file can hold one or more separate
38 ** BTrees.  Each BTree is identified by the index of its root page.  The
39 ** key and data for any entry are combined to form the "payload".  A
40 ** fixed amount of payload can be carried directly on the database
41 ** page.  If the payload is larger than the preset amount then surplus
42 ** bytes are stored on overflow pages.  The payload for an entry
43 ** and the preceding pointer are combined to form a "Cell".  Each
44 ** page has a small header which contains the Ptr(N) pointer and other
45 ** information such as the size of key and data.
46 **
47 ** FORMAT DETAILS
48 **
49 ** The file is divided into pages.  The first page is called page 1,
50 ** the second is page 2, and so forth.  A page number of zero indicates
51 ** "no such page".  The page size can be anything between 512 and 65536.
52 ** Each page can be either a btree page, a freelist page or an overflow
53 ** page.
54 **
55 ** The first page is always a btree page.  The first 100 bytes of the first
56 ** page contain a special header (the "file header") that describes the file.
57 ** The format of the file header is as follows:
58 **
59 **   OFFSET   SIZE    DESCRIPTION
60 **      0      16     Header string: "SQLite format 3\000"
61 **     16       2     Page size in bytes.
62 **     18       1     File format write version
63 **     19       1     File format read version
64 **     20       1     Bytes of unused space at the end of each page
65 **     21       1     Max embedded payload fraction
66 **     22       1     Min embedded payload fraction
67 **     23       1     Min leaf payload fraction
68 **     24       4     File change counter
69 **     28       4     Reserved for future use
70 **     32       4     First freelist page
71 **     36       4     Number of freelist pages in the file
72 **     40      60     15 4-byte meta values passed to higher layers
73 **
74 ** All of the integer values are big-endian (most significant byte first).
75 **
76 ** The file change counter is incremented when the database is changed
77 ** This counter allows other processes to know when the file has changed
78 ** and thus when they need to flush their cache.
79 **
80 ** The max embedded payload fraction is the amount of the total usable
81 ** space in a page that can be consumed by a single cell for standard
82 ** B-tree (non-LEAFDATA) tables.  A value of 255 means 100%.  The default
83 ** is to limit the maximum cell size so that at least 4 cells will fit
84 ** on one page.  Thus the default max embedded payload fraction is 64.
85 **
86 ** If the payload for a cell is larger than the max payload, then extra
87 ** payload is spilled to overflow pages.  Once an overflow page is allocated,
88 ** as many bytes as possible are moved into the overflow pages without letting
89 ** the cell size drop below the min embedded payload fraction.
90 **
91 ** The min leaf payload fraction is like the min embedded payload fraction
92 ** except that it applies to leaf nodes in a LEAFDATA tree.  The maximum
93 ** payload fraction for a LEAFDATA tree is always 100% (or 255) and it
94 ** not specified in the header.
95 **
96 ** Each btree pages is divided into three sections:  The header, the
97 ** cell pointer array, and the cell content area.  Page 1 also has a 100-byte
98 ** file header that occurs before the page header.
99 **
100 **      |----------------|
101 **      | file header    |   100 bytes.  Page 1 only.
102 **      |----------------|
103 **      | page header    |   8 bytes for leaves.  12 bytes for interior nodes
104 **      |----------------|
105 **      | cell pointer   |   |  2 bytes per cell.  Sorted order.
106 **      | array          |   |  Grows downward
107 **      |                |   v
108 **      |----------------|
109 **      | unallocated    |
110 **      | space          |
111 **      |----------------|   ^  Grows upwards
112 **      | cell content   |   |  Arbitrary order interspersed with freeblocks.
113 **      | area           |   |  and free space fragments.
114 **      |----------------|
115 **
116 ** The page headers looks like this:
117 **
118 **   OFFSET   SIZE     DESCRIPTION
119 **      0       1      Flags. 1: intkey, 2: zerodata, 4: leafdata, 8: leaf
120 **      1       2      byte offset to the first freeblock
121 **      3       2      number of cells on this page
122 **      5       2      first byte of the cell content area
123 **      7       1      number of fragmented free bytes
124 **      8       4      Right child (the Ptr(N) value).  Omitted on leaves.
125 **
126 ** The flags define the format of this btree page.  The leaf flag means that
127 ** this page has no children.  The zerodata flag means that this page carries
128 ** only keys and no data.  The intkey flag means that the key is a integer
129 ** which is stored in the key size entry of the cell header rather than in
130 ** the payload area.
131 **
132 ** The cell pointer array begins on the first byte after the page header.
133 ** The cell pointer array contains zero or more 2-byte numbers which are
134 ** offsets from the beginning of the page to the cell content in the cell
135 ** content area.  The cell pointers occur in sorted order.  The system strives
136 ** to keep free space after the last cell pointer so that new cells can
137 ** be easily added without having to defragment the page.
138 **
139 ** Cell content is stored at the very end of the page and grows toward the
140 ** beginning of the page.
141 **
142 ** Unused space within the cell content area is collected into a linked list of
143 ** freeblocks.  Each freeblock is at least 4 bytes in size.  The byte offset
144 ** to the first freeblock is given in the header.  Freeblocks occur in
145 ** increasing order.  Because a freeblock must be at least 4 bytes in size,
146 ** any group of 3 or fewer unused bytes in the cell content area cannot
147 ** exist on the freeblock chain.  A group of 3 or fewer free bytes is called
148 ** a fragment.  The total number of bytes in all fragments is recorded.
149 ** in the page header at offset 7.
150 **
151 **    SIZE    DESCRIPTION
152 **      2     Byte offset of the next freeblock
153 **      2     Bytes in this freeblock
154 **
155 ** Cells are of variable length.  Cells are stored in the cell content area at
156 ** the end of the page.  Pointers to the cells are in the cell pointer array
157 ** that immediately follows the page header.  Cells is not necessarily
158 ** contiguous or in order, but cell pointers are contiguous and in order.
159 **
160 ** Cell content makes use of variable length integers.  A variable
161 ** length integer is 1 to 9 bytes where the lower 7 bits of each
162 ** byte are used.  The integer consists of all bytes that have bit 8 set and
163 ** the first byte with bit 8 clear.  The most significant byte of the integer
164 ** appears first.  A variable-length integer may not be more than 9 bytes long.
165 ** As a special case, all 8 bytes of the 9th byte are used as data.  This
166 ** allows a 64-bit integer to be encoded in 9 bytes.
167 **
168 **    0x00                      becomes  0x00000000
169 **    0x7f                      becomes  0x0000007f
170 **    0x81 0x00                 becomes  0x00000080
171 **    0x82 0x00                 becomes  0x00000100
172 **    0x80 0x7f                 becomes  0x0000007f
173 **    0x8a 0x91 0xd1 0xac 0x78  becomes  0x12345678
174 **    0x81 0x81 0x81 0x81 0x01  becomes  0x10204081
175 **
176 ** Variable length integers are used for rowids and to hold the number of
177 ** bytes of key and data in a btree cell.
178 **
179 ** The content of a cell looks like this:
180 **
181 **    SIZE    DESCRIPTION
182 **      4     Page number of the left child. Omitted if leaf flag is set.
183 **     var    Number of bytes of data. Omitted if the zerodata flag is set.
184 **     var    Number of bytes of key. Or the key itself if intkey flag is set.
185 **      *     Payload
186 **      4     First page of the overflow chain.  Omitted if no overflow
187 **
188 ** Overflow pages form a linked list.  Each page except the last is completely
189 ** filled with data (pagesize - 4 bytes).  The last page can have as little
190 ** as 1 byte of data.
191 **
192 **    SIZE    DESCRIPTION
193 **      4     Page number of next overflow page
194 **      *     Data
195 **
196 ** Freelist pages come in two subtypes: trunk pages and leaf pages.  The
197 ** file header points to the first in a linked list of trunk page.  Each trunk
198 ** page points to multiple leaf pages.  The content of a leaf page is
199 ** unspecified.  A trunk page looks like this:
200 **
201 **    SIZE    DESCRIPTION
202 **      4     Page number of next trunk page
203 **      4     Number of leaf pointers on this page
204 **      *     zero or more pages numbers of leaves
205 */
206 #include "sqliteInt.h"
207 #include "pager.h"
208 #include "btree.h"
209 #include "os.h"
210 #include <assert.h>
211 
212 /* Round up a number to the next larger multiple of 8.  This is used
213 ** to force 8-byte alignment on 64-bit architectures.
214 */
215 #define ROUND8(x)   ((x+7)&~7)
216 
217 
218 /* The following value is the maximum cell size assuming a maximum page
219 ** size give above.
220 */
221 #define MX_CELL_SIZE(pBt)  (pBt->pageSize-8)
222 
223 /* The maximum number of cells on a single page of the database.  This
224 ** assumes a minimum cell size of 6 bytes  (4 bytes for the cell itself
225 ** plus 2 bytes for the index to the cell in the page header).  Such
226 ** small cells will be rare, but they are possible.
227 */
228 #define MX_CELL(pBt) ((pBt->pageSize-8)/6)
229 
230 /* Forward declarations */
231 typedef struct MemPage MemPage;
232 typedef struct BtLock BtLock;
233 
234 /*
235 ** This is a magic string that appears at the beginning of every
236 ** SQLite database in order to identify the file as a real database.
237 **
238 ** You can change this value at compile-time by specifying a
239 ** -DSQLITE_FILE_HEADER="..." on the compiler command-line.  The
240 ** header must be exactly 16 bytes including the zero-terminator so
241 ** the string itself should be 15 characters long.  If you change
242 ** the header, then your custom library will not be able to read
243 ** databases generated by the standard tools and the standard tools
244 ** will not be able to read databases created by your custom library.
245 */
246 #ifndef SQLITE_FILE_HEADER /* 123456789 123456 */
247 #  define SQLITE_FILE_HEADER "SQLite format 3"
248 #endif
249 
250 /*
251 ** Page type flags.  An ORed combination of these flags appear as the
252 ** first byte of on-disk image of every BTree page.
253 */
254 #define PTF_INTKEY    0x01
255 #define PTF_ZERODATA  0x02
256 #define PTF_LEAFDATA  0x04
257 #define PTF_LEAF      0x08
258 
259 /*
260 ** As each page of the file is loaded into memory, an instance of the following
261 ** structure is appended and initialized to zero.  This structure stores
262 ** information about the page that is decoded from the raw file page.
263 **
264 ** The pParent field points back to the parent page.  This allows us to
265 ** walk up the BTree from any leaf to the root.  Care must be taken to
266 ** unref() the parent page pointer when this page is no longer referenced.
267 ** The pageDestructor() routine handles that chore.
268 **
269 ** Access to all fields of this structure is controlled by the mutex
270 ** stored in MemPage.pBt->mutex.
271 */
272 struct MemPage {
273   u8 isInit;           /* True if previously initialized. MUST BE FIRST! */
274   u8 nOverflow;        /* Number of overflow cell bodies in aCell[] */
275   u8 intKey;           /* True if intkey flag is set */
276   u8 leaf;             /* True if leaf flag is set */
277   u8 hasData;          /* True if this page stores data */
278   u8 hdrOffset;        /* 100 for page 1.  0 otherwise */
279   u8 childPtrSize;     /* 0 if leaf==1.  4 if leaf==0 */
280   u16 maxLocal;        /* Copy of BtShared.maxLocal or BtShared.maxLeaf */
281   u16 minLocal;        /* Copy of BtShared.minLocal or BtShared.minLeaf */
282   u16 cellOffset;      /* Index in aData of first cell pointer */
283   u16 nFree;           /* Number of free bytes on the page */
284   u16 nCell;           /* Number of cells on this page, local and ovfl */
285   u16 maskPage;        /* Mask for page offset */
286   struct _OvflCell {   /* Cells that will not fit on aData[] */
287     u8 *pCell;          /* Pointers to the body of the overflow cell */
288     u16 idx;            /* Insert this cell before idx-th non-overflow cell */
289   } aOvfl[5];
290   BtShared *pBt;       /* Pointer to BtShared that this page is part of */
291   u8 *aData;           /* Pointer to disk image of the page data */
292   DbPage *pDbPage;     /* Pager page handle */
293   Pgno pgno;           /* Page number for this page */
294 };
295 
296 /*
297 ** The in-memory image of a disk page has the auxiliary information appended
298 ** to the end.  EXTRA_SIZE is the number of bytes of space needed to hold
299 ** that extra information.
300 */
301 #define EXTRA_SIZE sizeof(MemPage)
302 
303 /* A Btree handle
304 **
305 ** A database connection contains a pointer to an instance of
306 ** this object for every database file that it has open.  This structure
307 ** is opaque to the database connection.  The database connection cannot
308 ** see the internals of this structure and only deals with pointers to
309 ** this structure.
310 **
311 ** For some database files, the same underlying database cache might be
312 ** shared between multiple connections.  In that case, each contection
313 ** has it own pointer to this object.  But each instance of this object
314 ** points to the same BtShared object.  The database cache and the
315 ** schema associated with the database file are all contained within
316 ** the BtShared object.
317 **
318 ** All fields in this structure are accessed under sqlite3.mutex.
319 ** The pBt pointer itself may not be changed while there exists cursors
320 ** in the referenced BtShared that point back to this Btree since those
321 ** cursors have to do go through this Btree to find their BtShared and
322 ** they often do so without holding sqlite3.mutex.
323 */
324 struct Btree {
325   sqlite3 *db;       /* The database connection holding this btree */
326   BtShared *pBt;     /* Sharable content of this btree */
327   u8 inTrans;        /* TRANS_NONE, TRANS_READ or TRANS_WRITE */
328   u8 sharable;       /* True if we can share pBt with another db */
329   u8 locked;         /* True if db currently has pBt locked */
330   int wantToLock;    /* Number of nested calls to sqlite3BtreeEnter() */
331   Btree *pNext;      /* List of other sharable Btrees from the same db */
332   Btree *pPrev;      /* Back pointer of the same list */
333 };
334 
335 /*
336 ** Btree.inTrans may take one of the following values.
337 **
338 ** If the shared-data extension is enabled, there may be multiple users
339 ** of the Btree structure. At most one of these may open a write transaction,
340 ** but any number may have active read transactions.
341 */
342 #define TRANS_NONE  0
343 #define TRANS_READ  1
344 #define TRANS_WRITE 2
345 
346 /*
347 ** An instance of this object represents a single database file.
348 **
349 ** A single database file can be in use as the same time by two
350 ** or more database connections.  When two or more connections are
351 ** sharing the same database file, each connection has it own
352 ** private Btree object for the file and each of those Btrees points
353 ** to this one BtShared object.  BtShared.nRef is the number of
354 ** connections currently sharing this database file.
355 **
356 ** Fields in this structure are accessed under the BtShared.mutex
357 ** mutex, except for nRef and pNext which are accessed under the
358 ** global SQLITE_MUTEX_STATIC_MASTER mutex.  The pPager field
359 ** may not be modified once it is initially set as long as nRef>0.
360 ** The pSchema field may be set once under BtShared.mutex and
361 ** thereafter is unchanged as long as nRef>0.
362 */
363 struct BtShared {
364   Pager *pPager;        /* The page cache */
365   sqlite3 *db;          /* Database connection currently using this Btree */
366   BtCursor *pCursor;    /* A list of all open cursors */
367   MemPage *pPage1;      /* First page of the database */
368   u8 inStmt;            /* True if we are in a statement subtransaction */
369   u8 readOnly;          /* True if the underlying file is readonly */
370   u8 pageSizeFixed;     /* True if the page size can no longer be changed */
371 #ifndef SQLITE_OMIT_AUTOVACUUM
372   u8 autoVacuum;        /* True if auto-vacuum is enabled */
373   u8 incrVacuum;        /* True if incr-vacuum is enabled */
374 #endif
375   u16 pageSize;         /* Total number of bytes on a page */
376   u16 usableSize;       /* Number of usable bytes on each page */
377   u16 maxLocal;         /* Maximum local payload in non-LEAFDATA tables */
378   u16 minLocal;         /* Minimum local payload in non-LEAFDATA tables */
379   u16 maxLeaf;          /* Maximum local payload in a LEAFDATA table */
380   u16 minLeaf;          /* Minimum local payload in a LEAFDATA table */
381   u8 inTransaction;     /* Transaction state */
382   int nTransaction;     /* Number of open transactions (read + write) */
383   void *pSchema;        /* Pointer to space allocated by sqlite3BtreeSchema() */
384   void (*xFreeSchema)(void*);  /* Destructor for BtShared.pSchema */
385   sqlite3_mutex *mutex; /* Non-recursive mutex required to access this struct */
386 #ifndef SQLITE_OMIT_SHARED_CACHE
387   int nRef;             /* Number of references to this structure */
388   BtShared *pNext;      /* Next on a list of sharable BtShared structs */
389   BtLock *pLock;        /* List of locks held on this shared-btree struct */
390   Btree *pExclusive;    /* Btree with an EXCLUSIVE lock on the whole db */
391 #endif
392   u8 *pTmpSpace;        /* BtShared.pageSize bytes of space for tmp use */
393 };
394 
395 /*
396 ** An instance of the following structure is used to hold information
397 ** about a cell.  The parseCellPtr() function fills in this structure
398 ** based on information extract from the raw disk page.
399 */
400 typedef struct CellInfo CellInfo;
401 struct CellInfo {
402   u8 *pCell;     /* Pointer to the start of cell content */
403   i64 nKey;      /* The key for INTKEY tables, or number of bytes in key */
404   u32 nData;     /* Number of bytes of data */
405   u32 nPayload;  /* Total amount of payload */
406   u16 nHeader;   /* Size of the cell content header in bytes */
407   u16 nLocal;    /* Amount of payload held locally */
408   u16 iOverflow; /* Offset to overflow page number.  Zero if no overflow */
409   u16 nSize;     /* Size of the cell content on the main b-tree page */
410 };
411 
412 /*
413 ** Maximum depth of an SQLite B-Tree structure. Any B-Tree deeper than
414 ** this will be declared corrupt. This value is calculated based on a
415 ** maximum database size of 2^31 pages a minimum fanout of 2 for a
416 ** root-node and 3 for all other internal nodes.
417 **
418 ** If a tree that appears to be taller than this is encountered, it is
419 ** assumed that the database is corrupt.
420 */
421 #define BTCURSOR_MAX_DEPTH 20
422 
423 /*
424 ** A cursor is a pointer to a particular entry within a particular
425 ** b-tree within a database file.
426 **
427 ** The entry is identified by its MemPage and the index in
428 ** MemPage.aCell[] of the entry.
429 **
430 ** When a single database file can shared by two more database connections,
431 ** but cursors cannot be shared.  Each cursor is associated with a
432 ** particular database connection identified BtCursor.pBtree.db.
433 **
434 ** Fields in this structure are accessed under the BtShared.mutex
435 ** found at self->pBt->mutex.
436 */
437 struct BtCursor {
438   Btree *pBtree;            /* The Btree to which this cursor belongs */
439   BtShared *pBt;            /* The BtShared this cursor points to */
440   BtCursor *pNext, *pPrev;  /* Forms a linked list of all cursors */
441   struct KeyInfo *pKeyInfo; /* Argument passed to comparison function */
442   Pgno pgnoRoot;            /* The root page of this tree */
443   CellInfo info;            /* A parse of the cell we are pointing at */
444   u8 wrFlag;                /* True if writable */
445   u8 atLast;                /* Cursor pointing to the last entry */
446   u8 validNKey;             /* True if info.nKey is valid */
447   u8 eState;                /* One of the CURSOR_XXX constants (see below) */
448   void *pKey;      /* Saved key that was cursor's last known position */
449   i64 nKey;        /* Size of pKey, or last integer key */
450   int skip;        /* (skip<0) -> Prev() is a no-op. (skip>0) -> Next() is */
451 #ifndef SQLITE_OMIT_INCRBLOB
452   u8 isIncrblobHandle;      /* True if this cursor is an incr. io handle */
453   Pgno *aOverflow;          /* Cache of overflow page locations */
454 #endif
455 #ifndef NDEBUG
456   u8 pagesShuffled;         /* True if Btree pages are rearranged by balance()*/
457 #endif
458   i16 iPage;                            /* Index of current page in apPage */
459   MemPage *apPage[BTCURSOR_MAX_DEPTH];  /* Pages from root to current page */
460   u16 aiIdx[BTCURSOR_MAX_DEPTH];        /* Current index in apPage[i] */
461 };
462 
463 /*
464 ** Potential values for BtCursor.eState.
465 **
466 ** CURSOR_VALID:
467 **   Cursor points to a valid entry. getPayload() etc. may be called.
468 **
469 ** CURSOR_INVALID:
470 **   Cursor does not point to a valid entry. This can happen (for example)
471 **   because the table is empty or because BtreeCursorFirst() has not been
472 **   called.
473 **
474 ** CURSOR_REQUIRESEEK:
475 **   The table that this cursor was opened on still exists, but has been
476 **   modified since the cursor was last used. The cursor position is saved
477 **   in variables BtCursor.pKey and BtCursor.nKey. When a cursor is in
478 **   this state, restoreCursorPosition() can be called to attempt to
479 **   seek the cursor to the saved position.
480 **
481 ** CURSOR_FAULT:
482 **   A unrecoverable error (an I/O error or a malloc failure) has occurred
483 **   on a different connection that shares the BtShared cache with this
484 **   cursor.  The error has left the cache in an inconsistent state.
485 **   Do nothing else with this cursor.  Any attempt to use the cursor
486 **   should return the error code stored in BtCursor.skip
487 */
488 #define CURSOR_INVALID           0
489 #define CURSOR_VALID             1
490 #define CURSOR_REQUIRESEEK       2
491 #define CURSOR_FAULT             3
492 
493 /* The database page the PENDING_BYTE occupies. This page is never used.
494 ** TODO: This macro is very similary to PAGER_MJ_PGNO() in pager.c. They
495 ** should possibly be consolidated (presumably in pager.h).
496 **
497 ** If disk I/O is omitted (meaning that the database is stored purely
498 ** in memory) then there is no pending byte.
499 */
500 #ifdef SQLITE_OMIT_DISKIO
501 # define PENDING_BYTE_PAGE(pBt)  0x7fffffff
502 #else
503 # define PENDING_BYTE_PAGE(pBt) ((Pgno)((PENDING_BYTE/(pBt)->pageSize)+1))
504 #endif
505 
506 /*
507 ** A linked list of the following structures is stored at BtShared.pLock.
508 ** Locks are added (or upgraded from READ_LOCK to WRITE_LOCK) when a cursor
509 ** is opened on the table with root page BtShared.iTable. Locks are removed
510 ** from this list when a transaction is committed or rolled back, or when
511 ** a btree handle is closed.
512 */
513 struct BtLock {
514   Btree *pBtree;        /* Btree handle holding this lock */
515   Pgno iTable;          /* Root page of table */
516   u8 eLock;             /* READ_LOCK or WRITE_LOCK */
517   BtLock *pNext;        /* Next in BtShared.pLock list */
518 };
519 
520 /* Candidate values for BtLock.eLock */
521 #define READ_LOCK     1
522 #define WRITE_LOCK    2
523 
524 /*
525 ** These macros define the location of the pointer-map entry for a
526 ** database page. The first argument to each is the number of usable
527 ** bytes on each page of the database (often 1024). The second is the
528 ** page number to look up in the pointer map.
529 **
530 ** PTRMAP_PAGENO returns the database page number of the pointer-map
531 ** page that stores the required pointer. PTRMAP_PTROFFSET returns
532 ** the offset of the requested map entry.
533 **
534 ** If the pgno argument passed to PTRMAP_PAGENO is a pointer-map page,
535 ** then pgno is returned. So (pgno==PTRMAP_PAGENO(pgsz, pgno)) can be
536 ** used to test if pgno is a pointer-map page. PTRMAP_ISPAGE implements
537 ** this test.
538 */
539 #define PTRMAP_PAGENO(pBt, pgno) ptrmapPageno(pBt, pgno)
540 #define PTRMAP_PTROFFSET(pgptrmap, pgno) (5*(pgno-pgptrmap-1))
541 #define PTRMAP_ISPAGE(pBt, pgno) (PTRMAP_PAGENO((pBt),(pgno))==(pgno))
542 
543 /*
544 ** The pointer map is a lookup table that identifies the parent page for
545 ** each child page in the database file.  The parent page is the page that
546 ** contains a pointer to the child.  Every page in the database contains
547 ** 0 or 1 parent pages.  (In this context 'database page' refers
548 ** to any page that is not part of the pointer map itself.)  Each pointer map
549 ** entry consists of a single byte 'type' and a 4 byte parent page number.
550 ** The PTRMAP_XXX identifiers below are the valid types.
551 **
552 ** The purpose of the pointer map is to facility moving pages from one
553 ** position in the file to another as part of autovacuum.  When a page
554 ** is moved, the pointer in its parent must be updated to point to the
555 ** new location.  The pointer map is used to locate the parent page quickly.
556 **
557 ** PTRMAP_ROOTPAGE: The database page is a root-page. The page-number is not
558 **                  used in this case.
559 **
560 ** PTRMAP_FREEPAGE: The database page is an unused (free) page. The page-number
561 **                  is not used in this case.
562 **
563 ** PTRMAP_OVERFLOW1: The database page is the first page in a list of
564 **                   overflow pages. The page number identifies the page that
565 **                   contains the cell with a pointer to this overflow page.
566 **
567 ** PTRMAP_OVERFLOW2: The database page is the second or later page in a list of
568 **                   overflow pages. The page-number identifies the previous
569 **                   page in the overflow page list.
570 **
571 ** PTRMAP_BTREE: The database page is a non-root btree page. The page number
572 **               identifies the parent page in the btree.
573 */
574 #define PTRMAP_ROOTPAGE 1
575 #define PTRMAP_FREEPAGE 2
576 #define PTRMAP_OVERFLOW1 3
577 #define PTRMAP_OVERFLOW2 4
578 #define PTRMAP_BTREE 5
579 
580 /* A bunch of assert() statements to check the transaction state variables
581 ** of handle p (type Btree*) are internally consistent.
582 */
583 #define btreeIntegrity(p) \
584   assert( p->pBt->inTransaction!=TRANS_NONE || p->pBt->nTransaction==0 ); \
585   assert( p->pBt->inTransaction>=p->inTrans );
586 
587 
588 /*
589 ** The ISAUTOVACUUM macro is used within balance_nonroot() to determine
590 ** if the database supports auto-vacuum or not. Because it is used
591 ** within an expression that is an argument to another macro
592 ** (sqliteMallocRaw), it is not possible to use conditional compilation.
593 ** So, this macro is defined instead.
594 */
595 #ifndef SQLITE_OMIT_AUTOVACUUM
596 #define ISAUTOVACUUM (pBt->autoVacuum)
597 #else
598 #define ISAUTOVACUUM 0
599 #endif
600 
601 
602 /*
603 ** This structure is passed around through all the sanity checking routines
604 ** in order to keep track of some global state information.
605 */
606 typedef struct IntegrityCk IntegrityCk;
607 struct IntegrityCk {
608   BtShared *pBt;    /* The tree being checked out */
609   Pager *pPager;    /* The associated pager.  Also accessible by pBt->pPager */
610   Pgno nPage;       /* Number of pages in the database */
611   int *anRef;       /* Number of times each page is referenced */
612   int mxErr;        /* Stop accumulating errors when this reaches zero */
613   int nErr;         /* Number of messages written to zErrMsg so far */
614   int mallocFailed; /* A memory allocation error has occurred */
615   StrAccum errMsg;  /* Accumulate the error message text here */
616 };
617 
618 /*
619 ** Read or write a two- and four-byte big-endian integer values.
620 */
621 #define get2byte(x)   ((x)[0]<<8 | (x)[1])
622 #define put2byte(p,v) ((p)[0] = (u8)((v)>>8), (p)[1] = (u8)(v))
623 #define get4byte sqlite3Get4byte
624 #define put4byte sqlite3Put4byte
625 
626 /*
627 ** Internal routines that should be accessed by the btree layer only.
628 */
629 int sqlite3BtreeGetPage(BtShared*, Pgno, MemPage**, int);
630 int sqlite3BtreeInitPage(MemPage *pPage);
631 void sqlite3BtreeParseCellPtr(MemPage*, u8*, CellInfo*);
632 void sqlite3BtreeParseCell(MemPage*, int, CellInfo*);
633 int sqlite3BtreeRestoreCursorPosition(BtCursor *pCur);
634 void sqlite3BtreeGetTempCursor(BtCursor *pCur, BtCursor *pTempCur);
635 void sqlite3BtreeReleaseTempCursor(BtCursor *pCur);
636 void sqlite3BtreeMoveToParent(BtCursor *pCur);
637