xref: /sqlite-3.40.0/src/btreeInt.h (revision 8a29dfde)
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 1.20 2008/03/29 16:01:04 drh 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 idxShift;         /* True if Cell indices have changed */
275   u8 nOverflow;        /* Number of overflow cell bodies in aCell[] */
276   u8 intKey;           /* True if intkey flag is set */
277   u8 leaf;             /* True if leaf flag is set */
278   u8 zeroData;         /* True if table stores keys only */
279   u8 leafData;         /* True if tables stores data on leaves only */
280   u8 hasData;          /* True if this page stores data */
281   u8 hdrOffset;        /* 100 for page 1.  0 otherwise */
282   u8 childPtrSize;     /* 0 if leaf==1.  4 if leaf==0 */
283   u16 maxLocal;        /* Copy of BtShared.maxLocal or BtShared.maxLeaf */
284   u16 minLocal;        /* Copy of BtShared.minLocal or BtShared.minLeaf */
285   u16 cellOffset;      /* Index in aData of first cell pointer */
286   u16 idxParent;       /* Index in parent of this node */
287   u16 nFree;           /* Number of free bytes on the page */
288   u16 nCell;           /* Number of cells on this page, local and ovfl */
289   struct _OvflCell {   /* Cells that will not fit on aData[] */
290     u8 *pCell;          /* Pointers to the body of the overflow cell */
291     u16 idx;            /* Insert this cell before idx-th non-overflow cell */
292   } aOvfl[5];
293   BtShared *pBt;       /* Pointer to BtShared that this page is part of */
294   u8 *aData;           /* Pointer to disk image of the page data */
295   DbPage *pDbPage;     /* Pager page handle */
296   Pgno pgno;           /* Page number for this page */
297   MemPage *pParent;    /* The parent of this page.  NULL for root */
298 };
299 
300 /*
301 ** The in-memory image of a disk page has the auxiliary information appended
302 ** to the end.  EXTRA_SIZE is the number of bytes of space needed to hold
303 ** that extra information.
304 */
305 #define EXTRA_SIZE sizeof(MemPage)
306 
307 /* A Btree handle
308 **
309 ** A database connection contains a pointer to an instance of
310 ** this object for every database file that it has open.  This structure
311 ** is opaque to the database connection.  The database connection cannot
312 ** see the internals of this structure and only deals with pointers to
313 ** this structure.
314 **
315 ** For some database files, the same underlying database cache might be
316 ** shared between multiple connections.  In that case, each contection
317 ** has it own pointer to this object.  But each instance of this object
318 ** points to the same BtShared object.  The database cache and the
319 ** schema associated with the database file are all contained within
320 ** the BtShared object.
321 **
322 ** All fields in this structure are accessed under sqlite3.mutex.
323 ** The pBt pointer itself may not be changed while there exists cursors
324 ** in the referenced BtShared that point back to this Btree since those
325 ** cursors have to do go through this Btree to find their BtShared and
326 ** they often do so without holding sqlite3.mutex.
327 */
328 struct Btree {
329   sqlite3 *db;       /* The database connection holding this btree */
330   BtShared *pBt;     /* Sharable content of this btree */
331   u8 inTrans;        /* TRANS_NONE, TRANS_READ or TRANS_WRITE */
332   u8 sharable;       /* True if we can share pBt with another db */
333   u8 locked;         /* True if db currently has pBt locked */
334   int wantToLock;    /* Number of nested calls to sqlite3BtreeEnter() */
335   Btree *pNext;      /* List of other sharable Btrees from the same db */
336   Btree *pPrev;      /* Back pointer of the same list */
337 };
338 
339 /*
340 ** Btree.inTrans may take one of the following values.
341 **
342 ** If the shared-data extension is enabled, there may be multiple users
343 ** of the Btree structure. At most one of these may open a write transaction,
344 ** but any number may have active read transactions.
345 */
346 #define TRANS_NONE  0
347 #define TRANS_READ  1
348 #define TRANS_WRITE 2
349 
350 /*
351 ** An instance of this object represents a single database file.
352 **
353 ** A single database file can be in use as the same time by two
354 ** or more database connections.  When two or more connections are
355 ** sharing the same database file, each connection has it own
356 ** private Btree object for the file and each of those Btrees points
357 ** to this one BtShared object.  BtShared.nRef is the number of
358 ** connections currently sharing this database file.
359 **
360 ** Fields in this structure are accessed under the BtShared.mutex
361 ** mutex, except for nRef and pNext which are accessed under the
362 ** global SQLITE_MUTEX_STATIC_MASTER mutex.  The pPager field
363 ** may not be modified once it is initially set as long as nRef>0.
364 ** The pSchema field may be set once under BtShared.mutex and
365 ** thereafter is unchanged as long as nRef>0.
366 */
367 struct BtShared {
368   Pager *pPager;        /* The page cache */
369   sqlite3 *db;          /* Database connection currently using this Btree */
370   BtCursor *pCursor;    /* A list of all open cursors */
371   MemPage *pPage1;      /* First page of the database */
372   u8 inStmt;            /* True if we are in a statement subtransaction */
373   u8 readOnly;          /* True if the underlying file is readonly */
374   u8 maxEmbedFrac;      /* Maximum payload as % of total page size */
375   u8 minEmbedFrac;      /* Minimum payload as % of total page size */
376   u8 minLeafFrac;       /* Minimum leaf payload as % of total page size */
377   u8 pageSizeFixed;     /* True if the page size can no longer be changed */
378 #ifndef SQLITE_OMIT_AUTOVACUUM
379   u8 autoVacuum;        /* True if auto-vacuum is enabled */
380   u8 incrVacuum;        /* True if incr-vacuum is enabled */
381   Pgno nTrunc;          /* Non-zero if the db will be truncated (incr vacuum) */
382 #endif
383   u16 pageSize;         /* Total number of bytes on a page */
384   u16 usableSize;       /* Number of usable bytes on each page */
385   int maxLocal;         /* Maximum local payload in non-LEAFDATA tables */
386   int minLocal;         /* Minimum local payload in non-LEAFDATA tables */
387   int maxLeaf;          /* Maximum local payload in a LEAFDATA table */
388   int minLeaf;          /* Minimum local payload in a LEAFDATA table */
389   u8 inTransaction;     /* Transaction state */
390   int nTransaction;     /* Number of open transactions (read + write) */
391   void *pSchema;        /* Pointer to space allocated by sqlite3BtreeSchema() */
392   void (*xFreeSchema)(void*);  /* Destructor for BtShared.pSchema */
393   sqlite3_mutex *mutex; /* Non-recursive mutex required to access this struct */
394   BusyHandler busyHdr;  /* The busy handler for this btree */
395 #ifndef SQLITE_OMIT_SHARED_CACHE
396   int nRef;             /* Number of references to this structure */
397   BtShared *pNext;      /* Next on a list of sharable BtShared structs */
398   BtLock *pLock;        /* List of locks held on this shared-btree struct */
399   Btree *pExclusive;    /* Btree with an EXCLUSIVE lock on the whole db */
400 #endif
401   u8 *pTmpSpace;        /* BtShared.pageSize bytes of space for tmp use */
402 };
403 
404 /*
405 ** An instance of the following structure is used to hold information
406 ** about a cell.  The parseCellPtr() function fills in this structure
407 ** based on information extract from the raw disk page.
408 */
409 typedef struct CellInfo CellInfo;
410 struct CellInfo {
411   u8 *pCell;     /* Pointer to the start of cell content */
412   i64 nKey;      /* The key for INTKEY tables, or number of bytes in key */
413   u32 nData;     /* Number of bytes of data */
414   u32 nPayload;  /* Total amount of payload */
415   u16 nHeader;   /* Size of the cell content header in bytes */
416   u16 nLocal;    /* Amount of payload held locally */
417   u16 iOverflow; /* Offset to overflow page number.  Zero if no overflow */
418   u16 nSize;     /* Size of the cell content on the main b-tree page */
419 };
420 
421 /*
422 ** A cursor is a pointer to a particular entry within a particular
423 ** b-tree within a database file.
424 **
425 ** The entry is identified by its MemPage and the index in
426 ** MemPage.aCell[] of the entry.
427 **
428 ** When a single database file can shared by two more database connections,
429 ** but cursors cannot be shared.  Each cursor is associated with a
430 ** particular database connection identified BtCursor.pBtree.db.
431 **
432 ** Fields in this structure are accessed under the BtShared.mutex
433 ** found at self->pBt->mutex.
434 */
435 struct BtCursor {
436   Btree *pBtree;            /* The Btree to which this cursor belongs */
437   BtShared *pBt;            /* The BtShared this cursor points to */
438   BtCursor *pNext, *pPrev;  /* Forms a linked list of all cursors */
439   struct KeyInfo *pKeyInfo; /* Argument passed to comparison function */
440   Pgno pgnoRoot;            /* The root page of this tree */
441   MemPage *pPage;           /* Page that contains the entry */
442   int idx;                  /* Index of the entry in pPage->aCell[] */
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 };
456 
457 /*
458 ** Potential values for BtCursor.eState.
459 **
460 ** CURSOR_VALID:
461 **   Cursor points to a valid entry. getPayload() etc. may be called.
462 **
463 ** CURSOR_INVALID:
464 **   Cursor does not point to a valid entry. This can happen (for example)
465 **   because the table is empty or because BtreeCursorFirst() has not been
466 **   called.
467 **
468 ** CURSOR_REQUIRESEEK:
469 **   The table that this cursor was opened on still exists, but has been
470 **   modified since the cursor was last used. The cursor position is saved
471 **   in variables BtCursor.pKey and BtCursor.nKey. When a cursor is in
472 **   this state, restoreOrClearCursorPosition() can be called to attempt to
473 **   seek the cursor to the saved position.
474 **
475 ** CURSOR_FAULT:
476 **   A unrecoverable error (an I/O error or a malloc failure) has occurred
477 **   on a different connection that shares the BtShared cache with this
478 **   cursor.  The error has left the cache in an inconsistent state.
479 **   Do nothing else with this cursor.  Any attempt to use the cursor
480 **   should return the error code stored in BtCursor.skip
481 */
482 #define CURSOR_INVALID           0
483 #define CURSOR_VALID             1
484 #define CURSOR_REQUIRESEEK       2
485 #define CURSOR_FAULT             3
486 
487 /*
488 ** The TRACE macro will print high-level status information about the
489 ** btree operation when the global variable sqlite3BtreeTrace is
490 ** enabled.
491 */
492 #if SQLITE_TEST
493 # define TRACE(X)   if( sqlite3BtreeTrace ){ printf X; fflush(stdout); }
494 #else
495 # define TRACE(X)
496 #endif
497 
498 /*
499 ** Routines to read and write variable-length integers.  These used to
500 ** be defined locally, but now we use the varint routines in the util.c
501 ** file.
502 */
503 #define getVarint    sqlite3GetVarint
504 #define getVarint32(A,B)  ((*B=*(A))<=0x7f?1:sqlite3GetVarint32(A,B))
505 #define putVarint    sqlite3PutVarint
506 
507 /* The database page the PENDING_BYTE occupies. This page is never used.
508 ** TODO: This macro is very similary to PAGER_MJ_PGNO() in pager.c. They
509 ** should possibly be consolidated (presumably in pager.h).
510 **
511 ** If disk I/O is omitted (meaning that the database is stored purely
512 ** in memory) then there is no pending byte.
513 */
514 #ifdef SQLITE_OMIT_DISKIO
515 # define PENDING_BYTE_PAGE(pBt)  0x7fffffff
516 #else
517 # define PENDING_BYTE_PAGE(pBt) ((PENDING_BYTE/(pBt)->pageSize)+1)
518 #endif
519 
520 /*
521 ** A linked list of the following structures is stored at BtShared.pLock.
522 ** Locks are added (or upgraded from READ_LOCK to WRITE_LOCK) when a cursor
523 ** is opened on the table with root page BtShared.iTable. Locks are removed
524 ** from this list when a transaction is committed or rolled back, or when
525 ** a btree handle is closed.
526 */
527 struct BtLock {
528   Btree *pBtree;        /* Btree handle holding this lock */
529   Pgno iTable;          /* Root page of table */
530   u8 eLock;             /* READ_LOCK or WRITE_LOCK */
531   BtLock *pNext;        /* Next in BtShared.pLock list */
532 };
533 
534 /* Candidate values for BtLock.eLock */
535 #define READ_LOCK     1
536 #define WRITE_LOCK    2
537 
538 /*
539 ** These macros define the location of the pointer-map entry for a
540 ** database page. The first argument to each is the number of usable
541 ** bytes on each page of the database (often 1024). The second is the
542 ** page number to look up in the pointer map.
543 **
544 ** PTRMAP_PAGENO returns the database page number of the pointer-map
545 ** page that stores the required pointer. PTRMAP_PTROFFSET returns
546 ** the offset of the requested map entry.
547 **
548 ** If the pgno argument passed to PTRMAP_PAGENO is a pointer-map page,
549 ** then pgno is returned. So (pgno==PTRMAP_PAGENO(pgsz, pgno)) can be
550 ** used to test if pgno is a pointer-map page. PTRMAP_ISPAGE implements
551 ** this test.
552 */
553 #define PTRMAP_PAGENO(pBt, pgno) ptrmapPageno(pBt, pgno)
554 #define PTRMAP_PTROFFSET(pBt, pgno) (5*(pgno-ptrmapPageno(pBt, pgno)-1))
555 #define PTRMAP_ISPAGE(pBt, pgno) (PTRMAP_PAGENO((pBt),(pgno))==(pgno))
556 
557 /*
558 ** The pointer map is a lookup table that identifies the parent page for
559 ** each child page in the database file.  The parent page is the page that
560 ** contains a pointer to the child.  Every page in the database contains
561 ** 0 or 1 parent pages.  (In this context 'database page' refers
562 ** to any page that is not part of the pointer map itself.)  Each pointer map
563 ** entry consists of a single byte 'type' and a 4 byte parent page number.
564 ** The PTRMAP_XXX identifiers below are the valid types.
565 **
566 ** The purpose of the pointer map is to facility moving pages from one
567 ** position in the file to another as part of autovacuum.  When a page
568 ** is moved, the pointer in its parent must be updated to point to the
569 ** new location.  The pointer map is used to locate the parent page quickly.
570 **
571 ** PTRMAP_ROOTPAGE: The database page is a root-page. The page-number is not
572 **                  used in this case.
573 **
574 ** PTRMAP_FREEPAGE: The database page is an unused (free) page. The page-number
575 **                  is not used in this case.
576 **
577 ** PTRMAP_OVERFLOW1: The database page is the first page in a list of
578 **                   overflow pages. The page number identifies the page that
579 **                   contains the cell with a pointer to this overflow page.
580 **
581 ** PTRMAP_OVERFLOW2: The database page is the second or later page in a list of
582 **                   overflow pages. The page-number identifies the previous
583 **                   page in the overflow page list.
584 **
585 ** PTRMAP_BTREE: The database page is a non-root btree page. The page number
586 **               identifies the parent page in the btree.
587 */
588 #define PTRMAP_ROOTPAGE 1
589 #define PTRMAP_FREEPAGE 2
590 #define PTRMAP_OVERFLOW1 3
591 #define PTRMAP_OVERFLOW2 4
592 #define PTRMAP_BTREE 5
593 
594 /* A bunch of assert() statements to check the transaction state variables
595 ** of handle p (type Btree*) are internally consistent.
596 */
597 #define btreeIntegrity(p) \
598   assert( p->pBt->inTransaction!=TRANS_NONE || p->pBt->nTransaction==0 ); \
599   assert( p->pBt->inTransaction>=p->inTrans );
600 
601 
602 /*
603 ** The ISAUTOVACUUM macro is used within balance_nonroot() to determine
604 ** if the database supports auto-vacuum or not. Because it is used
605 ** within an expression that is an argument to another macro
606 ** (sqliteMallocRaw), it is not possible to use conditional compilation.
607 ** So, this macro is defined instead.
608 */
609 #ifndef SQLITE_OMIT_AUTOVACUUM
610 #define ISAUTOVACUUM (pBt->autoVacuum)
611 #else
612 #define ISAUTOVACUUM 0
613 #endif
614 
615 
616 /*
617 ** This structure is passed around through all the sanity checking routines
618 ** in order to keep track of some global state information.
619 */
620 typedef struct IntegrityCk IntegrityCk;
621 struct IntegrityCk {
622   BtShared *pBt;    /* The tree being checked out */
623   Pager *pPager;    /* The associated pager.  Also accessible by pBt->pPager */
624   int nPage;        /* Number of pages in the database */
625   int *anRef;       /* Number of times each page is referenced */
626   int mxErr;        /* Stop accumulating errors when this reaches zero */
627   char *zErrMsg;    /* An error message.  NULL if no errors seen. */
628   int nErr;         /* Number of messages written to zErrMsg so far */
629 };
630 
631 /*
632 ** Read or write a two- and four-byte big-endian integer values.
633 */
634 #define get2byte(x)   ((x)[0]<<8 | (x)[1])
635 #define put2byte(p,v) ((p)[0] = (v)>>8, (p)[1] = (v))
636 #define get4byte sqlite3Get4byte
637 #define put4byte sqlite3Put4byte
638 
639 /*
640 ** Internal routines that should be accessed by the btree layer only.
641 */
642 int sqlite3BtreeGetPage(BtShared*, Pgno, MemPage**, int);
643 int sqlite3BtreeInitPage(MemPage *pPage, MemPage *pParent);
644 void sqlite3BtreeParseCellPtr(MemPage*, u8*, CellInfo*);
645 void sqlite3BtreeParseCell(MemPage*, int, CellInfo*);
646 #ifdef SQLITE_TEST
647 u8 *sqlite3BtreeFindCell(MemPage *pPage, int iCell);
648 #endif
649 int sqlite3BtreeRestoreOrClearCursorPosition(BtCursor *pCur);
650 void sqlite3BtreeGetTempCursor(BtCursor *pCur, BtCursor *pTempCur);
651 void sqlite3BtreeReleaseTempCursor(BtCursor *pCur);
652 int sqlite3BtreeIsRootPage(MemPage *pPage);
653 void sqlite3BtreeMoveToParent(BtCursor *pCur);
654