xref: /sqlite-3.40.0/src/btreeInt.h (revision a408adc5)
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 1.4 2007/05/16 17:28:43 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 more
77 ** than once within the same second.  This counter, together with the
78 ** modification time of the file, allows other processes to know
79 ** when the file has changed and thus when they need to flush their
80 ** cache.
81 **
82 ** The max embedded payload fraction is the amount of the total usable
83 ** space in a page that can be consumed by a single cell for standard
84 ** B-tree (non-LEAFDATA) tables.  A value of 255 means 100%.  The default
85 ** is to limit the maximum cell size so that at least 4 cells will fit
86 ** on one page.  Thus the default max embedded payload fraction is 64.
87 **
88 ** If the payload for a cell is larger than the max payload, then extra
89 ** payload is spilled to overflow pages.  Once an overflow page is allocated,
90 ** as many bytes as possible are moved into the overflow pages without letting
91 ** the cell size drop below the min embedded payload fraction.
92 **
93 ** The min leaf payload fraction is like the min embedded payload fraction
94 ** except that it applies to leaf nodes in a LEAFDATA tree.  The maximum
95 ** payload fraction for a LEAFDATA tree is always 100% (or 255) and it
96 ** not specified in the header.
97 **
98 ** Each btree pages is divided into three sections:  The header, the
99 ** cell pointer array, and the cell area area.  Page 1 also has a 100-byte
100 ** file header that occurs before the page header.
101 **
102 **      |----------------|
103 **      | file header    |   100 bytes.  Page 1 only.
104 **      |----------------|
105 **      | page header    |   8 bytes for leaves.  12 bytes for interior nodes
106 **      |----------------|
107 **      | cell pointer   |   |  2 bytes per cell.  Sorted order.
108 **      | array          |   |  Grows downward
109 **      |                |   v
110 **      |----------------|
111 **      | unallocated    |
112 **      | space          |
113 **      |----------------|   ^  Grows upwards
114 **      | cell content   |   |  Arbitrary order interspersed with freeblocks.
115 **      | area           |   |  and free space fragments.
116 **      |----------------|
117 **
118 ** The page headers looks like this:
119 **
120 **   OFFSET   SIZE     DESCRIPTION
121 **      0       1      Flags. 1: intkey, 2: zerodata, 4: leafdata, 8: leaf
122 **      1       2      byte offset to the first freeblock
123 **      3       2      number of cells on this page
124 **      5       2      first byte of the cell content area
125 **      7       1      number of fragmented free bytes
126 **      8       4      Right child (the Ptr(N) value).  Omitted on leaves.
127 **
128 ** The flags define the format of this btree page.  The leaf flag means that
129 ** this page has no children.  The zerodata flag means that this page carries
130 ** only keys and no data.  The intkey flag means that the key is a integer
131 ** which is stored in the key size entry of the cell header rather than in
132 ** the payload area.
133 **
134 ** The cell pointer array begins on the first byte after the page header.
135 ** The cell pointer array contains zero or more 2-byte numbers which are
136 ** offsets from the beginning of the page to the cell content in the cell
137 ** content area.  The cell pointers occur in sorted order.  The system strives
138 ** to keep free space after the last cell pointer so that new cells can
139 ** be easily added without having to defragment the page.
140 **
141 ** Cell content is stored at the very end of the page and grows toward the
142 ** beginning of the page.
143 **
144 ** Unused space within the cell content area is collected into a linked list of
145 ** freeblocks.  Each freeblock is at least 4 bytes in size.  The byte offset
146 ** to the first freeblock is given in the header.  Freeblocks occur in
147 ** increasing order.  Because a freeblock must be at least 4 bytes in size,
148 ** any group of 3 or fewer unused bytes in the cell content area cannot
149 ** exist on the freeblock chain.  A group of 3 or fewer free bytes is called
150 ** a fragment.  The total number of bytes in all fragments is recorded.
151 ** in the page header at offset 7.
152 **
153 **    SIZE    DESCRIPTION
154 **      2     Byte offset of the next freeblock
155 **      2     Bytes in this freeblock
156 **
157 ** Cells are of variable length.  Cells are stored in the cell content area at
158 ** the end of the page.  Pointers to the cells are in the cell pointer array
159 ** that immediately follows the page header.  Cells is not necessarily
160 ** contiguous or in order, but cell pointers are contiguous and in order.
161 **
162 ** Cell content makes use of variable length integers.  A variable
163 ** length integer is 1 to 9 bytes where the lower 7 bits of each
164 ** byte are used.  The integer consists of all bytes that have bit 8 set and
165 ** the first byte with bit 8 clear.  The most significant byte of the integer
166 ** appears first.  A variable-length integer may not be more than 9 bytes long.
167 ** As a special case, all 8 bytes of the 9th byte are used as data.  This
168 ** allows a 64-bit integer to be encoded in 9 bytes.
169 **
170 **    0x00                      becomes  0x00000000
171 **    0x7f                      becomes  0x0000007f
172 **    0x81 0x00                 becomes  0x00000080
173 **    0x82 0x00                 becomes  0x00000100
174 **    0x80 0x7f                 becomes  0x0000007f
175 **    0x8a 0x91 0xd1 0xac 0x78  becomes  0x12345678
176 **    0x81 0x81 0x81 0x81 0x01  becomes  0x10204081
177 **
178 ** Variable length integers are used for rowids and to hold the number of
179 ** bytes of key and data in a btree cell.
180 **
181 ** The content of a cell looks like this:
182 **
183 **    SIZE    DESCRIPTION
184 **      4     Page number of the left child. Omitted if leaf flag is set.
185 **     var    Number of bytes of data. Omitted if the zerodata flag is set.
186 **     var    Number of bytes of key. Or the key itself if intkey flag is set.
187 **      *     Payload
188 **      4     First page of the overflow chain.  Omitted if no overflow
189 **
190 ** Overflow pages form a linked list.  Each page except the last is completely
191 ** filled with data (pagesize - 4 bytes).  The last page can have as little
192 ** as 1 byte of data.
193 **
194 **    SIZE    DESCRIPTION
195 **      4     Page number of next overflow page
196 **      *     Data
197 **
198 ** Freelist pages come in two subtypes: trunk pages and leaf pages.  The
199 ** file header points to first in a linked list of trunk page.  Each trunk
200 ** page points to multiple leaf pages.  The content of a leaf page is
201 ** unspecified.  A trunk page looks like this:
202 **
203 **    SIZE    DESCRIPTION
204 **      4     Page number of next trunk page
205 **      4     Number of leaf pointers on this page
206 **      *     zero or more pages numbers of leaves
207 */
208 #include "sqliteInt.h"
209 #include "pager.h"
210 #include "btree.h"
211 #include "os.h"
212 #include <assert.h>
213 
214 /* Round up a number to the next larger multiple of 8.  This is used
215 ** to force 8-byte alignment on 64-bit architectures.
216 */
217 #define ROUND8(x)   ((x+7)&~7)
218 
219 
220 /* The following value is the maximum cell size assuming a maximum page
221 ** size give above.
222 */
223 #define MX_CELL_SIZE(pBt)  (pBt->pageSize-8)
224 
225 /* The maximum number of cells on a single page of the database.  This
226 ** assumes a minimum cell size of 3 bytes.  Such small cells will be
227 ** exceedingly rare, but they are possible.
228 */
229 #define MX_CELL(pBt) ((pBt->pageSize-8)/3)
230 
231 /* Forward declarations */
232 typedef struct MemPage MemPage;
233 typedef struct BtLock BtLock;
234 
235 /*
236 ** This is a magic string that appears at the beginning of every
237 ** SQLite database in order to identify the file as a real database.
238 **
239 ** You can change this value at compile-time by specifying a
240 ** -DSQLITE_FILE_HEADER="..." on the compiler command-line.  The
241 ** header must be exactly 16 bytes including the zero-terminator so
242 ** the string itself should be 15 characters long.  If you change
243 ** the header, then your custom library will not be able to read
244 ** databases generated by the standard tools and the standard tools
245 ** will not be able to read databases created by your custom library.
246 */
247 #ifndef SQLITE_FILE_HEADER /* 123456789 123456 */
248 #  define SQLITE_FILE_HEADER "SQLite format 3"
249 #endif
250 
251 /*
252 ** Page type flags.  An ORed combination of these flags appear as the
253 ** first byte of every BTree page.
254 */
255 #define PTF_INTKEY    0x01
256 #define PTF_ZERODATA  0x02
257 #define PTF_LEAFDATA  0x04
258 #define PTF_LEAF      0x08
259 
260 /*
261 ** As each page of the file is loaded into memory, an instance of the following
262 ** structure is appended and initialized to zero.  This structure stores
263 ** information about the page that is decoded from the raw file page.
264 **
265 ** The pParent field points back to the parent page.  This allows us to
266 ** walk up the BTree from any leaf to the root.  Care must be taken to
267 ** unref() the parent page pointer when this page is no longer referenced.
268 ** The pageDestructor() routine handles that chore.
269 */
270 struct MemPage {
271   u8 isInit;           /* True if previously initialized. MUST BE FIRST! */
272   u8 idxShift;         /* True if Cell indices have changed */
273   u8 nOverflow;        /* Number of overflow cell bodies in aCell[] */
274   u8 intKey;           /* True if intkey flag is set */
275   u8 leaf;             /* True if leaf flag is set */
276   u8 zeroData;         /* True if table stores keys only */
277   u8 leafData;         /* True if tables stores data on leaves only */
278   u8 hasData;          /* True if this page stores data */
279   u8 hdrOffset;        /* 100 for page 1.  0 otherwise */
280   u8 childPtrSize;     /* 0 if leaf==1.  4 if leaf==0 */
281   u16 maxLocal;        /* Copy of Btree.maxLocal or Btree.maxLeaf */
282   u16 minLocal;        /* Copy of Btree.minLocal or Btree.minLeaf */
283   u16 cellOffset;      /* Index in aData of first cell pointer */
284   u16 idxParent;       /* Index in parent of this node */
285   u16 nFree;           /* Number of free bytes on the page */
286   u16 nCell;           /* Number of cells on this page, local and ovfl */
287   struct _OvflCell {   /* Cells that will not fit on aData[] */
288     u8 *pCell;          /* Pointers to the body of the overflow cell */
289     u16 idx;            /* Insert this cell before idx-th non-overflow cell */
290   } aOvfl[5];
291   BtShared *pBt;       /* Pointer back to BTree structure */
292   u8 *aData;           /* Pointer back to the start of the page */
293   DbPage *pDbPage;     /* Pager page handle */
294   Pgno pgno;           /* Page number for this page */
295   MemPage *pParent;    /* The parent of this page.  NULL for root */
296 };
297 
298 /*
299 ** The in-memory image of a disk page has the auxiliary information appended
300 ** to the end.  EXTRA_SIZE is the number of bytes of space needed to hold
301 ** that extra information.
302 */
303 #define EXTRA_SIZE sizeof(MemPage)
304 
305 /* Btree handle */
306 struct Btree {
307   sqlite3 *pSqlite;
308   BtShared *pBt;
309   u8 inTrans;            /* TRANS_NONE, TRANS_READ or TRANS_WRITE */
310 };
311 
312 /*
313 ** Btree.inTrans may take one of the following values.
314 **
315 ** If the shared-data extension is enabled, there may be multiple users
316 ** of the Btree structure. At most one of these may open a write transaction,
317 ** but any number may have active read transactions. Variable Btree.pDb
318 ** points to the handle that owns any current write-transaction.
319 */
320 #define TRANS_NONE  0
321 #define TRANS_READ  1
322 #define TRANS_WRITE 2
323 
324 /*
325 ** Everything we need to know about an open database
326 */
327 struct BtShared {
328   Pager *pPager;        /* The page cache */
329   BtCursor *pCursor;    /* A list of all open cursors */
330   MemPage *pPage1;      /* First page of the database */
331   u8 inStmt;            /* True if we are in a statement subtransaction */
332   u8 readOnly;          /* True if the underlying file is readonly */
333   u8 maxEmbedFrac;      /* Maximum payload as % of total page size */
334   u8 minEmbedFrac;      /* Minimum payload as % of total page size */
335   u8 minLeafFrac;       /* Minimum leaf payload as % of total page size */
336   u8 pageSizeFixed;     /* True if the page size can no longer be changed */
337 #ifndef SQLITE_OMIT_AUTOVACUUM
338   u8 autoVacuum;        /* True if auto-vacuum is enabled */
339   u8 incrVacuum;        /* True if incr-vacuum is enabled */
340   Pgno nTrunc;          /* Non-zero if the db will be truncated (incr vacuum) */
341 #endif
342   u16 pageSize;         /* Total number of bytes on a page */
343   u16 usableSize;       /* Number of usable bytes on each page */
344   int maxLocal;         /* Maximum local payload in non-LEAFDATA tables */
345   int minLocal;         /* Minimum local payload in non-LEAFDATA tables */
346   int maxLeaf;          /* Maximum local payload in a LEAFDATA table */
347   int minLeaf;          /* Minimum local payload in a LEAFDATA table */
348   BusyHandler *pBusyHandler;   /* Callback for when there is lock contention */
349   u8 inTransaction;     /* Transaction state */
350   int nRef;             /* Number of references to this structure */
351   int nTransaction;     /* Number of open transactions (read + write) */
352   void *pSchema;        /* Pointer to space allocated by sqlite3BtreeSchema() */
353   void (*xFreeSchema)(void*);  /* Destructor for BtShared.pSchema */
354 #ifndef SQLITE_OMIT_SHARED_CACHE
355   BtLock *pLock;        /* List of locks held on this shared-btree struct */
356   BtShared *pNext;      /* Next in ThreadData.pBtree linked list */
357 #endif
358 };
359 
360 /*
361 ** An instance of the following structure is used to hold information
362 ** about a cell.  The parseCellPtr() function fills in this structure
363 ** based on information extract from the raw disk page.
364 */
365 typedef struct CellInfo CellInfo;
366 struct CellInfo {
367   u8 *pCell;     /* Pointer to the start of cell content */
368   i64 nKey;      /* The key for INTKEY tables, or number of bytes in key */
369   u32 nData;     /* Number of bytes of data */
370   u32 nPayload;  /* Total amount of payload */
371   u16 nHeader;   /* Size of the cell content header in bytes */
372   u16 nLocal;    /* Amount of payload held locally */
373   u16 iOverflow; /* Offset to overflow page number.  Zero if no overflow */
374   u16 nSize;     /* Size of the cell content on the main b-tree page */
375 };
376 
377 /*
378 ** A cursor is a pointer to a particular entry in the BTree.
379 ** The entry is identified by its MemPage and the index in
380 ** MemPage.aCell[] of the entry.
381 */
382 struct BtCursor {
383   Btree *pBtree;            /* The Btree to which this cursor belongs */
384   BtCursor *pNext, *pPrev;  /* Forms a linked list of all cursors */
385   int (*xCompare)(void*,int,const void*,int,const void*); /* Key comp func */
386   void *pArg;               /* First arg to xCompare() */
387   Pgno pgnoRoot;            /* The root page of this tree */
388   MemPage *pPage;           /* Page that contains the entry */
389   int idx;                  /* Index of the entry in pPage->aCell[] */
390   CellInfo info;            /* A parse of the cell we are pointing at */
391   u8 wrFlag;                /* True if writable */
392   u8 eState;                /* One of the CURSOR_XXX constants (see below) */
393   void *pKey;      /* Saved key that was cursor's last known position */
394   i64 nKey;        /* Size of pKey, or last integer key */
395   int skip;        /* (skip<0) -> Prev() is a no-op. (skip>0) -> Next() is */
396 #ifndef SQLITE_OMIT_INCRBLOB
397   u8 isIncrblobHandle;      /* True if this cursor is an incr. io handle */
398   Pgno *aOverflow;          /* Cache of overflow page locations */
399 #endif
400 };
401 
402 /*
403 ** Potential values for BtCursor.eState.
404 **
405 ** CURSOR_VALID:
406 **   Cursor points to a valid entry. getPayload() etc. may be called.
407 **
408 ** CURSOR_INVALID:
409 **   Cursor does not point to a valid entry. This can happen (for example)
410 **   because the table is empty or because BtreeCursorFirst() has not been
411 **   called.
412 **
413 ** CURSOR_REQUIRESEEK:
414 **   The table that this cursor was opened on still exists, but has been
415 **   modified since the cursor was last used. The cursor position is saved
416 **   in variables BtCursor.pKey and BtCursor.nKey. When a cursor is in
417 **   this state, restoreOrClearCursorPosition() can be called to attempt to
418 **   seek the cursor to the saved position.
419 */
420 #define CURSOR_INVALID           0
421 #define CURSOR_VALID             1
422 #define CURSOR_REQUIRESEEK       2
423 
424 /*
425 ** The TRACE macro will print high-level status information about the
426 ** btree operation when the global variable sqlite3_btree_trace is
427 ** enabled.
428 */
429 #if SQLITE_TEST
430 # define TRACE(X)   if( sqlite3_btree_trace ){ printf X; fflush(stdout); }
431 #else
432 # define TRACE(X)
433 #endif
434 
435 /*
436 ** Routines to read and write variable-length integers.  These used to
437 ** be defined locally, but now we use the varint routines in the util.c
438 ** file.
439 */
440 #define getVarint    sqlite3GetVarint
441 #define getVarint32(A,B)  ((*B=*(A))<=0x7f?1:sqlite3GetVarint32(A,B))
442 #define putVarint    sqlite3PutVarint
443 
444 /* The database page the PENDING_BYTE occupies. This page is never used.
445 ** TODO: This macro is very similary to PAGER_MJ_PGNO() in pager.c. They
446 ** should possibly be consolidated (presumably in pager.h).
447 **
448 ** If disk I/O is omitted (meaning that the database is stored purely
449 ** in memory) then there is no pending byte.
450 */
451 #ifdef SQLITE_OMIT_DISKIO
452 # define PENDING_BYTE_PAGE(pBt)  0x7fffffff
453 #else
454 # define PENDING_BYTE_PAGE(pBt) ((PENDING_BYTE/(pBt)->pageSize)+1)
455 #endif
456 
457 /*
458 ** A linked list of the following structures is stored at BtShared.pLock.
459 ** Locks are added (or upgraded from READ_LOCK to WRITE_LOCK) when a cursor
460 ** is opened on the table with root page BtShared.iTable. Locks are removed
461 ** from this list when a transaction is committed or rolled back, or when
462 ** a btree handle is closed.
463 */
464 struct BtLock {
465   Btree *pBtree;        /* Btree handle holding this lock */
466   Pgno iTable;          /* Root page of table */
467   u8 eLock;             /* READ_LOCK or WRITE_LOCK */
468   BtLock *pNext;        /* Next in BtShared.pLock list */
469 };
470 
471 /* Candidate values for BtLock.eLock */
472 #define READ_LOCK     1
473 #define WRITE_LOCK    2
474 
475 /*
476 ** These macros define the location of the pointer-map entry for a
477 ** database page. The first argument to each is the number of usable
478 ** bytes on each page of the database (often 1024). The second is the
479 ** page number to look up in the pointer map.
480 **
481 ** PTRMAP_PAGENO returns the database page number of the pointer-map
482 ** page that stores the required pointer. PTRMAP_PTROFFSET returns
483 ** the offset of the requested map entry.
484 **
485 ** If the pgno argument passed to PTRMAP_PAGENO is a pointer-map page,
486 ** then pgno is returned. So (pgno==PTRMAP_PAGENO(pgsz, pgno)) can be
487 ** used to test if pgno is a pointer-map page. PTRMAP_ISPAGE implements
488 ** this test.
489 */
490 #define PTRMAP_PAGENO(pBt, pgno) ptrmapPageno(pBt, pgno)
491 #define PTRMAP_PTROFFSET(pBt, pgno) (5*(pgno-ptrmapPageno(pBt, pgno)-1))
492 #define PTRMAP_ISPAGE(pBt, pgno) (PTRMAP_PAGENO((pBt),(pgno))==(pgno))
493 
494 /*
495 ** The pointer map is a lookup table that identifies the parent page for
496 ** each child page in the database file.  The parent page is the page that
497 ** contains a pointer to the child.  Every page in the database contains
498 ** 0 or 1 parent pages.  (In this context 'database page' refers
499 ** to any page that is not part of the pointer map itself.)  Each pointer map
500 ** entry consists of a single byte 'type' and a 4 byte parent page number.
501 ** The PTRMAP_XXX identifiers below are the valid types.
502 **
503 ** The purpose of the pointer map is to facility moving pages from one
504 ** position in the file to another as part of autovacuum.  When a page
505 ** is moved, the pointer in its parent must be updated to point to the
506 ** new location.  The pointer map is used to locate the parent page quickly.
507 **
508 ** PTRMAP_ROOTPAGE: The database page is a root-page. The page-number is not
509 **                  used in this case.
510 **
511 ** PTRMAP_FREEPAGE: The database page is an unused (free) page. The page-number
512 **                  is not used in this case.
513 **
514 ** PTRMAP_OVERFLOW1: The database page is the first page in a list of
515 **                   overflow pages. The page number identifies the page that
516 **                   contains the cell with a pointer to this overflow page.
517 **
518 ** PTRMAP_OVERFLOW2: The database page is the second or later page in a list of
519 **                   overflow pages. The page-number identifies the previous
520 **                   page in the overflow page list.
521 **
522 ** PTRMAP_BTREE: The database page is a non-root btree page. The page number
523 **               identifies the parent page in the btree.
524 */
525 #define PTRMAP_ROOTPAGE 1
526 #define PTRMAP_FREEPAGE 2
527 #define PTRMAP_OVERFLOW1 3
528 #define PTRMAP_OVERFLOW2 4
529 #define PTRMAP_BTREE 5
530 
531 /* A bunch of assert() statements to check the transaction state variables
532 ** of handle p (type Btree*) are internally consistent.
533 */
534 #define btreeIntegrity(p) \
535   assert( p->inTrans!=TRANS_NONE || p->pBt->nTransaction<p->pBt->nRef ); \
536   assert( p->pBt->nTransaction<=p->pBt->nRef ); \
537   assert( p->pBt->inTransaction!=TRANS_NONE || p->pBt->nTransaction==0 ); \
538   assert( p->pBt->inTransaction>=p->inTrans );
539 
540 
541 /*
542 ** The ISAUTOVACUUM macro is used within balance_nonroot() to determine
543 ** if the database supports auto-vacuum or not. Because it is used
544 ** within an expression that is an argument to another macro
545 ** (sqliteMallocRaw), it is not possible to use conditional compilation.
546 ** So, this macro is defined instead.
547 */
548 #ifndef SQLITE_OMIT_AUTOVACUUM
549 #define ISAUTOVACUUM (pBt->autoVacuum)
550 #else
551 #define ISAUTOVACUUM 0
552 #endif
553 
554 
555 /*
556 ** This structure is passed around through all the sanity checking routines
557 ** in order to keep track of some global state information.
558 */
559 typedef struct IntegrityCk IntegrityCk;
560 struct IntegrityCk {
561   BtShared *pBt;    /* The tree being checked out */
562   Pager *pPager;    /* The associated pager.  Also accessible by pBt->pPager */
563   int nPage;        /* Number of pages in the database */
564   int *anRef;       /* Number of times each page is referenced */
565   int mxErr;        /* Stop accumulating errors when this reaches zero */
566   char *zErrMsg;    /* An error message.  NULL if no errors seen. */
567   int nErr;         /* Number of messages written to zErrMsg so far */
568 };
569 
570 /*
571 ** Read or write a two- and four-byte big-endian integer values.
572 */
573 #define get2byte(x)   ((x)[0]<<8 | (x)[1])
574 #define put2byte(p,v) ((p)[0] = (v)>>8, (p)[1] = (v))
575 #define get4byte sqlite3Get4byte
576 #define put4byte sqlite3Put4byte
577 
578 /*
579 ** Internal routines that should be accessed by the btree layer only.
580 */
581 int sqlite3BtreeGetPage(BtShared*, Pgno, MemPage**, int);
582 int sqlite3BtreeInitPage(MemPage *pPage, MemPage *pParent);
583 void sqlite3BtreeParseCellPtr(MemPage*, u8*, CellInfo*);
584 void sqlite3BtreeParseCell(MemPage*, int, CellInfo*);
585 u8 *sqlite3BtreeFindCell(MemPage *pPage, int iCell);
586 int sqlite3BtreeRestoreOrClearCursorPosition(BtCursor *pCur);
587 void sqlite3BtreeGetTempCursor(BtCursor *pCur, BtCursor *pTempCur);
588 void sqlite3BtreeReleaseTempCursor(BtCursor *pCur);
589 int sqlite3BtreeIsRootPage(MemPage *pPage);
590 void sqlite3BtreeMoveToParent(BtCursor *pCur);
591