SQLite

Check-in [47ec874947]
Login

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview
Comment:Fix a failing assert in btree.c. The same bug was causing a spurious SQLITE_CORRUPT return when compiled without SQLITE_DEBUG. (CVS 6800)
Downloads: Tarball | ZIP archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA1: 47ec8749470af7cab9f3ef15effce1a7ba79a654
User & Date: danielk1977 2009-06-23 11:22:29.000
Context
2009-06-23
11:53
Add a test to verify that #3929 is fixed. (CVS 6801) (check-in: 16c1ae9bde user: danielk1977 tags: trunk)
11:22
Fix a failing assert in btree.c. The same bug was causing a spurious SQLITE_CORRUPT return when compiled without SQLITE_DEBUG. (CVS 6800) (check-in: 47ec874947 user: danielk1977 tags: trunk)
2009-06-22
19:05
Simplifications to vdbe.c and it service routines in support of coverage testing. (CVS 6799) (check-in: 308f2e6152 user: drh tags: trunk)
Changes
Unified Diff Ignore Whitespace Patch
Changes to src/btree.c.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
/*
** 2004 April 6
**
** The author disclaims copyright to this source code.  In place of
** a legal notice, here is a blessing:
**
**    May you do good and not evil.
**    May you find forgiveness for yourself and forgive others.
**    May you share freely, never taking more than you give.
**
*************************************************************************
** $Id: btree.c,v 1.638 2009/06/22 18:03:52 danielk1977 Exp $
**
** This file implements a external (disk-based) database using BTrees.
** See the header comment on "btreeInt.h" for additional information.
** Including a description of file format and an overview of operation.
*/
#include "btreeInt.h"












|







1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
/*
** 2004 April 6
**
** The author disclaims copyright to this source code.  In place of
** a legal notice, here is a blessing:
**
**    May you do good and not evil.
**    May you find forgiveness for yourself and forgive others.
**    May you share freely, never taking more than you give.
**
*************************************************************************
** $Id: btree.c,v 1.639 2009/06/23 11:22:29 danielk1977 Exp $
**
** This file implements a external (disk-based) database using BTrees.
** See the header comment on "btreeInt.h" for additional information.
** Including a description of file format and an overview of operation.
*/
#include "btreeInt.h"

6448
6449
6450
6451
6452
6453
6454
6455
6456

6457
6458
6459
6460
6461
6462
6463
6464
6465
6466
6467
6468
6469
6470
6471
6472
6473
6474


6475

6476
6477
6478
6479
6480
6481
6482
  /* If no error has occured and pPage has an overflow cell, call balance() 
  ** to redistribute the cells within the tree. Since balance() may move
  ** the cursor, zero the BtCursor.info.nSize and BtCursor.validNKey
  ** variables.
  **
  ** Previous versions of SQLite called moveToRoot() to move the cursor
  ** back to the root page as balance() used to invalidate the contents
  ** of BtCursor.apPage[] and BtCursor.aiIdx[]. This is no longer necessary,
  ** as balance() always leaves the cursor pointing to a valid entry.

  **
  ** There is a subtle but important optimization here too. When inserting
  ** multiple records into an intkey b-tree using a single cursor (as can
  ** happen while processing an "INSERT INTO ... SELECT" statement), it
  ** is advantageous to leave the cursor pointing to the last entry in
  ** the b-tree if possible. If the cursor is left pointing to the last
  ** entry in the table, and the next row inserted has an integer key
  ** larger than the largest existing key, it is possible to insert the
  ** row without seeking the cursor. This can be a big performance boost.
  */
  pCur->info.nSize = 0;
  pCur->validNKey = 0;
  if( rc==SQLITE_OK && pPage->nOverflow ){
    pCur->atLast = 0;
    rc = balance(pCur);

    /* Must make sure nOverflow is reset to zero even if the balance()
    ** fails.  Internal data structure corruption will result otherwise. */


    pCur->apPage[pCur->iPage]->nOverflow = 0;

  }
  assert( pCur->apPage[pCur->iPage]->nOverflow==0 );

end_insert:
  return rc;
}








|
|
>













<



|
>
>

>







6448
6449
6450
6451
6452
6453
6454
6455
6456
6457
6458
6459
6460
6461
6462
6463
6464
6465
6466
6467
6468
6469
6470

6471
6472
6473
6474
6475
6476
6477
6478
6479
6480
6481
6482
6483
6484
6485
  /* If no error has occured and pPage has an overflow cell, call balance() 
  ** to redistribute the cells within the tree. Since balance() may move
  ** the cursor, zero the BtCursor.info.nSize and BtCursor.validNKey
  ** variables.
  **
  ** Previous versions of SQLite called moveToRoot() to move the cursor
  ** back to the root page as balance() used to invalidate the contents
  ** of BtCursor.apPage[] and BtCursor.aiIdx[]. Instead of doing that,
  ** set the cursor state to "invalid". This makes common insert operations
  ** slightly faster.
  **
  ** There is a subtle but important optimization here too. When inserting
  ** multiple records into an intkey b-tree using a single cursor (as can
  ** happen while processing an "INSERT INTO ... SELECT" statement), it
  ** is advantageous to leave the cursor pointing to the last entry in
  ** the b-tree if possible. If the cursor is left pointing to the last
  ** entry in the table, and the next row inserted has an integer key
  ** larger than the largest existing key, it is possible to insert the
  ** row without seeking the cursor. This can be a big performance boost.
  */
  pCur->info.nSize = 0;
  pCur->validNKey = 0;
  if( rc==SQLITE_OK && pPage->nOverflow ){

    rc = balance(pCur);

    /* Must make sure nOverflow is reset to zero even if the balance()
    ** fails. Internal data structure corruption will result otherwise. 
    ** Also, set the cursor state to invalid. This stops saveCursorPosition()
    ** from trying to save the current position of the cursor.  */
    pCur->apPage[pCur->iPage]->nOverflow = 0;
    pCur->eState = CURSOR_INVALID;
  }
  assert( pCur->apPage[pCur->iPage]->nOverflow==0 );

end_insert:
  return rc;
}