xref: /sqlite-3.40.0/src/vacuum.c (revision 554cb87d)
1 /*
2 ** 2003 April 6
3 **
4 ** The author disclaims copyright to this source code.  In place of
5 ** a legal notice, here is a blessing:
6 **
7 **    May you do good and not evil.
8 **    May you find forgiveness for yourself and forgive others.
9 **    May you share freely, never taking more than you give.
10 **
11 *************************************************************************
12 ** This file contains code used to implement the VACUUM command.
13 **
14 ** Most of the code in this file may be omitted by defining the
15 ** SQLITE_OMIT_VACUUM macro.
16 */
17 #include "sqliteInt.h"
18 #include "vdbeInt.h"
19 
20 #if !defined(SQLITE_OMIT_VACUUM) && !defined(SQLITE_OMIT_ATTACH)
21 
22 /*
23 ** Execute zSql on database db.
24 **
25 ** If zSql returns rows, then each row will have exactly one
26 ** column.  (This will only happen if zSql begins with "SELECT".)
27 ** Take each row of result and call execSql() again recursively.
28 **
29 ** The execSqlF() routine does the same thing, except it accepts
30 ** a format string as its third argument
31 */
32 static int execSql(sqlite3 *db, char **pzErrMsg, const char *zSql){
33   sqlite3_stmt *pStmt;
34   int rc;
35 
36   /* printf("SQL: [%s]\n", zSql); fflush(stdout); */
37   rc = sqlite3_prepare_v2(db, zSql, -1, &pStmt, 0);
38   if( rc!=SQLITE_OK ) return rc;
39   while( SQLITE_ROW==(rc = sqlite3_step(pStmt)) ){
40     const char *zSubSql = (const char*)sqlite3_column_text(pStmt,0);
41     assert( sqlite3_strnicmp(zSql,"SELECT",6)==0 );
42     /* The secondary SQL must be one of CREATE TABLE, CREATE INDEX,
43     ** or INSERT.  Historically there have been attacks that first
44     ** corrupt the sqlite_schema.sql field with other kinds of statements
45     ** then run VACUUM to get those statements to execute at inappropriate
46     ** times. */
47     if( zSubSql
48      && (strncmp(zSubSql,"CRE",3)==0 || strncmp(zSubSql,"INS",3)==0)
49     ){
50       rc = execSql(db, pzErrMsg, zSubSql);
51       if( rc!=SQLITE_OK ) break;
52     }
53   }
54   assert( rc!=SQLITE_ROW );
55   if( rc==SQLITE_DONE ) rc = SQLITE_OK;
56   if( rc ){
57     sqlite3SetString(pzErrMsg, db, sqlite3_errmsg(db));
58   }
59   (void)sqlite3_finalize(pStmt);
60   return rc;
61 }
62 static int execSqlF(sqlite3 *db, char **pzErrMsg, const char *zSql, ...){
63   char *z;
64   va_list ap;
65   int rc;
66   va_start(ap, zSql);
67   z = sqlite3VMPrintf(db, zSql, ap);
68   va_end(ap);
69   if( z==0 ) return SQLITE_NOMEM;
70   rc = execSql(db, pzErrMsg, z);
71   sqlite3DbFree(db, z);
72   return rc;
73 }
74 
75 /*
76 ** The VACUUM command is used to clean up the database,
77 ** collapse free space, etc.  It is modelled after the VACUUM command
78 ** in PostgreSQL.  The VACUUM command works as follows:
79 **
80 **   (1)  Create a new transient database file
81 **   (2)  Copy all content from the database being vacuumed into
82 **        the new transient database file
83 **   (3)  Copy content from the transient database back into the
84 **        original database.
85 **
86 ** The transient database requires temporary disk space approximately
87 ** equal to the size of the original database.  The copy operation of
88 ** step (3) requires additional temporary disk space approximately equal
89 ** to the size of the original database for the rollback journal.
90 ** Hence, temporary disk space that is approximately 2x the size of the
91 ** original database is required.  Every page of the database is written
92 ** approximately 3 times:  Once for step (2) and twice for step (3).
93 ** Two writes per page are required in step (3) because the original
94 ** database content must be written into the rollback journal prior to
95 ** overwriting the database with the vacuumed content.
96 **
97 ** Only 1x temporary space and only 1x writes would be required if
98 ** the copy of step (3) were replaced by deleting the original database
99 ** and renaming the transient database as the original.  But that will
100 ** not work if other processes are attached to the original database.
101 ** And a power loss in between deleting the original and renaming the
102 ** transient would cause the database file to appear to be deleted
103 ** following reboot.
104 */
105 void sqlite3Vacuum(Parse *pParse, Token *pNm, Expr *pInto){
106   Vdbe *v = sqlite3GetVdbe(pParse);
107   int iDb = 0;
108   if( v==0 ) goto build_vacuum_end;
109   if( pParse->nErr ) goto build_vacuum_end;
110   if( pNm ){
111 #ifndef SQLITE_BUG_COMPATIBLE_20160819
112     /* Default behavior:  Report an error if the argument to VACUUM is
113     ** not recognized */
114     iDb = sqlite3TwoPartName(pParse, pNm, pNm, &pNm);
115     if( iDb<0 ) goto build_vacuum_end;
116 #else
117     /* When SQLITE_BUG_COMPATIBLE_20160819 is defined, unrecognized arguments
118     ** to VACUUM are silently ignored.  This is a back-out of a bug fix that
119     ** occurred on 2016-08-19 (https://www.sqlite.org/src/info/083f9e6270).
120     ** The buggy behavior is required for binary compatibility with some
121     ** legacy applications. */
122     iDb = sqlite3FindDb(pParse->db, pNm);
123     if( iDb<0 ) iDb = 0;
124 #endif
125   }
126   if( iDb!=1 ){
127     int iIntoReg = 0;
128     if( pInto && sqlite3ResolveSelfReference(pParse,0,0,pInto,0)==0 ){
129       iIntoReg = ++pParse->nMem;
130       sqlite3ExprCode(pParse, pInto, iIntoReg);
131     }
132     sqlite3VdbeAddOp2(v, OP_Vacuum, iDb, iIntoReg);
133     sqlite3VdbeUsesBtree(v, iDb);
134   }
135 build_vacuum_end:
136   sqlite3ExprDelete(pParse->db, pInto);
137   return;
138 }
139 
140 /*
141 ** This routine implements the OP_Vacuum opcode of the VDBE.
142 */
143 SQLITE_NOINLINE int sqlite3RunVacuum(
144   char **pzErrMsg,        /* Write error message here */
145   sqlite3 *db,            /* Database connection */
146   int iDb,                /* Which attached DB to vacuum */
147   sqlite3_value *pOut     /* Write results here, if not NULL. VACUUM INTO */
148 ){
149   int rc = SQLITE_OK;     /* Return code from service routines */
150   Btree *pMain;           /* The database being vacuumed */
151   Btree *pTemp;           /* The temporary database we vacuum into */
152   u32 saved_mDbFlags;     /* Saved value of db->mDbFlags */
153   u64 saved_flags;        /* Saved value of db->flags */
154   i64 saved_nChange;      /* Saved value of db->nChange */
155   i64 saved_nTotalChange; /* Saved value of db->nTotalChange */
156   u32 saved_openFlags;    /* Saved value of db->openFlags */
157   u8 saved_mTrace;        /* Saved trace settings */
158   Db *pDb = 0;            /* Database to detach at end of vacuum */
159   int isMemDb;            /* True if vacuuming a :memory: database */
160   int nRes;               /* Bytes of reserved space at the end of each page */
161   int nDb;                /* Number of attached databases */
162   const char *zDbMain;    /* Schema name of database to vacuum */
163   const char *zOut;       /* Name of output file */
164 
165   if( !db->autoCommit ){
166     sqlite3SetString(pzErrMsg, db, "cannot VACUUM from within a transaction");
167     return SQLITE_ERROR; /* IMP: R-12218-18073 */
168   }
169   if( db->nVdbeActive>1 ){
170     sqlite3SetString(pzErrMsg, db,"cannot VACUUM - SQL statements in progress");
171     return SQLITE_ERROR; /* IMP: R-15610-35227 */
172   }
173   saved_openFlags = db->openFlags;
174   if( pOut ){
175     if( sqlite3_value_type(pOut)!=SQLITE_TEXT ){
176       sqlite3SetString(pzErrMsg, db, "non-text filename");
177       return SQLITE_ERROR;
178     }
179     zOut = (const char*)sqlite3_value_text(pOut);
180     db->openFlags &= ~SQLITE_OPEN_READONLY;
181     db->openFlags |= SQLITE_OPEN_CREATE|SQLITE_OPEN_READWRITE;
182   }else{
183     zOut = "";
184   }
185 
186   /* Save the current value of the database flags so that it can be
187   ** restored before returning. Then set the writable-schema flag, and
188   ** disable CHECK and foreign key constraints.  */
189   saved_flags = db->flags;
190   saved_mDbFlags = db->mDbFlags;
191   saved_nChange = db->nChange;
192   saved_nTotalChange = db->nTotalChange;
193   saved_mTrace = db->mTrace;
194   db->flags |= SQLITE_WriteSchema | SQLITE_IgnoreChecks;
195   db->mDbFlags |= DBFLAG_PreferBuiltin | DBFLAG_Vacuum;
196   db->flags &= ~(u64)(SQLITE_ForeignKeys | SQLITE_ReverseOrder
197                    | SQLITE_Defensive | SQLITE_CountRows);
198   db->mTrace = 0;
199 
200   zDbMain = db->aDb[iDb].zDbSName;
201   pMain = db->aDb[iDb].pBt;
202   isMemDb = sqlite3PagerIsMemdb(sqlite3BtreePager(pMain));
203 
204   /* Attach the temporary database as 'vacuum_db'. The synchronous pragma
205   ** can be set to 'off' for this file, as it is not recovered if a crash
206   ** occurs anyway. The integrity of the database is maintained by a
207   ** (possibly synchronous) transaction opened on the main database before
208   ** sqlite3BtreeCopyFile() is called.
209   **
210   ** An optimisation would be to use a non-journaled pager.
211   ** (Later:) I tried setting "PRAGMA vacuum_db.journal_mode=OFF" but
212   ** that actually made the VACUUM run slower.  Very little journalling
213   ** actually occurs when doing a vacuum since the vacuum_db is initially
214   ** empty.  Only the journal header is written.  Apparently it takes more
215   ** time to parse and run the PRAGMA to turn journalling off than it does
216   ** to write the journal header file.
217   */
218   nDb = db->nDb;
219   rc = execSqlF(db, pzErrMsg, "ATTACH %Q AS vacuum_db", zOut);
220   db->openFlags = saved_openFlags;
221   if( rc!=SQLITE_OK ) goto end_of_vacuum;
222   assert( (db->nDb-1)==nDb );
223   pDb = &db->aDb[nDb];
224   assert( strcmp(pDb->zDbSName,"vacuum_db")==0 );
225   pTemp = pDb->pBt;
226   if( pOut ){
227     sqlite3_file *id = sqlite3PagerFile(sqlite3BtreePager(pTemp));
228     i64 sz = 0;
229     if( id->pMethods!=0 && (sqlite3OsFileSize(id, &sz)!=SQLITE_OK || sz>0) ){
230       rc = SQLITE_ERROR;
231       sqlite3SetString(pzErrMsg, db, "output file already exists");
232       goto end_of_vacuum;
233     }
234     db->mDbFlags |= DBFLAG_VacuumInto;
235   }
236   nRes = sqlite3BtreeGetRequestedReserve(pMain);
237 
238   sqlite3BtreeSetCacheSize(pTemp, db->aDb[iDb].pSchema->cache_size);
239   sqlite3BtreeSetSpillSize(pTemp, sqlite3BtreeSetSpillSize(pMain,0));
240   sqlite3BtreeSetPagerFlags(pTemp, PAGER_SYNCHRONOUS_OFF|PAGER_CACHESPILL);
241 
242   /* Begin a transaction and take an exclusive lock on the main database
243   ** file. This is done before the sqlite3BtreeGetPageSize(pMain) call below,
244   ** to ensure that we do not try to change the page-size on a WAL database.
245   */
246   rc = execSql(db, pzErrMsg, "BEGIN");
247   if( rc!=SQLITE_OK ) goto end_of_vacuum;
248   rc = sqlite3BtreeBeginTrans(pMain, pOut==0 ? 2 : 0, 0);
249   if( rc!=SQLITE_OK ) goto end_of_vacuum;
250 
251   /* Do not attempt to change the page size for a WAL database */
252   if( sqlite3PagerGetJournalMode(sqlite3BtreePager(pMain))
253                                                ==PAGER_JOURNALMODE_WAL
254    && pOut==0
255   ){
256     db->nextPagesize = 0;
257   }
258 
259   if( sqlite3BtreeSetPageSize(pTemp, sqlite3BtreeGetPageSize(pMain), nRes, 0)
260    || (!isMemDb && sqlite3BtreeSetPageSize(pTemp, db->nextPagesize, nRes, 0))
261    || NEVER(db->mallocFailed)
262   ){
263     rc = SQLITE_NOMEM_BKPT;
264     goto end_of_vacuum;
265   }
266 
267 #ifndef SQLITE_OMIT_AUTOVACUUM
268   sqlite3BtreeSetAutoVacuum(pTemp, db->nextAutovac>=0 ? db->nextAutovac :
269                                            sqlite3BtreeGetAutoVacuum(pMain));
270 #endif
271 
272   /* Query the schema of the main database. Create a mirror schema
273   ** in the temporary database.
274   */
275   db->init.iDb = nDb; /* force new CREATE statements into vacuum_db */
276   rc = execSqlF(db, pzErrMsg,
277       "SELECT sql FROM \"%w\".sqlite_schema"
278       " WHERE type='table'AND name<>'sqlite_sequence'"
279       " AND coalesce(rootpage,1)>0",
280       zDbMain
281   );
282   if( rc!=SQLITE_OK ) goto end_of_vacuum;
283   rc = execSqlF(db, pzErrMsg,
284       "SELECT sql FROM \"%w\".sqlite_schema"
285       " WHERE type='index'",
286       zDbMain
287   );
288   if( rc!=SQLITE_OK ) goto end_of_vacuum;
289   db->init.iDb = 0;
290 
291   /* Loop through the tables in the main database. For each, do
292   ** an "INSERT INTO vacuum_db.xxx SELECT * FROM main.xxx;" to copy
293   ** the contents to the temporary database.
294   */
295   rc = execSqlF(db, pzErrMsg,
296       "SELECT'INSERT INTO vacuum_db.'||quote(name)"
297       "||' SELECT*FROM\"%w\".'||quote(name)"
298       "FROM vacuum_db.sqlite_schema "
299       "WHERE type='table'AND coalesce(rootpage,1)>0",
300       zDbMain
301   );
302   assert( (db->mDbFlags & DBFLAG_Vacuum)!=0 );
303   db->mDbFlags &= ~DBFLAG_Vacuum;
304   if( rc!=SQLITE_OK ) goto end_of_vacuum;
305 
306   /* Copy the triggers, views, and virtual tables from the main database
307   ** over to the temporary database.  None of these objects has any
308   ** associated storage, so all we have to do is copy their entries
309   ** from the schema table.
310   */
311   rc = execSqlF(db, pzErrMsg,
312       "INSERT INTO vacuum_db.sqlite_schema"
313       " SELECT*FROM \"%w\".sqlite_schema"
314       " WHERE type IN('view','trigger')"
315       " OR(type='table'AND rootpage=0)",
316       zDbMain
317   );
318   if( rc ) goto end_of_vacuum;
319 
320   /* At this point, there is a write transaction open on both the
321   ** vacuum database and the main database. Assuming no error occurs,
322   ** both transactions are closed by this block - the main database
323   ** transaction by sqlite3BtreeCopyFile() and the other by an explicit
324   ** call to sqlite3BtreeCommit().
325   */
326   {
327     u32 meta;
328     int i;
329 
330     /* This array determines which meta meta values are preserved in the
331     ** vacuum.  Even entries are the meta value number and odd entries
332     ** are an increment to apply to the meta value after the vacuum.
333     ** The increment is used to increase the schema cookie so that other
334     ** connections to the same database will know to reread the schema.
335     */
336     static const unsigned char aCopy[] = {
337        BTREE_SCHEMA_VERSION,     1,  /* Add one to the old schema cookie */
338        BTREE_DEFAULT_CACHE_SIZE, 0,  /* Preserve the default page cache size */
339        BTREE_TEXT_ENCODING,      0,  /* Preserve the text encoding */
340        BTREE_USER_VERSION,       0,  /* Preserve the user version */
341        BTREE_APPLICATION_ID,     0,  /* Preserve the application id */
342     };
343 
344     assert( SQLITE_TXN_WRITE==sqlite3BtreeTxnState(pTemp) );
345     assert( pOut!=0 || SQLITE_TXN_WRITE==sqlite3BtreeTxnState(pMain) );
346 
347     /* Copy Btree meta values */
348     for(i=0; i<ArraySize(aCopy); i+=2){
349       /* GetMeta() and UpdateMeta() cannot fail in this context because
350       ** we already have page 1 loaded into cache and marked dirty. */
351       sqlite3BtreeGetMeta(pMain, aCopy[i], &meta);
352       rc = sqlite3BtreeUpdateMeta(pTemp, aCopy[i], meta+aCopy[i+1]);
353       if( NEVER(rc!=SQLITE_OK) ) goto end_of_vacuum;
354     }
355 
356     if( pOut==0 ){
357       rc = sqlite3BtreeCopyFile(pMain, pTemp);
358     }
359     if( rc!=SQLITE_OK ) goto end_of_vacuum;
360     rc = sqlite3BtreeCommit(pTemp);
361     if( rc!=SQLITE_OK ) goto end_of_vacuum;
362 #ifndef SQLITE_OMIT_AUTOVACUUM
363     if( pOut==0 ){
364       sqlite3BtreeSetAutoVacuum(pMain, sqlite3BtreeGetAutoVacuum(pTemp));
365     }
366 #endif
367   }
368 
369   assert( rc==SQLITE_OK );
370   if( pOut==0 ){
371     nRes = sqlite3BtreeGetRequestedReserve(pTemp);
372     rc = sqlite3BtreeSetPageSize(pMain, sqlite3BtreeGetPageSize(pTemp), nRes,1);
373   }
374 
375 end_of_vacuum:
376   /* Restore the original value of db->flags */
377   db->init.iDb = 0;
378   db->mDbFlags = saved_mDbFlags;
379   db->flags = saved_flags;
380   db->nChange = saved_nChange;
381   db->nTotalChange = saved_nTotalChange;
382   db->mTrace = saved_mTrace;
383   sqlite3BtreeSetPageSize(pMain, -1, 0, 1);
384 
385   /* Currently there is an SQL level transaction open on the vacuum
386   ** database. No locks are held on any other files (since the main file
387   ** was committed at the btree level). So it safe to end the transaction
388   ** by manually setting the autoCommit flag to true and detaching the
389   ** vacuum database. The vacuum_db journal file is deleted when the pager
390   ** is closed by the DETACH.
391   */
392   db->autoCommit = 1;
393 
394   if( pDb ){
395     sqlite3BtreeClose(pDb->pBt);
396     pDb->pBt = 0;
397     pDb->pSchema = 0;
398   }
399 
400   /* This both clears the schemas and reduces the size of the db->aDb[]
401   ** array. */
402   sqlite3ResetAllSchemasOfConnection(db);
403 
404   return rc;
405 }
406 
407 #endif  /* SQLITE_OMIT_VACUUM && SQLITE_OMIT_ATTACH */
408