Well your test code has no leak with the free added -- so you need to post better test code to demonstrate your leak. Otherwise the suspicion would be with YOUR code. Michael D. Black Senior Scientist Northrop Grumman Mission Systems
________________________________ From: [email protected] on behalf of liubin liu Sent: Fri 4/23/2010 9:19 PM To: [email protected] Subject: Re: [sqlite] Is there any memory leak in the code while being busy? Thanks :) But in my real code, the sqlite3_free(sql) is in the right place. So I think that there are other reasons causing the leak. Black, Michael (IS) wrote: > > You need to sqlite3_free(sql) after you use the sql from your > sqlite3_mprintf(). > > sql = sqlite3_mprintf (sql_f, i); > ret = sqlite3_prepare_v2 (db1, sql, -1, &p_stmt, NULL); > sqlite3_free(sql); > > > Michael D. Black > Senior Scientist > Northrop Grumman Mission Systems > > > ________________________________ > > From: [email protected] on behalf of liubin liu > Sent: Fri 4/23/2010 7:16 AM > To: [email protected] > Subject: Re: [sqlite] Is there any memory leak in the code while being > busy? > > > > > sorry for multi-send the message. > > I just test the code again. And sqlite3_finalize() may free the memory. > I'm > wrong in the first post. > > > But I test the routine of sqlite3_prepare_v2() + sqlite3_step() + > sqlite3_finalize() in my real code. > > And the test result say when sqlite3_step() is shadowed, the leak is zero. > When doing the sqlite3_step(), the leak is about 1k byte. And another > curious phenomenon is that while there are many datas in the data file, > the > leak is more bigger than while there are few datas in the data file. > > > > > > Marcus Grimm wrote: >> >> it is not necessary to send your question multible times... ;) >> >> to answer: what makes you think that sqlite3_finalize can't >> free the prepared statement ? >> >> liubin liu wrote: >>> Is there any memory leak in the code? >>> >>> Below is the code. Is there any memory leak in the pthread2? >>> >>> While pthread1 is using test.db exclusively, the sqlite3_prepare_v2() of >>> pthread2 still prepares the p_stmt pointer to a piece of memory malloced >>> by >>> sqlite3_preapare_v2(). And then the sqlite3_finalize() can't free the >>> memory >>> still because pthread1 is using test.db exclusively. Does it cause a >>> memory >>> leak? >>> >>> >>> >>> ______code__________________________________________________________ >>> >>> #include <stdio.h> >>> #include <unistd.h> // for usleep() >>> #include <sys/time.h> // for gettimeofday() >>> #include <pthread.h> >>> #include <sqlite3.h> >>> >>> void pthread1 (void); >>> void pthread2 (void); >>> >>> >>> int main (void) >>> { >>> int ret = -1; >>> >>> sqlite3 *g_db = NULL; >>> ret = sqlite3_open ("test.db", &g_db); >>> ret = sqlite3_exec (g_db, "CREATE TABLE t1 (id INTEGER PRIMARY KEY, >>> d1 >>> TEXT)", NULL,NULL,NULL); >>> ret = sqlite3_close (g_db); >>> >>> usleep (500000); >>> >>> >>> pthread_t pthr1, pthr2; >>> ret = pthread_create (&pthr1, NULL, (void *) pthread1, NULL); >>> ret = pthread_create (&pthr2, NULL, (void *) pthread2, NULL); >>> >>> >>> ret = pthread_join (pthr1, NULL); >>> printf ("thread1 end\n"); >>> ret = pthread_join (pthr2, NULL); >>> printf ("thread2 end\n"); >>> >>> return 0; >>> } >>> >>> >>> void pthread1 (void) >>> { >>> int ret = -1; >>> sqlite3 *db1 = NULL; >>> ret = sqlite3_open ("test.db", &db1); >>> >>> char *sql_f = "INSERT OR REPLACE INTO t1 VALUES (%d, >>> 'aaabbbcccddd1122')"; >>> char *sql = NULL; >>> >>> sqlite3_stmt *p_stmt = NULL; >>> >>> struct timeval tv1, tv2; >>> ret = gettimeofday (&tv1, NULL); >>> >>> ret = sqlite3_prepare_v2 (db1, "BEGIN TRANSACTION", -1, &p_stmt, >>> NULL); >>> ret = sqlite3_step (p_stmt); >>> ret = sqlite3_finalize (p_stmt); >>> >>> // n=1000000, test.db is 25843712 bytes, 25M; >>> int i=0, n=1000000; >>> for (i=0; i<n; i++) >>> { >>> sql = sqlite3_mprintf (sql_f, i); >>> ret = sqlite3_prepare_v2 (db1, sql, -1, &p_stmt, NULL); >>> ret = sqlite3_step (p_stmt); >>> ret = sqlite3_finalize (p_stmt); >>> } >>> >>> printf ("pthread1: ret: %d\n", ret); >>> ret = sqlite3_prepare_v2 (db1, "COMMIT TRANSACTION", -1, &p_stmt, >>> NULL); >>> ret = sqlite3_step (p_stmt); >>> ret = sqlite3_finalize (p_stmt); >>> >>> ret = gettimeofday (&tv2, NULL); >>> printf ("time is : %ds\n", (int) (tv2.tv_sec - tv1.tv_sec)); >>> >>> >>> ret = sqlite3_close (db1); >>> } >>> >>> >>> >>> void pthread2 (void) >>> { >>> int ret = -1; >>> >>> sqlite3 *db2 = NULL; >>> ret = sqlite3_open ("test.db", &db2); >>> >>> usleep (2000000); >>> >>> >>> sqlite3_stmt *p_stmt = NULL; >>> int i=0, n=10; >>> for (i=0; i<n; i++) >>> { >>> char *sql_f = "INSERT OR REPLACE INTO t1 VALUES (%d, >>> 'bbbbbbbbbbbb1122')"; >>> char *sql = NULL; >>> >>> sql = sqlite3_mprintf (sql_f, i); >>> ret = sqlite3_prepare_v2 (db2, sql, -1, &p_stmt, NULL); >>> printf ("pthread2: prepare: %d, p_stmt: %p, errmsg: %s\n", >>> ret, >>> p_stmt, sqlite3_errmsg(db2)); >>> ret = sqlite3_step (p_stmt); >>> printf ("pthread2: step: %d, p_stmt: %p, errmsg: %s\n", >>> ret, >>> p_stmt, sqlite3_errmsg(db2)); >>> ret = sqlite3_finalize (p_stmt); >>> printf ("pthread2: finalize: %d, p_stmt: %p, errmsg: %s\n", >>> ret, >>> p_stmt, sqlite3_errmsg(db2)); >>> printf ("\n"); >>> >>> usleep (300000); >>> } >>> >>> ret = sqlite3_close (db2); >>> printf ("pthread2: close: %d, p_stmt: %p, errmsg: %s\n", ret, >>> p_stmt, sqlite3_errmsg(db2)); >>> } >> _______________________________________________ >> sqlite-users mailing list >> [email protected] >> http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users >> >> > > -- > View this message in context: > http://old.nabble.com/Is-there-any-memory-leak-in-the-code-while-being-busy--tp28337646p28340567.html > Sent from the SQLite mailing list archive at Nabble.com. > > _______________________________________________ > sqlite-users mailing list > [email protected] > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > > > > _______________________________________________ > sqlite-users mailing list > [email protected] > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > > -- View this message in context: http://old.nabble.com/Is-there-any-memory-leak-in-the-code-while-being-busy--tp28337646p28347901.html Sent from the SQLite mailing list archive at Nabble.com. _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
_______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
