Since I love to debug code and I needed a distraction....
Problem was introduced on 2009-06-05 update
Several SQLITE_CORRUPT_BKPT were inserted (so this corruption may have been
going for a while).
And here's where the corruption is returned:
iPage = get4byte(&aData[8+closest*4]);
printf("iPage=%p aData=%p\n",iPage,aData);
if( iPage>mxPage ){
printf("X7 iPage=%p, closest=%d ",iPage,closest);
rc = SQLITE_CORRUPT_BKPT;
goto end_allocate_page;
}
./sqlite3 test.db <log.txt
delete
X7 iPage=0x1200000, closest=0 SQL error near line 297: database disk image is
malformed
X7 iPage=0x1200000, closest=0 SQL error near line 301: database disk image is
malformed
X7 iPage=0x1200000, closest=0 SQL error near line 303: database disk image is
malformed
X7 iPage=0x2ba01e1, closest=0 SQL error near line 305: database disk image is
malformed
With the added printout of iPage it suddenly changes values....
iPage=0x2136 aData=0x5a04458
iPage=0x1c9c aData=0x5989ac8
iPage=0x1f6e aData=0x5c1c5e8
iPage=0x2136 aData=0x5a04458
iPage=0x1200000 aData=0x5a80508
X7 iPage=0x1200000, closest=0 SQL error near line 297: database disk image is
malformed
The aData where it bombs here is the first time this address has shown up.
Michael D. Black
Senior Scientist
Advanced Analytics Directorate
Northrop Grumman Information Systems
________________________________
From: sqlite-users-boun...@sqlite.org on behalf of Richard Hipp
Sent: Mon 8/23/2010 8:57 AM
To: General Discussion of SQLite Database; sqlite-annou...@sqlite.org
Subject: EXTERNAL:[sqlite] Possible database corruption bug in SQLite
A user has provided us with a script that appears to result in SQLite
database corruption. The problem has existed in all versions of SQLite
going back to 3.6.16 in June of 2009. A bisect shows that the problem was
injected on 2009-06-17.
The problem appears to be associated with incremental vacuum. We are still
working to characterize the problem more precisely.
The discovery of this bug so close to the release of version 3.7.1 is a
coincidence. Nevertheless, we will likely be issuing version 3.7.2 within a
few days in order to fix the problem. So if you are thinking of upgrading
to 3.7.1, you might want to hold back for a day or two to see what our
investigation of this new problem turns up.
--
D. Richard Hipp
d...@sqlite.org
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
