Issue Details (XML | Word | Printable)

Key: DERBY-1716
Type: Bug Bug
Status: Closed Closed
Resolution: Fixed
Priority: Major Major
Assignee: Yip Ng
Reporter: Yip Ng
Votes: 0
Watchers: 0
Operations

If you were logged in you would be able to see more operations.
Derby

Revoking select privilege from a user times out when that user still have a cursor open.

Created: 17/Aug/06 08:26 PM   Updated: 13/Dec/07 09:05 AM
Return to search
Component/s: SQL
Affects Version/s: 10.2.1.6, 10.2.2.0, 10.3.1.4
Fix Version/s: 10.2.2.0, 10.3.1.4

Time Tracking:
Not Specified

File Attachments:
  Size
Text File Licensed for inclusion in ASF works derby1716-trunk-diff01a.txt 2006-09-23 02:15 AM Yip Ng 11 kB
Text File Licensed for inclusion in ASF works derby1716-trunk-diff02.txt 2006-09-26 08:44 PM Yip Ng 8 kB
Text File Licensed for inclusion in ASF works derby1716-trunk-diff03.txt 2006-10-06 05:45 PM Yip Ng 8 kB
Text File Licensed for inclusion in ASF works derby1716-trunk-stat01a.txt 2006-09-23 02:15 AM Yip Ng 0.5 kB
Text File Licensed for inclusion in ASF works derby1716-trunk-stat02.txt 2006-09-26 08:44 PM Yip Ng 0.4 kB
Text File Licensed for inclusion in ASF works derby1716-trunk-stat03.txt 2006-10-06 05:45 PM Yip Ng 0.4 kB
Environment: Sun JDK 1.4.2
Issue Links:
Reference
 

Issue & fix info: Patch Available
Resolution Date: 10/Oct/06 05:42 AM


 Description  « Hide
Revoking table select privilege from a user will time out if that user still have an open cursor on that table.
Hence, a database owner will not be able to revoke select privilege from any user(s) if they still have a cursor
open. i.e.:

ij version 10.2
ij> connect 'jdbc:derby:cs1;create=true' user 'user1' as user1;
WARNING 01J14: SQL authorization is being used without first enabling authentication.
ij> connect 'jdbc:derby:cs1' user 'user3' as user3;
WARNING 01J14: SQL authorization is being used without first enabling authentication.
ij(USER3)> set connection user1;
ij(USER1)> create table t1001 (c varchar(1));
0 rows inserted/updated/deleted
ij(USER1)> insert into t1001 values 'a', 'b', 'c';
3 rows inserted/updated/deleted
ij(USER1)> grant select on t1001 to user3;
0 rows inserted/updated/deleted
ij(USER1)> set connection user3;
ij(USER3)> autocommit off;
ij(USER3)> GET CURSOR crs1 AS 'select * from user1.t1001';
ij(USER3)> next crs1;
C
----
a
ij(USER3)> set connection user1;
ij(USER1)> -- revoke select privilege while user3 still have an open cursor
revoke select on t1001 from user3;
ERROR 40XL1: A lock could not be obtained within the time requested
ij(USER1)> select * from syscs_diag.lock_table;
XID |TYPE |MODE|TABLENAME |LOCKNAME |STATE|TABLETYPE|LOCK&|INDEXNAME
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
130 |TABLE|IS |SYSTABLEPERMS |Tablelock |GRANT|S |4 |NULL
130 |ROW |S |SYSTABLEPERMS |(1,7) |GRANT|S |2 |NULL
130 |TABLE|IS |T1001 |Tablelock |GRANT|T |1 |NULL

3 rows selected
ij(USER1)> set connection user3;
ij(USER3)> next crs1;
C
----
b
ij(USER3)> next crs1;
C
----
c
ij(USER3)> close crs1;
ij(USER3)>

Is there a reason why Derby still keep shared locks on SYS.SYSTABLEPERMS during fetch?


sysinfo:

------------------ Java Information ------------------
Java Version: 1.4.2_12
Java Vendor: Sun Microsystems Inc.
Java home: C:\Program Files\Java\j2re1.4.2_12
Java classpath: derby.jar;derbytools.jar
OS name: Windows XP
OS architecture: x86
OS version: 5.1
Java user name: Yip
Java user home: C:\Documents and Settings\Yip
Java user dir: C:\work3\derby\tests\derby-10.2.1.0\lib
java.specification.name: Java Platform API Specification
java.specification.version: 1.4
--------- Derby Information --------
JRE - JDBC: J2SE 1.4.2 - JDBC 3.0
[C:\work3\derby\tests\derby-10.2.1.0\lib\derby.jar] 10.2.1.0 beta - (430903)
[C:\work3\derby\tests\derby-10.2.1.0\lib\derbytools.jar] 10.2.1.0 beta - (430903)
------------------------------------------------------
----------------- Locale Information -----------------
Current Locale : [English/United States [en_US]]
Found support for locale: [de_DE]
         version: 10.2.1.0 - (430903)
Found support for locale: [es]
         version: 10.2.1.0 - (430903)
Found support for locale: [fr]
         version: 10.2.1.0 - (430903)
Found support for locale: [it]
         version: 10.2.1.0 - (430903)
Found support for locale: [ja_JP]
         version: 10.2.1.0 - (430903)
Found support for locale: [ko_KR]
         version: 10.2.1.0 - (430903)
Found support for locale: [pt_BR]
         version: 10.2.1.0 - (430903)
Found support for locale: [zh_CN]
         version: 10.2.1.0 - (430903)
Found support for locale: [zh_TW]
         version: 10.2.1.0 - (430903)
------------------------------------------------------


 All   Comments   Work Log   Change History   Subversion Commits      Sort Order: Ascending order - Click to sort in descending order
Yip Ng added a comment - 20/Sep/06 05:19 PM
The shared lock on SYS.SYSTABLEPERMS is actually obtained by the activation itself during
execution time (at the start of fillResultSet() of the activation class) since the grantee's privileges
are not in the permission cache yet. Thus, the shared locks will last till end of the transaction.
This is done once and those privileges that were loaded from the data dictionary(DD) will be stored
in the permission cache for use later so that the system does not need to load those info again
from the DD. Perhaps these privilege info can be loaded at compilation time instead as an improvement,
so that at execution time, privilege related metadata can be retrieved from the permission cache instead
of from the DD.


protected ResultSet fillResultSet() throws StandardException
{
    ((Activation) this).getLanguageConnectionContext().getAuthorizer().authorize( (Activation) this, 1 );
    return getResultSetFactory().getScrollInsensitiveResultSet( <snip> );
}

Yip Ng added a comment - 20/Sep/06 06:28 PM
Note that since the statement may be cached by the system and reuse by other users of different
connections, the suggested scheme of loading privilege(s) at compilation time may not eliminate the
DD access in all cases if the system were to load the privilege info based on the auth id that compiles
the statement.

Bryan Pendleton added a comment - 20/Sep/06 11:15 PM
Yip, are you saying that we use the *user's* transaction to read table and column
permissions information from the data dictionary? I would have expected us
to use some sort of internal "system" transaction for this kind of catalog fetch.

Yip Ng added a comment - 21/Sep/06 12:06 AM
Bryan, at compilation time, Derby uses an internal system nested transaction. After compilation is completed, this nested system
transaction is committed; therefore, the shared locks for system tables are released. However, at execution time of this jira case,
the activation class is querying the data dictionary for permission since those descriptors are not in the permission cache, therefore,
the shared locks are obtained for those system tables using the user transaction.

Yip Ng added a comment - 21/Sep/06 01:29 AM
I debugged into the code abit today, I don't think the system is fetching the permission descriptor(s)
from the system catalog at compilation time for the stated problem of this jira. During bind phase,
the system just collects various permission(s) that are required to execute the select statement. It is
at the execution time of the statement that the descriptors are fetched from the data dictionary.

Yip Ng added a comment - 23/Sep/06 02:15 AM
Attaching initial patch derby1716-trunk-diff01a.txt for DERBY-1716. The patch fixes the stated problem of this jira where it now loads the auth id's permission descriptors into the permission cache at compilation time instead, so that at execution time, the system does not have to go through the system table(s) again and hold the shared locks till the end of user transaction. derbyall passes. Appreciate if someone can review the patch. Thanks.

Bryan Pendleton added a comment - 23/Sep/06 03:14 PM
Hi Yip, thanks for the patch!

I read through the code and it looks good. Thanks for the clear comments in the
addToPermissionsCache method and for the good tests.

I noticed as part of studying DERBY-1847 that DataDictionaryImpl.clearCaches()
does not clear the DataDictionary permissions cache. This seems wrong to me,
and although it probably isn't *directly* related to this JIRA issue, I still wanted to
bring it up since you've been studying this code, and get your opinion.

Regarding your patch, I noticed you called it an "initial patch". Do you feel that this
patch is ready for commit? Or do you think that there is additional work to do?

Are other reviewers looking at this patch?

Bryan Pendleton added a comment - 23/Sep/06 03:22 PM
Addendum to previous comment: the patch applies cleanly for me, and I
verified that, without the code change, the test case demonstrates the
bug, and with the code change, the test case passes. I am +1 to commit,
and I propose to commit this change if Yip feels that it is ready for commit.

Yip Ng added a comment - 24/Sep/06 06:00 PM
Hi Bryan, thanks for taking the time for reviewing the patch. For the clearCaches() part, it looks like the
permission cache handling are somewhat different from other caches. For instance, when a table gets
dropped, its permission descriptor gets removed from the cache in its constant action(DROP TABLE) via
DataDictionary's dropAllTableAndColPermDescriptors(). So they do get removed from the permission
cache.

I am trying another different approach so please do not commit the current patch just yet. I will post my findings
soon. Thanks.

Mike Matrigali added a comment - 24/Sep/06 10:07 PM
unchecking patch available, contributer has asked for this patch to not be applied yet.

Yip Ng added a comment - 26/Sep/06 08:44 PM
Attaching patch derby1716-trunk-diff02.txt for DERBY-1716.

After further analysis, I think the previous patch does not address all cases.
In particular, unlike other descriptors, when privilege(s) get revoked from user,
the statement is not subject to recompilation, so then we are back to square one
since the previous patch attempts to bring in the permission descriptor(s) into
the permission cache at compilation time to avoid reading from system tables at
execution time.

I believe the proper proposal fix is to use internal nested read-only transaction
when the system is reading permission descriptors from the system tables. At a
high level, a statement undergoes the following typical steps for it to get executed
by the system:

1. Statement Compilation Phase
 a) Parse the statement
 b) Bind the statement and collects required permissions for it to be executed.
 c) Optimize the statement
 d) Generate the activation for the statement
 

2. Statement Execution Phase
 a) Check if the authoration id has the required privileges to execute the statement.
 b) Execute the statement


The problem lies in permissions checking step at statement execution phase. Before a statement can be executed in SQL authorization mode, the authorization id's privileges needs to be check against the permission cache or if the privileges are not available in the cache, the system needs to read this metadata information from the system tables. But the system is using *user transaction* to do this, so the shared locks that got acquired by the user transaction may not get released immediately; therefore, leading to lock timeout when the grantor attempts to revoke the user's privilege. To resolve this issue, the system now will start an internal read-only nested transaction(same lock space as the parent transaction) to read permission related info from the system tables and release the shared locks
as soon as the permissions check is completed before statement execution. This tackles the root of the stated problem. derbyall passes. Appreciate if someone can review the code changes. Thanks.

Other interesting observations while going through the code:

(1) In the current implementation, privileges collection actually ends at the time where the constant action is created and not at bind time. e.g.: DROP TABLE has schema permission collected at that time.

(2) Permissions cache handling are different from other data dictionary caches. The other caches gets *ALL* cleared out either at startReading() or startWriting() time in Data Dictionary depending on the current cache mode. However, only the "affected" cached items in the permissions cache are removed and this logic is handled by the statements themselves and not the data dictionary.


Mike Matrigali added a comment - 06/Oct/06 05:07 PM
could you regenerate this patch, I am getting failures in applying the test portions of the patch -
grantRevokeDDL.sql has been a hot file.

Yip Ng added a comment - 06/Oct/06 05:45 PM
Previous patch got stale, attaching derby1716-trunk-diff03.txt.

Mike Matrigali added a comment - 07/Oct/06 03:34 PM
Committed this patch to the trunk.

m3_ibm142:20>svn commit

Sending java\engine\org\apache\derby\impl\sql\conn\GenericAuthorizer.java

Sending java\testing\org\apache\derbyTesting\functionTests\master\grantRevokeDDL.out
Sending java\testing\org\apache\derbyTesting\functionTests\tests\lang\grantRevokeDDL.sql
Sending java\testing\org\apache\derbyTesting\functionTests\tests\lang\grantRevokeDDL_app.properties
Transmitting file data ....
Committed revision 453935.

Rick Hillegas added a comment - 28/Nov/06 06:20 PM
Ported to 10.2 branch at subversion revision 480148.

Myrna van Lunteren added a comment - 12/Dec/06 10:46 PM
updated the 10.2 j9_foundation canon with revision 486380.

Andrew McIntyre added a comment - 13/Dec/07 09:05 AM
This issue has been resolved for over a year with no further movement. Closing.