Mailing-List: contact derby-dev-help@db.apache.org; run by ezmlm
Precedence: bulk
Reply-To: <derby-dev@db.apache.org>
Message-ID: <1591634360.1211406355831.JavaMail.jira@brutus>
Date: Wed, 21 May 2008 14:45:55 -0700 (PDT)
From: "Dag H. Wanvik (JIRA)" <jira@apache.org>
To: derby-dev@db.apache.org
Subject: [jira] Issue Comment Edited: (DERBY-3681) When authenticating a
 user at connect time, verify that the user provided is not also a defined
 role name.
In-Reply-To: <711980761.1211309156309.JavaMail.jira@brutus>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/DERBY-3681?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12598799#action_12598799 ] 

dagw edited comment on DERBY-3681 at 5/21/08 2:44 PM:
---------------------------------------------------------------

Uploading patch derby-3681-1 which implements the check. Running
regression tests now.

Patch details:

M      java/engine/org/apache/derby/impl/jdbc/EmbedConnection.java

An extra test is added to checkUserCredentials when applicable
(dictionary >= 10.4 and sqlAuthorization is in use). A user name which
equals a role name will lead to connection level exception state
NET_CONNECT_AUTH_FAILED, i.e. 08004 and the connect will fail. I chose
not to add a new error message here, in the spirit of not providing
too much information when authentication fails. Admittedly, this could
cause some user confusion ("this user is defined, why can't I
connect?"), so a more helpful message is arguably better.

M      java/testing/org/apache/derbyTesting/functionTests/tests/lang/RolesTest.java

Added a test case for the new check.

M      java/testing/org/apache/derbyTesting/junit/DriverManagerConnector.java

For the client, openConnection(String, String, String) can't reliably
determine if a 08004 means the database does not exist or if it is a
authentication problem, so before this fix, my new test failed because
the fallback here tried to create a new database in such cases. This
should also fail when using the provided user, but it did not, since the
fallback used a call to getConnectionByAttributes(url, "create",
"true"). This uses the default user, not the one provided. 
So i modified this code to try to create using the
provided user which seems more logical anyway..

M      java/testing/org/apache/derbyTesting/junit/CleanDatabaseTestSetup.java

Added code to remove any roles when cleaning a database. This was
necessary since the RolesTest left a role in the database called
"bar", which the subsequent SQLGrantREvokeDDLTest tried to use as a
user name. The new check introduced with this issue made that fail.


      was (Author: dagw):
    Uploading patch derby-3681-1 which implements the check. Running
regression tests now.

Patch details:

M      java/engine/org/apache/derby/impl/jdbc/EmbedConnection.java

An extra test is added to checkUserCredentials when applicable
(dictionary >= 10.4 and sqlAuthorization is in use). A user name which
equals a role name will lead to connection level exception state
NET_CONNECT_AUTH_FAILED, i.e. 08004 and the connect will fail. I chose
not to add a new error message here, in the spirit of not providing
too much information when authentication fails. Admittedly, this could
cause some user confusion ("this user is defined, why can't I
connect?"), so a more helpful message is arguably better.

M      java/testing/org/apache/derbyTesting/functionTests/tests/lang/RolesTest.java

Added a test case for the new check.

M      java/testing/org/apache/derbyTesting/junit/DriverManagerConnector.java

For the client, openConnection(String, String, String) can't reliably
determine if a 08004 means the database does not exist or if it is a
authentication problem, so before this fix, my new test failed because
the fallback here tried to create a new database in such cases. Thus
should also when using the provided user, but it did not, since the
fallback used a call to getConnectionByAttributes(url, "create",
"true"); which uses the default user, not the one provided, which I
want to fail. So i modified this code to try to create using the
provided user which seems more logical anyway..

M      java/testing/org/apache/derbyTesting/junit/CleanDatabaseTestSetup.java

Added code to remove any roles when cleaning a database. This was
necessary since the RolesTest left a role in the database called
"bar", which the subsequent SQLGrantREvokeDDLTest tried to use as a
user name. The new check introduced with this issue made that fail.

  
> When authenticating a user at connect time, verify that the user provided is not also a defined role name.
> ----------------------------------------------------------------------------------------------------------
>
>                 Key: DERBY-3681
>                 URL: https://issues.apache.org/jira/browse/DERBY-3681
>             Project: Derby
>          Issue Type: Sub-task
>          Components: Security
>            Reporter: Dag H. Wanvik
>            Assignee: Dag H. Wanvik
>             Fix For: 10.5.0.0
>
>         Attachments: derby-3681-1.diff, derby-3681-1.stat
>
>
> Although we try to avoid creating role that are not also valid Derby users (see DERBY-3673), we cannot
> in general know for sure that no such user exists; it could be added to derby.properties after
> the role has been created, authentication could be LDAP or user-defined, in which cases
> the check at role creation time will not work. So, in order to avoid collisions between user identifiers and role identifiers, we shoudl check at connect time that there is no role by same name as the supplied user name.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.