PostgreSQL: 2013-04-04 Security Release FAQ

The following versions make up this release:

• v9.2.4
• v9.1.9
• v9.0.13
• v8.4.17

While this FAQ covers the 2013-04-04 PostgreSQL Security Update in general, most of its contents focus on the primary security vulnerability patched in the release, CVE-2013-1899.

Are there any known exploits "in the wild" for this vulnerability?

There are no known exploits at the time of release.

Who is particularly vulnerable because of this issue?

Any system that allows unrestricted access to the PostgreSQL network port, such as users running PostgreSQL on a public cloud, is especially vulnerable. Users whose servers are only accessible on protected internal networks, or who have effective firewalling or other network access restrictions, are less vulnerable.

This is a good general rule for database security: do not allow port access to the database server from untrusted networks unless it is absolutely necessary. This is as true, or more true, of other database systems as it is of PostgreSQL.

What is the nature of the vulnerability?

The vulnerability allows users to use a command-line switch for a PostgreSQL connection intended for single-user recovery mode while PostgreSQL is running in normal, multiuser mode. This can be used to harm the server.

What potential exploits are enabled by this vulnerability?

Persistent Denial of Service: an unauthenticated attacker may use this vulnerability to cause PostgreSQL error messages to be appended to targeted files in the PostgreSQL data directory on the server. Files corrupted in this way may cause the database server to crash, and to refuse to restart. The database server can be fixed either by editing the files and removing the garbage text, or restoring from backup. Configuration Setting Privilege Escalation: in the event that an attacker has a legitimate login on the database server, and the server is configured such that this user name and the database name are identical (e.g. user web, database web), then this vulnerability may be used to temporarily set one configuration variable with the privileges of the superuser. Arbitrary Code Execution: if the attacker meets all of the qualifications under 2 above, and has the ability to save files to the filesystem as well (even to the tmp directory), then they can use the vulnerability to load and execute arbitrary C code. SELinux will prevent this specific type of exploit.

Which major versions of PostgreSQL are affected?

Versions 9.0, 9.1 and 9.2.

Users of version 8.4 are not affected. Users of version 8.3 and earlier are not affected by this issue, but are vulnerable to other unpatched security vulnerabilities, since those versions are EOL.

How can users protect themselves?

• Download the update release and update all of your servers as soon as possible.
• Ensure that PostgreSQL is not open to connections from untrusted networks.
• Audit your database users to be certain that all logins require proper credentials, and that the only logins which exist are legitimate and in current use.

Use of advanced security frameworks, such as SELinux with PostgreSQL's SEPostgres extension, also lessen or eliminate the exposure and potential damage from PostgreSQL security vulnerabilities.

Who was given access to the information about the vulnerability?

Specifics about the vulnerability were first disclosed to our security team.

The PostgreSQL Global Development Group (PGDG) has had, for several years, a policy granting engineers who build PostgreSQL binary packages to be distributed to the public (such as RPMs and Windows installers) early access to be able to release information and code so that packages can be ready on the official release date. This applied to both minor and major releases. Given the increasing prevalence of PostgreSQL-as-a-Service (PGaaS) as a distribution mechanism, we are revising this policy to accomodate the case of the cloud providers. The new policy is still being edited and should be available soon.

When was the vulnerability discovered?

This vulnerability was first reported to the PostgreSQL Global Development Group (PGDG) security team on March 12, 2013.

We filed for the CVE, with the assistance of the Red Hat security team, on March 27.

Who discovered the vulnerability?

Mitsumasa Kondo and Kyotaro Horiguchi of NTT Open Source Software Center while conducting a security audit. NTT is a longtime contributor to PostgreSQL.

How was the vulnerability reported?

Kondo-san and Horiguchi-san sent email to security@postgresql.org.

As reported by TechCrunch and Hacker News, some entities including cloud platform provider Heroku were given early access. Why did this occur?

Heroku was given access to updated source code which patched the vulnerability at the same time as other packagers. Because Heroku was especially vulnerable, the PostgreSQL Core Team worked with them -- to secure their infrastructure and to use their deployment as a test-bed for the security patches. This helped to verify that the security update did not break any application functionality. Heroku has a history both of working closely with community developers, and of testing experimental features in their PostgreSQL service.

Who was given access to the code before the official release?

We have two teams that communicate on private lists hosted on the PGDG infrastructure. Both teams had access to the source code prior to the release of any packages for analyzing the security patch and then creating packages for distributing PostgreSQL binaries. These are our Security Team and our Packagers List. In both cases, these groups had early access in order to participate in patching the security hole.

How can end-users with large deployments or security-sensitive applications obtain early access security information?

At this time, the PostgreSQL project does not provide users who are not directly involved in patching security vulnerabilities or packaging PostgreSQL for other users early access to security information, patches, or code. It is possible that at some time in the future we may be in a position to offer such access, but we are not able to now.

Was taking the repository private while this security discussion was ongoing the proper thing to do?

Given the severity of the vulnerability, the PostgreSQL Core team deliberated and determined the security risk posed by having the source code for the fix available before the packages were made available outweighed the public’s interest in having immediate access.

Normal procedure for sharing information about security releases is to send an announcement our developer mailing list, pgsql-hackers@postgresql.org, a week before a new release. Tom Lane did this. Then, due to the severity of the security vulnerability, we also sent an announcement to pgsql-announce@postgresql.org and to our RSS News feed on our website homepage. We did this because we wanted to give DBAs sufficient time to plan for a maintenance window to upgrade.

The timing of the announcements and the release was based on the availability of volunteer packagers and release managers to conduct the release.

How is the PostgreSQL project organized?

PostgreSQL Global Development Group (PGDG) is a volunteer-run, global organization. We have a six-person core team, a number of Major Contributors and several mailing lists that make up the centralized portion of our community. See here for details about contributors.

How are new members added to the Security Team or Packagers?

Membership in both groups is maintained by the Core Team.

How often does PostgreSQL find new security vulnerabilities?

We find zero to seven minor security issues a year. This is the first security issue of this magnitude since 2006: the "backslash escape encoding issue", which affected MySQL and a few other database systems as well.

How was the vulnerability introduced?

It was created as a side effect of a refactoring effort to make establishing new connections to a PostgreSQL server faster, and the associated code more maintainable.

Who discovers vulnerabilities in PostgreSQL?

We are fortunate to have a large pool of security engineers who test PostgreSQL regularly and responsibly report security issues so that they can be fixed. This includes:

• QA staff at contributing companies like NTT Open Source, EnterpriseDB and 2ndQuadrant
• Security researchers at the Japanese federal security agency
• Security researchers at security companies, such as Secunia
• Coverity’s Scan Project
• and our large pool of participating community users, who report bugs.

What else is included in this release?

This release also updates four other, minor, security issues which are detailed on the security page and in the release announcement. It includes a number of bug fixes for PostgreSQL as well, most notably fixes for two potential data corruption issues with binary replication.