Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 6 Jun 2018 13:33:59 +0200
From: Tomas Hoger <>
To: Denis Magda <>
Cc:, dev <>, "Rai,
 Harendra" <>
Subject: Re: [CVE-2014-0114]: Apache Ignite is vulnerable to
 existing CVE-2014-0114

Hi Denis!

On Fri, 1 Jun 2018 10:16:50 -0700 Denis Magda wrote:

> [CVE-2014-0114]: Apache Ignite is vulnerable to existing CVE-2014-0114
> Severity: Important
> Vendor: The Apache Software Foundation
> Versions Affected: Apache Ignite 2.4 or earlier
> Impact:
> An attacker can execute arbitrary code on Ignite nodes in the case
> when Ignite classpath contains arbitrary vulnerable classes.
> Description:
> Apache Ignite used commons-beanutils-1.8.3.jar library which did not
> suppress the class property, which allowed remote attackers to
> "manipulate" the ClassLoader and execute arbitrary code via the class
> parameter, as demonstrated by the passing of this parameter to the
> getClass method of the ActionForm object in Struts 1.

This announcement is very light on details.  Would it be possible to
provide more details, ideally a link to the fix that was applied to
address this issue?

Searching for more information, I found out that the upstream Jira
ticket for this issue should be:

The ticket is non-public, but its content is leaked via a mailing list:

This has some important info, indicating that the problem (only?)
affects Ignite for .NET.  The reported problem basically seems to be:
Ignite for .NET bundles commons-beanutils 1.8.3 and that should be
upgraded to 1.9.2.  Looking into apache.ignite.2.4.0.nupkg and
apache.ignite.2.5.0.nupkg, I can see that commons-beanutils upgrade as
requested did happen in 2.5.0.

Note that I do not see any commons-beanutils jar in and  Are those, as well as source
distribution, considered unaffected?

Now back to the CVE - I do not believe that your re-use of the old
CVE-2014-0114 is correct.  In the report, there was some ambiguity
whether Struts or Commons-BeanUtils should be blamed for the flaw,
however it seems to be explicit enough that the CVE-2014-0114 is for

As noted in the mail, the problem wasn't fixed in Commons-BeanUtils,
which only added mechanisms to make it easy for applications using
Commons-BeanUtils to easily disable processing of the "class"
property.  It did not even disable processing by default, as noted in
the release notes:

Release 1.9.2 mainly addresses a potential security issue when accessing
properties in an uncontrolled way. In a nutshell, if an application that uses
Commons BeanUtils passes property paths from an external source directly to
the getProperty() method of BeanUtilsBean, an attacker can access the class
loader via the class property available on all Java objects.

In version 1.9.2 now a special BeanIntrospector class was added which allows
suppressing this property. Note that this BeanIntrospector is NOT enabled by
default! Commons BeanUtils is a low-level library, and on this layer it cannot
be decided whether access to a certain property is legal or not. Therefore,
an application has to activate this suppressing BeanIntrospector explicitly.
This can be done with the following lines of code:

BeanUtilsBean bub = new BeanUtilsBean();

Now all access to properties has to be done via the specially configured
BeanUtilsBean instance. More information about this issue can be found at or in section 2.5
of the user's guide.

Note that there was a request to assign a separate CVE for the
BeanUtils part that was rejected (actually, CVE-2014-3540 was assigned
and later rejected), see this post from Mitre for details:

It has few parts that are relevant to Ignite:

In particular, the 1597344 change has this documentation:

   Adding this instance as BeanIntrospector to an instance of
   PropertyUtilsBean suppresses the class property; it can then no
   longer be accessed.

This is an additional step that would need to be followed for any
currently shipped product that relies on commons-beanutils. Simply
picking up version 1.9.2 does not solve the problem. The product's
source code must additionally be modified by (for example) changing
or adding an addBeanIntrospector method call.

Did Ignite get any other changes related to this issue apart from
upgrading Commons-BeanUtils?  If not, Commons-BeanUtils upgrade should
not be expected to solve the problem (if Ignite actually was affected /
used Commons-BeanUtils in a vulnerable way, which isn't demonstrated in
the IGNITE-8472).

Another relevant part is:

If any other product makes a security announcement that they have
or equivalent code as a change to the default behavior, then there can
be an individual CVE ID for that product. However, if any other product
simply makes a security announcement that they have decided to ship
commons-beanutils 1.9.2 -- but the class property remains exposed in
the product as it is shipped and installed by default -- then a CVE ID
would not be assigned.

If Ignite got/gets a fix that leverages the SUPPRESS_CLASS from
Commons-BeanUtils 1.9.2 to disable processing of the class property, it
should get its own CVE assigned.

Tomas Hoger / Red Hat Product Security

Powered by blists - more mailing lists

Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.