The Apache Logging Services PMC has received the security report
referenced below. After analysis, we classified it as INVALID (works as
designed). In the interest of transparency, and so that the community
and other researchers can benefit from the analysis, we are disclosing a
summary of it.
Report reference: public GitHub issue, opened 24 October 2026
https://github.com/apache/logging-log4j2/issues/4426
Disposition: INVALID -- works as designed, out-of-scope adversary; no CVE
TLP:CLEAR (this message may be redistributed without restriction)
The report was filed publicly, together with a public exploit lab,
before any triage. The reporter was informed of this classification on
the issue itself.
== Summary of the report ==
At initialization, `log4j-api` reads every `log4j2.system.properties`
and `log4j2.component.properties` resource found on the classpath. The
former is copied into the Java system properties, the latter supplies
Log4j's own configuration properties, and neither leaves a trace in the
status logger. The reporter showed that a jar placed on the classpath of
an application without a Log4j configuration of its own can carry such a
file setting `log4j.configurationFile` and
`log4j2.Script.enableLanguages`, so that a configuration from the same
jar runs a script during initialization. The precondition, in the
reporter's words, is "a writable classpath directory or control of a
dependency artifact". The reporter proposed logging every properties
file loaded, a warning whenever system properties are written, and
documentation stating that the classpath is a trust boundary.
== PMC assessment ==
This is NOT a vulnerability. The report requires an adversary that our
threat model lists as out of scope, twice:
https://logging.apache.org/security.html#threat-common-adversary
First, both files are configuration properties, and an adversary able to
modify configuration properties or configuration files is out of scope.
Second, and more fundamentally, an adversary able to place a jar on the
classpath can already execute arbitrary code in the application, which
is also out of scope ("including code introduced through plugins, custom
appenders, or other application extensions"). Such an adversary does not
need a properties file or a scripting engine: the same jar can carry a
`log4j2-test.xml`, which takes precedence over the application's own
`log4j2.xml`, a Log4j plugin, a `META-INF/services` registration of an
`org.apache.logging.log4j.spi.Provider`, which `log4j-api` instantiates
on the first call to `LogManager`, or simply a class with a static
initializer. The properties files add no capability the adversary does
not already have.
Reading `log4j2.system.properties` from the classpath and copying it
into the Java system properties is documented, and is the purpose of the
file:
https://logging.apache.org/log4j/2.x/manual/systemproperties.html
The proposed warning would fire on every legitimate use of a documented
feature, and would not help against the adversary described, who also
controls the logging configuration, including the status logger.
Verifying the integrity of the dependencies that make up an application
is the responsibility of whoever assembles it; the part that concerns
Log4j artifacts is described in the "Supply chain attacks" section of
the threat model.
A `DEBUG` status message naming each properties file Log4j loads, and
where it was found, would nevertheless be a reasonable troubleshooting
aid. Contributions are welcome as a regular enhancement.
Finally, we remind researchers that public issues are not a supported
channel for security reports, and that publishing an exploit before
triage puts users at risk if the issue turns out to be real. Please
follow the instructions at
https://logging.apache.org/security.html#reporting and e-mail
[email protected].
== References ==
Threat model:
https://logging.apache.org/security.html
Public report:
https://github.com/apache/logging-log4j2/issues/4426
System properties documentation:
https://logging.apache.org/log4j/2.x/manual/systemproperties.html
Related report by the same reporter:
https://github.com/apache/logging-log4j2/issues/4425
Questions and follow-up are welcome on this list or GitHub Discussions.
On behalf of the Apache Logging Services PMC,
Piotr P. Karwasz
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]