[ 
https://issues.apache.org/jira/browse/WW-5352?focusedWorklogId=900606&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-900606
 ]

ASF GitHub Bot logged work on WW-5352:
--------------------------------------

                Author: ASF GitHub Bot
            Created on: 19/Jan/24 07:32
            Start Date: 19/Jan/24 07:32
    Worklog Time Spent: 10m 
      Work Description: kusalk commented on code in PR #832:
URL: https://github.com/apache/struts/pull/832#discussion_r1458480297


##########
core/src/main/java/com/opensymphony/xwork2/config/impl/DefaultConfiguration.java:
##########
@@ -395,6 +396,7 @@ public static ContainerBuilder 
bootstrapFactories(ContainerBuilder builder) {
                 .factory(SecurityMemberAccess.class, Scope.PROTOTYPE)
                 .factory(OgnlGuard.class, StrutsOgnlGuard.class, 
Scope.SINGLETON)
                 .factory(ProviderAllowlist.class, Scope.SINGLETON)
+                .factory(ThreadAllowlist.class, Scope.SINGLETON)

Review Comment:
   This bean was introduced as a way to mutate the allowlist read by 
`SecurityMemberAccess` - if it were `prototype` scope, `ParametersInterceptor` 
and `SecurityMemberAccess` would not share the same instance of 
`ThreadAllowlist` and thus this wouldn't be possible.



##########
apps/showcase/src/main/java/org/apache/struts2/showcase/action/ParamsAnnotationAction.java:
##########
@@ -0,0 +1,133 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *  http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+package org.apache.struts2.showcase.action;
+
+import com.opensymphony.xwork2.ActionSupport;
+import org.apache.struts2.interceptor.parameter.StrutsParameter;
+import org.apache.struts2.showcase.model.MyDto;
+
+import java.lang.reflect.Field;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import static java.util.Collections.singletonList;
+import static java.util.Collections.singletonMap;
+
+/**
+ * This class supports {@link 
com.atlassian.confluence.stateless.webdriver.selenium3.security.StrutsParametersTest}
+ * which prevents critical security regressions. Do NOT modify without 
understanding the motivation behind the tests and
+ * the implications of any changes.
+ */
+public class ParamsAnnotationAction extends ActionSupport {
+
+    @StrutsParameter

Review Comment:
   So this PR doesn't add any new injection capabilities - OGNL has always been 
able to set public fields without setters. It has also never required a getter 
to set a value, only a setter.
   
   A getter is only used when you want to access a nested setter (or public 
field). For example, the parameter name `kusal.lukasz` translates to 
`getKusal()`, then on the returned object, `setLukasz(_)`. 
   
   What this PR does is use annotations to clearly mark the code path OGNL will 
take to perform parameter injection. The `ParametersInterceptor` is predicting 
the field or method OGNL will invoke, and if it is not annotated, it will be 
stripped out of the acceptable parameters map before OGNL attempts injection. 
In this way, when inspecting the source of an Action class, it is perfectly 
clear which methods and fields are exposed to the internet. We are essentially 
making OGNL more intuitive for developers - because right now, attackers seem 
to understand OGNL better than the developers using it! 😂



##########
core/src/main/java/org/apache/struts2/ognl/ThreadAllowlist.java:
##########
@@ -0,0 +1,51 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *  http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+package org.apache.struts2.ognl;
+
+import java.util.HashSet;
+import java.util.Set;
+
+import static java.util.Collections.emptySet;
+import static java.util.Collections.unmodifiableSet;
+
+/**
+ * Allows any bean to allowlist a class for use in OGNL expressions, for the 
current thread only. The allowlist can be
+ * cleared once any desired OGNL expressions have been evaluated.
+ *
+ * @since 6.4.0
+ */
+public class ThreadAllowlist {
+
+    private final ThreadLocal<Set<Class<?>>> allowlist = new ThreadLocal<>();

Review Comment:
   Yeah this is a good question - the reason I went with this design is because 
by isolating the allowlists on a per-request basis, it should make the 
application even more secure. Request A will not be able to access a class that 
might only be required by request B, and vice-versa. Note that we clear the 
allowlist at the end of every request. The less options attackers have to craft 
a payload, the better. Atlassian have seen some very creative payloads where 
our internal classes are chained together in unconventional ways.
   
   There is also no performance benefit by making this a global cache as we 
don't currently cache the annotation lookups either. I don't expect the 
overhead of checking annotations per-request to be significant, but this is 
something we can reevaluate in the future.





Issue Time Tracking
-------------------

    Worklog Id:     (was: 900606)
    Time Spent: 8h 10m  (was: 8h)

> Implement annotation mechanism for injectable fields via parameters
> -------------------------------------------------------------------
>
>                 Key: WW-5352
>                 URL: https://issues.apache.org/jira/browse/WW-5352
>             Project: Struts 2
>          Issue Type: Improvement
>          Components: Core, Core Interceptors
>            Reporter: Kusal Kithul-Godage
>            Priority: Minor
>             Fix For: 6.4.0
>
>          Time Spent: 8h 10m
>  Remaining Estimate: 0h
>
> struts.parameters.requireAnnotations
>  
> Require an explicit annotation '@StrutsParameter' on one of: 
> Getter/Setter/Field/ReturnType for injecting parameters.
>  
> This mechanism is intended to be a more usable replacement for 
> 'ParameterNameAware'



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to