[ 
https://issues.apache.org/jira/browse/COLLECTIONS-701?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17938768#comment-17938768
 ] 

Volodymyr Siedlecki commented on COLLECTIONS-701:
-------------------------------------------------

 

If you add the set to itself once, no stack overflow occurs, but, if you add it 
twice, it does.    Not a very realistic scenario in practice, but still 
possible for an attacker to abuse it potentially.  Example code: 
{code:java}
jshell> import org.apache.commons.collections4.list.*;
jshell>  final List<Object> source = new ArrayList<>();
   ...>  final List<Object> list = SetUniqueList.setUniqueList(source);
   ...>  list.add(list);
source ==> []
list ==> []
$6 ==> true
jshell> list.add(list);
|  Exception java.lang.StackOverflowError
|        at AbstractListDecorator.hashCode (AbstractListDecorator.java:75)
|        at ArrayList.hashCodeRange (ArrayList.java:595)
|        at ArrayList.hashCode (ArrayList.java:582)
|        at AbstractListDecorator.hashCode (AbstractListDecorator.java:75)
|        at ArrayList.hashCodeRange (ArrayList.java:595)
{code}
 

> StackOverflowError in SetUniqueList.add() when it receives itself
> -----------------------------------------------------------------
>
>                 Key: COLLECTIONS-701
>                 URL: https://issues.apache.org/jira/browse/COLLECTIONS-701
>             Project: Commons Collections
>          Issue Type: Bug
>          Components: Collection
>    Affects Versions: 3.2.2
>            Reporter: Shin Hong
>            Priority: Critical
>             Fix For: 4.3
>
>
> Hi.
> We found that the following test case fails with a StackOverFlowError 
> exception:
> {code:java}
> test() {        
>    SetUniqueList l = new SetUniqueList(new LinkedList<Object>()) ;        
>    l.add((Object) l) ;    
> }{code}
> The add() execution traps into an infinite recursion which crashes the 
> program.
> From the stack trace, we found that the infinite recursion occurs
> at AbstractList.hashCode() since it invokes hashCode() of each of its 
> elements.
>  



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

Reply via email to