[JAMES-2278] IMAP QRESYNC (RFC-5162) is buggy

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.0.0, master, 3.0.1
• Fix Version/s: 3.7.0
• Component/s: IMAPServer
• Labels: bugfix
• Assignee: Unassigned
• Reporter: Benoit Tellier
• Created: 04/Jan/18 07:20
• Updated: 08/Sep/21 03:10
• Resolved: 08/Sep/21 03:10

Description

As reported by lmilev on the gitter chat, passing known sequence sets to QRESYNC leads to an error in the IMAP layer:

INFO | jvm 1 | 2018/01/03 12:25:15 | [imapserver-executor-17] ERROR org.apache.james.imap.processor.base.AbstractChainedProcessor - Error while processing IMAP request
INFO | jvm 1 | 2018/01/03 12:25:15 | java.lang.IndexOutOfBoundsException: Index: 2, Size: 2
INFO | jvm 1 | 2018/01/03 12:25:15 | at java.util.ArrayList.rangeCheck(ArrayList.java:653)
INFO | jvm 1 | 2018/01/03 12:25:15 | at java.util.ArrayList.get(ArrayList.java:429)
INFO | jvm 1 | 2018/01/03 12:25:15 | at org.apache.james.imap.processor.AbstractSelectionProcessor.respond(AbstractSelectionProcessor.java:240)
INFO | jvm 1 | 2018/01/03 12:25:15 | at org.apache.james.imapserver.netty.ImapChannelUpstreamHandler.messageReceived(ImapChannelUpstreamHandler.java:194)

I tried to write a MPT test but it is not trivial as it depends on random values (UIDVALIDITY) and non fixed values (MODSEQ). Moreover the code is messy, dirty and not explicit.

The incriminated lines seems to be:

if (knownUidsList.size() > index++) {
     int msnAsInt = msn.intValue();
     MessageUid knownUid = knownUidsList.get(index);
     // Complicated sequential stuff
}

As this little test shows it:

    @Test
    public void test() {
        int i = 0;
        System.out.println(i++);
        System.out.println(i++);
    }

// Will output
// 0
// 1

The index can clearly be out of range.

if (knownUidsList.size() > ++index) {
     int msnAsInt = msn.intValue();
     MessageUid knownUid = knownUidsList.get(index);
     // Complicated sequential stuff
}

Should fix the issue.

We should find a way to provide at least unit tests for that issue.

Issue Links

• links to: GitHub Pull Request #635