Critical security vulnerability related with Netty 4.2.16.Final

Hello,

As part of a security review, we identified that recent Openfire releases are using Netty version 4.2.16.Final, as indicated in the project’s pom.xml.

We have identified a publicly disclosed security vulnerability affecting the Netty version used by Openfire:

CVE-2025-58056

Could you please clarify whether the vulnerable Netty component/code path is actually used by Openfire in a way that makes Openfire susceptible to this vulnerability?

If Openfire is affected, could you also advise whether there is a planned Openfire release that will update the Netty dependency to a patched version?

According to the Netty project, the relevant fix is included in Netty 4.2.17.Final.

More generally, would it be possible to update Openfire’s Netty dependency to the latest security-fixed version in an upcoming release, in order to address this and any other security vulnerabilities affecting the current Netty version?

Thanks for raising this. We appreciate people keeping an eye on Openfire’s dependencies.

As far as we can tell, Openfire is not affected by CVE-2025-58056: According to the Netty advisory (GHSA-fghv-69vj-qj49), the issue was fixed in 4.2.5.Final. The 4.2.16.Final version that Openfire currently uses should already include that fix.

Perhaps a different CVE was intended, since Netty 4.2.17.Final did address several other issues? It’s also common for scanning tools to match affected versions incorrectly. If you can tell us which finding you’re working from, we’re happy to take another look.

As for your more general question: yes, we try to keep our dependencies current, and Netty updates are part of our regular release work.

A wider note, offered in a constructive spirit: Openfire is run by a small group of volunteers. A report like this takes real time to triage. We have to find the advisory, check the affected version ranges, and work out whether the vulnerable code is even reachable in Openfire. That time comes from the same limited pool we use for fixing bugs and building features. As this report shows, sometimes that work ends with “not affected.”

This is becoming increasingly pressing. With the rise of AI-assisted tooling, we’re receiving more and more security-related reports, and the triage workload is rising with them. We’re glad to keep improving the project, and we take security seriously. But most of this work is unpaid.

Organizations that build on Openfire have the most to gain from it staying healthy, and they’re also best placed to help. We’d love to see more of them take that step. Here are some ways to do so:

  • An initial assessment. Include links to the advisory, the affected version ranges, and, if you can, whether Openfire uses the affected module or code path. This turns a research task into a quick review.
  • Pull requests. A dependency bump is often a one-line change, and we’re happy to review it.
  • Testing. Running pre-release builds in your environment and reporting back is very valuable to us.
  • Private reporting. For anything that might really be exploitable, please use our security reporting process rather than the public forum.
  • Commercial support. If your organization needs formal answers, assessments, or timelines for its own compliance processes, paid support or sponsorship is a good fit for that, and it directly supports the people doing this work.

One small request as well: this forum is public, so we’d ask that “critical” be reserved for issues confirmed to affect Openfire. Other users read these titles, and a misplaced one can cause unnecessary alarm.

Thanks again, and we hope to see you around!

Apologies for the mix-up—I realized I made a typo in the CVE ID in my previous post.

The correct vulnerability I am referring to is CVE-2026-75595

Hi @guus

Thank you so much for the detailed and constructive response.

First, I want to sincerely apologize for the confusion - you are absolutely right that my previous post referenced the wrong CVE. I made a mistake while copying the ID.

The vulnerability I am concerned about is actually CVE-2026-75595, which was patched very recently in Netty 4.2.17.Final.

To save you some research time, here is the initial assessment based on your helpful guidelines:

Advisory Link: NVD - Home

Affected Version Range: Netty 4.2.0.Final through 4.2.16.Final (meaning Openfire’s current 4.2.16.Final is affected).

The Issue: A bounds-checking offset error in SslClientHelloHandler#decode during SNI parsing can cause an IndexOutOfBoundsException, forcing a fallback to the default SSL context. This potentially allows a mutual TLS (mTLS) bypass if the default context does not require client authentication.

I completely understand and respect the workload constraints on volunteer-run projects. I really appreciate the work you do to keep Openfire safe and healthy.

If a PR to bump Netty to 4.2.17.Final would be helpful, I would be glad to submit one to help save you the development time.

Thanks again for your patience and for guiding me on how to make better reports.

I have created this ticket for the upgrade: https://igniterealtime.atlassian.net/browse/OF-3404

The change is being PR’ed here: OF-3404: Bump Netty from 4.2.16.Final to 4.2.19.Final by guusdk · Pull Request #3516 · igniterealtime/Openfire · GitHub