CVE-2026-109549: OmniFaces: SSE push channels are not bound to their owning HTTP session
Medium CVSS 5.9
- OSV id
- GHSA-w3w6-9m25-g88p
- Also known as
- CVE-2026-109549
- Published
- Last modified
Affected packages
| Package | Affected versions | Fixed in |
|---|---|---|
| org.omnifaces:omnifaces Maven |
>= 5.2.0, < 5.4.3 | 5.4.3 |
How to fix it
Upgrade org.omnifaces:omnifaces to 5.4.3 or later.
# in pom.xml: set the <version> of org.omnifaces:omnifaces (or the property that holds it) to 5.4.3
mvn versions:use-dep-version -Dincludes=org.omnifaces:omnifaces -DdepVersion=5.4.3 -DforceVersion=true
# if it is an indirect dependency, pin it in <dependencyManagement>, then: mvn dependency:tree
- If it is an indirect dependency, pin it in <dependencyManagement> (Maven) or with a dependency constraint (Gradle), or upgrade the library that brings it in; `mvn dependency:tree` / `./gradlew dependencies` show which one.
- Check the changelog for breaking changes (especially across major versions), update the lockfile with the package manager, and run your tests.
Not sure whether your project uses an affected version? Check your lockfile below for free.
Details
Summary
GHSA-fp43-vj7g-pg92 (fixed 5.4.2) bound session- and view-scoped WebSocket (o:socket) push channels to the HTTP session that registered them, so a leaked channel ID can no longer be connected to from a foreign or absent HTTP session. The fix (d5cae243c4) touched only the Socket* classes.
The parallel SSE (o:sse, added in 5.2) push transport was left unchanged. SseEndpoint.doGet accepts a connection whenever the presented channel ID is in the application-wide registry, without ever checking that the caller's HTTP session owns it. A leaked session/view-scoped SSE channel ID can therefore be replayed from any (or no) HTTP session to receive the messages pushed to it — exactly the behaviour the advisory removed for WebSocket, still present on SSE at the latest release.
Root Cause
Both push transports share the same base classes. A session- or view-scoped channel ID is minted as channel + "?" + UUID.randomUUID() and registered only in an application-scoped, session-agnostic registry: PushChannelManager#L123 mints it and #L136 calls getPushSessions().register(channelId); the registry's membership test is PushSessionManager#L137-L139:
protected boolean isRegistered(String channelId) {
return sessions.containsKey(channelId); // keyed by channelId only — no HTTP-session dimension
}
WebSocket (fixed). SocketChannelManager.register additionally records session/view channel IDs on the HTTP session, and the handshake refuses anything not owned by the caller's session, at SocketEndpoint.java#L133-L142:
if (
!SocketChannelManager.isApplicationScopedChannelId(channelId)
&& !SocketChannelManager.isChannelIdRegisteredInSession((HttpSession) request.getHttpSession(), channelId)
) {
throw new IllegalStateException(ERROR_UNAUTHORIZED_CHANNEL);
}
SSE (unfixed). SseChannelManager extends the base register() unchanged (no HTTP-session registration), and the endpoint's sole accept gate is the global isRegistered check, at SseEndpoint.java#L59-L66:
var channel = getChannel(request);
var channelId = channel + "?" + request.getQueryString();
var sseSessions = Beans.getReference(SseSessionManager.class);
if (channel == null || !sseSessions.isRegistered(channelId)) {
response.getOutputStream().write(SSE_NOT_FOUND_EVENT);
return;
}
// ...accepts: startAsync() and streams events for this channel
doGet never calls request.getSession(...); no Sse* class consults the HTTP session for the accept decision. So for a session/view-scoped channel ID the fix intends to bind, WebSocket refuses a foreign/absent session while SSE accepts it. (Application-scoped channels are unbound by design on both transports — not at issue here.)
Related, same root cause: the idle-timeout and max-sessions-per-channel bounds the fix added to SocketEndpoint / SocketSessionManager were also not ported to SSE — SseEndpoint.doGet sets asyncContext.setTimeout(0) and SseSessionManager inherits the unbounded getMaxSessionsPerChannel().
Steps to Reproduce
The PoC drives the real SseEndpoint.doGet from omnifaces 5.4.2 and shows it accepts a leaked session-scoped channel ID from a request carrying no owning HTTP session (and never consults the session), while the WebSocket fix refuses the same class of ID.
Prerequisites
- JDK 21 or newer (tested on Temurin/Zulu 21 and OpenJDK 25)
- Apache Maven 3.9+
Step 1: Create the project
```bash mkdir -p /tmp/omnifaces-sse-poc/src/test/java/org/omnifaces/cdi/push cd /tmp/omnifaces-sse-poc cat > pom.xml <<'EOF' <?xml version="1.0" encoding="UTF-8"?> <project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd"> <modelVersion>4.0.0</modelVersion> <groupId>poc</groupId> <artifactId>omnifaces-sse-poc</artifactId> <version>1.0</version> <packaging>jar</packaging>
References
- GHSA-w3w6-9m25-g88p on OSV.dev
- Web: https://github.com/omnifaces/omnifaces/security/advisories/GHSA-w3w6-9m25-g88p
- Advisory: https://nvd.nist.gov/vuln/detail/CVE-2026-109549
- Web: https://github.com/omnifaces/omnifaces/commit/78fe2b6989387f93718bdb932fb4497e338dd638
- Package: https://github.com/omnifaces/omnifaces
Am I affected?
Upload or paste your lockfile. We check every package version in it against OSV.dev, this advisory included, and email you the vulnerable ones with the version that fixes them. Free, and no account needed.
About this data
Source: GitHub Advisory Database via OSV.dev (CC-BY 4.0). Severity comes from the advisory's own rating when it has one, else from its CVSS score.