CVE-2026-59379
Patch Available.
This Vulnerability has been fixed in the Never-Ending Support (NES) version offered by HeroDevs.
Overview
Spring Web Services (Spring-WS) is the Spring project for building contract-first SOAP web services, providing server-side endpoint mappings and endpoints as well as the client-side WebServiceTemplate, with support for WS-Addressing and transports such as HTTP, JMS, email and XMPP.
A server-side request forgery (SSRF) vulnerability (CVE-2026-59379) has been identified in the WS-Addressing support of Spring Web Services, which allows attackers to make the server send out-of-band reply messages to internal-only hosts, cloud metadata endpoints, or other destinations that are not otherwise reachable from the attacker. The delivery is one-way: Spring Web Services does not read back or expose the response of the targeted destination.
Per OWASP: In a Server-Side Request Forgery (SSRF) attack, the attacker can abuse functionality on the server to read or update internal resources.
This issue affects multiple versions of Spring Web Services.
Details
Module Info
- Product: Spring Web Services
- Affected packages:
org.springframework.ws:spring-ws-core - Affected versions: >=5.0.0 <=5.0.3, >=4.1.0 <=4.1.4, >=4.0.0 <=4.0.19, <=3.1.9
- GitHub repository: https://github.com/spring-projects/spring-ws
- Published packages: https://central.sonatype.com/artifact/org.springframework.ws/spring-ws-core
- Package manager: Maven
- Fixed in:
- OSS Spring Web Services 5.0.4
- NES for Spring Web Services
Vulnerability Info
This Medium-severity vulnerability is found in the org.springframework.ws:spring-ws-core package in multiple versions of Spring Web Services. When a request carries a non-anonymous wsa:ReplyTo or wsa:FaultTo address, AbstractAddressingEndpointMapping hands the reply to one of its configured WebServiceMessageSender instances, which opens a connection to that peer-supplied address. Before connecting, AbstractWebServiceMessageSender.supports(URI, UriSource) consults an optional, application-supplied DestinationPolicy, and when none is configured it falls back to the sender's default checks for UriSource.REMOTE destinations:
@Override
public boolean supports(URI uri, UriSource uriSource) {
if (!this.supportedUriFilter.test(uri)) {
return false;
}
D details = createDescriptor(uri, uriSource);
Predicate<D> defaultChecks = defaultChecks(uriSource);
return (this.destinationPolicy != null) ? this.destinationPolicy.accept(details, defaultChecks)
: defaultChecks.test(details);
}For HTTP, those default checks in AbstractHttpWebServiceMessageSender screen the destination by its form rather than by where it actually leads:
protected boolean remoteDefaultChecks(HttpDestinationDescriptor descriptor) {
String hostName = descriptor.hostName();
if (hostName == null) {
return false;
}
if ("localhost".equals(hostName) || hostName.endsWith(".localhost")) {
return false;
}
if (!this.allowSiteLocalIpv4 && descriptor.isSiteLocalIpv4Literal()) {
return false;
}
InetAddress hostAddress = descriptor.hostAddress(this.allowDnsResolution);
if (hostAddress != null) {
if (isLocalAddress(hostAddress)) {
return false;
}
if (!this.allowSiteLocalIpv4 && hostAddress instanceof Inet4Address && hostAddress.isSiteLocalAddress()) {
return false;
}
}
return true;
}Only localhost names and loopback, link-local, any-local and RFC 1918 IPv4 literals are rejected, and host names are not resolved by default, so a reply address whose host name resolves to an internal address passes the checks. The JMS, mail and XMPP senders apply similar format-based screening (JNDI-style destination names, local mail hosts and the XMPP service domain). Screening a peer-supplied destination for reachability in this way cannot be made correct, so a remote attacker who controls the wsa:ReplyTo or wsa:FaultTo header can choose a destination that passes the default checks and still reaches an internal-only host or a cloud metadata endpoint. Nothing requires a sender to be configured with a DestinationPolicy before AbstractAddressingEndpointMapping uses it for out-of-band replies.
Deployments are affected when all of the following are true: an AbstractAddressingEndpointMapping subclass is registered with one or more WebServiceMessageSender instances for out-of-band replies that rely on the default checks rather than an application-supplied DestinationPolicy; the service accepts WS-Addressing headers from untrusted callers; and no restrictive destination validation or network-level egress control compensates for the insufficient default checks.
Mitigation
Only recent versions of Spring Web Services receive community support. Older lines are End-of-Life and will not receive public updates to address this issue.
Users of the affected components should apply one of the following mitigations:
- Upgrade to a currently supported version of Spring Web Services.
- Leverage a commercial support partner like HeroDevs for post-EOL security support.
Credits
- FEARIS (finder)