New MCP server — Bring vulnerability intelligence into your AI
Back to feed

CVE-2026-44328

ExploitedNVDCIRCLENISA

EUVD-2026-32570 · Published May 8, 2026

free5GC's SMF UPI DELETE /upi/v1/upNodesLinks/{ref} panics on AN-node deletion via nil UPF dereference; unauthenticated, state-mutating

Summary

free5GC's SMF mounts the UPI management route group without inbound OAuth2 middleware (same root cause as the broader UPI auth gap reported in free5gc/free5gc#887). On top of that, the DELETE /upi/v1/upNodesLinks/{upNodeRef} handler unconditionally dereferences upNode.UPF after the type-guarded async release, even though AN-typed nodes are constructed without a UPF object. As a result, a single unauthenticated DELETE /upi/v1/upNodesLinks/gNB1 request crashes the handler with a nil-pointer panic AND mutates the in-memory user-plane topology before panicking (the UpNodeDelete(upNodeRef) line runs first). This is an unauthenticated, state-mutating panic-DoS sink that an off-path network attacker can trigger by name against any AN entry.

Details

Validated against the SMF container in the official Docker compose lab.

  • Source repo tag: v4.2.1
  • Running Docker image: free5gc/smf:v4.2.1
  • Runtime SMF commit: 8385c00a
  • Docker validation date: 2026-03-22 local (container log timestamp 2026-03-21T23:43:17Z)
  • SMF endpoint: http://10.100.200.6:8000

Control comparison on the same SMF instance:

  • GET /nsmf-oam/v1/ (no token) -> 401 Unauthorized
  • DELETE /upi/v1/upNodesLinks/gNB1 (no token) -> 500 Internal Server Error (panic)

The sibling nsmf-oam returning 401 proves OAuth middleware IS wired in for other SMF route groups; the UPI group specifically is mounted without it.

Vulnerable handler logic (paths in free5gc/smf):

// NFs/smf/internal/sbi/api_upi.go:94..99
if upNode.Type == smf_context.UPNODE_UPF {
    go s.Processor().ReleaseAllResourcesOfUPF(upNode.UPF)
}
upi.UpNodeDelete(upNodeRef)
upNode.UPF.CancelAssociation()   // <-- panics for AN-typed nodes; nil UPF

The Type == UPNODE_UPF guard only protects the asynchronous ReleaseAllResourcesOfUPF call. After that, UpNodeDelete(upNodeRef) runs unconditionally (so the topology mutation lands first), and then upNode.UPF.CancelAssociation() is called unconditionally on a *UPF that is nil for AN nodes by construction.

Code evidence:

  • UPI group mounted WITHOUT auth middleware:
    • NFs/smf/internal/sbi/server.go:76
    • NFs/smf/internal/sbi/server.go:78
  • Protected control comparison (other SMF groups DO use auth):
    • NFs/smf/internal/sbi/server.go:99
    • NFs/smf/internal/sbi/server.go:105
  • Delete handler (panic site):
    • NFs/smf/internal/sbi/api_upi.go:94
    • NFs/smf/internal/sbi/api_upi.go:99
  • AN nodes are constructed without a UPF object (root cause of the nil deref):
    • NFs/smf/internal/context/user_plane_information.go:95
    • NFs/smf/internal/context/user_plane_information.go:97
PoC

Reproduced end-to-end against the running SMF at http://10.100.200.6:8000.

  1. Control: protected sibling OAM route returns 401:
curl -i http://10.100.200.6:8000/nsmf-oam/v1/
HTTP/1.1 401 Unauthorized
  1. Trigger: unauthenticated DELETE on the default AN node gNB1:
curl -i -X DELETE http://10.100.200.6:8000/upi/v1/upNodesLinks/gNB1
HTTP/1.1 500 Internal Server Error
  1. SMF container logs (docker logs --tail 120 smf) show topology mutation landing BEFORE the panic, and the panic stack pointing at api_upi.go:99:
[INFO][SMF][Init] UPNode [gNB1] found. Deleting it.
[INFO][SMF][Init] Delete UPLink [UPF] <=> [gNB1].
[ERRO][SMF][GIN] panic: runtime error: invalid memory address or nil pointer dereference
github.com/free5gc/smf/internal/sbi.(*Server).DeleteUpNodeLink
    /go/src/free5gc/NFs/smf/internal/sbi/api_upi.go:99 +0x298
[INFO][SMF][GIN] | 500 | DELETE | /upi/v1/upNodesLinks/gNB1

The lab state was manually restored after validation by re-creating the AN entry; that POST is restoration-only and is NOT a mitigation.

Impact

Three compounding defects on the same SMF SBI surface:

  1. Missing inbound authentication (CWE-306) and authorization (CWE-862) on the UPI route group, so the trigger is reachable to any off-path network attacker who can reach SMF on the SBI -- no token, no session, no UE state needed. The same-instance nsmf-oam returning 401 proves the middleware is wired in elsewhere and only missing on UPI.
  2. NULL pointer dereference (CWE-476) in DeleteUpNodeLink: the Type == UPNODE_UPF guard only covers the async release call, then upNode.UPF.CancelAssociation() runs unconditionally on AN-typed nodes that have a nil UPF field by construction.
  3. Order of operations (CWE-755 / CWE-754): UpNodeDelete(upNodeRef) mutates the in-memory user-plane topology BEFORE the dereference panics, so the topology change lands even though the request returns 500. This makes the bug state-mutating, not just a plain panic.

Any party that can reach SMF on the SBI can:

  • Delete arbitrary named entries (e.g. gNB1) from SMF's in-memory user-plane topology anonymously via a single DELETE /upi/v1/upNodesLinks/{ref} request, denying SMF's ability to consider that AN/UPF in subsequent UPF selection / PFCP path establishment for legitimate UE sessions.
  • Trigger a panic on the SMF goroutine for the deleted-AN case, even though Gin recovers the goroutine, leaving the topology in the mutated state above.
  • Repeat the trigger by name against any AN entry, sustaining the topology denial without ever authenticating.

This is a strict superset of the impact in free5gc/free5gc#887 for this specific code path: same auth bypass, plus a concrete request-triggerable nil deref, plus state mutation that survives the panic.

Affected: free5gc v4.2.1.

Upstream issue: https://github.com/free5gc/free5gc/issues/905
Upstream fix: https://github.com/free5gc/smf/pull/199

Technologies

Free5Gc Free5GcFree5Gc Smf

Weaknesses

CWE-862CWE-476CWE-306

CVSS scores

  • v3.18.2CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:H

MITRE ATT&CK

Inferred

Where this vulnerability fits in an attack

Initial Access Persistence Stealth Defense Impairment Credential Access

References

Potentially impacted assets

See if this affects your attack surface

Latest trending attack

Criticalyesterday

Google Brotli decoder - Remote Code Execution via compressed requests (Exploitation Check)

This control is the second step of the Patrowl response to CVE-2026-111497 (Node2Shell). It complements the first control of 22 October 2026, which focused on the Node.js runtime, and does not replace it. Why a second control CVE-2026-111497 is an out-of-bounds write in the Google Brotli decoder, the library that decompresses Brotli ("br") data. It was first published as a Node.js vulnerability, because Node.js bundles this decoder. Later publications showed that the flaw is in the library itself. Every product that embeds an affected version of the decoder is concerned, whatever its language or vendor. A public proof of concept demonstrates code execution. Any server that decompresses request bodies sent with "Content-Encoding: br" exposes the decoder to unauthenticated attackers. Known cases include Express 5 and NestJS 11 on Node.js, Koa and Fastify setups, Bun, Envoy and Istio with request decompression enabled, and any application or reverse proxy that decodes Brotli request bodies with an affected library. The flaw also works in the other direction. A malicious server can compromise an unpatched client that decodes a Brotli response, for example Node.js fetch(), Python clients (requests, httpx, aiohttp, Scrapy) with the Brotli or brotlicffi package, or libcurl builds with Brotli support. What this control checks Patrowl sends small, valid Brotli-compressed requests to each asset and observes whether the server decodes them. The check does not depend on the technology: it detects the behaviour, whatever the stack behind it. These requests are what any legitimate client could send. They never reach the vulnerable code path and cannot affect the availability of the service. How to read the result Potentially impacted means the asset decodes Brotli request bodies: the attack surface is exposed. It does not confirm that the asset is vulnerable. The version of the Brotli decoder in use is not visible from the outside. This check covers applications that parse request bodies on their root path. A Not impacted result does not rule out exposure on other routes, nor on the client side. Checking the versions on the servers remains the only exhaustive verification. Affected and fixed versions Google Brotli 1.0.3 to 1.2.0 are affected, fixed in 1.0.10, 1.1.1 and 1.2.1. Fixed releases of the main products embedding it: Node.js 22.23.4, 24.22.1 and 26.12.1 (end-of-life lines 20.x and older and 25.x receive no fix), Bun 1.4.4, Envoy 1.37.8, 1.38.6, 1.39.3 and 1.40.0, Istio 1.29.9, 1.30.6 and 1.31.2, Python Brotli 1.2.1 and brotlicffi 1.2.1.0. Products that link against the system library depend on the Brotli package of the Linux distribution. Recommendations Update every component that embeds the Brotli decoder (only real fix). Start with the assets flagged by this control. Restart the processes after the update. Rebuild containers, single-executable applications and appliances that bundle an affected runtime or library. Reduce the exposure until the update is deployed. Reject requests with "Content-Encoding: br" at the reverse proxy, load balancer or WAF, case-insensitively and including lists such as "gzip, br". On Envoy and Istio, disable request decompression for Brotli. Review outbound calls. List the features that fetch content from servers you do not control, and patch the services that host them first.

Take 15 minutes to discover our platform with our experts

Patrowl Intel platform screenshot