llamaCVEVulnerability Intelligence
/
Advisory · GitHub · PyPI · vllm

GHSA-935W-9G4M-P28P

vLLM: Harmony tool continuations drop `cache_salt` — restoring a cross-tenant prefix-cache membership oracle

Parche disponible
Descargar PNG JSON
CVSS
—
low
Fuente: —
ExplotaciónSin evidenciaNo figura en CISA KEV
Exploits públicos0Ninguno conocido
Parche1 versiónPrimer parche a -14 d de la divulgación
Exposición44 dDesde la divulgación, 5 oct 2026, 7:02 p. m.
Detección42 dDesde la divulgación hasta su ingreso al índice
Timeline

Ciclo de vida de la vulnerabilidad

Hoy
Divulgación
Publicada
Modificada
Explotación
Remediación
Parche 0.30.0
Monitoreo
Detectada
22 sept 26
6 oct 26
15 oct 26
17 nov 26
19 nov 26
22 sept 26
6 oct 26
21 oct 26
5 nov 26
19 nov 26
FechaFaseEventoDetalle
Remediación Parche 0.30.0 vllm · -14 d desde divulgación
Divulgación Publicada GHSA-935W-9G4M-P28P divulgada
Divulgación Modificada Última actualización del registro
Monitoreo Detectada Ingresó al índice
Resumen
Descripción del advisory
Texto original · GitHub · en inglés

Affected

  • Ecosystem / package: pip / `vllm`
  • Affected versions: vLLM ≤ 0.25.1 (confirmed on 0.25.1, commit `752a3a504485` https://github.com/vllm-project/vllm/tree/752a3a504485790a2e8491cacbb35c137339ad34). The lower bound predates 0.25.1; maintainers can confirm how far back the tool-continuation re-submission has omitted the salt.

Summary

On the GPT-OSS "Harmony" path (`POST /v1/responses`), a request that uses a built-in or MCP tool runs as a multi-turn loop: after each tool call vLLM re-renders the full next-turn Harmony prompt and re-submits it to the engine. Turn 1 correctly carries `request.cache_salt`, but the tool-continuation re-submission rebuilds the engine input via `tokens_input(token_ids)` with no `cache_salt`. The continuation prefix is therefore cached in the global *unsalted* namespace even though the caller opted into salting. A second tenant who can guess the low-entropy post-tool history submits the reconstructed continuation (unsalted) and reads exact per-turn cached-token counts from the Responses usage — restoring the prompt-membership oracle that `cache_salt` is documented to prevent.

Silently dropping a preserved salt *after* the supported tool workflow is enabled is a broken isolation control: the caller enabled salting and every turn should stay isolated, but continuation turns leak into the shared cache.

This is distinct from GHSA-4qjh-9fv9-r85r https://github.com/vllm-project/vllm/security/advisories/GHSA-4qjh-9fv9-r85r (CVE-2025-46570 https://nvd.nist.gov/vuln/detail/CVE-2025-46570): that advisory is the prefix-cache membership oracle for which `cache_salt` is the documented mitigation, and its PR-17045 fix does not close this site — the Harmony tool continuation silently drops the preserved salt, caching in the unsalted namespace and leaking exact `cached_tokens_per_turn` counts from a different sink (the Responses serving continuation, not general TTFT timing).

Affected code

Links pinned to the confirmed commit `752a3a504485` https://github.com/vllm-project/vllm/tree/752a3a504485790a2e8491cacbb35c137339ad34 (v0.25.1):

  • The drop (sink): `vllm/entrypoints/openai/responses/serving.py#L712-L713` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/entrypoints/openai/responses/serving.py#L712-L713 — `token_ids = context.render_for_completion()` then `engine_input = tokens_input(token_ids)`, with no `cache_salt`.
  • Correct turn-1 call for contrast: `vllm/entrypoints/openai/responses/serving.py#L755` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/entrypoints/openai/responses/serving.py#L755 — `tokens_input(prompt_token_ids, cache_salt=request.cache_salt)`.
  • `tokens_input` stores the salt only if passed: `vllm/inputs/engine.py#L51-L66` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/inputs/engine.py#L51-L66 (`if cache_salt is not None: inputs["cache_salt"] = cache_salt`).
  • The engine request copies only the current input's salt: `vllm/v1/engine/input_processor.py#L380` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/v1/engine/input_processor.py#L380 (`cache_salt=decoder_inputs.get("cache_salt")` → `None` for the continuation).
  • Prefix-cache hashing keys on the salt only when present: `vllm/v1/core/kv_cache_utils.py#L560-L561` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/v1/core/kv_cache_utils.py#L560-L561 (`[request.cache_salt] if (start_token_idx == 0 and request.cache_salt) else []`).
  • The oracle the attacker reads: `vllm/entrypoints/openai/responses/serving.py#L909` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/entrypoints/openai/responses/serving.py#L909 (`cached_tokens_per_turn`).
  • The documented control being defeated: `vllm/entrypoints/openai/responses/protocol.py#L235` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/entrypoints/openai/responses/protocol.py#L235 (`cache_salt` field).

The tool-continuation re-submission rebuilds the engine input with no `cache_salt`:

# vllm/entrypoints/openai/responses/serving.py Lines 711-715
            if isinstance(context, HarmonyContext):
                token_ids = context.render_for_completion()
                engine_input = tokens_input(token_ids)

                sampling_params.max_tokens = max_model_len - len(token_ids)

Contrast with the correct turn-1 call, which does preserve the caller's salt:

# vllm/entrypoints/openai/responses/serving.py Lines 754-755
        prompt_token_ids = render_for_completion(messages)
        engine_input = tokens_input(prompt_token_ids, cache_salt=request.cache_salt)

`tokens_input` stores the salt on the engine input only when it is passed, so the continuation input carries none and lands in the unsalted namespace:

# vllm/inputs/engine.py Lines 51-66
def tokens_input(
    prompt_token_ids: list[int],
    *,
    prompt: str | None = None,
    cache_salt: str | None = None,
) -> TokensInput:
    """
    Construct [`TokensInput`][vllm.inputs.engine.TokensInput]
    from optional values.
    """
    inputs = TokensInput(type="token", prompt_token_ids=prompt_token_ids)

    if prompt is not None:
        inputs["prompt"] = prompt
    if cache_salt is not None:
        inputs["cache_salt"] = cache_salt

Impact

An authenticated tenant of a shared deployment can recover whether a guessed post-tool prompt or history was processed by another tenant, with exact cached-token counts rather than noisy latency — the exact prompt-membership oracle `cache_salt` is documented to prevent. It defeats the multi-user prefix-cache isolation guarantee for salted Harmony tool sessions.

Preconditions: a GPT-OSS Harmony model on `/v1/responses`; prefix caching enabled (default); an operator-enabled built-in or MCP tool server; the victim sets `cache_salt` and triggers at least one tool continuation; and the attacker can reconstruct the post-tool history closely enough to match the token prefix. The `AC:H` metric reflects that guessable-history precondition.

Suggested Fix

Propagate `request.cache_salt` into every Harmony (and Parsable) tool-continuation re-submission — at the continuation call site call `tokens_input(token_ids, cache_salt=request.cache_salt)`, mirroring the correct turn-1 call. Carry the salt on the `HarmonyContext` (thread the originating `request` into the context) so no continuation path can omit it:

# vllm/entrypoints/openai/responses/serving.py
             if isinstance(context, HarmonyContext):
                 token_ids = context.render_for_completion()
-                engine_input = tokens_input(token_ids)
+                engine_input = tokens_input(
+                    token_ids,
+                    cache_salt=(
+                        context.request.cache_salt
+                        if context.request is not None
+                        else None
+                    ),
+                )

with `HarmonyContext.__init__` gaining a `request: ResponsesRequest | None = None` parameter (stored as `self.request`) that `_create_responses` passes when constructing the context. The continuation prefix is then cached in the victim's salted namespace, mirroring turn 1.

Suggested regression test: assert `cached_tokens_per_turn == 0` for a different-salt probe against a salted victim continuation (the four-way control from the proof of concept).

Credit

Reported by: Patch the Planet (Trail of Bits + OpenAI collaboration)

This vulnerability was discovered using GPT-5.5-Cyber as part of the Patch the Planet security initiative.

---

Proposed fix: a fix for this issue is proposed in a public pull request: https://github.com/vllm-project/vllm/pull/51818

Incluye detalles técnicos publicados por el investigador. Úselos solo para validar y corregir sus propios sistemas.

Versiones

Rangos afectados

Afectada Corregida Última
vllm< 0.30.0
0.30.0
0.34.0
Remediación

Estado del parche por paquete

Paquete / productoEstadoCorregida enDesde divulgaciónÚltima versión
pypivllm
Corregida 0.30.0 -14 d 0.34.0 no afectada
Versiones corregidas0.30.0
Explotación

Arsenal público

Sin exploits, módulos ni PoC públicos conocidos a la fecha de verificación.

Referencias
Otrahttps://github.com/vllm-project/vllm/security/advisories/GHSA-935w-9g4m-p28p
Otrahttps://github.com/vllm-project/vllm/pull/50195
Otrahttps://github.com/vllm-project/vllm/pull/51818
Otrahttps://github.com/vllm-project/vllm/commit/6a2a2bb02b563b83f946012959fd3927984d072a
Otrahttps://github.com/vllm-project/vllm
Otrahttps://github.com/vllm-project/vllm/releases/tag/v0.30.0

Las referencias externas se muestran para copiar: cve.llama.ec no lo redirige a otros sitios.