llamaCVEVulnerability Intelligence
/
Advisory · GitHub · PyPI · vllm

GHSA-PH3R-5JFG-F84F

vLLM: Mirrored multimodal IPC caches desync after a rejected request — a later request reusing the same media hash trips a receiver assertion in the engine core

Parche disponible
Descargar PNG JSON
CVSS
—
medium
Fuente: —
ExplotaciónSin evidenciaNo figura en CISA KEV
Exploits públicos0Ninguno conocido
Parche1 versiónPrimer parche a -41 d de la divulgación
Exposición45 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 0.28.0
Monitoreo
Detectada
26 ago 26
6 oct 26
8 oct 26
17 nov 26
20 nov 26
26 ago 26
16 sept 26
8 oct 26
29 oct 26
20 nov 26
FechaFaseEventoDetalle
Remediación Parche 0.28.0 vllm · -41 d desde divulgación
Divulgación Publicada GHSA-PH3R-5JFG-F84F 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 mirrored sender/receiver cache protocol reaches.

Summary

vLLM's default multimodal cache (`mm_processor_cache_type="lru"`) mirrors state across two processes: the frontend (P0) holds only metadata (`MultiModalProcessorSenderCache`) while the engine core (P1) holds the real payload (`MultiModalReceiverCache`). The design invariant is that `get_and_update()` runs on P0 and P1 in lockstep for every request, so eviction order stays mirrored and P0 can answer "is this cached in P1?" without talking to P1.

That invariant breaks when a request is rejected after P0 has rendered and hashed the multimodal input (populating the P0 cache) but before P1 receives the item — for example, an oversized chat prompt rejected on `max_model_len` *after* rendering. P0 now believes the media is cached while P1 never got it. A later request reusing the same media hash gets a P0 hit, so P0 sends `None` instead of the payload, and P1 — which has nothing cached — trips `assert mm_item is not None, f"Expected a cached item for {mm_hash=}"`.

This is a remotely reachable, request-controlled cache-mirroring desync on the standard multimodal inference path. It requires only the default cache configuration.

Affected code

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

  • P0 metadata cache — `MultiModalProcessorSenderCache` at `vllm/multimodal/cache.py#L379` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/multimodal/cache.py#L379; `get_and_update_item` at `#L410-L421` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/multimodal/cache.py#L410-L421, commit assertion at `#L418` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/multimodal/cache.py#L418.
  • P1 payload cache (the actual sink) — `MultiModalReceiverCache` at `vllm/multimodal/cache.py#L630` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/multimodal/cache.py#L630; `get_and_update_item` at `#L652-L663` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/multimodal/cache.py#L652-L663, with the failing `assert mm_item is not None, f"Expected a cached item for {mm_hash=}"` at `#L660` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/multimodal/cache.py#L660.
  • Default `mm_processor_cache_type = "lru"` at `vllm/config/multimodal.py#L132` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/config/multimodal.py#L132; dispatch in `vllm/multimodal/registry.py#L294-L307` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/multimodal/registry.py#L294-L307 (sender) and `#L322-L331` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/multimodal/registry.py#L322-L331 (receiver). The `processor_only`, disabled-caching, and `shm` paths are not affected.
  • Rejection-after-render window — rendering happens before length validation in `vllm/entrypoints/openai/chat_completion/serving.py#L206-L231` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/entrypoints/openai/chat_completion/serving.py#L206-L231 (`render_chat_request`), and the length check raises after the render in `vllm/entrypoints/serve/utils/api_utils.py#L171-L189` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/entrypoints/serve/utils/api_utils.py#L171-L189.
  • Cache-commit call site — `vllm/multimodal/processing/processor.py#L1347` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/multimodal/processing/processor.py#L1347 (`_merge_mm_kwargs`) commits the P0 sender entry during render.
  • The separate stale-order eviction hang is already fixed via `vllm/utils/cache.py#L120-L121` https://github.com/vllm-project/vllm/blob/752a3a504485790a2e8491cacbb35c137339ad34/vllm/utils/cache.py#L120-L121 (`LRUCache.touch()` guarding `if key in self:`) — a different mechanism that does not touch the sender/receiver commit-ordering protocol and does not remediate this assertion.

The P1 receiver sink — the assertion that fires when P1 receives `None` for a hash it never cached:

# vllm/multimodal/cache.py Lines 651-663
    @override
    def get_and_update_item(
        self,
        mm_item: MultiModalKwargsItem | None,
        mm_hash: str,
    ) -> MultiModalKwargsItem:
        if (cached_item := self._cache.get(mm_hash)) is not None:
            return cached_item

        assert mm_item is not None, f"Expected a cached item for {mm_hash=}"

        self._cache[mm_hash] = mm_item
        return mm_item

The P0 sender — on a hit it drops the payload (returns `None`) and, on a miss during render, unconditionally commits the metadata entry with no rollback tied to admission:

# vllm/multimodal/cache.py Lines 409-422
    @override
    def get_and_update_item(
        self,
        mm_item: MultiModalProcessorCacheInItem,
        mm_hash: str,
    ) -> MultiModalProcessorCacheOutItem:
        if (cached_item := self._cache.get(mm_hash)) is not None:
            return None, cached_item.prompt_updates

        assert mm_item is not None, f"Expected a cached item for {mm_hash=}"

        self._cache[mm_hash] = MultiModalProcessorCacheItemMetadata(*mm_item)

        return mm_item

Impact

A remote client submitting multimodal requests can poison a cache identity — render a media item successfully, then have that request rejected — so a later request reusing the same media hash fails on the P1 receiver assertion. This is an availability failure against a shared serving instance. No code execution, memory corruption, or data disclosure is claimed.

On this revision the failure is scoped as a request-level preprocessing error (the engine core catches around `preprocess_add_request`); public reports show the same assertion cascading into further engine-loop assertions on other revisions. It applies to multimodal models running the default mirrored `lru` cache.

Suggested Fix

Two complementary changes:

  1. Make the mirrored commit atomic with admission — insert into the P0 sender cache only after the request has passed all admission checks (length, limits) and P1 has acknowledged the item, or roll back the P0 insert on rejection.
  2. Defense in depth — convert the P1 receiver `assert mm_item is not None` into a checked, request-scoped error (fetch-on-miss from P0) so a desync degrades a single request rather than asserting in the engine loop.

The core of the rollback half: wrap the post-render length check so a `ValueError` rejection discards the P0 entries the render just committed, before re-raising. Add a `discard_sender_cache_item()` on the processor cache (no-op default, `pop` on the sender) and a `Renderer.discard_mm_cache_entries()` that walks a rendered request's `mm_hashes`:

# vllm/entrypoints/openai/chat_completion/serving.py (_create_chat_completion)
-            max_tokens = get_max_tokens(
-                max_model_len,
-                ...,
-                truncate_prompt_tokens=request.truncate_prompt_tokens,
-            )
+            try:
+                max_tokens = get_max_tokens(
+                    max_model_len,
+                    ...,
+                    truncate_prompt_tokens=request.truncate_prompt_tokens,
+                )
+            except ValueError:
+                for rendered_input in engine_inputs:
+                    if mm_hashes := rendered_input.get("mm_hashes"):
+                        self.renderer.discard_mm_cache_entries(mm_hashes)
+                raise
# vllm/multimodal/cache.py (MultiModalProcessorSenderCache)
+    @override
+    def discard_sender_cache_item(self, mm_hash: str) -> None:
+        self._cache.pop(mm_hash, None)

This closes the `max_model_len` rejection path; because any other rejection-after-render path reopens the same window, pairing it with the defense-in-depth change above (making the P1 `assert` a checked, request-scoped error) is recommended.

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/51897

Versiones

Rangos afectados

Afectada Corregida Última
vllm< 0.28.0
0.28.0
0.34.0
Remediación

Estado del parche por paquete

Paquete / productoEstadoCorregida enDesde divulgaciónÚltima versión
pypivllm
Corregida 0.28.0 -41 d 0.34.0 no afectada
Versiones corregidas0.28.0
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-ph3r-5jfg-f84f
Otrahttps://github.com/vllm-project/vllm/pull/46747
Otrahttps://github.com/vllm-project/vllm/pull/51897
Otrahttps://github.com/vllm-project/vllm/commit/396204230423b7cc6798300926b8fa30190d26a9
Otrahttps://github.com/vllm-project/vllm
Otrahttps://github.com/vllm-project/vllm/releases/tag/v0.28.0

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