GHSA-PF3H-QJGV-VCPR

Vulnerability from github – Published: 2026-04-03 21:51 – Updated: 2026-04-06 23:20
VLAI?
Summary
vLLM: Server-Side Request Forgery (SSRF) in `download_bytes_from_url `
Details

Summary

A Server Side Request Forgery (SSRF) vulnerability in download_bytes_from_url allows any actor who can control batch input JSON to make the vLLM batch runner issue arbitrary HTTP/HTTPS requests from the server, without any URL validation or domain restrictions.

This can be used to target internal services (e.g. cloud metadata endpoints or internal HTTP APIs) reachable from the vLLM host.

Details

Vulnerable component

The vulnerable logic is in the batch runner entrypoint vllm/entrypoints/openai/run_batch.py, function download_bytes_from_url:

# run_batch.py Lines 442-482
async def download_bytes_from_url(url: str) -> bytes:
    """
    Download data from a URL or decode from a data URL.

    Args:
        url: Either an HTTP/HTTPS URL or a data URL (data:...;base64,...)

    Returns:
        Data as bytes
    """
    parsed = urlparse(url)

    # Handle data URLs (base64 encoded)
    if parsed.scheme == "data":
        # Format: data:...;base64,<base64_data>
        if "," in url:
            header, data = url.split(",", 1)
            if "base64" in header:
                return base64.b64decode(data)
            else:
                raise ValueError(f"Unsupported data URL encoding: {header}")
        else:
            raise ValueError(f"Invalid data URL format: {url}")

    # Handle HTTP/HTTPS URLs
    elif parsed.scheme in ("http", "https"):
        async with (
            aiohttp.ClientSession() as session,
            session.get(url) as resp,
        ):
            if resp.status != 200:
                raise Exception(
                    f"Failed to download data from URL: {url}. Status: {resp.status}"
                )
            return await resp.read()

    else:
        raise ValueError(
            f"Unsupported URL scheme: {parsed.scheme}. "
            "Supported schemes: http, https, data"
        )

Key properties:

  • The function only parses the URL to dispatch on the scheme (data, http, https).
  • For http / https, it directly calls session.get(url) on the provided string.
  • There is no validation of:
  • hostname or IP address,
  • whether the target is internal or external,
  • port number,
  • path, query, or redirect target.
  • This is in contrast to the multimodal media path (MediaConnector), which implements an explicit domain allowlist. download_bytes_from_url does not reuse that protection.

URL controllability

The url argument is fully controlled by batch input JSON via the file_url field of BatchTranscriptionRequest / BatchTranslationRequest.

  1. Batch request body type:
# run_batch.py Line 67-80
class BatchTranscriptionRequest(TranscriptionRequest):
    """
    Batch transcription request that uses file_url instead of file.

    This class extends TranscriptionRequest but replaces the file field
    with file_url to support batch processing from audio files written in JSON format.
    """

    file_url: str = Field(
        ...,
        description=(
            "Either a URL of the audio or a data URL with base64 encoded audio data. "
        ),
    )

# run_batch.py Line 98-111
class BatchTranslationRequest(TranslationRequest):
    """
    Batch translation request that uses file_url instead of file.

    This class extends TranslationRequest but replaces the file field
    with file_url to support batch processing from audio files written in JSON format.
    """

    file_url: str = Field(
        ...,
        description=(
            "Either a URL of the audio or a data URL with base64 encoded audio data. "
        ),
    )

There is no restriction on the domain, IP, or port of file_url in these models.

  1. Batch input is parsed directly from the batch file:
# run_batch.py Line 139-179
class BatchRequestInput(OpenAIBaseModel):
    ...
    url: str
    body: BatchRequestInputBody
    @field_validator("body", mode="plain")
    @classmethod
    def check_type_for_url(cls, value: Any, info: ValidationInfo):
        url: str = info.data["url"]
        ...
        if url == "/v1/audio/transcriptions":
            return BatchTranscriptionRequest.model_validate(value)
        if url == "/v1/audio/translations":
            return BatchTranslationRequest.model_validate(value)

# run_batch.py Line 770-781
   logger.info("Reading batch from %s...", args.input_file)

    # Submit all requests in the file to the engine "concurrently".
    response_futures: list[Awaitable[BatchRequestOutput]] = []
    for request_json in (await read_file(args.input_file)).strip().split("\n"):
        # Skip empty lines.
        request_json = request_json.strip()
        if not request_json:
            continue

        request = BatchRequestInput.model_validate_json(request_json)

The batch runner reads each line of the input file (args.input_file), parses it as JSON, and constructs a BatchTranscriptionRequest / BatchTranslationRequest. Whatever file_url appears in that JSON line becomes batch_request_body.file_url.

  1. file_url is passed directly into download_bytes_from_url:
# run_batch.py Line 610-623
def wrapper(handler_fn: Callable):
        async def transcription_wrapper(
            batch_request_body: (BatchTranscriptionRequest | BatchTranslationRequest),
        ) -> (
            TranscriptionResponse
            | TranscriptionResponseVerbose
            | TranslationResponse
            | TranslationResponseVerbose
            | ErrorResponse
        ):
            try:
                # Download data from URL
                audio_data = await download_bytes_from_url(batch_request_body.file_url)

So the data flow is:

  1. Attacker supplies JSON line in the batch input file with arbitrary body.file_url.
  2. BatchRequestInput / BatchTranscriptionRequest / BatchTranslationRequest parse that JSON and store file_url verbatim.
  3. make_transcription_wrapper calls download_bytes_from_url(batch_request_body.file_url).
  4. download_bytes_from_url’s HTTP/HTTPS branch issues aiohttp.ClientSession().get(url) to that attacker-controlled URL with no further validation.

This is a classic SSRF pattern: a server-side component makes arbitrary HTTP requests to a URL string taken from untrusted input.

Comparison with safer code

The project already contains a safer URL-handling path for multimodal media in vllm/multimodal/media/connector.py, which demonstrates the intent to mitigate SSRF via domain allowlists and URL normalization:

# connector.py Lines 169-189
 def load_from_url(
        self,
        url: str,
        media_io: MediaIO[_M],
        *,
        fetch_timeout: int | None = None,
    ) -> _M:  # type: ignore[type-var]
        url_spec = parse_url(url)

        if url_spec.scheme and url_spec.scheme.startswith("http"):
            self._assert_url_in_allowed_media_domains(url_spec)

            connection = self.connection
            data = connection.get_bytes(
                url_spec.url,
                timeout=fetch_timeout,
                allow_redirects=envs.VLLM_MEDIA_URL_ALLOW_REDIRECTS,
            )

            return media_io.load_bytes(data)

and:

# connector.py Lines 158-167
  def _assert_url_in_allowed_media_domains(self, url_spec: Url) -> None:
        if (
            self.allowed_media_domains
            and url_spec.hostname not in self.allowed_media_domains
        ):
            raise ValueError(
                f"The URL must be from one of the allowed domains: "
                f"{self.allowed_media_domains}. Input URL domain: "
                f"{url_spec.hostname}"
            )

download_bytes_from_url does not reuse this allowlist or any equivalent validation, even though it also fetches user-provided URLs.

Severity ?
5.4 (Medium)
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:L
Show details on source website
To clipboard 

{
  "affected": [
    {
      "package": {
        "ecosystem": "PyPI",
        "name": "vllm"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0.16.0"
            },
            {
              "fixed": "0.19.0"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2026-34753"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-918"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-04-03T21:51:00Z",
    "nvd_published_at": "2026-04-06T16:16:36Z",
    "severity": "MODERATE"
  },
  "details": "### Summary\n\nA Server Side Request Forgery (SSRF) vulnerability in `download_bytes_from_url` allows any actor who can control batch input JSON to make the vLLM batch runner issue arbitrary HTTP/HTTPS requests from the server, without any URL validation or domain restrictions.\n\nThis can be used to target internal services (e.g. cloud metadata endpoints or internal HTTP APIs) reachable from the vLLM host.\n\n------\n\n### Details\n\n#### Vulnerable component\n\nThe vulnerable logic is in the batch runner entrypoint `vllm/entrypoints/openai/run_batch.py`, function `download_bytes_from_url`:\n\n```\n# run_batch.py Lines 442-482\nasync def download_bytes_from_url(url: str) -\u003e bytes:\n    \"\"\"\n    Download data from a URL or decode from a data URL.\n\n    Args:\n        url: Either an HTTP/HTTPS URL or a data URL (data:...;base64,...)\n\n    Returns:\n        Data as bytes\n    \"\"\"\n    parsed = urlparse(url)\n\n    # Handle data URLs (base64 encoded)\n    if parsed.scheme == \"data\":\n        # Format: data:...;base64,\u003cbase64_data\u003e\n        if \",\" in url:\n            header, data = url.split(\",\", 1)\n            if \"base64\" in header:\n                return base64.b64decode(data)\n            else:\n                raise ValueError(f\"Unsupported data URL encoding: {header}\")\n        else:\n            raise ValueError(f\"Invalid data URL format: {url}\")\n\n    # Handle HTTP/HTTPS URLs\n    elif parsed.scheme in (\"http\", \"https\"):\n        async with (\n            aiohttp.ClientSession() as session,\n            session.get(url) as resp,\n        ):\n            if resp.status != 200:\n                raise Exception(\n                    f\"Failed to download data from URL: {url}. Status: {resp.status}\"\n                )\n            return await resp.read()\n\n    else:\n        raise ValueError(\n            f\"Unsupported URL scheme: {parsed.scheme}. \"\n            \"Supported schemes: http, https, data\"\n        )\n```\n\nKey properties:\n\n- The function only parses the URL to dispatch on the scheme (`data`, `http`, `https`).\n- For `http` / `https`, it directly calls `session.get(url)` on the provided string.\n- There is no validation of:\n  - hostname or IP address,\n  - whether the target is internal or external,\n  - port number,\n  - path, query, or redirect target.\n- This is in contrast to the multimodal media path (`MediaConnector`), which implements an explicit domain allowlist. `download_bytes_from_url` does not reuse that protection.\n\n#### URL controllability\n\nThe `url` argument is fully controlled by batch input JSON via the `file_url` field of `BatchTranscriptionRequest` / `BatchTranslationRequest`.\n\n1. Batch request body type:\n\n```\n# run_batch.py Line 67-80\nclass BatchTranscriptionRequest(TranscriptionRequest):\n    \"\"\"\n    Batch transcription request that uses file_url instead of file.\n\n    This class extends TranscriptionRequest but replaces the file field\n    with file_url to support batch processing from audio files written in JSON format.\n    \"\"\"\n\n    file_url: str = Field(\n        ...,\n        description=(\n            \"Either a URL of the audio or a data URL with base64 encoded audio data. \"\n        ),\n    )\n```\n\n```\n# run_batch.py Line 98-111\nclass BatchTranslationRequest(TranslationRequest):\n    \"\"\"\n    Batch translation request that uses file_url instead of file.\n\n    This class extends TranslationRequest but replaces the file field\n    with file_url to support batch processing from audio files written in JSON format.\n    \"\"\"\n\n    file_url: str = Field(\n        ...,\n        description=(\n            \"Either a URL of the audio or a data URL with base64 encoded audio data. \"\n        ),\n    )\n```\n\nThere is no restriction on the domain, IP, or port of `file_url` in these models.\n\n1. Batch input is parsed directly from the batch file:\n\n```\n# run_batch.py Line 139-179\nclass BatchRequestInput(OpenAIBaseModel):\n    ...\n    url: str\n    body: BatchRequestInputBody\n    @field_validator(\"body\", mode=\"plain\")\n    @classmethod\n    def check_type_for_url(cls, value: Any, info: ValidationInfo):\n        url: str = info.data[\"url\"]\n        ...\n        if url == \"/v1/audio/transcriptions\":\n            return BatchTranscriptionRequest.model_validate(value)\n        if url == \"/v1/audio/translations\":\n            return BatchTranslationRequest.model_validate(value)\n```\n\n```\n# run_batch.py Line 770-781\n   logger.info(\"Reading batch from %s...\", args.input_file)\n\n    # Submit all requests in the file to the engine \"concurrently\".\n    response_futures: list[Awaitable[BatchRequestOutput]] = []\n    for request_json in (await read_file(args.input_file)).strip().split(\"\\n\"):\n        # Skip empty lines.\n        request_json = request_json.strip()\n        if not request_json:\n            continue\n\n        request = BatchRequestInput.model_validate_json(request_json)\n```\n\nThe batch runner reads each line of the input file (`args.input_file`), parses it as JSON, and constructs a `BatchTranscriptionRequest` / `BatchTranslationRequest`. Whatever `file_url` appears in that JSON line becomes `batch_request_body.file_url`.\n\n1. `file_url` is passed directly into `download_bytes_from_url`:\n\n```\n# run_batch.py Line 610-623\ndef wrapper(handler_fn: Callable):\n        async def transcription_wrapper(\n            batch_request_body: (BatchTranscriptionRequest | BatchTranslationRequest),\n        ) -\u003e (\n            TranscriptionResponse\n            | TranscriptionResponseVerbose\n            | TranslationResponse\n            | TranslationResponseVerbose\n            | ErrorResponse\n        ):\n            try:\n                # Download data from URL\n                audio_data = await download_bytes_from_url(batch_request_body.file_url)\n```\n\nSo the data flow is:\n\n1. Attacker supplies JSON line in the batch input file with arbitrary `body.file_url`.\n2. `BatchRequestInput` / `BatchTranscriptionRequest` / `BatchTranslationRequest` parse that JSON and store `file_url` verbatim.\n3. `make_transcription_wrapper` calls `download_bytes_from_url(batch_request_body.file_url)`.\n4. `download_bytes_from_url`\u2019s HTTP/HTTPS branch issues `aiohttp.ClientSession().get(url)` to that attacker-controlled URL with no further validation.\n\nThis is a classic SSRF pattern: a server-side component makes arbitrary HTTP requests to a URL string taken from untrusted input.\n\n#### Comparison with safer code\n\nThe project already contains a safer URL-handling path for multimodal media in `vllm/multimodal/media/connector.py`, which demonstrates the intent to mitigate SSRF via domain allowlists and URL normalization:\n\n```\n# connector.py Lines 169-189\n def load_from_url(\n        self,\n        url: str,\n        media_io: MediaIO[_M],\n        *,\n        fetch_timeout: int | None = None,\n    ) -\u003e _M:  # type: ignore[type-var]\n        url_spec = parse_url(url)\n\n        if url_spec.scheme and url_spec.scheme.startswith(\"http\"):\n            self._assert_url_in_allowed_media_domains(url_spec)\n\n            connection = self.connection\n            data = connection.get_bytes(\n                url_spec.url,\n                timeout=fetch_timeout,\n                allow_redirects=envs.VLLM_MEDIA_URL_ALLOW_REDIRECTS,\n            )\n\n            return media_io.load_bytes(data)\n```\n\nand:\n\n```\n# connector.py Lines 158-167\n  def _assert_url_in_allowed_media_domains(self, url_spec: Url) -\u003e None:\n        if (\n            self.allowed_media_domains\n            and url_spec.hostname not in self.allowed_media_domains\n        ):\n            raise ValueError(\n                f\"The URL must be from one of the allowed domains: \"\n                f\"{self.allowed_media_domains}. Input URL domain: \"\n                f\"{url_spec.hostname}\"\n            )\n```\n\n`download_bytes_from_url` does not reuse this allowlist or any equivalent validation, even though it also fetches user-provided URLs.",
  "id": "GHSA-pf3h-qjgv-vcpr",
  "modified": "2026-04-06T23:20:36Z",
  "published": "2026-04-03T21:51:00Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-pf3h-qjgv-vcpr"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-34753"
    },
    {
      "type": "WEB",
      "url": "https://github.com/vllm-project/vllm/pull/38482"
    },
    {
      "type": "WEB",
      "url": "https://github.com/vllm-project/vllm/commit/57861ae48d3493fa48b4d7d830b7ec9f995783e7"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/vllm-project/vllm"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:L",
      "type": "CVSS_V3"
    }
  ],
  "summary": "vLLM: Server-Side Request Forgery (SSRF) in `download_bytes_from_url `"
}


Log in or create an account to share your comment.




Tags
Taxonomy of the tags.




Sightings

Author Source Type Date

Nomenclature

  • Seen: The vulnerability was mentioned, discussed, or observed by the user.
  • Confirmed: The vulnerability has been validated from an analyst's perspective.
  • Published Proof of Concept: A public proof of concept is available for this vulnerability.
  • Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
  • Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
  • Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
  • Not confirmed: The user expressed doubt about the validity of the vulnerability.
  • Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.



Detection rules are retrieved from Rulezet.