GHSA-mrw7-hf4f-83pf CVE-2025-62164 PYSEC-2026-2018
vLLM deserialization vulnerability leading to DoS and potential RCE
High
PyPI
- Severity
- 8.8 / 10
- High · CVSS 3.1
- Exploit chance (EPSS)
- 0.96%
- Chance of exploitation in the next 30 days. Higher than 61.67% of scored vulnerabilities.
- Known exploited
- No
- Not in CISA's catalog
- Weakness
- CWE-123 Write-what-where Condition
- CWE-20 Improper Input Validation
- CWE-502 Deserialization of Untrusted Data
- CWE-787 Out-of-bounds Write
- Fix
0.11.1- Upgrade vllm to this version or later
Description
Summary
A memory corruption vulnerability that leading to a crash (denial-of-service) and potentially remote code execution (RCE) exists in vLLM versions 0.10.2 and later, in the Completions API endpoint. When processing user-supplied prompt embeddings, the endpoint loads serialized tensors using torch.load() without sufficient validation.
Due to a change introduced in PyTorch 2.8.0, sparse tensor integrity checks are disabled by default. As a result, maliciously crafted tensors can bypass internal bounds checks and trigger an out-of-bounds memory write during the call to to_dense(). This memory corruption can crash vLLM and potentially lead to code execution on the server hosting vLLM.
Details
A vulnerability that can lead to RCE from the completions API endpoint exists in vllm, where due to missing checks when loading user-provided tensors, an out-of-bounds write can be triggered. This happens because the default behavior of torch.load(tensor, weights_only=True) since pytorch 2.8.0 is to not perform validity checks for sparse tensors, and this needs to be enabled explicitly using the torch.sparse.check_sparse_tensor_invariants context manager.
The vulnerability is in the following code in vllm/entrypoints/renderer.py:148
def _load_and_validate_embed(embed: bytes) -> EngineEmbedsPrompt:
tensor = torch.load(
io.BytesIO(pybase64.b64decode(embed, validate=True)),
weights_only=True,
map_location=torch.device("cpu"),
)
assert isinstance(tensor, torch.Tensor) and tensor.dtype in (
torch.float32,
torch.bfloat16,
torch.float16,
)
tensor = tensor.to_dense()
Because of the missing checks, loading invalid prompt embedding tensors provided by the user can cause an out-of-bounds write in the call to to_dense .
Impact
All users with access to this API are able to exploit this vulnerability. Unsafe deserialization of untrusted input can be abused to achieve DoS and potentially remote code execution (RCE) in the vLLM server process. This impacts deployments running vLLM as a server or any instance that deserializes untrusted/model-provided payloads.
Fix
https://github.com/vllm-project/vllm/pull/27204
Acknowledgements
Finder: AXION Security Research Team (Omri Fainaro, Bary Levy): discovery and coordinated disclosure.
Affected versions
PyPI
| Package | Vulnerable | Fixed in |
|---|---|---|
| vllm | >= 0.10.2, < 0.11.1 |
0.11.1 |
How to fix
Move to the fixed version or later with your package manager, update the lockfile, and run your tests. If the package only arrives through another dependency, upgrade that dependency first. If none of its releases pulls a fixed version yet, force the version with your package manager's override.
vllm PyPI → 0.11.1
- pip
pip install --upgrade "vllm>=0.11.1"- poetry
poetry add "vllm>=0.11.1"- uv
uv add "vllm>=0.11.1"
Severity breakdown
CVSS 3.1 8.8
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
- Attack vector
- Network
- Attack complexity
- Low
- Privileges required
- Low
- User interaction
- None
- Scope
- Unchanged
- Confidentiality
- High
- Integrity
- High
- Availability
- High
Weakness
CWE-123: Write-what-where Condition
Any condition where the attacker has the ability to write an arbitrary value to an arbitrary location, often as the result of a buffer overflow.
How to prevent it
- Architecture and Design: Use a language that provides appropriate memory abstractions.
- Operation: Use OS-level preventative functionality integrated after the fact. Not a complete solution.
CWE-20: Improper Input Validation
The product receives input or data, but it does not validate or incorrectly validates that the input has the properties that are required to process the data safely and correctly.
How to prevent it
- Architecture and Design: Consider using language-theoretic security (LangSec) techniques that characterize inputs using a formal language and build "recognizers" for that language. This effectively requires parsing to be a distinct layer that effectively enforces a boundary between raw input and internal data representations, instead of allowing parser code to be scattered throughout the program, where it could be subject to errors or inconsistencies that create weaknesses. [REF-1109] [REF-1110] [REF-1111]
- Architecture and Design: Use an input validation framework such as Struts or the OWASP ESAPI Validation API. Note that using a framework does not automatically address all input validation problems; be mindful of weaknesses that could arise from misusing the framework itself (CWE-1173).
- Architecture and Design, Implementation: Understand all the potential areas where untrusted inputs can enter the product, including but not limited to: parameters or arguments, cookies, anything read from the network, environment variables, reverse DNS lookups, query results, request headers, URL components, e-mail, files, filenames, databases, and any external systems that provide data to the application. Remember that such inputs may be obtained indirectly through API calls.
- Implementation: Assume all input is malicious. Use an "accept known good" input validation strategy, i.e., use a list of acceptable inputs that strictly conform to specifications. Reject any input that does not strictly conform to specifications, or transform it into something that does. When performing input validation, consider all potentially relevant properties, including length, type of input, the full range of acceptable values, missing or extra inputs, syntax, consistency across related fields, and conformance to business rules. As an example of business rule logic, "boat" may be syntactically valid because it only contains alphanumeric characters, but it is not valid if the input is only expected to contain colors such as "red" or "blue." Do not rely exclusively on looking for malicious or malformed inputs. This is likely to miss at least one undesirable input, especially if the code's environment changes. This can give attackers enough room to bypass the intended validation. However, denylists can be useful for detecting potential attacks or determining which inputs are so malformed that they should be rejected outright.
- Architecture and Design: For any security checks that are performed on the client side, ensure that these checks are duplicated on the server side, in order to avoid CWE-602. Attackers can bypass the client-side checks by modifying values after the checks have been performed, or by changing the client to remove the client-side checks entirely. Then, these modified values would be submitted to the server. Even though client-side checks provide minimal benefits with respect to server-side security, they are still useful. First, they can support intrusion detection. If the server receives input that should have been rejected by the client, then it may be an indication of an attack. Second, client-side error-checking can provide helpful feedback to the user about the expectations for valid input. Third, there may be a reduction in server-side processing time for accidental input errors, although this is typically a small savings.
- Implementation: When your application combines data from multiple sources, perform the validation after the sources have been combined. The individual data elements may pass the validation step but violate the intended restrictions after they have been combined.
- Implementation: Be especially careful to validate all input when invoking code that crosses language boundaries, such as from an interpreted language to native code. This could create an unexpected interaction between the language boundaries. Ensure that you are not violating any of the expectations of the language with which you are interfacing. For example, even though Java may not be susceptible to buffer overflows, providing a large argument in a call to native code might trigger an overflow.
- Implementation: Directly convert your input type into the expected data type, such as using a conversion function that translates a string into a number. After converting to the expected data type, ensure that the input's values fall within the expected range of allowable values and that multi-field consistencies are maintained.
- Implementation: Inputs should be decoded and canonicalized to the application's current internal representation before being validated (CWE-180, CWE-181). Make sure that your application does not inadvertently decode the same input twice (CWE-174). Such errors could be used to bypass allowlist schemes by introducing dangerous inputs after they have been checked. Use libraries such as the OWASP ESAPI Canonicalization control. Consider performing repeated canonicalization until your input does not change any more. This will avoid double-decoding and similar scenarios, but it might inadvertently modify inputs that are allowed to contain properly-encoded dangerous content.
- Implementation: When exchanging data between components, ensure that both components are using the same character encoding. Ensure that the proper encoding is applied at each interface. Explicitly set the encoding you are using whenever the protocol allows you to do so.
CWE-502: Deserialization of Untrusted Data
The product deserializes untrusted data without sufficiently ensuring that the resulting data will be valid.
How to prevent it
- Architecture and Design, Implementation: If available, use the signing/sealing features of the programming language to assure that deserialized data has not been tainted. For example, a hash-based message authentication code (HMAC) could be used to ensure that data has not been modified.
- Implementation: When deserializing data, populate a new object rather than just deserializing. The result is that the data flows through safe input validation and that the functions are safe.
- Implementation: Explicitly define a final object() to prevent deserialization.
- Architecture and Design, Implementation: Make fields transient to protect them from deserialization. An attempt to serialize and then deserialize a class containing transient fields will result in NULLs where the transient data should be. This is an excellent way to prevent time, environment-based, or sensitive variables from being carried over and used improperly.
- Implementation: Avoid having unnecessary types or gadgets (a sequence of instances and method invocations that can self-execute during the deserialization process, often found in libraries) available that can be leveraged for malicious ends. This limits the potential for unintended or unauthorized types and gadgets to be leveraged by the attacker. Add only acceptable classes to an allowlist. Note: new gadgets are constantly being discovered, so this alone is not a sufficient mitigation.
- Architecture and Design, Implementation: Employ cryptography of the data or code for protection. However, it's important to note that it would still be client-side security. This is risky because if the client is compromised then the security implemented on the client (the cryptography) can be bypassed.
- Operation: Use an application firewall that can detect attacks against this weakness. It can be beneficial in cases in which the code cannot be fixed (because it is controlled by a third party), as an emergency prevention measure while more comprehensive software assurance measures are applied, or to provide defense in depth [REF-1481].
CWE-787: Out-of-bounds Write
The product writes data past the end, or before the beginning, of the intended buffer.
How to prevent it
- Requirements: Use a language that does not allow this weakness to occur or provides constructs that make this weakness easier to avoid. For example, many languages that perform their own memory management, such as Java and Perl, are not subject to buffer overflows. Other languages, such as Ada and C#, typically provide overflow protection, but the protection can be disabled by the programmer. Be wary that a language's interface to native code may still be subject to overflows, even if the language itself is theoretically safe.
- Architecture and Design: Use a vetted library or framework that does not allow this weakness to occur or provides constructs that make this weakness easier to avoid. Examples include the Safe C String Library (SafeStr) by Messier and Viega [REF-57], and the Strsafe.h library from Microsoft [REF-56]. These libraries provide safer versions of overflow-prone string-handling functions.
- Operation, Build and Compilation: Use automatic buffer overflow detection mechanisms that are offered by certain compilers or compiler extensions. Examples include: the Microsoft Visual Studio /GS flag, Fedora/Red Hat FORTIFY_SOURCE GCC flag, StackGuard, and ProPolice, which provide various mechanisms including canary-based detection and range/index checking. D3-SFCV (Stack Frame Canary Validation) from D3FEND [REF-1334] discusses canary-based detection in detail.
- Implementation: Consider adhering to the following rules when allocating and managing an application's memory: Double check that the buffer is as large as specified. When using functions that accept a number of bytes to copy, such as strncpy(), be aware that if the destination buffer size is equal to the source buffer size, it may not NULL-terminate the string. Check buffer boundaries if accessing the buffer in a loop and make sure there is no danger of writing past the allocated space. If necessary, truncate all input strings to a reasonable length before passing them to the copy and concatenation functions.
- Operation, Build and Compilation: Run or compile the software using features or extensions that randomly arrange the positions of a program's executable and libraries in memory. Because this makes the addresses unpredictable, it can prevent an attacker from reliably jumping to exploitable code. Examples include Address Space Layout Randomization (ASLR) [REF-58] [REF-60] and Position-Independent Executables (PIE) [REF-64]. Imported modules may be similarly realigned if their default memory addresses conflict with other modules, in a process known as "rebasing" (for Windows) and "prelinking" (for Linux) [REF-1332] using randomly generated addresses. ASLR for libraries cannot be used in conjunction with prelink since it would require relocating the libraries at run-time, defeating the whole purpose of prelinking. For more information on these techniques see D3-SAOR (Segment Address Offset Randomization) from D3FEND [REF-1335].
- Operation: Use a CPU and operating system that offers Data Execution Protection (using hardware NX or XD bits) or the equivalent techniques that simulate this feature in software, such as PaX [REF-60] [REF-61]. These techniques ensure that any instruction executed is exclusively at a memory address that is part of the code segment. For more information on these techniques see D3-PSEP (Process Segment Execution Prevention) from D3FEND [REF-1336].
- Implementation: Replace unbounded copy functions with analogous functions that support length arguments, such as strcpy with strncpy. Create these if they are not available.
References
Advisories
Packages
Other links
Timeline
- Published
- Last updated by the source
- Latest EPSS score
Related advisories
| Advisory | Severity | Why it's related |
|---|---|---|
| GHSA-8397-8qf6-f2gm vLLM: Phi-4 multimodal audio encoder materializes and retains full attention matrices: a single max-length audio OOM-crashes the vLLM engine | Medium | Also affects vllm |
| GHSA-x567-49g5-p4cf vLLM: Responses API bypasses trust_request_chat_template for chat_template_kwargs | Medium | Also affects vllm |
| GHSA-vvpg-rrww-wpmh vLLM: Permissive default CORS origins allow cross-origin requests to the vLLM HTTP server | Medium | Also affects vllm |
| GHSA-q97c-p368-75c4 vLLM: Out-of-vocabulary prompt token id passes input validation and triggers a CUDA device-side assert (engine crash / DoS) | Medium | Also affects vllm |
| GHSA-h3rc-j82p-2c3h vLLM: Untrusted EPS inputs can trigger denial of service or Ghostscript RCE | High | Also affects vllm |
| GHSA-gq5q-89c7-vhvx vLLM DeepSeek-OCR image processing vulnerable to denial of service | Medium | Also affects vllm |
| GHSA-6757-hggm-hp2w vLLM: Public multimodal SHM handle forgery reaches worker-side `pickle.loads()` with insecure serialization disabled | High | Also affects vllm |
| GHSA-542m-9234-8gvf vLLM: DoS via unbounded chat_template Jinja2 rendering exhausts the renderer thread pool | Medium | Also affects vllm |
MergeHarbor checks every repository's dependencies against this database, ranks what matters by exploitability and reachability, routes it to the owning team and opens the fix pull request.
Check your repos with MergeHarbor