vllm-project vllm version 0.6.0 contains a vulnerability in the AsyncEngineRPCServer() RPC server entrypoints. The core functionality run_server_loop() calls the function _make_handler_coro(), which directly uses cloudpickle.loads() on received messages without any sanitization. This can result in remote code execution by deserializing malicious pickle data.

vLLM is an inference and serving engine for large language models (LLMs). From versions 0.10.2 to before 0.11.1, a memory corruption vulnerability could lead to a crash (denial-of-service) and potentially remote code execution (RCE), exists 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. This issue has been patched in version 0.11.1.

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Versions starting from 0.6.5 and prior to 0.8.5, having vLLM integration with mooncake, are vulnerable to remote code execution due to using pickle based serialization over unsecured ZeroMQ sockets. The vulnerable sockets were set to listen on all network interfaces, increasing the likelihood that an attacker is able to reach the vulnerable ZeroMQ sockets to carry out an attack. vLLM instances that do not make use of the mooncake integration are not vulnerable. This issue has been patched in version 0.8.5.

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. When vLLM is configured to use Mooncake, unsafe deserialization exposed directly over ZMQ/TCP on all network interfaces will allow attackers to execute remote code on distributed hosts. This is a remote code execution vulnerability impacting any deployments using Mooncake to distribute KV across distributed hosts. This vulnerability is fixed in 0.8.0.

vLLM, an inference and serving engine for large language models (LLMs), has an issue in versions 0.6.5 through 0.8.4 that ONLY impacts environments using the `PyNcclPipe` KV cache transfer integration with the V0 engine. No other configurations are affected. vLLM supports the use of the `PyNcclPipe` class to establish a peer-to-peer communication domain for data transmission between distributed nodes. The GPU-side KV-Cache transmission is implemented through the `PyNcclCommunicator` class, while CPU-side control message passing is handled via the `send_obj` and `recv_obj` methods on the CPU side.​ The intention was that this interface should only be exposed to a private network using the IP address specified by the `--kv-ip` CLI parameter. The vLLM documentation covers how this must be limited to a secured network. The default and intentional behavior from PyTorch is that the `TCPStore` interface listens on ALL interfaces, regardless of what IP address is provided. The IP address given was only used as a client-side address to use. vLLM was fixed to use a workaround to force the `TCPStore` instance to bind its socket to a specified private interface. As of version 0.8.5, vLLM limits the `TCPStore` socket to the private interface as configured.

vLLM is a library for LLM inference and serving. vllm/model_executor/weight_utils.py implements hf_model_weights_iterator to load the model checkpoint, which is downloaded from huggingface. It uses the torch.load function and the weights_only parameter defaults to False. When torch.load loads malicious pickle data, it will execute arbitrary code during unpickling. This vulnerability is fixed in v0.7.0.

vLLM is an inference and serving engine for large language models. In a multi-node vLLM deployment using the V0 engine, vLLM uses ZeroMQ for some multi-node communication purposes. The secondary vLLM hosts open a `SUB` ZeroMQ socket and connect to an `XPUB` socket on the primary vLLM host. When data is received on this `SUB` socket, it is deserialized with `pickle`. This is unsafe, as it can be abused to execute code on a remote machine. Since the vulnerability exists in a client that connects to the primary vLLM host, this vulnerability serves as an escalation point. If the primary vLLM host is compromised, this vulnerability could be used to compromise the rest of the hosts in the vLLM deployment. Attackers could also use other means to exploit the vulnerability without requiring access to the primary vLLM host. One example would be the use of ARP cache poisoning to redirect traffic to a malicious endpoint used to deliver a payload with arbitrary code to execute on the target machine. Note that this issue only affects the V0 engine, which has been off by default since v0.8.0. Further, the issue only applies to a deployment using tensor parallelism across multiple hosts, which we do not expect to be a common deployment pattern. Since V0 is has been off by default since v0.8.0 and the fix is fairly invasive, the maintainers of vLLM have decided not to fix this issue. Instead, the maintainers recommend that users ensure their environment is on a secure network in case this pattern is in use. The V1 engine is not affected by this issue.

GPT Academic upload Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of GPT Academic. Authentication is not required to exploit this vulnerability. The specific flaw exists within the upload endpoint. The issue results from the lack of proper validation of user-supplied data, which can result in deserialization of untrusted data. An attacker can leverage this vulnerability to execute code in the context of root. Was ZDI-CAN-27957.

Foundation Agents MetaGPT deserialize_message Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Foundation Agents MetaGPT. Authentication is not required to exploit this vulnerability. The specific flaw exists within the deserialize_message function. The issue results from the lack of proper validation of user-supplied data, which can result in deserialization of untrusted data. An attacker can leverage this vulnerability to execute code in the context of the service account. Was ZDI-CAN-28121.

Inductive Automation Ignition ParameterVersionJavaSerializationCodec Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Inductive Automation Ignition. Authentication is not required to exploit this vulnerability. The specific flaw exists within the ParameterVersionJavaSerializationCodec class. The issue results from the lack of proper validation of user-supplied data, which can result in deserialization of untrusted data. An attacker can leverage this vulnerability to execute code in the context of SYSTEM. Was ZDI-CAN-20290.

Upsonic Cloudpickle Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Upsonic. Authentication is not required to exploit this vulnerability. The specific flaw exists within the add_tool endpoint, which listens on TCP port 7541 by default. The issue results from the lack of proper validation of user-supplied data, which can result in deserialization of untrusted data. An attacker can leverage this vulnerability to execute code in the context of the service account. Was ZDI-CAN-26845.

GPT Academic run_in_subprocess_wrapper_func Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of GPT Academic. Authentication is not required to exploit this vulnerability. The specific flaw exists within the run_in_subprocess_wrapper_func function. The issue results from the lack of proper validation of user-supplied data, which can result in deserialization of untrusted data. An attacker can leverage this vulnerability to execute code in the context of root. Was ZDI-CAN-27958.

Vulnerability in the Oracle Coherence product of Oracle Fusion Middleware (component: Caching,CacheStore,Invocation). Supported versions that are affected are 3.7.1.0, 12.1.3.0.0, 12.2.1.3.0 and 12.2.1.4.0. Easily exploitable vulnerability allows unauthenticated attacker with network access via T3 to compromise Oracle Coherence. Successful attacks of this vulnerability can result in takeover of Oracle Coherence. CVSS 3.0 Base Score 9.8 (Confidentiality, Integrity and Availability impacts). CVSS Vector: (CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H).

Apache Camel's Jackson and JacksonXML unmarshalling operation are vulnerable to Remote Code Execution attacks.

This vulnerability allows remote attackers to execute arbitrary code on affected installations of SolarWinds Network Performance Monitor 2020.2.1. Authentication is not required to exploit this vulnerability. The specific flaw exists within the SolarWinds.Serialization library. The issue results from the lack of proper validation of user-supplied data, which can result in deserialization of untrusted data. An attacker can leverage this vulnerability to execute code in the context of SYSTEM. Was ZDI-CAN-12213.

IdeaBlade Breeze Breeze.Server.NET before 1.6.5 allows remote attackers to execute arbitrary code, related to use of TypeNameHandling in JSON deserialization.

Akamai CloudTest before 58.30 allows remote code execution.

Mescius ActiveReports.NET ReadValue Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Mescius ActiveReports.NET. Interaction with this library is required to exploit this vulnerability but attack vectors may vary depending on the implementation. The specific flaw exists within the implementation of the ReadValue method. The issue results from the lack of proper validation of user-supplied data, which can result in deserialization of untrusted data. An attacker can leverage this vulnerability to execute code in the context of the current process. Was ZDI-CAN-25246.

Mescius ActiveReports.NET TypeResolutionService Deserialization of Untrusted Data Remote Code Execution Vulnerability. This vulnerability allows remote attackers to execute arbitrary code on affected installations of Mescius ActiveReports.NET. Interaction with this library is required to exploit this vulnerability but attack vectors may vary depending on the implementation. The specific flaw exists within the TypeResolutionService class. The issue results from the lack of proper validation of user-supplied data, which can result in deserialization of untrusted data. An attacker can leverage this vulnerability to execute code in the context of the current process. Was ZDI-CAN-25397.

IBM Partner Engagement Manager 2.0 could allow a remote attacker to execute arbitrary code on the system, caused by an unsafe deserialization flaw. By sending specially-crafted data, an attacker could exploit this vulnerability to execute arbitrary code on the system. IBM X-Force ID: 203091.

A deserialization vulnerability exists in h2oai/h2o-3 versions <= 3.46.0.8, allowing attackers to read arbitrary system files and execute arbitrary code. The vulnerability arises from improper handling of JDBC connection parameters, which can be exploited by bypassing regular expression checks and using double URL encoding. This issue impacts all users of the affected versions.

A vulnerability in the h2oai/h2o-3 repository allows attackers to exploit deserialization of untrusted data, potentially leading to arbitrary code execution and reading of system files. This issue affects the latest master branch version 3.47.0.99999. The vulnerability arises from the ability to bypass regular expression filters intended to prevent malicious parameter injection in JDBC connections. Attackers can manipulate spaces between parameters to evade detection, allowing for unauthorized file access and code execution. The vulnerability is addressed in version 3.46.0.8.

In Pivotal Spring AMQP versions prior to 1.7.4, 1.6.11, and 1.5.7, an org.springframework.amqp.core.Message may be unsafely deserialized when being converted into a string. A malicious payload could be crafted to exploit this and enable a remote code execution attack.

A deserialization of untrusted data vulnerability with a malicious payload can allow an unauthenticated remote code execution (RCE).

A vulnerability in infiniflow/ragflow versions v0.12.0 allows for remote code execution. The RPC server in RagFlow uses a hard-coded AuthKey 'authkey=b'infiniflow-token4kevinhu'' which can be easily fetched by attackers to join the group communication without restrictions. Additionally, the server processes incoming data using pickle deserialization via `pickle.loads()` on `connection.recv()`, making it vulnerable to remote code execution. This issue is fixed in version 0.14.0.

Dell SRM versions prior to 4.5.0.1 and Dell SMR versions prior to 4.5.0.1 contain an Untrusted Deserialization Vulnerability. A remote unauthenticated attacker may potentially exploit this vulnerability, leading to arbitrary privileged code execution on the vulnerable application. The severity is Critical as this may lead to system compromise by unauthenticated attackers.

IBM WebSphere Application Server Network Deployment 7.0, 8.0, 8.5, and 9.0 could allow a remote attacker to execute arbitrary code on the system with a specially-crafted sequence of serialized objects from untrusted sources. IBM X-Force ID: 181228.

HTTPServerILServlet.java in JMS over HTTP Invocation Layer of the JbossMQ implementation, which is enabled by default in Red Hat Jboss Application Server <= Jboss 4.X does not restrict the classes for which it performs deserialization, which allows remote attackers to execute arbitrary code via crafted serialized data.

This vulnerability allows remote attackers to execute arbitrary code on affected installations of NEC ESMPRO Manager 6.42. Authentication is not required to exploit this vulnerability. The specific flaw exists within the RMI service. The issue results from the lack of proper validation of user-supplied data, which can result in deserialization of untrusted data. An attacker can leverage this vulnerability to execute code in the context of SYSTEM. Was ZDI-CAN-10007.

The JIRA Workflow Designer Plugin in Atlassian JIRA Server before 6.3.0 improperly uses an XML parser and deserializer, which allows remote attackers to execute arbitrary code, read arbitrary files, or cause a denial of service via a crafted serialized Java object.

Synacor Zimbra Collaboration Suite 8.7.x through 8.8.11 allows insecure object deserialization in the IMAP component.

An issue was discovered in the serialize-to-js package 0.5.0 for Node.js. Untrusted data passed into the deserialize() function can be exploited to achieve arbitrary code execution by passing a JavaScript Object with an Immediately Invoked Function Expression (IIFE).

ColdFusion versions Update 1 and earlier, Update 7 and earlier, and Update 15 and earlier have a deserialization of untrusted data vulnerability. Successful exploitation could lead to arbitrary code execution.

ColdFusion versions Update 3 and earlier, Update 10 and earlier, and Update 18 and earlier have a deserialization of untrusted data vulnerability. Successful exploitation could lead to arbitrary code execution.

An issue was discovered in Joomla! before 3.9.3. The phar:// stream wrapper can be used for objection injection attacks because there is no protection mechanism (such as the TYPO3 PHAR stream wrapper) to prevent use of the phar:// handler for non .phar-files.

An issue was discovered in NumPy before 1.16.3. It uses the pickle Python module unsafely, which allows remote attackers to execute arbitrary code via a crafted serialized object, as demonstrated by a numpy.load call. NOTE: third parties dispute this issue because it is a behavior that might have legitimate applications in (for example) loading serialized Python object arrays from trusted and authenticated sources.

A remote deserialization of untrusted data vulnerability in HPE Intelligent Management Center (IMC) PLAT version 7.2 E0403P06 was found.

An attacker could send a specifically crafted payload to the XML-RPC invocation script and trigger the unserialize() call on the "what" parameter in the "openads.spc" RPC method. Such vulnerability could be used to perform various types of attacks, e.g. exploit serialize-related PHP vulnerabilities or PHP object injection. It is possible, although unconfirmed, that the vulnerability has been used by some attackers in order to gain access to some Revive Adserver instances and deliver malware through them to third party websites. This vulnerability was addressed in version 4.2.0.

A Remote Code Execution vulnerability in HPE Intelligent Management Center (iMC) PLAT version 7.3 E0504P2 was found.

Dell EMC Storage Monitoring and Reporting version 4.3.1 contains a Java RMI Deserialization of Untrusted Data vulnerability. A remote unauthenticated attacker may potentially exploit this vulnerability by sending a crafted RMI request to execute arbitrary code on the target host.

The Java implementations of AMF3 deserializers in WebORB for Java by Midnight Coders, version 5.1.1.0, derive class instances from java.io.Externalizable rather than the AMF3 specification's recommendation of flash.utils.IExternalizable. A remote attacker with the ability to spoof or control an RMI server connection may be able to send serialized Java objects that execute arbitrary code when deserialized.

The option-tree plugin before 2.7.3 for WordPress has Object Injection because the + character is mishandled.

The option-tree plugin before 2.7.0 for WordPress has Object Injection by leveraging a valid nonce.

The Virim plugin 0.4 for WordPress allows Insecure Deserialization via s_values, t_values, or c_values in graph.php.

VMware vRealize Automation (7.3 and 7.2) and vSphere Integrated Containers (1.x before 1.3) contain a deserialization vulnerability via Xenon. Successful exploitation of this issue may allow remote attackers to execute arbitrary code on the appliance.

PharMetaDataInterceptor in the PharStreamWrapper (aka phar-stream-wrapper) package 2.x before 2.1.1 and 3.x before 3.1.1 for TYPO3 mishandles Phar stub parsing, which allows attackers to bypass a deserialization protection mechanism.

VMware vSphere Data Protection (VDP) 6.1.x, 6.0.x, 5.8.x, and 5.5.x contains a deserialization issue. Exploitation of this issue may allow a remote attacker to execute commands on the appliance.

A remote code execution vulnerability was identified in HPE Intelligent Management Center (IMC) PLAT earlier than version 7.3 E0506P09.

Unauthenticated RCE is possible when JMeter is used in distributed mode (-r or -R command line options). Attacker can establish a RMI connection to a jmeter-server using RemoteJMeterEngine and proceed with an attack using untrusted data deserialization. This only affect tests running in Distributed mode. Note that versions before 4.0 are not able to encrypt traffic between the nodes, nor authenticate the participating nodes so upgrade to JMeter 5.1 is also advised.

In Apache Solr versions 5.0.0 to 5.5.5 and 6.0.0 to 6.6.5, the Config API allows to configure the JMX server via an HTTP POST request. By pointing it to a malicious RMI server, an attacker could take advantage of Solr's unsafe deserialization to trigger remote code execution on the Solr side.